Seafood Research From Fish To Dish

Seafood
research
from fish
to dish
edited by:
J.B. Luten
C. Jacobsen
K. Bekaert
Quality, safety and processing A. Sæbø
of wild and farmed fish J. Oehlenschläger
Seafood research from fish to dish
binnenwerk.indd 1 21-08-2006 12:32:50

binnenwerk.indd 2 21-08-2006 12:32:50
Seafood research from
fish to dish
Quality, safety and processing of wild and
farmed fish
edited by
J.B. Luten
C. Jacobsen
K. Bekaert
A. Sæbø
J. Oehlenschläger
Wageningen Academic
P u b l i s h e r s
binnenwerk.indd 3 21-08-2006 12:32:50

This work is subject to copyright. All rights
are reserved, whether the whole or part of
the material is concerned. Nothing from this
ISBN: 978-90-8686-005-0 publication may be translated, reproduced,
e-ISBN: 978-90-8686-581-9 stored in a computerised system or published in
DOI: 10.3920/978-90-8686-581-9 any form or in any manner, including electronic,
mechanical, reprographic or photographic,
without prior written permission from the
Cover design: Oddvar Dahl, publisher, Wageningen Academic Publishers, P.O.
Fiskeriforskning Box 220, 6700 AE Wageningen, the Netherlands,
Cover photo: Frank Gregersen, www.WageningenAcademic.com
Fiskeriforskning
The individual contributions in this publication
and anyliabilities arising from them remain the
First published, 2006 responsibility of the authors.
The publisher is not responsible for possible

© Wageningen Academic Publishers damages, which could be a result of content
The Netherlands, 2006 derived from this publication.
binnenwerk.indd 5 21-08-2006 12:32:50
binnenwerk.indd 6 21-08-2006 12:32:50
Foreword
The route from fish to the final product ready to be eaten by the consumer is a long one.
Therefore product quality, safety and processing have always been important issues in seafood
research.
However, over the last years seafood research has undergone a number of changes. Integration
of various scientific disciplines in former traditional technological oriented seafood research has
taken place. The importance of consumer science in the seafood research area has increased.
Most consumers consider fish as healthy and nutritious. In spite of this, some European
countries experience a trend towards declining seafood consumption. It is therefore crucial to
understand consumer behaviour and well-being to seafood and to adapt seafood products to
consumer demands.
Scientific and technological developments in the field of food have led to a marked shift in
the way consumers deal with food and health. There is a growing awareness that the dietary
source and form of food may affect the overall health of the consumer. In case of seafood there
are established roles of vitamin D and calcium in bone health promotion and of omega-3 poly-
unsaturated fatty acids (PUFA) in reducing the risk of cardiovascular disease. At the same time,
PUFA also pose a great challenge to food and fish technologists. This is due to the fact that
PUFA are very susceptible to oxidation due to their polyunsaturated nature. Thus, to maintain
the healthy properties of the PUFA, lipid oxidation must be prevented in fish products and
fish oil enriched foods. Scientifically based knowledge about processing technology and lipid
chemistry is necessary to obtain this goal.
Provision of seafood from capture fish is declining and partly not sustainable. Seafood from
aquaculture can potentially overcome this problem. It can deliver a product of defined quality
and composition to the market in all seasons of the year enabling a greater penetration of
‘healthy foods’ in the diet of consumers. With increasing intensification the ability to determine
the quality of the product emerges in several ways, which lends itself to tailor-made seafood
products. In addition, high seafood quality should be linked to ethically acceptable husbandry
practices and aquaculture systems, in reality and as perceived by the consumers. Furthermore, it
is important to diversify farming to various white fish species, such as cod and carp. However,
these ‘new species’ are likely to be more susceptible to quality problems. Therefore new
knowledge about quality of farmed fish is essential.
In 2005 the unique opportunity was taken to combine two important international meetings
in the seafood research area. The Institute for Marine Resources & Ecosystem Studies (IMARES,
former RIVO)) and the Unit Animal Sciences – Fisheries (D-VI, former SFD) from the Belgium
Institute for Agricultural and Fisheries Research were the organising host institutes for 35th
annual meeting of the West European Fish Technologists Association (WEFTA). Karen Bekaert
(D-VI) and my self were in charge of the organisation. The decision to combine this event with
the yearly meeting of the European section of the American Oil Chemists Society was initiated
by Charlotte Jacobsen from the Danish Institute for Fisheries Research.
Quality, safety and processing of wild and farmed fish
binnenwerk.indd 7 21-08-2006 12:32:50

In this combined meeting we decided to focus on quality, safety and seafood consumer’s issues
and to give room for presentations from aquaculture. In this way we were able to cover the
scientific topics in the total seafood chain area from fish to dish.
With this book the editorial team is aiming to reach a broad group of readers, from students,
experienced scientists to actors in the seafood chain. In the eight chapters of this book
scientists from various disciplines address the advances in seafood research with respect
to quality, safety, consumer’s demand and processing of wild and farmed fish. Each chapter
contains applied research papers and research notes. Many papers are a reflection from on-going
national and European collaborative projects. Experts have refereed all papers. The editorial
team of this book wish to express their gratitude for their referee work.
My special thanks go to my co-editors Charlotte Jacobsen, Karen Bekaert, Asgeir Sæbø (Natural
Lipids LtD AS, Norway) and Jörg Oehlenschläger (Federal Research Centre for Nutrition and
Food, Germany). The two intensive editorial working days of Charlotte, Karen, Jörg and I in
Madrid showed the good team-spirit we have had from the beginning to realise this book.
I would like to thank my two colleagues at Fiskeriforksning, Tromsø, Norway: Oddvar Dahl for
his creative work to design the cover of the book and Frank Gregersen for providing the photos
for the cover.
Mieke van der Putte (IMARES) is thanked for her help in collecting all the papers.
Last but not least my thanks to the management of Fiskeriforskning for giving me the opportunity
to realise this book here in the beautiful Northern part of Norway within the context of my
combined job at Fiskeriforskning and IMARES
Joop Luten
Fiskeriforskning, Tromsø, Norway

IMARES, IJmuiden, The Netherlands
binnenwerk.indd 8 21-08-2006 12:32:50

Table of contents
Foreword 7
Chapter 1: Nutritional properties and oxidation of marine lipid 15
Marine phospholipids (MPL): Resources, applications and markets 17

Erik Løvaas
Effects of dietary triacylglycerol structure on plasma and liver lipid levels in rats fed
low-fat diets containing n-3 polyunsaturated fatty acids of marine origin 29
Trine Porsgaard, Xuebing Xu and Huiling Mu
Cholesterol content in seafood, data from the last decade: A review 41

Jörg Oehlenschläger
Capelin oil for human consumption 59

Margrét Bragadóttir, Ása Þorkelsdóttir, Irek Klonowski and Helga Gunnlaugsdóttir
Oxidative stability of fish oil enriched yoghurts 71

Charlotte Jacobsen, Mette B. Let, Gitte Andersen and Anne S. Meyer
Antioxidant synergy effect between α-tocopherol and ascorbate on the autoxidation

of liposomes 87
Harald Barstad, Anne Cecilie Alvik and Erik Løvaas
Effect of grape antioxidant dietary fibre on the prevention of lipid oxidation in minced
fish: Evaluation by different methodologies 95
Isabel Sánchez-Alonso, Antonio Jiménez-Escrig, Fulgencio Saura-Calixto and
Javier Borderías
Effects of antioxidants on copper induced lipid oxidation during salting of cod (Gadus
morhua L.) 105
Kristin Lauritzsen and Ragnar L. Olsen
Rapid assessment of storage quality of cliff-fish from saithe by fluorescence spectroscopy 119
Agnar Sivertsen, Kristin Lauritzsen, Annette Veberg and Jens Petter Wold
Natural antioxidants in cod liver oil: Pitfalls during oxidative stability assessment 127
Eva Falch, Anders Øverby and Turid Rustad
Quality, safety and processing of wild and farmed fish
binnenwerk.indd 9 21-08-2006 12:32:50

Chapter 2: Quality of farmed fish 137
Pre-slaughter starvation of farmed Atlantic cod fed vegetable proteins: Effects on

quality parameters 139
Gunn Berit Olsson, Bjørn Gundersen and Margrethe Esaiassen
The effect of pre rigor processing of cod (Gadus morhua L.) on quality and shelf life 149
Torbjørn Tobiassen, Leif Akse, Kjell Midling, Kåre Aas, Reidun Dahl and Guro Eilertsen
Gelatinolytic activity in muscle of farmed and wild Atlantic cod (Gadus morhua)
related to muscle softening 161
Gunn Berit Olsson, Marie Cooper, Tone Jakobsen Friis and Ragnar L. Olsen
Relevance of storage temperature for contraction and gaping of pre rigor filleted
farmed cod (Gadus morhua L.) 173
Turid Mørkøre, Sølvi J. Hansen and Kjell-Arne Rørvik
Brining of farmed cod fillets: Effects on quality aspects 185

Margrethe Esaiassen, Grete Lorentzen, Reidun Dahl, Guro Eilertsen, Bjørn Gundersen,
Torbjørn Tobiassen and Morten Sivertsvik
Enrichment of functional selenium in farmed African catfish (Clarias goriepinus) by

dietary modulation 193
Joop Luten and Edward Schram
Comparison of commercial and experimental slaughter of farmed carp (Cyprinus carpio)

with respect to development of rigor mortis and flesh quality 201
Hans van de Vis, Henryk Bialowas, Maciej Pilarczyk, Marcel Machiels, Henny Reimert,
Martine Veldman and Bert Lambooij
Chapter 3: Consumers knowledge, perception, need for information about seafood 211
Too much or too little information? The importance of origin and traceability for
consumer trust in seafood in Norway and Germany 213
Arne Dulsrud, Hans Martin Norberg and Thorsten Lenz
Consumer knowledge and interest in information about fish 229

Zuzanna Pieniak, Wim Verbeke, Karen Brunsø and Svein Ottar Olsen
The importance of bacalhau consumption in Portugal and a preliminary product

consumer test in Lisboa 241
Jens Østli, Morten Heide, Mats Carlehög and Guro Eilertsen
Traceability: Simulated recall of fish products 251

Kine Mari Karlsen and Gunnar Senneset
10 Seafood research from fish to dish
binnenwerk.indd 10 21-08-2006 12:32:50

Chapter 4: Quality seafood 263
Effect of catch location, season and quality defects on value of Icelandic cod (Gadus
morhua) products 265
Sveinn Margeirsson, Allan A. Nielsen, Gudmundur R. Jonsson and Sigurjon Arason
Protein degrading enzymes in herring (Clupea harengus) muscle and stomach 275
Hanne Solvang Felberg and Iciar Martinez
Characterization of the quality of frozen sea products commercialized in Portugal 283

Susana Gonçalves, Helena Lourenço, Claudia Afonso, Maria Fernanda Martins and Maria
Leonor Nunes
Development of a Quality Index Method scheme to evaluate freshness of tub gurnard

(Chelidonichthys lucernus) 289
Karen Bekaert
Spoilage of herring (Clupea harengus) under chilled conditions and offal under ambient
and chilled conditions when treated with commercial preservative or additive products 297
Michael Gallagher, Marianne Green and Frank Trearty
Effect of freezing and different heat treatments on Anisakis larvae: Preliminary study 309
Margarita Tejada, Maria Teresa Solas, Alfonso Navas and Angel Mendizábal
Use of “filtered smoke” and carbon monoxide with fish: A review 317
Reinhard Schubring
Chapter 5: Microbial quality of seafood 347
Assessing the time-temperature history of aseptically filleted cod from predicted and
measured contents of sulphide producing bacteria 349
Taran Skjerdal and Sissel Ranneklev
Growth kinetic of Staphylococcus aureus during cod (Gadus morhua) rehydration 359
Sónia Pedro, Ireneu Batista and Maria Leonor Nunes
Aeromonas spp. and potential indicators in mussels in Northern Greece 365

Amin Abrahim, Eleni Iossifidou, Nikolaos Soultos, Dimitrios Koutsopoulos, Ioannis
Tzavaras and Pavlos Koidis
Characterisation of Vibrio parahaemolyticus isolated from seafood products 371

Sónia Pedro, Susana Castro, Marisa Santos and Ana Teia Santos
Inactivation of Listeria innocua isolated from fish products by pulsed light 381
Amaia Lasagabaster and Iñigo Martínez de Marañón
Quality, safety and processing of wild and farmed fish 11
binnenwerk.indd 11 21-08-2006 12:32:51

Antimicrobial effect of chitosan on micro-organisms isolated from fishery products 387
Ziortza Cruz, Helene Lauzon, Juan Carlos Arboleya, Maider Nuin, Iñigo Martínez de
Marañón and Felix Amarita
Selection of psychotrophic bacteria active against spoilage and pathogenic micro-

organisms relevant for seafood products 395
Sébastien Matamoros, Marie-France Pilet, Frédérique Gigout, Hervé Prevost and
Françoise Leroi
Selection of non-tyramine producing Carnobacterium strains for the biopreservation of

cold-smoked salmon 403
Anne Brillet, Sébastien Matamoros, Christine Blanchet-Chevrollier, Françoise Leroi,
Hervé Prevost and Marie-France Pilet
Chapter 6: Full utilisation of the catch 411
Production and characterization of a sockeye salmon (Oncorhynchus nerka) liver meal

and dried powders from stickwaters 413
Sébastien Plante, Alexandra C.M. Oliveira, Scott Smiley and Peter J. Bechtel
Evaluation of antioxidant activities in by-product hydrolysates, fractionation

and concentration of active molecules using separation technologies (ultra- and
nanofiltration technologies) 419
Aurélie Chabeaud, Laurent Vandanjon, Pascal Jaouen, Patrick Bourseau, Charles
Delannoy, Ragnar Johannsson, Gudjon Thorkelsson and Fabienne Guerard
Acid and alkaline-aided protein recovery from Cape hake by-products 427
Ireneu Batista, Carla Pires, Rute Nelhas and Vera Godinho
Quality evaluation of silver smelt (Argentina silus) and its suitability for seafood
products: Compilation of results from three European research centres 439
Iren Skjåstad Stoknes, Jörg Oehlenschläger and Ronan Gormley
Chapter 7: Nutrients and contaminants in seafood 457
Effect of sample preparation with or without shell liquor on contents and retentions of
macro and micro elements in three species of bivalve molluscs 459
Anna Badiani, Silvia Testi, Sergio Ghidini, Mavina Silvi, Chiara Foschi and
Pier Paolo Gatta
Seasonal variation of proximate and fatty acid class composition of wild and cultured
Brown Meagre (Sciena umbra), a new species for aquaculture 469
Şükran Caklı, Tolga Dincer, Aslı Cadun, Şahin Saka and Kürşat Firat
Composition and nutritional value of fishery products consumed in Portugal 477

Maria Leonor Nunes, Narcisa Bandarra, Luisa Oliveira, Ireneu Batista and
Maria Antónia Calhau
binnenwerk.indd 12 21-08-2006 12:32:51

Polychlorinated biphenyls and organochlor pesticides in brown shrimp (Crangon
crangon) of the Belgian continental shelf 489
Marc Raemaekers, Sabrine Derveaux and Koen Parmentier
Concentrations of mercury, lead and cadmium in bivalves from the Portuguese coast 497
Helena Lourenço, Carmen Lima, Ana Oliveira, Susana Gonçalves, Claudia Afonso, Maria
Fernanda Martins and Maria Leonor Nunes
Contaminant metals in cod products 503

Claudia Afonso, Helena Lourenço, Maria Fernanda Martins and Maria Leonor Nunes
Chapter 8: Advanced methods for quality determination of seafood 507
Instrumental quality control of stockfish 509

Heidi Nilsen, Agnar Sivertsen, Sjurd-ur Joensen, Ingebrigt Bjørkevoll and Karsten Heia
Time temperature indicators as quality and shelf life indicators for fresh turbot (psetta
maxima) 519
Maider Nuin, Begoña Alfaro, Ziortza Cruz and Nevea Argarate
Instrumental colour analysis of Atlantic salmon (Salmo salar L.) muscle 525
Lars Helge Stien, Asbjørn Høyem Amundsen, Turid Mørkøre, Simon Nesse Økland and
Ragnar Nortvedt
Revision of analytical methodologies to verify the production method of fish 541

Iciar Martinez
Detection of noroviruses: Comparison of viral extraction methods in bivalve molluscs 551

Leen Baert, Lieselot Bontinck, Mieke Uyttendaele and Johan Debevere
Keyword index 561
binnenwerk.indd 13 21-08-2006 12:32:51

binnenwerk.indd 14 21-08-2006 12:32:51
Chapter 1:
Nutritional properties and oxidation of marine lipid
Seafood is a very important source of healthy lipids. Particularly the long chain omega-3 fatty
acids have been shown to have numerous health effects in the human body. At the same time,
marine lipids also pose a great challenge to food and fish technologists. This is due to the fact
that omega-3 fatty acids are very susceptible to oxidation due to their polyunsaturated nature.
Thus, to maintain the healthy properties of the omega-3 fatty acids, lipid oxidation must be
prevented in fish products and fish oil enriched foods. Scientifically based knowledge about
processing technology and lipid chemistry is necessary to obtain this goal.
This chapter covers several of the aspects mentioned above. For example, the nutritional
properties of marine phospholipids originating from krill are discussed together with possible
nutritional and technological applications of marine phospholipids. The effect of using so-called
structured lipids containing omega-3 fatty acids on the plasma and liver levels of triglycerides
and cholesterol is reported in another paper. A third paper concerns the level of cholesterol in
seafood, which has been determined in a very large number of fish and shellfish.
The possibilities of enriching foods with omega-3 fatty acids with the aim of increasing the
consumer’s intake of these healthy fatty acids are reported in two papers. One dealing with fish
oil enriched mayonnaise and the other dealing with fish oil enriched yoghurt.
Protection against lipid oxidation by the addition of antioxidative compounds to fish oil,
liposomes or fish products is reported in a number of different papers. Promising results with
natural compounds such as rosemary extract, tea catechins, tocopherol and grape fibres are
presented. In addition, synergistic effects between tocopherol and ascorbate are shown in
another paper. Finally, both pro-oxidative and antioxidative effects of ascorbate in salted cod
muscle are also presented.
When evaluating lipid oxidation, it is necessary to use reliable and reproducible methods. A wide
range of standard methods have been available for many years. Recently developed methods
include fluorescence spectroscopy for measuring TBARS and electron spin resonance (ESR) for
measuring free radicals. Two papers discuss the applicability of these two methods and the
pitfalls for the ESR method is also dealt with.
binnenwerk.indd 15 21-08-2006 12:32:51

binnenwerk.indd 16 21-08-2006 12:32:51
Marine phospholipids (MPL): Resources, applications and markets
Erik Løvaas
University of Tromsø, Department of Pharmacy, N-9037 Tromsø, Norway
Abstract
Marine phospholipids (MPL) are extremely valuable components that can be applied within
diverse areas like nutrition, pharmacy, medicine and basic research. MPL can be extracted from
diverse marine raw material, in particular from roe and krill but also from by-products from the
fishery industry. This review describes the chemical characteristics of marine phospholipids,
their natural resources, and some present and future applications. The possible market size for
MPL is estimated by using the omega-3 market as an indicator.
Keywords: marine phospholipids, omega-3 fatty acids, nutraceuticals, pharmaceuticals,

antioxidants
Introduction
There is a current interest in omega-3-lipids with a high content of EPA (C20:5 n-3) and
DHA (C22:6 n-3). The interest is based on the observations that omega-3 fatty acids have
a number of interesting biological actions, like anti-inflammatory effects (James and others
2000; Ross 1999; Adam 2003), prevention of cardiovascular diseases (Siscovick and others
2000; Marckmann and others 1999; Anonymous 1999), improved neuronal function (Loudes and
others 1983), learning behaviour (Wainwright 2000; Zimmer and others 2000; Uauy and others
1996) and visual function (Horrocks and others 1999; Jacobson 1999), retardment of cognitive
impairment in old people (Kalmijn and others 1997; Bourre 2004; Haag 2003; Horrocks and
others 1999; Markesbery 1997; Youdim and others 2000), and preventive effects of breast
(Bartsch and others 1999) and prostate cancer (Leitzmann and others 2004). There is evidence
to suggest that omega-3 LCPUFA (long chain polyunsaturated fatty acids), DHA in particular,
increase the sensitivity of cancer cells to anticancer agents and pro-oxidants (Timmer-Bosscha
and others 1998; Germain and others 1998).
The demand for polyunsaturated fatty acids in health related products is increasing (Anonymous
2005), and there are ongoing efforts to identify methods for supplying fish oils in forms, which
are compatible with established consumer habits and consumer trends. Late developments have
been to produces soft capsules and to produce microencapsulated oils that can be mixed with
dry material (like flour) and used in composite products (like bread).
Phospholipids extracted from certain marine organisms have a high content of omega-3-fatty
acids, with EPA and DHA as main components. Marine phospholipids have potential applications
in human and animal nutrition, in pharmacology and in drug delivery. Liposomes made of
marine phospholipids may also serve as convenient targets for oxidation processes, and can be
used for high throughput screening of antioxidants.
binnenwerk.indd 17 21-08-2006 12:32:51

Erik Løvaas
Characteristics of marine phospholipids (MPL)
Phospholipids are the building blocks of biological membranes, and constitute the border
between the internal milieu of the cell and the external environment. Phospholipids are the
most fundamental prerequisite for organizing matter into living, self-reproducing entities. When
phospholipids are mixed with water they spontaneously aggregate into micro-spheres (liposomes)
where a phospholipid membrane encloses an internal space (Lasic and Papahadjopoulos 1998).
Phospholipids have numerous applications, some related to their physical properties, and some
to their biochemical properties.
Phospholipids are formed from four individual components: (1) fatty acids, (2) a negatively
charged phosphate group, (3) an alcohol and (4) a backbone (Figure 1). There is variability
in the alcohol component and in the fatty acid components, giving rise to a great number of
different phospholipid species. The biggest variability is in the fatty acid components.
Phospholipids have a hydrophilic head and a hydrophobic tail. The hydrophobic tail tends
to aggregate itself in non-polar associations, while the hydrophilic head group forms stable
interactions in polar (aqueous) environments. When phospholipids are introduced into an
aqueous environment they spontaneously form nano-structures (liposomes). Liposomes are
versatile structures that are applied as delivery vehicles (e.g. site specific drug delivery agents).
They are also realistic models for biological membranes. In particular one should note that fatty
acids attached to a phospholipid are “water soluble”.
The term “Marine Phospholipids” is used to describe a subgroup of phospholipids that has
a high content of marine omega-3 fatty acids. Such phospholipids are abundant in marine
organisms. In particular they are present in high concentrations in krill and in fish roe. They
are also present in high amounts in the human brain (Svennerholm 1968; Clandinin 1999; Erren
and others 2004) and in specific cell membranes (Uauy and others 2001).
Glycerol
C O C H2
O Choline
O O C H3
C CH
+
C H2 O P O C H2 C H2 N C H3
O
–
O C H3
Fatty acid
Phosphate
Hydrophobic tail Polar head group
Figure 1. Structure of a phospholipid.
binnenwerk.indd 18 21-08-2006 12:32:52

For many years there have been efforts to produce phospholipids with a high content of omega-3
fatty acids. One method has been to force feed hens with marine oils, and subsequently isolate
the phospholipids from the egg yolk (Surai and others 2000). Such drastic methods generates
phospholipids that maximum contain a few percent of omega-3 fatty acids. In contrast, marine
phospholipids from roe typically contain 50% omega-3 fatty acids (Figure 2). The omega-3
fatty acids are generally esterified in the 2-position of the phospholipid (Kuksis 1972; Amate
and others 1999).
A Fatty acid composition

18:1
of DHA Egg Lecithin
16:0
18:2
18:0
22:1 22:6
5 6 7 8 9 10 11 12 13 14 15 16 17
16:0
Fatty acid composition
B
of cod roe
18:1
22:6
20:5
16:1
14:0 18:3
18:2
18:0
5 6 7 8 9 10 11 12 13 14 15 16 17
Figure 2. Fatty acid profile of (A) DHA egg Lecithin, and (B) marine phospholipids from cod roe.
binnenwerk.indd 19 21-08-2006 12:32:54

Erik Løvaas
Natural resources of MPL
Marine phospholipids can be isolated from a variety of raw materials. Current raw materials
include krill, fish roe and fish meal, which all have characteristic properties making them more
or less suitable for phospholipid extraction. Important factors to consider when choosing a
raw material are:
• Price of raw material
• Level of marine phospholipids in raw material
• Quality and freshness of raw material
• Catch volume and stability of catch over an extended period
• Available process technology
• Pollutants in raw material
• Laws and regulations (e.g. relating to “novel food”)
Antarctic krill (Euphausia superba) is a small shrimp like zooplankton (crustacean) that plays
a key role in the Antarctic food web. Krill is one of the biggest biological resources on earth.
The biomass of Antarctic krill is estimated to 1.350 million tons, about five times the weight
of the human population (Stevens 1995). Krill swarms can occupy an area of 500 sq km, and
hold up to 20,000 individuals per cubic meter. Biomass densities attain 10 – 16 kg/m3. Unlike
other Euphausiid species, Euphausia superba feeds almost exclusively upon phytoplankton.
This contributes to the very high abundance and potential catch of krill, since there are
comparatively small losses between the primary production and the potential harvest by man.
Krill occurs in groups or large swarms and occupies a niche similar to that of the herring in the
North Atlantic, since large schools of pelagic fish are absent. They attain a size of 7 cm (normal
range 4.5 – 5.0 cm) and feed primarily on phytoplankton or sea ice algae.
The catch limit on krill in the South Atlantic Region is set to 4 million tons (Hewitt and others
2004), but the annual krill catch has never amounted to more than 550,000 tons (Dommarsnes
and others 2004; Anonymous 2006). It is still the largest crustacean fishery in the world. The
catch is by trawls with small mesh size and big openings, and catch efficiency can be 5 – 20
tons hour. The price for krill is approx. 700 US$/ton delivered at the Falkland Islands. Transport
from South America to Canada or North Europe is about 150 US$/ton.
There is a considerable variation in the amount of lipid and the lipid composition of krill. The
lipid content may vary between 12.5 and 0.5% (w/w wet mass) with about 25% as phosphatidyl
choline (Pond and others 1995). Specific production technologies give rise to a phospholipid
enriched oil, containing as much as 40 – 60% phospholipids, of which phosphatidyl choline is
the dominating (>75%) species. The phospholipids have a high level of omega-3 fatty acids
(approx. 40%), and the EPA:DHA ratio is 25:15 (which makes it similar to 18:12 oil from South
America). Such oils are used as dietary supplements.
Recently roe has been identified as a good source for marine phospholipids. Developments on
process technology will also make fish meal attractive in the future.
Applications of MPL
Marine phospholipids have potential applications in diverse areas like nutraceuticals (functional
food, dietary supplements, sports nutrition, maternal nutrition, and baby food), pharmaceuticals
binnenwerk.indd 20 21-08-2006 12:32:54

(liposomes – site specific drug delivery, clinical nutrition), cosmeceuticals (liposomes), and
biochemistry (model membranes, liposomal test systems for antioxidants).
The word “potential” is used deliberately, because many of the mentioned applications have not
been tested at the present time. Marine phospholipids are new at the market place, and their
range of applications has yet to be determined. However, intensive development work is going
on, and it is probable that soon a number of new products will become available.
Test system for antioxidants

The best-documented applications of MPL are related to liposomes. Liposomes made of MPL
have been developed as a test system for antioxidants, and as model systems for oxidation of
biological membranes. Some basic principles for these developments will be reviewed here.
Liposomes made from marine phospholipids oxidize fast in the presence of trace amounts of
Fe2+, and the oxidation can be followed by UV/VIS spectroscopy.
A summary of the assay is presented in Figure 3a and is reviewed by Barstad (2006). Basically, the
liposomes are prepared by mixing MPL and lipid soluble antioxidants in a common organic solvent
(e.g. methanol:chloroform). The ratio of antioxidant to phospholipid is typically 1:100 (mol:mol).
Buffer is added after removal of the solvent under a stream of nitrogen, and the phospholipids
are allowed to swell at ambient temperature for 60 minutes. Shaking the solution produce milky
white multilamellar vesicles (MLV) (Chatterjee and others 1988), which are sonicated to give
small unilamellar vesicles (SUV). Liposome oxidation is then initiated by addition of a catalytic
Step 1: Preparation of liposomes incorporated with antioxidants
100 µl PC-MPL (10 mg/ml CHCl3) 1250 nmol

10 µl antioxidant (1 mM) in CHCl3 10 nmol
CHCl3 removed under a stream of N2 at 40°C
6.0 ml 50 mM MES-buffer, pH 5.5 0.167 mg PC-MPL/ml
Shake up the solution, flush with N2 and hydrate for 2 hours at room temperature
Sonicate for 20 seconds with a RMS effect of 2-3 watt
Flush the tubes with N2 to keep the liposomes intact until the time of measurement
Step 2: Measurement of lipid peroxidation in the plate reader
280 µl liposomes/antioxidant
20 µl Fe(II) (150 µM)
Read absorbance at 232 nm
Figure 3a. Outline of procedures for the liposome antioxidant assay.
binnenwerk.indd 21 21-08-2006 12:32:55

Erik Løvaas
amount of Fe2+, and the oxidation is monitored by following the development of dienes (UV/VIS
measurements at 232 nm). By doing the assay in a plate reader it is possible to investigate a
great number of samples at the same time. As an example is shown the simultaneous testing of
13 antioxidants, all added in the same concentration (Figure 3b).
The system has a number of merits that makes it an attractive method for screening of new
compounds for antioxidative effects:
• It is easy to prepare liposomes that have lipid soluble test compounds incorporated into a
membrane. The test system does not experience the same problems as do for example LDL,
where lipid soluble antioxidants are difficult to incorporate.
• Liposomes are realistic models for biological membranes, and thus represent a test system
which is more likely to pick up antioxidative effects which are relevant in medicine and
pharmacology.
• Antioxidative effects can be measured in true time as formation of dienes can be measured
by UV-spectroscopy. No derivatization is necessary, or any need to manipulate samples
prior to measurement. This increases the reliability of the method, and makes it useful as
a routine screening assay.
• Antioxidants are incorporated into the membrane in a predetermined stoichiometry.
By manipulating the fluidity of the membrane the effect of dynamic movements of the
membrane components (lipids and antioxidants) can be studied. Detailed information on
the mechanistic relationships can thus be obtained.
• Liposomes represent an interphase where synergistic effect of water-soluble and lipid soluble
antioxidants conveniently can be studied. The system is well suited for studying complex
relationships between antioxidative compounds.
• Oxidation is initiated by addition of a catalytic amount of initiator (Fe2+), and a free radical
process is initiated. Different types of antioxidants (including both metal chelators and free
radical scavengers) retard the oxidation process in a concentration dependent fashion. The
extent and mode of inhibition is characteristic of the antioxidant under investigation.
Nutraceutical applications
There is an increasing attention on marine phospholipids because of their potential nutritional
value and their unique biophysical properties. Central to this interest is the fact that enzymes
that are involved in lipid metabolism (lipases) essentially acts in an aqueous environment.
Water soluble lipases have better access to water-soluble substrates (like phospholipids) than to
water insoluble substrates (like triglycerides). In sum this means that omega-3 fatty acids from
marine phospholipids are more easily accessible for catabolic processes than triglycerides.
The present imbalance in the fatty acid intake represent serious trouble for the health situation
in the general population, and there is a strong desire to improve the situation by introducing
new products on the market with a high level of omega-3 fatty acids and low level of omega-6
fatty acids. Numerous studies have demonstrated that omega-3 fatty acids have positive effects
on preventing or curing diseases like heart disease, cancer, rheumatoid arthritis, diabetes,
ulcerative colitis, allergies, eczema, skin thickening, weight gain and a host of other diseases.
Researchers are now also linking inadequate intake of these omega-3 fats in pregnant women
to premature birth and low birth weight (Al and others 2000; Crawford 2000; Koletzko 2002;
Smuts and others 2003), and to hyperactivity in children (Burgess and others 2000).
binnenwerk.indd 22 21-08-2006 12:32:55

Figure 3b. Simultaneous testing of antioxidants by the liposome diene conjugation assay. The assay is run
at room temperature (25 °C).
Pregnancy nutrition and infant formula
Important brain growth occurs during pregnancy and the first few years of life, and there are
several nutritional building blocks that are vital for brain growth and visual function (Bryan and
others 2004; Smuts and others 2003; Lauritzen and others 2001; Crawford 2000; Neuringer 2000;
Uauy and others 2000; Carlson 1999). The most important of these are omega-3 fatty acids (DHA)
and omega-6 fatty acids (arachidonic acid, or ARA). Many expectant mothers are not aware that
what they eat makes a substantial difference in the health of their unborn child.
The DHA content of breast milk depends on the mother’s diet. Most pregnant women have a
low intake of omega-3 fat, which ultimately may have consequences for the unborn and the
suckling child. It is thus advised that the expectant mother should eat fish several times a week
to increase brain building DHA, or alternatively that dietary supplements should be considered
when fish intake is not sufficient. Marine phospholipids should be likely candidates for such
products.
There has long been a concern that lipids in infant formulas may be absorbed to a lesser extent
than breast milk. It has been suggested that absorption of fish oil in pre-term infants may be
poor as the amount of DHA required in formulas was almost a magnitude greater than usually
provided by human milk (Liu and others 1987). A low absorption means that more lipids are
left in the intestine, giving rise to adverse effects. In particular more triacylglycerols (TAG) and
binnenwerk.indd 23 21-08-2006 12:33:49

Erik Løvaas
fatty acids will appear in the colon and lead to conditions like intestinal obstruction, increase
in colon fatty acid soap content, and altered microbial composition (Ling and others 1997;
Verkade and others 1991).
Addition of nutritional marine phospholipids to infant formula may have several advantageous
effects in neonates. Firstly, there is evidence that TAG absorption may be impaired when the
supply of exogenous PL is insufficient for micelle formation during fat absorption (Levy and
Roy 2005). Secondly, nutritional phospholipids would only require the action of phospholipase
A2 on one fatty acid to produce free fatty acids and 1-lysophospholipid that would be readily
absorbed, utilizing more of the dietary lipid, leaving fewer lipids left in the intestine. In
premature infants it has been found that DHA from egg PL was better absorbed than DHA from
breast milk and TAG from single cell sources. Improved absorption of lipid and long chain PUFA
from egg phospholipids has also been shown in other studies (Amate and others 2001; Makrides
and others 2002). It was also found that giving omega-3 enriched eggs (presumably high
contents of PL DHA) to infants of 6-12 months of age increased erythrocyte DHA concentration
to 30-40% above that of infants fed breast milk and can be used to maintain a high omega-
3 PUFA level. Phospholipids are also required for intestinal lipoprotein formation and lipid
distribution outside the enterocytes (Tso and others 1984). Lack of nutritional PL could lead
to impaired enterocyte lipoprotein synthesis causing lipid accumulation in the cells.
Omega-3 and the brain
Though the brain is able to make some of what it needs, the brain is surprisingly dependent
upon raw materials from the diet, notably fatty acids (Bryan and others 2004; Finley and others
2001; Lauritzen and others 2001; Innis 1994). If the right fat is not supplied, brain structure
and brain function is altered. The brains requirement for highly specified fats, coupled with
the poor dietary choices of most people living on modern diets, has created a problem in our
society. Particular interest has been ascribed to omega-3 fatty acids. Recent findings show the
positive effects of these fatty acids on:
• Bipolar depression. It is known that EPA and DHA gives a significant reduction in the
symptoms of depression (Logan 2004)
• Age-related memory loss and Alzheimer’s disease. It has been shown that adults with
dementia, Alzheimer’s and cognitive problems have a low level of DHA in the blood, and it
is suggested that deficiency of these fatty acids in the diet may be a risk factor for cognitive
impairment (Bryan and others 2004; Markesbery 1997; Calon and others 2005; Morris and
others 2003)
• Reading and learning problems in children. Children with fatty acid deficiency have been
found to have poorer reading abilities (Burgess and others 2000; Hals and others 2000).
• Intelligence in children. Breastfed children were compared with formula fed children who
had none of the fatty acid DHA in their formula. The breast fed group had a significant
higher IQ score at 8 years of age (Koo 2003).
It is highly desirable to produce food ingredients that meet the needs of the consumers by
having an attractive texture and flavour profile and good oxidation stability, and at the same
time having a healthy composition.
binnenwerk.indd 24 21-08-2006 12:33:52

Market trends for MPL (human nutrition)
The market for MPL is still in its infancy. However, an increasing activity in the field is observed,
and a number of companies are preparing market introduction of either natural MPL, derivatives
of natural MPL, or synthetic MPL. The development of entirely new market segments, as well as
a penetration into existing markets on omega-3 products is expected. It seems reasonable to
assume that the MPL market will follow the general trends of omega-3 fish oils.
The global market for omega-3 fatty acids is estimated to 15-20,000 tons, derived from a total
world production of fish oil of approximately 300,000 tons/year. Of this 10-12,000 tons are
refined fish oils, 6-7,000 tons cod liver oil, 1000 tons tuna oil, and 300 tons DHA/algae oil.
About 1.500 tons of the market is concentrates with 50-70% omega-3 fatty acids, and a variable
ratio between EPA and DHA (Anonymous 2005; Anonymous 2003). A market share of 1% equals
a demand of 150 tons marine phospholipids/year.
The nutraceutical industry has identified “health” as one of the top 10 “mega trends” in the
years to come. The industry on a worldwide basis is presently valued to US$ 7.1 billion/year,
and is growing by 6.1% annually (Anonymous 2005). In North America and in Europe, fish oils
are the fastest growing supplement, with an estimated annual growth rate between 15 and
18%. From 1994, where the sales were approximately $27 million, to $190 million in 2003, the
overall growth was an impressive 603% (Anonymous 2003).
Conclusion
There is a search for new products. Marine phospholipids have a range of biochemical and
biophysical properties that is attractive for many applications and market segments, and certainly
new products that will find applications in food and pharma will be developed. At present
there is a need to scale up production of these compounds and to identify natural resources
that comply with regulatory issues. The business prospects for this class of compounds will
undoubtedly attract attention from a diverse group of actors, including producers, distributors,
pharmaceutical companies, researchers, and investors.
References
Adam O. 2003. Dietary fatty acids and immune reactions in synovial tissue. Eur J Med Res 8(8):381-387.
Al MDM, Van Houwelingen AC, Hornstra G. 2000. Long-chain polyunsaturated fatty acids, pregnancy, and pregnancy
outcome. Am J Clin Nutr 71(1):285S-291S.
Amate L, Ramírez M, Gil A. 1999. Positional analysis of triglycerides and phospholipids rich in long-chain polyunsaturated
fatty acids. Lipids 34(8):865-871.
Amate L, Gil A, Ramirez M. 2001. Feeding infant piglets formula with long-chain polyunsaturated fatty acids as
triacylglycerols or phospholipids influences the distribution of these fatty acids in plasma lipoprotein fractions.
J Nutr 131 (4):1250-1255.
Anonymous. 1999. Dietary supplementation with n-3 polyunsaturated fatty acids and vitamin E after myocardial
infarction: results of the GISSI-Prevenzione trial. Gruppo Italiano per lo Studio della Sopravvivenza nell’Infarto
miocardico. Lancet 354 (9177):447-455.
Anonymous. 2003. Rapport nr. 4613/111: Internasjonal markeds- og industrianalyse for marine ingredienser. RUBIN
Trondheim, Norway.
Anonymous. 2005. Report No. 1818: World Nutraceuticals. The Fredonia Group Inc Cleveland, USA.
binnenwerk.indd 25 21-08-2006 12:33:52

Erik Løvaas
Anonymous. 2006. Statistical Bulletin (1995 - 2004). (17), CCAMLR Hobart, Australia.
Barstad H, Alvik AC, Løvaas E. 2006. Antioxidant synergy effect betweenn α-tocopherol and ascorbate on the
autoxidation of liposomes. In: Luten JB, Jacobsen C, Bekaert K, Sæbø A, Oehlenschläger J, editors. Seafood
research from fish to dish: Quality, safety, and processing of wild and farmed fish. Wageningen: Wageningen
Academic Publishers. pp.87-94.
Bartsch H, Nair J, Owen RW. 1999. Dietary polyunsaturated fatty acids and cancers of the breast and colorectum:
emerging evidence for their role as risk modifiers. Carcinogenesis 20(12):2209-2218.
Bourre JM. 2004. Roles of unsaturated fatty acids (especially omega-3 fatty acids) in the brain at various ages and
during ageing. J Nutr Health Aging 8(3):163-174.
Bryan J, Osendarp S, Hughes D, Calvaresi E, Baghurst K, van-Klinken JW. 2004. Nutrients for cognitive development in
school-aged children. Nutr Rev 62(8):295-306.
Burgess JR, Stevens L, Zhang W, Peck L. 2000. Long-chain polyunsaturated fatty acids in children with attention-deficit
hyperactivity disorder. Am J Clin Nutr 71(1):327S-330S.
Calon F, Lim GP, Morihara T, Yang F, Ubeda O, Salem NJ, Frautschy SA, Cole GM. 2005. Dietary n-3 polyunsaturated
fatty acid depletion activates caspases and decreases NMDA receptors in the brain of a transgenic mouse model of
Alzheimer’s disease. Eur J Neurosci 22(3):617-626.
Carlson SE. 1999. Long-chain polyunsaturated fatty acids and development of human infants. Acta Paediatrica
Supplement 88:72-77.
Chatterjee SN, Agarwal S. 1988. Liposomes as membrane model for study of lipid peroxidation. Free Radic Biol Med
4:51-72.
Clandinin MT. 1999. Brain development and assessing the supply of polyunsaturated fatty acid. Lipids 34(2):131-
137.
Crawford MA. 2000. Placental delivery of arachidonic and docosahexaenoic acids: implications for the lipid nutrition of
preterm infants. Am J Clin Nutr 71(1):275S-284S.
Dommarsnes A, Iversen SA, Løbach T. 2004. Fiskerier i Antaktis. Havets ressurser 2004 160-163. Havforskningsinsituttet
Bergen, Norway.
Erren TC, Erren M. 2004. Can fat explain the human brain’s big bang evolution? Horrobin’s leads for comparative and
functional genomics. Prostaglandins Leukot Essent Fatty Acids 70(4):345-347.
Finley JW, Shahidi F. 2001. The chemistry, processing, and health benefits of highly unsaturated fatty acids. An
overview. ACS Symposium Series 788:2-11.
Germain E, Chajes V, Cognault S, Lhuillery C, Bougnoux P. 1998. Enhancement of doxorubicin cytotoxicity by
polyunsaturated fatty acids in the human breast tumor cell line MDA-MB-231: relationship to lipid peroxidation.
Int J Cancer 75(4):578-583.
Haag M. 2003. Essential fatty acids and the brain. Can J Psychiatry 48(3):195-203.
Hals J, Bjerve KS, Nilsen H, Svalastog AG, Ek J. 2000. Essential fatty acids in the nutrition of severely neurologically
disabled children. Br J Nutr 83:219-225.
Hewitt RP, Watkins J, Naganobu M, Sushin V, Brierley AS, Demer D, Kasatkina S, Takao Y, Goss C, Malyshko A. 2004.
Biomass of Antarctic krill in the Scotia Sea in January/February 2000 and its use in revising an estimate of
precautionary yield. Deep Sea Research Part II: Topical Studies in Oceanography 51(12-13):1215-1236.
Horrocks LA, Yeo YK. 1999. Health benefits of docosahexaenoic acid (DHA). Pharmacol Res 40:211-225.
Innis SM. 1994. The 1993 Borden award lecture. Fatty acid requirements of the newborn. Can J Physiol Pharmacol
72:1483-1492.
Jacobson SW. 1999. Assessment of long-chain polyunsaturated fatty acid nutritional supplementation on infant
neurobehavioral development and visual activity. Lipids 34(2):151-160.
James MJ, Gibson RA, Cleland LG. 2000. Dietary polyunsaturated fatty acids and inflammatory mediator production.
Am J Clin Nutr 71(1 Suppl):343S-348S.
Kalmijn S, Feskens EJM, Launer LJ, Kromhout D. 1997. Polyunsaturated fatty acids, antioxidants, and cognitive function
in very old men. Am J Epidemiol 145:33-41.
Koletzko B. 2002. Parenteral lipid infusion in infancy: physiological basis and clinical relevance. Clin Nutr 21:53-65.
binnenwerk.indd 26 21-08-2006 12:33:52

Koo WW. 2003. Efficacy and safety of docosahexaenoic acid and arachidonic acid addition to infant formulas: can one
buy better vision and intelligence? J Am Coll Nutr 22(2):101-107.
Kuksis A. 1972. Newer developments in determination of structure of glycerides and phosphoglycerides. Progr Chem
Fats Lipids 12:1-163.
Lasic D, Papahadjopoulos D. 1998. Medical Applications of Liposomes. Amsterdam: Elsevier. 779p.
Lauritzen L, Hansen HS, Jørgensen MH, Michaelsen KF. 2001. The essentiality of long chain n-3 fatty acids in relation
to development and function of the brain and retina. Prog Lipid Res 40(1-2):1-94.
Leitzmann MF, Stampfer MJ, Michaud DS, Augustsson K, Colditz GC, Willett WC, Giovannucci EL. 2004. Dietary intake
of n-3 and n-6 fatty acids and the risk of prostate cancer. Am J Clin Nutr 80(1):204-216.
Levy E, Roy CC. 1989. Developmental aspects of intestinal lipoprotein synthesis and secretion In: Lebenthal M, editor.
Human Gatrointestinal Development. New York: Raven Press. p 491-502.
Ling SC, Weaver IT. 1997. The fate of fat in the infant’s colon. Q J Med 90:553-555.
Liu CC, Carlson SE, Rhodes PG, Rao VS, Meydrech EF. 1987. Increase in plasma phospholipid docosahexaenoic and
eicosapentaenoic acids as a reflection of their intake and mode of administration. Pediatr Res 22(3):292-296.
Logan AC. 2004. Omega-3 fatty acids and major depression: A primer for the mental health professionals. Lipids Health
Disease 3(25):1-8.
Loudes C, Faivre-Bauman A, Barret A, Grouselle D, Puymirat J, Tixier-Vidal A. 1983. Release of immunoreactive TRH in
serum-free cultures of mouse hypothalamic cells. Brain Res 285(2):231-234.
Makrides M, Hawkes JS, Neumann MA, Gibson RA. 2002. Nutritional effect of including egg yolk in the weaning diet of
breast-fed and formula-fed infants: a randomized controlled trial. Am J Clin Nutr 75(6):1084-1092.
Marckmann P, Grønbæk M. 1999. Fish consumption and coronary heart disease mortality. A systematic review of
prospective cohort studies. Eur J Clin Nutr 53(8):585-590.
Markesbery WR. 1997. Oxidative stress hypothesis in Alzheimer’s disease. Free Radic Biol Med 23:134-147.
Morris MC, Evans DA, Bienias JL, Tangney CC, Bennett DA, Wilson RS, Aggarwal N, Schneider J. 2003. Consumption of
fish and n-3 fatty acids and risk of incident Alzheimer disease. Arch Neurol 60(7):940-946.
Neuringer M. 2000. Infant vision and retinal function in studies of dietary long-chain polyunsaturated fatty acids:
methods, results, and implications. Am J Clin Nutr 71(1):256S-267S.
Pond D, Watkins J, Priddle J, Sargent J. 1995. Variation in the lipid content and composition of Antarctic krill Euphausia
superba at South Georgia. Mar Ecol Prog Ser 117:49-57.
Ross R. 1999. Atherosclerosis is an inflammatory disease. Am Heart J 138(5):S419-S420.
Siscovick DS, Raghunathan TE, King I, Weinmann S, Bovbjerg VE, Kushi L, Cobb LA, Copass MK, Psaty BM, Lemaitre R,
Retzlaff B, Knopp RH. 2000. Dietary intake of long-chain n-3 polyunsaturated fatty acids and the risk of primary
cardiac arrest. Am J Clin Nutr 71(1):208S-212S.
Smuts CM, Borod E, Peeples JM, Carlson SE. 2003. High-DHA eggs: Feasibility as a means to enhance circulating DHA
in mother and infant. Lipids 38(4):407-414.
Stevens JE. 1995. Dismantling the Myths of the Southern Ocean: The Secret Lives of Krill. Sea Frontiers 20(2):26-31.
Surai PF, MacPherson A, Speake BK, Sparks N-HC. 2000. Designer egg evaluation in a controlled trial. Eur J Clin Nutr
54(4):298-305.
Svennerholm L. 1968. Distribution and fatty acid composition of phosphoglycerides in normal human brain. J Lipid
Res 9(5):570-579.
Timmer-Bosscha H, de-Vries EG, Meijer C, Oosterhuis JW, Mulder NH. 1998. Differential effects of all-trans-retinoic acid,
docosahexaenoic acid, and hexadecylphosphocholine on cisplatin-induced cytotoxicity and apoptosis in a cisplantin-
sensitive and resistant human embryonal carcinoma cell line. Cancer Chemother Pharmacol 41(6):469-476.
Tso P, Drake DS, Black DD, Sabesin SM. 1984. Evidence for separate pathways of chylomicron and very low-density
lipoprotein assembly and transport by rat small intestine. Am J Physiol 247(6):G599-G610.
Uauy R, Peirano P, Hoffman D, Mena P, Birch D, Birch E. 1996. Role of essential fatty acids in the function of the
developing nervous system. Lipids 31 Suppl:S167-S176.
Uauy R, Hoffman DR. 2000. Essential fat requirements of preterm infants. Am J Clin Nutr 71(1):245S-250S.
Uauy R, Mena P. 2001. Lipids and neurodevelopement. Nutr Rev 59(8 Pt 2):S34-S46.
binnenwerk.indd 27 21-08-2006 12:33:52

Erik Løvaas
Verkade HJ, Hoving EB, Muskiet FA, Martini IA, Jansen G, Okke A, Vonk RJ, Bijleveld CM. 1991. Fat absorption in
neonates: comparison of long-chain-fatty-acid and triglyderide compositions of formula, feces, and blood. Am J
Clin Nutr 53:643-651.
Wainwright P. 2000. Nutrition and behaviour: the role of n-3 fatty acids in cognitive function. Br J Nutr 83(4):337-
339.
Youdim KA, Martin A, Joseph JA. 2000. Essential fatty acids and the brain: Possible health implications. Internat J
Develop Neurosci 18(4-5):383-399.
Zimmer L, Delpal S, Guilloteau D, Aioun J, Durand G, Chalon S. 2000. Chronic n-3 polyunsaturated fatty acid deficiency
alters dopamine vesicle density in the rat frontal cortex. Neurosci Lett 284(1-2):25-28.
binnenwerk.indd 28 21-08-2006 12:33:52

Effects of dietary triacylglycerol structure on plasma and liver lipid
levels in rats fed low-fat diets containing n-3 polyunsaturated fatty
acids of marine origin
Trine Porsgaard1, Xuebing Xu2 and Huiling Mu1
1Biochemistry and Nutrition Group, 2Food Biotechnology and Engineering Group, BioCentrum-DTU,
Technical University of Denmark, 2800 Lyngby, Denmark
Abstract
For three weeks rats were fed low-fat diets of which 25% was n-3 polyunsaturated fatty acids
(PUFAs). The n-3 PUFAs originated either from two structured triacylglycerols (TAGs) differing
in intramolecular structure or from fish oil. A group fed chow was included as control. Dietary
inclusion of n-3 PUFAs increased the level of n-3 PUFAs in plasma and liver. TAG structure
influenced the content of individual n-3 PUFAs meaning that enrichment of the sn-2 position
with a specific fatty acid increased the level of this fatty acid. Differences in dietary TAG
structure did not influence plasma and liver levels of TAG and cholesterol.
Keywords: cholesterol, n-3 polyunsaturated fatty acids, rats, structured lipids, triacylglycerols
Introduction
Marine oils contain n-3 polyunsaturated fatty acids (PUFAs), especially 20:5n-3 and 22:6n-3.
The sn-2 position of the glycerol moiety in fish oil is enriched with PUFAs, especially 22:6n-3,
whereas in oils of marine mammals, including whale and seal oil, this is reversed, i.e. there is
less PUFA enrichment in the sn-2 position (Ackman 1988). A high intake of n-3 PUFAs has been
associated with a beneficial effect on the plasma lipid profile, leading to a low incidence of
coronary heart disease (Kris-Etherton and others 2002). Furthermore, n-3 PUFAs were shown to
influence the brain and visual development during infancy (Uauy and others 2001) and to have
immunomodulating effects (Yaqoob 2003). In addition, positive effects of fish oil have been
reported on mental health (Peet and Stokes 2005) and in patients with rheumatoid arthritis
(Kremer 2000).
In this paper the term structured triacylglycerols (TAGs) will cover TAGs that have been modified
or restructured from natural oils and fats, and from fatty acids for specific nutritional purposes.
Structured TAGs can be produced by chemical modification or enzyme technology (Xu 2000). By
using enzymatic modification with lipases having specific selectivity, specific structured TAGs
can be produced, whereas randomized TAGs may be produced both by chemical and enzymatic
modification. TAGs containing both medium-chain fatty acids (MCFAs) and long-chain fatty
acids (LCFAs) on the same glycerol molecule combine the advantages of the easily digested and
absorbed MCFAs with the delivery of various PUFAs with specific effects in the body. TAGs with
this structure (MLM-type, M=MCFA, L=LCFA) were shown to improve the absorption of essential
fatty acids during malabsorption (Hubbard and Mckenna 1987; Christensen and others 1995b; Tso
and others 1999; Rubin and others 2000; Straarup and Høy 2000) and to increase the uptake of
PUFAs for tissue regeneration after surgery (Kenler and others 1996; Kruimel and others 2001).
binnenwerk.indd 29 21-08-2006 12:33:53

Studies have shown that the absorption of dietary TAGs in the immediate absorptive phase
depends on the intramolecular structure with enhanced absorption of fatty acids located
in the sn-2 position (Ikeda and others 1991; Christensen and others 1995a; Straarup and
Høy 2000; Porsgaard and others 2005b). This results from the preferential hydrolysis by the
pancreatic lipase in the digestive tract of fatty acids located in the sn-1,3 positions resulting
in sn-2 monoacylglycerols (sn-2 MAGs) and free fatty acids (FFA) (Mattson and Volpenhein
1964). Studies investigating the longer-term effects of structured TAGs have shown that the
intramolecular structure might influence plasma and liver lipid profiles in rats (Nagata and
others 2003, 2004), while no differences in amounts of 22:6n-3 in brain phospholipids of rat
pups from dams fed different dietary TAG structures and n-3 sources were observed (Hartvigsen
and others 2004).
The aim of the present study was to compare the effects of differences in dietary TAG structure
on liver and plasma lipid levels of TAG and cholesterol, and on the fatty acid compositions, when
n-3 PUFAs were included in the dietary oils. Due to the many positive results reported with n-3
PUFAs, it is of outmost importance to investigate if the effects of n-3 PUFAs are influenced by
their position on the glycerol backbone. For this purpose, two structured TAGs containing n-3
PUFAs and differing in intramolecular structure were produced by enzymatic interesterification
and included as part of low-fat diets for rats, while two other groups were fed fish oil or ordinary
rat chow. After 3 weeks feeding, the contents of TAG and cholesterol in plasma and livers were
measured together with the fatty acid compositions.
Materials and methods
Diets
Two specific structured TAGs, LML (L=LCFA, M=MCFA) and MLM, were produced by lipase-
catalysed interesterifications as described previously (Porsgaard and others 2005a). Safflower
oil was from Urtekram A/S (Mariager, Denmark), olive oil from FDB (Albertslund, Denmark),
and fish oil from Aarhus United A/S (Aarhus, Denmark). The overall composition of the diets
containing LML, MLM, or fish oil was as follows (in wt %): 56 corn starch (Bestfoods Nordic
A/S, Skovlunde, Denmark); 20 casein (Miprodan milk proteins, Arla Foods amba, Viby, Denmark),
10 sucrose (Danisco Sugar, Copenhagen, Denmark), 5 salt mixture (including trace elements),
4 fat, 4 cellulose powder (MN 100, Machery-Nagel GmbH & Co, Düren, Germany), 0.5 vitamin
mixture, and 0.5 choline chloride (Merck, Darmstadt, Germany). The vitamin and salt mixtures
were composed as described by Aaes-Jørgensen and Hølmer (1969).
The diets were balanced with respect to content of saturated, monounsaturated, n-6, and n-3
PUFA. The LML diet contained 53% LML, 21.5% safflower oil, and 25.5% olive oil; the MLM diet
52% MLM, 21.5% safflower oil, and 26.5% olive oil; and the fish oil diet 79% fish oil and 21%
safflower oil. The diets were powdered, stored at –20 °C and provided fresh every day. One group
of rats was fed ordinary pelleted rat chow (Altromin No. 1324, Chr. Petersen A/S, Ringsted,
Denmark) with 4 wt % fat. All diets and water were supplied ad libitum.
Animals
Thirty-two male specific-pathogen free Wistar rats (Taconic M & B A/S, Ll. Skensved, Denmark)
were included in the study. They were housed two per plastic cage in a temperature (21 °C)
and humidity (50%) controlled environment on a 12 hour – 12 hour light/dark cycle. They were
acclimatized to the housing conditions for 10 days, in which they all were fed the rat chow
binnenwerk.indd 30 21-08-2006 12:33:53

Effects of dietary triacylglycerol structure on plasma and liver lipid levels in rats
diet. The rats were then randomised to the four dietary groups for the following 3 weeks. At
the end of the feeding period, rats were fasted over-night, killed by decapitation, blood was
collected in heparin glasses and the livers were excised, weighed, and immediately frozen in
liquid nitrogen. Plasma isolated by centrifugation and livers were stored at -80 ºC until analysis.
The experiment was approved by the Danish Committee for Animal Experiments.
Lipid analyses
Lipid from diets, plasma, and livers were extracted with chloroform and methanol (Folch and
others 1957). Fatty acid composition of total TAG and the structure of TAG represented by
the fatty acid composition in sn-2 MAG of dietary fats were measured as described previously
(Porsgaard and Høy 2000). Lipid extracts from plasma and liver were methylated with a
borontriflourid (BF3) catalysed method (Morrison and Smith 1964) and analysed by gas-
chromatography using the following procedure: a Hewlett-Packard 5890 chromatograph was
equipped with a split/splitless injector, SP2380 capillary column (60 m, I.D. 0.25 mm; Supelco
Inc., Bellefonte, PA), flame-ionisation detection and helium as carrier gas. Conditions were as
follows: injector temperature 270 °C, flame ionisation detector 270 °C, helium carrier gas at
1.2 ml/min, injector split ratio 1:11. Initial oven temperature was 70 °C which was held for
5 min followed by temperature programming: 15 °C/min until 160 °C followed by 1.5 °C/min
until 200 °C, which was maintained for 15 min and finally the temperature was raised to 225
°C and maintained for 10 min. Peak areas were calculated using a Hewlett-Packard computing
integrator. The fatty acids were identified by comparing the retention time with standards of
known fatty acid composition.
Plasma TAG and cholesterol were measured enzymatically using commercial kits (Roche
Diagnostics GmbH, Mannheim, Germany). Liver TAG and cholesterol contents were measured by
quantitative high-performance thin layer chromatography (HPTLC) using a slight modification
of a method described by Müller and others (2004). Briefly, livers added cholesteryl acetate
(Sigma, St. Louis, MO) during extraction as internal standard were redissolved in a defined
volumen of chloroform/methanol (95/5, vol/vol). Samples and a quantitative standard mixture
composed of cholesterol, triolein, and cholesteryl acetate (a six-point standard curve was
applied on each plate) were applied on prewashed and activated silica HPTLC plates (Silica Gel-
60, Merck GmbH, Darmstadt, Germany) using a Desage AS-30 HPTLC applicator (Desage GmbH,
Wiesloch, Germany). The HPTLC plates were developed in a Camag Horizontal TLC-chamber
(Camag, Muttenz, Switzerland) in the first run in a solvent system consisting of hexane/diethyl
ether/formic acid (80/20/1, vol/vol/vol) and in the second run in heptane/diethyl ether (95/5,
vol/vol) increasing the length of the development by 2 cm in comparison with the first run.
Following development, the plates were soaked with charring reagent (633 mM CuSo4·5H2O
in 8% phosphoric acid) and heated at 160 ºC for 8 min. Quantification was performed through
densitometry at 420 nm, using a Desaga CD-60 HPTLC densitometer in reflectance mode. R2
values for the regression lines were never below 0.99.
Statistics
Results are expressed as mean values with their standard errors of the mean (SEM), n=8.
Differences between groups were tested by one-way ANOVA (GraphPad PRISM version 3.02,
GraphPad software, San Diego, CA). Tukey’s Multiple Comparison post test was used to determine
which means differed. Differences were considered significant at p<0.05.
binnenwerk.indd 31 21-08-2006 12:33:53

Results and discussion
Dietary fats
Fatty acid compositions of the dietary fat mixtures, shown in Table 1, were balanced with
respect to content of saturated (24.2-26.5 wt %), monounsaturated (30.4-34.6 wt %), n-6
PUFAs (18.0-20.6 wt %), and n-3 PUFAs (22.9-24.9 wt %). In contrast, the chow diet was
enriched in 18:2n-6 and with a low content of n-3 PUFAs. The content of 20:5n-3 was higher
in the MLM and LML diets than in the fish oil diet and the content of 22:6n-3 was higher in the
LML and fish oil diets than in the MLM diet, but overall the n-3 PUFA contents in the diets were
comparable. In the LML diet, the sn-2 position of dietary TAGs was enriched in 10:0, whereas
the MLM diet was enriched in 20:5n-3 in the sn-2 position, and the fish oil diet in 22:6n-3.
We used low-fat diets (4 wt %) with fat contents similar to ordinary rat chow in contrast to
comparable studies where higher dietary fat loads were given (10 wt %) (Nagata and others
2003, 2004). On the other hand, the structured TAGs used in the studies by Nagata and others
were purified by preparative high-performance liquid chromatography leading to a higher purity
of the structured TAGs. The cost of making highly purified structured TAGs is too high for using
the oils as commercial products, but they are of scientific value when studying absorption
parameters. In the present study, we chose to use non-purified structured TAGs leading to the
presence of other TAG species than the desired ones, but with a relative enrichment of the MLM
and LML species in the two oils, respectively.
Body weight gains and liver weights of rats

The rats weighed 249 ± 2 g at the beginning of the experiment and gained on average 81 ± 3,
82 ± 6, 85 ± 3, and 73 ± 7 g in the LML, MLM, fish oil, and chow groups, respectively (no
differences between groups). After 3 weeks feeding, liver weights were 8.9 ± 0.3, 8.8 ± 0.4,
8.7 ± 0.2, and 7.9 ± 0.4 g in the LML, MLM, fish oil, and chow groups, respectively (no
differences between groups). Thus, weight gain and final liver weights were not influenced by
dietary treatment.
TAG and cholesterol contents of plasma and liver

The highest plasma TAG concentration after 3 weeks feeding (Figure 1A) was measured in the fish
oil group (1.3 ± 0.1 mmol/l) and it was significantly higher than in the chow group (0.8 ± 0.2
mmol/l, p<0.05). Parallel to the plasma TAG levels, the highest liver TAG content (Figure 2A)
was measured in the fish oil group (33.8 ± 5.9 mg/g liver) and it was significantly higher than in
the chow group (17.2 ± 1.6 mg/g liver, p<0.05). No differences were measured in either plasma
or liver TAG concentrations between the three groups fed n-3 PUFAs. In many studies, fish oil
was shown to lower plasma TAG concentrations both in humans (reviewed by Kris-Etherton
and others 2002) and in some animal species, including rats, in comparison with lard (Otto
and others 1992), compared with corn oil (Herman and others 1991; Kim and others 2004),
or with coconut and safflower oils (Mohan and others 1991). Some studies showed greater
plasma TAG lowering effect the higher the dietary fish oil concentration (Bourre and others
1990; Nieuwenhuys and others 1998). On the contrary, Willumsen and others (1993) showed
that 22:6n-3 in contrast to 20:5n-3 did not possess a hypotriglyceridemic effect when fed as
highly purified free fatty acids to rats. In our experiment, the highest 22:6n-3 contents were
found in fish oil and LML and could partly explain that the highest plasma TAG concentrations
were found in these groups, although only statistically significant difference between the fish
oil and chow groups was found. Another factor that might have influenced the results was
that the effects of a non-purified diet (chow) were compared with the effects of purified diets.
binnenwerk.indd 32 21-08-2006 12:33:53

Table 1. Fatty acid composition of total TAG and in the sn-2 position of dietary fats (wt %)1.
LML MLM Fish oil Chow LML MLM Fish oil Chow
FA total TAG sn-2 position
10:0 16.1 10.8 2.9 -2 32.2 2.5 1.5 -

14:0 0.5 1.2 4.6 0.2 0.7 1.9 7.4 0.3
16:0 6.1 8.5 13.6 17.4 2.0 5.6 15.4 3.8
16:1n-7 0.5 1.4 4.0 0.2 0.6 1.8 4.5 0.3
18:0 3.1 3.1 2.7 2.8 1.6 2.5 1.4 0.5
18:1n-9 25.9 24.7 16.0 17.4 30.1 28.9 15.3 19.9
18:1n-7 1.6 2.1 1.9 1.2 1.1 2.5 1.3 0.5
18:2n-6 17.0 19.3 17.1 54.2 20.6 22.2 20.2 68.8
18:3n-3 0.6 0.9 2.1 5.5 0.9 1.2 2.1 5.8
18:4n-3 0.5 1.8 3.0 - 0.1 2.0 3.1 -
20:0 0.5 0.4 0.2 0.3 0.3 0.3 0.1 -
20:1n-9 1.4 1.9 5.0 0.4 1.1 2.0 2.0 0.2
20:4n-6 0.8 1.0 0.5 - 0.2 1.0 0.5 -
20:4n-3 0.6 0.7 0.6 - 0.1 0.9 0.6 -
20:5n-3 13.0 14.0 8.2 0.2 4.3 17.0 9.4 -
22:1n-11 0.8 2.0 7.5 - 0.6 1.7 1.9 -
22:5n-3 1.8 0.8 0.7 - 0.7 1.1 1.1 -
22:6n-3 8.3 4.8 8.6 0.1 1.2 4.5 11.7 -
Others3 0.9 0.6 0.8 0.1 1.6 0.4 0.5 -
SFA4 26.5 24.2 24.2 20.7 37.3 12.9 26.1 4.6
MUFA4 30.4 32.3 34.6 19.2 34.5 37.0 25.0 20.9
n-6 PUFA4 18.2 20.6 18.0 54.5 21.0 23.5 21.0 68.8
n-3 PUFA4 24.9 22.9 23.2 5.7 7.2 26.6 27.9 5.8
1 Values are the mean of three determinations.

2A dash (-) means not detected.
3 Others represent fatty acids that contributed less than 0.5 wt % of total fatty acids.
4 SFA, MUFA, n-6 PUFA, and n-3 PUFA represent the sum of total saturated, monounsaturated, n-6
polyunsaturated, and n-3 polyunsaturated fatty acids, respectively.
Feeding of non-purified diets versus semi-purified diets was previously found to influence the
atherogenicity in hamsters (Nicolosi and others 1998). Furthermore, we used low-fat diets in
contrast to most other studies that have been performed with medium or high dietary fat loads
(Bourre and others 1990; Mohan and others 1991; Herman and others 1991; Otto and others
1992). A factor that could also influence the outcome of the experiment.
The plasma cholesterol concentrations after 3 weeks feeding (Figure 1B) were between 1.4 ±
0.3 and 2.0 ± 0.2 mmol/l in rats from the 4 dietary groups (no differences between groups).
In livers, the cholesterol contents (Figure 2B) were between 2.7 ± 0.3 and 3.5 ± 0.4 mg/
g liver (no differences between groups). In humans, dietary fish oil has limited effect on
plasma cholesterol contents (reviewed by Kris-Etherton and others 2002), while in rats serum
cholesterol levels were decreased in some experiments (Mohan and others 1991; Otto and others
binnenwerk.indd 33 21-08-2006 12:33:54

A
1.5 a,b a
TAG concentration (mmol/l)

a,b
1.0 b
0.5
0.0
LML MLM Fish oil Chow
B
2.5
Cholesterol concentration (mmol/l)
2.0
1.5
1.0
0.5
0.0
Figure 1. Plasma concentrations of TAG (A) and cholesterol (B) (mmol/l) in rats following 3 weeks feeding
(LML, MLM, fish oil, and chow dietary groups). Data are mean ± SEM of 8 rats. Values not sharing a common
letter are significantly different (P<0.05).
1992; Nieuwenhuys and others 1998), or the effect depended on the background diet (Herman
and others 1991).
In the present study, TAG structure had no influence on plasma and liver TAG and cholesterol
concentrations. This is in contrast to the studies by Nagata and others (2003, 2004). With
highly purified structured TAGs of the MLM- and LML-types where L was 18:2n-6 and M either
8:0 or 10:0 (purity 90-92%), serum TAG levels were lower in rats fed LML-type TAGs than in
those fed MLM-type TAGs (Nagata and others 2003). All structured TAGs decreased serum
cholesterol compared with the control corn oil with a tendency for lower levels in serum from
binnenwerk.indd 34 21-08-2006 12:33:54

A
a
40
TAG content (mg/g liver)
30 a,b
a,b
20 b
10
0
B
4
Cholesterol content (mg/g liver)
0
Figure 2. Liver contents of TAG (A) and cholesterol (B) (mg/g liver) in rats following 3 weeks feeding (LML,
MLM, fish oil, and chow dietary groups). Data are mean ± SEM of 8 rats. Values not sharing a common letter
are significantly different (P<0.05).
rats fed TAGs containing 10:0 in comparison with those fed 8:0. No differences were observed
in liver cholesterol levels, while liver TAG levels were lower in rats fed TAGs containing 10:0
than in those fed corn oil. When purified TAGs of the structure MML or MLM, where M was 8:0
and L was either 20:5n-3 or 22:6n-3 (purity 68-82%) was fed to rats, all structured TAGs led to
lower serum cholesterol, TAG, and phospholipid levels than the control soybean oil (Nagata and
others 2004). Minor differences were observed between the structured TAGs with the highest
levels observed in the 22:6n-3/8:0/8:0 fed rats. In hamsters fed either fish or seal oil, serum
TAG and cholesterol levels were not influenced by differences in dietary TAG structure, while
binnenwerk.indd 35 21-08-2006 12:33:54

liver cholesterol levels were highest and TAG levels were lowest after fish oil feeding where
the sn-2 position of the glycerol moiety in fish oil was enriched with n-3 PUFAs (Yoshida and
others 2001). In rats fed seal or squid oil no effects of differences in dietary TAG structure
were observed on plasma and liver TAG and cholesterol levels (Ikeda and others 1998). On
the basis of these studies, no overall conclusion can be drawn regarding the effect of dietary
intramolecular TAG structure on levels of TAG and cholesterol in plasma and liver. Factors like
purity of dietary structured TAGs, dietary fat levels and fatty acid compositions, and species
may influence the results.
Fatty acid composition of plasma and liver

No 10:0 was detected in either plasma or liver of rats fed MLM and LML (Table 2 and 3)
indicating that this MCFA was either oxidized in the liver or metabolized to a longer-chain
derivative. Previous studies have shown that some MCFAs are absorbed through the lymphatic
system (Porsgaard and others 2005b; Straarup and others 2005) with decreasing absorption
the shorter the chain length of the MCFA (Mu and Høy 2000), but the major part of MCFAs is
absorbed through the portal vein for further oxidation in the liver (Bernard and Carlier 1991).
Table 2. Fatty acid composition of plasma from rats (wt %)1.
FA Mean SEM Mean SEM Mean SEM Mean SEM
14:0 0.6 0.0a,b 0.5 0.0a,b 0.6 0.1a 0.4 0.0b

16:0 22.4 0.3a 22.2 0.3a 23.0 0.2a 20.6 0.4b
16:1n-7 2.8 0.1a 2.9 0.2a 3.3 0.5a 1.3 0.1b
18:0 6.2 0.4b 6.4 0.3a,b 6.0 0.5b 8.0 0.5a
18:1n-9 11.6 0.6a 11.1 0.3a 10.6 0.2a 7.2 0.5b
18:1n-7 2.4 0.1a 2.4 0.1a 2.6 0.1a 2.1 0.1b
18:2n-6 19.3 0.5b,c 18.6 0.5c 21.8 0.7b 25.5 0.9a
18:3n-3 0.4 0.0b 0.4 0.0b 0.8 0.1a 1.0 0.1a
20:4n-6 14.8 0.9b 16.0 0.5b 12.2 0.7b 30.0 1.7a
20:5n-3 9.8 0.5b 11.7 0.5a 8.0 0.6b 0.7 0.0c
22:5n-3 1.6 0.1b 2.1 0.1a 1.4 0.1b 0.4 0.1c
22:6n-3 7.5 0.2b 5.4 0.2c 9.3 0.2a 2.7 0.1d
Others2 0.6 0.0 0.3 0.0 0.4 0.0 0.1 0.0
SFA3 29.1 0.5 29.1 0.4 29.6 0.6 29.0 0.5
MUFA3 16.8 0.7a 16.4 0.5a 16.5 0.6a 10.6 0.6b
n-6 PUFA3 34.5 0.7b 35.0 0.8b 34.3 0.8b 55.6 0.9a
n-3 PUFA3 19.5 0.5a 19.6 0.6a 19.5 0.9a 4.8 0.2b
1Values represent the mean ± SEM of 8 rats in each group. Values in rows not sharing a common
superscript letter were significantly different (P<0.01).
2Others represent fatty acids that contributed less than 0.5 wt % of total fatty acids.
3SFA, MUFA, n-6 PUFA, and n-3 PUFA represent the sum of total saturated, monounsaturated, n-6
binnenwerk.indd 36 21-08-2006 12:33:55

Table 3. Fatty acid composition of total lipids from rat livers (wt %)1.
FA Mean SEM Mean SEM Mean SEM Mean SEM
16:0 20.2 0.3a,b 19.6 0.3b,c 20.8 0.3a 19.0 0.3c

16:1n-7 2.2 0.1b 2.5 0.2b 3.5 0.1a 0.9 0.1c
18:0 11.3 0.4b 10.4 0.5b 9.9 0.3b 13.5 0.7a
18:1n-9 7.8 0.5a 7.1 0.2a 7.9 0.3a 5.3 0.5b
18:1n-7 2.6 0.1b 2.6 0.1b 3.0 0.1a 2.4 0.1b
18:2n-6 16.6 0.3b 16.7 0.3b 18.3 0.6b 23.3 0.7a
18:3n-3 0.3 0.0b 0.3 0.0b 0.7 0.1a 0.7 0.1a
20:3n-6 0.6 0.0a,b 0.6 0.0b 0.7 0.0a 0.3 0.0c
20:4n-6 15.9 0.7b,c 17.8 0.6b 13.9 0.6c 28.1 1.0a
20:5n-3 8.2 0.6b 10.6 0.7a 7.0 0.4b 0.4 0.0c
22:5n-3 2.3 0.1b 2.9 0.2a 1.8 0.1c 0.8 0.0d
22:6n-3 11.3 0.3a 8.7 0.2b 12.0 0.3a 4.7 0.3c
Others2 0.7 0.1 0.2 0.0 0.5 0.1 0.6 0.1
SFA3 31.9 0.6a,b 30.4 0.4b 31.3 0.5a,b 32.9 0.6a
MUFA3 12.6 0.6a,b 12.2 0.5b 14.4 0.4a 8.7 0.6c
n-6 PUFA3 33.1 0.6b 35.0 0.9b 32.9 0.8b 51.8 0.6a
n-3 PUFA3 22.4 0.8a 22.4 0.8a 21.4 0.6a 6.6 0.3b
1Values represent the mean ± SEM of 8 rats in each group. Values in rows not sharing a common
superscript letter were significantly different (P<0.01).
2Others represent fatty acids that contributed less than 0.5 wt % of total fatty acids.
3SFA, MUFA, n-6 PUFA, and n-3 PUFA represent the sum of total saturated, monounsaturated, n-6
In plasma, the major differences were observed in fatty acid compositions between rats fed n-3
PUFAs and rats fed chow. The higher content of n-6 PUFAs in the chow diet was reflected in
plasma leading to significantly higher contents of 18:2n-6, 20:4n-6, and total n-6 PUFA in the
chow group and lower contents of 16:0, 16:1n-7, 18:1n-9, 18:1n-7, 20:5n-3, 22:5n-3, 22:6n-3,
total monounsaturated fatty acids, and total n-3 PUFAs compared with the other three groups
(p<0.01). In liver, similar differences were observed between the groups fed n-3 PUFA and the
chow group. This means that the dietary fatty acid composition was reflected both in plasma
and in liver. Furthermore, the dietary TAG structure was reflected in the plasma and liver fatty
acids. MLM that was enriched in 20:5n-3 in the sn-2 position of TAG led to a higher content
of this fatty acid, while fish oil, which was enriched in 22:6n-3 in the sn-2 position, resulted
in the highest amount of this fatty acid in plasma and liver. A higher content of 22:5n-3 was
found both in plasma and in livers of rats fed MLM in comparison with the other n-3 PUFA diets.
This fatty acid was partly provided by the diet and partly formed through elongation of 20:5n-3
originating from enrichment of the sn-2 position in the MLM oil. Although the individual n-3
PUFAs in plasma and liver differed between the three groups fed dietary n-3 PUFAs, the overall
n-3 PUFA content was similar as were the overall contents of saturated, monounsaturated, and
n-6 PUFAs. Åkesson and others (1978) found that approximately 25% of the fatty acid initially
located in the sn-2 position of dietary fat had migrated to the sn-1,3 positions during hydrolysis
binnenwerk.indd 37 21-08-2006 12:33:55

and lymphatic absorption. This implies a general conservation of approximately 75% of the fatty
acids located in the sn-2 position, which is important when considering the possible advantages
in tailor-making fats with particular TAG structures and maintaining the location of fatty acids
in specific positions following absorption. From the present study, it seems that enrichment of
specific LCFAs in the sn-2 position of dietary fat will lead to a higher level of that specific fatty
acid in the organs; thus tailor-making is indeed possible.
Conclusion
From the results presented in this study, it can be concluded that dietary TAG structure influenced
fatty acid composition of livers and plasma, but had no influence on the concentrations of
TAG and cholesterol, when fed as part of low-fat diets to rats. The structured TAGs used in the
present study contained MCFAs and n-3 PUFAs of marine origin and were produced by enzymatic
interesterification without purification.
Acknowledgements
We greatly appreciate the technical help of Karen Jensen and Jannie Agersteen. Lillian Vile and
Nina Kjeldsen are thanked for taking care of the rats. Financial support was obtained from The
Danish Technological Research Council.
References
Aaes-Jørgensen E, Hølmer G. 1969. Essential fatty acid-deficient rats: I. Growth and testes development. Lipids
4(6):501-506.
Ackman RG. 1988. Some possible effects on lipid biochemistry of differences in the distribution on glycerol of long-
chain n-3 fatty acids in the fats of marine fish and marine mammals. Atherosclerosis 70(1-2):171-173.
Bernard A, Carlier H. 1991. Absorption and intestinal catabolism of fatty acids in the rat: effect of chain length and
unsaturation. Exp Physiol 76(3):445-455.
Bourre JM, Bonneil M, Dumont O, Piciotti M, Calaf R, Portugal H, Nalbone G, Lafont H. 1990. Effect of increasing
amounts of dietary fish oil on brain and liver fatty composition. Biochim Biophys Acta 1043(2):149-152.
Christensen MS, Høy C-E, Becker CC, Redgrave TG. 1995a. Intestinal absorption and lymphatic transport of
eicosapentaenoic (EPA), docosahexanoic (DHA), and decanoic acids: dependence on intramolecular triacylglycerol
structure. Am J Clin Nutr 61(1):56-61.
Christensen, MS, Müllertz A, Høy C-E. 1995b. Absorption of triglycerides with defined or random structure by rats with
biliary and pancreatic diversion. Lipids 30(6):521-526.
Folch J, Lees M, Stanley GHS. 1957. A simple method for the isolation and purification of total lipids from animal
tissues. J Biol Chem 226(1):497-509.
Hartvigsen MS, Mu H, Hougaard KS, Lund SP, Xu X, Høy C-E. 2004. Influence of dietary triacylglycerol structure and
level of n-3 fatty acids administered during development on brain phospholipids and memory and learning ability
of rats. Ann Nutr Metab 48(1):16-27.
Herman S, Sediaoetama AD, Karyadi D, Beynen AC. 1991. Influence of background composition of the diet on the
lipemic effect of fish oil vs. corn oil in rats. J Nutr 121(5):622-630.
Hubbard VS, McKenna MC. 1987. Absorption of safflower oil and structured lipid preparations in patients with cystic
fibrosis. Lipids 22(6):424-428.
Ikeda I, Tomari Y, Sugano M, Watanabe S, Nagata J. 1991. Lymphatic absorption of structured glycerolipids containing
medium-chain fatty acids and linoleic acid, and their effect on cholesterol absorption in rats. Lipids 26(5):369-
373.
binnenwerk.indd 38 21-08-2006 12:33:56

Ikeda I, Yoshida H, Tomooka M, Yosef A, Imaizumi K, Tsuji H, Seto A. 1998. Effects of long-term feeding of marine
oils with different positional distribution of eicosapentaenoic and docosahexaenoic acids on lipid metabolism,
eicosanoid production, and platelet aggregation in hypercholesterolemic rats. Lipids 33(9):897-904.
Kenler AS, Swails WS, Driscoll DF, DeMichele SJ, Daley B, Babineau TJ, Peterson MB, Bistrian BR. 1996. Early enteral
feeding in postsurgical cancer patients. Fish oil structured lipid-based polymeric formula versus a standard
polymeric formula. Ann Surgery 223(3):316-333.
Kim H-K, Choi S, Choi H. 2004. Suppression of hepatic fatty acid synthase by feeding α-linolenic acid rich perilla oil
lowers plasma triacylglycerol level in rats. J Nutr Biochem 15(8):485-492.
Kremer JM. 2000. n-3 Fatty acid supplements in rheumatoid arthritis. Am J Clin Nutr 71(1 suppl):349S-351S.
Kris-Etherton PM, Harris WS, Appel LJ. 2002. Fish consumption, fish oil, omega-3 fatty acids, and cardiovascular
disease. Circulation 106(21):2747-2757.
Kruimel JW, Naber TH, van der Vliet JA, Carneheim C, Katan MB, Jansen JB. 2001. Parenteral structured triglyceride
emulsion improves nitrogen balance and is cleared faster from the blood in moderately catabolic patients. J
Parenteral Enter Nutr 25(5):237-244.
Mattson FH, Volpenhein RA. 1964. The digestion and absorption of triglycerides. J Biol Chem 239(9):2772-2777.
Mohan PF, Phillips FC, Cleary MP. 1991. Metabolic effects of coconut, safflower, or menhaden oil feeding in lean and
obese Zucker rats. Br J Nutr 66(2):285-299.
Morrison WR, Smith LM. 1964. Preparation of fatty acid methyl esters and dimethylacetals from lipids with boron
fluoride-methanol. J Lipid Res 5:600-608.
Mu H, Høy C-E. 2000. Effects of different medium-chain fatty acids on intestinal absorption of structured triacylglycerols.
Lipids 35(1):83-89.
Müller H, Hellgren LI, Olsen E, Skrede A. 2004. Lipids rich in phosphatidylethanolamine from natural gas-utililzing
bacteria reduce plasma cholesterol and classes of phospholipids: a comparison with soybean oil. Lipids 39(9):833-
841.
Nagata J, Kasai M, Watanabe S, Ikeda I, Saito M. 2003. Effects of highly purified structured lipids containing medium-
chain fatty acids and linoleic acid on lipid profiles in rats. Biosci Biotechnol Biochem 67(9):1937-1943.
Nagata J, Kasai M, Negishi S, Saito M. 2004. Effects of structured lipids containing eicosapentaenoic or docosahexaenoic
acid and caprylic acid on serum and liver lipid profiles in rats. BioFactors 22(1-4):157-160.
Nicolosi RJ, Wilson TA, Lawton C, Rogers EJ, Wiseman SA, Tijburg LBM, Kritchevsky D. 1998. The greater atherogenicity
of nonpurified diets versus semipurified diets in hamsters is mediated via differences in plasma lipoprotein
cholesterol distribution, LDL oxidative susceptibility, and plasma α-tocopherol concentration. J Nutr Biochem
9(10):591-597.
Nieuwenhuys CMA, Béguin S, Offermans RFG, Emeis JJ, Hornstra G, Heemskerk JWM. 1998. Hypocoagulant and lipid-
lowering effects of dietary n-3 polyunsaturated fatty acids with unchanged platelet activation in rats. Arterioscler
Thromb Vasc Biol 18(9):1480-1489.
Otto DA, Baltzell JK, Wooten JT. 1992. Reduction in triacylglycerol levels by fish oil correlates with free fatty acid levels
in ad libitum fed rats. Lipids 27(12):1013-1017.
Peet M, Stokes C. 2005. Omega-3 fatty acids in the treatment of psychiatric disorders. Drugs 65(8):1051-1059.
Porsgaard T, Høy C-E. 2000. Lymphatic transport in rats of several dietary fats differing in fatty acid profile and
triacylglycerol structure. J Nutr 130(6):1619-1624.
Porsgaard T, Kánský J, Mason S, Mu H. 2005a. Size and number of lymph particles measured by a particle sizer during
absorption of structured oils in rats. Lipids 40(3):273-279.
Porsgaard T, Xu X, Göttsche J, Mu H. 2005b. Differences in the intramolecular structure of structured oils do not affect
pancreatic lipase activity in vitro or the absorption by rats of (n-3) fatty acids. J Nutr 135(7):1705-1711.
Rubin M, Moser A, Vaserberg N, Greig F, Levy Y, Spivak H, Ziv Y, Lelcuk S. 2000. Structured triacylglycerol emulsion,
containing both medium- and long-chain fatty acids, in long-term home parenteral nutrition: a double-blind
randomized cross-over study. Nutrition 16(2):95-100.
Straarup EM, Høy C-E. 2000. Structured lipids improve fat absorption in normal and malabsorbing rats. J Nutr
130(11):2802-2808.
binnenwerk.indd 39 21-08-2006 12:33:56

Straarup EM, Porsgaard T, Mu H, Hansen CH, Høy C-E. 2005. Lymphatic transport in rats of interesterified oils containing
conjugated linoleic acids. Lipids 40(7):677-684.
Tso P, Lee T, DeMichele SJ. 1999. Lymphatic absorption of structured triglycerides vs. physical mix in a rat model of fat
malabsorption. Am J Physiol 277(2 pt 1):G333-G340.
Uauy R, Hoffman DR, Peirano P, Birch DG, Birch EE. 2001. Essential fatty acids in visual and brain development. Lipids
36(9):885-895.
Willumsen N, Hexeberg S, Skorve J, Lundquist M, Berge RK. 1993. Docosahexaenoic acid shows no triglyceride-lowering
effects but increases the peroxisomal fatty acid oxidation in liver of rats. J Lipid Res 34(1):13-22.
Xu X. 2000. Production of specific-structured triacylglycerols by lipase-catalyzed reactions: a review. Eur J Lipid Sci
Technol 102(4):287-303.
Yaqoob P. 2003. Lipids and the immune response: from molecular mechanisms to clinical applications. Curr Opin Clin
Nutr Metab Care 6(2):133-150.
Yoshida H, Ikeda I, Tomooka M, Mawatari M, Imaizumi K, Seto A, Tsuji H. 2001. Effect of dietary seal and fish oils on
lipid metabolism in hamsters. J Nutr Sci Vitaminol 47(3):242-247.
Åkesson B, Gronowitz S, Herslof B, Ohlson R. 1978. Absorption of synthetic, stereochemically defined acylglycerols in
the rat. Lipids 13(5):338-343.
binnenwerk.indd 40 21-08-2006 12:33:56

Cholesterol content in seafood, data from the last decade: A review
Federal Research Centre for Nutrition and Food, Department for Fish Quality, Palmaille 9, 22767
Hamburg, Germany
Abstract
An extensive review is given about the content of cholesterol in aquatic animals. Specimen
of fish, crustaceans and molluscs have been collected onboard of the fishery research vessel
“Walther Herwig III” and at local fish markets in the time period between 1992 and 2004.
Most specimen are from North Atlantic waters (including Spitzbergen (Svalbard) and Greenland
waters) some are of freshwater and brackish water origin. The data presented here are mainly
on the cholesterol content in the edible part of the seafood (fish fillet, tail muscle and claw
meat of crustaceans, whole mussels) but also about other parts of the animals used as a food
item as the gonads (caviar, caviar preparations as spread). It is surprising how low the average
cholesterol content in marine fish is (30 – 40 mg/kg). Furthermore, it is astonishing that the
cholesterol content in edible part of marine fish (muscle) and in organs is not correlated with
the respective lipid content. Contents in muscle of freshwater fish are somewhat higher, while
the edible part of crustaceans and molluscs exhibit high cholesterol contents (up to over 200
mg/kg). Highest in cholesterol are fish eggs and food prepared from fish eggs as caviar and
caviar derived products.
Keywords: cholesterol, marine fish, fishery products, crustaceans
Introduction
In this review cholesterol contents in edible part (fillet) and gonads of a number of aquatic
species are reported. Data provided can help to improve the existing data in food composition
tables and assist all those who have to eat a low cholesterol diet because of medical advice.
This review will not deal with the physiological implications and metabolism of cholesterol in
fish or in human beings.
Knowledge about the content of cholesterol in food in general and especially in fish and
other types of seafood is of utmost importance for all those who have to take care for a
reduction of cholesterol intake via food. Also for those who prepare foods or diets or who
give recommendations about food intake profound knowledge about cholesterol content in
the raw material used for food preparation is essential. This can be achieved effectively and
reliable only when at least the average cholesterol content in fish and seafood based on a
robust database is known, but even better when also information about its possible variation
with season, catching ground and e.g. state of maturity is known. Furthermore, in fatty fish
the interrelationship between fat content and cholesterol content must be known. Cholesterol
content in processed fishery products can vary from the original cholesterol content present
in the raw material due to fat and water processing losses, deterioration of cells and to some
extent oxidation. The cholesterol molecule, however, is stable and is not degraded by the
principal technological processes.
binnenwerk.indd 41 21-08-2006 12:33:56

Unfortunately, there is still a lack in knowledge about cholesterol contents in fish and seafood.
Food composition tables (e.g. Anon 1999; Souci, Fachmann and Kraut 2000; Piironen V and
others 2002) show a lot of gaps and systematic investigations are rare or lacking completely.
A limited number of reviews and contributions covering a broad number of species have been
published (Feeley and others 1972; Kritchevsky and others 1967; Krzynowek 1985; Idler and
Wiseman 1971; Teshima 1991; Kanazawa 2001; King and others 1990).
Since 1996 the occurrence and amount of cholesterol in the edible part of seafood is investigated
quantitatively in the Department of Fish Quality of the Federal Research Centre for Nutrition
and Food (formerly Institute for Biochemistry and Technology of the Federal Research Centre
for Fisheries). A first report about the cholesterol content in lean fish species and in some
crustaceans was given in Tromsö in 1997 during the 28th WEFTA meeting and a second later in
Thessaloniki in 1998 during the 29th meeting (Oehlenschläger 1998). To broaden the basis of
our knowledge these investigations have been continued and now new data about cholesterol
content in gadoids, other seawater species, fatty pelagic species, flat fish species, crustacean
and molluscan shellfish and other species can be presented.
Cholesterol is the main sterol in marine fish like haddock, pollock, salmon and in crustaceans
like shrimp and lobster with over 90% of all sterols (Kritchevsky and others 1967). Oyster show
only 41% cholesterol of sterols, crab 57% and scallop and clam 26% and 37%, respectively. A
good review about the physiology and biochemistry of sterol in crustaceans, molluscs and fish
can be found by Teshima (1991).
Cholesterol has been analysed in a number of seafood species from different areas of the
world. Most of the investigations, however, have not been performed in a systematic way,
do not cover more than one catching area and do not show seasonal variations in content.
Ackman and McLeod (1988) have investigated cholesterol in fish and shellfish food products
from Nova Scotia. They reported low cholesterol contents in ground fish and pelagic fish
species, higher contents in molluscs and crustaceans (approx. 70-75 mg/100 g). About sterols
in seafood from Gilbert Bay in Southern Labrador was reported by Copeman and Parrish (2004).
They found that cholesterol was the major sterol in the molluscs investigated. Data about the
cholesterol content of sharks and rays have been reported by Lytle and Lytle (1994), who found
a variation in cholesterol content between 16 to 69 mg/100g. Commercial fish and crustacean
species from the Arabian Gulf were assayed for cholesterol by Ewaidah (1993). The cholesterol
content ranged from 65.2 mg/100 g to 146.2 mg/100 g. El. Sayed and others (1984) measured
the cholesterol content in Tilapia nilotica and Sparus auratus. The cholesterol content varied
between 50 and 170 mg/100g with the highest content in summer and the lowest in spring in
T. nilotica. 10 selected marine fish in Malaysian waters were investigated by Osman and others
(2001), for the cholesterol content. The total cholesterol content was generally low between
37 and 49 mg/100 g. Mathew and others (1999), found in 97 samples of 43 families differing
cholesterol contents. The work, however, is only based on one specimen per species. DeKonig
and others (1993) found in a number of South African fatty fish species an average cholesterol
content of 130 mg/100 g.
Imre and Saglik (1998) found in nine common fish species purchased on the central fish market
in Istanbul cholesterol contents between 43 mg/100 g and 75 mg/100 g. Two groups worked
on the cholesterol content in sardines. Krzynowek and others (1992) determined in Maine
sardines an average cholesterol content of 90 mg/100 g with little seasonal variation, while
binnenwerk.indd 42 21-08-2006 12:33:56

Cholesterol content in seafood, data from the last decade
De Leonardis and Macciola (2004) reported cholesterol contents of 93 mg/100 g (variation 67

to 131 mg/100 g) in Adriatic sardine fillets.
There are only few data on cholesterol in freshwater species. Bieniarz and others (2000)
analysed the cholesterol content in some farmed freshwater fish species in Poland and found the
lowest cholesterol content in European catfish (21 mg/100 g) while rainbow trout (muscle+skin)
exhibited the highest content with 190 mg/100 g. 20 mg Cholesterol/100 g was detected by
Lahti (1987) in vendace (Coregonus albula) from a Finish lake. Three Brazilian farmed freshwater
species from the Platina basin were reported by Moreira and others (2001) to contain cholesterol
contents of approx. 50 mg/100.
Concerning differences in the cholesterol content of wild and farmed fish species, Nettleton and
Exler (1992) could demonstrate that there are no significant differences in the wild and farmed
form of Channel catfish (Ictalurus punctatus) 58 mg/100 g and 61 mg/100 g, respectively, Coho
salmon (Oncorhynchus kisutch) 48 mg/100 g and 51 mg/100 g, respectively and rainbow trout
(Oncorhynchus mykiss) 60 mg/100 g and 59 mg/100 g, respectively.
The cholesterol content of red shrimp (Aristeus antennatus), pink shrimp (Parapenaeus
longirostris) and Norway lobster (Nephrops norvegicus) caught off the South coast (Algarve)
of Portugal was found to be 61 and 58 mg/100g for the shrimps and 60 mg/100 g for Norway
lobster by Rosa and Nunes (2003). Cholesterol content in winter samples was higher than
in summer samples. Cholesterol in several species of shrimp was assayed by Krzynowek and
Panunzio (1989). The overall average was 152 mg/100g of edible portion of shrimp. Mohd.Omar
and others (1995) published the cholesterol content of 5 Malaysian prawn species. The variation
was not big among species ranging from 157 to 186 mg/100 g. The highest content was found
in Rainbow prawn. Takada and others (1988) analysed 18 shrimp and prawn species for their
cholesterol content and found that the total cholesterol was in the range of 90-150 mg/100
g in the edible portion of the shrimp. From 100 g shrimp tissue (Penaeus aztecus) isolated
Johnston and others (1983) 201 mg/100 g cholesterol and 49 mg/100 g cholesterol esters.
Also Uddin and others (2001) found high cholesterol values in crustacean shellfish from the
Bay of Bengal varying between 130 mg/100 g in Metapenaeus monocerus to 161 mg/100 g in
Penaeus indicus. The same authors also found very high contents in cephalopods (200 mg/100
g in Loligo duvauceli and 169 mg/100 g in Sepia aculeate. In Dungeness crab and in pink shrimp
King and others (1990) found an average cholesterol content of 72 mg/100 g and 147 mg/100
g, respectively. In the same publication the authors analysed in molluscs: 48 mg/100 g in
Pacific oyster, 37 mg/100 g in blue mussel, 36 mg/100 g in Manila clam and 231 (197-293)
mg/100 g in California squid. Krzynowek and others (1989) investigated also 2 species of squid,
Loligo pealei and Illex illecebrosus, and reported cholesterol content of 300 mg/100g (171 – 449
mg/100 g variation) and 212 mg/100g (108 – 336 mg/100 g variation), respectively. Extreme
high cholesterol content in fish roe have been published firstly by Iwasaki and Harada (1984),
who published cholesterol contents up to 655 mg/100 g in roe of Black Sea bream. The same
authors investigated later (1985) also the roe of squid (Dorytheutis bleekeri) and crab (Portunus
trituberculatus) which showed 374 mg/100 g and 494 mg/100 g, respectively.
binnenwerk.indd 43 21-08-2006 12:33:56

Material and methods
Fish and shellfish

The fish used for the experiments were collected during 17 research cruises of the Fishery
Research Vessels “Walther Herwig II” and “Walther Herwig III” in North Atlantic waters and
“Clupea” in the Baltic Sea. Table 1 gives an overlook about cruises, areas and seasons. A
compilation of the species investigated and their scientific names are given in Table 2.
The marine specimens were taken from hauls mostly made by bottom trawling. Life fish was
killed by a blow on the head followed by gill cut and gutting. After bleeding the fish were
cleaned in cold seawater and subsequently filleted and skinned. Whole fillets of both sides of
the fish were chopped into small cubes of approx. 1 cm3. To get a sub-sample representative
of the whole fillet the cubes were mixed thoroughly and a sub-sample of ca. 50 g was taken.
This was put in a polyethylene jar covered with a polyethylene lid and frozen in a blast freezer
down to –39 °C. Until being analysed in the laboratory at land the samples were stored
frozen at –25 °C. Roe was carefully removed from the opened belly and care was taken not to
contaminate the roe with remains of liver or kidney.
At land samples were put in a freeze dryer and dried until water content was constant and
< 0.5% of dry weight. The dried samples were finely milled in a ball mill and then used for
cholesterol analysis.
Samples of cephalopods (octopus and squid) were obtained from German traders of frozen
fish and brown shrimp from the Wadden Sea off Schleswig-Holstein (Crangon crangon) from
a commercial shrimp processing establishment in North Germany. Samples of tropical and
Table 1. Cruises with fishery research vessels for sample collection for cholesterol analysis.
vessel cruise no year month area no of samples
Walther Herwig II 110 1990 Sep/Oct East/West Greenland 40

Walther Herwig II 115 1991 May/June North Sea 27
Walther Herwig II 126 1992 July North Sea 10
Walther Herwig II 137 1993 Sep/Oct East/West Greenland 23
Walther Herwig III 150 1994 Aug Barents Sea 64
Walther Herwig III 184 1997 May/June Faroe Islands 79
Walther Herwig III 187 1997 Aug Barents Sea 187
Walther Herwig III 194 1998 Mar South Ireland/Biscay 42
Walther Herwig III 195 1998 Apr Faroe Islands 232
Walther Herwig III 204 1999 Apr Faroe/Shetland Islands 67
Walther Herwig III 210 1999 Sep Skagerak/Kattegatt 79
Walther Herwig III 214 2000 Feb North Sea 13
Walther Herwig III 217 2000 May/June Barents Sea 185
Walther Herwig III 232 2001 Sep Faroe/Shetland Islands 165
Walther Herwig III 243 2002 Sep Westbritish/Ireland 69
Clupea 112 2001 Mar Usedom/Baltic Sea 116
Clupea 143 2003 July Usedom/Baltic Sea 103
binnenwerk.indd 44 21-08-2006 12:33:57

Table 2. List of species investigated in this review.
Eel Anguilla anguilla

Eelpout Zoaces viviparus
Burbot Lota lota
Blue Whiting Micromesistius poutassou
Blue ling Molva dypterygia
Bream Abramis brama
Conger Conger conger
American plaice Hippoglossoides platessoides
Dogfish Squalus acanthias
Megrim Lepidorhombus whiffiagonis
Flounder Plathichthys flesus
Perch Perca fluviatils
Trout Salmo trutta
Pout Trisopterus luscus
Forkbeard Phycis blennoides
Brown shrimp Crangon crangon
Brill Scophthalmus rhombus
Argentine Argentina sp.
Grenadier Coryphaenoides rupestris
Herring Clupea harengus
John Dory Zeus faber
Ocean quahaug Arcta islandica
Scallop Pectinidae
Cod Gadus morhua
Norway lobster Nephrops norvegicus
Cold water shrimp Pandalus borealis
Wolffish Anarhichas spec.
Pope, ruffe Gymnocephalus cernuus
Dab Limanda limanda
Gurnard Trigla spec.
Salmon Salmo salar
Ling Molva molva
Lemon sole Microstomus kitt
Capelin Mallotus villosus
Cusk Brosme brosme
Mackerel Scomber scombrus
Blue mussel Mytilus edulis
Mora moro Mora moro
Octopus Octopus vulgaris
Flying squid Todarodes sagittatus
Polar cod Boreogadus saida
Ocean perch Sebastes marinus, mentella
Roach Rutilus rutilus
Witch Glyptocephalus cynoglosus
Sandeel Ammodytae
Anchovy Engraulis encrasicolus
Sardine Sardina Pilchardus
binnenwerk.indd 45 21-08-2006 12:33:57

Table 2. Continued.
Haddock Melanogrammus aeglefinus

Houting Coregonus oxyrhynchus
Plaice Pleuronectes platessa
Greenland Halibut Rheinhardtius hippoglossoides
Black Scabbardfish Aphanopus carbo
Lumpsucker Cyclopterus lumpus
Hake Merluccius merluccius
Saithe Pollachius virens
Angler Lophius piscatorius
Sole Solea solea
Sprat Sprattus sprattus
Turbot Psetta maxima
Pollack Pollachius pollachius
Smelt Osmerus eperlanus
Horse mackerel Trachurus trachurus
Black Sea bream Mylio macrocephalus
Edible crab Cancer pagurus
White halibut Hippoglossus hippoglossus
Whelk Buccinum undatum
Whiting Merlangius merlangus
Pikeperch Stizostedion Lucioperca
subtropical shrimps and prawns were bought frozen at Hamburg fishmarket. These samples were
treated in the same way as the marine fish samples prior to analysis.
All freshwater fish samples originated from catches off the coast of Mecklenburg-Vorpommern
(Bodden) or were angled in Danish lakes South of Aarhus (Jutland) and have been treated as
the marine fish samples.
In total approx. 1600 fish samples comprising 68 species have been analysed for cholesterol
content. Additionally 75 shellfish samples, 171 commercial samples and 92 gonads (roe) have
been analysed. The total number of analyses was approx. 4200.
Cholesterol determination by capillary gas chromatography

The method used is a modification of the method proposed by Naeemi and others (1995). The
method was modified by Oehlenschläger (1999, 2000). Modified and newly added steps in the
procedure are designated by (MOD).
200 – 250 mg (MOD) of freeze-dried, finely homogenised sample were weighed into a test
tube. 25 μl of internal standard solution (1 mg 5-α-cholestane/ml cyclohexane) and 2.5 ml
saturated methanolic potassium hydroxide solution (KOH) were added. The test tube was closed
by a screw cap and shaken well by hand. Subsequently it was heated in a water bath at +80 °C
for 30 min. It is advisable for complete saponification to shake the glass 1-2 times during
tempering. The test tube was then cooled down under running tap water until temperature was
binnenwerk.indd 46 21-08-2006 12:33:57

below 50 °C. Then 0.5 ml of magnesium chloride solution (25.41 g MgCl2 * 6 H2O/250 ml H2O)
(MOD) and 2.5 ml cyclohexane were added. The test tube was well shaken on a vortex mixer
for 2 min and then centrifuged in a table centrifuge at maximum speed. The upper phase was
completely transferred into another test tube filled with some anhydrous sodium sulphate (tip
of a spatula) to remove residual water (MOD). Again the test tube was well shaken by hand.
According to the cholesterol concentration in the sample the solution was diluted 1:10 (MOD)
by cyclohexane prior to application or directly applied into a vial for the auto-sampler of the
gas chromatograph.
Gas chromatograph: Hewlett Packard 5890 Series II plus with auto sampler HP 7673 and HP
chemstation 3365; capillary column 30 m * 0.32 mm Ø with HP-1 or HP-5 (cross linked methyl
siloxane); carrier and make-up gas helium; FID: synthetic air generated by Whatman zero air
generator Model 75-83-220 and hydrogen generated by Packard hydrogen generator 9100;
split inlet temperature 270 °C, FID temperature 300 °C, Helium flow 1.5 ml/min; temperature
programme: 180 °C to 280 °C at 20 °C/min, then 10 min at 280 °C followed by 10 min at
300 °C (MOD).
The limit of detection was 0.5 μg cholesterol/g sample corresponding to 0.5 mg cholesterol/
kg. The recovery rate was 95.7% with external standard and 97% with internal standard.
This method can also measure other sterols but in this investigation cholesterol only was
determined. Data are presented as median, 25% and 75% percentiles, respectively, and minimal
and maximal values.
Result and discussion
Gadoid fish species

The cholesterol content of gadoid species (Table 3) was generally low. The species with the
lowest (median) cholesterol content between 27 and 36 mg/100 g were: hake, blue ling, ling,
pollack, grenadier, haddock and whiting. Blue whiting and cod were also low and only saithe
and pout were higher than 40 mg/100 g. The variation was lowest in grenadier, ling, saithe
and blue whiting, whereas blue ling, pout, cod, hake and Pollack differed more depending of
catching area and season. These findings correspond well with those of Ackman and McLeod
(1988) reporting the following cholesterol levels for gadoid fish species: cod (35 mg/100 g),
cusk (32 mg/100 g), haddock (29 mg/100 g), hake (23 mg/100 g) and pollack (50 mg/100 g,
but are lower than the cholesterol contents given by Candela and others (1997) with approx. 73
mg/100 g and Copeman and Parrish (2004) with 69 mg/100 g (using the Iatroscan technique)
for cod.
Flat fish species

The cholesterol content of the flat fish species (Table 3) were in the same range like that of
the gadoid species. Most flat fish species showed cholesterol contents between 30 and 40
mg/100 g only flounder, dab, Greenland halibut and White halibut were just above 40 mg/100
g. The variation was biggest in plaice, sole and White halibut – in these species the cholesterol
content can be as low as 10 to 15 mg/100 g. Some specimen of Greenland halibut showed
cholesterol values above 60 mg/100 g.
binnenwerk.indd 47 21-08-2006 12:33:57

Table 3. Cholesterol contents (mg/100 g wet weight) in edible part of marine and freshwater fish, crustaceans
and molluscs, median, 25% and 75% percentiles and min/max values.
Species median 25% percentile 75% percentile minimum maximum
American plaice 38 35 42 22 50
Anchovy 30 19 30
Angler 33 28 42 19 54
Argentine 56 47 62 32 99
Atlantic Salmon 26 24 27
Black Scabbardfish 34 33 35 32 38
Blue ling 28 22 47 19 57
Blue mussel 18
Blue Whiting 38 37 43 34 50
Bream 52 46 57 44 66
Capelin 73 71 77
Cod 39 34 46 20 64
Cold water shrimp 107 193 126 81 119
Dab 43 40 47 39 50
Dogfish 42 37 47 29 69
Edible crab 29 30 31 23 42
Eel 51 31 70
Eelpout 85 71 01 65 93
Flounder 42 40 46 30 60
Flying squid 140 132 153 75 227
Forkbeard 31 27 33 24 33
Greenland Halibut 42 36 45 31 68
Grenadier 35 33 37 26 44
Gurnard 38 32 40 28 52
Haddock 36 35 42 27 54
Hake 27 23 34 15 47
Herring 31 27 37 22 103
Horse mackerel 44 41 51 23 59
Houting 38 34 41 25 47
John Dory 52 48 57
Lemon sole 33 24 39 17 45
Ling 31 28 34 21 44
Mackerel 33 24 43 14 55
Megrim 40 37 43 32 54
Mora moro 31 30 31
Norway lobster 97 88 103 81 120
Ocean perch 41 35 47 20 53
Octopus 85 75 103 66 119
Perch 71 60 78 42 84
Pikeperch 63 53 72 42 79
Plaice 31 23 40 12 52
Pollack 31 25 37 23 39
Pope, ruffe 88 84 94 83 99
Pout 47 40 41 20 57
binnenwerk.indd 48 21-08-2006 12:33:58

Table 3. Continued.
Species median 25% percentile 75% percentile minimum maximum
Roach 44 40 48 30 57
Saithe 45 40 50 27 69
Sandeel 37 36 38 32 44
Sardine 24 23 25 18 27
Smelt 70 64 81 54 109
Sole 33 28 38 8 47
Sprat 40 35 43 25 52
Trout 27 22 30 17 32
Turbot 39 32 47
White halibut 45 41 51 14 62
Whiting 36 32 44 23 50
Witch 36 33 37 31 48
Wolffish 43 38 48 15 68
Other marine fish species

The fish species with the lowest cholesterol content was identified as Atlantic salmon with 26
mg/100 g (Table 3). Extremely low were Mora moro, angler, black scabbardfish. Most species were
in the range between 30 and 40 mg/100g, but some species had significant higher cholesterol
contents like argentine (56 mg/100 g), smelt (70 mg/100 g) and capelin (73 mg/100 g). Both
latter species belong to the family Osmeridae and it might be a characteristic of this family
to contain more cholesterol in their edible parts than others. There is also evidence that the
cholesterol content in fish from tropical areas is higher than in fish from temperate climate
areas. Ewaidah (1993) assayed 11 species from the Arabian Gulf and found cholesterol contents
between 54 mg/100 g and 94 mg/100 g with an average of 65 mg/100 g. This is clearly above
the average for fish species from the Northeast Atlantic which is approx. 40 mg/100 g.
In this investigation the cholesterol content of sharks and rays was not included. However,
Lytle and Lytle (1994) have analysed four shark and ray species for their cholesterol content.
Cholesterol contents in Southern Stingray (Menticirrhus americanus) ranged from 57 to 69
mg/100 g and in Atlantic stingray (Dasyatis Sabina) from 30 to 39 mg/100 g. In blacktip
shark (Carcharhinus limbatus) the range was between 34 and 58 mg/100 g and in Atlantic
sharpnose shark (Rhizoprionodon terraenovae) between 16 and 23 mg/100 g. This shows that
the cholesterol contents in sharks and rays (cartilaginous fish species) is of the some order as
in bony fish species.
Small fatty pelagic fish

The pelagic species herring, mackerel, anchovy and sardine were all very low in cholesterol
(24 to 33 mg/100 g) despite of the fact that they are fatty fish species (Table 3). The low
cholesterol content found in sardine is in contrast to the high values (93 mg/100 g average)
reported for this species from the Adriatic coast by other authors (DeLeonardis A and Macciola V,
2004). Sprat and horse mackerel were somewhat higher but still in the low range. The variation
in all species was low with the exception of herring. The cholesterol content of herring can
binnenwerk.indd 49 21-08-2006 12:33:58

vary from 20 mg/100 g to just above 100 mg/100 g as a function of catching area, season,
state of maturity and other factors. However, the majority of specimen analysed do not vary
too much.
Freshwater fish species

The cholesterol content of the edible part of freshwater species (Table 3) was found to be
generally higher than the cholesterol content of marine fish species. Trout and houting are
exceptions, the other species exhibit cholesterol contents above 50 mg/100 g. High contents
were especially found in eelpout, pope or ruffe, which is not a commercially used food fish,
perch and pikeperch. Most of the freshwater species are lean fish species comparable in fat
content to gadoids. The higher cholesterol content in freshwater fish seems to be characteristic
for them. This is supported by cholesterol levels found in perch (92 mg/100 g), pike (Esox lucius)
(63 mg/100g), pikeperch (79 mg/100 g), vendace (74 mg/100 g) and whitefish (Coregonus
lavaretus) (49 mg/100 g) by Piironen and others (2002).
Roe (gonads) of fish species

In Figure 1 are shown the cholesterol contents in roe (fish eggs) of all species investigated. At
a first glance it can be seen that the cholesterol contents of the roe was 5 to 10 times higher
than the content in the edible part of fish (muscle, fillet). Most fish roe scattered around 200
mg/100g, but some species like hake and plaice were much higher. The reason for the extremely
high cholesterol content in roe is the fact that in the single egg enough cholesterol has to be
present to form the lipid bi-layers in the cells during nuclear division of the developing fish
embryo. Consequently, all fishery products processed on the basis of roe (caviar, smoked or
canned cod roe, caviar substitute, keta-caviar, trout-caviar, caviar spread etc.) always contain
high amounts of cholesterol. Very high cholesterol contents in the roe of rainbow trout and
vendace have been determined by Piironen and others (2002) with 409 mg/100 g and 743
mg/100 g, respectively.
Shellfish
The mussels investigated, scallop, blue mussel and ocean quahaug were with content around
or lower 30 mg/100 g very low in cholesterol. The crustaceans, Norway lobster and cold water
shrimp were all high in cholesterol (Table 3). The welk lies in the same area (120 mg/ 100
g). An exception forms the edible crab (Cancer pagurus), which is the only crustacean in
this investigation with a very low cholesterol content (29 mg/100 g). The two cephalopods
measured, flying squid and common octopus exhibit high cholesterol contents, 144 and 84.5
mg/100 g, respectively (Table 3).
Crustaceans and molluscs contain a number of other sterol as cholest-5,22-dien-3-ß-ol and 24-
methylcholesta-5,22-dien-3-ß-ol, but the main sterol component is always cholesterol. Piretti
and Serrazanetti (1980) could show that the prawn Penaeus cheraturus contained 94.5% (173.6
mg/100 g) of its total sterols as cholesterol and Sica (1980) demonstrated that 91.1% of all
sterols in Loligo vulgaris were cholesterol. In other crustacean species like in Squilla mantis
from the Mediterranean Sea (Serrazanetti and others 1990) the cholesterol content can be a
much lower fraction of the total sterol content (55.7%).
Cholesterol and fat content

Figure 2 is a plot of the cholesterol contents of all fish specimen investigated versus their
respective dry matter. Since dry matter in fish increases strictly parallel with fat content this
binnenwerk.indd 50 21-08-2006 12:33:59

550
500
Cholesterol (mg/100 g wet weight)
450
400
350
300
250
200
150
100
50
Sebastes sp.
Gadus morhua
Gadus morhua, Baltic
Platichthys flesus
Perca fluviatilis
Anarhichas minor
Clupea harengus
Gymnocephalus cernuus
Rutilus rutilus
Coregonus sp.
Cyclopterus lumpus
Psetta maxima
Stizostedion lucioperca
Molva dypterygia
Pollachius pollachius
Molva molva
Merlangius merlangus
Anarhichas lupus
Melanogrammus aeglefinus
Merluccius merluccius
Pollachius virens
Pleuronectes platessa
Rheinhardtius hippoglossoides
Hippoglossoides platessoides
Figure 1. Cholesterol content in gonads (roe) of some fish species investigated, 25% and 75% percentiles
and min/max values.
figure can be used for statements about cholesterol and fat content. In the figure there are
three areas. Area 1 to the lower left is an area formed by specimen with a low cholesterol
content and a low to medium fat content. The second area to the lower right, is formed by
the specimen with high fat content and low cholesterol content and the third area, to the
upper left is formed by species with a high cholesterol content but a low fat content. Area 3 is
formed mainly by freshwater samples, area 2 by fatty pelagic species and area 1 by the marine
fish samples. It is obvious that there are no samples in this investigation, which are found in
the area “high cholesterol, high fat content”, since this is not existing. It must be stressed
that almost all fatty fish samples are low in cholesterol and that somewhat higher cholesterol
contents are never found in fatty fish, but in lean freshwater fish species.
Figure 3 supports these findings. The plot of cholesterol contents in all mackerel measured versus
dry matter content (s.a.) shows that the cholesterol content decreases the fattier the fish is. This
can be explained by the fact that the cholesterol is a component of the cell membranes of the
fish tissue. The number of cells is fixed and only subject to minor variations. Since there is no
accumulation of cholesterol in depot fat, subcutaneous fat layers etc. the cholesterol content is
almost constant and if the fish accumulates fat the cholesterol content measured goes down.
binnenwerk.indd 51 21-08-2006 12:33:59

120
100
Cholesterol (mg/100 g)
80
60
40
20
0
10 20 30 40 50
dry matter (% wet weight)
Figure 2. Plot of cholesterol content and dry matter (fat) in edible part of all fish specimen investigated.
60
50
40
30
20
15 20 25 30 35 40 45
dry matter (%)
Figure 3. Plot of cholesterol content and dry matter of mackerel specimen with different fat contents.
binnenwerk.indd 52 21-08-2006 12:33:59

The situation, however, is totally different in fish roe (Figure 4). The plot shows that the
fattier the roe is (dry matter) the more the content of cholesterol. Here cholesterol has another
function. It is accumulated in a depot to be present when the cell division takes place to form
a part of the cell membrane. The fattier the fish roe, the more cholesterol it contains.
Commercial samples
The results of the commercial samples (Table 4) support the findings in the collected samples
onboard the vessels. Fish roe products (Russian caviar, caviar spread, canned cod roe) are all
high in cholesterol with Russian caviar being on top. Commercial quick frozen octopus is also
high in cholesterol exceeding the content of the octopus sampled onboard. North Sea shrimps
are highest in cholesterol of all crustaceans. Smoked salmon produced from Norwegian farmed
salmon is very low with 26 mg/100g. This corresponds well with the value for raw salmon in
Table 3 (20 mg/100 g) and can be explained by the water loss during processing of the smoked
product.
Imported tropical and subtropical shrimps gave almost identical contents in two independent
trials (140 and 150 mg/kg). High cholesterol contents in tropical and subtropical shrimps and
prawns have been reported by other authors: Krzynowek and Panunzio (1989) found in Northern
shrimp (Pandalus borealis) 135 - 186 mg/100 g, in Georgia white shrimp (Peneaus setiferous)
139 - 147 mg/100 g, in Honduras pink shrimp (Penaeus durarum notialis) and in Ecuador white
shrimp (Penaeus vannamei) 150 mg/100 g, in Texas brown shrimp (P. aztecus) 151 mg/100 g.
Also Gopakumar and Nair (1975) reported for Metapenaeus monocerus from brackish water
90 mg/100 g and for the marine species M. dobsoni 170 mg/100 g, M. affinis 133 mg/100 g, P.
indicus 118 mg/100 g and Parapenaeopsi stylifera 130 mg/100 g. This is supported by Mohd.
45
40
35
dry matter (%)
30
25
20
15
dry matter = 8.4805+0.0679*cholesterol
10
5
50 100 150 200 250 300 350 400 450 500 550
Figure 4. Plot of cholesterol content and dry matter of gonads (roe) of all species analysed.
binnenwerk.indd 53 21-08-2006 12:33:59

Table 4. Cholesterol content (mg cholesterol/100 g wet weight) in commercial samples of fishery products,
arithmetic mean and standard deviation.
Fishery product Arithmetic mean Standard deviation
Caviar, Russia, Malossol 273 29

Caviar spread, Scandinavia, different brands 97 83
Cod roe, canned, Denmark 154
Octopus, deep frozen 114 5
Deep frozen tropical and subtropical shrimps 141 19
North Sea shrimps 194 9
Deep frozen shrimps, imported, 1997 151 16
Smoked salmon fillets, Norway 27 4
Omar and others (1995) who also found significant amounts in Malaysian prawn species: P.
monodon 178 mg/100 g, P. merguiensis 157 mg/100 g, Metapenaeus brevicornis 160 mg/100
g, Parapenaeopsis hardwickii 168 mg/100 g and Parapenaeopsis sculptilis 186 mg/100 g. P.
aztecus was also investigated by Johnston and others (1983) and exhibited 200 mg/100 g
of free cholesterol. Uddin and others (2001) confirmed these findings by analysing shrimps
and prawns from the Bay of Bengal: P. indicus 161 mg/100 g, P. monodon 132 mg/100 g and
Metapenaeus monocerus 130 mg/100 g. In this paper also the extremely high cholesterol
contents in cephalopods were confirmed in two species: squid (Loligo duvauceli) 200 mg/100
g and cuttle fish (Sepia aculeate) 169 mg/100 g.
Seasonal and area variation

Cod was caught in 5 different areas from March to October in different years. From Figure 5
it seems that the lowest cholesterol content was found in cod caught in Greenland waters or
Faroese/Shetland waters. The highest content was found in the Barents-Sea. A closer look,
however to the large variations and the min/max values – with the exception of the cod caught
in coastal Baltic Sea – shows that there is no great variation in cholesterol content as a factor
of season and catching ground in cod. In Figure 6 a frequency distribution plot of cholesterol
content in approx. 200 specimen of cod from different catching grounds and seasons is shown
proving that the cholesterol content in edible part of cod is normally distributed if a sufficient
large number of specimen is analysed.
As shown the cholesterol content in marine fish is varying considerably inter and intra
species, therefore it is not possible to use the cholesterol content in edible part of fish for
the determination of the fish species as reported earlier and proposed by Wurziger and Hensel
(1967).
Conclusion
From the results presented it can be concluded that:

• fish muscle is low in cholesterol;
• fatty fish is very often low in cholesterol;
• freshwater fish is higher in cholesterol compared with marine fish;
• shrimps and prawns, squid and octopus are high in cholesterol;
binnenwerk.indd 54 21-08-2006 12:34:00

70
65
60
55
Cholesterol (mg/100 g)
50
45
40
35
30
25
20
15
Mar Apr May June Aug Sep Oct
Figure 5. Cholesterol content in edible part of cod (Gadus morhua) caught in different catching areas in
different months.
45
40
35
30
Observations
25
20
15
10
0
15 20 25 30 35 40 45 50 55 60 65 70
Figure 6. Frequency distribution of cholesterol content (mg/100 g on a wet weight basis) in edible part of
cod (Gadus morhua) from different catching areas, seasons and years.
binnenwerk.indd 55 21-08-2006 12:34:00

• mussels and edible crab are low in cholesterol;

• roe (gonads) are always very high in cholesterol;
• no significant correlations was found with season and catching ground;
• there is an intrinsic variation between specimen of a given species.
In commercial samples the cholesterol content of smoked salmon is low, in imported tropical
and subtropical shrimps and in local shrimps always high, in canned roe products and caviar
products always high and Russian caviar has the highest cholesterol content.
Acknowledgements
The author wants to thank the crews and the masters of the vessels for their assistance and
help and my co-workers for their enthusiasm in the work and their continuous efforts to create
this huge data base. Special thanks to Hans-Jürgen Knaack for sampling onboard and to Tanja
Pieplow for all the gas chromatographic analyses.
References
Ackman RG, McLeod C. 1988. Total lipids and nutritionally important fatty acids of some Nova Scotia fish and shellfish
food products. Can Inst Food Sci technol J 21:390-398.
Anon 1999. Nutrient value of some common foods. Canadian Government Publishing, Ottawa, 54 p.
Bieniarz K, Koldras M, Kaminski K, Mejza T. 2000. Fatty acids and cholesterol in some freshwater fish species in Poland.
Folia Univ Agric Stetin. 214(27):21-44.
Candela M, Astiasaran I, Bello J. 1997. Effects of frying and warmholding on fatty acids and cholesterol of sole (Solea
solea), codfish (Gadus morhua) and hake (Merluccius merluccius). Food Chem 58:227-231.
Copeman LA, Parrish CC. 2004. Lipids classes, fatty acids, and sterols in seafood from Gilbert Bay, Southern Labrador.
J Agric Food Chem 52:4872-4881.
DeKoning AJ, Hearshaw KD, van der Merwe G. 1993. Free and esterified cholesterol in a number of South African fish
oils and their corresponding meals. Fat Sci Technol 95:27-31.
De Leonardis A, Macciola V. 2004. A study on the lipid fraction of Adriatic sardine filets (Sardina pilchardus). Nahrung/
Food 48:209-212.
El-Sayed MM, Ezzat AA, Kandeel KM, Shaban FA. 1984. Biochemical studies on the lipid content of Tilapia nilotica and
Sparus auratus. Comp Biochem Physiol 79B:589-594.
Ewaidah EH. 1993. Cholesterol, fat and food energy content of selected raw and cooked commercial fish species from
the Arabian Gulf. Ecol Food Nutr 30:283-292.
Feeley RM, Criner PE, Watt BK. 1972. Cholesterol content of foods. J Amer Diet Ass 61:134-149.
Gopakumar K and Rajendranathan Nair M. 1975. Lipid composition of five species of Indian prawns. J Sci Food Agric
26:319-325.
Idler DR, Wiseman P. 1971. Sterols of crustacea. Int J Biochem 2:91-98.
Imre S, Saglik S. 1998. Fatty acid composition and cholesterol content of some Turkish fish species. Turk J Chem
22:321-324.
Iwasaki M, Harada R. 1984. Cholesterol content of fish gonads and livers. Bull Jap Soc Sci Fish 50:1623.
Iwasaki M, Harada R. 1985. Proximate and amino acid composition of the roe and muscle of selected marine species.
J Food Sci 50:1585-1587.
Johnston JJ, Ghanbari HA, Wheeler WB, Kirk JR. 1983. Characterization of shrimp lipids. J Food Sci 48:33-35.
Kanazawa A. 2001. Sterols in marine invertebrates. Fisheries Sci 67:997-1007.
King I, Childs MT, Dorsett C, Ostrander JG, Monsen ER. 1990. Shellfish: Proximate composition, minerals, fatty acids,
and sterols. J Am Diet Ass 90:677-685.
binnenwerk.indd 56 21-08-2006 12:34:00

Kritchevsky D, Tepper SA, DiTullo NW, Holmes WL. 1967. The sterols of seafood. J Food Sci 32:64-66.
Krzynowek J. 1985. Sterols and fatty acids in seafood. Food Technol 39:61-68.
Krzynowek J, D’Entrement DL, Murphy J. 1989. Proximate composition and fatty acid and cholesterol content of squid,
Loligo pealei and Illex illecebrosus. J Food Sci 54:45-48.
Krzynowek J,, Panunzio LJ. 1989. Cholesterol and fatty acids in several species of shrimp. J Food Sci 54:237-239.
Krzynowek J, Uljua DS, Panunzio LJ, Maney RS. 1992. Factors affecting fat, cholesterol, and Omega-3 fatty acids in
Maine sardines. J Food Sci 57:63-65+111.
Lahti E. 1987. Total lipid and cholesterol contents of liver and muscle in some fish species, especially vendace
(Coregonus albula L.) in Finland. Arch Hydrobiol. 110:133-142.
Lytle JS, Lytle TF. 1994. Fatty acid and cholesterol content of sharks and rays. J Food Comp Anal 7:110-118.
Mathew S, Ammu K, Viswanathan Nair PG, Devadasan K. 1999. Cholesterol content of Indian fish and shellfish. Food
Chem 66:455-461.
Mohd. Omar AK, Abd Malik O, Ismail N. 1995. Cholesterol content of some Malaysian marine prawn species. ASEAN
Food J 10:39-40.
Moreira AB, Visentainer JV, de Souza NE, Matsushita M. 2001. Fatty acids profile and cholesterol contents of three
Brazilian Brycon freshwater fishes. J Food Comp Analysis 14:565-574.
Naeemi ED, Ahmad N, Al-sharrah TKM, Behbahani M. 1995. Rapid and simple method for determination of cholesterol
in processed food. JAOAC Int. 78:1522-1525
Nettleton JA, Exler J. 1992. Nutrients in wild and farmed fish and shellfish. J Food Sci 57:257-260.
Oehlenschläger J. 2000. Cholesterol content in edible part of marine fatty pelagic fish species and other seafood. In:
Georgakis SA, editor, Proceedings 29th WEFTA Meeting 1999,. Thessaloniki: Greek Society of Food Hygienists and
Technologists. p 107-115
Oehlenschläger J. 1999. Gaschromatographische Bestimmung des Gesamtcholesterolgehaltes im verzehrbaren Anteil
(Filet) von Meeresfischen und Garnelen. Lebensmittelchemie 53:86.
Oehlenschläger J. 1998. Cholesterol in Krebstieren – Analytik und Gehalte, Angewandte Wissenschaft, Heft 469:75-
81.
Osmann H, Suriah AR, Law EC. 2001. Fatty acid composition and cholesterol content of selected marine fish in Malaysian
waters. Food Chem 73:55-60.
Piironen V, Toivo J, Lampi A-M. 2002. New data for cholesterol contents in meat, fish, milk, eggs and their products
consumed in Finland. J Food Comp Analysis 15:705-713.
Piretti MV, Serrazanetti GP. 1980. Investigation of the sterol constituents of the prawn Penaeus cheraturus. J Amer
Diet Ass 61:134-149.
Rosa R, Nunes ML. 2003. Nutritional quality of red shrimp, Aristeus antennatus (Risso), pink shrimp, Parapenaeus
longirostris (Lucas) and Norway lobster, Nephrops norvegicus (Linnaeus). J Sci Food Agric 84:89-94.
Serrazanetti GP, Conte LS, Baraldi R. 1990. Sterol content in muscular tissue of Squilla mantis. Comp Biochem Physiol
96B:811-814.
Sica D. 1980. Sterols from some molluscs. Comp Biochem Physiol 65B:407-410.
Souci SW, Fachmann W, Kraut H. 2000. Food Composition and Nutrition Tables. CRC Press, New York, 1182 p.
Takada K, Aoki T, Kunisaki N. 1988. Proximate composition, free amino acid, fatty acid, mineral and cholesterol content
in imported frozen shrimps. Nippon Suisan Gakkaishi 54:2137-2179.
Teshima S-I 1991. Sterols of crustaceans, molluscs and fish. In: Patterson GW, Nes WD, editors, Physiology and
biochemistry of sterols. Champaign: Amer Oil Chem Soc. p 229-256
Uddin M, Jahan P, Ahmad MU. 2001. Fat and cholesterol content in some fish and shellfish of Bay of Bengal. Asian J
Chem 13:1227-1230.
Wurziger J, Hensel G. 1967. Über den Cholesterin-Gehalt in Fischfilets zur Ermittlung der Fischart. Fette Seifen
Anstrichm 69:937-942.
binnenwerk.indd 57 21-08-2006 12:34:00

binnenwerk.indd 58 21-08-2006 12:34:00
Capelin oil for human consumption
Icelandic Fisheries Laboratories, Skulagata 4, 101 Reykjavik, Iceland
Abstract
Oil-in-water emulsion with capelin oil was made as egg-free mayonnaise containing whey
protein emulsifier. Mayonnaise was enriched with 20% fish oil, with and without antioxidant
(tocopherol mixture (200 mg/kg)), in addition to a control sample without fish oil. The quality
and stability at 10 °C was evaluated by conjugated dienes/ trienes and sensory analysis,
besides stability tests at 60 °C. Lipid oxidation of the mayonnaise was not increased by fish
oil addition, but the fish oil flavour was apparent. Tocopherol was ineffective as antioxidant
in the fish oil enriched mayonnaise and tended to increase oxidative degradation. Processing
optimisation is necessary in order to conclude if capelin oil is suitable in functional foods with
sensory acceptance, but this study indicated that adequate stability could be obtained.
Keywords: fish oil, mayonnaise, lipid oxidation, sensory evaluation
Introduction
The significance of fish oils as important dietary source of valuable fatty acids of long chained
omega-3 type has been confirmed by research. The diverse beneficial health effects of a diet
high in long chain n-3 polyunsaturated fatty acids has been demonstrated by epidemiological
studies and reviewed extensively (Uauy and Valenzuela 2000; Simopoulos and Cleland 2003;
Schmidt and others 2004). In order to increase the consumption of marine fatty acids several
attempts have been done to incorporate fish oil into different foods (Young 1990; Kolanowski
and others 1999; Medina and others 2003). Incorporation of omega-3 fatty acids and fish
oils into functional food is limited by their high susceptibility to oxidative degradation. Oil-
in-water emulsions may be an effective method to deliver omega-3 fatty acids into foods,
rather than bulk lipids, because emulsions are more frequently found in actual food products
(Coupland and McClements 1996; McClements and Decker 2000). An emulsion consists of thee
regions: the interior of a droplet, the continuous phase and the interfacial region, where
systems consisting of oil droplets dispersed in an aqueous phase are known as an oil-in water
emulsion (Coupland and McClements 1996). Emulsifiers are situated at the oil-water interface
because they contain both hydrophilic and hydrophobic groups, and their function is to prevent
emulsions from separating into oil and water phases. Mayonnaise is an example of oil-in-water
emulsion that is traditionally made of egg yolk as the emulsifying agent. Other ingredients
in mayonnaise are oil, water, vinegar, salt, sugar and mustard. Potassium sorbate and sodium
benzoate are often added to mayonnaise to inhibit microbial growth. Vinegar, salt, sugar and
mustard are added to mayonnaise as flavouring ingredients, but all of these ingredients also
seem to play an important role for the physical stability of emulsions (McClements and Decker
2000). Besides egg yolk, many emulsifying agents have been applied in oil-in-water emulsions.
Whey proteins have gained much attention as they have been found to increase oxidative
stability of emulsions, including those containing omega-3 fatty acids (Hu and others 2003,
Djordjevic and others 2004b). Whey proteins are believed to inhibit oxidation by chelating
binnenwerk.indd 59 21-08-2006 12:34:01

prooxidant transition metal ions, inactivate free radicals or by forming physical barriers between
water-soluble prooxidants and lipids at the lipid-water interface (Donnelly and others 1998;
McClements and Decker 2000).
The fish oil consumption in Iceland has to date almost entirely been in the form of a daily spoon
of cod liver oil as a health remedy. Other sources of fish oils derive from fish meal production
on small, whole fish, such as capelin (Mallotus villosus), which is the major part of the Icelandic
fish oil production. To date little efforts have been made in order to exploit the capelin oil for
human consumption. The capelin oil produced in Iceland is mainly exported as an ingredient for
use in aquaculture and animal feeds, and the value of this feed grade fish oil is less than half
the price of fish oil for human consumption (e.g. cod-liver oil). Consequently, the possibilities
to increase the value of capelin oil by incorporation it into food products are encouraging. In
order to achieve food grade quality capelin oil it is necessary to take into account the natural
variation in the capelin stock as raw material and it might also require optimisation of handling
and processing techniques. The oil content in capelin can vary from 2–20% depending on
season, and the quality of the crude oil can vary as well, as reflected in the content of free
fatty acids, the content of natural antioxidants like tocopherol and astaxanthin, as well as in
fatty acid profile (Bragadóttir and others 2002). The fatty acid profile of capelin is unusual,
with extraordinary high concentration of long chained monounsaturated fatty acids (MUFA´s),
ranging from 46-57% and omega-3 fatty acids (C20:5 + C22:5 + C22:6) ranging from 12.5 to
18% of total fatty acids (Bragadóttir and others 2002).
The present investigation was undertaken to evaluate the possibilities to produce food products
containing capelin oil with emphasis on the best suitable means to prevent lipid oxidation.
Capelin oil was included in egg-free mayonnaise, containing commercial emulsifier mixture with
milk proteins. Mayonnaise made with milk proteins was selected as a food product for capelin
oil incorporation, as oil-in-water emulsions made with milk proteins have been extensively
studied for the past years, and found to be promising to prevent lipid oxidation and deliver
omega-3 fatty acids into foods.
Materials
The crude fish oil from capelin (Mallotus villosus) was provided by Síldarvinnslan hf., fish meal
factory in Siglufjördur, during winter season. The fish oil was distilled in a bench top molecular
distiller, type: KDL (UIC GmbH, Alzenau-Hörstein, Germany) at 185 ± 2 °C. The oil was packed
under nitrogen gas into 1 L brown flasks and kept at -24 °C until used. The composition of fatty
acids was; saturated fatty acids: 16.8%, monounsaturated fatty acids: 73.3%, polyunsaturated
fatty acids: 9.9%, and thereof n-3 fatty acids: 5.9%. Soybean oil was obtained from Kjarnavörur
hf, Garðabær, Iceland, the importer of Victoria-refined and deodorized soya oil (produced in
Holland by Vereenigde Oil Fabrieken). The composition of fatty acids was; saturated fatty acids:
14.2%, monounsaturated fatty acids: 25.2%, polyunsaturated fatty acids: 60.7%, and thereof
n-3 fatty acids: 6.7%. Coviox T-70, natural 70% tocopherol mixture was purchased from Cognis
GmbH (Düsseldorf, Germany). The vinegar, mustard, salt (NaCl) and sugar were purchased from
a local supermarket. Sodium benzoate was purchased from Merck (Darmstadt, Germany). The
Grindsted™ FF1110 stabiliser system was provided by Danisco A/S (Langebrogade, Copenhagen),
containing; milk protein (whey protein isolate), acetylated distarch adipate (E1422), guar gum
(E412) and sodium alginate (E401).
binnenwerk.indd 60 21-08-2006 12:34:01

Preparation of mayonnaise
The mayonnaise was made by a recipe from Danisco, Culinary Manual (Langebrogade,
Copenhagen), on 70% egg free mayonnaise, by a cold batch process in a mixer (Braun Electronic,
type 4265, Germany). Each batch contained by weight; distilled water (18.15%), salt (1.00%),
sugar (2.00%) and sodium benzoate (0.10%) that were mixed together. Grindsted™ FF1110
(1.40%) was pre-mixed with oil in a ratio of 1:2 and added to the mixture. The oil (70.00%)
was continuously emulsified into the water phase for 25 min. Finally, 10% vinegar (3.50%)
and mustard (1.50%) were blended together and added to the emulsion. Three samples were
prepared; one with soya oil and two with soya oil mixed with fish oil, with and without addition
of tocopherol as antioxidant (Table 1). For shelf life testing, the mayonnaise samples were
vacuum packed into glass jars (100 mL) for storage in the dark at 10-12 °C, to induce oxidation
and resemble inadequate refrigeration during storage and in the chill chain of the product.
Free fatty acids

Free fatty acids (FFA) were determined in the oils that were solubilized in alcohol/diethyl ether
(1:1) and titrated with diluted NaOH (AOCS 1998).
Peroxide value
Peroxide value of oils was measured by iodometric titration according to AOAC official method
965.33 (AOAC 1990).
Anisidine value
Anisidine value of oils was determined by the reaction of aldehydic compounds in oil and
p-anisidine, and absorbance measured at 350 nm, according to standard methods (IUPAC
1987).
Conjugated dienes (CD) and conjugated trienes (CT)

The determination of the absorbance in the UV spectrum of the samples was measured at 232
nm as conjugated dienes and at 268 nm as conjugated trienes according to standard methods
(IUPAC 1987), with minor modifications. Mayonnaise emulsions (0.150 g) were dissolved in
methanol (20-25 mL) and mixed for 20 seconds with Ultra-Turrax homogenizer (type T25, IKA
Werke, Staufen, Germany). The samples were centrifuged at 6500g for 5 min and the absorbance
of the supernants measured. Two measurements were performed on duplicate samples and the
results expressed according to the following formula: CD or CT = A232 or A268 / c x d; where A
is the absorbance reading at 232 or 268 nm, c denotes the concentration of the solution in g
per 100 mL and d is the length of the cell, in cm.
Table 1. Combinations of oils for the mayonnaise samples.
Code name Soya oil (%) Fish oil (%) Tocopherol (mg/kg oil)
S 100 - -
F 80 20 -
T 80 20 200
binnenwerk.indd 61 21-08-2006 12:34:01

pH
The pH was measured using a puncture, combination electrode (SE 104, Mettler Toledo, Greifensee,
Switzerland) connected to a pH meter (Knick-Portamess 913 pH, Berlin, Germany).
Sensory analysis
Samples were evaluated by Quantitative Descriptive Analysis (QDA) method (Stone and Sidel
1985). The method assumes detailed description of a product, such as odour, flavour, appearance
and texture. List of attributes are defined and used with unstructured scale. The Icelandic
Fisheries Laboratories sensory panel was trained in two sessions. Members had several years of
experience in evaluating rancidity of fish, fish oils and vegetable oils and have been trained
according to international standards (ISO 1993). Freshly made mayonnaise with pure soya oil,
soya and fish oil (70:30) as well as soya and rancid fish oil (70:30) was used for training the
panel. The panel compiled 16 descriptive attributes for mayonnaise, ranging from not present to
strong; each for both odour and flavour: acetic acid, oily, musty, painty, fish oil, rancid, acidic,
and for texture and appearance; creamy, clammy and colour (white/yellow).
Sensory assessments were carried out by seven to eight assessors (age range 30 - 60). Six
samples were evaluated (three different samples each in duplicate) in two sessions, three at
each. Sensory analysis, data collection and data analysis were done in the sensory program Fizz
version 1.3 (Biosystemes, France). The order of presentation of samples to the panelists was
balanced to minimize possible carry-over effects between samples. All observations of samples
were conducted under standardized conditions, with as little interruption as possible, at room
temperature, and under white fluorescent light. The mayonnaise was presented to the panelists
in small, transparent, disposable plastic cups covered with an aluminium foil, along with water
and crackers for oral rinsing between samples.
Oxidative stability
The oxidative stability of mayonnaise was measured electronically under oxygen pressure (5
bars) in an Oxipres apparatus (Mikrolab Aarhus A/S, Højbjerg, Denmark). Samples (7 g) were
weighed into reaction flasks (125 mL) and the pressure signal was recorded at 60 °C. Each
sample was measured in duplicate and the results presented as mean values.
Data analysis
Statistical analysis was done on the data by analysis of variance (ANOVA) on Number Cruncher
Statistical Software (NCSS 2000 and Pass Trial, Kaysville, Utah). Duncan comparison test used
to determine differences between samples (P < 0.05).
Measurements on oils as raw materials in mayonnaise

Efforts were made to ensure that the oils used as raw materials for the mayonnaise were of
high quality. The peroxide value (PV) of the soya oil was 1.0 meq/kg, whereas the PV of the
distilled fish oil was not detected. The anisidine value (AV) of the fish oil was therefore also
measured to verify its quality, and was found to be low, with AV of 2.5. The range of AV in
commercial capelin oil (58 samples) measured at our lab were approximately 2 to 17, with an
average of roughly 6. These samples were mostly made up of crude capelin oil that usually
has lower AV than more processed oil. Thus, the AV of crude capelin oil was found to increase
from 7.4 to 8.5 in alkali-refined oil, increasing further to 16.2 after bleaching of the oil, and
binnenwerk.indd 62 21-08-2006 12:34:01

the AV ended in 21.2 after deodorization (Bragadóttir and others 1992). The fish oil in this
study was therefore only refined by molecular distillation in order to minimize the negative
effects of advanced processing on its stability. Oil purification by molecular distillation has
been described as an effective method for deodorizing and improving flavour of fish oil for
human consumption (Dinamarca and others 1990). Furthermore, it removes contaminants like
chlorinated hydrocarbons, free fatty acids, oxidation products and cholesterol (Bimbo 1998). In
this study, the distillation of the fish oil decreased the content of fatty acids (FFA) from 2.04%
in the crude oil to 0.3%, and the AV decreased from 4.6 to 2.5.
Conjugated dienes and conjugated trienes

Absorbance around 232 nm is a measure of conjugated dienes (CD) which may result from
decomposition of linoleic hydroperoxides (IUPAC 1987), an initial product of oxidation. This
measurement has shown high correlation with other measurements of primary oxidation like
peroxide value in olive oil triacylglycerols (Gómez-Alonso and others 2004), as well as with
headspace oxygen in soybean oil (Chung and others 2004). Secondary products of autoxidation
and, particularly ethylenic diketones, show an absorption band at approximately 268 nm together
with conjugated trienes (CT) (IUPAC 1987). This measurement has shown good correlation with
other measures of oxidation products as dimers and polymers of triacylglycerols from olive oil
(Gómez-Alonso and others 2004). The CD values of mayonnaise in this study increased during
the first week of storage from approximately 0.4 to 1.3-1.6 (Figure 1). Little difference was
observed in CD-values between samples, although the control sample S (mayonnaise without
fish oil) ended at a higher value of approximately 2, compared to 1.8 in the F sample, consisting
of fish oil enriched mayonnaise without antioxidant addition (P < 0.05).
The CT values showed similar behaviour, with an incensement in the first week of storage, a lag
phase between 1 and 3 weeks and a subsequent incensement (Figure 2). The values increased
2.5
1.5
S
CD
T
1 F
0.5
0
0 1 2 3 4 5 6
Storagetime at 10°C (weeks)
Figure 1. Changes in conjugated dienes (CD) during storage of mayonnaise (n = 2). S: mayonnaise with
100% soya oil, F: mayonnaise with soya and fish oil (80:20), T: same as F with tocopherol as antioxidant
(200 mg/kg).
binnenwerk.indd 63 21-08-2006 12:34:01

0.8
0.6
S
CT
T
0.4 F
0.2
0
0 1 2 3 4 5 6
Figure 2. Changes in conjugated trienes (CT) during storage of mayonnaise (n = 2). For abbreviations, see
Figure 1.
from approximately 0.15 to 0.6-0.8, and the control sample (S) had in fact higher CT values
than the F sample after 1 and 6 weeks storage (P < 0.05). While the tocopherol addition
(sample T) did not improve the stability of the fish oil mayonnaise as measured by CD or CT.
Sensory analyses
The “home made”, egg-free mayonnaise prepared with diverse ingredients according to producer’s
recipe, contained 3.5% vinegar of 10% concentration, resulted in rather sour mayonnaise with
pH of approximately 3.8 in all samples. The vinegar odour and flavour were also the most
dominant sensory attributes for all samples, along with a thick and creamy texture (Table 2).
The pH of emulsions has been found to play an important role for the stability of emulsions
as the activity of proteins, such as whey proteins, has been found to be greatest at pH values
below the pI of the proteins (Hu and others 2003). This effect has been attributed to the ability
of the proteins to generate a positive electrical charge on the oil droplets, thereby repelling
positively charged transition metal ions (McClements and Decker 2000). The same research team
working with whey protein isolate at pH 3 found that it could be used effectively to protect
polyunsaturated lipids from oxidation (Djordjevic and others 2004a, 2004b).
The colour of fish oil added mayonnaise (both F and T) was more yellow than soya oil mayonnaise
(S) (P < 0.05). This colour difference between the two oils was obvious, soya oil being very
pale yellow and the fish oil almost orange in colour, which resulted in a yellow mayonnaise in
combination with soya oil, which alone produced almost white mayonnaise.
Odour attributes of mayonnaise revealed little differences between samples except for fish oil
odour, where the S sample showed tendencies to be lower than the other samples, although only
significantly lower than sample T after 6 weeks storage (P < 0.05). The S sample had values for
fish oil odour ranging from 1 to 6, where as the F sample had values ranging from 11-22 and
the T sample had values from 11 and reaching 27 at the end of the storage time (Table 2).
binnenwerk.indd 64 21-08-2006 12:34:02

Table 2. Sensory scores (means, n = 2) for descriptive attributes of mayonnaise during storage on scale
from 0-100 (not present to strong). Different superscripted letters indicate significant difference between
samples within column (P < 0.05).
Odour Flavour Texture

Sample*
clammy
vinegar
vinegar
creamy
fish oil
fish oil
painty
painty
colour
rancid
rancid
Weeks
musty
musty
thick
oily
oily
0 S 60 9 4 8 1a 1 55 17 10 9 0a 4a 64 66 19 13a
T 57 16 5 7 11 4 57 17 11 8 27b 8 57 57 18 56b
F 57 16 5 7 11 4 57 17 11 8 27b 8 57 57 18 56b
1 S 57 20 2 4 4ab 0 60 29 2 5 4a 0 66 68 19 12a
T 56 18 2 10 15 3 57 21 6 11 30b 6 64 58 20 59b
F 52 17 3 12 22bc 3 51 22 6 12 31bc 7 60 59 19 64b
3 S 61 26 3 8 6ab 2 58 30 9 13 9ac 2 65 63 22 12a

T 57 18 10 13 13 5 60 23 9 19 28b 11 67 66 28 53b
F 59 21 4 9 10 2 58 22 6 6 28b 3 53 57 22 61b
6 S 53 22 4 9 2a 1 56 28 9 16 2a 5 63 66 25 9a
T 55 16 6 11 27c 8 58 22 12 13 33b 16b 65 65 24 57b
F 58 20 6 13 14 2 56 22 11 17 30bc 6 56 58 23 54b
*S: mayonnaise with soya oil, F: mayonnaise with soya and fish oil (80:20), T: same as F with
tocopherol as antioxidant (200 mg/kg).
The score for fish oil flavour was higher for both F and T samples than the S sample throughout
the storage time (P < 0.05). The fish oil flavour did however not increase in these samples
during the storage, ranging from 0 to 9 in the S sample, but around 30 for the F and T samples
throughout the storage time (Figure 3).
There was a more rancid tendency during storage of the T sample than in S sample (P = 0.07),
ending in rancidity score of 16 for T but 5 for the S sample (Figure 4). The F sample was
surprisingly more in range with the S sample, with rancidity scores from 3 to 8.
Comparable observations have been reported from studies with conventional egg yolk
mayonnaise, where mayonnaise containing 16% fish oil did not oxidise faster than mayonnaise
without fish oil, judged from chemical parameters such as peroxide value and anisidine values
(Jacobsen and others 1999). However, unpleasant off-odours and off-flavours developed much
faster in the fish oil enriched mayonnaise. The authors proposed that the fishy off-flavour
compounds, in fish oil enriched mayonnaise might be caused by volatile compounds in trace
amounts, with low sensory threshold, present in the water phase of the mayonnaise. That
might explain why the oxidation was not higher in fish oil enriched mayonnaise, because the
measurements were done on the lipid fraction and not in the water phase. Furthermore, the
binnenwerk.indd 65 21-08-2006 12:34:02

45
40
Sensory score (0-100) 35
30
25 S
T
20
F
15
10
5
0
0 1 2 3 4 5 6
Figure 3. Evaluation of fish oil flavour by sensory panel (n = 2). For abbreviations, see Figure 1.
20
18
Sensory score (0-100)
16
14
12
S
10 T
8 F
6
4
2
0
0 1 2 3 4 5 6
Figure 4. Evaluation of rancid flavour by sensory panel (n = 2). For abbreviations, see Figure 1.
measurement on volatile compounds may not be as sensitive to small amounts of off-flavour

compounds as the sensory panel.
The tocopherol concentration chosen in this study (200 mg/kg of mixed tocopherols) was the
same as conventional addition to cod liver oil, in the fish oil industry. The tocopherol addition
did not improve the stability or the sensory quality of the mayonnaise in this study, but showed
tendencies to induce rancidity during storage according to sensory evaluation. These results
were in fair agreement with the work of Jacobsen and others (2000) who found no difference
in the development of volatile off-flavours as evaluated by GC-MS, but the peroxide values were
slightly increased in tocopherol (200 mg/kg) added mayonnaise. The sensory perception of the
binnenwerk.indd 66 21-08-2006 12:34:03

fish oil enriched mayonnaise was on the other hand not affected by the tocopherol addition
in their study. Likewise, tocopherol did not appear to be an efficient antioxidant in another
study with both water-dispersible tocopherols and oil-soluble tocopherols in 20-280 mg/kg
concentrations added to 16% fish oil enriched mayonnaise (Jacobsen and others 2001). In a
study with fish oil enriched milk emulsion, addition of a tocopherol mixture did not increase the
stability of the milk emulsion, while rapeseed oil containing natural tocopherols in combination
with fish oil, inhibited oxidation (Let and others 2005). Research with microencapsulated fish
oil, produced in emulsions with coating materials and microencapsulated by spray drying, showed
that samples containing DL-α-tocopherol (1000 mg/kg) were more stable than unprotected
samples (Jónsdóttir and others 2005). These results indicate that addition of tocopherols to
fish oil emulsions is a delicate balance between antioxidative and pro-oxidative factors, and
the content of endogenous tocopherols in the lipids needs to be considered.
Comparison of testing methods

Stability test of mayonnaise samples at 60 °C in Oxipres revealed very little difference between
samples, as the pressure drop occurred at almost the same time in all samples (results not
shown).
The results from the chemical measurements (CD and CT), as well as the Oxipres stability
test, indicated that there was little difference in the oxidative stability of the control soya
oil mayonnaise and the fish oil enriched mayonnaise. In fact, the results give the impression
that the fish oil enriched mayonnaise was slightly more stable during storage. The sensory
analysis gave on the other hand clear message; the enrichment of fish oil into mayonnaise
could be detected by fish oil flavour (and odour) with sensory scores around 4 for the control
mayonnaise and 30 (on a scale from 0–100) for fish oil enriched mayonnaise. Rancid flavour
was more pronounced already in the freshly made fish oil enriched mayonnaise, which indicates
that it might have been confused with the fish oil flavour, which remained essentially stable
throughout the storage experiment. The capelin oil used in this study was only refined by
molecular distillation in lab-scale in order to use as few intervening processing steps as possible
in order to maintain endogenous antioxidants and minimise oxidation. Because processes like
bleaching and deodorisation cause major loss of retinols, tocopherols and other constituents
with antioxidant activity as well as health beneficial effects (Scott and Latshaw 1991; Dunford
2001). As a result, the fish oil flavour might have been more pronounced than by conventional
fish oil production involving alkali refinement, bleaching and deodorisation. More optimal
conditions regarding refinement of the capelin oil, concentration of the capelin oil as well as
incorporation of the mayonnaise into some tasteful seafood, are only to mention few of the
factors that could be investigated in order to verify if capelin oil could be generously applied
for human consumption.
Conclusions
The focus in this preliminary study was to investigate the use of capelin oil in mayonnaise,
with emphasis on sensory acceptance. The lab-scale refinement of capelin oil and production
of mayonnaise replacing 20% of soya oil with capelin oil resulted in a mayonnaise with a
significant fish oil flavour. Lipid oxidation of the mayonnaise was not increased by fish oil
addition although sensory scores for rancidity were higher for the fish oil enriched mayonnaise
that contained tocopherol as antioxidant. Tocopherol was therefore ineffective as antioxidant in
the fish oil enriched mayonnaise using 200 mg/kg of mixed tocopherols, but tended to increase
binnenwerk.indd 67 21-08-2006 12:34:03

oxidative degradation in this study. Optimisation of processing parameters by studies on the

effect of raw material qualities and refinement techniques are necessary in order to conclude
if capelin oil can be used in functional foods with adequate sensory acceptance. Potential
possibilities are, however, in sight for capelin oil, as this study indicated that acceptable
stability regarding lipid oxidation could be obtained.
Acknowledgements
We gratefully thank the AVS fund for financial support, Guðjón Rúnarsson at Kjarnavörur Ltd for
advice and for providing the soya oil, Þórhallur Jónasson at SVN Ltd for his cooperation and
for providing the fish oil, Danisco A/S for providing the Grindsted stabiliser system and Jón
Ögmundsson and Eiríkur Kristinsson at Lýsi Ltd for their collaboration. Special thanks to the
sensory panellists of IFL for their generous contribution.
References
AOAC. 1990. Official methods of analysis, 15th ed. Arlington, Virginia: Association of Official Analytical Chemists.
AOCS. 1998. Official methods and recommended practices of the American Oil Chemists Society. 5th ed. Champaign,
Illinois: American Oil Chemists´ Society.
Bimbo, AP. 1998. Guidelines for characterizing food-grade fish oil. Inform 9(5):473-483.
Bragadóttir M, Pálmadóttir H, Kristbergsson K. 2002. Seasonal changes in chemical composition and quality parameters
of capelin (Mallotus villosus). J Aquatic Food Prod Technol 11(3/4):87-103.
Bragadóttir M, Þórisson S, Hjaltason B. 1992. Þránun Lýsis. Rannsóknastofnun fiskiðnaðarins. Rit Rf 32:1-35.
Chung HJ, Colakoglu AS, Min DB. 2004. Relationships among headspace oxygen, peroxide value, and conjugated diene
content of soybean oil. J Food Sci 69(2):FTC83-FTC88.
Coupland JN, McClements DJ. 1996 Lipid oxidation in food emulsions. Trends Food Sci Technol 7:83-91.
Dinamarca E, Garrido F, Valenzuela A. 1990. Simple high vacuum distillation equipment for deodorizing fish oil for
human consumption. Lipids 25(3):170-171.
Djordjevic D, Kim HJ, McClements DJ, Decker EA. 2004a. Physical stability of whey protein-stabilized oil-in-water
emulsions at pH 3: Potential ω-3 fatty acid delivery systems (Part A). J Food Sci 69(5):C351-C355.
Djordjevic D, McClements DJ, Decker EA. 2004b. Physical stability of whey protein-stabilized oil-in-water emulsions at
pH 3: Potential ω-3 fatty acid delivery systems (Part B). J Food Sci 69(5):C356-C362.
Donnelly JL, Decker EA, McClements DJ. 1998. Iron-catalyzed oxidation of menhaden oil as affected by emulsifiers. J
Food Sci 63(6):997-1000.
Dunford NT. 2001 Health benefits and processing of lipid-based nutritionals. Food-Technol 55(11):38, 40-44.
Gómez-Alonso S, Salvador MD, Fegapane G. 2004. Evolution of the oxidation process in olive oil triacylglycerol under
accelerated storage conditions (40-60°C). J Am Oil Chem Soc 81(2):177-184.
Hu M, McClements DJ, Decker EA. 2003. Impact of whey protein emulsifiers on the oxidative stability of salmon oil-in-
water emulsions. J Agric Food Chem 51:1435–1439.
ISO. 1993. Sensory analysis - general guidance for the selection, training and monitoring of assessors. Part 1: Selected
assessors, 8586-1. Genf, Switzerland: The International Organization for Standardization.
IUPAC. 1987. Standard Methods for the Analysis of Oils, Fats and Derivatives. 7th ed. Oxford, UK: Blackwell Scientific
Publications.
Jacobsen C, Hartvigsen K, Lund P, Meyer AS, Adler-Nissen J, Holstborg J, Hølmer G. 1999. Oxidation in fish-oil-enriched
mayonnaise. 1. Assessment of propyl gallate as an antioxidant by discriminant partial least squares regression
analysis. Eur Food Res Technol 210:13-30.
binnenwerk.indd 68 21-08-2006 12:34:03

Jacobsen C, Hartvigsen K, Lund P, Meyer AS, Adler-Nissen J, Holstborg J, Hølmer G. 2000. Oxidation in fish-oil-enriched
mayonnaise. 2. Assessment of the efficacy of different tocopherol antioxidant systems by discriminal partial least
squares regression analysis. Eur Food Res Technol 210:242-257.
Jacobsen C, Hartvigsen K, Lund P, Thomsen MK, Skibsted LH, Hølmer G, Adler-Nissen J, Meyer AS. 2001. Oxidation
in fish oil-enriched mayonnaise: 4. Effect of tocopherol concentration on oxidative deterioration. Eur Food Res
Technol 212:308-318.
Jónsdóttir R, Bragadóttir M, Arnarson GÖ. 2005. Oxidatively derived volatile compounds in microencapsulated fish oil
monitored by solid-phase microextraction (SPME). J Food Sci 70(7):C433-C440.
Kolanowski W, Swiderski F, Berger S. 1999. Possibilities of fish oil application for food products enrichment with omega-
3 PUFA. Int J Food Sci Nutr 50:39-49.
Let MB, Jacobsen C, Pham KA, Meyer AS. 2005. Protection against oxidation of fish-oil-enriched milk emulsions through
addition of rapeseed oil or antioxidants. J Agric Food Chem 53:5429-5437
McClements DJ, Decker EA. 2000. Lipid oxidation in oil-in water emulsions: Impact of molecular environment on
chemical reactions in heterogeneous food systems. J Food Sci 65:1270-1282.
Medina I, González MJ, Pazos M, Medaglia DD. 2003. Activity of plant extracts for preserving functional food containing
n-3-PUFA. Eur Food Res Technol 217(4):301-307.
Schmidt EB, Arnesen H, De Caterina R, Rasmussen LH, Kristensen SD. 2004. Marine n-3 polyunsaturated fatty acids
and coronary heart disease: Part 1. Background, epidemiology, animal data, effects on risk factors and safety.
Thromb Res 115(3):163-170.
Scott KC, Latshaw JD. 1991 Effects of commercial processing on the fat-soluble vitamin content of menhaden fish oil.
J Am Oil Chem Soc 68(4):234-236.
Simopoulos AP, Cleland LG. 2003 Omega-6/omega-3 essential fatty acid ratio: The scientific evidence. Basel, Switzerland:
Karger AG. 174 p.
Stone H, Sidel JL. 1985. Sensory evaluation practices. Orlando, Florida: Academic Press. 311 p.
Uauy R, Valenzuela A. 2000. Marine oils: The health benefits of n-3 fatty acids. Nutrition 16:680-684.
Young V. 1990. The usage of fish oils in food. Lipid Technol 2(1):7-10.
binnenwerk.indd 69 21-08-2006 12:34:03

binnenwerk.indd 70 21-08-2006 12:34:04
Oxidative stability of fish oil enriched yoghurts
Charlotte Jacobsen1, Mette B. Let1, Gitte Andersen1 and Anne S. Meyer2
1Department of Seafood Research, Danish Institute for Fisheries Research, B.221, DTU, DK-2800
Kgs. Lyngby, Denmark
2BioCentrum-DTU, B.221, DTU, DK-2800 Kgs. Lyngby, Denmark
Abstract
In this study the oxidative deterioration of fish oil enriched yoghurt with or without strawberry
jam was investigated. Two different emulsifiers (citric acid ester or milk protein) were evaluated.
Moreover, the oxidative stability of yoghurts prepared with pure fish oil or a mixture of fish oil
and rapeseed oil was compared. The sensory off-flavour, concentration of lipid hydroperoxides
and volatile secondary oxidation products, determined by dynamic headspace GC/MS, were
monitored during cold storage. Fish oil enriched yoghurt only oxidised slightly during storage.
Yoghurts prepared with a mixture of fish oil and rapeseed oil were more stable than yoghurts
prepared with pure fish oil. The choice of emulsifier did not seem to affect oxidative stability.
Keywords: omega-3 fatty acids, rapeseed oil, yoghurt, volatile oxidation products, sensory
analysis
Introduction
A still increasing amount of evidence supports the proposed beneficial health effects of the
marine n-3 long chain polyunsaturated fatty acids (PUFA) (Kris-Etherton and others 2002;
Hardman 2002; Trebble and others 2003). The two most potent n-3 PUFA seem to be C20:5 n-3
(EPA) and C22:6 n-3 (DHA). The most well documented effect is the apparent ability of these
n-3 PUFA to protect against cardiovascular death (Kris-Etherton and others 2002). Recently, a
high intake of n-3 PUFA has also been associated with lower risk of developing Alzheimer and
depressions (Mischoulon and Fava 2000). Marine oils are rich sources of EPA and DHA. Therefore,
many efforts have been made to incorporate marine oils into various food products. Several
functional food products based on dairy ingredients such as probiotic yoghurts are already on
the market and these products are perceived as being healthy by the consumers. Therefore,
yoghurts and other dairy products may be good vehicles for incorporation of n-3 PUFA into food
products. However, due to their highly polyunsaturated nature marine oils are very susceptible
to lipid oxidation, which will lead to the formation of unhealthy compounds and undesirable
fishy and rancid flavours.
Previous studies have dealt with the oxidative stability of fish oil enriched milk and different
strategies to reduce lipid oxidation were investigated (Let and others 2003, 2004, 2005a and
b). These studies showed that the otherwise significant lipid oxidation in fish oil enriched milk
could be prevented by using a mixture of fish oil and rapeseed oil instead of using pure fish oil
(Let and others 2004, 2005b).
The rate of lipid oxidation in complex food emulsions depends on numerous factors (Frankel
2005; Jacobsen 2004). Several studies in simple oil-in-water emulsions have suggested that
Quality, safety and processing of wild and farmed fish71
binnenwerk.indd 71 21-08-2006 12:34:04

the choice of emulsifier used to prepare the emulsion can significantly affect lipid oxidation
(Mei and others 1999; Hu and others 2003). Thus, Mei and others (1999) suggested that
positively charged emulsifiers/surfactants could reduce oxidation, because positively charged
metal ions would be repelled from the oil-water interface where they would otherwise catalyze
oxidation. In contrast, negatively charged surfactants were suggested to increase oxidation.
Recently, Djordjevic and others (2004) suggested that fish oil-in-water emulsions prepared
with whey proteins as emulsifiers could be used as n-3 PUFA carriers to be incorporated into
food products.
Lipid oxidation of n-3 PUFA rich products give rise to the formation of highly undesirable fishy
and rancid flavours (Jacobsen 1999; Let and others 2003). This flavour is extremely penetrating
in fish oil enriched milk, because milk in it-self has a mild flavour that cannot mask the
fishy flavour. The volatiles formed in fish oil enriched milk during oxidation has previously
been characterised (Venkateshwarlu and others 2004a). Later, four compounds, namely 1-
penten-3-one, 2,4-t,t-heptadienal, c-4-heptenal and 2,6-t,c-nonadienal, were suggested to
play a significant role for the intensity of fishy and metallic flavours in fish oil enriched milk
(Venkateshwarlu and others 2004b). Yoghurt products are often flavoured with fruits and it
is likely that the fruity taste could mask fishy flavours. Moreover, berries used in fruit jams
(e.g. strawberry, blueberry) have been shown to have antioxidative properties (Klopotek and
others 2005). The effect of antioxidants in lipid systems has been suggested to depend on
their polarity. According to the so-called “polar paradox” polar antioxidants such as trolox
will be more effective in non-polar systems like bulk oils while less polar antioxidants such as
tocopherol will be more effective in oil-in-water emulsions (Huang and others 1996).
The objective of this study was therefore to determine the oxidative stability of strawberry
flavoured yoghurt enriched with fish oil. A second objective was to investigate if yoghurt
based on a mixture of rapeseed oil and fish oil would be less susceptible to oxidation than
yoghurt based on pure fish oil. Oils were added to the yoghurt in the form of an oil-in-water
emulsion. The effect of using two structurally very different emulsifiers to prepare this oil-in-
water emulsion was also investigated. A final objective was to investigate whether the addition
of strawberry jam to fish oil enriched yoghurt affected its oxidative stability compared with
yoghurt without strawberry jam.
Fresh non-flavoured yoghurt with a fat content of 1.5 wt % was purchased locally. Refined cod
liver oil as well as refined rapeseed oil without added antioxidants were provided by Maritex
A/S, Århus, Denmark. A mixture of cod liver oil and rapeseed oil (1:1) was deodorised at
Biocentrum-DTU, Technical University of Denmark, Lyngby, DK, as previously described (Let and
others 2004). Data on oils are shown in Table 1. The fatty acid composition was determined by
preparation of methyl esters that were in turn analyzed by gas chromatography (AOCS Official
method Ce 2-66) The levels of tocopherols were determined by HPLC (AOCS Official method Ce
8-89). Tocopherol standards were purchased from Calbiochem, San Diego, CA. Chemicals and
external standards for identification of volatile oxidation products were all purchased from
Sigma Aldrich, Steinheim, Germany. All solvents were of HPLC grade from Lab Scan, Dublin,
Ireland. Citric acid ester, Citrem LR10 extra (based on glycerolmonooleate) was kindly donated
from Danisco (Brabrand, Denmark) from and milk protein Nutrilac D8080 was from Arla Foods
Ingredients (Aarhus, Denmark).
binnenwerk.indd 72 21-08-2006 12:34:04

Table 1. Data on oils used for experiment 1 and 2.
Fish oil 1a Rapeseed/fish oil Rapeseed oil Fish oil 2a

mixture
PV (meq/kg) 0.10 0.64 1.43 2.0

Free fatty acids (%) 0.05 0.01 0.02 0.01
Alpha-tocopherol (mg/kg) 312 244 159 330
Beta-tocopherol (mg/kg) ND 33 71 ND
Gamma-tocopherol (mg/kg) 3.6 158 344 NA
Delta-tocopherol (mg/kg) 16.4 3.4 7 NA
14:0 3.7 1.9 0.0 3.5
16:0 9.8 6.9 4.2 10.3
18:0 2.0 1.8 1.7 2.3
Sum SFA 15.5 10.6 5.8 16.1
16:1 (n-7) 6.3 3.2 0.2 6.7
18:1 (n-9) 16.7 37.3 57.6 18.9
18:1 (n-7) 3.7 3.3 2.8 4.4
20:1 (n-9) 12.2 6.2 1.4 14.3
22:1 (n-11) 7.1 3.7 ND 8.7
Sum MUFA 46.0 53.6 62.0 53.0
18:2 (n-6) 1.7 10.2 18.6 1.9
18:3 (n-3) 0.9 4.6 8.3 1.1
18:4 (n-3) 2.7 1.3 ND 3.0
20:5 (n-3) 8.6 4.2 ND 9.7
22:5 (n-3) 1.0 0.5 ND 1.2
22:6 (n-3) 11.9 5.7 ND 13.4
Sum PUFA 26.8 26.5 27.0 30.3
a Fishoil 1 was used in experiment 1 and Fish oil 2 was used in experiment 2, ND = not detectable, NA
= Not analysed.
Production of pre-emulsion for yoghurt

The fish oil was added to the yoghurt as a pre-emulsion. This oil-in-water emulsion (71% oil)
was prepared in a Stephan Mixer (Hameln, Germany). Different procedures were used for the
production of the emulsions based on citric acid and milk protein, respectively. In case of the
emulsion with citric acid, 360 g oil was heated and stirred in a beaker and 4.051 g citric acid
was slowly added. The heating was continued until a temperature of 65 °C was reached. 144 g
distilled water was poured into a Stephan mixer pre-heated by hot water. Subsequently, the oil-
emulsifier mixture was gradually (3 min.) added to the water while stirring at a speed of 1200
rpm. Then the stirring speed was increased to 1500 rpm and stirring was continued for another
2 ½ min. In the milk protein emulsion the amount of emulsifier (Nutrilac D-8080) was 5.401 g
and the emulsifier was manually mixed with water in the Stephan mixer bowl. The emulsification
of the water-emulsifier mixture with the oil was as described above for citric acid.
binnenwerk.indd 73 21-08-2006 12:34:04

Production of yoghurt
Yoghurts were prepared both with and without strawberry jam in accordance with the
experimental plan in Table 2a and b. In case of strawberry-flavoured yoghurt, 14.0 g emulsion
was added to 136.0 g strawberry jam in a beaker and mixed by hand. The strawberry-oil
emulsions was subsequently mixed into 850.0 g yoghurt by hand. This yoghurt contained
2.27% fat of which 0.99% was from the oil added and 1.28% was from milk fat. In case of the
non-flavoured yoghurt, 14.0 g emulsion was added directly to 986.0 g yoghurt and mixed as
previously described. This yoghurt contained 2.47% fat of which 0.99% was from the oil added
and 1.48% was from milk fat.
Storage experiment and preparing samples for analysis

Yoghurts from experiment 1 were stored in 1 L blue cap bottles while yoghurts from experiment
2 were stored in 250 mL blue cap bottles. In both experiments yoghurts were stored at 5 °C in
the dark for up to four weeks. Yoghurts were subjected to sensory evaluation (only experiment
1), peroxide value (PV) determination, and dynamic headspace GC-MS analyses. As for the
yoghurts from experiment 1, samples for chemical analysis were transferred to separate 250
mL blue cap bottles, flushed with nitrogen, and stored at –80 °C until analyses, while samples
for sensory analyses were evaluated directly at sampling. Bottles from experiment 2 were
transferred directly to the –80 °C freezer after sampling and kept at this temperature until
chemical analyses were performed.
Analyses of primary oxidation products

Yoghurts were thawed and 15 g sample was taken for lipid extraction by chloroform:methanol
(1:1 w/w) using a reduced amount of solvent ( Iverson and others 2001; Bligh and Dyer 1959)
PV was measured directly on the oils or on the fat extract from the yoghurt by colorimetric
determination of iron-thiocyanate (International IDF Standard 103A:1986).
Dynamic headspace analysis of volatile secondary oxidation products

Volatile secondary oxidation products from 10.0 g of yoghurt were purged and trapped on
Tenax GR® tubes with nitrogen (150 mL/min) for 30 min at 50 °C using 4-methyl-1-pentanol
as internal standard. The volatiles were desorbed (200 °C) from the trap in an automatic
thermal desorber (ATD-400, Perkin Elmer, Norwalk, CN) and cryofocused on a Tenax GR cold
Table 2a. Design of experiment 1.
Pure rapeseed oil Mixture of fish oil and rapeseed oil (1:1) Pure fish oil
Milk protein ROMP RFMP FOMP

Citric acid ester ROCE RFCE FOCE
Table 2b. Design of experiment 2.
With strawberry jam Without strawberry jam
Milk protein StrawMP MP

Citric acid ester StrawCE CE
binnenwerk.indd 74 21-08-2006 12:34:04

trap. Volatiles were separated by gas chromatography (HP 5890 IIA, Hewlett Packard, Palo Alto,
CA) and analyzed by mass spectrometry (HP 5972 mass-selective detector). Oven temperature
programme: 45 °C held for 5 min, 1.5 °C/min to 55 °C, 2.5 °C/min to 90 °C, 12 °C/min to
220 °C and finally held at 220 °C for 4 min. The individual compounds were identified by both
MS-library searches (Wiley138K, John Wiley and Sons, Hewlett Packard, US) and by authentic
external standards. The individual compounds were quantified through calibration curves.
Sensory evaluation
Yoghurts were evaluated by descriptive analysis by 9-13 panelists trained in descriptive analysis
of yoghurt samples with fishy off-flavours. ISO standards 6658, 8586, and 6564 were generally
followed for training and sensory analysis methods, respectively. The descriptors used for odour
and flavour assessment were fishy, rancid, astringent/yoghurt, strawberry/sweet and others,
and these were evaluated on a continuous intensity scale ranging from zero intensity to a
maximum intensity of 9. Samples (25 mL) were served randomized at 5 °C with crisp bread and
cold water in blind trials after 1, 8, 12 and 22 days of storage. Data were collected on PSION
mini computers (PSION, London, UK).
Statistical analysis
Experiment 1
Sensory data were not normaly distributed and could therefore not be analysed by a two-ways
ANOVA analysis. Instead, Kruskal Wallis test was used. In addition, all data were analysed
by ANOVA Partial Least Squares Regression (APLSR) analysis using the software programme
Unsrambler version 7.6 (Oslo, Norway). Design variables (Rapeseed, Rapeseed and fish, Fish oil
for oil types and Emulsifier type) were used as X-data and the measured variables (sensory data,
PV and volatiles) were used as Y-data. Only volatiles that seemed to change during storage were
included in the model (nonanal, pentanal, hexanal, t-2-hexenal, t,t-2,4-heptadienal, penten-
3-one, penten-3-ol). Likewise, only those sensory variables, which were significantly different
among the 6 yoghurts as evaluated by a Kruskal-Wallis test (p < 0.05) were included in the
APLSR model (fishy odour and flavour, rancid odour and flavour, sweet odour and flavour). A
so-called long thin data matrix was used for the model, where the yoghurts at each sampling
point was used as subjects and only one variable for each type of sensory descriptor/PV/volatile
compound was used. All data were weighted by 1/standard deviation and full cross validation
was used to validate the model. By using the jack knifing facility in Unscrambler it was possible
to calculate the regression coefficients for the variables.
Experiment 2
PV and volatiles data were analysed by APLSR analysis as described above. Only volatiles
that developed during storage and that seemed to differ between the different samples
were included in the model (2-ethylfuran, 2-pentanone, 2-t-hexenal, 1-hexanol, 2-heptanol,
hexanal, 2-heptanone, hexanoic acid, 2-nonanone, octanoic acid). Design variables used in this
experiment were yoghurt type and emulsifier type.
Experiment 1
In experiment 1, the oxidative stability of strawberry flavoured yoghurts produced from different
oil types and two different emulsifiers was assessed by sensory evaluation and analysis of
peroxide values (PV) and volatile secondary oxidation products.
binnenwerk.indd 75
the left within the inner ellipse suggested that sweetness was more intense in yoghurts made
from either rapeseed oil or from the rapeseed oil/fish oil mixture.
Figure 2 shows the intensity of fishy flavour of yoghurts during storage. The intensity of fishy
flavour was low in all samples (< 1.5) and generally it did not seem to develop during storage.
However, fishy flavour was higher in yoghurts with pure fish oil. The Kruskal-Wallis test showed
that on day 1 and 22, FOMP and FOCE were both significantly more fishy than the four yogurts
with rapeseed oil or rapeseed and fish oil mixture and also on day 12, FOCE was significantly
more fishy than these four yoghurts. The data for fishy odour were generally in agreement with
the fishy flavour data, although the intensity of the fishy odour was even lower (< 1.0). The
rancid odour and flavour in all samples were low throughout storage (< 0.5) (data not shown).
Sweetness did not change significantly over time and its intensity was between 2.5 and 3.5
for the odour and between 2.8 and 3.8 for the flavour (data not shown). Moreover, these data
confirmed that FOMP and FOCE were less sweet than the other yoghurts. The conclusions above
were further confirmed by the regression coefficients obtained from the APLSR model (Table 3).
These regression coefficients thus showed that across all sampling times, fishy odour and
flavour were significantly positive in yoghurts with fish oil while sweet odour and flavour were
significantly negative in these samples. Interestingly, all the data pointed to the conclusion
that the sensory panel could not discriminate between yoghurts made from pure rapeseed oil
or a mixture of rapeseed oil and fish oil. This observation is in agreement with findings from
previous studies on milk enriched with fish oil (1% milk fat and 0.5% oil added), where the
sensory panel also found that addition of pure fish oil gave rise to significantly more fishy
products whereas milk with pure rapeseed oil or a mixture of fish oil/rapeseed oil were not
significantly different (Let and others 2003, 2004). The fact that the emulsifier type did not
5.0
4.5
4.0
3.5
3.0
2.5
2.0
Intensity
1.5
1.0
0.5
Day 1
0.0
Day 22
RFMP FOCE FOMP
ROMP RFCE
ROCE
Figure 2. Development in fishy flavour in yoghurts from experiment 1 during storage at 5 °C. For interpretation
of code names refer to Table 2a.
binnenwerk.indd 77 21-08-2006 12:34:05

Table 3. Regression coefficients from APLSR model on data from experiment 1.
Rapeseed oil Rapeseed/Fish oil Fish oil Emulsifier type

O-fishy - ns + ns
O-Rancid ns ns ns ns
O-Sweet + ns - ns
F –fishy - ns + ns
F-Rancid ns ns ns ns
F-Sweet ns ns - ns
PV - ns + ns
Nonanal ns ns ns ns
Pentanal ns ns ns ns
Hexanal ns ns ns ns
t-2-hexenal ns ns + ns
t,t,-2,4-heptadienal - ns + ns
Penten-3-one - ns + ns
Penten-3-ol - ns + ns
+ denotes significantly positive regression coefficient, - denotes significantly negative regression

coefficient, ns denotes that the regression coefficient was insignificant
have any significant positive regression coefficients confirmed that the emulsifier type did not
significantly influence the sensory properties of the yoghurts.
Peroxide value
PV was located to the right in the APLSR plot in the 4th quadrant, which indicated that PV
was highest in yoghurts with fish oil (Figure 1). PV’s negative PC2 value also suggested that
PV was higher in the RF yoghurts compared with yoghurts made from pure rapeseed oil. These
conclusions were confirmed by the raw data, which clearly showed that PV were highest in
yoghurts with fish oil, followed by yoghurts with the rapeseed oil/fish oil mixture, while
yoghurts with pure rapeseed oil had the lowest PV (Figure 3). The regression coefficients also
supported these conclusions (Table 3). PV in yoghurts with pure rapeseed oil remained stable
around 1 meq/kg throughout storage, while fish oil yoghurts increased from 3-4 meq/kg to
approx 6.5 meq/kg. Yoghurts with the rapeseed oil/fish oil mixture had intermediate PVs that
increased from 2 to 4-5 meq/kg during storage. Thus, the PV data did not to the same extent
indicate a protective effect of the rapeseed oil against oxidation as was found from the sensory
data. Rather, PV data suggested that the effect of rapeseed oil was mainly a dilution effect.
Again no clear effect of the emulsifier type could be observed.
Volatiles
In the APLSR plot, penten-3-one, penten-3-ol, t,t-2,4-heptadienal and t-2-hexenal were found
to the right (Figure 1). While the former three volatiles had slightly positive PC2 values, t-
2-hexenal had a negative PC2 value. Hence, all four compounds appeared to be formed to a
greater extent in yoghurts with fish oil compared to the other samples. Pentanal was also
located to the right in the model, but much closer to the origin and hexanal and nonanal
were located slightly to the left in the model. Hence, the concentrations of these compounds
did not seem to be elevated in the FO yoghurts and especially hexanal and nonanal were not
well explained by the model. These interpretations of the model were further supported by the
binnenwerk.indd 78 21-08-2006 12:34:06

8.0
7.0
6.0
5.0
meq/ kg oil
4.0
3.0
2.0
1.0
0.0
0 5 10 15 20 25
Storage time (days)
ROMP RFMP FOMP

ROCE RFCE FOCE
Figure 3. PV during storage at 5 °C in experiment 1. For interpretation of code names refer to Table 2a.
regression coefficients, which showed that t-2-hexenal, t,t,-2,4-heptadienal and penten-3-

one had significantly positive regression coefficients for the fish oil design variable (Table 3).
Heptadienal also had a significant negative regression coefficient for rapeseed oil as was also
the case for penten-3-ol. Pentanal on the other hand had a significant negative regression
coefficient for the rapeseed/fish oil design variable showing that RF yoghurts had significantly
lower concentrations of pentanal than the other yoghurts. The regression coefficients for
nonanal and hexanal were not significant, which supported the interpretation above that
there was no clear pattern in the development of these volatiles. The raw data showed that
concentrations of all the above mentioned compounds generally were very low, but that they did
increase during storage. In fact, concentrations were so low that it was impossible to quantify
the compounds by the use of calibration curves. Instead, ion peak area/g for each compound
was calculated. The development of penten-3-ol is shown in Figure 4 to illustrate the difference
between the FO yoghurts and the other yoghurts. Similar differences were seen for penten-3-
one. Interestingly, t,t-2,4-heptadienal could only be found in the FO yoghurts. The results for
penten-3-ol, penten-3-one and heptadienal indicated that volatiles were formed to a higher
extent in fish oil yoghurts with citric acid compared to fish oil yoghurts with milk protein.
However, this “prooxidative” effect of citric acid was not a general trend in yoghurts with the
rapeseed/fish oil mixture or in yoghurts with pure rapeseed oil and could also not be seen for
the other compounds (nonanal, hexanal, pentanal and t-2-hexenal). Moreover, the increased
formation of penten-3-ol, penten-3-one and heptadienal may not be due to a pro-oxidative
effect of the citric acid ester as such, but could be due to the fact that the fish oil was heated
before the production of the pre-emulsion.
binnenwerk.indd 79 21-08-2006 12:34:06

25000
20000
15000
Area/g
10000
5000
0
1 8 12 22
Storage time (days)
ROMP ROCE RFMP

RFCE FOMP FOCE
Figure 4. Development in 1-penten-3-ol in yoghurts from experiment 1 during storage at 5 °C. For
interpretation of code names refer to Table 2a.
Other volatile compounds than the abovementioned were also identified in the yoghurts. Thus,
t-2-pentenal, t,c-2,4-hexadienal, t,c-2,6-nonadienal, 1-pentanol, 2-penten-1-ol were also
found, but their concentrations did not increase during storage (data not shown). Furthermore,
2-heptanol, 2-heptanone, 2-nonanone, 1-hexanol and 2-pentanone were identified. These
compounds did not show a clear pattern with respect to the effect of oil type and emulsifier.
Some yoghurts, particularly ROMP and FOCE had elevated concentrations of these compounds
at day 22. Maybe, this was an indication of microbial growth in these yoghurts.
Experiment 2
In experiment 2, the oxidative stability of fish oil yoghurts with or without strawberry jam
produced with two different emulsifiers was assessed by PV and volatiles and data were analysed
by ANOVA PLSR analysis.
ANOVA Partial Least Squares Regression analysis

The APLSR model was only validated by 1 PC, which explained 50% of the variation in the X-
data and 33% of the variation in the Y-data.
Scores plot and design variables in loadings plot

In the scores plot, strawberry yoghurts were placed to the right in the plot and non-flavoured
yoghurts to the left (plot not shown). Moreover, yoghurts with citric acid as emulsifier were
located in the bottom of the plot and yoghurts with milk protein in the top of the plot. In
accordance herewith, the yoghurt type design variable was located to the far left in the loadings
plot and the emulsifier type design variable was located in the top of the loadings plot (Figure
5). When a design variable only has two levels (e.g. strawberry and non-flavoured) only one
variable will be shown in the loadings plot. In Figure 5, the location of the yoghurt type design
variable thus indicated that yoghurts without strawberry were located to the left. Likewise,
binnenwerk.indd 80 21-08-2006 12:34:06

1.0 PC2
Emulsifier
0.5
2-hexenal
PV
2-ethyl-furan
Hexanoic acid
Octanoic acid Hexanal
2-pentanone
0 2-heptanon
2-nonanone
Yoghurt type 1-hexanol
-0.5
-1.0 PC1
-1.0 -0.5 0 0.5 1.0
Figure 5. APLSR correlation loadings plot from experiment 2. Design variables were used as X-data and
measured variables as Y-data. Yoghurt type is non-flavoured yoghurt versus strawberry flavoured yoghurt.
Emulsifier type is the emulsifier.
the location of the emulsifier design variable indicated that yoghurts with milk protein were
located in the top of the plot. Hence, PC1 mainly explained differences between yoghurts
with and without strawberry jam and PC2 explained differences caused by the two different
emulsifiers. Since PC2 was not validated cautious interpretation of the data in the PC2 direction
is, however, necessary.
Peroxide value
In the APLSR plot, the PV variable was located in the 1st quadrant close to the origin indicating
that this variable was not well explained by the model (Figure 5). Moreover, PV did not have
significant regression coefficients (Table 4). The raw data showed, however, that PV developed
differently in yoghurts with and without strawberry jam (Figure 6). PVs in yoghurts with
strawberry jam were thus higher than in non-flavoured yoghurts at day 1 (between 7 and 12
meq/kg versus 3 meq/kg, respectively). During storage, PVs in strawberry yoghurts remained
fairly stable, while PVs in non-flavoured yoghurts increased significantly to maximum values
that were higher than the maximum PVs in the corresponding strawberry yoghurts, which were
20.3 meq/kg in MP after 22 days and 14.2 meq/kg in CE after 29 days.
Volatiles
Generally, volatiles were located to the right in the loadings plot, except for 2-ethylfuran,
hexanoic acid and octanoic acid (Figure 5). Hence, most volatiles correlated negatively to
the yoghurt type design variable, indicating that across all sampling times concentrations of
volatiles were lowest in yoghurts without strawberry jam. The concentrations of 2-ethylfuran,
hexanoic acid and octanoic acid appeared, however, to be highest in these yoghurts. The
regression coefficients generally confirmed these conclusions. Thus, 2-pentanone, 1-hexenol,
2-heptanol, hexanal, 2-heptanone had significantly negative regression coefficients for yoghurt
type (i.e. yoghurt without strawberry flavour), while hexanoic acid and octanoic acid had positive
regression coefficients for this design variable (Table 4). No volatiles had significant regression
coefficients for the emulsifier type, indicating that as in experiment 1 the emulsifier type did not
binnenwerk.indd 81 21-08-2006 12:34:07

Table 4. Regression coefficients from APLSR model on data from experiment 2.
Yoghurt type Emulsifier type
PV ns ns
2-ethylfuran ns ns
2-pentanone - ns
2-t-hexenal ns ns
1-hexenol - ns
2-heptanol - ns
Hexanal - ns
2-heptanone - ns
Hexanoic acid + ns
2-nonanone ns ns
Octanoic acid + ns
ns = not significant
25.00
20.00
PV (meq/kg)
StrawCE
15.00
StrawMP
CE
10.00
MP
5.00
0.00
0 5 10 15 20 25 30
Storage days
Figure 6. Development of PV during storage at 5 °C in experiment 2. For interpretation of code names refer
to Table 2b.
have a clear effect on the oxidative stability of the yoghurts. Closer inspection of the raw data
revealed a more complex relationship between yoghurt type and the development of volatiles
than could be interpreted from the APLSR modeling. A general trend was that the initial level
of volatiles was highest in the strawberry yoghurts, which could also explain the negative
correlation between yoghurt type and several of the volatiles above. However, ethylfuran, 2-
t-hexenal, hexanoic acid and octanoic acid only developed during storage in yoghurts without
strawberry flavour as exemplified in Figure 7 for t-2-hexenal. For 2-pentanone, 1-hexanol, 2-
heptanol, 2-heptanone and 2-nonanone concentrations were fairly stable throughout storage for
all yoghurts, except at day 29 where concentrations increased either in both strawberry yoghurts
or in strawberry yoghurt with CE only. This increase in the volatiles may be due to microbial
binnenwerk.indd 82 21-08-2006 12:34:07

30,000
25,000
20,000
Area/g
15,000
10,000
5,000
0
0 5 10 15 20 25 30
Storage time (days)
StrawCE StrawMP CE MP
Figure 7. Development in t-2-hexenal during storage at 5 °C in experiment 2. For interpretation of code

names refer to Table 2b.
growth and not to oxidation. Hence, a clear development in volatiles during storage was only
found for 2-ethylfuran, 2-t-hexenal, hexanoic acid and octanoic acid and this development was
more pronounced in yoghurts without strawberry jam.
Only two of the volatiles quantified in experiment 2 were the same as in experiment 1. However,
almost all the compounds found in experiment 1 were also identified in experiment 2, but
several of them were not selected for further data treatment either because they did not
develop during storage, because their development did not appear to be dependent on the
experimental design or because the peak area was very low. The latter two explanations was
the reason why heptadienal, penten-3-one and penten-3-ol were not selected for further data
treatment in experiment 2. In experiment 1, these three compounds were found in significantly
higher concentrations in the fish oil yoghurts compared to the other yoghurts despite the fact
that peak areas were also very low in this experiment. The combination of the low peak areas
for these three compounds and the fact that all yoghurts in experiment 2 contained fish oil
probably explained why no clear differences between samples was obvious in this study.
In summary, experiment 1 and 2 both showed that the oxidative stability of yoghurts was
relatively high irrespective of the oil type used to prepare them. Thus, both fishy and rancid
flavours, PV and the concentrations of volatiles remained low throughout the storage period.
The results of experiment 2 were generally in agreement with this conclusion. Importantly, a
general comment from the sensory panel, which previously had also evaluated fish oil enriched
milk, was that it was much more difficult to detect fishy flavours in strawberry yoghurts with
pure fish oil than in milk enriched with pure fish oil. This conclusion was also verified by the
fact that the maximum score obtained for fishy flavour in the yoghurt experiment was 1.35 (for
FOCE at day 1) compared to e.g. 2.5 in fish oil enriched milk (Let and others 2005). Moreover,
among the volatiles (1-penten-3-one, 4-c-heptenal, 2,4-t,t-heptadienal and 2,6-t,c-nonadienal)
suggested to be important for fishy off-flavour in fish oil enriched milk (Venkateshwarlu and
others 2004a and b) only 1-penten-3-one and heptadienal increased during storage in the
present study. Especially, 2,6-nonadienal seems to contribute significantly to the fishy flavour
in neat fish oil (Macfarlane and others 2001) and in fish oil enriched milk (Venkateshwarlu and
binnenwerk.indd 83 21-08-2006 12:34:08

others 2004b). The fact that this compound did not develop during storage in fish oil enriched
yoghurt may suggest that fish oil enriched yoghurt oxidized slower than fish oil enriched milk.
The higher intensity off fishy flavour and lower oxidative stability in the fish oil enriched milk
was not due to a higher total fat or fish oil concentration in the milk. In fact, the opposite was
the case as the milk contained 1.5% fat of which 0.5% was fish oil and the yoghurt contained
2.27% fat of which 0.99% was fish oil. Taken together these results may indicate that n-3 PUFA
are less susceptible to oxidation in strawberry yoghurt compared with milk. However, more
data are necessary to confirm this conclusion. Moreover, the strawberry and yoghurt flavour
may also have masked the fishy flavour. The possible better oxidative stability of yoghurts
compared with milk may be due to the fact that the processing of yoghurts was milder than for
milks. For example, yoghurts were not heated or homogenized after addition of fish oil as was
the case for milk. Lipid oxidation is pH dependent, especially in the presence of proteins. This
is due to the fact that pH will affect the charge of the proteins and this could either lead to
increased repulsion or attraction between proteins and metal ions depending on the pH in the
emulsion and the pI of the proteins. The increased oxidative stability in yoghurt compared with
milk may therefore partly be due to the lower pH in yoghurt (pH 4.2), which perhaps lead to
decreased attraction of metal ions at the fat membrane. The difference in physical structure and
rheological properties between fish oil enriched milk and yoghurt may also partly explain their
different oxidative stabilities. Further studies are required to obtain a better understanding of
these issues.
Experiment 1 also showed that addition of rapeseed oil to fish oil increased the oxidative
stability of yoghurts. This was most obvious from the sensory and volatiles data. This conclusion
is in accordance with data from our previous studies with fish oil enriched milk (Let and others
2004, 2005). In the later study it was suggested that the anti-oxidative effect of rapeseed oil
is due to its high content of gamma-tocopherol.
In both experiment 1 and 2, the emulsifier type did not significantly affect the oxidative
stability of yoghurts. Further experimental work is required to elucidate whether emulsifier
addition can be used to control partitioning of antioxidants and/or pro-oxidants between the
different phases in composite dairy products and if such control in turn provides extra oxidative
protection of fish oil enriched yoghurts.
In experiment 2, yoghurts with strawberry jam had higher PV and concentrations of several
volatiles at day 1 compared to yoghurts without strawberry jam. Since it is not likely that
strawberry jam contained lipid hydroperoxides, the high PV at day 1 suggests that the strawberry
jam contained other compounds that reacted with the reagent used for the determination of PV
(Fe2+). Alternatively, the jam contained compounds that absorbed light at the same wavelength
as the Fe(SCN)3 complex formed in the process for the analysis of lipid hydroperoxides (500
nm). The high level of volatiles at day 1 suggests that strawberry jam had a natural content of
several of the volatiles that was used as markers of oxidation. Only PV and 2-ethylfuran, 2-t-
hexenal, hexanoic acid and octanoic acid developed significantly during storage and this only
happened in yoghurts without strawberry jam suggesting that these yoghurts oxidized faster
than yoghurts with strawberry jam. Hence, strawberry jam may have an anti-oxidative effect,
which could be ascribed to the strawberries. Further investigations are necessary to confirm
this hypothesis.
binnenwerk.indd 84 21-08-2006 12:34:08

The high levels of some ketones and alcohols at the end of the storage experiment in some
yoghurts could indicate microbial growth. In experiment 1, some yoghurts were clearly spoiled
by microbial growth after 29 days and were therefore not analysed. Further precautions are
therefore necessary to prevent microbial growth in future experiments. However, it should be
mentioned that in Denmark shelf life of commercial yoghurt products is only 21 days, so a shelf
life longer than 21 days will not be required for fish oil enriched yoghurts.
Conclusion
In conclusion, yoghurt seems to be a good vehicle for the incorporation of n-3 PUFA. Thus,
it was possible to produce fish oil enriched yoghurts with satisfactory oxidative stability and
sensory properties. In complete accordance with our previous findings with milk, lipid oxidation
of fish oil can be significantly reduced when the fish oil is blended with rapeseed oil before
use. This effect is most likely a result of the γ-tocopherol present in rapeseed oil rather than
due to a special effect of the rapeseed lipids themselves.
Acknowledgements
Lis Berner is thanked for skilful lab work and Else Green and the sensory panel at BioCentrum is
acknowledged for their skilled sensory work. This work was financially supported by the Danish
FØTEK III program, Maritex A/S Denmark and Arla Foods a.m.b.a.
References
AOCS Official method Ce 2-66. Preparation of methyl esters of long-chain fatty acids. 1992.
AOCS Champaign. IL; AOCS Official method Ce 1b-89. Fatty acid Composition by GLC. Marine Oils. 1992. AOCS Champaign. IL.
AOCS Official method Ce 8-89. Determination of Tocopherols and Tocotrienols in Vegetable Oils and Fats by HPLC. 1992.
AOCS Champaign. IL.
Bligh EG, Dyer WJ. 1959. A rapid method of total lipid extraction and purification. Can J Biochem Physiol 37:900-917.
J Food Sci 69(5):C356-C362.
Frankel EN. 2005. Lipid oxidation. Dundee, Scotland. The Oily Press Ltd.
Hardman WE. 2002. Omega-3 fatty acids to augment cancer therapy. J Nutr 132(11):3508-3512.
Hu M, McClements DJ, Decker EA. 2003. Impact of whey protein emulsifiers on the oxidative stability of salmon oil-in-
water emulsions. J Agric Food Chem 51(5):1435-1439
Huang S-W, Hopia A, Schwarz K, German JB, Frankel EN. 1996. Antioxidant activity of γ-tocopherol and Trolox in
different lipid substrates: bulk oils vs oil-in-water emulsions. J Agric Food Chem 44:444-452.
International IDF Standard 103A:1986. Milk and Milk Products; Determination of the iron content. 1986. International
Dairy Federation, Brussels, Belgium.
Iverson SJ, Lang SLC, Cooper MH. 2001. Comparison of the Bligh and Dyer and Folch methods for total lipid determination
in a broad range of marine tissue. Lipids 36:1283-1287.
Jacobsen, C. 1999. Sensory impact of lipid oxidation in complex food systems. Fett/Lipid 12:484-492.
Jacobsen, C. 2004. Developing polyunsaturated fatty acids as functional ingredients. In: Arnoldi, A. (ed) Functional
foods, diet, cardiovascular disease and diabetes. Cambridge, England, Woodhead Publishing Ltd.
Klopotek Y, Otto K, Bohm V. 2005. Processing strawberries to different products alters contents of vitamin C, total
phenolics, total anthocyanins, and antioxidant capacity. J Agric Food Chem 53(14):5640-5646.
Kris-Etherton PM, Harris WS, Appel LJ. 2002. Fish consumption, fish oil, omega-3 fatty acids, and cardiovascular
diseases. American Heart Association, Circulation 106:2747-2757.
binnenwerk.indd 85 21-08-2006 12:34:08

Let MB, Jacobsen C, Frankel EN, Meyer AS. 2003. Oxidative flavour deterioration of fish oil enriched milk: Effects of oil
type and Ethylenediaminetetraacetate (EDTA). Eur J Lipid Sci Technol 105:518-528.
Let MB, Jacobsen C, Meyer AS. 2004. Effects of fish oil type, lipid antioxidants and presence of rapeseed oil on oxidative
flavour deterioration of fish oil enriched milk. Eur J Lipid Sci Technol 106:170-182.
Let MB, Jacobsen C, Meyer AS. 2005a. Sensory stability and oxidation of fish oil enriched milk is affected by milk
storage temperatures and oil quality. Internat Dairy J 15:173-182.
Let MB, Jacobsen C, Pham KA, Meyer AS. 2005b. Protection against oxidation of fish oil enriched milk emulsions through
addition of rapeseed oil or antioxidants. J Agric Food Chem 53:5429-5437.
Macfarlane N, Salt J, Birkin R. 2001. The FAST index – A fishy scale. Int News Fats Oils Relat Mater 12:244-249.
Mei LY, McClements DJ, Decker EA.1999. Lipid oxidation in emulsions as affected by charge status of antioxidants and
emulsion droplets. J Agric Food Chem 47(6):2267-2273 Mischoulon D, Fava M. 2000. Docosahexanoic acid and
omega-3 fatty acids in depression. Psychiatric Clinics North America 23(4):785-794
Trebble TM, Wootton SA, Miles EA, Mullee M, Arden NK, Ballinger AB, Stroud MA, Burdge GC, Calder PC. 2003.
Prostaglandin E2 production and T cell function after fish-oil supplementation. Am J Clin Nutr 78:376-382
Venkateshwarlu G, Let MB, Meyer AS, Jacobsen C. 2004a. Chemical and olfactometric characterization of volatile flavor
compounds in a fish oil enriched milk emulsion. J Agric Food Chem 52:311-317.
Venkateshwarlu G, Let MB, Meyer, AS, Jacobsen C. 2004b. Modelling the sensory impact of defined combinations of
volatile lipid oxidation products on fishy and metallic off-flavors. J Agric Food Chem 52:1635-1641.
binnenwerk.indd 86 21-08-2006 12:34:08

Antioxidant synergy effect between α-tocopherol and ascorbate on
the autoxidation of liposomes
Harald Barstad1, Anne Cecilie Alvik2 and Erik Løvaas2

1Fiskeriforskning, PO Box 6122, N-9291 Tromsø, Norway
2Department of Pharmacy, University of Tromsø, NO-9037 Tromsø, Norway
Abstract
The antioxidant effects of α-tocopherol and ascorbate were evaluated in a model system based
on liposomes made from highly unsaturated phospholipids of marine origin. The peroxidation
level of the membrane fatty acids was monitored in a microplate spectrophotometer. Tocopherol
was shown to be an effective antioxidant when incorporated in the liposome membrane.
Ascorbate was shown to have antioxidant effect when distributed in the water phase. The
delay in onset of fatty acid peroxidation was greater when tocopherol and ascorbate was used
together than the added effect of tocopherol and ascorbate used separately. This indicates that
a synergy between α-tocopherol and ascorbate was observed.
Keywords: tocopherol, ascorbate, liposomes, marine phospholipids, synergy
Introduction
Liposomes are spherical particles formed from lamellar lipid-water systems. They consist of closed
bilayers of membrane lipids, with water inside. Liposomes mimic biological membranes and are
useful as a model system when studying reactions taking place in cell membranes. Liposomes
made from marine phospholipids (MPL) have a high amount of long chain n-3 polyunsaturated
fatty acids. They are therefore prone to oxidation. Like natural occurring cell membranes, the
liposome membrane could be protected by incorporated antioxidants. Liposomes from MPL have
been proposed as oral route vectors for n-3 fatty acids and tocopherol by Nacka and others
(2001). More on the use of MPL can be found in the manuscript by Løvaas (2006).
Oxidation of liposomes can be studied by monitoring of specific reaction products like aldehydes
or by measuring oxygen consumption. A convenient way of monitoring the early stages of
oxidation of marine lipids is to measure the formation of conjugated dienes from polyunsaturated
fatty acids.
Conjugated dienes are formed as a first product in the peroxidation of polyunsaturated fatty
acids. Their characteristic absorption band at about 232 nm is the basis for the IUPAC method
for determination of purity and deterioration of oils and fats (IUPAC 1987). The formation of
dienes can be monitored in real time by a spectrophotometric method. By applying a microplate
spectrophotometer multiple experiments can be performed simultaneously.
Tocopherols are effective inhibitors of lipid oxidation. They are known as the most important
lipophilic antioxidants in living cells (Packer and Obermüller-Jevic 2002). Incorporated in
binnenwerk.indd 87 21-08-2006 12:34:08

biological membranes they protect their functionality by inhibiting the oxidation of unsaturated
fatty acids in membrane lipids.
Ascorbate is a well known water soluble antioxidant both in vivo and in vitro. It can act both
as a metal chelator, as an oxygen scavenger and as a reducing agent (Frankel 2005).
Tappel (1968) suggested that the two vitamins act synergistic, tocopherol acting as the primary
antioxidant and the resulting tocopheroxyl radical reacting with ascorbate to regenerate tocopherol.
This was confirmed experimentally in organic solvent by Packer and others (1979). Scarpa and
others(1984) showed that ascorbate was able to regenerate the tocopheroxyl radical incorporated
in a soy lecithin liposome bilayer when ascorbate was in the surrounding aqueous phase.
The objective of this study was to use MPL liposomes as a model to investigate the combined
antioxidant effect of membrane bound α-tocopherol and ascorbate. If synergy could be
detected for tocopherol and ascorbate, the assay would be useful for screening other potential
antioxidants for synergy effects.
Standards and chemicals

DL-α-tocopherol, ascorbate, ammonium iron(II)sulfate and all solvents was of PA grade.
Marine phospholipids
MPL used for liposome preparation were extracted by the method of Bligh and Dyer (1959)
from cod roe. The phospholipids were purified on silica gel. The extract consisted of more than
90% phosphatidylcholine (PC) quantified by the method described by Bartlett (1959). The
long-chain polyunsaturated fatty acids docosahexaenoic acid and eicosapentaenoic acid were
present in concentrations of 12.8 and 10.9 g/100g PC extract, respectively. The MPL was stored
as a stock solution of 10 mg/ml in chloroform under nitrogen at -18 ºC.
Preparation of liposomes
Portions of 120 µl 10 mg/ml solution of MPL in chloroform were added to conical glass test
tubes, making a total of 1.5 µmol MPL using a molecular weight of 800. The chloroform was
evaporated by nitrogen flushing at 40 ºC. Then dry 6 ml 5 mM MES buffer solution at pH 5.5 was
added. An ultrasonic cell diruptor was inserted into the test tubes and the sample sonicated at
1 W for 60 s (Microson ultrasonic cell disruptor from Misonix Inc. Farmingdale, NY). The test
tubes were topped with nitrogen, sealed and placed in the refrigerator. The liposome solutions
were used the day they were made.
Incorporation of tocopherol to liposome membranes

0 – 1 mole % of α-tocopherol relative to the amount of MPL, was added to the MPL /chloroform
solution before evaporation and liposome preparation as described above.
Addition of ascorbate to the inter-liposome water phase

The ascorbate dissolved in 5.5 mM MES buffer was added to 5 ml liposome batches with or
without tocopherol incorporated. The volumes were adjusted to 6 ml with buffer. Portions of 280
µl liposome solution was applied to the microwells of a 96 wells UV-transparent microplate. The
microplate was then placed in the drawer of a SpectraMAX 190 Microplate Spectrophotometer.
binnenwerk.indd 88 21-08-2006 12:34:08

Antioxidant synergy effect between α-tocopherol and ascorbate on the autoxidation of liposomes
Initiation of oxidation
The peroxidation was initiated by adding 20 µl of a 300 µM ammonium iron(II)sulfate solution to
the microwells immediately before spectrophotometric monitoring. The solution was prepared in
1 M HCl. The addition of the acidic solution did not affect the pH in the microwells by more than
0.02 units. Not all 96 microwells were filled. Usually only 4 rows of 8 microwells were used. The
wells were filled with initiator one row at a time with a 8 channel pipette. When four rows were
filled the time shift between the first and the last addition was less than 20 seconds. The time
from addition of initiator to the first spectroscopic measurement was less than 15 seconds.
Peroxidation measurement
The incubation and measurements took place in the dark at a temperature between 25 and
26 ºC. The absorbance at 232 nm was measured every 15 seconds. Prior to every reading the
microplate was shaken.
Results
Results are presented as multiple line graphs. The lines in the same graph derive from
experiments carried out simultaneous in the wells of the same microplate. This ensures that
no differences of freshness or dilutions of the liposome solutions affect the results. Included
in all experimental runs are a number of controls with and without initiator and antioxidant
present. Each experiment is performed in duplicate and each curve represents the average of the
readings from two individual microwells. The first measurements are performed 15-30 seconds
after the admixture of initiator. This implies that the actual induction time is somewhat longer
than what is indicated on the time axis. The absorbance values for the first measurements after
initiation is set to zero units.
Peroxidation of marine liposomes with no added antioxidants (Figure 1)

The polyunsaturated fatty acids of the liposomes are highly unstable once oxidation is initiated.
In Figure 1 the peroxidation can be followed as an immediately rise in absorbance after the
addition of Fe2+. The vertical axis shows the increase in absorbance after the first measurement
set as 0. The time for the peroxidation curve to reach halfway up to its plateau is 1 min 15
seconds. The two parallels are in agreement within 5 seconds. The line for the absorbance
of liposomes with no admixture of initiator stays flat, showing no peroxidation during the
experimental timescale.
Antioxidant effects of α-tocopherol (Figure 2)

The lipid soluble α-tocopherol was incorporated into the liposome membranes in concentrations
of 0.1 to 0.7 mole %. The effect is visualised both as a delayed offset time and as a lower
peroxidation level. As the concentration of α-tocopherol increases, the slope decreases and the
peroxidation curve appears sigmoid. The concentration of 0.1 mole % α-tocopherol delays the
onset of peroxidation by 15 seconds while the concentration of 0.7 mole % delays the onset
by 9 min.
Antioxidant effects of ascorbate (Figure 3)

The absorption curves of liposomes protected by ascorbate added to the water phase is illustrated
in Figure 3. The vertical axis this time present absolute values of absorbance. The curves for
5 mole % ascorbate shows a delay in the onset of peroxidation of 7 min compared to the no-
antioxidant curve. 50 mole % shows a delay of 28 min, and 500 mole % ascorbate 86 min.
binnenwerk.indd 89 21-08-2006 12:34:09

0.5
Peroxidation level (absorbance units)

No antioxidant
0.4
0.3
0.2
0.1
No initiator
0.0
0 5 10 15 20 25
Reaction time (minutes)
Figure 1. Initiation of peroxidation of marine liposomes by Fe2+, no antioxidant present. The peroxidation
level is expressed as increase in absorbance at 232 nm.
0.5
mole% TOH
Peroxidation level (absorbance units)
0.0
0.1
0.4 0.3
0.5
0.3 0.7
0.2
0.1
0.0
0 5 10 15 20 25
Figure 2. Protective effects of α-tocopherol on the peroxidation of marine liposomes. The peroxidation level
is expressed as increase in absorbance at 232 nm. Tocopherol concentrations are mole % relative to the
amount of MPL.
binnenwerk.indd 90 21-08-2006 12:34:09

(absorbance units) 1.4

Peroxidation level
1.2
1.0
5% Ascorbate
0.8 50% Ascorbate
500% Ascorbate
0.6 No antioxidant
0 20 40 60 80 100
Figure 3. Protective effects of ascorbate on the peroxidation of marine liposomes. The peroxidation is
expressed as an increase in the overall absorbance at 232 nm. Ascorbate concentrations are mole % relative
to the amount of MPL.
Figure 3 also illustrates some of the limitations of the method. Absorption at 232 nm is not
specific for peroxidised fatty acids. Ascorbate itself has absorption in the UV range. The starting
point for the absorption curve of ascorbate at the highest concentration is therefore off scale.
The curve falls as ascorbate is consumed, and seems to level out before the characteristic
steep rise associated to the peroxidation chain reaction. However the actual concentrations of
dienes, ascorbate and the various degradation products cannot be calculated from the curve.
The prolonged time of the experiments opens for a number of side reactions having a lower
reaction rate.
The combined effect of α-tocopherol and ascorbate (Figure 4)

The combined effect of 0.4 mole % α-tocopherol and 5 or 50 mole % ascorbate is shown in
Figure 4. When using 0.4 mole % α-tocopherol and 5 mole % ascorbate, the induction period
(25 min) is longer than the simple sum of the induction periods of tocopherol (5 min) and
ascorbate (7 min). Likewise, when using 0.4 mole % α-tocopherol and 50 mole % ascorbate, the
induction period of 79 min is longer than the simple sum of the induction periods of tocopherol
(7 min) and ascorbate (31 min). The duplicate parallels differ by less than 1 min for the longer
induction periods. This means that synergy is observed.
Discussion
Antioxidant effects of α-tocopherol

The most important antioxidant action of tocopherol in the assay is as a scavenger of free
radicals. Tocopherol inhibits the propagation state of autoxidation by donating a phenolic
hydrogen atom to lipid peroxy radicals. The reaction of lipid peroxy radicals with tocopherol is
several orders of magnitude faster than the reaction with acyl lipids (Kamal-Eldin and Appelqvist
1996). Tocopherol is thus effective in small concentrations. The assay shows antioxidant
effect of 0.1 mole % α-tocopherol, or two tocopherol molecules per thousand molecules of
polyunsaturated fatty acids.
binnenwerk.indd 91 21-08-2006 12:34:09

0.6
Peroxidation level (absorbance units) 5 mole% AA

0.5
No antioxidant
0.5 mole% TOH
0.4
0.5 mole% TOH
+ 5 mol% AA
0.3
0.2
0.1
0.0
0 10 20 30 40 50 60
Figure 4. Synergetic protective effect of ascorbate and α-tocopherol on the peroxidation of marine liposomes.
The peroxidation level is expressed as increase in absorbance at 232 nm. Concentrations are mole % relative
to the amount of MPL.
As the lipid free radical is stabilised by the hydrogen atom from tocopherol, the latter is
transformed into a tocopheroxyl radical. The tocopheroxyl radical adds readily to other radicals,
forming stable products. This means that each molecule of tocopherol can act as an antioxidant
two times.
Antioxidant effects of ascorbate

Ascorbate is a well known water soluble antioxidant both in vivo and in vitro. It can act both as
an oxygen scavenger, as a metal chelator, and as a reducing agent. In the presence of metals (as
Fe3+) it can also act like a pro-oxidant (Frankel 2005). From our experiments it is not possible
to draw any conclusions on which reaction mechanisms are dominant.
The combined effect of tocopherol and ascorbate

The use of liposomes to study the kinetics of antioxidants on the oxidation of membrane
lipids has been performed by several authors. Most of the studies have been performed with
liposomes of soy bean lecithin. There is a great variety of methods used both for initiation of
the oxidation, the timescale of the experiments, and the methods of measurement of oxidation
level.
Scarpa and others (1984) showed that ascorbate was able to regenerate the tocopheroxyl
radical incorporated in a soy lecithin liposome bilayer when ascorbate was in the surrounding
aqueous phase. The oxidation was initiated by the Fe3+-triethylenetetramine complex and the
peroxidation monitored as oxygen consumption. The ascorbyl radical and tocopheroxyl radical
were monitored with Electron Paramagnetic Resonance (EPR). When the two antioxidants were
binnenwerk.indd 92 21-08-2006 12:34:10

used together, the tocopheroxyl radical did not appear before ascorbate was fully oxidised. The
authors concluded that α-tocopherol is regenerated by ascorbate. When ascorbate was used
alone they could not detect a protective effect measured as a lower oxygen consumption.
The interaction between α-tocopherol and ascorbate in liposomes of soybean phospholipids

has also been investigated by Niki and others (1985). When oxidation was induced by
radicals produced in the lipid phase with the lipofilic diazo compound AMVN (2,2’-azobis(2,4-
dimethylpentanenitrile)), ascorbate alone had no effect, while α-tocopherol had a greater
effect. When used together the effect was synergistic.
In the presence of radicals produced in the water phase with the hydrofilic diazo compound
AAPH (2,2’-azobis-(2-amidinopropane hydrochloride)) both ascorbate and α-tocopherol showed
inhibiting effects. The effect was additive rather than synergetic when the two was used
together. The degree of oxidation was measured as oxygen uptake.
Waters II and others (1997) showed that intravesicular ascorbate also protected α-tocopherol
in the membranes of soy lecithin liposomes when a water-soluble free radical initiator was
added to the outside. The oxidation was initiated by AAPH, and the peroxidation level analysed
after 2 and 4 hours.
Based on the thermodynamic properties of free radicals Buettner (1993) has proposed a “pecking
order” of free radicals and antioxidants. Using one-electron reduction potentials, the predicted
pecking order is in agreement with experimentally observed free radical electron (hydrogen
atom) transfer reactions. These potentials are also in the agreement with the cooperation of
tocopherol and ascorbate in the protection of lipids against peroxidation. Synergism is thought
to involve (1) a chain-breaking reduction by α-tocopherol of fatty acid free peroxyl radicals
inside the lipid bilayer, (2) migration of the resulting tocopheroxyl free radical to the membrane
surface and (3) recycling of the tocopheroxyl free radical by ascorbate in the aqueous phase.
Although the synergistic antioxidant effect of ascorbate and α-tocopherol has been demonstrated
both in vitro and in vivo before, the simplicity of the marine liposome model implies that lesser
known water and lipid soluble antioxidants can be easily tested the same way. The effectiveness
and possibility for high throughput screening analysis makes the model no less attractive.
Possible improvements of the method

The method is still under revision. A major problem is that the true values for the induction
periods are not known as the time from application of initiator to the first measurement is not
measured in these experiments. The method could be improved by using a 96 channel pipette,
ensuring all microwells being filled and reactions initiated at the same time. Using a pipette
robot the time for the induction periods also could be stated more accurately and improve the
reproducibility. A measurement of the wells before inclusion of initiator should also be included
for quantitative calculations including the increase of absorbance and rate of peroxidation.
References
Bartlett GR. 1959. Phosphorous assay in column chromatography. J Biol Chem 234(3):466-468.
Bligh EG, Dyer WJ. 1959. A rapid method of total lipid extraction and purification. Can J Biochem Physiol 37(8):911-
917.
binnenwerk.indd 93 21-08-2006 12:34:10

Buettner, GR. 1993. The pecking order of free radicals and antioxidants: Lipid peroxidation, α-Tocopherol, and ascorbate.
Arch Biochem Biophys 300(2):535-543.
Frankel EN. 2005. Lipid Oxidation. 2nd edition. England: The Oily Press. p 234.
Fukuzawa K, Tokumura A, Ouchi S, Tsukatani H. 1982. Antioxidant activities of tocopherols on Fe2+-ascorbate-induced
lipid peroxidation in lecithin liposomes. Lipids 17(7):511-513.
IUPAC 1987. Method 2.505 Evidence of purity and deterioration from ultraviolet spectrophotometry. In: IUPAC Standard
Methods for the Analysis of Oils, Fats and Derivatives, 7th edition p 212-213.
Kamal-Eldin A, Appelqvist L-Å. 1996. The chemistry and antioxidant properties of tocopherols and tocotrienols. Lipids
7(7):671-701.
Løvaas E. 2006. Marine Phospholipids (MPL): Resources, applications and markets. In: Luten JB, Jacobsen C, Bekaert
K, Saebo A, Oehlenschläger J, editors. Seafood research from fish to dish: Quality, safety and processing of wild
and farmed fish, Wageningen: Wageningen Academic Publishers, pp.17-28.
Nacka F, Cansell M, Méléard P, Combe N. 2001. Incorporation of α-tocopherol in Marine Lipid-Based Liposomes: In Vitro
and in Vivo Studies. Lipids 36(12):1313-1320.
Niki E, Kawakami A, Yamamoto Y, Kamiya Y. 1985. Oxidation of Lipids. VIII. synergistic inhibition of oxidation of
phosphatidylcholine liposome in aqueous dispersion by Vitamin E and Vitamin C. Bull Chem Soc Jpn 58:1971-
1975.
Packard L, Obermüller-Jevic UC. 2002. Vitamin E: An Introduction. In Packer L, Traber MG, Kraemer K, Frei B, editors.
The Antioxidant Vitamins C and E. Illinois: AOCS Press. p 133.
Packer JE, Slater TF, Willson RL.1979. Direct observation of a free radical interaction between vitamin E and vitamin
C. Nature 278:737-738.
Scarpa M, Rigo A, Maiorino M, Ursini F, Gregolin C. 1984. Formation of alpha-tocopherol radical and recycling of
alpha-tocopherol by ascorbate during peroxidation of phosphatidylcholine liposomes. An electron paramagnetic
resonance study. Biochim Biophys Acta 801(2):215-219.
Tappell AL. 1968. Will antioxidant nutrients slow aging process? Geriatrics 23:97-105.
Waters II, RE, White LL, May JM. 1997. Liposomes containing alpha-tocopherol and ascorbate are protected from an
external oxidant stress. Free Radic Res 26(4):373-379.
binnenwerk.indd 94 21-08-2006 12:34:10

Effect of grape antioxidant dietary fibre on the prevention of lipid
oxidation in minced fish: Evaluation by different methodologies
Isabel Sánchez-Alonso, Antonio Jiménez-Escrig, Fulgencio Saura-Calixto and Javier Borderías

Instituto del Frío (CSIC), José Antonio Nováis Nº 10, 28040 Madrid, Spain
Abstract
The aim is to study the effect of grape antioxidant dietary fibre (GADF) addition to minced
fish muscle (MFM) on lipid stability during frozen storage for up to 6 months. Concentrations
of 0, 2, and 4% w/w GADF were added to MFM samples. Analyses were carried out immediately
after preparation of samples and after storage at -20 ºC for different time periods. GADF
was characterized in terms of dietary fibre, total polyphenols and antioxidant capacity,
and multifunctional antioxidant assays (DPPH• Assay, FRAP, conjugated diene and triene
hydroperoxides and thiobarbituric acid index) were tested in all the MFM samples. The addition
of red grape fibre considerably delayed lipid oxidation in minced horse mackerel muscle during
the first three months of frozen storage.
Keywords: antioxidant dietary fibre, grape polyphenols, frozen storage, lipid oxidation
Introduction
Seafoods possess high nutritional value, and also functional properties thanks to their readily-
digested protein. As such, they are a good source of vitamins and minerals, and fatty fish
further contain high concentrations of polyunsaturated fatty acids (PUFA), eicosapentaenoic
acid (EPA) and docosahexaenoic acid (DHA) (Ackman 1999). Omega-3 fatty acids (particularly
EPA) have been shown to have protective effects, helping to prevent coronary heart disease,
reduce arrhythmias and thrombosis (Kinsella and others 1990), lower plasma triglyceride levels
(Harris 1997), and reduce blood clotting tendency (Mori and others 1997; Food and Nutrition
Board 2002). However, due to the high content of unsaturated lipid, fish products are very
susceptible to loss of quality through lipid oxidation. Lipid oxidation has long been recognized
as an important cause of food quality deterioration during the processing and storage of fatty
fish (Cheftel and Cheftel 1976; Hsieh and Kinsella 1989). There has been increasing interest in
identifying plant extracts to minimize or retard lipid oxidation in lipid-based food products for
example some natural phenolic compounds that are effective in preventing rancidity in many
lipid systems like fish muscle (Ikawa 1998; Medina and others 1999, 2003).
Recent research has stressed the importance of vinification by-products as plant materials that
are particularly rich in a wide range of polyphenols (Alonso and others 2002; Torres and others
2002). Grape by-products contain significant amounts of phenolic compounds, mostly flavonoids
(Escribano-Bailón and others 1995; Mazza 1995), which have been reported as potent inhibitors
of LDL oxidation (Teissedre and others 1996; Bravo 1998). Wine by-products are also rich in
dietary fibre (DF) (Bravo and Saura-Calixto 1998). It was recently proposed to define antioxidant
dietary fibre as a natural product that combines the beneficial effects of dietary fibre and natural
binnenwerk.indd 95 21-08-2006 12:34:10

antioxidants such as polyphenol compounds (Saura-Calixto 1998). This opens up a new field of
potential for applications of these by-products in the field of healthy foods.
The objective of this work was to study the effect of grape antioxidant dietary fibre (GADF)
addition to minced fish muscle (MFM) on lipid stability during frozen storage. Horse mackerel
(Trachurus trachurus), an under-utilized semi-fatty fish species, was selected due to the interest
in stabilizing its quality and developing high-added-value products. Different methods were
used (DPPH• Assay, FRAP, conjugated diene and triene hydroperoxides and thiobarbituric acid
index) in order to assess the multifunctional antioxidant activity of GADF on MFM during frozen
storage and to compare the results produced by different methods.
Preparation of fish and samples

Minced muscle was prepared from ice-stored horse mackerel (Trachurus trachurus) fillets.
Individuals were filleted without removing the skin by a local seafood company and transported
at 2 ºC to the pilot plant. The muscle was extracted using a Baader model 694 deboner machine
(Lübeck, Germany) equipped with a drum with 3 mm holes. Red grape pomace (peels and
seeds) (Vitis vinifera var. undetermined) obtained from a winery (Vinícola de Castilla, S.A.,
Manzanares, Ciudad Real, Spain) was processed following a patented procedure (Saura-Calixto
and Larrauri 1997), freeze-dried, milled to a particle size less than 0.5 mm and stored at -20 ºC
until analysis. The final product, named grape antioxidant dietary fibre (GADF), was mixed into
the minced fish muscle (MFM) in a mixing machine model RM-20 (Mainca, Granollers, Spain).
Concentrations of 0, 2, and 4% GADF were added to MFM samples, resulting in 0-GADF, 2-GADF,
and 4-GADF samples, respectively, which were then stored at -20 ºC for 6 months. GADF was
characterized in terms of DF, total polyphenols and antioxidant capacity, and multifunctional
antioxidant assays were tested in all the MFM samples. Analyses were performed immediately
after preparation of samples and during and after storage at -20 ºC.
Chemical analysis
Dietary Fibre
The AOAC enzymatic-gravimetric method (Prosky and others 1988) was modified in our
laboratory: dialysis against water was used instead of ethanol precipitation of soluble dietary
fibre (sDF) (Mañas and others 1995) for DF analysis. After enzymatic hydrolysis of digestible
components, insoluble DF (iDF) and soluble DF (sDF) fractions were separated and chemically
hydrolysed. iDF fractions were hydrolysed with 12 M sulphuric acid (30 ºC, 1 hour) then diluted
to 1 M sulphuric acid and hydrolysed at 100 ºC during 90 min. The remaining residues were
gravimetrically quantified as Klason lignin (KL) after drying at 105 ºC to constant weight. sDF
dialysates were hydrolysed with 1 M sulphuric acid (100 ºC, 90 min). Constituent neutral sugars
(NS) and uronic acids (UA) were quantified in the hydrolysates. NS were quantified by Gas-
Cromatography (Jiménez-Escrig and others 2001). UA were quantified spectrophotometrically
by the Scott method (Scott 1979) using galacturonic acid as standard. iDF was calculated as
(NS+UA+KL), and sDF as (NS+UA).
Extraction of Phenolics
One g of ground freeze-dried GADF sample was placed in a test tube; 40 mL methanol/water
(50:50) was added, plus HCl to obtain a final pH of 2.0. The solution was thoroughly shaken
at room temperature for 1 hour and centrifuged at 2500 x g for 10 min, and the supernatant
binnenwerk.indd 96 21-08-2006 12:34:10

Effect of grape antioxidant dietary fibre on the prevention of lipid oxidation in minced fish
was separated. 40 mL of acetone/water (70:30) was added to the residue, and shaking and
centrifugation were repeated. Both extracts were mixed. This procedure has been used by our
group and is described elsewhere (Jiménez-Escrig and others 2001). Extractions were performed
in triplicate and used to calculate the total phenolics content and the antioxidant capacity.
The total phenolics was determined spectrophotometrically according to the Folin-Ciocalteu
procedure (Singleton and others 1999) using gallic acid as standard (concentration range 5-25
mg per 100 mL), and the results were expressed as gallic acid equivalents (GAE).
Lipid damage measurements

All the antioxidant analyses were performed in 0-GADF, 2-GADF and 4-GADF samples.
DPPH• Assay
This assay is based in the reaction of a radical scavenger towards DPPH• radical (the dot indicates
that DPPH is a free radical), mainly by hydrogen donating. The depletion in the absorbance of
515 nm indicates the formation of DPPH-H from DPPH• radical, and consequently the radical
scavenging activity of the extracts assayed. The antioxidant activity of the GADF plus MFM was
measured in terms of radical scavenging activity, according to the DPPH• method (Sánchez-
Moreno and others 1998). Briefly, an aliquot of sample extract at different concentrations was
added to a DPPH• solution, and the absorbance at 515 nm was measured until the reaction
reached a plateau. A calibration curve at 515 nm was plotted with DPPH• to calculate the
remaining DPPH• concentration in the reaction medium. The parameter EC50, which reflects
the depletion of DPPH• to 50%, was expressed in terms of grams of GADF (dry matter) plus
MFM equivalent per gram of DPPH• in the reaction medium. The lower the value the higher the
antioxidant activity.
FRAP Assay
The ferric ion reducing antioxidant power (FRAP) is a method based on a single electron transfer
reaction between an oxidant and antioxidants, i.e. phenol antioxidants are oxidized by the
oxidant Fe (III). As a result a single electron is transferred from the antioxidant molecule to the
oxidant. The standard redox potential of Fe (III)/Fe(II) is 0.77 V. Any compound with lower redox
potential can theoretically reduce Fe(III) to Fe(II) (Ou and others 2002). It is more relevant
in the description of the method. The reducing power of samples was estimated following the
procedure described by Benzie and Strain (1996) with some modifications introduced in our
laboratory (Jiménez-Escrig and others 2003). Briefly, 900 µL of FRAP reagent, freshly prepared
and warmed at 37 ºC, was mixed with 90 µL distilled water and either 30 µL of test sample or
standard or appropriate reagent blank. The FRAP reagent contained 2.5 mL of a 10 mM TPTZ
solution in 40 mM HCl, plus 2.5 ml of 20 mM FeCl3-6H2O, plus 25 ml 0.3 mM acetate buffer pH
3.6. Readings at the absorption maximum (595 nm) were taken every 15 s using a Beckman
DU-640 spectrophotometer (Beckman Instruments Inc., Fullerton, CA, USA) equipped with a
thermostatic autocell-holder. Temperature was maintained at 37 ºC. Readings were taken at 30
min for calculation of FRAP values. Methanolic solutions of known Trolox concentrations were
used for calibration. The antioxidant results were expressed as µmol equivalents of Trolox per
g of sample.
Measurement of conjugated diene and triene hydroperoxides

Lipids were extracted from mackerel muscle (Bligh and Dyer 1959) and the lipid content was
determined gravimetrically in triplicate (Herbes and Allen 1983). Conjugated hydroperoxides were
measured from fish oil samples dissolved in hexane, and absorbance was measured at 234 nm
binnenwerk.indd 97 21-08-2006 12:34:11

and 268 nm. Concentrations of hydroperoxides were calculated as mmol of hydroperoxides per
kilogram of oil as described by Frankel and others (1994).
The thiobarbituric acid index (TBA-index)

The TBA-index was determined according to Vyncke (1970) on a 5% trichloracetic acid extract
of the restructured fish muscle. Results were expressed as mg malondialdehyde per kilogram
of sample. The spectrophotometer used was a UV/VIS Spectrophotometer Perkin-Elmer
Lambda 15.
Antioxidant effectiveness
In both instances, conjugated hydroperoxides and thiobarbituric acid index, the antioxidant
effectiveness was calculated as per cent inhibition of oxidation (% I) as described by Frankel
(1998): % I = (c-s/c)*100, where c= high increment of 0-GADF in the experiment and s=
increment of sample with added GADF at the same time. High levels of % I indicate greater
antioxidant effectiveness.
Statistical analyses
Tukey´s test (p<0.05.) was used to determine significant differences in the mean values at
different time points. The Statgraphics Plus 2.1 program was used for this.
Characterization of grape antioxidant dietary fibre

The chief characteristics of this natural product are that it is rich in both DF and polyphenolic
compounds and also presents relatively (Jimenez-Escrig and others 2001, 2003) high antioxidant
activity (ferric reducing ability and radical scavenging capacity as measured in vitro (Table 1).
It is worth noting that the sDF content of this GADF is relatively higher in comparison with
total DF content. The physical and chemical characteristics of the DF fractions will dictate the
specific local response in the gut and the associated systemic reactions. sDF is distinguished
by its ability to form viscous gels in the intestinal tracts. iDF does not exhibit viscosity but
instead is characterized by faecal-bulking capacity. Both forms of fibre share the ability to bind
water or mineral cations and can be used by the colonic microflora as a fermentable substrate
(Schieber and others 2001).
Table 1. Proximate composition of grape antioxidant dietary fibre (GADF), radical scavenging capacity
(DPPH•), and ferric reducing ability (FRAP)1.
Protein (% d.m.) 8.1 ± 0.4

Fat (% d.m.) 9.4 ± 0.1
Ash (% d.m.) 2.4 ± 0.1
Extractable Polyphenols (g GAE/ 100 g d.m.) 5.6 ± 0.2
Insoluble Dietary Fibre (% d.m.) 53.2 ± 3.4
Soluble Dietary Fibre (% d.m.) 20.78 ± 0.76
DPPH• CE50% (g GADF (d.m.)/g DPPH•) 153 ± 9
FRAP (µmol TE/ g of GADF (d.m.)) 525 ± 28
1 Valuesare Mean ± Standard Deviation of three replicate determinations. d.m.: dry matter. GAE = Gallic
Acid Equivalents; TE = Trolox Equivalents
binnenwerk.indd 98 21-08-2006 12:34:11

Antioxidant assays
A multifunctional methodology was assayed to assess how well GADF protected MFM against
oxidation during storage (six months) at -20 ºC. Some authors distinguish between the antioxidant
capacity and the reactivity (Roginsky and Lissi 2005) of polyphenols. As they describe them,
while the antioxidant capacity gives information about the duration of antioxidative action
(ability to retard oxidative degradation), the reactivity characterizes the starting dynamics of
antioxidation at certain concentrations of an antioxidant or an antioxidant mixture (determined
by the reactivity of an antioxidant to active free radicals in the corresponding reaction). For this
reason two kinds of analytical methods were used: indirect methods to examine the ability of
GADF in MFM to scavenge some free radicals (DPPH•) and the ferric reducing ability of plasma
(FRAP); and direct methods basically entailing studies of chain peroxidation and the effect of a
tested food containing antioxidants on the oxidative degradation of a testing system (formation
of conjugated hydroperoxides and aldehydes).
At the end of the third and the sixth months of storage, the ability to reduce Fe(III) to
Fe(II) was greater in both 4-GADF and 2-GADF samples than in MFM without GADF (Table 2).
Effectiveness was greatest (225%) at the end of the third month in the 4-GADF sample. The
same results were found for radical scavenging activity (DPPH•), in which antioxidant capacity
at the end of the third month of storage was higher in both MFMs with GADF than in the MFM
without GADF (Table 3). According to the DPPH• assay, 4-GADF provided greater protection
than 2-GADF. This assay may be used to indicate the scavenging effect of the polyphenols in
the fish matrix. A high decrease in antioxidant activity during the first three months is shown
in both assays, and it thus seems that a major part of antioxidants from GADF were consumed
Table 2. Ferric reduction ability (FRAP) of MFM added with GADF during six months of frozen storage
(-20 ºC)1.
Sample Day 0 Day 90 Day 180
0-GADF 27 ± 3.1a 4 ± 0.2a 3.4 ± 0.5ª

2-GADF 38 ± 3.8b 6 ± 0.6b 5.3 ± 0.8b
4-GADF 53 ± 3.2c 13 ± 1.1c 8.1 ± 0.8c
1 Values are mean ± standard deviation of three replicate determinations. d.m.: dry matter. Different
letters in the same column indicate significant difference (P<0.05).
Table 3. Radical scavenging capacity (DPPH•) expressed as CE50% (g GADF (d.m.)/g DPPH•) of MFM added
with GADF during three months of frozen storage (-20 ºC)1.
Sample Day 0 Day 90
0-GADF 12.1 ± 1.1a 350 ± 17a

2-GADF 11.0 ± 0.2b 193 ± 12b
4-GADF 7.1 ± 0.5c 156 ± 9.8c
1 Values are Mean ± Standard Deviation of three replicate determinations. d.m.: dry matter. Different
letters in the same column indicate significant difference (P<0.05).
binnenwerk.indd 99 21-08-2006 12:34:11

(a) Dienes
90
mmolhydrp/kglip 60
30
0 30 60 90 120 150 180
Days
30 (b) Trienes
mmolhydrp/kglip
20
10
0
0 30 60 90 120 150 180
Days
0-GADF 2-GADF 4-GADF
Figure 1. Formation of conjugated dienes (a) and trienes (b) during frozen storage at -20 ºC of MFM samples
with 0, 2 or 4% GADF added.
to inhibit lipid oxidation of the MFM sample. No detectable effect was observed after 90 days
of storage in all the samples.
By continuous monitoring, during frozen storage, of conjugated hydroperoxides and aldehydes

formed during lipid oxidation, the process can be observed in close detail. Figure 1 depicts the
formation of conjugated dienes (a) and trienes (b) in minced horse mackerel after six months
of frozen storage. During frozen storage, the samples with dietary fibre added (2-GADF and
4-GADF) exhibited less formation of dienes and trienes than the control lot (0-GADF) until 60
days. The rate of oxidation inhibition (Table 4) when the control increments were highest (at
30 days of storage) was 26.42 ± 0.08% for the lot with 2% GADF and 62.34 ± 1.02% for the lot
with 4% GADF added. From these values it would seem that GADF protects the initial formation
of oxidation compounds at 30 days of frozen storage. TBA-index values were lower in the
samples with dietary fibre added for most of the storage period (Figure 2). The rate of inhibition
at 90 days of frozen storage (higher control values of TBA-index) was 57.28 ± 9.17% for the
samples with 2% added fibre and 54.13 ± 12.16% for samples with 4% added fibre (Table 4).
The antioxidant action was not significantly different for the two levels of dietary fibre at this
time of frozen storage. There was less inhibition of conjugated hydroperoxide formation than of
binnenwerk.indd 100 21-08-2006 12:34:12

Table 4. Percentage of inhibition of formation of conjugated dienes and trienes and thiobarbituric index in
minced fish samples (mean±sd)1.
Sample Dienes Trienes i-TBA

(mmolhydrop/kglip) (mmolhydrop/kglip) (mgMDA/Kgmus)
Day 30 Day 30 Day 90
0-GADF 0.00 ± 8.96a 0.00 ± 6.32a 0.00 ± 4.49a

2-GADF 26.42 ± 0.08b 32.16 ± 4.25b 57.28 ± 9.17b
4-GADF 62.34 ± 1.02c 89.43 ± 7.79c 54.13 ± 12.16b
1 Different letters in the same column indicate significant differences (P<0.05).
TBA-i
3
mgMDA/Kgmuscl
0
0 30 60 90 120 150 180
Days
0-GADF 2-GADF 4-GADF
Figure 2. Formation of aldehydes during frozen storage at -20 ºC of MFM samples with 0, 2 or 4% GADF
added.
aldehyde formation; the same effect has been described in the case of tocopherols, which can
increase primary lipid oxidation products by donating hydrogen to a peroxyl radical to form a
lipid hydroperoxide while simultaneously decreasing formation of low molecular weight volatile
secondary compounds of oxidation (Decker and others 2005). The grape flavonoids in GADF
have considerable potential as antioxidants based on the combined actions of nonextractable
poplyphenols (NEPP) (polymeric proanthocyanidins and high molecular weight hydrolysable
tannins) and extractable polyphenols (EPP) (anthocyanins, flavonols, flavan-3-ols and phenolic
acids), both of which show some promise in the fields of nutrition and health. Use of EPPs as
antioxidants has been widely reported (Macheix and others 1991, Escribano-Bailón and others
1995). Hagerman and others (1998) reported that NEPPs were 15 to 30 times more effective
at quenching peroxyl radicals than were simple phenols. Flavonoids are able to scavenge the
radicals of hydroxyl, peroxyl, superoxide and nitric oxide. Besides free radical scavenging
activities, flavonoids possess metal chelating properties (Bors and others 2000, Yusuf and
Romeo 2004) which are very important in the development of rancidity in fish (Ramanathan
and Das 1993; Richards and Hultin 2002). Pazos and others (2005a) have reported that an
optimal combination of degree of procyanidin polymerization and percentage of galloylation, as
binnenwerk.indd 101 21-08-2006 12:34:13

in GADF, may help account for the high antioxidant efficacy of grape polyphenols in frozen fish
muscle. Pazos and others (2005b) also suggested that grape procyanidins are able to stabilize
frozen fatty fish and preserve vitamin E (endogenous antioxidant of fish). These results are
consistent with our own findings of high GADF antioxidant capacity during frozen storage in a
matrix made of mince muscle. We found high antiradical activity in the samples with GADF, but
this was not always accompanied by high antioxidant activity in foods. This study shows that
antiradical activity (measured as H-donating activity) in samples with GADF and antioxidant
capacity retarding the formation of degradation products from lipids in MFM are correlated. The
results suggest that GADF possesses notable antioxidant properties and is able to significantly
inhibit the development of rancidity in frozen fish muscle during the first three months. All
methodologies produced similar results, which suggest that methods of this kind are valid for
assessing the antioxidant capacity of GADF added to frozen mince muscle.
As reported above, GADF is a good antioxidant which could not only preserve MFM from oxidation
during storage but could also produce health benefits for the consumer thanks to its bioactive
compounds and DF content. It should be noted in this connection that in vitro activities can
only be considered potentially relevant in biological systems and that in vivo activities also
depend on bioavailability and biotransformation. It is known that polyphenols such as gallic
acid, caffeic acid, malvidin, catechine, and rutine, which are contained in red grape, can be
partly bioavailable (Manach and others 2005), and consequently their in vivo activities could
be significant.
Conclusions
All methods are valid and concordant in the evaluation of antioxidant capacity of GADF added to
frozen minced muscle during six months of storage. The addition of red grape fibre considerably
inhibits oxidation in horse mackerel minced muscle during the first three months of frozen
storage. These results indicate that GADF could be used as an ingredient to prevent oxidation
in minced fish during frozen storage. The reason for this effect may be either the chelant action
of fibre on some prooxidant metals or the action of polyphenols associated with dietary fibre.
List of abbreviations
DF dietary fibre
DM dry matter
FRAP ferric-reducing antioxidant power
GAE gallic acid equivalent
iDF insoluble dietary fibre
KL Klason lignin
NS neutral sugars
sDF soluble dietary fibre
TPTZ 2,4,6-tri(2-pyridyl)-s-triazine
Trolox 6-hydroxy-2,5,7,8-tetramethylchroman-2-carboxylic acid
UA uronic acids
MFM minced fish muscle
GADF grape antioxidant dietary fibre
TBA-i thiobarbituric acid index.
binnenwerk.indd 102 21-08-2006 12:34:13

Acknowledgements
This work has been supported by the Spanish Ministerio de Educación y Ciencia under Projects
AGL2002-04104-C04-03 and AGL2002-04104-C04-01, and by the EU under Integrated Project
SEAFOODplus (Ref. FP6/506359).
References
Ackman RG. 1999. Marine biogenic lipids, fats and oils. Boca Raton, FL: CRC Press.
Alonso AM, Guillén DA, Barroso CG, Puertas B, García A. 2002. Determination of antioxidant activity of wine byproducts
and its correlation with polyphenolic content. J Agric Food Chem 50:5832-5836.
Benzie II, Strain JJ. 1996. The ferric reducing ability of plasma (FRAP) as a measure of “Antioxidant Power”: The FRAP
Assay. Anal Biochem 239(1):70-76.
Bligh EG, Dyer WJ. 1959. A rapid method of total lipid extraction and purification. Can J Biochem Physio 37(8):911-917.
Bors W, Michel C, Stettmeier K. 2000. Electron paramagnetic resonance studies of radical species of proanthocyanidins
and gallate esters. Archives Biochem Biophys, 374:347-355.
Bravo L. 1998. Polyphenols: Chemistry, dietary sources, metabolism, and nutritional Significance. Nutr Rev 56(11):317-333.
Bravo L, Saura-Calixto F. 1998. Characterization of dietary fiber and the in vitro indigestible fraction of grape pomace.
Am J Enol Vitic 49:135-41.
Cheftel J, Cheftel H. 1976. Introducción a la Biología y Tecnología de Alimentos; Acribia: Zaragoza, Spain.
Decker EA, Warner K, Richards M, Shahidi F. 2005. Measuring antioxidant effectiveness in food. J Agric Food Chem
53:4303-4310.
Escribano-Bailón MT, Guerra MT, Rivas-Gonzalo JC, Santos-Buelga C. 1995. Proanthocyanidins in skins from different
grape varieties. Z Lebensm Unters Forsch 200:201-224.
Food and Nutrition Board, Institute of Medicine of the National Academies. 2002. Dietary reference intakes for energy
carbohydrate, fiber, fat, fatty acids, cholesterol, protein, and amino acids. Washington, DC: The National Academies
Press.
Frankel EN. 1998. Antioxidants. In Lipid Oxidation. Dundee, Scotland: The Oily Press, 303 p.
Frankel EN, Huang SW, Kanner J, Bruce German JB. 1994. Interfacial phenomena in the evaluation of antioxidants: Bulk
oils versus emulsions. J Agric Food Chem 42:1054-1059.
Hagerman AE, Rield KM, Jones GA. 1998. High molecular weight plant polyphenolics (tannins) as biological antioxidants.
Harris WS. 1997. N-3 Fatty acids and serum lipoproteins: human studies. Am J Clin Nutr 65(5):1645S-1654S.
Herbes S, Allen C. 1983. Lipid quantification of freshwater invertebrates: Method modification for microquantitation.
Can J Fish Aquat Sci (14):1315-1317.
Hsieh R, Kinsella J. 1989. Oxidation of polyunsaturated fatty acids: Mechanisms, products and inhibition with emphasis
on fish. Adv Food Nutr Res 33:233-341.
Ikawa Y. 1998. Use of tea extracts (sanfood) in fish paste products. New Food Ind 40:33-39.
Jimenez-Escrig A, Rincon M, Pulido R, Saura-Calixto F. 2001. Guava fruit (Psidium guajava L.) as a new source of
antioxidant dietary fiber. J Agric Food Chem 49(11):5489-5493.
Jiménez-Escrig A, Dragsted LO, Daneshvar B, Pulido R, Saura-Calixto F. 2003. In vitro antioxidant activities of edible
artichoke (Cynara scolymus L.) and effect on biomarkers of antioxidants in rats. J Agric Food Chem 51:5540-5545.
Kinsella JE, Lokesh B, Stone RA. 1990. Dietary n-3 polyunsaturated fatty acids and amelioration of cardiovascular
disease: possible mechanisms. Am J Clin Nutr 52:1-28.
Macheix JJ, Sapis CJ, Fleuriet A. 1991. Phenolic compounds and polyphenoloxidase in relation to browning in grapes
and wines. CRC Crit Rev Food Sci Nutr 30:441-486.
Manach C, Wiliamson G, Morand C, Scalbert A, Remesy C. 2005. Bioavailability and bioefficacy of polyphenols in humans.
I. Review of 97 bioavailability studies. Am J Clin Nutr 81:230s-242s.
binnenwerk.indd 103 21-08-2006 12:34:13

Mañas E, Saura-Calixto F. 1995. Dietary fibre analysis: methodological error sources. Eur J Clin Nutr 49:S158-S162.
Mazza G. 1995. Anthocyanins in grapes and grapes products. Crit Rev Food Sci Nutr 35-341-371.
Medina I, González MJ, Pazos M, Medaglia DD, Sacchi R, Gallardo JM. 2003. Activity of plant extracts for preserving
functional food containing n-3 PUFA. Euro Food Res Technol 217:301-307.
Medina I, Satué-García MT, German JB, Frankel, EN. 1999. Comparison of natural polyphenol antioxidants from extra virgin
olive oil with synthetic antioxidants in tuna lipids during thermal oxidation. J Agric Food Chem 47(12):4873-4879.
Mori TA, Beilin LJ, Burke V, Morris J, Ritchie J. 1997. Interactions between dietary fat, fish, and oils and their effects
on platelet function in men at risk of cardiovascular disease. Arterioscler Thomb Vasc Biol 17:279-286.
Ou B, Huang D, Hampsch-Woodill M, Flanagan JA, Deemer EK. 2002. Analysis of antioxidant activities of common
vegetables employing oxygen radical absorbance capacity (ORAC) and ferric reducing antioxidant power (FRAP)
assays: a comparative study. J Agric Food Chem 50:3122–3128.
Pazos M, Gallardo JM, Torres JL, Medina I. 2005a. Activity of grape polyphenol as inhibitors of the oxidation of fish
lipids and frozen fish muscle. Food Chem 92:547-557.
Pazos M, González MJ, Gallardo JM, Torres JL. 2005b. Preservation of the endogenous antioxidant system of fish muscle
by grape polyphenols during frozen storage. Eur Food Res Technol 220:514-519.
Prosky L, Asp NG, Schweizer TF, Devries JW, Furda IJ. 1988. Determination of insoluble, soluble and total dietary fiber
in foods and food products: interlaboratory study. J Assoc Anal Chem 71:1017-1023.
Ramanathan L, Das NP. 1993. Effect of natural copper chelating compounds on the prooxidant activity of ascorbic-acid
in steam-cooked ground fish. Int J Food Sci Tech 28(3):279-288.
Richards, MP, Hultin HO. 2002. Contributions of blood and blood components to lipid oxidation in fish muscle. J Agric
Food Chem 50(3):555-564.
Roginsky V, Lissi EA. 2005. Review of methods to determine chain-breaking antioxidant activity in food. Food Chem
92:235-254.
Sanchez-Moreno C, Larrauri JA, Saura-Calixto F. 1998. A procedure to measure the antiradical efficiency of polyphenols.
J Sci Food Agric 76:270-276.
Saura-Calixto F. 1998. Antioxidant Dietary Fiber Product: A new concept and a potential food ingredient. J Agric Food
Chem 48:4303-4306.
Saura-Calixto F, Larrauri J, inventors; CSIC. 1997. Concentrado de fibra dietética antioxidante natural de uva y su
procedimiento de obtención. Patente Española 9702397.
Schieber A, Stintzing FC, Carle R. 2001. By-products of plant food processing as a source of functional compounds-
recent developments. Trends Food Sci Tech 12(11):401-413.
Scott RW. 1979. Colorimetric determination of hexauronic acids in plant materials. Anal Chem 51:936-941.
Singleton VL, Orthofer R, Lamuela-Raventós RM. 1999. Analysis of total phenols and other oxidation substrates and
antioxidants by means of Folin- Ciocalteau reagent. Meth Enzymol 299:152-78.
Teissedre PL, Frankel EN, Waterhouse AL, Peteg H, German B. 1996. Inhibition of in vitro human LDL oxidation by
phenolic antioxidants from grapes and wines. J Sci Food Agric 70:55-61.
Torres JL, Varela B, García MT, Carilla J, Matito C, Centelles, JJ, Cascante M, Sort X and Bobet R. 2002. Valorization
og grape (Vitis vinifera) byproducts. Antioxidant and biological properties of polyphenolic fractions differing in
procyanidin composition and flavomol content. J Agric Food Chem 50:7548-7555.
Vyncke W. 1970. Direct determination of the thiobarbituric acid value in trichloracetic acid extracts of fish as measure
of oxidative rancidity. Fette Seifen Anstrichmittel 72(12):1084-1087.
Yusuf Y, Romeo TT. 2004. Health aspects of functional grape seed constituents. Trends Food Sci Tech 15:422-433.
binnenwerk.indd 104 21-08-2006 12:34:13

Effects of antioxidants on copper induced lipid oxidation during
salting of cod (Gadus morhua L.)
Kristin Lauritzsen1 and Ragnar L. Olsen2

2The Norwegian College of Fishery Science, University of Tromsø, Norway
Abstract
A model system for studying lipid oxidation of salted cod muscle was used for investigating the
effects of antioxidants and copper in the brine. The results showed that ascorbate might have
pro-oxidative or anti-oxidative effects depending on the ascorbate and metal concentrations.
Without added copper in the brine, concentrations of ≤ 500 mg/kg ascorbate had a pro-oxidative
effect. With 5 mg/kg copper added in the brine, low concentrations of ascorbate (≤ 50 mg/kg)
reduced the formation of TBARS in the cured product. At slightly higher concentrations (100-
200 mg/kg), the anti-oxidative properties were lost. Above 200 mg/kg the added ascorbate
reduced the oxidation level in the salt-ripened product. The application of ascorbate as an
antioxidant in salt curing of cod, requires mostly the use of high concentrations (≥ 1000 mg/
kg) in the brine. When similar concentrations (0.5 mM) of EDTA (ethylenediaminetetraacetic
acid), citrate or ascorbate were included with 3 mg/kg copper in the brine, EDTA was the only
compound that efficiently inhibited copper-induced lipid oxidation.
Keywords: cod, antioxidants, copper, lipid, oxidation
Introduction
Heavily salted cod are traditional products from the North-Atlantic fisheries and are highly
regarded as ripened fish products in many countries. Today, thoroughly washed split or filleted
cod are usually pickle-salted or brine-cured for a short period and then salt ripened, i.e.
kench cured in stacks, for at least 10 days. A major quality problem in such skin-on products
is yellow/brown discolouration of the flesh surface. Seafood containing n-3 polyunsaturated
fatty acids are highly susceptible to oxidation even when the fat content is very low as in
the muscle of cod fish (Dulavik and others 1998). Lipid oxidation causes undesirable colour
and flavours changes and a decrease in the n-3 polyunsaturated fatty acids that are known to
be beneficial to human health. Lipid oxidation products may even accelerate atherosclerotic
processes, coronary heart disease (Kubow 1992) and carcineogenesis (McBrien and Slater 1982).
In previous studies yellow/brown discolouration of the muscle surface was investigated in
downgraded heavily salted cod fillets from a processing plant (Lauritzsen and others 1999). The
results showed that discoloured areas were relatively evenly distributed on the muscle surface
and that lipid oxidation correlated with the copper content in the muscle. A model system for
studying lipid oxidation and yellow discolouration of salted cod muscle was established and
used for investigating the effects of transition metals during the salt curing process. The study
showed that the reduced form of copper is particularly pro-oxidative.
binnenwerk.indd 105 21-08-2006 12:34:13

In processed food, the pro-oxidative transition metals may originate from the raw material
itself or from equipment used during processing. Due to the fairly unrefined and bulk nature of
solar salt most often used during salt curing, it is not always easy to comply with the limits
of 0.1 mg/kg copper and 10 mg/kg iron in the salt. In the food industry, metal chelators are
commonly used to minimize the catalytic effect of transition metals in the peroxidation process
(Pokorný 1987; Løliger 1991). The most important chelators are water-soluble compounds
such as citric acid, ethylenediaminetetraacetic acid (EDTA), phosphoric acid derivatives and
some amino acids (Hølmer 1995). These function by coordinating the metals and change their
potential by suppressing the redox reactions producing peroxyl and alkoxylradicals. Chelators
also block complex formation with hydroperoxides and prevent their decomposition. Ascorbic
acid is another commonly used water soluble antioxidant that exerts its antioxidative effect
by radical scavenging via direct reaction with hydrophilic free radicals (Niki 1991) or by acting
as an O2 scavenger (Klaui and Pongracz 1981). On the other hand, ascorbic acid may be a
prooxidant in the presence of oxidised forms of transition metals. The effectiveness of various
antioxidants appears to be influenced by either the phase of reaction systems or the type of
radicals (Lin and Liang 2002). Antioxidant effects on transition metal induced lipid oxidation
have been extensively studied using oils or fatty acids/proteins/polysaccharides model systems
(Kanner and others 1977; Pokorný 1987; Harel 1994; Jacobsen and others 1999). Blended or
intact muscle tissues are however more complex. In the present study, a model system based
on brining and subsequent kench curing of fresh cod muscle was used to study the effects of
ascorbate, citrate and EDTA on copper induced lipid oxidation.
Model system
A model system simulating normal processing conditions of salted cod was used. The cod was
caught in November, June, August and September in the coastal waters of Northern Norway
and had a round weight of 2 to 5 kg. The fish was gutted, headed and stored in ice for 2-3
days before filleting and skinning. In each experiment, approximately 15-20 deboned fillets
were manually cut into 1-2 cm cubes of muscle tissue and mixed together to minimize the
variation among individuals. The cubes were brined for 5 days with a saturated salt solution
(approximately 26% (w/w) NaCl in distilled water). The saturated salt solution had added
copper (Cu) and antioxidants (ascorbate, citrate and EDTA) at different concentrations. Copper
was added as CuCl2, ascorbate as sodium ascorbate, citrate as sodiumcitrate and and EDTA was
disodium salt. Subsequently, the cubes were removed from the brine, drained for 30 seconds
and kench cured for 16 days using solid NaCl in a weight ratio of 1 part fish and 2 parts salt.
No antioxidants or metals were included during the kench curing. The temperature during the
experiments was 8-10 °C. All chemicals used for brine preparations, dry salting and chemical
analyses were obtained from Merck (Darmstadt, Germany).
The effects of increasing ascorbate concentration in the brine

The effect of increasing ascorbate concentration in the brine on copper induced lipid oxidation
was investigated in the model system using 200 g muscle cubes and 1600 g saturated salt
solution for each ascorbate concentration. The fish was caught in June, August and September
and the post rigor pH of the fillets was in the range of 6.42-6.72 and the pH of the brines
containing ascorbate with or without copper, was in the range of 6.31 to 6.59 prior to salting.
To the brines were added 0 or 5 mg/kg (0.079mM) of Cu2+ and 0, 10, 30, 50, 100, 200, 500,
1000, 2000, 5000, 7,000 or 10,000 mg/kg (w/w) ascorbate. The experiments were replicated
binnenwerk.indd 106 21-08-2006 12:34:14

Oxidation in salted cod
two times for each ascorbate concentration. Lipid oxidation as 2-thiobarbituric acid reactive
substances (TBARS) and instrumental yellow colour (Minolta Camera Co. Ltd., Osaka, Japan) of
salted muscle cubes were determined after 21 days.
The effects of using different antioxidants in the brine

The effect of using different antioxidants in the brine on copper induced lipid oxidation was
studied in the model system using 1400 g muscle cubes and 5000 g saturated brine for each
brine combination. The fish was caught in June and November and the post rigor pH of the
fillets was in the range of 6.52 to 6.73 and the pH of the brines containing antioxidant with or
without copper was in the range of 5.25 to 5.66 prior to salting. To the brines were added 0 or
3 mg/kg (0.047 mM) Cu2+ and 100 mg/kg ascorbate, 150 mg/kg citrate or 193 mg/kg of EDTA
giving approximately 0.5 mM antioxidant in the respective brines. 3 mg/kg copper was used
due to the low antioxidant concentration investigated. Both from the salted fish producer’s and
consumer’s point of view, the use of low concentrations of antioxidants in the process would be
preferable. The experiments were made to compare the effects from EDTA, citrate and ascorbate
at low levels on salted cod. They were carried out twice for each type of antioxidant. Lipid
oxidation as TBARS and instrumental yellow colour, muscle pH, copper, NaCl and water contents
of the muscle cubes were determined after 0, 0.5, 1, 2, 3, 4, 5 and 21 days of salting.
Sample preparation
Lipid oxidation and the chemical analysis were performed on pooled muscle samples. The pooled
samples were prepared by mincing the muscle cubes in a Dito Sama (K55) Food Processor
(Aubusson, France) for approximately 1 min. The samples were either analysed immediately
(Instrumental yellow colour and muscle pH) or packed in sealed plastic bags and stored for 1 to
4 weeks at –80 °C before analysis. Samples were either taken at the start (0 days) and during
the brining period (0, 0.5, 1, 2, 3, 4, 5 days) or at the end of the salting process (21 days).
At each sampling, 50-100 g of muscle cubes (approximately 25 – 50 pieces) were removed and
homogenized as described above.
Lipid oxidation
Lipid oxidation was estimated by determining the amount of TBARS in an aqueous trichloroacetic
acid extract (Witte and others 1970) as described previously (Dulavik and others 1998). Two
replicate extractions were made for each pooled sample of muscle cubes. Instrumental yellow
colour of the cured muscle mince was determined using a Minolta Chromameter, CR-200 (Minolta
Camera Co. Ltd., Osaka, Japan). The detector was placed on the surface of the plastic bag
containing muscle mince, at four different areas of the sample recording the L* a* b* modus,
obtaining a mean value.
Copper concentration of the fish muscle

The concentrations of copper in the muscle were determined after wet digestion; i.e. minced
muscle, 8-10 g, was boiled in 25 mL HCl (25% V/V). The metals were analysed by flame atomic-
absorption spectrometry as described by Simpson and Blay (1966) using a Perkin Elmer 3110
spectroscope (Perkin Elmer Co. Ltd., Norwalk, CT). Standards of copper and iron (Titrisol,
Merck, Darmstadt, Germany) were also diluted in 25% (w/w) HCl. Three replicates for each
pooled muscle sample were made. The mean and standard deviation were calculated from these
values.
binnenwerk.indd 107 21-08-2006 12:34:14

Muscle-pH
The pH of fresh muscle tissue was measured in a 1:1 mixture of muscle homogenate and 0.15
M KCl, while the pH of salted muscle was recorded after mixing the muscle homogenate with 5
parts of distilled water. The pH of the brines prior to salting was measured in a 1:4 dilution of
the brine homogenate with distilled water. A PHM 80 Radiometer (Copenhagen, Denmark) with
a glass electrode was used and 3 – 6 replicates of each sample were made to obtain a mean
value. Two or three replicates of each sample were analysed.
Water and NaCl content of the fish muscle

The water content of the muscle samples was determined by drying to a constant weight at
105 °C (AOAC [950.46] 1990) and the NaCl content was determined by the standard procedure
(AOAC [937.09] 1990). Three replicates of each sample were analysed for water and salt
determinations.
Statistical analysis of the analytical data from muscle samples was performed on the software
program SAS version 6.12 (SAS Institute, Cary, NC). Data were subjected to one-way analysis
of variance (ANOVA) by using the general linear model procedure. Where statistical differences
were noted for a measurement, differences among sample means were determined using the
Tukey’s Multiple comparison test. The level of significance was set at p = 0.05 for all tests.
Results
Effects of increasing concentration of ascorbate in the brine

A model system consisting of small muscle cubes of fresh cod and brines simulating normal
processing conditions was used to study copper-induced lipid oxidation. Figure 1 shows the
effects of an increasing ascorbate concentration when the model system was used without
addition of copper in the brine. An ascorbate concentration of <500 mg/kg in the brine induced
lipid oxidation as reflected in both TBARS and instrumental b* values of the salt ripened
product. At 1000 mg/kg ascorbate or higher, no pro-oxidant activity could be detected. The
TBARS value at this concentration was actually lower (not significantly, p>0.05) than in the
samples without ascorbate added.
Figures 2a and b show the TBARS and instrumental yellow colour (b*) values of kench cured
cod muscle cubes brined in saturated sodium chloride solution containing a fixed Cu2+ (5
mg/kg) and different concentrations of ascorbate (10-2000 mg/kg). With 5 mg/kg Cu2+ and
no antioxidant added, the TBARS values were approximately 280 nmol MDA/g after 21 days
of curing (Fig 2a). When the ascorbate concentration increased from 0 to 50 mg/kg, the
TBARS values decreased to approximately 40 nmol MDA/g, indicating antioxidative effects on
copper-induced lipid oxidation in the product. However, further increase in the concentration of
ascorbate to 100 mg/kg, sharply raised the TBARS values significantly (p<0.05) to approximately
the same level as found when no ascorbate was added. Higher concentrations of ascorbate
present during the initial period inhibited the formation of copper induced TBARS (Fig 2b).
With 1000 mg/kg the lowest value of oxidation products was reached. In general, the results of
the instrumental yellow colour (b*) measurements followed the TBARS values. However, with
the lowest concentrations of ascorbate (10-30 mg/kg) b* values appeared to be constant or
increasing instead of decreasing as the TBARS did.
binnenwerk.indd 108 21-08-2006 12:34:14

260 13
TBARS (nmol MDA equiv./g wet weight)

240 12
220 11
200 10
Instrumental Yellow (b*)

180 9
160 8
140 7
120 6
100 5
80 4
60 3
40 2
20 1
0 0
0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200
Sodium ascorbate (mg/kg)
Figure 1. Lipid oxidation in salt ripened cod. The effects of increasing ascorbate (0-2000 mg/kg)
concentrations during brining without copper added. TBARS (2-thiobarbituric acid reactive substances),
expressed as malondialdehyde (MDA) equivalents (®) and instrumental yellow colour values, b*, (l) were
measured at the end of the kench curing (21 days). Error bars represent mean standard deviations for pooled
muscle samples (N=8) made from 30-40 cod fillets.
Effects of including different antioxidants during brining

The effects of including three different antioxidants (EDTA, citrate and ascorbate) in the brine
on copper induced lipid oxidation were studied using the model system described earlier.
In these experiments, the copper concentration in the brines was 3 mg/kg Cu2+ while the
concentrations of the antioxidants were in the range of 100 to 193 mg/kg (0.5 mM). Muscle
cubes were sampled at the start (day 0), during the brining period (days 1-5) and at the end of
kench curing (day 21). Lipid oxidation was measured as TBARS values and instrumental yellow
colour values (b*) and the results are presented in Figures 3a and b, respectively. The presence
of antioxidants, inhibits copper induced oxidation during the brining period. However, at the
end of the brining period (day 5) the citrate samples showed TBARS values of 23 nmol MDA/g,
indicating enhanced copper-induced lipid oxidation.
The three antioxidants had clearly different effects on the TBARS of the salt ripened products
after 21 days of curing. Only EDTA inhibited copper induced lipid oxidation giving a TBARS
value of 15 compared to 111 nmol MDA/g in the samples brined with only copper added to the
brine. The TBARS values in the EDTA samples were actually in the same range as in samples not
oxidized with copper. Citrate on the other hand, accelerated copper induced lipid oxidation in
the cured muscle tissue producing TBARS values approximately 1.5 times higher than in the
samples brined with only Cu2+ added to the brine. Ascorbate appeared not to have any effect
on the copper induced lipid oxidation, giving TBARS values of 94 nmol MDA/g of the salt ripen
product.
binnenwerk.indd 109 21-08-2006 12:34:14

350 20

18
300
16

250 14
200 12
10
150 8
100 6
4
50
a 2
0 0
0 20 40 60 80 100 120 140 160 180 200 220
400 20
350 18
16

300
14
250 12
200 10
150 8
6
100
4
50 2
b
0 0
0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200
Sodium ascorbate (mg/kg)
Figure 2. Lipid oxidation in salt ripened cod. The effects of increasing ascorbate concentrations with 5 mg/kg
Cu2+ included in the brine. (a): 0-200 mg/kg ascorbate (N=4-8), (b): 200-2000 mg/kg ascorbate (N=8).
TBARS (2-thiobarbituric acid reactive substances), expressed as malondialdehyde (MDA) equivalents (®)
and instrumental yellow colour values, b*, (l) were measured at the end of the kench curing (21 days).
Error bars represent mean standard deviations for pooled muscle samples made from 20-40 cod fillets.
Instrumental yellow colour values (b*) of the minced muscle samples were also analysed as a
measure of copper-induced lipid oxidation in the cured muscle (Figure 3b). In general, the b*
values of the salt ripened products followed the TBARS values, except that of the ascorbate
samples which had b* values significantly (p<0.05) higher than the samples with only Cu2+
added in the brine, after 21 days of salting.
binnenwerk.indd 110 21-08-2006 12:34:15

200
TBARS (nmol MDA equiv. / g wet weight)

a
180
160
140
120
100
80
60
40
20
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
22
20 b
18
Instrumental yellow (b*)
16
14
12
10
8
6
4
2
0
-2 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
-4 Salting time (days)
Figure 3. The effects of different antioxidants in the brines on lipid oxidation measured as TBARS (2-
thiobarbituric acid reactive substances) (a) and instrumental yellow colour values (b) in cod muscle during
21 days of salting. Brines with 3 mg/kg Cu2+ added (l). Brines without Cu2+ added (r). Brines with 100
mg/kg ascorbate and 3 mg/kg Cu2+ added (u). Brines with 150 mg/kg citrate and 3 mg/kg Cu2+ added
(€). Brines with 193 mg/kg EDTA and 3 mg/kg Cu2+ added (n). Mean TBARS, expressed as malondialdehyde
(MDA) equivivalents, and mean instrumental colour values, b*, of pooled muscle samples (N=8) made from
30-40 cod fillets at each time of sampling.
To determine if the presence of antioxidants in the brine could affect the absorption of copper
ions from the brine and into the muscle, the concentration of the metal was determined in
the muscle samples during the brining process and in the salt ripened products (Figure 4). The
uptake of copper ions was significantly (p<0.05) affected by the different antioxidants. The
copper content of the EDTA samples increased just slightly during the brining period. During
binnenwerk.indd 111 21-08-2006 12:34:15

16
15
14
13
12
11
Copper (mg/kg)
10
9
8
7
6
5
4
3
2
1
0
0 0.5 1 2 3 4 5 21
Salting time (days)
Figure 4. The copper content of the cod muscle during salt curing with different antioxidants included in
the brine. Brines with 3 mg/kg Cu2+ added (l). Brines without Cu2+ added (r). Brines with 100 mg/kg
ascorbate and 3 mg/kg Cu2+ added (u). Brines with 150 mg/kg citrate and 3 mg/kg Cu2+ added (€). Brines
with 193 mg/kg EDTA and 3 mg/kg Cu2+ added (n). Mean copper content of pooled muscle samples (N=8)
made from 30-40 cod fillets at each time of sampling.
most of the process, it was lower than the content of the samples brined with only Cu2+ added.
The copper content of the EDTA samples approached its maximum of 5-6 mg/kg during the
first day of brining. On the other hand, the copper content of the citrate samples increased
more rapidly during the brining period and ended at a significantly (p<0.05) higher level than
the EDTA samples. The final copper concentration of the citrate samples was lower than the
concentration of the salt ripened samples brined with only copper added. The copper content
of the ascorbate samples increased less rapidly than in the citrate samples during the brining
but appeared significantly higher than in all other muscle samples after 21 days of curing. The
copper content of the control samples i.e. samples brined without any additives, decreased
slightly during the first day of brining and remained at this level during the rest of the curing
process.
In our previous study, it was indicated that copper induced lipid oxidation increased by a low
post rigor muscle-pH prior to heavy salt curing. The pH of the muscle was therefore determined
at start, during the brining and at the end of the curing process. It can be seen that the pH
of the EDTA samples remained at a significantly higher level than those of the citrate samples
during salt curing (Figure 5a). The muscle-pH decreased rapidly from 6.65 in fresh muscle to
values in the range of 6.1 to 6.2 during the first day of brining. Further on in the curing process,
the muscle-pH decreased just slightly (0.1-0.2). After 21 days of curing the range of the muscle-
pH values varied from 5.9 to 6.1, with the citrate samples at the lower level (5.93) and the EDTA
binnenwerk.indd 112 21-08-2006 12:34:16

6.8
a
6.7
6.6
6.5
Muscle pH 6.4
6.3
6.2
6.1
6.0
5.9
5.8
0 0.5 1 2 3 4 5 21
84 b
82
80
Moisture (%)
78
76
74
72
70
68
66
64
62
60
0 0.5 1 2 3 4 5 21
24
22
20
18
NaCl (%)
16
14
12
10
8
6
4 c
2
0
0 0.5 1 2 3 4 5 21
Salting time (days)
Figure 5. The pH (a), moisture content (b) and NaCl content (c) of the cod muscle during salt curing. Effects
of different antioxidants and copper in the brine. Brines with 3 mg/kg Cu2+ added (l). Brines without Cu2+
added (r). Brines with 100 mg/kg ascorbate and 3 mg/kg Cu2+ added (u). Brines with 150 mg/kg citrate
and 3 mg/kg Cu2+ added (€). Brines with 193 mg/kg EDTA and 3 mg/kg Cu2+ added (n). Mean muscle
pH, moisture and NaCl content of pooled muscle samples (N=8) made from 30-40 cod fillets at each time
of sampling.
samples at the upper level (6.13). Prior to salting, the pH of the brines were: control, 5.32;
Cu2+, 5.33; Cu2+ and citrate, 5.66; Cu2+ and ascorbate, 5.25 and Cu2+and EDTA, 5.50.
Conformational changes of the proteins, protein aggregation and losses of water are all
processes taking place during salt curing. The water content of the muscle samples was therefore
binnenwerk.indd 113 21-08-2006 12:34:17

determined initially, during the brining, and at the end of the curing process as shown in Figure
5b. The water content was reduced from 83% in fresh muscle to values in the range of 69 to
72% after one day of brining. During the remaining brining period, the water content of the
citrate samples decreased to a significantly (p<0.05) lower level than the EDTA and ascorbate
samples. After 21 days of curing, the three antioxidants had clearly different effects (p<0.05)
on the water contents of the muscle samples. The highest water content was found in the EDTA
samples (66.9%) and lowest value in the citrate samples (60.6%).
The NaCl concentration of the muscle samples was also determined initially, during the brining,
and at the end of the curing process as shown in Figure 5c. It can be seen that the NaCl content
increased rapidly from 0.15% in fresh muscle to concentrations in the range of 15 to 17% during
the first 12 hours of the brining. At the end of the brining process and after salt ripening,
the NaCl concentrations of the muscle were in the range of 17.6 to 18.2% and 18.6 to 22.7%,
respectively. The samples with lowest water content such as those brined with citrate, had the
highest NaCl content during the brining process. At the end of the curing, the samples brined
without additives had the highest NaCl concentration followed by the citrate samples.
Discussion
Effects of increasing ascorbate concentration in the brine

The model system simulating normal production of salted cod was used to investigate the
interaction effects of ascorbate and copper ions in the brine and the effects of ascorbate alone
on lipid oxidation in salt ripened products. The results clearly showed that asorbate used during
production of heavily salted cod might have pro-oxidative or antioxidative effects depending
on the ascorbate and metal concentrations. Without added copper in the brine, concentrations
of ≤ 500 mg/kg ascorbate had a prooxidative effect on the cured product. This pro-oxidative
effect is often explained by the ability of ascorbate to reduce endogenous transition metals.
Reduced transition metals are known to be particularly pro-oxidative (Haase and Dunkley 1969a
and b; Kanner and others. 1977; Frankel 1998). We suggest that at higher concentrations
(≥1000 mg/kg), ascorbate are able to reduce hydroperoxide radicals formed by the reduced
metals. In addition, it is known that ascorbate is able to chelate transition metals thereby
inactivating them (Khan and Martell 1967; Kanner and others. 1977). The concentrations we
found to be pro-oxidative and those found anti-oxidative, are in accordance with the findings
of Ramanathan and Das (1993) studying the effect of ascorbate on steam-cooked ground fish
and of Deng and others (1978) studying the effect of ascorbate on cold stored mullet.
With 5 mg/kg copper added in the brine, increasing concentration of ascorbate resulted in
different lipid oxidation levels. When no ascorbate was included in the brine, the salt ripened
product had, as expected, a high oxidation level measured both as TBARS and instrumental
yellow colour (b*) (Lauritzsen and others 1999). Under such pro-oxidative conditions the
presence of relatively low concentrations of ascorbate (≤50 mg/kg) reduced the formation of
malondialdehyd-bis(diethylacetal)-like substances in the cured product. The explanation for
this may be that ascorbate functions mainly as a radical scavenger. At higher concentrations
(100-200 mg/kg) the antioxidative properties are lost and pro-oxidative effects may even
be present (Figure 2a and b). One plausible explanation for this may be that the ascorbate
concentration must increase to a critical level before it reduces the transition metals. The
antioxidative properties of ascorbate at higher levels may be explained as described earlier.
These results appear similar to those obtained by Haase and Dunkley (1969a and b) and Kanner
binnenwerk.indd 114 21-08-2006 12:34:17

and others (1977) who studied oxidation in a potassiumlinoleate and in a β-carotene-linoleate

model system, respectively, and found pro-oxidative effects with low copper and high ascorbic
acid concentrations. As copper concentrations were raised in these investigations antioxidative
effects were found. On the basis of the present results for ascorbate concentrations 1000-2000
mg/kg, we may conclude that application of ascorbate as an antioxidant in salt curing of cod,
requires the use of high concentrations (≥1000 mg/kg) in the brine. However, the oxidation was
only measured at day 21 after salting and for new experiments it will be necessary to increase
the number of measurements.
Effects of including different antioxidants during brining

The effects of added EDTA, citrate and ascorbate in the brine on copper-induced lipid
oxidation were studied by the same model system. At concentrations used, EDTA was the only
effective antioxidant while citrate behaved as a pro-oxidant. EDTA is a chelator that forms
thermodynamically stable complexes with transition metal ions inhibiting electron transfer and
thus oxidation (Pribil 1972; Pokorný 1987). The spatial structure of the anion of EDTA, which
has six donor atoms, allows it to satisfy the coordination number of six frequently encountered
among metal ions. The effectiveness of EDTA in inhibiting copper-induced lipid oxidation has
been shown in several studies. Examples are inhibition of the development of off-flavour in
margarines during storage (Melniek 1961) inhibition of copper induced rancidity in cod muscle
blends at 12% NaCl concentrations (Castell and others 1965) and inhibition of metal-induced
lipid oxidation of fish oil enriched mayonnaise products (Jacobsen and others 2001).
Citric acid is a weaker chelating agent than EDTA. The lack of antioxidative effects, are probably
caused by a relatively low citrate concentration compared to the amount of copper. Reports
have suggested that a very high ratio of citric acid to copper is needed (Castell and others
1965; Flider and Orthoeffer 1981; Aubourg and others, 2004). It is also tempting to suggest
that the high NaCl concentration may have contributed to loss of chelating properties of
the antioxidant. The results in this experiment, using ascorbate at 100 mg/kg in the brine,
confirmed the high lipid oxidation found in salt ripened product in the experiments described
earlier. The lower TBARS level is probably caused by the use of 3 mg/kg copper in the second
experiments (Figure 3a and b) while 5 mg/kg was used in the first experiments (Figure 2a and
b). During most of the brining period (day 0-4) it was noticed that the TBARS values in all
samples treated with anti-oxidants were lower than in the samples brined with only copper.
It was surprising to find that citrate gave such low values initially particularly since it also
appeared to facilitate the absorption of copper into the muscle (Figure 4). In salt cured cod
we have previously shown that lipid oxidation, measured as TBARS correlated with the yellow
colour of the product (Lauritzsen and others 1999). In general, the results in Figures 3a and b
confirmed these results.
The results indicated that the uptake of copper into the muscle was influenced by the antioxidant
added during brining. EDTA appeared to retard the absorption of copper ions from the brine into
the muscle. The main reason for its effectiveness as an antioxidant is, however, suggested to
be due to its chelating properties. In addition to its ineffectiveness as an antioxidant, citrate
stimulated the uptake of copper into the muscle. The reason why citrate appeared to facilitate
the uptake of copper is not known. One can speculate that the binding of copper to citrate
neutralizes the positive charge of the metal and thereby makes it easier to penetrate into the
muscle (Goldstein and Czapski 1986; Kanner 1994). At the end of the curing the increased
binnenwerk.indd 115 21-08-2006 12:34:17

copper concentration observed in some of the samples could be explained by the loss of
moisture during kench curing.
Normally, salt curing induces conformation changes in the proteins and a reduction in the
muscle pH is observed (Lauritzsen and others 1999; Thorarinsdottir and others 2001). As
expected, the results with antioxidants included in the brine also showed that the muscle-pH
decreased by heavy salt curing. In our previous study, it was indicated that the copper induced
lipid oxidation increased by low pH of the fish muscle post rigor (Lauritzsen and others 1999).
In the present study we found that the citrate samples, the samples with the highest lipid
oxidation, had the lowest pH (Figure 5a). Recent reports have shown that pH affected lipid
oxidation of washed cod muscle when trout hemolysate was used as a catalyst (Richards and
Hultin 2000). The level of pro-oxidative deoxyhemoglobin was found to sharply increase with
pH reduction from 7.6 to 6.0 (Richards and others 2002). However, in heavy curing of cod this
mechanism is probably not relevant since the level of hemoglobin is very low in light muscle of
cod. The reduction in the pH of the EDTA samples appeared to occur more slowly than in other
samples. The binding of the EDTA to proteins may contribute to this (Hamm 1960).
The samples brined with citrate showed lower water contents than other samples (Figure 5b).
It has been reported earlier that the ultimate water content of cured meat muscle depends
mainly on the NaCl concentration and the pH of the brine (Hamm 1960). It is well known that
heavy salt curing causes denaturation of the muscle proteins due to “salting-out” effects (von
Hippel and Schleich 1969). Although the brine with citrate added had a slightly higher pH
value than the other brines, the pH of the cured muscle was among the lowest. The low water
content may be explained by increased “salting out” effects of the muscle proteins due to a
lower pH. Conversely, it is also known that enhanced interactions between muscle proteins
and lipid oxidation end products, lead to protein denaturation and subsequently dehydration
(Mackie 1993; Howell 1995; Li and King 1996; Badii and Howell 2002).
The lowered water content of the citrate samples was thought to be explained by an increased
uptake of NaCl. The NaCl content was therefore monitored during the experiments (Figure 5c).
As expected, the salt content changed inversely with the water content due to large osmotic
forces. A slightly higher NaCl content in the citrate samples during the brining period compared
to the other samples was observed. However, this may not explain all the existing differences in
the water content and lipid oxidation level between the samples brined with anti-oxidants.
Conclusions
The results from this work demonstrated that EDTA was the only antioxidant/chelator that
efficiently inhibited copper-induced lipid oxidation of the final cured product. Further
investigations are required on how the use of antioxidants during heavy curing of cod can
minimize problems associated with lipid oxidation and discolouration.
Acknowledgements
This work was financed by the Norwegian research Council.
binnenwerk.indd 116 21-08-2006 12:34:17

References
AOAC. 1990. Moisture in Meat. 1990. Official Methods of Analysis 950.46, 11:931.
AOAC. 1990. Salt (Chlorine as Sodium chloride) in Seafood. Official Methods of Analysis 937.09, 11:870.
Aubourg SP, Pérez-Alonso F, Gallardo JM. 2004. Studies on rancidity inhibition in frozen horse mackerel (Trachurus
trachurus) by citric ascorbic acid. Eur J Lipid Sci Technol 106:232-240
Badii F, Howell NK. 2002. Effect of antioxidants, citrate and cryoprotectants on protein denaturation and texture of
frozen cod (Gadus morhua). J Agric Food Chem. 50:2053-2061.
Castell CH, Maclean J, Moore B. 1965. Rancidity in lean fish muscle. IV. Effect of sodium chloride and other salts. J
Fish Res Bd Canada 22:929-944.
Deng JC, Watson M, Bates RP, Schroeder E. 1978. Ascorbic acid as an antioxidant in fish flesh and its degradation. J
Food Sci 43:457-460.
Dulavik B, Sørensen NK, Barstad H, Horvli O, Olsen RL. 1998. Oxidative stability of frozen light and dark muscles of
saithe (Pollachius virens L.). J Food Lipids 5:233-245.
Flider FJ, Orthoeffer FT. 1981. Metals in soybean oil. J Am Oil Chem Soc 58:270-272.
Frankel, EN. 1998. Lipid oxidation. Dundee, Scotland: The Oily Press Ltd., 303 pages.
Goldstein S, Czapski G. 1986. The role and mechanism of metal ions and their complexes in enhancing damage in
biological systems or in protecting these systems from the toxicity of O2-. Free Rad. Biol. and Med 2:3-11.
Haase G, Dunkley WL. 1969a. Ascorbic acid and copper in linoleate oxidation. 2. Ascorbic acid and copper as oxidation
catalysts. J Lipid Res 10:561-567.
Haase G, Dunkley WL. 1969b. Ascorbic acid and copper in linoleate oxidation. 3. Catalysts in combination. J Lipid Res
10:568-576.
Hamm R. 1960. Biochemistry of meat hydration. Adv Food Res 10:355-463.
Harel S. 1994. Oxidation of ascorbic acid and metal ions as affected by NaCl. J Agric Food Chem 42:2402-2406.
Howell N. 1995. Interaction of proteins with small molecules. In: Gaoucar A, editor. Ingredient interactions-effects on
food quality New York:Marcel Dekker. p 269-289.
Hølmer, G. 1995. Antioxidative systems, with special regard to membranes. In: Jessen F, editor. Fish quality-role
of biological membranes, Proceeding from the meeting 23rd and 24th of March at Hillerød. Danish Institute for
Fisheries Research. TemaNord; 1995:624, 201pages.
Jacobsen C, Adler-Nissen J, Meyer A S. 1999. Effect of ascorbic acid on iron release from the emulsifier interface and
on the oxidative flavour deterioration in fish oil enriched mayonnaise. J Agric Food Chem. 47:4917-4926.
Jacobsen C, Timm M, Meyer A S. 2001. Oxidation in fish oil enriched mayonnaise: ascorbic acid and low pH increase
oxidative deterioration. J Agric Food Chem 49:3947-3956.
Kanner J, Mendel H, Budowski P. 1977. Prooxidant and antioxidant effects of ascorbic acid and metal salts in a β-
carotene-linoleate model system. J Food Sci 42:60-64.
Kanner J. 1994. Oxidative processes in meat and meat products: Quality implications. Meat Sci. 36:169-189.
Khan TMM, Martell AE. 1967. Metal ion and metal chelate catalysed oxidation of ascorbic acid by molecular oxygen. 1.
cupric and ferric ion catalysed oxidation. J Am Chem Soc. 89:4176.
Kläui H, Pongracz G. 1981. Ascorbic acid and its derivates as antioxidants in oils and fats. In: Consell JN and Horning
DH, editors. Vitamin C, Ascorbic Acid. London: Applied Science Publishers. p 139-166.
Kubow S. 1992. Routes of formation and toxic consequences of lipid oxidation products in foods. Free Rad Biol Med
12:63-81.
Lauritzsen K, Martinsen G, Olsen RL. 1999. Copper induced lipid oxidation during salting of cod (Gadus morhua L). J
Food Lipids 6:299-315.
Li SJ, King AJ. 1996. Lipid oxidation and myosin denaturation in dark chicken meat. J Agric Food Chem 44:3080-
3084.
Lin CC, Liang JH. 2002. Effect of antioxidants on the oxidative stability of chicken breast meat in a dispersion system.
J Food Chem Toxicol. 67(2):530-533.
binnenwerk.indd 117 21-08-2006 12:34:17

Löliger J. 1991. The use of antioxidants in foods. In: Arouma OL and Halliwell B, editors. Free Radicals and Food
Additives. London: Taylor and Francis. p. 121-150.
Mackie IM. 1993. The effects of freezing on flesh proteins. Food Rev Int, 9:575-610.
McBrien DCH, Slater TF. 1982. Lipid peroxidation and activated intermediates. In: McBrien DCH and Slater TF. Free
Radicals, Lipid Peroxidation and Cancer. London: Academic Press, XII, 447 p.
Melniek D. 1961. Flavor stabilized salted margarine and process of producing the same. Corn Products Company, a
corporation of Delware. US Patent 2, 983:615.
Niki E. 1991. Vitamin C as an antioxidant. In: Simopoulos AP, editor. Selected vitamins, minerals and functional
consequences of maternal malnutrition. Basel, Switzerland: World Rev Nutr. Diet 4:1-3.
Pokorny J. 1987. Major factors affecting the autoxidation of lipids. In: Chan HW-S, editor. Autooxidation of unsaturated
lipids.London:Academic Press Inc. p 141-206
Pribil R. 1972. Analytical applications of EDTA and related compounds. Pergamon Press, Oxford, UK.
Ramanathan L and Das NP. 1993. Effect of natural copper chelating compounds on the pro-oxidant activity of ascorbic
acid in steam-cooked ground fish. J Food Sci Technol 28:279-288.
Richards MP and Hultin HO. 2000. Effect of pH on lipid oxidation using trout hemolysate as a catalyst: a possible role
for deoxyhemoglobin. J Agric Food Chem 48:3141-3147.
Richards MP, Østdal H, Andersen HJ. 2002. Deoxyhemoglobin-mediated lipid oxidation in washed fish muscle. J Agric
Food Chem 50:1278-1283.
Simpson GR and Blay RA. 1966. Rapid method for determination of the metals copper, zink, tin, iron and calcium in
food stuffs by atomic absorption spectroscopy. Food Trade Rev. August:35-37.
Thorarinsdottir KA, Arason S, Bogason SG, Kristbergsson K. 2001. Effects of phosphate on yield, quality, and water-
holding capacity in the processing of salted cod (Gadus morhua). J Food Sci 66:821-826.
von Hippel PH and Schleich T. 1969. The effects of neutral salts on the structure and conformational stability of
macromolecules in solution. In: Timasheff SN and Fasman GD, editors. Structure and stability of biological
macromolecules, New York: Marcel Dekker. p 417-574.
Witte VC, Krause GT, Bailey ME. 1970. A new extraction method for determining 2-thiobarbituric acid values for pork
and beef during storage. J Food Sci 35:582-585.
binnenwerk.indd 118 21-08-2006 12:34:18

Rapid assessment of storage quality of cliff-fish from saithe by
fluorescence spectroscopy
Agnar Sivertsen1, Kristin Lauritzsen1, Annette Veberg2,3 and Jens Petter Wold2
2MATFORSK AS, Osloveien 1, N-1430 Ås, Norway
3Norwegian University of Life Sciences, PO Box 5040, N-1432 Ås, Norway
Abstract
Minced samples of cliff-fish from saithe during cold storage at 4 ºC have been analyzed with
fluorescence spectroscopy, and the amount of thiobarbituric acid reactive substances (TBARS)
and total amount of volatile nitrogen (TVN) have been measured. Partial least square regression
models have been built between the fluorescence spectra and the chemical measured TBARS and
TVN values. Both models were evaluated using full cross validation and test set validation. The
correlation between the fluorescence spectra and TVN was 91% and the correlation between
the fluorescence spectra and TBARS was 79%, both using test set validation. The work has
showed that fluorescence spectroscopy can be used as a rapid and non-destructive method for
measuring TBARS and TVN in minced samples of cliff-fish from saithe with a high degree of
accuracy.
Keywords: fluorescence spectroscopy, lipid oxidation, cliff-fish, saithe, discoloration, non-

destructive methods
Introduction
Cliff-fish is an old and traditional product, which has been and still is popular due to its good
preservation properties and characteristic taste. The Norwegian export of cliff-fish from saithe
had a value of 660 million NOK in 2004. The main part of this is shipped to Brazil and the
Caribbean islands in cold storage containers. The transport can take up to 30 days, and yellow,
brown and red discolouration of the cliff-fish during transportation has frequently been observed.
This degradation often leads to customer complaints and economical losses to the producers.
Earlier studies on salted cod have shown a strong correlation between yellow-brown discoloration
and non-enzymatic lipid oxidation (Lauritzsen and others 1999). It is well known that protein
degradation and lipid oxidation give rise to a large number of volatile components (Toldra
and others 1998). It is likely that the protein degradation taking place in cliff-fish products
during long time storage, causes an increase in the volatile nitrogen fraction. Analyses of
volatile organic molecules (Headspace GC-MS analyses) have therefore been applied for quality
evaluation of heated fish powders (Mjøs and Solvang 2006). Schiff-bases polymerize by aldol
condensation producing dimers and complex high-molecular weight brown macromolecules
known as melanoidins, that are not well characterised. Schiff base compounds formed by
the interaction of oxidation products with proteins, phospholipids and nucleic acids, produce
chromophores showing characteristic fluorescence spectra (Veberg 2005; Frankel 1998).
binnenwerk.indd 119 21-08-2006 12:34:18

A rapid and non-destructive method for measuring the lipid oxidation, before the discoloration
is visible, would be a valuable tool for the cliff-fish industry. Such a tool could be used as a
method for monitoring the lipid oxidation during production and storage, or it could also be
used as an automatic grading criterion for whether the cliff-fish is suitable for long distance
transportation or should be sold at the local markets. Fluorescence spectroscopy has shown
promising results as a rapid and non-destructive method for measuring lipid oxidation in various
food products, such as sour crème, cheese and turkey meat (Wold and others 2002; Wold 2000).
Correlation of 87% between the fluorescence spectra and the amount of thiobarbituric acid
reactive substances (TBARS) in minced chicken meat has been reported (Wold and Mielnik
1999). An excitation wavelength of 380 nm, has given very informative spectra regarding lipid
oxidation in minced poultry meat (Wold 2000). Light at 380 nm was used as an excitation
source in this work, but there might be other excitation wavelengths that would be better
suited for assessing lipid oxidation in cliff-fish from saithe.
The aim of this study was to investigate fluorescence spectroscopy as a rapid and non-
destructive tool for measuring the storage quality of cliff-fish from saithe. Lipid oxidation has
been measured using fluorescence spectroscopy and the amount of TBARS. The total amount
of volatile nitrogen (TVN) was also determined. Partial least squares (PLS) regression models
were built between the fluorescence spectra and the chemical measured TBARS and TVN. Both
models were evaluated using full cross validation and test set validation.
The fish used in this work was saithe (Pollachius virens) caught by hand line in close proximity
to Tromsø in February and April 2004. The species were divided in 16 experimental groups with
10 fish per group and filleted by hand. The picklesalting was done in plastic containers with
dry salt covering the bottom of the containers and between the layers of fillets. 1.5 kg salt
per kg fillet was used during picklesalting, which lasted for five days. The fish was then dry
salted for a total of 20 days were the fillets was taken out and the salt was replaced with fresh
salt after 10 days. The salted fillets were then dried in a commercial cliff-fish dryer at Tromvik
Fiskeindustri AS 50 km west of Tromsø.
Storage experiment and sampling

All samples used in this work are from a 24 full factorial design, where the variables light,
oxygen availability, moisture content and salting temperature were varied during production.
This resulted in 16 different batches of cliff-fish, and the combinations of production and
storage parameters are shown in Table 1. After production the batches were stored at 4 ºC for
8 months in commercial cliff-fish packages. The fillets marked with 10% oxygen during storage
were vacuum packed with 90% vacuum in plastic bags. The 16 batches were stored for 1, 4 and 8
months at 4 ºC, which resulted in a total of 48 experiments to analyse. A sample containing five
random fillets from each experiment were taken out, and 0.5 cm slices were cut for every 3-4
cm along the fillet. The first 3 cm of the neck region was not used. All the slices from the five
fillets were homogenized, into one pooled sample, using a Dito Sama homogeniser (Aubusson,
France) at lowest speed for 0.5-1 minute. The homogenizing was done at room temperature. The
minced samples were stored in plastic zip lock bags for 1-2 weeks at -80 ºC before the chemical
and spectroscopic measurements were performed.
binnenwerk.indd 120 21-08-2006 12:34:18

Rapid assessment of storage quality of cliff-fish from saithe by fluorescence spectroscopy
Table 1. Experimental setup for the 24 full factor storage experiment.
Experiment nr. Light during Oxygen during Salting Moisture content

pickle salting storage (%) temperature (˚C)
E1 Yes 100 2-4 <50%

E2 Yes 100 2-4 <47%
E3 Yes 100 18-20 <50%
E4 Yes 100 18-20 <47%
E5 No 100 2-4 <50%
E6 No 100 2-4 <47%
E7 No 100 18-20 <50%
E8 No 100 18-20 <47%
E9 No 10 2-4 <50%
E 10 No 10 2-4 <47%
E 11 No 10 18-20 <50%
E 12 No 10 18-20 <47%
E 13 Yes 10 2-4 <50%
E 14 Yes 10 2-4 <47%
E 15 Yes 10 18-20 <50%
E 16 Yes 10 18-20 <47%
Thiobarbituric acid reactive substances (TBARS)

The TBARS were measured spectrophotometrically using the method by Dulavik and others
(1998). Lipid phase was extracted from the cliff-fish into trichloroacetic acid. Thiobarbituric
acid was added the extract, forming a pink colour complex with the oxidised components, and it
was quantified measuring the absorbance at 532 nm. The unit used is nanomole malondialdehyde
per gram muscle (nmol MDA/g). Two replicates were extracted for each homogenized sample,
and the TBARS value was measured twice for each extract. Average TBARS value and standard
deviation were calculated based on these four measurements for each of the 48 experiments.
Total amount of volatile nitrogen (TVN)

TVN was measured using direct distillation by Kjeltech 1026 and 1035 (Tecator instruments,
Stockholm, Sweden). The unit used is milligram of nitrogen per 100 grams of muscle (mg
N/100g). Three replicates were measured per homogenized sample, and the average TVN value
and standard deviations were calculated based on these three measurements.
Fluorescence spectroscopy
Measurements of fluorescence were performed on all the 48 samples with the optical system
shown in Figure 1. The excitation light source was generated by a 300 W xenon light source
(Oriel 6258, Oriel Corporation, Stratford, CT) mounted in a lamp housing (Oriel 66028) and
a 380 nm band pass filter with a bandwidth of 10 nm and diameter of 50 mm. The light was
directed onto the sample at an angle of 45º. The samples were placed into specially designed
sample cuvettes with a diameter 50 mm and depth of 10 mm exposing a flat circular surface
for measurements. A sharp 400 nm long pass filter was placed in front of the spectrometer to
prevent reflected excitation light from overlapping the autofluorescence light. The CCD detector
binnenwerk.indd 121 21-08-2006 12:34:18

A
E
D B
Figure 1. Optical system for fluorescence spectroscopy. A: Light source, B: band pass filter, C: sample cuvette,
D: long pass filter, E: spectrometer.
(Princeton TEA/CCD-512-TKBM1, Princeton Instruments Inc., Trenton, NJ) was mounted on a

spectrograph (Acton SP-150, Acton Research Group., Acton, MA). The entrance slit used in
the spectrograph was 0.3 mm wide, and the grating used had 300 grooves/mm with a grating
blaze at 300 nm. An exposure time of two seconds were used per sample. The final spectrum is
given in raw counts, i.e. the number of photons registered, and results from first subtracting
the dark current from the CCD and then taking the mean value over 300 horizontal lines on the
CCD. Each spectrum has a resolution of 5 nm in the region 410 to 750 nm. The black painted
laboratory was kept at 18 ºC and had a minimum of stray light, and the temperature of the
samples was approximately 18 ºC.
The fluorescence spectra from each sample were used to predict the chemical measured reference
measurements TBARS and TVN values. PLS regression (Martens and Næs 1991) was used to
generate calibration models for both TBARS and TVN. The models were evaluated using full cross
validation (leave one out) and test set validation. For test set validation the experiments E1-E5
in Table 1 were used, stored for 1, 4 and 8 months, corresponding to 15 samples, to build the
calibration model. The models were then tested on the rest of the 11 samples. All analysis was
performed using version 9.12 of the Unscrambler (CAMO AS, Norway).
Fluorescence measurements
Fluorescence spectra from sample E1 stored for 1, 4 and 8 months are shown in Figure 2. The
intensity of the fluorescence spectra is increasing with storage time, and some part of the
spectra, especially the peak at 468 nm, increases more than the others. This trend was observed
for all the samples. The emission at 468 nm is believed to be due to the Schiff base originated
from oxidized phospholipids and amino components (Frankel 1998). The intensity at 468 nm
for all the samples varied between 7824 and 24120 CCD counts.
Chemical measurements
The TVN values for all the 48 samples ranged from 11.9 – 41.0 mg N/100g, and the values
with standard deviation are shown in Figure 3. The samples had a steady increase in TVN
binnenwerk.indd 122 21-08-2006 12:34:19

20.000
1 month
4 months
8 months
CCD counts
10.000
0
450 500 550 600 650 700 750
Wavelength (nm)
Figure 2. Fluorescence spectra of sample E1 for 1, 4, and 8 months of storage.
50
0 month 1 month 4 months 8 months
45
40
TVN (mg N/100 g)
35
30
25
20
15
10
5
0
E1 E2 E3 E4 E5 E6 E7 E8 E9 E10 E11 E12 E13 E14 E15 E16
Figure 3. TVN content with standard deviation as error bars for the 16 different experiments at 0, 1, 4 and
8 months of storage. Refer to Table 1 for explanation of legends.
as a function of storage time for all the experiments except of two (E8 and E9), and the
mean standard deviation was 1.8. Preliminary studies using headspace GC-MS analyses have
shown an increase in the diamide and strecker-aldehyde fractions and a reduction in the
sulphur components with increasing discolouration of cliff-fish from saithe (Mjøs and others,
unpublished data). The degradation in the protein fraction of cliff-fish muscle and the increase
in the diamide concentration are believed to explain the increase in TVN.
binnenwerk.indd 123 21-08-2006 12:34:20

The TBARS values for the same samples ranged from 19.6 – 90.1 nmol MDA/g, and the mean
values with standard deviation are shown in Figure 4. The samples had highest TBARS values after
four months of storage for all the experiments except one (E4), and the mean standard deviation
was 8.1. The experiment E7 stored for eight months was corrupted due to an unpredictable error
during storage, and was therefore removed from the dataset.
Partial least square regression

Partial least square regression models were built between the fluorescence spectra and both
TBARS and TVN, and the results are summarized in Table 2. The models were evaluated using
both full cross validation (FCV) and test set validation (TSV). The best model for predicting
TVN from the fluorescence spectra was obtained using three principal components and the
correlation was found to be 0.91 for both the FCV and the TSV model. The root mean square
error prediction (RMSEP) was found to be 2.8 when using FCV and 3.3 using TSV.
The best model for predicting TBARS was found using four principal components, and the
correlation was 0.82 and 0.79 for FCV and TSV respectively. It is clear that the relationship
between the fluorescence spectra and TBARS was non-linear, since the fluorescence spectra had
a steady increase over the whole storage period, while the TBARS values declined after four
months of storage. PLS is not suited to model non-linear relationships. However, it has earlier
140
0 month 1 month 4 months 8 months
120
TBARS (nmol MDA/g)
100
80
60
40
20
0
E1 E2 E3 E4 E5 E6 E7 E8 E9 E10 E11 E12 E13 E14 E15 E16
Figure 4. TBARS values with standard deviation as error bars for the 16 different experiments at 0, 1, 4 and
8 months of storage. Refer to Table 1 for explanation of legends.
Table 2. PLS regression between fluorescence spectra and chemical measured parameters. All the models
are validated using full cross validation (FCV) and test set validation (TSV). RMSEP is the root mean square
prediction error of the PLS model. The mean value is calculated from the chemical measured parameters.
Fluorescence vs. Correlation (FCV/TSV) RMSEP (FCV/TSV) Mean Value (N=48)
TVN (3 PCs) r = 0.91/0.91 RMSEP = 2.8/3.5 24.8

TBARS (4 PCs) r = 0.82/0.79 RMSEP = 9.0/10.8 47.2
Log(TBARS) (4 PCs) r = 0.84/0.78 RMSEP = 8.5/10.1 47.2
binnenwerk.indd 124 21-08-2006 12:34:20

been shown that by transforming the TBARS values using the logarithm, i.e. Log (TBARS),
some of the non-linearity can be suppressed and therefore improve the PLS model (Wold and
Mielnik 1999). A model between the fluorescence spectra and the logarithm of TBARS was built
both using full cross validation and test set validation. An inverse logarithmic transform of the
output of these models were performed and compared with the models for predicting TBARS
directly. This improved the RMSEP with approximately 6%. Despite the models predicting the
TBARS value had a lower correlation than the TVN models, the RMSEP was only 5% larger then
the mean standard deviation of the chemical measured TBARS.
Fluorescence spectroscopy showed to be a promising method for rapid assessment of both the
TVN and TBARS values in minced samples from cliff-fish of saithe. This can be done with a
high degree of accuracy, cheaper and much faster than the traditional chemical method. The
non-linear relationship between the fluorescence spectra and TBARS values makes it difficult to
use PLS as a prediction method. This can be improved by using a better transformation of the
TBARS values or by using other regression methods, which are more suited for finding non-linear
relations, i.e. neural networks and support vector machines (Haykin 1999). It is likely that
fluorescence spectroscopy would perform better when predicting the TBARS values in an earlier
stage of the oxidation process, and earlier work showed this for various other food products
(Wold 2000). This can be verified with similar experiments as used in this work, but were the
cliff-fish is sampled earlier in the oxidation process.
Earlier results have shown that lipid oxidation and discoloration are strongly linked (Lauritzsen
and others 1999). A method for measuring the lipid oxidation rapid, non-destructive and at
an early stage, before it is visible, could provide the industry with a valuable tool for online
grading of cliff-fish with respect to rest storage time before discoloration. It is likely that
fluorescence spectroscopy also would perform well for predicting the TBARS and TVN values at
the surface of an intact cliff-fish.
Conclusions
Both lipid oxidation and the amount of total volatile nitrogen are important quality factors.
This work has shown that fluorescence spectroscopy can be used as a rapid and non-destructive
tool to measure the storage quality of minced cliff-fish from saithe.
References
Dulavik B, Sørensen N, K, Barstad H, Horvli O, Olsen RL. 1998. Oxidative stability of frozen light and dark muscles of
saithe (Pollachius virens L.). J Food Lipids 5:233-245.
Frankel EN. 1998. Lipid Oxidation. Dundee: The Oily Press. 303 p
Haykin S. 1999. Neural netwoks, a comprehensive foundation. 2nd ed. New Jersey: Prentice Hall inc.
Lauritzsen K, Martinsen G, Olsen RL. 1999. Copper induced lipid oxidation during salting of cod (Gadus morhua L.). J
Food Lipids 6:299-315.
Martens H, Næs, T. 1991. Multivariate Calibration. Chichester, UK: John Wiley and Sons, 438 p
Mjøs S, Solvang M. 2006. Patterns in volatile components over heated fish powders. Food Res Internat 39:190-202.
Toldra F, Flores M. 1998. The role of muscle proteases and lipases in flavour development during the processing of
dry-cured ham. Crit Rev Food Sci 38:331-352.
Veberg A, Vogt G, Wold, JP. 2005. Fluorescence in aldehyde model systems related to lipid oxidation. LWT - Food Sci
Technol 39(5):562-570
binnenwerk.indd 125 21-08-2006 12:34:21

Wold JP, Jørgensen K, Lundby F. 2002. Non-destructive measurement of light-induced oxidation in dairy products by
fluorescence spectroscopy and imaging. J Dairy Sci 85:1693-1704.
Wold JP. 2000. Rapid quality assessment of meat and fish by using near-infrared spectroscopy, autofluorescence
spectroscopy and image analysis. Ph.D thesis from the Agricultural University of Norway. ISBN 82-575-0413-0.
Wold JP, Mielnik M. 1999. Non-destructive assessment of lipid oxidation in minced poultry meat by autofluorescence
spectroscopy. J Food Sci 63(3):377-383.
binnenwerk.indd 126 21-08-2006 12:34:21

Natural antioxidants in cod liver oil: Pitfalls during oxidative
stability assessment
Eva Falch1, Anders Øverby2 and Turid Rustad2

1SINTEF Fisheries and Aquaculture, Trondheim, Norway
2The Norwegian University of Science and Technology, Department of Biotechnology, Trondheim,
Norway
Abstract
Tocopherols are natural antioxidants present in cod liver oil and are commonly added for
further improvement of the oxidative stability. The influence of a selection of antioxidants
on the oxidative stability of cod liver oil is investigated during oxidation at 40 °C. Rosemary
extract gave the highest oxidative stability index, but the highest amount of radicals trapped
by ESR spectroscopy. Low levels of trapped spin adducts were generally not correlating with
low oxidation levels obtained by conventional methods. Moreover, this study demonstrates
the antioxidative effect of the spin trap and the persistence of the spin adducts in samples
containing different antioxidants. These findings demonstrate the importance of using oxidation
stability tests with precautions when studying complex marine lipids.
Keywords: lipid oxidation, ESR, spin trapping, PBN, cod liver oil
Introduction
The complex nature of marine lipids requires advanced analytical techniques for compositional
analysis and for assessment of degradation products affecting their sensory attributes and
nutritional value. Lipid oxidation is the main factor limiting the shelf life of marine oils due to
formation of reaction products with unpleasant organoleptic properties. Marine lipids contain
high levels of long chain polyunsaturated fatty acids (PUFA) such as eicosapentaenoic acid
(EPA) and docosahexaenoic acid (DHA). The latter fatty acid is reported to oxidize five times
faster than linoleic acid (18:2), a principal fatty acid in vegetable oil (Cosgrove and others
1987). Among different methods to prevent lipid oxidation, the addition of antioxidants is
preferentially considered and applied by omega-3 producers and in food formulations. Recently,
interest in natural antioxidants has increased due to the concern about the long-term safety of
synthetic antioxidants. However, the synthetic antioxidants are still the most commonly used
antioxidants in food systems because of their chemical stability, low cost and availability (Yang
and others 2002). Natural antioxidative agents such as tocopherols and vitamin A are widely
used by fish oil producers and some commercial antioxidant products based on plant extracts
from rosemary, tea catechin, or other antioxidant mixtures are used occasionally. Some of these
antioxidants, rosemary in particular, are tested and reported to be effective in oil systems
(Nogala-Kalucka and others 2005; Frankel and others 1996a,b). However, the effect is highly
influenced by the different chemical constituents in oil.
Auto-oxidation proceeds through a chain propagation reaction with free radicals formed during
the initial stages. A wide variety of methods are available for assessment of lipid oxidation
binnenwerk.indd 127 21-08-2006 12:34:21

and oxidative stability of bulk oil, but they all have their drawbacks. Most traditional methods
available are based on determination of the labile hydroperoxides combined with measurement
of secondary reaction products from lipid oxidation (primarily aldehydes), however methods for
early assessment of lipid oxidation are needed (Falch and Aursand 2005). Effort has therefore
been put on development of more advanced methods for analysis of lipid oxidation such as
free radical assessment by electron spin resonance (ESR) spectroscopy. ESR has recently been
applied as an indicator of early lipid oxidation events in different food systems (Thomsen and
others 1999, 2000; Andersen and Skibsted 2002; Andersen and others 2005; Jensen and others
2005). However, the lipid-derived radicals are so reactive that their steady state concentrations
fall below the ESR detection limit, which has been reported to be 10-9-10-8 M under optimal
conditions (Schaich and Borg 1990; Andersen and Skibsted 2002). The spin trapping of free
radicals solves this problem by forming ESR detectable spin adducts. PBN have previously been
reported to be a valuable spin trap for bulk oil systems (Thomsen and others 2000; Velasco and
others 2004; Falch 2005; Falch and Aursand 2005; Falch and others 2005). This spin trap has
previously been used successfully in the study of radical development in vegetable oil systems
(Thomsen and others 2000; Velasco and others 2004) and in a recent study, also reported to
be successful in demonstrating the oxidative stability directly in salmon oil and cod liver oil
(Falch and others 2005).
Our previous investigations has demonstrated that the persistence of spin adducts are highly
variable in different marine oils (Falch and others 2005). In oxidation stability studies,
ESR therefore requires highly standardised assays to prevent the wrong conclusions to be
drawn. Interactions between spin trapping agents and specific compounds in the oil (radical
scavengers) might affect the spin trapping of lipid derived radicals. In this study, different
natural antioxidative agents are therefore evaluated for their influence on the spin trapping in
general, and on the persistence of spin adducts in particular, in crude cod liver oil (CLO). Since
there is a need for optimal methods for assessment of antioxidant effect and stability of marine
oils conventional analysis are also performed and evaluated.
Materials
Cod liver from partially farmed cod (Helgeland, Norway) from November was the basis for the
preparation of cod liver oil. The following antioxidative agents were used: (1) Tea catechin
powders, an extract of green tea (86% w/w purity) were supplied by Kinglong Natural Plant
Products Industry Ltd. (Changsha, Hunan, China), (2) Rosemary extracts as powder (approx.
30% w/w purity) were supplied by Guiness Chemical Ltd. (Portlaoise, Co. Laois, Ireland) (3)
liquid γ-tocopherol and (4) liquid α–tocopherol were supplied by GO-Johnson (Oslo, Norway)
and (5) liquid mixture of 90% tocopherols (Mixed tocopherols) (18% α-, 6% β-, 54% γ-, and
12% δ-tocopherol) were supplied by Carl Rasmussen AS (Bergen, Norway). All chemicals used
for analysis of lipid oxidation were ‘AnalaR’ grade obtained from Sigma- Aldrich Co. Ltd. (Dorest,
UK, Steinheim, Germany). The spin trap: α- phenyl-tert-butylnitrone (PBN) (purity 98%) was
also obtained from Sigma-Aldrich.
Sample material
The cod (Gadus morhua) were slaughtered, bled and transported on ice to the laboratory. Within
30 hours after slaughtering, the liver from 30 individuals were minced and frozen at -80 °C. The
livers were thawed and cod liver oil was released by heating in a microwave at 750W to a core
binnenwerk.indd 128 21-08-2006 12:34:21

Natural antioxidants in cod liver oil: Pitfalls during oxidative stability assessment
temperature of 95 °C, and then separated by centrifugation (2250g for 15 min). The oil was
frozen in small tubes, which were flushed with nitrogen and kept at -80 °C until analysis.
Experimental storage of CLO

The combinations of antioxidants and spin traps are presented in Table 1. The samples were
prepared in 30 g portions enabling analysis of oil stability and radical assessment by ESR. Parts
of this sample were transferred in duplicates into ESR tubes (quarts 0.5 mm, Wilmad Glass
Buena, NJ), capped and immediately stored in the dark at 40 °C. Samples were withdrawn after
120 hours (5 days) for analysis of primary peroxide value (Undeland and others 1998), and
conjugated diene hydroperoxides (AOCS) and secondary (thiobarbituric-acid-reactive substances
(TBARS) (Ke and Woyewoda 1979). All methods are based on spectroscopic absorbance. Oil
stability was determined in the oil before the experimental storage by using the Oil Stability
Index (OSI) using an Oil stability Instrument (Oil Stability Instrument, Omnion, USA) (Jebe
and others 1993) at 60 and 80 °C. This is a common method for determination of antioxidative
effect in oil systems.
Oil stability of natural antioxidants at different concentrations

Oil stability Index at 60 °C were analysed on CLO containing 0.1, 0.25 and 0.5% (w/w) of
Rosemary extract, Green tea catechin, α–tocopherol, γ-tocopherol and mixed tocopherols (α-
/β-/γ-/δ-tocopherol). The antioxidants were added directly to the CLO (no solvents) before the
OSI measurements. In order to eliminate the effect of potential differences in the oxidative
status of the CLO (control) used, the efficiency of these antioxidants was calculated on a
relative basis compared to the control sample (level 1) containing no antioxidant. The levels
are therefore the presented as the ratio of the OSI time of the sample with and without
antioxidant added. Levels above 1 show an increased stability compared to the control. The
OSI measurements were performed in triplicates.
Table 1. Experimental plan showing the addition of antioxidants and spin trap (PBN/α-phenyl-tert-
butylnitrone). Those samples that were analysed by conventional methods for determining lipid oxidation
are specified. PBN was added at a concentration of 1 mg/g oil and free radical assessment was performed
on those samples.
Samples Antioxidant (% w/w) Conventional methods Radical assessment
Control No X
Control / PBN No X X
γ-toc 0.5 X
γ-toc / PBN 0.5 X X
Rosemary 0.5 X
Rosemary / PBN 0.5 X X
α–toc 0.5 X
α–toc / PBN 0.5 X X
α–toc / PBN 0.25 X
α–toc / PBN 0.10 X
Tea catechin / PBN 0.5 X
binnenwerk.indd 129 21-08-2006 12:34:21

Chemical characterization of CLO

Fatty acids were analysed by gas chromatography after methylation of the fatty acids as
described in Falch and others (2006). Lipid classes (polar lipids, triacylglycerols, free fatty
acids, sterols and steryl esters) were separated by thin layer chromatography using an Iatroscan
device (Iatron Laboratories, Tokyo, Japan) and a polar solvent system as described in Falch and
others (2006). Pro-oxidative metals (Fe and Cu) were analysed by high resolution inductively
coupled plasma mass spectroscopy (ICP-MS) according to the method described by Falch and
others (2005).
Free radical assessment

The antioxidants presented in Table 1 were added to CLO with 1 mg PBN/g. The tocopherols were
added as a viscous liquid. Rosemary extracts, tea catechin and PBN were added as powder. The
additives (antioxidants and PBN) were dissolved by sonication directly in 5 g of the oil. Then,
the samples were dissolved in the rest of the oil (10 g oil in total). Samples with no additives
were analysed by the same procedures. The samples were then stored in the dark at 40 °C and
samples were withdrawn for ESR recording after 1, 2, 3, 4 and 5 hours. In order to study the
persistence of PBN spin adducts in these systems, the samples were further stored for 22, 42,
62 and 120 hours before recording the trapped radicals.
Instrumental settings of the ESR instrument

A Jeol Jes-FR30, Free Radical Monitor (Jeol Ltd., Tokyo, Japan), was used to conduct the ESR
analyses under the following conditions: Power 4mW, central field 335.6mT, sweep with 5mT,
modulation width 0.2mT and the receiver gain within the range 50-2000. Manganese oxide was
used as a reference maker to calculate the relative intensity of the central peak of the signal
from the spin adduct.
Data from conventional analysis were analysed using a single factor ANOVA analysis performed
in Excel (Microsoft). The results were presented as means ± standard deviations.
Chemical composition
The levels of n-3 fatty acids in the CLO were 19.2% of total fatty acids, with 6.7% EPA and
10.5% DHA. The lipid classes were determined as percentage of total lipids and showed that
the CLO contained 99.4% triacylglycerols, 0.46% cholesterol esters and 0.14% cholesterol. No
detectable levels of free fatty acids or phospholipids were found in the CLO. The CLO contained
0.029 μg 57Fe/g and 0.023 μg 63Cu/g.
Effect of antioxidants
Table 2 presents the OSI levels of natural antioxidants at different concentrations demonstrating
the effectiveness of rosemary extracts as antioxidant in CLO. Also tea catechin, γ-tocopherol
and mixed tocopherols (α-/β-/γ-/δ-tocopherol) were improving the oxidative shelf life. No or
minor improvement of the OSI were obtained by using α-tocopherol at concentrations between
0.5 and 0.025% (w/w) (data obtained with 0.025% α-tocopherol are not shown).
binnenwerk.indd 130 21-08-2006 12:34:22

Table 2. Oil stability of crude cod liver oil added natural antioxidants at different concentrations. Level 1
indicate no stabilising effect compared to non added oil, while higher levels indicate better effect of the
antioxidant at the chosen concentration. The levels are calculated from oil stability indexes performed at
60 °C.
Oil Stability
Concentration added to CLO (w/w): 0.1 (%) 0.25 (%) 0.5 (%)
Rosemary extract 3.6 ± 0.3 4.7 ± 0.1 5.7 ± 0.6

Tea catechin 3.3 ± 0.2 4.2 ± 0.3 3.7 ± 0.0
α-tocopherols 1.0 ± 0.0 1.1 ± 0.0 1.2 ± 0.0
γ-tocopherols 2.4 ± 0.1 2.6 ± 0.0 3.0 ± 0.2
Mixed tocopherols 2.2 ± 0.3 3.2 ± 0.5 2.8 ± 0.8
Antioxidative effect by conventional methods

The trapping of free radicals that are active in the lipid oxidation will as a consequence affect the
progress of lipid oxidation. This is demonstrated by the results from analysis of lipid oxidation
by conventional methods (Figure 1). Samples with PBN had significantly lower peroxide- and
TBARS values compared to the control with no additives, which demonstrates the antioxidative
effect of PBN. The conjugated diene hydroperoxides showed no significant differences (p >0.05)
between any of the samples analysed (results not shown) and were therefore not a recommended
method for analysis of lipid oxidation in these complex oil systems. While the PV generally did
not differentiate between the oxidation levels for the different antioxidants, the TBARS showed
a clear decrease of oxidation levels in the samples containing PBN in combination with the
antioxidant. The Oil Stability Indexes (Figure 2) of CLO without additives also demonstrated the
antioxidative effect of PBN. However, some of the OSI curves of combinations of antioxidants
and PBN were disturbed and complicated the OSI measurements.
ESR spin trapping

The radical intensity assessed with ESR and the spin trapping technique is presented in Figure 3.
The signal intensities increased during the first 5 hours of oxidation at 40 °C in all samples
(Figure 4A). While low intensity of PBN spin adducts are reported as an indicator of low lipid
oxidation and used to study the efficiency of antioxidants (Velasco and others 2004), our current
studies show that only tea catechin had lower intensity of adducts than the control sample.
Thus, the intensity of trapped radicals found in this study did not point to the same conclusions
as the results obtained from the conventional methods and oil stability measurements. For
example α-tocopherol, which had lower peroxide- and TBARS values, showed higher intensity
of spin adducts compared to the control. This is contrary to results from previous studies where
reduced ESR signal intensity in the samples containing α-tocopherol where found when PBN
were used as a spin trap for 30 min (Guo and other 2002). Velasco and others (2004), who
studied vegetable oils, reported that the higher the α-tocopherol concentrations the slower
the PBN adduct formation. They observed differences between the intensity of spin adducts in
samples with different concentrations of α-tocopherol after 1 to 5 hours of oxidation at 60 °C.
They further showed that the rate of spin adduct formation depended on the stability of the
samples before and after an induction point. The induction point was previously defined as
binnenwerk.indd 131 21-08-2006 12:34:22

Peroxide value
20
18 a a a
16 ab
b b b
14 b
mEq/kg
12
10
8
6
4
2
0
TBARS
1.8
a
1.6
1.4
1.2 b bc
c
ug/g oil
1 d bc bcd
0.8 e
0.6
0.4
0.2
0
l
)
)
)
ro
%
%
%
PB
nt
.5
.5
.5
.5
.5
.5
co
(0
(0
(0
(0
(0
(0
y
y
c
c
ar
ar
to
to
to
-to
γ-
/γ-
α-
em
em
/α
N
N
os
os
PB
PB
R
/R
N
PB
Figure 1. Oxidation levels of CLO with combinations of antioxidants and PBN after incubation at 40 °C
for 120 hours. Control is cod liver oil with no addition of antioxidants, toc denotes tocopherols. PBN
concentration: 1mg PBN/g oil. The letters a-e denotes the result of the ANOVA analysis. Different letters
indicate significantly different samples.
the period of time during which radicals are formed very slowly before a sharp linear increase
is observed (Thomsen and others 2000). In our present study very mild conditions were used
(40 °C) and the observable changes in intensity of PBN spin adducts between samples were
detected between 3 to 5 hours after addition of PBN. These findings emphasise the importance
of optimising the time for the reaction between spin trap and radical. Since tocopherols are
natural constituents of fish lipids their influence in the spin trapping is highly relevant during
free radical assessment. Some antioxidants such as α-tocopherol are reported to be pro-oxidants
at high concentrations.
binnenwerk.indd 132 21-08-2006 12:34:22

120 d
d
100 80C
OSI (hours)
80 60C
60
c c
40 b c c
a
20
0
l
)
)
)
ro
%
%
%
PB
nt
.5
.5
.5
.5
.5
.5
co
(0
(0
(0
(0
(0
(0
y
y
c
c
ar
ar
to
to
to
-to
γ-
/γ-
α-
em
em
/α
N
N
os
os
PB
PB
R
/R
N
PB
Figure 2. Oxidative stability (OSI) of CLO with combinations of antioxidants and PBN at 60 °C (60C)
and 80 °C (80C). Control is cod liver oil with no addition of antioxidants, toc denotes tocopherols. PBN
concentration: 1mg PBN/g oil. The letters a-e denotes the result of the ANOVA analysis. Different letters
indicate significantly different samples.
Signal
amplitude
Reference
(Mn)
Magnetic field (mT) 334 336 338
Figure 3. An ESR spectrum of PBN spin adducts from CLO containing 0.5% rosemary extract. The figure
illustrates how the relative intensity is measures (% of reference standard Mn-manganese).
The different results obtained with marine oils compared to previous results on vegetable oil
systems might be explained by the difference in sensitivity of the free radicals detected since
the sensitivity is highly affected by the microenvironment (Nishide and Miyoshi 2002) and the
spin trapping agent (Zhao and others 2001; Okada and others 2002). The spin trapping agent
binnenwerk.indd 133 21-08-2006 12:34:23

3.0 A Control
Rosemary extract
2.5 Tea catechin
γ-tocopherol
ESR spin adducts
2.0 α-tocopherol
0.25% α-tocopherol
1.5 0.1% α-tocopherol
1.0
0.5
0.0
0 1 2 3 4 5
Time (hours)
70
B
60
50
ESR spin adducts
40
30
20
10
0
0 20 40 60 80 100 120
Time (hours)
Figure 4.Relative intensity of PBN spin adducts (% of Manganese as reference) of CLO containing different
antioxidants. (A) First 5 hours of oxidation at 40 °C and (B) From 0 to 120 hours of oxidation at 40 °C.
Standard deviations; 1-5 hours: 0.0 – 0.13, 20-120 hours: 0.7 – 3.5 (ESR recordings were not performed in
duplicates on all selections).
has high specificity towards the radicals and some of the spin traps possess a kinetic specificity
meaning that some radicals are trapped faster than others (Kopáni and others 2006). Marine
lipids and their characteristic highly unsaturated fatty acids have a more complex composition
of oxidation products compared to the less unsaturated vegetable oils. Concretely, more than
ten different resonances from peroxyl groups (-OOH) of hydroperoxides are reported by 1H
nuclear magnetic resonance analysis of docosahexaenoate (Falch and others 2004).
binnenwerk.indd 134 21-08-2006 12:34:24

Persistence of PBN spin adducts

Figure 4B illustrates the radical intensity during the prolonged storage up to 120 hours and
demonstrates the difference in persistence of PBN spin adducts. The intensity of spin adducts in
the control sample, CLO added tea catechin and CLO added 0.25% α-tocopherol are increasing
during the whole experiment while the persistence of spin adducts with the other additives
vary between samples with shortest persistence in CLO with rosemary. Similar observations are
reported in salmon oil and docosahexaenoate, with persistence in similar conditions of <10
hours and <20 minutes (Falch and others 2005). As pointed out in previous investigations,
these results may be related to the interaction between spin adducts and radicals, formed as
oxidation proceeds, to give species not detectable by ESR (Velasco and others 2004; Quian
and others 2000). These findings underline the need for precautions when using the PBN spin
trapping in assessment of oxidative stability of cod liver oil with different additives.
Finally, it is important to note that conventional methods also have their limitations, especially
when introducing aromatic ring structures affecting the spectrophotometric absorption. We
have, in these investigations, added much higher concentrations of natural antioxidants than
what is commonly used by the industry, mainly to study the effect both on lipid oxidation
and on how such additives might influence the methodology. These high concentrations are,
however, generally found to improve the oxidative shelf life in the systems tested.
Conclusions
Rosemary extracts, tea catechin and tocopherols are all improving the oxidative shelf life
of cod liver oil which is clearly demonstrated by conventional methods and the oil stability
index. ESR and the PBN spin trapping technique directly in oils should be used with precaution
when studying oxidative stability in cod liver oil due to: (1) Free radical intensity from PBN
spin trapping directly in oils were negatively correlated with the other methods, (2) PBN has
a significant antioxidative effect and (3) Some of the PBN spin adducts (particularly rosemary
extract and high concentrations of tocopherols) had short half-life as ESR visible compound.
A definite assignment of ESR spectra to specific radicals is necessary in the study of radicals
involved in lipid oxidation.
Acknowledgements
This work has been funded by The Research Council of Norway in the project “Bærekraftig
verdiskaping fra biprodukter og bifangst“. Thor Bernt Melø is thanked for guidance in ESR.
References
Andersen ML, Skibsted LH. 2002. Detection of early events in lipid oxidation by electron spin resonance spectroscopy.
Eur J Lipid Sci 104:65-68.
Andersen ML, Velasco J, Skibsted LH. 2005. Analysis of lipid oxidation by ESR spectroscopy. In: Kamal-Eldin A and
Pokorny J, editors. Analysis of lipid oxidation. Champaign, Illinois: AOCS Press. p 127-151.
AOCS Official Methods and Recommended Practices of the AOCS. Official method Ti 1a-64 Spectrophotometric
determination of conjugated dienoic acid. AOCS Press, Champaign, 1997
Cosgrove JP, Church DF, Pryor WA. 1987. The kinetics of the autoxidation of polyunsaturated fatty acids. Lipids 22:299-
304.
binnenwerk.indd 135 21-08-2006 12:34:24

Falch, E. Lipids from residual raw materials – Quality assessment by advanced analytical methods. PhD thesis at the
Norwegian University of Science and Technology. Trondheim, Norway. Doctoral thesis 2005:238. ISBN 8247173808
Falch E and Aursand M. 2005. Resonance spectroscopy to study lipid oxidation in fish and fish products. In: Graham
Webb, editor. Handbook of modern magnetic resonance. New York, USA: Kluwer Academic/Plenum Publisher.
Falch E, Anthonsen H, Axelson D, Aursand M. 2004. Correlation between 1H NMR and traditional analytical methods for
determining lipid oxidation in ethylesters of Docosahexaenoic acid. J Am Oil Chem Soc 81(12):1105-1109.
Falch E, Velasco J, Aursand M, Andersen M. 2005a. Detection of radical development by ESR spectroscopy techniques
for assessment of oxidative susceptibility of fish oils. Eur J Food Res 221(5):667-674
Falch E, Rustad T, Aursand M. 2006. By-products from gadiform species as a raw material for production of marine lipids
as an ingredient for food and feed. Proc. Biochem. 41:666-674.
Frankel EN, Huang S-W, Aeschbach R, Prior E. 1996a- Antioxidant activity of rosemary extract and its constituents,
carsonic acid, carnosol and rosemarineic acid, in bulk and oil-in-water emulsion, J Agric Food Chem 44:131-135.
Frankel EN, Huang S-W, Prior E, Aeschbach R. 1996b. Evaluation of antioxidant activity of rosemary extracts, carnosol
and carsonic acid in bulk vegetable oils and fish oil and their emulsions. J Sci Food Agric 72:201-208.
Guo Q, Rimbach G, Moini H, Weber S, Packer L. 2002. ESR and cell culture studies on free radical-scavenging and
antioxidant activities of isoflavonoids. Toxicology 179(1-2):171-180.
Jebe TA, Matlock MG, Sleeter RT. 1993. Collaborative study of the oil stability index analysis, J Am Oil Chem Soc
70:1055-1061.
Jensen PN, Danielsen B, Bertelsen G, Skibsted LH, Andersen M. 2005. Storage stabilities of pork scratchings, peanuts,
oatmeal and muesli: comparison of ESR spectroscopy, headspace-GC and sensory evaluation for detection of
oxidation in dry foods. Food Chem 91:25-38.
Ke PJ and Woyewoda AD. 1979. Microdetermination of thiobarbituric acid values in marine lipids by a direct
spectrophotometric method with a monophasic reaction system. Analytica Chimica Acta 106:279-284.
Kopáni M, Celec P, Danišovič L, Michalka P, Biró C. 2006. Oxidative stress and electron spin resonance. Clinica Chimica
Acta 364(1-2):61-66
Nishide N, and Miyoshi N. 2002. Singlet oxygen trapping by DRD156 in micellar solutions. Life Sci 72(3):321-8.
Nogala-Kalucka M, Korczak J, Dratwia L-S, Siger A, Buchowski M. 2005. Changes in antioxidant activity and free
radical scavenging potential of rosemary extract and tocopherols in isolated rapeseed oil triacylglycerols during
accelerated tests. Food Chem 93:227-235.
Okada Y, Okajima H, Shima Y, Ohta H. 2002. Hydroxyl radical scavenging action of capsaicin. Redox Rep 7(3):153-7.
Qian SY, Wang HP, Schafer FQ, Buettner GR. 2000. EPR detection of lipid-derived free radicals from PUFA, LDL, and cell
oxidations. Free Rad Biol Med 29:568-579.
Schaich K M and Borg DC. 1990 Solvent effects in the spin trapping of lipid oxyl radicals. Free Radic Res Commun
9:267-278.
Thomsen MK, Vedstesen H, Skibsted LH. 1999. Quantification of radical formation in oil-in-water food emulsions by
electron spin resonance spectroscopy. J Food Lipids 6:149-158.
Thomsen MK, Kristensen D, Skibsted LH. 2000. Electron spin resonance spectroscopy for determination of oxidative
stability of food lipids. J Am Oil Chem Soc 77:725-730.
Undeland I, Stading M, Lingnert H. 1998. Influence of skinning on lipid oxidation in different horizontal layers of
herring (Clupea harengus) during frozen storage. J Sci Food Agric 78:441-450.
Velasco J, Andersen ML, Skibsted LH. 2004. Evaluation of oxidative stability of vegetable oils by monitoring the
tendency to radical formation. A comparison of electron spin resonance spectroscopy with the Rancimat method
and differential scanning calorimetry. Food Chem 85:623-632.
Yang M-H, Lin H-J, Choong Y-M. 2002. A rapid gas chromatographic method for direct determination of BHA, BHT and
TBHQ in edible oils and fats. Food Res Int 35(7):627-633.
Zhao H, Joseph J, Zhang H, Karoui H, Kalyanaraman B. 2001. Synthesis and biochemical applications of solid cyclic
nitrone spin trap: a relatively superior trap for detecting superoxide anions and glytathiyl radicals. Free Radic Biol
Med 32(5):599-606.
binnenwerk.indd 136 21-08-2006 12:34:24

Chapter 2:
Quality of farmed fish
The proportion of farmed fish and of other aquatic species of the total world fish production
has been steadily and is still increasing. Today about 30% of the world fish production comes
from aquaculture while the amount of captured fish has reached a relatively stable plateau of
about 100 million tons/year.
While the efforts in aquaculture research and industry in the past have been mainly aiming at
to increase the amount of fish produced nowadays the quality of farmed fish in the widest sense
and new species are in the focus.
More and more species are emerging and every year new successful candidates appear on the
market. An important species in this respect is Atlantic cod (Gadus morhua), one of the most
highly appreciated food fishes in the Northern hemisphere (fresh or salted), which stocks have
been partly overexploited.
This chapter starts with five papers investigating farmed cod. The effects of fed vegetable
proteins to the fish on quality parameters when starved before slaughter is the first, another
deals with a comparison of the gelatinolytic activity in muscle tissue of farmed and captures
cod. Followed by a paper in which different brining methods and their effects on quality of
farmed cod was investigated.
Two contributions concentrate on the pre rigor processing of cod and the influence of this early
processing on quality and shelf-life as well as on the relevance of the storage temperature on
the contraction and gaping of the fish.
Functional food products based on fish as raw material are actually an interesting and important
research area. Here a contribution on the development of a selenium enriched functional African
catfish is presented.
One of the many aspects of aquaculture is the problem of animal welfare in all steps of
aquaculture from the egg to slaughter. A comparison of commercial and experimental slaughter
of farmed carp in relation to rigor mortis development and fish flesh quality is an actual example
of work carried out in this field.
binnenwerk.indd 137 21-08-2006 12:34:24

binnenwerk.indd 138 21-08-2006 12:34:24
Pre-slaughter starvation of farmed Atlantic cod fed vegetable
proteins: Effects on quality parameters

Fiskeriforskning, PO Box 6122, N-9291 Tromsø, Norway
Abstract
Atlantic cod (Gadus morhua) is an emerging species for cold-water aquaculture in Europe.
Currently, diets for cod are largely based upon fish meal and fish oil. The use of alternative feed
ingredients is however expected and highly relevant. In this study we have been investigating
how pre-slaughter starvation affect muscle quality attributes, measured as muscle pH, water-
holding capacity (WHC) and shear force, of farmed Atlantic cod that have been fed diets
containing vegetable proteins. One group of fish was starved for two weeks prior to slaughter
and the other group was starved for three weeks. The starvation time influenced significantly
the attributes muscle pH, WHC and shear force, which may indicate that pre-slaughter starvation
could be used as a tool for modulating required quality attributes.
Keywords: farmed cod, vegetable proteins, starvation, water-holding capacity, shear force
Introduction
The interest for intensive aquaculture is increasing and has become the fastest growing
food production sector of the world. Fish feed producers have depended profoundly on fish
meal and fish oil as their source of protein and lipid. However, the feed industry is currently
encountering shortfalls in availability of these ingredients due to a decline in wild fish catching
and to increased human demands for some species currently being used for fish meal and oil
production. The need for alternative protein sources to replace fish meal in fish feed was even
strongly recommended by the Second International Symposium on Sustainable Aquaculture
(1998) in Oslo, Norway.
Cod (Gadus morhua) is an emerging species for cold-water aquaculture in Europe. Currently, diets
for cod are largely based upon fish meal and fish oil. The use of alternative feed ingredients is,
however, expected and highly relevant.
Among the alternative protein sources available for fish feeds, defatted soybean meal (SBM) has
been universally accepted. SBM has a favourable amino acid profile. It is consistently available
and reported to be palatable to most fish species (Akiyama 1988). By exchanging a part of the
fish protein in the feed with a combination of soybean meal and corn gluten meal, the protein
quality can easily be maintained to give optimal growth and feed efficiency (ongoing study at
Fiskeriforskning). The amino acid profile of the proteins from soybeans balances the amino acid
profile of corn gluten meal, especially with respect to arginine, lysine and sulphur containing
amino acids (Pongmaneerat and Watanabe 1993; Watanabe and others 1993; Akiyama and
others 1995).
binnenwerk.indd 139 21-08-2006 12:34:25

Substitution of marine proteins by vegetable proteins has been tested for several fish species
(Kaushik and others 1995; Watanabe and others 1998; Regost and others 1999; Kissil and others
2000; Storebakken and others 2000; Gómez-Requeni and others 2003). There is a variation in the
published results with respect to successful replacement of fish protein in the feed depending
on fish species, size and type of ingredients used. Furthermore, the success has primarily been
based on growth rate and feed efficiency rather than the quality of the end product.
Several studies have been performed to investigate the effects of periods of pre-slaughter
starvation on muscle quality of salmon (Einen and others 1998; Einen and Thomassen 1998;
Krogdahl and Bakke-McKellep 2005). A few starvation studies have been performed on cod with
the focus on physiological alterations (Guderley and others 1996; Lambert and Dutil, 1997;
Guderley and others 2003). Knowledge about the effect of starvation periods on muscle quality
is however limited. Annual cycles in food abundance force many fish species in their natural
habitats to subsist over long periods on little food. Atlantic cod are able to modify their muscle
metabolic machinery in response to changes in food availability (Gildberg 2004; Pelletier and
others 1993). There are indications that changes in the glycolytic apparatus are coordinated
with changes in growth rate (Guderley and others 1996). Prolonged starvation leads to white
muscle degradation, as muscle glycogen and proteins are used as energy sources. Starvation can
markedly increase the water content in white muscle, which can reach 90% during prolonged
starvation (Lambert and Dutil 1997). During natural starvation periods of wild cod the glycogen
content (ante mortem) is reduced. This results in decreased lactate and increased pH in the
muscle in the post rigor state. One consequence of this is a reduction in fillet gaping (Love
1988). In this study it was investigated how periods of pre-slaughter starvation affect muscle
quality attributes, measured as muscle pH, water-holding capacity (WHC) and shear force, of
farmed Atlantic cod that have been fed diets containing vegetable proteins.
Muscle texture is an important quality attribute of fish. Texture of raw fillets may be measured
objectively using instrumental equipments. The main techniques applied for fish may be
classified into puncture, compression, shear and tensile techniques (Sigurgisladottir and others
1999). Textural properties of fish muscle may also be measured indirectly as muscle pH and
water-holding capacity. The water-holding capacity (WHC) of the muscle is shown to be a useful
tool for measuring textural quality in fish and meat (Offer and Trinick, 1983; Fennema 1990;
Ofstad and others 1996; Wilsson III and van Laack 1999; Schäfer and others 2002; Olsson and
others 2003a; Huff-Lonergan and Lonergan 2005). A high WHC (low liquid loss, LL) is of great
importance both to the industry and the consumers. Limited knowledge is available about how
novel feed ingredients affect the muscle pH, WHC and texture and how these attributes change
during periods of non-feeding. The effects of replacing marine based protein with vegetable
protein in diets for cod on growth and quality in general were also investigated in our project,
but will be reported elsewhere. This paper is focusing on the effect of pre-slaughter starvation
on the muscle pH, WHC and shear force when cod have been fed vegetable based diets.
Fish
Farmed Atlantic cod were kept in nets at the aquaculture station at Austevoll, Bergen during
the feeding trial. The fish with an average individual weight of 760 g were divided into 3
different groups and were fed 3 different diets over a period of 6 months. One control group
was fed purely marine based feed (“0-diet”), one group was feed a diet where 15% (“15-diet”)
binnenwerk.indd 140 21-08-2006 12:34:25

Pre-slaughter starvation of farmed Atlantic cod fed vegetable proteins
of the total protein was vegetable proteins (soya and corn gluten) and finally one group was
feed a diet where 45% (“45-diet”) of the total protein was replaced by vegetable proteins. The
fish were starved for 2 or 3 weeks prior to slaughtering. The average fish weight at the end of
the feeding trial were in group “0-diet”, “15-diet” and “45-diet”, 1432 g, 1465 g and 1427 g,
respectively.
The fish were slaughtered at the aquaculture station and packed in ice in “airplane-packages”.
The fish were then transported by plane to Tromsø the next morning for further preparations in
the laboratories. This procedure was repeated one week later (for the fish starved for 3 weeks).
Fifteen fish from each group were filleted and stored on ice for two weeks. At sampling, on
storage day 2, 7 and 14, the fish (five individuals each time) were individually analysed for
muscle pH, WHC (LL%) and instrumentally measured shear force.
Muscle pH
Muscle pH was measured in chopped muscle suspended in 0.15 M KCl using a glass electrode.
Water-holding capacity
The water-holding capacity was measured individually using the net test according to Ofstad
and others (1993). Coarsely chopped muscle (15 g) was centrifuged at 320 x g for 15 min at
5 °C. The liquid loss was determined after the gentle centrifugation and expressed as percent
of the weight of liquid released pr. 15 g of sample (LL%). Mean values were calculated.
Shear force
The shear force was measured according to Sigurgisladottir and others (1999). A TA.HDi texture
analyzer with a load cell of 25 kg (Stable Micro System, England) was used. A 4 cm wide skin-
and bone free sample (longitudinal of the filets) were cut from the loins. The samples were
placed in a sample-holder and cut perpendicular to the longitude. The blade (knife edge, 60 °)
had a thickness of 3 mm and width of 70 mm which cut through the sample at a speed of
2 mm/sec. The shear force was measured as the maximum force required cutting through the
samples. Six replicates were made from each sample.
Data analysis
The data were analysed by use of multivariate techniques (PCA and PLS1-regression), applying
the software Unscrambler version 9.1.2 (CAMO Process AS, Oslo, Norway). The variables were
weighted with the inverse of the standard deviation of all the objects in order to compensate
for the different scales of the variables.
As an initial study of the effect of vegetable proteins, storage time and starvation time on the
selected quality attributes (pH, LL% and shear force) a principal component analysis (PCA)
was run. It was found that four principal components explained 93% of the variation in the
data set. The scores and correlation loadings of PC1 and PC2, representing 34% and 27% of
the total variation, are presented in Figure 1a and b, respectively. The score plot shows a clear
distinction between the fish starved for 2 and 3 weeks. Hence, the two groups were further
investigated separately as well as being analysed together. The correlation loadings plot shows
that the variation in starvation time, storage time, pH and LL% are explained more than 50%
by the model.
binnenwerk.indd 141 21-08-2006 12:34:25

PC2 Scores
3 2 weeks of
starvation
2
-1
-2
-3 3 weeks of
starvation
-4
PC1
-3 -2 -1 0 1 2 3
X-expl: 34%, 27%
Figure 1a. Score plot from the PCA of pH, liquid loss, % vegetable protein, shear force, storage time and
starvation time.
2 = 2 weeks of starvation
3 = 3 weeks of starvation
Correlation loadings (X)

PC2
1.0
+ storage time
+ LL (%)
0.5 + pH
+ shear force
+ % vegetable pro starvation time +

-0.5
-1.0
PC1
-1.0 -0.5 0 0.5 1.0
X-expl: 34%, 27%
Figure 1b. Correlation loading plot from the PCA of the quality attributes, storage time and % vegetable
proteins. The outer and inner ellipses indicate 100% and 50% explained variance, respectively.
binnenwerk.indd 142 21-08-2006 12:34:25

To identify the effects of starvation time, storage time and vegetable protein on the measured
variables (pH, LL% and shear force), and to reveal which variables significantly affect the
quality parameters, PLS1-regression models were made. Starvation time, vegetable protein and
storage time were applied as the X-matrix, while the individual quality parameters were used as
Y-matrices. By applying the Martens Uncertainty Test the influence of vegetable protein, storage
time and starvation time on each single quality parameter were examined. Table 1 summarises
the results of the analyses. Variables that significantly affect the quality parameters are shown,
both when all samples are analysed together, and when cod starved for two and three weeks
were analysed separately. When all samples are used in the model, starvation time did clearly
influence significantly all three attributes.
Muscle pH
When the starvation time increased the muscle pH did increase in all the diet groups (Table 2
and 3). As glycogen and triglycerides has a primary role as energy reserves, they are the first to
be used during starvation (Black and Love 1986). Consequently the muscle pH increases.
Independent of starvation time muscle pH decreased with increasing levels of vegetable protein.
Decreasing pH did not correspond to increased LL (see Table 2 and 3) which supports the
suggestion that pH alone does not influence changes in WHC (Rustad 1992; Olsson and others
2003b; Huff-Lonergan and Lonergan 2005). As expected, muscle pH within the diet groups
increased with increasing storage time with one exception. The muscle pH of cod that were fed
45% vegetable protein did not change during storage. This implies that the composition of the
muscle in this group could be altered due to the high amount of vegetable protein. Increased
inclusion of vegetable protein sources have been shown to reduce growth in salmon (Olli
and others 1994, 1995; Opstvedt and others 2003; Mundheim and others 2004) seabass and
trout (Gomes and others 1995a, 1995b; Dias and others 1999). When reduced growth is larger
than what can be explained by digestibility alone it has been suggested that some additional
physiological effects may be of influence (Mundheim and others 2004). This may be the case
for the group fed the highest amount of vegetable protein.
Table 1. Variables with significant effect on quality parameters. Significance is identified by Martens
uncertainty test (P = 0.05).
Variables with significant effect on quality parameters
Quality parameter All samples 2 weeks starvation 3 weeks starvation
pH storage time storage time storage time

% vegetable protein % vegetable protein % vegetable protein
starvation time
Liquid loss (%) storage time

% vegetable protein % vegetable protein
starvation time
Shear force storage time storage time

% vegetable protein % vegetable protein % vegetable protein
starvation time
binnenwerk.indd 143 21-08-2006 12:34:26

Table 2. Muscle pH, LL% and shear force in cod starved for 2 weeks prior to slaughter. Effects of level of
vegetable protein in the diets and storage on ice.
Storage time (days) Diet 0 Diet 15 Diet 45
pH 2 6.15a 6.19a/I 6.14I

7 6.28a/I 6.22a/I 6.16I
14 6.31a/I 6.30a/II 6.17I,II
LL(%) 2 12.9a 11.8ab 11.3

7 14.1 13.8a 12.0
14 14.5a/I 14b 9.6I
Shear force 2 4897 4489 4666a

7 4207a 4395 3939b
14 4837a/I 5048II 3143ab/I,II
Diet 0 = marine based diet, diet 15 = 15% vegetable proteins, diet 45 = 45% vegetable proteins of total
protein content.
Samples with the same letters within the column are significantly different.
Samples with the same roman numerals are significantly different.
Table 3. Muscle pH, LL% and shear force in cod starved for 3 weeks prior to slaughter. Effects of level of
vegetable protein in the diets and storage on ice.
Storage time (days) Diet 0 Diet 15 Diet 45
pH 2 6.25a/I 6.23a/II 6.18a/I,II

7 6.31a/I 6.27a/I 6.22a/I
14 6.40a/I 6.36a/I 6.31a/I
LL(%) 2 7.6ab/I 6.2a/I 7.8

7 10.1a 8.9a 8.3
14 10.1b/I 7.6I 8.7
Shear force 2 5581a 5371 5383ab

7 4743a/I 4917II 4335 a/I,II
14 5031I 4766 4262b/I
Diet 0 = marine based diet, diet 15 = 15% vegetable proteins, diet 45 = 45% vegetable proteins of total
protein content.
Samples with the same letters within the column are significantly different.
Samples with the same roman numerals are significantly different.
binnenwerk.indd 144 21-08-2006 12:34:26

WHC, measured as LL%

When all samples are used in the regression model, both starvation time and level of vegetable
protein influenced significantly the LL (Table 1). Increased starvation time resulted in decreased
LL (Table 2 and 3). The level of vegetable protein influenced significantly the LL when the
fish have been starved for only two weeks. When the fish were starved for one more week the
effect of vegetable protein is minimised and the storage time is the factor that significantly
influences the LL. Increasing storage time resulted in poorer WHC (increased LL%) in the fish
fed “0-diet” and “15-diet”. The LL% of fish fed the “45-diet” were however unchanged during
the storage period.
Shear force
The shear force was significantly influenced by starvation time, storage time and vegetable
protein when all samples were used in the model. Increased starvation time resulted in a
firmer muscle texture (higher shear force values). With respect to storage time, there was no
systematic trend in how this influenced the shear force. As can be seen from Table 1 there
was no significant effect of storage time on the shear force when the fish were starved for two
weeks. After one more week of starvation this was however changed. In the fish fed “0”- and
“45-diet” the shear force decreased significantly from day 2 to day 7 (see Table 3). In the fish
fed “15-diet” there were no significant effect of increased storage time on ice.
Conclusion
Periods of starvation before slaughtering (Johansson and Kiessling 1991; Einen and Thomassen
1998) and “feeding up” or “feeding down” (Einen and others 1999; Rasmussen and others 2000)
have all been shown to be possible tools to tune to required quality attributes. The results
from the present paper indicate that selected quality attributes may be modulated by periods
of pre-slaughter starvation. However, experiments better designed for investigating effects of
pre-slaughter starvation is needed before reliable conclusions can be made. Such experiments
are currently being performed at Fiskeriforskning.
Acknowledgements
The Norwegian Research Council is thanked for financial support.
References
Akiyama DM. 1988. Soybean utilization in fish feed. Korean Feed Association Conference, August 1988, Seoul, Korea.
Akiyama T, Unuma T, Yamamoto T, Marcouli P, Kishi S. 1995. “Combination use of malt protein flour and soybean meal
as alternative protein sources of fish meal in fingerling rainbow trout diets”. Fish Sci 61:828-832.
Black D, Love RM. 1988. Estimating the carbohydrate reserves in fish. J Fish Biol 32:335-340.
Dias J, Arzel J, Corraze G, Bautista JM, Alvarez MJ, Lopez-Bote CJ, Kaushik SJ. 1999. Effect of dietary protein sources on
lipid metabolism in seabass and rainbow trout. Conference paper, Workshop on fish nutrition, Brest, 9-10 March.
Einen O, Waagan B, Thomassen, MS. 1998. Starvation prior to slaughter in Atlantic salmon (Salmo salar) I. Effects on
weight loss, body shape, slaughter- and fillet-yield, proximate and fatty acid composition. Aquaculture 166:85-
104.
Einen O, Thomassen, MS. 1998. Starvation prior to slaughter in Atlantic salmon (Salmo salar) II. White muscle
composition and evaluation of freshness, texture and colour characteristics in raw and cooked fillets. Aquaculture
169:37-53.
binnenwerk.indd 145 21-08-2006 12:34:26

Einen O, Mørkøre T, Rørå AMB, Thomassen MS. 1999. Feed ration prior to slaughter – a potential tool for managing
product quality of Atlantic salmon (Salmo salar). Aquaculture 178:149-169.
Fennema OR. 1990. Comparative water holding properties of various muscle foods. A critical review relating to
definitions, methods of measurements, governing factors, comparative data and mechanistic matters. J Muscle
Foods 1:363-381.
Gildberg A. 2004. Digestive enzyme activities in starved pre-slaughter farmed and wild-captured, Atlantic cod (Gadus
morhua). Aquaculture 238:343-353.
Gomes EF, Rema P, Kaushik SJ. 1995a. Replacement of fish meal by plan proteins in the diet of rainbow trout
(Oncorhynchus mykiss): digestibility and growth performance. Aquaculture 130:177-186.
Gomes EF, Rema P, Gouveia A, Teles AO. 1995b. Replacement of fish meal by plant proteins in diets for rainbow trout
(Oncorhynchus mykiss): effect of the quality of the fishmeal based control diets on digestibility and nutrient
balances. Water Sci Technol 31(10):205-211.
Gómez-Requeni P, Mingarro M, Kirchner S, Calduch-Giner JA, Medale F, Corraze G, Panserat S, Martin SAM, Houlihan
DF, Kaushik SJ, Perez-Sanchez J. 2003. Effects of dietary amino acid profile on growth performance, key metabolic
enzymes and somatotropic axis responsiveness of gilthead sea bream (Sparus aurata). Aquaculture 220:749-767.
Guderley H, Dutil J-D, Pelletier D. 1996. The physiological status of Atlantic cod, Gadus morhua, in the wild and the
laboratory: estimates of growth rates under field conditions. Can J Fish Aquat Sci 53:550-557.
Guderley H, Lapointe D, Bédard M, Dutil J-D. 2003. Metabolic priorities during starvation: enzyme sparing in liver and
white muscle of Atlantic cod, Gadus morhua L. Comp Biochem Physiol A 135:347-356.
Huff-Lonergan E, Lonergan SM. 2005. Mechanisms of water-holding capacity of meat: The role of post mortem
biochemical and structural changes. Meat Sci 71:194-204.
Johansson L, Kiessling A. 1991. Effects of starvation in rainbow trout: II. Eating and storage quality of iced and frozen
fish. Acta Agric Scand 41:207-216.
Kissil GW, Lupatch I, Higgs DA, Hardy RW. 2000. Dietary substitution of soy and rapeseed protein concentrates for fish
meal, and their effects on growth and nutrient utilization in gilthead seabream Sparus aurata L. Aquaculture Res
31:595-601.
Kaushik SJ, Cravedi JP, Lalles JP, Sumpter J, Fauconneau B, Laroche M. 1995. Partial or total replacement of fish meal
by soybean protein on growth, protein utilization, potential estrogenic or antigenic effects, cholesterolemia and
flesh quality in rainbow trout, Oncorhynchus mykiss. Aquaculture 133:257-274
Krogdahl Å, Bakke-McKellep AM. 2005. Fasting and refeeding cause rapid changes in intestinal tissue mass and digestive
enzyme capacities of Atlantic salmon (Salmo salar L.). Comp Biochem Physiol A 141:450-460.
Lambert Y, Dutil J-D. 1997. Can simple condition indices be used to monitor and quantify seasonal changes in the
energy reserves of Atlantic cod (Gadus morhua) ? Can J Fish Aquat Sci 54(S1):104-112.
Love RM. 1988. The food fishes, their intrinsic variation and practical implications. Farrand Press, London, UK.
Mundheim H, Aksnes A, Hope B. 2004. Growth, feed efficiency and digestibility in salmon (Salmo salar L.) fed different
dietary proportions of vegetable protein sources in combination with two fish meal qualities. Aquaculture 237:315-
331.
Offer G, Trinick J. 1983. On the mechanism of water-holding in meat: The swelling and shrinking of myofibrils. Meat
Sci 8:245-281.
Ofstad R, Kidman S, Myklebust R, Hermansson AM. 1993. Liquid holding capacity and structural changes during heating
of fish muscle; cod (Gadus morhua L.) and salmon (Salmo salar). Food Structure 12:163-174
Ofstad R, Egelandsdal B, Kidman S, Myklebust R, Olsen RL, Hermansson A-M. 1996. Liquid loss as affected by post
mortem ultrastructural changes in fish muscle: Cod (Gadus morhua) and salmon (Salmo salar). J Sci Food Agric
71:301-312.
Olli JJ, Krogdahl Å, Ingh TSGAM, Brattås IE. 1994. Nutritive value of four soybean products in diets for Atlantic salmon
(Salmo salar L.). Acta Agric Scand A 44:50-60.
Olli JJ, Krogdahl Å, Våbenø A. 1995. Dehulled solvent extracted soybean meal as a protein source in diets for Atlantic
salmon Salmo salar L. Aquac Res 26:167-174.
binnenwerk.indd 146 21-08-2006 12:34:26

Olsson GB, Olsen RL, Carlehög M, Ofstad R. 2003a. Seasonal variation in chemical and sensory characteristics of farmed
and wild Atlantic halibut (Hippoglossus hippoglossus). Aquaculture 217:191-205
Olsson GB, Olsen RL, Ofstad R. 2003b. Post-mortem structural characteristics and water-holding capacity in Atlantic
halibut muscle. Lebensm-Wiss Technol 36:125-133.
Opstvedt J, Aksnes A, Hope B, Pike IH. 2003. Efficiency of feed utilization in Atlantic salmon (Salmo salar L.) fed diets
with increasing substitution of fish meal with vegetable proteins. Aquaculture 221:365-379.
Pelletier D, Guderley H, Dutil J-D. 1993. Does the aerobic capacity of fish muscle change with growth rates? Fish
Physiol Biochem 12:83-93.
Pongmaneerat, J. and Watanabe, T. 1993. “Utilization of soybean meal as protein source in diets for rainbow trout”
Nippon Suisan Gakkaishi 58:1983-1990.
Rasmussen RS, Ostenfeld TH. Rønsholdt B, McLean E. 2000. Manipulation of end-product quality of rainbow trout with
finishing diets. Aquaculture Nutr 6:17-23
Regost CJ, Arzel J, Kaushik SJ. 1999. Partial or total replacement of fish meal by corn gluten meal in diet for turbot
(Psetta maxima). Aquaculture, 180:99-117.
Rustad T. 1992. Muscle chemistry and the quality of wild and farmed cod. In: Huss, H. H., Jacobsen, M., and Liston,
J. (Eds.) Quality Assurance in the Fish Industry. London, Elsevier: 19-27.
Schäfer A, Rosenlund K, Purslow PP, Andersen HJ, Henckel P. 2002. Physiological and structural events post mortem
of importance for drip loss in pork. Meat Sci 61:355-366.
Sigurgisladottir S, Hafsteinsson H, Johnson A, Lie Ø, Nortvedt R, Thomassen M, Torrisen O. 1999. Textural Properties
of Raw Salmon Fillets as Related to Sampling Method. J Food Sci 64(1):99-104.
Storebakken T, Shearer KD, Baeverfjord G, Nielsen BG, Åsgård T, Scott T, De Laporte A. 2000. Digestibility of
macronutrients, energy and amino acids, absorption of elements and absence of intestinal entiritis in Atlantic
salmon, Salmo salar, fed diets with wheat gluten. Aquaculture 184:115-132.
Watanabe T, Pongmaneerat J, Satoh S, Takeuchi T. 1993. Replacement of fish meal by alternative protein sources in
rainbow trout diets. Nippon Suisan Gakkaishi 59:1573-1579.
Watanabe T, Verakunpiriya V, Watanabe K, Viswanath K, Satoh S. 1998. Feeding of rainbow trout with non-fish meal
diets. Fish Sci 63:258-266.
Wilsson III GG, van Lack RLJM. 1999. Sarcoplasmic proteins influence water-holding capacity of pork myofibrils. J Sci
Food Agric 79:1939-1942.
binnenwerk.indd 147 21-08-2006 12:34:26

binnenwerk.indd 148 21-08-2006 12:34:27
The effect of pre rigor processing of cod (Gadus morhua L.) on
quality and shelf life
Abstract
The objective was to determine how time of processing affected quality and shelf life of cod
fillets. Wild Atlantic cod were kept alive in net cages, with and without feeding, until slaughter
and filleting pre, in and post rigor. Fillets from fed cod obtained low ultimate muscle pH and
high weight reduction. There were only minor variations in pH and weight reduction depending
on the time of filleting. Rigor status affected contraction of the fillets. The reduction in length
from pre rigor fillets varied from 10% to 14%. Pre rigor fillets had less gaping, more firm texture
and slower bacterial growth after filleting. This leads to the conclusion that pre rigor processing
was a better concept for distribution and sale of high quality fresh cod fillets.
Keywords: wild cod, pre rigor, quality, shelf life
Introduction
Traditionally pre rigor processing is known from the production of frozen fillets onboard
factory ships. For most of the land based industry pre rigor processing was not an option
until aquaculture provided a stable supply of live raw material. In Norway, both production of
farmed cod from hatched fingerlings and capture based aquaculture are growing businesses.
In capture based aquaculture commercially sized wild cod is caught and stored alive with or
without feeding, until processing.
To obtain optimal yield and product quality, filleting while the fish still is in rigor is not an option.
When the processors have to store the raw material 3-5 days before filleting, until the fish is out
of rigor, valuable time for distribution and sale will be lost. Further on this means shorter shelf
life post filleting and less freshness in the products when they are bought by the consumers.
The growing interest for pre rigor filleting is based upon a belief that this concept could provide
a stable supply of very fresh fillets to the food industry and consumers. Pre rigor processing of
fillets close to the fish farms, and distribution of fillets instead of whole fish, may also provide a
more predictable product quality and will reduce the volume that has to be handled throughout
the distribution chain.
Time of filleting may significantly affect the quality and deterioration pattern of fish fillets.
Shrinkage of pre rigor fillets is well known, and texture and colour are reported to be different in
pre and post rigor processed salmon fillets (Skjervold 2002). The studies on salmon also showed
that early processing significantly reduced the severity of fillet gaping, (loss of muscle integrity)
and had effects on colour and texture (Skjervold 2002). Most of these differences are regarded
positive, i.e. that pre rigor filleting significantly reduced gaping in fresh salmon fillets.
binnenwerk.indd 149 21-08-2006 12:34:27

Other quality parameters such as weight changes (drip loss) and microbial activity may be
different when comparing fillets processed pre or post rigor during chilled storage. It is also
shown that low pH is associated with reduced water-holding capacity and increased myofibrillar
shrinkage during heating of cod (Ofstad and others 1996).
The difference in filleting time, 4-5 days post mortem for post rigor filleting and a few hours for
pre rigor filleting, is likely to affect the level of contamination and the microbiological growth
rate. Indeed, pre rigor processing might increase the sensitivity of the fillets to microbial
spoilage. The reason for this is the early exposure of the interior of the muscle to contamination
and growth of micro organisms. In addition it is assumed that the micro flora may be different
in pre and post rigor processed cod fillets.
It has also been shown that when using pre or post rigor salmon or rainbow trout fillets different
product qualities are obtained after salting and smoking (Rørå and others 2003; Tobiassen and
others 2003). Generally, the yield is lower when brining or salting pre rigor fillets.
Not many studies have taken place on the effects from pre rigor filleting of cod. However, it
is likely that the behaviour of a lean fish such as cod will differ significantly from fatty fish
species like salmon and rainbow trout. Therefore, the objectives of these two experiments
were to determine how pre rigor processing of wild Atlantic cod, obtained from capture based
aquaculture, affected fillet quality and shelf life of the fillets during chilled storage after
processing.
Fish raw material

Wild Atlantic cod (Gadus morhua L.) were caught by Danish seine and kept alive in net cages
until slaughtering and filleting. Two experiments were carried out in November 2004 and March
2005. In the two experiments the live cod had been treated differently.
• Cod fed with capelin: The raw material in experiment 1 was live cod that had been kept in
net cages and fed with capelin for six months before slaughtering and filleting.
• Cod not fed (starved): The raw material in experiment 2 was live cod that after capture was
stored in a net cage for 30 days without feeding.
Slaughtering, filleting and storage

The cod were killed by a blow to the head, bled, gutted, washed and stored on ice until filleting
pre, in and post rigor (<3 hours, 2 days and 4 days post mortem, respectively). After processing
the fillets were wrapped in plastic film and stored chilled on ice in 10 kg styropor boxes.
The following quality parameters were measured 6 days after filleting: reduction of fillet weight
(drip loss), fillet contraction (length reduction), muscle-pH, gaping and texture. To evaluate
shelf life, analyses of total viable counts (TVC) and sulphide-producing bacteria were carried
out during 12, 10 or 8 days after filleting the fish pre rigor, in rigor or post rigor respectively.
For all groups these correspond to 12 days after slaughtering.
binnenwerk.indd 150 21-08-2006 12:34:27

The effect of pre rigor processing of cod (Gadus morhua L.) on quality and shelf life
Analyses
• Fulton´s (K) condition factor; K = 100*(W/L3), where W is total weight (g) and L is the fork
length (cm). Numbers of fishes within each batch were 10.
• pH in blood and muscle was measured at slaughtering and 6 days after filleting by using a
hand-held WTW 330/Set-1 pH Meter (Wissenschaftlich-Technische Werkstatten, Weilheim,
Germany) equipped with a Hamilton double pore glass electrode (Hamilton Bonaduz,
Switzerland). Numbers of fillets within each batch were 5.
• Drip loss was estimated as weight changes after 6 days ice storage. Numbers of fillets within
each batch were 10.
• Fillet contraction during chilled storage was measured as length reduction at day 6 after
filleting, calculated as percentage of the initial length of the fillets measured at time of
processing. Numbers of fillets within each batch were 10.
Gaping and texture were evaluated by sensory methods. Three trained assessors examined 10
fillets within each batch. Fillet gaping was determined by visual inspection, and texture by
finger pressure. A scale of 0 denotes no gaping or a firm and natural texture, while 3 denotes
extreme gaping or very soft texture.
Raw material for microbial analyses was cut from the interior of the fillets using sterile technique.
Five samples, from different fillets were measured in duplicate at each sampling point. The total
viable count in the fillets (TVC) was measured using standard plate count agar (Oxide CM463).
Sulphide-producing bacteria were determined as black colonies on Iron agar Lyngby (Oxide
CM964) after 2-3 days incubation at 22 ºC.
Analyses of variance were carried out to establish effects of filleting time within each of the
experiments, and differences between the two experiments. A t-test was used to determine
significance between the samples. The significance level was set to P < 0.05.
Fish length and weight in the two experiments were not very different (Table 1). However, the
cod fed capelin for six months in experiment 1 had a rather high condition factor (K is 1.1)
compared to the condition factor (K is 0.8) of the starved cod in experiment 2 (Table 1).
Table 1. Fish length (cm), weight not gutted (g), Fultons condition factor (K-factor), pH in blood and muscle
measured at time of slaughter, average values and standard deviations (N = 10).
Parameters measured at slaughter Cod fed capelin (Experiment 1) Cod not fed (Experiment 2)
Fish length (cm) 74.1 ± 6.0 83.3 ± 7.5

Fish weight, not gutted (g) 4933 ± 1210 4820 ± 1530
K-factor (not gutted) 1.1 ± 0.1 0.8 ± 0.1
pH in the blood of live fish 7.6 ± 0.1 7.8 ± 0.1
pH in the muscle at death 7.7 ± 0.2 7.8 ± 0.1
binnenwerk.indd 151 21-08-2006 12:34:27

pH at slaughter and post rigor

Average pH values measured in blood and muscle immediately after death were slightly higher
in the starved fish in experiment 2 (Table 1). In both experiments the pH values measured
at slaughter (Figure 1) show some variation but are well within normal levels for rested, not
stressed cod (Kristoffersen and others 2006). The low final muscle pH 6.2 ± 0.1 in experiment
1 and the higher final muscle pH 6.8 ± 0.1 in experiment 2 correspond well with previously
reported values for fed cod and starved cod (Rustad 1992; Akse and others 1997). Esaiassen
and others (2004) showed that the pH post rigor and K-factor were negatively correlated in
cod; high K-factor supported low pH. The low K-factor probably explains the significantly higher
post rigor pH found in wild cod that had not been fed in experiment 2 (Figure 1). Only minor
variation in post rigor pH due to time of filleting was observed in the two experiments.
Drip loss (weight reduction)

There was a significantly higher weight loss of fillets from the well fed cod in experiment 1
during six days of chilled storage, indicating a higher drip loss compared to the fillets from
the starved cod in experiment 2 (Figure 2). Average muscle pH measured in experiment 1 was
low compared to experiment 2 (Figure 1). Within each of the experiments, there were small
differences in weight reduction during storage of fillets processed pre, in or post rigor, except
for the in rigor processed fillets in experiment 1 which tended to loose less weight (Figure 2).
8.0
7.8 Pre rigor
In rigor
7.6
Post rigor
7.4
pH in fish muscle
7.2
7.0
6.8
6.6
6.4
6.2
6.0
captured wild cod fed capelin captured wild cod starved
6 days after filleting
Figure 1. Muscle pH measured 6 days after filleting. The 3 batches in each of the two experiments were
filleted pre rigor (<3 hours post mortem), in rigor (2 days post mortem) or post rigor (4 days post mortem)
(N = 5).
binnenwerk.indd 152 21-08-2006 12:34:27

12 Pre rigor
In rigor
10 Post rigor
8
Weight loss %
0
Figure 2. Drip loss measured as weight reduction 6 days after filleting, calculated in % of the initial fillet
weight at slaughtering. The 3 batches in each of the two experiments were filleted pre rigor (< 3 hours post
mortem), in rigor (2 days post mortem) or post rigor (4 days post mortem) (N = 10).
Fillet contraction
In both experiments average length reduction was significantly higher in pre rigor processed
fillets compared to fillets processed in rigor or post rigor. As can be seen from Figure 3 there
was 14% average length reduction of pre rigor processed fillets in experiment 1 (wild cod fed
capelin) and 10% length reduction in experiment 2 (wild cod not fed) after 6 days storage at
0 ºC. Fillets processed in or post rigor showed hardly any reduction in length at all (Figure 3).
Skjervoll and others (2002) found that pre rigor filleting of salmon significantly affected fillet
length and thickness, and resulted in an average final length reduction of 8.6%. Sørensen and
others (1997) reported 15% length reduction of pre rigor processed salmon fillets, maximum
contraction was reached after 12 hours when fillets were stored at 20 ºC and after close to
40 hours when stored at 0 ºC and 10 ºC. Mørkøre and others (2004) observed contractions
immediately after filleting of pre rigor farmed cod, and in that study maximum contraction was
reached after 16 hours for fillets stored at 20 ºC (28% length reduction), and after 30 hours for
fillets stored at 0 ºC (20% length reduction). Mørkøre (2006) found 21% contraction of pre rigor
processed farmed cod fillets, and the contraction rate was faster in fillets from cod fed a diet
containing 60% fish oil and 40% soybean oil, compared to a diet containing 100% fish oil.
Gaping and texture

Gaping and texture were evaluated 6 days after filleting. This measurement day was chosen
since six days post filleting is within the normal distribution and sale time that is needed to
supply major European consumer markets with chilled cod fillets processed in Northern Europe
binnenwerk.indd 153 21-08-2006 12:34:28

20 Pre rigor
Fillet contraction (% length reduction)

18 In rigor
Post rigor
16
14
12
10
8
6
4
2
0
Figure 3. Fillet contraction measured as reduction of fillet length 6 days after filleting, calculated in % of
initial fillet length at slaughtering. The 3 batches in each of the experiments were filleted pre rigor (< 3
hours post mortem), in rigor (2 days post mortem) or post rigor (4 days post mortem) (N = 10).
(Norway). In traditional post rigor processing, the fillets could be 9 – 10 days old when sold,
compared to only 6 days when processed pre rigor.
Fillet gaping was graded from score 0 (no gaping) to score 3 (disintegrating, strong gaping) by
visual inspection. In both experiments it was found that pre rigor filleting significantly reduced
gaping scores, compared to post rigor filleting. Six days after processing the fillets from the
cod that had been fed capelin in experiment 1 obtained higher gaping scores (more gaping),
compared to the starved cod in experiment 2 (Figure 4). Post rigor muscle pH was low in the
cod in experiment 1 (Figure 1). The mechanisms behind gaping are not well documented but it
is known that the strength of the myocommata in cod and salmon can be profoundly weakened
by low post rigor pH (Love and others 1972; Lavéty and others 1988). Førde-Skjærvik and others
(2006) found significant but limited influence of muscle pH on gaping score in farmed cod
fillets. Love and Haq (1970) considered muscle pH to be the major factor regarding gaping,
but other factors influencing gaping in fillets are probably also present (Love and Haq 1970;
Fletcher and others 1997).
Texture was evaluated by finger pressure and graded from score 0 (normal texture) to score
3 (very soft). In both experiments in rigor and post rigor processed fillets were evaluated to
be more soft 6 days post filleting, compared to the pre rigor processed fillets (Figure 5). The
fact that the in and post rigor batches were older post mortem and obtained higher gaping
scores partly explain this. However, the significant higher contraction (shrinkage) of pre rigor
processed fillets is probably the most important reason behind the observed differences in
texture.
binnenwerk.indd 154 21-08-2006 12:34:28

3 Pre rigor
In rigor
2.5 Post rigor
2
Gaping score
1.5
0.5
0
Figure 4. Gaping score determined by visual inspection at day 6 after filleting. Gaping score 0 denotes no
gaping while score 3 denotes extreme gaping. The 3 batches in each experiment were filleted pre rigor (<
3 hours post mortem), in rigor (2 days post mortem) or post rigor (4 days post mortem) (N = 10).
3 Pre rigor
In rigor
2.5 Post rigor
Texture score
1.5
0.5
0
Figure 5. Texture score determined 6 days after filleting by pressing a finger into the cut side of the fillet.
Gaping score 0 denotes firm, natural texture while score 3 denotes very soft texture. The 3 batches in each
experiment were filleted pre rigor (< 3 hours post mortem), in rigor (2 days post mortem) or post rigor
(4 days post mortem) (N = 10).
binnenwerk.indd 155 21-08-2006 12:34:28

Shelf life
The analyses of total viable counts (TVC) and sulphide-producing bacteria in experiment 2 aimed
at evaluating how time of filleting affected the shelf life of the chilled cod fillets. Norwegian
food safety regulations establish TVC 5.0x105 CFU/g as a recommended limit that should not
be exceeded in seafood like raw lean fish or fish fillets, chilled or frozen. The same regulations
establish TVC > 5x106 CFU/g as an absolute limit where such seafood no longer is allowed for
human consumption (Anon 2004).
However, only a few microbial organisms give rise to off-flavours associated with seafood
spoilage. The specific spoilage organisms (SSOs) are typically present in low numbers (Gram
and Dalgaard 2002). Detection of SSOs is therefore more appropriate than TVC when assessing
fish quality. The fishy, rotten H2S-off-odours in spoiled, aerobically stored fish from cold and
temperate waters are generally caused by sulphide-producing bacteria (SPB), mainly Shewanella
putrefaciens.
As can be seen from Figure 6, pre rigor filleting reduced the total post mortem shelf life,
compared to in rigor and post rigor filleting. A probable explanation is that during in or post
rigor processing the raw material was stored as intact whole fish for 2 and 4 days post mortem
before filleting exposed the interior of the muscle to microbial contamination. At day 12 post
mortem, the average TVC of pre rigor processed fillets was 1.5x106, of in rigor processed fillets
5.1x105 and of post rigor processed fillets only 9.6x104.
However, regarding the shelf life of the products after filleting (Figure 7), the pattern is
different. As mentioned above, fresh cod fillets that are distributed from processors in Norway
to major seafood markets in Europe usually will be presented for sale to consumers 4 – 8 days
after processing. At day 4 after filleting TVC were < 1x103 CFU/g in all the three parallels (pre, in
and post rigor), which indicates high freshness. At day 8 after filleting, average TVC in post rigor
1.0E+07
Pre rigor
1.0E+06 In rigor
TVC (CFU/g muscle)
Post rigor
1.0E+05
1.0E+04
1.0E+03
1.0E+02
0 2 4 6 8 10 12
1.0E+01
1.0E+00
Days post mortem
Figure 6. Total viable counts (TVC) in raw cod fillets processed either pre, in or post rigor, stored chilled on
ice (≈ 0 ºC) for 12 days after slaughtering. Values at CFU=100 indicate CFU values below detectable level
(N = 5).
binnenwerk.indd 156 21-08-2006 12:34:29

1.0E+07
Pre rigor
1.0E+06 In rigor
TVC (CFU/g muscle) Post rigor
1.0E+05
1.0E+04
1.0E+03
1.0E+02
0 2 4 6 8 10 12
1.0E+01
1.0E+00
Days after filleting
Figure 7. Total viable counts (TVC) in raw cod fillets processed either pre, in or post rigor, stored chilled on
ice (≈ 0 ºC) after filleting. Values at CFU=100 indicate CFU values below detectable level (N = 5).
fillets already reached 9.6x104 CFU/g, while average TVC in the pre rigor processed fillets still
was 1.6x103 CFU/g, which can be denoted as high freshness. Figure 8 shows that towards the
end of the storing time, sulphide-producing bacteria are rapidly becoming a major component
in the micro flora of the products.
Sulphide-producing bacteria (CFU/g muscle)
1.0E+07
Pre rigor
1.0E+06 In rigor
Post rigor
1.0E+05
1.0E+04
1.0E+03
1.0E+02
0 2 4 6 8 10 12
1.0E+01
1.0E+00
Days after filleting
Figure 8. Sulphide-producing bacteria in raw cod fillets processed either pre, in or post rigor, stored chilled
on ice (≈ 0 ºC) after filleting. Values at CFU=100 indicate CFU values below detectable level (N = 5).
binnenwerk.indd 157 21-08-2006 12:34:29

Conclusions
Fillets from wild cod fed capelin in experiment one obtained a low ultimate muscle pH and high
weight reduction (drip loss) during storage, compared to starved cod in experiment two. Within
each of the experiments, only minor variations in pH and weight reduction were registered
dependent on whether the fillets were processed pre, in or post rigor. Rigor state when filleted
significantly affected contraction of the fillets post processing. Average reduction in length
of pre rigor processed fillets varied from 10% to 14%, the highest reduction in fillets from fed
cod in experiment one. None or minor contractions were observed in fillets processed in or post
rigor. Pre rigor processed fillets in both experiments had less gaping and a more firm texture,
compared to in rigor and post rigor processed fillets. The bacterial growth after filleting was
slower in pre rigor processed fillets compared to fillets processed from in rigor and post rigor
raw material that had been stored on ice for 2 and 4 days prior to filleting. Low rate of gaping
in the fillets, better texture and prolonged shelf life after filleting all leads to the conclusion
that pre rigor processing is a better concept for distribution and sale of high quality fresh cod
fillets, compared to traditional post rigor processing.
Acknowledgements
The present work has been financially supported by the Norwegian Research Council. Also the
company Gunnar Klo AS and the fishermen at Stø in Vesterålen are thanked for good and skilful
cooperation.
References
Akse L, Midling K. 1997. Live capture and starvation of capelin cod (Gadus morhua L.) in order to improve the quality.
In: Luten JB, Børresen T, Oehlenschlager J, editors. Seafood from Producer to Consumer, Integrated Approach to
Quality. Amsterdam: Elsevier. p 47-58.
ANON. 2004. Norwegian Food Safety Authority: Microbiological Guidelines, Fish Products.
Gram L, Dalgaard P. 2002. Fish spoilage bacteria – problems and solutions. Curr Opin Biotech 13:262-266.
Esaiassen M, Nilsen H, Joensen S, Skjerdal T, Carlehøg M, Eilertsen G, Gundersen B, Elvevoll E. 2004. Effects of catching
methods on quality changes during storage of cod (Gadus morhua L.). Lebensm Wiss Technol 37:643-648.
Fletcher GH, Hallet IC, Jerret AR, Holland A.J. 1997. Changes in the fine structure of the myocommata-muscle fibre
junction related to gaping in rested and exercised muscle from king salmon (Oncorhynchus tshawytscha). Food Sci
Technol-Leb 30:246-252.
Førde-Skjærvik O, Skjærvik O, Mørkøre T, Thomassen MS, Rørvik K-A. 2006. Dietary influence on quality of farmed Atlantic
cod (Gadus morhua): Effect on glycolysis and buffering capacity in white muscle. Aquaculture. Forthcoming.
Kristoffersen S, Tobiassen T, Steinsund V, Olsen R. 2006. Slaughter stress, post-mortem muscle pH and rigor development
in farmed Atlantic cod (Gadus morhua L.). Int J Food Sci Tech. Forthcoming.
Lavéty J, Afolabi OA, Love RM. 1988. The connective tissues of fish: IX. Gaping in farmed species. Int J Food Sci Tech
23:23-30.
Love RM, Haq MA. 1970. The connective tissue of fish III. The effect of pH on gaping in cod entering rigor mortis at
different temperatures. J Food Tech 5:241-248.
Love RM, Lavéty J, Garcia NG. 1972. The connective tissues of fish VI. Mechanical studies on isolated myocommata.
J Food Techol 7:291-301.
Mørkøre T, Hansen SJ, Rørvik K-A. 2004. Quality of pre rigor cod fillets. Storage temperatures affect contraction and
gaping. Poster presentation. Program conference (NRC), Fish farming 23-24 March. Gardermoen. In Norwegian.
binnenwerk.indd 158 21-08-2006 12:34:29

Mørkøre T. 2006. Relevance of dietary oil source for contraction and quality of pre rigor filleted Atlantic cod, Gadus
morhua. Aquaculture 251:55-65.
Ofstad R, Kidman S, Hermanson AM. 1996. Ultramicroscopial structures and liquid loss in heated cod (Gadus morhua)
and salmon (Salmo salar) muscle. J Sci Food Agric 72:337-347.
Rustad T. 1992. Muscle chemistry and quality of wild and farmed cod. In: Huss HH, Jakobsen M, Liston J, editors.
Quality Assurance in the Fish Industry. Amsterdam: Elsevier. p 19-27.
Rørå AMB. 2003. Raw material characteristics and treatment – effects on yield and quality of cold-smoked Atlantic
salmon. PhD thesis. Norwegian University of Life Science, Ås, Norway. ISBN 82-575-0564-1.
Skjervold PO. 2002. Live chilling and pre rigor filleting of salmonids – Technology affecting physiology and product
quality. PhD thesis. Norwegian University of Life Science, Ås, Norway.
Sørensen NK, Brataas R, Nyvold TE, Lauritzen K. 1997. Influence of early processing (pre rigor) on fish quality. In: Luten
JB, Børresen T, Oehlenschlager J, editors. Seafood from Producer to Consumer, Integrated Approach to Quality.
Amsterdam: Elsevier. p 253-263.
Tobiassen T, Olsen JV, Akse L. 2003. Prosessering av pre- og post-rigor ørret, effekter i produksjon av røkte produkter.
Report 5/2003, Norwegian Institute of Fisheries and Aquaculture Research.
binnenwerk.indd 159 21-08-2006 12:34:30

binnenwerk.indd 160 21-08-2006 12:34:30
Gelatinolytic activity in muscle of farmed and wild Atlantic cod
(Gadus morhua) related to muscle softening
Gunn Berit Olsson1, Marie Cooper1, Tone Jakobsen Friis1 and Ragnar L. Olsen2
2The Norwegian College of Fishery Science, University of Tromsø, N-9037 Tromsø, Norway
Abstract
Farmed cod differ in several respects from the wild cod. One observation is that muscle of raw,
farmed cod appears to be softer and to lose its integrity faster than that of the wild fish. By
using zymography we examined if these differences were due to differences in gelatinolytic
activity between the farmed and the wild fish. Possible correlations between the content of
trace metals in the muscle and the gelatinolytic activity were also investigated. The activity of
matrix metalloproteinase – 2 (MMP-2) was higher in the muscle of wild cod then in farmed cod.
High MMP-2 activity corresponded to low content of the selected trace metals and vice versa.
Keywords: farmed/wild cod, matrix metalloproteinases, trace metals, muscle softening
Introduction
The large-scale production of cod in aquaculture is a relatively new and developing industry. Cod
is a highly prized, premium food fish popular throughout Scandinavia and the rest of Europe.
Cod farming has experienced a variety of problems during the course of development. In several
respects, the producers have failed to attain the quality standards required by both the industry
and the consumer. It has been observed that farmed cod differ in numerous respects from their
wild counterparts. The central focus in this project has been the differences in muscle texture
between farmed and wild cod. Muscle of farmed cod appears to be softer and to lose its integrity
faster than that of the wild fish.
Published results about the post mortem textural changes in fish have suggested several
mechanisms explaining muscle softening, but still these are not fully understood. The muscle
structure is complex and we do not know the location, concentration and function of many
minor components responsible for muscle integrity. For example, Brüggemann and Lawson
(2005) have only recently reported the presence of types III, V VI and IV collagen in addition
to type I in cod muscle after this work was completed. Differing mechanisms have been
proposed according to the techniques used and the different species studied by researchers.
Recent publications indicate that the softening of cod muscle during ice storage is caused
more by collagenolytic activity (Hernandez-Herrero and others 2003) or by degradation of
fibre to connective tissue attachment in salmon (Taylor and others 2002) than by proteolysis
of myofibrils. Observations by Hatae and others (1989), Busconi and others (1989), Bremner
(1992) and Hernandez-Herrero and others (2003) indicate that the myofibrillar proteins are
largely unaffected during post mortem storage. They suggest that the softening of muscle is
not caused by the breakdown of myofibrils, but is probably due to proteolytic digestion of
minor cell components that link the major structural units together. Additionally it is likely that
binnenwerk.indd 161 21-08-2006 12:34:30

changes in the elements of the cytoskeleton weaken the overall structure (Busconi and others
1989; Morrison and others 2000). It has been shown that gradual disintegration of extracellular
matrix (ECM) structure between muscle fibres may be of major importance for the post mortem
softening of fish muscle (Ando and others 1991b; Bremner 1999; Kubota and others 2001).
This disintegration is probably related to degradation of proteoglygans (PG’s) that link the
collagen fibrils and stabilise the structure rather than degradation of the collagen fibres as
such (Nishimura and others 1996).
One structural alteration observed post mortem in soft muscle is gaps in the extracellular matrix
(ECM) (Olsson and others 2003) a problem of considerable commercial importance. This may be
a result of ECM degradation which has been reported in fish (Bremner and Hallet 1985; Ando
and others 1991a). Previous work has shown that this degradation proceeds faster in farmed
than in wild fish (Ofstad and others 1996; Olsson and others 2003). Furthermore, it has been
proposed that degradation of proteoglycans in the ECM, involved in linking the collagen fibrils,
may be a major factor in weakening of the intramuscular connective tissue (Kim and Haard
1992; Nishimura and others 1996). The degradation of ECM components, including collagen, has
been suggested to result from activity of matrix metalloproteinases (Bracho and Haard 1995;
Sato and others 1997; Saito and others 2000a; 2000b; Kubota and others 2003) and probably
matrix serine proteases are also involved (Kubota and others 2001; Lødemel and Olsen, 2003).
Matrix metalloproteinases (MMP’s) are calcium- and zinc-dependent neutral/alkaline proteinases
with gelatinolytic activity that play an important role in extracellular matrix (ECM) turnover
(Woessner 1991; Birkedal-Hansen and others 1993; Massova and others 1998). The MMP’s are
secreted into the extracellular space as proenzymes containing a regulatory pro-peptide motif.
They are kept inactive by an interaction between a cystein-thiol (SH) group in the propeptide
domain and a zinc ion in the catalytic domain. The presence of MMP’s has been described in
muscle of Atlantic cod, spotted wolffish and Atlantic salmon (Lødemel and Olsen 2003). The
ECM is primarily a collection of fibrous proteins embedded in a hydrated polysaccharide gel. The
macromolecules (collagens and glucosaminoglycans) which make up the ECM are secreted at a
local level by cells, in particular fibroblasts. The role of the matrix is to provide a physiological
framework for the cells that are responsible for its production. Gelatinolytic enzymes in fish are
of interest due to their ability to degrade ECM components, and hence the possibility that they
contribute to early post mortem degradation of fish muscle (Bremner and Hallet 1985; Hallet
and Bremner 1988; Ando and others 1991b; Lødemel and Olsen 2003). MMP’s are known to play
an important role in the ECM-turnover in general (Woessner 1991; Birkedal-Hansen and others
1993; Massova and others 1988). Despite the very limited information currently available on
MMP’s and their inhibitors in skeletal muscle, it is becoming increasingly clear that they have
important physiological functions in maintenance of the integrity and homeostasis of muscle
fibres and of the extracellular matrix (Carmeli and others 2004). As a group, the MMP’s are
able to cleave all the structural components of the ECM in their native forms at neutral pH.
Maintenance of ECM integrity appears to be their primary function. The two gelatinases, MMP-
2 and MMP-9, bind strongly to gelatine due to fibronectin type II domains in their catalytic
domain (Strongin and others 1993). Gelatin SDS-PAGE has been shown to be a valuable tool for
investigating the presence of MMP’s and other proteinases with gelatinolytic activity.
Many of the enzymes involved in extracelluar matrix synthesis and degradation are metal
dependent (Pena and others 1999; Shim and Harris 2003). In general, trace metals are essential
for health, forming integral components of proteins involved in all aspects of biological function.
However, in excess they are toxic as they may bind inappropriately to biologically sensitive
binnenwerk.indd 162 21-08-2006 12:34:30

Gelatinolytic activity in muscle of farmed and wild Atlantic cod related to muscle softening
molecules or form dangerous free radicals. Trace metal availability is influenced by a number
of factors including the levels of other micronutrients in feed, concentrations in water, tissue
distribution and mechanisms of absorption and catabolism. Information exists regarding the
requirements for trace metals in some fresh water species (Handy and others 2000; Kamunda
and others 2000; Gaitlin and Wilson 1986), but very little research has been done into the
precise nutritional requirements of the marine species now being domesticated (Young Cho and
others 1993). The levels and tissue distribution of micronutrients in free living species have
been published (Falandysz and Lorenc-Biaka 1984; Teeny and others 1984; Márquez and others
1998; Shiau and others 2003; Celik and Oehlenschläger 2004). The main source for essential
metals is the diet, but in aquatic organisms an alternative uptake route is available from the
water (Bury and others 2003).
Commercially available cod feed contains high levels of trace metals. There are indications
that deposition of melanin in blood vessels (black stripes in the fillets) observed especially in
farmed cod, are related to high levels of copper which again corresponds to high copper levels
in the diet (Cooper and Midling 2005). Since several trace metals are able to displace each
other in enzymatic reactions it is possible that such displacement can take place in the MMPs,
with effects on enzyme activity.
Accordingly, this project had two objectives:

1. To investigate possible differences in the activity of matrix metalloproteinases in farmed
versus wild Atlantic cod that could explain the differences in muscle texture between the
two groups.
2. To investigate whether differences in the enzyme activity could be related to the levels
of trace metals present in the muscles, since the matrix metalloproteinases are metal
dependent.
Fish
Farmed Atlantic cod of 2663 ± 462 g was obtained from the Aquaculture station at Skulgambukt,
Tromsø. The fish were killed by a blow to the head, bled and transported rapidly, in ice, to the
institute for sampling. The wild cod (4750 ± 870 g) were caught by long line and live-stored
for a few days at the aquaculture station before slaughtering. The wild fish were slaughtered
in the same way as the farmed cod. Muscle samples were cut approximately 3 to 4 hours after
slaughter.
Muscle extraction
Muscle samples from ten farmed and six wild cod were dissected from the sixth muscle segment
of the fillet, counted from the head region. Duplicates of 200 mg of muscle were added to 2 ml
cold extraction buffer (50 mM Tris-HCl, pH8.0, 10 mM CaCl2, 2.5g l-1 Triton X-100 and 0.2 g l-1
NaN3) and mixed thoroughly for 2 x 30 sec by an Ultra Turrax. The extracts were transferred to
Eppendorf tubes and centrifuged (10,000 x g, 4 °C) for 10 min. The last step was repeated once.
The extracts were stored at –80 °C prior to use. To ensure applying equal amounts of protein
to each well of the gel, protein concentrations in the muscle extracts were determined using a
NanoDrop®3.1.0 ND 1000 spectrophotometer (Saveen Werner).
binnenwerk.indd 163 21-08-2006 12:34:30

Gelatin SDS polyacrylamide gel electrophoresis (zymography)

Substrate gel electrophoresis was performed according to Lødemel and Olsen (2003). Samples
of the muscle extracts were thawed on ice and diluted by an equal amount of non-reducing
sample buffer (62.5 mM Tris-HCl, pH 6.8, 4% SDS, 35% glycerol and 0.01% bromphenol blue).
Non-heated, non-reduced samples (20 µl) were applied to the wells of the SDS polyacrylamide
gels (50 g l-1 polyacrylamide in the stacking gel and 90 g l-1 polyacrylamide in the separation
gel); the separation gel contained 1 mg ml-1 gelatin. The gels were run for 30 min at 100 V
followed by 75 min at 150 V in a Mini-PROTEAN 3 electrophoresis cell (Bio-Rad) placed in an ice
bath. After electrophoresis the gels were washed in 25 g l-1 Triton X-100 for 2 x 15 min prior to
incubation in 50 mM Tris-HCl, pH 8.0, 10 mM CaCl2 and 0.2 g l-1 NaN3 for 20 h at 37 °C. After
incubation the gels were stained in 1.0 g l-1 Coomassie Brilliant Blue in 450 ml l-1 methanol and
100 ml l-1 acetic acid and destained in 100 ml l-1 methanol and 100 ml l-1 acetic acid. From each
individual two extract samples were run twice on two different gels. The gelatinolytic activities
were identified as clearing zones. The net intensity and areas of the band were analysed using
a Kodak DS Image station 440 CF scientific imaging system.
Metal analysis
Samples for metal analysis were prepared according to the method recommended by the
Norwegian committee on food analysis (No 161, 1998). Briefly, muscle samples (dissected
from the same area as for muscle protein extracts as in section 2.2) were weighed into Teflon
tubes containing 4 ml 25% hydrochloric acid (Merck) and 2 ml 30% H202 (Sigma) and digested
by boiling in a Perkin-Emer multiwave microwave oven. Extracts were removed from the Teflon
tubes and filtered through ashless filter circles. Metal concentrations were determined using
a Perkin–Elmer 3110 flame atomic absorption spectrophotometer and a standard curve using
Tritisol (Merck) metal standard solutions diluted in extraction solution was done with each
analysis. Extractions were performed in duplicate. Blanks containing extraction solution without
sample and samples of a standard (Merck Titrisol standards of Cu, Zn, Fe, Mg or Ca) were
extracted in parallel with experimental samples and measured in the course of each assay.
A t-test was used to determine significant differences in trace metal content between the
farmed and the wild cod. The data were analysed by use of multivariate techniques (PLS1-
regression), applying the software Unscrambler version 9.1.2 (CAMO Process AS, Oslo, Norway).
The variables were weighted with the inverse of the standard deviation of all the objects in
order to compensate for the different scales of the variables.
Gelatin SDS polyacrylamide gel electrophoresis (zymography)

The presence of MMP’s in fish tissue has been described by using zymography (Bracho and Haard
1995; Kubota and others 1998a, 1998b, 2000; Lødemel and Olsen 2003). Gelatin Sepharose 4B
has also been shown to be an effective tool for separating gelatine degrading enzymes in fish
muscle (Lødemel and Olsen 2003). In Figure 1 and 2 the zymography gels on muscle extracts
from 6 wild and 10 farmed cod, respectively are shown. Gelatinolytic activities are recognised
as clearing zones in the gels. Purified MMP-2 and MMP-9 were used as positive controls.
Visual analysis of the zymograms shows that the pattern in the clearing zones was in several
respects different between the farmed and the wild cod. The zymograms show the presence of
gelatinolytic enzymes of 225, 92, 72 and 62 kDa in muscle extracts of both farmed and wild cod.
binnenwerk.indd 164 21-08-2006 12:34:30

1 2 3 4 5 6 7 8 9
225kDa
92kDa
72kDa
62kDa
Figure 1. Gelatin zymogram showing the presence of gelatine degrading enzymes in muscle of wild cod.
Lane 1: prestained standards (kDa), lane 2-7: individuals of wild cod; lane 8: MMP-2; lane 9: MMP-9.
1 2 3 4 5 6 7 8 9 10 11 12 13
225kDa
92kDa
72kDa
62kDa
Figure 2. Gelatin zymogram showing the presence of gelatine degrading enzymes in muscle of farmed cod.
Lane 1: prestained standards (kDa), lane 2-11: individuals of farmed cod; lane 12: MMP-2; lane 13: MMP-9.
This correlates well with the results of Lødemel and Olsen (2003) and probably represents the
homodimer of MMP-9 (225kDa), the pro-MMP-9 (92kDa), the pro-(72kDa) and active (62kDa)
form of MMP-2. In addition to the MMP’s, matrix serine proteinases may also play an important
role in degradation of the ECM (Koshikawa and others 1992). Some of the bands that did not
correspond to the controls (MMP-2 and MMP-9) could therefore be serine proteinases.
binnenwerk.indd 165 21-08-2006 12:34:31

The intensity of the 72kDa band appeared to be higher in the wild cod muscle compared to
the muscle of farmed cod. The activity in the 225 and 92kDa band appears to vary greatly
both between the two groups and within the groups. The individual differences were highly
reproducible. Several authors have reported increased proteolytic activity in fish muscle during
sexual maturation (Yamashita and Konagaya 1990; Toyohara and others 1991; Kubota and others
2000). In our experiment both the wild and the farmed cod had reached sexual maturation
and this could not explain the different enzyme activity between the farmed and the wild fish.
Differences in gelatinolytic activity could be related to the level of copper and / or other trace
metal in the muscle. The reactive nature of ionic copper makes it a toxic metal if not properly
handled by the cells (Rae and others 1999). Under conditions of copper overload, free copper
ions can accumulate and react to generate hydroxyl radicals that can engage in reactions that
can adversely modify proteins, lipids and nucleic acids (Halliwell and others 1984). Free copper
ions may also exert their toxic property by displacing other essential metal co-factors (for
example Zn) from metalloenzymes. It has been reported that copper can substitute for Zn(II)
in zinc-finger transcription factors, a substitution that renders the proteins unable to bind to
their target sequence (Predki and Sarkar 1992).
Crude extracts from muscle of both farmed and wild cod were also purified by using gelatine
Sepharose 4B as affinity matrix (data not shown). From this purification we could conclude
that our 72 kDa band corresponded to MMP-2 and was therefore used as the effect variable in
the statistical analysis described further down.
Metal analysis
The levels of copper, zinc, iron, magnesium and calcium were measured in muscle extracts from
9 wild and 10 farmed cod (including the same individuals as used in the zymografy) and in 15
different commercial cod feeds. The results are given in Table 1. Except for iron, the levels of
all the trace metals were significantly higher in farmed then in wild cod. The commercial feed
contains high levels of all the assessed trace metals and may thus explain the differences in
trace metal content between the farmed and the wild fish. Aquatic animals can take up trace
metals from sea water, but the main source is the diet (Bury and others 2003). There is very
little knowledge about the real dietary requirements, mineral status and metal metabolism of
farmed fish in general. Some requirements for channel catfish, rainbow trout, Pacific salmon,
common carp and tilapia are presented by Young Cho and others (1993) (see Table 1). For cod
no such information is available. We may infer from the information given in Table 1 that trace
mineral nutrition in farmed cod is not as close to the natural situation as it should be.
Correlations between trace metal content and gelatinolytic activities in the muscle
To investigate possible correlations between the trace metal levels and the MMP-activities, a
partial least squares regression (PLS1) was used. By image analysis (Kodak DS Image station
440 CF) the clearing zones in the zymografy gels were quantified by net intensity and the
MMP-2 band was selected as the Y-variable. The measured trace metal levels were used as the
X-variables. The correlation loadings of two principal components (PC’s), representing 59%
and 20% of the explanation of X and 57% and 16% of the explanation of Y, are presented in
Figure 3a. The correlation loadings plot shows that the variations in Cu, Zn, Mg and Ca content
are mainly explained by PC 1, and that they are correlated to each other. This supports the fact
that trace metals work interactively and may even replace each other. The level of iron is mainly
explained by PC 2. Furthermore, the plot shows that content of Cu, Zn, Mg and Ca are negatively
correlated to the MMP-2 activity. In other words, muscle containing high levels of the measured
binnenwerk.indd 166 21-08-2006 12:34:32

Table 1. Mean values and standard deviation of trace metal content in muscle extracts from wild and farmed
cod, commercial cod feed and sea water (mg/l).
Cu (mg/kg) Zn (mg/kg) Fe (mg/kg) Mg (mg/kg) Ca (mg/kg)
Wild cod (n=9) 0.34 ± 0.4a 4.4 ± 2.8a 11.9 ± 12.5a 204 ± 16a 129 ± 84a
(p<0.01) (p<0.001) (n.s.) (p<0.001) (p<0.001)
Farmed cod (n=10) 0.50 ± 0.15b 11.6 ± 3.0b 10.9 ± 8.3a 248 ± 20b 524 ± 186b
Commercial cod feed 18.5 223 183 1009 15344
mg/kg (n=15)
Sea water mg/l 0.0003 0.01 0.01 1350 450
Requirements mg/kg 3 30 30 500 n.a.
of diet
Significant differences between wild and farmed cod are indicated with letter superscript. Samples with
different letters in the same column are significantly different.
Significance are identified by a t-test and are given between brackets. n.s.= not significant
n.a.= not analysed
copper, zinc, magnesium and calcium showed suppressed activity of MMP-2. The corresponding
score plot from the PLS1 regression is given in Figure 3b. The farmed cod samples, with only
one exception, appear on the left part of the plot and are thus characterised by high levels of
the trace metals and low enzyme activity. The wild samples on the right-hand side of the plot
are thus characterised by low levels of trace metals and high enzyme activity.
Correlation loadings (X and Y)

PC2
1.0
+ Fe
0.5
+ Cu + 72 kD
+ Zn
0
+ Ca
+ Mg
-0.5
-1.0
PC1
-1.0 -0.5 0 0.5 1.0
Figure 3a. Correlation loading plot from PLS1-regression of trace metals and intensity of the 72kDa-band
(MMP-2). The outer and the inner ellipse indicate 100% and 50% explained variance, respectively.
binnenwerk.indd 167 21-08-2006 12:34:32

Scores
PC2
3
+ SK B9
2
+ WF B4
1
+ SK B5
+ SK B8
+WFWF
B1
B3+
0 +SK B7 + WF B2
+ SK B10
+ SK B4
+ WF B5
-1 + SK B3 + SK B2
+ SK B1
+ SK B6
-2
PC1
-3 -2 -1 0 1 2 3
Figure 3b. Score plot from PLS1-regression of trace metals and intensity of the 72kDa-band (MMP-2). SK
B1-10 = farmed cod samples WF1-5 = wild cod samples.
The results of this experiment show that there are differences in gelatinolytic activity between
the muscle of farmed and of wild cod. Whether these differences can be related to the observed
differences in muscle texture is still not clear. Farmed and wild fish live under quite different
conditions. This is a challenge when designing experiments to compare the two groups. Farmed
fish have a much more consistent history than their wild counterparts. For example, their
age is known and they have all been exposed to the same environmental conditions. The
most prominent difference is however, that farmed fish are provided with a constant supply
of nutrient-dense formulated feed. The wild fish are subjected to considerable environmental
variation and fluctuations in both the availability and composition of feed. It is likely that
the main source for the differences between farmed and wild fish is to be found in the food
intake. The results from this experiment showed that the farmed fish received high levels of
trace metals via the feed and this corresponded with high levels in the muscle. The decreased
MMP-2 activity could be explained by alterations in the enzyme structure and / or inhibitors
due to interactions from excess incorrect trace metals. However, the actual activities of MMP’s
are under complex pre- and post-translational regulation including activation of proMMP’s
to MMP’s by enzymatic cleavage and inhibition by tissue inhibitors of the MMP’s (Gomez
and others (1997) which makes it difficult to identify possible influencing factors. It should
be emphasised that sampling of the fish in our experiment was done only 2 – 3 hours after
slaughter. Hence the results are probably more representative of ante-mortem differences than
post-mortem degradation. As the difference in muscle texture can be observed immediately
post mortem the differences observed between farmed and wild cod muscle texture are probably
due to differences in the muscle structure of living fish. The synthesis of ECM components
may be lower in the farmed than in the wild fish. It has been shown in experiments with rats
that exercising increased both mRNA and protein levels of MMP-2 in skeletal muscle (Koskinen
binnenwerk.indd 168 21-08-2006 12:34:32

and others 2001; Carmeli and others 2005). Exercise has been shown to cause changes in
synthesis and degradation of basement membrane type IV collagen and to proteins regulating
its degradation in rat skeletal muscle (Koskinen and others 2001). The influence of exercise
on MMP-2 expression is apparently dominant in muscle containing a high percentage of fast
fibres (Carmeli and others 2005). It is possible that minimal exercise of the fish in captivity in
addition to feed related differences may generate alterations in the ECM components. Hence
the differences in post-mortem quality of farmed and wild fish are probably due to variations
in the physiological status of the living fish as much as they are due to post-mortem changes
in enzyme activity.
Conclusion
It is believed that the best approach to finding causes for the diet related differences observed
between farmed and wild cod is to identify which fundamental mechanisms are involved
rather than to search for dietary components. To further increase the understanding of these
mechanisms recently proteome analysis of the muscles of farmed and wild cod was introduced
Acknowledgements
The Norwegian Research Council is thanked for financial support.
References
Ando M, Toyohara H, Shimizu Y, Sakaguchi M. 1991a. Post mortem changes of fish muscle proceeds independently of
resolution of rigor mortis. Nippon Suisan Gakkaishi 57:1165-1169.
Ando M, Toyohara H, Shimizu Y, Sakaguchi M. 1991b. Post-mortem tenderization of rainbow trout (Oncohynchus mykiss)
muscle caused by gradual disintegration of the extracellular matrix structure. J Sci Food Agric 55:589-597.
Birkedal-Hansen H, Moore WGI, Bodden MK, Windsor LJ, Birkedal-Hansen B, Decarlo A, Engler JA. 1993. Matrix
metalloproteinases – a review. Crit Rev Oral Biol Med 4:197-250.
Bracho GE, Haard NF. 1995. Identification of two matrix metalloproteinases in the skeletal muscle of Pacific rockfish
(Sebastes sp). J Food Biochem 19:299-319.
Bremner HA and Hallet IC. 1985. Muscle fibre - connective tissue junctions in the fish blue grenadier (Macruronus
novaezelandiae). A scanning electron microscope study. J. Food Sci. 50(4):975-980.
Bremner H.A. 1992. Fish flesh structure and the role of collagens – its post-mortem aspects and implications for fish
processing. In: Huss H.H., Jakobsen M. and Liston J. (Eds.): Quality assurance in the fish industry. Amsterdam,
Elsevier Science Publishers B.V. pp. 39-62
Bremner HA. 1999. Gaping in fish flesh. In: K. Sato, Sakaguchi, M. and Bremner, H.A. (Eds) Extracellularmatrix of fish
and shellfish. Trivandrum, Research Signpost: 81-94.
Brügermann DA, Lawson MA. 2005. The extracellaular matrix of Gadus morhua muscle contains types III,V, VI and IV
collagens in addition to type I. J Fish Biol 66:810-821.
Bury NR, Walker PA, Glover CN. 2003. Nutritive metal uptake in teleost fish. J Exp Biol 206:11-23.
Busconi L, Folco EJ, Martone CB, Sanchez J. 1989. Post mortem changes in cytoskeletal elements of fish muscle. J
Food Biochem 13:443-451.
Carmeli E, Moas M, Reznick AZ, Coleman R. 2004. Matrix metalloproteinases and skeletal muscle: a brief review. Muscle
Nerve 29(2):191-197.
Carmeli E, Moas M, Lennon S, Powers SK. 2005. High intensity exercise increases expression of matrix metalloproteinases
in fast skeletal muscle fibres. Exp Physiol 90(4):613-619.
binnenwerk.indd 169 21-08-2006 12:34:32

Celik U, Oehlenschlager J. 2004. Determination of zinc and copper in fish samples collected from Northeast Atlantic
by DPSAV. Food Chem 87:343-347.
Cooper M, Midling KØ. 2005. Blood vessel melanosis: a physiological detoxification mechanism in Gadus morhua? In
Review, Aquaculture International.
Falandysz J, Lorenc-Biaka H. 1984. Trace metals in fish from the Southern Baltic. Meeresforsch 30:111-119.
Gatlin DM III, Wilson RP. 1986. Dietary copper requirements for channel catfish. Aquaculture 54:277-285.
Gomez DE, Alonso DF, Yoshiji H, Thorgeirsson UP. 1997. Tissue inhibitors of metalloproteinases: structure regulation
and biological functions. Eur J Cell Biol 74:111-122.
Hallet IC, Bremner HA. 1988. Fine structure of the myocommata – muscle fiber junction in hoki (Macruronus
novaezelandiae). J Sci Food Agric 44:245-261.
Halliwell B, Gutteridge JM. 1984. Oxygen toxicity, oxygen radicals, transition metals and disease. Biochem J 219:1-14.
Handy RD, Musonda MM, Phillips C, Falla SJ. 2000. Mechanisms of gastrointestinal copper absorption in the African
walking catfish: dose-effects and a novel anion-dependent pathway in the intestine. J Exp Biol 203:2365-2377.
Hatae K, Lee KH, Tsuchiya T, Shimada A. 1989. Textural properties of cultured and wild fish meat. Bull Jap Soc Sci Fish
55:363-368 (from English abstract).
Hernández-Herrero MM, Duflos G, Malle P, Bouquelet S. 2003. Collagenase activity and protein hydrolysis as related to
spoilage of iced cod (Gadus morhua). Food Res Int 36:141-147.
Kamunda CN, Grosell M, Lott JNA, Wood CM. 2000. Copper metabolism and gut morphology in rainbow trout (Oncorhynchus
mykiss) during chronic sublethal dietary copper exposure. Can J Fish Aquat Sci 58:293-305.
Kim K. and Haard NF. 1992. Degradation of proteoglycans in the skeletal muscle of pacific rockfish (Sebastes Sp) during
ice storage. J Muscle Foods 3:103-121.
Koshikawa N, Yasumitsu H, Umeda M, Miyazaki K. 1992. Multiple secretion of matrix serine proteinases by human
gastric-carcinoma cell-lines. Cancer Res 52:5046-5053.
Koskinen SOA, Wang W, Ahtikoski AM, Kjær M, Han XY, Komulainen J, Kovanen V, Takala TES. 2001. Acute exercise
induced changes in rat skeletal muscle mRNAs and protein regulating type IV collagen content. Am J Physiol
Regulatory Integr Comp Physiol 280:R1292-R1300.
Kubota S, Kinoshita M, Toyohara H, Sakaguchi M. 1998a. Gelatinolytic activities in ayu muscle in the spawning stage.
Fish Sci 64:1001-1002.
Kubota S, Toyohara H, Sakaguchi M. 1998b. Occurrence of gelatinolytic activities in yellowtail tissues. Fish Sci 64:439-
442.
Kubota S, Kinoshita M, Yokohama Y, Toyohara H, Sakaguchi M. 2000. Induction of gelatinolytic activities in ayu muscle
at the spawning stage. Fish Sci 66:574-578.
Kubota M, Kinoshita M, Kubota S, Yamashita M, Toyohara H, Sakaguchi M. 2001. Possible implication of metalloproteinases
in post-mortem tenderization of fish muscle. Fish Sci 67:965-968.
Kubota M, Kinoshita M, Takeuchi K, Kubota S, Toyohara H, Sakaguchi M. 2003. Solubilization of type I collagen from
fish muscle connective tissue by matrix metalloproteinase-9 at chilled temperature. Fish Sci 69:1053-1059.
Lødemel JB, Olsen RO. 2003. Gelatinolytic activities in muscle of Atlantic cod (Gadus morhua), spotted wolffish
(Anarhicas minor) and Atlantic salmon (Salmo salar). J SciFood Agric 83:1031-1036.
Márquez M, Vodopivez C, Causaux R, Curtosi A. 1998. Metal (Fe, Zn, Mn and Cu) levels in the Antarctic fish Notothenia
coriiceps. Polar Biol 20:404-408.
Massova I, Kotra LP, Fridman R, Mobashery S. 1998. Matrix metalloproteinases: Structures, evolution and diversification.
FASEB J 12:1075-1095.
Morrison EH, Bremner HA, Purslow PP. 2000. Location and post-mortem changes in some cytoskeletal proteins in pork
and cod muscle. J Sci Food Agric 80:691-697.
Nishimura T, Hattori A, Takahashi K. 1996. Relationship between degradation of proteoglycans and weakening of the
intramuscular connective tissue during post-mortem ageing of beef. Meat Sci 42(3):251-260.
Ofstad R, Egelandsdal B, Kidman S, Myklebust R, Olsen RL, Hermansson A-M. 1996. Liquid loss as affected by post-
mortem ultrastructural changes in fish muscle; cod (Gadus morhua L.) and salmon (Salmo salar). J Sci Food Agric
71:301-312.
binnenwerk.indd 170 21-08-2006 12:34:33

Olsson GB, Olsen RL, Ofstad R. 2003. Post-mortem structural characteristics and water-holding capacity in Atlantic
halibut muscle. Lebensm-Wiss Technol 36:125-133.
Pena MM, Lee J, Thiele DJ. 1999. A delicate balance : Homeostatic control of copper uptake and distribution. J Nutr
129:1251-1260.
Predki PF, Sarkar B. 1992. Effect of replacement of “zinc finger” zinc on estrogen receptor DNA interactions. J Biol
Chem 267:5842-5846.
Rae TD, Schmidt PJ, Pufahl RA, Culotta VC, O’Halloran TV. 1999. Undetectable intracellular free copper: the requirement
of a copper chaperon for superoxide dismutase. Science 285:805-808.
Saito M, Kunisaki N, Urano N, Kimura S. 2000a. Characterisation of cDNA clone encoding the matrix metalloproteinase
2 from rainbow trout fibroblasts. Fish Sci 66:334-342.
Saito M, Sato K, Kunisaki N, Kimura S. 2000b. Characterisation of a rainbow trout matrix metalloproteinase capable of
degrading type I collagen. Eur J Biochem 267:6943-6950.
Sato K, Ando M, Kubota S, Origasa K, Kawase H, Toyohara H. 1997. Involvement of type V collagen in softening of fish
muscle during short-term chilled storage. J Agric Food Chem 45:343-348.
Shiau SY, Ning YC. 2003. Estimation of dietary copper requirements of juvenile tilapia Oreochromis nilticus X O. aureus.
Animal Sci 77:287-292.
Shim H, Harris ZL. 2003. Genetic defects in copper metabolism. J Nutr 133(5):1527S-1531S.
Strongin AY, Collier IE, Krasnov PA, Genrich LT, Marmer BL, Goldberg GI. 1993. Human92kDa Type IV collagenase:
functional analysis of fibronectin and carboxyl end domains. Kidney Intern 43:158-162.
Taylor RG, Fjæra SO, Skjervold PO. 2002. Salmon fillet texture is determined by myofiber-myofiber and myofiber-
myocommata attachment. J Food Sci 67(6):2067-2071.
Teeny FM, Gauglitz EJ, Hall AS, Houle, CR. 1984. Mineral composition of the edible muscle tissue of seven species of
fish from the Northeast Pacific. J Agric Food Chem 32:852-855.
Toyohara H, Ito K, Ando M, Kinoshita M, Shimizu Y, Sakaguchi M. 1991. Effect of maturation on activities of various
proteases and protease inhibitors in the muscle of ayu (Plecoglossus altivelis). Comp Biochem Physiol B 99:419-
424.
Woessner JF. 1991. Matrix metalloproteinases and their inhibitors in connective-tissue remodelling. FASEB J 5:2145-
2154.
Yamashita M, Konagaya S. 1990. High activities of cathepsin-B, cathepsin-D, cathepsin-H and cathepsin-L in the white
muscle of chum salmon in spawning migration. Comp Biochem Physiol B 95:149-152.
Young Cho C, Cowey CD, Dabrowski K, Hughes S, Lall S, Lovell RT, Murai T, Wilson RP. 1993. Nutrient requirements of
fish. Pub National Research Council, subcommittee on fish nutrition, Washington DC.
binnenwerk.indd 171 21-08-2006 12:34:33

binnenwerk.indd 172 21-08-2006 12:34:33
Relevance of storage temperature for contraction and gaping of pre
rigor filleted farmed cod (Gadus morhua L.)

AKVAFORSK, Institute of Aquaculture Research, N-1432 Ås, Norway
Abstract
Fish that are filleted immediately after slaughtering (pre rigor) can reach the consumers as super
fresh products. Pre rigor processing demands controlled rigor contraction, and that freshness
and quality are preserved. The present study showed that contraction of pre rigor cod fillets
was accelerated and stronger when they were kept at temperatures above 6 °C. After 24 hours,
the contraction was 20% and 28% for fillets kept at 0 °C and 20 °C, respectively. Increasing
temperature caused faster pH drop and more gaping. Pre rigor fillets had less gaping than their
post rigor counterparts. The results suggested that optimum storage temperature of pre rigor
farmed cod is 0-6 °C to avoid gaping and to obtain delayed and reduced rigor contraction.
Keywords: farmed cod, pre rigor fillets, gaping, storage temperature, muscle pH
Introduction
Improved shelf-life is essential for maintaining high eating quality. Cod filleted before rigor
onset (pre rigor fillets) can reach the consumers in a fresher state than cod filleted after rigor
resolution (post rigor fillets). Hence, early processed farmed cod has a great potential supplying
the increasing demand for super-fresh products in the EU-market.
Rigor mortis is a strong muscle contraction that occurs when the energy reserves in the muscle
are depleted post mortem. Stress during slaughtering accelerates ATP degradation and rigor
development after death (Erikson, 1997; Thomas and others 1999; Robb and others 2000;
Skjervold and others 1999, 2001; Black and others, 2004), although the sensitivity to handling
stress seems to depend on factors such as hatchery temperature (Jittinandana and others
2003), feed formulation (Mørkøre 2006), and acclimation temperature before slaughtering
(Hwang and others 1991; Skjervold and others 2001). Furthermore, it is well recognised that
storage temperature affects rigor mortis progress and related biochemical changes of gutted
fish (Hwang and others 1991; Mochizuki and others 1999; Tome and others 2000; Jerret and
others 2002; Mishima and others 2005). There is, however, limited knowledge on the impact
of temperature on rigor contraction of pre rigor fish fillets. It is not recommended that fish are
handled when they are stiff in rigor. Therefore, pre rigor filleting requires that rigor onset occurs
after a certain period of time after slaughtering. Cod farming facilitates pre rigor processing
since stress in conjugation with slaughter can be minimised and because of the short distance
between slaughterhouses and processing plants. Nonetheless, pre rigor filleting is only an
option if the suitability for further value adding and eating quality is acceptable.
Fillet gaping is a phenomenon where slits or holes appear in the fillet due to rupturing of the
connective tissue (Love 1970). Fillets with soft texture and gaping give decreased yield during
binnenwerk.indd 173 21-08-2006 12:34:33

processing and they cannot be sold as high quality products (Love 1988; Mitchie 2001). Among
factors that promote gaping and texture deterioration in farmed fish are high feeding intensity
(Einen and others 1999) and stress during slaughtering (Jerret and others 1996; Sigholt and
others 1997; Skjervold 2002). Gaping frequency also varies seasonally (Love 1988, Mørkøre
and Austreng 2004), and there is evidence that seasonal changes in gaping co-vary with
changes in post rigor muscle pH (Love 1988). It is well known that high storage temperatures
accelerate quality deterioration, but additional documentation is needed on the impact of
storage temperatures on the rate and degree of contraction of pre rigor cod fillets and fillet
gaping, and on the relationship between rate of rigor contraction and fillet gaping. This study
was undertaken to examine contraction, muscle pH and gaping frequency of pre rigor fillets of
farmed Atlantic cod kept at temperatures ranging from 0 – 20 °C.
The fish examined were farmed Atlantic cod (Gadus morhua L.) reared at AKVAFORSK research
station at the Norwegian West coast (Averøy). The feed used was a commercial extruded
dry feed (CO-7 Europa 18%, manufactured by Trouw Espania S.A., delivered by Skretting AS,
Norway). Sampling was performed in October (n = 54) and November 2003 (n = 64). The fish
were anaesthetized by metakain (MS222), using 150 ml master batch per 60 l of seawater.
Thereafter the gill arches were cut, and the fish were bled out in a tub of seawater. Finally
a blow to the head killed the fish. In October, 24 cod were filleted max 20 minutes after
slaughtering (pre rigor) while 30 cod were filleted following 4 days of ice storage (post rigor).
In November, 34 cod were filleted max 20 minutes after slaughtering while 30 cod were filleted
following 4 days of ice storage. All filleting was performed by hand. The study performed in
October is referred to as Experiment 1 whereas the study performed in November is referred
to as Experiment 2. Immediately after slaughtering, muscle samples for glycogen and lactate
determination were cut from the neck region in cod used for analyses in the post rigor state,
and muscle pH was analysed in the cut area. After post rigor filleting, pH and fillet gaping
were analysed, and samples for determination of dry matter content were cut from the dorsal
part from each left fillet side. The average temperature in the sea was 10.5 °C and 8.7 °C in
Experiment 1 and Experiment 2, respectively.
Experiment 1
The pre rigor fillets were stored at either 0 °C, 6 °C, 12 °C or 20 °C; six fillets at each temperature.
Fillet contraction was measured at 0, 3, 6, 8 and 10 hours after slaughtering as 100% x fillet
length after storage x initial fillet length-1. The fillets were subjected to pH and gaping analyses
10 hours after filleting.
Experiment 2
The pre rigor fillets were stored at 0 °C or 20 °C, 17 fillets at each temperature. The reason that
the number of fillets was increased at each temperature was the experience from experiment
1 showing a high variation between fillets within the same temperature treatment. Fillet
contraction was analysed at 0, 3, 6, 8, 10, 16, 20, 24 and 30 hours after slaughtering. The fillets
were subjected to pH and gaping analyses at 10 hours and 24 hours after filleting.
Analyses
Dry matter content was analysed in duplicates in the dorsal fillet section by drying approximately
5 g muscle homogenate from pooled samples per temperature treatment at 105 °C overnight. The
binnenwerk.indd 174 21-08-2006 12:34:33

Relevance of storage temperature for contraction and gaping of pre rigor filleted farmed cod
pH was measured using a muscle electrode (Schott pH-elektrode, Blueline 21 pH, WTW, Weilheim,
Germany) connected to a pH-meter (330i SET - Wissenschaftlich- Technische- Werkstätten GmbH
& Co. KG WTW, Weilheim, Germany). The muscle electrode was inserted directly into the muscle.
Glycogen and lactate were analysed as described by Einen and Thomassen (1998). Gaping was
evaluated according to a scale ranging from 0 to 4, where score 0 = no gaping, score 1 = less
than 5 small slits, score 2 = less than 10 small slits or less than 2 large slits, score 3 = more
than 10 small slits and/or less than 2 – 5 large slits, score 4 = severe gaping (more than 15
small slits and/or more than 5 large slits) (modified according to Andersen and others 1994).
Statistics
ANOVA analyses were carried out with temperature as treatment using the SAS software package
(SAS Institute, 1990). Significant differences among means within sampling time were ranked
by Duncan’s multiple range test. Regression analyses were used to examine changes over time
within temperature treatment and to describe relationships between variables. Proportion of
the total variation explained by the model is expressed by R2 and the level of significance was
set at p<0.05.
Results
The mean body weight of the fish was 650 grams in both experiments, and the individual fillet
weight was 150 grams on average. The liver index was 11.8% in experiment 1 and 11.9% in
experiment 2, and all fish examined were immature (gonad index 0.8 – 2.1) (Table 1).
Experiment 1
Fillet contraction (pre rigor fillets)
The fillet contraction was similar for the different temperature treatments during the first six
hours of storage, but for fillets stored at 20 °C the contraction was significantly highest after
8 hours (p<0.001) and 10 hours of storage (p<0.001) (Figure 1). After 10 hours of storage the
contraction of the fillets stored at 12 °C was significantly higher compared with those stored
Table 1. Slaughter data, fillet pH at slaughtering and in post rigor fillets (5 days post mortem), and
composition of muscle sampled immediately after harvesting.
Experiment 1 Experiment 2
Mean SD Mean SD
Whole body weight, g 655 52 653 63

Condition factor 1.31 0.12 1.23 0.12
Liver index, % 11.9 1.6 11.8 2.1
Gonad index, % 0.8 0.4 2.1 0.8
Fillet weight, g 147 15.7 147 16.8
pH at slaughtering 7.19 0.18 7.11 0.07
pH 5 days post-mortem 6.25 0.09 6.15 0.13
Glycogen 5.3 0.6 4.8 0.7
Lactate 3.1 0.5 2.9 0.6
Dry matter content, % 20.7 0.5 20.7 0.6
binnenwerk.indd 175 21-08-2006 12:34:34

30
0°C
6°C
25 12°C
20°C
20
Fillet contraction, %
15
10
0
0 2 4 6 8 10
Hours after slaughtering
Figure 1. Development in contraction of pre rigor cod fillets stored at 0 °C, 6 °C, 12 °C or 20 °C for 10 hours.
The points and error bars denote mean ± standard deviation of six fillets (Experiment 1).
at 6 °C (p<0.05). The contraction of fillets stored at 0 °C or 6 °C did not differ significantly

at any sampling time. The contraction after 10 hours of storage was 7.6%, 6.5%, 11.6% and
23.3% for the fillets stored at 0 °C, 6 °C, 12 °C and 20 °C, respectively.
Regression analyses revealed that the contraction increased linearly for the fillets stored at 0 °C
(y = 0.70x + 0.13; R2 = 0.98), 6 °C (y = 0.65x + 0.27; R2 = 0.98) and 12 °C (y = 1.13x - 0.06;
R2 = 0.997) during the 10 hours of the storage period. For the fillets stored at 0 °C or 6 °C only
the contraction rate at 10 hours storage differed significantly from that recorded at 3 hours.
For the fillets stored at 12 °C, the contraction increased significantly after 8 hours storage.
For the fillets stored at 20 °C the contraction rate followed the curve linear the equation: y =
0.19x2 + 0.41x + 0.29; R2 = 0.997), and for these fillets the contraction increased significantly
after 6 hours.
The contraction within temperature treatment showed highest variation between fillets after
3 hours of storage, independent of temperature treatment (CV 61 – 92%). Thereafter the
individual variation within storage time decreased, and at 10 hours of storage, the CV was 26
– 34%. The variation between fillets within storage time was consistently highest for the fillets
stored at 20 °C.
Muscle pH
The muscle pH at slaughtering was 7.19 on average and the glycogen and lactate content
averaged 5.3 mg g-1 and 3.1 mg g-1 wet muscle, respectively (Table 1). After 10 hours storage,
binnenwerk.indd 176 21-08-2006 12:34:34

the muscle pH was significantly lower in pre rigor fillets stored at 20 °C (pH 6.05) and higher
in fillets stored at 0 °C (pH 6.67) or 6 °C (pH 6.53) (Figure 2). There was a significant negative
relationship between storage temperature and muscle pH (y = -0.0011x2 – 0.008 + 6.67; R2 =
0.994) (Figure 2). The muscle pH of post rigor fillets differed significantly from pre rigor fillets
stored at 0 °C but not from the muscle pH of fillets stored at 20 °C (Table 2).
Fillet gaping
Gaping analysed in pre rigor fillets after 10 hours storage ranged from 1.0 to 2.25 and a
significant positive relationship was found between storage temperature and gaping frequency
(y = 0.0044x2 – 0.03x + 0.998; R2 = 0.99) (Figure 3). ANOVA analyses revealed no significant
difference between gaping frequencies of fillets stored at 0 °C (gaping score 1.0), 6 °C (gaping
score 1.0) or 12 °C (gaping score 1.3). Gaping score of post rigor fillets (gaping score 1.4)
tended to be higher than of pre rigor fillets stored at 0 °C or 6 °C, similar with that of pre
rigor fillets stored at 12 °C, but significantly higher compared with pre rigor fillets stored at 20
°C (p<0.05) (Table 2). Degree of gaping related negatively to muscle pH (y = 5.1x2 - 68.6x +
229.9; R2 = 0.996).
0h 10h
7.2
Muscle pH 10h following slaughtering
7.0
6.8
6.6
6.4
6.2
6.0
0 4 8 12 16 20
Storage temperature, °C
Figure 2. Muscle pH in pre rigor cod fillets stored at different temperatures. The dotted line indicates the
muscle pH analysed immediately after slaughtering (Experiment 1).
binnenwerk.indd 177 21-08-2006 12:34:34

Table 2. Muscle pH and fillet gaping in pre rigor cod fillets stored at 0 °C or 20 °C and in gutted cod stored
on ice until post rigor filleting four days after slaughtering.
Pre rigor Post rigor
Following 10 h storage Following 24 h storage 4 days after

slaughtering
0 ˚C 20 ˚C 0 ˚C 20 ˚C
Experiment 1, October
Muscle pH 6.7 ± 0.1 6.1 ± 0.1 - - 6.3 ± 0.1
Fillet gaping, score 1.0 ± 0.1 2.3 ± 0.1 - - 1.4 ± 0.1
Experiment 2, November
Muscle pH 6.7 ±0.0 5.8 ± 0.0 6.1 ± 0.0 5.8 ± 0.1 6.2 ± 0.0
Fillet gaping, score 0.8 ± 0.1 2.6 ± 0.2 0.6 ± 0.1 3.3 ± 0.1 1.8 ± 0.1
3.0
2.5
Gaping frequency, score
2.0
1.5
1.0
0.5
0.0
0 4 8 12 16 20
Storage temperature, °C
Figure 3. Gaping frequency (score 0 – 4) of pre rigor cod fillets as a function of storage temperature. The
analyses were performed 10 hours after filleting and the points and error bars denote mean ± standard
deviation of six fillets (Experiment 1).
binnenwerk.indd 178 21-08-2006 12:34:35

Experiment 2
Fillet contraction (pre rigor fillets)
Degree of contraction differed significantly between pre rigor fillets stored at 0 °C and 20 °C
after 3 hours storage (p<0.001) and onwards (p<0.001) (Figure 4). The fillet contraction after 3
hours storage was 3.4% and 6.7% for the fillets stored at 0 °C and 20 °C, respectively, and after
10 hours storage the contraction averaged 6.9% and 22.6%. Contraction of the fillets stored
at 20 °C increased significantly until 16 hours (contraction = 27%) and the final contraction
after 30 hours was 28.1% (Figure 5). The contraction of the fillets stored on ice increased
significantly until 24 hours (contraction = 18%). The final contraction of these fillets was 19.6%
after 30 hours of storage. The variation between individual fillets within storage temperature
was highest after 3 hours storage, independent of temperature treatment (CV 40 – 51%).
Thereafter the individual variation within storage time decreased, and at 30 hours storage, the
CV was 9.3 – 11.4%. The variation between fillets within storage time was generally higher for
the fillets stored at 20 °C compared with those stored at 0 °C.
Muscle pH
The mean muscle pH at slaughtering was 7.11 and the glycogen and lactate content averaged
4.8 mg g-1 and 2.9 mg g-1 wet muscle, respectively (Table 1). The mean muscle pH was 6.7
and 5.8 for pre rigor fillets stored for 10 hours at 0 °C and 20 °C, respectively (Table 2). After
24 hours storage period the muscle pH averaged 6.1 and 5.8 for fillets stored at 0 °C and
30
25
20
15
10
0°C
5
20°C
0
0 5 10 15 20 25 30
Hours after slaughtering
Figure 4. Development in contraction of pre rigor cod fillets stored at 0 °C or 20 °C for 30 hours. The points
and error bars denote mean ± standard deviation of 17 fillets (Experiment 2).
binnenwerk.indd 179 21-08-2006 12:34:35

Figure 5. Pre rigor cod fillet stored at 20 °C for 24 hours. The dotted line shows the original shape
of the fillet.
20 °C, respectively. The mean muscle pH of post rigor fillets analysed 4 days post mortem was
6.2. Muscle pH of the fillets stored at 0 °C increased significantly during the storage period
(p<0.001) whereas the muscle pH of the fillets stored at 20 °C was similar after 10 hours and
24 hours storage. The muscle pH was consistently lower (p<0.001) in the fillets stored at 20 °C
compared with 0 °C storage temperature. Post rigor fillets had significantly higher pH compared
with pre rigor fillets stored at 20 °C (p<0.001). No significant difference between pH in pre rigor
fillets kept at 0 °C for 24 hours and post rigor fillets was observed (Table 2).
Data on pH values from both Experiment 1 and Experiment 2 were pooled in order to evaluate
the overall relationship between contraction of pre rigor fillets and muscle pH. The average
muscle pH was calculated for fillets with a contraction averaging 0, 5, 10, 15, 20, 25 and 30%
(n = 57 for fillets with 0% contraction, otherwise n = 11 – 15). As it appears from Figure 6,
there was an overall negative relationship between fillet contraction and muscle pH (y = 11.8x2
– 174x + 641; R2 = 0.98).
Gaping
After 10 hours storage, the mean gaping score in pre rigor fillets were 0.8 and 2.6 for fillets stored
at 0 °C and 20 °C, respectively (Table 2). Following 24 hours storage the gaping score averaged
0.6 and 3.3. Gaping score of the fillets stored at 0 °C did not increase significantly during the
storage period, whereas the gaping score for the fillets stored at 20 °C was significantly higher
after 24 hours compared with that after 10 hours (p<0.001). The gaping score of post rigor
fillets was significantly higher compared with pre rigor fillets stored at 0 °C and significantly
lower compared with fillets stored at 20 °C.
binnenwerk.indd 180 21-08-2006 12:34:36

30
25
20
15
10
0
5.7 5.9 6.1 6.3 6.5 6.7 6.9 7.1
Muscle pH
Figure 6. Overall relationship between contraction and muscle pH in pre rigor filleted farmed cod. The points
and error bars denote mean ± standard deviation of 11 – 15 fillets, except for fillets with 0% contraction,
where the point denote mean ± standard deviation of 57 fillets (Pooled data from Experiment 1 and
Experiment 2).
Discussion
The present study showed that contraction of pre rigor cod fillets is highly temperature dependent,
where especially fillets stored at room temperature have accelerated rigor contraction. Rigor
contraction averaged 3% of the initial fillet length after 3 hours storage. This corresponds with
Mørkøre (2006) who also reported an instantaneous contraction of farmed cod fillets stored at
6 °C. The variation in contraction among fillets was highest after 3 hours storage, and fillets
stored at 20 °C had consistently highest individual variation. Fewer fillets analysed at each
storage temperature in Experiment 1 (n = 6) than in Experiment 2 (n = 17), probably explain
the lack of significant difference between temperature treatments in Experiment 1 shortly after
filleting.
Fillets stored at 20 °C had a consistent contraction of 23% after 10 hours storage, whereas
those stored at 0 °C contracted 7-8%. Fillets stored at 20 °C reached a maximum contraction
after 16 hours storage (28% of initial length), in comparison to 25-30 hours for fillets stored
at 0 °C (20% of initial length). Similarly, Mørkøre (2006) reported a final contraction of 21%
after 24 hours in cold stored farmed cod. Limited contraction is positive since such fillets are
thicker without having altered surface, while fillets with a rigor contraction of nearly 30% have
a severely altered shape and a rough surface.
binnenwerk.indd 181 21-08-2006 12:34:36

Sørensen and others (1997) and Einen and others (2002) reported 14 – 16% to be the maximum
contraction in cold stored pre rigor salmon, and Slinde and others (1998) reported stronger
contraction in cod than in salmon fillets. Sørensen and others (1997) observed a difference in
contraction of only 2%-units for pre rigor salmon fillets stored at 20 °C compared 0 °C. Therefore,
shrinkage of pre rigor cod fillets appear to be stronger and more extensively accelerated at high
temperatures compared with those of salmon.
The contraction of fillets stored at 6 °C or 12 °C was only recorded until 10 hours. Therefore, this
study do not provide absolute maximum contraction for these storage temperatures. However,
the calculated contraction, based on the regression equations given in Experiment 1, is 17%
and 16% after 24 hours for fillets stored at 0 °C and 6 °C, respectively. Consequently, it
is suggested that fillets stored at 6 °C would have reached a similar or slightly lower final
contraction compared with those stored at 0 °C after 24 hours.
It is generally believed that rigor mortis relates to falling ATP levels in muscles post mortem.
Correspondingly, Mørkøre (2006) found a linear relationship between rigor contraction and ATP
depletion in farmed cod. Studies have documented that activities of enzymes involved in ATP
consumption are temperature dependent (Tanaka 1991). Since contraction of fillets stored at 0
°C and 6 °C was relatively similar, the temperature dependency of enzyme activity in relation to
ATP consumption does not seam to be linear in post-mortem cod muscle. Iwamoto and others
(1985; 1988; 1990) reported a longer delay in the progress of rigor-mortis and ATP degradation
when kept at 10 °C than when kept at 0 °C in the red sea bream, the bartailed flathead, the
yellowtail, the plaice, and the Japanese striped knifejaw. Mochizuki and other (1999) reported
similar results for chub mackerel, thus advised to store this specie at 5 °C to delay post-mortem
changes. Mishima and others (2005) suggested that horse mackerel should be stored at 10 °C
for about 24 hours, whereas Yamanaka (2002) proposed storing the plaice at 5 or 10 °C initially
in order to delay rigor mortis and thereafter at 0 °C to prevent deterioration of freshness. The
present results suggest that the temperature dependency of rigor contractions of cod fillets
differs from those species discussed above, and that the optimum temperature for delaying
rigor development in farmed cod is 0 °C – 6 °C.
Development and ultimate pH was significantly affected by storage temperature in a non-linear

manner. It is assumed that higher temperature probably accelerated the glycogen degradation
and lactate production, hence caused the faster pH drop and accelerated contraction. Similarly,
Mørkøre (2006) also found an inverse relationship between fillet contraction and muscle pH
in cod, although the author reported that pH started dropping after 8 hours storage, at which
time the contraction was 9 – 12%. Additional factors than pH per se are also influencing the
rigor contraction rate and strength. When the pH is low enough, certain critical enzymes are
inhibited and glycolysis ceases (Foegeding and others 1996). Moreover, it is possible that
glycogen degradation is so slow at low temperatures that enzymes involved in the glycolysis
deteriorate before all the glycogen is broken down into lactate, thus a higher ultimate pH is
obtained.
Gaping and fillet softening have been related to lower strength of the connective tissue of the
myocommata (Love 1988), but also a rapid decline in pH might cause flesh softening in cod
(Ang and Haard 1985). The strength of myocommata can be reduced many-fold by the reduction
of post rigor pH (Love 1988). Moreover, whole fish entering rigor mortis at raised temperatures
can have increased gaping (Lavèty and others 1988). In the present study, increased gaping
binnenwerk.indd 182 21-08-2006 12:34:36

coincided with high storage temperature and low muscle pH. Degree of gaping was significantly
lower in ice stored pre rigor fillets than in post rigor cod kept on ice. Pre rigor fillets are free
to shrink because the muscle is not attached to the skeleton. The macroscopic structure of
pre rigor fillets is therefore protected against disruption of muscle cell-myocommata junctions
during rigor development, and the risk for gaping is reduced.
Conclusion
The desirable temperature to avoid gaping and to obtain delayed and less strong rigor contraction
in pre rigor farmed cod fillets was 0 – 6 °C. Rigor development seems to be related to a multitude
of factors with a high degree of interaction among them. These include the in vivo ATP and
glycogen level, and the rate of degradation of these components. Specie specific enzyme
kinetics should be further elucidated in order to improve the understanding of the underlying
mechanisms related to rigor development.
Acknowledgements
The Norwegian Research Council supported the study. The staff at AKVAFORSK research station
at Averøy is especially thanked for their technical assistance.
References
Andersen UB, Strømsnes AN, Steinsholt K, Thomassen MS. 1994. Fillet gaping in farmed Atlantic salmon (Salmo salar).
Norwegian J Agric Sci 8:165-179.
Ang J, Haard NF. 1985. Chemical composition and post-mortem changes in soft textured muscle from intensively feeding
cod, Gadus morhua. J Food Biochem 9:49-64.
Black SE, Jerret AR, Forster ME. 2004. Extension of the pre-rigor period in ischemic white muscle from yellow-eye
mullet (Aldrichetta forsteri) and New Zealand snapper (Pagrus auratus) as affected by hyperbaric oxygen treatment.
J. Food Sci 69:297-C307.
Einen O, Thomassen MS. 1998. Starvation prior to slaughter in Atlantic salmon (Salmo salar). II. White muscle
composition and evaluation of freshness, texture and colour characteristics in raw and cooked fillets. Aquaculture
169:37-53.
Einen O, Mørkøre T, Rørå AMB, Thomassen MS. 1999. Feed ration prior to slaughter – a potential tool for managing
product quality of Atlantic salmon (Salmo salar). Aquaculture 178:149-169.
Einen O, Guerin T, Fjæra SO, Skjervold PO. 2002. Freezing of pre-rigor fillets of Atlantic salmon. Aquaculture 212:129-
140.
Erikson U. 1997. Muscle quality of Atlantic salmon (Salmo salar) as affected by handling stress. Dr. Ing. Thesis. 1997:68.
Norway: Norwegian University of Science and Technology. 132 p.
Foegeding EA, Lanier TC, Hultin HO. 1996. Characteristics of edible muscle tissues. In: Hoar WS, Randall DJ, editors.
Food Chemistry. London: Academic Press. p 503-543.
Hwang GC, Ushio H, Watabe S, Iwamoto M, Hashimoto K. 1991. The effect of thermal-acclimation on rigor mortis
progress of carp stored at different temperatures. Nippon Gakkaishi 57:541-548.
Iwamoto M, Ioka H, Saito M, Yamanaka H. 1985. Relation between rigor mortis of sea bream and storage temperature.
Iwamoto M, Yamanaka H, Abe H, Ushio H, Watabe S, Hashimoto K. 1988. ATP and creatine phosphate breakdown in
spiked plaice muscle during storage, and activities of some enzymes involved. J Food Sci 53:1662-1665.
Iwamoto M, Yamanaka H, Abe H, Watabe S, Hashimoto S. 1990. Rigor-mortis progress and its temperature-dependency
in several marine fishes. Nippon Suisan Gakkaishi 56:93-99.
binnenwerk.indd 183 21-08-2006 12:34:36

Jerret AR, Stevens J, Holland AJ. 1996. Tensile properties of white muscle in rested and exhausted Chinook salmon
(Oncorhynchus tshawytscha). J Food Sci. 61:527-532.
Jerret AR, Law RE, Holland AJ, Black SE. 2002. Profiles of New Zealand snapper (Pagrus aurats) post mortem metabolism
as affected by acclimated temperature and post mortem storage temperature. J. Food Sci 68:2843-2850.
Jittinandana S, Kenney PS, Slider SD, Mazik P, Bebak-Williams J, Hankins JA. 2003. Effect of attributes and handling
stress on quality of smoked Arctic char fillets. J Food Sci. 68:57-63.
Lavéty J, Afolabi OA, Love RM. 1988. The connective tissues of fish. IX. Gaping in farmed species. Int J Food Sci Tech
23:23-30.
Love RM. 1970. The Chemical Biology of Fishes. London and New York:Academic Press. 547 p.
Love RM. 1988. The Food fishes – Their intrinsic variation and practical implications. London: Farrand press. 276 p.
Mochizuki S, Ueno Y, Satoh K, Hida N. 1999. Effects of storage temperature on post-mortem changes in the muscle of
chub mackerel. Nippon Gakkaishi 65:495-500.
Mishima T, Nonaka T, Okamoto A, Tsuchimoto M, Ishiya T, Tachibana K, Tsuchimoto M. 2005. Influence of storage
temperature and killing procedure on post-mortem changes in the muscle of horse mackerel caught near Nagasaki
Prefecture, Japan. Fisheries Sci. 71:187-194.
Mittchie I. 2001. Causes of downgrading in the salmon farming industry. In: Kestin SC, Warris PD, editors. Fishing News
Books. Oxford: Blackwell Science Ltd. p 129-136.
Mørkøre T. 2006. Relevance of dietary oil source for contraction and quality of pre-rigor filleted Atlantic cod, Gadus
morhua. Aquaculture 251:56-65.
Mørkøre T, Austreng E. 2004. Temporal changes in texture, gaping, composition and copper status of Atlantic salmon
(Salmo salar, L) fed moist feed or extruded dry feed. Aquaculture 230:439-455.
Robb DHF, Kestin SC, Warriss PD, 2000. Muscle activity at slaughter: I. Changes in flesh colour and gaping in rainbow
trout. Aquaculture 182:261-269.
Sigholt T, Erikson U, Rustad T, Johansen S, Nordtvedt TS, Seland A. 1997. Handling stress and storage temperature
affect meat quality of farm-raised Atlantic salmon (Salmo salar). J Food Sci 62:898-905.
Skjervold PA. 2002. Live-chilling and pre-rigor filleting of salmonids – technology affecting physiology and product
quality. Dr. Agric. thesis. Ås: Norwegian University of Life Sciences. 150 p.
Skjervold PO, Fjæra SO, Ostby PB. 1999. Rigor in Atlantic salmon as affected by crowding stress prior to chilling before
slaughter. Aquaculture 175:93-101.
Skjervold PO, Fjæra SO, Ostby PB, Einen O. 2001. Live-chilling and crowding stress before slaughter of Atlantic salmon
(Salmo salar). Aquaculture 192:265-280.
Slinde E, Torrissen O, Thorsheim G. 1998. Muscle contraction forces, cold shortening and onset of rigor mortis in fish.
Poster presentation Biokjemisk Vintermøte, Hamar, Norway.
Sørensen NK, Brataas R, Nyvold TE, Lauritzen K. 1997. Influence of early processing (pre-rigor) on fish quality. In: Luten
JB, Børresen T, Oehlenschlager J, editors. Seafoods from Producer to Consumer, Integrated Approach to Quality.
Amsterdam: Elsevier. p 253-263.
Tanaka T. 1991. Transportation condition and market price of spiked fish. In: Yamanaka Y, editor. Rigor mortis in fish.
Tokyo: Kouseishakouseikaku. p. 832-855.
Thomas PM, Pankhurst NW, Bremner HA. 1999. The effect of stress and exercise on post-mortem biochemistry of Atlantic
salmon and rainbow trout. J Fish Biology 54:1177-1196.
Tome E, Iglesias M, Kodaira M, Gonzalez A. 2000. Effect of storage temperature on the onset of rigor mortis and stability
of cultured tilapia (Oreochromis sp.). Revista Cientifica-Facultad de Ciencias Veterinarias 10:339-345.
Yamanaka H. 2002. Relation between post mortem biochemical changes and quality in the muscle of fish and shellfish.
binnenwerk.indd 184 21-08-2006 12:34:36

Brining of farmed cod fillets: Effects on quality aspects
Margrethe Esaiassen1, Grete Lorentzen1, Reidun Dahl1, Guro Eilertsen1, Bjørn Gundersen1,
Torbjørn Tobiassen1 and Morten Sivertsvik2
2Norconserv AS - Seafood Processing Research, PO Box 327, N-4002 Stavanger, Norway
Abstract
Farmed cod is often characterised by high liver-somatic index, low ultimate muscle pH and
a “soft” muscle, and one might expect reduced juiciness and altered texture in farmed cod
compared to wild. This study was focused on whether quality parameters of ice stored farmed
cod fillets could be influenced by brining using a mixture of salt, phosphates, sodium-ascorbate
and glucose. It is shown that brining has a significant influence on water-holding capacity and
b*-value of the fillets, whilst whiteness and water content are significantly affected by both
brining and storage time. On the other hand, brining showed no significant influence on TMAO-
and TMA-values, total viable count, count of sulphide producing bacteria and Photobacterium
phosphoreum, as well as lightness (L*) and the a*-value from the colour measurement. These
parameters were significantly influenced by storage time only.
Keywords: farmed cod, brining, storage
Introduction
It is well known that the quality of wild cod may vary according to several factors like seasonal
variation, feed, environment and post harvest treatment. It is expected that the quality of
farmed cod will become more uniform than the quality of wild cod, since factors like body size,
growth rate and sexual maturation can be manipulated through breeding programmes, feed and
controlled farming conditions. But despite being more uniform, the quality of farmed cod may
differ considerably from the quality of wild cod. Farmed Atlantic cod reach a weight of 3 kg two
year after hatching, while wild cod will reach the same weight after three to five years (Otterå
and others 2005). This faster growth will most probably affect the structure and texture of the
muscle (Johnston 2001; Bugeon and others 2004).
Farmed cod is often characterised by high liver-somatic index and a “soft” muscle (Losnegard and
others 1986; Otterå and others 2005). Farmed cod have higher muscle glycogen than wild cod
(Rustad 1992), resulting in low ultimate muscle pH. The muscle pH of farmed cod 24 hours post
mortem is generally 0.5-0.9 units lower than in wild cod (Rustad 1992). The pH of farmed cod
is low throughout the year while the pH of wild cod is low during the summer. A low ultimate
muscle pH is probably reflecting heavy feeding prior to slaughter (Ang and Haard 1985). Since pH
is shown to have significant effects on the texture of codfish (Love 1979; Ang and Haard 1985;
Segars and Johnson 1987), it is likely that the texture of farmed cod will differ from wild cod.
It is also shown that low pH is associated with reduced water-holding capacity and increased
myofibrillar shrinkage during heating of cod (Ofstad and others 1996), and consequently one
might expect reduced juiciness and altered texture in farmed cod compared to wild.
binnenwerk.indd 185 21-08-2006 12:34:37

M. Esaiassen, G. Lorentzen, R. Dahl, G. Eilertsen, B. Gundersen, T. Tobiassen and M. Sivertsvik
However, a wide variety of compounds have been shown to improve properties like juiciness and
texture in different processed dairy -, meat – and fish products, and among these salt, sucrose,
glucose and phosphates (Krivchenia and Fennema 1988; Dziezak 1990; Craig and others 1991;
MacDonald and Lanier 1997; Park and others 1997; Zheng and others 1999; Badii and Howell
2002; Herrera and others 2002). Recent studies have demonstrated that the intensity of sensory
attributes like juiciness, flakiness and whiteness could be heightened in products of frozen and
thawed cod by applying brining (Esaiassen and others 2004b, 2005). In these reports, the cod
fillets were treated with brine mixtures consisting of salt, phosphates, sodium-ascorbate and
glucose in a vacuum-tumbler.
The objective of the present work was to determine if brining with a mixture of salt, phosphates,
sodium-ascorbate and glucose would influence selected quality attributes of farmed cod fillets.
Since altering the composition of the fillets by addition of ingredients may affect the storage
stability of the products, the cod fillets were also subjected to ice storage in order to compare
the quality of brined and non-brined fillets during storage.
Raw materials and sampling

Atlantic cod, with an average live weight of 3.5 kg were obtained from Fjord Marin Helgeland AS
(Brønnøysund, Norway). The fry were produced by Lofilab (Leknes, Norway), and the fish were
cultivated for two years in a fjord close to Brønnøysund. The fish were fed a dry pelleted feed
provided by BioMar (Myre, Norway) and starved for 13 days at 11 °C water temperature prior to
slaughtering. The fish were killed by a blow to the head, followed by exsanguination. Bleeding
was conducted in seawater for 30 minutes after cutting the isthmus. The fish were then gutted,
cleaned and put in ice before transportation to the laboratory. After 5 days of ice storage, the
fish were filleted and divided into two batches. One batch was kept as control group while the
other batch was brined. Both the brined and the non-brined fillets (control group) were wrapped
in plastic and kept in ice until the measurement day. On the day of measurements, eight fillets
from each treatment group were sampled. Four fillets from each group were applied to microbial
analyses, water-holding capacity and colour measurements, while the remaining four fillets were
subjected to chemical analyses.
Brining
The brine consisted of 5% (w/v) NaCl and 5% (w/v) SFK 428 (SFK Food AS, Viborg, Denmark)
dissolved in cooked and chilled water, and the pH of the brine was 7.9. The SFK 428 consists of
polyphosphate, triphosphate, diphosphate, glucose and sodium-ascorbate, with a P2O5 content
of 23.5%. Cod fillets and brine were mixed in a 1:1 ratio. Batches of approximately 2.0 kg were
tumbled in a Scanio vacuum tumbler Type VTO (Scanio A/S, Denmark) for 3 minutes at 0.1 bar
and 4 rpm. After tumbling the air inlet was opened, and the samples were left in the brine for
2 minutes before removal. The samples were then drained for 5 minutes before weighing and
packaging.
Microbial analyses
Samples for microbial analyses were cut from four fillets in each batch using sterile technique.
Fish muscle samples (3-10 g) from the separate fishes were homogenized with 9 parts of
peptone water containing 0.9% NaCl using a stomacher (Lab Blender 400) for approximately 4
minutes and diluted tenfolds in peptone water. Sulphide producing bacteria were determined
binnenwerk.indd 186 21-08-2006 12:34:37

as black colonies on Iron agar Lyngby (Oxoid) after 2-3 days incubation at 22 °C. Total viable
counts (TVC) were determined using standard plate count agar (Oxoid CM463) supplemented
with 1.5% NaCl, and incubated for 5 days at 12 °C. Numbers of Photobacterium phosphoreum
were determined by a conductance method as described by Dalgaard and others (1996).
Water-holding capacity and colour measurements

The water-holding capacity was determined in four fillets from each batch as liquid loss and
expressed as percentage of weight of liquid released (Ofstad and others 1993). Flesh colour
was measured in three points per fillet using a Minolta Croma Meter CR 200 calibrated to a
white standard. L*a*b* measurement mode was used, and whiteness (W) was calculated as W=
(L*-3b*) as described by Park (1994).
Chemical analyses
Two hundred grams were cut from the loin of each of four fillets. The samples from the separate
fillets were homogenized for 3 x 1 min in a Hallda Blender. Water content was determined by
drying samples of mixed, minced muscle overnight at 103 °C. The mixed, minced muscle was
also used for obtaining perchloric acid - extracts to determine trimethylamine oxide (TMAO),
trimethylamine (TMA), dimethylamine (DMA) and monomethylamine (MMA) as described by
Oetjen and Karl (1999). The amines were determined in triplicate. Muscle pH was measured in
minced muscle suspended in 0.15 M KCl, using a glass electrode.
The data were analysed by use of multivariate techniques, applying the software Unscrambler
version 9.2 (CAMO Process AS, Oslo, Norway). Prior to the analyses, a logarithmic transformation
was applied to the microbial counts in order to compare values that range over several orders
of magnitude (Martens and Martens 2001). In addition, the variables were weighted with the
inverse of the standard deviation of all objects in order to compensate for the different scales of
the variables. Principal Component Analysis, PCA (Martens and Næs 1989), was used to identify
similarities and differences amongst samples on the basis of microbiological, chemical and
physical data. Partial Least Square Regression, PLSR1, (Martens and Næs 1989) with Martens
Uncertainty Test (Martens and Martens 2001) was applied to identify the effect of brining and
storage time on the different quality attributes.
In order to evaluate how the different quality parameters were affected by brining and storage,
a principal component analysis was performed using the mean values of the quality parameters
for the different batches (brined/non-brined at different storage time). It was found that
four principal components (PCs) explained 94% of the variations in the data set. The scores
and correlation loadings of PC1 and PC2, representing 44 and 28% of the total variation, are
presented in Figure 1. The score plot shows a distinction between brined and non-brined cod,
and the discrimination is mainly along PC2. The correlation loadings plot shows that brining /
non-brining is described by PC2, while the variation in storage time is described by PC1. From
the correlation loading plot it is seen that variations in both water-holding capacity (%LL) and
b*-values from the colour measurements are mainly described in PC2, and thus correlated to
brining. Brined cod have lower liquid loss and lower b*-values than the non-brined samples.
The results from assessing water-holding capacity (liquid loss) and b*-value during storage are
shown in Figure 2 and Figure 3, respectively. According to the principal component analysis,
binnenwerk.indd 187 21-08-2006 12:34:37

PC2 Scores
4
2 + brined + brined
+ brined + brined
0
+ non-brined
-2 + non-brined
+ non-brined + non-brined
-4
PC1
-4 -2 0 2 4
Correlation loadings (X)

1.0 PC2
+ Brining
+ W
+ Water content
+ MMA
+ TMA
+ pH Storage time ++
+ a Photobact
0 Sul-proc-bact + + L+
+ TMAO + TVC
+ DMA
+ %LL
-1.0 + b
PC1
-1.0 -0.5 0 0.5 1.0
Figure 1. Score plot and correlation loading plot from PCA of the quality parameters measured for the
different brined/non-brined batches at different storage time. The outer and the inner ellipse indicate 100%
and 50% explained variance, respectively.
whiteness and water content are influenced by both brining and storage time. The changes in
whiteness and water content during storage are shown in Figures 4 and 5, respectively.
The correlation loadings plot also shows that the variations in TMAO- and TMA-values, total
viable count, as well as the count of sulphide producing bacteria and P. phosphoreum are
mainly described by PC1, and that these parameters, with the exception of TMAO, are positively
correlated to storage time. The TMAO content is negatively correlated to storage time. These
findings are in accordance with previously reported results (Oehlenschläger 1992; Botta 1995;
Gram and Huss 1996; Olafsdottir and others 1997). Lightness (L*) is also positively correlated
to storage time, and the a*-value from the colour measurement is negatively correlated to
storage time.
binnenwerk.indd 188 21-08-2006 12:34:37

10
8
WHC (% LL)
6
0
0 5 10 15 20
Storage time (days after brining)
Figure 2. Water-holding capacity (expressed as %LL) of brined and non-brined farmed cod fillets as a function
of storage days after filleting/brining. ° = non-brined fillets, l = brined fillets.
-2
b*-value
-4
-6
-8
-10
0 5 10 15
Figure 3. b*-value from colour measurements of brined and non-brined farmed cod fillets as a function of
storage days after filleting/brining. ° = non-brined fillets, l = brined fillets.
80
Whiteness (L*-3b*)
75
70
65
60
55
50
0 5 10 15
Figure 4. Whiteness of brined and non-brined farmed cod fillets as a function of storage days after filleting/
brining. ° = non-brined fillets, l = brined fillets.
binnenwerk.indd 189 21-08-2006 12:34:38

84
Water content (%)

82
80
78
76
74
0 5 10 15
Figure 5. Water content of brined and non-brined farmed cod fillets as a function of storage days after
filleting/brining. ° = non-brined fillets, l = brined fillets.
In order to identify the effect of brining and storage time on the measured quality attributes,
and to reveal if brining could significantly influence the quality parameters during storage,
PLSR1 models were made. Brining and storage time data were applied as the X-matrix, while
the individual quality parameters were used as Y-matrices. By applying the Martens Uncertainty
Test, which is based on the jack-knifing principle (CAMO 2002), the influence of brining and
storage time upon every single quality parameter is examined. Table 1 summarizes the results
of the analyses.
Table 1. Variables with significant effect on quality parameters. Significance is identified by Martens
Uncertainty Test (5% level).
Quality parameter Variables with significant effect on quality

parameters of post rigor fillets
pH -none-
L* Storage time
a* Storage time
b* Brining
Whiteness Storage time
Brining
Water content Storage time
Brining
Water holding capacity Brining
MMA -none-
DMA -none-
TMA Storage time
TMAO Storage time
TVC Storage time
Sulph. prod. bact. Storage time
P.phosphoreum Storage time
binnenwerk.indd 190 21-08-2006 12:34:38

Neither brining nor storage had significant influence on pH. The finding that storage time
did not have significant influence on pH is in accordance results from Esaiassen and others
(2004a). Regarding the influence of brining, the pH of both the non-brined and the brined
fillets ranged from 6.2 to 6.5. The brine uptake of the fillets were approximately 8-10% (data
not shown), and this uptake does not seem to be high enough to influence the pH of the fillets
significantly despite the brine having a slightly alkaline pH. This is in contrast to results by
Thorarinsdottir and others (2004). However, the discrepancies could be due to the generally
higher pH-values in wild cod used by Thorarinsdottir and others compared to the pH-values in
the farmed cod in the present work. Or it could be due to different phosphate blends used. In
both works, the phosphate blends have a blurred composition, and exact comparisons could
thus not be made.
From Table 1 it is also seen that brining has significant influence on water-holding capacity,
water content, the b*-value and the whiteness of the fillets. This is in accordance with results
from Esaiassen and others (2004b, 2005), where it was shown that both whiteness and juiciness
could be improved in cod fillets, as assessed by sensory analyses. Thorarinsdottir and others
(2004) also showed that the water-holding capacity of cod fillets could be improved both by
salt and phosphates, and especially when salt and phosphates were combined. Table 1 shows
that the whiteness and water content were significantly influenced by storage time as well
as brining. On the other hand, the L*-value, a*-value, TMA- and TMAO-content, as well as
total viable count and the count of sulphide producing bacteria and P. phosphoreum were not
significantly influenced by brining, but by storage time only.
Conclusion
By applying brining, the water-holding capacity and the water content of fillets from farmed
cod could be significantly increased. In addition, the b*-value from the colour measurements
could be significantly reduced, and the whiteness of the fillets could be significantly increased.
The whiteness and water content are also significantly affected by storage time.
Storage time has significantly influence on TMAO- and TMA-values, total viable count, the count
of sulphide producing bacteria and Photobacterium phosphoreum, as well as lightness (L*) and
the a*-value from the colour measurement. Neither brining nor ice storage had significant
influence on pH, MMA- or DMA-values.
References
Ang JF, Haard NF. 1985. Chemical composition and post-mortem changes in soft texture muscle from intensively feeding
Atlantic cod (Gadus morhua, L.). J Food Biochem 9:9-64.
Badii F, Howell NK. 2002. Effect of antioxidants, citrate, and cryoprotectants on protein denaturation and texture of
frozen cod (Gadus morhua). J Agric Food Chem 50(7):2053-2061.
Botta JR. 1995. Evaluation of seafood freshness quality. New York: VHC Publishers Inc.
Bugeon J, Lefevre F, Fauconneau B. 2004. Correlated changes in skeletal muscle connective tissue and flesh texture during
starvation and re-feeding in brown trout (Salmo trutta) reared in seawater. J Sci Food Agric 84:1433-1441.
CAMO. 2002. The Unscrambler user guide. Oslo: CAMO Process AS.
Craig J, Bowers JA, Seib P. 1991. Sodium triphosphate and sodium ascorbate monophosphate as inhibitors of off-flavour
development in cooked, vacuum-packed, frozen turkey. J Food Sci 56(6):1529-1531.
Dalgaard, P, Mejlholm O, Huss HH. 1996. Conductance method for quantitative determination of Photobacterium
phosphoreum in fish products. J Applied Bacteriol 81(1):57-64.
binnenwerk.indd 191 21-08-2006 12:34:38

Dziezak JD. 1990. Phosphates improve many foods. Food Technology 44(4):80-92.
Esaiassen M, Østli J, Joensen S, Prytz K, Olsen JV, Carlehog M, Elvevoll EO, Richardsen R. 2005. Brining of cod fillets:
Effects of phosphate, salt, glucose, ascorbate and starch on yield, quality parameters and consumers liking. Food
Sci Technol/LWT 38(6):641-649.
Esaiassen M, Nilsen H, Joensen S, Carlehög M, Skjerdal T, Gundersen B, Elvevoll E. 2004a. Effects of catching methods
on quality changes during storage of cod (Gadus morhua). Food Sci Technol/LWT 37(6):643-648.
Esaiassen M, Østli J, Elvevoll EO, Joensen S, Prytz K, Richardsen R. 2004b. Brining of cod fillets: influence on sensory
properties and consumers liking. Food Quality and Preference, 15:421-428.
Gram L, Huss HH. 1996. Microbiological spoilage of fish and fish products. Int J Food Microbiol 33(1):121-137.
Herrera JJR, Pastoriza L, Sampedro G. 2002. Effects of various cryostabilisers on protein functionality in frozen-
stored minced blue whiting muscle: The importance of inhibiting formaldehyde production. Eur Food Res Techn,
214(5):382-387.
Johnston IA. 2001. Implications of muscle growth patterns for the colour and texture of fish flesh. In: Kestin SC, Warriss
PD, editors. Farmed fish quality. Oxford: Blackwell. p 13-30.
Krivchenia M, Fennema O. 1988. Effect of cryoprotectants on frozen whitefish fillets. J Food Sci 53(4):999-1003.
Losnegard N, Langmyhr E, Madsen D. 1986. Farmed cod, quality and utilisation, chemical composition as a function of
season . NFFR no V-709.001. 11. 1986. Norwegian Directorate of Fisheries.
Love RM. 1979. The post mortem pH of cod and haddock muscle and its seasonal variation. J Sci Food Agric 30:433-
438.
MacDonald GA, Lanier TC. 1997. Cryoprotectants for improving frozen-food quality. In: Erickson MC and Hung YC,
editors. Quality in Frozen Food. New York: Chapman & Hall. p 197-232.
Martens H, Næs T. 1989. Multivariate calibration. Chichester, UK: John Wiley and Sons.
Martens H, Martens M. 2001. Multivariate analysis of quality - An introduction. West Sussex, UK: John Wiley and Sons.
Oehlenschläger J. 1992. Evaluation of some well established and some underrated indices for the determination of
freshness and/or spoilage of ice stored wet fish. In: Huss HH, Jakobsen M, Liston J, editors. Quality assurance in
the fish industry. Amsterdam: Elsevier Science Publishers. p 339-350.
Oetjen K, Karl H. 1999. Improvement of gas chromatographic determination methods of volatile amines in fish and
fishery products. Deutsche Lebensmittel-Rundschau, 95(10):403-407
Ofstad R, Kidman S, Myklebust R, Hermansson AM. 1993. Liquid holding capacity and structural changes during heating
of fish muscle; cod (Gadus morhua L.) and salmon (Salmo salar). Food Structure 12:163-174.
Ofstad R, Kidman S, Hermansson AM. 1996. Ultramicroscopical structures and liquid loss in heated cod (Gadus morhua)
and salmon (Salmo salar) muscle. J Sci Food Agric 72:337-347.
Ólafsdóttir G, Martinsdóttir E, Oehlenschlager J, Dalgaard P, Jensen B, Undeland I, Mackie IM, Henehan G, Nielsen J,
Nilsen H. 1997. Methods to evaluate freshness in research and industry. Trends Food Sci Technol 8:258-265.
Otterå H, Carlehög M, Karlsen Ø, Akse L, Borthen J, Eilertsen G. 2005. Quality in farmed Atlantic cod (Gadus morhua
L.) Submitted to Aquaculture International.
Park JW. 1994. Functional protein additives in surimi gels. J Food Sci 59:525-527.
Park JW, Lin TM, Yongsawatdigul K. 1997. New developments in manufacturing of surimi and surimi seafood. Food Rev
Internat 13(4):577-610.
Rustad T. 1992. Muscle chemistry and the quality of wild and farmed cod. In: Huss HH, Jacobsen M, Liston J, editors.
Quality Assurance in the Fish Industry. London: Elsevier. p 19-27.
Segars RA, Johnson EA. 1987. Instrumental measurement of textural quality of fish flesh: Effect of pH and cooking
temperature. In: Kramer D, Liston J, editors. Seafood quality determination. Amsterdam: Elsevier Science Publishers,
p 49-61.
Thorarinsdottir KA, Gudmundsdottir G, Arason S, Thorkelsson G, Kristbergsson K. 2004. Effects of added salt, phosphates,
and proteins on the chemical and physiochemical characteristics of frozen cod (Gadus morhua) fillets. J Food Sci
69:144-152.
Zheng M, Toledo R, Wicker L 1999. Effect of phosphate and pectin on quality and shelf-life of marinated chicken breasts.
J Food Quality 22(5):553-564.
binnenwerk.indd 192 21-08-2006 12:34:39

Enrichment of functional selenium in farmed African catfish
(Clarias goriepinus) by dietary modulation
Joop Luten1 and Edward Schram2

2Institute for Marine Resources & Ecosystem Studies (Wageningen IMARES), PO Box 68, 1970 AB
IJmuiden, the Netherlands
Abstract
An important trace element to human and animal health concern is selenium. Fish is one of the
main sources for selenium in the diet of humans. Enrichment of selenium in farmed fish gives
opportunities to produce tailor-made seafood with functional selenium. The form of selenium
in the diet of farmed fish may influence the final concentration in the edible part of fish as well
it biological activity. A few studies with African catfish (Clarias goriepinus) given a selenium
enriched diet were carried out African catfish was fed with a selenomethionine or with garlic
enriched in organoselenium during a period of six weeks. The selenium dose in the feed varied
between the normal dose of approx. 1 mg selenium/kg to 8 mg selenium/kg feed. Besides, the
effect of methionine concentration in the feed on the incorporation of selenomethionine in the
muscle tissue of African catfish was studied. It was shown that African catfish growth was not
affected by the dietary selenium level. The selenium content in the edible part of the African
catfish increased linearly with increasing dietary selenium levels. The increase was approx. 3
times higher using selenomethionine in the feed then with the organoselenium compounds
from selenium enriched garlic supplemented to the feed.
Keywords: selenium, African catfish, dietary modulation, enrichment, farming
Introduction
Selenium is identified to be an essential trace element and is of fundamental importance to

human health (Rayman, 2000). As a constituent of selenoproteins, selenium has structural and
enzymatic roles. Selenium is a component of the enzyme glutathione peroxidase, which removes
hydrogen peroxide and lipid peroxidases from cells. The range between selenium deficiency and
selenium toxicity is narrow (0.1 - 4.0 mg/kg in plants). Diets containing with a concentration
of more than 5 mg Se /kg are considered to be poisonous to man and animals. An elevated
intake (1-5 mg selenium/kg) may be associated with reduced cancer risk.
Selenium metabolism, the chemical forms of selenium and anti-carcinogenic activity are
extensively reviewed by Ip (1998). The most successful human cancer intervention study is
the selenium trial by Clark and others (1996). They performed a double-blind, randomized,
placebo-controlled trial involving 1312 patients in which treated individuals received 200
µg Se/day from selenized yeast for on average 4.5 years. It was found that in patients that
received the Se supplement the relative risk of cancer incidence in lung, colon and prostate
was reduced to almost 50% on average. Cancer inhibition activity is known to depend on the
chemical form of selenium. For example the anti-carcinogenic activity of selenomethionine is
binnenwerk.indd 193 21-08-2006 12:34:39

limited due to non-specific incorporation in body proteins in place of methionine (Ip 1998).
Unlike selenomethionine, Se-methylselenocysteine cannot be non-specifically incorporated in
body proteins. In addition, it is a good precursor for generating monomethylated selenium. It
is these precursors that are likely to have good chemopreventive activity (Ip 1998).
Garlic is an excellent source of functional selenium. Garlic converts and accumulates inorganic
selenium in the soil to organic selenium compounds following the sulphur assimilatory pathway
(Shrift 1973). The main form of organoselenium in garlic are γ-glutamyl-Se-methylselenocysteine
and Se-methylselenocysteine (Ip and others 1995, 2000).
Ip and Lisk (1996) showed that the anti-carcinogenic efficacy of selenium in garlic was superior
to selenomethionine. As result garlic enriched with high levels of selenium with superior anti-
carcinogenic properties can be produced (Ip and others 1992). However, the high selenium
levels in enriched garlic make it unsuitable for direct human consumption and a vehicle is
required to incorporate the selenium enriched garlic in human diets. Production of selenium
rich farmed fish through dietary modulation is potential solution but possible toxicity effects
for the fish needs to be avoided.
The effect of selenium (added as Na2SeO3) dietary enhancement on fish has been studied
for rainbow trout (Oncorhyngus mykiss) by Hilton and others (1980). They showed that the
minimum requirement of selenium in feed is 0.07 mg/kg. Chronic dietary selenium toxicity
occurred at 13 mg selenium per kg dry feed. Tissue selenium analysis indicated that trout can
maintain homeostasis with dietary selenium levels up to 1.25 mg/kg dry feed. The selenium
uptake and accumulation in tissues of trout on diets containing in excess of 3 mg/kg dry feed
may ultimately be toxic to trout for if maintained for a longer period. Gatlin and Wilson (1984)
studied the dietary selenium (added as Na2SeO3) requirement of fingerling catfish. A minimum
requirement of 0.25 selenium /kg dry feed was established. Selenium concentrations in edible
muscle tissue increased almost linearly to 3.5 mg/kg with increasing dietary selenium levels
(15 mg/kg). At a level of 15 mg selenium/kg dry feed intoxication effects in the channel
catfish was observed. In order to raise eviscerated body burdens of hatchery-reared coho
salmon to their wild counterparts Felton and others (1996) gave elevated levels of selenium
(as Na2SeO3). The highest dose given was 13.6 mg selenium/kg feed. A dietary level of 8.6
mg selenium/kg feed was capable of inducing eviscerated body burdens similar to those found
in wild fish. Lorentzen and others (1994) investigated whether selenium supplementation is
required in apractical fish-meal based feed for Atlantic salmon. Further the metabolic effects
of supplementing selenite or selenomethionine up to 2 mg/kg was compared with emphasis on
the selenium content of the fillets. Only a minor increase in muscle selenium was concentration
was observed by supplementing selenite. In case of selenomethionine selenium in the muscle
tissue increased from 0.48 mg selenium/kg to 2.51 mg selenium/kg.
In this paper the results are presented of a few studies of farming African catfish (Clarias
goriepinus) with selenium enriched feed. African catfish was fed with a selenomethionine or
with organoselenium enriched in garlic. Besides, the effect of methionine concentration in
the feed on the incorporation of selenomethionine in the muscle tissue of African catfish was
studied.
binnenwerk.indd 194 21-08-2006 12:34:39

Enrichment of functional selenium in farmed African catfish by dietary modulation
Farming studies
For the study with selenomethionine enriched feed 450 individuals of African catfish (Clarias
goriepinus), were distributed in nine tanks of 35 litres. The water temperature was kept at 25 ±
0.9 °C. The fish were conditioned during two weeks by feeding basal fish meal diets (Table 1)
containing no supplemental selenium. The initial average weight of the African catfish per tank
after acclimatization was 51.5 ± 6.4 gram. The tanks with African catfish were individually
supplied with tap-water from IJmuiden. Water was continuously changed (120 l/tank/day).
Once every three days the content of one tank was changed completely.
In case of the study with organoselenium enriched garlic feed the initial average weight of
the African catfish per tank after acclimatization was 100.7 ± 2.7 gram. Each of 18 tanks of
32 litres contained 24 individuals of African catfish. Fishes were allowed to acclimatize to the
experimental conditions for four days prior to the start of the trial. During acclimatization
all fish were fed with a basal fish meal diet (Table 2) containing no supplemental selenium.
Treatments were assigned to the tanks using a random number table, with triplicate tanks for
each of the six treatments. Tanks were flown through with 25 °C tap water at a rate of 0.6 tot
0.8 L/min, yielding a renewal rate of approximately 1 L/hr for each tank. Tank effluents were
discharged as recirculation possibly resulted in transfer of selenium compounds between tanks
that had leached from the experimental feeds.
African catfish were fed in all experiments daily with the experimental diets for six weeks equal
to 2% of biomass. This ratio ensured complete uptake of the feed. The uptake of the feed was
followed by mirrors underneath the tanks. The fish were bulk weighed at the start, after two
and four weeks, whereupon the amount of feed was adjusted. Every two weeks, five fish from
each tank were sampled randomly. The sampled fish were weighed individually. Muscle tissue
was sampled and the selenium content was determined.
Table 1. The composition of the basal fish meal diet (g/kg dry diet) for African catfish, used for the
experimental diets with increasing Se content.
Experiment 1 Experiment 2 Experiment 3
Fish meal1 478 239 450

Soy concentrate - 252 125
Blood meal 43 43 -
Fish oil 58 58 60
Wheat starch 62 62 -
Wheat - - 213
Wheat gluten meal - - 100
Binder 21 21 25
Vitamin/Mineral mix 21 21 20
Monocalciumphoshate - - 7
1 Contained 1.2% dry weight methionine (experiment 1) and 0.9% dry wt methionine (experiment 2).
binnenwerk.indd 195 21-08-2006 12:34:39

Table 2. Final body weight, weight gain and selenium concentration and corresponding standard deviation
in African catfish, fed a diet supplemented with varying selenium concentrations for 6 weeks (n=5)
(Experiment 1, selenomethionine).
Group 1 2 3 4 5 6
Supplemental Se (mg/kg) 0 0.3 0.6 1.0 2.0 4.0

Final body weight (g) 171 ± 24 196 ± 19 166 ± 40 169 ± 36 166 ± 24 183 ± 19
Weight gain (%) 333 ± 47 381 ± 38 324 ± 77 329 ± 69 322 ± 48 35 ± 38
Se muscle (mg/kg) 0.26 ± 0.02 0.36 ± 0.02 0.43 ± 0.04 0.48 ± 0.08 0.74 ± 0.05 1.34 ± 0.04
Experimental diets
Experiment 1 with selenomethionine enriched feed
The composition of the formulated basal fish meal diet was on dry weight basis 60.2% protein,
16.3% fat, 12.5% carbohydrates and 9.1% ash (Table 1). The basal fish meal diet contained 3.4
mg Se/kg which implies that selenium is added to the commercial fish meal. Five additional
experimental diets were prepared with 0.3, 0.6, 1.0, 2.0 and 4.0 mg selenium per kg diet as
selenomethionine as a supplement. Therefore selenomethionine was dissolved in water and
added to the moist pellet (30% moist) which was pelleted through a 2.0 mm dice. The fish
were fed with frozen feed. Analyses showed that the selenomethionine-supplemented diets
contained 3.5, 3.8, 4.1, 5.3 and 6.5 mg Se/kg, respectively.
Experiment 2 with methionine enriched feed

The influence of methionine content in the diet on the incorporation of selenomethionine was
studied in African catfish by varying the methionine content in the feed. Three experimental
diets were prepared with a low (0.9% dry weight methionine), normal (1.2% dry weight
methionine) and high (1.5% dry weight methionine) methionine content. The supplemented
selenium content, as selenomethionine, was 1.0 mg selenium per kg diet. The composition of
the basal fish meal diet for this experiment was on dry weight basis 58.3% protein, 12.8% fat,
18.2% carbohydrates, and 7.3% ash (Table 1). Analyses showed that the selenium content in
the low, normal and high methionine diet was 2.5-2.8 mg/kg.
Experiment 3 with selenium garlic enriched feed

Treatments consisted of five groups with different dietary levels of organoselenium from
selenium enriched garlic (Table 3). One group of African catfish was fed with feed with no garlic
added. Organoselenium originated from garlic grown in selenium enriched soil. Garlic converts
and accumulates inorganic selenium present in the soil to organoselenium compounds following
the sulfur assimilatory pathway. The different selenium levels in the experimental feeds were
established by including garlic with high and low selenium content in different ratios. The total
amount of garlic, 12.1 g dry matter/kg, was equal in all these experimental feeds.
The composition of the formulated basal fish meal diet was on dry weight basis 46.8% protein,
13.6% fat, 14.8% carbohydrates and 10.1% ash (Table 1). The basal fish meal diet contained
0.14 mg selenium/kg which implies that selenium is added to the commercial fish meal.
Five additional experimental diets were prepared with 1.9 - 8.5 mg selenium per kg diet as
organoselenium in garlic. The fish were fed with frozen feed.
binnenwerk.indd 196 21-08-2006 12:34:40

Table 3. Final body weight, weight gain and selenium concentration and corresponding standard deviation
in African catfish, fed a diet supplemented with varying selenium concentrations for 6 weeks (n=5)
(Experiment 3, organoselenium from garlic).
Group 1 2 3 4 5 6
Se in feed (mg/kg) 1.92 2.77 3.94 5.14 8.51 1.92

Final body weight (g) 450 ± 19 438 ± 15 406 ± 21 432 ± 13 423 ± 15 395 ± 45
Weight gain (%) 344 ± 27 340 ± 4 310 ± 25 330 ± 21 315 ± 10 289 ± 38
Se muscle (mg/kg) 0.24 ± 0.01 0.35 ± 0.03 0.47 ± 0.03 0.60 ± 0.01 0.85 ± 0.06 0.26 ± 0.02
Selenium analyses
Homogenized samples of the muscle tissue were microwave digested in microwave lined
digestion vessels using 70% (m/v) nitric acid (5 ml) and 30% (m/v) hydrogen peroxide (5
ml) as the oxidant mixture (3). After digestion, the samples were reduced in the microwave
by adding 5 ml 37% (m/v) hydrochloric acid in order to convert Se6+ into Se4+. The resulting
solution was diluted to a final volume of 25 ml using 10% hydrochloric acid. Blanks (5 ml
70% (m/v) nitric acid and 5 ml 30% (m/v) hydrogen peroxide) were treated in the same way.
Total selenium was determined by hydride generation Flow Injection Analysis System - Atomic
Absorption Spectrometry (FIAS-AAS). NaBH4 (0.2%) in NaOH (0.05%) was used to generate the
H2Se. The accuracy of the analyses was established by analyzing the selenium content in the
BCR certified Cod-CRM422 reference sample.
Farming study
In all experiments, mortality was negligible. In Table 2 and 3, the final body weight, percent
weight gain and selenium concentration in the muscle tissue of African catfish, are presented
in relation to the selenium dose supplemented in the diet for six weeks. African catfish growth
was not affected by the dietary selenium level. Selenium content in the muscle tissue increased
with increasing dietary selenium levels. The presence of garlic has a significant positive effect
on growth rate. The final body weight of African catfish fed without garlic (395 ± 45 g) is on
the average approx. 8% lower then in groups of African catfish receiving garlic in their diet.
Figure 1 and 2 shows a linear relationship between the added dietary selenium as
selenomethionine (Experiment 1) respectively organoselenium from selenium enriched garlic
(Experiment 3) and the muscle selenium content in the African catfish. The increase of selenium
content in the muscle tissue is approx. 3 times higher in case of selenomethionine then in case
organoselenium compounds from selenium enriched garlic. This might be due to differences in
selenium metabolism of the different selenium compounds.
This study shows that supplementing the diet for African catfish with selenium using
selenomethionine or organo selenium present in selenium enriched garlic can increase the
selenium content in the edible which might increase the intake of humans who eat regularly
African catfish. This increase is in agreement with the results from Gatlin and Wilson (1984),
who also demonstrated a progressive selenium increase in channel catfish muscle during a
binnenwerk.indd 197 21-08-2006 12:34:40

1.4
Muscle tissue selenium content (mg/kg) y = 0.2693x + 0.218
1.2 R2 = 0.9929
0.8
0.6
0.4
0.2 after conditioning
0
0 0.5 1 1.5 2 2.5 3 3.5 4
Supplemented dietary selenium content (mg/kg)
Figure 1. Selenium content in muscle tissue of African catfish farmed with selenomethionine enriched feed
(Experiment 1).
1.00
0.90 y = 0.0911x + 0.0953
Total [Se] in fillet (mg/kg)
R2 = 0.9851
0.80
0.70
0.60
0.50
0.40
0.30
0.20
0.10
0.00
0 2 4 6 8 10
Total [Se] in feed (mg/kg)
Figure 2. Selenium content in muscle tissue of African catfish farmed with organoselenium from selenium
enriched garlic feed (Experiment 3).
farming period of 15 weeks. However Lorentzen and others (1994) showed no increase in the
selenium muscle content of salmon in case selenite was added to the feed. But the addition
of selenomethionine in the feed for salmon increased the level of selenium in the edible part
approximately four times after a feeding period of 8 weeks.
binnenwerk.indd 198 21-08-2006 12:34:40

In Table 4 the final body weight, percent weight gain and the selenium concentration in the
fish muscle are shown of the feeding trial with different content of methionine. African catfish
growth was not effected by dietary methionine level and neither was the concentration of
muscle selenium.
In the metabolism of animals, selenomethionine may have two major pathways, that of methionine
and that of selenium (Burk 1976). In protein synthesis, no difference is made between methionine
and selenomethionine, and therefore selenomethionine will be incorporated into proteins as
an analogue of methionine (Sunde 1984). These two selenomethionine pathways may explain
the large increase of selenium found in African catfish muscle farmed with selenomethionine
enriched feed. Besides, the main protein pools of fish are present in the muscle tissue, so it is
expected, especially in fast growing fish, that muscle selenium increases when selenomethionine
is incorporated into proteins (Lorentzen and other 1994)). The methionine content of the diet
and the rate of protein synthesis of the fish are critical for the pathway selenomethionine
will follow. A diet, which is low in methionine content, will enhance the incorporation of
selenomethionine into proteins. The results of experiment 2 show that there is no difference in
muscle selenium concentration between a low methionine diet and a normal or high methionine
diet. An explanation for this result could be that the normal methionine diet provided already
more than sufficient in the methionine need of the African catfish.
The absorption of γ-glutamyl-Se-methylselenocysteine and Se-methylselenocysteine, being the

main forms in garlic, has been studied in rats (Dong and others 2001) given a diet with 3 mg
/kg of each component in their feed for 50 days. Liver and kidney were able to retain more
selenium (~ 3-fold increase) when compared with mammary gland, muscle and plasma (~ 2-fold
increase or less). No speciation was done of the selenium in the various samples investigated.
Ip (1998) concluded that Se-methylselenocysteine can not be incorporated nonspecifically into
proteins. In our study the content of Se-methylselenocysteine at the highest selenium dose
was 0.85 mg/kg dry feed and of γ-glutamyl-Se-methylselenocysteine 2.8 mg/kg dry feed. An
on-going study about the speciation of the organoselenium compounds in the muscle tissue of
African catfish shows that selenium γ-glutamyl-Se-methylselenocysteine is not detectable in
an extract of the muscle tissue. But Se-methylselenocysteine is present in detectable amounts
besides selenomethionine.
Table 4. Final body weight, weight gain, methionine content and selenium concentration and the
corresponding standard deviation in African catfish muscle, fed with a diet supplemented with 1 mg Se/kg
feed as selenomethionine and varying methionine concentrations for 6 weeks (n=5) (Experiment 2).
Methionine diet low normal high
Methionine % dry wt 0.9 1.2 1.5

Selenium added as SeMet (mg/kg) 1 1 1
Final body weight (g) 140 ± 23 156 ± 30 156 ± 11
Weight gain (%) 272 ± 44 303 ± 64 304 ± 21
Se muscle (mg/kg) 0.42 ± 0.02 0.41 ± 0.05 0.47 ± 0.03
binnenwerk.indd 199 21-08-2006 12:34:40

Conclusions
African catfish growth was not affected by the dietary selenium level. The selenium content
in the edible part of the African catfish increased linearly with increasing dietary selenium
levels. The increase was approx. 3 times higher using selenomethione in the feed then with the
organoselenium (γ-glutamyl-Se-methylselenocysteine and Se-methylselenocysteine) compounds
from selenium enriched garlic supplemented to the feed. However possible health beneficial
effect of selenium from fish will also depends on the chemical form in which the selenium
compounds are present.
In this study, there was no difference in the muscle selenium concentration of African catfish
fed with a low methionine diet, a normal or high methionine diet.
Acknowledgements
This work was in part financed by the EU-Commission under FAIR PL-95.771 “Speciation and
bioavailability of selenium from processed and tailor-made fishery products” and the FP6
Integrated Project SEAFOODplus, contract No FOOD-CT-2004-506359.
References
Burk RF. 1976. Selenium in man. In: Prasad AS. Trace elements in human health and disease. London. Academic press.
P 105-134.
Clark LC, Combs Jr GF, Turnbull BW, Slate HE, Chalker DK, Chow J, Davis LS, Glover RA, Graham GF, Gross EG, Krongrad
A, Lesher Jr JL, Park HK, Sanders Jr BB, Smith CL, Taylor JR 1996. Effects of selenium supplementation for cancer
prevention in patients with carcinoma of the skin. A randomized controlled trial. Nutritional Prevention of Cancer
Study Group. JAMA 276:1957-1963
Dong Y, Lisk D, Block E, Ip C. 2001. Characterization of the biological activity of γ-glutamyl-Se-methylselenocysteine.
Cancer Research 61:2923-2928
Felton SP, Landolt ML, Grace R. 1996. Effects of selenium dietary enhancement on hatchery-reared coho salmon,
Oncorhynchus kisutch (Walbaum), when compared with wild coho: hepatic enzymes and seawater adaptation
evaluated. Aquaculture Research 27:135-142.
Gatlin DM, Wilson RP. 1984. Dietary selenium requirement of fingerling channel catfish. J. Nutr., 114:627-633.
Hilton JW, Hodson PV, Slinger SJ. 1980. The requirement and toxicity of selenium in Rainbow trout (Salmo gairdneri).
J. Nutr. 110:2527-2535.
Ip C, Lisk DJ, Stoewsand GS. 1992. Mammary cancer prevention by regular garlic and selenium enriched garlic. Nutr.
Cancer 17:279-286.
Ip C, Lisk DJ. 1995. Efficacy of cancer prevention by high selenium-garlic is primarily depent on the action of selenium.
16:2649-2652
Ip C, Lisk DJ. 1996. The attributes of selenium-enriched garlic in cancer prevention. Adv Exp Med Biol 401:179-187
Ip C. 1998. Lessons from basic research in selenium and cancer prevention. J. Nutr. 128:1845-1854.
Ip C, Birringer M, Block E, Kotrebai M, Tyson JF, Uden PC, Lisk DJ. 2000. Chemical speciation influences comparative
activity of selenium-enriched garlic and yeast in mammary cancer prevention. J. Agr. Food Chem., 48:2062-2070
Lorentzen M, Maage A, Julshamn K. 1994 Effects of dietary selenite or selenomethionine on tissue selenium levels of
Atlantic salmon (Salmo salar ). Aquaculture 121:359-367.
Rayman MP. 2000. The importance of selenium to human health. Lancet 356:233-241
Sunde RA. 1984. The biochemistry of selenoproteins. J. Am. Oil Chem. Soc. 61:1891-1900.
binnenwerk.indd 200 21-08-2006 12:34:41

Comparison of commercial and experimental slaughter of farmed
carp (Cyprinus carpio) with respect to development of rigor mortis
and flesh quality
Hans van de Vis1, Henryk Bialowas2, Maciej Pilarczyk2, Marcel Machiels1, Henny Reimert3,
Martine Veldman1 and Bert Lambooij3
1Institute for Marine Resources & Ecosystem Studies (Wageningen IMARES), PO Box 68, 1970 AB
IJmuiden, The Netherlands
2Polish Academy of Sciences, Institute of Ichthyobiology and Aquaculture, Zaborze, ul. Kalinowa 2,
43-520 Chybie, Poland
3Animal Sciences Group, Wageningen University and Research Centre, Division of Food and
Nutrition, PO Box 65, 8200 AB, Lelystad, The Netherlands
Abstract
The objective of the study was to compare the commercial slaughter procedure, asphyxia
followed by a blow on the head, to an experimental method, electrical stunning followed by
chilling. Farmed carp (Cyprinus carpio) was used in the study. Effects of the slaughter methods
were investigated by analysis of onset and resolution of rigor mortis (rigor index values).
Changes in quality profile during storage of the fillets at 0 °C were assessed by analysis of colour
(L* (lightness), a* (green-redness), and b* (blue-yellowness values), pH value and K-values.
The commercial method did not result in a more rapid onset and intensity of rigor mortis for
the commercial method, compared to the experimental one. Maximal rigor index values were
recorded after 24 h of storage for both methods and a complete resolution of rigor mortis did
not seem to occur as after 5 days of storage at 0 °C the rigor index curve levelled off at 72%.
Persistently lower L* and higher a* values were not found for fillets that were obtained by
the experimental method, compared to the commercial one. It appeared, however, that flesh
obtained by application of the commercial method had persistently higher b* values (p< 0.05)
than the commercial method. During the entire storage period of the fillets pH values were
higher after electrical stunning (p < 0.05), compared to the commercial method. The difference
in pH may be attributed to stress that occurred during asphyxia. However, this was not reflected
into the evolution of the K-values, as no persistent differences (p< 0.05) were found between
the methods. The observations that the slaughter methods only influenced the pH values
persistently suggests that other mechanisms might have influenced the evolutions the other
product quality parameters
Keywords: carp, rigor mortis, pH, colour
Introduction
Since the last decades consumers, supermarkets, governments, farmers, processors of fish and
animals welfare and consumers’ associations have become increasingly concerned about fish
welfare. This concern also comprises the protection of welfare of fish at slaughter.
binnenwerk.indd 201 21-08-2006 12:34:41

H. van de Vis, H. Bialowas, M. Pilarczyk, M. Machiels, H. Reimert, M. Veldman and B. Lambooij
Having in mind the concern that mammals and birds should be spared unnecessary suffering
at slaughter or killing, the question is whether this concern is also relevant for fish. For man
it is known that awareness of pain, fear and stress apparently depends on functions of specific
regions of the massively developed cerebral cortex, i.e. a well-developed frontal lobe and
neocortex. As fish lack these brain regions, it can be argued that fish do not have a capacity
to experience pain and fear (Bermond 1997; Rose 2002). It should be emphasized that using
the logic of Rose (2002) and Bermond (1997), it has to be concluded that the capacity of
consciously experience pain and fear is likely to be absent in very young children and all
animals except higher primates.
In reviews of Braithwaite and Huntingford (2004) and Chandroo and others (2004) evidence
is presented that the suggestion fish cannot experience pain and fear is not supported by the
current literature. For feeling of pain fish should, besides a nociceptive system, which is needed
for detection of potentially injurious stimuli, also possess cognitive capacities. Cognitive
capacities are also needed to consciously experience fear. With respect to cognitive capacities
Braithwaite and Huntingford (2004) and Chandroo and others (2004) discussed literature that
suggests that fish have an ability to detect, conceptualize and respond to nociceptive stimuli.
In the context of the issue on welfare, it has to be emphasized that there are big gaps in our
knowledge on fish welfare. Nevertheless, current literature suggests that from an ethical point
of view welfare aspects should be taken into account during the production of food fish.
In order to protect welfare of fish at slaughter the animals should be rendered unconscious prior
to killing. For eel we have established that electrical stunning is suitable to avoid unnecessary
stress during the process of slaughter (Lambooij and others 2002). In addition, Morzel and
Van de Vis (2003) showed that reduction of stress at slaughter can improve product quality
parameters. With respect to farmed carp (Cyprinus carpio) little is known about effects of
slaughter methods on product quality. Carp is appreciated by consumers in various European
countries and is therefore intensively farmed. Annually 445 000 tons of carp is produced in
Europe. Farmers are becoming interested in research that could help them to improve the
quality of their products.
Therefore the objective of the study was evaluation of flesh quality of farmed carp (Cyprinus
carpio) immediately following electrical stunning in combination with chilling. Flesh quality was
compared to that of a batch slaughtered by the currently applied commercial method (asphyxia
for 30 min followed by a blow on the head). Asphyxia is performed in practice to exhaust the
carps to facilitate the application of the blow on the head. Product quality of carp was evaluated
by analysis of rigor mortis, pH, colour, and ATP and its breakdown products during storage.
K-value, first defined by Saito and others (1959), is a widely accepted indicator of fish
freshness. Increasing K-values have been correlated with loss of freshness in many fish species.
Furthermore, pre-slaughter and slaughter stress has been repeatedly reported to lead to faster
ATP degradation and higher K-value (Lowe and others 1993; Morzel and Van de Vis 2003; Sigholt
and others 1997).
binnenwerk.indd 202 21-08-2006 12:34:41

Comparison of commercial and experimental slaughter of farmed carp
Fish
Carps (Cyprinus carpio) were kept in a recirculation tank in the Institute of Ichthyobiology
and Aquaculture of the Polish Academy of Sciences. For both slaughter methods fishes were
collected from the same tank, using a hand net. The average live weight of the fishes was 1308
+ 183 g. Prior to slaughter the fishes were fasted for 4 days.
Slaughter methods for carp

Each individual carp was electrically stunned for 5 s in a Plexiglas box filled with (size 50 x 20
x 70 cm) tap water. The bottom and top plate electrodes measured an area of 648 cm2 each
at 16 cm distance. The power supply (Stork RMS, Lichtenvoorde, the Netherlands) delivered
a constant voltage of 400 V (600 mA, 50 Hz a.c). Immediately after stunning, the carp was
removed from the tank and placed in flake ice for a period of 15 min to kill them. In another
study we assessed by registration of EEGs that under these conditions carps could be stunned
immediately by electricity. Recovery could be prevented when the stunned carp were placed in
ice (Lambooij and Van de Vis unpublished results). For assessment of the experimental method
25 fishes were stunned and killed.
The second batch (n=25) was slaughtered by application of a commercial method, which
consisted of asphyxia for 30 min at room temperature in combination with a manually applied
blow on the head of each individual fish.
Rigor mortis
The rigor mortis Index values (RI) were measured for 10 whole fishes of each batch. The
measurements were performed 0, 15, 23.5, 39, 47, 63, 71, 87 and 111 hours after application
of the experimental or commercial method. During this time, fishes were kept in polystyrene
boxes filled with melting flake ice.
The method to calculate RI values is the following (Bito and others 1983). The sag of the tail
is measured when the front half of the fish’s body is placed on a horizontal table. The RI value
is obtained from the equation:
(Dt - Do)
RIt(%) = 100 * Do
where Do and Dt represent the distance of the base of the caudal fin from the horizontal line of
the table, as measured pre rigor and at subsequent intervals during storage, respectively. The
fishes were stored flat between measurements.
pH measurements
The pH was followed 0, 12, 24, 48, 72, 96, 120, 144, 168 hours post slaughter. Analysis of pH
was performed directly in the tissue of each right-hand side fillet on the visceral side using a
spear electrode. The pH was measured with a pH meter at three places at the dorsal side of the
fillet: rostral, mid-axial and caudal. From each batch 15 fillets were analysed.
Colour measurements
The colour of the fillets was monitored using a Chroma-meter CR-200 (Minolta Konica, Japan).
Measurements were taken in triplicate on the visceral and skin side of the right-hand fillet of
binnenwerk.indd 203 21-08-2006 12:34:41

the 15 fillets of each batch, which were also used for analysis of pH. The measurements were
carried out mid-axial on the dorsal, lateral and ventral sides. Chromatic values are given in the
CIE-L*a*b* system. The L* value is a measure for lightness. The a* and b* are the chromacity
coordinates. The a* and b* indicate colour directions: +a* is the red direction, -a* is the green
direction, +b* is the yellow direction and –b* is the blue direction (Warris 1996).
ATP and its degradation products for K-value

Samples were taken from the left-hand side fillets 0, 24, 72, 120 and 169 hours after slaughter.
From each batch 15 left-hand side fillets were stored in flake ice for analysis. For the sampling
we selected fillets from 10 fishes of each batch that have been stunned prior by applying the
commercial or experimental method. The samples were taken mid-axial from the dorsal side. The
samples were immediately put into liquid nitrogen to freeze them as quickly as possible to stop
enzymatic activity. Subsequently, they were stored in the freezer at –80 ˚C until analysis.
Prior to extraction, the samples were thawed. Care was taken to keep the temperature of
the thawed samples below 4 °C. Extraction and analysis of the extracts were performed, as
described by Veciana-Nogues and others (1997).
Data analysis
Rigor mortis index (RI) values were analysed by ANOVA. For the pH evolution a forth-order
polynomial regression model with time after slaughter as independable variable was used to fit
the observations. The various coefficients were tested for a difference with 0. If a coefficient
did not differ significantly from 0 the term was omitted from the model.
The evolution of the L*, a* and b* values were analysed by linear regression. For the evolution
of K-value we established whether there was a significant difference between the mean values
obtained at each measurement interval. For each measurement interval the least significant
difference was calculated and compared with the difference of the mean values of each
treatment. This approach was used, as it appeared that K values could not be related to the
measurements by using non-linear regression. In the data analysis no distinction was made
between male and female carps.
Ethics
The experiments were approved beforehand by a governmental experimental committee.
Rigor mortis
Statistical analysis revealed a significant difference between the batch slaughtered by the
experimental method and the commercial one. It appeared that for the commercial method the
RI values were persistently lower than for the experimental method (p < 0.0014). A graphical
representation of measured data is shown in Figure 1. For both slaughter methods the fishes
were in full rigor 24 h post mortem. A complete resolution of rigor mortis did not seem to occur,
as after 5 days of storage at 0 °C the rigor index curve levelled off at approximately 70% for
the both methods.
Development of rigor in relation to various stressors has been studied by many authors. Pre-
slaughter stress due to crowding, capture or harvesting always accelerated the process of rigor
binnenwerk.indd 204 21-08-2006 12:34:41

100
80
RI values (%)
60
40 electricity +
chilling
asphyxia + blow
20 on the head
0
0 24 48 72 96 120
Time (hours)
Figure 1. Evolution of rigor mortis in carp slaughtered by electricity followed by chilling or asphyxia followed
by a blow on the head. RI means rigor mortis index.
(Lowe and others 1993; Skjervold and others 2001). Contrary to these studies, we observed that
stress, which might occur during asphyxia, did not accelerate the onset of rigor in the carps,
compared to the use of electricity. This suggests that during asphyxia other sources of energy
for the movements of the carps might have been used than ATP, as the onset of rigor mortis
occurs when the ATP is nearly depleted (Erikson and others 1997).
As to the effect of slaughter itself, fewer studies have been performed. Morzel and others
(2002) observed that turbot stunned by electricity entered more rapidly into rigor mortis than
the fishes that were stunned slowly by bleeding.
pH
Average pH values over a seven-day period of post mortem storage of carps slaughtered by the
experimental and commercial method are presented in Figure 2. In Figure 2 for each batch
average pH values of rostral, mid-axial and caudal measurements are shown. Statistical analysis
revealed that flesh of carps killed by the experimental method had significantly higher pH
values during the entire storage period, compared to the batch slaughtered by the commercial
method. This was observed for the rostral, mid-axial and caudal measurements of the pH
performed with carps from both batches (p<0.05). The established models are shown as lines
in Figure 2. The markers depict the averages of the measured values.
Low pH is classically used as an indicator of acute stress and activity at the time of slaughter
in many animal species including fish. Our results suggest that the commercial method may be
stressful to carp, as its application resulted in significantly lower pH values, compared to the
experimental method (p < 0.05). Lower initial pH associated with higher ante mortem stress
has been described by numerous authors in salmonids (Azam and others 1989; Sigholt and
others 1997; Thomas and others 1999), in turbot (Morzel and others 2002) or in eel (Morzel
and Van de Vis 2003). This was related to depletion of energy reserves, mainly glycogen, with
production and accumulation of lactate. However, electro stimulation might have occurred
binnenwerk.indd 205 21-08-2006 12:34:42

rostral electricity + chilling

6.90
asphyxia + blow on
6.80 the head
model experimental
6.70 model commercial
6.60
pH
6.50
6.40
6.30
0 50 100 150 200
Time (hours)
mid-axial electricity + chilling

6.90 asphyxia + blow on
6.80 the head
model experimental
6.60
pH
6.50
6.40
6.30
0 50 100 150 200
Time (hours)
caudal electricity + chilling

6.90 asphyxia + blow on
6.80 the head
model experimental
6.60
pH
6.50
6.40
6.30
0 50 100 150 200
Time (hours)
Figure 2. pH changes in carp slaughtered by electricity followed by chilling or asphyxia followed by a blow
on the head: rostral, mid-axial and caudal measurements.
binnenwerk.indd 206 21-08-2006 12:34:42

during the application of the experimental method, as this results in an immediate fall of pH
in warm-blooded animals (Hwang and others 2003).
In our study we observed for both batches that after 48 hours of storage pH values increased
and maximum values were reached at 72 hours. In the later stage of storage for both batches
the pH values decreased. Sikorski and others (1990) reported that an increase to neutral
alkaline pH values can be due to the accumulation of nitrogenous compounds generated by both
endogenous and bacterial enzymes. However, it is not likely that this occurred in our study, as
the increase in pH values was observed after 2 to 3 days of storage at 0 °C.
Colour
Contrary to our expectations, the experimental slaughter method did not result in persistently
lower L* and higher a* values compared to the commercial method (p <0.05), as this occurred
for both eel (Morzel and Van de Vis 2003) and turbot (Morzel and others 2002) when electricity
was applied and compared to a slow commercial slaughter method.
For blue-yellowness persistently higher values were obtained for mid-axial lateral, and mid-axial
ventral analysis of fillets slaughtered by the commercial method, compared to those obtained
by the experimental method (p < 0.05). However, the differences between the average b*
values of each batch ranged from 0.5 to 0.8 and from 0.4 to 0.8 for the mid-axial lateral and
mid-axial ventral measurements, respectively. It is not likely that these small differences can
be distinguished visually, as humans are only able to detect differences ranging from 1 to 2
(R Kranen personal communication).
Assessment of the colour measurements taken on the three different locations at the visceral
side of the fillets revealed that the range of differences for the average L*, a* and b* was,
in general, larger than the differences related to the slaughter methods (see Table 1 and 2).
For the comparison the largest differences were selected for each colour parameter in Table
1 (indicated in italics) and compared to the differences in Table 2. The differences that are
larger than those in Table 1 are indicated in bold in Table 2. It appeared when the L*, a* and
Table 1. Range of differences in L*, a* and b* values related to the location of measurements on carp
fillets.
Colour parameter Location Range of difference between commercial and

experimental methods
L*value mid-axial dorsal -1.4 to 0.3

L*value mid-axial lateral 0.1 to 1.2
L*value mid-axial ventral 0.2 to 1.7
a*value mid-axial dorsal -0.6 to 0.8
a*value mid-axial lateral -0.2 to 0.8
a*value mid-axial ventral -0.1 to 0.9
b*value mid-axial dorsal -0.6 to 1.2
b*value mid-axial lateral 0.5 to 0.8
b*value mid-axial ventral 0.4 to 0.8
binnenwerk.indd 207 21-08-2006 12:34:42

Table 2. Range of differences in L*, a* and b* values according to the slaughter technique used.
Slaughter method Colour Range of difference of colour parameters measured at different locations
parameter on the fillet
mid-axial dorsal - mid-axial dorsal - mid-axial lateral -

mid-axial lateral mid-axial ventral mid-axial ventral
Commercial L*value 0.5 to 1.0 -3.6 to -0.6 -4.3 to -1.2

a*value -1.1 to -0.2 -2.7 to -0.3 -2.1 to -0.2
b*value -0.5 to -0.1 -2.7 to -0.1 -2.5 to 0.0
Experimental L*value 0.9 to 1.6 -2.2 to 0.9 -3.8 to -0.6
a*value -0.8 to 0.6 -2.5 to 0.2 -2.2 to 0.0
b*value -0.5 to 0.8 -2.8 to 0.1 -2.3 to -0.1
b* average values that were obtained from mid-axial dorsal and mid-axial lateral measurements
were compared to those in Table 1 no differences in the ranges were found.
With regard to storage in practice we would like to point out that in Poland carp is in general
eaten on the day of slaughter. Nevertheless, we decided to measure colour parameters as
function of time, as the carp processing industry in Poland is maturing, which will result in
storage of products at retailers.
K value
Evolution of K-values with time is shown in Figure 3. Contrary to our expectation, K-values
were not consistently and significantly (p<0.05) higher in carps slaughtered according to the
70
60
50
K value (%)
40
30
electricity + chilling
20 asphyxia + blow on
the head
10
0
0 50 100 150 200
Time (hours)
Figure 3. Mean K value measured over a 168 hours period post mortem in raw carp muscle according to the
slaughter technique.
binnenwerk.indd 208 21-08-2006 12:34:43

commercial method. Only for the samples taken after 120 h of storage we established that the
mean K-value of the batch slaughtered by the experimental method was significantly lower (p
< 0.05) than the mean value for the batch slaughtered by the commercial method.
Conclusion
The present study indicates that the alternative experimental slaughter technique resulted
in significantly higher pH values, compared to the commercial method. However, this is not
reflected into a more rapid onset and intense of rigor mortis for the commercial method,
compared to the experimental one. We also observed no persistent effect of slaughter on the
evolution of L*, a* and b* values and K-values of the fillets with time. These results suggest
that stress or electro-stimulation may have affected the outcome, however, other mechanisms
may also contribute. Further research is required to evaluate the relationship between slaughter
methods and product quality parameters of carp in more detail.
Acknowledgements
The study was financed by the European Integrated Research Project SEAFOODplus (contract
number 506359) and the Dutch Ministry of Agriculture, Nature and Food Quality.
References
Azam K, Mackie IM, Smith J. 1989. The effect of slaughter method on the quality of rainbow trout (Salmo gairdneri)
during storage on ice. Int J Food Sci Technol 24:69-79.
Bermond B. 1997. The myth of animal suffering. In: M. Dol and S. Kasamoentalib, S. Lijmbach, E. Rivas and R. van den
Bos, editors. Animal consciousness and animal ethics. Van Gorcum and Co., Assen, the Netherlands. 125-143.
Bito M, Yamada K, Mikumo Y, Amano K. 1983. Studies on rigor mortis of fish – I. Difference in the mode of Rigor mortis
among some varieties of fish by modified Cutting’s method. Bull Tokai Reg Fish Res Lab 109:89-96.
Braithwaite VA, Huntingford FA. 2004. Fish and welfare: do fish have the capacity for pain perception and suffering?
Animal Welfare 13:87-92.
Chandroo KP, Duncan IJH, Moccia RD. 2004. Can fish suffer?- perspectives on sentience, pain, fear and stress. Appl
Animal Behav Sci 86:225–250.
Erikson U, Beyer AR, Sigholt T. (1997). Muscle high-energy phosphates and stress affect K-values during ice storage of
Atlantic salmon (Salmo salar). J Food Sci 62:43-47.
Hwang IH, Devin CE, Hopkins DL. 2003. The biochemical and physical electrical stimulation of beef and sheep
tenderness. Meat Science 65:677-691.
Lambooij E, Van de Vis JW, Kuhlmann H, Münkner W, Oehlenschläger J, Kloosterboer RJ, Pieterse C. 2002. A feasible
method for humane slaughter of eel (Anguilla anguilla L.): electrical stunning in fresh water prior to gutting.
Aquaculture Res 33:643-652.
Lowe TE, Ryder JM, Carragher JF, Wells RMG. 1993. Flesh quality in snapper, Pagurus auratus, affected by capture stress.
J Food Sci 58:770-773, 796.
Morzel M, Sohier D, van de Vis H. 2002. Evaluation of slaughtering methods for turbot with respect to animal welfare
and flesh quality. J Sci Food Agric 82:19-28.
Morzel M, van de Vis JW. 2003. Effect of the slaughter method on the quality of raw and smoked eels (Anguilla anguilla
L.). Aquaculture Res 34:1-11.
Rose JD. 2002. The neurobehavioural nature of fishes and the question of awareness and pain. Fish Sci 10(1):1–38.
Saito T, Arai K, Matsuyoshi M. 1959. A new method for estimating the freshness of fish. Bull Jap Soc Sci Fish 24:759-
751
binnenwerk.indd 209 21-08-2006 12:34:43

Sigholt T, Erikson U, Rustad T, Johansen S, Nordtvedt TS, Seland A. 1997. Handling stress and storage temperature
affect meat quality of farmed-raised Atlantic salmon (Salmo salar). J Food Sci 62:898-905.
Sikorski ZE, Kolakowska A, Burt JR. 1990. Post harvest biochemical and microbial changes. In Z.E. Sikorski, editor,
Seafood: resources, nutritional composition and preservation, CRC Press Inc., Boca Raton, 55-75.
Skjervold PO, Fjæra SO, Øtsby PB, Einen O. 2001. Live-chilling and crowding stress before slaughter of Atlantic salmon
Thomas PM, Pankhurst NW, Bremner HA. 1999. The effect of stress and exercise on post-mortem biochemistry of Atlantic
salmon and rainbow trout. J Fish Biol 54:1177-1196.
Veciana-Nogues MT, Izquierdo-Pulido M, Vidal-Carou MC. 1997. Determination of ATP related compounds in fresh and
canned tuna fish by HPLC. Food Chem 59(3):467-47.
Warris PD. 1996. Instrumental measurements of colour. In: SA Taylor, A Raimundo, M Severini and FJM Smulders, editors:
Meat Quality and Meat Packaging. ECCEAMST, Utrecht, The Netherlands, 221-231.
binnenwerk.indd 210 21-08-2006 12:34:43

Chapter 3:
Consumers knowledge, perception, need for
information about seafood
Food problems such as BSE, dioxine and food- and mouth disease have increasingly made the
consumer worried about their quality and safety of food. Consumers trust, knowledge and
interest information about seafood are important issues nowadays.
In the first paper of this chapter answers are given on the importance of traceability for
establishing consumer trust in seafood? How much and what type of information is needed
to establish consumer trust? Will consumer trust be influenced by the source of information?
In the last paper a striking example is presented how the actual situation is regarding the
traceability in Norwegian fish industry and the food retail trade. Additionally the companies
readiness for recall of products from wild and farmed fish is tested.
Consumer product knowledge is an important factor in understanding consumer decision-making

because of its role in information search and information processing. In the second paper
European consumer’s objective and subjective knowledge about fish, their interest in and
need for information related to fish and to formulate recommendations for more effective
communication is assessed. Results will be given whether providing more information aiming at
increasing objective or subjective knowledge would help to increase the consumption of fish.
Salted and dried products of cod, in Portugal named “bacalhau”, are in general not very well
known outside the few, but for Norway important markets where the product is consumed. In
third paper of this chapter a historical overview is given of the production of “bacalhau” and
use in Portugal. For Norway as a main supplier of “bacalhau” to the Portuguese market it is
important to know how the Portuguese customers react to the increased number of “bacalhau”
and try to optimize the product portfolio in such a way that the consumers’ quality demands
are met. In the third paper the results of a preliminary product consumer test in Lisboa are
presented in which consumers evaluated the same fish both as a whole product (imitating the
buying process) and through tasting of a desalted sample from the same lot.
binnenwerk.indd 211 21-08-2006 12:34:43

binnenwerk.indd 212 21-08-2006 12:34:43
Too much or too little information? The importance of origin and
traceability for consumer trust in seafood in Norway and Germany
Arne Dulsrud1, Hans Martin Norberg2 and Thorsten Lenz3

1Statens institutt for forbruksforskning (National Institute for Consumer Research), PO box 4682
Nydalen, N-0405 Oslo, Norway
2Fiskeriforskning, PO box 6122, N-9291 Tromsø, Norway
3Bundesforschungsanstalt für Ernährung und Lebensmittel (Federal Research Center for Nutrition
and Food), Haid-und-Neu Straße 9, D-76131 Karlsruhe, Germany
Abstract
The significance of information to establish consumer trust is disputed. Consumers of seafood

depend on complex chains of distribution. Using qualitative data from consumer focus groups
and interviews with representatives from the fish industry in Norway and Germany, opinions
about what type of information consumer’s desire were compared. It was found that the
conditions for consumer trust vary considerably between the countries. The findings do not
only indicate that the source of information influences consumer trust, but also that the
significance of the source varies between Norway and Germany. These results will have relevance
for trust-building strategies both among private actors and public food safety authorities in
Germany and Norway.
Keywords: consumers, trust, traceability, origin, focus groups
Introduction
As consumers of fish we depend on complex and dynamic systems of food provision consisting
of long chains of impersonal and often unknown actors. This situation does not only pose a
challenge to producers and processors of seafood but also to supervisory (regulatory) authorities
whose objective is to ensure that food is perceived to be safe. For a food safety authority,
information and documentation of the quality of food is one of the most important means to
create consumer trust. Consumer trust implies that consumers perceive the risk handling of the
product sector, retailers and food safety authority as acceptable. In that respect it is important
to understand what conditions confidence-inspiring information is based on.
The consequence of information related to the establishment of consumer trust is not clear,
neither empirically nor theoretically. Type of trust and circumstances the consumer face seem
to matter (Kjærnes and Dulsrud 1998). In that respect an exploration of consumers’ trust in
seafood is of interest. As a perishable food, seafood imposes demands on a comprehensive
quality control. In contrast with other foodstuffs there are few reliable standards for quality
on seafood and variations in quality seem to be more random and unpredictable for seafood
(Anderson and Anderson 1991). Seafood is both an industrial input factor and a finished product
which is distributed over long distances and through many connecting links. Therefore, in many
cases the distribution becomes complex and is often perceived as diffuse. Given the fact that
the trade in seafood is international and global, the establishment of consumer trust requires
binnenwerk.indd 213 21-08-2006 12:34:43

communication across national boundaries which, in the next place, challenges perceptions
about risk and responsibility issues which are country specific. There is empirical evidence
that the degree of trust and what source(s) people have confidence in is different among
nations (Berg 2000), but findings do not explain why and how. Moreover, there is empirical
evidence that consumers’ need for labelling and other product information is considerably
different among the Nordic countries (TemaNord 2001). This may imply that a procedure for
documentation of quality which inspires confidence in one country will not necessarily be
confidence-inspiring in another country. Hence, it is of interest to explore consumer trust in
two countries, Norway and Germany.
Two aspects of trust influence the importance of information: trust can be regarded as both
reflexive and non-reflexive. Reflexive trust presupposes an experienced uncertainty and is
expressed as an active scepticism (Giddens 1991), in that a reflexive choice is made between
trust and distrust. Non-reflexive trust, on the other hand, implies that trust is taken for
granted. In case of non-reflexive trust, the option between distrust and trust is not an issue.
In this regard, trust is a part of a normalcy, where deviations from the habitual situation are
not regarded as a concern. Based on this theoretical distinction, a distinction will be made
between trust and confidence (Luhmann 1988), where trust presupposes an active stand from
the consumer. Confidence, on the other hand, refers to a situation where trust is characteristic
for everyday practice and routines and seldom needs to be verified. Information will be an
important element in the formation of trust in the former case (i.e. reflexive trust). If reflexive,
trust is understood as a decision-making problem for which relevant information may be a
decision-support factor. On the other hand, information is not crucial in cases where confidence
exists. Too much information could create suspicion in the same way as if there were reasons
for uncertainty (Lagerspetz 1998). This means that information conveyed by a traceability
system could be very important in cases were trust is reflexive, as this reliable information
is fundamental for trust to prevail. However, extensive information about the history of the
product could be redundant or, in some extreme cases, such knowledge could produce the reverse
of the desired effect. Therefore, it is crucial to identify various types of trust and confidence.
Such knowledge will highlight the significance of information in general and traceability in
particular. Moreover, there is a need to explore whether the source of information affects trust
formation. Since knowledge and information about the food item in the distribution chain is
provided by a compound of organisations and units, it is fruitful to question whether the source
of information affects trust formation.
Based on this background the following research questions are raised: What is the importance of
traceability for establishing consumer trust in seafood? How much and what type of information
is needed to establish consumer trust? Will consumer trust be influenced by the source of
information? That is, traceability as a possible confidence-inspiring flow of information, type of
confidence-inspiring information as well as the importance of information provider are central
to this research.
In order to study the importance of information for trust two contexts were selected: Norway
and Germany. By contrasting two nations it is possible to explore more specifically what kind
of trust is identified among consumers, and how information interacts with trust and distrust.
binnenwerk.indd 214 21-08-2006 12:34:43

Too much or too little information?
There are differences between the two countries when it comes to consumption, distribution,
trust and the organisation of food authorities. This is illustrated in Table 1.
The major difference in consumption patterns of fish in Norway and Germany is explained
by the fact that Norway is major fishing nation while Germany is not. Norwegian consumers
consume more fish than most other consumers in Europe, due to among others the proximity
to fishing areas. The fact that older people tend to eat more fish than the young ones – and
that people in coastal areas have a higher fish consumption than inland consumers – seems
to be the case both in Germany and Norway (DGE 2004; FIZ 2004; Myrland and others 2000;
Olsen and Kristoffersen 1999). The share of fresh fish consumption in Norway exceeds Germany
by almost five times, as only 10 percent of the fish consumed in Germany is fresh. Most fish
in Germany is imported, i.e. more than 80 per cent, and Germany is an important market for
Norwegian seafood. On the other hand, only a small part of fish consumed by Norwegians is
imported. In Germany, fish is sourced as any other global commodity, although most of the
imported fish originate from fishing areas in northern Europe (FIZ 2004). Most fish purchased by
Norwegian consumers has a national origin. The distribution pattern in grocery retailing is quite
similar in Norway and Germany, except that fresh fish in Norwegian supermarkets is supplied by
independent wholesalers at a local level. Surveys among European consumers indicate that the
level of trust in food varies across national borders. Norwegian consumers have high trust in
institutional actors, such as food authorities and media. German consumers, on the other hand,
express scepticism towards institutional actors. According to a recent survey, Norway stands
out as a high-trust country while German regions prove to be middle or lower- trust area (Trust
in food 2004). In contrast with other countries German consumers and media seem to express
more scepticism to food (Berg 2000). Yet, according to the previous theoretical discussion on
types of trust, there is little knowledge on what kind of trust and distrust is prevalent in Norway
and Germany respectively. Furthermore, it is not known to what extent information counts in
trust relationships, nor it is known if various sources of information affect trust. In general, an
analysis of seafood as a case in Germany and Norway may generate variations and contrasts in
Table 1. Consumption patterns, seafood import, distribution and trust in Norway and Germany.
Norway Germany
Consumption pr capita (purchased fish) 23 kg 14.4 kg

Consumption share represented by fresh 59% 10%
fish
Share of seafood imported Approx. 10% 82%
Distribution Retail distribution of frozen Retail distribution of both
fish; fresh fish distributed by fresh fish and frozen fish
independent wholesalers
Consumer trust* High trust Sceptical
Reorganisation of food authorities Non-controversial Controversial
*Consumer trust is based on a truth-telling index for consumers in Denmark, Germany, Norway, Italy, UK
and Portugal (Poppe and Kjærnes 2003).
(Sources: DGE 20004, FIZ 2004, Lien 2005)
binnenwerk.indd 215 21-08-2006 12:34:44

the understanding of the concept of consumer trust as well as in the significance of design of
information (message framing).
Social phenomena regarded as non-reflected and taken for granted – such as confidence – are
suitable to be explored with qualitative research techniques. Qualitative techniques allow
a flexible indirect research strategy adapted to examine understandings and meanings at a
deeper level. In this case, focus groups and qualitative interviews were used to get behind the
statements given by people. In order to understand the everyday experience of the consumer a
phenomenological approach for focus-group interviews was followed (Calder 1977).
Empirical data were collected in Norway and Germany by means of focus-group interviews
among consumers and key-informant interviews among suppliers (manufacturers, distributors
and retailers). A comparison between consumers and suppliers gave the opportunity to examine
both seller’s and buyer’s views about product information. Especially it was of interest to find
out if major consumer concerns were on the agenda for distributors and suppliers of fish. In
other words, the focus was put on the “downstream” information from the consumer to the food
chain, i.e. the flow of information that has to be handled for an actor to be market oriented
and responsive to the end-user. Although the questionnaire guides for focus-group and key-
informant interviews were different, both focused on identical main issues.
In the next place, interviews were interpreted in terms of theoretical distinctions like
“confidence” and “trust”. Are understandings of crucial issues symmetric between consumers
and suppliers, or does it exist discrepancies and, thus, contradictions? Symmetry between
suppliers and consumers expectations indicates a state of consumer confidence. Discrepancy,
on the other hand, would point to controversies. Whether contradictions imply scepticisms and
distrust among consumers remains to be examined.
In order to attain various experiences from fish consumption, the participants were recruited
according to four strata: consumption frequency, age, gender and number of children. Eight
focus-group interviews were carried out, four groups in each country.
The same recruitment criteria applied to all groups: fish-consumption experience, age, gender
and children. Different degree of consumption experience should be represented, females should
constitute two thirds, participants should be from 20 to 65 years, and minimum half of them
should have children living at home. The reason for this specification was as follows: seafood
consumption increases with increasing age; most women are responsible for doing shopping to
the household and, besides, men tend to dominate discussions (undesirable behaviour. Persons
having responsibility for children are assumed to be more conscious about what food to purchase
and the content thereof. In Norway two groups were arranged in a town in the north (Tromsö)
and two were arranged in the capital (Oslo), i.e. locations representing coast and relatively high
consumption versus inland and relatively low consumption respectively. In Germany all groups
were arranged in the city of Karlsruhe (south-west Germany) due to practical reasons (i.e. site
of our collaboration institute). For the two first German groups consumption frequency was used
as a recruitment criterion in order to secure enough experienced participants, whereas for the
rest of the groups the frequency of seafood consumption was not a criterion for participation. In
both countries the principle of self-recruitment (“snowball method”) was used. That is, a person
familiar to one of the researchers was contacted in order to recruit one male or female (strange
to the researcher) in his/her circle of acquaintances. Thereafter, this participant was asked to
binnenwerk.indd 216 21-08-2006 12:34:44

recruit one male or female (strange to the researcher) in his/her circle of acquaintances, and
so on until the desired composition of the group was achieved. In other words, participants
were recruited from different circles of acquaintances resulting group members unknown to each
other and unknown to the researcher (focus-group moderator). The number of participants per
group was six or seven. However, one group consisted of nine participants.
Key-informant interviews were carried out at three levels in the distribution chain: production,
wholesale and retail. The Norwegian part included eight interviews: production (4); wholesale
(1); retail (3). The German part included nine interviews: production (4); wholesale (2);
retail (3).
The questionnaire was based on five conversation topics: safety, quality, food control,
traceability, and information. How these issues were treated on a conversation level is the
starting point for further analysis. In this way, everyday experiences are linked to purchase and
consumption of fish. Correspondingly, representatives from the distribution chain were asked
to present their understanding of the research issues. In the first section the data from the
Norwegian case will be presented, in the second a comparison will be made between these and
the data from the German case.
The Norwegian case

Reflections by consumers
In what manner did the focus-group participants relate themselves to the issues of interest to
us? It was found that participants related themselves to the issues in terms of four classifications
of fish. In many cases these classifications can be related to one or more of our conversation
topics. As mentioned above, Norway is among the countries with the highest consumption of
fish per capita in Europe. It should therefore be expected a high degree of familiarity with fish
as food. Yet, there was a large portion of younger, urban participants that made connotations
of uncertainty when buying and preparing fish. As one woman in her thirties expressed, “there
is always something spooky about fish”. Although “spooky” is a dramatic term, she did not use
“spooky” in terms of fish being dangerous or harmful, she rather wanted to express her lack of
knowledge and inexperience. Something similar was uttered by another woman, “fish is fish, and
meat is meat”. She explained that fish was not the same as meat and other food ingredients,
because she with meat always knew what she could expect. “The fish in the counter is totally
anonymous, it does not signal anything”, said another woman. These statements remind us
that fish is a kind of “credence good” (Ford and others 1988). A credence good signifies that
trust will always be an issue because you will not know the quality, even not after purchase. It
can therefore be ascertained that trust is an issue. Still it is it is unclear what the perception
of uncertainty is related to and neither anything is known about its extension. Therefore a
search was done for distinctions that could signify the content of uncertainty more clearly. It
was found that the conversations turned on four main classifications.
One of the most basic categorisations was made between fish and shellfish. For some people
in the focus group shellfish such as clams and oysters was simply “horrible”, thus marking a
distinction between the edible and inedible. This finding is consistent with traditionalist norms
of eating behaviour and was particularly revealed among the elder participants. Others avoided
shellfish due to health risk. “Shells are trifles”, said a woman in her early thirties, “and I always
binnenwerk.indd 217 21-08-2006 12:34:44

try to avoid trifles”. She explained that trifles reside at the sea bottom, which according to her
was “dirty”. Like the other shellfish eaters in our focus groups she could refer to “unpleasant
experiences” and was concerned about origin and preparation.
Fish, on the other side, had no connotations of being inedible or involving health risk of any
kind. None of our participants had ever become sick. Nor could they remember that family
members or friends had any concerns. This indicated that fish and food safety was not an issue at
a conversational level. As in most western countries, European food scandals had raised a debate
in the Norwegian public media about food safety and the role of the food control. However,
very few reflect on this during their daily shopping routines, “It is simply something that never
strikes me when I do my shopping” a housewife said. The idea of a health risk was absent. The
non-reflected nature of food safety indicated a strong sense of confidence in fish.
This does not mean that a notion of health risk to fish during shopping was totally absent.
Instead the awareness of risk appeared in other contexts. The focus-group participants
distinguished unprocessed and processed fish. Many associated processed fish with minced fish.
Norwegians eating habits include several variants of minced fish, such as fish balls, puddings
etc. Some were suspicious of what had been “put into the fish mince”, others were sceptical
towards the vacuum packaging. “It’s without taste”, explained a young man, “and sometimes
it is sour”. The classification of processed and unprocessed contains distinctions of both health
risk and taste, and this raise the question of trust to highly processed fish.
A more complicated distinction was made between fresh and frozen fish. Although there is
in general switch from frozen food to fresh food in consumption, the consumption of frozen
filets has increased among Norwegian consumers in recent years (Lien 2005). The focus-group
participants did not consider frozen fish as inferior to fresh fish. On the contrary, there was a
strong expressed idea that frozen fish is good: “Fish is most fresh just before it is frozen” said
a middle-aged woman. Frozen filet represented something safe and predictable, i.e. you always
knew what you got. This was not the case with fresh fish, where a lot of differing statements
appeared. Also the notion of freshness was diffuse. When a young woman in her thirties was
asked what freshness meant, she answered bluntly “I have no idea, that is something I must
ask about”. Information from the persons behind the counter therefore became a key issue. In
many cases, asking simply left one none the wiser, said a young man; “sometimes I feel that I
could have moved behind the counter and said exactly the same my self”. Instead of exposing
oneself to the uncertainty of buying fresh fish, they stuck to the familiar, “when we need some
fish, we take it frozen” confirmed a woman. Of course, there were several participants who
regarded fresh fish as superior to frozen fish. Due to lack of knowledge about fresh fish, other
signs of quality were used. Lack of skills was compensated by proxy indicators, such as cleaning
and hygiene in the shop. The different understanding of fresh fish and frozen fish lead us to
believe in a strong division of trust.
There were other distinctions of quality. The focus-group participants had ideas about the
significance of origin, i.e. they distinguished domestic and foreign origin. However, this
distinction was not at all consistent. People seemed to prefer fish from domestic waters,
because it came from “clear and cold waters” in the north. Contrary to what have been found
about consumer trust in the agricultural sector (Poppe and Kjærnes 2003), the focus-group
participants were not particularly suspicious about imported fish. For some knowledgeable
participants in our group, origin was important in order to attain quality predictability. An
binnenwerk.indd 218 21-08-2006 12:34:44

elderly woman said she wished to know where fish proven to be excellent comes from, but
“normally you do not know that”. Information about fishing ground or the vessel could increase
reliability of own choices.
Another distinction was made between wild and farmed fish. The participants were relatively
indifferent to the negative media attention about the issue of toxic residuals in fodder for
farmed fish. Most food safety allegations were ignored: “Wild salmon tastes better than farmed”
was repeated several times in the groups. Previous studies have concluded that this distinction
of taste is related to ideas of wildness and natural habitat as superior to domestication. This
study indicates that the distinction is more concrete. During the conversation a women in her
fifties repeated “I’m constantly returning to this fat content in salmon. Why sell it like that?
The wild salmon isn’t like that, is it?” Farmed fish evoked scepticism – not due to health risk
or doubt in industrialised production of food but because of fat content and taste.
As a conclusion, the focus-group interviews in Norway suggest that health risk is not questioned
during everyday consumption of fish. From this it is assumed that there is a confidence in
food safety. People have a strong belief in the role of the food control. They believed that
Norwegian citizens are subject to so extensive public control that the same must be the case
for supermarkets. Neither was the sensory quality of frozen fish ever questioned. Frozen fish
represented predictability of taste – you always knew what you got. This sense of confidence
seemed to stem from a strong belief in freezing technology and domestic logistics in preserving
taste. The non-reflexive character of trust meant that information about traceability was not
a critical issue. The quality of fresh fish was however doubted. A basic scepticism seemed
to prevail. This does not necessarily mean a general distrust in fresh fish. However, trust in
quality seemed to depend on own experience and precautions, such as specific knowledge and
a selectivity on where to buy fish. In these cases, information about quality distinctions – such
as freshness – was regarded as critical.
Reflections by food chain

The purpose with the interviews was to elicit informants’ experiences and opinions as regards
their own enterprise’s position in the flow of product information (from source of origin via
processing to the market). Broadly speaking, the essence of interviews turned on food safety
and quality, perceived consumer uncertainty, and market- and food-chain communication.
Although all informants gave voice to the view that “quality” − implicitly sensory quality (taste,
texture, appearance and freshness) − is what the consumers want, consumers’ ability to judge
quality on fish and seafood were questioned. Moreover, informants had experienced weaknesses
and failures at different levels of the food chain resulting in unsatisfactory maintenance of
product quality. In what follows we will elaborate on this coincidence.
Despite quality claim an overriding attention in the food chain, the majority of informants
had experienced poor quality, e.g. due to seasonal variations affecting muscle quality, reckless
handling of the catch, or subsequent routine failure in the handling of goods. Particularly, a
recurrent problem at the retail level seems to be temperature variation in refrigerated display
counters. At the same time, however, informants acknowledged the retail level’s responsibility
to shoppers as regards the shop’s duty to maintain product quality. In particular this was the
case for fresh fish, a product all informants considered to be without known identity.
binnenwerk.indd 219 21-08-2006 12:34:44

Doubt about consumers’ ability to judge quality manifested itself as regards taste including
differentiation of taste. Consumers were considered to be different as regards their ability to
taste if fish is okay and, in particular, fresh. Moreover, they were regarded incompetent to
differentiate types of fish: “Sometimes I wonder if they [viz. the consumers] don’t know how
fish shall taste. I think consumers accept pretty much; they don’t know that fish can taste
different” was remarked by a representative for a producer of high value (modified atmosphere)
fish. Another aspect of consumer ignorance may be lack of complaints in cases where complain
would have been the expected response to dissatisfaction. Hence, absence of complaints −
which was the general impression passed on by the informants − should not necessarily be
interpreted as an indication of consumer satisfaction; they do not complain if they have no
expectations. To make a comparison, a representative for a shop claimed that they have more
complaints on meat. This is plausible if people are more certain about what to expect from
meat than from fish. For us it is reasonable to speculate if unsatisfactory maintenance of
quality maybe is related to customers’ non-ability to discover or reveal irregularities. Under such
circumstances, it is a risk that dishonest players may make the most of consumers’ ignorance
(i.e. information asymmetry).
While quality was perceived to be a consumer concern and the conditions of competition were
said to be improved by emphasising quality, food safety was not considered to be a consumer
concern but assumed to be an obviousness for consumers: “They almost take ‘hygiene quality’
for granted, consequently ‘safety quality’ is expected to be protected” as a representative for a
multiple stores put it. Extensive customer requirements in addition to directions from the food
safety authority implied that enterprises allocated considerable resources in order to attend to
food safety by means of routines, systems and standards.
The existence of a food safety authority implies that informants assumed consumer trust in
fish and seafood to be high in Norway. Based on the views expressed by producers high trust
seems to be the case regardless of the product’s country of origin. A representative for a
producer put it this way: “I think nobody cares if the origin of the product is Chilean, Chinese
or Norwegian. For instance, I don’t think people trust more in two days remaining shelf life if
the product is Norwegian compared with Bangladeshi. I think one trust in the public authorities
and the reliable major manufacturers.” However, unconditional trust is not just according to
a representative for a multiple stores: “I think consumers would have been more sceptical if
they knew how seldom inspections are done.” In our opinion this statement illustrates an
exaggerated responsibility for the authorities. Besides, it is illustrative of what has to be
a balance between basic requirements that now and then is controlled by authorities and
enterprises’ own responsibility to, at all times, run a business in due diligence with regard to
possible offences relating to food safety standards.
In this study it is found that representatives from the food chain were ambivalent as regards
willingness to inform consumers. On the one hand, a flow of information − both upstream
and downstream − was considered necessary for the market communication to be effective,
i.e. “tune” the industry in on the consumers and allow the food safety authority to mediate
relevant information. In that respect the afore-mentioned lack of consumer complaints is a
behaviour which brings about inadequate market communication: “We wish for feedback; poor
products may result in more consumers avoiding repurchase” a representative for a producer
claimed. In response to a question on what information consumers actually seek to make
an informed choice a representative for a shop emphasised the need for signals that guides
binnenwerk.indd 220 21-08-2006 12:34:44

behaviour, i.e. brands rather than origin would be useful and, as regards fresh fish, smell and
colour are important. Besides, local origin was deemed to be useful and preferred (provided
that the quality was not inferior) and the rationale would be a desire to secure vigorous local
communities. The impression of representatives for multiple stores was that specification of
fishing area was not desired by consumers and neither name of fishing vessel nor type of fishing
gear. Most consumers were not assumed to be able to draw conclusions from such information.
However, from producers’ point of view specification of origin would be desirable due to two
reasons. First, declaration of country of origin would prevent unintentional negative publicity,
e.g. if a food accident hits plants processing cod in China this should not affect cod processed
in Norway. Second, specification of origin is appropriate if the origin become important for the
quality of the product (taste, texture, state of nutrition etc.), otherwise not.
On the other hand, suppliers gave counter-arguments as regards communication of quality-

related information to consumers. Views on possible useful information made available via
traceability systems included origin but other information as well. However, the opinion was
clear: much of the information from traceability is needless for the consumer. A representative
for a producer claimed that it is sufficient to simply inform that a system for traceability is in
place (exists) and, by so doing, gain credibility. Especially views on date of catch and production
as well as use-by date were passed remarks on. The contrast between layman belief and expert
knowledge were expected to result in controversy. This is illustrated by a representative for a
producer who said: “I think many have a romantic idea that the fish is caught the day before. I
don’t think it is smart to print day of catch. I believe the fish is caught several more days ago
than the consumer thinks.”
This observation reflected the main contradiction between the understanding of trust and
information between consumers and suppliers. Consumers’ information seeking versus the food
chain’s communication brings to light a dilemma. Given the fact that fish quality − especially
fresh fish quality − is variable and players in the food chain assume consumers are ignorant as
regards judgment of fish quality, does the industry want to retain information (that consumers
evidently demand) for fear that release of information of interest will result in unintended
consequences? For instance, fear that fish consumption will be reduced when consumers
realise that fish sold as fresh is not caught the day before. Focus-group participants urged for
more information on freshness and origin in order to overcome the uncertainty of taste that
characterised fresh fish. This attitude to consumer demands evoked associations of paternalism.
In our opinion such attitude could lead to unintended and unwanted consequences: consumers
will not buy fresh fish. If that would happen, market communication surely has failed.
The German case

Reflections by consumers
German consumers are different from Norwegian both when it comes to the per capita consumption
of fish and consumption of fresh fish. Furthermore, most fish consumed in Germany is imported,
implying that sourcing, supply and consumption of fish takes place under different circumstances
than in Norway. Although we find some similarities in which issues like quality, safety and the
meaning of information are considered in our focus groups, there are essential differences. It was
found that trust is grounded on other criteria and this implies that the meaning of information
must be regarded to be different than in the Norwegian context. In the following, the main
topics as differences between the Norwegian and German focus groups were explored. That is,
due to practical reasons (space-saving) it was decided to present the German results relative to
binnenwerk.indd 221 21-08-2006 12:34:45

the Norwegian findings (rather than, more “correctly”, present the German findings first and then
discuss the major differences and similarities between the two countries).
There was more apprehension of health risk in the German focus groups, although it was not
strongly concretized. Even though Germans had an awareness of the origin of fish, this was
not related to a domestic vs. foreign dimension. One obvious explanation is that most fish
consumed in Germany is imported, thus making the distinction irrelevant. Still, the notion of
local origin makes sense. The German focus-group interviews took place in the hinterland of
southern Germany, far from the coast, where there is access to local freshwater fish, such as
carp and trout. However, none of the participants seemed to regard local origin as superior.
Although there is a “romantic” idea of buying fish directly from the producer – near the coast
from a fisherman or in the south from an aquaculture producer (carp and trout) – local origin is
not a decisive aspect of buying fish. “From my feelings”, said a man in the thirties, “it can be
that I regard salt-water fish as more fresh than freshwater fish”. People may prefer salt-water
fish, probably because the sea is not regarded as polluted as lakes and rivers. The distinction
freshwater and salt-water fish seemed to generate associations of both food safety and sensory
quality, at least in the eyes of fresh fish consumers.
There were other distinctions as well. People had ideas that linked taste to geographical origin.
“Origin is for me a question of taste, I think” said a 42 year old man, but otherwise he was
unable to explain the difference in taste between fish from the Atlantic and the Mediterranean.
What seemed to be the case was that this distinction had a more emotional foundation. A
mother in the forties explained that there is “information of origin that can be sympathetic and
unsympathetic. I don’t have very much knowledge, but fish from the Mediterranean makes me
question whether it is proper or not”. The distinction sympathetic and unsympathetic seemed to
cover a number of concerns that many participants related to origin. The feeling of sympathy
seemed to capture a number of issues, such as whether the sea was regarded as “proper”,
transport distance, sustainability of fishing and content of foreign ingredients in fish. Thus,
the distinction seemed to entail a more complex mixture of ethical concerns and health risk.
The sympathy dimension, on the other hand, was represented by fish from imagined “virgin”
oceans, such as the Pacific or waters surrounding Greenland, being a symbol of the healthy fish
in clean nature. The Victorian perch, imported fresh and airborne from Tanzania, and is quite
common in supermarket shelves, seemed to evoke opposite associations. Participants referred
to critical media coverage on both this African perch and the Asian freshwater catfish – raising
questions about third-world ethical issues. Neither was the third world considered to be reliable
as regards pesticide spill in fishing waters. In spite of expressed doubts, however, the declared
sceptics did not avoid buying the relevant fish. This indicates a trust among consumers, but
that trust is questioned and considered as conditional. As a mother in the late thirties stated
“I don’t want to know too much, [...] because each food could have a bad aspect.”
The German participants did distinguish wild and farmed fish, although this distinction is
more complicated than the Norwegian one. Due to extensive media coverage during recent
years, feeding stuff and farmed fish was regarded a source of health risk. People were afraid
of medication of salmon, and they were suspicious about fish-farming practices. Norwegian
salmon gave rise to concern, as “the Scandinavians normally blind off a fjord for fish farming”.
In contrast, farmed salmon from Scotland and Ireland was regarded more favourable due to
farming practices more in harmony with nature. Norwegian fish farming evoked notions of
industrialised farming (fed with “pigments and I do not know what”) and intensive farming
binnenwerk.indd 222 21-08-2006 12:34:45

(equal to the industrialised poultry production). Wild fish, having stronger connotations to
pureness and nature, was not unproblematic either. People questioned the danger of over-fishing
(exploitation) in the North Sea and, thus, gave voice to environmental concerns. They were
also worried about fish-catching practices in the Atlantic which involves disposing of by-catch.
Once again, these objections made by the participants did not prevent them from buying fish.
Instead, there is an open reflexivity on the conditions of trust, making consumers vulnerable to
negative media publicity. This indicates that traceability information on environmental issues
and health risk could increase trust.
In conformity with the Norwegian focus groups, the Germans made a distinction between
fresh and frozen fish. Frozen fish represents something safe and predictable. Similar to the
Norwegians, a woman in her forties maintained that “I normally prefer frozen fish, as it is deep-
frozen shortly after it is caught”. Frozen fish did not provoke any reflections on risk, “when the
cold chain is not broken, everything is okay” said a 45 year old woman − “when I buy my frozen
filets from Lidl I have trust”. Another person referred to “standards” that she always expected
the retailers to comply with.
Buying fresh fish, on the other hand, seemed to require a competence. Persons eating fresh
fish emphasised the significance of selecting the proper retailer. “When I do my shopping at a
specialist shop I always have a feeling that the fish is fresh”, said an elderly male participant
experienced with self-caught fish. Like the other fresh-fish consumers, he had selected the outlet
based on informal information, word of mouth and reputation. Another man said that he mostly
bought his fresh fish at the weekly market, “the fish is fresh, and in spite of being a little more
expensive, it always tastes better”. More than in Norway, the consumers who preferred fresh fish
attached importance to taste. Not only did they emphasise sensory quality before price, some of
them travelled out of town in order to get the superior quality. To choose fresh fish was a matter
of connoisseurship – it demonstrated both time and competence to prepare.
These differences are important in terms of trust and information. Contrary to the frozen fish, the
trust in fresh fish was particularistic in the sense that it depended on the trust in one specific
shop and the personal advice from the shop’s staff – not trust in shops in general. For frozen
fish, the trust was generalized and linked to a belief in the superiority of the cold chain. So far,
there are conformities with the Norwegian case. Different from Norway, however, the role of the
food authorities was less present in their trust considerations. The Germans did not have a belief
in an extensive food control as a guarantor of trust. Some of them had diffuse ideas that “some
people up there” controlled the fish, without specifying whether “some” were private or public.
Others maintained that the food authorities carried out controls in the food chain in order to
“protect the consumers”. Yet, the role of the food authorities was regarded as more limited; the
participants assumed that the food authorities were “overloaded with work” and not capable of
ensuring a complete food safety. These findings are consistent with results from other studies
(Poppe and Kjærnes 2003), arguing that the restructuring of food policy in Germany has been a
controversial policy issue, challenging the legitimacy of the regulatory framework. “I regard the
retailer as the prime responsible for trust”, said a young woman. Consumers’ trust rested on a
systemic level, guaranteed by the standardized logistics of the retailer.
To summarize, the trust in fish among German focus-group participants appeared more
conditional than among the Norwegian participants. Trust was to a lesser degree taken for
granted. The Germans raised more questions and were more doubtful about food safety and
binnenwerk.indd 223 21-08-2006 12:34:45

ethical issues. This attitude does not necessarily imply distrust. The presentation of frozen fish
in supermarkets entailed an immediate trust similar to what we defined as confidence. Still,
the consumers were more sensitive to negative media publicity on issues like over-fishing and
medication of farmed fish. In our opinion, this signifies that communication of traceability
information to consumers should be seen as more important in Germany than in Norway. In
the communication process, the retailers were more important than the food authorities. The
retailers were the reliable key actor “in charge of” the cold chain.
Reflections by food chain

The German key-informant interviews mainly covered the same topics as the corresponding
Norwegian part and the views were basically identical with regard to food safety not considered
to be a consumer issue and limited consumer relevance of traceability information. However,
German and Norwegian views differed somewhat with regard to reflections about aspects of
quality and communication of information including origin. Below, similarities and differences
will be referred to.
As previously mentioned, German suppliers - contrary to the Norwegian suppliers – depend

heavily on global sourcing of seafood and more than 80 percent of the fish is imported.
In general, the distribution of fish in Germany is more complex than in Norway because it
involves longer supply lines. This structural difference implies that access to information pose a
challenge to German distributors. “When you import fresh prawns from south-east Asia you can
guarantee that from Frankfort the cold chain is perfect whereas no one knows what happened
before” said an importer of frozen prawns. In other words, a lack of knowledge prevailed as
regards product history (handling, temperature etc.) previous to arrival of the consignment
in Frankfort. Like this, complexity lead to increased uncertainty and confined the importer’s
possibility to communicate reliable information. A consequence is that information about how
many days the fish is when sold to consumers is inaccurate, or even incorrect. This seems to
be due to two reasons: lack of knowledge of the history of the batches caused by imperfect
traceability, and unwillingness to communicate exactly how many days since catch in order
to avoid rejection (non-purchase). The latter, which we will emphasise below, is a result of a
strategy to communicate freshness.
Our empirical material disclosed two different strategies to communicate freshness. One strategy
was to communicate information about days since catch as a “guarantee” that the fish in the
counter would be maximum 48 hours − a lapse of time which was known by experience to be
acceptable to consumers. Professionals knew that the sensory acceptability of fresh fish could
be much longer than 48 hours (provided proper temperature etc.) but they anticipated that
longer indication of time would definitely not be met by consumer acceptance. In other words,
this is a communication strategy where expert knowledge submits to layman’s understanding.
Another strategy was to communicate remaining shelf life (rather than days since catch). That
is, recommended time till preparation/consumption was passed on to the consumer and this
information was supported with the expert’s subjective judgment of quality indicators (eyes, gills
etc.). This was a communication strategy in use for personal service and an effort to establish
trust by letting expert knowledge dominate layman’s idea about the limit of consumption for
fresh fish. It is reasonable to consider the latter strategy as an effort which attends to the fact
that the average German’s competence to judge the quality of fish is bad and, accordingly, the
need to educate consumers is considerable. Based on that particular aspect of trust which refers
to layman belief vs. expert knowledge we see that consumer education as an “active” strategy
binnenwerk.indd 224 21-08-2006 12:34:45

is a different solution than the afore-mentioned information retention as a “passive” strategy,

i.e. the German and Norwegian case respectively. The former leads to trust in seafood, the latter
is vulnerable in case of a bad consumer experience.
Correspondingly to the dominating view among the Norwegian key informants food safety
was not deemed to be a concern to German consumers. Like in Norway, this was primarily
due to the fact that the enterprises were subject to self inspection and customers’ extensive
standards. Besides, brand owners claimed that to secure a well-known brand was seen as an
adequate incentive to operate in accordance with high standards on food safety and quality
standards which they claimed exceeded the basic requirements of the food safety authority.
Brand protection, demanding customer requirements, and some enterprises’ ideas that own
competence exceeded the competence of the food-safety-authority staff gave the impression
that the consumer-no-concern situation was relatively more attributed to (/thanks to) the
enterprises’ own efforts than the controls by the food safety authority.
Most informants acknowledged that traceability was a legal requirement primarily in use
to demonstrate an enterprise’s product liability, e.g. in case of recall. So, extensive and
detailed information made technically available via a traceability system was not meant to be
communicated to consumers.
Communication by the food chain vs. information search by consumers brought about some
interesting aspects. The discussion of possible use of information traced through the food chain
included views on the usefulness of knowing the origin of the product. According to a fish broker
origin could be a means to differentiate fish by its freshness, implicitly less fresh fish originates
from distant waters. Besides, origin indicated by ocean area would make it possible to specify
endangered stocks. Moreover, indication of geographical origin would be useful to select preferred
qualities. As regards farmed salmon, knowledge about country of origin made a wholesaler able
to avoid salmon from suppliers known to deliver fish with high fat content and fish farmed
under conditions were innate behaviour of the species is considerably suppressed. According to
such considerations this wholesaler used country-of-origin information to discriminate between
salmon farmed in Norway − regarded as more fat − and salmon farmed in Scotland and Ireland:
“a Norwegian farmed salmon is, in any case, considerably more fatty […] and, for certain, a
Norwegian farmed salmon is not so high-grade as a Scottish salmon which has lived in a sort
of half freedom” The aforementioned represented possible usefulness of origin information for
actors in the food chain. Consumers’ interest seemed to be different.
Consumers’ interest in product information in general was to know from where a product
originates. According to a wholesaler supplying restaurants, consumers’ interest was more
specific in a restaurant context. Restaurant guests liked to relate origin to a particular region
which gave them associations (e.g. high quality food from the French region Brittany associated
with holiday in the area). Besides, a retailer’s experience was that consumers who bought
salmon used knowledge about country of origin to make probable information about the fish
being wild or farmed. Contrary to this experience, a producer of high value (ready to eat)
products argued that consumers have got used to a particular quality and colour on (farmed)
salmon and this quality and colour cannot be reproduced by wild salmon (implicitly a fish with
natural variation in quality and colour would neither be recognised nor preferred). This view
was corroborated by the opinion of a fish broker. Hence, wild vs. farmed was not really an issue
to consumers. Accordingly, only a slight share of consumers was believed to take an interest in
binnenwerk.indd 225 21-08-2006 12:34:45

origin, like expressed by an informant in a fish retail and restaurant chain: “However, among
the consumers who daily are behind our door, i.e. weekly about 1.5 million people, I guess not
more than 10 would ask from where the fish originates”.
To summarize, compared with the Norwegian case the German key informants were more
reflected regarding range of application for use of information on origin. This may be due
to many reasons of which the above-mentioned examples have called attention to Germans’
concern for environmental and ethical issues, and their knowledge about qualities related
to various product attributes − all indicated by our focus groups to be more top of mind for
Germans than for Norwegians.
Conclusion
In this explorative study on consumer views on food safety and quality on fish in Norway and
Germany, the meaning of information on trust was analyzed. By making a distinction between
confidence and trust, it is argued that information is not a precondition for confidence to
prevail. Trust, on the other hand, is reflexive by nature and contingent on information in
order to be sustained. As regards the research questions, the findings permit to make some
comparative conclusions related to the importance of information, types of trust-inspiring
information and the meaning of information source in Norway and Germany. In Norway, neither
health risk nor ethical issues were questioned. This signifies a high degree of confidence that
food is safe. More information on food safety issues was regarded as redundant. However, the
sensory quality of fresh fish was disputed. While the consumers demanded more information
on freshness, fat content in salmon etc, the suppliers and supermarkets retained relevant
information and this caused a “vacuum of consumer trust”. This may have caused that consumers
nowadays seem to prefer frozen fish before fresh (Lien 2005). Consumers had an undisputed
trust in the Norwegian Food Safety Authority and regarded it as the guarantor of food safety.
Accordingly, consumers called for a single party that could guarantee satisfactory sensory
quality. At present, no immediate candidate seemed capable of entering this position. Focus-
group participants themselves questioned whether they, at the point of purchase, were able
to process and make use of complex traceability information (as conveyed by labelling or by
other means). The Norwegian Food Safety Authority, regarded by consumers as a legitimate
representative of consumer interest, could have an important role to play as a trust enhancing
actor. Having set regulation covering sensory qualities aside in recent years, it remains to be
seen whether the Norwegian Food Safety Authority regards sensory quality issues as a relevant
regulatory activity.
Although the German focus-group participants seemed to have trust in food safety, they
questioned issues like health risk and ethical concern to a higher degree than the Norwegians.
This indicates that confidence is not taken for granted. The focus-group participants were
sensitive to negative media publicity on these issues and they expressed reflexivity on a number
of quality and ethical issues. Documentation – especially related to aquaculture practices
– was understood as a consumer issue. Although they did not call for an extension of existing
product-labelling schemes, they wished relevant information to be available to the public for
further validation. Therefore transparency throughout the distribution chain was considered
as important. In general, industrial standards and cool chain management were regarded as
a guarantee of food safety. Although the regulatory activities of the German food safety
authorities were acknowledged, public authority were not expected to take a full responsibility
binnenwerk.indd 226 21-08-2006 12:34:46

of consumer issues in general, implying that information traced through the food chain and
made available by private actors could be valued. To a larger extent than in the Norwegian case,
manufacturers and retailers were found to be important providers of information. To a lesser
degree, private actors in Germany can rely on trust-inspiring efforts by the public authorities.
This indicates that information from manufactures and retailers is crucial in order to sustain
trust.
The findings in this study indicate that national contexts should be regarded as important in
order to understand the relationship between information and consumer trust. Furthermore, it is
questioned to what extent it is feasible to launch a Pan-European message in order to provide
trust-inspiring information. Consumers define their own role, the responsibility of public control
and the tasks of the private actors differently. In order to develop categorisations that involve
more countries, further research should aim to clarify whether the specific national findings are
transferable to other national contexts.
Acknowledgements
The authors thank Oddrun Bjørklund at Fiskeriforskning for the help we received collecting the
data of the Norwegian part of this research project. Funding for this research was received from
the Research Council of Norway.
References
Anderson JG, Anderson JL. 1991. Seafood Quality: Issues for Consumer Researchers. J Consumer Affairs 25(1):144-
163.
Berg L. 2000. Tillit til mat i kugalskapens tid. En komparativ kartlegging med fokus på forbrukertillit og matsikkerhet
i Norge, England og Belgia. [Trust in Food in the Age of Mad Cow’s Disease] SIFO report no. 5. Lysaker, Norway:
The National Institute for Consumer Research. 146 p.
Calder BJ. 1977. Focus Groups and the Nature of Qualitative Marketing Research. J Marketing Research. XIV:353-364.
DGE (Deutschen Gesellschaft für Ernährung e.V.). 2004. Ernährungsbericht 2004. Bonn.
FIZ (Fisch-Informationszentrum) (ed.). 2004. Fischwirtschaft - Daten und Fakten 2004. Hamburg. (www.fischinfo.
de).
Ford GT, Smith DB, Swasy JL. 1988. An Empirical Test of the Search, Experience and Credence Attributes Framework.
Advances in Consumer Research 15:239-243.
Giddens A. 1991. Modernity and Self-Identity. Self and Society in the Late Modern Age. Cambridge: Polity Press.
Kjærnes U, Dulsrud A. 1998. Consumption and Mechanisms of Trust. Paper presented at the ESA subgroup The Sociology
of Consumption, University of Milan, Italy, 16-17 September 1998.
Lagerspetz O. 1998. Trust. The Tacit Demand. Dordrecht: Kluwer Academic Publishers. 177 p.
Lien K. 2005. Markedsrapport norsk konsum av sjømat. Mai. Tromsø: Eksportutvalget for fisk.
Luhmann N. 1988. Familiarity, Confidence, Trust: Problems and Alternatives. In: Gambetta D, editor. Trust Making and
Breaking Cooperative Relations. Oxford: Basil Blackwell Ltd. p 94-107.
Myrland Ø, Trondsen T, Johnston, RS, Lund E. 2000. Determinants of seafood consumption in Norway: lifestyles,
revealed preferences, and barriers to consumption 11:169-188.
Olsen SO, Kristoffersen EM. 1999. Sjømat i norske husholdninger. Forbruk, anskaffelse og holdninger til fersk versus
frosset fisk. Report no. 19. Tromsø, Norway: Fiskeriforskning. 73 p.
Poppe C, Kjærnes U. 2003. Variations in Trust in Food in Europe. Oslo: SIFO Professional Report no. 5. 160 p.
TemaNord. 2001. Forbrukernes krav til fødevaremærkning og vareinformation. Report no. 501. Copenhagen, Denmark:
Nordic Council of Ministers. 120 p.
binnenwerk.indd 227 21-08-2006 12:34:46

binnenwerk.indd 228 21-08-2006 12:34:46
Consumer knowledge and interest in information about fish
Zuzanna Pieniak1, Wim Verbeke1, Karen Brunsø2 and Svein Ottar Olsen3
1Department of Agricultural Economics, Ghent University, Coupure links 653, 9000 Gent,
Belgium
2MAPP Centre, Aarhus School of Business, Haslegaardsvej 10, DK-8210 Aarhus, Denmark
3Fiskeriforskning, PO Box 6122, NO-9291 Tromsø, Norway
Abstract
This paper focuses on European consumer’s objective and subjective knowledge about fish.
Cross-sectional data were collected through the SEAFOODplus pan-European consumer survey
(n=4.786) with samples representative for age and region in Belgium, the Netherlands, Denmark,
Spain and Poland. Objective and subjective knowledge, as measured using multi-item constructs,
are poorly correlated and actual levels differ strongly between countries. Subjective knowledge
is found to be a better predictor of fish consumption frequency than objective knowledge,
particularly so among the populations with the highest subjective knowledge. With respect to
fish consumption decisions, what consumers believe to know about fish matters more than how
much they actually know.
Keywords: knowledge, consumer, fish, information
Introduction
Consumer product knowledge is an important factor in understanding consumer decision-

making because of its role in information search (Brucks 1985; Rao and Sieben 1992) and
information processing (Bettman and Park 1980; Rao and Monroe 1988). With respect to
consumer knowledge, two constructs are to be distinguished, namely objective knowledge
and subjective knowledge. Subjective or self-assessed knowledge measures relate to people’s
perceptions of what or how much they know about a product class and are based on consumer’s
interpretation of what s/he knows, while objective measures are the accurate information about
the product class stored in long-term memory (Selnes and Gronhaug 1986; Park and others
1994). Although findings about the impact of knowledge on information processing are often
contradictory, for example whether or not perceived (subjective) knowledge is a better predictor
of behaviour than actual (objective) knowledge (Radecki and Jaccard 1995) there is a consensus
that knowledge is a key construct in information processing. Furthermore, subjective knowledge
affects information processing activities differently than objective knowledge (Brucks 1985).
Consumers develop product knowledge through the search and use of information as well as
through experience (Howard and Sheth 1969). This search of information is external, thus
involves seeking information from the environment such as packaging information, brochures,
advertisements, newspapers, personal selling, friends and others.
As far as known, no study has been performed before with respect to consumer’s knowledge
and interest in information about fish. Fish is food product with a predominantly healthy image
among consumers. Nevertheless, fish consumption remains below recommended intake levels
in several European countries. One of the potential barriers to higher fish intake may relate
binnenwerk.indd 229 21-08-2006 12:34:46

to consumer knowledge about the product. Verbeke (2005) indicated that consumers must
have a sufficient level of knowledge, based on reliable information, in order for information
to have a favourable impact on consumer’s food choice. As such, a better understanding of
consumer’s knowledge about fish, and its relation with behavioural characteristics can provide
valuable insights in consumer decision making towards fish, and may yield recommendations
for information provision through future communications about fish.
The overall objective of this paper is to assess European consumer’s objective and subjective
knowledge about fish, their interest in and need for information related to fish and to formulate
recommendations for more effective communication. The aim of this paper is to investigate
whether providing more information aiming at increasing objective or subjective knowledge
would help to increase the consumption of fish.
Data for this study were collected within the EU FP6 Integrated Project SEAFOODplus. First, in
order to gain preliminary insight in the consumer’s information needs, exploratory primary data
were collected through qualitative focus group discussions in May 2004 in Spain and Belgium.
Second, a quantitative cross-sectional consumer survey was carried out in November-December
2004 in five European countries: Belgium, Denmark, the Netherlands, Poland and Spain. A
quota sampling procedure with age and region as quota control variables was used. A total
sample of 4.786 consumers (n=800-1.100 respondents per country) was obtained. Samples were
representative within each country for age and region. All respondents were responsible for food
purchasing within their household. The questionnaire measured a wide variety of constructs
including behaviour, attitude, beliefs, perceptions, subjective and objective knowledge and
cognition with respect to fish, and interest in information cues and traceability. The focus of
this paper is on consumer knowledge of fish, use of and trust in information sources, interest
in information cues and perceived benefits from fish traceability and labelling.
Five questions assessed consumer’s need for cognition (NFC). Need for cognition is the tendency
to derive “intrinsic enjoyment” from “engaging in effortful information processing” (Cacioppo
and others 1986). The items were adapted from the short form of the NFC scale (Cacioppo and
others 1984). They were selected from the 34-items NFC scale as the five items with the highest
factor loadings (Cacioppo and Petty 1982). In a previous study, Steward and others (2003)
have used only the three items with the highest factor loading from the NFC scale. Responses
on the three items were significantly correlated, had adequate internal consistency (α= 0.76),
and were combined to create one NFC score.
The internal consistency reliability of the five NFC items was tested using Cronbach’s alpha
which for these items was 0.80. Additionally, a second internal consistency measure, namely
the mean inter-item correlation was computed. The value of the mean inter-item correlation
was 0.43.
Four statements assessed consumer’s subjective knowledge. Respondents were asked to rate how
much they felt they knew about fish in general, compared to an average person and compared to
their friends. Additionally, two more items were “I have a lot of knowledge about how to prepare
fish for dinner”; and “I have a lot of knowledge about how to evaluate the quality of fish”. For
all items, a 7-point Likert scale ranging from “totally disagree” to “fully agree” was used. This
binnenwerk.indd 230 21-08-2006 12:34:46

measure is consistent with measures used in previous studies (Brucks 1985; Park and others
1994). The internal consistency reliability of the subjective knowledge scale was 0.89, with a
mean inter-item correlation of 0.66. In the case of subjective knowledge, the value of alpha is
high and, at the same time, the mean-inter item correlation is high, above 0.5, which indicates
that the high alpha is due to strongly intercorrelated items, i.e. the topical scale includes some
items that are highly redundant (known as the attenuation paradox in psychometric theory)
(Clark and Watson, 1995). Therefore, a mean inter-item correlation above a threshold indicates
that scale items ought to be more differentiated in order to form a more reliable measure of
the construct of current interest
Next, consumer’s level of objective knowledge about fish was measured with five statements that
are either true or false. It was assumed that those statements should be common knowledge
among at least half of the population. Three of the statements were false: “More than half of the
fish we can buy is farmed fish” (the average market share of aquacultured fish is in the range of
25-30%, depending on the definition, data source and country); “Fish is a source of dietary fibre”
(fish does not contain any dietary fibre, although many consumers believe so because of some
fish’s fibrous texture (Verbeke and others 2005), and “Cod is a fatty fish” (cod is classified as a
lean fish). The remaining two statements were true: “Fish is a source of omega-3 fatty acids”;
and “Salmon is a fatty fish”. For the five statements, a “true” / “false” scale was used (Park and
others 1994). It was opted for not including a “don’t know” answer, which forced respondents to
think and make up their mind about the proposed statements. Through providing the opportunity
to indicate a certainty level (measured by 5-point Likert scales ranging from “very uncertain”
to “very certain”), respondents could reflect how sure/unsure they felt about their answer.
Through this procedure a large share of “don’t know” responses were avoided. Respondents who
really did not know the answer could still report a low level of certainty. It is assumed that
certain general psychological factors such as self-confidence may also influence self-assessed
knowledge judgements. Although Park and others (1994) have not supported this idea, other
studies suggest that consumers’ knowledge judgements may be influenced by a general feeling
of self-confidence, independent of what is actually known (DeNisi and Shaw 1977).
The measurements of self-assessed or subjective as well as objective knowledge will allow

measuring the link between knowledge, pre-purchase information search depending on the
information source, and fish consumption behaviour.
Objective and subjective knowledge about fish across Europe

Comparison of both objective and subjective knowledge between the countries revealed
significant differences. The multiple comparisons (Post Hoc) Duncan test was used to determine
differences in the subjective and objective knowledge between the countries. In Table 1 a, b,
c, d indicate significantly different means of objective knowledge on a 6-point scale (0=none
answers correct; 5=all answers correct) and of subjective knowledge on a 7-point scale (1=totally
disagree; 7=fully agree) between the countries. Respondents from Denmark have the highest
factual knowledge about fish. Danish gave on average 3.67 correct answers on five questions
asked. For all included items the majority of Danish respondents gave a correct answer. Danes
account for 57.2% of the respondents who answered all objective knowledge items correctly,
who constitute 10.7% of the total sample. Danish were followed by Spanish respondents and
further by Dutch and Belgians with respect to objective knowledge. Polish respondents have the
binnenwerk.indd 231 21-08-2006 12:34:46

lowest objective knowledge with regard to fish, with an average of only 2.52 correct answers
on five questions asked (Table 1).
The most common knowledge is that “fish is a source of omega-3 fatty acids” (77% correct in
the total sample), whereas most of the respondents failed to provide a correct answer (“no” in
this case) to the statement that “fish is a source of dietary fibre” (59.7% wrong in the total
sample) (Table 2).
With regard to the second measure of knowledge, i.e. subjective knowledge, Polish and Spanish
consumers show the highest subjective, thus self-esteemed knowledge. On the contrary, Dutch
consumers estimate their knowledge about fish as the lowest one (Table 1).
Women are found to have significantly higher objective knowledge about fish in comparison
with men. Furthermore, it is found that knowledge increases with age: the older the respondent,
the higher their objective knowledge about fish. With respect to the education, higher educated
respondents are found to have significantly higher objective knowledge in general, and in
particular higher correct knowledge about the items: “fish is a source of dietary fibre”; “fish is a
Table 1. Mean score on the two knowledge constructs, comparison between the countries.
Country F-value p-value
Belgium Denmark Netherlands Poland Spain
Objective knowledge* 2.73b 3.67d 2.73b 2.52a 3.05c 156.447 < 0.001
Subjective 3.25b 3.40c 2.96a 3.77d 3.79d 53.388 < 0.001
knowledge**
* 0=none answers correct; 5=all answers correct

**7-point scale (1= totally disagree; 7= totally agree)
Table 2. Percentage of correct answers on the objective knowledge items, comparison between countries.
Country Pearson p-value

Chi-Square
Belgium Denmark Netherlands Poland Spain
More than half of the 31.3 56.1 54.3 34.6 46.6 193.455 < 0.001
fish we can buy is
farmed fish
Fish is a source of 33.5 59.8 28.6 33.9 40.4 256.160 < 0.001
dietary fibre
Cod is a fatty fish 64.1 76.6 63.4 51.3 59.2 153.962 < 0.001
Fish is a source of 69.5 90.9 71.8 67.9 81.3 218.483 < 0.001
omega-3 fatty acids
Salmon is a fatty fish 75.1 83.4 54.9 64.5 77.1 233.365 < 0.001
binnenwerk.indd 232 21-08-2006 12:34:47

source of omega-3 fatty acids”; and “salmon is a fatty fish”. Finally, the comparison between the
income classes also revealed significant differences. Respondents from the upper income class
have the highest objective knowledge about fish in general, whereas analyses of the separate
items showed that the lowest income class score significantly worse than middle and/or upper
income classes on almost all separate items, except for the item: “more than half of the fish
we can buy is farmed fish”.
Associations between knowledge and behaviour towards fish

Next, the correlation coefficients between the knowledge constructs, behavioural intention and
behaviour towards fish were calculated (Table 3). Behaviour towards fish was measured as total
fish consumption frequency per week, which we have defined as the sum of fish consumed at
home and fish consumed away from home. The respondents were asked to state “How often do
you eat fish (at home, away from home)” on a 9-point frequency scale ranging from “never” to
“daily or almost every day”.
All correlations have the expected positive sign (Table 3), indicating that higher knowledge
associates with higher fish consumption (frequency or intention). The strongest correlation
is found between behaviour and intention (r=0.533) and between intention and subjective
knowledge (r=0.356). The latter correlation indicates that people who believe that their fish
knowledge is high are more likely to plan, desire or expect to eat fish in the following days.
The correlation coefficient between subjective and objective knowledge for the total sample
(n=4.786) is found to be significant but very small (r=0.097), thus what people believe to know
about fish matches only poorly with their actual objective knowledge. These results support the
findings from the study by Radecki and Jaccard (1995) where also a weak relationship between
actual knowledge and perceived knowledge was observed. However, the same analysis across the
countries shows that only for the respondents from Belgium (r=0.161), Denmark (r=0.123) and the
Netherlands (r=0.182) the correlation coefficient between the objective and subjective knowledge
is significant. In the case of Spanish and Polish respondents – the two samples with the highest
subjective knowledge – objective knowledge is not correlated with subjective knowledge.
Table 3. Bi-variate correlation coefficients between measured constructs for the total sample (n=4.786).
Objective Subjective Certainty Need for Intention Behaviour1 Behaviour2

knowledge knowledge cognition
Objective knowledge 1
Subjective knowledge 0.097** 1
Certainty 0.259** 0.249** 1
Need for cognition 0.144** ns 0.117** 1
Intention 0.113** 0.356** 0.168** ns 1
Behaviour 1 0.120** 0.294** 0.125** ns 0.533** 1
Behaviour 2 0.044** 0.104** 0.057** 0.054** 0.195** 0.222** 1
ns= not significant ** correlation is significant at the 0.01 level (2-tailed)

1 fish consumption at home * correlation is significant at the 0.05 level (2-tailed)
2 fish consumption out of home
binnenwerk.indd 233 21-08-2006 12:34:47

With regard to the correlations between knowledge and behaviour, the subjective knowledge
was found to have a low and positive correlation with fish consumption at home both in the
total sample (r=0.294) (Table 3) and in all five countries (Table 4), meaning that respondents
who perceive their fish knowledge as high, eat more fish at home. This correlation was the
strongest in Belgium and in Spain. Additionally, subjective knowledge was significantly, but
lower correlated with fish consumption out of home in four of the countries (Table 4), namely
Belgium, the Netherlands, Poland and Spain. Objective knowledge is found to be non-significant
as determinant of fish consumption at home in Poland and away from home in Denmark, the
Netherlands and Spain. For the total fish consumption subjective knowledge was significantly
correlated with consumption frequency in all countries (Table 4), whereas objective knowledge
only in Belgium, the Netherlands and Denmark. Thus, these results support the findings by
Radecki and Jaccard (1995) that subjective knowledge is a better predictor of behaviour (in
our case fish consumption frequency) than objective knowledge.
Knowledge and information source usage and trust

In line with information processing theory, people gain knowledge from their experience and
from the use of external information sources. Therefore, it is important to check what are the
most frequently used information sources with respect to fish. The analyses of the frequency
of use of information sources across the countries shows that for four countries: Belgium,
Denmark, the Netherlands and Spain, the fish monger is the most frequently consulted, while
family and friends is the second most frequently used source of information about fish. In the
case of Poland, family and friends are the most frequently used, whereas the fish monger is
ranked second. However, it should be noted that the mean scores for those items are relatively
low and below the neutral point, meaning that in general the use of information sources about
fish is rather low.
Additionally, respondents with higher NFC use significantly more often government as a source
of information about fish, which indicates that these consumers process information more
Table 4. Correlation between knowledge and fish consumption across countries.
Knowledge Belgium Denmark Netherlands Poland Spain

(n=852) (n=1110) (n=809) (n=1015) (n=1000)
Consumption Objective 0.161** 0.157** 0.101** ns 0.077*

at home
Subjective 0.329** 0.250** 0.285** 0.201** 0.299**
Consumption Objective 0.077* ns ns -0.069* ns
out of home
Subjective 0.208** ns 0.112** 0.130** 0.135**
Total fish Objective 0.171** 0.154** 0.104** ns ns
consumption
Subjective 0.365** 0.218** 0.271** 0.214** 0.300**
** correlation is significant at the 0.01 level (2-tailed)

* correlation is significant at the 0.05 level (2-tailed)
ns = not significant
binnenwerk.indd 234 21-08-2006 12:34:47

carefully and systematically considering the content of the message and evaluating the merits
of the arguments. On the other hand, respondents who are low in NFC are more likely to use
the sources in which they more easily put their trust of information: family and friends, doctor,
dietician and public health recommendations. Our findings fit with the results obtained by
Cacioppo and Petty (1982) who claimed that individuals who are high in NFC, as opposed to
those low in NFC, are “thought to be more likely to expend effort on information acquisition,
reasoning, and problem solving to cope with a wide variety of predicaments in their world”
(Cacioppo and others 1996).
In order to give an indication of both use and trust in information sources, a new variable has
been created by multiplying consumer’s use and trust scores for each information source. The
results show that both for the entire sample and for the individual countries, the most trusted
and used sources of information about fish are the fish monger and family and friends. On the
contrary, the least trusted and used sources of information about fish are government, radio
and scientists.
Correlations between use of and trust in information sources and different levels of actual
knowledge were calculated. First, a factor analysis using maximum likelihood estimation and
Promax rotation was performed based on respondent’s scores on use of 15 information sources.
This yielded three distinct factors, explaining 61.7% of the variance in the initial data. The
first factors can be typified as the objective sources of information (government, scientists,
consumer organisations, newspapers and fish / food industry). The second factor includes the
commercial sources of information (television, advertising and supermarkets); while the third
contains the information sources directly related to health (doctor, dietician and public health
recommendations). Commercial sources of information about fish are most frequently used, and
fish monger
family & friends
public health recomm
doctor
supermarket
dietician
fisherman
TV
industry
consumer org.
newspapers
advertising
scientists
radio
government
1 6 11 16 21 26
Figure 1. Trust × use of information sources about fish (scale from 1 to 49) for the total sample
(n=4.786).
binnenwerk.indd 235 21-08-2006 12:34:48

their use increases with increasing objective knowledge, except for the most knowledgeable
consumers. Information sources related to health are most frequently used among consumers
with an average level of objective knowledge about fish. The use of objective information is
low, particularly among the least knowledgeable fish consumers (Figure 2).
Second, a factor analysis with regard to trust in information sources was performed, using
the same procedure as indicate above. Three factors were identified, namely trust in mass
media and commercial information sources (advertising, television, supermarkets and radio),
trust in personal information sources (doctor, dietician, public health recommendations,
fish monger, family and friends, fisherman/fish farmers, fish or food industry) and trust in
independent information sources (government, scientists and consumer organisation). Then,
the three recognised factors were correlated with the objective knowledge. The higher objective
knowledge about fish, the higher consumer’s trust in personal, independent and commercial
/ mass media information sources. Trust in independent information sources about fish is
particularly stronger among the most knowledgeable consumers as compared to consumers with
average or low levels of objective knowledge (Figure 3).
Use of information cues

With respect to consumer’s use of fish information cues, expiry date was found to be the
most used cue in all countries. For almost all information cues (fish species, price, weight,
expiry date, date of capture, and capture area), except for “nutritional composition”, a higher
objective knowledge significantly increases the use of those information cues. Only in case of
the “brand name”, respondents with lower objective knowledge indicate to use this information
cue more often.
The analysis of the interest in potential information cues showed that respondents from all
countries were most interested in a fish safety guarantee (mean on 7-point scale = 5.51)
3.4
Use of information sources
3.2
3.0 objective
commercial
2.8 related to health
2.6
2.4
2.2
2.0
0 1 2 3 4 5
Number of correct answers
Figure 2. Association between objective knowledge and use of information sources about fish for the total
sample (n=4.786).
binnenwerk.indd 236 21-08-2006 12:34:48

Trust in information sources 5.5
4.5
mass media/commercial
4 personal
independent
3.5
3
1 2 3 4 5 6
Figure 3. Association between objective knowledge and trust in information sources about fish for the total
sample (n=4.786).
and in quality marks (mean = 5.43). The least interesting cue for the consumer was the
batch identification number (mean = 4.04) and information on the feed used during farming
(mean = 4.25). Furthermore, significant differences for all potential information cues were
found for different knowledge levels. A general tendency is recognised that the higher both
the objective and the subjective knowledge, the stronger consumer’s interest in each of the
potential information cues.
With regard to the traceability issues, the respondents from all countries were most in favour
of the idea that the retailer should keep all the necessary information about fish and make
it available upon consumer’s request (respondents scored highest on this traceability item).
Additionally, significant differences were found for five issues with respect to traceability. In all
cases the same tendency was seen: consumers with higher subjective and objective knowledge
about fish indicate to be more willing to pay for a fish with better documentation; express a
stronger desire to have a direct access to as much information as possible about fish; and have
a stronger preference for retailers to keep the information about the fish and make it available
upon request. Figure 4 illustrates that especially the most knowledgeable consumers express
more interest in information from traceability.
binnenwerk.indd 237 21-08-2006 12:34:48

Interest in information from traceability

6
5.5
5
willing to pay for extra fish
4.5 information
important to have direct access
4 to information
retailer keeps the information
3.5 about fish
3
0 1 2 3 4 5
Figure 4. Association between objective knowledge and interest in traceability issues (7-point scale) about
fish for the total sample (n=4,786).
Conclusions
This paper has focused on the impact of consumer knowledge about fish on interest in
information and fish consumption behaviour. Two measures of knowledge were distinguished:
subjective, i.e. self-assessed knowledge and objective or factual knowledge. The correlation
coefficient between those two knowledge constructs for the total sample (n=4.786) was found
to be significant but very small (r=0.097), meaning that what people believe to know about
fish matches poorly with their actual objective knowledge.
The results of this study show that subjective knowledge is a better predictor of behaviour
than objective knowledge, in this particular case, of fish consumption frequency. Subjective
knowledge was found to be a good determinant of total fish consumption in all countries,
whereas the objective knowledge was found to be non-significant as determinant of fish
consumption frequency in Poland and Spain, which are the two countries with the highest
subjective knowledge.
Improving consumers’ subjective knowledge is more likely to cause an increase in the fish
consumption (particularly at home) as compared to the strategies aiming at increasing
consumers’ objective knowledge. Therefore, future communications should concentrate on
consumers’ self-assessed knowledge, e.g. improving consumers’ self-confidence in evaluating
fish quality; because it appears important what people believe to know rather than what they
actually know.
binnenwerk.indd 238 21-08-2006 12:34:48

Acknowledgements
This work was performed within of the FP6 Integrated Research Project SEAFOODplus, contract
No FOOD-CT-2004-506359. The financing of the work by the European Union is gratefully
acknowledged.
References
Bettman J, Park CW. 1980. Effects of prior knowledge and experience and phase of the choice process on consumer
decision processes: a protocol analysis. J Cons Res 7:234-248.
Brucks M. 1985. The effect of product class knowledge on information search behaviour. J Cons Res 12:1-16.
Cacioppo JT, Kao CF, Petty RE, Rodriguez R. 1986. Central and peripheral routes to persuasion – an individual difference
perspective. J Pers Soc Psychol 51(5):1032-1043.
Cacioppo JT, Petty RE, Kao CFChuan Feng Kao. 1984. The efficient assessment of Need for Cognition. J Pers Assess
48(3):306-307.
Cacioppo JT, Petty RE. 1982. The need for cognition. J Pers Soc Psychol 42(1):116-131.
DeNisi AS, Shaw JB. 1977. Investigation of the uses of self-reports of abilities. J Appl Psychol 62:641-644.
Howard JA, Sheth JN. 1969. The theory of buyer behaviour, New York, John Wiley & Sons, Inc.
Park CW, Mothersbaugh DL, Feick L. 1994. Consumer knowledge assessment. J Cons Res 21:71-82.
Radecki CM, Jackard J. 1995. Perceptions of knowledge, actual knowledge, and information search behaviour. J Exp
Soc Psych 31:107-138.
Rao AR, Monroe KB. 1988. The moderating effect of prior knowledge on cue utilization in product evaluations. J Cons
Res 15:253-264.
Rao AR, Sieben WA. 1992. The effect of prior knowledge on price acceptability and the type of information examined.
J Cons Res 19(2):256-270.
Selnes F, Gronhaug K. 1986. Subjective and objective measures of product knowledge contrasted. Adv Cons Res 13:67-71.
Verbeke W, Sioen I, Pieniak Z, Van Camp J, De Henauw S. 2005. Consumer perception versus scientific evidence about
health benefits and safety risks from fish consumption. Public Health Nutr 8(4):422-429.
Verbeke W. 2005. Agriculture and the food industry in the information age. Eur Rev Agr Econ 32:347-368.
binnenwerk.indd 239 21-08-2006 12:34:49

binnenwerk.indd 240 21-08-2006 12:34:49
The importance of bacalhau consumption in Portugal and a
preliminary product consumer test in Lisboa

Abstract
In this project 120 Lisboans were recruited to taste and rank nine different types of salted and
dried cod, in Portuguese named ”bacalhau”. The objective was to identify a possible correlation
between the whole “bacalhau” and the overall liking after desalting and cooking. Only two
products showed a clear connection between how they were ranked and how they scored as
desalted and cooked product.
Keywords: salted and dried cod, bacalhau, consumer test
Introduction
Salted and dried products of cod, in Portugal named “bacalhau”, are in general not very well
known outside the few, but for Norway important markets where the product is consumed.
Even though the history is unclear about who actually discovered and started the cod fisheries
outside of Newfoundland, the result was a vast quantity of salted and dried cod brought back to
Europe, especially to Spain (at first to the north and north-west region) and Portugal (Kurlansky
1999). The tradition of using salt as a way to conserve seafood was already known in these
areas and the southern coasts of Spain and Portugal have for centuries produced large quantities
of salt from sea water using the heat from the sun to evaporate the water (Gallart-Jornet and
others 2004). The practice of salting cod can probably be dated back to the 13th century, but
it is not known where or who actually started the practice of first salting and then drying cod
(Johansen and others 2003). From around 1520 salted and dried cod was an important export
product from the east coast of North-America and was distributed to many European markets.
The transport of slaves from Africa to the Caribbean islands facilitated this distribution and the
less worth products/offal were used as food items for the poor expatriates (Kurklansky 1999).
Throughout the centuries the Caribbean community appreciated the salted and dried products.
This is the reason behind the present interest in and consumption pattern of salted and dried
fish products (cod, saith, alaskan pollack and shark) in Brazil and the Caribbean area (Fjørtoft
2005a, 2005b, 2005c). The catholic tradition of avoiding meat on Fridays and during Lent
also helped to boost the demand because “bacalhau” was an appropriate and accepted food
item made available the whole year around. Even today, the main consumption of salted and
dried cod takes place in communities dominated by Catholics and in most markets the peak
consumption is around Lent/Easter.
“Bacalhau” is traditionally a cod that is bled, gutted and deheaded. Then it is split along the
backbone making it possible to lay the fish flat down (one skinside and one meatside). Half the
backbone, approximately from the vent and forward is removed. For about 30 days, the fish is
pickle salted or left in a saturated salt brine or a combination of the two salting methods. In
binnenwerk.indd 241 21-08-2006 12:34:49

former day only pickle salting was used. After salt curing, the fish was brought out in the open
air to dry while laying with flesh side up on cleaned cliffs and rocks. Therefore the word clip-fish
or even cliff-fish is used. This work required a lot of manpower because the fish had to be spread
out in the morning and collected every afternoon and protected in case of showers. Along the
Norwegian coasts the late spring/early summer offers ideal drying conditions and the famous cod
fisheries in the Lofoten region takes place in the same period. Stockfish (dried cod) was a well
known product, but due to the fierce competition from North-America, many producers in Lofoten
and elsewhere started to make salted and dried fish instead. An advantage was also that the
fish could be salted throughout the year and dried when the weather conditions were favourable
(Johansen and others 2003). Nowadays salted cod is produced regularly on a year around basis
and the drying takes place in kilns. From 100 kg of fresh, gutted and deheaded cod one ends up
with 50-55 kg of “bacalhau” depending on water content (Joensen personal communication).
After drying the fish is ready for shipping and distribution to final consumer. Salted and dried
cod has become a voluminous product and for a lot of Norwegian fish processing plants this
product represents the most important source of income (Bendiksen 2004). The export value of
salted and dried cod from Norway exceeded 200 millions Euros in 2004 and Portugal imported
the same year 21.300 tonnes worth 130 millions Euros. Norway is responsible for producing more
than 50% of the bacalhau consumed in Portugal (www.seafood.no).
“Bacalhau” is the Portuguese name for the Atlantic and the Pacific cod (Gadus morhua and
Gadus macrocephalus), but it is also a protected term for salted and dried products made from
the same two species with a water content of 47% or lower (Anon 2005). In Portugal all other
salted and dried seafood products must be marketed and sold under another name/term. This
dual use of the same name could lead to misunderstandings, but other products of G. morhua
and G. macrocephalus than salted and dried varieties have been very rare in the marketplace.
For most Portuguese people the term “bacalhau” is therefore synonymous with salted and dried
cod (Jensen personal communication). The situation is probably similar in Spain, even though
the import of fresh cod products seems to increase. These products are usually marketed as
“bacalao fresco”, indicating that they are fresh and not salted. The main “bacalao”- product
in Spain today is salted cod, and as with the Portuguese, most people in Spain will associate
“bacalao” with salted (and sometimes dried) cod (Østli and Heide 2004).
Portuguese fishermen were entitled to fish cod among other species along the English coast
from 1353 and from the beginning of the 1500-century the fishing activity also included
crossing the Atlantic Ocean. During the first decades of the 15th century the fisheries at
Newfoundland was well established, and the main product brought back to Europe was salted
and dried cod. Portugal experienced harsh times throughout the 15th and 16th century with
famine and social instability. In this situation the salted and dried cod played an important
role and the monarch supported overseas initiatives that could bring sufficient quantities
of “bacalhau” to ease the domestic lack of cheap and nutritious food items. From then on,
the “bacalhau” became sort of a “hero” and gradually became a very important part of the
Portuguese cultural and gastronomic identity. The “bacalhau”-business has even had important
influence on contemporary Portuguese politics (Garrido 2004). Today, when Portuguese people
or their heirs gather together outside of Portugal, they often meet in “Academia do Bacalhau”
(the bacalhau-academy or –club”). Another example is the Norwegian export of salted and dried
cod to France. Almost all of it is consumed by people living and working in France but with a
Portuguese origin (www.seafood.no).
binnenwerk.indd 242 21-08-2006 12:34:49

The importance of bacalhau consumption in Portugal
It is not possible to find trustworthy statistical information that could be used to estimate
the domestic consumption of “bacalhau” in Portugal. Records from the years 1955-56 show a
consumption rate of about 2 kg in the south and in the southern and eastern interiors while the
habitants of suburban areas like Porto and Lisbon consumed 17 kg pr year on average (Garrido
2004). A qualified guess of the contemporary average rate of consumption is 6-7 kg pr capita
(Jensen personal communication). If one translates these figures into meals, each Portuguese
will consume “bacalhau” for dinner approximately once a week. By visiting Portugal one will
soon discover that “bacalhau” is a regular choice even in the tiniest rural cafes and the piles of
“bacalhau” offered in the super- and hypermarkets further emphasize the importance of these
products. Even if there is no such Portuguese saying, it is often said that Portuguese women are
not ready for marriage before they can prepare at least 365 different plates of “bacalhau”.
Except for ready meals and a few other value added products (all branded), the “bacalhau” in
Portugal must be classified as a commodity or a generic product where the customer is left to
choose in a combination of the (often inaccurate) information given at point-of-sale and by
touching, smelling and selecting one among many other unwrapped whole “bacalhau”. In shops
some customers just grab a fish but many customers do some serious and often time consuming
evaluation of many fish before they finally select one. The customer then usually brings the
whole fish to a shop attendant operating an electrical band saw. The fish is weighted then cut
the way the customer wants. Buying “bacalhau” in Portugal means that the customers must
touch and handle the product. This way of self-service shopping forces the customers to do most
of the product evaluation themselves, in contrast to the way fresh seafood is sold.
During the last decades the category “bacalhau” has changed dramatically (Jensen personal
communication). Until 2-3 decades ago most “bacalhau” were 12 months or more by time of
consumption and it was a quite homogenous product: Yellowish in colour, well dried and with a
characteristic mature taste. Nowadays a lot of different products can be found. It is not unusual
that the biggest retailers stock more than 15 different whole “bacalhau” products varying in
size, species, price, colour, processing, dryness, taste and information at point of sale. If one
adds the different cuts and the value added products (desalted, ready meals) the number of
different “bacalhau”-products exceeds one hundred.
The strong tradition of eating “bacalhau” together with a market situation offering a much more
heterogeneous product portfolio has confused the consumers and concerned the legislators.
Whether the product heterogeneity is a result of a more diverse consumer demand or the result
of undermining an informal agreement on product standards will probably never be known.
While there were more consensuses earlier about how to define and recognize “bacalhau” the
amount of varieties has influenced the consumers’ evaluation and choice to a degree that the
authorities in 2005 issued a law standardizing some of the marketing aspects of the product. The
law is very detailed and regulates the offerings made from retailers. The main paragraphs deals
with maximum water content, product definitions/terms, packaging and storage temperature
in the shop (Anon 2005).
From the background given above it is quite clear that “bacalhau” holds a special status in
Portugal, both as a cultural “identity-tag” and a very much appreciated food stuff for weekdays
and festivities. For Norway as a main supplier of “bacalhau” to the Portuguese market it is
important to know how the Portuguese customers react to the increased number of “bacalhau”
binnenwerk.indd 243 21-08-2006 12:34:49

and try to optimize the product portfolio in such a way that the consumers’ quality demands
are met.
In this study experienced consumers of “bacalhau” evaluated the same fish both as a whole
product (imitating the buying process) and through tasting of a desalted sample from the same
lot of “bacalhau” (resembling the situation when eating the fish). By comparing the results, it
was possible to identify to what degree both evaluations correlate.

The consumer panel
The test took place in Lisboa, Portugal in the spring of 2004 with 120 pre-recruited consumers.
Market intelligence (Jensen personal communication) showed that the majority of the participants
should be middle-aged/old because it was presumed that they had a longer relevant experience.
In addition, two other age groups were chosen so that a cross age comparison could be made.
The general trend in Portugal is that the women are responsible for the shopping/preparing
of food. By watching couples doing retail shopping, one can sometimes observe that the man
is responsible for the selection of the “bacalhau”. So at least in some parts of the Portuguese
society the evaluation and selection of “bacalhau” seems to be a male affair. Therefore, 20%
males were included in the test. On this background, it was decided that the panel should be
have the following characteristics:
• Each participant had to buy and prepare “bacalhau” at least once the last 30 days before
recruitment.
• 50% of the participant was to be 50 years of age or older. The group aged 35-49 and the
group aged 25-34 should each represent 25% of the participants.
• The gender issue was taken into account by using 20% men and 80% women.
• The professional recruiting company provided a panel very close to the characteristics given
above.
The test set-up

The pre-recruited consumers were organized in groups of ten. After a short introduction given
by a Portuguese co-worker they were led to the sensory laboratory. Here they were given
a number tag making it possible to identify each participant throughout the whole test.
A questionnaire had to be filled out before the tasting session started. This questionnaire
contained questions about age, consumption rate, where the “bacalhau” was bought, prepared
and eaten, competence on evaluating whole “bacalhau” and other topics. After the tasting
session the consumers were led back to the rest room where refreshments were available and
information was given about the ranking procedure. The term ranking was chosen to describe
the procedure of assessing the whole “bacalhau” and putting them in order from best to least
liked. Each session with 10 consumers lasted for about 90 minutes. In total 13 sessions, carried
out over 3 days, were necessary. Only parts of the results collected in this study is presented
in this paper.
The products, the “bacalhau”

Due to satiety when doing consumer product tests, the maximum amount of samples in a tasting
session should not exceed 10-12. The normal procedure also includes a “warming-up” sample
equal for all consumers. The results of this sample are not used in the analyses. It was decided
to use 9 different products in addition to the “warming-up” sample made from a non-matured
binnenwerk.indd 244 21-08-2006 12:34:49

but fully salted cod; neutral colour and neutral taste. The “bacalhau” samples were aimed to
reflect the span of “bacalhau” in the Portuguese market. Therefore 4 types of “bacalhaus”
bought in retail-stores in Lisboa (P1, P2, P3 and P4) were selected. Four samples were made at
Fiskeriforskning in Tromsø, Norway (F1, F3, F4 and F8) and the last sample (OLD) was bought
from a producer of traditional “bacalhau” in Norway. This product was regarded to be quite
similar to “bacalhau” in Portugal two decades ago. All the products were made from Atlantic
cod, Gadus morhua. Each lot contained 20 fish of size “Crescido” (1-2 kg), the most popular
size in Portugal. Average size was about 1.8 kg.
Preparing the samples

From each “bacalhau” the belly flaps and tail part were cut away. Then the skin was removed
and the piece was cut in two similar pieces leaving a right and a left loin. Each of the loins
was cut (vertically on the direction of the backbone) in samples of approximately the same
thickness (2 cm). All bones and discoloured parts were removed. The samples were then desalted
through a process developed at Fiskeriforskning. The desalting vessel had a size that enabled
the samples to be completely covered by water (samples and water in a 1:10 weight ratio, no
water exchange). Chicken wire on the bottom secured that all sample sides were exposed to
the water. The vessels were put in the cold store for 24 hours. Excess water was wiped of the
desalted samples before they were put in sealed plastic boxes and put back in the cold store.
All the samples to be used for the same day were desalted simultaneously. To facilitate the
serving, each of the 10 sets of samples was put in a separate aluminium container. 20 ml of
water was added to each container to avoid the samples sticking to the bottom. Each container
was covered using folio of aluminium and steamed in a professional kitchen steam oven for 12
minutes before the samples were served.
Tasting
Each consumer in the group of ten occupied a separate box in the sensory laboratory. For each
product the consumers were asked to tick a position on a 9-point scale ranging from “Detesto”
(I dislike extremely) to “Adoro” (I love). The midpoint of the scale was titled “Não gosto nem
desgosto” (neither like nor dislike). Each product had to be evaluated on a separate sheet
of paper and the serving order for each consumer was randomly selected. The samples were
given a three-digit code as identity and to each consumer water and neutral tasting crackers
were provided in order to cleanse the palate between each sample. It was emphasised to the
consumers to follow this procedure. A spittoon was also provided.
Ranking
Each type of “bacalhau” consisted of 20 fish and the fish used for the ranking were picked in
such a way that they were representative for each type. It was also aimed to pick fish with
approximately the same overall size, weight and thickness, the latter being an important cue for
customers when choosing the optimal “bacalhau” product (Jensen personal communication).
Two separate rooms were prepared for the ranking session making it possible for two consumers
to do the ranking simultaneously. Each room contained two tables and the “bacalhau“ were
randomly piled on one. The consumers were asked to rank the “bacalhau” from best to least liked
by placing the fish on the other table. The consumers were not instructed how to do the ranking.
This meant that all the consumers were forced to touch and move the products, hopefully
resembling what happens in the shop. Each fish was coded with a three-digit number different
from the numbers used in the tasting session. After the consumer had finished the ranking was
registered by a co-worker and the fish were piled on the other table in a random order.
binnenwerk.indd 245 21-08-2006 12:34:50

Panel evaluation
The consumption frequency of “bacalhau” among the participants is shown in Figure 1. As can
be seen from the figure, more than half of the consumers reported a consumption of “bacalhau”
of once a week or more frequent, 10% of the consumers reported to consume “bacalhau”
twice a week or more. On average, the consumption of “bacalhau” is a bit less than once a
week. The consumption pattern was not influenced by age. When comparing competence and
consumption, no connection was found, indicating that other factors than competence drives
the consumption.
As can be seen from Figure 2, more than 55% of the consumers in each age group thought their
competence on evaluating whole “bacalhau” was neither good nor bad. About 40% of those in
the age group above 50 years described their competences as good, and the data indicate that
age as expected results in a higher competence. For the consumer group of 50 years or older,
who have been consuming (and often buying) “bacalhau” about once a week for decades, it
is hard to explain why the majority of them do not report a higher level of knowledge on this
matter. Due to skewed number of participants in each age group one must however be careful
about generalizing the findings.
In all consumer groups (Figure 1) the majority (more than 60%) reported that they consumed
most of the “bacalhau” at home. The percentage increased with increasing consumption. 75%
of those reporting a consumption of 1-2 times a week consumed the majority in home. All
consumers reported that whole “bacalhau” was the most regular product bought. 36% reported
that they sometimes were responsible for the preparation of “bacalhau” at home while 64%
reported that they always were responsible for the preparation.
Factors like consumption pattern and experience both in buying and preparing clearly indicates
that the participating consumers in this test can be characterized as experienced with respect
to buying and preparing “bacalhau”.
50
40
percentage
30
20
10
0
> twice 1-2 times 1-2 times 1-2 times
a week per week per 2 weeks per month
Figure 1. Self-reported consumption of ”bacalhau” (N=120).
binnenwerk.indd 246 21-08-2006 12:34:50

Competence versus age

(each group = 100%)
70
60
50
Percentage
Neither good nor bad

40
Good
30
Very good
20
10
0
> 50 years 35-49 years 25-34 years
(N=63) (N-23) (N=26)
Figure 2. Self-reported competence in evaluating whole ”bacalhau” separated by age.
Tasting
In this study 118 complete sets of tasting evaluations were accepted. For most products, the
scoring alternatives from 1 to 9 were used. The mean scores for the overall liking are presented in
Table 1. The mean score for the products F3 and P3 were significantly different from F4 and OLD
(pairwise t-test, p<0.05). OLD was significantly different from all the other “bacalhau” except
F4 and P2. P2 was not significantly different from any of the other “bacalhau”. Although the
differences were significant it should be remarked that the differences were small in particular
taking into the account that the complete range of the scale was 9 points. On the other hand,
by using the mean one can disguise the fact that all the products were liked and disliked by
different consumer segments.
Table 1. Score distribution of “overall liking” for the “bacalhau”-products (N = 118).
Product Mean scores
F3 6.37a
P3 6.32 a
P1 6.3 ab
F1 6.27 ab
F8 6.26 ab
P4 6.07 ab
P2 6.03 abc
F4 5.95bc
OLD 5.69c
Values marked with different letters were significantly different (p<0.05)
binnenwerk.indd 247 21-08-2006 12:34:50

Ranking
The ranking (Figure 3) is presented as the distribution in percentages of each product ranked
as best and least liked product. Ranking scales 2-8 are not presented in Figure 3. The ranking
results show that F4 and P4 were both ranked as the most favourable products (26% and 22%
respectively). At the same time very few consumers ranked the same products as the least
liked product (1% and 2% respectively). OLD was regarded as the least liked product (32%),
only 4% of the consumers ranked OLD as their favourite. The products P3, F3 and F1 were more
anonymous, i.e. these products were mostly ranked in the scales 2-8. P2, F8 and to a greater
extent P1 seem to be products that were both best and least liked. 24% of the consumers ranked
P1 as the least liked and 13% ranked the product as best liked.
Comparison of taste and ranking

A comparison of the results from the tasting (Table 1) and the ranking (Figure 3.) showed that
F4, although the best liked product assessed in the ranking, scored significantly lower on taste
compared to the products F3 and P3 with the highest scores on taste. In the ranking experiment
F3 and P3 was not preferred by many participants (7% best liked F3 and 3% best liked P3). P1
was ranked as the second lowest product but got one of the highest mean scores for overall
liking by tasting. P4 was ranked as the second best product and got an overall liking score not
significantly different from the best scoring products. OLD was the least preferred product in
both ranking and taste.
Conclusion
The results indicate that the preferences of Lisboan consumers on taste and appearance of
bacalhau are not very well correlated. Only for two of the nine samples investigated a weak
connection was observed between ranking on outer appearance of the “bacalhau” and the
overall liking based on tasting after desalting and cooking. An explanation for this mismatch
Ranking of 9 whole “bacalhau”-products

35
30
25
Percentage
20 ranked first
15 ranked last
10
0
F4 P4 P2 P3 F3 F1 F8 P1 OLD
Figure 3. Ranking the best liked (ranked first) and the least liked (ranked last) among the 9 different
“bacalhau”-products (N=120).
binnenwerk.indd 248 21-08-2006 12:34:51

in evaluation between taste and appearance could be that the number of different products in
the market place has increased a lot the last decade, making it hard for the customers to know
which cues they should use to ensure the best tasting product. The fact that so many of the
participating consumers evaluated their own knowledge of “bacalhau” as neither good nor bad
could support this suggested explanation.
An important lesson from this study is that the producers and distributors of “bacalhau” in and
to the Portuguese market should strive to minimize the gap between the expectations created
by appearance and the performance on the palate. The limited (and randomly selected) number
of products used in this project shows a large discrepancy not to be ignored by the industry.
Otherwise there is a fair chance that consumers will turn away from “bacalhau” simply because
they find the product to difficult to evaluate.
Acknowledgements
The project was financed by Bacalaoforum. Helpful support was given by Øyvind Jensen, The
Norwegian Seafood Export Council in Lisboa and Andreia Martins, Carlos Cardoso and Salomé
Magalhãez at IPIMAR (Instituto Português de Investigacão Marítima) in Lisboa.
References
Anon. 2005. Decreto-Lei n.º 25/2005 de 28 de Janeiro. Diário da República – I Série-A. N.º 20 – 28 de Janeiro, Lisboa,
Portugal
Bendiksen, BI. 2005. Driftsundersøkelsen i fiskeindustrien - Oppsummering av inntjening og lønnsomhet i 2004. (The
economic situation in the Norwegian fishing industry: The annual survey 2004) . Fiskeriforskning. Report nr 19.
(in Norwegian). ISBN-13 978-82-7251-569-9 - ISBN-10 82-7251-569-5
Fjørtoft, L. 2005a. Kartlegging av markedene for konvensjonelle produkter: Trinidad & Tobago, St.Vincent, Grenada og
Barbados (Market survey: Salted and dried fish products in Trinidad & Tobago, St.Vincent, Grenada and Barbados).
FHL report. 20 p.
Fjørtoft, L. 2005b. Kartlegging av markedene for konvensjonelle produkter: Puerto Rico (Market survey: Salted and dried
fish products: Puerto Rico). FHL report. 13 p.
Fjørtoft, L. 2005c. Kartlegging av markedene for konvensjonelle produkter: Jamaica. (Market survey: Salted and dried
fish products: Jamaica). FHL report. 10 p.
Gallart-Jornet L., Escriche-Roberto I., Fito-Maupoey, P. 2004. La salazón de pescado, una tradición en la dieta
mediterránea (Salted fish, a tradition within the Mediterranean diet). Editorial Universidad Politécnica de Valencia,
Spain. 222 p.
Garrido, Á. 2004. O Estado Novo e a Campanha do Bacalhau (The New State and the role of the bacalhau). Circulo de
Leitores. 454 p.
Johansen, E., Mangseth, M. Moe, I. 2003. Bacalao, bacalhau, baccalà. Orkana. ISBN 82-91233-98-5. 130 p.
Kurlansky, M.1999. Cod. Vintage. ISBN 0-09-926870-1. 294 p.
Østli, J.and Heide M. 2004. Markedstest av oppdrettet torsk i det spanske restaurantsegmentet (Market survey: Farmed
cod in Spain). Fiskeriforskning Report nr 4.
binnenwerk.indd 249 21-08-2006 12:34:51

binnenwerk.indd 250 21-08-2006 12:34:51
Traceability: Simulated recall of fish products
Kine Mari Karlsen1 and Gunnar Senneset2
2SINTEF Fisheries and Aquaculture, 7465 Trondheim, Norway
Abstract
This paper describes results from a simulated recall of sixteen different fish products in the
Norwegian fish industry and food retail trade. The objective of the project was to evaluate the
traceability systems in the supply chains of the companies and to test the companies’ readiness
in case of recall of fish products. Products with different degrees of processing of wild and
farmed fish were bought in shops and attempts were made to trace them back to the origin (fish
vessel or breeder) of the products. The study shows that the situation within the Norwegian
fish industry and food retail trade is unsatisfactory. Of the selected fish products almost 40%
of the fish products could not be traced back to the fishing vessel or the breeder.
Keywords: traceability, traceability system, recall, food safety
Introduction
In recent years there has been increased focus on traceability in food supply chains because of
food scandals such as the Bovine Spongiform Encephalopathy (BSE) and dioxin scandal (Elbers
and others 2001; Fallon 2001; Pettitt 2001; Van Dorp 2003; CIES 2005). This has resulted in
new demands from the markets and the authorities for food safety and traceability (EC 178/02
2002; Børresen 2004). The European Union (EU) Food Law (EC 178/02) came into effect on
January 1st 2005 and demands one up-one down traceability. This means that every company
has to know from whom they have received goods and when, and to whom they dispatched the
goods. A traceability system can be paper based or electronically based (Moe 1998; Derrich
and Dillon 2004). Today the fish industry uses paper based traceability systems (Pàlsson and
others 2000; Dreyer and others 2004).
The objective of the project was to establish the status of traceability systems in the Norwegian
fish industry and food retail trade, and additionally to test the companies’ readiness for recall of
fish products. Products with different degrees of processing of wild and farmed fish were bought
in shops and attempts were made to trace them back to the origin (fish vessel or breeder) of
the products.
Definitions
Traceability
The International Organization for Standardization (ISO) defines traceability as follows (ISO
2000):
‘Ability to trace the history, application or location of an entity by means of
recorded identifications.’
binnenwerk.indd 251 21-08-2006 12:34:51

In a product sense, it may relate to:

• the origin of materials and parts;
• the product processing history;
• the distribution and location of the product after delivery.
There are two types of traceability (Moe 1998; Dreyer and others 2004):
• Internal Traceability is the ability to trace the product information internal in a company.
• Chain traceability is the ability to trace the product information through links in a supply
chain, in other words the product information a company gets and gives away. Traceability
is not the product information itself, but is a tool which makes it possible to trace this
information through the supply chain.

Tracefish
Tracefish was a concerted action EU project from 2000-2002 called ”Traceability of Fish
Products” (QLK1-2000-00164). The outcome of the project was three standards for voluntary
electronically systems for chain traceability, in other words recording and exchanging of
traceability information in the seafood chains:
1. The farmed fish standard describes what information should be recorded and where in the
farmed fish supply chain.
2. The captured fish standard describes what information should be recorded and where in the
captured fish supply chain.
3. The technical standard describes how the information should be coded, transmitted or made
available in electronic form.
The information in the farmed and captured fish standards are categorized in “shall”, “should”
and “may”. “Shall” refers to information necessary to identify and trace the movement of the
products trade through the supply chain. “Should” refers to information required by law related
to food safety, labelling and quality and “may” refers to further information required by law
and trade information.
Trade unit
Trade item, hereafter called trade unit, is defined by the European Article Number (EAN) and
the Uniform Code Council (UCC) as any item for which there is a need to retrieve predefined
information and that may be priced, ordered, or invoiced at any point in any supply chain (EAN
and UCC 2005).
Logistic unit
Logistic unit is defined by the EAN and the UCC as an item of any composition established for
transport and/or storage that needs to be identified and managed through the supply chain
(EAN and UCC 2005).
Methods
Survey
There are no standard accepted methods for conducting a survey to test the companies’ readiness
to recall fish products; hence it was necessary to design such a method.
binnenwerk.indd 252 21-08-2006 12:34:51

Fish products
The first step in the survey was to decide which fish products to map (Figure 1). In the survey
the selection of fish products was based on different types of products: farmed fish, wild fish,
different species, degrees of processing and randomly selected retailers, fresh food counters
and producers. It was important to have many different types of fish products in order to be
able to draw reliable conclusions.
Mapping method
The next step in the survey was to decide how to map the fish products. The method used in this
survey was based on the farmed fish standard (CEN 2003 a) and the captured fish standard (CEN
2003 b) of the Tracefish concerted action project, hereafter called the Tracefish standards.
The links in the farmed fish standard are breeders, hatcheries, fish farms, live fish transporters,
processors, transporters, stores and retailers. The links in the captured fish standard are fishing
vessel, vessel landing businesses, auction markets, processors, transporters, stores, traders,
wholesalers, retailers and caterers. The tables of all the links in the Tracefish standards were
extended with columns containing the following information: other name of the information
element, element value, information received from link and comments (Table 1). These tables
were filled out during the mapping of the fish products.
4A 5A
NO Information Interviews with
from the each link in the
personnel supply chain
1 2 3
6
Decide which Decide how Purchase Consumer
Information
fish products to map the of the fish package?
analysis
to map fish products products
4B 5B
YES Information on Interviews with
the consumer each link in the
package supply chain
Figure 1. Overview of the survey.
Table 1. An example of a table used to map fresh Norwegian haddock at the fishing vessel.
Information element Element value Information Comments Shall Should May

(other name of the received
information element) from link
Fishing vessel
CFV01 Food business ID1 Name and address 2 Recorded x
manually
CFV02 Vessel ID Registration number x
CFV03 GMP2 certification x
binnenwerk.indd 253 21-08-2006 12:34:52

Table 1. Continued.
Information element Element value Information Comments Shall Should May

(other name of the received
information element) from link
For each trade unit created

Identity
CFV04 Trade unit ID 7030075050 2 x
(Internal code)
Description
CFV05 Type of unit Norway haddock, 1 x
round
CFV06 Net weight 761 kg 1 x
CFV07 Species Norway haddock x
CFV08 Area/country of origin 0304 1 x
CFV09 Product form x
CFV10 Size grade x
CFV11 Product condition Fresh 2 x
Production history
CFV12 Date of capture or sailing x
CFV13 Fishing method Net 1 x
CFV14 Trawl or soak time x
CFV15 Ethical aspects of fishery x
CFV16 Size grading method x
CFV17 Weighing method x
CFV18 Stowage method x
CFV19 Storage temperature x
control method
CFV20 Storage temperature x
record
For each logistic unit created
Identities
CFV21 Logistic unit ID x
CFV22 Trade unit IDs x
For each unit dispatched (either as a logistic unit or a separate trade unit)
Identity
CFV23 Unit ID x
Destination
CFV24 Next food business ID Name and address 2 Recorded x
manually
CFV25 Date and time of dispatch 21.09.2004 1 Recorded x
manually
CFV26 Place of dispatch Address 2 Recorded x
manually
The information from the survey is confidential, thus the table above is only an illustration how the
table is used to map fish products. 1. Identification. 2. Good Manufacturing Practice.
binnenwerk.indd 254 21-08-2006 12:34:52

Another aid during the mapping of the fish products was logs consisting of information about
the fish products, contact persons, date, time and information received. An example of one
log is shown in Table 2, which describes information noted during the mapping of the fresh
Norwegian haddock. For each fish product one log was made. All communications as visit,
telephone, e-mail and fax with the companies during the survey were included in the logs. The
time used to get the information was recorded. Traceability readiness and systems of companies
were also noted.
Purchase of the fish products

The different fish products were purchased one at a time and attempts were made to trace them
back to the fishing vessel or the breeder. The starting point for the tracing of a fish product
was the label on the consumer package or the information from the personnel at the fresh
food counter (Figure 2). Based on the information received from the retailers or the fresh food
counter the next company in the supply chain was contacted by telephone. For fish products
Table 2. An example of a log used to map fresh Norwegian haddock.
Date of purchase 15.08.2004

Place of purchase Name and address
Link number 0: Information on the consumer package

Product Fresh Norway haddock
Product number No labelling
Producer No labelling
Production date No labelling
Best before date No labelling
Contact information No labelling
Link Company Aid Contact Date Time start Time end Information
person received
Retailer Name Purchase in Name 15.08.2004 10:07 10:16

the shop
Transport Name Telephone Name 15.08.2004 10:23 10:25
Producer Name Telephone Name 15.08.2004 11:00 11:04 Contact
quality
management
leader
Producer Name Telephone Name 15.08.2004 14:25 14:40
Transport Name Telephone Name 15.08.2004 13:00 13:10
Auction Name Telephone Name 16.08.2005 09:00 09:10
market
Fishing vessel Name Telephone Name 16.08.2004 09:15 09:16 No contact
Fishing vessel Name Telephone Name 16.08.2005 14:00 14:10
The information from the survey is confidential, thus the table above is only an illustration how to use
the log to map fish products.
binnenwerk.indd 255 21-08-2006 12:34:52

Fish product
Raw material
Retailer Transport Producer Transport
supplier
The material flow

The focus during the survey
Figure 2. Overview of the fish product mapping.
in consumer package with contact information, this was used as a basis for tracing, and the
producer was contacted directly. The aids mentioned above were used to map the fish products.
The different fish products were also documented with figures describing the material flow in
the supply chains. In this survey the focus was on attempting to trace back the fish products
to the fishing vessel or the breeder, thus the companies were not required to document the
information about the received and shipped fish products with orders or invoices. Information
received from the companies by telephone or e-mail was accepted as a basis for assessment of
the status of the traceability systems in the Norwegian fish industry and food retail trade.
Information analysis
The mapped fish products with their respective supply chains were described with figures, logs
and tables. The possibility of tracing back the fish products to the fishing vessel or the fish
farm was assessed.
Tested fish products

Sixteen different fish products were purchased from retailers and fresh food counters (Table 3).
Twelve of these products originated from wild fish and four originated from farmed fish. The
fish products were either chilled or frozen. Most of the fish products were chilled. Ten of these
products were processed. The number of fish products in the simulated recall was too small to
assess if there is a difference between the product categories, the results will thus be presented
together.
Identification of the fish products

According to the Tracefish standards unique identification of each unit is essential in order
to be able to trace the fish products (EAN 2003 a, b). In the survey no record was made of
whether the recommendations in the Tracefish standards for labelling trade units and logistic
units to transmit information between the companies in a supply chain were followed. Several
of the companies used the date when they received and dispatched the fish products as a
key to further tracing in the respective supply chains. Only five of sixteen fish products were
binnenwerk.indd 256 21-08-2006 12:34:53

Table 3. Overview over the fish products in the survey.
Farmed fish Captured Chilled or

fish frozen in the
shop
Product Product description Unprocessed Processed Unprocessed Processed
number product product product product
1 Frozen saithe fillet, x Frozen

consumer package
2 Fish cake, consumer x Chilled
package
3 Lightly salted Norwegian x Frozen
haddock, consumer
package
4 Frozen salmon steak, x Frozen
consumer package
5 Fresh Norwegian haddock, x Chilled
from fresh food counter
6 Fresh cod, from fresh food x Chilled
counter
7 Frozen halibut, consumer x Frozen
package
8 Peppered mackerel, x Chilled
consumer package
9 Smoked salmon, a x Chilled
producer’s brand
10 Smoked salmon, a retailer’s x Chilled
brand
11 Fresh shrimp x Chilled
12 Shrimp in brine x Chilled
13 Saithe steak with onion x Frozen
14 Spiced herring fillet x Chilled
15 Warm-smoked mackerel x Chilled
16 Ready meal, salmon x Chilled
labelled with batch numbers (Table 4). A batch is defined as a quantity that has gone through
the same processes (ERC 2004). The other fish products in consumer packages were labelled
with best before dates, and the producer used this information to find the production date and
the batch. The date was an important key in transmitting information between the transport
companies and other companies.
Tracing of the fish products

Traceability is based on a clear relationship between batch, trade units and logistic units, thus
it must be presented to fulfil traceability (Kim and others 1995, Moe 1998). In the survey the
relationship between the units varied within the companies in the supply chains. Typical batch
size for a fish producer was one day’s production where deliveries from several fishing vessels
binnenwerk.indd 257 21-08-2006 12:34:53

Table 4. Type of traceability information on the fish products.
Type of traceability key
Product Product description Production date Best before date Batch number Not labelled
number
1 Frozen saithe fillet, x

consumer package
2 Fish cake, consumer x
package
3 Lightly salted Norwegian x x
haddock, consumer
package
4 Frozen salmon steak, x
consumer package
5 Fresh Norwegian haddock, x
from fresh food counter
6 Fresh cod, from fresh food x
counter
7 Frozen halibut, consumer x
package
8 Peppered mackerel, x x
consumer package
9 Smoked salmon, a x
producer’s brand
10 Smoked salmon, a retailer’s x x
brand
11 Fresh shrimp x
12 Shrimp in brine x
13 Saithe steak with onion x
14 Spiced herring fillet x x
15 Warm-smoked mackerel x x
16 Ready meal, salmon x x
were combined. In the farmed fish supply chain one production batch could be one delivery
from a fish farm or it could be deliveries from several fish farmers. Some companies had an
overview of received and shipped fish products, but could not trace the products internally
because they did not record information about the fish products during the manufacturing.
Hence the information was lost and it was not possible to trace the fish product further back
in the supply chain to find the fishing vessel or the breeder.
In the survey it was not possible to find the raw material sources for six of the tested fish
products. Hence almost 40% of the products could not be traced back to the fishing vessel or
the breeder (Table 5).
binnenwerk.indd 258 21-08-2006 12:34:53

Table 5. The results from the tracing of the fish products.
Product Product description Trace back to the fishing Number of communication

number vessel or the breeder for each product
1 Frozen saithe fillet, consumer package Yes 25

2 Fish cake, consumer package Yes 7
3 Lightly salted Norwegain haddock, No 5
consumer package
4 Frozen salmon steak, consumer No 21
package
5 Fresh Norwegain haddock, from fresh Yes 23
food counter
6 Fresh cod, from fresh food counter Yes 10
7 Frozen halibut, consumer package Yes 9
8 Peppered mackerel, consumer package Yes 4
9 Smoked salmon, a producer’s brand No 3
10 Smoked salmon, a retailer’s brand Yes 11
11 Fresh shrimp Yes 6
12 Shrimp in brine No 14
13 Saithe steak with onion No 30
14 Spiced herring fillet Yes 6
15 Warm-smoked mackerel No 9
16 Ready meal, salmon Yes 11
Extent of the work

Table 5 gives an overview over the total number of visits, telephones, e-mails and faxes that
was necessary to trace back the fish product to the fishing vessel, the breeder or to the company
where it was not possible to trace further back. The number of communications for each fish
product during the survey varied a lot. The average number to trace back the fish products
with the respective supply chains was eight. It was necessary to have on average twelve visits,
telephones, e-mails and/or faxes to get an overview of all links in the different supply chains.
The mapping was finished when all the companies in a supply chain were identified. For the fish
products in consumer package with contact information, the producer was contacted directly.
In practice this meant that it was not necessary to contact all companies in all supply chains.
The date and the time use during the survey for all communications were recorded, but this does
not give a realistic picture of the time necessary for tracing back to the raw material source,
because the companies would have prioritised differently in the case of a real recall. In other
companies it was necessary to speak with for example the quality management leader to get
the information necessary to be able to trace the fish product further back in the supply chain.
If this person was on holiday nobody else could give the information about the fish products.
Figure 3 shows that there is not a clear relation between the number of visits, telephones, e-
mails and faxes made in the survey for each fish product (Y-axis) and the number of contacted
companies in the respective supply chains (X-axis).
binnenwerk.indd 259 21-08-2006 12:34:53

35
Number of visits, telephones, 30
25
e-mails and faxes
20
15
10
0
0 2 4 6 8 10 12
Number of contacted companies in the respective supply chains
Figure 3. Relation between the number of visits, telephones, e-mails and faxes made in the survey for each
fish product and the number of contacted companies in the respective supply chains.
Conclusions
During the survey a large number of companies were contacted. Even though the selection of
fish products in the simulated recall was relatively small, the study shows that the situation
within the Norwegian fish industry and the food retail trade is unsatisfactory. Of the selected
fish products, 40% could not be traced back to the fishing vessel or the breeder.
Similarities of the traceability systems in the companies were as follows:

• The date was used as a traceability key in reception, production and delivery of the fish
products. In the EAN.UCC system a Global Trade Item Number (GTIN) is used to identify trade
units and there are several numbering options to construct a GTIN (bar code symbology)
(EAN and UCC 2005). The fish products in the survey were not identified with GTIN except
in the labelling of consumer package.
• Some companies had routines for internal traceability and used batch numbers to identify the
raw material deliveries and their own production batches. There were also other companies
which did not have internal traceability and could not document the connection between
the raw materials, the production batches and dispatched products.
• The knowledge of the Tracefish standards was low. In the survey there was no record of any
companies that actively used these standards.
• Electronic transfer of information between the companies in a supply chain was infrequent.
The transfer of information was mainly based on paper, fax or e-mail. A lot of information
was stored in paper based archives or the information was not stored. In the companies
where tracing was possible, this was largely based on the knowledge and presence of key
persons rather than on systematic registration. A paper based system can be very time
consuming depending on the size and identification of batches, trade units, in addition
to systematically records, documentations, the actual paper system and management. As a
consequence of a time consuming system is a large product quantities may be recalled and
binnenwerk.indd 260 21-08-2006 12:34:54

withdrawn because it is not practically possible to get enough information to recall and
withdraw a limited product volume.
Acknowledgements
The simulated recall of fish products conducted in the Norwegian fish industry and food retail
trade was collaboration between the Norwegian Institute of Fisheries and Aquaculture, SINTEF
Fisheries and Aquaculture, the Norwegian Fisherman’s Association (NFA) and the Norwegian
Seafood Association (NSA). The authors want to thank Knut Eriksen from NFA and Frode
Kvamstad from NSA for participation in the project and fruitful discussions.
References
Børresen T. 2004. Traceability in the seafood chain. INFOFISH International. 3:48-51.

CEN (European Committee for standardization). 2003 a. Traceability of fishery products – Specification of the information
to be recorded in farmed fish distribution chains. CWA (CEN Workshop Agreement) 14659.
CEN (European Committee for standardization). 2003 b. Traceability of fishery products – Specification of the information
to be recorded in captured fish distribution chains. CWA (CEN Workshop Agreement) 14660.
CIES (The food business forum). 2005. Implementing traceability in the food supply chain. Madrid: CIES. 21 p.
Derrich S, Dillon M. 2004. A guide to traceability within the fish industry. Switzerland and Denmark: SIPPO/EUROFISH.
ISBN 1900134187. 78 p.
Dreyer HC, Wahl R, Storøy J, Forås E, Olsen P. 2004. Traceability standards and supply chain relationships. In:
Aronsson, H., editor. Proceedings of the 16th Annual Conference for Nordic Researchers in Logistics, NOFOMA
2004, Challenging Boundaries with Logistics. Sweden p 155-170.
EAN and UCC (European Article Number and Uniform Code Council). 2005. Basics and principles of the EAN.UCC system.
In: EAN and UCC, editor. General EAN.UCC Specifications v.6.0. Belgium and USA. 18 p.
EC (European Communities) 178/02. 2002. Regulation (EC) No 178/2002 of the European parliament and of the council
of 28 January 2002 laying down the general principles and requirements of food law, establishing the European
Food Safety Authority and laying down procedures in matters of food safety. http://europa.eu.int/eur-lex/pri/en/
oj/dat/2002/l_031/l_03120020201en00010024.pdf. Article 18. p. 11.
Elbers ARW, Moser H, Ekker HM, Crauwels PAA, Stegman JA, Smak JA, Pluimers FH. 2001. Tracing systems used during
the epidemic of classical swine fever in the Netherlands, 1997-1998. Rev. sci. tech. Off. Int. Epiz. 20(2):614-
629.
ERC (Efficient Consumer Response). 2004. ECR-Using traceability in the supply chain to meet consumer safety
expectations. Europe: ECR Europe. 75 p.
Fallon M. 2001. Traceability of poultry and poultry products. Rev. sci. tech. Off. Int. Epiz. 20(2):538-546.
ISO (International Organization of Standardization). 2000. Quality management systems. Fundamentals and vocabulary.
(ISO 9000:2000). Section 3.5.4. p. 21.
Kim HK, Fox MS, Gruninger M. 1995. An ontology of quality for enterprise modelling. In Proceedings on Enabling
Technologies: Infrastructure for collaborative enterprises 1995 of WET ICE. Los Albamitos, USA: IEEE. p 105-116.
Moe T. 1998. Perspectives on traceability in food manufacture. Trends in Food Science and Technology. 8:211-214
Pàlsson PG, Storøy J, Frederiksen M, Olsen P. 2000. Traceability and electronic transmission of qualitative data for fish
products. Status report nr. 3. Project nr. 66031400 in The Nordic Minster Council. 22 p.
Pettitt RG. 2001. Traceability in the food animal industry and supermarket chains. Rev. sci. tech. Off. Int. Epiz.
20(2):584-597.
Van Dorp KJ. 2003. Beef labelling: the emergence of transparency. Supply Chain Management. 8(1):32-40.
binnenwerk.indd 261 21-08-2006 12:34:54

binnenwerk.indd 262 21-08-2006 12:34:54
Chapter 4:
Quality seafood
Quality of seafood is an important issue in the fish sector that covers various topics. As fish
is a very perishable product, a lot of work has been done to understand the quality changes
occurring post mortem is fish and the mechanisms of fish deterioration. From the moment
the fish is caught, the spoilage process is started, and the quality and freshness of the fish
is affected. Based on this knowledge, research has been done to improve methods to store
fish in an optimal way and thus prolong the shelf life. For the fish industry, fish authorities
and consumers, the development of satisfying methods to assess the quality, the safety and
the authenticity of fish products is of major importance. The articles in this chapter deal with
these themes.
In the first paper the effects of location the catch, season on the quality of high value cod
products produced by the Icelandic seafood industry. Data were collected on board of fishing
vessels as well in the industry.
Two articles deal with the spoilage patterns occurring in herring. One treats with finding the
enzymes responsible for the belly bursting phenomenon. This would be the first step to provide
the means to decrease its impact. The second paper discusses the effects of various products
which appeared on the market to reduce spoilage rates on fresh fish and fish destined for
fishmeal.
To evaluate the freshness of fish, sensory methods are commonly used in the fish sector. They
have the advantage to be fast, non-destructive and easy-to-use. The development of a QIM
scheme for tub gurnard is described in one article. Chemical methods are as well useful to assess
fish quality. The quality of frozen products on the Portuguese market is evaluated by various
chemical methods in another article.
Protection of the consumer against misleading or unsafe food is the topic of the last
presentations. An overview is given of the use of the additive carbon monoxide (CO) in fish
which can mislead the consumer. CO inhibits the usually occurring brown colouring of fish flesh
so that bright-red coloured fish meat (for example tuna) is not longer indication of freshness.
Finally, a preliminary study was done to obtain information on the effect of culinary treatments
on the allergens from Anisakidae. This allergy can occur in individuals who were previously
sensitised to the parasite.
binnenwerk.indd 263 21-08-2006 12:34:54

binnenwerk.indd 264 21-08-2006 12:34:54
Effect of catch location, season and quality defects on value of
Icelandic cod (Gadus morhua) products
Sveinn Margeirsson1,2, Allan A. Nielsen3, Gudmundur R. Jonsson1 and Sigurjon Arason2,1

1University of Iceland, Department of Mechanical and Industrial Engineering, Hjardarhagi 2-6,
107 Reykjavik, Iceland
2Icelandic Fisheries Laboratories (IFL), Skulagata 4, PO Box 1405, IS-101 Reykjavik, Iceland
3Technical University of Denmark, Informatics and Mathematical Modelling (IMM), Richard
Petersens Plads, Building 321, DK-2800 Kgs. Lyngby, Denmark
Abstract
Effect of location of catch, season and quality defects of cod on the proportion of high-value
cod products was studied in close cooperation with two Icelandic fisheries companies. Data
from 2001-2004 on season and high-value cod products were collected in a fish processing
plant in West-Iceland. Seasonal-, spatial-, quality defects-, and high-value product data were
collected onboard fishing vessels and in another fish processing plant in North-Iceland. Results
show seasonal and spatial variation in proportion of high-value cod products. Close correlation
between quality defects factors and proportion of high-value products indicates that lower
defect rate of cod fillets would result in significantly higher product value.
Keywords: cod, product value, gaping, seasonal variation, spatial variation
Introduction
The return from cod catching and processing has been connected to variables like fillet yield,
gaping and parasites (Birgisson 1995). Another variable, which is considered of importance for
the return of cod processing, is the proportion of different products, i.e. the proportion between
“cheap” and “expensive” products. An example of “cheap” cod product is mince, while different
fillet portion products are “expensive” products. The product types in Table 1 are widespread
in Icelandic seafoood industry.
Table 1. Common product types in cod processing.
1 Fish fillets, fresh. Chilled or on ice

2 Tail (sometimes classified with 3)
3 Frozen fish fillets / frozen portions
4 Frozen fish fillets, in blocks
5 Minced or strained fish, frozen
binnenwerk.indd 265 21-08-2006 12:34:55

A common indicator for the value of the fillet is the Fillet Portion ratio or FP-ratio, where the
first 3 product types in Table 1 are taken into one category, called fillet portions. FP-ratio is
the proportion of the fillet that becomes fillet portions after trimming.
Previous results indicate that return of Icelandic fishing industry can be increased considerably
by managing catching and processing simultaneously (Eyjolfsson and others 2001; Teitsson
1990). Managing catching and processing unabridged and use of prior knowledge along with
the use of previous data to synchronize fisheries and processing can simplify planning for both
fisheries and processing managers (Bjarnason 1997).
Over 40% of the annual cod catch in Iceland is caught by trawlers (Anonymous 2004). It has
long been assumed that both length and size the haul could effect on the quality of the catch.
One haul is the process of dropping a trawl, towing it by the trawler and hauling it in again.
The length of a haul is the time-lag from when the trawl is dropped until it is hauled in again.
The size of the haul is the weight of the catch. It is obligatory for all trawlers catching in
Icelandic waters to register how long the haul is, the size of it, proportion of different fish
species, depth, weather conditions, date and the location of the catch (in GPS coordinates and
square numbers shown in Figure 1).
During the last years increased emphasis has been put on increasing the FP-ratio (Asgeirsson
2005). However, there is limited literature available on the matter. Other important variables
for the return of cod catching and processing have been studied more thoroughly. Love (1975)
found less gaping in large than in small cod. This was contradicted by Birgisson (1995) who
found more gaping in larger cod. Rikhardsson and Birgisson (1996) concluded that gaping
Figure 1. It is obligatory for all boats, fishing in Icelandic waters to register the location of their catch.
binnenwerk.indd 266 21-08-2006 12:34:55

Effect of catch location, season and quality defects on value of icelandic cod products
correlated with condition factor (weight/(length)3) and proportion of viscera. Morphology of

the fish has been related to fillet yield (Cibert and others 1999) and it is likely that condition
factor can be used as an indicator for fillet yield. Eyjolfsson and others (2001) and Margeirsson
and others (2003) found a significant correlation between fillet yield and condition factor and
stated that there was a considerable difference in condition factor and fillet yield between
catching areas in Icelandic waters.
The aim of this study was to map FP-ratio in Icelandic cod, both with respect to catching date
and the location of catching ground and to find out what variables affect FP-ratio most. The
study was carried out in cooperation with two Icelandic companies, Samherji and Gudmundur
Runolfsson (GR). Both companies own trawlers which serve as resources for their land-based
processing.
Methods
Data in this study were twofold. The first data set was collected from the intern production
system of GR in W-Iceland. This data set ranged from January 2001 - June 2004 and showed
how fillets were utilised into different products in the company. Total number of production
days under investigation was 619. All data were scaled with the mean of corresponding year, as
shown in Equation 1. This was done to minimize possible error because of different machines,
labour etc.
FP
FPiknew = meanik(FP ) (1)
k
FPik = FP-ratio from measurement i at year k
FPk = All FP-ratio measurements at year k
Data from Samherji were collected from November 2003 - May 2005. Measurements were done
on 512 cods in total. They all came from the same trawler, Bjorgulfur EA-312. The date and
location of each haul were registered, along with the length and size of the haul. The location
was registered as GPS coordinates at the time the trawl was dropped. The depth and ocean
temperature were registered. After hauling the fish were gutted and iced in tubs. The size of
the haul was estimated by counting number of tubs. After unloading the catch, but before
processing, samples of four cods were taken from 2-3 tubs (hauls) from each fishing tour. The
total weight of cod in the tubs and the length and weight of the sampled cod were measured
and registered. Marel® PV 1740 (d = 20 g) was used for weighing and the length of the fish
was measured with a steel yardstick. The fish was headed, weighed (using Marel PV 1740),
filleted and skinned. Heading was done with Baader® 434 and filleting with Baader 189. The
skinning machine was Baader 51. The fillets were weighed using Marel PL 2010 (d = 1 g) and
all visual parasites counted and plucked out. Gaping was measured by putting a transparent
plastic card with a grid on the fillets (Figure 2). The grids were 4cm x 4cm. If a gaping area
on the fillet covered one grid cell, it counted as one gaping unit. One unit was also counted if
the aggregated area of two or more gaping areas covered one grid cell. The same measurement
was used for bloodstains and redness. At last, the fillets were cut into different product types
and the amount in product types weighed.
binnenwerk.indd 267 21-08-2006 12:34:55

Figure 2. A plastic card with a grid, used for measurements on bruises and gaping.
The geostatistical analysis was mainly based on semivariograms and kriging. Variowin® 2.21,
software available on the internet (Pannatier 1999) was used for making semivariograms, which
show if there is spatial correlation in the data (Nielsen 2004; Anselin 2003).
N(h)
1
Semivariogram (experimental): γ̂(h) = 2N(h) Σ[z(rk) - z(rk+h)]2,
k=1
where z(rk) and z(rk+h) are scalar measurements from the points r and r+h and N(h) is the
number of point pairs separated by the vector h. It is custom to calculate average of γ̂(h)over
the interval h+∆h in order to obtain N(h) high enough for attaining low estimation of the γ̂(h)
variance.
On later stages, a smaller dataset was obtained from the whole dataset, called sub-dataset A.
Sub-dataset A included measurements from 15.11.2003-6.2.2004 (measurements from 20.1.2004
- 26.1.2004 though discarded).
Multivariate regression analysis was used to search for a functional relationship (multivariate
linear model) between the response variables and the independent variables after a thorough
outlier detection of all variables. S-PLUS® 6.1 (MathSoft) and Matlab® (MathWorks) were used
for statistical analysis.
GR data - Seasonal effect

The data from GR were analysed with respect to time of processing only. Figure 3 (left) shows
how the FP-ratio changed with the season. It is evident from the figure that the FP-ratio is low
in April and the late summer months. Figure 3 (right) shows the catching location of GR’s trawlers
from 2.3.2004-10.5.2004. It can be assumed that approximately the same area applies to the
catching locations in the March-May period in 1999-2003 and therefore also that GR catches
close to known spawning areas out of the West and South coast (Begg and Marteinsdottir 2003).
Gaping has been connected to chemical changes right after spawning in cod (Anonymous 2005),
and spawning might therefore explain the decrease in FP-ratio in April. The most likely reasons
for the late-summer drop in FP-ratio were considered labour force connected (Gudmundur Smari
binnenwerk.indd 268 21-08-2006 12:34:56

1.1
1.05
66°N
1
Scaled FP-ratio
0.95
0.9
Reykjavik
0.85 64°N
2001
0.8 2002
2003
2004
0.75
0 2 4 6 8 10 12
26°W 24°W 22°W
Number of month
Figure 3. Left: Scaled FP-ratio with respect to number of month in 2001-2004. Right: Catch locations of GR
trawlers 2.3.2004-10.5.2004.
Gudmundsson, personal communication). Emphasis on processing of other fish species than cod
in the summer along with vacations may reduce the cod-processing competence of the labour
force. It was considered unlikely that such reasons were connected to the FP-drop in April.
Samherji data – spatial and seasonal effect

Figure 4 shows all FP-ratio measurements from Samherji (November 2003-May 2005). A visual
inspection reveals a considerable internal spatial variability in each area. Despite this variability
one can find significant difference in FP-ratio between areas. The area marked as 1 on Figure 4
had for instance a mean FP-ratio of 0.695 ± 0.018 (95% confidence limit) while area 2 had a
mean FP-ratio of 0.744 ± 0.017. The rise in the semivariogram for sub-dataset A (Figure 5) implies
more similarity between measurements that are spatial neighbours than between measurements
that are distant from each other, which also indicates spatial difference in FP-ratio.
To analyze the effect of season, the data were plotted as a function of time (Figure 6).
Although having some effect, season did not have nearly as much effect on FP-ratio here as
by the measurements at GR. The reason for less effect of season on FP-ratio at Samherji than
GR is probably twofold. Firstly, Samherji’s processing is focused on cod all year round and
should therefore not suffer from lack of labour force training at any time. Secondly, Samherji’s
trawlers are catching in other areas than GR’s trawlers (see Figure 4 and Figure 3 (right)). The
GR-trawlers catch close to spawning areas while the Samherji-trawlers do not, at least not to
the same extent.
Effect of quality defects, environmental conditions and catching method

Even though season and location of catch may be important factors in terms of FP-ratio,
it is evident that other variables may come into play or even be underlying reasons for the
importance of season and location. ANOVA (Analysis of variance) was used in order to find
relationship between the measured independent variables (such as weight of the cod and haul
binnenwerk.indd 269 21-08-2006 12:34:56

Figure 4. FP-ratio from all measurements in Samherji, November 2003 – May 2005.
γ(|h|)
50
0.0048
0.0042
0.0036 15
0.003
118
0.0024
96 77
0.0018 47
0.0012 53
0.0006
0
0 9000 18000 27000 36000 45000 54000 63000 72000 81000
|h|
Figure 5. Semivariogram for FP-ratio, sub-dataset A.
size) and FP-ratio. Thereafter a multivariate linear regression model was made with the most
important independent variables (variables with lowest p-values). As a start of the analysis,
assumption of normality of FP-ratio was checked. There was some deviation from normality
(skewness), but it was however not considered necessary to sacrifice the simplicity of the data
by transforming them.
binnenwerk.indd 270 21-08-2006 12:34:57

Measurements Average
0.85
0.8
0.75
0.7
FP-ratio
0.65
0.6
0.55
0.5
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Number of months from start
Figure 6. FP-ratio as a function of month from start of measurements. Month number one is November 2003,
month number 19 is May 2005.
Table 2. Analysis of variance of the measured variables (main effect). The variables with p<0.05 have
significant effect on FP-ratio (95% confidence).
Source d.f. Sum of squares F p
Nematodes/kg 1 0.0022 0.79 0.37

Bloodstains/kg 1 0.0030 10.85 0.001
Redness/kg 1 0.0022 0.80 0.37
Gaping/kg 1 0.18 66.61 <0.0001
Age 1 0.00008 0.03 0.87
Haul length 1 0.00035 0.12 0.72
Haul size 1 0.00038 0.14 0.71
Weight of cod 1 0.0061 2.19 0.14
Condition Factor 1 0.0047 1.68 0.20
Head proportion 1 0.000077 0.028 0.87
Residual error 377 1.05 0.0028
The ANOVA was perfomed on the whole Samherji dataset. The effect of bloodstains/kg, gaping/
kg and weight (the variables with the lowest p-values) was estimated with a classical linear
regression model (Johnson and Wichern 2002) explained in Equation 2.
Y = 0.74 - 0.0061x1 - 0.0046x2 + 0.0073x3 (2)
binnenwerk.indd 271 21-08-2006 12:34:59

As the low correlation coefficient indicates, only a small amount of the variability in the data
can be explained by the model. This is caused by at least two factors. Firstly, spatial and
seasonal effects are not modelled. As previous results show, both catch location and season
affect FP-ratio. It was however not considered feasible to model the effect of location and
season with such few data. Secondly, there are a number of factors that are not registered or
measured in this study which may increase variability in the data. Different crews onboard the
vessel, different weather conditions, feed and genetic diversity are examples of such factors.
However, even though only a little amount of the variability is modelled, it is important to
know that bloodstains, gaping and weight affect FP-ratio, e.g. when explaining to vessel crews
why and how they shall avoid bloodstains and gaping.
Table 3 shows that bloodstains/kg and gaping/kg have significant (p<0.05) negative effect on
FP-ratio. The mean values for bloodstains/kg and gaping/kg were 0.47 and 2.83 respectively
(the Samherji data). It can therefore be estimated that the average cod has lost over 1.5
percentage points in FP-ratio only because of bloodstains and gaping. On the other hand can
the processing manager expect 0.7 percentage points increase in FP-ratio for every kg’s increase
in weight of the (gutted) cod (p=0.059). It may though be a dangerous long-term strategy only
to go after the large cods (i.e. by size-selective fishing gear), since it has been shown that large
cod in Icelandic waters produces greater quantity of eggs and more viable larvae than younger
and smaller cod (Marteinsdottir and others 2000a; Marteinsdottir and Begg 2002).
Table 3. Coefficients in linear regression for FP-ratio.
Symbol in Value Std. Error t-value p(⎮t⎮)

Equation 2
(Intercept) 0.74 0.01 74.9 <0.0001

Bloodstains/kg x1 -0.0061 0.0025 -2.44 0.015
Gaping/kg x2 -0.0046 0.0010 -4.87 <0.0001
Weight of cod x3 0.0073 0.0038 1.90 0.059
R-squared = 0.093
Conclusion
Seasonal and spatial difference in FP-ratio in Icelandic cod was found to some extent. More data
is required for confirmation. The spatial difference might be caused by a certain regionality of
individual cods. That would concur with recent studies indicating differential regional spawning
components (Begg and Marteinsdottir 2000; Marteinsdottir and others 2000b) and older tag-
recapture studies where most tags turned up in close proximity to where the fish was tagged
(Thorsteinsson and Marteinsdottir 1992; Jónsson 1996). Gaping, bloodstains and the weight
of the cod were the variables found to affect FP-ratio most, but modelling with those variables
only explained a small amount of the variability in the data. The knowledge gained in the study
may be of use in managing catching and processing as a whole in Icelandic fisheries companies,
binnenwerk.indd 272 21-08-2006 12:34:59

although much is to be done in order to gain a full understanding of the factors that affect
value of cod products.
Acknowledgements
The help and advice of Dora Gisladottir, quality manager in Samherji hf, is highly appreciated.
We would like to thank the quality department in Samherji for collection of data; GR’s CEO for
access to internal production data and Gudjon Thorkelsson for discussion on the study and its
results. This study was a part of the project COD PROCESSING FORECAST and was supported
financially by the Icelandic Centre for Research (Rannis) and the AVS fund.
References
Anonymous. 2004. Aflahefti 2003-2004 (Catch report 2003-2004). Fiskistofa, Ingolfsstræti 1, IS-101, Reykjavik.
Anonymous. 2005. Gaping of fillets. http://www.fao.org/wairdocs/tan/x5935e/x5935e01.htm Visited 23.8.2005. Fao.
org. Source of website: R. M. Love. Ministry of Agriculture, Fisheries and Food. Torry research station (Torry advisory
note no. 61)
Anselin L. June 24, 2003. An Introduction to Variography using Variowin. Available at: http://sal.uiuc.edu/csiss/pdf/
variowin.pdf. Visited 19.5.2005
Asgeirsson A. 2005. Increased value of seafood through new and improved technology. Lecture in a congress on
increased value of seafood, organised by the ministry of fisheries in Iceland, 8.9.2005.
Begg GA, Marteinsdottir G. 2000. Spawning origins of pelagic juvenile cod Gadus morhua inferred from spatially explicit
age distributions: potential influences on year-class strength and recruitment. Mar Ecol Progr Ser 202:193-217.
Begg GA, Marteinsdottir G. 2003. Spatial partitioning of relative fishing mortality and spawning stock biomass of
Icelandic cod. Fish Res 59:343-362.
Birgisson R. 1995. AFLABÓT. Náttúrulegur breytileiki þorsks með tilliti til eiginleika í vinnslu (Natural variability of cod
regarding production properties). Icelandic Fisheries Laboratories. Report 111:6-14.
Bjarnason KG. 1997. Upplýsingakerfi skipstjóra. Aflakort og aflaspár. (Information systems for captains). University of
Iceland, Mechanical Engineering. Reykjavik. M.Sc. thesis, p.60
Cibert C, Fermon Y, Vallod D, Meunier FJ. 1999. Morphological screening of carp Cyprinus carpio: relationship between
morphology and fillet yield. Aquatic Living Resources, 12(1):1-10.
Eyjolfsson B, Arason S, Þorkelsson G, Stefánsson G. 2001. Holdafar þorsks, vinnslunýting og vinnslustjórnun (Condition
factor of cod, production yield and production management). Report 2-01. Icelandic Fisheries Laboratories.
Reykjavik.
Johnson RA, Wichern DW. 2002. Applied Multivariate Statistical Analysis. Prentice Hall, New Jersey.
Jónsson J, 1996. Tagging of cod (Gadus morhua) in Icelandic waters 1948-1986. Rit Fiskideildar 14:1-82.
Karlsdottir H. (supervision), 2005. Statistical Series Fisheries. Statistics Iceland, Borgartuni 21a, 150 Reykjavik. p.
11-12.
Love RM. 1975. Variability in Atlantic Cod (Gadus morhua) from the Northeast Atlantic: a Review of Seasonal and
Environmental Influences on Various Attributes of the Flesh. J. Fish. Res. Board Can 32:2333-2342.
Margeirsson S, Jónsson GR, Arason S, Þorkelsson G. 2003. Yield, physical characteristics and quality of cod catch
(Nýting, eðliseiginleikar og gæði þorskafla). M.Sc. thesis, University of Iceland, department of engineering, 53-67
(in Icelandic).
Marteinsdottir G, Guðmundsdóttir A, Þorsteinsson V, Stefánsson G. 2000a. Spatial variation in abundance, size
composition and viable egg production of spawning cod (Gadus morhua L.) in Icelandic waters. ICES J. Mar Sci
57:824-83.
Marteinsdottir G, Gunnarsson B, Suthers IM. 2000b. Spatial variation in hatch date distribution and origin of pelagic
juvenile cod in Icelandic waters. ICES J Mar Sci 57:1184-1197.
binnenwerk.indd 273 21-08-2006 12:35:00

Marteinsdottir G, Begg GA. 2002. Essential relationship incorporating the influence of age, size, and condition on
variables required for estimation of reproductive potential in Atlantic cod Gadus morhua stocks. Mar Ecol Progr
Ser 235(2003):235-256.
Nielsen AA. 2004. Geostatistik og analyse af spatielle data. Course material. IMM, Technical University of Denmark.
p. 1-5.
Pannatier Y. 1999. VARIOWIN: Software for Spatial Data Analysis in 2D. Available at http://www-sst.unil.ch/research/
variowin/#DOWNLOAD. Visited 19.5.2005.
Rikhardsson JH, Birgisson R. 1996. Aflabót (Improved catch). Icelandic Fisheries Laboratory. Reykjavík. Report no.
48:44-51.
Thorsteinsson V, Marteinsdottir G. 1992. Tagging of cod at spawning sites on the North and East Coast in 1991. Ægir
2:60-64 (in Icelandic).
binnenwerk.indd 274 21-08-2006 12:35:00

Protein degrading enzymes in herring (Clupea harengus) muscle and
stomach
Hanne Solvang Felberg1 and Iciar Martinez1,2

1SINTEF Fisheries and Aquaculture, Ltd., N-7465 Trondheim, Norway
2Norwegian College of Fishery Science, University of Tromsø, N-9037 Tromsø, Norway
Abstract
The activities of pepsin from stomach and gelatinolytic enzymes from blood and skeletal muscle
isolated from herring (Clupea harengus) caught in the spring were examined by zymography.
Extracted pepsin activity was low. Gelatine zymography revealed the presence of several intense
bands with gelatinolytic activity in the ventral muscle, but not in the dorsal muscle. Blood
had one main band with activity that displayed similar electrophoretic mobility to one of the
weakest bands from ventral muscle. These results indicate that the identified gelatinolytic
activities in ventral muscle contribute to the belly bursting of this species during the heavy
feeding spring period.
Keywords: herring, pepsin, matrix metalloproteinases, gelatinolytic activities, zymography,

belly bursting
Introduction
“Belly bursting” is the post mortem rapid tissue degradation that ends up in the disruption of
the abdominal wall in some species of pelagic fish. This process occurs during the heavy feeding
season, usually in spring. The degradation may be so severe that a few hours after capture
may be enough for the fish to become unsuitable for human consumption. Belly bursting is
commonly attributed to the effect of proteases which either originate in the digestive system
of the fish, the zooplankton, the intestinal flora or the fish muscle (Gildberg 1982; Martinez
and Gildberg 1988; Huss 1995).
Belly bursting has been coupled to weakening of the collagen due to gastric acid leakage and
pepsin activities in capelin (Mallotus villosus) (Gildberg 1982), and to tryptic and chymotryptic
activities in anchovy (Engraulis encrasicolus) (Martinez and Gildberg 1988) and it is known that
collagen is degraded during belly bursting (Gildberg 1982).
In fish, extracellular matrix components, including collagen, have been suggested to be

degraded by matrix metalloproteinases (MMP’s) (Bracho and Haard 1995; Teruel and Simpson
1995; Saito and others 2000a, b; Kinoshita and others 2002; Kubota and others 2003; Lødemel
and Olsen 2003; Lødemel and others 2004; Olsson and others 2006), a group of calcium
and zinc-dependent endopeptidases important in the turnover of some extracellular matrix
components (Woessner 1991; Birkedal-Hansen and others 1993; Massova and others 1998).
Finding enzymatic causes for the belly bursting phenomenon would be the first step to provide
the means to remedy or decrease its impact, which may have important economical consequences
binnenwerk.indd 275 21-08-2006 12:35:00

for fishermen and the seafood industry. To our knowledge, the only previous work published
on belly bursting in herring is the one of Almy (1926), and no other published work on the
involvement of gelatinolytic activities in this phenomenon has been found. The present work
contains the first part of the activities undertaken by our group to address possible enzymatic
causes for belly bursting in herring, namely analyses of i) pepsin activity in stomach extracts
and ii) gelatinolytic activities in ventral and dorsal white muscle extracts and blood of herring
captured during the heavy feeding spring period in the North Sea. This report is believed to be
the first on the activities of pepsin and gelatine degrading enzymes from herring tissues.

Fish samples
Herring (Clupea harengus) had been captured by purse seiner in the North Sea in the spring of
2005. Immediately after capture samples of blood were taken and frozen at -20 ºC. The rest of
the fish were placed in ice and samples were taken after 24 hours, when they started to show
signs of bursting. The stomach and samples of dorsal and ventral white muscle were dissected
and stored at -20 °C on board the fishing vessel and at -80 ºC upon arrival to our laboratory.
Five fish with an average weight ofh 168.0 ± 15.0 g and 26.1 ± 0.6 cm in length were used for
these studies. All samples were analyzed within 4 months of capture.
Pepsin extraction
The stomach, which was filled with prey, was dissected, cut open, rinsed with distilled water
and the gastric mucosa was scraped. The gastric mucosa from 3-4 fish were pooled due to their
small size. The pools were homogenised with distilled water (4 °C, pH 6.0) in a ratio sample:
water of 1:10 (w/v) and the homogenate was centrifuged at 16000 g for 30 min at 4 °C. The
pH of the supernatant was lowered to approximately 2 with 1 M glycine-HCl (Gly-HCl) and it
was left overnight at 4 °C to activate the pepsinogen to pepsin. The amount of protein in the
supernatant was estimated by the Bradford method (Bradford 1976). Porcine pepsin (Sigma
P-7000) was used for comparison purposes.
Pepsin zymography
Fish pepsin activity was visualised in home-made mini gels (7 cmx8 cm) after electrophoresis
under neutral conditions (pH 7.5) as described by Díaz-López and others (1998). The composition
of the gels is shown in Table 1. The polyacrylamide solution consisted of a mixture of acrylamide
and piperazine diacrylamide in the ratio 30:0.8 (Laemmli 1970; Hochstrasser and others 1988).
Electrophoresis was performed in the MiniProtean III electrophoresis cell (Biorad) for 1 hour
at constant 100 V. At this point, the tracking dye was about 2 cm from the bottom of the gel
and the run was stopped.
After electrophoresis, the gels were incubated first for 15 min in 0.1 M HCl to activate the
enzyme, followed by 30 min in a solution containing 0.25% hemoglobin (Sigma) in 0.1 M
Gly-HCl, pH 2.0 at 4 °C, and finally for 90 min in a fresh Hb solution at 37 °C. Gels were then
washed in distilled water, fixed for 15 min in a 12% TCA solutions, stained with 0.1% Coomassie
Brilliant Blue R250 in 45% ethanol and 10% acetic acid and destained with 10% ethanol 10%
acetic acid. During destaining, clear zones, indicating protease activity can be visualised after
a few minutes, but well-defined zones are obtained after 2-4 hours of staining. Gels were dried
between two sheets of cellophane, scanned and examined by visual inspection.
binnenwerk.indd 276 21-08-2006 12:35:00

Protein degrading enzymes in herring muscle and stomach
Table 1. Composition of gels for pepsin zymography (Díaz-López and others 1998).
Stacking gel Resolving gel
Polyacrylamidea 5% 15%
Gel buffer composition 0.1 M Tris-H3PO4, pH 5.5 0.07 M Tris-HCl, pH 7.5
Ammonium persulfate 0.04% 0.075%
TEMED 0.15% 0.08%
a Polyacrylamide
consisted of an acrylamide-piperazine diacrylamide mixture in the ratio 30:0.8
(Laemmli 1970; Hochstrasser and others 1988).
Extraction of gelatinolytic enzymes from muscle

Gelatinolytic activities were extracted and analyzed as described by Lødemel and Olsen (2003).
In short, dorsal (n = 5, analyzed individually) and ventral (the ventral muscles of 5 fish were
pooled prior to extraction due to their small size) samples of muscle and of coagulated blood
(n= 2, analyzed individually) were homogenised on ice with the extraction buffer (50 mM Tris-
HCl pH 8.0, 10 mM CaCl2 and 0.25% Triton X-100, in a ratio sample:buffer 1:10 (w/v). The
samples were agitated for 4 hours at 4 ºC prior to centrifugation (6000 x g, 30 min, 4 ºC). The
supernatant was decanted and stored on ice. The pellet was resuspended in 10 ml of extraction
buffer and centrifuged for 20 min at 20000 x g and 4 ºC. The last step was repeated and the
three supernatants were pooled (yielding a total volume of 40 ml) and stored at -80 ºC until
further use. Protein concentrations were determined by the Bradford method (Bradford 1976).
Gelatine SDS polyacrylamide gel electrophoresis (Gelatine zymography)

Substrate gel electrophoresis were performed according to Lødemel and Olsen (2003) using
the Laemmli’s system (Laemmli 1970). Samples were diluted with a modified Laemmli buffer
(62.5 mM Tris-HCl pH 6.8, 4% SDS, 25% glycerol and some bromophenol blue). Non-heated,
non-reduced samples were loaded into to wells of 0.5 mm thick home-made mini gels whose
composition is shown in Table 2.
Electrophoresis took place for 30 min at 100 V followed by 75 min at 150 V in a MiniProtean
III electrophoresis cell (Biorad) placed in an ice bath. After electrophoresis, the gels were
Table 2. Composition of gels for gelatine zymography (Lødemel and Olsen 2003).
Stacking gel Resolving gel
Polyacrylamidea 5% 9%
Gelatine - 0.1%
Gel buffer composition 0.125 M Tris-HCl, pH 6.8 0.375 M Tris-HCl pH 8.8
SDS 0.1% 0.1%
Ammonium persulfate 0.05% 0.075%
TEMED 0.1% 0.08%
a Polyacrylamide consisted of an acrylamide-bis acrylamide mixture in the ratio 30:0.8 (Laemmli 1970).
binnenwerk.indd 277 21-08-2006 12:35:00

washed twice in 2.5% Triton X-100 for 15 min prior to incubation overnight in 50 mM Tris-
HCl pH 8.0 and 10 mM CaCl2 at 38 ºC. The gels were then stained with 0.1% Coommassie
Brilliant Blue R250 in 45% ethanol and 10% acetic acid. To maximise the contrast between
clear bands and the background the gels were destained in 40% ethanol and 10% glycerol. The
gelatinolytic activities were identified as clear zones against a blue background. The molecular
mass markers used were the Low Molecular Weight Calibration kit for SDS Electrophoresis of
Amersham Biosciences (cat. No. 17-0446-01), and they were treated as the samples: dissolved
in modified Laemmli buffer with no reductants or heat denaturation. These standards were
used as internal standards to verify that there were no variations attributable to run-to-run
differences. Gels were dried between two sheets of cellophane, scanned as described above and
examined by visual inspection.
Pepsin activity
Zymography was the selected technique to examine proteolytic activities due to its convenience
and ability to render fast and reproducible results for both pepsin (Díaz-López and others
1998) and gelatinolytic activities (Heussen and Dowdle 1980; Birkedal-Hansen and Taylor 1982;
Snoek-van Beurden and Von den Hoff 2005).
Figure 1 shows the pepsin zymogram of pig and herring pepsin extracts. Activity in the herring
extract was only detected in the area corresponding to the front of the migration and increasing
the acrylamide percentage of the resolving gel from 12% (which was the concentration used
in the first gels, results not shown) to 15% did not permit a better resolution of the activity.
Porcine pepsin activity was observed in two bands. Other fish pepsins has been examined
using the same technique, and how the enzyme activities appear on a pepsin zymogram varies
depending on the species (Díaz-López and others 1998). While some of the species examined
by these authors had two or three bands (seabream, Pagus and Pagellus) others had only one
(Sarpa and Boops).
The pepsin activity detected was weak, which agrees with the results published (Díaz-López and
others 1998) for other species using the same system described here. These authors noted that
P H
Figure 1. Pepsin zymogram. Only the lower part of the gel displaying pepsin activities is shown. P, pig pepsin
and H, herring pepsin extract. The arrow indicates the only band with activity found in herring extracts,
which co-migrated with the electrophoretic front.
binnenwerk.indd 278 21-08-2006 12:35:00

omnivorous and herbivorous species showed a single proteolytic fraction with low activity and
migrating with the front, as was to the case in herring. In addition, (Gildberg 1982) showed
that there is a negative correlation between feed content and amount of extractable pepsin
from fish stomach, which would further explain why the activity registered was low, since the
stomachs used for this work were filled with prey. It is possible that the levels of pepsinogen
were low if most of it was being secreted, activated to pepsin and bound to the substrate,
which filled the stomachs.
Gelatinolytic activity
Fish with signs of belly bursting (stored for 24 hours on ice after capture) were selected for this
first work in order to identify proteolytic activities in action during the bursting phenomenon.
Further work is going on in our laboratory to examine ventral muscles of herring in the autumn,
when belly bursting does not occur. For all seasons, however, with or without belly bursting, we
use dorsal muscle as control, since no phenomenon analogous to that of belly bursting occurs
in the dorsal muscle.
The gelatinolytic activities of the extracts from dorsal and ventral muscles and blood are shown
in Figure 2. Blood was included in this analysis in order to identify activities originating in this
tissue, which may have contributed to the gelatinolytic activities from skeletal muscles. This was
done because we can be confident that there was blood retained in the capillary vessels in the
muscles since the fish had been captured by purseseiner and were therefore not properly bled.
There were three very strong bands of gelatinolytic activity in the extracts from ventral muscles,
in contrast to only one very weak band in extracts from the dorsal muscle (labelled with an
arrow in Figure 2). Interestingly, there were very few proteins detected by Coomassie Blue
m D V m B m
Figure 2. Gelatine zymography of total extracts of herring, D, dorsal and V, ventral muscles and B, blood.
Approximately 10 µg of total protein of dorsal extract was loaded, 5 μg of ventral extract and 25 μg of blood
extract. The arrow indicates the only band with activity detected in the dorsal muscle. m, low molecular
mass standards.
binnenwerk.indd 279 21-08-2006 12:35:01

staining in the ventral extracts, indicating that there was strong proteolytic activity, in addition
to the gelatinolytic for which the assay was designed. One possible explanation for these strong
gelatinolytic activities may be the presence-removal of inhibitors in the ventral muscle or in the
extracts. As reviewed by Snoek-van Beurden and Von den Hoff (2005), SDS dissociates the tissue
inhibitors of metalloproteinases (TIMPs) from the MMP’s, which may result in higher activity
registered by zymography than the real in vivo activities. In addition, refolding of MMP’s by
treatment with Triton X-100 only partially restores the original activity of the enzymes. Thus,
the natural activity of these enzymes cannot be accurately established by this method. The
presence of inhibitors in the tissue should vary according to the physiological status and higher
activities would correspond to lower levels of the natural inhibitors.
An alternative explanation may be that the gelatinolytic activities in the ventral muscles used
for this study may have been activated during the post mortem storage period due, among
other factors, to leakage of trypsin-like enzymes. Trypsin has been shown to activate animal
collagenase (Birkedal-Hansen and others 1976) and dissection of the abdominal area of these
fish showed weakening of the intestinal wall and possible leakage of the contents of pyloric
caeca which contain trypsin. Interestingly, Almy suggested the implication of trypsin as the
main responsible enzyme for belly bursting in herring already in 1926 (Almy 1926). Leakage of
trypsin-like enzymes would also explain the disappearance of most of the protein bands from
the total protein extract from the ventral muscle that were present in the extract from the dorsal
(Figure 2, compare lanes V and D). Trypsin itself, if contaminating the dissected muscle, may
have contributed to some of the gelatinolytic activity shown in Figure 2.
Blood had also considerable gelatinolytic activities, but since (1) the amount of activity in the
blood contained in capillaries in the muscle must be much less than the activity visualized here,
which corresponds to a sample of whole blood and (2) the main activity from blood seemed to
have similar electrophoretic mobility to that of one of the minor bands and different from that
of the three very intense bands from ventral muscle (Figure 2), we believe that it is possible to
exclude blood contamination as the major source for the strong gelatinolytic activities detected
in ventral muscle.
Gelatinolytic, collagenolytic and MMP-activities have been described in several species of fish
(Bracho and Haard 1995; Teruel and Simpson 1995; Kubota and others 1998a, b; Kubota and
others 2000; Matsui and others 2000; Saito and others 2000a, b; Kinoshita and others 2002;
Takeuchi and others 2002; Kubota and others 2003; Lødemel and Olsen 2003; Lødemel and
others 2004; Olsson and others 2006). It is possible that belly bursting in herring may be
due to high trypsin-like activity and the activation of gelatinolytic enzyme(s) in muscle by
these proteases. Lødemel and others (2004) have investigated gelatine degrading enzymes
from female gonad of Atlantic cod (Gadus morhua) activated by trypsin. They found that
this activation led to appearance of several new gelatine degrading enzyme forms with lower
molecular weight, which is to be expected in enzymes activated by the loss of inhibitory
peptides and is also illustrated by our results.
Conclusion
These are the first results in the work to map the relevance of several possible proteolytic
activities on the belly bursting of herring. Apparently low levels of pepsin activity were registered
in extracts from stomach. Results from gelatine zymography, on the other hand, indicate that
binnenwerk.indd 280 21-08-2006 12:35:01

gelatinolytic activities may be very relevant for the disruption of ventral muscle during the heavy
feeding period and current work is in progress to further characterize these activities, including
by affinity purification of the extracts and inhibitor studies. Thus, although the gelatinolytic
activities seen cannot be assigned to any specific enzyme(s) yet, it was shown for the first time
that these activities are very strong in the ventral muscle of spring herring compared to dorsal
muscle. As far as we are aware of, these preliminary results are the first report of gelatinolytic
activities in herring and their possible implications in belly bursting.
Acknowledgements
We are grateful to the crew of Libas and to Bergen fiskeindustri, for the help in the collection
of the fish used in this work and to the Norwegian Research Council for the financial support
to projects 135332/I10;146932/130 and 151648/120.
References
Almy LH. 1926. The role of the proteolytic enzymes in the decomposition of the herring. J Ann Chem Soc 48:2136-
2146.
Birkedal-Hansen H, Cobb CM, Taylor RE, Fullmer HM. 1976. Synthesis and release of procollagenase by cultured
fibroblasts. J Biol Chem 251(10):3162-3168.
Birkedal-Hansen H, Taylor RE. 1982. Detergent-activation of latent collagenase and resolution of its component
molecule. Biochem Biophys Res Commun 107(4):1173-1178.
Birkedal-Hansen H, Moore WGI, Bodden MK, Windsor LJ, Birkedal-Hansen B, DeCarlo A, Engler JA. 1993. Matrix
metalloproteinases: A review. Crit Rev Oral Biol M 4(2):197-250.
Bracho G, Haard N. 1995. Identification of two matrix metalloproteinases in the skeletal muscle of Pacific rockfish
(Sebastes sp.). J Food Biochem 19:299-319.
Bradford M. 1976. A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the
principle of protein-dye binding. Anal Biochem 72:248-254.
Díaz-López M, Moyano-López FJ, Alarcón-López FJ, Garcia-Carreño FL, Navarrete del Toro MA. 1998. Characterization of
fish acid proteases by substrate-gel electrophoresis. Comp Biochem Phys B 121:369-377.
Gildberg A (1982). Autolysis of fish tissue - general aspect, Institute of Fisheries, University of Tromsø, Norway. Ph.D
thesis
Heussen C, Dowdle EB. 1980. Electrophoretic analysis of plasminogen activators in polyacrylamide gels containing
sodium dodecyl sulfate and copolymerized substrates. Anal Biochem 102:196-202.
Hochstrasser DF, Patchornik A, Merril CR. 1988. Development of polyacrylamide gels that improve the separation of
proteins and their detection by silver staining. Anal Biochem 173(2):412-423.
Huss HH (1995). Quality and quality changes in fresh fish. FAO Fisheries Technical Paper 348:195.
Kinoshita M, Yabe T, Kubota M, Takeuchi K, Kubota S, Toyohara H, Sakaguchi M. 2002. cDNA cloning and characterization
of two gelatinases from Japanese flounder. Fish Sci 68:618-626.
Kubota M, Kinoshita M, Takeuchi K, Kubota S, Toyohara H, Sakaguchi M. 2003. Solubilization of type I collagen from
fish muscle connective tissue by matrix metalloproteinase-9 at chilled temperature. Fish Sci 69:1053-1059.
Kubota S, Kinoshita M, Toyohara H, Sakaguchi M. 1998a. Gelatinolytic activities in ayu muscle at spawning stage. Fish
Sci 64:1001-1002.
Kubota S, Toyohara H, Sakaguchi M. 1998b. Occurence of gelatinolytic activities in yellowtail tissues. Fish Sci 64:439-
442.
Kubota S, Kinoshita M, Yokoyama Y, Toyohara H, Sakaguchi M. 2000. Induction of gelatinolytic activities in ayu muscle
at the spawning stage. Fish Sci 66:574-578.
binnenwerk.indd 281 21-08-2006 12:35:01

Laemmli UK. 1970. Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature
227:680-685.
Lødemel JB, Olsen RL. 2003. Gelatinolytic activities in muscle of Atlantic cod (Gadus morhua), spotted wolffish
(Anarhichas minor) and Atlantic salmon (Salmo salar). J Sci Food Agric 83:1031-1036.
Lødemel JB, Mæhre HK, Winberg J-O, Olsen RL. 2004. Tissue distribution, inhibition and activation of gelatinolytic
activities in Atlantic cod (Gadus morhua). Comp Biochem Phys B 137:363-371.
Martinez A, Gildberg A. 1988. Autolytic degradation of abdominal tissue in anchovy (Engraulis encrasicholus). Int J
Food Sci Technol 23:185-194
Massova I, Kotra L, Fridman R, Mobashery S. 1998. Matrix metalloproteinases: structures, evolution, and diversification.
FASEB Journal 12:1075-1095.
Matsui H, Ogiwara K, Ohkura R, Yamashita M, Takahashi T. 2000. Expression of gelatinases A and B in the ovary of the
medaka fish Oryzias latipes. Eur J Biochem 267:4658-4667.
Olsson GB, Cooper M, Friis TJ, Olsen RL. 2006. Gelatinolytic activity in muscle of farmed and wild Atlantic cod (Gadus
morhua) related to muscle softening. In: Luten JB, Jacobsen C, Bekaert K, Saebo A, Oehlenschläger J, editors.
Seafood research from fish to dish: Quality, safety and processing of wild and farmed fish, Wageningen: Wageningen
Academic Publishers pp.161-171.
Saito M, Kunisaki N, Urano N, Kimura S. 2000a. Characterization of cDNA clone encoding the matrix metalloproteinase
2 from rainbow trout fibroblast. Fish Sci 66:334-342.
Saito M, Sato K, Kunisaki N, Kimura S. 2000b. Characterization of a rainbow trout matrix metalloproteinase capable of
degrading type I collagen. Eur J Biochem 267:6943-6950.
Snoek-van Beurden PAM, Von den Hoff JW. 2005. Zymographic techniques for the analysis of matrix metalloproteinases
and their inhibitors. BioTechniques 38:73-83.
Takeuchi K, Kubota S, Kinoshita M, Toyohara H, Sakaguchi M. 2002. Cloning and characterization of cDNA for carp
matrix metalloproteinase 9. Fish Sci 68:610-617.
Teruel S, Simpson B. 1995. Characterization of the collagenolytic enzyme fraction from winter flounder (Pseudopleuronectes
americanus). Comp Biochem Physiol 112B(1):131-136.
Woessner J. 1991. Matrix metalloproteinases and their inhibitors in connective tissue remodeling. FASEB Journal
5:2145-2154.
binnenwerk.indd 282 21-08-2006 12:35:01

Characterization of the quality of frozen sea products
commercialized in Portugal
Susana Gonçalves, Helena Lourenço, Claudia Afonso, Maria Fernanda Martins and Maria
Leonor Nunes
INIAP-IPIMAR, Av. Brasília, 1449-006 Lisbon, Portugal
Abstract
Total volatile basic nitrogen (TVB-N), trimethylamine (TMA) and peroxide value (PV) of about 300
samples of frozen sea products were evaluated. The samples were collected in several retailers in
Portugal since 2000. The lowest values of TVB-N were registered in tuna, octopus and mollusc
bivalve samples, while the highest concentrations were found in squids and shrimps (about 50
and 41 mg N/100 g, respectively). Conversely, the maximum levels of TMA were observed in
squids and scabbardfish (16.5 mgN/100 g and 9.4 mg N/100 g, respectively). Relatively to PV,
scabbardfish and tuna samples presented the highest levels (about 30 meq O2/kg).
Keywords: Total volatile basic nitrogen, trimethylamine, peroxide value, frozen sea products
Introduction
In a balanced diet fishery products are consistently considered as an essential food. However,
they are very perishable and even along the frozen distribution chain they can undergo several
changes if the preserving, packaging and retailing conditions are not adequate.
The shelf life of frozen fish depends on the product characteristics (raw materials and ingredients),
pre-freezing treatment, freezing process, packaging film and processes, and of course storage
conditions (Aubourg 1999; Schubring 1999). However, the quality deterioration and potential
hazards are usually aggravated or complicated by a fluctuating time-temperature environment
(e.g. freeze/thaw cycle) during storage.
In most of frozen fish products the main changes are related to rancidity, toughening (protein
denaturation), discoloration and desiccation (freezer burn) (Sikorski 1976; Bechmann 1998;
Tokur and others 2006). To follow the intensity of main changes several biochemical methods
are used alone or together with sensory analysis. Dimethylamine, formaldehyde, thiobarbituric
acid-related substances and individual nucleotides are very extensively mentioned in literature
(Gallardo and others 1991; Bechmann 1998; Ruiz-Capillas and Moral 2001; Jacober and Rand
1982; Özogul and others 2004). However, some inspection services and fishing industry sector
still request information on total volatile bases (TVB-N), trimethylamine (TMA) and peroxide
value (PV) to characterise and compare the quality of frozen fish products.
Taking into consideration that the per capita consumption of seafood (about 60 kg) in Portugal
is one of the highest within European countries and that more than 40% of the fish products is
frozen, the aim of this work was to characterise the quality of frozen seafood commercialised
in Portugal. For this characterisation three different categories were considered –fish, molluscs
and crustaceans- by using the three quality indicators mentioned above: TVB-N, TMA and PV.
binnenwerk.indd 283 21-08-2006 12:35:01

Susana Gonçalves, Helena Lourenço, Claudia Afonso, Maria Fernanda Martins and Maria Leonor Nunes
Material
The frozen samples (Table 1) used in this study, about 300, were collected in Portugal directly
from several commercial retailers, between 2000 and 2004. All samples were considered to have
commercial quality.
Collected frozen samples were immediately transported in isotermic conditions to the laboratory
where they were held at -20 ºC. Each sample was composed by 5 or 10 specimens or portions.
According to their availability, the samples were constituted by whole fish units or portions,
eviscerated cephalopods, shucked mollusc bivalve and crustacean meat. After thawing, muscle
was removed, homogenized in a food blender with stainless steel cutters and stored in plastic
bags at 5 ºC for some hours prior to analysis.
Methods
For TVB-N and TMA, sample muscles were homogenized with 5% trichloroacetic acid (TCA),
ratio 1:2 (w/v) for 1 min in a blender (Polytrom 3100, Kinematica, Littau, Switzerland). The
homogenates were filtered through a filter paper (Whatman no.1) and filtrates were analysed in
duplicate for TMA (previous addition of formaldehyde 37% 1:1 v/v) and TVB-N, by the modified
Conway microdiffusion method as described by Cobb and others (1973). Results were expressed
as mg N/100 g of wet sample.
The PV levels were determined using Bligh and Dyer (1959) fat extraction method, followed by
titration (EEC Regulation 183/93). Results were expressed as meq oxygen/kg lipid.
TVB-N levels in fish samples did never exceed 35 mg N/100 g (Figure 1). The highest mean
concentration was obtained in tuna (21.2 mg N/100 g) and the lowest was found in monkfish
(8.0 mg N/100 g). Several authors found similar results in fresh fish like Ruíz-Capillas and Moral
Table 1. Studied products and respective number of analyzed samples.
Fish Molluscs and crustacean
Cod – Gadus morhua (N=8) Shrimp – Penaeus spp. (N=37)

Hake – Merluccius merluccius (N=56) Ox crab – Cancer pagurus (N=3)
Horse Mackerel – Trachurus trachurus (N=8) Squid – Loligo vulgaris (N=10)
Meagre – Argyrosomus regius (N=11) Octopus – Octopus vulgaris (N=17)
Monkfish – Lophius piscatorius (N=3) Shucked mussel – Mytilus edulis (N=6)
Redfish – Sebastes marinus (N=5) Shucked cockle – Cerastoderma edule (N=4)
Sardine – Sardina pilchardus (N=27) Shucked clam – Spisula solida (N=19)
Scabbardfish – Aphanopus carbo (N=13) Cocktail (N=7)
Tuna – Thunnus spp. (N=19) Clam – Spisula solida (N=9)
Mixture of different shellfish species (mussel, clam, shrimp, hake) and analogue of crab legs.
binnenwerk.indd 284 21-08-2006 12:35:02

Characterization of the quality of frozen sea products commercialized in Portugal
35
TVB-N (mgN/100g)
30
25
20
15
10
5
0 Cod (N=8)
Hake (N=56)
Horse Mackerel (N=8)
Meagre (N=11)
Monkfish (N=3)
Redfish (N=5)
Sardine (N=27)
Scabbardfish (N=13)
Tuna (N=19)
Figure 1. Value range for TVB-N in the studied fish species. The dot represents the mean value.
(2001) in hake (Merluccius merluccius) captured in the Galician platform, Özogul and others
(2004) in sardines (Sardina pilchardus) obtained from Mallaig in Scotland and Castro and others
(2006) for European sea bass. Regarding tuna samples, higher values were observed by Ruíz-
Capillas and Moral (2005) for whole bigeye tuna (Thunnus obesus). Nevertheless, the highest
value was found in one hake sample (32.7 mg N/100 g) and did not exceed the limit value of
35 mg N/100 g indicated by EU (Directive 95/149/EEC).
In molluscs and crustaceans (Figure 2), the highest mean concentration of TVB-N was obtained
in ox crab (24.1 mg N/100 g), and the maximum value of 49.8 mg N/100 g was observed in one
squid sample. Generally, the values found in this work are in accordance with the ones reported
by other authors, namely Matsumoto and others (1990), López-Caballero and others (2002) for
deep water pink shrimp (Parapenaeus longirostris), Ruíz-Capillas and others (2002) for some
cephalopods and Manousaridis and others (2005) for shucked mussels.
The TMA mean values in fish samples (Figure 3) ranged between 1.0 mg N/100 g (horse
mackerel) and 3.5 mg N/100 g (scabbardfish). The highest content (9.4 mg N/100 g) found
in scabbardfish is considered to be important according to Connel (1975). However, it did
not exceed the limit value indicated by EU of 12 mg N/100 g (Directive 91/493/EEC). Similar
values had also been reported by other authors (Ruíz-Capillas and others 2001; Özogul and
others 2004).
The lowest levels, below 1.5 mg N/100 g, were found in shucked bivalve samples (Figure 4).
Higher values were observed by Manousaridis and others (2005) in similar material. Some
octopus and squid samples presented very high maximum concentrations, respectively 8.5 mg
N/100 g and 16.5 mg N/100 g, indicating probably a loss of quality before frozen processing.
Regarding the PV (Figure 5), all cephalopod samples and most shrimp samples showed values
below the detection limit (0.4 meq O2/kg of lipid). The highest ones were registered in a few
scabbardfish and tuna samples, respectively 30 and 26 meq O2/kg of lipid.
binnenwerk.indd 285 21-08-2006 12:35:02

60
TVB-N (mgN/100g)
50
40
30
20
10
S. mussel
S. cockle
Ox Crab
Octopus
S. clam
Shrimp
(N=10)
(N=37)
(N=9)
(N=19)
Cocktail
(N=17)
Clam
Squid
(N=3)
(N=7)
(N=4)
(N=6)
Figure 2. Value range for TVB-N in the studied mollusc and crustacean species. The dot represents the mean
value.
10
TMA-N (mgN/100g)
9
8
7
6
5
4
3
2
1
0
Cod (N=8)
Hake (N=56)
Horse Mackerel (N=8)
Meagre (N=11)
Redfish (N=5)
Sardine (N=27)
Scabbardfish (N=13)
Tuna (N=19)
Figure 3. Value range for TMA-N in the studied fish species. The dot represents the mean value.
Considering the acceptable limit levels proposed by Directive 95/149/EEC and Connel (1975),
the results indicated a total of 15 products with values above those limits (Table 2). They
were mainly found in the mollusc and crustacean group for the TVB-N parameter: 7 shrimp
samples and 1 squid sample exceeded the TVB-N limit. Regarding TMA levels, values above the
suggested limit were found in scabbardfish (3 samples) and cephalopods (2 samples). The PV
analyses detected 1 tuna sample and 1 scabbardfish sample with values above the acceptable
limit. However, the results mentioned above constitute only 3.3% of the total sample analysis
performed within this study.
binnenwerk.indd 286 21-08-2006 12:35:03

Characterization of the quality of frozen sea products commercialized in Portugal
18
16
TMA-N (mgN/100g)
14
12
10
8
6
4
2
0
S. mussel
S. cockle
Ox Crab
Octopus
S. clam
Shrimp
(N=10)
(N=37)
(N=9)
(N=19)
Cocktail
(N=17)
Clam
Squid
(N=3)
(N=7)
(N=4)
(N=6)
Figure 4. Value range for TMA-N in the studied mollusc and crustacean species. The dot represents the mean
value.
35.0
30.0
PV (meq O2/kg)
25.0
20.0
15.0
10.0
5.0
0.0
Cephalopods Hake Scabbardfish Shrimp Tuna
(N=8) (N=33) (N=3) (N=7) (N=8)
Figure 5. Value range for PV in the studied species. The dot represents the mean value.
Table 2. Number of samples with levels above the acceptable limit and respective percentage of total samples
in each group/parameter.
Fish Mollusc and crustacean Total
Samples % Samples % Samples %
TVB-N (>35 mgN/100g) 0 0.0 8 7.3 8 3.1

TMA-N (>5 mgN/100g) 3 3.9 2 3.5 5 3.7
PV (>20 meqO2/kg) 2 4.5 0 0.0 2 3.4
Total 5 1.9 10 5.5 15 3.3
binnenwerk.indd 287 21-08-2006 12:35:03

Conclusions
As a final consideration, it can be said that the quality of frozen sea products commercialized in
Portugal is acceptable, in spite of some samples presenting TVB-N, TMA and PV levels slightly
above the recommended maximum level (about 3.3% of the samples).
References
Aubourg SP. 1999. Lipid damage detection during the frozen storage of an underutilized fish species. Food Res Int
32:497-502.
Bechmann IE. 1998. Comparison of the formaldehyde content found in boiled and raw mince of frozen saithe using
different analytical methods. Lebensm Wiss Technol 31:449-453.
Bligh E, Dyer W. 1959. A rapid method of total lipid extraction and purification. Can J Biochem Physiol 37:911-917.
Castro P, Padrón JCP, Cansino MJC, Velázquez ES, Larriva RM. 2006. Total volatile base nitrogen and its use to assess
freshness in European sea bass stored in ice. Food Control 17(4):245-248.
Cobb BF III, Alaniz I, Thompson JrC. 1973. Biochemical and microbial studies on shrimp: volatile nitrogen and amino
nitrogen analysis. J Food Sci 38:431-436.
Connel JJ. 1975. Control of fish quality. Surrey: Fishing News (Books) Ltd. 179p.
EU. 1991. Commission Regulation (EEC) No 493/91 of 22 July 1991 laying down the health conditions for the production
and the placing on the market of fishery products. Official Journal L 268, 24/09/1991. p 0015–0034.
EU. 1993. Commission Regulation (EEC) No 183/93 of 29 January 1993 amending Regulation (EEC) No 2568/91 on the
characteristics of olive oil and olive-residue oil and on the relevant methods of analysis. Official Journal L 022,
30/01/1993. p 0058–0068.
EU. 1995. 95/149/EC: Commission Decision of 8 March 1995 fixing the Total volatile bases nitrogen (TVB-N) limit
values for certain categories of fishery products and specifying the analysis methods to be used. Official Journal
L 097, 29/04/1995. p 0084‑0087.
Gallardo JM, Pérez-Martin R, Sotelo CG, Aubourg S, Banga JR. 1991. Evolución de las aminas volátiles en dos tipos de
merluza (Merluccius australis, Merluccius capensis) y en rosada (Xiphiurus capensis) durante el almacenamiento a
18 ºC. Rev Agroquím Tecnol Aliment 31:103-110.
López-Caballero ME, Gonçalves A, Nunes ML. 2002. Effect of CO2/O2 – containing modified atmospheres on packed
deepwater pink shrimp (Parapenaeus longirostris). Eur Food Res Technol 214:192-197.
Manousaridis G, Nerantzaki A, Paleologos EK, Tsiotsias A, Savvaidis IN, Kontominas MG. 2005. Effect of ozone on
microbial, chemical and sensory attributes of shucked mussels. Food Microbiol 22:1-9.
Matsumoto M, Yamanaka H. 1990. Post-mortem biochemical changes in the muscle of kuruma prawn during storage and
evaluation oh the freshness. Nippon Suisan Gakkaishi 56:1145‑1149.
Özogul F, Polat A, Özogul Y. 2004. The effects of modified atmosphere packaging and vacuum packaging on chemical,
sensory and microbiological changes of sardines (Sardina pilchardus). Food Chem 85:49-57.
Ruiz-Capillas C, Moral A. 2001. Correlation between biochemical and sensory quality indices in hake stored in ice. Food
Research International 34:441-447.
Ruiz-Capillas C, Moral A. 2005. Sensory and biochemical aspects of quality of whole bigeye tuna (Thunnus obesus)
during bulk storage in controlled atmospheres. Food Chem 89:347-354.
Ruiz-Capillas C, Moral A, Morales J, Montero P. 2002. Characterisation of non-protein nitrogen in the Cephalopods
volador (Illex coindetii), pota (Todaropsis eblanae) and octopus (Eledone cirrhosa). Food Chem 76:165-172.
Schubring R. 1999. DSC studies on deep frozen fishery products. Thermochim Acta 337:89-95.
Sikorski Z, Olley J, Kostuch S. 1976. Protein changes in frozen fish. CRC Crit Rev Food Sci Nutr 8:97-125.
Tokur B, Ozkütük S, Atici E, Ozyurt G, Ozyurt CE. 2006. Chemical and sensory quality changes of fish fingers, made from
mirror carp (Cyprinus carpio L., 1758), during frozen storage (-18 ºC). Food Chem 99:335-341.
binnenwerk.indd 288 21-08-2006 12:35:03

Development of a Quality Index Method scheme to evaluate
freshness of tub gurnard (Chelidonichthys lucernus)
Karen Bekaert
Belgium Institute for Agricultural and Fisheries Research, Unit Animal Sciences - Fisheries,
Ankerstraat 1, B-8400, Oostende, Belgium
Abstract
This paper describes the development of a Quality Index Method (QIM) scheme for tub gurnard
(Chelidonichthys lucernus) and its evaluation in a shelf life study. QIM is one of the most
interesting sensory methods at the moment for assessment of fish freshness. It is based on a
scoring system for the evaluation of characteristic changes that occur during fish spoilage. As
a result of this study, a validated QIM scheme is proposed for this species. The appearance of
skin and slime, clarity and shape of the eyes, odour, colour and slime of the gills, texture and
appearance of the fins are the parameters included which gave a total of 22 points. The global
correlation coefficient was r=0.9861. The end of shelf life was reached at day 18-19 of storage.
Keywords: sensory evaluation, Quality Index Method, tub gurnard, fish freshness, shelf life
Introduction
The importance of tub gurnard caught by Belgian fishing vessels has been increasing through
the last decade. As tub gurnard is one of the few fish species not subject to total allowable
catches and quotas, the catches of tub gurnard more than doubled from 231 tons in 1996
to 528 tons in 2004. The yearly landings of fish, shellfish and molluscs of Belgian fisherman
increased to 20 000 tons. This does not represent a massive amount, but the Belgian fisheries
are characterized by the very varied assortment landed at the auctions. The Belgian authorities
undertook actions (Anonymous 1994, Anonymous 2002) to promote less-consumed fish species
such as tub gurnard to diversify the consumption pattern of customers and to get a better
equilibrium between offer and demand. This resulted in an increase of the landings and the
consumption of gurnard.
At the moment, there is a strong demand for rapid and easy-to-use methods to evaluate the
freshness of fish that can be easily implemented in the fish industry. As characteristic changes
occur in the appearance, odour, texture, flavour during deterioration of fish, sensory methods
can be used to determine the freshness of fish. Sensory methods utilize one or more of the five
senses to judge some aspect of quality. They have the advantage to be fast, simple, reliable
and non destructive on raw fish (Olafsdottir and others 1997).
The recommended sensory method for the evaluation of whole fish in the industry and the
inspection authorities in Europe is the EU-scheme. In this scheme, 3 grades of freshness are
determined: E for extra quality, A for average quality, and B for bad quality. Below B is the level
where fish is unfit for human consumption (EU 1996). However, the validity of this scheme has
been questioned because it does not take into account differences between species as only
binnenwerk.indd 289 21-08-2006 12:35:03

Karen Bekaert
general parameters are used (Luten and Martinsdottir 1997). Today, one of the most interesting
alternative sensory methods is the Quality Index Method (QIM). This sensory method was first
developed by Bremner and other researchers (Bremner 1985, Bremner and others 1987) and
has received much attention since (Baixas-Nogueras and others 2003; Herrero and others 2003;
Huidobro and others 2000; Hyldig and Nielsen 1997; Larsen and others 1992; Martinsdottir 2000;
Martinsdottir and others 2001; Nielsen and Jessen 1997; Sveinsdottir and others 2003). It is
based upon a points scoring system for the evaluation of the key parameters in deterioration
of individual fish species. The set of characteristic attributes of each parameter is assigned a
range of demerit points (≤ 3). The QIM is so-designed that the sum of the demerit points (the
quality index, QI) given to each of the evaluated parameters correlates linearly with storage
time in ice. This permits to predict the remaining storage time and provides thus an alternative
to the sensory methods used before.
The aim of this study was to develop a Quality Index Method (QIM) scheme for raw tub gurnard
(Chelidonichthys lucernus). Furthermore, the maximum storage time in ice was determined by
the sensory evaluation of cooked samples of tub gurnard. To validate the proposed scheme,
complementary sensory and chemical parameters (TVB-N and pH) were monitored during ice
storage.
Fish samples
For the design of the preliminary QIM scheme, fresh tub gurnard was obtained from a cruise
with the Belgian research vessel to the North Sea. Immediately after the catch, the fish were
washed and stored on board in flake ice. After landing, they were transported to the laboratory
where they were stored in ice in a refrigerator (2 ± 1 °C) in self-draining boxes for 22 days. Ice
was added to the boxes as required. Second batch of tub gurnard was purchased from a Belgian
fishing vessel for the evaluation of the QIM scheme, the determination of shelf life and for
the chemical analyses. The samples had been caught the night before and were immediately
transported and stored at the lab in the same manner as the first batch after landing.
Development of QIM
The procedure for the development of a Quality Index scheme described by Martinsdottir (2000)
was followed.
One expert from the Sea Fisheries Department (SFD) participated in the preliminary examination
of tub gurnard. The changes in appearance, odour and texture were observed after 0-22 days
of storage in ice and were listed in a preliminary QIM scheme. A total of 30 fish were used for
the development of the initial scheme.
For the development of the final QIM scheme, 4 assessors were selected from among the staff
of the SFD and from the Flemish fish auctions (Zeebrugge and Oostende), on the basis of their
ability to identify odours and familiarity with sensory analysis. The aim to select assessors from
the auctions was to know their opinion on the selected parameters and to be sure no important
quality parameter would be forgotten. Three sessions of 2 hours each were used for explaining
and adapting the initial scheme. Tub gurnards between 1 and 22 days in ice were assessed.
All suggestions of improvement were included in the final scheme. The ultimate version of the
scheme was finalized by the panel leader and used in the shelf life study.
binnenwerk.indd 290 21-08-2006 12:35:04

Development of a quality index method scheme to evaluate freshness of tub gurnard
The shelf life study experiment was carried out with a trained sensory panel (n=8) from SFD
and the fish auctions. Training sessions were given during 4 full days, spread over 3 weeks. On
the third and fourth day of training, panellists, unaware of the fish storage time, determined
freshness with QIM on fish between 2 and 22 days old, originating from different batches. Five
fish per storage day were assessed and their QI was averaged to diminish the influence of the
biological variation between individual fish.
A linear regression analysis of sensory changes in contrast to time of ice storage was performed
with the data obtained (GraphPad Prism).
Cooked tub gurnard

The sensory assessment score sheet for cooked white fish was used following the Torry Scheme
(Shewan and others 1953) during the shelf life study. In this scheme, the flavour and odour
of cooked fish is evaluated, using a gradation from 10 (for fresh fish) to 3 (for spoiled fish).
The rejection point was set at < 5.5, which corresponds to the detection of off-odours and
off-flavours in the fish. The portions of tub gurnard were filleted and skinned by hand and then
cooked in a glass jar in a microwave at 600 W during 2 minutes. They were immediately served
to the sensory panel participating in the shelf life study. A linear regression analysis between
the average Torry score for each day in contrast to storage time in ice was performed with the
data obtained (GraphPad Prism).
Chemical analysis
TVB-N
Total volatile basic nitrogen determination were made using the direct distillation method
according to Antonacopoulos (1968). This method is based on distillation and titration of the
volatile basic nitrogen.
pH
The pH was measured from the samples using a pH meter (Model 744, Metrohm, Switzerland),
by inserting a needle glass electrode (Metrohm, Switzerland) directly into the fish.
Development of QIM
In the preliminary observation, the changes in appearance of skin, colour of mucus on the skin,
texture of flesh, colour, odour and mucus of the gills, appearance of the fins, clarity of cornea
and pupil, form of the eyes were described during 22 days. The fish was cut open to describe
the colour of the blood, of the cut-surface and the appearance and odour of the entrails. These
data were utilized for the development of the preliminary QIM scheme, which gave a QI score
of maximum 42 points.
This scheme was presented to the sensory panel (n=4) from the auctions and SFD. It was
decided to omit all the parameters for which the fish had to be opened, since the assessment
was destructive and increased significantly the evaluation speed. The parameters “clarity of
cornea and pupil” were grouped to become “clarity of the eye”. Scores were omitted or added
and descriptions of several parameters were adapted using more precise and/or common words
to facilitate comprehension. This resulted in a QI scheme for tub gurnard with a maximum score
of 22 (Table 1), describing 9 sensory parameters for appearance of skin, mucus on skin, form and
clarity of the eyes, odour, colour and mucus of the gills, texture and appearance of the fins.
binnenwerk.indd 291 21-08-2006 12:35:04

Karen Bekaert
Table 1. Final Quality Index scheme developed for raw tub gurnard.
Attribute Description QI
Skin appearance Fresh, bright, red-orange colours, metallic/bronze sideline, cream-white belly 0
Less bright, less red-orange, some greyish discolouration 1
Dull, greyish, putrefaction colour 2
Skin mucus Clear, not clotted 0
Milky, slighted clotted 1
Yellow, brown, clotted 2
Eye form Convex 0
Flat, slightly sunken 1
Sunken 2
Eye clarity Clear cornea; clear, black pupil; clear yellow border around pupil 0
Cornea slightly clouded, dull pupil, yellow border less defined 1
Milky cornea, milky pupil 2
Grey or red cornea, grey pupil 3
Gill odour Fresh, seaweed, metallic, grass 0
Neutral, metallic, slightly musty 1
Musty 2
Sour, rotten, faecal 3
Gill colour Bright, red 0
Less bright, red 1
Discoloured, yellow spots 2
Yellow/ brown discolouration 3
Gill mucus No mucus or clear mucus, lamellae slight sticky 0
Milky mucus, lamellae sticky 1
Yellow-brown mucus, lamellae sticky 2
Texture In rigor 0
Firm, elastic 1
Less firm, less elastic 2
Soft 3
Fins Bright colours, colours well differentiated 0
Less bright, colours less differentiated 1
Grey-brown, colours not differentiated 2
Total score 0- 22
Before using this scheme in the shelf life study, it was verified that all the selected parameters
increased with storage time in ice (Table 2). The average scores for the individual attributes
all correlated significantly with storage time in ice and the correlation coefficient for each
individual parameter was high for most attributes (> 0.85). However, the correlation coefficient
of the QI scheme based on different parameters gave a better relation with storage time
in ice (0.99). As a result the prediction about storage time for different parameters would
be more precise than a QI scheme based on a single parameter and a sufficient number of
attributes is needed to give a total possible score of reasonable magnitude. In addition, as
more attributes are evaluated minor differences in judgements does not merely influence the
total score (Bremner 1987). It was therefore considered that an index of a maximum of 22
binnenwerk.indd 292 21-08-2006 12:35:04

Table 2. Average score for each quality attribute assessed with the QIM scheme for tub gurnard stored in ice
and the correlation to days in ice.
Days in ice 1 2 5 7 8 9 10 13 17 22 Correlation
Skin appearance 0.13 0.00 0.56 1.00 1.22 1.07 1.56 1.44 2.00 2.00 0.94
Skin mucus 0.00 0.33 0.67 0.67 0.67 0.67 0.67 1.67 2.00 2.00 0.94
Eye form 0.00 0.00 0.89 1.11 1.33 1.00 1.11 1.67 1.80 2.00 0.93
Eye clarity 0.00 0.00 0.89 0.89 1.00 0.93 2.00 1.56 3.00 3.00 0.95
Gill odour 0.07 0.00 0.78 0.78 0.89 1.27 1.56 2.33 3.00 3.00 0.97
Gill colour 0.20 1.00 0.44 0.78 0.78 1.20 1.67 1.78 2.20 2.78 0.94
Gill mucus 0.00 0.33 0.56 0.78 0.67 1.00 1.78 1.67 2.00 2.00 0.92
Texture 0.40 0.67 1.22 1.33 1.00 1.33 1.33 1.78 2.00 2.67 0.97
Fins 0.07 0.00 0.56 0.89 0.78 1.27 0.89 1.11 1.40 1.78 0.94
Total 0.87 2.33 6.22 7.78 8.00 9.20 12.11 14.22 19.40 21.23 0.99
demerit points would be suitable for tub gurnard. The scores for the mucus of the gills seemed
to correlate least with storage time in ice. This score remained constant from day 5 to day 10.
The scores for the skin, the eyes, the gills (odour and colour), texture and fins increased rather
constantly throughout the storage time, even though the scores showed some fluctuation with
storage time in ice. The texture and the odour of the gills seemed to give the best correlation.
At the end of storage time, most of the attributes reached the maximum score.
Shelf life study

A global linear relationship (y =1.0015x + 0.3286) between the average QI for each storage day
and storage time on ice was determined (Figure 1) with the results of the shelf life study. The
regression model gave a high correlation (r = 0.9856) and was highly significant (P < 0.0001),
which means that the QI scheme is an excellent instrument for freshness measurement of tub
gurnard.
25
y = 1.0015x + 0.3286
r = 0.9861
20
QI score
15
10
0
0 5 10 15 20 25
Days in ice
Figure 1. Average QI of each storage day analysed versus days in ice.
binnenwerk.indd 293 21-08-2006 12:35:05

Karen Bekaert
Figure 2 shows that the individual panellists participating in the QIM evaluation of tub gurnard
performed differently during the shelf life study. Panellists 1 and 8 gave consistently lower
scores throughout storage time while panellists 4 and 7 scored highest. The variation between
the scores the panellists gave was highest in the middle of storage time (from day 6 to day 18).
This indicates that panellists scored more unanimous in their evaluation at the beginning and
the end of storage time. This might be because in the early stage of storage, the characteristics
of the evaluated parameters are very prominent and at the end of storage time, changes become
more pronounced than in the middle of storage time.
The Torry scheme for cooked white fish was used to determine the end of shelf life of tub
gurnard (Figure 3). Odour and taste in cooked gurnard decreased gradually with storage time
(r = 0.9595, p < 0.0001). The rejection point for the cooked fillets was established when the
Torry score was lower than 5. At day 18-19 of storage, the point at which the cooked fillets
were judged unacceptable was reached, corresponding to a quality index of 18. The flavour of
the tub gurnard was then described as slightly sour and presence of traces of off-flavours, the
odour was described like “milk jug odours and boiled potato”. Thereafter, the tub gurnard was
no longer fit for human consumption.
Furthermore, chemical parameters (TVB-N and pH) were examined to verify whether they were
consistent with the suggested sensory rejection value and if they supported the suitability of
the proposed QIM scheme. Table 3 shows the results of the evolution of chemical parameters
during ice storage. TVB-N did not increase during the first 15 days of storage and reached 30
mg N/100 after 18 days of storage, which is close to the limit of acceptability (30-35 mg N/100
g) established by the European Union regulation (EU 1995). Even though pH values are proved
to be a poor freshness indicator due to wide variation between species, individual specimens
and biological variation, a pH-value of 7 can in most tests be considered as an indication of
spoilage (Oehlenschlager 1992). The measured pH values varied widely at the beginning of
storage time, and in our study, a value close to 7 was reached at day 18 of storage. The pH
value could be correlated with the number of days in ice over the whole storage period (pH =
6.20 + 0.041 days, r = 0.9165 ). The chemical rejection values seemed to confirm the end of
shelf life determined with the sensory assessment of cooked fish.
25 1
2
20
3
4
QI score
15
5
6
10
7
8
5
0
0 5 10 15 20 25
Days in ice
Figure 2. Average QI given by each QIM panellist in the shelf life study for each day of analysis versus days
in ice.
binnenwerk.indd 294 21-08-2006 12:35:05

Quality Index Torry score

25 12 QI
Torry
20 10
trendline Torry
8 trendline QI
15
6
10 rejection threshold
4
5 2
0 0
0 5 10 15 20 25
Days in ice
Figure 3. Changes in QI and Torry Score during ice storage of tub gurnard.
Table 3. Results of the chemical analysis and sensory evaluation during ice storage of tub gurnard.
Days in ice QI score Torry score TVB (mg N/100 g) pH
2 1.2 9.7 14.4 6.25

5 6.2 8.8 12.0 6.49
6 5.9 8.7 16.0 6.62
7 6.2 7.7 12.2 6.64
8 8.4 7.5 14.3 6.41
9 10.4 6.7 15.3 6.47
11 12.1 6.5 16.0 6.47
15 17.0 6.0 16.2 6.75
18 18.2 5.6 30.6 6.92
20 18.8 5.0 36.9 7.09
21 21.2 4.2 41.3 7.17
Conclusion
As a result of this study, a validated QIM scheme is proposed for tub gurnard. The appearance
of skin and slime, clarity and shape of the eyes, odour, colour and slime of the gills, texture
and appearance of the fins are the parameters included, which gave a total Quality Index of 22
points. The total scores for quality attributes gave a linear relationship with storage time in ice.
The global correlation coefficient was r=0.9856. This means that the method offers a reliable
procedure for freshness evaluation of tub gurnard. The maximum storage time was determined
by sensory analysis of cooked fish and was reached after 18 days of storage in ice, corresponding
to a theoretical quality index of 18. Consequently, the method provides information about the
remaining shelf life of the fish in ice. Furthermore, the end of shelf life determined by the
sensory assessment of cooked fish was in accordance with the suggested spoilage indicators
of the chemical parameters.
binnenwerk.indd 295 21-08-2006 12:35:06

Karen Bekaert
References
Anonymous. 1994. Activiteitenverslag VLAM, Vlaams Promotiecentrum voor Agro- en Visserijmarketing, Brussel.
Anonymous. 2002. Activiteitenverslag VLAM, Vlaams Promotiecentrum voor Agro- en Visserijmarketing, Brussel.
Antonacopoulos N. 1968. Bestimmung des flüchtigen Basenstickstoffs. In: Acker I. Handbuch der Lebensmittelchemie.
Vol. III/2. Berlin: Springer Verlag. p 1493-1494.
Baixas-Nogueras S, Bover-Cid S, Veciana-Nogués T, Nunes, ML, Vidal-Carou MC. 2003. Development of a Quality Index
Method to evaluate freshness in Mediterranean hake (Merluccius merluccius). J Food Sci 68(3):1067-1071.
Bremner HA. 1985. A convenient easy-to-use system for estimating the quality of chilled seafoods. In: Scott DN, Summers
G, editors. Fish Processing bulletin, proceedings of the fish processing conference ’85. Wellington: DSIR. p 59-70.
Bremner HA, Olley J, Vail AMA. 1987. Estimating time-temperature effects by a rapid systematic sensory method. In:
Kramer DE, Liston J, editors. Seafood Quality Determination. Amsterdam: Elsevier Science Publishers. p 413-434.
EU. 1995. 95/149/EC: Commission Decision of 8 March 1995 fixing the total volatile basic nitrogen (TVB-N) limit values
for certain categories of fishery products and specifying the analysis methods to be used. Official Journal L097,
29/04/1995, p 0084-0087.
EU. 1996. Council Regulation (EC) No 2406/96 of 26 November 1996 laying down common marketing standards for
certain fishery products. Official Journal L 334, 23/12/1996, p1.
Herrero AM, Huidobro A, Careche M. 2003. Development of a quality index method for frozen hake (M. capensis and M.
paradoxus). J Food Sci 68(3):1086-1092.
Huidobro A, Pastor A, Tejada, M. 2000. Quality index method development for raw gilthead seabream (Sparus aurata).
J Food Sci 65(7):1202-1205.
Hyldig G, Nielsen J. 1997. A rapid sensory method for quality management. In: Olafsdottir G, Luten J, Dalgaard P,
Careche M, Verrez-Bagnis V, Martinsdottir E, Heia K, editors. Methods to determine the freshness of fish in research
and industry, Proceedings of the Final Meeting of the Concerted Action “Evaluation of Fish Freshness”, FAIR
Programme of the EU. Paris: Institut International du froid. p 297-305.
Larsen L, Heldbo J, Jespersen CM and Nielsen J. 1992. Development of a method for quality assessment of fish for
human consumption based on sensory evaluation. In: Huss HH, Jakobsen M, Liston J, editors. Quality assurance
in the fish industry. Amsterdam: Elsevier Science Publishers. p 351-358.
Luten J, Martinsdottir E. 1997. QIM: a European tool for fish freshness evaluation in the fishery chain. In: Olafsdottir
G, Luten J, Dalgaard P, Careche M, Verrez-Bagnis V, Martinsdottir E, Heia K, editors. Methods to determine the
freshness of fish in research and industry, Proceedings of the Final Meeting of the Concerted Action “Evaluation of
Fish Freshness”, Fair Programme of the EU. Paris: Institut International du Froid. p 287-296.
Martinsdottir E. 2000. Quality management of stored fish. In: Bremner HA, editor. Safety and quality issues in fish
processing. Cambridge: Woodhead Publishing Ltd. p 360-378.
Martinsdottir E, Sveinsdottir K, Luten JB, Schelvis-Smit R, Hyldig G. 2001. Reference manual for the fish sector; sensory
evaluation of fish freshness. QIM Eurofish. IJmuiden: QIM Eurofish. 49 p.
Nielsen J, Jessen K. 1997. New developments in sensory analysis for fish and fishery products. In: Luten JB, Børresen
T, Oehlenschläger J, editors. Seafood from producer to consumer, integrated approach to quality. Amsterdam:
Elsevier Science Publishers. p 537-547.
Olafsdottir G, Martinsdottir E, Oehlenschläger J, Dalgaard P, Jensen B, Undeland I, Mackie IM, Henehan G, Nielsen J,
Nilsen H. 1997. Methods to evaluate fish freshness in research and industry, Trends Food Sci Technol 8:258-265.
Oehlenschlager J. 1992. Evaluation of some well established and some underrated indices for the determination of
freshness and/or spoilage of ice stored wet fish. In: Huss HH, Jakobsen M, Liston J, editors. Quality assurance in
the fish industry. Amsterdam: Elsevier Science Publishers. p 339-350.
Shewan JM, Macintosh RG, Tucker CG and Ehrenberg ASC. 1953. The development of a numerical scoring system for the
sensory assessment of the spoilage of wet white fish stored in ice. J Sci Food Agric (4):283-298.
Sveinsdottir K, Hyldig G, Martinsdottir E, Jorgensen B, Kristbergsson K. 2003. Quality index method (QIM) developed
for farmed Atlantic Salmon (Salmo salar). Food Qual Prefer 14:237-245.
binnenwerk.indd 296 21-08-2006 12:35:06

Spoilage of herring (Clupea harengus) under chilled conditions
and offal under ambient and chilled conditions when treated with
commercial preservative or additive products
Michael Gallagher1, Marianne Green1 and Frank Trearty2
1Bord Iascaigh Mhara, Irish Sea Fisheries Board, PO Box no. 12, Crofton Road, Dun Laghaire, Co.
Dublin, Ireland
2Aqualab, Donegal Road, Killybegs, Co. Donegal, Ireland
Abstract
Four commercially available preservatives; acetic acid, Citrox™, Fishform and XyRex were
tested to ascertain their impacts at reducing spoilage on whole fresh herring (Clupea harengus)
stored for up to seven days in a chilled environment and herring offal stored both in chilled
and ambient environments for up to seven days. Samples were taken on a daily basis to assess
spoilage rates. Temporal trends in levels of TVC and TVB-N indicated that each of the products
generally had an impact at reducing levels relative to the controls. The impacts were particularly
evident for both Fishform and acetic acid up to day seven in the chilled environment. In
addition acetic acid and Fishform also had the largest impact at maintaining lower TVB-N levels
in the offal under both temperature regimes. With regard to biogenic amine levels the addition
of products had an affect relative to the control in both environments, and acetic acid appeared
to exert the largest control in this regard for both whole herring and offal.
Keywords: spoilage, herring, processing aids, quality, fishmeal
Introduction
Humans utilize over 1000 species of fish and shellfish, as a highly nutritious and readily
available source of food (Fraser and Sumar 1998). Fish are however, a very perishable commodity
(Efiuvwevwere and Ajuboye 1996) and in today’s consumer driven world, many fish products
on the supermarket shelves are sourced from international markets (Venugopal 2002) often
involving complex distribution networks (Lehane and Olley 2000). This poses major challenges
in terms of ensuring that a fresh, safe product is delivered to the consumer.
The storage life of fresh fish products depends on several factors: storage conditions (time,
temperature, presence and concentration of gases, and relative humidity of the environment),
intrinsic factors of fish (species, age, size, fat content, moisture content, pH, feeding, and
physiological status), and number of initial microflora (Efiuvwevwere and Ajuboye 1996; Kim
and others 2001; Venugopal 2002). How fresh the fish are will determine if they can be used
for human consumption or what the quality grade of fishmeal will be (Fraser and Sumar 1998).
Shelf life of fish for human consumption varies from batch to batch (Dalgaard and others 1997)
however predictions are still required for consumer confidence in the product.
The loss of fish freshness and associated spoilage is a complex combination of autolytic and
microbiological processes (Waliszewski and Avalos 2001). Reducing the storage temperature
binnenwerk.indd 297 21-08-2006 12:35:06

can slow these processes down, which in consequence retards the appearance of characteristics
associated with reduced quality such as off odours (Gram and Dalgaard 2002; Kim and others
2001; Gelman and others 2001), gaping of the muscle (Delbarre-Ladrat and others 2004),
congealing blood and colour loss in the gills (Fraser and Sumar 1998). In a study by Kim and
others (2001) it was demonstrated that high storage temperatures of 25 °C exacerbated the
deterioration of quality through spoilage. The link between temperature and spoilage rates is
indeed very tangible and as a consequence proper chill chain management is recommended as
the main way of retarding microbial spoilage of fish (Waliszewski and Avalos 2001).
As well as good management practises (Huss and others 2000), there are different interventions
that can prolong the shelf life of fresh fish following their harvest. These include rapid chilling,
storage at very low temperatures, smoking, freezing, modified atmosphere storage and using
ionising radiation, as well as different preservatives such as organic acid salts or antibacterial
agents (Efiuvwevwere and Ajuboye 1996; Gelman and others 2001). The addition of chemical
preservatives such as potassium sorbate has been found to reduce the development rate of
bacteria in silver perch (Gelman and others 2001) and extends shelf life in shrimp (Al-Dagal
and Bazarra 1999). However, many of the commercially available preservatives are not effective
on fresh fish (Waliszewski and Avalos 2001).
Various products have appeared on the market over recent years, which are claimed to reduce
spoilage rates of both fresh fish and fish destined for fishmeal if applied to RSW tanks at
sea. Certain sectors of industry in Ireland have expressed interest in utilizing some of these
products to assist in optimising quality and financial returns. The potential benefits of these
products are obvious, particularly where extended sea trips are undertaken, where extended
storage pre-processing is required or where subsequent complex distribution chains exist. In
the present trial temporal trends in sensory assessments, total viable count (TVC), total volatile
basic nitrogen (TVB-N), the levels of biogenic amines as histamine, putrescine, cadaverine
and tyramine were compared. The objectives were to establish if four commercially available
products claiming to reduce spoilage rates maintained freshness of whole fish destined for
human consumption and fish offal destined for fishmeal for extended periods of time.

Four commercially available preservatives; acetic acid, Citrox™ (Global Spill Control Pty Ltd),
Fishform (Hydro Formates AS) and XyRex (XyRex Ltd.) were selected to ascertain their
impacts at reducing spoilage on whole fresh herring (Clupea harengus) stored in a chilled
environment and herring offal stored both in chilled and ambient environments. Fresh whole
herring destined for human consumption and herring offal destined for fishmeal were sourced
from a local processor in August 2004 who obtained the herring from a single refrigerated
seawater vessel. In total 15 labelled bins of one thousand litres were used for trial purposes.
Eighty kg (100 kg for XyRex) of whole herring were placed in 5 bins and each of the test
products were added to four of these bins following the manufacturers’ instructions. The fifth
bin of whole herring was left untreated for control purposes. The 5 bins were placed in a chilled
environment and samples were taken on selected days for quality and laboratory analysis (TVC,
TVB-N, and the biogenic amines histamine, putrescine, cadaverine and tyramine). Similarly, for
fish offal trials 80 kg (100 kg for XyRex) of fish offal were placed in each of the remaining 10
bins. Each of the products was added to two random bins and the two remaining bins were left
untreated as the control. Four of the treated bins (one for each treatment) and an untreated
binnenwerk.indd 298 21-08-2006 12:35:06

Spoilage of herring and offal when treated with commercial preservative or additive products
bin were placed in a chilled environment for the duration of the trial. The remaining 5 bins
were left in sheltered ambient conditions for the duration of the trial. As with the whole fresh
herring, sampling was carried out on selected days for quality and laboratory analysis (TVB-N
and the biogenic amines). Tidbit® (Onset Computer Corporation) temperature loggers were
placed in a selection of containers in both environments to monitor temperature profiles
throughout the trial.
The TVB-N levels were determined according to Antonacopoulos and Vyncke (1989). The TVC
contents were determined by ISO 4833:2003 method. The biogenic amines were analysed
using the Torry method, HPLC (Schmidtlein 1980) (Post column derivitisation using OPA). All
laboratory analysis was carried out at an Irish National Accreditation Board (INAB) accredited
laboratory.
Sensory analysis was carried out using a herring quality scoring system developed by the Irish
Sea Fisheries Development Board. Score allocation for each of the attributes was based on a
five point decreasing scale with 5 for very fresh and 1 for decomposition developed. The sensory
qualities of the fish were determined by the appearance of the eyes, gills, skin, the state of
rigor and the odour of the gills by an experienced quality assessor.
Results
Temporal trends of temperature profiles in chilled environments revealed that after an initial
elevated temperature of 15 °C, temperatures reduced to 0 °C after 30 hours in all treatments
monitored and remained at this level for the duration of the trials (Figure 1). In terms of
freshness quality of chilled whole fish, no marked benefit was discernible from using any of
the products (Table 1). In fact quality scores from both, acetic acid and Fishform treated fish
were lowest largely due to the fact that the preservation processes of both products altered the
sensory attributes. In contrast, TVB-N (Figure 2a), Histamine (Figure 2c) and TVC (Figure 2b)
testing revealed that both acetic acid and Fishform delayed spoilage most effectively. Citrox
and XyRex were only marginally better than the control in this regard. The findings for tyramine,
putrescine and cadaverine were less clear for each treatment (Table 2). By day 7 Fishform and
acetic acid generally maintained lower levels than the two remaining treatments and the control
for these tests.
Table 1. Temporal trends in sensory scores allocated to treated herring.
Treatment Day
One Two Three Four Seven
Control 3 4 3 3 2
Acetic acid 3 3 3 2 2
Citrox 3 4 3 2 3
Fishform 3 2 3 2 2
XyRex 4 3 3 3 3
binnenwerk.indd 299 21-08-2006 12:35:06

40
logger inserted
35 logger removed
Day 1
Day 7
30
Chill Fish Tank 3
25
Temp (°C)
Chill Offal Tank 3

20
Chill Offal Tank 5
15
Ambient Offal Tank 3
10
Ambient Offal Tank 5
5
0
-5
0 20 40 60 80 100 120 140 160
Cumulative hours
Figure 1. Temporal temperature profile for herring stored in chill conditions and herring offal stored in chill
and ambient conditions.
Table 2. Temporal trends of tyramine, putrescine and cadaverine levels of whole treated herring and offal
stored under chilled and ambient conditions (mg/100 g wet weight). Levels of tyramine, putrescine and
cadaverine were not determined for offal stored under ambient conditions on day 7.
Whole herring chilled Offal chilled Offal ambient
Day 1 Day 4 Day 7 Day 1 Day 4 Day 7 Day 1 Day 4
Tyramine Control 0 0 0.44 0 0.95 3.94 0 6.71

Acetic acid 0 0.09 0.44 0 1.01 1.71 0 8.47
Citrox 0 0.03 0.14 0 1.07 4.12 0 7.65
Fishform 0 0.04 0.3 0 1.1 2.09 0 9.82
XyRex 0 0.02 0.3 0 0.59 4.84 0 8.09
Putrescine Control 0.2 1.04 0.82 0.2 2.26 3.05 0.2 26.07
Acetic acid 0.1 0.43 0.61 0.1 1.95 2.15 0.1 7.06
Citrox 0.1 0.4 0.46 0.1 1.98 3.04 0.1 19.78
Fishform 0.2 0.23 0.74 0.2 1.79 2.01 0.2 10.27
XyRex 0.09 0.34 1.2 0.09 1.95 3.53 0.09 27.1
Cadaverine Control 0.6 0.57 4.76 0.6 8.1 10.77 0.6 20.14
Acetic acid 0.08 1.01 3.28 0.08 5.45 7.36 0.08 11.35
Citrox 0.06 1.28 4.17 0.06 9.21 11.83 0.06 21.06
Fishform 0.1 0.43 1.71 0.1 2.99 4.6 0.1 12.35
XyRex 0.08 1.11 4.88 0.08 6.9 12.01 0.08 22.49
binnenwerk.indd 300 21-08-2006 12:35:07

a. 40
35 Control
TVB-N (mgN/100g)
30 Acetic acid
25 Citrox
20
Fishform
15
Xyrex
10
5 Acceptable
Limits
0
1 2 3 4 5 6 7
Day of storage
b. 2500
2000 Control
TVC (cfu/g)
Acetic acid
1500
Citrox
1000 Fishform
Xyrex
500
0
1 2 3 4 5 6 7
Day of storage
c. 5
Histamine (mg/100g)
4
Control
3 Acetic acid
Citrox
2 Fishform
Xyrex
1
0
1 2 3 4 5 6 7
Day of storage
Figure 2. Temporal trends from laboratory analysis of treated herring when stored in chill conditions. a.
Temporal trends in TVB-N levels of treated herring stored in chilled conditions. b. Temporal trends in TVC
levels of treated herring stored in chilled conditions. c. Temporal trends in histamine levels of treated herring
stored in chilled conditions.
binnenwerk.indd 301 21-08-2006 12:35:08

TVB-N (Figure 3a) and histamine levels for chilled offal treatments (Figure 3b) followed very
similar trends to those of chilled whole fish, with both acetic acid and Fishform maintaining
lowest levels. For the biogenic amines putrescine, cadaverine and tyramine (Table 2), the trend
was more evident in fish offal than for whole fish, with acetic acid and Fishform retarding the
development of these biogenic amines. The benefits of Citrox and XyRex relative to the control
were only marginal in this regard.
With increased temperatures under ambient conditions, averaging at 17 °C for the duration of
the trial (Figure 1), spoilage rates increased in fish offal, as is evidenced from the different
test regimes. Acetic acid and Fishform proved most effective at maintaining TVB-N (Figure 4a)
and histamine levels (Figure 4b) and also clearly maintained lower putrescine, cadaverine and
tyramine levels relative to the control and the two remaining treatments, Citrox and XyRex
(Table 2).
a. 40
Control
35
TVBN (mgN/100g)
30 Acetic acid
25 Citrox
20
Fishform
15
10 Xyrex
5 Acceptable
Limits
0
1 2 3 4 5 6 7
Day of storage
b. 10
Control
8
Histamine (mg/100g)
Acetic acid
6 Citrox
Fishform
4
Xyrex
2
Acceptable
Limits
0
1 2 3 4 5 6 7
Day of storage
Temporal trends in TVB-N levels of treated herring offal stored in chilled conditions. b. Temporal trends in
histamine levels of treated herring offal stored in chilled conditions.
binnenwerk.indd 302 21-08-2006 12:35:09

a. 180
Control
150
TVBN (mgN/100g)
Acetic acid
120 Citrox
90 Fishform
60 Xyrex
30 Acceptable
Limits
0
1 2 3 4
Day of storage
b. 18
Control
Histamine (mg/100g)
15
Acetic acid
12
Citrox
9
6 Fishform
3 Xyrex
0
1 2 3 4
Day of storage
Temporal trends in TVB-N levels of treated herring offal stored in ambient conditions. b. Temporal trends in
Histamine levels of treated herring offal stored in ambient conditions.
For this trial a large amount of fresh whole fish destined for human consumption and fish offal
destined for fishmeal were used (1260 kg) to mimic commercial conditions as closely as possible.
It has been widely documented that focussing on specific tests to derive spoilage trends can
often be misleading (Poulter and others 1981; Dawood 1988). Therefore a number of different
tests commonly used to assess spoilage trends were carried out resulting in a total of 25 sensory
assessments, 60 TVB-N, 42 TVC and 120 biogenic amine tests being undertaken. Unfortunately due
to the scale of the trial and resource constraints it was not possible to undertake replicate laboratory
testing within sample batches collected for each test regime, therefore results are not statistically
robust. Furthermore only one assessor undertook daily sensory assessments. Nonetheless temporal
trends can be gleaned from the derived data from the various tests undertaken, giving an initial
appraisal of the performance of each of the products during the trial.
binnenwerk.indd 303 21-08-2006 12:35:10

Sensory scores of whole chilled herring revealed that spoilage occurred under each of the
treatments with progression of time. Whole fish treated with Fishform and acetic acid had the
poorest scores largely due to the fact that the preservation processes which were similar to
that observed for pickling or marinating, altered the freshness attributes used to derive the
scores. Both Fishform and acetic acid concentrations tested, are used only on fishmeal and
obviously given the fact that they altered the characteristics of the fish are unsuitable for fresh
whole fish destined for human consumption. The freshness trends from XyRex and Citrox, both
products that have been specifically marketed for maintaining freshness when used on whole
fish destined for human consumption, proved inconclusive.
As spoilage progresses in fish volatile basic nitrogen compounds develop and their increasing
levels are regularly used as an indicator of spoilage (Venugopal 2002). For chilled whole fish
slight increases in TVB-N levels were evident in the control relative to the other treatments by
day 3 and 4 indicating that spoilage was proliferating more rapidly in untreated fish. However
by day 7 only whole fish treated with acetic acid and Fishform maintained acceptable levels of
below 25 mg N/100 g. Total viable count is the most common method for the determination of
bacteriological quality of seafood (Huss and others 1974). TVC testing of whole fish revealed
that levels were maintained within acceptable limits for the duration of the trial in the chilled
environment. Nonetheless a trend was evident, with highest levels observed in untreated
whole fish and lowest levels in both whole fish treated with acetic acid and Fishform. Generally
Citrox and XyRex maintained slightly lower TVC levels relative to the control. Histamine is
rarely found in fresh whole fish, but increases with the progression of fish decomposition in
certain fish species (Frank and others 1981) and levels have been used as a chemical index
of spoilage (Shakila 2003). It is obvious that the whole fish were fresh at the commencement
of these trials, with only 0.02 mg/100 g of histamine noted on day 1 of the trial where the
fish would have been approximately 48 hours since capture. By day 4 a maximum of 0.48
mg/100 g was noted in whole fish treated with Citrox, however, little differences were evident
between each of the treatments at this stage. By day seven the chilled environment maintained
histamine levels within acceptable limits, with a maximum of 4.3 mg/100 g evident for the
control and as with the above laboratory tests acetic acid and Fishform maintained the lowest
histamine levels of 1.1 and 0.8 mg/100 g respectively. As with histamine the proliferation of
other biogenic amines, including putrescine, cadaverine and tyramine steadily increase once
bacterial spoilage is initiated and thus are used as tests for fish spoilage (Fenandez-Salguero
and Mackie 1987). However no discernible temporal trend was evident for the biogenic amines
putrescine, cadaverine and tyramine, indicating that advanced decomposition did not progress
under chilled conditions for whole fish for the duration of the trial.
Although fish offal were retained under identical chilled conditions as that of whole fish, results
from the various analyses indicated that spoilage occurred more rapidly and in addition the
effectiveness of certain products were more readily evident. The rapidity of spoilage of fish
offal compared to whole fish is due to autolytic enzymes from the digestive tract and spoilage
bacteria on the surface impacting on the exposed tissue accelerating the rate of degradation
(Borquez and others 1994). Although general sensory changes were noted during the trial no
specific scores were recorded due to a lack of an available quality scoring system. TVB-N levels
steadily increased for the first three days of the trial however from day 4 onwards levels were
highest in the control fish offal and in offal treated with Citrox and XyRex. TVC tests were not
undertaken as it was felt that levels would be excessively high due to bacterial degradation from
the gut contents and exposed flesh. Histamine levels were not detectable on day 1 indicating
binnenwerk.indd 304 21-08-2006 12:35:11

that the offal was very fresh, however by day 4, levels were above 2 mg/100 g in all treatments,
and were highest in untreated offal and offal treated with XyRex and Citrox. By day seven
levels remained below 4 mg/100 g for offal treated with acetic acid and Fishform, whereas in
all other treatments including the control, levels were at least double, with a maximum of 9.2
mg/100 g for offal treated with XyRex. The levels of putrescine, cadaverine and tyramine tested
were very low (0-0.6 mg/100 g) on day one indicating the freshness of the offal. By day seven
differences in each treatment was discernible with lowest levels detected in offal treated with
acetic acid and Fishform.
The impact of high temperatures on spoilage rates and the influence of certain products were
clearly more discernible on offal stored under ambient conditions. By day 2, TVB-N levels
were between 41 and 46 mg N/100 g, apart from offal treated with acetic acid and Fishform,
which had levels of 12.5 and 24.6 mg N/100 g respectively. TVB-N levels increased to 70.0 and
74.2 mg N/100 g by day four in offal treated with acetic acid and Fishform respectively, but
these levels were half of those observed in the remaining treatments. Histamine levels were
barely discernible on day 1, as offal was fresh, however histamine development occurred very
rapidly under ambient temperatures, with a maximum of 16.8 mg/100 g in offal treated with
XyRex on day 4. As with TVB-N, histamine levels on day four were lowest in offal treated with
acetic acid and Fishform, with levels of 5.4 and 5.8 mg/100 g noted respectively. The levels
of putrescine, cadaverine and tyramine were between 0.2 and 0.6 mg/100 g on day one. By
day four however levels of cadaverine and putrescine had increased to over 20 mg/100 g in all
treatments with the exception of offal treated with acetic acid and Fishform which had levels
of 10 mg/100 g.
The health implications of consuming spoiled fish are well documented. Particularly scrombroids
that can readily accumulate high levels of histamine, as the toxicity manifests itself as ‘histamine
poisoning’ in certain individuals (Venugopal 2002). It has also been noted that other biogenic
amines, such as putrescine and cadaverine exacerbate histamine toxicity in addition to being
carcinogenic (Kim and others 2001; Mah and others 2002). The presence of histamine levels in
fishmeal also presents problems for animals, such as gizzard erosion and weight loss in poultry.
There are also potentially serious implications when fishmeal- like protein preparations are used
as supplements known as ‘protein concentrate’ in human food (Hose and others 2002).
It is readily evident that temperature played a major role in the spoilage rates of fish examined
in this study, with spoilage occurring five times in order of magnitude in fish offal stored under
ambient conditions compared to chilled whole fish. It has previously been suggested that an
increase of 10 ºC increases spoilage rates by an order of 5-6 (Venugopal 2002). In terms of the
performance of the various products tested, especially where effective chill chain management
has not been implemented e.g. fishmeal production, it is clear that the organic acids such as
acetic acid and Fishform performed most effectively. Borques and others (1994) found that
acetic acid, at concentrations 5 mg/kg was most effective at inhibiting spoilage, with TVB-
N levels 50% lower than untreated fish after 36 hours storage. Furthermore acetic acid as a
preservative has the advantage of being abundant in nature and a normal natural metabolite
in man and animal (Huges 1960). Therefore compared to all products tested acetic acid proved
most effective for fishmeal, particularly if chill chain management can not be achieved. In terms
of the two products XyRex and Citrox, marketed for reducing spoilage in whole fish destined for
human consumption the benefits were not readily evident in either whole fish or fish destined
for fishmeal.
binnenwerk.indd 305 21-08-2006 12:35:11

The most prudent recommendation to minimise spoilage and optimise quality is to implement
Good Manufacturing Practices, whereby pre-determined high standards of hygiene and handling
are adhered to throughout the chill chain. Under this regime no products are required, which
is increasingly becoming a requirement as consumers primarily purchase fish, as they feel its
wholesome and free from additives or any unnecessary interventions. For fishmeal production,
products are required, but the amount can be minimised by effective chill chain management
from catching through to final processing.
This preliminary study provides an initial appraisal of the performance of various products
under different testing regimes. Further work should however be undertaken whereby sufficient
replicate testing is carried out to ensure statistical validity of the outcomes of the study.
Conclusions
Acetic acid and Fishform proved most effective at reducing spoilage rates of fish offal stored
under both chilled and ambient environments. Although both these products also maintained
lowest spoilage levels in whole chilled fish, the fact that these altered the sensory attributes of
whole fish through preservation, somewhat similar to that observed for pickling or marinating,
renders them ineffective for fresh whole fish destined for human consumption. The two remaining
products Citrox and XyRex did not markedly outperform the control in terms of reducing spoilage
and it is recommended that GMP is a more effective way of minimising spoilage. Further tests
should be undertaken; ensuring replicate analyses are carried out.
References
Ababouch L, Afilal ME, Benabdeljelil H, Busta FF. 1991. Quantitative changes in bacteria, amino acids and biogenic
amines in sardine (Sardina pilchardus) stored at ambient temperature (25-28 °C) and in ice. Int J Food Sci Technol
26:297-306.
Al-Dagal MM, Bazarra WA. 1999. Extension of shelf life of whole and peeled shrimp with organic acid salts and
bifidobacteria. J Food Prot 62(1):51-6.
Antonacopoulos N, Vyncke W. 1989. Determination of volatile basic nitrogen in fish: a third collaborative study by
WEFTA. Z Lebensm Unters For 189:309-316.
Delbarre-Ladrat C, Verrez-Bagnis V, Noel J, Fleurence J. 2004. Relative contribution of calpain and cathepsins to protein
degradation in muscle of sea bass (Dicentrarchus labrax L.). Food Chem 88(3):389-395.
Borquez R, Espinoza M, Ormeno R. 1994. Effects of storage time and chemical preservatives on the total volatile
basic nitrogen-content in Chilean mackerel (Trachurus murphy) prior to fish-meal production. J Sci Food Agric
66(2):181-186.
Council Directive 1991 Laying down the health conditions for the production and the placing on the market of fishery
products 91/493/EEC. O J L 268, 24.09.1991, 15p.
Dalgaard P, Mejlholm O, Huss HH. 1997. Application of an iterative approach for development of a microbial model
predicting the shelf-life of packed fish. Int J Food Micro 38(2):169-179.
Dawood AA, Karkalas J, Roy RN, Williams CA. 1988. The occurrence of non-volatile amines in chilled-stored rainbow
trout (Salmo irideus). Food Chem 27:33-45.
Efiuvwevwere BJO, Ajiboye MO. 1996. Control of microbiological quality and shelf-life of catfish (Clarias gariepinus) by
chemical preservatives and smoking. J Appl Bacteriol 80(5):465-470.
Fernandez-Salguero J, Mackie IM. 1987. Comparative rates of spoilage of fillets and whole fish during storage of haddock
(Melanogrammus aeglefinus) and herring (Clupea hearengus) as determined by the formation of non-volatile amines.
Int J Food Sci Technol 22:385-390.
binnenwerk.indd 306 21-08-2006 12:35:11

Fraser O, Sumar S. 1998. Compositional changes and spoilage in fish - an introduction. Nutrition and Food Science
98:5-275.
Gelman A, Glatman L, Drabkin V, Harpaz S. 2001. Effects of Storage Temperature and Preservative Treatment on Shelf
Life of the Pond-Raised Freshwater Fish, Silver Perch (Bidyanus bidyanus). J Food Prot 64(10):1584-1591.
Gram L, Dalgaard P. 2002. Fish spoilage bacteria-problems and solutions. Curr Opin Biotech 13(3):262-6.
Huss HH. 1995. Quality and quality changes in fresh fish. FAO Fish. Tech Paper 348.
Huss HH, Reilly A, Ben Embarek PK. 2000. Prevention and control of hazards in seafood. Food Control 11(2):149-
156.
Kim SH, Field KG, Chang DS, Wei CI, An H. 2001. Identification of Bacteria Crucial to Histamine Accumulation in Pacific
Mackerel during Storage. J Food Prot 64(10):1556-1564.
Kose S, Quantick P, Hall G. 2003. Changes in the levels of histamine during processing and storage of fishmeal. Anim
Feed Sci Tech 107(1):161-172.
Lehane L, Olley J. 2000. Histamine fish poisoning revisited. Int J Food Microbiol 58:1-37.
Mah JH, Han HK, Oh YJ, Kim MG, Hwang HJ. 2002. Biogenic amines in Jeotkals, Korean salted and fermented fish
products. Food Chem 79(2):239-243.
Schmidtlein H. 1980. Determination of biogenic amines by means of high pressure liquid chromatography. Translation
Ser (Torry Research Station); T1289, T1292.
Veciana-Nogues MTM, Vidal-Carou MC, Marine-Font A. 1990. Histamine and tyramine during storage of anchovy, Engraulis
encrasicholous relationships with other fish spoilage indicators. J Food Sci 55:1192-1193.
Venugopal V. 2002. Biosensors in fish production and quality control. Biosens Bioelectron 17(3):147-157.
Waliszewski KN, Avalos M 2001. Shelf-life extension of tilapia in crushed ice with additives at-3C. J Food Proc Pres
25(4):299-307.
binnenwerk.indd 307 21-08-2006 12:35:11

binnenwerk.indd 308 21-08-2006 12:35:11
Effect of freezing and different heat treatments on Anisakis larvae:
Preliminary study
Margarita Tejada1, Maria Teresa Solas2, Alfonso Navas3 and Angel Mendizábal4
1Instituto del Frío (IF), Consejo Superior de Investigaciones Científicas (CSIC)
C/ José Antonio Novais, 10. 28040 Madrid, Spain
2Departamento de Biología Celular, Facultad de Biología, Universidad Complutense, Madrid,
Spain
3Museo Nacional de Ciencias Naturales (MNCN), Consejo Superior de Investigaciones Científicas
(CSIC), Spain
4Ayuntamiento de Madrid, Instituto de Salud Pública, Departamento de Seguridad Alimentaria,
Sección Inspección Veterinaria de Mercamadrid, Spain
Abstract
Ingestion of fish infested with Anisakidae larvae by humans can cause infestation (anisakidosis)
when the live larva is eaten and/or allergy which occurs in individuals previously sensitised
to the live parasite. Treatments in which the larvae are killed prevent anisakidosis. However,
there is no information on the effect of these treatments on the parasites allergens (somatic,
excretory/secretory and surface allergens). Some allergens are temperature resistant. Hake
steaks, to which parasites were added (Anisakis sp.) were chilled, frozen, heated and microwave
treated. The results of Electron Microscopy and fluorescence emission of the parasites showed
differences in relation to the technological treatments.
Keywords: Anisakis, allergy, Scanning Electron Microscopy (SEM), Environmental Scanning

Electron Microscopy (ESEM)
Introduction
The intake of fish infested with Anisakidae larvae is a problem in countries where fish is
traditionally eaten raw or undercooked and is an emerging problem in Spain. Ingestion of
parasite larvae by humans can cause infestation with the parasite (anisakiasis or anisakidosis)
when the live larva is eaten. Allergy of varying intensity can occur in individuals who have
been previously sensitised to the live parasite. Clinical symptoms have been described mainly
as gastrointestinal symptoms. When these symptoms are associated with dermatological
manifestation or anaphylaxis, the disorder is called gastro allergic anisakiasis. Anisakis
hypersensitivity is described when patients present exclusively cutaneous and/or anaphylactic
manifestations. The clinical signs depend on the zone of the gastrointestinal tract where the
larva is located and also if the response is to an acute or chronic infection (Kitadai and others
1987; Matsuoka and others 1994; Daschner and others 2000; López Serrano and others 2000).
Clinical symptoms due to the consumption of fish infested with live Anisakidae larvae have been
described since decades, mainly in countries where raw, undercooked fish or fish subjected to
mild treatments, where the larvae remain alive is consumed (VanThiel and others 1960; Williams
1965). However, studies relating human allergy to consumption of fish parasitized with Anisakis
sp. are more recently published (Kasuya and others 1990; Kasuya and Koga 1992; Ishikura and
binnenwerk.indd 309 21-08-2006 12:35:11

others 1993; Audicana and others 1995). The prevalence, geographical distribution, frequency
and extent of sensitization among the general population, are unknown in spite of the large
amount of work done in these topic at present.
Anisakidosis can be prevented by subjecting the fish to treatments in which the larva is killed
(freezing, frozen storage, high-temperature or prolonged cooking, microwave processing, high
pressure, pH, salt concentration, irradiation etc.) (Van Mameren and Houwing 1968; Smith and
Wooten 1975; Hauck 1977; Panebianco and Lo Schiavo 1985; FAO/IAEA 1992; Karl and others
1995; Howgate 1998; Adams and others 1999; FDA 2001; Beldsoe and Oria 2001; Molina García
and Sanz 2002; Schlüter and others 2004; Sánchez-Monsalvez and others 2005) although there
is disagreement about the intensity of the treatments to kill the larvae. Freezing of fish before
eating raw or subjected to mild treatments is mandatory in many countries. Nevertheless, there
is no information on the effect of various technological or culinary treatments on the allergens
from these parasites. Three groups of allergens from Anisakis simplex have been described:
somatic, excretory/ secretory (ES) and surface allergens. Most of the described allergens are
proteins of molecular weight between 13 to 150 kDa. Some of them have been proved to be
heat resistant (Moneo and Caballero 2002; Caballero and Moneo 2004; Valls and others 2005).
There is controversy among researchers about the allergy caused by the dead parasite (Audicana
and others 1997, 2002; Valiñas and others 2001; Caballero and Moneo 2004). This disagreement
may be due to the fact that differences in the origin of the parasite, killing methods and
ante mortem conditions, may cause a different release of allergens from the parasite to the
gastrointestinal tract.
The objective of this preliminary study is to detect the effect of chilling, freezing, cooking and
microwave treatments on the surface structure of Anisakis simplex to determine later the effect
of such treatments on recognition of allergens from the parasites.
Preparation of the lots

Live Anisakis simplex larvae were obtained from parasitized hake (Merluccius merluccius) ovaries
(Figure 1). Forty to sixty live larvae were sandwiched between two hake steaks (1 cm thick)
of thawed hake (Fig 2a). Each sandwich was vacuum sealed in heat-resistant bags (Figure 2b)
and stored in chilled conditions (5 ± 1 ºC) (Lot 1) or immediately frozen and stored at -20 ºC
(Lot 4). After 20 hours of chilled storage five sandwiches were opened to observe if migration
of the larvae to the flesh had occurred. Five other sandwiches were individually cooked for 8
min. in a water bath set at 95 ± 1 ºC (Lot 2). Five sandwiches were taken out of the sealed
bags, and processed individually in a microwave resistant container covered with microwave
resistant film using a commercial 900 W microwave oven operating at maximum (100%) power
for 3 minutes (Lot 3). Time setting for treating (water bath or microwave treatments) of the
fish sandwiches was adjusted to the size of the sandwiches. After the treatments the sandwiches
were again vacuum packed and stored at 5 ± 1 ºC. After 24 hours one sandwich subjected to
each treatment was opened to detect if movement of the larvae took place.
Movement of the larvae were observed at room temperature by comparison of pictures taken of
the ovaries and the infested sandwiches. Pictures were taken in a fixed position up to 30 min
at 5 min intervals.
binnenwerk.indd 310 21-08-2006 12:35:12

Effect of freezing and different heat treatments on Anisakis larvae
Figure 1. Hake ovaries parasitized with Anisakis larvae.
a b
c d
Figure 2. Hake steaks inoculated with Anisakis larvae: a. Immediately after inoculation; b. Vacuum packed
sandwich; c. after 20 hours chilled storage; d. after 5 days chilled storage.
binnenwerk.indd 311 21-08-2006 12:35:12

Emission of fluorescence of the nematodes in the treated and control lots was measured after
exciting by UV light (366nm) was evaluated.
Electron microscopy Anisakis simplex from the ovaries and from the four lots were observed
by Scanning Electron Microscopy (SEM) and Environmental Scanning Electron Microscopy
(ESEM).
SEM
Individual Anisakis larvae and small blocks of infested tissue (1.5 - 2 cm cubes) were cut from
the sandwiches in all lots. The samples were fixed with a mixture (1:1 v/v) of paraformaldehyde
(4%) and glutaraldehyde (0.2%) in 0.1 M phosphate buffer pH 7.2, post-fixed with OsO4, washed,
dehydrated in increasing concentrations of acetone, critical-point-dried, sputter-coated with
gold/palladium in a metallizer (Balzer, SCD004) and scanned by SEM (Jeol, JSC,6400, Akishima,
Tokyo, Japan) at 20 kV.
ESEM
Pieces of infested tissue (muscle (Lot 1 to Lot 4) and ovaries) were immersed in water for a
period of approximately 30 min. to extract the larvae. Than, these Anisakis were separated from
Lot 1 and ovaries were immersed in acetic acid solution (pH 2) (acid treatment) or in tap water
(water treatment) for 7 days after which, selected specimens were examined. Larvae removed
from treated lots (Lot 2, Lot 3 and Lot 4) were maintained in water until examined (maximum
time 60 min) using an Environmental Scanning Electron Microscope (FEI-Quanta 200 Scanning
Electron Microscope; SEM-ESEM), operating up to 30 kV in low vacuum mode (environmental
mode) in which the column is under high pressure range of 0.133 to 53.32 mbar using water
vapour from a built-in water reservoir for preserving the integrity of the hydrated samples.
Migration of the larvae was observed from the internal part of the sandwich into the hake
muscle after 20 hours of chilled storage (Figure 2c). Movement and migration of the larvae
were evident in the ovaries and in the chilled lot after 5 days storage, the longer the storage
period the deeper the penetration in the muscle (Figure 2 d).
The temperature in the thermal centre reached in the fish sandwiches in Lots 2 and 3 (86.3 ºC
and 66.9 ºC) is considered as sufficient to kill the Anisakis simplex larvae (Council Directive 493
1991; Adams and others 1999). Lot 4 was frozen stored at -20 ºC for 48 hours. No movement
of the parasites was observed in the heat or microwave treated lots neither in the frozen lot
after thawing the sandwiches overnight at 5 ºC.
Emission of fluorescence by the larvae when the sandwiches were examined under UV light
was clearly observed in Lot 4. Nevertheless the emission of fluorescence in this lot decreased
or even disappeared after heating the thawed sandwiches in the conditions of Lot 2. No clear
emission of fluorescence by the Anisakis larvae was observed in the other lots, irrespective of
the treatment. Fluorescence of Anisakis larvae has been associated with the death of the larvae
developing a fluorescent white colour in a UV chamber (Karl and others 1995).
No apparent changes in the Anisakis larvae were detected by ESEM or SEM in the chilled stored
lot (Lot 1) even in the harsh acidic conditions employed (Figure 3a, b, c)
binnenwerk.indd 312 21-08-2006 12:35:12

b c
Figure 3. a. Anisakis larvae from hake muscle chilled stored (Lot 1) for 44 h, observed by SEM; b-c. Anisakis
larvae extracted from hake ovaries observed by ESEM. b. acid treatment (pH 2, 7 days); c. water treatment
(7 days). Bars: a. 800 µm; b. 300 µm; c. 1 mm.
Changes detected in the surface of the Anisakis larvae in the heated (Lot 2) (Figure 4 a, b)
and microwave (Lot 3) (Figure 5a, b) treated lots were similar, with coagulated and disrupted
zones and emission of internal material out of the body cavity. The body shape changed from
a cylindrical to a dehydrated appearance, more evident when examined by ESEM.
In the frozen stored lot (Lot 4) (Figure 6a, b, c) Anisakis with no apparent changes were found
together with parasites with changes in the shape similar to the ones observed by ESEM in
treatments were the temperature was raised. Nevertheless no apparent rupture of the cuticle
was evident.
Conclusions
Changes in emission of fluorescence, in the surface and in the body shape of Anisakis simplex
larvae after freezing and/or heat treatments were observed. When the parasite cuticle is damaged
by heating and microwave heating it is possible that compounds with allergic potential may
be released.
Research work is needed to determine the effect of freezing and heat treatments on the death
of the larvae in the context of potential allergens.
binnenwerk.indd 313 21-08-2006 12:35:13

a b
Figure 4. Anisakis larvae after heating (Lot 2) observed by ESEM. Bars: a. 1 mm; b. 500 µm
a b
Figure 5. Anisakis larvae after microwave treatment (Lot 3). a. SEM; b. ESEM. Bars: a. 3 mm; b. 500 μm.
b c
Figure 6. Anisakis larvae after freezing (Lot 4). a. SEM; b and c. ESEM. Bars: a. 100 µm; b and c. 200 µm
binnenwerk.indd 314 21-08-2006 12:35:13

Acknowledgements
Thanks are due to Dr. J. Ignacio Moneo from the Immunology Department, Hospital Carlos III,
Madrid, and Prof. Dr. Juan A. Balbuena from the Marine Zoology Unit, University of Valencia
for their help in this work.
References
Adams AM, Miller KS, Wekell MM, Dong FM. 1999. Survival of Anisakis simplex in microwave-processed arrowtooth
flounder (Atheresthes stomias). J Food Prot 62(4):403-409.
Audicana MT, Fernandez de Corres L, Muñoz D, Fernandez E, Navarro JA, Del Pozo MD. 1995. Recurrent anaphylaxis
caused by Anisakis simplex parasitizing fish. J Allergy Clin Immunol 96:558-560.
Audicana L, Audicana MT, Fernandez de Corres L, Kennedy MW. 1997. Cooking and freezing may not protect against
allergenic reactions to ingested Anisakis simplex antigens in humans. Vet Rec 140:235.
Audicana MT, Ansotegui IJ, Fernandez de Corres L, Kennedy MW. 2002. Anisakis simplex: dangerous-dead and alive?
Trends Parasitol 18:20-5.
Beldsoe GE, Oria MP. 2001. Potential hazards in cold smoked fish: Parasites. J Food Sci Supplement to vol 66(7):S-
1100-S-1103.
Caballero ML, Moneo I. 2004. Several allergens from Anisakis simplex are highly resistant to heat and pepsin treatments.
Parasitol Res 93:248-51.
Council Directive 493 of 22 July 1991 laying down the health conditions for the production and the placing on the
market of fishery products (91/493/EEC). Official Journal L 286, 24/09/1991. p 15-34.
Daschner A, Alonso Gómez A, Cabanas R, Suárez de Parga, JM, López-Serrano, MC. 2000. Gastroallergic anisakiasis:
Borderline between food allergy and parasitic disease-clinical and allergologic evaluation of 20 patients with
confirmed acute parasitism by Anisakis simplex. J Allergy Clin Immunol 105:176-181.
FAO/IOEA. 1992. Final FAO/IOEA research co-ordination meeting on the use of irradiation to control infectivity of
foodborne parasites. Food Irradiat Newsl 16(1):5-14.
FDA/CFSAN. 2001. Processing Parameters Needed to Control Pathogens in Cold Smoked Fish - Chapter V. Potential
Hazards in Cold-Smoked Fish: Parasites. U. S. Food and Drug Administration Center for Food Safety and Applied
Nutrition.
Hauck AK. 1997. Occurrence and survival of the larval nematode Anisakis sp. in the flesh of fresh, frozen, brined and
smoked pacific herring, Clupea harengus Pallasi. J Parasitol 63(3):197-208.
Howgate P. 1988. “Freezing to kill nematode parasites in fish products: Implications for HACCP.” May 1998. http://
seafood.ucdavis.edu/Pubs/nematodes.htm.
Ishikura H, Kikuchi K, Nagasawa K, Ooiwa T, Takamiya H, Sato N, Sugane K. 1993. Anisakidae and anisakidosis. Prog
Clin Parasitol 3:43-102.
Karl H, Roepstorff A, Huss HH, Bloemsma B. 1995. Survival of Anisakis larvae in marinated herring fillets. Int J Food
Sci Technol 29(6):661-70.
Kasuya S, Hamano H, Izumi S. 1990. Mackerel induced urticaria and Anisakis. Lancet 335:665.
Kasuya S, Koga K. 1992. Significance of detection of specific IgE in Anisakis-related diseases. Arerugi 41:106-10.
Kitadai Y, Haruma K, Sumii K, Uemura N, Tari A, Tokumo K, Kajiyama G. 1987. Two cases of vanishing tumor of the
stomach due to anisakiasis. Jpn J Intern Med 76:1434-1438.
López Serrano MC, Alonso-Gómez A, Moreno-Ancillo A, Daschner A, Suárez de Parga J. 2000. Gastroallergic anisakiasis:
immediate hypersensitivity due to Anisakis simplex infestation. Allergol Immunol Clin 15:230-236.
Matsuoka H, Nakama T, Kisanuti H, Uno H, Tachibana N, Tsubouchi H, Horii Y, Nawa, Y. 1994. A case report of
serologically diagnosed pulmonary anisakiasis with pleural effusion and multiple lesions. Am J Trop Med Hyg
51(6):819–822.
binnenwerk.indd 315 21-08-2006 12:35:14

Molina García AD, Sanz PD. 2002. Anisakis simplex larva killed by High-Hydrostatic-Pressure Processing. J Food Prot
65(2):383-388.
Moneo I, Caballero ML. 2002. Anisakis simplex larvae release allergens during a short incubation in diluted acid which
can be used for clinical diagnosis. Allergol Immunol Clin 17:201-7.
Panebianco A, Lo Schiavo A. 1985. Indagine sulla presenza di larve Anisakidi in aringhe salate e affumicate del
commercio. Considerazioni d’ordine insopectivo. Clin Vet 108(3):80-184.
Sánchez-Monsalvez I, De Armas-Serra C, Martínez J, Dorado M, Sánchez A , Rodríguez-Caabeiro F. A New Procedure
for Marinating Fresh Anchovies and Ensuring the Rapid Destruction of Anisakis Larvae. J. Food Prot 68(5):1066–
1072.
Schlüter O, Boguslawski S, Meyer C, Schubring R, Knorr D. 2004. Potential of moderate high hydrostatic pressure to
control fish parasites (Anisakis simplex) and specific microorganisms. ICEF9 Meeting (International Congress on
Engineering and Food) Montpellier, France.
Smith JW, Wooten R. 1975. Experimental studies on the migration of Anisakis sp. Larvae (Nematoda: Ascaridida) into
the flesh of herring, Clupea harengus L. Int J Parasitol 5:133-136.
Valiñas B, Lorenzo S, Eiras A, Figueiras A, Sanmartín ML, Ubeira FM. 2001. Prevalence of and risk factors for IgE
sensitization to Anisakis simplex in a Spanish population. Allergy 56:667-671.
Valls A, Pascual CY, Martín Esteban M. 2005. Anisakis allergy: an update. Rev Fr Allergol Immunol Clin 45:108–113.
Van Mameren J, Houwing H. 1968. Effect of irradiation on Anisakis larvae in salted herring Elimination of harmful
organisms from food and feed by irradiation. Vienna. IAEA 73-80.
Van Thiel PH, Kuipers FC, Roskam TH. 1960. A nematode parasitic to herring causing acute abdominal syndromes in
man. Trop Geogr Med 12:97-113.
Williams HH. 1965. Roundworms in fishes and so-called “Herring-worm disease”. Br Med J 5440:964-967.
binnenwerk.indd 316 21-08-2006 12:35:14

Use of “filtered smoke” and carbon monoxide with fish: A review
Reinhard Schubring
Federal Research Centre for Nutrition and Food, Department of Fish Quality, Palmaille 9, D-22767
Hamburg, Germany
Abstract
This review deals with the importance of flesh colour for quality assessment of muscle food
and possibilities of influencing the colour via muscle colour pigments. Carbon monoxide as
reagent to stabilise muscle colour and processes producing carbon monoxide for this purpose
are highlighted concentrating on those using filtered smoke to treat fish muscle. The scientific
background is discussed and an overview is given on the situation concerning food law in this
respect in different countries.
Keywords: colour, heme proteins, carbon monoxide, filtered smoke
Introduction
Since a couple of years attentive consumers could notice that bright-red coloured tuna is
offered in sushi bars or at fishmongers indicating excellent freshness of the fish offered.
However, this assumption could be wrong because a technology was used that imparts a fresh
colour even when frozen/thawed fish was used. This technology uses the strong affinity of
carbon monoxide (CO) that is an natural component of smoke against both ferric atoms of
the blood pigment hemoglobin (Hb) and of the muscle pigment myoglobin (Mb) to avoid the
usually occurring brown colouring of the fish flesh. By variation of the traditional technology of
smoking in such a way that the flavour components of the smoke are removed via filtration an
unhindered access is facilitated for CO to the blood muscle pigments. The compounds carboxy
myoglobin (CO-Mb) and carboxy hemoglobin (CO-Hb) developed on this way turned out to be
stable even during prolonged frozen storage and convey the colour impression of a quasi-fresh
product to the consumer.
This technologically caused “progress”, that at present offers only the US consumers freshly
looking tuna loins because of the toleration by the U.S. authorities, is confronted in Europe as
well as in other countries with the existing food legislation. These problems will be discussed
subsequently.
CO is a non-permitted additive for food in Europe. Therefore, technologies which only support
the colour-imparting properties of CO have to be seen as non-conform with the food law in
force. It must be assumed that the aim of using “filtered” smoke or CO for colour stabilisation
of fish is considered as a deception of the consumer. The natural indicator of the deterioration
of quality of tuna fish, the colour change from red to brown, is masked by this manipulation,
thus also spoiling parts of fish impart the impression of perfect freshness to the consumer.
binnenwerk.indd 317 21-08-2006 12:35:14

Reinhard Schubring
Importance of colour of flesh for quality evaluation
Colour is a prime sensory parameter that determines consumer acceptance of a meat (Nam
and Ahn 2003). Consumers want to have the colour of their meat within a normal range, and
discoloured meat is considered as being inferior in quality or as being contaminated. The
estimated loss of value in beef due to discolouration at the retail market in the U.S. is over
700 million U.S.$ per year. While in warm-blooded meat the consumer relates the colour of
lean to freshness (Seideman and others 1984), expectations of consumers for fish flesh are
somewhat different. Fish fillet has to be as white as possible otherwise it will not be accepted
as appropriate to the quality in question. For example, saithe (Pollachius virens), in Germany a
well-liked species available at a reasonable low price, is hardly marketable in U.K. because of
its gray flesh (Love 1997). In U.K. fresh fish fillets are compared with cod and haddock (“white”
fish). Not before the introduction of new species like salmon and tuna fish with pronounced
flesh colour German consumers became familiar with those species. Salmon itself is unable to
synthesise carotinoids (Baker and others 2002) and therefore the feed introducing the pigments
astaxanthin and canthaxanthin are responsible for flesh colour (Matsuno 2001). In contrast, the
heme proteins (Mb and Hb) cause the red muscle colour found in tuna and other dark-fleshed
fish species. As already mentioned for beef the bright red colour of tuna muscle is an economical
important factor (Ochiai and others 1988) due to preference by consumers who consider red
muscle colour as an indicator of freshness (Ishiwata and others 1996).
Possibilities of influencing the colour
The intensity of colour of the salmon flesh can be regulated by the amount of carotinoids added
to feed (Nickell and others 2001). Beside the amount of carotinoids added to feed also the date
of feeding as well as the fat content are important for colouration (Love 1997). In contrast, the
colour stability of heme pigments is affected by other factors. Extensive investigations on the
influence of chilled and frozen storage on the stability of colour in tuna have been performed
earlier in Japan (Bito 1964, 1965a, b, 1967, 1968, 1969a, b, 1970, 1975, 1976, 1980; Bito and
Kiriyama 1973 a, b). It became clear that the freezing rate was meaningful only during short-
term frozen storage and that the storage temperature must be below -35 °C, in general. To avoid
a cracking of the surface in meat frozen at very low temperature e.g., by using liquid nitrogen,
it appeared recommendable to pre-freeze sample at -10 to -40 °C. Colour changes during frozen
storage at -20 °C could be minimised when samples were pre-frozen at -60 to -80 °C for 50 s.
To minimise discolouration during defrosting it was recommended that a temperature range of
-5 to -1.5 °C should be passed as fast as possible. For packaging the fish a packaging material
with low oxygen permeability but high water vapour permeability was recommended. When
samples were stored in the temperature range of -30 to 2 °C colour changes were reduced with
decreasing temperature. Compared with frozen storage at -2 to -10 °C discolouration of meat
was markedly lower in chilled stored tuna. Colour deterioration of muscle in iced and frozen
stored bonito, yellow fin and skipjack tuna (Euthynnus affinis, Thunnus albacares, Katsuwonus
pelamis, respectively) caught in Seychelles waters was determined by Matthews (1983). Storage
life on ice before noticeable browning occurred was 12-14 days for yellow fin and 7 days for
bonito and skipjack. At cold storage temperature above -30 °C colour deterioration was very
rapid. Dorsal and ventral meat cuts of blue fin tuna (Thunnus thynnus), yellow fin tuna (T.
albacares), and big eye tuna (T. obesus) were stored in PVC bags in ice water or in air blast
freezer at -20 °C for 2 weeks and, in the case of blue fin tuna, at -80 °C for 2 months to
investigate discolouration profiles (Chow and others 1988). Discolouration of frozen/thawed
binnenwerk.indd 318 21-08-2006 12:35:14

Use of “filtered smoke” and carbon monoxide with fish
meat proceeded faster than that of meat in ice water storage, irrespective of tuna species. No
significant differences in discoloration were observed in dorsal and ventral meat. Discoloration
of blue fin tuna meat stored at -20 to -80 °C for 1 month was slower when storage temperature
was lower. Due to the limited possibilities of chilling and freezing with exception of very low
temperature to stabilise the colour of tuna meat further possibilities to avoid discolouration
were investigated.
Interestingly, irradiation, which was known to cause colour changes attracted attention of
researchers. Formation of met-myoglobin (met-Mb) by irradiation of meat causes obviously its
discolouration. However, regeneration of oxy-myoglobin (oxy-Mb) from met-Mb was found, too,
and seen as responsible for improvement of colour. It was unfavourable that irradiation doses
of 0.2 to 1 Mrad were needed for improvement of colour because already irradiation with a dose
> 0.2 Mrad caused deviation of the taste. Gamma irradiation converted the brown met-Mb to
a red pigment, which is similar but not identical to oxy-Mb. The colour changes in irradiated
meat vary depending on various factors such as irradiation dose, animal species, muscle type,
and packaging conditions. Irradiation increased redness regardless of pork-quality type, and
the increases were proportional to irradiation dose. Irradiation and subsequent storage of
pork improved the red colour even in PSE pork, indicating that irradiation can be used to
increase the acceptability of low-quality pork. Packaging environment is an important factor
that influences the colour of irradiated meat during storage. Irradiated pork loin muscle showed
increased redness, and the red colour was stable during refrigerated storage even under aerobic
conditions. The red colour formed in pork steaks by irradiation, however, was more intense and
stable under anaerobic than aerobic conditions. Irradiation increased the redness (a-value) of
both aerobically and vacuum packaged turkey breast, but vacuum-packaged meat was redder
than aerobically packaged meat and was stable during storage. Irradiated raw pork and turkey
became redder in anaerobic conditions, and their reflectance spectra showed that irradiation
induced the formation of an oxy-Mb-like pigment in pork. During the frozen storage, irradiation
increased pink colour in both aerobically and vacuum packaged turkey breast, and the pink
colour was stable (Nam and Ahn 2003). Colour changes in irradiated fresh meat occur because
of the susceptibility of the Mb molecule, especially the iron, to alterations in the chemical
environment and to energy input. The potential for iron electrons to exist in various states
makes the environment adjacent to the iron atom particularly vulnerable to the presence of
electron-donating compounds and high energy inputs (irradiation). Initial condition of the Mb
(Fe++, =O2, Fe+++), modification of oxidation–reduction potential of the tissue, and generation of
ligand-forming compounds (CO) from endogenous organic compounds and water are enhanced
or suppressed depending on the gas atmosphere, temperature, pH and Mb concentration of
the system. Generation of stable red pigments or brown pigments which become red over time
appears to be due to binding of irradiation-generated reactive oxygen species (·O2¯ ) or gases
(CO) which become ligands bound by iron under altered reducing conditions (Brewer 2004).
Considerable amounts of CO were produced by radiolysis of organic components in irradiated
frozen meat and poultry. CO can be produced from various types of organic compounds such as
alcohols, aldehydes, ketones, carboxylic acids, amides, and esters as a radiolytic product. The
production of CO in meat was proportional to irradiation dose in turkey breast meat. Radiolytic
hydrogen, carbon monoxide and methane are largely formed from acetone at extremely high
electron dose rates (Nam and Ahn 2003).
Already early a connection between the discolouration of tuna meat and enzymatic activity was
supposed. Following the development of a method for determination of the met-Mb reductase
binnenwerk.indd 319 21-08-2006 12:35:14

Reinhard Schubring
activity (Yamanaka and others 1973a) changes of enzymatic activity during frozen storage and
subsequent thawing of tuna meat was investigated (Yamanaka and others 1973b). During frozen
storage at –20 °C the activity of met-Mb reductase decreased rapidly with marked formation of
met-Mb. At –30 °C and –40 °C the decrease in the activity was only little and the red colour
of meat was satisfactorily retained. The activity decreased remarkably during thawing and
subsequent storage at 3 °C for 18 hours in the tuna meat stored at –20 °C and –30 °C, but
not in the meat stored at –40 °C. In the meat samples, the formation of met-Mb in the course
of thawing and subsequent storage was remarkable. These results suggest that the activity of
met-Mb reductase could be a useful index in the quality evaluation of tuna meat, especially in
respect of the colour (Yamanaka and others 1973b). While Livingston and Brown (1981) still
question the physiological importance of such enzymatic activities, it has been established
by the working group of Jiang (Pong and others 2000; Chiou and others 2001) that met-Mb
reductase influences the accumulation of met-Mb in chilled-stored tuna meat as an essential
factor. Furthermore, it was found that immersion in met-Mb reductase could extend the colour
stability of tuna meat, and that the enzymatic reduction of met-Mb by its reductase occurred
during refrigerated storage (Chiou and others 2001).
It appears to be obvious that close correlations between fat oxidation, formation of met-Mb
and colour changes of tuna meat during chilled storage which are manifested by increasing
met-Mb content, decreasing redness and decreasing acceptance of odour (Lee and others 2003)
are existing. Therefore, the addition of antioxidants for colour stabilisation seems to be useful
(Xiong and others 1993; Houben and others 2000).
Muscle pigments and their reactions
Hemoglobin is the iron-containing oxygen-transport in the red cells of the blood in mammals
and other animals. Hb transports oxygen from the lungs to the rest of the body, such as to the
muscles, where it releases the oxygen load. The name hemoglobin is the concatenation of heme
and globin, reflecting the fact that each subunit of Hb is a globular protein with an embedded
group; each heme group contains an iron atom, and this is responsible for the binding of
oxygen. The Hb molecule is an assembly of four globular protein subunits. Each subunit is
composed of a protein chain tightly associated with a non-protein heme group. A heme group
consists of an iron atom held in a heterocyclic ring, known as a porphyrin (Figure 1). This iron
atom is the site of oxygen binding. The iron atom is bonded equally to all four nitrogens in
the centre of the ring, which lie in one plane. Two additional bonds perpendicular to the plane
on each side can be formed with the iron to form the fifth and sixth positions, one connected
strongly to the protein, the other available for binding of oxygen. The iron atom can either be
in the Fe2+ or Fe3+ state, but ferrihemoglobin (met-Hb) (Fe3+) cannot bind oxygen.
In adult humans, the most common Hb type is a tetramer (which contains 4 subunit proteins),
consisting of two α and two β subunits non-covalently bound. The subunits are structurally
similar and about the same size. Each subunit has a molecular weight of about 16,000 daltons,
for a total molecular weight of the tetramer of about 64,000 daltons. The four polypeptide
chains are bound to each other by salt bridges, hydrogen bonds and hydrophobic interaction.
In the tetrameric form of normal adult Hb, the binding of oxygen is a cooperative process.
The binding affinity of Hb for oxygen is increased by the oxygen saturation of the molecule.
Hemoglobin’s affinity for oxygen is decreased in the presence of CO because both gases compete
for the same binding sites on Hb, CO binding preferentially to oxygen. Hb binding affinity for CO
binnenwerk.indd 320 21-08-2006 12:35:14

N N
Fe
N N
o- o-
o o
Figure 1. Structure of heme.
is 200 times greater than its affinity for oxygen, meaning that small amounts of CO dramatically
reduces hemoglobin’s ability to transport oxygen. When Hb combines with CO, it forms a very
bright red compound called carboxy-Hb. When inspired air contains CO levels as low as 0.02%
headache and nausea occur; if the CO concentration is increased to 0.1%, unconsciousness will
follow. In heavy smokers, up to 20% of the oxygen active sites can be blocked by CO.
Hb also has competitive binding affinity for sulfur monoxide (SO), nitrogen dioxide (NO2) and
hydrogen sulfide (H2S). The iron atom in the heme group must be in the Fe2+ oxidation state to
support oxygen transport. Oxidation to Fe3+ state converts Hb into met-Hb which cannot bind
oxygen. Nitrogen dioxide and nitrous oxide are capable of converting Hb to met-Hb.
Myoglobin, a protein with a rich and varied history in science, was distinguished as the first
protein for which a three-dimensional structure was determined, Mb has been studied extensively
in relation to its roles in O2 storage, PO2 (oxygen partial pressure) buffering and facilitated O2
diffusion. The structure of Mb was first delineated over 40 years ago (Kendrew and others 1958,
1960; Kendrew 1963). Mb is a cytoplasmic hemoprotein consisting of a single polypeptide chain
of 154 amino acids. Mb was so named because of its functional and structural similarity to Hb
(Kendrew and other 1954). Evolutionarily, Mb and Hb arose from a common ancestral gene over
500 million years ago. Like Hb, Mb reversibly binds O2 and thus may facilitate O2 transport from
red blood cells to mitochondria during periods of increased metabolic activity or serve as an O2
reservoir during hypoxic or anoxic conditions. Mb binds oxygen by its heme residue, a porphyrin
ring:iron ion complex. The polypeptide chain is folded and cradles the heme prosthetic group,
positioning it between two histidine residues. The iron ion interacts with six ligands, four of
which are provided by the nitrogen atoms of the four pyroles and share a common plane. The
imidazole side chain of His93 provides the fifth ligand, stabilizing the heme group and slightly
displacing the iron ion away from the plane of the heme. The sixth ligand position, unoccupied
in deoxymyoglobin, serves as the binding site for O2, as well as for other potential ligands
such as CO or NO. When O2 binds, the iron ion is partially pulled back toward the porphyrin
plane. Although this displacement is of little consequence in the function of monomeric Mb,
it provides the basis for the conformational changes that underlie the allosteric properties of
tetrameric Hb.
binnenwerk.indd 321 21-08-2006 12:35:15

Reinhard Schubring
Mb is perhaps best known as an O2-storage protein in muscle (Ordway and Garry 2004). This
role is especially evident in marine mammals and birds that undergo extended periods of
apnea when diving. In the absence of inspired O2, stored O2 (oxy-Mb) becomes available to
supply locomotor muscles involved in diving-related activities. The role of Mb as a store of O2
is supported by the observation that diving mammals and birds can have muscle Mb contents
that are increased 10- to 30-fold compared with those seen in animals that do not experience
prolonged apnea (Table 1).
Mb, a mobile carrier of oxygen, is developed in red muscle in response to mitochondrial demand
for oxygen and transports oxygen from the sarcolemma to the mitochondria of vertebrate heart
and red muscle cells (Wittenberg and Wittenberg 2003).
Antarctic ice-fishes of the family Channichthyidae lack blood Hb and circulating red blood
cells. Some species lack cardiac Mb as well. Mb is developed in skeletal muscle more or less in
proportion to the cytochrome oxidase content of the muscle (Figure 2).
Mb is a relatively small protein (MW 17,600) that is little affected by environmental conditions.
The molecules are slippery in the sense that they slide past one another with little frictional
interaction. Mb may reach 400–500 µmol kg-1 wet mass in skeletal muscles. Mb concentration
increases with the work to which the muscle is put and should be regarded as optimised for the
particular muscle at a particular rate of work output. Oxygen affinity is also subject to genetic
selection pressure. The oxygen affinities and oxygen dissociation rate constants of Mb from
predacious, oceanic fish that maintain muscle temperatures well above ambient are similar to
those of a related fish whose muscle operates at cool ambient temperature, when compared at
the operating temperature of the muscle (Cashon and others 1997; Marcinek and others 2001).
By acting as a scavenger of the bioactive molecule NO, oxy-Mb regulates both oxygen supply
and utilization. NO is generated continuously in the myocyte. Oxy-Mb reacts with NO to form the
innocuous product nitrate, with concomitant formation of ferric Mb, which is recycled through
the action of intracellular met-Mb reductase. Sarcoplasmic NO concentration is determined by
the balance between the rates of these two processes.
Progressive conversion of intracellular oxy-Mb to CO-Mb now abolishes about one-third of the
oxygen consumption. The oxy-Mb-dependent portion of the oxygen uptake decreases linearly
with increasing mole fraction of intracellular CO-Mb. Muscle and heart, despite access to almost
unlimited oxygen, operate in steady states close to the oxygen pressure (0.33 kPa) required for
half-saturation of sarcoplasmic Mb with oxygen (Wittenberg and Wittenberg 2003).
Table 1. Myoglobin content in skeletal muscle (Noren and Williams, modified, 1999).
Model Content (mg g-1 wet mass) Tissue
Northern elephant seal 64 Longissimus dorsi m.

Bottlenose dolphin 26 Longissimus dorsi m.
New Zealand white rabbit 8 Skeletal muscle
Mouse (129) 1.9 Skeletal muscle
Human 2.1 Psoas m.
binnenwerk.indd 322 21-08-2006 12:35:15

0.5
0.4
Myoglobin (mmol kg-1)

Horse
Horse Ox
0.3 (L. dorsi)
Elephant Pig
0.2 Sheep
0.1 Hare
0 Rabbit
1000 2000 3000
Cytochrome oxidase activity (QO2)
Figure 2. Relation of myoglobin concentration in muscles of various animals to their cytochrome oxidase
activity (Wittenberg and Wittenberg 2003).
Handling of carcasses post mortem obviously influences the relative rates of Hb and Mb. For fish
an influence of the catching method was found (Barrett and others 1965). Molecular weight
of the Mb of fish are smaller than for humans. For milkfish-Mb a Mw of 15,900 is reported that
is smaller than that of yellow fin tuna with 16,200 but larger than that for mackerel Mb and
sardine Mb with 14,900 and 14,600, respectively (Chen and Chow 2001). The molecular weight
of trout Mb was 16,017 Da based on electro spray ionisation mass spectrometry (Richards and
others 2005).
Concentration of Mb varied dependent on species. Typical concentration for longissimus dorsi

muscles are: 0.02% (rabbit), 0.06% (pig), 0.25% (sheep), 0.5% (bull) and 0.91 (blue whale).
In meat of seals even 4.8% was determined (Pegg and Shahidi 1997). In dark muscles of tuna
the Mb content was found with 1.6%, 2.3% and 1.6% for skipjack, yellow fin and skipjack,
respectively, while those found in the ordinary muscles of the same species were smaller with
0.05%, 0.06% and 0.2% (Kanoh and others 1986). Beside species and muscle groups further
factors, like age, sex, stress, nutrition and environment influence the Mb content (Livingston
and Brown 1981; Pegg and Shahidi 1997).
Structure and chemistry of the iron in heme molecule is the key of understanding the reactions
and colour changes taking place at Mb as mentioned above. Detailed information are given in
reviews published by Giddings (1977), Livingston and Brown (1981), Pegg and Shahidi (1997),
Nam and Ahn (2003) and Brewer (2004). Table 2 shows the large variety of colour properties
of Mb depending on the chemical state of iron atom and a colour-generating ligand and Table
3 the colour and colour values of derivates of milkfish Mb.
While red colour of meat results mainly from the co-operation between oxy-Mb and deoxy-
Mb, discolouration is chiefly caused by oxidation from Mb to met-Mb (Nam and Ahn 2003;
Livingston and Brown 1981). To stabilise Mb in the Fe2+ form NO-Mb as well as CO-Mb are
binnenwerk.indd 323 21-08-2006 12:35:15

Reinhard Schubring
Table 2. The colour properties of myoglobin depending on the chemical state of iron atom and a colour-
generating ligand (Nam and Ahn 2003).
Heme protein Heme iron Ligand Colour Name
Globin Ferric H2O Brown Metmyoglobin

Globin Ferrous O2 Bright red Oxymyoglobin
Globin Ferrous - Purple Deoxymyoglobin
Globin Ferrous NO Pink NO-myoglobin
Globin Ferric NO Brown NO-metmyoglobin
Globin Ferrous CO Pink CO-myoglobin
Globin Ferrous S Green Sulfmyglobin
- Ferrous NO Pink Hemochrome
- Ferric NO Brown Hemichrome
Denatured globin Ferrous NO Pink Nitrosyl hemochrome
Denatured globin Ferric NO Brown Denatured hemichrome
Denatured globin Ferrous CO Pink Denatured CO-hemochrome
Table 3. Colour and tristimulus values (Hunter L, a and b values) of milkfish Mb derivates1 in 50 mM sodium
phosphate buffer (pH 6.25) (Chen and Chow 2001).
Derivates Colour Hunter’s colour values
L a b
Deoxy Mb magenta 57.1 26.6 4.6

Oxy Mb bright red 58.8 29.2 15.2
Carbonyl Mb deep pink 57.7 29.9 6.4
Met Mb brown 66.9 11.5 18.1
Cyanmet Mb orange 54.2 13.7 15.2
1 protein concentration: 0.802 mg ml-1
used (Yamanaka 1973b). CO binds very strong to Mb. This is documented by its high affinity
and very low dissociation behaviour. For example, the affinity of CO to Mb is figured 100 times
higher than that of met-Mb (Nam and Ahn 2003). CO dissociates 1000 times slower than
oxygen from Mb (Livingston and Brown 1981). In Hb, affinity of CO excels that of oxygen by
twohundertfold. Against heme the affinity of CO is even 25,000 times stronger than that of
oxygen. This possibility, to transfer Mb by addition of CO in a less reactive state compared
with the interaction between oxygen and Mb under simultaneous stabilisation of colour of
fresh meat, has been found already years ago so attractive that several technological variants
of practice-oriented use have been developed.
binnenwerk.indd 324 21-08-2006 12:35:16

Use of carbon monoxide for influencing the colour of meat and fish
Packaging in modified atmosphere

According to Church (1994) the stabilising effect of CO on the colour of meat has been known
and also patented for more than 100 year. One of the first applications reported by Wolfe
(1980) was the use of CO as part of protective gases for meat and fishery products. With the
introduction of reduced oxygen packaging and modified atmospheric packaging (MAP) during
the 1960’s, it became more obvious that product exposure to gas blends with carbon monoxide
(CO) yielded favourable colour development and retention (Otwell and others 2003). In a
review (Sørheim and others 1997) on technological, hygienic and toxicological aspects of the
use of CO as part of protective gases for packaging of meat in modified atmosphere CO2 (60-
70%), N2 (30-40%) and CO (<0.5%) is identified as an appropriate gas blend. The shelf life of
meat packaged in the CO mixture is longer than that of meat packaged in the commonly used
atmospheres with high oxygen (O2) content (approximately 70% O2 and 30% CO2). A possible
negative aspect of using CO in the MAP of retail meat is concern that consumers might misjudge
the quality of a product, because its true microbiological status may be masked by its stable,
cherry red carboxy-Mb colour. However according to the authors, consumers will be able to
detect spoilage by the presence of off-odours. At the current low concentrations, < 0.5%, CO per
se seems to have no or only minor effects on bacteria and the shelf life of the meat. Furthermore
it is pointed out by the authors that CO containing gas blends are in use in the Norwegian
meat industry since mid of 1980s and had a marked share of 50 to 60% at this time (1997).
These positive aspects were supported by further investigations of the same authors (Sørheim
and others 1999). Ground beef, beef loin steaks and pork chops were packaged in modified
atmospheres of 0.4% CO/60% CO2/40% N2 and 70% O2/30% CO2. The packs were stored in the
dark at 4 or 8 °C for up to 21 days. Meat in 0.4% CO/60% CO2/40% N2 had a stable bright red
colour that lasted beyond the time of spoilage. The storage lives in this gas mixture at 4 °C,
as limited by off-odours, were 11, 14 and 21 days for ground beef, beef loin steaks and pork
chops, respectively. The 70% O2/30% CO2 atmosphere resulted in an initially bright red to red
colour of the meat, but the colour was unstable and off-odours developed rapidly. Despite this
obviously scientific-based statements the MAP packaging of fresh meat under addition of CO
was until recently only in Norway permitted by food law. To enter the EU market with such
products, the Norwegian Meat Co-operative and the Norwegian Independent Meat Association,
representing the Norwegian meat industry, have applied for the use of a gas mixture containing
60%-70% CO2, 30%-40% nitrogen N2 and < 0.5% CO as components of the packaging gas in
MAP for fresh red meat (mainly beef, pork, lamb but also horse, goat, reindeer, game etc.). For
clarification the Scientific Committee on Food (SCF) was asked to evaluate the safety of CO as
a packaging gas for meat in a mixture with CO2 and N2 by the European Commission (EC). In
their Opinion adopted on 13 December 2001, the SCF concluded that (a) meat packaged in MAP
containing a high concentration of CO2 and 0.3%-0.5% CO remains microbiologically stable for
11-21 days when stored at a maximum temperature of 4 °C, (b) the use of meats packaged in
MAP containing 0.3%-0.5% CO contributes in negligible amounts to the overall exposure to CO
and the CO-Hb concentration in humans, (c) there is no health concern associated with the use
of 0.3%-0.5% CO in a gas mixture with CO2 and N2 as a MAP gas for fresh meat provided the
temperature during storage and transport does not exceed 4 °C. However the Committee wished
to point out that, should products be stored under inappropriate conditions, the presence of
CO may mask visual evidence of spoilage (European Commission 2001). Therefore no permission
was given by EC to the Norwegian meat industry to enter their products on EU markets.
binnenwerk.indd 325 21-08-2006 12:35:16

Reinhard Schubring
In the USA the situation had changed and the use of CO as component of packaging gases was
also not permitted. Since February 2002 the FDA (Food and Drug Administration) has CO accepted
as GRAS (generally recognised as safe) for the use as a component of a gas mixture in a MAP
system with an level of 0.4%. The other components of the MAP system which should be used
for packaging fresh cuts of case ready muscle meat and ground case ready meat are CO2 (30%)
and N2 (69.4%). However, the case ready meats would be removed from the MAP system prior to
retail display (FDA 2002). Investigations performed under these conditions were reported by Hunt
and others (2004). Colour and microbiology of steaks and ground beef stored in 0.4% CO, 30%
CO2, and 69.6% N2 (but removed from the modified atmosphere before display) were compared
with product displayed immediately after packaging in polyvinyl chloride film (only atmospheric
oxygen). Compared with products exposed to atmospheric oxygen only in ground beef longissimus
dorsi muscle and outside semimembranosus muscle colour and shelf life will decrease during
prolonged MAP storage, whereas in psoas major muscle and inside semimembranosus muscle the
properties of colour and shelf life are either not affected or will increase slightly. Nevertheless, the
increase in colour life due to the addition of CO to MAP of beef is not drastic enough to prevent
the negative effects of abusive storage temperatures and extended storage times. CO addition to
MAP of beef will not mask product spoilage and microbial unsoundness in beef. In between FDA
has extended their permission for using CO in MAP of fresh meat (FDA 2004, 2005) in that the
meat is placed on a tray within a chamber, the chamber is then filled with the desired atmosphere,
and finally, a barrier film is affixed to the package. The packages are then labelled with a validated
open date code at a central location and will be subject to no further processing or manipulation
at retail. The open date code established for products packed in the MAP system will not exceed
35 days following the date of packaging for intact muscle cuts and 28 days for ground beef.
However, it appears that this acceptance of CO application for processing MAP packaged fresh
meat by FDA is not uncontradicted in the USA. Very recently a Citizen Petition requesting FDA
to enforce ban on CO gas in fresh meat packaging has been launched by Kalsec, Inc. (Kalsec
2005). Kalsec are producers of spice, herb, hop, and vegetable extracts for use in food, beverage,
and pharmaceutical products. This Citizen Petition requests that FDA take immediate action to
enforce a ban on CO in fresh meat packaging, and specifically, to terminate the agency’s unlawful
acceptance of the GRAS notifications submitted by GRAS Notice Nos. GRN 000083 and 000143.
According to Kalsec “the ban requested by this Citizen Petition is necessary to prevent serious
food safety harms to the public, and preserve consumer confidence in the safety and integrity of
the U.S. meat supply. Moreover, FDA is obligated to enforce the ban requested under the Federal
Food, Drug and Cosmetic Act and current FDA regulations, as a matter of law. The use of CO gas
in fresh meat packaging produces an artificially intense, persistent red colour in meat that can
simulate the look of fresh meat and mask the natural signs of aging and spoilage that consumers
depend upon in making safe food choices, including browning and tell-tale odours. Consumers
have no way to tell the difference between meat packaged with CO gas that may merely look fresh
and safe, and genuinely fresh and wholesome meat. As a result, CO presents serious consumer
deception and food safety risks which jeopardize the public health” (Kalsec 2005).
In contrast to meat, no scientific investigations are known which CO use as component of MAP
packaging in fish (Sivertsik and others 2002). This review cited a reference (Rosnes and others
1998) which reported the successful application of small amounts of CO (0.2 to 1.0%) in gas
blends together with CO2 and N2 to avoid discolouration of gills of salmon. The main reason
of lacking commercial attempts may be seen in that mainly lean fish and consequently white-
fleshed fish is used for MAP packaging. On the other hand, when salmon is used colouration is
based not on Mb as discussed above.
binnenwerk.indd 326 21-08-2006 12:35:16

Application of carbon monoxide on fish via modified smoke processing using traditional
and modified processing methods
Smoked fish is by definition fish and part of fish which are treated by wood smoke for
characteristic colour and flavour. Aims of both principal variants of the traditional smoking
process include uptake of smoke and drying when cold smoking at temperatures < 30 °C as
well as hot smoking where a thermal treatment (cooking by air) additionally takes place. In
both cases results a significant reduction of moisture. Actual reviews on processes applied
world-wide as well as in Germany are given by Doe (1999) and Tülsner (1994). However, since
the beginning of this millennium a new technology keeps the fish processing industry and
traders busy. Announcements in newspapers and TV rattle consumers more often than giving
appropriate information. Though the legal position appears to be clear, possibly caused by
lacking appropriate analytical methods for determination of CO in fish products, the food control
laboratories did not push it with the necessary consequence.
Therefore it appears necessary to comment on the problem from the scientific standpoint. What
is special on the modified smoking methods? As mentioned above, except the use of liquid
smoke smoking is seen as treatment of the food with smoke that resulted from burning the
wood. Smoke is an emulsion of droplets in a continuous phase of air and vapours stabilised by
electrostatic charges on the droplets. For flavouring, colouring and microbistatic purposes, the
vapours are of greatest importance in smoking (Horner 1997). Over 200 components have been
identified in the vapours including CO which generation is amounted to 80 to 370 g/kg wood
(Larson and Koenig 1993). In traditional smoking the colouring effect is obviously caused by
other smoke components. Colour imparted to the fish by the smoking process is due to carbonyl-
amino reactions of the Maillard type and has been correlated with a quantitative decrease in
carbonyl groups in the smoke. Brown pigment forming in the surface tissue were said to inhibit
further penetration of carbonylic groups and other smoke components to underlying tissues.
Ribose from the degradation of nucleotide and ribonucleic acids and free amino compounds ,
such as anserine and taurin, in the fish muscle extractives also contribute to browning at the
surface of drying fish. It was found that as the fish muscle spoils, β-alanin-1-methyl histidine
and lysine become increasingly important in the extent of browning. Thus for a standard
smoking operation, the condition and extent of spoilage of the raw material can vary the extent
of brown colour formation (Horner 1997).
After almost complete removal of all others colour-active components from smoke only CO
is as reaction partner at disposal and imparts, as discussed above, the typical colour to the
fish flesh. On this knowledge several processing technologies are based. According to Otwell
and others (2003) the first patent that dealt with colour stabilisation by use of CO has been
granted to Woodruff and Silliker (1985) in the USA. It claims that good colour in fresh meat,
fresh poultry, and fresh fish is established and maintained by subjecting such meat, poultry
and fish to an atmosphere containing a low oxygen concentration to convert oxy-Mb on the
surface of the meat and poultry to reduced Mb, and both oxy-Mb and oxy-Hb in fish to reduced
Mb/Hb, respectively, then subjecting the fresh meat, fresh poultry and fresh fish to a modified
atmosphere containing a small amount of CO to convert the reduced Mb to CO-Mb to a depth
of not more than about 0.375 inch below the surface of the meat and poultry, and to convert
the reduced Mb/Hb to reduced CO-Mb/CO-Hb in the fish. The modified atmosphere is a new
composition of matter. During or after the conversion, the fresh meat, fresh poultry and fresh
fish may be maintained at temperatures above freezing in an atmosphere that contains more
than about 10% CO2 by volume to inhibit bacterial growth, or, alternatively, the fresh meat,
binnenwerk.indd 327 21-08-2006 12:35:16

Reinhard Schubring
fresh poultry and fresh fish may be frozen and maintained frozen in normal air atmosphere.
Another patent was granted to Yamaoka and others (1996) that claims that raw fish and
meat are smoked to sterilize and prevent decomposition and discoloration without losing
their freshness. The smoked fish and meat pick up agreeable taste and flavour, and remain as
wholesome as fresh ones when kept at easily obtainable cold-storage or freezing temperatures,
even during long transportation. The smoke generated by burning a smoking material at 250
to 400 °C is passed through a filter to remove tar. The smoke retaining ingredients, exerting
highly preservative and sterilizing actions passed through the filter, are cooled to between 0
and 5 °C in a cooling unit. Fish or meat is processed by exposure to the smoke at the extra-low
temperature thus obtained.
The probably most used patent was granted to Kowalski (1999) that claimed the manufacture
of tasteless super-purified smoke to treat seafood and meat to preserve the freshness, colour,
texture, and natural flavour, particularly after the food is frozen and thawed. The smoke is
generated by burning an organic smoking material at preferably 260 to 571 °C in a smoke
generator. It is then passed through a precipitation filtering tower comprised of filters of ice,
cloth, and activated carbon to remove taste imparting, and carcinogenic, particulates and
vapours. The super-purified smoke is then stored and aged in a temporary pressure pot or in
canisters for treatment at the same time or at another place and time. The super-purified smoke
is used to treat seafood or meat in plastic bags at temperatures between its variable freezing
point and 7.8 °C for 12 to 48 hours, or until the desired effect is achieved. The product is then
frozen, stored for up to one year, and quickly or slowly thawed with little degradation of the
treated seafood or meat.
The last patent relevant to this subject has been granted to Olson and Brinsmade (2004) and
claimed a seafood preservation process in that preservation of meat products is accomplished
utilizing a combination of smoke, ozone and freezing preservation techniques. Particularly,
fish products are sized into portions that are first treated with smoke, followed by treatment
with ozone and then optionally frozen. The preservation system extends the shelf life of
the fish products and permits the fish to maintain its freshness and freedom from bacterial
decomposition for a longer period of time following catch. The preservation process further
maintains the characteristics of day caught fish, such as taste, texture and colour, making the
refreshed fish products produced by the present system more appealing to consumers.
Using the description in the patent of “tasteless smoke” process (Kowalski 1999) as an example
the processing steps and specialities should be explained. “The intention of treatment with
our tasteless super-purified smoke is to preserve the vitality of the seafood so it appears and
tastes similar to fresh after it is frozen and thawed. In all cases seafood should be wholesome.
However, seafood that is consumed uncooked for sashimi must be visually attractive in its raw
form. The purpose of improving the aesthetic qualities is to create a seafood product which
is visually suitable for sashimi after freezing and thawing. The result will be a sashimi quality
seafood product delivered to the consumer equal to, or superior to, fresh seafood in regards to
vitality, quality, safety, and convenience”.
The manufacturing process starts with the smoke generating part of the apparatus using a
natural gas or electric burner to combust wood sawdust packed into a multiple cylinder retort
at temperatures in an operable range of 204 to 510 °C. The pyrolysis of the wood sawdust
into smoke creates by-products of tar, moisture, and particulate residue at the outlet of the
binnenwerk.indd 328 21-08-2006 12:35:17

smoke generating subsystem. These by-products are collected in liquid form in a tar/moisture/
residue condensation chamber and drained out a purge valve near the end of the process. The
smoke is next super-purified such that the phenols in both particulate and gaseous vapour
phases are reduced to concentrations below recognition thresholds for odour and taste that
impart a smoked flavour to the treated food. The smoke is most efficiently super-purified by
flowing through a precipitation tower which washes and filters the smoke through ice and
a combination of adsorbent and molecular sieve filters of cloth and activated carbon. This
washed smoke next passes through activated carbon and cloth filters to adsorb odour and taste
imparting phenols and other carcinogenic particulates and gases. At this point, substantially
tasteless super-purified smoke can be used to directly flood a smoking treatment chamber
filled with seafood or other meats to produce an acceptable result. Alternatively, the smoke
can be pumped into an expandable, or fixed, storage chamber for short term storage, or into
a canister for long term storage. If immediate treatment of seafood by the super-purified
smoke is desired directly from the inner accordion bladder, the aging time is in an operable
range of 1 to 72 hours, and an optimal range of 24 to 48 hours. If treatment at another time
or place from storage canisters is desired, the aging time is in an operable range of greater
than 1 hour, and an optimal range of 2 weeks to 6 months. If immediate treatment is desired,
each whole tuna or other seafood species is taken from cold storage, filleted into loins, and
then filleted into sashimi slices and steaks (smaller fish can be treated whole). Sashimi slices
are placed in a dipping solution to stabilize colour, enhance flavour and firm the texture of
the fish. Steaks, which are ultimately cooked and not consumed raw, do not require this step.
The filleted fish is then placed in plastic bags. The air in each bag is substantially removed,
a hose and dispensing nozzle from the pressure pot are inserted, and the valve is opened to
flush the seafood with an operable rage of 0.05:1 or greater, and an optimal range of 1.5:1 to
20:1 ratios of volume of tasteless, super-purified smoke to volume of seafood. The bag is then
sealed. The super-purified smoke treatment occurs until the desired penetration of tasteless
super-purified smoke into the fish is complete. This desired penetration is complete after
approximately 12 to 48 hours. The minimum temperature during treatment varies with the type
of seafood being treated and is approximately 0.1 °C above its variable freezing point. The
treatment temperature is an operable range from above the variable seafood freezing point to
7.8 °C, and an optimal range from above the variable seafood freezing point to 1.7 °C. The
carboxymyoglobin, nitric oxide myoglobin, and nitrogen dioxide myoglobin present in the
treated product, as well as absorbed CO2 and phenols from the super-purified smoke, result
in both stable organoleptic freshness characteristics and stable red colour after freezing and
thawing. Such product will organoleptically stay fresh longer than untreated product before
decomposition begins. However, this characteristic of enhanced freshness longevity does not
often come into play since the product is thawed in small quantities only as needed and does
not require extended shelf life in its thawed state. Figure 3 shows an preferred embodiment of
a tasteless super-purified smoke manufacturing apparatus used in the practiced process.
In the latest patent the effect of purified smoke is supported by treatment with ozone (Olson
and Brinsmade 2004). This technology should be characterised briefly in the following. In
general, the process includes the steps of smoking of fresh fish, treating it with ozone and
optionally freezing the fish. When a smoke and ozone process is utilized, the shelf life is
extended and the fish retains more of its “fresh” colour. The smoke/ozone process retains the
“fresh” colour and extends shelf life of the fish flesh by binding the CO molecule to the heme
pigment in the Hb molecule in such a way that it takes much greater than normal oxidative force
to oxidize the Hb molecule. Furthermore, the smoke/ozone process aids in the prevention of
binnenwerk.indd 329 21-08-2006 12:35:17

Reinhard Schubring
precipitation tower
smoke generator
storage chamber
Figure 3. Preferred embodiment of a tasteless super-purified smoke manufacturing apparatus (Kowalski

1999, modified).
bacterial decomposition and maintains the Hb molecule (red colour) during freezing and frozen
storage by binding it with a CO molecule. Optionally, the smoke/ozone process can include the
steps of wiping the flesh of the fish with alcohol one to three times during the preservation
process, before or after smoking the fish. The application of alcohol to the exterior of the fish
kills surface and shallow bacteria on contact with the alcohol. The fish would be placed in a
modified “smoke” atmosphere for 1 to 72 hours, with the length of time depending on the
thickness of the fish product, with thicker products requiring more time than thinner products.
If the smoke is applied to the fish while in a vacuum chamber, the time required for the smoke
application can be reduced to less than a minute. During the smoking step, a vast majority of
“aerobic” bacteria die as there is no oxygen available for them to survive. The smoking step
additionally creates an acidic pH in the fish by the dissolution of free CO2, present in the
smoke, into the fish. The acidic pH prevents the growth of bacteria during the “fresh” stages
of the process. An optional final step can involve freezing of the product to kill an additional
percentage of the bacteria present on the product. The fish product can be initially prepared
into appropriately sized sections or fillets in order to accelerate the smoke/ozone application
steps. With the use of this fish preservation process, the shelf life of the product is increased,
usually from about 2-3 days after the product is landed to about 10-12 days. This increase in
shelf life after the product has been treated allows the product to be shipped to remote areas
requiring longer shipping times. Also, the final processing of the product into consumer-ready
forms, including cutting, portioning and packing the product, can be performed at the central
processing facility. This avoids the necessity of having to perform the final processing of the
product at the store level.
binnenwerk.indd 330 21-08-2006 12:35:17

Different objectives of traditional and modified smoke processes
According to the Guidelines of the German Food Code (Anonymous 2003) smoked fish is a
product of differently pre-processed fresh fish, deep frozen fish or parts of fish, salt fish or
slightly salted fish and parts of fish which are processed by treatment with freshly generated
smoke. Hot smoked fish has depending on the species used a concolorus surface (golden to
brown). Skin should be undamaged, translucent or silky and shiny. Meat colour is according to
species bright however for tuna reddish to dark red and for salmon and salmon trout slightly
pink, orange to red. Cold smoked fish should be look according to species uniformly golden
to brown with skin being undamaged. Colour of meat is expected to be bright to brownish.
For salmon and salmon trout the meat should be slightly pink, orange to red. Odour and taste
should be species-specific, smoky and salty.
When reading the following statement given by the Cryofresh Co., one user of the “tasteless”
smoke processing it becomes clear that the objectives compared to traditional smoking are
completely different: “ The tasteless smoke is intended to be used on raw seafood, such as
tuna and salmon, before it is frozen. The tasteless smoke is added to preserve the taste, aroma,
texture and colour of the frozen seafood. Without the addition of tasteless smoke, frozen
tuna and other red-meat seafood is prone to browning, the development of off-odours and
decreased palatability during freezing” (http://www.cryofresh.com/fda.doc). The treatment
does not intend to produce smoked fishery products but tries to conserve quality properties
during freezing and frozen storage. Therefore, it is failed to compare “filtered smoke” treated
products with those processed in the traditional way of smoking.
Scientific background of the application of carbon monoxide and filtered smoke

on fish
While the effect of CO as component of gas blends used for MAP packaging of meat, where
it is comparable with filtered smoke regarding to its role in stabilisation of colour, has been
investigated extensively (Gee and Brown 1978; Wolfe 1980; Aasgaard 1993; Church 1994;
Sørheim and others 1997, 1999; Luno 1998; Cornforth 1998; Klettner 2001; Kusmider 2002;
Narasimha Rao and Sachindra 2002; Krause 2003; Hunt 2004, John and others 2005), scientific
papers on application of filtered smoke or CO on fish are rather scarce. In East Asia were
obviously some scientific activities initiated by the ban of imports of CO treated tilapia (Tilapia
spp., Oreochromis spp.) through the Japanese government 1997 (http://www.seafoodbusiness.
com/buyguide/issue_tilapia.htm).
These papers (Abe and others 1994; Takeda and others 1995; Miyazaki and others 1997; Chow
and others 1997, 1998a, 1998b; Hsieh and others 1998; Yoshida and Hori 1998) deal mainly
with the determination of CO in fish and were unfortunately mostly written in Japanese. Also the
probably most often applied method for CO determination originated from that time (Ishiwata
and others 1996). For determination of CO in fish flesh this was homogenised with water under
ice cooling and the homogenate was centrifuged. After separating the supernatant sulphuric
acid was added and the mixture shaken vigorously. Part of the headspace gas was injected into
the gas chromatograph. CO was separated from other gases on the column and was reduced
to methane in the methaniser for detection by FID. The retention time and peak area were
compared with those obtained with the calibration CO gas. The recovery of CO from Tilapia was
85.2% and that from tuna 92.2%. The detection limit of this method was 2 μg of CO/kg fish
binnenwerk.indd 331 21-08-2006 12:35:17

Reinhard Schubring
flesh (Ishiwata and others 1996). It has been reported recently (Jonker and others 2003) that
the CO content in tuna can exactly and reproducibly be determined by using this method. If the
determination should be repeated the use of same homogenate within a very short period of
time is recommended. In medical research comparable methods are used to determine e.g. the
CO content of blood (Iffland and others 1989; Widdop 2002).
The same detection limit as mentioned above was also obtained by Miyazaki and others (1997)
with their method developed for the analysis of CO in fish meat. The frozen (or raw) sample was
minced and weighed into a screw cap vial. After addition of l-octanol and 10% sulfuric acid,
the vial was incubated at 40 °C in a water bath for 5 min, then shaken vigorously for 15 min
to release the CO. CO was analyzed by head-space gas chromatography. Gas chromatography
equipped with a reducing column after the main column to change CO to methane before entering
the FID, was used to measure the CO residue (Chow and others 1998a). Several tuna flesh having
different grade of red colour were treated with CO for 5 days. Higher CO residue was found in
the flesh that contained more Mb. The molar ratio of CO to Mb was about 11-13%. In order to
understand the characteristics of the reaction between CO and Mb in tuna flesh, yellow fin tuna
steaks (thickness 1.5-2.0 cm) were treated with CO (Chow and others 1997). After treatment,
the absorption spectrum of the Mb and the penetration of CO into the tuna flesh were analyzed.
The specific spectrum of CO-Mb within the visible range was observed after bubbling CO directly
into Mb extracts for 10 seconds. However, the characteristic CO-Mb spectrum was not found for
the Mb extracts from the tuna steaks treated with CO for 30 min. Neither was it found for the
extracts from tuna steaks treated with CO for 120 hours. The penetration of CO into the tuna flesh
was very slow. It took about 1-4 hours for CO to penetrate to 2 to 4 mm under the surface. The
penetration time to reach 4 to 6 mm under the surface was about 8 hours. The Mb extracts from
tuna flesh treated with CO had higher absorbance at 570 nm than at 580 nm. This characteristic
could be used to determine whether a sample had been treated with CO. The big eye tuna steaks
(ca. 8×6×2 cm), after being treated with CO for 5 days, were stored together with the untreated
control fish steaks under two different conditions, in ice at 0 °C for 7 days and frozen stored at
-20 °C for 6 months (Chow and others 1998b). For each storage condition, the changes in met-
Mb formation, colour appearance and CO residue in the surface layer (2 mm thick), the middle
layer (2-4 mm from surface) and the inner layer (4-6 mm from surface) of the CO-treated and
the control were measured and compared. The results revealed that, during the 7-day ice storage
period, the met-Mb values in the three layers of the CO-treated tuna steak increased gradually,
but all remained below 10%. For the control, the met-Mb in the middle and inner layers were as
high as 50% before storage, obviously being higher than that of the surface layer, but increased
only slowly in all three layers throughout the iced storage. In views of the measured colour
values (Hunter L, a and b) among the three layers of tuna steaks, the +a value of the CO-treated
steak was lower than that of the control. The amount of the CO residue in the CO-treated steak
decreased drastically during iced storage and became undetectable in both the middle and the
inner layers after 7 days of storage. In the case of frozen storage at -20 °C for 6 months, the
values of met-Mb and Hunter a for CO-treated steak remained almost constant, whereas the
value of met-Mb of control increased gradually up to 40-60% while the +a value decreased from
11 to 5. The CO residue in the surface and middle layers decreased during the early stage of the
six-month freezing storage followed by no apparent change. There was no apparent variation in
CO residue in the inner layer of tuna steak throughout the frozen storage.
The yellow fin tuna steaks (ca. 8×5×2 cm) sealed in a package and treated with 99.5% CO at
4 °C for 5 days were compared with the control steaks without CO treatment stored at same
binnenwerk.indd 332 21-08-2006 12:35:17

condition (Hsieh and others 1998). The surface layer (2 mm thick), middle layer (2-4 mm from
surface) and inner layer (4-6 mm from surface) were separately cut from the fish steak. For
each layer, the Hunter’s colour value (L, a, b), met-Mb formation, total volatile basic nitrogen
(TVB-N) and CO residue were measured and compared. The results revealed that Hunter +a value
of tuna steaks treated with CO for 4 h increased higher than that of the control steak for all
three layers. However the values of L and +b between the CO-treated steak and the control
steak differed slightly. The met-Mb in the surface layer showed no significant difference with
or without the CO treatment whereas the met-Mb in the middle and inner layers of CO-treated
steak were apparently lower than that of the control steak. The TVB-N of the steaks remained
below 20 mg/100 g during the 5-day chilling storage regardless of CO treatment. The amount of
CO residue in the surface layer increased rapidly in 24 hours of the treatment and consequently
reached the highest value at 96 hours, while those in the middle and inner layers increased
quickly in 48 hours. The CO residue in the inner layer remained constant after 48 hours of
treatment at a value about a half as high as that in the surface layer.
In 2003, objectives and first results of US researchers were presented. Otwell and others
(2003) defined the open questions, that are to be answered by research as follows: Can inferior
fish quality, as denoted by development or loss of particular meat colours, be reversed by
CO applications that enhance product appearance? What degree of transition is possible and
does this involve potential food safety concerns? Can controls be developed to monitor CO
applications to suit commercial and regulatory guidelines for product quality and safety? Would
these controls vary by fish species, product types (fillets vs. whole fish) and product forms (fresh
vs. frozen)? Kristinsson and others (2003) demonstrated that CO significantly stabilized tilapia
Hb in terms of denaturation. CO-Hb also was significantly more stable towards autoxidation (and
thus colour changes) and had decreased pro-oxidative activity. Fish treated with CO and filtered
smoke (FS) had significantly improved colour on treatment and colour stability during storage.
Using digital machine vision analysis a detailed colour analysis could be achieved identifying
representative colour and colour changes for each treatment. This was directly related to the
heme proteins stability which was enhanced on CO/FS treatment. Pure CO had a more dramatic
effect on colour compared to FS. Lipid oxidation was reduced for some of the CO/FS treatments,
however the treatments had little effect on histamine formation which could be of concern.
When complexed with CO Hb was substantially stabilized compared to oxy-Hb and met-Hb (which
was the least stable). CO-Hb exhibited greatly increased stability towards thermal denaturation,
was able to largely retain its structure down to pH 3.5 and up to pH 12 and was more resistant
to chemical denaturation. UV- visible spectroscopy results show that this stabilization comes
from the ability of CO to stabilize the heme environment in Hb. CO-Hb did also have substantially
higher oxidative stability (i.e. resisted the formation of met-Hb) and thus colour stability
compared to oxy-Hb at all temperatures and pH’s tested. This oxidative stability was greatly
enhanced at pH 8 vs. pH 7 and 6 also was significantly more stable towards autoxidation due
to an enhancement in the binding of CO in the heme pocket of Hb. Complexing Hb with CO
also lowered its ability to oxidize fish membrane lipids, thus serving an anti-oxidative purpose.
This lower pro-oxidative ability is likely due to increased stability towards autoxidation of the
CO-Hb compared to oxy-Hb. Fish treated with CO and FS had significantly enhanced a-values on
treatment, which was favoured by sensory panellists over untreated control. Gas treatment led
to significant colour stabilization on freezing, thawing and cold storage compared to untreated
controls. Colour enhancement and stability was directly proportional to the percent CO used to
treat the fish and subsequently the stability of heme proteins after treatment as assessed by Hb’s
UV-visible spectra. Employing digital machine vision analysis it was possible to obtain a detailed
binnenwerk.indd 333 21-08-2006 12:35:18

Reinhard Schubring
colour analysis and colour stability kinetic analysis by identifying classes of colours and hues
representative of each treatment. Lipid oxidation was reduced for the CO/FS treatments except
for the 4% CO treatment. The increased oxidative stability correlated with increased stability of
the heme proteins in the muscle to autoxidation. CO and FS treatment lowered aerobic microbial
counts and extended microbial shelf life of both mahi mahi and tuna, likely as a result of
reduction in oxygen. CO/FS treated samples developed histamine both during gas treatment and
on storage, however histamine developed slower as percent CO increased, indicating an effect
on histidine forming micro organisms.
Results obtained since than can be derived from the abstract of presentations at IFT Annual
Meetings 2004 and 2005. Mony and Kristinsson (2004) reported that CO-Hb was to be substantially
more stable than oxy-Hb at all pH values and temperatures tested, as the CO molecule greatly
stabilized the heme group from oxidation. As pH increased the stability towards autoxidation
increased for both Hb types. Thermal-denaturation experiments revealed that CO-Hb was far
by the most stable of the three types tested, followed by Oxy-Hb, with met-Hb being the least
stable. Conformational results suggest that stabilization of the heme environment dictated
the onset of overall protein unfolding and aggregation. These results suggest that CO binding
stabilizes the heme group and thereby the whole protein molecule. This suggests that treatment
with CO could be useful in prolonging shelf-life of fish muscle. Ludlow and others (2004) found
that CO and FS treatments significantly altered red colour of tuna muscle. Red colour stability
was greatly affected during cold storage after thawing. Colour stability was directly related to
the amount of CO bound to the heme proteins and their oxidative stability, which was increased
on CO/FS treatment. Lipid oxidation was reduced by the CO/FS treatments compared to control.
Only a 100% CO treatment led to a significant delay in microbial growth after thawing. Freezing
inhibited histamine formation, while abuse studies indicated that high levels of histamine
can form after CO treatment while colour is still acceptable. At the same CO percentage FS did
not show any benefit over a commercial gas mixture. The results indicate that certain CO gas
mixtures may extend shelf life of tuna, while the possibility of abuse remains and warrants
further research. Comparing the quality profiles, untreated fresh tuna steaks was found to be
significantly higher in TVB-N, TMA-N, and bacterial numbers over CO treated ones stored on
ice, but the sensory rejection point of both was similar (Leydon and others 2004). The notable
exception was lipid oxidation which was lower for treated tuna at all storage temperatures.
Colour of treated tuna remained constant for all temperatures evaluated. Indicators evaluated
for chemical, microbiological and sensory quality showed that treated and untreated tuna steaks
had comparable profiles with similar trends. However, results showed that oxidation is retarded
by the treatment. Also when using mahi mahi fillet CO/FS treatment led to a significant increase
in redness (a*) but had less effect on lightness (L*) and yellowness (b*). Increase and stability
in a* was proportional to % CO and amount of CO bound to heme proteins. Lipid oxidation
was reduced for the CO/FS treatments compared to control, except for 4% CO. Treatment with
CO gas mixtures can significantly extend quality of mahi mahi (Demir and others 2004). As in
mahi mahi also in Spanish mackerel treatment by CO significantly (p<0.001) influenced a* value
on treatment and red colour stability on storage. This stability was directly correlated with
the binding of CO to heme proteins in muscle, with 100% CO treatment giving most binding.
Development of b* was minimized for the samples with most red colour stability. Lipid oxidation
development was retarded on treatment with CO, most for the 100% CO treatment, while it was
significantly suppressed for all samples when kept in the gases for 8 days. Treatment with CO
led to a significant delay in microbial growth on storage. Therefore, it is stated that CO and FS
can have multiple positive effects on the quality of fatty species rich in dark muscle (Garner and
binnenwerk.indd 334 21-08-2006 12:35:18

Kristinsson 2004). When comparing changes in the quality profiles of FS treated and untreated
(UT) tilapia fillets (Oreochromis spp.) differences (p<0.05) were observed in sensory assessment
(Leydon and others 2005), where FS samples held at refrigerated temperature and on ice were
rejected sooner. While microbiological assessment generally showed no differences for FS and
UT product samples, biochemical markers at time zero showed FS samples significantly higher in
apparent ammonia and TVB-N. Ammonia and TVB-N continued to be significantly higher over 6-9
days of storage for iced samples. At time zero, TBARS and POV values were lower for FS samples
and continued to have lower trends throughout storage for all temperatures. Colour changes
over time were noted for a* for FS (all storage temperatures) and UT (room and refrigerated
temperature) samples. In addition to filtered smoke, fillets appeared to have been treated with
ozone and UV which would decrease the microbiological load. However, treatments would not
mask markers used for quality assessment. Although lipid oxidation was retarded, chemical and
sensory measurements indicated that the treated fillets were of lower quality. According to
Mantilla and others (2005) currently the application of CO into fish muscle via euthanasia is
being performed by tilapia processors. There is a lack of information on how application of CO
with euthanasia will affect fish quality compared to conventional applications. Live tilapia was
(a) euthanized in water saturated with CO or (b) killed following industry guidelines. Fillets (b)
were treated with 100% CO for 30 minutes. Both treatments led to a significant increase in a*
of the fillet white and dark muscle. This distinctive cherry red colour was maintained for a long
period of time for both treatments while a brown colour developed for the controls. There was
no significant difference in a* between the gas treated and the euthanized fillets. The UV-Vis
spectra confirmed the intake of CO by heme proteins by both methods and also demonstrated
improvements in heme protein stability. These results suggest both treatments have a positive
effect on colour and heme stability. The similarity of results between the treatments imply
that the industry could adopt this new method of processing which has advantages such as
shorter processing and less product handling. However, it has to be taken into consideration
that handling of CO can bear a risk for the personal.
Beside of fish also the effect of CO and FS treatment on lamb meat colour and oxidative rancidity
was investigated by Demir and Kristinsson (2005a, b). CO/FS treatment led to a significant
increase in a* and red colour stability, which was found to be proportional to amount of CO
bound to heme proteins in muscle. Lipid oxidation was reduced for the CO/FS treatments
compared to control. Treatment with CO gas mixtures can significantly extend the quality on
lamb steaks. Higher concentrations of CO prevented discoloration of steaks. When applied to
goat steaks CO/FS treatment led to a significant increase in a* but decreased L* and had less
effect on b*. Red colour increase and stability was proportional to CO bound in muscle. Lipid
oxidation was reduced for the treatments compared to control. Treatment with CO gas mixtures
can significantly extend quality of goat meat.
Kristinsson and others (2005) examined and compared the properties of tilapia haemoglobin
complexed to either O2 (Oxy-Hb) or CO (CO-Hb) at pH 6.5, which reflects the tilapia muscle
post-mortem pH. CO-Hb was significantly more stable against autoxidation compared to Oxy-Hb
when kept at 4 and -30 °C for 23 days. Almost no loss of CO was detected for both temperatures
according to the UV-vis spectra of Hb. This stabilization was also believed to play a role in
increased protein structure stabilization since less protein aggregation was seen for CO-Hb.
The higher protein stabilization for Hb was linked to the heme group, which was maintained
in its reduced state longer for CO-Hb vs Oxy-Hb and was likely less exposed to solvent. CO-Hb
had significantly less peroxidase activity than Oxy-Hb and thus reactivity with H2O2. The pro-
binnenwerk.indd 335 21-08-2006 12:35:18

Reinhard Schubring
oxidative activity of CO-Hb was significantly reduced in a linoleic acid micelle system compared
to that of Oxy-Hb, while smaller differences in activity were seen in a washed cod and tilapia
muscle model system. Taking into account that 4 different types of Hb have been isolated from
blood of rainbow trout that are distinguishable based on electrophoretic mobility. Components
I, II, and III are cathodic Hbs, and component IV is classified as an anodic Hb. The objective
of investigations performed by Richards and others (2005) was to compare the pro-oxidative
characteristics of trout Hb and trout Mb. At pH 6.3, Mb autoxidized more rapidly (3.5-fold) as
compared to anodic Hb. Anodic Hb was a better catalyst of lipid oxidation in washed cod muscle
as compared to Mb at pH 6.3. This suggested that some process other than met heme protein
formation was the rate-limiting step in lipid oxidation processes. Anodic Hb released its heme
group much more rapidly than Mb. In comparisons of anodic and cathodic Hb, heme loss rate
better predicted the onset of lipid oxidation than autoxidation rate. The authors conclude that
heme dissociation has a primary role in the ability of different heme proteins to promote lipid
oxidation processes.
With the aim to develop a simple method to detect CO treated products based on the spectral
characteristics of extracted heme proteins a model system consisting of tilapia Hb and washed
tilapia muscle was constructed to find conditions to optimally extract CO-Hb. Different forms of
Hb were produced (Oxy-Hb, CO-Hb and MetHb) and mixed to construct a standard curve based
on heme peak absorbance wavelength. Using the model system it was possible to find the
optimal conditions to extract CO-Hb and maintain its stability. The overall procedure was less
than 30 min and could be done at room temperature. Higher CO-Hb recoveries and stability were
found when Hb was extracted at pH 8 vs. pH 6. More CO-Hb was recovered at high Hb additions
vs. low additions. The method could be successfully used to identify products treated with CO
when it was tested on several treated seafoods (Huo and Kristinsson 2005). The problem of
treatment of fish with CO or filtered smoke is briefly discussed by Balaban and others (2005).
Evaluation of colour can be done subjectively by sensory analysis or by objective measurements
with colorimeters or machine vision as described by Balaban and others (2005). Machine vision
can differentiate and quantify colour distributions in food samples with uneven colours. In the
case of fresh tuna, the hue values offered less variability and more monotonous change with
storage time in this experiment. The colour of fresh tuna exposed to 4% CO for 48 hours under
refrigeration, whether subsequently irradiated or not, stayed substantially unchanged with
refrigerated storage time. The colours of controls or irradiated samples did change substantially,
and turned brown. A simple and confirmative method for quantitative determination of carbon
monoxide in tuna and mahi-mahi tissues using GC/MS, following chemical liberation of CO into
headspace, is described by Anderson and Wu (2005). They used a molecular sieve column to
allow for separation of CO from nitrogen and oxygen, thereby eliminating the need to flush the
headspace of each sample bottle with helium prior to CO liberation. Investigating commercially
CO treated frozen tuna from multiple sources, only one source of frozen tuna steak was found
that had not been treated with CO with dark-brown appearance. In addition, only one sample of
frozen/vacuum packed mahi-mahi was found to be CO treated. The untreated frozen tuna sample
was found to be in the 150 ng/g range, while the treated samples were near or above 1 μg/g.
The treated mahi-mahi sample was in the 500 ng/g range. It is a common practice to remove
CO-treated/vacuum-packed frozen tuna steaks from the original packages before display directly
on ice or repackage in over-wrap packs for retail sale at refrigeration temperature. Therefore
authors were concerned whether significant loss of CO might occur once the frozen steaks were
exposed to atmosphere and began thawing as it might lead to ambiguity from the regulatory
aspect. It was found that upon homogenizing the fish in a food processor, it is important to
binnenwerk.indd 336 21-08-2006 12:35:18

place the flesh in a closed system immediately to prevent CO loss. Figure 4 displays the loss
of CO from fish homogenate held a 4 °C over the course of 24 hours. In addition to physically
damaging the protein and increasing the surface area of the meat exposed to the atmosphere,
homogenization also damages cellular compartments allowing for the release of enzymes and
other chemicals. Any or all of the above factors could explain the high rate of loss of CO in fish
homogenate (Anderson and Wu 2005).
The development of an easy and fast method for carbon monoxide detection is also described
by Haase and others (2005) and Heyer and Schubring (2005). They use suitable sensor
technology and can avoid harsh conditions for CO extraction from fish flesh. The electrochemical
DrägerSensors XS R CO are electrochemical measuring transducers for measuring the partial
pressure of gases under atmospheric conditions. The ambient air being monitored diffuses
through a membrane into the liquid electrolyte in the sensor. The electrolyte contains a
scanning electrode, a counter electrode and a reference electrode. An electronic potentiostat-
circuit ensures a constant voltage between sensing electrode and reference electrode. Voltage,
electrolyte and electrode material are selected to suit the gas being monitored so that it is
transformed electrochemically on the sensing electrode. The flow of e¯ electrons generated by
the reaction is a measure of the gas concentration. The CO concentrations obtained on both CO
treated and untreated tuna muscle were comparable with those reported when determination
was performed by the use GC or GC/MS. In these connection investigation was performed on the
effect of CO treatment on thermal stability of muscle proteins taken from tuna muscle (Heyer
and Schubring 2005). Figure 5 shows the DSC pattern obtained and reveals that the attachment
of CO to blood and muscle pigments obviously not significantly affect their thermal stability.
However, freezing caused a modification of the myosin peak. The shoulder to be seen on the
low-temperature side of the peak disappeared causing a lower differentiation and broadening
of the myosin peak. Transition temperatures as well as transition enthalpies did not change
significantly.
1600
1200
CO (ng/g)
800
400
0
0 5 10 15 20 25 30
Time (hours)
Figure 4. Depletion of CO from two commercially treated tuna samples that were homogenized in-lab and
subsequently held at 4 °C in an open system (Anderson and Wu 2005).
binnenwerk.indd 337 21-08-2006 12:35:19

Reinhard Schubring
-6.33 a
-6.38
-6.43
-6.48 b
-6.53
Heat flow (mW)
-6.58
-6.63 c
-6.68
-6.73
-6.78 d
-6.83
-6.88
-6.93
-6.98
-7.03
30 35 40 45 50 55 60 65 70
Furnace temperature (°C)
Figure 5. DSC curves of fresh (a, b) and frozen (c, d) tuna muscle with (a, c) and without CO treatment (b,
d) (Heyer and Schubring 2005).
Legal situation concerning the treatment of fish and parts of fish with CO and
“filtered smoke”
As already mentioned the stabilisation of colour of dark-fleshed fish species by CO treatment

has been performed in South East Asia at the beginning of the nineties of the last century to
saturate the big Japanese demand on high-grade fresh fish (http://www.seafoodbusiness.com/
buyguide/issue_tilapia.htm). Cases of fraud occurred in that CO-treated tilapia was offered as
cheap substitute. Labelled “izumi-dai,” the Japanese word for snapper, that was a hot seller in
sushi bars, where it was an inexpensive substitute for real Japanese snapper. However, Japanese
authorities, largely at the behest of the sashimi tuna industry, banned the importation of CO-
treated seafood in 1997 by setting a limit of CO content in fish flesh. 200 μg CO /kg fish flesh
were permitted. At the same time fish processing industry and trade installed requirements
necessary for a distribution cold chain that was able to guaranty extreme low temperatures in the
range of -60 °C. This low temperature appears to be necessary to avoid discolouration (Wheaton
and Lawson 1985). Due to the ban in Japan, Taiwanese fish (tilapia) farmers were forced to
develop markets in the United States, where the attractive vacuum-packed fillets with bright-
red fat lines became a big hit. The first year after CO tilapia fillets were banned in Japan, U.S.
imports of frozen tilapia fillets from Taiwan almost doubled, to 1,300 metric tons. Additionally
processing methods have been improved. Instead using CO to treat the fish, variants of the
traditional smoking process “tasteless smoke” or “clean smoke” or general “filtered smoke”
treated product were processed and offered with enhanced colour properties. On this way it was
easier to obtain the permission of authorities in the USA because traditional food processes are
seen in general as safe (GRAS). In the GRN No 000015, the FDA’s response letter (FDA 2000) to
Hawaii International’s GRAS determination it is written that based on the information provided
by Hawaii International, as well as other information available to FDA, the agency has no
questions at this time regarding the conclusion of Hawaii International that tasteless smoke is
binnenwerk.indd 338 21-08-2006 12:35:19

GRAS for use on raw tuna, before it is frozen, to preserve its taste, aroma, texture, and colour.
It is pointed out that the agency has not made its own determination regarding the GRAS status
of this use of tasteless smoke. It is demanded that the ingredient statement on labels of tuna
treated with tasteless smoke must comply with the labelling regulations. Tuna treated with
tasteless smoke should not be represented as a product characterized by “smoke” as a flavour.
Further, tuna that are treated and preserved with tasteless smoke may not be identified as “fresh
frozen” because the product has been preserved by the tasteless smoke and the regulations do
not provide for the use of the term “fresh” on products containing a preservative. It is further
written if the application of tasteless smoke causes the colour of the tuna flesh to be enhanced,
potentially causing consumers to be misled about the true nature or value of the tuna, the
product may be adulterated. In their USFDA Import Bulletin 16B-95 (USFDA 1999) on “tasteless
smoke and/or carbon monoxide” the FDA obviously reacts on various obscurities: “Imported
tuna treated with TS or CO should be: (i) labeled as processed foods that have been treated
with CO or TS, (ii) and not misrepresented as fresh frozen seafood by their label, and (iii) near
normal in flesh colour”. In an effort to determine criteria to objectively describe “near normal
in flesh colour”, the USDC Seafood Inspection Program collected colour data over five years. As
result a standard value of a* ≥ 16.2 was recommended (http://sst.ifas.ufl.edu/26thAnn/file10.
pdf) estimated with a colour measurement equipment HunterLab MiniScan XE Plus 45/0 LAV.
There is a list of foreign facilities verified by USDC (2005) for the production of filtered smoke
treated fishery products: Indonesia (10), Philippines (11), Ecuador (1), Sri Lanka (1), Sultanate
von Oman (1); and for the production of carbon monoxide treated fishery products: Indonesia
(1), Sultanate von Oman (1).
Irrespective of the clear legal status of using filtered smoke or CO to impair colour of fish and
meat in the USA there are actions to register to ban this use of CO mainly for fresh meat. As
mentioned above a Citizen Petition has been submitted by Kalsec Inc. (2005). On a special
website (http://www.co-meat.com/index.html) the actual situation can be followed.
In several other countries, however, the treatment of fish with CO is prohibited. A clear
standpoint was announced by the Canadian Food Inspection Agency already in 1999 with
the following information to all importers of fish: “…Please be informed that no shipments
of fish treated with CO will be allowed entry into Canada. The fish will be rejected for non-
permitted additive (CO). The primary methodology used will be sensory evaluation (unnatural
colour)…”. However, the reservation against tasteless smoke treated tuna appeared to be
less on condition that a correct labeling of the product has been done. Tasteless smoked
seafood cannot be classified as fresh/frozen but as preserved. This must be clearly indicated
on the label. Consumers are not familiar with the terms “tasteless smoke” or “modified smoke”
(http://sst.ifas.ufl.edu/25thAnn/file10.pdf).
A comparable situation appears to be in Singapore where the treatment of fresh, chilled or

thawed out frozen tuna cuts with CO to obtain the desired colour effect on appearance is seen
as a malpractice and is not condoned by Agri-food and Veterinary Authority (AVA) as CO-treated
tuna meat could mislead consumers into thinking that it is fresher or of higher value than it
actually is. However, tasteless smoke process is an approved processing technique whereby
smoke generated as in a normal food smoking process is first filtered to remove the characteristic
smoke flavour before application. This process is allowed to be used in the preservation of tuna.
Such products are required to be clearly labelled as smoked so that consumers will know that it
has been preserved by smoking (http://www.ava.gov.sg/JAVASCRIPT/carbonMTuna.htm).
binnenwerk.indd 339 21-08-2006 12:35:19

Reinhard Schubring
In Japan, a notice was issued to ban fish that have an initial CO content ≥ 500 µg/kg, or an
initial content ≥ 200 µg/kg which decreases significantly during two days of refrigerated storage
(Japan, Food Sanitation Law – Article 6). Also in Australia there seems to be concern about the
reappearance of CO-treated tuna, also known as “tasteless smoked tuna” (Lee 2005).
In our global world such new processing technologies are not only known in Asia and Oversea.
Therefore, it is not astonishing that Europe as well as Germany has been confronted with this
problem. Numerous information concerning CO treated fishery products detected on the market
in the different Member States of the EU were exchanged between the control authorities
during the last years within Rapid Alert System for Food and Feed (RASFF). The nearness to The
Netherlands where one of the leading companies (Anova Food B.V.) is domiciled in that use
filtered smoke (clear smoke) and is also assignee (http://www.anovafood.com/page.asp?lInt
MenuStyle=5&lIntLevel=11&lStrLang=EN&lStrBuyer=&lStrPagePath=Clearsmoke®) of the patent
granted to Olson and Brinsmade (2004) can be seen as a cause of the enormous presence of such
products on the market and underlines the special position of The Netherlands in this respect.
The Dutch companies, that import tuna, was announced that the Food and Consumer Product
Safety Authority (VWA) would start inspecting tuna fish and would reject consignments treated
with clear smoke or similar treatments. One company concerned challenged this decision and
made a court case. This case was lost by the VWA. The VWA did all they could to prohibit the entry
of this product on the European market. Since they lost three cases, they are at the end of their
possibilities of enforcement. The Netherlands are the only country in Europe where the trade
with filtered smoke (Clearsmoke) treated fish is not banned. The Section on Toxicological Safety
of the Standing Committee on the Food Chain and Animal Health, responsible within the EU
expressed concern that consumers were being misled as to the freshness of the product in 2003.
This issue was specifically addressed by the Council and the European Parliament during the co-
decision procedure leading to Directive 2003/114/EC (Anonymous 2004) and neither institution
supported the authorisation of CO as a food additive. CO would fall under the definition of a food
additive and was thus not authorised. The Committee also agreed that if a product was made
to Directive 91/493/EEC (Anonymous 1991) on fishery products, which requires that treatments
applied to inhibit development of pathogenic micro-organisms or constituting an important
element in the preservation of the product must be scientifically recognised or formally approved.
The Committee agreed furthermore to the conclusion that “clear-smoke technology” was an
indirect way of adding carbon monoxide to food. It also reiterated its previous conclusion that if
food is labelled as “smoked”, it should have a smoky flavour. The Directorate General for Health
and Consumer Affairs of the European Commission on request of a Dutch processor asked the
European Food Safety Authority (EFSA) on 8 October 2004 to evaluate the process of treatment
fishery products with the clearsmoke process. The EFSA Panel on food additives, flavourings,
processing aids and materials in contact with food (AFC) that is responsible has not answered
the question at present and is still looking for further information. Therefore, national authorities
in Europe could not find support by the European Commission at present.
Finally, the legal situation in Germany should be mentioned briefly. It has been discussed
recently in detail (Stenzel and Feldhusen 2004, Schubring 2004).
In contrast to EU regulations which are directly valid and legally binding, an EU directive has first
to be implemented in national law. This means that the Member States are obliged to transfer
the objectives and requirements of the directive into national law. Both the Law on Food and
Commodities (Lebensmittel- und Bedarfsgegenständegesetz, LMBG) and its follow-up law the
binnenwerk.indd 340 21-08-2006 12:35:19

new Act on Reclassification of Food and Animal Feed Law (Lebensmittel-, Bedarfsgegenstände-
und Futtermittelgesetzbuch, LFBG) which came into force on 1 September 2005 lay down legal
conditions that can be applied to the use of CO or CO containing gas blends like filtered smoke
in the processing of fish and fishery products. According to chapter 2, LFGB § 6 No (1) it is
evident that legal conditions are existing which prohibit the trade of fish and fishery products
which are CO treated or are treated with gas blends containing CO in Germany. Based on the
actual RASFF notification or information in 2006 it can be concluded that at present fishery
products with CO treatment or CO gas blend treatment are not of importance in Germany as well
as in the other member states of EU.
References
Aasgaard J. 1993. Farbstabilität bei verpackten Fleischerzeugnissen. Einfluss von Verpackungsmethoden,

reduziertem Druck und Kohlenmonoxid in modifizierten Atmosphären auf die Farbstabilität von Brühwurst- und
Kochschinkenaufschnitt. Fleischwirtschaft 73:441-444.
Abe M, Miyazaki H, Nagai Y, Nakajima M, Mijabe M. 1994. Determination of carbon monoxide in dark-fleshed fish by
gas chromatography. Ann Rep Nagoya City Publ Health Res Inst 40:10-15.
Anderson CR, Wu W-H. 2005. Analysis of carbon monoxide in commercially treated tuna (Thunnus spp.) and mahi-mahi
(Coryphaena hippurus) by gas chromatography/mass spectrometry. J Agric Food Chem 53:7019-7023.
Anonymous 1991. Council Directive 91/493/EEC of 22 July 1991 laying down the health conditions for the production
and placing on the market of fishery products. OJ L268 24.09.1991 p 15-34.
Anonymous 2003. Leitsätze für Fische, Krebs- und Weichtiere und Erzeugnisse daraus, Neufassung vom 27.November
2002 (Beilage Nr.46 b zum BAnz. vom 7.März 2003, GMBl. Nr.8-10 S.157 vom 20.Februar 2003)
Anonymous 2004. Directive 2003/114/EC of the European Parliament and of the Council
of 22 December 2003 amending Directive 95/2/EC on food additives other than colours and sweeteners. OJ 47: L 24
29.1.2004 p 59-64.
Baker RTM, Pfeiffer A-M, Schöner F-J, Smith-Lemmon L. 2002. Pigmenting efficacy of astaxanthin and canthaxanthin
in fresh-water reared Atlantic salmon, Salmo salar. Anim Feed Sci Technol 99:97-106.
Balaban MO, Kristinsson HG, Otwell WS. 2005. Evaluation of color parameters in a machine vision analysis of carbon
monoxide-treated fish–part I: fresh tuna. J Aquat Food Prod Technol 14(2):5-24.
Barrett I, Brinner L, Brown WD, Dolev A, Kwon TW, Little A, Olcott HS, Schaefer MB, Schrader P. 1965. Changes in tuna
quality and associated biochemical changes during handling and storage aboard fishing vessels. Food Technol
19:1846-1855.
Bito, M. 1964. Studies on the retention of meat colour of frozen tuna. I. Absorption spectra of the aqueous extract of
frozen tuna meat undergoing discoloration. Bull Jap Soc Sci Fish 30:847-857.
Bito M. 1965a. Studies on the retention of meat colour of frozen tuna. II. Effect of storage temperature on preventing
discoloration of tuna meat during freezing storage. Bull Jap Soc Sci Fish 31:534-539.
Bito M. 1965b. Studies on the retention of meat colour of frozen tuna. III. The discoloration of tuna meat during the
storage around at temperature of freezing point. Bull Jap Soc Sci Fish 31:540-545.
Bito M. 1967. Studies on the retention of meat colour of frozen tuna. IV. Acceleration of discoloration of tuna meat
by freezing and its relation to storage temperature. Bull Jap Soc Sci Fish 33:33-40.
Bito M. 1968. Studies on the retention of meat colour of frozen tuna. V. Effect of freezing rate. Bull Jap Soc Sci Fish
34:608-612.
Bito M. 1969a. Studies on the retention of meat colour of frozen tuna. VI. Effect of plastic film packaging and ice-
glazing on the rate of discoloration. Bull Jap Soc Sci Fish 35:218-226.
Bito M. 1969b. Studies on the retention of meat colour of frozen tuna. VII. Effect of freezing at ultra-low temperatures.
Bull Jap Soc Sci Fish 35:1193-1200.
Bito M. 1970. Studies on the retention of meat colour of frozen tuna. VIII. Discoloration of frozen meat during
defrosting in water or air. Bull Jap Soc Sci Fish 36:402-406.
binnenwerk.indd 341 21-08-2006 12:35:20

Reinhard Schubring
Bito M. 1975. Effect of thawing and packaging material on discoloration of frozen skipjack meat. Bull Jap Soc Sci Fish
41:1031-1037.
Bito M. 1976. Retention of meat colour of frozen tuna. Bull Tokai Reg Fish Res Lab 84:51-113.
Bito M. 1980. Effect of storage temperature on discoloration during frozen storage in packaged meat prepared from
tuna frozen-stored aboard ships. Bull Tokai Reg Fish Res Lab 103:73-82.
Bito M, Kiriyama H. 1973a. Penetration of NaCl into the meat, and the meat colour of brine-frozen skipjack tuna. Bull
Tokai Reg Fish Res Lab 75:75-86.
Bito M, Kiriyama H. 1973b. Effect of storage temperature on discoloration of skipjack meat during frozen storage. Bull
Tokai Reg Fish Res Lab 75:87-94.
Brewer S. 2004. Irradiation effects on meat color – a review. Meat Sci 68:1-17.
Bundesministerium für Ernährung, Landwirtschaft und Verbraucherschutz (BMELV). 2005. Lebensmittel-,
Bedarfsgegenstände- und Futtermittelgesetzbuch. BGBl I 2005, 2618 (3007) http://www.verbraucherministerium.
de/index-uuid=000840DE53CB10168F766521C0A8D816&print=yes.html
Bundesministerium für Gesundheit. 1995. Das Lebensmittel- und Bedarfsgegenständegesetz. Bonn: Bundesministerium
für Gesundheit 50 p.
Cashon RE, Vayda ME, Sidell BD. 1997. Kinetic characterization of myoglobins from vertebrates with vastly different
body temperatures. Comp Biochem Physiol B 117:613-620.
Chen W-L, Chow C-J. 2001. Studies on the physicochemical properties of milkfish myoglobin. J Food Biochem 25:157-
174.
Chiou T-K, Pong C-Y, Nieh F-P, Jiang S-T. 2001. Effect of met-myoglobin reductase on the color stability of blue fin tuna
during refrigerated storage. Fisheries Sci 67:694-702.
Chow C-J, Ochiai Y, Watabe S, Hashimoto K. 1988. Effect of freezing and thawing on the discoloration of tuna meat.
Chow CJ, Liu SM, Tsai ML. 1997. Characteristics of reaction between carbon monoxide gas and myoglobin in tuna flesh.
J Food Drug Anal 5:199-206.
Chow CJ, Hsieh PP, Hwang MS. 1998a. Quantitative determination of carbon monoxide residue in tuna flesh. J Food
Drug Anal 6:439-446.
Chow CJ, Hsieh PP, Tsai ML, Chu YJ. 1998b. Quality changes during iced and frozen storage of tuna flesh treated with
carbon monoxide gas. J Food Drug Anal 6:615-623.
Church N. 1994. Developments in modified-atmosphere packaging and related technologies. Trends Food Sci Technol
5:345-352.
Cornforth DP, Rabovitser JK, Ahuja S, Wagner JC, Hanson R, Cummings B, Chudnovsky Y. 1998. Carbon monoxide, nitric
oxide, and nitrogen dioxide levels in gas ovens related to surface pinking of cooked beef and turkey. J Agric Food
Chem 46:255-261.
Demir N, Kristinsson HG, Balaban MO. 2004. Quality changes in mahi mahi (Coryphaena hippurus) fillets treated
by different carbon monoxide concentrations and filtered smoke as assessed by color machine vision and lipid
oxidation. http://ift.confex.com/ift/2004/techprogram/paper_24707.htm
Demir N, Kristinsson HG. 2005. Objective quality assessments of goat meat steaks treated by different carbon monoxide
concentrations and filtered smoke using color machine vision system. http://ift.confex.com/ift/2005/techprogram/
paper_31477.htm
Demir N, Kristinsson HG. 2005. Quality changes in lamb steaks treated by different carbon monoxide concentrations
and filtered smoke as assessed by color machine vision and lipid oxidation. http://ift.confex.com/ift/2005/
techprogram/paper_31491.htm
Doe PE. editor. 1998. Fish drying & smoking: Production and quality. Lancaster, Basel: Technomic Publishing Co. 267 p.
European Commission. 2001. European Commission SCF/CS/ADD/MSAd/204 Final. Opinion of the Scientific Committee
on Food on the use of carbon monoxide as component of packaging gases in modified atmosphere packaging for
fresh meat. Brussels 18 12 2001. http://europa.eu.int/comm/food/fs/sc/scf/out112_en.pdf.
FDA. 2000. US Food and Drug Administration. 2000. GRAS notice nr GRN 000015. March 20, 2000, Washington, D.C.
USFDA, http://www.cfsan.fda.gov/~rdb/opa-g015.html.
binnenwerk.indd 342 21-08-2006 12:35:20

FDA. 2002. US Food and Drug Administration. 2002. GRAS notice nr GRN 000083. February 21, 2002, Washington, D.C.
FDA. 2004. US Food and Drug Administration. 2004. GRAS notice nr GRN 000143. July 29, 2004, Washington, D.C.
FDA. 2005. US Food and Drug Administration. 2005. GRAS notice nr GRN 000167. September 29, 2005, Washington,
D.C. USFDA, http://www.cfsan.fda.gov/~rdb/opa-g167.html.
Garner KS, Kristinsson HG. 2004. Quality of Spanish mackerel (Scomberomorous maculatus) muscle as affected by carbon
monoxide and filtered smoke gas treatment. http://ift.confex.com/ift/2004/techprogram/paper_24776.htm
Gee DL, Brown WD. 1978. Extension of shelf life in refrigerated ground beef stored under an atmosphere containing
carbon dioxide and carbon monoxide. J Agric Food Chem 26:274-276.
Giddings GG. 1977. The basis of color in muscle foods. Crit Rev Food Sci Nutr 8:81-114.
Haase T, Heyer H, Schubring R. 2005. Entwicklung einer einfachen und schnellen Analysenmethode für Kohlenmonoxid
in Fisch und Durchführung orientierender Untersuchungen. Inf Fischereiforsch - Inf Fish Res 52:61-65.
Heyer H, Schubring R. 2005. Entwicklung einer einfachen und schnellen Analysenmethode für Kohlenmonoxid in Fisch
– Teil 2: Methodenoptimierung. Inf Fischereiforsch – Inf Fish Res 52:106–114.
Horner WFA. 1997. Preservation of fish by curing (drying, salting and smoking). In: Hall GM. editor. Fish processing
technology. 2nd ed. London, Weinheim, New York, Tokyo, Melbourne, Madras: Blackie Acad & Prof. p 119-159.
Houben JH, van Dijk A, Eikelenboom G, Hoving-Bolink AH. 2000. Effect of dietary vitamin E supplementation, fat level
and packaging on colour stability and lipid oxidation in minced beef. Meat Sci 55:331-336.
Hsieh PP, Chow CJ, Chu YJ, Chen WL. 1998. Change in color and quality of tuna during treatment with carbon monoxide
gas. J Food Drug Anal 6:605-613.
Hunt MC, Mancini RA, Hachmeister KA, Kropf DH, Merriman M, Delduca G, Milliken G. 2004. Carbon monoxide in modified
atmosphere packaging affects color, shelf life, and microorganisms of beef steaks and ground beef. J Food Sci 69:
FCT45-52.
Huo L, Kristinsson HG. 2005. Rapid detection of carbon monoxide treated seafood products based on spectral properties
of heme proteins. http://ift.confex.com/ift/2005/techprogram/paper_31446.htm
Iffland R, Balling P, Eiling G. 1989. Optimierung des Kohlenmonoxidnachweises im biologischen Material. Beitr Gerichtl
Med 47:89-94.
Ishiwata H, Takeda Y, Kawasaki Y, Yoshida R, Sugita T, Sakamoto S, Yamada T. 1996.
Concentration of carbon monoxide in commercial fish flesh and in fish flesh exposed to carbon monoxide gas for color
fixing. J Food Hyg Soc Japan 37:83-90.
John L, Cornforth D, Carpenter CE, Sorheim O, Pettee BC, Whittier DR. 2005. Color and thiobarbituric acid values of
cooked top sirloin steaks packaged in modified atmospheres of 80% oxygen, or 0.4% carbon monoxide, or vacuum.
Meat Sci 69:441-449.
Jonker KM, Geertsen JAM, Bos HPGM. 2003. Untersuchung auf Kohlenmonoxid in Thunfisch. SANCO-2003-02727-00-
00-DE-TRA-00 (NL)
Kalsec, Inc. 2005. Letter regarding Citizen Petition Requesting FDA to Enforce Ban on Carbon Monoxide Gas in Fresh
Meat Packaging. November 15, 2005. http://www.fda.gov/ohrms/dockets/dockets/05p0459/05p-0459-cp00001-
01-vol1.pdf
Kanoh S, Suzuki T, Maeyama K, Takewa T, Watabe S, Hashimoto K. 1986. Comparative studies on ordinary and dark
muscles of tuna fish. Bull Jap Soc Sci Fish 52:1807-1816.
Kendrew JC, Parrish RG, Marrack JR, Orlans ES. 1954. The species specificity of myoglobin. Nature 174:946 -949.
Kendrew JC, Bodo G, Dintzis HM, Parrish RG, Wyckoff H. 1958. A three-dimensional model of the myoglobin molecule
obtained by x-ray analysis. Nature 181:662 -666.
Kendrew JC, Dickerson RE, Strandberg BE, Hart RG, Davies DR, Phillips DC, Shore VC. 1960. Structure of myoglobin. A
three-dimensional Fourier synthesis at 2 Å resolution. Nature 185:422 -427.
Kendrew JC. 1963. Myoglobin and the structure of proteins. Science 139:1259 -1266.
Klettner PG. 2001. Schutzatmosphärenverpackung bei Hackfleisch vom Schwein und Rind unter besonderer
Berücksichtigung von Kohlenmonoxid. Mitteilungsblatt BA Fleischforschung 40(152):149-157.
binnenwerk.indd 343 21-08-2006 12:35:20

Reinhard Schubring
Kowalski WR, inventor; 1999 Oct. 26. Process for manufacturing tasteless super-purified smoke (TSPS) for treating
seafood to be frozen and thawed. U.S. patent 5,972,401.
Krause TR, Sebranek JG, Rust RE, Honeyman MS. 2003. Use of carbon monoxide packaging for improving the shelf life
of pork. J Food Sci 68:2596-2603.
Kristinsson HG, Mony S, Demir N, Balaban MO, Otwell WS. 2003. The effect of carbon monoxide and filtered smoke on
the properties of aquatic muscle and selected muscle components. Reykjavik, Iceland: Taft 2003, Proc of the TAFT
2003 Conf p 27-29.
Kristinsson HG, Mony SS, Petty HT. 2005. Properties of tilapia carboxy- and oxyhemoglobin at post mortem pH. J Agric
Food Chem 53:3643-3649.
Kusmider EA, Sebranek JG, Lonergan SM, Honeyman MS. 2002. Effects of carbon monoxide packaging on color and lipid
stability of irradiated ground beef. J Food Sci 67:3463-3468.
Larson TV, Koenig JO. 1993. A summary of the emission characterization and noncancer respiratory effects of wood
smoke. Epa-453/R-93-036. Environmental Protection Agency. Washington USA.
Lee S, Joo ST, Alderton AL, Hill DW, Faustman C. 2003. Oxymyoglobin and lipid oxidation in yellowfin tuna (Thunnus
albacares) loins. J Food Sci 68:1664-1668.
Lee, D. 2005. Carbon monoxide treated tuna: why is it here? Prof Fisherman 27(2):18.
Leydon NL, Suman SP, Ellis PC, Palmer C, Faustman LC, Pivarnik LF. 2005. Quality and safety assessment of time/
temperature storage conditions for filtered smoked tilapia fillets produced for retail distribution. http://ift.confex.
com/ift/2005/techprogram/paper_29637.htm
Leydon NL, Suman SP, Ellis PC, Rossi S, Palmer C, Faustman C, Pivarnik LF. 2004. Quality assessment of commercially
produced tasteless smoked seafood products. http://ift.confex.com/ift/2004/techprogram/paper_23353.htm
Livingston DJ, Brown WD. 1981. The chemistry of myoglobin and its reactions. Food Technol 35:244-252.
Love RM. 1997. Biochemical dynamics and the quality of fresh and frozen fish. In: Hall GM. editor. Fish processing
technology. 2nd ed. London, Weinheim, New York, Tokyo, Melbourne, Madras: Blackie Acad & Prof. p 1-31.
Ludlow N, Kristinsson HG, Balaban MO, Otwell WS, Welt BA. 2004. Effect of different carbon monoxide and filtered
smoke treatments on the quality and safety of yellowfin tuna (Thunnus albacares) muscle. http://ift.confex.com/
ift/2004/techprogram/paper_24784.htm
Luno M, Beltran JA, Roncales P. 1998. Shelf-life extension and colour stabilisation of beef packaged in a low O2
atmosphere containing CO: loin steaks and ground meat. Meat Sci 48:75-84.
Mantilla D, Kristinsson HG, Otwell WS, Balaban MO, Chapman FA. 2005. Euthanasia of tilapia (Oreochromis niloticus) using
carbon monoxide for red color stabilization. http://ift.confex.com/ift/2005/techprogram/paper_30798.htm
Marcinek DJ, Bonaventura J, Wittenberg JB, Block BA. 2001. Oxygen affinity and amino acid sequence of myoglobins
from endothermic and ectothermic fish. Am J Physiol Reg Integ Comp Physiol 280:R1123-R1133.
Matthews AD. 1983. Muscle colour deterioration in iced and frozen stored bonito, yellowfin and skipjack tuna caught
in Seychelles waters. J Fd Technol 18:387-392.
Matsuno T. 2001. Aquatic animal carotenoids. Fisheries Sci 67:771-783.
Miyazaki H, Abe M, Asanoma M, Nagai Y, Nakajima M, Miyabe M. 1997. Simple detection of carbon monoxide in fish
meat by GC. J Food Hyg Soc Japan 38:233-239.
Mony S, Kristinsson HG. 2004. Thermal denaturation and oxidative stability of carboxy-, oxy- and met-hemoglobin from
tilapia. http://ift.confex.com/ift/2004/techprogram/paper_24831.htm
Nam KC, Ahn DC. 2003. Effects of irradiation on meat color. Food Sci Biotechnol 12(2):198-205.
Narasimha Rao D, Sachindra NM. 2002. Modified atmosphere and vacuum packaging of meat and poultry products.
Food Rev Int 18:263-293.
Nickell DC, Springate JRC. 2001. Pigmentation of farmed salmonids. In: Kestin SC, Warriss PD. editors. Farmed fish
quality. Oxford: Fishing News Books. p. 59-75.
Noren SR, Williams TM. 1999. Body size and skeletal muscle myoglobin of cetaceans: adaptations for maximizing dive
duration. Comp Biochem Physiol A 126:181-191.
Ochiai Y, Chow C-J, Watabe S, Hashimoto K. 1988. Evaluation of tuna meat discoloration by Hunter color difference
scala. Nippon Suisan Gakkaishi 54:649-653.
binnenwerk.indd 344 21-08-2006 12:35:20

Olson BE, Brinsmade DB, inventors; 2004 Aug. 17. Seafood preservation process. U.S. patent 6,777,012.
Ordway GA, Garry DJ. 2004. Myoglobin: an essential hemoprotein in striated muscle. J Exp Biol 207:3441-3446.
Otwell WS, Balaban M, Kristinsson H. 2003. Use of carbon monoxide for color retention in fish. Reykjavik, Iceland: Taft
2003, Proc of the TAFT 2003 Conf p 24-26.
Pegg RB, Shahidi F. 1997. Unraveling the chemical identity of meat pigments. Crit Rev Food Sci Nutr 37:561-589.
Pong C-Y, Chiou T-K, Ho M-L, Jiang S-T. 2000. Effect of polyethylene package on the metmyoglobin reductase activity
and color of tuna muscle during low temperature storage. Fisheries Sci 66:384-389.
Richards MP, Dettmann MA, Grunwald EW. 2005. Pro-oxidative characteristics of trout hemoglobin and myoglobin: a
role for released heme in oxidation of lipids. J Agric Food Chem 53:10231-10238.
Rosnes JT, Sivertsik M, Skipsnes D, Nordtvedt TS, Corneliussen C, Jakobsen O. 1998. Transport of superchilled salmon
in modified atmosphere. In: International Institute of Refrigeration, Editor. Proceed. from Hygiene, Quality and
Safety in the Cold Chain and Air-Conditioning. Paris: IIF-IIR-Commission C2/E1 p 229-236.
Schubring R. 2004. Applikation von “tasteless smoke” und Kohlenmonoxid bei Fisch - Technologiefolgenabschätzung.
Arch Lebensmittelhyg 55:85-93.
Seideman SC, Cross HR, Smith GC, Durland PR. 1984. Factors associated with fresh meat color: a review. J Food Qual
6:211-237.
Sivertsvik M, Jeksrud WK, Rosnes JT. 2002. A review of modified atmosphere packaging of fish and fishery products
- significance of microbial growth, activities and safety. Int J Food Sci Technol 37:107-127.
Sørheim O, Aune T, Nesbakken T. 1997. Technological, hygienic and toxicological aspects of carbon monoxide used in
modified-atmosphere packaging of meat. Trends Food Sci Technol 8:307-312.
Sørheim O, Nissen H, Nesbakken T. 1999. The storage life of beef and pork packaged in an atmosphere with low carbon
monoxide and high carbon dioxide. Meat Sci 52:157-164
Stenzel W-R, Feldhusen F. 2004. Aspekte der Verwendung von Kohlenmonoxid bei Fleisch- und Fischerzeugnissen.
Fleischwirtschaft 84:131-136.
Takeda Y, Kawasaki Y, Sugita T, Sakamoto S, Sato K, Yoshida R, Maitani T, Ishiwata H, Yamada T. 1995. Inspection of
carbon monoxide in imported tilapia. Eisei Shikenjo Hokuko 113:74-76.
Tülsner M. 1994. Fischverarbeitung. Bd. 1: Rohstoffeigenschaften und Grundlagen der Verarbeitungsprozesse. Hamburg:
Behr’s Verlag. p 263-279.
USDC. 2005. USDC participants list for firms, facilities and products. U.S. Department of Commerce, NOAA, 20(2):
September 2005, revised 1/9/06.
USFDA 1999. Import Bulletin 16B-95. Tasteless smoke and/or carbon monoxide. May 1999. http://seafood.ucdavis.
edu/guidelines/fdabulletin16b.htm.
Wheaton FW, Lawson TB. 1985. Processing aquatic food products. New York: John Wiley and Sons p 518.
Widdop B. 2002. Analysis of carbon monoxide. Ann Clin Biochem 39:378-391.
Wittenberg JB, Wittenberg BA. 2003. Myoglobin function reassessed. J Exp Biol 206:2011-2020.
Wolfe SK. 1980. Use of CO- and CO2-enriched atmospheres for meats, fish, and produce. Food Technol 34:55-58, 63.
Woodruff RE, Silliker J, inventors; TransFRESH Corporation (Salinas, CA) assignee. 1985 juni 11. Process and composition
for producing and maintaining good colour in fresh meat, fresh poultry and fresh fish. U.S. patent 4,522,835.
Xiong YL, Decker EA, Robe GH, Moody WG. 1993. Gelation of crude myofibrillar protein isolated from beef heart under
antioxidative conditions. J Food Sci 58:1241-1244.
Yamanaka H, Takamizawa M, Amano K. 1973a. Relation between the colour of tuna meat and the activity of metmyoglobin
reductase. I. A method for the determination of metmyoglobin reductase activity of the extract from tuna meat.
Bull Jap Soc Sci Fish 39:667-671.
Yamanaka H, Takamizawa M, Amano K. 1973b. Relation between the colour of tuna meat and the activity of metmyoglobin
reductase. II. Changes in the activity of metmyoglobin reductase during frozen storage and subsequent thawing of
tuna meat. Bull Jap Soc Sci Fish 39:673-681.
Yamaoka K, Adachi T, Ohta S, inventors; 1996 Aug. 17. Method for curing fish and meat by extra-low temperature
smoking. U.S. patent 5,484,619.
Yoshida M, Hori S. 1998. Effect of carbon monoxide on colour difference and K values in fresh fish muscle. Jap J Food
Chem 5:19-23.
binnenwerk.indd 345 21-08-2006 12:35:20

binnenwerk.indd 346 21-08-2006 12:35:20
Chapter 5:
Microbial quality of seafood
In this chapter, papers about the microbial quality aspects related to safety and spoilage of
seafood are grouped. Microbial activity is responsible for spoilage of most fresh and lightly
preserved seafood. Today, as the interest in those foods is increasing, it is important to obtain
detailed information about microorganisms to reduce losses due to spoilage. In addition,
microorganisms in seafood can be responsible for the outbreak of seafood-borne human diseases.
The amount of diseases due to seafood is rather high compared to other food and should be
reduced for the safety of the consumer. These issues represent a challenge for research. To
inhibit the growth of pathogens in food, it is important to understand the reproduction pattern
of the different organisms and to have good methods to characterize them.
In the first paper a comparison is made between the predicted and measured contents of sulphide
producing bacteria in aseptically filleted cod during the assessment the time temperature
history. The second article in this chapter deals with the growth kinetic of Staphylococcus
aureus during cod desalting in order to evaluate its response. The importance of the incidence
of the pathogenic Aeromonas spp. and coliforms in mussles in Greece was investigated.
Furthermore the characterisation of Vibrio parahaemolyticus was tested by conventional and
novel methods.
Different methods for inhibiting the growth of spoilage and pathogenic microorganisms
are described. One paper deals with inactivation of microorganisms by pulsed light for the
preservation of food. Three papers are dealing with biopreservation as an innovative way of
reducing microbial risks in lightly preserved food. Chitosan, a natural biopolymer, has been
suggested as a novel food biopreservative. In addition, the selection of specific bacteria active
against spoilage and pathogenic target bacteria are discussed.
binnenwerk.indd 347 21-08-2006 12:35:20

binnenwerk.indd 348 21-08-2006 12:35:21
Assessing the time-temperature history of aseptically filleted
cod from predicted and measured contents of sulphide producing
bacteria
Taran Skjerdal1 and Sissel Ranneklev2
1DNV Research, Veritasveien 1, N-1322 Høvik, Norway
2Colifast AS, Strandveien 35, N-1366 Lysaker, Norway
Abstract
A protocol for assessing fish time-temperature history has been developed and tested on
aseptically filleted cod. The protocol is based on traceability data, a predictive model for growth
of sulphide producing bacteria in fish, and analysis of these bacteria. The results showed that
abuse temperature periods could be proved for time-temperature loads corresponding to 8 hours
at room temperature (app 22 °C) and 24 hours at 10 °C, if these conditions were applied later
than four days after catch. The rapid test kit (FAST Fish Test) gave similar analysis results to
the reference agar plate method (Iron Agar Lyngby), indicating that the methodology could be
used without access to laboratory facilities. The protocol developed has potential uses within
product assessment, verification of supply chain information, and can also give an indication
about the food safety of fish intended for sushi, but needs to be further developed and tested
for commercially produced fish.
Keywords: shelf life, time-temperature control, sulphide producing spoilage bacteria, rapid
analysis, risk management, cod
Introduction
Two recent surveys of fresh fish offered in supermarkets and specialised fish shops in Oslo and
Bergen showed that the fresh fish was not as fresh as many consumers assumed (The Norwegian
Broadcasting, 2005). It was found that 17 out of 23 shops in Oslo, some of them high profiled
fish shops, sold fish with sensory unacceptable levels of spoilage bacteria. More than 50% of the
“fresh” fish samples analysed contained spoilage bacteria corresponding to a week or older fish.
These observations indicate that methods for fish quality and freshness assessments developed
during the last decades have not been implemented, at least not in the parts of the “fjord-
to-fork chain” being nearest to the consumers. There may be several reasons for this. The fish
supply chain is complicated and as a result, the product history information is often dubious or
lost completely. This limits the possibilities to select better suppliers or ask for improvements in
the supply chain. Further, consumption of traditionally prepared fish does not imply particular
health risks, so as long as the consumers accept the quality level, there seems to be little to
win by measuring the fish freshness. A third reason is that shops have no laboratory facilities,
and external analyses are costly. Based on these points, it is reasonable to ask whether quality
control of fish is useful or needed. There are at least two new developments in recent years
which indicate that there is. First, implementation of traceability has made it easier to identify
and thereby avoid using producers and suppliers that do not deliver desired quality fish. The
second is the increased popularity of sushi, which has higher food safety risks than heat treated
binnenwerk.indd 349 21-08-2006 12:35:21

fish. Producers, restaurants, and retailers as well as consumers have to take the new risks
related to consumption of raw fish into consideration, meaning that more attention should be
paid to the food safety aspects of fish. Tools that can easily provide relevant information for
assessment of food quality and safety issues are therefore needed.
Degradation of fish during post mortem storage has been studied for decades, and it is
well known that both storage time and storage temperature are essential elements for the
degradation rate. Several methods for fish quality assessment have been developed, (Olafsdottir
and others 1998). Development of sensory analysis of attributes like odour, texture, colour,
appearance of skin, slime, eyes, and gills started already in 1953 by Shewan and co-workers,
and was further developed to the Torry scheme and Quality Index Method (QIM) (Bremner 1985;
Luten and Martinsdottir 1997; Jonsdottir and others 2005). The sum of scores (QIM score) of
each attribute correlates linearly with the ice-storage time of the fish. The content of specific
spoilage bacteria can also be correlated to storage time. Jørgensen and others (1988) have
reported that white fish become sensory spoiled when the level of sulphide producing bacteria
(SPBs) reaches 108 cfu/g fish. The growth rate of these and other spoilage bacteria in various
fish species at different temperatures have been measured, and integrated in a Seafood spoilage
predictor (Dalgaard and others 2002). This program estimates the storage time of the fish,
and predicts the remaining shelf life of the fish at various temperature scenarios. SPBs can
be measured with a traditional agar plate method (Gram and others 1987), and with a rapid
and simple method, called FAST fish Test, based on the similar growth medium and detection
principle as the agar plate method (Skjerdal and others 2004; Lorentzen and others 2003;
Ranneklev and Berg 2005).
In the present work, the suitability of SPBs and QIM scores used as test parameters to identify
storage periods at abuse temperature for cod fillets has been investigated. The test protocols
have been designed based on traceability data, available predictive models and analysis of QIM
and SPBs. The experimental testing in this work was limited to aseptically filleted cod.
Protocol concept
Two protocols for assessment of the time-temperature history of cod fillets were suggested, based
on SPBs and QIM measurements, respectively. The protocol concept based on SPB measurement
is shown in Figure 1. It is based on three modules; (1) date and place of the catch, -which
should be possible to obtain from traceability data, (2) a predictive model for a measurable
quality parameter or score during storage, and, (3) an analysis method for determination of this
quality parameter. If fish has been stored on ice during the entire period, and no contamination
has been introduced, the measured value of the quality parameter should be identical to the one
predicted for ice stored control. If the measured value is significantly higher than the predicted
value of ice stored control, it is likely that the fish has been stored at abuse temperature,
alternatively, the fish has been contaminated or is older than assumed.
In the present work, the protocol concept was tested for its suitability to prove abuse
temperature periods during storage. Aseptically filleted cod with a known history was used as
test material in order to avoid interfering effects of contamination and incorrect traceability
information.
binnenwerk.indd 350 21-08-2006 12:35:21

Assessing the time-temperature history from contents of sulphide producing bacteria
Log SPB/g fish
-4 -2 0 2 4 6 8 10
Storage time (days)
Figure 1. Protocol concept. ° Catch date, obtained from traceability data; ——— predicted growth of SPB
in icestored fish; Measured SPB content in accordance with predicted content, confirm ice stored fish;
Measured SPB content significantly higher than predicted for ice stored fish, indicate abuse temperture
periods (……), or ice stored fish but more contaminated or older than assumed (----).
Protocol based on sulphide producing bacteria (SPB)

Protocol 1 was based on sulphide producing bacteria (SPBs), which limit the sensory shelf life
in most white fish species from temperate waters (Gram and others 1987, 1996). The Seafood
spoilage predictor (Dalgaard and others 2002) was used to predict the content of SPB in the
fish at various time-temperature scenarios. The model was downloaded from http://www.dfu.
min.dk/micro/ssp/. The analysis kit FAST (‘FAST Fish Test, Colifast AS, Lysaker, Norway) and Iron
Agar Lyngby (Oxoid Ltd) were used to determine the content of sulphide producing bacteria
in the fish. The FAST kit consists of a tube (50 ml) containing 7 ml of broth, with similar
composition to Iron Agar Lyngby (Gram and others 1987; Skjerdal and others 2004). In brief,
ferric iron in the medium is used as the terminal electron acceptor by the bacteria, and the
ferrous iron reacts with sulphide produced by the bacteria. During growth a black precipitate
of iron sulphide is formed when the number of bacteria in the tube has reached a certain level.
During analysis approximately 1 g of fish is transferred to the tube, and incubated at 30 °C in
a portable incubator with a web camera attached (www.Colifast.no, Lysaker, Norway). The time
until the medium turns black correlates with the initial level of sulphide producing bacteria in
the sample, and this is the basis for the enumeration.
Protocol based on QIM

Protocol 2 was based the QIM index. The QIM score was determined according to the scheme
for assessment of gutted cod and the QIM Rating System as downloaded from http://www.dfu.
min.dk/QIMRS/. After filleting, a simplified assessment of the parameters odour, appearance,
colour and texture was applied. Each parameter was described corresponding to “good”, “fair”
or “poor” quality, transferred to respective scores 1, 2 or 3.
Experimental testing of protocols

Fish raw materials
Gutted cod, caught in the Oslo fjord was bought from a local fisherman. The fish was iced on
board within 5 hours after catch, and transported to the lab within 15 hours. The fish was
binnenwerk.indd 351 21-08-2006 12:35:21

filleted under aseptic conditions, cut in pieces of approximately 25 grams, and put in sterile
Petri dishes, marked with individual number and stored on ice for 11 days. Six individual fish,
approximately 1 kg each, were used in each experiment.
Time-temperature storage scenarios

Two kinds of storage regimes including abuse temperatures were applied. In the first series of
experiments the fish was stored on ice but with a period at abuse temperature, with various
temperature loads applied on day 4 in the storage period. Fish pieces were then moved from ice
to either 4, 10 or 22 °C, but 10 pieces were kept on ice as a control. The samples at 4 °C were
moved back on ice after 8 or 24 hours, those at 10 °C after 4, 8 or 24 hours, and those at 20
°C after 2, 4 or 8 hours, and stored further for one week. In the second series of experiments,
the period at abuse temperature was applied on various days throughout storage, but the time-
temperature load in these periods was constant. The temperature loads used were 10 °C for 8
or 24 hours, and room temperature (app 22 °C) for 2, 4 or 8 hours. Another three fillet pieces
were ice stored for the entire period, as a control.
All experiments were done in triplicate. The content of sulphide producing bacteria was
determined on day 1, 5, and 11 using Iron Lyngby Agar and the FAST method, and predicted
using the Seafood spoilage predictor based on the applied time-temperature conditions.
The difference between ice stored control samples and samples stored periods at abuse
temperature were analysed using a one sided hypothesis test (Box and others 1978). The
bacterial load in the test samples (n=3 at each abuse storage condition) and ice stored control
samples (n=15) were compared. The H0 hypothesis that there was no difference in SPB content
between the sets was rejected at a 95% significance level.
3. Results and discussion
The protocols were tested on aseptically filleted cod. Fish pieces were stored on ice, but
exposed to abuse temperatures for various periods during storage.
Analysis of ice stored control samples

Initial studies were performed with ice stored cod fillets in order investigate the correspondence
between predicted and measured quality parameters. The results are shown in Figure 2. The
measured SPB levels corresponded well with predicted values obtained from the Seafood
Spoilage Predictor, assuming initial contamination level of SPB in the fish fillets between 1
and 5 cfu/g. This was in accordance with earlier reported studies (Dalgaard 1993; Skjerdal and
others 2004), numerous storage trials of commercially caught cod from the Barents Sea in the
period 1998-2002 (Skjerdal, unpublished results), and analysis of commercial samples at two
fish processing plants in Norway (Lorentzen and Tryland 2003). A protocol concept based on
SPB levels seemed to give relevant results, and cod from the Oslo fjord suitable raw material
for testing. The two analysis methods for SPBs gave similar results, indicating that the protocol
could be used without access to laboratory facilities.
The distribution of SPB contents in ice stored fillets were described based on analysis of
15 samples (results not shown). It was found that the SPB content in the test samples was
binnenwerk.indd 352 21-08-2006 12:35:21

5
Predicted level
log(SPB)
4
Iron Lyngby Agar
3
FAST
2
0
day 1 day 5 day 11
Figure 2. Predicted and measured contents of sulphide producing bacteria in cod fillets during ice storage.
The predicted values were obtained using the Seafood Spoilage Predictor.
significantly higher than the ice stored control if the mean values were 0.5 or 0.75 log units
higher, using 95 and 99% significance level, respectively.
The fish samples were analysed sensorially during storage. The scores of odour, colour,
appearance and texture increased with storage time (results not shown), but the assessors
tended to overestimate the scores compared to the expected scores around day 5 of the storage
period. According to Grete Hylding (DIFRES, personal communication), assessors need more
training for judging fillets than for gutted fish, and a complete QIM scheme for cod fillets is
still not available. Also, the knowledge about QIM scores at other temperature conditions than
ice storage is limited (Dalgaard 2000), and such knowledge is needed to make analysis of the
time-temperature history possible.
Based on these initial experiments, further studies focused on development of a SPB based
protocol for proving of storage periods at abuse temperatures.
SPB content in ice stored fillets after abuse temperature periods on day 4
Fish samples were stored in ice, with the exception of a period at abuse temperature at
day 4, and the SPB content analysed. The abuse temperatures periods applied were chosen to
give predicted SPB contents 0.1-2.2 log units higher than in the ice stored control samples.
The measured and predicted SPB contents after day 5 and 11 are shown in Figure 3 and 4,
respectively.
On day 5, only the samples that had been exposed to the highest temperature loads, i.e 10 °C
for 24 hours or room temperature for 8 hours, had significantly higher SPB levels than the
ice stored control sample (95% significance level). The measured SPB levels in these samples
were, however, far below the predicted levels (Figure 3). It was assumed that the low measured
values could be due to measurement inaccuracy for low levels of SPBs, as the accuracy of the
FAST method for SPB contents around 1000 SPB/g fish is rather low (Colifast 2004), and the
detection level for agar plate methods is 50-100 cfu/g fish. The fish samples were therefore
stored for one more week and analysed again (Figure 4). In this set of results, none of the
binnenwerk.indd 353 21-08-2006 12:35:22

9
8
log(SPB)/g fish 7
6
5
4
3
2
1
0
e
oo h
oo h
8h
10 h
10 8h
8h
h
ag
24
4
4
24
m
°C
°C
C
or
°C
oo
C
5°
st
10
5°
R
e
ic
Estimated level Iron Lyngby Agar FAST
Figure 3. Predicted and measured contents of sulphide producing bacteria in cod fillets after 5 days of
ice storage, with a period at abuse temperature on day 4. The 95 and 99% significance levels for proving
difference between ice stored samples and abuse temperature samples are marked with horizontal dotted
lines.
9
8
log(SPB)/g fish
7
6
5
4
3
2
1
0
e
2h
4h
8h
4h
°C h
8h
h
ag
24
8
24
m
°C
°C
C
or
oo
oo
oo
5°
C
st
10
10
5°
10
R
e
ic
predicted level FAST
Figure 4. Predicted and measured contents of sulphide producing bacteria in cod fillets after 11 days of ice
storage, with a period at abuse temperature on day 4. The 95 and 99% significance levels for proving difference
between ice stored samples and abuse temperature samples are marked with horizontal dotted lines.
samples at abuse temperature was found different from the ice stored control. Further, no
sensory difference between any of the samples was detected (results not shown).
Based on these results it was concluded that the applied temperature loads, which theoretically
should lead to up to 2.2 log units increased SPB content compared to ice stored control, only
have limited effect on the growth rate of the SPB content in aseptically filleted cod when
applied on day 4 of storage, and as a consequence, SPB measurements can not be used to prove
binnenwerk.indd 354 21-08-2006 12:35:22

abuse temperature storage with temperature loads in this range. The Seafood Spoilage Predictor
has been extensively tested, so the observed lack of correlation between time-temperature
history and SPB level in this experiment seems surprising. On the other hand, the fish used
in this experiment was carefully handled, and the leakage of cellular water within the fish, or
other factors that may have an impact on the growth conditions for SPBs during the first days
of storage, may not be representative for industrially processed fish. Further, the SPB content
in ice stored cod fillets during the first 3-5 days of storage is usually below the detection level
for agar plate determination methods even for industrially processed fish (Skjerdal unpublished
results), and realistic growth experiments in this period are therefore difficult to perform.
SPB contents in ice stored fish after abuse temperature periods after day 4
Further experiments were performed in order to investigate whether the protocol could prove
abuse temperature periods when applied later in the storage period than on day 4. Some of the
results are shown in Figure 5. For the highest temperature loads, i.e. 24 hours at 10 °C and 4
or 8 hours at room temperature, the predicted and measured SPB values were significantly (99%
significance level) higher than in the ice stored control. Further, the final SPB contents in the
fish samples after 11 days of storage were better correlated to the predicted values the later
the abuse temperature period was applied. The smell intensity of the samples with the highest
temperature loads was also slightly higher than the ice stored control and the samples exposed
to lower temperature loads (results not shown).
These observations indicate that storage periods at abuse temperature late in the storage period
have a higher impact on the SPB content compared to the same abuse temperature conditions
applied shortly after catch, and that SPB measurements can be used to prove abuse temperature
storage periods later than day 4.
7
Log spb/g fish
4
control
day 4
day 5
day 7
day 4
day 6
day 4
day 8
day 4
day 8
day 4
day 8
10°C 8h 10°C 24h Room 2h Room 4h Room 8h

predicted level FAST
Figure 5. Predicted and measured contents of sulphide producing bacteria in cod fillets after 11 days of ice
storage, but with one period at abuse temperature. The 95 and 99% significance levels for proving difference
between ice stored samples and abuse temperature samples are marked with horizontal dotted lines.
binnenwerk.indd 355 21-08-2006 12:35:22

Relevance of developed protocol in assessment of time-temperature-contamination history

in commercial samples
The required difference in SPB content between test samples and the ice stored controls to
prove abuse temperature periods was in this work found to be 0.5-0.75 log units. This result
was based on parallel storage trials of test and control samples, but these types of trials are not
relevant or possible to perform for a fish company or supermarket. The SPB content in ice stored
fish has to be predicted from the Seafood Spoilage Predictor. The estimate will necessarily be
based on fish from other geographical areas, seasons and batches, and a higher significance
level is needed to prove deviation in the test samples. Differences of 1 log unit are considered
significant in traditional quantitative microbiology, and a threshold value in the range 1-1.5 log
units could be appropriate. In the present study, the time-temperature conditions that leaded
to significantly higher SPB levels than in the ice stored control would be found significant also
with this criterion (Figure 5).
In aseptically filleted cod, abuse temperature storage and long storage period are the only
reasons for elevated levels of SPB. In analysis of commercial samples, however, contribution
from contamination and the risk that incorrect traceability data have to be taken into account
in the interpretation of the results (Figure 1). The protocol described in this work can not
distinguish between these reasons for elevated SPB levels. However, all the three reasons
represent negative deviations in the supply chain. From a practical point of view, this kind of
information is often the most important one in supplier selection and assessment.
The developed protocol can be used to give an indication of the food safety of the fish. Sulphide
producing bacteria grow even in ice stored fish, while pathogenic bacteria like Listeria spp.
usually need temperatures above 2 °C to give measurable growth (Gimenes and Dalgaard 2003).
Periods at abuse temperatures therefore leads to potential growth of Listeria. It may appear
as a drawback for the developed protocol that it can prove abuse temperature conditions after
day 4 of storage only. Concerning food safety, this is actually a benefit, as contamination of
pathogenic bacteria in cod is most likely to occur during filleting and processing when the
fish flesh is exposed to bio films on equipment and bacteria on hands and knifes. Commercial
filleting is according to the fish industry usually done approximately 3-4 days after catch or
slaughtering, and periods at abuse temperatures after this date are therefore more relevant
for the food safety than abuse temperatures earlier in the storage period. Growth of Listeria
has been extensively studied in products that are consumed raw, like cold smoked salmon. To
our knowledge, there are few similar studies in fresh fish, and there is little information about
Listeria in these types of products. The increasing use of sushi, combined with the decreasing
knowledge of food hygiene and food quality among many modern consumers, the risks initiated
by contamination and abuse temperature storage should be taken seriously, and the developed
protocol can be a tool for risk evaluation.
Growth of spoilage bacteria at dynamic temperature conditions has been studied for a number of
fish species, and good correlations between predicted and measured levels have been obtained
(Dalgaard 1993; Taoukis and others 1999; Giannakourou and others 2005). A monitoring system
based on prediction and measurement of spoilage bacteria, like the one developed in this work,
seems therefore to be relevant for other species than cod. However, the protocol can not be
transferred to species where the growth of SPBs is slow, missing or unpredictable, for instance
halibut (Huss 1995). Further, it should not be used for preserved products where the SPBs have
been killed or inhibited, for instance in modified atmosphere packed fillets, some kinds of
binnenwerk.indd 356 21-08-2006 12:35:23

frozen and thawed fillets, and salt-cured cod products (Barat and others 2005; Bjørkevoll and
others 2003; Skjerdal and others 1999; Dalgaard 1995). It could however be used for lightly
salted cod products, as low salt content does not inhibit growth of SPBs significantly (Skjerdal
unpublished results).
Conclusions
The present study has indicated that deviations in the supply chain of ice stored cod can be
proved from predicted and measured contents of SPBs. A protocol has been developed and found
useful to prove abuse temperature conditions corresponding to temperature loads 8 hours at
room temperature or 24 hours at 10 °C when applied after day 4. The protocol can also be used
to give an indication of the food safety of raw fish products, like sushi.
Acknowledgements
The development of the methodology has been financed by DNV, while the experimental
verification studies have been financed both by DNV and Colifast. The authors thank dr Sigrid
Brynestad and dr Thomas Mestl for valuable comments during the work.
References
Barat JM, Gallart-Jornet L, Andrés A, Akse L, Carlehög M, Skjerdal OT. 2006. Influence of cod freshness on salting,
drying and desalting stages J Food Engin, 73:9-19
Bjørkevoll I, Olsen RL, Skjerdal T. 2003. Spoilage potential and origin of Psychrobacter in salt-cured cod (Gadus morhua).
Int J Food Microbiol 84(2):175-187.
Box GE, Hunter WG, Hunter JS. 1978. Statistics for experimenters. John Wiley & Sons, Inc. 652 p.
Bremner HA. 1985, A convenient easy-to-use system for estimating the quality of hilled seafood. Fish Process Bull, Vol
7. Department of Scientific and Industrial Research, Wellington, NZ, p 59-70
Colifast AS, 2004. Technical note No 5 FAST, Colifast AS
Dalgaard P, Buch P, Silberg S. 2002. Seafood Spoilage Predictor – development and distribution of product specific
application software. Intern J Food Microbiol 73:343-349.
Dalgaard P. 2000. Fresh and lightly preserved seafood. In: Man CMD, Jones AA. Editors, Shelf life evaluation of foods.
2nd ed Maryland: Aspen Publishing, p 110-139
Dalgaard P. 1995. Qualitative and quantitative characterization of spoilage bacteria from packed fish. Intern J Food
Microbiol 26:319-333
Dalgaard P. 1993. Evaluation and prediction of microbial fish spoilage. Ph.d thesis. The Technological Laboratory of
the Danish Ministry of Fisheries and the Royal Veterinary and Agricultural University, Denmark, as cited by Huss
(1995).
Giannakourou MC, Koutsoumanis K, Nychas GJE, Taoukis PS. 2005. Field evaluation of the application of time temperature
integrators for monitoring fish quality in the chill chain. Intern J Food Microbiol 102:323-336
Gimenez B, Dalgaard P. 2004. Modelling and predicting the simultaneous growth of Listeria monocytogenes and spoilage
micro-organisms in cold-smoked salmon. J Appl Microbiol 96(1):96-109
Gram L, Huss HH. 1996. Microbiological spoilage of fish and fish products. Intern J Food Microbiol 33:121–137.
Gram L, Trolle G, Huss HH. 1987. Detection of specific spoilage bacteria from fish stored at low (0 ºC) and high (20 ºC)
temperatures. Intern J Food Microbiol 4:65-72.
Huss HH. (ed), 1995. Quality and quality changes in fresh fish. FAO Fisheries Technical Paper No. 348 FAO, Rome, 195 p.
Jonsdottir SM, Hylding G, Nielsen J, Bleechmore T, Silberg S. 2005. Rapid PC based sensory method rating system.
http://www.dfu.min.dk/QIMRS/qim_0202.htm
binnenwerk.indd 357 21-08-2006 12:35:23

Jørgensen BR, Gibson DM, HH Huss. 1988. Microbiological quality and shelf life prediction of chilled fish. Intern J
Food Microbiol 6:295-307.
Lorentzen G, Skjerdal OT, Berg J. 2003. Rapid method of detection and enumeration of sulphide producing bacteria in
fish products. United states Patent, patent No US 6,632,632 BI
Lorentzen G, Tryland I. 2003. FAST-Utvikling og ferdigstillelse av en hurtigmetode for dokumentasjon av råvare og
produktkvalitet i sjømatindustrien. Fiskeriforskning report 7/2003, Tromsø (In Norwegian).
Luten JB, Martinsdottir E. 1997. QIM, an European tool for fish freshness evaluation in the fishery chain. In: Olafsdottir,
G., Luten, J., Dalgaard, P., Careche, M., Verez-Bagnis, V., Martinsdottir, E., Heia, K. editors, Methods to Evaluate
Fish Freshness in Research and Industry. International Institute of Refrigeration (IIR), Paris, 247-354.
Olafsdottir G, Luten J, Dalgaard P, Careche M, Verres-Bagnis V, Martinsdottir E, Heia K. 1998. Methods to determine the
freshness of fish in research and industry, Proceedings of the final meeting of the concerted action “evaluation of
fish freshness” AIR3CT94 2283. Paris: Institute International du Froid.
Ranneklev S, Berg J. 2005. FAST – a rapid test for determination of spoilage bacteria in fish. Forum News, Microbial
methods forum, Campden & Chorleywood Food Research Association Group, p 1-4
Shewan JM, Macintosh RG, Tucker CG, Ehrenberg ASC. 1953. The development of a numerical scoring system for the
sensory assessment of the spoilage of wet white fish stored in ice. J Sci Food Agric 4:283-298
Skjerdal OT, Lorentzen G, Tryland I, Berg JD. 2004. New method for rapid and sensitive quantification of sulphide
producing bacteria in fish from arctic and temperate waters. Intern J Food Microbiol 93(3):325-333.
Skjerdal OT, Esaiassen M, Løkken GB. 1999. The quality and marketing possibilities of thawed cod fillets can be improved
by adaptation the freezing conditions. In Kennedy, C. editor. The preservation of frozen food quality and safety
throughout the distribution chain. Milan meeting 23-24 April 1999-Plenary 6 proceedings. University of Leeds,
p 89-92.
Taoukis PS, Koutsoumanis K, Nychas GJE. 1999. Use of time-temperature integrators and predictive modelling for shelf
life control of chilled fish under dynamic storage conditions. Intern J Food Microbiol 53:21-31
binnenwerk.indd 358 21-08-2006 12:35:23

Growth kinetic of Staphylococcus aureus during cod (Gadus morhua)
rehydration

Departamento de Inovação Tecnológica e Valorização dos Produtos da Pesca INIAP/IPIMAR,
Avenida de Brasília, 1449-006 Lisboa, Portugal
Abstract
Staphylococcus aureus is able to produce thermostable staphylococcal enterotoxins leading
to food poisoning. It has been identified as the main contaminant of dried/wet salted cod
products and unsafe levels have been detected in samples desalted at 20 ºC. So, growth trials
with different inocula were performed in nutrient broth, cod juice and rehydrated dried salted
cod, at three temperatures and growth curves calculated in order to evaluate S. aureus response
during cod desalting. Levels of approximately 106 S. aureus/g could be reached after desalting
at 20 ºC for periods usually necessary to obtain a palatable product. These findings should be
considered in the implementation of HACCP plans for desalting units.
Keywords: Staphylococcus aureus, food poisoning, cod, Gadus morhua, desalting, growth
kinetic
Introduction
Staphylococcus aureus is a facultative anaerobic Gram-positive coccus. It is non-motile, catalase

and coagulase positive. Some strains are able to produce enterotoxins which are thermostable
and the causative agents of staphylococcal food poisoning (SFP). The enterotoxins generally do
not affect the sensory characteristics of the contaminated food and may therefore go unnoticed
by consumers. The symptoms are abdominal cramps, nausea, vomiting, sometimes followed by
diarrhoea. The onset of symptoms is rapid (from 30 min to 8 hours) and usually spontaneous
remission is observed after 24 hours. Due to the self-limiting nature of this intoxication, its
incidence is thought to be severely underreported. Nevertheless, SFP remains as one of the
leading causes of food-borne illness worldwide (Sandel and McKillip 2004).
This food-borne pathogen is not part of the natural microbial flora of fish. Its habitat is the
skin and mucous membranes of animal and man, being the carrier rate among normal healthy
individuals about 50 percent or more (Arbuthnott 1990). Seafood is usually contaminated by
this bacterium during processing or food-service operations. The organism is extremely salt
tolerant, can survive in brines and was found to survive during several weeks, in salted fish
(Huss and Valdimarsson 1990). S. aureus is a poor competitor with the normal food saprophytic
biota flora and will not multiply in raw fish. However, in seafood products where the growth
of competing organisms is inhibited the presence of staphylococci constitutes a health hazard
under favourable conditions. Enterotoxins do not normally reach levels that will cause food
poisoning until the numbers of the pathogen reach 105 to 106/g (ICMSF 1996). S. aureus
has been shown to grow at temperatures as low as 7 °C, but the lower limit for enterotoxin
production has been shown to be 10 °C.
binnenwerk.indd 359 21-08-2006 12:35:23

Dried salted fish products prepared with cod (Gadus morhua) are popular foods in Mediterranean,
Latin American and South American countries and have NaCl concentrations of approximately
20%. They are usually considered shelf-stable and are therefore often sold unrefrigerated
and unpacked. These products are manipulated with minimum protection against casual
contamination and Gram-positive cocci represent frequently the predominant microflora
(Vilhelmsson and others 1997; Pedro and others 2002). Before consumption, these products
have to be rehydrated (desalted) during 24-72 hours and sometimes this process is performed
at room temperature, leading occasionally to products with unsafe levels of S. aureus, with
possible toxin formation (Pedro and others 2004).
As there is limited data on this subject, it is necessary to obtain additional growth data in
this food system in order to formulate recommendations to industry. The kinetic parameters of
microbial growth can be described by mathematical models and the modified Gompertz model has
been commonly used. However, some authors have reported that this model can overestimate
the growth rate by 10-20% (Dalgaard and others 1994) and in some cases provides negative
lag times. The log transformed Logistic model with four parameters has been referred to as the
best model for growth estimated by viable counts (Dalgaard and Koutsoumanis 2001).
The aim of this work was to study the growth kinetic of S. aureus during the rehydration of
dried salted cod under different temperature, inoculum and substrate conditions, in order to
evaluate its response during cod desalting.
In the experiments S. aureus ATCC 25923 was cultivated at different temperatures in three
substrates: Brain Heart Infusion containing 3% NaCl (BHI) (Oxoid, England); Desalted Cod Juice
(CJ), prepared according to Gram and others (1987) using desalted cod at 4 ºC for 48 hours; and
rehydrated dried salted cod portions (CD, Desalted Cod), prepared according to Barat and others
(2004). For BHI and CJ, the inoculation level was 103 and the growth temperatures were 37, 20
and 10 ºC. For CD, the inoculation levels were 103, 104, 105 and the growth temperatures were
20 ºC for all the contamination levels and 10 ºC for the lower contamination level. Experiments
were run in triplicate.
Growth in the different substrates was periodically measured with plate count on duplicate
spread plates of Baird Parker Medium (Merck, Germany). This assessment was performed until
cultures were well over 106 bacteria/ml or g, when applicable. The measured viable counts were
transformed to a base 10 logarithm; averages and standard deviations of the transformed values
were then calculated (Schaffner 1998).
The growth rate was estimated from the slope of the experimental curve at the exponential
phase, using linear regression analysis. The lag phase was estimated from the period between
the initial point and the point where the regression line for the exponential phase intersected
a horizontal line through the initial point on the semi-logarithmic plot (Dalgaard 1995).
The kinetic parameters of the predicted curves were determined using the modified Gompertz
model (Eq. (1)) and the log-transformed four-parameter Logistic model (Eq. (2)). Gompertz
modified equation estimates the logarithm of the relative population size [y = ln(N/N0)] and
describes a sigmoidal curve containing parameters that are microbiologically relevant, such as
binnenwerk.indd 360 21-08-2006 12:35:23

Growth kinetic of Staphylococcus aureus during cod rehydration
the lag phase (λ), maximum growth rate (µm) and maximal value reached at the stationary
phase (A) (Zwietering and others 1990). In the log-transformed form of the four-parameter
Logistic model the number of microorganisms at time (t) is estimated and the parameters
minimum (Nmin) and maximum (Nmax) asymptotic numbers of microorganisms, maximum growth
rate (µm) and time at inflection point (ti) are calculated.
{ [
y = A • exp -exp m
μ •e
A
(λ - t) + 1 ]} (1)
Log (Nt) = Log {Nmin + 1 + exp

Nmax - Nmin
[-m (t-t )]
} (2)
m i
Fitting of the predicted curves was performed by minimizing the sum of the squares of the
differences between the predicted values and those measured. The statistical significance of
the difference between models in terms of goodness-of-fit to the same set of data was assessed
by using the F test described by Motulsky and Ransnas (1987) for comparing two models
with different number of parameters. The following equation was used: F = [(SS1 - SS2)/(df1
- df2)]/(SS2/df2), where SS is the residual sum of squares of each model, df is the number of
degrees of freedom, subscript 1 refers to the model with fewer parameters and 2 to the model
with more parameters.
The comparison of data fits showed that the more complex model did not show a significantly
superior ability to fit experimental data than the simpler extensively used modified Gompertz
model. In this context, microbiological parameters were calculated and growth curves constructed
using the latter model. The microbiological parameters for S. aureus growth curves obtained in
the three substrates, and using the fitted growth curves general equations of modified Gompertz
model are presented in Table 1.
Table 1. Microbiological parameters determined using the fitted growth curves for S. aureus BHI (brain heart
infusion), CJ (desalted cod juice) and CD (rehydrated dried salted cod) under different conditions.
Lag phase λ (h) Maximum growth rate µm Maximal value reached at the
(h-1) stationary phase A (Log cfu)
BHI, 103, 37 ºC 2.7 1.79 9.05

BHI, 103, 20 ºC 7.3 0.33 8.33
BHI, 103, 10 ºC 42.1 0.07 7.86
CJ, 103, 37 ºC 1.7 1.13 8.72
CJ, 103, 20 ºC 7.7 0.22 8.92
CJ, 103, 10 ºC 45.3 0.06 8.75
CD, 105, 20 ºC 17.8 0.21 8.25
CD, 104, 20 ºC 11.5 0.09 8.05
CD, 103, 20 ºC 14.3 0.07 7.44
CD, 103, 10 ºC 44.2 0.03 4.50
binnenwerk.indd 361 21-08-2006 12:35:24

In general, under the same conditions the liquid substrates (BHI and CJ) had superior ability to
support the growth of S. aureus than the rehydrated dried salted cod (CD), presenting shorter
lag phases, higher maximum growth rates and reaching higher maximal values at the stationary
phase.
As expected temperature markedly influenced the microbiological parameters determined in all

growth substrates. This effect was particularly evident in the lag phase and maximum growth
rates in liquid substrates and maximal values in solid substrate (CD). In fact, the lag phases
obtained in the former substrates increased from approximately 2 to 45 hours with the lowering
of temperature. On the other hand, µm decreased from approximately 1.8 to 0.06 h-1 when
reducing the temperature from 37 to 10 ºC (Figure 1). In the case of CD, the maximal values
increased from 4.5 to 7.4 log units by changing temperature from 10 to 20 ºC.
The comparison of the growth parameters at 20 ºC in liquid and solid substrates showed that in
the latter substrate the lag phase and the maximum growth rate of bacteria were approximately
the double and one fourth of those observed in the liquid substrates, respectively. However, at
10 ºC minor differences in the lag phases were observed within substrates and the maximum
growth rate of bacteria in CD was only one half of those registered in BHI and CJ. Concerning
maximal values reached at 20 ºC differences of one log unit were observed between solid and
liquid substrates, whereas at 10 ºC the differences were about 3.5 log units.
7
Log CFU/ml
1
0 12 24 36 48 60 72 84 96 108 120 132 144 156 168
Time (hours)
Figure 1. Experimental data of S. aureus growth in brain heart infusion (close symbol) and in desalted cod
juice (open symbol) at 37 ºC (¢,£), 20 ºC (p,r) and 10 ºC (ò,ô) and fitted growth curves obtained
from the Gompertz model.
binnenwerk.indd 362 21-08-2006 12:35:24

Growth kinetic of Staphylococcus aureus during cod rehydration
In CD the initial microbial load did not seem to influence both the lag phase and the maximum
growth rate at 20 ºC. On the other hand, higher inoculum levels led to higher maximal values at
the stationary phase for the same temperature (Figure 2). In the absence of data obtained with
similar substrates, results were compared with those referred for steak and kidney pie mix at 21
ºC with contamination levels of 103 (ICMSF 1996). In rehydrated dried salted cod the values of
lag phase (14.3 hours) and generation time (3.6 hours) were slightly higher than those reported
in meat pie, where those parameters were 8 and 2.3 hours respectively.
In general, the data obtained with CD as growth substrate were rather different from those
observed in both liquid substrates, in spite of one of them being prepared from cod. These
differences may be due to the lower NaCl concentration in the liquid substrates (BHI and CJ)
and also to the presence of some competing bacteria in the experiments with desalted cod
(CD). These findings suggest that nor BHI neither CJ are suitable to model S. aureus growth
during cod desalting.
The growth curves of S. aureus in desalted cod (CD) may allow to predict the time needed to
attain unsafe levels of 106 organisms/g, with possible toxin formation. Data showed that this
level was reached approximately after 24, 35 and 46 hours for initial contamination levels
of 105, 104 and 103 respectively, at 20 ºC. However, desalting at 10 ºC of contaminated raw
material (S. aureus ≤ 103 organisms/g) does not allow attaining unsafe levels.
Conclusion
Desalting at room temperature (20 ºC) can lead to unsafe levels of S. aureus (106 organisms/g)
for periods usually necessary to obtain a palatable product. Desalting at 10 ºC of contaminated
raw material (S. aureus ≤ 103 organisms/g) does not allow attaining unsafe levels. It is
recommended desalting cod at temperatures below 10 ºC.
10
9
8
7
Log CFU/g
6
5
4
3
2
1
0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170
Time (hours)
Figure 2. Experimental data of S. aureus growth in rehydrated dried salted cod at 20 ºC (close symbol) and
at 10 ºC (open symbol) for initial inoculum levels of 105 (¢) 104 (p) and 103 (ò,ô) and fitted growth
curves obtained from the Gompertz model. Growth level for potential toxin production (–––).
binnenwerk.indd 363 21-08-2006 12:35:25

Acknowledgements
This work has been financed by the Project 22-05-01-FEDER-000008 “Surveillance, Quality and
Food Safety” of the QCAIII-MARE Program.
References
Arbuthnott JP. 1990. Staphylococcal toxins in human disease. J Appl Bacteriol Suppl 101S–107S.
Barat JM, Rodryguéz-Barona S, Andres A, Visquert M. 2004. Mass transfer analysis during the cod desalting process.
Food Res Int 37:203-208.
Dalgaard P. 1995. Modelling of microbial activity and prediction of shelf life for packed fresh fish. Int J Food Microbiol
26:305-317.
Dalgaard P, Ross T, Kamperman L, Neumeyer K, McMeekin TA. 1994. Estimation of bacterial growth rates from
turbidimetric and viable count data. Int J Food Microbiol 23:391-404.
Dalgaard P, Koutsoumanis K. 2001. Comparison of maximum specific growth rates and lag times estimated from
absorbance and viable count data by different mathematical models. J Microbiol Methods 43:183-196.
ICMSF (International Commission on Microbiological Specifications for Foods). 1996. Microorganisms in Food. Vol. 5:
Microbiological specifications of food pathogens. London: Blackie Academic and Professional. 513 p.
Gram L, Trolle G, Huss HH. 1987. Detection of specific spoilage bacteria from fish stored at low (0 ºC) and high (20 ºC)
temperatures. Int J Food Microbiol 4:65-72.
Huss HH, Valdimarsson G. 1990. Microbiology of salted fish. FAO Fish Tech News 10(1):1-2.
Motulsky HJ, Ransnas LA. 1987. Fitting curves to data using nonlinear regression: a practical and nonmathematical
review. FASEB J 1:365-374.
Pedro S, Magalhães N, Albuquerque MM, Batista I, Nunes ML, Bernardo FM. 2002. Preliminary observations on spoilage
potential of flora from desalted cod (Gadus morhua). J Aqua Food Prod Technol 11(3/4):143–150.
Pedro S, Albuquerque MM, Nunes ML, Bernardo FM. 2004. Pathogenic bacteria and indicators in salted cod (Gadus
morhua) and desalted products at low and high temperatures. J Aqua Food Prod Technol 13(3):39-48.
Sandel MK, McKillip JL. 2004. Virulence and recovery of Staphylococcus aureus relevant to the food industry using
improvements on traditional approaches. Food Control 15:5-10.
Schaffner DW. 1998. Predictive food microbiology Gedanken experiment: why do microbial growth data require a
transformation? Food Microbiol 15:185-189.
Vilhelmsson O, Hafsteinsson H, Kristjánsson JK. 1997. Extremely halotolerant bacteria characteristic of fully cured and
dried cod. Int J Food Microbiol 36:163-170.
Zwietering MH, Jongenburger I, Rombouts FM, Van’t Riet K. 1990. Modelling the bacterial growth curve. App Environ
Microbiol 56(6):1875-1881.
binnenwerk.indd 364 21-08-2006 12:35:25

Aeromonas spp. and potential indicators in mussels in Northern
Greece
Amin Abrahim, Eleni Iossifidou, Nikolaos Soultos, Dimitrios Koutsopoulos, Ioannis Tzavaras
and Pavlos Koidis
Laboratory of Hygiene of Foods of Animal Origin, Department of Hygiene and Technology of Foods
of Animal Origin, Faculty of Veterinary Medicine, Aristotle University of Thessaloniki, 54124
Thessaloniki, Greece
Abstract
The objective of the present study was to investigate the presence of potential indicators
(coliforms, faecal coliforms and Escherichia coli I) and Aeromonas spp. in mussels (Mytilus
galloprovincialis) in Northern Greece. A total of 59 mussel samples were analyzed. Most probable
number (MPN) of total coliforms, faecal coliforms and E. coli I were determined. Total coliforms,
faecal coliforms and E. coli I were detected in 80%, 54% and 44% of the samples and the MPN
values ranged from 2.30 to 3.69 log10/100 g, from 2.30 to 3.52 log10/100 g and from 2.30
to 3.36 log10/100 g respectively. Aeromonas hydrophila was isolated from 22% of the mussel
samples. Samples with high level of faecal coliforms and E. coli I have to be purified after
harvest according to the Directive 91/492/EEC.
Keywords: mussel, total coliforms, faecal coliforms, Escherichia coli I, Aeromonas spp.
Introduction
Mussels (Mytilus galloprovincialis) are the most efficient and indiscriminate feeders (Hackney
and Rippen 2000) and are notoriously liable to become contaminated with bacteria derived
from human sewage (ICMSF 1998).
Some strains of Aeromonas spp. are considered to be important pathogens to human,

amphibian, reptiles and fish (Goodwin and others 1983; Austin and Allen-Austin 1985).
Aeromonas (principally A. hydrophila) has the status of food borne pathogen of emerging
importance (Daskalov 2005). A. hydrophila is widely distributed in the environment (Hazen
and others 1978) and has been isolated almost in all food samples such as fish, seafood, red
meat, poultry (Palumbo and others 1985), raw milk and other milk products (Melas and others
1999). Aeromonas spp. have been isolated from such other aquatic animals as finfish, shellfish,
crustaceans, and amphibians. Both finfish and shellfish are known reservoirs of this bacterium
(Abeyta and others 1994).
Some strains of A. hydrophila produce a hemolysin and an enterotoxin and may cause disease
in humans ranging from mild diarrhea to life-threatening febrile gastroenteritis (Hubbert and
others 1996). A. hydrophila is of great concern to immunocompromised patients with underlying
malignancies and foods containing high levels of Aeromonas spp. destined for these individuals
should be considered as hazardous (Abeyta and others 1994).
binnenwerk.indd 365 21-08-2006 12:35:25

A. Abrahim, E. Iossifidou, N. Soultos, D. Koutsopoulos, I. Tzavaras and P. Koidis
Faecal coliforms are specific indicators exhibiting a high positive correlation with faecal
contamination from warm-blooded animals (Ŝolić and others 1999). Most countries use faecal
coliforms and E. coli I as indicators of faecal pollution or to establish the safety of shellfish or
the purity of water in the growing areas (Ahmed 1991; Hubbert and others 1996; Ho and Tam
2000). Analysis of faecal coliforms and E. coli are believed to have limited predictive value for
viral pathogens such as enteroviruses (Muniain-Mujika and others 2002).
In Greece, no documented information is available regarding the enumeration of total coliforms,

faecal coliforms and E. coli I and only few information on the presence of Aeromonas spp. in
mussels is available. The present study was hence undertaken to investigate possible presence
of these microorganisms in mussels harvested in shellfish growing areas of Northern Greece.
Sample collection and preparation

Fifty-nine samples of mussels (Mytilus galloprovincialis) were provided by the National Reference
Laboratory on Marine Biotoxins, Center of Veterinary Institution of Thessaloniki. All samples
came from Thermaikos Gulf (Northwest Aegean Sea), which is the largest shellfish culture area
of Greece (ca. 40,000 tonnes/year of mussels) (Grigoriadou and others 2005). The samples
were obtained during the eight-month period from October 2004 to May 2005. All samples,
1-2 days old, were transported to the laboratory in ice chest at 4o C and were analyzed within
1 hour of arrival.
Each sample (25 to 30 mussels, approximately 100 –180 g) was scrubbed with a clean brush
under running water to remove debris, allowed to dry, disinfected with 70% ethanol, and
opened aseptically using a sterile shucking knife. Both flesh and intervalvular fluids were
removed aseptically, pooled together and homogenized for 1-2 min using a Stomacher 400 Lab
Blender (Seward Medical, London, UK).
Bacteriological analyses
• Counts of total coliforms, faecal coliforms and Escherichia coli I: Total coliforms, faecal
coliforms and Escherichia coli I were analyzed with the most-probable number (MPN)
procedure based on the multiple-tube fermentation technique, five tubes with three dilutions
(Motes and others 1984; Hitchins and others 1995).
• Isolation of Aeromonas spp.: Twenty-five grams of mussel flesh and intervalvular fluid were
homogenized in 225 ml Tryptone Soy broth (Difco Laboratories, Detroit, USA), containing
30 μg/ml of ampicillin, using a Stomacher 400 Lab Blender (Seward Medical, London, UK)
for 2 min. The homogenate was incubated for 24 hours at 28 °C and one loopful of the
enriched culture was streaked onto Starch Ampicillin (SA) agar (Palumbo and others 1985)
and incubated at 28 °C for 24 hours.
SA agar was prepared according to Palumbo and others (1985) by using 31 g of phenol red
agar base (Difco Laboratories, Detroit, Mich.), 10 g of soluble starch (Riedel-de Haën, Seelze,
Germany) and 1000 ml of distilled water. After autoclaving at 121 °C for 15 min and tempering
to 50 °C, ampicillin (Sigma Chemical Co., St. Louis, USA) was added to achieve a concentration
of 10 μg/ml.
binnenwerk.indd 366 21-08-2006 12:35:25

Aeromonas spp. and potential indicators in mussels in northern Greece
SA agar plates were flooded with ca. 5 ml Lugol iodine solution (Palumbo and Buchanan 1988).
Yellow to honey coloured colonies surrounded by a clear zone were scored as presumptive A.
hydrophila. Five presumptive A. hydrophila were selected from each plate and transferred to
nutrient agar (Difco Laboratories, Detroit, USA) slants and incubated at 28 °C for 24 hours for
further studies. All Gram-negative, oxidase-positive isolates were further screened by using the
following biochemical tests: O/F test, esculin hydrolysis, growth in KCN broth, gas production
from glucose, motility test, resistance to the vibriostatic agent 0/129 and growth in nutrient
broth containing 0 or 6% sodium chloride (Popoff 1984). Isolates yielding a typical reaction
to one or more of the above test were characterized as “nonclassified” (Kirov 2001). Isolates
with positive reactions for the above tests were further confirmed by API-20E kit (bioMérieux,
Marcy-l’Etoil, France) as A. hydrophila (Carnahan and others 1991).
According to European Union regulation (EU Directive 91/492) live bivalve molluscs intended for
immediate human consumption must contain less than 300 (2.48 log10/100 g) faecal coliforms
or less than 230 (2.36 log10/100 g) E. coli I per 100 g of mollusc flesh and intravalvular liquid
based on a five-tube, three-dilution MPN-test or any other bacteriological procedure shown to
be equivalent accuracy. In the present study, 43 (73%) and 44 (75%) of the analyzed mussel
samples had less than 300 faecal coliforms/100 g and 230 E. coli/100 g, respectively. These
shellfish were considered suitable for direct consumption. Sixteen (27%) and 15 (25%) of
mussel samples had more than 300 faecal coliforms/100 g and 230 E. coli/100 g and the MPN
values ranged from 2.65 to 3.52 log10/100 g and from 2.65 to 3.36 log10/100 g, respectively.
Therefore, these shellfish should be relayed to reduce the bacterial load after harvesting by
sending them to the depuration center. A purification (depuration) process is a controlled
process using only sterilized sea water to allow live, filter feeding shellfish to purge themselves
naturally of most microorganisms which may have accumulated from the environment (Bird
1996). The use of faecal indicators provides the probability, not the certainty, of the presence
of pathogens. In fact, the detection of microbial indicators constitutes only the presumption of
the risk; the presence of the pathogens can be verified only by specific analyses or by evaluating
epidemiological data (Bonadonna and others 2002).
Total coliforms were not detected in 12 (20%) of mussel samples examined. In 28 samples
(47%) and 19 samples (32%), the MPN values of total coliforms ranged from 2.30 to 3.04
log10/100 g and from 3.11 to 3.69 log10/100 g respectively.
It is known that the concentration of faecal coliforms in edible shellfish tissue gives an
indication of the potential health hazard to consumers of shellfish due to pathogens of faecal
origin, which may have been present in the environment surrounding the mussel breeding areas
(Ŝolić and others 1999). The temporary contamination of the tested mussel samples might be
attributed to the bacterial loads of the main rivers in Greece and the drainage ditches that
flow in Thermaikos Gulf. In addition, the prevailing currents within the area of the Gulf could
disperse the discharge of the bacteria from the river mouth till the mussel breading areas
leading to the contamination of the mussels.
According to Guillon-Cottard and others (1998), on a number of occasions in the summer,

the faecal coliforms count in mussels from Saint-Gervias harbour greatly exceeded European
binnenwerk.indd 367 21-08-2006 12:35:25

guidelines for shellfish farming areas, i.e. 300 faecal coliforms per 100 g. This is due to the fact
that the boating activity was highest in the summer.
In the present study Aeromonas spp. were detected in 13 (22%) samples out of 59 total tested.
A total of 16 strains were isolated from 13 samples, coming from several samples having two
or three Aeromonas spp. All of the 16 strains were A. hydrophila.
A number of studies have reported the incidence of Aeromonas spp. in different shellfish
species. Colakoglu and others (2005) reported a higher frequency of A. hydrophila (32%) in
shellfish harvested off Dardanelles cost of Turkey; this is probably due to seasonal factors since
the samples were collected during the two-month period from May to June. In Italian study, out
of 144 of mussel samples examined, 32 (22%) were positive for Aeromonas species. In total,
22 strains belonged to A. hydrophila, 8 to HG2 unnamed species and 2 to A. caviae (Ottaviani
and others 2005). In a retail survey of New Zealand seafood, of the food categories tested,
shellfish had the highest incidence of motile aeromonads, 66% of samples being positive, and
58% positive for A. hydrophila and/or A. sorbia (Hudson and De Lacy 1991). In Greece, one out
of five mussel samples (Mytilus edulis) examined was reported to harbour A. hydrophila (Davies
and others 2001).
Conclusion
From the present work and the samples analyzed it is shown that in some cases (16% of mussel
samples) the concentrations of indicators of faecal contamination exceed the concentrations
established by European Legislation (Directive 91/492). Thus, it is of paramount importance
that the mussels with high level of faecal coliforms and E. coli are purified after harvest
according to the Directive E.C. 91/492. Therefore, the safety of shellfish growing areas is of
great concern in order to minimize the potential health risks associated with the consumption
of molluscan shellfish.
The present work highlights an important prevalence of Aeromonas hydrophila in mussels

collected from Thermaikos Gulf (Northwest Aegean Sea). During the eight month study period,
a homogeneous distribution of A. hydrophila positive samples was detected. The presence of
motile aeromonads in 22% of the mussel samples represents a possible public health concern,
since the Aeromonas group is considered as “emerging pathogen” and is autochthon of the
aquatic environment. This concern is magnified by the fact that motile aeromonads can grow
at refrigeration temperatures. When the products are cooked before consumption and no
contamination takes place afterwards, the risk will be completely eliminated since the bacteria
and their toxins are heat sensitive (Feldhuson 2000).
Acknowledgements
We thank the National Reference Laboratory on Marine Biotoxins, Center of Veterinary Institutions
of Thessaloniki for their contribution to the sampling.
References
Abeyta JRC, Palumbo SA, Stelma JRGN. 1994. Aeromonas hydrophila group. In: Hui YH, Gorham JR, Murrell KD, Cliver DO,
editors. Foodborne Disease Handbook. Disease caused by bacteria. Volume 1. New York: Marcel Dekker. p 1-27.
binnenwerk.indd 368 21-08-2006 12:35:26

Aeromonas spp. and potential indicators in mussels in northern Greece
Ahmed FE. 1991. Seafood safety. Committee on evaluation of the safety of fishery products. Washington DC: National
Academy Press. 432 p.
Austin B, Allen-Austin D. 1985. A review – bacterial pathogens of fish. J Applied Bacteriol 58:483-506.
Bird PD. 1996. HACCP assessment of oyster depuration in Australia. In: Martin RE, Collette RL, Slavin JW, editors. Fish
Inspection, Quality control, and HACCP. A global focus. Lancaster, Basel: Technomic Publishing Co. p 330-340.
Bonadonna L, Briancesco R, Coccia AM, Semproni M, Stewardson D. 2002. Occurrence of potential bacterial pathogens
in coastal areas of the Adriatic Sea. Environ Monit Assess 77:31-49.
Carnahan AM, Behram S, Joseph SW. 1991. Aerokey II: A Flexible key for identifying clinical Aeromonas species. J Clin
Microbiol 29:2843-2849.
Colakoglu FA, Sarmasik A, Koseoglou B. 2006. Occurrence of Vibrio spp. And Aeromonas spp. in shellfish harvested off
Dardanelles cost of Turkey. Food Control. Forthcoming.
Council Directive, 91/492/EEC. 1991. Laying down the health conditions for the production and the placing on the
market of live bivalve molluscs. Official Journal of the European Committees L268/1-14.
Daskalov H. 2006. The importance of Aeromonas hydrophila in food safety. Food Control. Forthcoming.
Davies AR, Capell C, Jehanno D, Nychas GJE, Kirby RM. 2001. Incidence of foodborne pathogens on European fish.
Food control 12:67-71.
Feldhuson F. 2000. The role of seafood in bacterial foodborne diseases. Microbes infect 2:1651-1660.
Goodwin CS, Harper WES, Stewart JK, Gracey M, Burke V, Robinson J. 1983. Enterotoxigenic Aeromonas hydrophila and
diarrhoea in adults. Med J Australia 1:25-26.
Grigoriadou IK, Mouratidou T, Katikou P. 2005. Investigation on the presence of domoic acid in Greek shellfish. Harmful
Algae 4:717-723.
Guillon-Cottard I, Augier H, Console JJ, Esmieu O. 1998. Study of microbiological pollution of a pleasure boat harbour
using mussels as bioindicators. Mar Environ Res 45:239-247.
Hackney CR, Rippen TE. 2000. The molluscan shellfish industry. In: Martin RE, Carter EP, Flick JR GJ, Davis LM, editors.
Marine and freshwater products handbook. Lancaster: Technomic Publishing Co. p 323-338.
Hazen TC, Fliemans CB, Hirsch RP, Esch GW. 1978. Prevalence and distribution of Aeromonas hydrophila in the United
States. Appl Environ Microbiol 36:731-738.
Hitchins AD, Feng P, Watkins WD, Rippey SR, Chandler LA. 1995. Escherichia coli and the coliform bacteria. In: FDA
Bacteriological Analytical Manual. 8th ed. Gaithersburg: AOAC International. p 4.01-4.29.
Ho BSW, Tam TY. 2000. Natural depuration of shellfish for human consumption: a note of caution. Water Res 34:1401-
1406.
Hubbert WT, Hagstad HV, Spangler E, Hinton MH, Huges KL. 1996. Food safety and Quality Assurance. Foods of Animal
Origin. 2nd ed. Iowa: Iowa State University Press. 72 p.
Hudson JA, De Lacy KM. 1991. Incidence of motile Aeromonads in New Zealand retail foods. J Food Protect 54:696-
699.
ICMSF. 1998. Microorganisms in foods, Microbial ecology of food commodities. London: Blackie Academic and
Professional. 615 p.
Kirov SM. 2001. Aeromonas and Plesiomonas species. In: Doyle MP, Beuchat LR, Montville TJ, Miliotis MD, Bier JW,
editors. Food Microbiology. Fundamentals and frontiers. 2nd ed. Washington DC: ASM Press. p 301-327.
Melas DS, Papageorgiou DK, Mantis AI. 1999. Enumeration and confirmation of Aeromonas hydrophila, Aeromonas
caviae, and Aeromonas sobria isolated from raw milk and other milk products in Northern Greece. J Food Protect
62:463-466.
Motes ML, McPherson RM, McPaola A. 1984. Comparison of three international methods with APHA method for
enumeration of Escherichia coli in estuarine waters and shellfish. J Food Protect 47:557-561.
Muniain-Mujika I, Calvo M, Lucena F, Girones R. 2002. Comparative analysis of viral pathogens and potential indicators
in shellfish. Int J Food Microbiol 83:75-85.
Ottaviani D, Santarelli S, Bacchiocchi S, Masini L, Ghittino C, Bacchiocchi I. 2006. Occurrence and characterization of
Aeromonas spp. in mussels from the Adriatic Sea. Food Microbiol. Forthcoming.
binnenwerk.indd 369 21-08-2006 12:35:26

Palumbo SA, Maxino F, Williams AC, Buchanan RL, Thayer DW. 1985. Starch ampicillin agar for quantitative detection
of Aeromonas hydrophila. Appl Environ Microbiol 50:1027-1030.
Palumbo SA, Buchanan RL. 1988. Factors affecting growth or survival of Aeromonas hydrophila in foods. J Food Safety
9:37-51.
Popoff M. 1984. Aeromonas. In: Krieg NR, Holt JG. editors. Bergey’s Manual of Systematic Bacteriology, Vol. 1. Baltimore:
The Williams and Wilkins Co. p 545-548.
Ŝolić M, Krstulović N, Jozić S, Curać D. 1999. The rate of concentration of faecal coliforms in shellfish under different
environmental conditions. Environ Int 25:991-1000.
binnenwerk.indd 370 21-08-2006 12:35:26

Characterisation of Vibrio parahaemolyticus isolated from seafood
products
Departamento de Inovação Tecnológica e Valorização dos Produtos da Pesca INIAP/IPIMAR,
Avenida de Brasília, 1449-006 Lisboa, Portugal
Abstract
Vibrio parahaemolyticus is a gram-negative halophilic autochthonous bacterium in the
coastal marine environments that occasionally is present in inland aquacultures and salt-
water environments. It is found in the gut and on the surface of marine animals and may be
accumulated by bivalves. The consumption of raw or undercooked contaminated seafood may
cause human illness, this pathogen being a leading cause of food borne gastroenteritis. The
pathogenicity of this bacterium is mainly associated with the possession of two haemolysins,
thermostable direct haemolysin (TDH) and thermostable related haemolysin (TRH). Besides,
a positive Kanagawa phenomenon (KP) and the production of urease have been described as
markers to predict the potential pathogenicity of this microorganism. The aim of this work was
to characterise selected strains of V. parahaemolyticus isolated from shellfish harvesting areas
and imported seafood products for biochemical and physiological properties and presence of
virulence factors. Isolates were assayed for biochemical properties, salt tolerance and identified
according to the analytical profile index of API 20E and 20NE. Species identification was also
assayed by PCR using primer sets for tlh and toxR. The detection of the haemolysin genes
was performed by PCR using two sets of primers. Selected correctly sized amplicons were
confirmed by sequencing. Isolates exhibited diverse analytical profiles and API 20NE yielded
the higher percentage of identification. Most of the strains identification was confirmed by PCR
results. Isolates trh-positive were detected. The obtained results indicate that pathogenic V.
parahaemolyticus may be present in food/environmental samples and suggest that a health risk
can be associated with their consumption.
Keywords: V. parahaemolyticus, identification, virulence factors, sequence analysis, toxR gene
Introduction
Vibrios are primarily associated with estuarine and coastal marine environments and can be
found in the gut of marine fish and on the surface of crustaceans. Usually these bacteria are
halophilic and easily destroyed by heat, but they possess the ability to develop very fast when
temperature and nutrient availability are adequate. The occurrence of halophilic vibrios in
seafood does not result from anthropogenic pollution and so, its presence cannot be estimated
by the values of faecal contamination used to monitor the quality of coastal waters.
Although most of marine vibrios are non pathogenic for humans there is a group present in
coastal marine waters that can cause illness to men. Currently there are 11 pathogenic species
to humans: V. alginolyticus, V. cholerae, V. cincinnatiensis, V. damsela, V. fluvialis, V. furnissii, V.
hollisae, V. metschnikovii, V. mimicus, V. parahaemolyticus, V. vulnificus (Elliot and others 1995).
binnenwerk.indd 371 21-08-2006 12:35:26

Clinically the more important pathogenic species to men are V. cholerae, V. parahaemolyticus
and V. vulnificus. Among these, V. parahaemolyticus has been identified as the major cause of
foodborne gastroenteritis in Japan (ICMSF 1996). The organism is of particular significance in
that country, and accounts for 40 to 70% of the nation’s food poisoning (Varnam and others
1996). In the U.S.A., 40 foodborne outbreaks by V. parahaemolyticus were reported to CDC
from 1973 to 1998 (Daniels and others 2000). In most European countries, this bacterium
is not considered as a major health problem and consequently has not been included in
routine foodborne outbreaks investigation. Nevertheless, a shellfish-associated outbreak by V.
parahaemolyticus was recently reported in Spain (Lozano-León and others 2003).
The pathogenesis of V. parahaemolyticus is still not clearly understood, but the relatively rapid
onset of the disease suggests the involvement of an enterotoxin. Usually, clinical isolates of V.
parahaemolyticus possess a thermostable direct haemolysin (TDH), which causes beta-haemolysis
of human erythrocytes in agar medium (Wagatsuma agar), i.e. Kanagawa phenomenon and is
used as virulence marker. V. parahaemolyticus ability to cause illness can also be related to
others virulence factors, such as the production of a thermostable direct haemolysin related
toxin (TRH) which correlates with the ability to hydrolyse urea (CCFH 2002). TDH and TRH
haemolysins, considered the major virulence factors of V. parahaemolyticus, are encoded by the
tdh and trh genes respectively, and have multiple biological functions like haemolytic activity,
enterotoxicity, citotoxicity and cardiotoxicity. The urease positive phenotype is relatively rare
and is usually associated with the possession of the trh gene, making the urea hydrolysis a
good virulence marker for diagnostics (Suthienkul and others 1995; Park and others 2000),
although this phenotype does not imply the presence of trh gene (Nakaguchi and others 2003).
The strains possessing the tdh gene, the trh gene or both genes are strongly associated with
clinical cases, regardless of their KP reaction, whereas the distribution frequency of these genes
in environmental strains is very low (Osawa and others 1993).
Current conventional methods for V. parahaemolyticus identification are time consuming and often
give variable results. Consequently, several laboratories have implemented DNA-based assays
for the detection of genes specific of these bacteria, such as tlh and toxR, being most methods
based in polymerase chain reaction (PCR). On the other hand, the detection of haemolysins
implies the use of agar containing fresh human blood, which is difficult to obtain. In this
context, the use of molecular biology techniques can also be a good alternative allowing the
direct detection of the genes encoding both haemolysins (tdh and trh genes). So, in this work
selected strains of V. parahaemolyticus, isolated from shellfish harvesting areas and imported
seafood products, were characterised using conventional biochemical and physiological tests,
rapid identification systems and DNA-based assays. The presence of virulence factors was also
determined.
Bacterial strains
Thirty environmental strains of Vibrio isolated from shellfish collected from Portuguese harvesting
areas - mussels (Mytillus edulis and M. galloprovincialis) and oysters (Crassostrea angulata) -
and frozen crustacean imported from third countries were further characterized by biochemical
tests, including cytochrome oxidase and catalase activities, motility, sensitivity to vibriostatic
agent O/129 (10 and 150 µg), growth in peptone water at different NaCl concentrations, Voges-
binnenwerk.indd 372 21-08-2006 12:35:26

Characterisation of Vibrio parahaemolyticus isolated from seafood products
Proskauer (VP) reaction, arginine dihydrolase and for the urea hydrolysis in Christensen’s agar
(Elliot and others 1995).
For identification of the isolated strains, API 20E and API 20NE (Biomérieux, France) galleries
were used. V. parahaemolyticus strains were confirmed by detection of the toxR and tlh genes,
using specific primers (Bej and others 1999; Kim and others 1999). For the production of
β-haemolysis (Kanagawa phenomenon) strains were tested in Wagatsuma agar. The presence
of the tdh and trh genes (encoding TDH and TRH haemolysins) was determined by PCR using
specific primers (Bej and others 1999). Three V. parahaemolyticus reference strains ATCC 17802
(tdh-; trh+); ATCC 27519 (tdh-; trh-) and ATCC 43996 (tdh+; trh-) were used as controls.
DNA extraction
Isolates were grown in alkaline peptone water at 30 ºC with agitation at 100 rpm, overnight.
Total genomic DNA was extracted according to the protocol for bacteria supplied with the
QIAmp® DNA Mini Kit (QIAGEN, Hilden, Germany).
Oligonucleotide primers and PCR amplifications

PCR assays were performed using the oligonucleotide primers pairs for tlh, tdh, trh and toxR
(Table 1), custom synthesized by Biomers (Ulm, Germany).
Each 50 µL of PCR mixture contained 1 μL of purified genomic DNA, 10 pmol/µL of each

oligonucleotide primer, 5 μL of a 10 x concentrated PCR reaction buffer (Amersham, England),
200 µM of each of the dNTP’s (Amersham, England) and 5 units of AmpliTaq DNA polymerase
(Amersham, England).
All PCR amplifications were performed in a thermo cycler (BioRad, iCycler, USA) using the
following temperature-cycling parameters to amplify the tlh, tdh and trh genes: initial
denaturation at 94 ºC for 3 minutes followed by 30 cycles of amplification; each cycle consisted
of denaturation at 94 ºC for 1 min., primer annealing at 58 ºC for 1 min, and primer extension at
72 ºC for 1 min. Following the amplification cycles, samples were kept at 72 ºC for 5 minutes to
allow final extension of the incompletely synthesized DNA. For the toxR gene, the amplification
conditions consisted in an initial denaturation step at 96 ºC for 5 minutes followed by 20 cycles
of amplification consisting of denaturation at 94 ºC for 1 min, annealing at 63 ºC for 1.5 min,
and extension at 72 ºC for 1.5 min. This was followed by incubation at 72 ºC for 7 minutes.
Table 1. Oligonucleotide primers used to amplify bacterial DNA.
Gene Primer Sequence Size amplicon Source
tlh Tlh D 5’ – aaa gcg gat tat gca gaa gca ctg– 3’ 450 pb Bej and others
Tlh R 5’ – gct act ttc tag cat ttt ctc tgc – 3’ 1999
tdh Tdh D 5’ – gta aag gtc tct gac ttt tgg ac – 3’ 269 pb Bej and others
Tdh R 5’ – tgg aat aga acc ttc atc ttc acc – 3’ 1999
trh Trh D 5’ – ttg gct tcg ata ttt tca gta tct – 3’ 500 pb Bej and others
Trh R 5’ – cat aac aaa cat atg ccc att tcc g – 3’ 1999
toxR Kim D 5’ – gtc ttc tga cgc aat cgt tg – 3’ 368 pb Kim and
Kim R 5’ – ata cga gtg gtt gct gtc atg – 3’ others 1999
binnenwerk.indd 373 21-08-2006 12:35:27

Detection of PCR amplicons

PCR amplicons were separated in a 2% agarose gel (Amersham, England) in which 2 x 10 -4 μg
per ml of ethidium bromide (Amersham, England) were added. Electrophoresis was performed
using 0.5 x concentrated Tris-Borate-EDTA (TBE) buffer pH 8.3 and a voltage of 5 V/cm for 45
min. Following electrophoretic separation, DNA fragments in the gels were visualized under a
GelDoc 2000 UV transilluminator (BioRad, Hercules, USA) and photographed in digital format.
toxR gene sequencing and sequence analysis

Separation of PCR amplified DNA was performed in a 1% agarose gel containing 2 x 10 -4 μg
per ml of ethidium bromide. Electrophoresis was performed using 0.5 x concentrated TBE buffer
pH 8.3 and a voltage of 5 V/cm for 45 min. Fragments of the expected size were cut, weighted
and purified using JETQUICK® - Spin kit Gel Extraction System (Genomed, Löhne, Germany)
according to the supplier’s recommendations.
Nucleotide sequences of purified DNA were determined using the Thermo Sequenase™ Primer
Cycle Sequencing Kit (Amersham, England). For sequencing, an aliquot (3µl) of a master mix
containing 12µl of DNA and 1µl of each Dye-primer (Kim D and Kim R - 10pmol/µl) was used.
The thermo cycler was programmed as follows: 30 cycles of 30s at 95 ºC, 30s at 55 ºC, 1 min at
72 ºC (Direct) and 30 cycles of 30 s at 95 ºC, 30 s at 50 ºC, 1 min at 72 ºC (Reverse).
After completion of the cycling program, the tubes were centrifuged briefly to collect the
contents at the bottoms of the tubes and placed on ice. Formamide loading dye (6µl) was
added to each tube and mixed well. Each sample was heated at 72 ºC for 3 min and immediately
placed on ice. Ten µl of each sample were loaded into separate lanes of the gel. An ALFexpress
II (Amersham Pharmacia Biotech, England) automated DNA sequencer was used and both DNA
strands were sequenced.
Nucleotide sequences were aligned with the CLUSTAL W (version 1.82) multiple sequence alignment
program (Thompson and others 1994), and the alignments were refined by visual inspection.
Differences were found between the two API systems used for Vibrio identification, for the same
confidence level (Figure 1). Using API 20E system seventeen environmental strains presented
no identification while using API 20NE all strains were identified. Besides, API 20NE system
yielded the higher percentage of very good identification. The results for API 20E and API 20NE
identifications and PCR amplifications of the tlh and toxR fragments were correlated for the
strains tested. Only 6 of 30 (20%) strains showed a correlation between API 20E identification
and amplification of the tlh gene. Five of these strains (17% of the total) showed a correlation
between API 20E identification and amplification of the toxR gene. One strain, identified by
API 20E as V. parahaemolyticus (92.8%) was confirmed by the presence of amplicon with the
correct molecular weight for the tlh gene but not for toxR gene. Concerning API 20NE, 21 of
the 30 (70%) strains showed a correlation between API identification and amplification of the
tlh gene; 20 of the 30 tested strains (67%) showed a correlation between API identification
and amplification of the toxR gene. One strain, identified by API 20NE as V. parahaemolyticus
(96.9%) was confirmed by the presence of fragments with the correct molecular weight for the
tlh gene but not for toxR gene. Moreover, API 20NE identified one strain as V. parahaemolyticus
(83.9%) but this result was not confirmed by either of the molecular methods used. Three
binnenwerk.indd 374 21-08-2006 12:35:27

API results
>99%
97-99%
90-97%
80-90%
Non Vp
Non ID
0 5 10 15 20 25 30
20E 20NE
Figure 1. Values of the percentage of the identification for V. parahaemolyticus by API systems 20E and
20NE (Non Vp - not identified as V. parahaemolyticus; Non ID - not identified).
strains, identified by both API systems as non-V. parahaemolyticus, yielded fragments with the
correct molecular weight for the tlh gene but not for toxR gene.
Given the considerable variation of the API identification results, the species confirmation by
PCR has shown to be useful. However, discrepancies were observed between the two PCR assays
results.
Although Bej and others (1999) claimed that the tlh gene was specific for V. parahaemolyticus
and could be used as a species-specific marker, Croci and others (2001) found 30% of V.
alginolyticus also produced thermolabile toxic substances. Also Robert-Pillot and others (2002)
reported that primers targeting the tlh gene cannot be used to differentiate V. parahaemolyticus
from V. alginolyticus, once they obtained amplicons with both strains, showing that these two
species display sequence similarities in the region binding to the oligonucleotide primers. This
can explain why three of our strains identified as V. alginolyticus yielded fragments with the
correct molecular weight for the tlh gene.
Of the strains identified as V. parahaemolyticus by the API systems, 55% were ONPG +, 60%
LDC +, 5% ODC +, 25% CIT +, 5% TDA +, 100% IND +, 100% GEL +, 100% GLU +, 100% MAN +,
10% RHA +, 10% SAC +, 100% MEL +, 35% AMY +, 95% ARA +, 100% OX +. Two URE + strains
were detected using Christensen’s agar containing 2% NaCl.
Relatively to the detection of virulence factors (TDH and TRH), no tdh+ strain was detected.
Two tested strains, isolated from live mollusc bivalves, presented an amplicon with the correct
molecular weight for the trh gene (500 base pairs) in agarose gel (Figure 2). All the reference
strains used as controls yielded results according to the expected. Both trh-positive strains had
the ability to hydrolyse urea, showing a high correlation between these factors, as described by
other authors. Besides, one of them produced β-haemolysis on Wagatsuma agar (KP-positive)
in spite of being tdh-negative.
binnenwerk.indd 375 21-08-2006 12:35:27

1 2 3 4 5 6 7
Figure 2. Agarose gel electrophoresis with PCR amplification of genomic DNA from V. parahaemolyticus, using
oligonucleotide primers specific for trh gene. Lanes: 1 – 50 bp DNA ladder; 2 – negative environmental strain;
3 – positive environmental strain; 4 – negative environmental strain; 5 – positive environmental strain;
6 – Positive control (reference strain); 7 – Negative control.
Although some authors (Nakaguchi and others 2003) claim that the distribution frequency of
the trh gene in environmental strains is very low, our data (still based on very few positive
strains) indicates that a considerable amount of the examined environmental strains can
exhibit virulence-related properties and suggest that urease-positive phenotype is common
in specific harvesting areas. Several authors have already reported the presence of the trh
gene in environmental isolates differing only in the percentage found. For instance, Shirai
and others (1990) reported 7% positive strains while Hervio-Heath and others in 2002 found
5% positive strains. Our results of 6.7% positive strains are in agreement with these authors.
Nevertheless, such values are much lower than those reported more recently in oysters along
the coast of India, with 59.3% trh positive samples (Deepanjali and others 2005). So, the use
of identification/quantification methods for V. parahaemolyticus, without considering their
virulence, underestimates the number of environmental pathogenic strains. These findings
reveal the presence of potentially pathogenic V. parahaemolyticus in the marine environment,
which can be easily introduced in the food chain. Moreover, some seafood (including filter-
feeding bivalves) are usually consumed raw or undercooked, representing a potential hazard
for the consumer’s health.
The resulting sequences for sixteen environmental strains (3, 4, 5, 6, 7, 8, 9, 10, 13, 14, 16,
18, 19, 21, 23, 27) and the reference strain B (ATCC 27519) were aligned and compared with
the reference strain A (ATCC 17802) sequence (Figure 3). All resulting nucleotide sequences
were submited to the BLAST program (Karlin and Altschul 1990) in GenBank, and aligned with
all V. parahaemolyticus strains existing in the database. The reference strain A used in this
study revealed 100% nucleotide identity to V. parahaemolyticus strain ATCC 17802 with the
acession number AY527396. Strain 21, isolated from frozen crustacean imported from a third
country, presented the lower nucleotide identity to strain A (96%), presenting at least 6 base
pair substitutions. The higher nucleotide identity to strain A (99%) was obtained for strain 14,
isolated from shellfish collected from a harvesting area, which differed only by one bp.
Although most sequences are very similar, they presented some common substitutions, being
the substitution of an A for a G in position 286 the most frequent, appearing in 11 sequences.
binnenwerk.indd 376 21-08-2006 12:35:28

strain bp
A CGTTGAACCA GAAGCGCCAG TAGTACCTGA AAAAGCACCT GTGGCTTCTG CTGTGAATCC 60
3 .......... .......... .......... .......... .......... .......... 60
4 .......... .......... .......... .......... .......... .......... 60
5 .......... .......... .......... .......... .......... .......... 60
6 .......... .......... .......... .......... .......... .......... 60
7 ......C... .......... .........- .......... .......... .......... 59
8 .......... .......... .........- .......... .......... .......... 59
9 .......... .......... .........- .......... .......... .......... 59
10 .......... .......... .........- .......... .......... .......... 59
13 .......... .......... .......... .......... .......... .......... 60
14 .......... .......... .......... .......... .......... .......... 60
16 .......... .......... .........- .......... .......... .......... 59
18 .......... .......... .......... .......... .......... .......... 60
19 .......... .......... .......... .......... .......... .......... 60
21 .........G AG........ .......... .......... .......... .......... 60
23 .......... .......... .........- .......... .......... .......... 59
27 ......C... .......... ....C..... .......... .......... ....N..... 60
B .......... .......... .......... .......... .......... .......... 60
****** ** ******** **** **** ********** ********** **** *****
A TTGGATTCCA CGCGTTATTT TATTTTTGGC ACTATTACTA CCGATTTGCG TACTGCTGTT 120
3 ......-... A....-.... .......... .......... .......... .......... 118
4 .......... .......... .......... .......... .......... .......... 120
5 .......... .......... .......... .......... .......... .......... 120
6 ..N....... .......... .......... .......... .......... .......... 120
7 .......... .......... .......... ...G...... .......... .......... 119
8 ..N....... .......... .......... ...G...... .......... .......... 119
9 .......... .......... .......... .......... .......... .......... 119
10 .......... .......... .......... .......... .......... .......... 119
13 A......... .......... .......... .......... .......... .......... 120
14 .......... .......... .......... .......... .......... .......... 120
16 .......... .......... .......... .......... .......... .......... 119
18 .......... .......... .......... .......... .......... .......... 120
19 ..N....... .......... .......... ...G...... .......... .......... 120
21 A......... .......... .......... .......... .......... .......... 120
23 .......... .......... .......... .......... .......... .......... 119
27 .......... .......... .......... .......... .......... .......... 120
B .......... .......... .......... .......... .......... .......... 120
* *** *** **** **** ********** *** ****** ********** **********
A TACAAACCCT GCGGAATCTC AGTTCCGTCA GATTGGTGAG TATCAGAACG TACCAGTGAT 180
3 .......... .......... .......... .......... .......... .......... 178
4 .......... .......... .......... .......... .......... .......... 180
5 .......... .......... .......... .......... .......... .......... 180
6 .......... .......... .......... .......... .......... .......... 180
7 .......... .......... .......... .......... .......... .......... 179
8 .......... .......... .......... .......... .......... .......... 179
9 .........A .......... .......... .......... .......... .......... 179
10 .........A .......... .......... .......... .......... .......... 179
13 .......... .......... .......... .......... .......... .......... 180
14 .......... .......... .......... .......... .......... .......... 180
16 .......... .......... .......... .......... .......... .......... 179
18 .......... .......... .......... .......... .......... .......... 180
19 .......... .......... .......... .......... .......... .......... 180
21 .......... .......... .......... .......... .......... .......... 180
23 .........A .......... .......... .......... .......... .......... 179
27 .......... .......... .......... .......... .......... .......... 180
B .........A .......... .......... .......... .......... .......... 180
********* ********** ********** ********** ********** **********
Figure 3. Nucleotide sequence alignment of a portion of the toxR fragment (amplified using primers Kim D
and Kim R) from 16 environmental and 2 reference strains of V. parahaemolyticus (A – ATCC 17802 and B
– ATCC 27519). “.” means identical bp to the reference strain A sequence, “-” means no base, and “*” means
identical bp in all strains. On the left are indicated the total number of bp of each sequence.
binnenwerk.indd 377 21-08-2006 12:35:28

A GACACCTGTA AATCACCCGC AAATCAACAA CTGGTTGCCT TCTATTGAGC AGTGCATTGA 240

3 .......... .......... .......... .......... .......... .......... 238
4 .......... .......... .......... .......... .......... .......... 240
5 .......... .......... .......... .......... .......... .......... 240
6 .........N .......... .......... .......... .......... .......... 240
7 .......... .......... .......... .......... .......... .......... 239
8 .......... .......... .......... .......... .......... .......... 239
9 .......... .......... .......... .......... .......... .......... 239
10 .......... .......... .......... .......... .......... .......... 239
13 .......... .......... .......... .......... .......... .......... 240
14 .......... .......... .......... .......... .......... .......... 240
16 .......... .......... .......... .......... .......... .......... 239
18 .......... .......... .......... .......... .......... .......... 240
19 .......... .......... .......... .......... .......... .......... 240
21 .......... .......... .......... .......... .......... .......... 240
23 .......... .......... .......... .......... .......... .......... 239
27 .......... .......... .......... .......... .......... .......... 240
B .......... .......... .......... ....C..... .......... .......... 240
********* ********** ********** **** ***** ********** **********
A ACGCTACGTT AAGCACCATG CAGAAGACTC GTTACCAGTG GAAGTAATTG CCACTGGCGG 300
3 .......... .......... .......... .......... .....G.... .......... 298
4 .......... .......... .......... .......... .....G.... .......... 300
5 .......... ..A....... .......... .......... .......... .......... 300
6 .......... ..A....... .......... .......... .......... .......... 300
7 .......... .......... .......... .......... .......... .......... 299
8 .......... .......... .......... .......... .......... .......... 299
9 .......... .......... .......... .......... .....G.... .......... 299
10 .......... .......... .......... .......... .....G.... .......... 299
13 .......... .......... .......... .......... .....G.... .......... 300
14 .......... .......... .......... .......... .....G.... .......... 300
16 .......... .......... .......... .......... .....G.... .......... 299
18 .......... .......... .......... T......... .....G.... .......... 300
19 .......... .......... .......... .......... .......... .......... 300
21 .......... .......... .......... C......... .....G.... .......... 300
23 .......... ..A....... .......... .......... .......... ......CG.. 299
27 .......... .......... .......... C......... .....G.... .T........ 300
B .......... .......... .......... .......... .....G.... .......... 300
********** ** ******* ********** ********* ***** **** * **** **
A ACAAAATAAC CA 312
3 .......... .. 310
4 ..GN...... .. 312
5 .......NN. .. 312
6 .......GG. .. 312
7 ..-....... .. 310
8 ..RV...NN. .. 311
9 ..N....... .. 311
10 ..-....... .. 310
13 .......... .. 312
14 .......... .. 312
16 .......... .. 311
18 .N-....... .. 311
19 ..-N...R.. .. 311
21 ..NN...DD. .. 312
23 ..-N...NN. .. 310
27 NN-....... .. 311
B .NBN...... .. 312
*** * **
Figure 3. Continued.
binnenwerk.indd 378 21-08-2006 12:35:28

In position 130, 4 sequences presented a substitution of a T for an A, and in only 3 sequences

there were substitutions of an A instead of a G in position 94, and a G instead of an A in
position 253. This data, obtained from sequence analysis, can be of great value for the design
of specific oligonucleotide primers for PCR. In spite of the detected substitutions, possibly point
mutations, no significant genetical diversity was detected among the characterized strains.
The percentage of nucleotide identity in a 310 bp portion obtained for all of the above strains
with primers Kim D and Kim R of the toxR fragment ranged from 94 to 100%. Strains 6 and
27, isolated from shellfish collected from different harvesting were the less closely related
sequences, as they presented 94% of nucleotide identity. The different geographical origin of
the isolates might be associated with different selection pressures and consequently favour
genetic diversity, justifying the differences detected in toxR sequences. On the other hand,
strains 14 and 16, isolated from shellfish harvested from the same production area, showed
100% nucleotide identity.
Conclusion
There are potentially pathogenic (trh-positive) V. parahaemolyticus strains in the marine

environment, which can easily be introduced in the food chain, creating a potential hazard for the
consumer’s health. The public health significance of V. parahaemolyticus environmental strains
should be further assessed. These facts should be taken into consideration when elaborating
plans for microbiological surveillance systems for harvesting areas. Currently, integrated
methodologies for the identification/quantification of (potentially) pathogenic strains in
environmental/food samples should be adopted, using molecular biology techniques.
Acknowledgements
This work was performed within the Project 44.03.13.IFP.0007 of the QCAIII-MARIS Program
and within the Integrated Project (IP) SEAFOODplus, granted by the EU Commission under
contract No FOOD-CT-2004-506359.
References
Bej AK, Patterson DP, Brasher CW, VickeryMCL, Jones DD, Kaysner CA. 1999. Detection of total and hemolysin-producing V.
parahaemolyticus in shellfish using multiplex PCR amplification of tl, tdh and trh. J Microbiol Meth 36:215-225.
CCFH (Codex Committee on Food Hygiene). 2002. Discussion paper on risk management strategies for Vibrio spp. in
seafood. CX/FH 03/5-Add.3.
Croci L, Serratore P, Cozzi L, Stacchini A, Milandri S, Suffredini E, Toti L. 2001. Detection of Vibrionaceae in mussels
and in their seawater growing area. Lett Appl Microbiol 32:57-61.
Daniels NA, MacKinnon L, Bishop R, Altekruse S, Ray B, Hammond RM, Thompson S, Wilson S, Bean NH, Griffin PM, Slutsker.
2000. L. Vibrio parahaemolyticus infections in the United States, 1973-1998. J Infect Dis 181(5):1661-1666.
Deepanjali A, Kumar HS, Karunasagar I, Karunasagar I. 2005. Seasonal variation in abundance of total and pathogenic Vibrio
parahaemolyticus bacteria in oysters along the Southwest Coast of India. Appl Environ Microbiol 71(7):3575-3580.
Elliot EL, Kaysner CA, Jackson L, Tamplin ML. 1995. Capítulo 9. Vibrio cholerae, Vibrio parahaemolyticus, Vibrio vulnificus,
and other Vibrio spp. In: FDA Bacteriological Analytical Manual. 8ª edição. Gaithersburg: AOAC International. p
9.01-9.27.
Hervio-Heath D, Colwell RR, Derrien A, Robert-Pillot A, Fournier JM, Pommepuy M. 2002. Occurrence of pathogenic
vibrios in coastal areas of France. J Appl Microbiol 92:1123-1135.
binnenwerk.indd 379 21-08-2006 12:35:29

ICMSF (International Commission on Microbiological Specifications for Foods). 1996. Microorganisms in Food. Volume
5: Microbiological specifications of food pathogens. London: Blackie Academic and Professional. 513 p.
Karlin S, Altschul SF. 1990. Methods for assessing the statistical significance of molecular sequence features by using
general scoring schemes. Proc Natl Acad Sci USA 87:2264-2268.
Kim YB, Okuda J, Matsumoto C, Takahashi N, Hashimoto S, Nishibuchi M. 1999. Identification of Vibrio parahaemolyticus
strains at the species level by PCR targeted to the toxR gene. J Clinical Microbiol 37(4):1173-1177.
Lozano-León A, Torres J, Osorio CR, Martínez-Urtaza J. 2003. Identification of tdh-positive Vibrio parahaemolyticus from
an outbreak associated with raw oyster consumption in Spain. FEMS Microbiol Lett 226:281-284.
Nakaguchi Y, Okuda J, Iida T, Nishibuchi M. 2003. The urease gene cluster of Vibrio parahaemolyticus does not influence
the expression of the thermostable direct hemolysin (TDH) gene or the TDH-Related hemolysin gene. Microbiol
Immunol 47(3):233-239.
Osawa R, Okitsu T, Morozumi H, Yamai S. 1996. Occurrence of urease–positive Vibrio parahaemolyticus in Kanagawa,
Japan, with specific reference to presence of thermostable direct hemolysin (TDH) and the TDH-Related hemolysin
genes. Appl Environ Microbiol 62(2):725-727.
Park K, Iida T, Yamaichi Y, Oyagi T, Yamamoto K, Honda T. 2000. Genetic characterization of DNA region containing the
trh and ure genes of Vibrio parahaemolyticus. Infect Immun 68(10):5742-5748.
Robert-Pillot A, Guenole A, Fournier JM. 2002. Usefulness of R72H PCR assay for differentiation between Vibrio
parahaemolyticus and Vibrio alginolyticus species: validation by DNA-DNA hybridization. FEMS Microbiol Lett 215:1-6.
Robert-Pillot A, Guénolé A, Lesne J, Delesmont R, Fournier JM, Quilici ML. 2004. Occurrence of the tdh and trh genes
in Vibrio parahaemolyticus isolates from waters and raw shellfish collected in two French coastal areas and from
seafood imported into France. Int J Food Microbiol 91:319-325.
ShiraiH, Ito H, Hirayama T, Nakamoto Y, Nakabayashi N, Kumagai K, Takeda Y, Nishibuchi M. 1990. Molecular
epidemiologic evidence for association of thermostable direct hemolysin (TDH) and TDH-related hemolysin of
Vibrio parahaemolyticus with gastroenteritis. Infect Immun 58(11):3568-3573.
Suthienkul O, Ishibashi M, Iida T, Nettip N, Supavej S, Eampokalap B, Makino M, Honda T. 1995. Urease production
correlates with possession of the trh gene in Vibrio parahaemolyticus strains isolated in Thailand. J Infect Dis
172(5) :1405-1408.
Thompson JD, Higgins DG, Gibson YJ. 1994. CLUSTALW: improving the sensitivity of progressive multiple sequence
alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids
Res 22:4673-4680.
Varnam AH, Evans MG. 1996. Foodborne Pathogens. England: Manson Publishing. 557 p.
binnenwerk.indd 380 21-08-2006 12:35:29

Inactivation of Listeria innocua isolated from fish products by
pulsed light

AZTI-Tecnalia, Food Research Division, Txatxarramendi Ugartea z/g, 48395 Sukarrieta (Bizkaia),
Spain
Abstract
The impact of pulsed light technology to inactivate microorganisms isolated from fish products
has been examined. Comparing the time constant of this process and the microbial inactivation
curves, it is shown that a brief exposure of bacteria to light can inactivate them. For instance,
a short treatment time (325 µs) at relatively low doses (0.7 J.cm-2) induced more than 7 Log
CFU.cm-2 reduction in the culturability/survival of Listeria innocua. Cell inactivation increased
with the intensity of light pulse and the number of pulses. Pulsed light efficacy would depend
on the light dose received by bacteria. Although additional studies are needed, pulsed light
technology appears to be a promising non thermal process to decontaminate the surface of
some fish products.
Keywords: pulsed light, microbial inactivation, non-thermal process, fish products, Listeria
Introduction
Foodstuffs have traditionally been thermally processed to preserve them. However, thermal
technologies may cause undesirable changes (Barbosa-Canovas and others 1998). Furthermore,
consumers have a growing preference for convenience and lightly preserved food products. The
need to enhance food preservation process efficiency and food quality has led to improvements
in existing processes and developments in emerging non-thermal technologies (Barbosa-
Canovas and others 1997, 1998). A novel non-thermal process such as pulsed light technology
could be applied to avoid the cooked appearance of food products resulting from conventional
thermal treatments. This technology could also be used as an alternative to traditional chemical
treatments to prevent the spoilage of foodstuffs. In particular, this process could improve safety
and increase shelf life of lightly preserved fish products (e.g. cold smoked fish products).
The pulsed light technology consists of high power pulses of UV light or broadband emission
light (approx. 200-1000 nm) with a considerable amount of light in the short-wave UV spectrum
(Dunn and others 1995; Wekhof 2000). The efficacy of pulsed light process is greater than
the continuous UV light (McDonald and others 2000), which may be caused by its biggest
dissipation and penetration power (Krishnamurthy and others 2004). Even though the peak
power of each pulse is high because of its short duration, the total pulse energy is relatively
low. Therefore, since the average power requirement for a pulsed light treatment is moderate,
it can be considered economical (Dunn 1996).
The pulsed light process has been approved by the US Food and Drug Administration (2003)
for applications in food products after being evaluated for both safety and effectiveness:
binnenwerk.indd 381 21-08-2006 12:35:29

treatments up to 12 J.cm-2 could be applied. This process is effective to inactivate a wide

broad of microorganisms (vegetative bacteria, yeast, moulds, microbial and fungal spores,
viruses and protozoan oocysts) and enzymatic systems involved in food products spoilage
(Dunn and others 1995; Dunn 1996; Martínez de Marañón and Gartzia 2002). The specific
mechanisms by which pulsed light causes cellular inactivation still remain unclear. However,
since the spectrum of pulsed light includes UV, it is expected that UV-inducing changes, such
as formation of pyrimidine dimers in DNA (McDonald and others 2000), play an important role.
Otherwise, Wekhof (2000) attributed the inactivation to cell disintegration after instantaneous
overheating of cellular constituents. Krishnamurthy and others (2004) found only minimal
heating after pulsed light treatments, inactivation would then occur because of the pulsed
light and not the temperature increase.
Although some studies have been carried out to inactivate microorganisms by pulsed light, only
a few have characterised the impact of process factors on microbial inactivation. Furthermore,
the inactivation of microorganisms isolated from seafood products by pulsed light has never
been studied. The aim of this work was to estimate the effectiveness of this decontamination
process and to establish efficient conditions to inactivate target micro-organisms, i.e. bacteria
isolated from fish products, inoculated onto solid food models.
Microorganism
Microorganisms isolated from fish products, were grown on BHI (Brain heart Infusion) broth. In
order to estimate the critical factors of the pulsed light process, the strain EU2174 of Listeria
innocua was selected from this collection. This strain could be considered as a surrogate for the
human pathogenic bacteria Listeria monocytogenes (pathogenic bacteria isolated from several
lightly preserved fish products).
Listeria innocua EU2174 was stored at -80 ºC in BHI broth cryotubes supplemented with 20%
glycerol. Prior to cell culture, a preculture was carried out at 37 ºC for 24 hours after thawing
the cryotubes. Cell culture tubes were inoculated at 103 cells.mL-1 by transferring the adequate
volume from the preculture tubes. This strain was then grown at 37 ºC in not shaken BHI culture
tubes until early stationary growth phase (20 hours; 109 cells.mL-1 ).
Stationary phase cells were harvested by centrifugation (10000xg, 4 ºC, 15 min), washed twice
in KPBS (Potassium Phosphate Buffered Saline; pH: 6.7) and finally resuspended in this buffer
at a cell density of approximately 109 cells.mL-1. Then, cell suspension was serially diluted in 1%
buffered peptone water and each dilution was surface plated onto BHI 1.5% agar petri dishes
or cold smoked salmon (CSS) 10 minutes before pulsed light treatment.
Pulsed light treatment

The pulsed light treatments were performed using a SteriBeam device (Germany). The emission
of a single light pulse is characterized by several consecutive steps (Figure 1). The electric
power is stored in an energy storage capacitor and later released quickly to a Xenon lamp
(Wekhof 2000). Then, this lamp emits short and high intensity light flashes that are transmitted
to the surface of the products in the treatment camera.
binnenwerk.indd 382 21-08-2006 12:35:29

Inactivation of Listeria innocua isolated from fish products by pulsed light
Reflector
Capacitor Switch
Xenon lamp
Energy
(electrical power supply)
Figure 1. Pulsed light process.
To characterize the pulsed light process, time constant of each pulse of light was measured by
a photodiode placed at the same position (5.5 cm from the light strobe) than samples (see
below). Before measurements, the photodiode was calibrated with a reference pulsed light
source with a pulse duration and spectra similar to the xenon flash lamp used in this work. This
time constant would allow to estimate the exposure time required to inactivate microorganisms
and cell death kinetics therefore.
Samples (Petri dishes) inoculated with L. innocua were subjected to different pulsed light
treatments. The impact of process factors such as pulse intensity, treatment time and number
of pulses was evaluated at room temperature (23 ºC) at a constant distance (5.5 cm) from the
light strobe.
Culturability-viability evaluation
Inoculated Petri dishes, untreated (control) and treated samples, were directly incubated at
37 ºC for 48 h. For preliminary validation tests with cold smoked salmon, Petri dishes containing
cells recovered from the surface of inoculated CSS were also incubated at 37 ºC for 48 h. After
counting colonies (Colony Forming Units), plates were left at 37 ºC for 7 days more in order to
verify that no additional increment in CFU occurred. The results were expressed in CFU.cm-2.
All experiments were repeated at least three times.
Pulsed light process consists of a successive repetition of light pulses. Figure 2 shows the
electrical signal (mV) modification measured by a photodetector placed at the same position
than inoculated samples. As it is shown in this figure, the duration of each pulse was 325
µs whereas 125 µs were required to reach the pulse peak. The period of time between two
consecutive pulses was approximately 27s for the device used in this work.
After characterizing the time constant of the process, the antimicrobial effectiveness of different
pulsed light treatments was evaluated. Samples (Petri dishes inoculated with L. innocua) were
exposed to two different pulse energy fluxes: 0.35 and 0.7 J.cm-2. Figure 3 shows that pulsed
light inactivated L. innocua (> 7 Log CFU) at relatively low doses (0.7 J.cm-2) and short
treatment time (325 µs). Similar results have been found for other spoilage and pathogen
microorganisms isolated from fish products (results not shown). This figure points out that cell
inactivation increased with the intensity of the single pulse. Hence, the culturability/survival
decreased with treatment time because of the increase in the number of pulses. Therefore,
microbial inactivation could depend on the intensity of the treatment.
binnenwerk.indd 383 21-08-2006 12:35:30

600
500 325 µs
Voltage (mV) 400
300
200
100
125 µs
0
0 50 100 150 200 250 300 350 400 450 500
Time (µs)
Figure 2. Time constants for a single light pulse.
9
8
Inactivation (Log CFU.cm-2)
7
6
5
4
3
2
1
0
0 1 2 3 4 5
Number of pulses
Figure 3. Effect of pulsed light treatments on L. innocua inactivation. ô: 0.35 J.cm-2 per pulse; ò: 0.7
J.cm-2 per pulse. Dashed line represents the limit of detection.
Results from Figure 3 have been taken into account to represent Figure 4, which shows L.
innocua inactivation for light doses ranging from 0.35 to 2.8 J.cm-2. This figure points out that
cell inactivation increased with the light dose, i.e. the higher the microbial exposure to light
the better efficacy of the decontamination process.
Tests performed with CSS pieces show that pulsed light is also effective in inactivating L.
innocua from the surface of lightly preserved fish products. The surface of treated CSS pieces is
almost completely decontaminated contrary to the control CSS surface (Figure 5).
binnenwerk.indd 384 21-08-2006 12:35:30

Inactivation of Listeria innocua isolated from fish products by pulsed light
9
8
Inactivation (Log CFU.cm-2)

7
6
5
4
3
2
1
0
0.00 0.35 0.70 1.05 1.40 1.75 2.10 2.45
Light dose (J.cm-2)
Figure 4. Effect of light dose on L. innocua inactivation. ô: 0.35 J.cm-2 per pulse; ò: 0.7 J.cm-2 per pulse.
Dashed line represents the limit of detection.
Control
Treated
Figure 5. L. innocua colonies, previously inoculated on the surface of cold smoked salmon pieces, before
and after treatment by pulsed light.
Conclusions
The pulsed light technology is a novel process which could be used as an efficient treatment to
inactivate Listeria inocolu isolated from fish products. The antimicrobial effectiveness depends
on the pulse intensity and the number of pulses. Additional studies should be performed to
prove the suitability of this process to reduce surface contamination (spoiling and pathogenic
bacteria) of lightly preserved fish products.
binnenwerk.indd 385 21-08-2006 12:35:31

Acknowledgements
This work was performed within the Integrated Project (IP) SEAFOODplus, partially granted by
the EU Commission under contract NºFOOD-CT-2004-506359, the Department of Agriculture
and Fisheries from the Basque Government and the Spanish Ministry for Education and Science
(FIT-060000-2004-285).
References
Barbosa-Canovas GV, Góngora-Nieto MM, Swanson BG. 1998. Nonthermal electrical methods in food preservation. Food
Science Technology International 4:363-370.
Barbosa-Canovas GV, Pothakamury UR, Palou E, Swanson BG. 1997. Nonthermal Preservation of foods. New York: Marcel
Dekker. 276 p.
Dunn J. 1996. Pulsed light and electric field for foods and eggs. Poultry Science 75:1133-1136.
Dunn J, Ott T, Clark W. 1995. Pulsed-light treatment of food and packaging. Food Technology 49:95-98.
FDA. 2003. Pulsed light for the treatment of food. Code of Federal Regulations 21CFR179.41:443.
Krishnamurthy K, Demirci A, Irudayaraj J. 2004. Inactivation of Staphylococcus aureus by pulsed UV-light sterilization.
Journal of Food Protection 67(5):1027-1030.
Martínez de Marañón I, Gartzia I. 2002. Non-thermal treatments to food preservation: Pulsed Light. Proceedings of
Symposium on Emerging Technologies for the Food Industry. Madrid. p 199
McDonald KF, Curry RD, Clevenger TE, Unklesbay K, Eisenstark A, Golden J, Morgan RD. 2000. A comparison of pulsed
and continuous ultraviolet light sources for the decontamination of surfaces. IEEE Transactions on Plasma Science
28(5):1581-1587.
Wekhof A. 2000. Disinfection with flash lamps. Journal of Pharmaceutical Science and Technology 54:264-276.
binnenwerk.indd 386 21-08-2006 12:35:31

Antimicrobial effect of chitosan on micro-organisms isolated from
fishery products
Ziortza Cruz1, Helene Lauzon2, Juan Carlos Arboleya1, Maider Nuin1, Iñigo Martínez de
Marañón1 and Felix Amarita1
1AZTI-Tecnalia, Txatxarramendi Ugartea z/g, 48395 Sukarrieta (Bizkaia), Spain
2Icelandic Fisheries Laboratories, Skulagata 4, IS-101 Reykjavik, Iceland
Abstract
Chitosan, a natural biopolymer, has been suggested as a novel food biopreservative. The
objective of this work was to identify the most suitable formulation of chitosan to inhibit or
retard growth of psychrotrophic bacteria isolated from fishery products. For this purpose, the
antimicrobial activity of several chitosan preparations towards spoilage bacteria and pathogens/
surrogates was evaluated at 8 and 15 °C. The different chitosan preparations influenced the
bacteria differently. Listeria monocytogenes was more sensitive than L. innocua, especially at
higher temperature, while Photobacterium phosphoreum was more sensitive than Shewanella
putrefaciens, but similar patterns were observed at both temperatures. Usually, the higher
chitosan concentration showed increased antimicrobial effectiveness and the use of acetic acid
was preferable.
Keywords: chitosan, antimicrobial, growth inhibition, spoilage bacteria, Listeria, organic acids
Introduction
Chitin is the second most abundant biopolymer on earth after cellulose. Chitin occurs in the
exoskeleton of invertebrates, particularly in crustacean, molluscs and insects, and in the cell
walls of certain fungi (Lower 1984). Biodegradation of chitin is very slow in crustacean shell
waste. The disposal of this waste is problematic and thus, production of value-added products
such us chitin and its derivatives and their posterior application in different fields is of utmost
interest (Shahidi and others 1999).
Chitosan is the principal derivative of chitin and is produced by alkaline deacetylation of chitin.
The typical commercial chitosan has approximately 85% deacetylation. Chitosan is composed
primarily of glucosamine, 2-amino-deoxy-β-D-glucose.
Chitosan has been proved to be non-toxic, biodegradable, biocompatible and it has been used
in the food industry as safe and natural fat digestion and trapped lipid compound (Coma and
others 2002). The antibacterial and antifungal activity of chitosan has been reported widely in
the scientific literature mainly on the basis of in vitro trials against individual microorganisms.
(Roller 2003). Such studies have led to suggestions that chitosan could be used as a novel food
preservative (Chen and others 1998; Shahidi 1999; Rhoades and Roller 2000; Roller and Covill
2000; Tsai and others 2000). However, the evidence in the literature regarding antimicrobial
activity is contradictory. Reported minimum inhibitory concentrations (MICs) for same species
vary several orders of magnitude (Roller 2002). These variations were suggested to be due to
binnenwerk.indd 387 21-08-2006 12:35:31

Z. Cruz, H.L. Lauzon, J.C. Arboleya, M. Nuin, I. Martínez de Marañón and F. Amarita
different types of chitosan (e.g. degree of acetylation, chain length and concentration used),
the conditions of testing (e.g. pH, temperature and medium) and target organism (Roller 2003).
Therefore, the success of its application will be a result of an appropriate combination of all
these parameters.
The objective of this work was to identify the most suitable formulation of chitosan to inhibit
or retard growth of psychrotrophic bacteria isolated from fishery products. For this purpose the
antimicrobial activity of each different chitosan preparation towards various spoilage bacteria
and pathogens/surrogates was evaluated at 8 and 15 °C.
Preparation and selection of chitosan formulations

Five different commercial chitosan products were selected based on their degree of deacetylation
and molecular weight. Solutions of all types of chitosans at different concentrations were
prepared by using acetic, lactic and ascorbic acid. Some of the formulations, especially those
with higher concentrations of chitosan, showed a low solubility during the sample preparation.
Thus, a specific method was developed, consisting basically of fast stirring and high temperature.
The selection of appropriate chitosan formulations was based on organoleptic properties, i.e.
odour and colour intensity.
Combination of two selected chitosans at low (1% w/v) and high (2% w/v) initial concentration
with two different organic acids (lactic and acetic; Panreac Química, S.A, Barcelona, Spain) as
solvent led to 8 chitosan formulations and 2 controls, named C1 to C10 (Table 1). Controls (C9
and C10) consisted of organic acid solutions without chitosan and were tested for antimicrobial
activity in order to verify whether they contributed to the inhibitory response observed.
Bacterial strains
Various spoilage bacteria and pathogens/surrogates (Table 2) were used for the assessment of
the antimicrobial activity of the chitosan preparations. The strains used in the project were
maintained frozen (-70 °C) with 30% (v/v) sterile glycerol in sterile vials (Sarstedt, Nümbrecht,
Germany). Working cultures were made by transferring the content of thawed vials to 3 ml of
Table 1. List of chitosan formulations.
Code Type of chitosan Concentration (w/v) Acid solution (v/v)
C1 Low viscosity 1% Lactic acid 1%

C2 Low viscosity 2% Lactic acid 1%
C3 Low viscosity 1% Acetic acid 0.5%
C4 Low viscosity 2% Acetic acid 0.5%
C5 High viscosity 1% Lactic acid 1%
C6 High viscosity 2% Lactic acid 1%
C7 High viscosity 1% Acetic acid 0.5%
C8 High viscosity 2% Acetic acid 0.5%
C9 - 0% Lactic acid 1%
C10 - 0% Acetic acid 0.5%
binnenwerk.indd 388 21-08-2006 12:35:31

Antimicrobial effect of chitosan on micro-organisms isolated from fishery products
Table 2. List of the 4 bacterial groups prepared as cocktails from various strains.
Type of bacteria Bacterial group Cocktail EU collection # Origin

strains
Gram-negative Photobacterium Pp 26 EU2181 air-stored haddock fillets, 0 °C

spoilage bacteria phosphoreum - Iceland
Pp 3 MAP haddock fillets, 0 °C - Iceland
Pp 6 MAP haddock fillets, 0 °C - Iceland
SF686 EU2183 CSS, 8 °C - France
Shewanella Sp 92 EU2185 Spoiled cod - Iceland
putrefaciens Sp 126 EU2186 Spoiled cod - Iceland
Pathogens/ Listeria A-5410 EU2143 CSS - Iceland
surrogates monocytogenes B35 EU2147 Whole prawn -Iceland
VC 40 EU2149 CSS - Netherlands
Rf 151 EU2162 CSS - France
UN608 EU2168 CSS – Spain
Listeria innocua H-670 EU2172 ‘Gravad’ salmon - Iceland
H-418 EU2173 Used brine - Iceland
CCUG 15531 T EU2174 Reference strain
appropriate broth. Gram-negative spoilage bacteria were grown in Marine broth (MB, Difco) at
15 °C (48 h) while other bacteria were cultured in Tryptic Soy Broth supplemented with 0.6%
(w/v) Bacto Yeast Extract (TSB-YE, Difco) at 22 °C (48 hours). Prior to the microplate inhibitory
assay, strains were grown separately in Veal Infusion Broth (VIB, Difco) (adjusted to 1% w/v
NaCl) at 15 °C for 4-5 days. All strains of each species were mixed (cocktails) by dilution into
VIB for the antimicrobial assays.
Microplate inhibitory assay

Microtiter plates (96 wells, Linbro, ICN Biomedicals Inc., Ohio, USA) were used to evaluate the
antimicrobial activity of the chitosan preparations towards the 4 bacterial cocktails (Table 2)
at 8 and 15 °C. The chitosan preparations were added to VIB (1% NaCl, pH 7.2) as 1% v/v,
resulting in a final concentration of 0.01 and 0.02% (w/v). Each chitosan preparation was
tested in duplicate. Growth of the bacterial cocktails in VIB with no added chitosan was used
as the reference growth behaviour. Bacterial growth was followed by absorbance measurements
at 590 nm using a Titertek Multiskan Plus spectrophotometer (model MKII, Flow Laboratories)
on days 0-1-2-3 for 15 °C and on days 0-3-6-13 for 8 °C. Absorbance measurements on day 0
were used as the blank values. Initial bacterial inoculation level was about log 3-4 CFU/ml.
binnenwerk.indd 389 21-08-2006 12:35:32

The final objective of chitosan application is to increase the shelf life of minimally processed
fishery products. It is therefore important to evaluate the organoleptic characteristics of chitosan
formulations. Chitosan formulations had been prepared from five different types of chitosan
and a selection was performed based on pre-determined organoleptic parameters. Chitosan-
acetic acid (1% w/v) solutions were rejected because they showed a strong odour. Solubility of
ascorbic acid solutions was quite low compared to the other acids and consequently, ascorbic
acid solutions were not further analysed. Moreover, some chitosan solutions that showed an
intense yellow colour were also rejected for undesirable organoleptic reasons. Based on this
evaluation, two types of food grade chitosan (Primex ehf, Siglufjordur, Iceland) were tested
for their antimicrobial activity. They showed considerable differences in molecular weight but
both presented a high deacetylation degree, a combination that appears to be crucial for their
antimicrobial activity (Omura and others 2002).
C1 to C4 formulations led to the development of turbidity when added to the microbiological

medium (pH 7) and therefore, testing their inhibitory properties towards bacteria by using a
spectrophotometric method would be improper. However due to low turbidity and the need
to assess a formulation based on low viscosity, C4 was also chosen for antimicrobial assays.
Therefore results shown in Figures 1-4 demonstrate the antimicrobial properties of the selected
chitosan formulations (C4 to C8) towards 4 bacterial groups at 8 and 15 °C, in comparison to
the effect of the organic acids alone (C9 and C10) or no added chitosan (none).
Figures 1 and 2 show the growth of pathogenic and surrogate bacteria, i.e. Listeria monocytogenes
and Listeria innocua, at 8 and 15 ºC, respectively. The relative increase in bacterial growth is
presented as time progressed. This step-wise increase indicates the immediate inhibitory properties
of the solutions tested as well as the overall sensitivity of the target bacterial groups with time.
L. innocua was less sensitive than L. monocytogenes, with formulations C4-C7-C8 being most
Listeria innocua Listeria monocytogenes

C 10 C 10
C9 C9
C8 Day 3 C8
C7 C7
C6 Day 6 C6
C5 C5
Day 13
C4 C4
None None
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
Absorbance (590 nm) Absorbance (590 nm)
Figure 1. Growth of pathogenic/surrogate bacteria at 8 ºC as influenced by the chitosan preparations (see

Table 1). C4 – C8 consisted of organic acid solutions with chitosan and C9 and C10 consisted of organic
acid solutions without chitosan. ‘None’ shows the microbial growth in culture medium, i.e. without chitosan
and organic acids, and its maximum absorbance value at day 3 is represented as a dotted line. The minimal
absorbance at day 3 is shown by a straight line, which represents therefore, the highest microbial growth
inhibition.
binnenwerk.indd 390 21-08-2006 12:35:32

Listeria innocua Listeria monocytogenes

C 10 C 10
C9 C9
C8 Day 1 C8
C7 C7
C6 Day 2 C6
C5 C5
Day 3
C4 C4
None None
0.0 0.2 0.4 0.6 0.8 0.0 0.2 0.4 0.6 0.8
Figure 2. Growth of pathogenic/surrogate bacteria at 15 ºC as influenced by the chitosan preparations (see
Table 1). C4 – C8 consisted of organic acid solutions with chitosan and C9 and C10 consisted of organic
acid solutions without chitosan. ‘None’ shows the microbial growth in culture medium, i.e. without chitosan
and organic acids, and its maximum absorbance value at day 3 is represented as a dotted line. The minimal
absorbance at day 3 is shown by a straight line, which represents therefore, the highest microbial growth
inhibition.
inhibitory. C5 was the least effective formulation towards L. innocua, followed by C6. Otherwise,
growth of L. monocytogenes was delayed by all the different chitosan solutions at 8 and 15 °C.
Overall, the organic acids alone did not apparently inhibit the bacterial groups tested.
Figures 3 and 4 show the growth of Gram-negative spoilage bacteria, i.e. Photobacterium
phosphoreum and Shewanella putrefaciens at 8 and 15 ºC, respectively. Differences were
observed among these two bacterial groups, P. phosphoreum being the most sensitive, but
similar inhibitory patterns were seen at both temperatures. In most cases, C4 and C8 were
the most inhibitory formulations at the initial stage of the incubation period while C4-C6-C8
contributed to the delayed growth of both spoilage bacteria upon completion of the incubation.
Again solvent controls clearly illustrated that the inhibitory effect of the acids alone was null
or poor.
Photobacterium phosphoreum Shewanella putrefaciens

C 10 C 10
C9 C9
C8 Day 3 C8
C7 C7
C6 Day 6 C6
C5 Day 13 C5
C4 C4
None None
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Figure 3. Growth of spoilage bacteria at 8 ºC as influenced by the chitosan preparations (see Table 1). C4
– C8 consisted of organic acid solutions with chitosan and C9 and C10 consisted of organic acid solutions
without chitosan. ‘None’ shows the microbial growth in culture medium, i.e. without chitosan and organic
acids, and its maximum absorbance value at day 13 is represented as a dotted line. The minimal absorbance
at day 13 is shown by a straight line, which represents therefore, the highest microbial growth inhibition.
binnenwerk.indd 391 21-08-2006 12:35:33

Photobacterium phosphoreum Shewanella putrefaciens

C 10 C 10
C9 C9
C8 Day 3 C8
C7 C7
C6 Day 6 C6
C5 C5
Day 13
C4 C4
None None
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Figure 4. Growth of spoilage bacteria at 15 ºC as influenced by the chitosan preparations (see Table 1). C4
– C8 consisted of organic acid solutions with chitosan and C9 and C10 consisted of organic acid solutions
without chitosan. ‘None’ shows the microbial growth in culture medium, i.e. without chitosan and organic
acids, and its maximum absorbance value at day 3 is represented as a dotted line. The minimal absorbance
at day 3 is shown by a straight line, which represents therefore, the highest microbial growth inhibition.
The different chitosan preparations (C4-C8) influenced the bacteria differently. In general and
for all strains, C6 was more effective compared to C5 as well as, C8 compared to C7. Therefore,
the higher chitosan concentration showed an increased antimicrobial effectiveness. In fact,
very low concentration (0.01-0.02% w/v) of chitosan did inhibit or retard the bacterial groups
investigated in this study. Similarly, Devlieghere and others (2004) reported low concentrations
of chitosan to inhibit various micro-organisms. Gram-negative bacteria were the most sensitive
(MIC less than or equal to 0.006% (w/v)) while the sensitivity of Gram-positive bacteria was
highly variable but intermediary for yeast.
In the same way different antimicrobial activity can be observed regarding the organic acid
used. For the same pH (5.5), acetic acid exhibited higher growth inhibition: C8 was more
inhibitory than C6 and in general, C7 was slightly more effective than C5. Therefore, it was
found that at same chitosan concentration and pH, formulations made with acetic acid were
more effective than the ones made with lactic acid
Finally, by comparing C4 and C8, no differences in antimicrobial activity among the type of
chitosan have been pointed out. Comparison of the other untested chitosan preparations (C1-
C3) could not be established due to the turbidity development.
Conclusion
This study shows that chitosan inhibits or retards growth of Gram-positive and Gram-negative
bacteria isolated from fishery products. Therefore, chitosan could be used to increase the shelf
life of fishery products.
Future studies will be focused on evaluating more concisely the most promising formulations, i.e.
high concentration of chitosan dissolved in acetic acid independently of the type of chitosan.
Selection of a proper fish model system and further testing on a wider pH range should be
attempted to assess the effect of such parameters on the antimicrobial properties of promising
chitosan solutions.
binnenwerk.indd 392 21-08-2006 12:35:34

Acknowledgements
This work was performed within the Integrated Project (IP) SEAFOODplus, partially granted by
the EU Commission under contract NºFOOD-CT-2004-506359, the Department of Agriculture and
Fisheries from the Basque Government, the Spanish Ministry for Education and Science (FIT-
060000-2004-285) and the Icelandic Fisheries Laboratories.
References
Chen CS, Liau WY, Tsai GJ. 1998. Antibacterial Effects of N-Sulfonated and N-sulfobenzoyl Chitosan and Application to
Oyster Preservation. Journal of Food Protection 61:1124-1128.
Coma V, Martial-Gros A, Garreau S, Copinet A, Salin, F. 2002. Edible Antimicrobial Films Based on Chitosan Matrix.
Journal of Food Science 67:1162-1168.
Devlieghere F, Vermeulen A, Debevere J. 2004. Chitosan: antimicrobial activity, interactions with food components and
applicability as a coating on fruit and vegetables. Food Microbiol. 21(6):703-714.
Lower SE. 1984. Polymers from the Sea Chitin and Chitosan I. Manufacturing Chemist 55:73-75
Omura Y, Shigemoto M, Akiyama T, Saimoto H, Shigemasa Y, Nakamura I, Tsuchido T. 2002. Reexamination of antimicrobial
activity of chitosan having different degrees of acetylation and molecular weights. In: Varum KM, Domard A,
Smidsrod O, editors. Advances in Chitin Science (volume VI), 5th International Conference of the European Chitin
Society, Trondheim: p 273-274.
Rhoades J, Roller S. 2000. Antimicrobial actions of degraded and native chitosan against spoilage organisms in
laboratory media and foods. Appl. Environ Microbiol. 66(1):80-86.
Roller S. 2002. Chitosan, a novel food preservative? In: Chitosan in Pharmacy and Chemistry., R. A. A. Muzzarelli and
C. Muzzarelli, editors., Atec, Italy. p177-182
Roller S. 2003. Chitosan: a new food preservative or laboratory curiosity? In: Natural antimicrobials for the minimal
processing of foods. S. Roller, editor, CRC Press. p159-175.
Roller S., and Covill N. 2000. The antimicrobial properties of chitosan in mayonnaise and mayonnaise-based shrimp
salads. J. Food Prot. 63(2):202-209.
Shahidi F., Kamil J., Arachchi V., Jeon YJ. 1999. Food aplications of chitin and chitosans. Trends in Food Science &
Tecnology 10:37-51
Tsai GJ, Wu ZY, Su WH. 2000. Antibacterial activity of a chitooligosacharide mixture prepared by cellulase digestion of
shrimp chitosan and its application to milk preservation. J. Food Prot. Jun 63(6):747-52.
binnenwerk.indd 393 21-08-2006 12:35:34

binnenwerk.indd 394 21-08-2006 12:35:34
Selection of psychotrophic bacteria active against spoilage and
pathogenic micro-organisms relevant for seafood products
Sébastien Matamoros1,2, Marie-France Pilet1, Frédérique Gigout2, Hervé Prevost1 and

Françoise Leroi2
1Laboratoire de Microbiologie Alimentaire et Industrielle, ENITIAA, Nantes, France
2Département de Sciences et Techniques Alimentaires Marines, IFREMER, Nantes, France
Abstract
In this study, 51 seafood products were screened to select psychrophilic inhibiting lactic
acid bacteria. 5575 colonies were tested for inhibition against four target strains : Listeria
monocytogenes, Staphylococcus xylosus, Pseudomonas sp. and Serratia liquefaciens. 456 colonies
(8.2%) showed inhibition and 132 (28.9%) of them were picked up. 54 isolates growing at
15 °C but not at 30 °C were selected. Basic phenotypic characteristics (Gram, catalase and
oxidase test) were then tested. Finally, 52 presumptive psychrophilic LAB strains were selected.
The inhibition spectrum of these strains was enlarged to 14 spoiling or pathogenic target strains
relevant for seafood products. The inhibition profiles were recorded, and clustering was made
to separate eight distinct groups.
Keywords: biopreservation, psychrophilic lactic acid bacteria, inhibition
Introduction
Seafood products, widely marketed throughout the world, are easily spoiled by micro-organisms.
Due to its aquatic environment, the microbiology of fish products is quite specific. The main
spoilage flora of fresh seafood products stored at low temperature is constituted of Gram
negative bacteria like Shewanella spp., Pseudomonas spp. and Photobacterium spp. (Gram and
Huss 1996). Vacuum packaging reduces the level of the respiratory bacteria such as Pseudomonas
spp. However Shewanella putrefaciens and Photobacterium phosphoreum develop easily under
vacuum conditions, due to their capacity to use trimethylamine oxide (TMAO) as an electron
acceptor. Reduction of TMAO into trimethylamine (TMA) leads to specific fishy off-odours. Modified
atmosphere packaging (MAP) with CO2 inhibits the development of respiratory organisms such
as Pseudomonas spp. and S. putrefaciens. This result explains the significant extension of the
product shelf-life. However, P. phosphoreum is resistant to CO2 atmosphere and is still a source
of spoilage (Gram and Huss 1996). Lightly preserved fish products (cold-smoked fish, pickled
and marinated fish, etc.) are generally distributed at chilled temperature and under vacuum or
modified atmosphere packaging. This kind of packaging favours the growth of lactic acid bacteria
(LAB) such as Lactobacillus spp., Carnobacterium spp. and Leuconostoc spp., which soon become
the predominant flora. They are mixed with typical Gram negative spoilage bacteria. They can
be part of the spoilage by producing off odours and biogenic amines. Enterobacteriaceae (Gram
negative) and Brochothrix thermosphacta (Gram positive) can also be found in lightly preserved
products (Gram and Huss 1996; Leroi and others 1998; Gram and Dalgaard 2002).
binnenwerk.indd 395 21-08-2006 12:35:34

Sébastien Matamoros, Marie-France Pilet, Frédérique Gigout, Hervé Prevost and Françoise Leroi
Pathogenic risks in lightly preserved seafood products occur mainly from Listeria monocytogenes
and biogenic amines production. L. monocytogenes is a Gram positive bacterium that can resist
salting, drying and cooling (Collins-Thompson 1980). This resistance allows L. monocytogenes
to grow in preserved products such as smoked salmon (Heinitz and Johnson 1998). Low numbers
of Clostridium botulinum, Bacillus spp. and Salmonella spp. have been reported in seafood (Huss
1997). High levels of biogenic amines, particularly histamine in scombroid fish, can lead to
food poisoning (ten Brink and others 1990). Their formation is mainly due to microbial activity
and can be related to the presence of P. phosphoreum, psychrotolerant Morganella morganii-like
bacteria and LAB, as for example in lightly preserved tuna (Emborg and others 2005).
Biopreservation is a novel and innovative way of reducing microbial risks and/or extending shelf
life of lightly preserved products. Biopreservation consists in inoculating in the food matrix
bacterial strains selected for their ability to inhibit growth of undesirable bacteria, while not
presenting any spoilage capacity (Rodgers 2001). Many studies have been conducted on the
biopreservation of food products (seafood, meat, dairy products). For example Vermeiren and
others (2004) investigated the preservative abilities of Lb. sakei on cooked meat products.
Brillet and others (2004) used Carnobacterium divergens to protect cold-smoked salmon from
growth of L. monocytogenes. According to Helander and others (1997), the control of Gram-
negative bacteria is problematic because they are resistant to bacteriocins, which are molecules
often produced by LABs and active against related species.
In this study, isolation of bacteria from different seafood products was carried out. The objective
was to isolate psychrophilic LAB cultures active against a wide range of Gram negative and Gram
positive spoilage or pathogenic target bacteria, to be used in the seafood industry.
Bacteria isolation from seafood products

Fifty one seafood products were obtained from different supermarkets (Table 3). They were
stored at 4 or 8 °C, and opened in a range of 10 days before to 10 days after the use by date.
A 30 g sample of each product was taken aseptically and diluted in 120 ml of physiological
water (1 g/l tryptone (Biokar Diagnostics, Beauvais, France), 8.5 g/l NaCl (Merck, Darmstadt,
Germany), 1 drop of bromocresol purple). The sample was crushed in a Stomacher 400 (Seward
Medical, London, UK) for 2 min. After 30 min for revivification, four decimal dilutions were
made in physiological water, and 0.1 ml of each dilution was spread on Elliker agar plates
(Elliker, Biokar Diagnostics, with 1.5% agar added). For each sample, 5 plates of each dilution
were made (total: 25 plates per sample). Plates were then incubated in anaerobic conditions at
8 °C for 10 to 15 days. At that time, plates showing growth of 10 to 15 colonies were selected
for the double layer inhibition test.
Double layer inhibition test

Four target strains have been selected for the test: Listeria monocytogenes (EU2160),
Staphylococcus xylosus (EU2178), Pseudomonas group I (EU2189) and Serratia liquefaciens
(EU2196). These strains were incubated twice successively for 72 and 24 hours in Brain Heart
Broth (Biokar Diagnostics), before being diluted to a previously determined concentration.
Information about culture conditions are presented in Table 1. One millilitre of selected dilution
was added to 15 ml of soft BHB agar (Brillet and others 2004) and then spread on an isolation
plate showing growth of 10 to 15 colonies. The four target strains were spread on different
binnenwerk.indd 396 21-08-2006 12:35:34

Selection of psychotrophic bacteria active against spoilage and pathogenic micro-organisms
plates of the same set (same product, same dilution). The fifth plate of the set was kept intact
as a negative control. Plates were then incubated for 24 hours at the specific strain incubation
temperature (Table 1). Presence of inhibition zone was checked visually. Based on colony
aspect (size, colour, appearance, presence or not of polysaccharide) and product of origin,
approximately 1 out of 3 colonies showing an inhibition zone was picked up and cultivated in
Elliker broth (Biokar Diagnostics) at 15 °C.
Selected strains were isolated twice successively on Elliker agar and then incubated in Elliker
broth at 15 °C for 4 days before being frozen for conservation at –80 °C in Elliker broth
containing 10% of glycerol (Panreac Quimicia SA, Barcelona, Spain).
Phenotypic tests
The selected strains were pre-cultivated for 72 hours in Elliker broth at 15 °C. In order to check
their psychrotrophic characteristic, each strain was then inoculated in two Elliker broth tubes.
One tube was incubated at 15 °C, while the other one was incubated at 30 °C. Growth was
watched visually by checking turbidity after 48 h and one week of culture. Strains presenting
growth at 15 °C but no growth at 30 °C after one week were tested for three characteristics:
Gram, catalase and oxidase. Their morphological characteristics were observed by phase contrast
microscopy. The Gram test was performed by the KOH method (Gregersen 1978). The Catalase
test was performed by dropping 10% H2O2 (Merck) on a colony. The Oxidase test was performed
on a BBL DrySlide (Becton Dickinson and Company, Le Pont de Claix, France). Only Gram
positive, catalase negative and oxidase negative strains were selected for further studies.
Inhibition spectrum
Research of the inhibitory capacities of the selected strains was enlarged to 14 spoilage and
pathogenic target strains listed in Table 2. The tested LAB strains were incubated for 48 h
in Elliker broth at 15 °C. Hundred-fold dilutions were made in physiological water. Strains
were spotted on Elliker agar plates (10 µl per spot) with 6 spots per plate. Plates were then
incubated anaerobically for 10 days at 8 °C.
The target strains were cultivated twice successively for 72 hours and 24 hours in specific
medium broth, at their specific temperature (Table 2). They were then diluted in physiological
water (Table 2), and 1 ml of final dilution was added to 15 ml of specific medium (Table 2)
containing 10 g/l of agar. Soft agar was then spread on previously spotted and incubated Elliker
agar plates. After 24 hours incubation at target strain appropriate temperatures (Table 2), the
plates were examined for evidence of inhibition. The size of the inhibition zone was recorded,
and a score from 0 to 4 was given as follows : 0 for no inhibition, 1 for an inhibition zone’s
Table 1. Target strains used in first screening and appropriate cultures conditions for double layer
realisation.
Strain Code Incubation temperature Used dilution
L. monocytogenes EU2160 20 °C 1/100

S. xylosus EU2178 37 °C 1/1000
Pseudomonas sp. EU2189 20 °C 1/1000
S. liquefaciens EU2196 20 °C 1/1000
binnenwerk.indd 397 21-08-2006 12:35:35

Table 2. Target strains used in the inhibition spectrum and appropriate cultures conditions for double layer
realisation.
Code* Species Medium** Temperature Dilution
EU2199 Psychrobacter sp. BHB 20 °C 10-2

EU2183 Photobacterium phosphoreum BHB + 15g/l NaCl 15 °C 0
EU2204 Lactobacillus farciminis BHB 20 °C 10-2
EU2206 Brochotrix thermosphacta BHB 20 °C 10-3
EU2187 Shewanella putrefaciens BHB 20 °C 10-2
EU2211 Bacillus subtilis BHB 37 °C 10-3
EU2160 Listeria monocytogenes BHB 20 °C 10-2
EU2178 Staphylococcus xylosus BHB 37 °C 10-3
EU2189 Pseudomonas group 1 BHB 20 °C 10-3
EU2196 Serratia liquefaciens BHB 20 °C 10-3
CIP81.3 Salmonella enterica BHB 37 °C 10-2
CIP76.25 Staphylococcus aureus BHB 37 °C 10-1
CIP76.24 Escherichia coli BHB 37 °C 10-2
ENITIAA Clostridium sporogenes RCM*** 37 °C 10-2
* EU: HURDLETECH collection, stored in IFREMER, Nantes, France ; CIP: Institut Pasteur Collection,
Paris, France; ENITIAA: ENITIAA collection, Nantes, France
** BHB: Brain Heart Broth, Biokar Diagnostic; RCM medium was prepared as follow : 3 g/l yeast extract
(Biokar Diagnostics), 10 g/l meat extract (Oxoid LTD., Basingstoke, Hampshire, England), 10 g/l peptone
(Oxoid LTD.), 5 g/l glucose (Merck), 1 g/l potato starch (La Bovida, Nanterre, France), 5 g/l NaCl (Merck),
3 g/l Sodium Acetate (Merck), 0.5 g/l cystein hydrochloride (Sigma-Aldrich, St Quentin Fallavier, France),
0.5 g/l agar (Biokar Diagnostics). The pH was adjusted at 6.9 and medium was then autoclaved
diameter of 1 cm, 2 for a diameter of 2 cm, 3 for a diameter of 3 cm and 4 for a diameter of 4
cm or more (meaning the inhibition zone extended to the next spot on the plate).
Ward’s hierarchical clustering method with squared Euclidian distance was used to separate
the 52 LAB isolates into groups according to the size of their inhibition zone for the fourteen
target strains (Uniwin plus, version 4.01, Sigma plus, Paris, France).
Isolation and characterisation

From the 51 seafood products analysed, a total of 5575 colonies have been covered with
indicative bacteria. 456 (8.2%) colonies showed an inhibition activity and 132 of these
(28.9%) were selected upon their appearance for further tests. All results concerning isolation
and characterisation are presented in Table 3. From the 132 selected strains, 54 (40.9%) were
able to grow at 15 °C and showed no growth at 30 °C. According to Shewan and Murray (1979),
psychrophilic bacteria have an optimum growth temperature of 15 °C and a maximum growth
temperature of 18 °C, whereas psychrotrophic strains have an optimum growth temperature
between 20 and 30 °C, and a maximum at 37 °C. The selection of strains showing no growth
at 30 °C orients the selection toward psychrophiles. The selection’s main purpose was to select
strains that will be capable of growing at chilled storage temperatures. Moreover, psychrophilic
binnenwerk.indd 398 21-08-2006 12:35:35

strains will not grow at body temperature, meaning no growth will occur in the human tractus.
If the product is submitted to abused temperature during storage, the biopreservative strains
will not overgrow and spoil the product. Fifty two (96.3%) out of the 54 psychrophilic strains
were Gram positive, catalase negative, oxidase negative rods or cocci, which means that they
probably belong to the LAB group.
As shown in Table 3, a lot of presumptive LAB colonies were obtained from MAP and smoked
fish products. This is coherent with previous publications (Leroi and others 1998; Gonzalez-
Rodriguez and others 2002). But very few of the colonies coming from smoked fish products
showed inhibition, except for the cold-smoked salmon. A significant number of colonies from
MAP products showed inhibition and were selected thereafter. In addition, 44 strains out of the
52 selected strains (84.6%) came from salmon: MAP salmon (39 strains) and smoked salmon
(5 strains). Moreover, the screening for psychrophilic bacteria eliminated most of the strains
isolated from smoked salmon. Of the 37 strains originally selected from smoked salmon, 32
(86.5%) grew at 30 °C and were not selected. On the opposite, 40 of 48 (83.3%) of the strains
Table 3. Number of colonies screened and selected per product.
Product name Number of Number of Number of Number of

colonies colonies isolated selected
covered showing colonies potential LAB
inhibition
MAP shrimp 157 26 11 0

MAP rough head grenadier 520 24 7 3
MAP salmon 718 151 48 39
MAP sea-bream 240 11 13 5
MAP whiting 456 14 3 0
Smoked haddock 217 4 4 0
Smoked mackerel 121 0 0 0
Smoked herring 300 2 2 0
Smoked trout 354 0 0 0
Smoked tuna fish 368 0 0 0
Smoked shark 476 2 2 0
Smoked salmon 554 221 37 5
Red mullet viscera 371 0 0 0
Mackerel viscera 198 0 0 0
Herring viscera 39 0 0 0
Sea-bream viscera 184 1 1 0
Shrimp (fresh) 0 0 0 0
Salmon carpaccio 128 0 0 0
Salted cod 0 0 0 0
Roe cod (tarama) 174 0 4 0
Roe lumpfish, roe salmon 0 0 0 0
Marinated sardines, anchovy, mussel, 0 0 0 0
pickled shell fish and pickled herring
Total 5575 456 132 52
binnenwerk.indd 399 21-08-2006 12:35:35

isolated from MAP salmon showed no growth at 30 °C and where selected. This is probably the
result of the food processing : during cold smoking, temperature rises to 20-25 °C, which may
inhibit psychrophilic strains. Fish viscera gave a smaller but still important number of colonies.
However, none of them showed any interesting anti-microbial potential.
Inhibition spectrum
The aim of this experiment was to test the inhibiting capacities of the 52 isolated strains. They
were tested against typical pathogenic and spoiling strains (later referred to as target strains).
Most of the target strains chosen are commonly found in seafood products (Gram and Huss 1996).
Others like E. coli, Salmonella or B. subtilis are not generally associated with seafood products, but
can in rare outbreaks be isolated from these products (Heinitz and Johnson 1998; Huss 1997). The
clustering analysis, based on the size of the inhibition zone of fourteen target strains, allowed
the distinction of eight clusters, containing from one to 16 LAB strains. Mean inhibitory spectrum
of each cluster is presented in Table 4. All strains showed inhibiting capacities against at least
two of the fourteen target strains. No strain showed inhibition against all of the fourteen target
strains. All strains showed inhibition against L. monocytogenes. This is important because L.
monocytogenes is a common pathogenic bacterium found in seafood products. This also confirms
the ability of LAB strains to easily inhibit L. monocytogenes. All strains inhibited at least one
Gram positive and one Gram negative strain. However, the mechanism of inhibition needs to be
investigated, as it is known that the nature of the antimicrobial compound (acid, bacteriocin,
hydrogen peroxide) influence the inhibition of some strains, especially Gram negative (Helander
1997). One of the clusters (cluster 8) regrouped five strains showing the weakest inhibiting
capacities, including negative control Lb. curvatus. Cluster 5 had a high inhibiting potential:
Table 4. Mean inhibitory spectrum of eight clusters of psychrophilic lactic acid bacteria againts fourteen
target strains. (the number represent the mean inhibition halo diameter of the group, in cm). Clusters are
obtained by the Ward’s hierarchichal clustering method. Brackets: number of isolates/cluster.
Target strain
Gram + Gram -
L. monocytogenes
B. thermosphacta
P. phosphoreum
S. putrefaciens
S. liquefaciens
Psychrobacter
C. sporogenes
Pseudomonas
L. farciminis
S. enterica
B. subtilis
S. xylosus
S. aureus
E. coli
Cluster
1 (16) 1 0 0 4 0 0 0 0 0 2.2 0 0 2 0
2 (12) 2 0 1 4 0 1 1 0 0 3 1 1 2 1.3
3 (6) 2 0 1 4 0 1 2 0 1 2 3 4 2 4
4 (9) 1.7 0 0 4 0 3 1 0 0 2 0 4 2 4
5 (2) 2 0 1 3 1 1 1 1 1 2 2 2 2 2
6 (1) 2 1 0 2 0 1 1 0 0 1 2 2 1 3
7 (2) 1 0 3 4 0 1 1 3 3 1 1 4 1 3
8 (5) 1 0 0 0 0 0.8 0 0 0 0.2 0.2 0.4 0.2 1
binnenwerk.indd 400 21-08-2006 12:35:36

although it had weaker inhibition (i.e. smaller inhibiting zone diameter) than what could be seen
in other clusters, it inhibited all target strains except Staphylococcus xylosus. Cluster 7 showed
approximately the same inhibiting pattern than cluster 5, but with stronger inhibition of E. coli,
Staphylococcus aureus and Salmonella enterica. It is also important to notice that typical seafood
product spoiling bacteria are frequently inhibited: all clusters (except cluster 8) presented at least
weak inhibition against Pseudomonas and Serratia liquefaciens. P. phosphoreum was inhibited by
all clusters except clusters 1 and 4.
Conclusion
Lightly preserved seafood products offer a good range of LAB adapted to biopreservation.
Smoked salmon and other MAP seafood products were the products that contained the highest
levels of interesting LAB, providing nearly 85% of our final isolates. True psychrophilic bacteria
are quite rare and it appeared much easier to isolate psychrophiles from products that have
never been heated. Around 8% of the colonies investigated showed an inhibition capacity
and less than 40% of the selected colonies finally matched the researched characteristics
(psychrophilic presumptive LAB). Some isolates, for example those from cluster 5 (coming from
sea bream) and cluster 7 (coming from smoked salmon), exhibited an interesting spectrum
against both pathogenic and spoiling microflora. These results are to be confirmed in seafood
products. The non spoiling capacity of the selected LAB has also to be checked and a complete
identification, using both phenotypic and molecular methods must be provided.
Acknowledgements
This work was performed within the Integrated Project (IP) SEAFOODplus granted by the EU
commission under contract n°FOOD-CT-2004-506359.
References
Brillet A, Pilet MF, Prevost H, Bouttefroy A, Leroi F. 2004. Biodiversity of Listeria monocytogenes sensitivity to
bacteriocin-producing Carnobacterium strains and application in sterile cold-smoked salmon. J Appl Microbiol
97:1029-1037.
Brillet A, Pilet MF, Prevost H, Cardinal M, Leroi F. 2005. Effect of inoculation of Carnobacterium divergens V41, a
biopreservative strain against Listeria monocytogenes risk, on the microbiological, chemical and sensory quality of
cold-smoked salmon. Int J Food Microbiol 104:309-324.
Collins-Thompson DL. 1980. Food spoilage and food-borne infection hazards. In: Graham HD, editor. Safety of foods.
2nd ed. Mayaguez : University of Puerto Rico Mayaguez. p 15-53.
Emborg J, Laursen BG, Dalgaard P. 2005. Significant histamine formation in tuna (Thunnus albacares) at 2 °C – effect of
vacuum- and modified atmosphere- packaging on psychrotolerant bacteria. Int J Food Microbiol 101(3):263-279.
Gonzalez-Rodriguez MN, Sanz JJ, Santos JA, Otero A, Garcia-Lopez ML. 2002. Numbers and types of microorganisms in
vacuum-packed cold-smoked freshwater fish at the retail level. Int J Food Microbiol 77:161-168.
Gram L, Dalgaard P. 2002. Fish spoilage bacteria – problems and solutions. Curr Opin Biotechnol 13:262-266.
Gram L, Huss HH. 1996. Microbiological spoilage of fish and fish products. Int J Food Microbiol 33:121-137.
Gregersen T. 1978. Rapid method for distinction of Gram-negative from Gram-positive bacteria. Eur J Appl Microbiol
Biotechnol 5:123-127.
Heinitz LM, Johnson JM. 1998. The incidence of Listeria spp., Salmonella spp., and Clostridium botulinum in smoked
fish and shellfish. J Food Prot 61(3):318-323.
binnenwerk.indd 401 21-08-2006 12:35:36

Helander IM, von Wright A, Mattila-Sandholm TM. 1997. Potential of lactic acid bacteria and novel antimicrobial against
Gram-negative bacteria. Trends Food Sci Technol 8(5):146-150.
Huss HH. 1997. Control of indigenous pathogenic bacteria in seafood. Food Control 8(2):91-98.
Leroi F, Joffraud JJ, Chevalier F, Cardinal M. 1998. Study of the microbial ecology of cold-smoked salmon during storage
at 8 °C. Int J Food Microbiol 39:111-121.
Rodgers S. 2001. Preserving non-fermented refrigerated foods with microbial cultures – a review. Trends Food Sci
Technol 12:276-284.
Shewan JM, Murray CK. 1979. The microbial spoilage of fish with special reference to the role of psychrophiles. In:
Russel AD, Fuller R, editors. Cold tolerant microbes in spoilage and the environment. London: Academic Press. p
117-136.
ten Brink B, Daminsk C, Joosten HMLJ, Huis in’t Veld JHJ. 1990. Occurrence and formation of biologically active amines
in foods. Int J Food Microbiol 11:73-84.
Vermeiren L, Devlieghere F, Debevere J. 2004. Evaluation of meat born lactic acid bacteria as protective cultures for
the biopreservation of cooked meat products. Int J Food Microbiol 96:149-164.
binnenwerk.indd 402 21-08-2006 12:35:36

Selection of non-tyramine producing Carnobacterium strains for the
biopreservation of cold-smoked salmon
Anne Brillet1, Sébastien Matamoros1, Christine Blanchet-Chevrollier1, Françoise Leroi2,

Hervé Prevost1 and Marie-France Pilet1
1Unité de Recherche Qualité Microbiologique et Aromatique des Aliments, ENITIAA, rue de la
Géraudière, BP 82225, 44322 Nantes Cedex 3, France
2Laboratoire de Sciences et Techniques Alimentaires Marines, IFREMER, rue de l’Ile d’Yeu, BP
21105, 44331 Nantes Cedex 3, France
Abstract
In this study, we have screened a collection of Carnobacterium strains that could be used
for biopreservation in order to find a natural tyramine negative strain. This screening was
performed using the detection of a part of the tyrosine decarboxylase gene by PCR test. On
35 strains of Carnobacterium tested, all showed the presence of the tdc gene suggesting that
they all produce tyramine. This was assessed by the quantification of tyramine production for
10 strains. In a second part, a mutation procedure using ethyl methyl sulfonate was used to
select a tyramine negative mutant of Carnobacterium divergens V41 that is a good candidate
for biopreservation applications. A mutant strain called C. divergens V41A8 was selected and
characterized. The mutant was identical to the wild strain concerning carbohydrates fermentation
profile, antibiogram spectrum, bacteriocin production, and bacteriocin spectrum towards Listeria
monocytogenes. The growth of strain C. divergens V41A8 was tested by comparison to the wild
strain on a sterile cold smoked salmon model. The mutant grew more slowly than the wild strain
on the product, but it reached nearly the same level after 28 days of storage. Moreover, the
production of tyramine detected on cold smoked salmon inoculated with C. divergens V41 (122
µg/g after 28 days of storage) was not detected at all when the product was inoculated with
the mutant strain C. divergens V41A8. This strain could be an interesting alternative for the
application of biopreservative Carnobacterium on food products naturally contaminated with
tyramine such as smoked fishes.
Keywords: salmon, Carnobacterium, biopreservation
Introduction
Biogenic amines are produced in several food products by decarboxylation of amino acids,
mainly due to the activity of microbial decarboxylases. Fish products are specially concerned by
the presence of biogenic amines as they have high amino acid content, and are contaminated by
decarboxylase positive bacterial flora. In cold smoked salmon, histamine, tyramine, cadaverine
and putrescine are the main biogenic amines that are produced by the bacterial flora (Jorgensen
and others 2000b). Whereas the production of histamine, cadaverine and putrescine can be
attributed to Gram negative spoilage flora e.g. Photobacterium phosphoreum or enterobacteria
like Morganella morganii or Serratia sp. (Kim and others 2000; Emborg and others 2002),
tyramine production on fish products is mainly the consequence of the presence and growth of
Photobacterium phosphoreum, and also lactic acid bacteria from the genus Carnobacterium and
binnenwerk.indd 403 21-08-2006 12:35:36

A. Brillet, S. Matamoros, C. Blanchet-Chevrollier, F. Leroi, H. Prevost and M.-F. Pilet
Lactobacillus (Jorgensen and others 2000b; Emborg and others 2002). Among lactic acid bacteria,
the genus Carnobacterium is well represented at the end of the shelf-life of cold smoked salmon
(Leroi and others 1998; Gonzalez-Rodriguez and others 2002). Many carnobacteria strains
are known for bacteriocin production, and this property could be exploited on such lightly
preserved product to control the development of food pathogens like Listeria monocytogenes.
We have isolated three strains, Carnobacterium divergens V41, Carnobacterium piscicola V1
(Pilet and others 1995) and SF668 (Duffes and others 1999a) that have been shown to reduce
the growth of inoculated Listeria monocytogenes in sterile cold-smoked salmon (Brillet and
others 2004) without affecting the sensory attributes of the product (Brillet and others 2005).
These strains could be used as biopreservative agents to control Listeria monocytogenes risk in
cold smoked salmon. However, production of tyramine ranging from 35 µg/g to 122 µg/g was
observed on cold smoked salmon for the three strains (Brillet and others 2005). These levels
are within the range of naturally contaminated cold smoked salmon that frequently contains
more than 200 µg/g of tyramine at the end of the storage (Jorgensen and others 2000a; Connil
and others 2002b). Tyramine can have toxicological effects on sensitive consumers when high
amounts are produced in food (Ten Brink and others 1990; Santos 1996) and although it has
never been implicated in fish poisoning, the production of this biogenic amine by a strain used
for biopreservation could be a barrier to its use in the European legislative context (Wessels and
others 2004). For that reason, we have performed in this study the screening of natural tyramine
negative strains among the genus Carnobacterium and the isolation and first characterization of
a mutant strain from Carnobacterium divergens V41 that produces no tyramine.
Bacterial strains and media

Carnobacterium divergens V41, Carnobacterium piscicola V1 and SF668 were isolated from
trout intestine and cold smoked salmon (Pilet and others 1995; Duffes and others 1999a).
Enterococcus faecalis JH2-2 was obtained from Université de Caen, and Listeria innocua 1 from
the ENITIAA collection.
A collection of 35 Carnobacterium strains isolated from cold smoked salmon was supplied by
IFREMER. Listeria monocytogenes strains were isolated from French salmon smoked plants and
were supplied by ASEPT (Laval, France). All the strains were propagated in Elliker broth at 30 °C
for carnobacteria and Enterococcus, at 37 °C for Listeria strains.
Screening of tyramine non producing strains by PCR- based test

Two PCR primers TDC1 (5’-GAAGCACAAATTCGTCATTTA-3’) and TDC2 (5’-TAACCAATCATTTTTTTCCAT-
3’) were designed from the alignment of sequences of available bacterial tyrosine decarboxylase
genes (program MULTALIN http://prodes.toulouse.inra.fr/multalin/; Corpet 1988). DNA was
extracted from Carnobacterium strains as described by Rachman and others (2004). The
amplification of potential tdc gene were performed in a total volume of 20 µl containing 1X PCR
buffer, 0.2 mM dNTPs, 0.4 µM of each primer, 1.5 mM MgCl2, 1 U Taq polymerase and 40 ng of
bacterial DNA. PCRs were carried out in a PT-100 thermocycler (MJ Research, Bio-Rad, Waltham,
USA) with a first denaturation step of 7 min at 94 °C, followed by 35 cycles of 1 min 94 °C,
45 s at 49 °C, 4 min at 72 °C and a final extension step of 7 min at 72 °C. Amplified products
were analysed by electrophoresis in a 3% (wt/v) agarose gel in Tris-acetate-EDTA buffer and
were subsequently visualized by UV illumination after ethidium bromide staining.
binnenwerk.indd 404 21-08-2006 12:35:37

Selection of non-tyramine producing Carnobacterium strains for the biopreservation of salmon
Mutagenesis and isolation of mutants

The experimental procedure used in this work was described previously (Richard and others
2003). Briefly, 10 to 120 µl of ethyl methyl sulfonate (Sigma M0880, d = 1.7 g/ml) was added
to 2 ml of exponential growth cultures of Carnobacterium divergens V41 and incubated at 30 °C
for 2 hours. After centrifugation, the cells were washed twice with physiological water and
appropriated dilutions were plated onto Elliker agar plates and incubated 48 hours at 30 °C.
Colonies obtained from culture with less than 10% survival rate were picked and studied for
tyramine production.
Tyramine production detection and quantification

The colonies obtained from the mutagenesis procedure were inoculated in microplates containing
150 µl modified Maijala broth (Tryptone 5 g/L, yeast extract 4 g/L, Meat extract 8 g/L, tyrosine
2 g/L, Tween 80 0.5 g/L, MgSO4 0.02 g/L, CaCO3 0.01 g/L, MnSO4 0.005 g/L, FeSO4 0.004
g/L, bromocresol purple 0.006 g/L) (Maijala 1993). After incubation at 30 °C for 48-72 hours,
apparition of violet colour suggesting alcalinisation of the media indicated tyramine producing
strains. All the isolates giving no alcalinisation were selected for tyramine quantification on
HPLC by the method described previously (Connil and others 2002a). Enterococcus faecalis
JH2-2 and Listeria innocua 1 were used respectively as tyramine producer and tyramine non
producer controls.
Characterization of the mutant

The tyrosine decarboxylase negative strain was characterized in comparison with the wild strain
by fermentation profile using API 50 CH (Biomérieux, Marcy l’Etoile, France). Antibiotic resistance
was checked on wild and mutant strains using 18 antibiotics (Biomérieux, Marcy l’Etoile, France)
by the disc diffusion test. Bacteriocin activity was tested as follows: Carnobacterium divergens
V41 and A8 were grown at 30 °C for 24 hours in Elliker broth. After centrifugation (8000 rpm, 6
min.), cell-free supernatant (CFS) was treated 10 min at 80 °C. CFS was serially two-fold diluted
in phosphate buffer (0.1 M, pH 6.5) and 10 µl of each dilution was spotted onto Elliker agar
plates containing 107 CFU/ml of Listeria innocua 1 indicator strain. The bacteriocin activity was
defined as the reciprocal of the lowest dilution showing no inhibition of the indicator strain in
arbitrary units AU/ml. Bacteriocin spectrum on 57 Listeria monocytogenes strains described in
a previous study (Brillet and others 2004) were checked by spotting filtered supernatant of C.
divergens V41 and A8 onto Elliker agar plates containing 107 CFU/ml of each Listeria strain.
Growth on sterile cold-smoked salmon and tyramine production

Sterile cold-smoked salmon (CSS) was prepared as described by Joffraud and others (1998).
Each Carnobacterium strain was inoculated in CSS as described before (Brillet and others 2004).
Briefly, appropriate dilutions of each Carnobacterium strain were mixed and inoculated (2% v/w)
in parts of 30 g of thawed sterile CSS pieces distributed in 15 polyamide polyethylene bags
(Bourdeau, Saint Etienne de Montluc, France). Pieces were gently mixed with the inoculating
solution and samples were then vacuum-packed and incubated for 28 days using the following
conditions: 9 days at 4 °C followed by 19 days at 8 °C, with a break during 2 hours at 20 °C
after 19 days of storage (industrial recommendations NF V 01-003, 2004). Microbial analysis
was done weekly in triplicates in Elliker plates incubated aerobically for 5 days at 20 °C. Results
are expressed as mean of three measurements ± 95% Confidence Interval.
Tyramine was determined by HPLC on salmon flesh as described before (Brillet and others 2005)
after 14 and 28 days of storage.
binnenwerk.indd 405 21-08-2006 12:35:37

Screening of tyrosine decarboxylase negative Carnobacterium strains by PCR

A collection of 35 Carnobacterium strains isolated from cold smoked salmon was checked for
the presence of potential tdc gene by PCR. A fragment of 300 bp (Figure 1) corresponding to
the amplification using specific tdc primers was obtained for all the Carnobacterium strains.
All the strains were also detected as tyramine positive by alcalinisation of modified Maijala’s
broth. These results suggest that all the strains are potential tyramine producers. To confirm
this hypothesis, tyramine production measured by HPLC on 10 strains chosen randomly among
the 35 showed that the strains produced around 1500 µg/ml tyramine after 24 to 48 hours of
growth in Maijala’s medium. These results showed that strains from the genus Carnobacterium
isolated from cold smoked salmon are tyramine producers as it has been shown previously for
carnobacteria isolated from meat products (Masson and others 1996). As no natural tyramine
negative mutants could be isolated, the strategy of mutagenesis was chosen for further
studies.
Isolation of tyramine – mutants from Carnobacterium divergens V41

5300 colonies of Carnobacterium divergens V41 were screened on modified Maijala’s broth
after exposure to ethyl methyl sulfonate (EMS). Twelve colonies showing no alcalinisation
of the media were streaked on Elliker agar and re-inoculated in modified Maijala to check
tyramine production. One isolate showing no tyramine production was selected and named
V41A8. Tyramine quantification on HPLC was checked for this mutant strain and the wild strain
1 2 3 4 5 6 7 8 9
1000 pb
500 pb
Figure 1. Amplification product of potential tdc gene on Carnobacterium strains. 1 : 100 bp molecular weight
ladder (Biolabs), 2 and 7 : tyramine negative lactic acid bacteria strains, 3 to 6 and 8 : Carnobacterium
SF2025, SF2051, SF2052, SF2053, SF2055 from the collection. 9 : negative control.
binnenwerk.indd 406 21-08-2006 12:35:38

after 1 and 5 days of incubation at 30 °C. Tyramine production was around 2051 µg/ml for the
wild strains after 1 days and 2069 µg/ml after 5 days, whereas no tyramine production was
detectable for V41A8. This strain was thus kept for the following experiments.
After PCR amplification of the potential tdc gene with the specific primers used before, a
positive result was therefore obtained for V41 and V41A8 (data not shown). The mutation has
probably affected separated bases on the tyrosine decarboxylase gene of V41A8. In this case,
the annealing of the primers remains possible leading to a positive result for the mutant strain.
The sequencing of the whole tyrosine decarboxylase gene of V41 and V41A8 would confirm this
hypothesis. This result demonstrated the limits of the PCR method for the screening of tyramine
positive strains. Positive PCR results should then be confirmed with tyramine detection or
quantification as it was done in this study with the Carnobacterium strain collection.
The screening of 5300 colonies was necessary for the isolation of one stable mutant after
EMS exposition, and this rate was in agreement with the study of Joosten who used the same
method on Enterococcus faecalis (Joosten 1995). The mutant V41A8 was considered as stable
as no revertant could be detected after several propagation of the strain.
Characterization of the mutant

The fermentation profile determined on 50 carbohydrates was identical between both strains:
glycerol, ribose, D-glucose, D-fructose, D-mannose, N-acetylglucosamine, amygdaline, arbutine,
esculine, salicine, cellobiose, maltose, sucrose, trehalose, b-gentiobiose, and gluconate were
positive, all the other sugars were not fermented. Both strains showed also the same antibiotic
resistance and sensitivity (Table 1). These results suggest that the mutation procedure did not
affect the main metabolic activities.
The comparison of inhibition properties of both strains is also important for the characterization
of the tyramine negative strain. The wild strain Carnobacterium divergens V41 produces a
bacteriocin named divercin V41 that is active against Listeria monocytogenes (Pilet and others
1995; Metivier and others 1998). This bacteriocin is involved in the inhibition effect of the
strain against Listeria monocytogenes in cold smoked salmon (Duffes and others 1999b; Richard
and others 2003; Brillet and others 2004). Bacteriocin activity was 12800 AU/ml on Listeria
innocua 1 for both mutant and wild strains showing that the mutation did not affect bacteriocin
production. Moreover, a study of the inhibition spectrum against a large number of Listeria
monocytogenes strains showed that both strains gave inhibition zones on all the 57 strains
tested. All these results show that the tyramine negative strains have kept all its inhibition
properties in liquid media.
Growth on sterile cold-smoked salmon and tyramine production

The growth of both strains during storage is shown on Figure 2. For the wild strain, the growth
began at 4 °C, from 1.2 ± 1.1 x 105 CFU/g to 4.3 ± 1.2 x 107 CFU/g after 4 weeks of storage.
The mutant strain grew more slowly than the wild strain during the three first weeks of storage,
but reached the similar level of 1.6 ± 1.2 x 107 CFU/g after four weeks. It is possible that the
mutation procedure had affected growth adaptability of the strain at low temperatures or on
fish muscle that is mainly composed of amino acids, peptides and proteins.
A small amount of tyramine (43 µg/g) was detected after 14 days when the product was
inoculated with the strain C. divergens V41 and it increased to 122 µg/g after 28 days of
binnenwerk.indd 407 21-08-2006 12:35:38

Table 1. Antibiogram spectrum of C. divergens V41 and V41A8. R:resistant; S:sensitive; I:intermediate.
Antibiotic Resistance or sensitivity of C. divergens V41 and V41A8
Pénicillin G R
Ampicillin S
Cefalotin I
Cefotaxim R
Streptomycin R
Gentamicin R
Kanamycin R
Chloramphénicol S
Tétracyclin S
Erythromycin S
Spiramycin S
Colistin R
Vancomycin S
Triméthoprim S
Nalidixic acid R
Norfloxacin I
Rifampicin S
Fusidic acid I
9
log10 (CFU/g salmon)
8
7
6
5
4
3
0 7 14 21 28
Time (days)
Figure 2. Growth of C. divergens V41 (£) or V41A8 (r) inoculated in sterile cold smoked salmon during
storage (9 days at 4 °C and 19 days at 8 °C with a break of 2 hours at 20 °C after 19 days). Bars indicate
95% confidence intervals.
storage (Table 2). This production could be directly attributed to the strain as no tyramine was
detected at all on the uninoculated control. On the opposite, when the mutant strain V41A8
was inoculated on the product, no tyramine could be detected after 14 days nor after 28 days
of storage.
binnenwerk.indd 408 21-08-2006 12:35:39

Table 2. Tyramine production in sterile batches of cold-smoked salmon in the presence of C. divergens V41
or C. divergens V41A8, stored under vacuum during nine days at 4 °C and 19 days at 8 °C. Each result
represents the mean of three measures (95% confidence interval).
Storage time (days) Tyramine (µg g-1)
Control (sterile CSS) D14 nd*

D28 nd*
C. divergens V41 D14 43 ± 23
D28 122 ± 9.4
C. divergens V41A8 D14 nd*
D28 nd*
*nd : not detected
Conclusion
In conclusion, we have shown that tyramine production is a characteristic of Carnobacterium

strains isolated from cold smoked salmon. However, it is possible to obtain by a simple chemical
mutation a tyramine negative strain that keeps interesting inhibition properties. The strain
is able to grow on cold smoked salmon at refrigerated temperatures without producing any
detectable amount of tyramine. Further studies are needed to determine the effect of this strain
on the inhibition of Listeria monocytogenes in cold smoked salmon and on the quality of the
product. As it has been obtained by simple mutagenesis without recombinant DNA, this strain
is not considered as a genetic modified organism (GMO) as defined in the Directive 2001/18/EC
(2001). However, the use of bacterial strains as Carnobacterium divergens V41 or V41A8 for the
biopreservation of non fermented foods will require complete information about their taxonomy,
safety and consumer perception (Wessels and others, 2004).
Acknowledgements
This work was supported by grants from the “Aliment Qualité Sécurité” project no. R01 ⁄ 05 from
the French Ministry of Agriculture and Fishery, and from partners CITPPM, ENITIAA, IFREMER
and ASEPT.
References
Brillet A, Pilet MF, Prevost H, Bouttefroy A, Leroi F. 2004. Biodiversity of Listeria monocytogenes sensitivity to
bacteriocin-producing Carnobacterium strains application in sterile cold-smoked salmon. J Appl Microbiol 97:1029-
1037.
Brillet A, Pilet MF, Prévost H, Cardinal M, Leroi F. 2005. Effect of inoculation of Carnobacterium divergens V41, a
biopreservative strain against Listeria monocytogenes risk, on the microbiological, chemical and sensory quality of
cold-smoked salmon. Int J Food Microbiol 104:309-324.
Connil N, Le Breton Y, Dousset X, Auffray Y, Rince A, Prevost H. 2002a. Identification of the Enterococcus faecalis
tyrosine decarboxylase operon involved in tyramine production. Appl Environ Microbiol 68:3537-3544.
binnenwerk.indd 409 21-08-2006 12:35:39

Connil N, Prevost H, Dousset X. 2002b. Production of biogenic amines and divercin V41 in cold smoked salmon
inoculated with Carnobacterium divergens V41, and specific detection of this strain by multiplex-PCR. J Appl
Microbiol 92:611-617.
Corpet F. 1988. Multiple sequence alignment with hierarchical clustering. Nucl Ac Res 16:10881-10890.
Directive 2001/18/EC of the European parliament and of the council of 12 March 2001 on the deliberated release into
the environment of genetically modified organisms. Official J Europ Com, 17/04/2001.
Duffes F, Leroi F, Boyaval P, Dousset X. 1999a. Inhibition of Listeria monocytogenes by Carnobacterium spp. strains in
a simulated cold smoked fish system stored at 4 degrees C. Int J Food Microbiol 47:33-42.
Duffes F, Corre C, Leroi F, Dousset X, Boyaval P. 1999b. Inhibition of Listeria monocytogenes by in situ produced and
semipurified bacteriocins of Carnobacterium spp. on vacuum-packed, refrigerated cold-smoked salmon. J Food
Prot 62:1394-1403.
Emborg J, Laursen BG, Rathjen T, Dalgaard P. 2002. Microbial spoilage and formation of biogenic amines in fresh and
thawed modified atmosphere-packed salmon (Salmo salar) at 2 degrees C. J Appl Microbiol 92:790-799.
Gonzalez-Rodriguez MN, Sanz JJ, Santos JA, Otero A, Garcia-Lopez ML. 2002. Numbers and types of microorganisms in
vacuum-packed cold-smoked freshwater fish at the retail level. Int J Food Microbiol 77:161-168.
Joffraud J.J, Leroi F, Chevalier F. 1998. Development of a sterile cold-smoked fish model. J Appl Microbiol 85:991-
998.
Joosten HM. 1995. Isolation of tyrosine decarboxylaseless mutants of a bacteriocin-producing Enterococcus faecalis
strain and their application in cheese. J Food Prot 58:1222-1226.
Jorgensen LV, Dalgaard P, Huss HH. 2000a. Multiple compound quality index for cold-smoked salmon (Salmo salar)
developed by multivariate regression of biogenic amines and pH. J Agric Food Chem 48:2448-2453.
Jorgensen LV, Huss HH, Dalgaard P. 2000b. The effect of biogenic amine production by single bacterial cultures and
metabiosis on cold-smoked salmon. J Appl Microbiol 89:920-934.
Kim SH, Ben-Gigirey B, Barros-Velazquez J, Price RJ, An H. 2000. Histamine and biogenic amine production by
Morganella morganii isolated from temperature-abused albacore. J Food Prot 63:244-251.
Leroi F, Joffraud JJ, Chevalier F, Cardinal M. 1998. Study of the microbial ecology of cold-smoked salmon during storage
at 8 degrees C. Int J Food Microbiol 39:111-121.
Maijala RL. 1993. Formation of histamine and tyramine by some lactic bacteria in MRS-broth and modified decarboxylation
agar. Lett Appl Microbiol 17:40-43.
Masson F, Talon R, Montel MC. 1996. Histamine and tyramine production by bacteria from meat products. Int J Food
Microbiol 32:199-207.
Metivier A, Pilet MF, Dousset X, Sorokine O, Anglade P, Zagorec M, Piard JC, Marion D, Cenatiempo Y, Fremaux C. 1998.
Divercin V41, a new bacteriocin with two disulphide bonds produced by Carnobacterium divergens V41: primary
structure and genomic organization. Microbiol 144:2837-2844.
Pilet MF, Dousset X, Barré R, Novel G, Desmazeaud M, Piard JC. 1995. Evidence for two bacteriocins produced by
Carnobacterium piscicola and Carnobacterium divergens isolated from fish and active against Listeria monocytogenes.
J Food Prot 58:256-262.
Rachman C, Kabadjova P, Valcheva R, Prévost H, Dousset X. 2004. Identification of Carnobacterium species by restriction
fragment length polymorphism of the 16S – 23S rRNA gene intergenic spacer region and species-specific PCR. Appl
Env Microbiol 70:4468-4477.
Richard C, Brillet A, Pilet MF, Prevost H, Drider D. 2003. Evidence on inhibition of Listeria monocytogenes by divercin
V41 action. Lett Appl Microbiol 36:288-292.
Santos MHS. 1996. Biogenic amines: their importance in foods. Int J Food Microbiol 29:213-231.
Ten Brink B, Damink C, Joosten HM, Huis in ‘t Veld JH. 1990. Occurrence and formation of biologically active amines
in foods. Int J Food Microbiol 11:73-84.
Wessels S, Axelsson L, Hansen EB, De Vuyst L, Laulund S, Lahteenmaki L, Lindgren S, Mollet B, Salminen S, von Wright
A. 2004. The lactic acid bacteria, the food chain, and their regulation. Trends Food Sci Technol 15:498-505.
binnenwerk.indd 410 21-08-2006 12:35:39

Chapter 6:
Full utilisation of the catch
Recognition of limited resources and increasing environmental pollution has emphasised the
need for better utilisation of fisheries by-products (e.g. heads, frames, viscera) or to exploit
underutilised fish species.
Consumers have become concerned about the limitation of seafood sources. There is an increase
in the amounts of seafood by-products due to growth of aquaculture. Recovery and utilization
of by-products can be improved and the potential health benefits of new components from
seafood by-products need to be tested for the efficacy.
In the first paper the possibilities are described to create novel ingredients from salmon liver
and stickwaters. It is found that salmon liver meal was rich in proteins, has a complete amino
acid profile and was rich in cholesterol, all qualities sought by shrimp feed producer. Stickwater
was found to be high in proteins, but the amino acids profiles did not meet the requirement
for fish nutrition.
The second paper in this chapter explores the technical feasibility of ultra-filtration and
nano-filtration membranes to fractionate and concentrate fish hydrolysates with antioxidant
properties. The selected allowed fractionating hydrolysates in several fractions of different
molecular weights. The progress of the fractionation was monitored with the changes in the
antioxidant activities.
In the third paper the characterisation of proteins recovered from cape hake by-products is
described with focus on solubility, emulsifying and foaming capacities.
Silver smelt (Argentinus silus) is an underutilised fish species caught in many locations in
Europe and elsewhere. It has a white flesh and a bland flavour which make it suitable for sale
as fresh fillets or in products. However, it has received relatively little exposure compared to
some other underutilised species despite its good flesh properties. In the last paper three
European institutes have made a compilation of the quality traits and its suitability for seafood
product development.
binnenwerk.indd 411 21-08-2006 12:35:39

binnenwerk.indd 412 21-08-2006 12:35:39
Production and characterization of a sockeye salmon (Oncorhynchus
nerka) liver meal and dried powders from stickwaters
Sébastien Plante1, Alexandra C.M. Oliveira1, Scott Smiley1 and Peter J. Bechtel2
1Fishery Industrial Technology Center, Univ. of Alaska, Fairbanks, School of Fisheries & Ocean
Sciences, 118 Trident Way, Kodiak, AK 99615-7401, USA
2USDA-ARS, SARU, Univ. of Alaska, Fairbanks, 245 O’Neill Bldg., Fairbanks, AK 99775-7220, USA
Abstract
Fisheries in Alaska produce more that 1.5 millions metric tons (mt) of waste annually. The goal
of this study was to create novel ingredients from salmon liver and stickwaters. Raw livers were
dried to a powder in a commercial vacuum dryer. It is found that salmon liver meal was rich
in proteins, has a complete amino acid profile and was rich in cholesterol, all qualities sought
by shrimp feed producer. Stickwater was found to be high in proteins, but the amino acids
profiles did not meet the requirement for fish nutrition. However, when dried on a drum dryer,
stickwater could be a potential binding agent or an apatite enhancer in feed formulation for
aquaculture.
Keywords: by-products, liver, salmon, feed
Introduction
More than 2.2 millions metric tons (mt) of fish were harvested in 2001 and 2002 in Alaska
(National Marine Fisheries Service 2003). Walleye pollock (Theragra chalcogramma) being the
most abundant with more than 1 million mt followed by the combined species of pacific salmons
for about 250,000 mt. Assuming that only 25% of the fish are used for human food, this leaves
a considerable amount of by-products to process into fishmeal (Smiley and others 2003). The
objectives of this study were to develop new uses for this enormous amount of seafood by-
products by making novel ingredients with two major Alaskan by-products: salmon livers and
stickwater.
Excellent markets exist for salmon roe from certain species. However, there are few markets
for other visceral organs including salmon livers, and seafood processors usually send them
to the waste stream. Livers make up roughly 2% of the weight of salmon, thus about 5,000
mt of salmon livers are available annually for use in the production of specialty aquacultural
feeds. Preliminary results suggested that salmon liver contained high levels of cholesterol, an
important component in shrimp metabolism (Smith and others 2001). The first objective of
this study was to develop and chemically characterize a novel feed ingredient for aquacultured
marine shrimp feeds derived from stabilized salmon livers.
Stickwater is a by-product of fishmeal production. Derived from the oil depleted liquid fractions
during processing, stickwater contains heat soluble proteins as well as soluble organic molecules.
The concentration in solids of Alaskan stickwater is typically around 7% and represent about
70% of the total weight of the by-products used for fishmeal production (Pederson and others
binnenwerk.indd 413 21-08-2006 12:35:39

2003). Alaska’s fishmeal production differs because food-grade processing byproducts are
employed instead of ‘whole fish’. Consequently, little is known about the chemical or nutritional
characteristics of stickwaters made from seafood by-products. The second objective of this
study was to produce dried powders from stickwater to function as an attractant, appetite
enhancer or a binding agent in feed formulations for aquacultured species.
Salmon liver
Approximately 60 kg of fresh sockeye salmon livers were collected on three different processing
days at a local seafood processor in Kodiak (Alaska). Three batches of livers were dried in
a Littleford vacuum dryer (Model FM 130-0) for about 5 hours or until moisture content of
products reached between 6 and 8%. A small quantity of livers was also lyophilized to serve as
a control. Chemical analyses were performed on the lyophilized and vacuum-dried meals.
Stickwaters
Approximately 50 kg of pollock, salmon, and rockfish stickwaters were collected on three
different processing days at the Kodiak Fishmeal Company in Alaska. The stickwaters were
dried using lyophilization (Virtis Virtual Model 52ES) and a drum dryer (Buflovak 6” Dia. X 8”
Atmospheric Double Drum Dryer). Steam pressure in the drums was ~80 P.S.I. The drums were
turning at 1.5 RPM with a 0.15 mm clearance between them. Chemical analyses were performed
on the dried stickwaters.
Common laboratory analysis

Moisture and ash content were determined using AOAC methods (1990). Protein analyses
were carried out using a LECO FP 2000 protein analyzer. Amino acid analyses were performed
on HPLC using AOAC methods (1990). Lipid content was determined using the Folch and
others method (1957). Fatty acids were prepared according to Oliveira and Bechtel (2005) and
chromatographically analyzed on an Agilant GC6850 GC/FID. Lipid classes were determined
using the method described by Oliveira and Bechtel (2005). Cholesterol analysis was performed
with modifications of the method of Kovacs and others (1979). Colour was determined using a
Minolta CR-300 chromameter and reported as L*a*b*.
Salmon liver
The protein content of vacuum-dried liver meal composition was comparable the standard
whitefish meal. Ash content of the vacuum dried liver meal was low and its protein content was
comparable to the standard whitefish meal (Smiley and others 2003) being a few percentages
above the industry standard of 65% (Table 1). The moisture content of the lyophilized liver meal
was lower than the vacuum-dried meal, which can explain the slightly higher protein content;
nonetheless, both meals were of similar composition. Some of the amino acids in the liver meals
were lower than those found in regular fishmeal, but except for phenylalanine and methionine,
they still meet the indispensable amino acid requirements for shrimp or coho salmon (Table 2).
Triglycerides, diglycerides and monoglycerides accounted for less than 1% of the total lipids
in both meals, while phospholipids and free fatty acid account for 85 and 7% respectively.
This suggests that the energy contribution from the lipids in both meals will be minimal. In
the freeze-dried and vacuum-dried meal the cholesterol content was found to be 1.2 ± 0.2 and
binnenwerk.indd 414 21-08-2006 12:35:40

Production and characterization of a sockeye salmon liver meal and dried powders from stickwaters
Table 1. Proximate composition of salmon liver meal (freeze-dried and vacuum-dried) as well as pollock,
salmon, and rockfish stickwater meals (freeze-dried and drum-dried). Values are mean ± standard deviation
(g/100 g).
Salmon liver Pollock stickwater Salmon Rockfish Typical

stickwater stickwater Alaskan
Fishmeal1
FD VD FD DD FD DD FD DD
Water (%) 2.4 7.3 4.2 3.4 4.2 2.8 3.3 2.9 6.1
±1.2 ±1.1 ±0.6 ±0.5 ±2.2 ±0.5 ±0.3 ±0.1 ±1.9
Protein (%) 76.4 71.8 79.7 76.7 81.6 82.9 76.0 76.1 69.3
±1.3 ±2.4 ±3.3 ±2.0 ±3.8 ±2.7 ±0.1 ±0.2 ±2.7
Lipid (%) 16.1 17.5 10.7 13.0 4.5 2.9 10.9 10.5 7.6
±1.0 ±2.5 ±2.5 ±2.9 ±2.2 ±0.6 ±0.1 ±0.2 ±1.8
Ash (%) 6.4 6.1 11.3 10.9 13.1 13.0 10.2 9.7 17.0
±0.1 ±0.6 ±0.4 ±0.2 ±1.5 ±0.8 ±0.0 ±0.2 ±3.9
FD: Freeze-dried; VD: Vacuum-dried; DD: Drum-dried

1 (Smiley and others 2003)
1.0 ± 0.2 g/100g of meal, respectively. These values were about 5-time more than the current
minimum recommendation for shrimp feed (Smith and others 2001). These results indicated
that sockeye salmon liver meal was of high quality and the lipid fraction contained high level
of cholesterol. Thus, it was demonstrated that salmon livers could be a low-cost/high-quality
ingredient for feed formulations in shrimp aquaculture.
Stickwaters
Stickwater meals were higher in protein content and lower in ash and moisture content
compared to regular fishmeal (Table 1). Pollock and rockfish stickwater were quite similar in
terms of proximate analysis, but salmon stickwater differed in terms of proteins and lipids
(Table 1). The concentration of most amino acids found in stickwater are below those found
in fish meal and most are below the indispensable amino acid requirement for coho salmon
(Table 2). Glycine concentration was very high in stickwater. This can be explained by the
high concentrations of collagen and soluble proteins in stickwater (Soares and others 1973).
The lyophilizer produced a very light colored powder and while the drum dryer produced a
more ‘yellowish’ colour (Table 3). Drum dryer technology has proven to be effective in drying
stickwater. It takes only minutes to dry the product compared to up to 10 days for freeze dyer.
Also, based on the similar composition between the freeze-dried and the drum-dried products,
one can argue that drum dryer technology does not negatively alter the chemical properties
of the product. In conclusion, due to the poor amino acid profiles, stickwater should not be
used as a fishmeal replacement in feed formulation. However, stickwater powder should be an
excellent ingredient to act as a binder agent or as an apatite enhancer (Forster and others
2004) for aquacultured species.
binnenwerk.indd 415 21-08-2006 12:35:40

Table 2. Common amino acids of salmon liver meal (freeze-dried and vacuum-dried), pollock, salmon, and
rockfish stickwater meals (freeze-dried and drum-dried) as well as published values for typical Alaskan
fishmeals. Published essential amino acid requirements for shrimp and coho salmon are also presented as
comparison. Values are mean ± standard deviation (g/100 g).
Salmon liver Pollock Salmon Rockfish Typical Essential amino

stickwater stickwater stickwater Alaskan acid requirements
Fishmeal1
FD VD FD DD FD DD FD DD Shrimp2 Coho
salmon3
ALA 4.2 4.1 4.6 4.5 4.4 4.2 5.0 5.0 5.3 - -
(%) ±0.4 ±0.3 ±0.2 ±0.1 ±0.4 ±0.4 ±0.1 ±0.2 ±0.3
ARG 6.1 5.7 7.3 7.5 7.7 7.4 7.7 7.6 6.3 5.8 3.2-5.8
(%) ±0.6 ±0.3 ±0.2 ±0.1 ±1.1 ±0.6 ±0.1 ±0.3 ±0.5
ASP 8.3 7.8 5.3 4.4 4.6 4.2 4.7 4.7 8.2 - -
(%) ±0.6 ±0.4 ±0.3 ±0.0 ±0.3 ±0.2 ±0.0 ±0.1 ±0.7
GLU 9.1 8.6 7.9 7.6 7.8 7.4 7.9 7.8 11.5 - -
(%) ±0.5 ±0.5 ±0.4 ±0.1 ±0.4 ±0.6 ±0.0 ±0.3 ±0.9
GLY 3.6 3.5 11.2 11.5 9.7 9.8 12.3 12.4 6.1 - -
(%) ±0.3 ±0.2 ±0.8 ±0.7 ±2.3 ±1.7 ±0.1 ±0.6 ±0.8
HIS 2.1 1.9 1.2 1.1 1.5 1.4 1.1 1.1 1.8 2.1 0.9-1.8
(%) ±0.2 ±0.0 ±0.0 ±0.0 ±0.1 ±0.0 ±0.0 ±0.0 ±0.2
ILEU 3.5 3.3 1.3 1.4 1.5 1.7 1.4 1.4 3.0 3.4 1.2
(%) ±0.3 ±0.2 ±0.1 ±0.1 ±0.3 0.1 ±0.0 ±0.1 ±0.3
LEU 6.1 5.8 2.8 3.0 2.9 3.0 3.0 3.0 6.4 5.4 3.4
(%) ±0.5 ±0.2 ±0.3 ±0.2 ±0.3 ±0.2 ±0.0 ±0.1 ±0.7
LYS 5.6 5.3 3.3 3.4 3.3 3.4 3.8 3.6 6.4 5.3 3.8
(%) ±0.4 ±0.4 ±0.3 ±0.1 ±0.1 ±0.0 ±0.0 ±0.1 ±0.6
MET 2.0 1.9 1.5 1.5 1.6 1.5 1.5 1.6 1.8 2.4 2.7
(%) ±0.2 ±0.1 ±0.0 ±0.0 ±0.1 ±0.1 ±0.0 ±0.1 ±0.3
PHE 3.5 3.4 1.4 1.5 1.6 1.6 1.8 1.8 3.3 4.0 4.5
(%) ±0.3 ±0.1 ±0.1 ±0.0 ±0.1 ±0.1 ±0.0 ±0.1 ±0.4
PRO 3.0 3.0 4.6 4.6 4.6 4.7 5.2 5.4 4.0 - -
(%) ±0.2 ±0.1 ±0.3 ±0.6 ±0.6 ±0.6 ±0.0 ±0.2 ±0.3
SER 3.1 2.9 3.2 1.7 2.9 1.4 1.8 1.8 4.5 - -
(%) ±0.3 ±0.1 ±0.1 ±0.0 ±0.3 ±0.0 ±0.0 ±0.1 ±0.3
THR 3.5 3.4 1.9 1.5 2.1 1.6 1.6 1.6 4.1 3.6 2.0
(%) ±0.3 ±0.1 ±0.1 ±0.0 ±0.0 ±0.0 ±0.0 ±0.1 ±0.4
TYR 3.0 2.8 0.7 0.8 0.9 0.9 0.7 0.7 3.2 - -
(%) ±0.3 ±0.1 ±0.1 ±0.0 ±0.2 ±0.1 ±0.0 ±0.0 ±0.4
VAL 4.3 4.2 1.7 1.8 2.0 2.2 2.0 2.0 3.6 4.0 2.2
(%) ±0.4 ±0.2 ±0.1 ±0.1 ±0.3 ±0.2 ±0.0 ±0.1 ±0.4
FD: Freeze-dried; VD: Vacuum-dried; DD: Drum-dried

1 Recalculated using data from Smiley (2003)
2 Recommendations from Akiyama and others (1991)
3 (Wilson 2002)
binnenwerk.indd 416 21-08-2006 12:35:41

Production and characterization of a sockeye salmon liver meal and dried powders from stickwaters
Table 3. L*, a*, and b* values for pollock, salmon, and rockfish stickwater meal (freeze-dried and drum-
dried). Values are mean ± SD.
Pollock stickwater Salmon stickwater Rockfish stickwater
FD DD FD DD FD DD
L* 76.8 73.5 86.5 77.0 84.5 87.3

±6.5 ±0.6 ±1.7 ±4.5 ±1.9 ±3.1
a* 0.8 4.3 2.0 2.8 0.8 3.5
±0.5 ±1.7 ±0.8 ±1.7 ±1.0 ±1.3
b* 24.3 39.5 24.3 28.8 24.0 36.0
±3.9 ±2.9 ±3.6 ±3.0 ±2.2 ±3.6
FD: Freeze-dried; DD: Drum-dried

L*: 0=black, 100=white
a*: negative=green, 0=gray, positive=red
b*: negative=blue, 0=gray, positive=yellow
Conclusions
In conclusion, the fractionation of the waste stream derived from the processing of seafood
in Alaska can allow the development of specialty products that have unusual and often unique
properties. These new specialty products, when used in appropriately designed feed formulations,
can have an elevated value when compared to that of traditional secondary products such as
fair to average fishmeal, bone meal, fish oil and stickwater.
Acknowledgements
The authors thank B. Pfutzenreuter, S. Coen, and K. Brenner for their technical assistance and
J.K. Babbitt for his critical comments on seafood processing. This project was funded by a
USDA-ARS grant #5341-31410-002-00D.
References
Akiyama DM, Dominy WG, Lawrence AL. 1991. Penaeid shrimp nutrition for the commercial feed industry revised. In:
Akiyama DM, Tan RKH, editors. Proceedings of the aquaculture feed processing and nutrition workshop, Thailand
and Indonesia, Sept 19-25. Singapore: American Soybean Association. p 80-98.
AOAC. 1990. Official methods of analysis (15th Edition). Washington D.C.: Association of Official Analytical Chemists,
1298 p.
Folch J, Lees M, Sloane Stanley GH. 1957. A simple method for the isolation and purification of total lipids from animal
tissues. J Biol Chem. 226:497-509.
Forster I., Dominy W., Smiley S., Bechtel P.J. 2004. Recent advances in utilization of fish byproducts. In aquaculture
feeds. World Aquaculture Society meeting 2004 - Abstract Book. p. 200.
Kovacs MIP, Anderson WE, Ackman RG. 1979. A simple method for the determination of cholesterol and some plant
sterols in fishery-based food products. J Food Sci. 44:1299-1305.
binnenwerk.indd 417 21-08-2006 12:35:41

National Marine Fisheries Service. 2003. Fisheries of the United States. Current fishery statistics no. 2002. Bethesda,
Maryland: NOAA, NMFS. 126 p.
Oliveira ACM, Bechtel PJ. 2005. Lipid composition of Alaska pink salmon (Oncorhynchus gorbuscha) and Alaska walleye
pollock (Theragra chalcogramma) by-products. J Aquatic Food Prod Technol. 14(1):73-91
Pederson LD, Crapo C, Babbitt J, Smiley S. 2003. Membrane filtration of stickwater. In: Bechtel PJ, editor. Advances in
seafood by-products 2002 conference proceeding. Anchorage, AK: Alaska Sea Grant College Program. p 359-369.
Smiley S, Babbitt J, Divakaran S, Forster I, de Oliveira A. 2003. Analysis of groundfish meals made in Alaska. In: Bechtel
PJ, editor. Advances in seafood by-products 2002 conference proceeding. Anchorage, AK: Alaska Sea Grant College
Program. p 431-454.
Smith DM, Tabrett SJ, Barclay MC. 2001. Cholesterol requirement of sub-adult black tiger shrimp Penaeus monodon
(Fabricius). Aquaculture Res. 32:399-405.
Soares J, Jr., Miller D, Cuppett S, Bauersfeld P, Jr. 1973. A review of the chemical and nutritive properties of condensed
fish solubles. Fish Bull. 71(1):255-265.
Wilson RP. 2002. Amino acids and proteins. In: Halver JE, Hardy RW, editors. Fish Nutrition. New York: Academic Press.
p 143-179.
binnenwerk.indd 418 21-08-2006 12:35:41

Evaluation of antioxidant activities in by-product hydrolysates,
fractionation and concentration of active molecules using
separation technologies (ultra- and nanofiltration technologies)
Aurélie Chabeaud1,2, Laurent Vandanjon2, Pascal Jaouen3, Patrick Bourseau2, Charles
Delannoy4, Ragnar Johannsson5, Gudjon Thorkelsson5 and Fabienne Guerard1
1Laboratory ANTiOX, University of West Brittany, Quimper, France
2Laboratory LET2E-EA 3373, University of South Brittany, Lorient, France
3Laboratory GEPEA-UMR CNRS 6144, University of Nantes, Saint-Nazaire, France
4Coopérative de Traitement des Produits de la Pêche (CTPP), Boulogne sur Mer, France
5Icelandic Fisheries Laboratories (IFL), Reykjavik, Iceland
Abstract
Recognition of limited resources and increasing environmental pollution has emphasised the
need for better utilisation of fisheries by-products (e.g. heads, frames, viscera). The present
investigation explores the technical feasibility of ultrafiltration and nanofiltration membranes
to fractionate and concentrate fish hydrolysates with antioxidant properties. The selected
4 kDa and 8 kDa UF membranes allowed fractionating hydrolysates in several fractions of
different molecular weights. The 0.3 kDa NF membrane allowed concentrating these fractions.
The combined study of chromatographic data, amount of peptides quantified in permeates
and retentates as well as antioxidant activities, allowed us to study the progression of the
fractionation in correlation with the increase or decrease in the antioxidant activities.
Keywords: by-products, ultrafiltration (UF), nanofiltration (NF), hydrolysates, antioxidant
Introduction
Global production from capture fisheries and aquaculture supplied about 130 million tonnes of
fish per year (Report FAO, 2004). 76% (100 million tonnes) of this catch is intended for human
consumption. However, depending on fish species and processing (canning, freezing), only
50-70% of this catch is really used in human diet. The remainder such as fishery by-products
(e.g. heads, frames, skin, viscera) is especially converted into animal feed and oil. However,
this underutilized material has a nutritional value almost as good as whole fish (Synowiecki and
Al Khateeb 2000; Ibrahim and others 1994) and has an obvious potential as food ingredient.
Although there are many approaches to further utilisation of these resources, interest has been
expressed in isolating or processing value-added components (Venugopal and Shahidi 1995).
Production of fish protein hydrolysate (FPH) by proteinase treatment is a mean to transform

by-products into new products with improved functional and biological properties. In recent
years, a large number of biologically active peptides from fishery by-products hydrolysates with
interesting and very promising new applications have been generated. For example, active
factors such as peptides inhibiting the angiotensin I-converting enzyme thus exhibiting an
antihypertensive effect (Wako and others 1996; Kohama and others 1991), gastrointestinal
peptides such as gastrin and cholecystokinin (CCKs) (Ravallec-Ple and others 2000; Cancre and
binnenwerk.indd 419 21-08-2006 12:35:42

A. Chabeaud, L. Vandanjon, P. Jaouen, P. Bourseau, C. Delannoy, R. Johannsson, G. Thorkelsson and F. Guerard
others 1999), cellular growth factors (Ravallec-Ple and others 2001), factors such as calcitonin
and calcitonin gene related peptide (CGRP) (Fouchereau-Peron and others 1999) were detected
in hydrolysate fractions. The presence of antioxidant compounds in marine hydrolysates has
also been reported (Guerard and others 2005).
Antioxidants, mainly present in fruit and vegetables, are reported to prevent oxidative damage
by free radical and active oxygen, and prevent the occurrence of disease and aging (Hirose and
others 1994). Since 1960, the antioxidant activity of proteins, peptides and amino acids has
been widely studied. Antioxidant activities have been demonstrated in hydrolysates from food
proteins such as milk casein (Suetsuna and others 2000), soy protein (Moures and others 2005),
egg-yolk protein (Sanakana and others 2004), and, in recent years, from marine by-products
proteins (Je and others 2005; Jun and others 2004; Guerard and others 2004; Jao and others
2002; Jeon and others 2000; Shahidi and others 1996).
The molecular weight of the hydrolyzed proteins is one of the most important factors in producing
protein hydrolysates with bioactive peptides. Accordingly, in order to obtain peptide fractions
with desired molecular size, in this study ultrafiltration (UF) and nanofiltration (NF) systems have
been used. These processes used porous membranes characterised by the molecular weight cut-
off (MWCO), which indicates the smallest molecular weight component retained at 90% (Cheryan
1998). UF and NF aim at concentrating and/or fractionating one or several components of a
solution in order to permit selective passage of components through the membrane.
In the present study, the potential of various membranes to fractionate a fish protein hydrolysate
using UF and NF membranes is described in a first step. In a second step, a process integrating
appropriate membranes was applied to fractionate and concentrate a fish protein hydrolysate
with antioxidant properties.
Materials
The spray-dried Blue Whiting hydrolysate BWH (0.1 <Mw< 10 kDa) was obtained from Primex
Company (Iceland). The commercial spray-dried FPH (0.1 <Mw< 5 kDa), named Collagen HM,
was obtained from CTPP (Boulogne sur Mer, France).
Protein content
The protein (N x 6.25) content of samples was determined using the Kjeldahl method (AOAC,
1995).
Process performances
The process performances are principally described by the permeation flux, and the selectivity of
the membrane. The permeation flux is defined by the volume of solution crossing the membrane
per unit of surface and of time (L h-1 m-2). The selectivity is defined as the proportion of
peptides retained by the membrane and can be expressed using the retention rate as follows:
RR = 1 - (Cp/Cr)
where Cp and Cr are the concentration of peptides in permeate and retentate respectively. RR
varies between 0 (no retention) and 1 (total retention).
binnenwerk.indd 420 21-08-2006 12:35:42

Evaluation of antioxidant activities using separation technologies
In addition, the membrane selectivity is evaluated by studying the molecular weight distribution
profile of permeate and retentate.
Molecular weight distribution profile

Molecular weight (MW) distribution profiles of permeates and retentates were analyzed by size
exclusion chromatography using a Superdex Peptide® HR 10/30 column (Pharmacia) in FPLC
mode (fractionation range of the column was 7000 to 100 Da) according to Guerard and others
(2001). The mobile phase (isocratic elution) consisted in water with TFA 0.1% and acetonitrile
(70:30). The flow rate was 0.5 mL/min. The liquid chromatographic system consisted in a
Waters 600 automated gradient controller pump and a Waters 996 photodiode array detector.
Millenium 32 software was used to collect, plot and process the chromatographic data.
Antioxidant activity
The antioxidant activity was determined using the beta-carotene/linoleate model system as
described by Marco (1968) with slight modifications. Five mL aliquots of a beta-carotene-
linoleate emulsion were transferred to glass tubes, containing 200 µL of sample at 1.25 mg/
mL protein. All samples were incubated simultaneously for 2 hours at 50 °C, under mixing.
Absorbance values at 470 nm were recorded every 30 min with microplate reader.
Ultrafiltration and nanofiltration processes

A UF/NF Microlab40 pilot plant (VMA Industry) with a maximum capacity of 5 L was used.
Four organic UF and NF membranes PCI with a tubular geometry were selected for treatment of
BWH in total recycling mode (i.e. permeate + retentate) (Table 1). Operating conditions were:
Temperature 15 °C, tangential velocity 2.5 m/s and pressure ranging from 8 to 35 bar. Although
operating conditions were not optimized, they are considered as standard conditions for the
membranes, thus allowing a comparison of membrane performances, which were described by
the evaluation of the retention rate.
Fractionation and concentration of the commercial FPH ‘Collagen HM’ was carried out as
described above. The process took place in two steps. Five litres of Collagen HM (50 g/L) were
fractionated on the MT PO4 UF membrane (MWCO of 4 kDa) at 40 °C, 25 bar. The experiment
was stopped after achievement of a volume reduction factor (VRF) of 5. Then, the 4 L of UF
permeate were further concentrated using the MT O4 NF membrane (with MWCO 300 Da) at
40 °C, 35 bar. The experiment was stopped after achievement of a VRF of 4. Permeates and
Table 1. Characteristics of UF and NF membranes tested.
Commercial designation Supplier Material Molecular weight cut off
MT 04 PCI PA* / PES 300 Da

MT P04 PCI PES** 4 kDa
MT 68 PCI PS*** 8 kDa
MT 120 PCI PS 20 kDa
* Polyamide
** Polyethersulfone
*** Polysulfone
binnenwerk.indd 421 21-08-2006 12:35:42

retentates were analysed in terms of protein content, molecular weight distribution profile and
antioxidant activities.
Study of potentiality of membranes to fractionate and concentrate FPH

The objective of this experimental work was to study the potentialities of membrane processes
for concentrating and fractionating hydrolysates with a wide range of MW distribution.
According to the retention rate, UF and NF membranes can be used to:
• separate peptides from enzyme or non hydrolysed proteins if the membrane has a low
retention rate, < 30% (in this case, peptides can pass almost freely through the membrane
but proteins are retained),
• concentrate peptides solutions or purify solutions in diafiltration mode to eliminate salts
or decrease undesirable odour (Vandanjon and others 2005a) if the membrane has a high
retention rate (> 90%) ,
• fractionate peptides with respect to their molecular weight distribution if the membrane
has an intermediate retention rate.
The retention rate of the tested membranes is shown in Figure 1. The NF membrane showed
a retention rate of 98% (measured here by nitrogen content). Consequently, this membrane
is well-suited for concentration or purification of hydrolysates. Conversely, the 20 kDa UF
membrane has a relatively low retention rate of about 30%. This membrane could be used for
separation of peptides from proteins or enzymes in a membrane enzymatic reactor, for example.
The UF membranes of 4 and 8 kDa presented intermediate retention rates ranging from 70% to
80%, and then could be used for peptide fractionation.
Fractionation and concentration of FPH with antioxidant properties

The UF membranes of 4 kDa and the NF membrane of 300 Da were chosen according to
their interesting performances for a given application (Vandanjon and others 2005b), and
the molecular weight pattern of Collagen HM (Figure 1). Figure 2 describes the experimental
procedure.
100
90
Peptides retention (%)
80
70
60
50
40
30 NF
UF UF
20
10 UF
0
300 4000 8000 20000
Membrane Cut-off (Da)
Figure 1. Retention rate versus molecular weight cut-off for four organic membranes.
binnenwerk.indd 422 21-08-2006 12:35:42

Substrate: 5L of Collagen HM (CTPP, France) at 50 g/L

0.1 kDa < MW < 5 kDa
T = 40°C
UF Fractionation
P = 25 bar
4 kDA - VRF = 5
u = 2.5 m.s-1
UF retentate
(V = 1L) 4L of UF permeate
T = 40°C
NF Concentration
P = 35 bar
300 Da - VRF = 4
u = 2.5 m.s-1
NF retentate
(V = 1L)
NF permeate (V = 3L)
Figure 2. Collagen HM fractionation and concentration flow sheet using successively UF 4 kDa and NF 0.3
kDa membranes.
During the first step, 5 L of Collagen HM were initially fractionated using the 4 kDa UF membrane.
The initial permeation flux was 120 L h-1 m-2, and then progressively decreased to a final value
of 40 L h-1 m-2 after 90 min. This flux can be considered as correct; however, it shows a non-
negligible fouling of the membrane.
It is well documented that fouling is a consequence of the interactions between membrane

surface and feed solution. Zeman and Zydney (1996) demonstrated that both the permeation
rate and the extent of fouling depend on the strength of the membrane-solute interactions,
together with the hydrodynamic forces acting on the macro solutes and the chemical nature of
the membrane. Therefore, surface chemistry, solute-solute and solute-membrane interactions
are key parameters in fouling.
Concerning the results of selectivity, the membrane retained 62% of peptides, thus demonstrating
that the fractionation was efficient. This result is confirmed by molecular weight profiles of the
collected permeate and retentate (Figure 3). The elimination of all the peptides of MW higher
than 4 kDa was observed in the UF permeate, thus demonstrating a good selectivity of the
membrane tested. Moreover, before fractionation, antioxidant activity of Collagen HM was about
80%. This activity was maintained in both fractions, which signifies that a 90 min treatment
does not damage peptides.
Then, the UF permeate was concentrated using 300 Da NF membrane. The initial permeation
flux of 200 L h-1m-2 remained constant during the 25 minutes of experiment. This result
demonstrated the absence of membrane fouling. The NF membrane retained 92% of peptides,
allowing an efficient concentration (Figure 4). The activity was maintained in the retentate
and permeate.
binnenwerk.indd 423 21-08-2006 12:35:43

1.8
Collagen HM 81.1%
UF retentate 79.5%
Absorbance (220nm) 1.5 UF permeate 77.2%
1.2
0.9
0.6
0.3
0.0
0 10 20 30 40 50 60
Time (min)
Figure 3. Elution of Collagen HM, UF retentate and UF permeate from Superdex Peptide® HR 10/30 after
fractionation using 4kDa UF membrane (see experimental procedure in Figure 2). The antioxidant activity
of samples ranged from 77.2 to 81.1%.
0.8
UF permeate 77.2%
NF permeate 82%
NF retentate 79.4%
Absorbance (220nm)
0.6
0.4
0.2
0.0
0 10 20 30 40 50 60
Time (min)
Figure 4. Elution of UF permeate, and NF permeate and retentate from Superdex Peptide® HR 10/30 after
fractionation using 0,3 kDa NF membrane (see experimental procedure in Figure 2). The antioxidant activity
of samples ranged from 77.2 to 82%.
Conclusion
The present study showed the good performances of the organic membranes tested to fractionate
industrial fish protein hydrolysates with antioxidant properties. Future works will explore
some fractionation criteria other than molecular weight, such as peptide charge, peptide
hydrophobicity. Then, the optimized process could be applied to concentrate peptide fractions
with wide range of biological activities.
binnenwerk.indd 424 21-08-2006 12:35:44

Acknowledgements
This work was performed within the Integrated Research Project SEAFOODplus contract no FOOD-
CT-2004-506359. The financing of the work by the European Union is gratefully acknowledged.
The financing of the work by the region of Brittany is also gratefully acknowledged (Activemb
project).
References
Cancre I, Ravallec R, Van Wormhoudt A, Stenberg E, Gildberg A, Le Gal Y. 1999. Secretagogues and growth factors in
fish and crustacean protein hydrolysates. Mar Biotechnol 1:489-494.
Cheryan, M. 1998. Ultrafiltration and microfiltration handbook. Technomic Publishing Co., Lancaster.
FAO Annual Report. The State of World Fisheries and Aquaculture (SOFIA), World review of fisheries and aquaculture,
2004.
Fouchereau-Peron M, Duvail L, Michel C, Gildberg A, Batista I, Le Gal Y. 1999. Isolation of an acid fraction from a
fish protein hydrolysate with a calcitonin-gene-related-peptide-like biological activity. Biotechnol Appl Biochem
29:87-92.
Guerard F, Dufosse L., De la Broise D, Binet A. 2001. Enzymatic hydrolysis of proteins from yellowfin tuna (Thunnus
albacares) wastes using Alcalase. J Mol Cat B : Enz 11:1051-1059.
Guerard F, Sumaya-Martinez MT, Thomas S, Linard B, Binet A. 2004. Enhancement of the radical scavenging activity of
tuna waste hydrolysate. TotalFood2004, Norwich, UK. p 93-98
Guerard F, Sellos D, Le Gal Y. 2005. Fish and Shellfish Upgrading, Traceability. In: Le Gal Y, Ulber R, Editors. Marine
Biotechnology, Advances In Biochemical Engineering/Biotechnology. Springer, 50 pp. p 127-164
Hirose M, Imaida K, Tamano S, Ito N. 1994. Cancer chemoprevention by antioxidants. In: Ho CT, Huang MT, Osawa T,
editors. Food phytochemicals for cancer prevention II. Washington DC : ACS, p 122-132.
Ibrahim HM, Salama MF, El banna HA. 1999. Shrimp’s waste: chemical composition, nutritional value and utilization.
Nahrung 43:418-423.
Jao C, Ko W. 2002. 1,1-Diphenyl-2-picrylhydrazyl (DPPH) radical scavenging by protein hydrolyzates from tuna cooking
juice. Fisheries Sci 68:430-435.
Je JY, Park PJ, Kim SK. 2005. Antioxidant activity of a peptide isolated from Alaska pollack (Theragra chalcogramma)
frame protein hydrolysate). Food Res Int 38(1):45-50.
Jeon YJ, Byun HG, Kim SK. 2000. Improvement of functional properties of cod frame protein hydrolysates using
ultrafiltration membranes. Process Biochem 35:471-478.
Jun SY, Park PJ, Jung WK, Kim SK. 2004. Purification and characterization of an antioxidative peptide from enzymatic
hydrolysate of yellowfin sole (Limanda aspera) frame protein. Eur Food Res Technol 219:20-26.
Kohama Y, Oka H, Kayamori Y, Tsujikawa T, Mimura T, Nagase Y, Satake M. 1991. Potent synthetic analogues of
angiotensin-converting enzyme inhibitor derived from tuna muscle. Agric Biol Chem 55:2169-2170.
Marco GJ. 1968. A rapid method for evaluation of antioxidants. JAOCS 45:594-598.
Moure A, Dominguez H, Parajo JC. 2005. Antioxidant properties of ultrafiltration-recovered soy protein fractions from
industrial effluents and their hydrolysates. Process Biochem 41:447-456.
Ravallec-Ple R, Gilmartin L, Van Wormhoudt A, Le Gal Y. 2000. Influence of the hydrolysis process on the biological
activities of protein hydrolysates from cod (Gadus morhua) muscle. J Sci Food Agric 80:2176-2180.
Ravallec-Ple R, Charlot C, Pires C, Braga V, Batista I, Van Wormhoudt A, Le Gal Y, Fouchereau-Peron M. 2001. The
presence of bioactive peptides in hydrolysates prepared from processing waste of sardine (Sardina pilchardus). J
Sci Food Agric 81(11):1120-1125.
Sanakana S, Tachibana Y, Ishihara N, Juneja LR. 2004, Antioxidant activity of egg-yolk protein hydrolyzates in a linoleic
acid oxidation system. Food Chem 86(1):99-103.
binnenwerk.indd 425 21-08-2006 12:35:44

Shahidi F, Amarowicz R. 1996. Antioxidant activity of protein hydrolysates from aquatic species. JAOCS 73(9):1197-
1999.
Suetsuna K, Ukeda H, Ochi H. 2000. Isolation and characterization of free radical scavenging activities peptides derived
from casein. J. Nutr. Biochem. 11:128-131.
Synowiecki J, Al-khateeb NAAQ. 2000. The recovery of protein hydrolysate during enzymatic isolation of chitin from
shrimp Crangon crangon processing discards. Food Chem 68:147-152.
Vandanjon L, Bourseau P, Jaouen P. 2005a. Concentration et désodorisation de solutions de peptides d’origine marine
par ultrafiltration et nanofiltration. 10ème Congrès SFGP, Toulouse, 20-22 sept 2005. Forthcoming
Vandanjon L, Johannsson R, Derouiniot M, Jaouen P. 2005b. Concentration and purification by membrane processes of
blue whiting hydrolysates. J Food Eng. Submitted.
Venugopal V, Shahidi F. 1995. Value-added products from underutilized fish species. Crit Rev Food Sci Nutr 35(5):431-
453.
Wako Y, Ishikawa Q, Muramoto K. 1996. Angiotensin I-converting enzyme inhibitors in autolysates of squid liver and
mantle muscle. Biosci Biotech Biochem 60:1353-1355.
Zeman LJ , Zydney AL. 1996. Microfiltration and Ultrafiltration: Principles and applications. Marcel Dekker, New York.
604 p.
binnenwerk.indd 426 21-08-2006 12:35:44

Acid and alkaline-aided protein recovery from Cape hake by-
products

Department of Technological Innovation and Upgrading of Fish Products, Fisheries and Sea
Research Institute - INIAP/IPIMAR, Avenida de Brasília, 1449-006 Lisboa, Portugal
Abstract
Cape hake whole muscle and myofibrillar proteins showed a minimal solubility at pH 5.5 and
maximal solubility at pH 2.5 and 12. The proteins were recovered by solubilisation of unwashed
and washed mince at pH 2 (UWAc and WAc) and 12 (UWAl and WAl) followed by precipitation at
pH 5.5. The yields achieved ranged between 43.6 and 63.0%. WAc had the highest emulsifying
and foaming capacity and UWAl exhibited the best textural properties. The folding test score of
all products was around 3, which could be considered of medium quality. The decrease of protein
extractability of recovered protein suggests significant protein denaturation and aggregation.
Keywords: alkali/acid solubilisation, yield, fish proteins, functional properties, solubility, Cape
hake
Introduction
The traditional production of surimi (concentrate of myofibrillar proteins) involves the removal
of unnecessary materials, such as fat, pigments, and sarcoplasmic proteins by several washings
of the fish mince. The main disadvantages of this process are the large amount of water required,
the loss of proteins in the washing water and the relatively low yields achieved. Moreover, as
pointed out by Kristinsson and Demir (2003), there is a great need to utilize processing
byproducts and underutilized species. Thus, to address these problems a new process for
increasing yield and utilizing unconventional fish sources was developed to produce functional
protein isolates (Hultin and Kelleher 1999, 2000a; Hultin and others 2000). The process is
based on the solubilisation of fish proteins at acidic (pH 2-3.5) or alkaline (pH 10.5-11.5)
conditions followed by precipitation of all soluble proteins at their isoelectric point (pH 5-6).
This process permits the utilization of low-value, whole, dark-muscle fish species and it is also
advantageous for white-fleshed species (Hultin and Kelleher 2000b). These authors also claim
that this new process improves the safety of the recovered proteins.
The production of fillets and fish portions from white fish generates a large volume of by-
products, which are reduced to fish meal or are just dumped. Thus, the objectives of this
work were to study the recovery of proteins from Cape hake by-products by acid and alkaline
processes and to measure the main functional properties of the proteins obtained.
binnenwerk.indd 427 21-08-2006 12:35:44

Material
Frozen by-products resulting from the portioning (fish ‘sawdust’ and cut offs) of Cape hake
(Merluccius capensis) were used.
Methods
Extraction of myofibrillar proteins
For the extraction of myofibrillar proteins, the Cape hake mince was washed twice with distilled
water using a mince:water ratio of 1:10. The final pellet obtained was kept frozen until further
utilisation.
Protein solubility as a function of pH

The solubility of mince and myofibrillar proteins in distilled water was studied. The same study
was done in myofibrillar proteins using hard water (60 ppm CaCO3). The mince or the proteins
were diluted in water (1:50), homogenised in a Polytron homogeniser and the pH of the protein
suspension was adjusted to the different pH values (2-12) using 1 M HCl or NaOH. The sample
solutions were centrifuged at 30000 x g at 5 ºC for 20 min and the protein content of the
supernatant was determined according to the method of Bradford (1976). Protein solubility (%)
was calculated by dividing the protein concentration in the supernatant after centrifugation
with the total protein concentration before centrifugation.
Protein recovery
The acid and alkaline protein recovery from Cape hake mince was done following the sequence
presented in Figure 1. The yield of protein recovery was calculated by dividing the total protein
content in the protein recovered by the protein in the starting material.
Protein content
The determination of protein content was done by the method of Kjeldahl as described in the
AOAC (1998).
Protein extractability
Protein solubility of muscle and all recovered proteins at low ionic-strength buffer (50 mM KCl
potassium phosphate, pH 7) (WSP = water-soluble protein) and high ionic-strength buffer (0.6
M KCl potassium phosphate, pH 7) (SSP = salt-soluble protein) was studied as described by
Yongsawatdigul and Park (2004). The protein concentration in the supernatant was determined
by the method of Bradford (1976). The protein extractability was expressed as mg of protein in
the supernatant per g of protein in the product.
Emulsifying capacity (EC)

A 5% protein solution was placed into a vessel and continuously stirred using a Polytron PT
3100 homogenizer at 20,000 rpm as small volumes of oil (approximately 1 ml/s) were titrated
into the vessel. Oil addition was stopped when the emulsion breaks down or inverts, which was
determined by carefully observing the change in viscosity. The EC was defined as the maximum
amount of oil that can be dispersed in a protein solution without the emulsion breaking down
or inverting.
binnenwerk.indd 428 21-08-2006 12:35:44

Acid and alkaline-aided protein recovery from cape hake by-products
Cape hake
(sawdust and cut offs)
Removal of skin and

bones by hand
Grinding
10 vol. distilled water
Homogenisation
(washing)
Centrifugation
Supernatant
(continuous centrifuge)
Sediment extracted
10 vol. distilled water
Homogenisation
(manual stirring)
Solubilisation at pH 2 Solubilisation at pH 12
(adding 1M HCl and stirring) (adding 1M NaOH and stirring)
2nd sol. 2nd sol.
Centrifugation Centrifugation
Pellet (continuous centrifuge) (continuous centrifuge) Pellet
Solubilized protein Solubilized protein
Precipitation at pH 5.5 Precipitation at pH 5.5

(adding 1M NaOH and stirring) (adding 1M NaOH and stirring)
Centrifugation Centrifugation
(continuous centrifuge) (continuous centrifuge)
Recovered protein Recovered protein
(UWAc or WAc) (UWAl or WAl)
pH ajust at 6.75 pH ajust at 6.75
Storage at -20 ºC Storage at -20 ºC
Figure 1. Acid and alkaline protein recovery procedure. UWAc, WAc - Proteins from acidic solubilisation of
unwashed and washed mince respectively. UWAl, WAl - Proteins from alkaline solubilisation of unwashed
and washed mince respectively.
Foaming capacity (FC) and foam stability (FS)

Foaming capacity was determined according to the method described by Diniz and Martin
(1997) with some modifications. A 40 ml quantity of 2% protein dispersion in 5% NaCl was
homogenized in a 50 ml graduated cylinder, for 3 min at 10000 rpm, using a Polytron PT 3100
homogenizer (Kinematica AG, Lucerne, Switzerland). FC was expressed as the volume of foam
formed immediately after homogenisation per g of protein. Foam stability (FS) was estimated
as the percent of foam remaining after 30 and 60 min.
binnenwerk.indd 429 21-08-2006 12:35:45

Gel forming ability

The preparations of gels and the puncture, compressing, and folding tests were done as described
by Mendes and others (1998).
Colour parameters
The colour of gellified products prepared with the recovered proteins was measured with a
colour meter Macbeth Color-eye 3000 (Kollmorgen Instruments Corporation, New York, USA).
Tristimulus colour coordinates were used to measure the degree of lightness (L*), redness (+a*)
or greenness (-a*), and yellowness (+b*) or blueness (-b*). Whiteness was calculated by the
equation: Whiteness = L*-3b* as published by Park and others (1994).
The process of protein recovery from acid and alkali-aided solubilisation requires good protein
solubility at low or high pH values to achieve high yields. Thus the solubility profile of the Cape
hake mince and myofibrillar proteins as a function of pH was studied. As described by Hultin
and others (1995) a wide range of conditions has been followed to study the solubility of fish
muscle proteins. In this work the same fish:extraction medium ratio as described by Stefansson
and Hultin (1994) was chosen but at lower centrifugal force. As it can be seen in Figure 2, that
all solubility profiles show the typical U-shape form with a minimum at pH 5.5. The maximum
solubility was measured at pH 2.5 and 12. Similarly, Choi and Park (2002) obtained maximum
solubility of Pacific whiting mince at pH 2 and 11 and minimum solubility at pH 5. Kim and
others (2003) working with the same species reported the lowest solubility at pH 5.5 and
maximum solubility at pH 3 and 12. The Cape hake myofibrillar proteins, however, exhibited
lower solubility than whole muscle proteins, particularly in the pH range between 4.5 and 9.0.
Similar results obtained by Yongsawatdigul and Park (2004) for the solubility profile of rockfish
100
80
% extracted protein
60
40
20
0
0 2 4 6 8 10 12 14
pH
Myo proteins (d.w.) Myo proteins (h.w.) Mince (d.w.)
Figure 2. Effect of pH on the solubility of Cape hake myofibrillar (myo) and mince. d.w. – distilled water;
h.w. - hard water (60 ppm).
binnenwerk.indd 430 21-08-2006 12:35:45

actomyosin and mince were reported. On the other hand, the presence of Ca2+ in water also
increased the protein solubility both in the acidic and alkaline pH range. A similar effect was
also recorded for the blue whiting myofibrillar proteins (Batista and others 2003) but only for
the lowest acidic pH values. In the case of minced muscle a second minimum of the protein
solubility was recorded at pH 9.5, which was also described by Kristinsson and Demir (2003)
for the solubility of catfish and croaker proteins.
The yields achieved in the protein recovery at a pilot plant scale are shown in Table 1. The lower
yields obtained in products WAc and WAl are due to the elimination of sarcoplasmic proteins
in the washing step. The objective of the initial mince washing was to evaluate the possible
influence of sarcoplasmic proteins on the functional properties of the recovered proteins. The
second solubilisation was done to recover the proteins present in the weak gel formed on
the top of the sediment after centrifugation. The alkaline process yielded higher recoveries
compared to the acid process. In both processes the yields achieved in the solubilisation step
were similar but the isoelectric precipitation was more effective in the case of the alkaline-
added process. Slightly higher yield was achieved in the alkaline-added sardine protein recovery
but in the case of blue whiting proteins the reverse was verified (Batista and others 2003).
Kim and others (2003) also obtained the highest recovery of Pacific whiting proteins at pH 12.
However, Kristinsson and Demir (2003) reported higher protein recoveries in the acid-added
processing of the fish muscle from different species.
The protein content of the different samples of protein recovered ranged between 9.3% and
12.2% depending on its water content. Some fat was released during the solubilisation process,
which was easily separated by centrifugation to recover the solubilised proteins. As described
by Schoen (1977), the functional properties of proteins are dependent on the isolation methods
used. The protein recovered after acidic and alkaline treatment exhibited significantly lower
(p<0.05) solubility in both low and high ionic-strength buffers than the muscle proteins
(Figure 3). Within the recovered proteins the extractability of both types of proteins from
product WAc was significantly lower (p<0.05) than those of the other products. The decrease of
protein solubility in the recovered proteins may be due to protein denaturation and aggregation
as referred by Yongsawatdigul and Park (2004), who also obtained similar results working with
rockfish muscle proteins.
Table 1. Yields achieved in the protein recovery after acid and alkali-added processing.
Yields (%)
Acidic solubilisation Alkaline solubilisation
UWAc WAc UWAl WAl
Washing - 76.8 - 78.4

1st solubilisation 80.6 60.7 78.2 61.3
2nd solubilisation 2.2 4.3 4.2 8.8
Isoelectric precipitation 70.9 87.3 76.5 91.2
Global yield 58.7 43.6 63.0 50.1
binnenwerk.indd 431 21-08-2006 12:35:45

200
x
160 a
Protein (mg/g)
120
y y
y
d
b z b
80 c
40
0
Muscle UWAc WAc UWAI WAI
Recovered proteins
Figure 3. Protein extractability of muscle and different products. – water-soluble proteins; – salt-soluble
proteins. Results with the same letter are not significantly different.
The emulsifying capacity of the four samples of proteins recovered under different conditions is
shown in Figure 4. The significantly higher (p<0.05) EC of the product WAc may be due to the
higher hydrophobicity of its proteins once this property is directly related to the capacity of
interactions between proteins and lipids. The EC of all products is generally low when compared
with the results published for fish proteins from different species (Borderías and others 1985;
Huidobro and Tejada 1993; Huidobro and others 1998). However, as pointed out by Cofrades and
others (1996) there are considerable differences in the protein functionality depending not only of
the myosystem characteristics but also on quantitative factors affecting the nature of the system
in which the functional properties are measured (protein concentration, pH, ionic strength).
100 b
E. capacity (goil/g prot.)
80
60 a a
c
40
20
0
UWAc WAc UWAI WAI
Recovered proteins
Figure 4. Emulsifying capacity of the recovered proteins. Results with the same letter are not significantly
different.
binnenwerk.indd 432 21-08-2006 12:35:46

The foaming capacity of the different recovered proteins (Figure 5) was relatively low and
the results did not present any clear trend. However, it is to stress the similarity between
the foaming and the emulsifying capacity of the recovered proteins, which is related to the
analogy between the formation of a foam and an emulsion. The foaming capacity of product
WAc was also significantly higher (p<0.05) than that of the other proteins. The data available
on the foaming capacity of fish proteins is relatively scarce but it has been reported that
certain species exhibit good aptitude for this functional property (Hermansson and others 1971;
Baldwin and Sinthavais 1974; Montero and Borderías 1991). However, the changes undergone by
fish proteins during frozen storage are responsible for changes in texture and losses of protein
functionality (Huidobro and Tejada 1993). The raw material used in this study is a by-product
from a processing factory and it was thawed during fish portioning, refrozen and frozen stored
until utilization, which accelerate protein changes, particularly aggregation. These changes
together with those eventually occurring during protein solubilisation and precipitation could
be responsible for the low foaming capacity of the recovered proteins.
The products UWAc, WAc, and WAl presented similar foam stability (Figure 6), which could be
considered fairly good. As published by Wilde and Clark (1996) the formation of highly elastic
adsorbed layers by globular proteins increase the viscosity and will contribute for the foam
stability. The proteins may maintain an extensive intermolecular network forming strong films
and consequently more stable foams.
The washing of mince before protein recovery led to a significant increase (p<0.05) of the
breaking force of the gel prepared with proteins from the acidic solubilisation (Figure 7) but
this effect was not observed in the proteins from the alkaline process. The latter result seems
that the presence of sarcoplasmic proteins did not interfere with the textural properties of
gels, which was previously referred by Yongsawatdigul and Park (2004). Product WAc was harder
than product UWAc but the products UWAl and WAl were similar. The product UWAl was slightly
harder than the product UWAc and showed the best textural properties. In general, the products
obtained in this work had similar textural properties to those reported by other authors (Kim
and others 2003; Yongsawatdigul and Park 2004). The lowest breaking force recorded for product
40
Foam volume/g protein
b
30 ab
20 a
a
10
0
UWAc WAc UWAI WAI
Recovered proteins
Figure 5. Foaming capacity of the recovered proteins. Results with the same letter are not significantly
different.
binnenwerk.indd 433 21-08-2006 12:35:46

140
120
Foam stability (%)

100
80
60
40
20
0
UWAc WAc UWAI WAI
Recovered proteins
30 min 60 min
Figure 6. Foam stability of the recovered proteins.
0.005 16
y
y
0.004 y
Breaking force (kN)
Deformation (mm)
a 12
a
0.003 ab
x b 8
0.002
4
0.001
0.000 0
UWAc WAc UWAI WAI
Recovered protein
Breaking force sd Deformation
Figure 7. Breaking force and deformation of the different recovered proteins. Results with the same letter
are not significantly different.
UWAc was also obtained by Choi and Park (2002) and Yongsawatdigul and Park (2004) in proteins
recovered from acid solubilisation. However, Kim and others (2003) reported that both Pacific
whiting proteins treated at pH 11 and pH 2 had good textural properties. The protein gels prepared
by Hultin and Kelleher (1999) from acid solubilisation of Atlantic mackerel and cod showed high
shear stress and shear strain.
The product UWAl showed the highest hardness and elasticity values (Figure 8), whereas product
UWAc had the lowest hardness value. The positive effect of the previous mince washing on the
hardness is clearly evidenced in the acid process and the reverse in the alkaline one. These
results are also in accordance with those obtained in the puncture test.
binnenwerk.indd 434 21-08-2006 12:35:47

The cohesiveness values were between 0.37 for the product WAc and 0.53 for the product WAl,
showing the other two products intermediate results. The effect of washing the mince on the
cohesiveness did not show a clear trend.
The various products had a relatively low folding test score (Figure 9) and the mince washing
and the solubilisation process used did not affect this property. The folding test is an empirical
test but it gives a good indication of the quality of the gels and the scores obtained were in
accordance with the texture measurements.
The L* values of all recovered proteins were higher than the mince as well as the whiteness
(Figure 10). This is due to the removal of myoglobin during the protein recovery in the different
processes. Higher b* values were measured in the products UWAc, UWAl, and WAl than in the
mince. This increase of the b* value was also referred in several works on fish species (Choi and
Park 2002; Kim and others 2003; Yongsawatdigul and Park 2004) and chicken thigh (Kelleher and
0.10 100
0.08 80
Hardness (kN)
Elasticity (%)
0.06 60
0.04 40
0.02 20
0.00 0
UWAc WAc UWAI WAI
Recovered proteins
Hardness Elasticity
Figure 8. Hardness and elasticity of the different recovered proteins.
4
Folding test score
0
UWAc WAc UWAI WAI
Recovered proteins
Figure 9. Folding test score of the different recovered proteins.
binnenwerk.indd 435 21-08-2006 12:35:47

6.0 100
80
4.0
Whiteness
60
b* value
40
2.0
20
0.0 0
Mince UWAc WAc UWAI WAI
Recovered proteins
b* value Whiteness
Figure 10. Whiteness and b* value of the different recovered proteins.
Hultin 2000) and it could result from the acceleration of myoglobin at low or high pH values. It
has also to be mentioned the decrease of the b* values in the products WAc and WAl relatively to
the correspondent ones prepared from unwashed mince, which evidences the removal of myoglobin
in the washing step. The a* values of all recovered proteins and mince were quite similar being
between –1.94 for mince and –1.79 for the product UWAc.
Conclusions
The following conclusions can be drawn from the results obtained in this work:
• The recovered proteins presented lower extractability in both low and high ionic-strength
buffers than the muscle suggesting that protein denaturation and aggregation occurred.
• The protein recovered after both treatments had relatively low emulsifying and foaming
capacity but the highest values were recorded in the product obtained after acidic
solubilisation of washed mince. However, the foam stability of all products could be
considered average.
• The proteins recovered after alkaline solubilisation of unwashed mince showed the best
textural properties but the gels obtained are of medium quality according to the results of
the folding test score.
Acknowledgements
This work was performed within the Integrated Project (IP) SEAFOODplus, granted by the EU
Commission under contract No FOOD-CT-2004-506359.
References
AOAC, 1998. Official Methods of Analysis. 16th Ed., 4th revision. Association of Official Analytical Chemists. Washington
DC. (cd-rom).
Baldwin R C, Sinthavais S. 1974. Fish protein concentrate foam. J Food Sci 39(5):880-882.
Batista I, Mendes R, Nelhas R, Pires C. 2003. Proteins from sardine and blue whiting recovered by new extraction
techniques: solubility and gelation properties. Proceedings of the TAFT 2003 Conference. Reykjavik: The Icelandic
Fisheries Laboratories. p 276-278.
binnenwerk.indd 436 21-08-2006 12:35:47

Borderías AJ, Jiménez-Colmenero F, Tejada M. 1985. Viscosity and emulsifying ability of fish and chicken muscle
protein. J Food Technol 20:31-42.
Bradford M M. 1976. A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the
principle of protein-dye binding. Anal. Biochem 72:248-254.
Choi Y J, Park J W. 2002. Acid-aided protein recovery from enzyme-rich Pacific whiting. J Food Sci 67(8):2962-2967.
Cofrades S, Carballo J, Careche M, Jiménez-Colmenero F. 1996. Emulsifying properties of actomyosin from several
species. Lebensm Wiss Technol 29:379-384.
Diniz M, Martin A M. 1997. Effects of the extent of enzymatic hydrolysis on functional properties of shark protein
hydrolysate. Lebensm Wiss Technol 30:266-272.
Hermansson A M, Sivik B, Skjoldebrand C. 1971. Functional properties of protein for food-factors affecting solubility,
foaming and swelling of fish protein concentrate. Z Lebensm Unters Forsch 4:201-205.
Huidobro A, Montero P, Borderías A J. 1998. Emulsifying properties of an ultrafiltered protein from minced fish water.
Food Chem 61(3):339-343.
Huidobro A, Tejada M. 1993. Emulsifying capacity of fish mince from several species during frozen storage. J Sci Food
Agric 61:333-338.
Hultin HO, Feng Y, Stanley DW. 1995. A re-examination of muscle protein solubility. J Muscle Foods 6:91-107.
Hultin HO, Kelleher SD. 1999. Process for isolating a protein composition from a muscle source and protein composition.
U.S. patent 6,005,073.
Hultin HO, Kelleher SD. 2000a. High efficiency alkaline protein extraction. U.S. patent 6,136,959.
Hultin HO, Kelleher SD. 2000b. Surimi processing from dark muscle fish. In: Park JW, editor. Surimi and surimi seafood.
New York: Marcel Dekker. p 59-77.
Hultin HO, Kelleher SD, Feng Y, Kristinsson HG, Richards P, Undeland I, Ke S. 2000. High efficiency alkaline protein
extraction. U.S. Patent Application No. 60/230,397.
Kelleher SD, Hultin HO. 2000. Functional chicken muscle isolates prepared using low ionic strength, acid solubilization/
precipitation. Proceedings of 53rd Annual Reciprocal Meat Conference. Savoy, Illinois: American Meat Science Assn.
p 76-81.
Kim YS, Park JW, Choi YJ. 2003. New approaches for the effective recovery of fish proteins and their physicochemical
characteristics. Fisheries Sci 69:1231-1239.
Kristinsson H, Demir N. 2003. Functional fish protein ingredients from fish species of warm and temperate waters:
comparison of acid- and alkali-aided processing vs. conventional surimi processing. In: Bechtel PJ, editor. Advances
in Seafood Byproducts: 2002 Conference Proceedings. Alaska Sea Grant College Program. Fairbanks: University of
Alaska. p 277-295.
Mendes R, Batista I, Kandando R, Howell N. 1998. Influence of washing parameters on the characteristics of horse
mackerel (Trachurus trachurus) mince. J Food Biochem 22(6):511-528.
Montero P, Borderías AJ. 1991. Emulsifying capacity of collagenous material from muscle and skin of hake. Food Chem
41:251-267.
Park JW, Yongsawatdigul J, Lin TM. 1994. Rhelogical behavior and potential cross-linking of Pacific whiting (Merluccius
productus) surimi gel. J Food Sci 59(4):773-776.
Schoen HM. 1977. Functional properties of proteins and their measurement. In: Whitaker JR, Tannenbaum SR, editors.
Food Proteins. Wesport: AVI Publishing Company. p 387-400.
Stefansson G, Hultin HO. 1994. On the solubility of cod muscle proteins in water. J Agric Food Chem 42(12):2656-
2664.
Wilde PJ, Clark DC. 1996. Foam formation and stability. In: Hall, G M, editor. Methods for Testing Protein Functionality.
London: Blackie Academic and Professional. p 110-148.
Yongsawatdigul J, Park JW. 2004. Effects of alkali and acid solubilisation on gelation characteristics of rockfish muscle
proteins. J Food Sci 69(7):499-505.
binnenwerk.indd 437 21-08-2006 12:35:48

binnenwerk.indd 438 21-08-2006 12:35:48
Quality evaluation of silver smelt (Argentina silus) and its
suitability for seafood products: Compilation of results from three
European research centres
Iren Skjåstad Stoknes1, Jörg Oehlenschläger2 and Ronan Gormley3
1Møre Research, PO Box 5075, N-6021, Aalesund, Norway,
2Federal Research Centre for Nutrition and Food, Department of Seafood Research, Palmaille 9,
D-22767 Hamburg, Germany
3Ashtown Food Research Centre (Teagasc), Ashtown, Dublin 15, Ireland
Abstract
The quality of silver smelt has been tested by three European research centres in Germany,
Ireland and Norway. Møre Research (Norway) examined the proximate composition, water
holding capacity, cook loss and colour of the fish flesh. Silver smelt had a high protein content
(18.7%) and a low water content (79.1%). It is a lean species with a fat content around 1%.
The water holding capacity of frozen and thawed fish flesh was found to be very good compared
to the reference species cod (Gadus morhua). Sensory tests on steamed fillet samples carried out
at Teagasc (Ireland) (previously iced) indicated that silver smelt odour and flavour ratings were
high and remained relatively constant over days 3 - 6 post-catching. Paired comparison tests
of steamed silver smelt versus steamed cod showed a preference ratio of 9/3 (12 panellists) in
favour of silver smelt. Most panellists agreed that silver smelt had a pleasant bland flavour and
a firmer texture than the sample of cod. Fishcakes from silver smelt were preferred to those
made from pollock, ling or cod and were much preferred to a commercial sample. With the
exception of crumb colour, silver smelt nuggets were preferred on all counts to commercial cod
nuggets. The preferences were statistically significant except for texture. However, some tasters
found the texture of the silver smelt nuggets to be rubbery or too firm. The strength of fish gels
is a good index of water binding and is important for the textural properties of fish products.
Full strength silver smelt gels were softer than those from ling or cod but were firmer than
pollock gels. Freezing increased centrifugal drip values in silver smelt fillets. However, frozen
mince from silver smelt had a much higher centrifugal drip value indicating that mince should
not be stored frozen but as whole fillets that are minced post-thawing. These data, overall
suggest that silver smelt has potential for products but that toughening during frozen storage
could be a potential problem. Quality is not only associated with technological properties and
sensory attributes but also with the concentrations of beneficial components and the burden
of toxic components and elements like heavy metals in the edible part of fish. Therefore at
the Federal Research Centre for Nutrition and Food (Germany) the content of the heavy metals
cadmium and lead in silver smelt from different fishing ground had been determined as well
as the content of the essential trace elements zinc, copper and selenium. While the cadmium
content was found to be extremely low, the lead content was found to be in the range of other
marine fish species as e.g. in small spotted cat shark (Scyliorhinus caniculus). The levels of the
essential trace elements copper, zinc and selenium are found to be in the lower range compared
with other marine fish species.
Keywords: greater argentine, silver smelt, Argentina silus
binnenwerk.indd 439 21-08-2006 12:35:48

Iren Skjåstad Stoknes, Joerg Oehlenschläger and Ronan Gormley
Introduction
Silver smelt (Argentinus silus) is an underutilised fish species caught in many locations in
Europe and elsewhere. It has a white flesh and a bland flavour which make it suitable for sale as
fresh fillets or in products. However, it has received relatively little exposure compared to some
other underutilised species despite its good flesh properties. Surveys on the population of silver
smelt have been published by a number of authors. Magnussen (1996) studied the occurrence
of silver smelt in Icelandic waters. Monstad and Johannessen (2003) reported that populations
of 200-400kt were found in 1989-94 along the shelf west of the UK and Ireland (via acoustic
measurements) and recordings (1990-92) from mid-Norway showed slightly higher values that
these. Johannessen and Monstad (2003) found that silver smelt was more congregated at
the continental slope and in the deeper parts of the shelf (mid Norway) in springtime while
in autumn it was scattered over a wide area. Immature specimens occurred mainly at depths
less than 300m while the mature individuals predominated at greater depths. Bergsted and
others (2003) studied the ecology of fish inhabiting the 300-700m deep shelf trough of the
central Skagerrak (North-eastern North Sea) and found that silver smelt was highly dominant
in this area. The diet of silver smelt was uncertain because a high fraction of stomach contents
were unidentifiable. However, a probable food source may be mesopelagic and benthopelagic
gelatinous plankton. Published data on the quality of silver smelt and on its suitability for
products is also limited. Gormley and others (1993) studied the effect of long-term frozen
storage on silver smelt flesh quality and also evaluated the properties of gels made from this
species (Gormley and others 1991; Gormley and others 1994). The overall outcome from these
studies was that silver smelt has good quality attributes and is suitable for seafood products.
Mackie and Hardy (1969) reported good acceptability of silver smelt by taste panels but stressed
that rancidity could occur during long-term storage because of its unsaturated oil content.
During the last years, there has been an increasing interest of exploring so far under-utilised
fish species that can replace or supplement the available commercial species. It is, therefore,
important that the new species can compete with the traditional species regarding both
taste and nutritional values. Synnes and Stoknes (2004) made a selection of under-utilised
fish species as possible raw material for product development. Silver smelt (Argentina silus),
northern wolffish (Anarhichas denticulatus) and polar cod (Boregadus saida) were chosen as
interesting species for further investigation of raw material properties. In this study, the
chemical composition and some physical attributes for silver smelt are reported.
The protein content and the water holding capacity of the proteins are important factors for the
texture of the fish flesh. Water holding capacity refers to the ability of the proteins to retain
water and retain it against gravitational force within a protein matrix. A high protein content
and a good water holding capacity of proteins is therefore important as it often improves the
texture of the fish flesh.
Furthermore it is necessary for a complete quality characterisation of a new food fish species to
investigate the concentrations of both, beneficial components as selenium which is an essential
element and is present in marine fish in considerably amounts as well as toxic components like
heavy metals in the edible part of the fish candidate.
The current study reports results on the quality testing of silver smelt from three European
research centres, i.e. Møre Research in Norway (Centre 1), the Federal Research Centre of
binnenwerk.indd 440 21-08-2006 12:35:48

Quality evaluation of silver smelt and its suitability for seafood products
Nutrition and Food, Department of Seafood Research in Germany (Centre 2) and Ashtown
Food Research Centre (Teagasc) in Ireland (Centre 3). The results from the three centres are
presented separately followed by a collated discussion. This combined output on silver smelt
was coordinated under the ongoing integrated research project SEAFOODplus of the EU 6th
Framework programme.
Source of the fish
Centre 1 (Møre Research, NO)

Silver smelt (Argentinus silus) was caught in the North Sea at the positions 63˚33.0’ N 10˚50.2’
E and 63˚43.4’ N 10˚57.7’ E off North West of Norway in November 2004 by the Institute of
Marine Research, Norway. The species was caught at depths of 54 to 59 meters. The length of
the fishes varied between 29 to 42 cm and the weight varied from 270 to 980 grams. Cod (Gadus
morhua) was used for comparison and was sourced from a commercial long liner and was caught
in the Western part of the Barents Sea at the position 73˚14.0’ N 14˚50.0’ E at depths of 518
to 602 meters. The average length was 65 cm and the average weight was 2.5 kg. All species
were frozen whole onboard and stored at -30 °C.
Centre 2 (Federal Research Centre for Nutrition and Food, DE)

Silver smelt for the experiments on the suitability for product processing was caught during
the 10th and 14th cruise of the fishery research vessel “Walther Herwig II” in seventies. During
two cruises with the FRV “Walther Herwig II” silver smelt was caught for proximate analysis
by bottom trawling in the West British waters. The cruises have been made during a long term
programme of Federal Research Centre for Nutrition and Food between 1980 and 1986 in which
the composition of under-utilised species has been investigated in correlation with catching
grounds, seasons and state of maturity of the fish. The fish were selected from the catch
immediately after the haul. The fish were killed, allowed to bled and than manually filleted.
Onboard of the vessel length, weight, pH, TVB-N and ammonia was measured in the edible
part of the fish. The filets were then thawed in a blast freezer to –30 °C and stored in frozen
condition (-25 °C) until analysed in the laboratory at land. Here the samples were analysed for
raw protein (nitrogen x 6.25), total fat content, total minerals (ash), chloride (measured as
NaCl), phosphorus, moisture and trimethylamineoxide TMAO.
During the cruise no. 195 of the FRV “Walther Herwig III” 5 specimen of Argentina silus were
caught west off Shetland Islands on 60° 40.58’ N 002° 58.69’ W by bottom trawling. Fish were
treated as described above. The fillets obtained were taken to a place where sub samples for
later trace elements analysis were taken in a way to avoid any contamination of the samples.
The sub samples were then frozen in special containers and kept frozen until sample preparation,
digestion and analysis in the laboratory at land. These samples were used for measuring the
content of toxic elements cadmium and lead and essential elements zinc and copper in edible
part of silver smelt. Selenium content in the edible part was determined in samples of Silver
smelt which were sent from Centre 1 to Centre 2.
Centre 3 (Ashtown Food Research Centre, IE)

Two lots of silver smelt (Argentinus silus) were caught off the west coast of Ireland and were
delivered to Ashtown Food Research Centre (ARFC) in ice. Catch 1 samples were received on
day 2 (day 0 = day of catching) and catch 2 on day 3 after catch. The fish were tested as
whole fish, fillets and mince, and samples were frozen and set aside for testing at a later date.
binnenwerk.indd 441 21-08-2006 12:35:48

Commercial block frozen fillets originating from the same catches were also received at AFRC for
testing. Samples of ling (Molva molva), pollock (Pollachius pollachius) and cod (Gadus morhua)
were used for some of the comparisons and were sourced from the Dublin fish market and
were delivered in ice to AFRC. Sub-samples of these fillets were also blast frozen (-35 °C/2 h).
Commercial samples of frozen cod fishcakes and fillets were purchased in a local supermarket
for use in comparative tests with silver smelt nuggets, fishcakes and breaded fillets/portions.
Methods

Fillet yield
Blocks of frozen whole fish were thawed over night in cold running water and the fishes
were then filleted. Skin remained on. Fillet yield was calculated from round fish weight of 10
individuals of silver smelt.
Sample preparation
Whole fillets from left side of the fish were homogenised by a food processor (Braun Combimax
600) to a homogenous mass. Whole fillets from right side of the fish were put in plastic bags
and frozen. Also part of homogenised samples were put in plastic containers and frozen. Frozen
samples were sent to the Federal Research Centre for Nutrition and Food in Germany for analysis
of selenium.
Moisture content
The moisture content was determined according to a modified version of the method described
by Børresen 1980. The moisture content of 5 g of homogenised sample was determined by
drying the sample in an oven at 105 ˚C in 22 h. The water content was determined as the
weight loss after drying. It was conducted three parallels for each individual and the mean
value was calculated.
Protein content
The total protein content was determined by the Kjeldahl method (Ritzman and Daniels 1975)
by which the concentration of nitrogen is measured. A conversion factor of 6.25 was used to
convert total nitrogen to crude protein. The analyses were performed by The Norwegian Food
Control and Animal Welfare Authority, Department Ålesund, Norway.
Fat content
The total fat content was determined gravimetrically after ethyl acetate extraction according
to Losnegard and others (1979) by The Norwegian Food Control and Animal Welfare Authority,
Department Ålesund, Norway.
Water holding capacity

Water holding capacity was determined according to a modified version of the method described
by Børresen 1980. Homogenised sample from each individual (2 g) was centrifuged at 4500
rpm for 15 min, through a filter that allowed the water to be removed from the muscle. After
centrifugation, the fillets were weighed again, and the difference before and after centrifugation
was calculated. The water content was determined and used to calculate the water holding
capacity. Water holding capacity is presented as percent of remaining sample after centrifugation.
The analysis was performed in triplicate for each individual and mean value was calculated.
binnenwerk.indd 442 21-08-2006 12:35:48

Cook loss
Cook loss was determined according to a modified version of the method described by Børresen
(1980). Homogenised sample (2 g) from each individual were put in a container with a filter
that allowed the water to be removed from the sample, and incubated at 80 ˚C for 15 minutes.
Then the containers were weighed. The water content was determined, and used to calculate the
cook loss. Cook loss is presented as percent of sample. The analysis was performed in triplicate
for each individual and mean value was calculated.
Colour measurement
The colour of the raw muscle was measured with a Minolta Chromameter CR20. The measurement
describes the whiteness (L=100 is white, L=0 is black), green-red (a*=-60 is green, a*=60 is
red) and yellow-blue (b*=-60 is blue, b*=60 is yellow) colours of the sample. Ten measurements
were conducted on each fillet from each individual, and mean value was calculated.

Selenium
After mincing and freeze drying the samples were digested by microwave and the selenium
content was determined by GF-AAS using Pd-modifier and standard-addition-technique. Sample
preparation for GF-AAS: 0.3 – 0.35 g of freeze dried and homogenised fish samples were
weighted exactly into the reacting tubes for wet digestion. 4 ml nitric acid (65% Suprapur,
Merck, Darmstadt, Germany) were added and the suspensions were shaken, then 1 ml hydrogen
peroxide (30%, Merck) was added, the tubes were covered with a lid and placed into the
microwave high pressure digestion system ultraCLAVE III (MLS GmbH, Leutkirch, Germany).
Certified standard reference material CRM 422 (cod muscle) was treated in the same way to
ensure analytical quality. The microwave program applied has the following characteristics:
Start temperature: 22 °C, loading pressure: 40 bar, 5 min heating up to 50 °C (100 bar,
1000 W), 25 min heating up to 110 °C (120 bar, 700 W), 10 min heating up to 170 °C (110
bar, 1000 W) and finally constant temperature for 45 min. Cooling down and pressure release
occurred within 50 min. After opening of the reaction vessel the reaction tubes were put under
a flue until next day. Then the clear solutions were quantitatively transferred and diluted to
50ml in a graduated flask. For GF-AAS (standard-addition-technique) the standard was added
manually to the samples.
Conditions for Atom-Absorption-Spectrometry: The samples, reference material were analysed

by GF- AAS (Perkin Elmer ZL 4100) equipped with an EDL for Se-analyses using Pd-modifier and
standard addition technique. The signal of peak area was measured at 196 nm with Zeeman
background correction using a specific time temperature furnace program. The slit width was
2.0 nm.
Heavy metals
Analysis of cadmium, lead, zinc and copper was made according to Celik, Caklı and Oehlenschläger
(2004).
Proximate composition
Analyses for raw protein (nitrogen x 6.25), total fat content, total minerals (ash), chloride
(measured as NaCl), phosphorus, moisture and trimethylamineoxide TMAO were performed using
the methods described earlier by Oehlenschläger (1991).
binnenwerk.indd 443 21-08-2006 12:35:49

Centre 3 (Ashtown Food research Centre, IE)

Cooking procedures for fillets and mince
Frozen whole fish were thawed in cold running water (3 hours) and were then filleted. Frozen
fillets and frozen mince were thawed overnight at 4 °C. The thawed fish was placed in light (250
gauge) polythene bags (500 g fish/bag) which were submerged in a water bath at 80 °C until an
internal fish temperature of 75 °C was achieved. This normally took 15-20 min. Any excess liquid
was drained off and the samples were presented to taste panels.
Preparation of fish gels

Fish mince, salt (0.5% based on fish weight) and water (where used) were mixed in a Kenwood
food blender for 3 min at full speed. The mixture was transferred to sausage casings and cooked
for 40 min at 90 °C as described by Gormley and others (1994). In the added water gels the
ratio of fish mince to added water was 2 parts fish to 1 part water (w/w). This procedure
produced gel cylinders 25 cm in length and 4 cm in diameter.
Preparation and cooking of products from silver smelt

Fishcakes were prepared from frozen fillets of pollock, ling, cod and silver smelt. The proportions
were fish mince (50%), water (33%), potato flake (16.2%), and seasoning (0.8%). The potato
was mixed with boiling water, the mince and seasoning were added, and the mixture was
blended, hand extruded, blast frozen (-35 °C/2 hours) and enrobed. This involved dipping in a
batter mix (2:1 water/powder, supplied by William Blake Ltd, Dublin), coating in Pandoro crumb
(type F1012), and flash frying in oil at 190 °C for 30 sec. The samples were blast frozen again
followed by storage at –18 °C until required for tasting. The frozen samples were finish-fried
in oil at 160 °C for 5 min and were presented to a taste panel for evaluation. Nuggets (3 x 3 x
2.5 cm) and portions (13 x 7 x 2.5 cm) were cut from frozen blocks of silver smelt mince. The
samples were breaded commercially and then blast frozen (-35 °C) until required. The samples
were finish-fried at 160 °C until they floated and were presented to taste panels together with
commercial brands of cod nuggets and breaded portions.
Physico-chemical tests
The moisture, protein and fat contents of the silver smelt tissue were measured by standard
procedures as outlined by Fagan and others (2003) as was total viable count (TVC). Total
volatile base nitrogen (TVB-N) was tested by the method of Malle and Tao (1986). Colour of
fish pieces and gels was measured with a Hunter D25A colour meter fitted with a 5 cm specimen
port and L and b values were recorded. Centrifugal drip was measured as outlined by Wierbiki
and Deatherage (1958). Gel texture was measured by cutting the gel cylinders into smaller
cylinders with dimensions 2 cm long and 4 cm in diameter. The cylinders were compressed by 8
mm (40%) using a shear press and a flat ended probe 5 cm in diameter as described by Gormley
and others (1994. The force was recorded in Newtons (N). Puncture force was measured by
the same shear press system fitted with a flat ended probe 12 mm in diameter. The probe was
allowed penetrate into the gel until a maximum reading was achieved.
Sensory tests
A range of sensory tests was conducted on the samples and the procedures are given below on
a test by test basis. Samples of steamed (3-5 min) flesh from iced silver smelt were evaluated
for odour and flavour on days 3, 4, 5 and 6 post-catching (test 1). Three judges expert in fish
evaluation were used and scored the samples from 10 (best) to 0 (worst). Test 2 involved a paired
comparison procedure (12 panellists) which compared steamed silver smelt and steamed cod.
binnenwerk.indd 444 21-08-2006 12:35:49

The cod was obtained from a local supermarket. Fish cake preference (test 3) was determined
using a rank type panel. Fishcake samples of silver smelt, ling, cod, pollock and a commercial
sample were ranked for preference by a 12-member panel with the most preferred sample
ranked 1 and the least ranked 5. In test 4, paired comparison taste panels evaluated silver smelt
nuggets versus commercial cod nuggets (12 panellists) for crumb colour, flesh colour, flavour,
texture and overall preference. Test 5 embraced similar comparisons (22 panellists) of breaded
portions of silver smelt versus breaded commercial cod portions.

Fillet yield and flesh composition
Silver smelt had an average fillet yield (skin on fillets) of 42.0% of whole fish (Table 1). This is
quite high compared to the reference fish cod, which had a fillet yield of 34.8%.
The content of moisture, protein and fat was examined in the muscle of individuals of the
species silver smelt. Silver smelt had an average composition of 79.1% moisture, 18.7% protein
and 1.6% fat (Table 2). This is a favourable composition compared to cod and the other species
studied. The water content of silver smelt was lower and the protein content higher than what
was the case for cod. The flesh of silver smelt had the lowest water content (79.1%) in this
study. This is in accordance with previous studies (Bykov 1983), reporting a water content in
flesh from silver smelt from 75,4 to 81.1%. Bykov 1983 reported also a protein content of 17.2
to 18.5%, slightly less than what was found in our study, and a fat content between 0.4 and
4% due to seasonal variations.
The protein content is important when considering quality and texture of the fish muscle. Fish
muscles that contain small amounts of protein tend to loose much water upon cooking, which
Table 1. Fillet yield (% weight of fillets with skin on as a percentage of whole fish weight).
Species Fillet yield (%)
Silver smelta 42.0 + 7.3

Codb 34.8 + 3.4
an=10, 1 male
bn=5, all female
Table 2. Proximate composition of muscle fraction (g/100g muscle).
Species Water Protein Fat
Silver smelta 79.1 + 1.1 18.7 + 0.9 1.6 + 0.8

Codb 82.5 + 0.3 16.7 + 0.4 0.3 + 0.1
an=10
bn=5
binnenwerk.indd 445 21-08-2006 12:35:49

ruins the texture of the meat. The present study showed that, with respect to the protein
content (Table 2), silver smelt seems to have even higher protein content than cod.
Water holding capacity, cook loss and colour

The water holding capacity of silver smelt (Table 3) was high (78.8%), while flesh from cod
had a water holding capacity of 66.3%.
Flesh from silver smelt had a low cook loss (27.9%). In comparison, flesh from cod had a cook
loss of 42.0%. The results from this study indicate that increasing water holding capacity gives
a lower cook loss, which is in accordance with results obtained by Sarma and others (2000).
The flesh from silver smelt had colour parameters more similar to cod, but was slightly less
white and had a bit more of the yellow colour than cod, probably because of a bit higher protein
content and lower water content (Table 4).

In studies with silver smelt performed at the former Institute for Biochemistry and Technology
of the Federal Research Centre for Fisheries in the 70ties the main focus had been put on
processing properties of this species and evaluation of preparation methods for its fillets.
(Christians O 1972, 1976). These early studies have only been communicated in German
language and restricted until now.
In three batches the lengths and weights have been measured. The average length was found
to be between 39-40 cm with a corresponding average weight of 427- 514 g. Water and fat
content were determined by standard operation procedures used at the Centre. The average
Table 3. Water holding capacity (g remaining muscle weight after centrifugation per 100 g muscle) and cook
loss (g lost weight after heat treatment per 100 g muscle).
Species Water holding capacity Cook loss
Silver smelta 78.8 ± 5.9 27.9 ± 4.4

Codb 66.3 ± 4.0 42.0 ± 2.7
an=10
bn=5
Table 4. Colour of muscle (L-lightness, a- green/red, b-blue/yellow).
Species L* a* b*
Silver smelta 54.0 + 2.9 0.2 + 0.6 0.9 + 0.2

Codb 57.0 + 0.7 -1.5 + 0.5 -0.5 + 2.4
an=10
bn=5
*L=100:white, L=0: black; a=-60:green, a=60: red; b=-60: blue, b=60:yellow
binnenwerk.indd 446 21-08-2006 12:35:49

water content was 76.6% ranging from 74.6 to 78.8%. The average fat content was found to
be 0.4% (0.2- 1.2%).
Processing and sensory properties

Silver smelt was filleted by the filleting machine Baader 181. Due to the thin and fragile skin
the removal of the skin could not be achieved on raw fish neither manually nor mechanically.
A machine which would remove the skin after freezing of the surface could solve the problem.
After blanching in hot water the fish the skin can also be removed easily.
Some preparation methods to transfer the raw Silver smelt into products intended for human
consumption were also tested. Silver smelt was cooked as fillet or as whole gutted fish. The
fish was also fried or hot and cold smoked. Additionally the fresh fillets as well as the smoked
products were canned. Additional tests to prepare minced fish flesh of acceptable quality using
the bone-separator 694 of the Baader company have been carried out. A sensory evaluation after
cooking described that the white flesh of Silver smelt is of firm structure. When the fish is eaten
the taste is predominated by components also found in cucumber or potato-juice (very similar
to smelt (Osmerus eperlanus)). This taste, which is not pleasant for most consumers, however,
can be covered by baking or cooking the fish with spices. When the fish is fried it losses this
strange taste. The taste of the fillets is then similar to the fillets of plaice. The whole gutted
fish and fillets were hot smoked. In the whole fish the cucumber-taste could still be noticed.
The hot smoked fillets had no cucumber taste anymore, the flesh is juicy and well tasting. A cold
smoking system with external smoke supply and controlled air-humidity led to a product with a
good quality in colour, texture and taste. The fillets were treated 18 hours before smoking with
1% pickling salt followed by keeping them in an atmosphere of 60% rel. air humidity at 30 °C
for two to four hours. The weight loss during smoking was found to be 17%.
Fresh, hot and cold smoked samples were canned and offered well-tasting products. In
combination with different sauces based on tomatoes, dill, mustard, onions the taste could be
adjusted as appropriate. Hot and cold smoked fillets in vegetable oil showed a good texture.
Finally attempts were made to prepare minced fish flesh of acceptable quality. The quality of
this product was strongly influenced by the preparation of the fish prior to application of bone
separator. The intention that the headed, gutted and washed fish could be used as starting
material for this trial had to be given up, because the resulting minced fish flesh was of inferior
quality due to remaining blood, extruded spinal marrow and remaining pieces of the black
peritoneum. The remaining blood which was present in the vascular system near the main bone
had a considerable negative influence on sensory quality and colour which increased even with
time of frozen storage. Therefore attempts had been made to remove blood spots by intensive
washing prior to processing, however, was not successful.
Other experiments were performed with raw material which had been frozen before being
separated. After heading and gutting the silver smelt was treated by the bone-separator. The
minced fish flesh was washed in a washing drum with small drum holes. During this procedure
the minced fish flesh took up considerable amounts of water, while beneficial components were
washed out by the washing water. The complete removal of the coagulated blood could not be
achieved completely. Due to the fact that flavour components were washed out also, the farce
was characterised by a watery tasteless flavour. These experiments were repeated with fresh
material but the complete washing out of remaining blood could not be achieved satisfactory.
To avoid the extrusion of the spinal marrow the pressure of the belt of the drum of the bone
binnenwerk.indd 447 21-08-2006 12:35:49

separator was reduced. But these technical changes had not the desired effect. The results could
be improved by application of the procedure to fillets. But under these conditions only the skin
was separated. The application to fillets results in a decreased yield (-10%).
Selenium content
In the samples of silver smelt obtained from Centre 2 an average selenium content of 0.20 mg
Se/kg wet weight (range: 0.19- 0.24 mg Se/kg) could be determined. This is somewhat lower
since the average content in edible part of marine fish species from the North-Eastern Atlantic
amounts to 0.35 mg Se/kg wet weight (Oehlenschläger 1995).
Proximate composition
The results of the composition of the two samplings during two cruises of FRV “Walther Herwig
II” are given in Table 5. The water content in both samples (81.2 and 80.3%) is higher than
that in the samples investigated by Centre 1 (79.1%). The raw protein content of the fish
caught on Lousy Bank is also somewhat lower (17.7%) but that of the fish from Porcupine Bank
which amounts to 18.7% has exactly the content measured in Centre 1 (18.7%). Concerning fat
content both fish samples from the Outer Banks are lean (0.57% and 0.78%) compared with the
fish caught in shallow Norwegian waters (1.6%). The fish from Lousy Bank was caught in late
spring while the fish from Porcupine Bank was caught in mid of October. Fish from Porcupine
Bank showed a higher pH in muscle, contained more raw protein and more fat but less moisture
than fish from Lousy Bank. This can be explained by the different seasons in which the fish was
caught. While fish in spring is often still recovering from the winter, fish caught in October is
very often in a good condition in respects of protein and fat content after feeding throughout
the summer period. The mineral content and the phosphorus content was almost the same.
Also the TVB-N content which were taken to know the basic values of this often used spoilage
indicator were similar on both sampling places (17.4 and 16.8 mg/100 g) and resemble the
TVB-N found at Centre 3 on day 1 (16.5 mg/100 g). The biggest difference was found in the
chloride content (0.14 versus 0.28 mg/100 g), in TMAO content (118 versus 147 mg/100 g and
ammonia content (9.8 versus 13.1 mg/100 g). This, however, can be explained by the different
depths in which the fish have been caught. The deeper the fish live the higher is the need of
osmotic regulation in the organs including muscle. TMAO is well known in fishes to function
as a regulator for osmotic pressure. And it seems that chloride and ammonia content play a
similar role.
Silver smelt from the Outer Banks of the West British waters is generally characterised by a
high protein content, a very low fat content and a high moisture content. Its mineral content
is as high as in other lean fish species and the phosphorus content is with 200 mg/ 100 g wet
weight typical for a lean fish species.
Heavy metals
The cadmium content (Table 6) was found to be extremely low in the edible part of silver smelt.
The cadmium content usually found in food fishes from the North East Atlantic varies between
1-5 µg/kg. The cadmium burden in silver smelt from this location is extremely low (average
of 0.75 µg Cd/kg). The lead content found is somewhat higher compared to other lean fish
species from the North East Atlantic varying between 2 – 6 µg Pb/kg, however, it can not be
considered as being extremely high and is similar to what was found in the small spotted cat
shark (15 mg Pb/kg) (Scyliorhinus caniculus). The zinc content measured (2.6 mg Zn/kg) is
somewhat lower than the zinc level in other species (about 3-4 mg Zn/kg) but is as high as
binnenwerk.indd 448 21-08-2006 12:35:50

Table 5. Results of the composition of silver smelt of two samplings during two cruises of FRV “Walther
Herwig II”.
Argentina silus, FRV Walther Herwig II, 08.05.1983, Lousy Bank, 602 m depth
chloride (NaCl)
Phosphorpus
(mg/100 g)
(mg/100 g)
(mg/100 g)
(mg/100 g)
(mg/100 g)
raw protein
Ammonia
moisture
minerals
Length
Weight
TVB-N
TMAO
(c)m
(%)
(%)
(%)
(%)
(g)
pH
fat
1 500 42 6.20 17.49 0.62 1.19 81.72 0.16 0.20 15.4 130 8.7
2 700 46 6.61 17.49 0.51 1.20 81.54 0.13 0.21 16.7 112 10.0
3 700 45 6.15 17.48 0.54 1.24 81.31 0.11 0.20 19.2 105 10.2
4 600 44 6.14 17.57 0.62 1.25 81.05 0.13 0.17 18.5 119 10.8
5 600 44 6.78 18.28 0.56 1.29 80.16 0.16 0.20 17.4 121 9.2

Mean 620 44.2 6.38 17.66 0.57 1.23 81.16 0.14 0.20 17.4 118 9.8
SD 84 1.5 0.30 0.35 0.05 0.04 0.61 0.02 0.02 1.5 9.5 0.8
Argentina silus, FRV Walther Herwig II, 17.10.1981, Porcupine Bank, 1000 m depth
chloride (NaCl)
Phosphorous
(mg/100 g)
(mg/100 g)
(mg/100 g)
(mg/100 g)
(mg/100 g)
raw protein
Ammonia
moisture
minerals
Length
Weight
TVB-N
TMAO
(c)m
(%)
(%)
(%)
(%)
(g)
pH
fat

1 330 36 6.70 17.97 0.83 1.29 79.8 0.34 0.19 16.8 162 13.3
2 370 38 6.76 18.37 0.57 1.31 80.09 0.19 16.8 163 12.7
3 410 38 6.74 19.56 0.61 1.25 80.04 0.22 0.17 16.8 127 11.4
4 380 38 6.70 18.13 1.13 1.24 81.37 0.19 16.8 136 9.8
5 290 34 6.78 19.44 0.74 1.24 79.94 0.20 16.8 148 13.1

mean 356 36.8 6.74 18.69 0.78 1.27 80.25 0.28 0.19 16.8 147 12.1
SD 47 1.8 0.04 0.75 0.22 0.03 0.64 0.08 0.01 16 1.5
measured e.g. in haddock (2.6 - 2.9 mg Zn/kg) and the copper content (0.1 mg Cu/kg) is quite
low but still comparable with the concentration found in other lean fish species (0.1 to 0.8 mg
Cu/kg) (Celik and Oehlenschläger, 2004).
Centre 3 (Ashtown Food Research Centre, IE)

Flesh composition, colour, odour and flavour
Composition was measured on raw silver smelt with and without skin. Ranges in skin-on samples
were 78.2-79.0% (moisture), 17.3-17.7% (protein) and 2.0-2.1% (fat). Corresponding ranges in
samples without skin were 79.9-81.4% (moisture), 17.3-19.1% (protein) and 0.3-0.8% (fat). The
most salient feature of these data was the higher fat content of the skin-on portions. Hunter L
binnenwerk.indd 449 21-08-2006 12:35:50

Table 6. Trace elements in edible part of Argentina silus caught West off Shetland Islands on a wet weight
basis.
Cadmium (µg/kg) Lead (µg/kg) Zinc (mg/kg) Copper (mg/kg)
1 1 14 2.4 0.04
2 Not determined 25 2.6 0.09
3 1 17 2.5 0.12
4 0.5 18 2.9 0.13
5 0.5 5 2.7 0.11
mean 0.75 15.8 2.62 0.10

SD 0.29 7.3 0.19 0.04
(whiteness) values for silver smelt were 61.0 (raw) and 76.1 (steamed) and for cod 65.1 (raw)
and 81.4 (steamed). The L/b (white/yellow) ratios were 26.5 (raw) and 8.0 (steamed) for
silver smelt and 12.3 and 9.7 for cod. These values show that silver smelt whiteness compared
favourably with that of cod. The L values for silver smelt were in the middle of those reported by
Fagan and others (2006) for other underutilised species, i.e. raw roundnose grenadier (L = 67),
faux siki (67), cardinal fish (55) and blue whiting (51). Sensory tests (test 1) on steamed samples
indicated that silver smelt odour ratings were high and remained relatively constant over days 3,
4, 5 and 6 post-catching with scores of 8.7, 8.1, 9.0 and 8.0 ( 10 = best; 0 = worst), respectively.
Corresponding scores for flavour were 7.5, 7.3, 8.0 and 8.0. Paired comparison tests of steamed
silver smelt versus steamed cod (test 2) showed a preference ratio of 9/3 (12 panellists) in
favour of silver smelt. Most panellists agreed that silver smelt had a pleasant bland flavour and
a firmer texture than the sample of cod. It is stressed that the cod sample was purchased in a
supermarket and had an unknown age/storage history.
TVB-N, TVC and centrifugal drip

The TVB-N values (Table 7) were well below the upper limit of 35 mg/100 g flesh (EC Directive
95/149/EC) with the exception of the day 6 values of 28.7 (catch 1; whole fish) and 34.7
mg/100 g (catch 1; fillet). The overall levels are similar to those reported for chilled whiting,
mackerel and farmed salmon fillets by Fagan and others (2003) and suggest that silver smelt
has a good shelf life at 2-4 °C. The TVCs showed a similar pattern (Table 7) to TVB-N. TVCs
for silver smelt fillets were above the upper limit of log10 5 cfu/g (proposed by the Irish
Department of Health and Children Guidelines 1992) on day 5 (for whole fish, catch 1) and on
days 3-5 for the fillets. The data (Table 7) suggest that catch 2 was more hygienic than catch
1. Centrifugal drip (CD) values varied widely both within and between species (Table 8). Fresh
silver smelt fillets had a similar CD to cod but was higher than that found for fresh pollock and
ling. The silver smelt values were also higher than those for chilled or frozen whiting, mackerel
and farmed salmon (Fagan and others 2003). Freezing caused an increase in CD for the four
species and values increased by 71 (silver smelt), 118 (cod), 132 (pollock) and 205% (ling)
[(frozen sample CD – fresh sample CD ÷ fresh sample CD) x 100]. Thawing a frozen fillet and
immediately refreezing it did not influence CD values but storing frozen fillets for 6 months gave
a large increase in CD (Table 8). This agrees with the findings of Gormley and others (1993)
who showed that water binding properties of silver smelt decreased steadily during frozen
storage for 150, 235 and 374 days. The changes during frozen storage were caused mainly by
binnenwerk.indd 450 21-08-2006 12:35:51

Table 7. Total volatile base nitrogen (TVB-N) (mg/100g) and total viable count (TVC) (log 10 cfu/g) values
for fresh silver smelt whole fish and fresh fillets.
Catch 1 Catch 2
Day Whole fish Fillet Whole fish
TVBN
1 --- --- 16.5
2 --- --- 17.1
3 14.6 16.4 19.2
4 23.4 16.7 18.9
5 17.1 21.7 ---
6 28.7 34.7 ---
TVC
1 --- --- 4.47
2 --- --- 4.29
3 3.13 6.10 3.88
4 4.42 5.99 3.95
5 6.08 6.29 ---
alterations in fish myofibrillar proteins during frozen storage as a result of the formation of
intermolecular cross-linkages, and consequently the aggregation and denaturation of actomysin
(Buttkus 1970). The CD of mince prepared from frozen fillets was similar to that from the fillets
(Table 8). However, frozen mince had a much higher CD value indicating that mince should not
be stored frozen but as whole fillets that are minced post-thawing. Anese and Gormley (1996)
showed the importance of cryoprotection when freezing fish minces.
Properties of silver smelt gels

The strength of fish gels is a good index of water binding and is important for the textural
properties of fish products (Gormley and others 1994). Full strength silver smelt gels had an
intermediate compression value relative to the other species (Table 9) being softer than ling or
cod but firmer than pollock gels. However, silver smelt gel was the softest of the added water
gels. The penetrometer values generally mirrored the compression data. It is noted that these
gels were made from frozen mince, and silver smelt mince is particularly prone to freezing
damage. However, a follow-on test indicated gel compression values of 92.5, 90.2 and 59.2 for
silver smelt gels made from fresh (i.e. not previously frozen) mince, mince from frozen fillets,
and frozen mince, respectively. These data show that the first or second options were best.
Acceptability of silver smelt products

Fishcakes from silver smelt were preferred to those made from pollock, ling or cod and were
much preferred to the commercial sample (Table 10). Comments by panellists for the silver
smelt samples were favourable and included ‘firm and fishy’, ‘good flavour/texture’, ‘very good
sample’. Two panellists commented that the silver smelt sample was slightly tough or chewy.
Cakes from cod also received good comments but a number of panellists cited ‘lacks fish’ as a
negative aspect. Pollock cakes got comments such as ‘good flavour’ but also negatives such
as ‘lacks fish’, ‘soft texture’, oily. Comments for ling fishcakes included ’lacks flavour’and ‘bad
mouthfeel’. The main comments for the commercial fishcakes were ‘lacks fish’, ‘bad mouthfeel’
binnenwerk.indd 451 21-08-2006 12:35:51

Table 8. Centrifugal drip loss (500 G/10 min) from fish fillets/pieces and mince.
Sample Centrifugal drip (%)
Silver smelt (catch 2; day 1; fresh fillet) 12.2

Silver smelt (catch 1; frozen fillet) 20.7
Silver smelt (catch 1; frozen fillet; tested 6 months later) 36.1
Silver smelt (catch 1; refrozen fillet) 20.4
Fresh pollock filleta 8.2
Frozen pollock filletb 19.0
Fresh ling filleta 6.5
Frozen ling filletb 19.8
Fresh cod filleta 11.3
Frozen cod filletb 24.6
Silver smelt (mince from frozen fillets) 20.0
Silver smelt (frozen mince) 38.0
a Freshfillet obtained from Dublin fish market

b Freshfillet obtained from Dublin fish market and blast frozen (-35 °C/2 hours) followed by storage
(1 week/-18 °C)
Table 9. Compression, penetrometer and Hunter colour valuesa for full strength and added water fish gelsb
made from fish which was minced prior to freezing.
Fish type Compression (N)c Penetrometer (N)c Hunter (L/b)a
Full strength gels

Silver smelt 59.2 9.7 9.9
Ling 85.8 15.0 7.4
Cod 67.3 10.3 9.0
Pollock 49.4 7.0 7.4
F-test p< 0.001 p< 0.001 p< 0.001

LSD 6.91 1.74 0.37
Added water gelsb

Silver smelt 5.2 1.03 11.2
Ling 15.4 2.85 7.9
Cod 12.1 2.60 8.6
Pollock 11.0 1.95 7.0
F-test p< 0.001 p< 0.001 p< 0.001

LSD 1.21 0.49 1.03
a Hunter whiteness/yellowness ratio

b2 parts fish mince + 1 part water
c Newtons force
binnenwerk.indd 452 21-08-2006 12:35:51

Table 10. Taste panel rank sumsa for fishcake preference.
Fishcake Rank sums a
Silver smelt 20 (1st) (p< 0.001)

Ling 41 (4th)
Cod 31 (2nd)
Pollock 36 (3rd)
Commercial sample 52 (5th) (p< 0.001)
a 12panellists; low rank sums = preferred samples; ranges for statistical significance = 25-47 (p< 0.05)
and 22-50 (p< 0.01)
and poor flavour. Comparisons of silver smelt nuggets and portions with commercial samples
of cod nuggets and portions are presented in Table 11. The commercial cod portions were in
a ‘waferlight’ batter. With the exception of crumb colour, silver smelt nuggets were preferred
on all counts to the commercial cod nuggets. The preferences were statistically significant
except for texture (Table 11). Some panellists found the texture of the silver smelt nuggets
to be rough, rubbery or too firm. However, they still preferred the silver smelt nugget texture.
In contrast to the nuggets, the silver smelt portions were preferred only for colour (Table
11) while the commercial portions were preferred for flavour, texture and overall preference.
The panel commented that the silver smelt portions lacked flavour and were too rubbery. The
reversal in preference for silver smelt nuggets and portions in comparison with their commercial
cod counterparts was unexpected. Possible explanations could be that the commercial nugget
samples were of very low quality and the commercial breaded portions of high quality and/or
that silver smelt nuggets were a ‘formed’ product whereas the portions were intact fillet tissue.
This may have had a bearing on the texture of the products in that toughening may not have
been so apparent in the ‘formed’ nuggets as in the intact tissue of the portions. These data
suggest that silver smelt has potential for products but that toughening during frozen storage
could be a potential problem. These findings agree with those of Mackie and Hardy (1969) who
reported good acceptability of silver smelt by taste panels but stressed that rancidity could
occur during long term storage because of its unsaturated oil content.
Table 11. Paired comparison taste panel preference ratios for silver smelt (SSN) versus commercial brand
cod nuggets (CCN) (panel 1)a and for breaded silver smelt (SSP) versus commercial brand cod portions
(CCP) (panel 2)b.
Parameter Panel 1 (SSN vs CCN)a Panel 2 (SSP vs CCP)b
Crumb colour 2/13 (p< 0.01) 22/2 (p< 0.001)

Flesh colour 13/2 (p< 0.01) 19/5 (p< 0.01)
Flavour 14/1 (p< 0.01) 5/19 (p< 0.01)
Texture 11/4 (NS)c 4/20 (p< 0.001)
Overall preference 13/2 (p< 0.001) 8/16 (NS)c
a 15 panellists
b 22 panellists
c not significant
binnenwerk.indd 453 21-08-2006 12:35:51

Overall outcomes from Centres 1-3
Outcomes from sensory tests on steamed fillet, fishcakes and breaded nuggets from silver smelt
were positive. Full strength silver smelt gels had good compression and penetrometer values
indicating good water binding properties. However, these gels were softer than those from ling
or cod but were firmer than pollock gels. Freezing increased centrifugal drip (CD) values in silver
smelt fillets. However, frozen mince from silver smelt had a much higher CD value indicating
that mince should not be stored frozen but as whole fillets that are minced post-thawing. These
data, overall suggest that silver smelt has potential for products but that toughening during
frozen storage could be a potential problem.
Considering the technological properties and the burden of the heavy metals cadmium and
lead silver smelt seems to be suitable as high quality raw material for the consumer driven
development of new tailor-made seafood products. In comparison with other marine fish species
the content of the essential elements zinc and copper in the edible part of silver smelt seems
to be rather low. Depending on depth of capture and provenance differences in chloride and
TMAO levels could be detected.
Conclusions
The results obtained in the three participating research centres show the silver smelt or greater
argentine (Argentina silus) is a valuable food fish with good sensory, compositional and physical
properties. It is low in inorganic contaminants. Due to these favourable characteristics it is
recommended that this species should be used much more for human consumption than for
production of fish meal (feed).
Acknowledgements
The study at Møre Research has partly been financed by the EU-Commission and the Norwegian
Research Council. We also want to thank The Norwegian longliners M/S Torita and M/S Sætring,
Eilert Hermansen (Institute of Marine Research) and Jan Erich Rønneberg (Møre Research)
for collecting samples to this study. Wenche E. Larssen (Møre Research) is acknowledged for
preparing samples and performing analytical laboratory work. Dr. Marianne Synnes and Ann
Helen Hellevik are gratefully acknowledged for a skilful and significant contribution in the
project, both with scientific work and as co-authors of reports. We also want to thank all the
partners in the SEAFOODplus-project for valuable information and a nice collaboration.
Thanks are due to the EU for part-funding aspects of the work at AFRC (Contract UP. 1. 299)
and to Patrick Ward and Joseph Somers [formerly of the Irish Sea Fisheries Board (BIM)] who
conducted some of the trials. This combined output from three research centres on silver smelt
was coordinated under the ongoing integrated research project SEAFOODplus.
The valuable input and the contributions of Dr. Otto Christians, Dr. Sabine Mierke-Klemeyer and
Hans-Jürgen Knaack (all affiliated with Centre 2) are gratefully acknowledged.
This work was partially performed within the Integrated Research Project SEAFOODplus, contract
No FOOD-CT-2004-506359. The partial financing of the work by the European Union is gratefully
acknowledged.
binnenwerk.indd 454 21-08-2006 12:35:52

References
Anese M, Gormley TR. 1996. Effect of dairy ingredients on some chemical, physico-chemical and functional properties
of minced fish during freezing and frozen storage. LWT-Food Sci Technol 29:151-157.
Bergstad OA, Wik AD, Hildre O. 2003. Predator-prey relationships and food sources of the Skagerrak deep-water fish
assemblage. J Northwest Atlantic Fish Sci 31:165-180.
Børresen T. 1980. New methods for determination of liquid holding capacity, Salt water holding capacity and cook loss
in fish muscle (in Norwegian). FTFI Report 663, Tromsø, Norway.
Brennan M, Gormley TR. 1999. The quality of underutilised deep-water species. Research Report, vol. 22. The National
Food Centre, Dublin.
Buttkus H. 1970. Accelerated denaturation of myosin in frozen solution. J Food Sci 32:558-562.
Bykov VP. 1983. Marine fishes: chemical composition and processing properties. Amerind Publishing Co. Pvt. Ltd., New
Dehli. 333p. (FishBase).
Celik U, Caklı S, Oehlenschläger J. 2004. Determination of the lead and cadmium burden in some North-eastern Atlantic
and Mediterranean fish species by DPSAV. Eur Food Res Technol. 218:298-305
Celik U, Oehlenschläger J. 2004. Determination of zinc and copper in fish samples collected from Northeast Atlantic
by DPSAV. Food Chem. 87:343-347
Christians O. 1972. Neue Konsumfische? Inf Fischw. 19:20-23
Christians O. 1976. Neue Nutzfische, Research Report (M 76-03) Meeresforschung , (Research Projekt MF 95 financed
by Bundesministerium für Forschung und Technologie), ZLDI/DFVLR, München
Department of Health and Children (Ireland) Guidelines. 1992. Microbiological Guidelines for fish. Total viable counts
for raw and cooked seafood products.
EC Directive 95/149/EC: Commission decision of 8 March 1995 fixing the total volatile base nitrogen (TVB-N) limit
values for certain categories of fishery products and specifying the analysis methods to be used. Official J L 097,
29/04/1995. 0084-0087
Fagan JD, Gormley TR, Uí Mhuircheartaigh M. 2003. Effect of freeze-chilling, in comparison with fresh, chilling and
freezing, on some quality parameters of raw whiting, mackerel and salmon portions. LWT-Food Sci Technol 36:647-
655.
Gormley TR, Ward P, Somers J. 1991. Silver smelt: a valued non quota species. Farm and Food 1(4):8-9.
Gormley TR, Ward P, Somers J. 1993. A note on the effect of long-term frozen storage on some quality parameters of
silver smelt (Argentinus silus). Irish J Agric Food Res 32:201-204.
Gormley TR, Connolly PL, Ward P. 1994. Evaluation of deep water fish species. Farm and Food 4(1):8-11.
Johannessen A, Monstad T. 2003. Distribution, growth and exploitation of greater silver smelt (Argentina silus) in
Norwegian waters. J Northwest Atlantic Fish Sci 31:319-332.
Losnegard N, Bøe B, Larsen T. 1979. Investigations of extraction buffers in determination of fat content. The Directorate
of Fisheries, Norway. Report No. 1, 12pp.
Mackie PR, Hardy R. 1969. A lipid analysis of greater silver smelt [Argentina silus (Ascanius)] and an evaluation of its
potential for food and fish meal production. J Food Technol 4(3):241-253.
Magnussen JV. 1996. Greater silver smelt, Argentinus silus in Icelandic waters. J Fish Biol 49(supplement A):259-275.
Malle P, Tao SH. 1986. Rapid quantitative determination of trimethylamine using steam distillation. J Food Protec
50:756-760.
Monstad T, Johannessen A. 2003. Acoustic recordings of greater silver smelt (Argentina silus) in Norwegian waters and
west of the British Isles, 1989-94. J Northwest Atlantic Fish Sci 31:339-351.
Oehlenschläger J. 1991. Chemical composition of the flesh and other tissues of Antarctic fish species of the families
Channichthyidae and Nototheniidae. Food Chem 40:159-167
Oehlenschläger J.1995. Selengehalte im Seefischmuskel. in: Ernährungsphysiologische Eigenschaften von
Lebensmitteln.
Angewandte Wissenschaft, Landwirtschaftsverlag Münster, 445:86-91
binnenwerk.indd 455 21-08-2006 12:35:52

Økland HMW, Stoknes IS, Remme J, Kjerstad M, Synnes M. 2004. Proximate composition, fatty acid and lipid class
composition of the muscle from deep-sea teleosts and elasmobranchs. Comp Biochem Physiol B 140:437-443.
Ritzman SE, Daniels JC. 1975. Introduction. In: Ritzman SE, Daniels JC (Eds.), Serum Protein Abnormalities: Diagnostic
and Clinical Aspects. Little, Brown and Company, Boston.
Sarma J, Reddy GVS, Srikar LN. 2000. Effect of frozen storage on lipids and functional properties of proteins of dressed
Indian oil sardine (Sardinella longiceps). Food Res Int 3(10):815-820.
Synnes M, Stoknes IS. 2004. Underutilised fish species. A selection of underutilised fish species as possible raw material
for product development in the project Consumerproducts which is part of the EU-project SEAFOODplus. Report
Å0419. Møre Research, Aalesund, Norway.
Wierbiki E, Deatherage FE. 1958. Determination of water holding capacity of fresh meats. J Agric Food Chem 6:387-392.
binnenwerk.indd 456 21-08-2006 12:35:52

Chapter 7:
Nutrients and contaminants in seafood
Fish and other seafood have a head with two faces, one positive and one negative. The positive
face are the beneficial nutrients and nutritious components encountered in seafood as highly
unsaturated long chain fatty acids (n-3 series), the essential elements iodine and selenium,
the vitamins A, D and E, the high concentration of taurine, the very favourable amino acid
composition, the good digestibility of the fish protein due to low connective tissue content etc.
The negative face are the undesirable inorganic and organic substances which can - if present
in elevated amounts in the aqueous environment - be accumulated in fish (bones, liver, kidney
and muscle). These both groups of substances can be of natural and anthropogenic origin.
Independent of the origin a too high concentration in seafood products intended for human
consumption can have an adverse effect on human health if ingested for long time periods.
Therefore a continuous monitoring of the concentration of these undesirable substances is a
prerequisite for a successful risk assessment and science based legislation. Sound knowledge
about the content of the beneficial components is essential for reliable food composition tables
which are used by dieticians, medical doctors, cooks, nutritionists and others.
In the section of this chapter covering the positive aspects macro and micro elements in bivalve
molluscs, the proximate and fatty acid composition of farmed and wild brown meagre (Sciena
umbra) and a large number of nutrients of fishery products typical for the Portuguese market
are reported.
In the section covering the negative aspects a compilation of the concentrations of

polychlorinated biphenyl (PCBs) and organochlorine pesticides in brown shrimp (Crangon
crangon) from the Belgian continental shelf is presented. Two other contributions describe the
situation in respect of heavy metals in bivalves from the Portuguese coasts and contaminant
heavy metals in cod products offered on the Portuguese market.
binnenwerk.indd 457 21-08-2006 12:35:52

binnenwerk.indd 458 21-08-2006 12:35:52
Effect of sample preparation with or without shell liquor on
contents and retentions of macro and micro elements in three
species of bivalve molluscs
Anna Badiani1, Silvia Testi1, Sergio Ghidini2, Mavina Silvi3, Chiara Foschi3 and Pier Paolo Gatta1
1Dipartimento di Morfofisiologia Veterinaria e Produzioni Animali, Alma Mater Studiorum –
Università di Bologna, Via Tolara di Sopra 50, 40064 Ozzano Emilia (BO), Italy
2Dipartimento di Produzioni Animali, Biotecnologie Veterinarie, Qualità e Sicurezza degli Alimenti,
Università di Parma, Via del Taglio 8, 43100 Parma, Italy
3Laboratorio di Acquacoltura, Polo Scientifico-Didattico di Cesena, Alma Mater Studiorum –
Università di Bologna, Viale Vespucci 2, 47042 Cesenatico (FC), Italy
Abstract
The macro and micro element contents of soft tissues from Tapes philippinarum, Callista chione,
and Mytilus galloprovincialis, either with (CL) or without (SL) shell liquor, were investigated
both at the raw and cooked state, to gain information on true (TR) and apparent (AR) retention
factors of the mineral component of these bivalve molluscs. On a per-serving basis, SL molluscs
were more significant sources of several minerals than their CL counterparts. TR proved to be a
more reliable coefficient than AR in the estimation of nutrient composition of cooked molluscs
based on data from the raw ones, probably due to the importance of the cooking yield for this
kind of food.
Keywords: bivalve molluscs, shell liquor, minerals, cooking, retentions
Introduction
Bivalve molluscs, a valuable source of minerals (Reilly 2002), are hugely popular in Mediterranean
countries, where they are usually eaten raw or only lightly cooked, to prevent meat toughening.
Shell liquor is customarily consumed with meat, when eating raw molluscs. Furthermore,
Mediterranean cuisine makes large use of shell liquor in shellfish-based recipes to add natural
taste and flavour, and possibly valuable nutrients leached from bivalves upon cooking (Davidson
2002). This notwithstanding, information is lacking regarding macro and micro element contents
of both raw and cooked bivalve tissues eaten with shell liquor as compared with the elemental
contents of the same tissues consumed without liquor, that is the standard matrix considered
in databases and food composition compilations.
True nutrient retention factors, first proposed by Murphy and others (1975), definitely proved to
be useful tools in calculating nutrient composition of cooked products based on data from raw
ones, which is a great asset when in need of making frequent updates of food composition tables,
as essential as they are expensive (Haytowitz and others 1996; Schakel and others 1997). Actually
the United States Department of Agriculture makes frequent use of true retention factors and
cooking yields to calculate nutrient data for cooked products from comparable raw items (USDA
2005a). In order to establish clear relationships between the composition of raw and cooked
shellfish, Murphy and others (1975) suggested to carefully draw sub samples for cooking and
binnenwerk.indd 459 21-08-2006 12:35:52

Anna Badiani, Silvia Testi, Sergio Ghidini, Mavina Silvi, Chiara Foschi and Pier Paolo Gatta
ensuing analyses from well-mixed lots of raw food, to analyse similarly chosen sub samples at the
raw state and to weigh the product before and after cooking to determine its cooking yield.
Although Murphy’s suggestion seems easy and sensible, to our knowledge it has still to be
implemented on molluscs. The same applies to apparent nutrient retention factors, where nutrient
contents of raw and cooked sub samples have both to be expressed on the moisture-free basis.
Apparent retentions, which are based on the assumption that solids are not lost to any practical
extent with cooking, offer the obvious advantage to circumvent difficulties associated with
obtaining weights. Working with animal foods, where both solids and moisture losses are expected
upon cooking, Murphy and others (1975) found that for protein, fat, ash, vitamins and minerals
apparent retentions were significantly higher than the corresponding true values whenever the
food was composed of more than one type of tissue, in this case apparent retentions being
not reliable measures of the true ones. Whether and to what extent true retentions of minerals
approximate their apparent counterparts in molluscs is presently unknown.
Therefore, the aims of this study were: to determine the content of several minerals in three
species of bivalve molluscs widely consumed in Italy, analysed both raw and cooked, and with
or without shell liquor; to devise and implement a simple method by which true retention
factors could be acquired for minerals in bivalve molluscs, either with or without shell liquor;
to compare true and apparent retention factors determined for each mineral in the presence
or absence of shell liquor.
Mollusc samples
Ten lots each (around 1 kg weight/lot) of freshly harvested Manila clam (Tapes philippinarum,
Adams & Reeve, 1850; TP in the following; 233 – 240 specimens/kg), smooth venus (Callista
chione, L., 1758; CC; 20 – 27 specimens/kg), and Mediterranean mussel (Mytilus galloprovincialis,
Lamarck, 1819; MG; 54 - 59 specimens/kg) were obtained in pairs during late spring from
a certified purification centre based in Cesenatico (Upper-middle Adriatic Sea). Following a
traditional Italian practice (Slow Food 2004), in order to make molluscs disgorge the imbedded
sand each lot was soaked for 4 hours in potable municipal pipe-borne water (8 L/kg molluscs)
to which coarse sea salt had been added (30 g salt/L water). Each lot of molluscs was then
halved on a weight basis: one half was shucked by hand, their soft tissues being pooled and
analysed in the raw state, with (CL, 5 lots) or without (SL, 5 lots) the strained shell liquor. The
other half was arranged in a single layer in a teflon-coated pan and sautéed on a hot plate
without any added cooking fat. Cooking was discontinued as soon as the molluscs opened. At
that moment cooking time was recorded and the final temperatures reached by the soft tissues
of three randomly selected specimens per lot were quickly checked with an iron-constantan
(type J) wire thermocouple connected to a digital thermometer (Extech Instruments, Waltham,
Mass.) and averaged. Within each lot, the soft tissues from cooked molluscs were pooled and
analysed either with (CL, 5 lots) or without (SL, 5 lots) the strained shell liquor.
Sample analyses
Moisture and ash were determined in duplicate following the AOAC (2000) methodologies (Sec
950.46B and 920.153, respectively). After freeze-drying 0.5 g of each homogenised sample
were mineralised in a microwave system (MLS-1200 MEGA Milestone, FKV, Sorisole, Italy) using
3 mL of nitric acid 65% (Romil Ltd, Cambridge, UK) and 0.5 mL hydrogen peroxide (Merck,
binnenwerk.indd 460 21-08-2006 12:35:52

Effect of sample preparation on macro and micro elements in three species of bivalve molluscs
Milano, Italy). After cooling to room temperature, the acid digests were made up to 50 mL in
volumetric flasks by adding bi-distilled water. The macro (sodium, Na; potassium, K; calcium,
Ca; magnesium, Mg; phosphorus, P; sulphur, S) and micro (iron, Fe; copper, Cu; zinc, Zn;
nickel, Ni; cobalt, Co) element contents of the mineralised samples were quantified using a
Jobin Yvon Ultima 2 Inductively Coupled Plasma Atomic Emission Spectrophotometer (ICP-
AES), equipped with a Jobin Yvon AS 421 auto sampler (HORIBA Jobin Yvon Srl, Opera, Italy),
adopting operating conditions, emission wavelength lines, and detection limits (Ghidini and
others 2005) as listed in Table 1.
The quality of the analytical results was controlled by analysing the standard reference material
“Mussel Tissue” (SRM 2976, National Institute of Standards & Technology, Gaithersburg, Md.)
for relevant macro and micro elements. For each element, duplicate determinations were carried
out five times during the project, following the analytical procedures used in this work. The
mean values observed were always within the certified (or reference) intervals.
Table 1. ICP-AES operating conditions, emission wavelenghts (nm) and detection limits (mg/kg wet
tissue).
Operating conditions
Radio frequency power (W) 1100
Gas flows (L/min)
plasma 12.0
auxiliary 0
sheat 0.2
nebuliser 0.73
Nebulisation pressure (bar) 2.9
Nebuliser type concentric
Spray chamber type cyclonic
Peristaltic pump speed (mL/min) 1
Rinse time (s) 15
Transfer time (s) 45
Stabilisation time (s) 15
Number of replicates 3
Emission wavelengths (EW) and detection limits (DL) for: EW DL

Ca 393.366 0.006
Co 228.616 0.005
Cu 324.754 0.003
Fe 259.940 0.004
K 766.490 0.057
Mg 279.553 0.008
Na 589.592 0.190
Ni 231.604 0.005
P 213.618 0.155
S 180.676 0.304
Zn 213.856 0.004
binnenwerk.indd 461 21-08-2006 12:35:53

Cooking yields and nutrient retentions

Within each lot, the percentage cooking yield should ideally be determined by dividing the
weight of soft tissues with (CL) or without (SL) shell liquor from the cooked molluscs by the
weight of soft tissues with (CL) or without (SL) shell liquor from the same molluscs at the raw
state, multiplied by 100. Being unfeasible to weigh the raw soft tissues (either CL or SL) from
the same molluscs destined for cooking, the denominator of the ratio was obtained as follows:
• as to the CL treatment, by subtracting the shell weight (as ascertained after cooking) from
the weight of the molluscs to be cooked;
• as to the SL treatment, by calculating the percentage of soft tissues in the molluscs to be
analysed at the raw state and applying that percentage to the weight of the molluscs to
be cooked.
The true retention factor (TR) was calculated for each nutrient using the following formula
(USDA 2005a):
TR = (Nc / Nr) * CY
where: Nc = nutrient content per g of cooked mollusc (either CL or SL);

Nr = nutrient content per g of raw mollusc (either CL or SL);
CY = % cooking yield.
The apparent retention factor (AR) was calculated for each nutrient using the formula first
proposed by Murphy and others (1975):
AR = (Ncdb / Nrdb) × 100
where: Ncdb = nutrient content per g of cooked mollusc (either CL or SL), on a dry basis;
Nrdb = nutrient content per g of raw mollusc (either CL or SL), on a dry basis.
Statistical analyses
One-way analysis of variance with repeated measures was used to determine the effect of the
treatment (presence/absence of shell liquor) on the ash and mineral contents of either raw or
cooked molluscs. Two-way analysis of variance was used to test the significance of the effect
of species and treatment on the true retention factors obtained for ash and minerals. Scheffé
post hoc test was used to separate means when a significant F-value was obtained (p ≤ 0.05).
Within species and treatment, the Student’s t-test for dependent samples was used to detect
statistically significant differences (p ≤ 0.05) between the true and apparent retention factors
calculated for the same nutrient. All statistical computation was performed using Statistica AX
software package (version 7.1, 2005; StatSoft Italia srl, Vigonza, Italy).
Nomenclature
AR apparent retention factors
CC smooth venus (Callista chione)
CL mollusc with shell liquor
MG Mediterranean mussel (Mytilus galloprovincialis)
SL mollusc without shell liquor
TP Manila clam (Tapes philippinarum)
TR true retention factors.
binnenwerk.indd 462 21-08-2006 12:35:53

Raw soft tissue with shell liquor (CL treatment) amounted to the following percentage of
mollusc weight: TP = 46.86 ± 0.24, CC = 45.22 ± 0.84, and MG = 67.09 ± 1.06, whereas the
incidence of soft tissue without shell liquor (SL treatment), that is the meat yield %, was: TP =
22.26 ± 0.66, CC = 23.75 ± 0.32, and MG = 26.72 ± 0.71. Therefore, MG emerged as the richest
of shell liquor amongst the three species.
Independently of how the mollusc meat was to be analysed (that is either CL or SL, since the
presence/absence of the shell liquor in the matrix could not possibly have any bearing on
the data), average cooking time (s) and final temperature reached by the soft tissues (°C)
were [mean ± standard error (s.e.)]: 254 ± 13 and 84.4 ± 1.0 for TP, 304 ± 15 and 81.9 ± 1.4
for CC, and 281 ± 9 and 80.8 ± 1.6 for MG, respectively. For TP only did not any significant
difference emerge between CL and SL cooking yields (overall mean ± s.e. = 84.57 ± 2.59%). On
the contrary, significant differences (p < 0.001) emerged between CL and SL cooking yields for
both CC (81.09 ± 1.03% and 59.16 ± 2.67%, resp.) and MG (90.37 ± 0.45% and 73.62 ± 2.72%,
resp.), thus revealing a considerable amount of meat shrinkage upon cooking.
On the whole, the macro elemental composition of SL-TP (Table 2) and SL-MG (Table 4) at the
raw state came within the combined ranges derived from several food composition tables and
data bases (Holland and others 1993; Carnovale and Marletta 2000; Souci and others 2000; USDA
2005b), though Na content tended to be higher than literature data as a possible consequence of
the domestic depuration previously described. As to the micro elemental composition of SL-TP and
Table 2. Ash and mineral contents of Tapes philippinarum, either with (CL) or without (SL) shell liquor
(mg/100 g edible portion, except where noted).
Nutrientd CL SL RSDe Stat. sign.f
Raw Cooked Raw Cooked Liquor (L) State (St) L * St
Ash, g 2.60a 2.81a 2.28b 2.35b 0.097 ** * n.s.

Calcium 76.8 69.4 69.0 63.2 25.554 n.s. n.s. n.s.
Phosphorus 125b 140b 218a 240a 14.738 *** * n.s.
Sulphur 185b 216b 325a 347a 22.649 *** * n.s.
Sodium 1090a 1139a 469b 554b 99.554 *** n.s. n.s.
Potassium 154b 176b 215a 198a 11.524 *** n.s. **
Magnesium 26.8b 28.6b 34.7a 36.1a 2.114 *** n.s. n.s.
Iron 3.12b 2.87b 5.40a 5.17a 0.648 *** n.s. n.s.
Copper 0.076b 0.073b 0.130a 0.121a 0.014 ** n.s. n.s.
Zinc 1.76b 1.90ab 2.08ab 2.44a 0.423 ** n.s. n.s.
Nickel 0.024 0.022 0.026 0.026 0.005 n.s. n.s. n.s.
Cobalt 0.013b 0.012b 0.016a 0.014ab 0.001 *** + n.s.
d Means on the same row followed by different superscripts differ significantly (a, b: p ≤ 0.05).
e RSD = Residual Standard Deviation.
f *** p ≤ 0.001;** p ≤ 0.01; * p ≤ 0.05; + p ≤ 0.10; n.s. = non significant.
binnenwerk.indd 463 21-08-2006 12:35:54

SL-MG, it was fairly comparable to the results of researches based on molluscs from the eastern
Mediterranean sea (Bogoni and Favretto 1988; Karakoltsidis and others 1995; Storelli and others
2000; Ghidini and others 2002). A total lack of recent compositional data emerged for CC.
Compared with CL molluscs, their SL counterparts had significantly lower contents of ash and
Na, whilst being richer in P, S, K, Mg (TP only), Fe, Cu (p < 0.10 for MG), Zn, Ni (TP excluded),
and Co. No significant differences emerged between CL and SL treatments for Ca contents
(Table 2, 3, and 4). Cooking did not induce significant changes in elemental contents of CL
molluscs. The same held true for SL-TP. A significant increase in concentration occurred upon
cooking for Zn, Ni, and Co in SL-CC, and for P, Fe, and Zn in SL-MG. The only significant decrease
in concentration was observed for K in SL-MG. From a nutritional standpoint and independently
of treatment, cooked TP and CC were richer sources of macro and micro elements than MG, TP
being significantly the richest in Cu and Zn, CC the best endowed with Mg, Ni, and Co.
The “species” effect was evident for the true retention factors (TR) of ash, Ca, P, S, K, and
Mg. TR of these minerals were significantly different when TP, CC, and MG were compared, the
lowest cooking yields of CC being responsible for the lowest values emerging for this species
(Table 5). The “liquor” effect proved to be significant for ash, S, K, and Fe: the CL treatment
generated significantly higher TR than the SL treatment with the only exception of Fe, where
the reverse was true. The most plausible explanation may be a propensity to leaching from the
meat into the liquor for S and K, and more in general for ash, against a concentration effect
for Fe. In general, K and Fe could be considered having different behaviour upon cooking: the
former is prone to leave the food with its own cooking juices, the latter tends to concentrate
Table 3. Ash and mineral contents of Callista chione, either with (CL) or without (SL) shell liquor (mg/100 g
edible portion, except where noted).
Ash, g 2.45a 2.69a 2.02b 2.09b 0.193 ** n.s. n.s.

Calcium 103 65.8 102 97.0 18.466 n.s. * +
Phosphorus 127b 110b 204a 237a 12.637 *** n.s. **
Sulphur 261b 287b 391a 441a 43.174 *** + n.s.
Sodium 765a 866a 355b 338b 129.019 ** n.s. n.s.
Potassium 167b 186b 224a 212a 11.256 *** n.s. *
Magnesium 76.1 78.2 71.0 77.0 5.099 n.s. n.s. n.s.
Iron 2.91b 2.71b 4.31ab 6.73a 1.111 *** + *
Copper 0.050b 0.048b 0.083a 0.100a 0.017 * n.s. n.s.
Zinc 0.62c 0.65c 1.00b 1.40a 0.087 *** *** ***
Nickel 0.051c 0.048c 0.075b 0.099a 0.008 *** * **
Cobalt 0.26b 0.23b 0.38b 0.55a 0.046 ** ** **
d Means on the same row followed by different superscripts differ significantly (a-c: p ≤ 0.05).
e,f Same footnotes as in Table 2.
binnenwerk.indd 464 21-08-2006 12:35:54

Table 4. Ash and mineral contents of Mytilus galloprovincialis, either with (CL) or without (SL) shell liquor
(mg/100 g edible portion, except where noted).
Ash, g 2.20ab 2.34a 1.81b 1.97ab 0.078 * ** n.s.

Calcium 38.7 33.6 32.4 34.8 5.363 n.s. n.s. n.s.
Phosphorus 69.2c 68.1c 146b 179a 8.185 *** ** **
Sulphur 188b 192b 314a 330a 19.647 *** n.s. n.s.
Sodium 956a 977a 353b 427b 86.827 * n.s. n.s.
Potassium 138c 141c 192a 165b 5.586 *** ** ***
Magnesium 67.6 69.2 55.1 65.2 5.164 + * n.s.
Iron 2.48bc 2.20c 4.40b 5.53a 0.279 * ** ***
Copper 0.037 0.032 0.071 0.070 0.004 + n.s. n.s.
Zinc 0.71c 0.70c 1.63b 2.09a 0.098 *** *** ***
Nickel 0.014b 0.012b 0.028a 0.028a 0.002 *** n.s. n.s.
Cobalt 0.011b 0.009b 0.023a 0.024a 0.001 *** n.s. n.s.
d Means on the same row followed by different superscripts differ significantly (a-c: p ≤ 0.05).
e,f Same footnotes as in Table 2.
because largely bound to components of the soft tissues. This observation was corroborated
by earlier results on mineral true retentions in other muscle foods (Badiani and others 1998,
1999). Nevertheless, it should be noted that the TR for Cu, Zn, Ni, and Co, micro elements
which are equally bound to heavy ligand molecules in organic matrixes (Reilly 2004), were not
affected by the presence/absence of liquor, as well as by the species. For these elements, as
well as for Na, a certain dispersion of the data could have prevented the emergence of any
significant difference.
No significant differences were observed between TR and apparent retention factors (AR) for
TP, either from the CL or the SL treatment (data not shown). Following the conclusions drawn
by Badiani and others (2002) about other nutrients from earlier research on muscle foods,
this could be ascribed to the high TP cooking yields, which did not differ between the two
treatments. It seemed as if whenever cooking yields were rather high (that is ≥ 80%), the AR
for any given nutrient in any given food would not differ significantly from its TR counterpart,
hence the former could be advantageously used in place of the latter. As a matter of fact, this
reasoning did not apply either to CL-CC or to CL-MG, where, notwithstanding high cooking
yields, AR was significantly (CC) or marginally (MG) higher than the corresponding TR for each
nutrient. Moreover, a statistically significant difference between AR and TR affected each and
every nutrient of SL-CC and SL-MG. As to CC, the overall difference (all nutrients considered)
between AR and TR (mean ± s.e.) was 8.39 ± 0.31% for the CL treatment and 16.34 ± 1.51%
for the SL treatment. The corresponding values for MG were 5.98 ± 0.15% and 12.99 ± 0.54%
for the CL and SL treatment, respectively.
In order to explain the clear-cut distinction observed between TR and corresponding AR values
for CC and MG, consideration should be given to the fact that their whole organism had
binnenwerk.indd 465 21-08-2006 12:35:54

Table 5. True nutrient retentions (%) for ash and minerals in Tapes philippinarum (TP), Callista chione (CC),
and Mytilus galloprovincialis (MG) analysed either with (CL) or without (SL) shell liquor.
Nutrientd Liquor (L) Species (Sp) Overall RSDe Stat. sign.f

mean
TP CC MG (om) Sp L Sp * L
Ash CL 88.9 89.8x 95.7 91.4x 8.418 ** *** **

SL 89.5a 61.6by 79.7ab 76.9y
om 89.2a 75.7b 87.7a
Calcium CL 74.8 54.4 79.8 69.7 21.774 * n.s. n.s.
SL 97.2 57.2 80.5 78.3
om 86.0a 55.8b 80.2a
Phosphorus CL 92.0 71.4 89.2 84.2 12.397 *** n.s. n.s.
SL 96.2 68.7 90.6 85.2
om 94.1a 70.1b 89.9a
Sulphur CL 96.5 88.9 92.5 92.6x 13.488 * ** n.s.
SL 92.9 65.9 77.2 78.6y
om 94.7a 77.4b 84.8 ab
Sodium CL 86.7 93.5 92.0 90.7 21.265 n.s. n.s. *
SL 108 62.0 92.2 87.6
om 97.6 77.8 92.1
Potassium CL 94.6 90.6x 92.1x 92.4x 9.663 * *** +
SL 80.2a 56.2by 63.8aby 66.8y
om 87.4a 73.4b 77.9ab
Magnesium CL 88.4 83.4 92.5 88.1 10.796 ** + +
SL 90.8a 64.0b 87.0a 80.6
om 89.6a 73.7b 89.7a
Iron CL 75.6 76.4 80.7 77.6y 15.911 n.s. * n.s.
SL 84.4 93.5 94.4 90.8x
om 80.0 85.0 87.6
Copper CL 80.4 80.3 75.7 78.8 16.083 n.s. n.s. n.s.
SL 82.6 70.4 72.5 75.2
om 81.5 75.4 74.1
Zinc CL 90.9 84.8 89.6 88.4 19.652 n.s. n.s. n.s.
SL 107 82.8 94.6 94.8
om 98.9 83.8 92.1
Nickel CL 80.4 78.2 82.8 80.5 19.883 n.s. n.s. n.s.
SL 93.9 79.3 73.4 82.2
om 87.2 78.7 78.2
Cobalt CL 80.9 75.3 79.8 78.7 13.705 n.s. n.s. n.s.
SL 76.3 87.0 75.9 79.7
om 78.6 81.2 77.9
e,f Samefootnotes as in Table 2.

d Means followed by different superscripts differ significantly (a, b: between groups; x, y: within group;
p ≤0.05).
binnenwerk.indd 466 21-08-2006 12:35:55

undergone heating, possibly displaying quite a distinct behaviour from that of a real muscle food
as, for instance, a beef cut or a fish fillet. These findings seem to corroborate the observations
emerged from the research of Murphy and others (1975). Nevertheless, the different behaviour
of TP retentions suggests that the response of bivalve organs and tissues to cooking could be
species-specific, probably as a consequence of their peculiar relative importance. However,
taking into account the research effort needed to clarify this issue even confining the study
to the most widely consumed species, it would be advisable to try and determine TR only,
whenever bivalve molluscs are under examination for ash and mineral retentions.
Conclusion
T. philippinarum, C. chione, and M. galloprovincialis were more significant sources of several

macro and micro elements when consumed without shell liquor than when eaten with it, which
was ascribed to a dilution effect caused by the liquor itself on nutrient contents within each
serving. Expected pitfalls in the implementation of the calculation of true retention factors for
bivalve molluscs could be easily overcome. From the examination of the true retention factors,
it emerged that: amongst the selected species, C. chione was affected by the lowest mineral
retentions, as a consequence of the lowest cooking yields; amongst the selected minerals,
sulphur and potassium showed a certain propensity to leaching from the meat into the shell
liquor upon cooking, whereas iron tended to be concentrated within the meat itself. Apparent
retention factors approximated the corresponding true retentions for T. philippinarum only. It
is therefore advisable to adopt true retention factors instead of their more easily attainable
apparent counterparts, when studying the mineral component of bivalve molluscs.
Acknowledgements
Financial support received from the Italian Ministry of Education, University and Research
(MIUR, “ex-60%” funds) is gratefully acknowledged.
References
AOAC. 2000. Official methods of analysis. 17th Ed. Vol. II. Gaithersburg, Md.: Association of Official Analytical Chemists.
Chap. 39. p 1-4.
Badiani A, Bitossi F, Marchetti S, Stipa S. 1999. Contents and retention of minerals in selected beef retail cuts cooked
by different household methods. In: Piva G, Bertoni G, Masoero F, Bani P, Calamari L, editors. Recent progress in
animal production science. 1. Milano, I: Franco Angeli s.r.l. p 683-685.
Badiani A, Nanni N, Gatta PP, Bitossi F, Tolomelli B, Manfredini M. 1998. Nutrient content and retention in selected
roasted cuts from 3-month-old ram lambs. Food Chem 61:89-100.
Badiani A, Stipa S, Bitossi F, Gatta PP, Vignola G, Chizzolini R. 2002. Lipid composition, retention and oxidation in fresh
and completely trimmed beef muscles as affected by common culinary practices. Meat Sci 60:169-186.
Bogoni P, Favretto L. 1988. Ricerche chemometriche su elementi in tracce nei mitili eduli. Riv Ital Sci Aliment 17(2):91-
98.
Carnovale E, Marletta L. 2000. Tabelle di composizione degli alimenti – Aggiornamento 2000. Roma, I: Istituto
Nazionale di Ricerca per gli Alimenti e la Nutrizione. p 37, 40, 99.
Davidson A. 2002. Mediterranean seafood. A comprehensive guide with recipes. Berkeley, Ca.: Ten Speed Press. p 195-
207, 245-351.
Ghidini S, Battaglia A, Zanardi E, Campanini G. 2002. Elementi in traccia in molluschi e crostacei dell’Alto Adriatico.
Ann Fac Med Vet Univ Parma 21:225-233.
binnenwerk.indd 467 21-08-2006 12:35:55

Ghidini S, Zanardi E, Conter M, Ianieri A, Campanini G. 2005. Possibilità di utilizzo dell’ICP-AES nell’ispezione degli
alimenti di origine animale. Electronic Proceedings of the SISVet National Congress; 2005 Sept 21-24; Viareggio,
I. Powered by Sinergia Arti Grafiche e Visive, www.sinergiaweb.it. p. 379-380.
Haytowitz DB, Pehrsson PR, Smith J, Gebhardt SE, Matthews RH, Anderson BA. 1996. Key foods: setting priorities for
nutrient analyses. J Food Comp Anal 9:331-364.
Holland B, Brown J, Buss DH. 1993. Fish and fish products. Third supplement to the fifth edition of McCance and
Widdowson’s “The composition of foods”. Cambridge, UK: The Royal Society of Chemistry. p 78-85.
Karakoltsidis PA, Zotos A, Constantinides SM. 1995. Composition of the commercially important Mediterranean finfish,
crustaceans, and molluscs. J Food Comp Anal 8:258-273.
Murphy EW, Criner PE, Gray BC. 1975. Comparison of methods for calculating retentions of nutrients in cooked foods.
Reilly C. 2002. Minerals. In: Henry CJK, Chapman C, editors. The nutrition handbook of food processors. Cambridge,
UK: Woodhead Publishing Ltd. p 97-116.
Reilly C. 2004. The nutritional trace elements. Oxford, UK: Blackwell Publishing Ltd. p 82-134, 211-233.
Schakel SF, Buzzard IM, Gebhardt SE. 1997. Procedures for estimating nutrient values for food composition databases.
J Food Comp Anal, 10:102−114.
Slow Food. 2004. Ricette di osterie d’Italia - Il pesce: 600 piatti di mare, di lago e di fiume. Bra, I: Centro Stampa.
p 462.
Souci SW, Fachmann W, Kraut H. 2000. Food composition and nutrition tables. 6th Ed. Stuttgart, D: Medpharm GmbH
Scientific Publishers. p 482-483.
Storelli MM, Storelli A, Marcotrigiano GO. 2000. Heavy metals in mussels (Mytilus galloprovincialis) from the Ionian Sea,
Italy. J Food Protect 63:273-276.
[USDA] U.S. Department of Agriculture. 2005a. Composition of foods - raw, processed, prepared. Release 18. Agricultural
Research Service, Beltsville Human Nutrition Research Center, Nutrient Data Laboratory, Beltsville, Md. Available
from: http://www.nal.usda.gov/fnic/foodcomp/Data/SR18/SR18_doc.pdf. Accessed September 2, 2005.
[USDA] U.S. Department of Agriculture, Agricultural Research Service. 2005b. USDA National Nutrient Database for
Standard Reference, Release 18. Nutrient Data Laboratory Home Page, NDB No. 15157, 15164. Available from:
http://www.nal.usda.gov/fnic/foodcomp/Data/SR18/reports/sr18page.htm. Accessed September 2, 2005.
binnenwerk.indd 468 21-08-2006 12:35:56

Seasonal variation of proximate and fatty acid class composition of
wild and cultured Brown Meagre (Sciena umbra), a new species for
aquaculture
Şükran Caklı1, Tolga Dincer1, Aslı Cadun1, Şahin Saka2 and Kürşat Firat2
1Ege University, Faculty of Fisheries, Department of Fishery and Fish Processing Technology, 35100
Bornova, Izmir, Turkey
2Ege University, Faculty of Fisheries, Department of Aquaculture, 35100 Bornova, Izmir, Turkey
Abstract
Mediterranean fishery and aquaculture industry is looking for alternative species for aquaculture
in sea cages. Different species can be chosen as potential aquaculture candidates because of their
eating qualities, their nature and ability to be held in captivity at high densities. In this regard,
brown meagre (Sciena umbra) is one of the candidate species, which offers good possibilities.
No studies about the quality of this species either wild or cultured exist. In the current study,
the proximate and fatty acid class composition of the flesh of cultured and wild brown meagre
(Sciena umbra) were analysed each month during a 8 months period of time. In addition, colour
measurements and sensory analyses were carried out. Clear seasonal differences were determined
in the proximate composition of wild and cultured fish. The seasonal dependent fatty acid
analysis done for each group of fish, seemed to indicate higher levels of monoenes, PUFA and
HUFA (n-3 and n-6) fatty acids in cultured fish and higher levels of saturated fatty acids in wild
fish. In the sensory analysis wild meagre was perceived as having less fresh odour and fresh
flavour and being less fat and less juicy. From the results obtained in the present work it can be
concluded that cultured brown meagre can be an addition to traditional cultured species.
Keywords: Sciena umbra, fatty acids, colour, proximate composition, new species
Introduction
The quality differences of fish being caught in the wild and fish from aquaculture continues
to be an important issue in many countries of the world. Many criteria have been established
and many studies have been executed about the quality of fish flesh (Iwamato and Yamanaka
1986, Khalil and others 1986, Aoki and others 1991, Nettleton and Exler 1992, Caklı and Alpbaz
1997, Jonsson 1997, Gallagher and others 1998, Jeong and others 2002, Alam and others 2002,
Fåhraeus-Van Ree and Spurrell 2003). Fish culture in the Mediterranean is essentially based on
two species, sea bream and sea bass, with productions which have increased in a spectacular
way e.g. from 37,179 mt in 1994 to 76,000 mt in 1998. This increase in production has led to
market saturation and a fall in price. One of the forms in which market supply may be increased
and a contribution be made to development and expansion of aquaculture is to diversify the
species being cultured. In this regard, brown meagre is one of the candidate species, which
offers good possibilities.
The species Sciena umbra and Dentex dentex have great potential for aquaculture in Turkey
where the Aegean aquaculture industry is looking for alternative species to culture in sea cages.
binnenwerk.indd 469 21-08-2006 12:35:56

According to the fishermen brown meagre (Sciena umbra) is highly prized as one of the best
eating fishes in Turkey. The species Sciena umbra is distributed from Western part to Southern
part of Turkey from Izmir to Antalya. These species can be chosen as a potential aquaculture
candidate because of their eating qualities, their nature and ability to be held in captivity at
high densities. However, understanding and controlling reproduction in brown meagre is the
fundamental challenge for their development as a new species for aquaculture. The aquaculture
of brown meagre, which is being done in small-size operations, is a new pilot activity. For that
reason, no studies about the fish quality of this species either wild or cultured exist.
Turkey is the second largest producer of farmed fish in the Mediterranean region. Fishery
production of Turkey is approximately 627,847 tons/year. Aquaculture production amounts to
61,165 tons/year (FAO 2002). Aquaculture of sea bass and sea bream has been done successfully
in Turkey since 1990. Also the aquaculture activity of new commercial species as common
dentex, sharpnout seabream, brown meagre and red sea bream began in 2000 with a few
foundations and is today still active with a successful production.
Brown meagre, which could be found easily along the shallow rocky bottoms until 50 years ago,
are now difficult to find in most of the coastal areas, due to spear fishing and environmental
changes. Continuous fishing pressure and pollution in some coastal areas were expected to
result in a serious depletion of several fish stocks over the past three decades. At that time
culture activities of this species became important. In this regard, cultured meagre is one of
the candidate economical species, which offers good financial possibilities to the fishermen and
good nutrient quality for the consumers.
In this study, a comparison of principal quality of cultured and wild meagre (Sciena umbra) was
aimed at. To achieve this aim, determinations of proximate composition (protein, fat, moisture)
and fatty acid composition, colour measurements and sensory analyses have been performed
monthly over a period of 8 months.
Material
A total of 40 samples of wild and 40 samples of aquacultured Sciena umbra were included in
this study. Cultured samples were taken from a single aquaculture farm and wild samples were
caught in the Aegean Sea. Over a period of 8 months 5 samples of each of wild and cultured
meagre were taken and analyzed every month. Of each batch of 5 samples 3 were used for the
determination of the proximate composition, the analysis of fatty acid composition as well as
colour measurements. The remaining two samples were subjected to a sensory analysis.
Methods
The first part of the method deals of the aquacultural activity and the second part deals of the
determination of the nutritive and sensory characteristics of the portion size fish.
Aquacultural activity
Brood stock and egg incubation
Brown meagre (Sciena umbra) broodstock, 5 females (1.4 ± 0.1 kg mean body weight) and 5
males (0.7 ± 0.1 kg mean weight), were selected from wild breeders and stocked in an 10 m3
tank with a seawater flow of 1.5 m3 per hour. The fish were hand-fed once a day to satiation
binnenwerk.indd 470 21-08-2006 12:35:56

Seasonal variation of proximate and fatty acid class composition of Brown Meagre
at noon, with approximately equal amounts by weight of a moist pellet brood fish diet and
cuttlefish (Sepia officialis) and shrimp (Palaemon elegans). The fish were subjected to natural
photoperiod, and the water temperature varied between 19-24 °C. Eggs spawned by fish group
were immediately collected in a recuperator. Following the fertilization, the viable buoyant eggs
were separated from the dead sinking eggs.
Larval rearing
Larvae were stocked at density of 80 individuals.l-1 in a circle shape tank (6 m3). Larval rearing
was carried out in a closed sea water system. Water temperature was maintained between 19
and 24 °C. During larval culture period, oxygen, salinity and pH were maintained at > 90%,
38‰ and 7.8, respectively. Ammonia and nitrite were kept constant always below 0.01 mg.1-1.
During the rearing conditions clear water technique was applied. From the day 3 to day 8, larvae
were fed on rotifers. From day 5 to day 18 the larvae were fed on freshly hatched Artemia nauplii
and from day 15 to 32 they were fed on enriched Artemia metanauplii. From day 30 until day
90 only dry food were distributed. Cages were in a circle shape and has a 12 m radius and 8 m
water depth. In these cages according to largeness of fish 8-18 mm nets without a bow was
used. Fish which were in 4.1 ± 0.2 g weight were stoked at the density of 20 individuals/m3.
Moisture pellets were used in feeding and feeding was done three times a day. Pellets were
distributed slowly, allowing all fish to eat. Fish were sampled in the cage conditions in the
middle of every month. Analytical part of this study started after the cultured fish became
nearly 200 gr (portion size).
Analytical methods
Analysis of proximate composition
Three of the wild and of the cultured meagre sampled each month were pooled and homogenized.
Dry matter was determined by drying the samples at 105 °C to a constant weight (AOAC
1990). Crude protein content was calculated by converting the nitrogen content determined
by Kjeldahl’s method (6.25xN). Total lipid (TL) was extracted and purified according to Bligh
and Dyer (1959), and TL content was determined gravimetrically. The analyses of the pooled
samples were all carried out in triplicate.
Analysis of fatty acid compositions

The lipids of the TL extract were saponified and esterified for fatty acid analysis by the method
of IUPAC II D19. Separation of fatty acid methyl esters was achieved on a SP-2330 Fused Silica
Capillary Column (30 m x 0.25 mm, i.d. 0.20 µm). The oven temperature was 120 ºC for 5 min,
programmed to 180 ºC at 10 ºC/min, then programmed to 220 ºC at 20 ºC/min and then held
there for 20 min. The carrier gas was helium with a linear flow rate 0.5 ml/min and split ratio
of 1/150.
Colour measurement
The colour measurement of fish samples were carried out with the spectral colour meter Spectro-
pen® (Dr. Lange, Düsseldorf, Germany). The colour was measured on the homogenates prepared
each month for the proximate analysis. The homogenate was placed in plastic Petri dishes, the
colour of each of homogenate was measured in ten different locations and the average values
of the measurements were determined. In the CIE Lab system L* denotes lightness on a 0 to
100 scale from black to white; a* (+) red or (-) green; and b* (+) yellow or (-) blue (Schubring
and Meyer 2002).
binnenwerk.indd 471 21-08-2006 12:35:56

Sensory analysis
The sensory panel was trained according to Carbonell and others (2002). Dorsal fillets were cut
from each meagre. The parts close to the head and the tail, respectively, were removed to obtain
portions of approximately 6x3 cm. For preliminary sessions (selection of descriptors) each
fillet was divided into two portions of approximately 3x3 cm. For the main study two cultured
and two wild brown meagre were analyzed each month and the whole fillets (approximately
3x3 cm) were presented to a trained panel of 8 people for sensory assessment. In the sensory
analysis, an acceptance test was performed using a scale form 0 to 10 and the 11 descriptors are
shown in Table 5. In the preliminary as well as in the main study, the samples were wrapped in
aluminium foil without any seasoning added and baked in a preheated conventional household
oven. Preheating took 30 min at 200 °C and cooking time was 5 min for approximately 3x3
cm portions and 8 min for whole fillets. Portions from the same sample batch were heated
together. The sensory analyses were carried out each month of the study with the main aim of
determining differences in the sensory quality of cultured and wild meagre.
Results
Changes in proximate composition of wild and farmed meagre are given in Table 1.
The chemical composition of wild brown meagre captured from nature varied as follows: moisture
from 71.99 ± 2.66% to 76.36 ± 0.66%, total lipid from 0.75 ± 0.04% to 1.89 ± 0.27% and protein
from 20.35 ± 0.13% to 23.53 ± 0.25% during the 8 months study. Chemical composition of
cultured brown meagre varied as follows: moisture from 68.11 ± 0.77% to 76.54 ± 0.19%, total
lipid from 0.55 ± 0.05% to 4.81 ± 0.01% and protein from 19.24 ± 1.44% to 21.82 ± 1.46% during
the 8 months study. Differences between cultured and wild meagre in their moisture, fat and
protein values are evident. The fat and moisture values had the expected inverse correlations.
Table 1. Proximate composition of brown meagre (%). Arithmetic means and standard deviation.
Months Moisture % Fat % Protein %
Wild meagre October 75.55 ± 0.44 1.89 ± 0.27 20.35 ± 0.13

November 75.32 ± 0.33 1.46 ± 0.07 20.76 ± 0.09
December 76.36 ± 0.66 1.23 ± 0.27 22.10 ± 0.04
January 74.30 ± 0.26 1.64 ± 0.30 23.53 ± 0.25
February 74.78 ± 0.23 0.75 ± 0.04 23.23 ± 0.20
March 71.99 ± 2.66 1.22 ± 0.05 22.22 ± 0.91
April 72.96 ± 0.53 1.48 ± 0.05 23.24 ± 0.05
May 73.66 ± 1.44 1.48 ± 0.05 23.05 ± 0.14
Cultured meagre October 76.54 ± 0.19 1.81 ± 0.01 19.24 ± 1.44
November 76.35 ± 0.35 0.55 ± 0.05 20.30 ± 0.35
December 75.02 ± 0.73 3.12 ± 0.25 20.40 ± 0.2
January 75.20 ± 1.81 2.85 ± 0.73 20.65 ± 0.82
February 68.11 ± 0.77 3.86 ± 0.27 21.82 ± 1.46
March 72.09 ± 1.15 4.13 ± 0.34 20.55 ± 0.57
April 72.21 ± 1.10 4.39 ± 1.08 19.60 ± 0.58
May 76.54 ± 0.19 4.81 ± 0.01 19.24 ± 1.24
binnenwerk.indd 472 21-08-2006 12:35:56

In Table 2 and 3 the fatty acid composition is shown of the wild and farmed brown meagre,
respectively. Cultured meagre contains higher levels of total n-3 highly unsaturated fatty acids
(HUFA) and polyunsaturated fatty acids (PUFA) compared to wild fish. The ratio of n-3 PUFA
to n-6 PUFA is higher in cultured meagre. Nutrient sources (feed) of cultured meagre have
obviously influenced their lipid composition, as frequently reported for the other species, even
if fatty acid composition of feed is not reported in this paper. Consumption of either wild or
cultured meagre will contribute to dietary n-3 PUFA intake, with benefits to human health.
As shown in the Tables 2 and 3 there are not general variations of fatty acid fractions from
the first to the eighth month of this study. This may be a result of diet combination and the
intake of the lipid via feed. Cultured fish seemed to show higher levels of monoenes and PUFA
and HUFA fatty acids (n-3 and n-6), but wild fish had higher levels of saturates fatty acids.
These results have some similarities with the study of Chang-Yang Peng and Hsueh-Chen Tsai
(2004). Values of colour measurement on wild and cultured meagre during the 8 months study,
are presented in Table 4.
For wild brown meagre the results for L* values varied from 51.5 ± 2.08 to 61.8 ± 3.15, for a*
from -2.8 ± 0.12 to 0.1 ± 0.18 and for b* values from 8.7 ± 0.41 to 11.9 ± 1.17. For cultured
brown meagre the results are as follows: L* values ranged from 51.5 ± 2.22 to 60.2 ± 1.10,
a* values of the cultured and wild samples were similar. Similar to the cultured fish b* values
ranged from 5.0 ± 1.15 to 11.6 ± 1.13. Similar results were reported in a study with different
species by Caklı and others (2005).
In the preliminary session, sample preparation and experimental conditions were established
and 13 sensory attributes were selected as descriptors (2 for odour, 2 for appearance, 2 for
flavour and 5 for texture). Over the 8 months time period clear differences between samples
were detected for seven attributes (odour intensity, fresh odour, colour, flavour intensity, fresh
flavour, juiciness and fatness). Cooked wild samples were perceived as having less fresh odour
and fresh flavour and being less fat and less juicy.
Conclusion
In a parallel evaluation of cultured and wild brown meagre flesh quality, some differences could
be detected. There were differences between cultured and wild meagre in their moisture, fat and
protein contents. Furthermore cultured fish seemed to have higher levels of monoenes, PUFA
and HUFA fatty acids (n-3 and n-6) while wild fish had higher levels of saturates fatty acids.
As the sensory results which were based mainly on odour and flavour attributes showed several
differences, the overall sensory quality was also considered better in cultured meagre.
From the results obtained in the present work it can be concluded that the cultured meagre can
be an alternative species in terms of the culture of traditionally Mediterranean species.
binnenwerk.indd 473 21-08-2006 12:35:57

Table 2. Wild meagre fatty acid composition. Arithmetic means and standard deviation.
binnenwerk.indd 474
474
Months October November December January February March April May
Saturates 50.39 ± 0.21 48.80 ± 0.05 49.45 ± 0.21 50.00 ± 0.34 56.87 ± 0.24 55.97 ± 0.21 50.95 ± 0.41 50.47 ± 0.25
Monoenes 34.27 ± 0.25 35.38 ± 0.10 36.41 ± 0.02 38.41 ± 0.13 25.99 ± 0.18 31.34 ± 0.16 35.71± 0.27 30.15 ± 0.13
Total (n-3) 9.62 ± 0.17 9.60 ± 0.13 9.86 ± 0.16 9.79 ± 0.10 7.52 ± 0.04 7.52 ± 0.027 9.65 ± 0.18 9.84 ± 0.54
Total (n-6) 2.65 ± 0.20 2.54 ± 0.21 2.74 ± 0.05 2.74 ± 0.04 3.14 ± 0.28 3.15 ± 0.18 2.77 ± 0.01 2.77 ± 0.34
HUFA (n-3) 8.88 ± 0.12 9.42 ± 0.14 9.43 ± 0.23 9.03 ± 0.07 6.05 ± 0.31 6.18 ± 0.31 9.23 ± 0.05 8.44 ± 0.52
PUFA 15.34 ± 0.14 15.82 ± 0.31 14.14 ± 0.02 13.20 ± 0.06 17.14 ± 0.02 12.68 ± 0.17 11.52 ± 0.41 19.38 ± 0.21
Table 3. Cultured meagre fatty acid composition. Arithmetic means and standard deviation.
Months October November December January February March April May
Saturates 43.14 ± 0.015 42.23 ± 0.06 41.89 ± 0.25 42.51 ± 0.42 36.93 ± 0.17 43.14 ± 0.015 42.23 ± 0.06 41.89 ± 0.25
Monoenes 35.10 ± 0.03 35.56 ± 0.04 41.14 ± 0.18 35.52 ± 0.16 40.36 ± 0.02 35.10 ± 0.03 35.56 ± 0.04 41.14 ± 0.18
Total (n-3) 13.75 ± 0.16 14.27 ± 0.54 9.21 ± 0.32 13.22 ± 0.06 11.71 ± 0.07 13.75 ± 0.16 14.27 ± 0.54 9.21 ± 0.32
Total (n-6) 7.92 ± 0.02 7.65 ± 0.15 8.29 ± 0.56 8.82 ± 0.16 8.35 ± 0.16 7.92 ± 0.02 7.65 ± 0.15 8.29 ± 0.56
HUFA (n-3) 10.65 ± 0.04 11.18 ± 0.07 8.85 ± 0.32 12.64 ± 0.05 9.65 ± 0.25 10.65 ± 0.04 11.18 ± 0.07 8.85 ± 0.32
PUFA 21.82 ± 0.51 22.08 ± 0.35 16.98 ± 0.01 22.33 ± 0.02 19.70 ± 0.12 21.82 ± 0.51 22.08 ± 0.35 16.98 ± 0.01
Table 5. Sensory assessment of cooked samples of meagre (average values for all sessions).
Sample Odour Fresh odour Colour Flakiness Flavour Fresh Firmness Chewiness Fibrousness Juiciness Fatness
intensity intensity flavour
Wild 6.87 ± 2.21 6.12 ± 2.43 2.66 ± 2.4 5.00 ± 2.04 6.32 ± 1.53 6.85 ± 1.2 6.95 ± 2.71 3.20 ± 1.89 5.35 ± 2.63 5.12 ± 2.37 5.12 ± 2.06
Cultured 6.17 ± 2.80 6.94 ± 2.07 3.24 ± 1.01 5.04 ± 3.06 8.97 ± 1.33 9.02 ± 1.46 6.91 ± 2.61 3.22 ± 2.48 5.40 ± 2.21 6.95 ± 2.54 6.74 ± 1.13
21-08-2006 12:35:57
Table 4. Colour measurements of wild and cultured meagre. Arithmetic means and standard deviation.
Months L* a* b*
Wild meagre October 52.5 ± 1.18 0.1 ± 0.18 10.6 ± 0.28

November 51.5 ± 2.08 -0.5 ± 0.12 8.7 ± 0.41
December 55.4 ± 2.13 -1.3 ± 0.01 9.8 ± 1.32
January 52.7 ± 0.85 -1.4 ± 0.20 9.1 ± 1.34
February 59.3 ± 1.12 -0.2 ± 0.14 10.0 ± 0.82
March 61.8 ± 3.15 -1.3 ± 0.14 11.9 ± 1.17
April 59.9 ± 1.12 -2.8 ± 0.12 9.3 ± 3.10
May 56.6 ± 1.02 -1.5 ± 0.01 11.4 ± 2.17
Cultured meagre October 57.4 ± 1.17 -1.7 ± 0.12 9.3 ± 1.21
November 57.5 ± 1.04 0.6 ± 0.13 11.6 ± 1.13
December 58.8 ± 1.16 -1.4 ± 0.23 6.3 ± 2.07
January 51.5 ± 2.22 -1.4 ± 0.05 5.0 ± 1.15
February 56.0 ± 1.14 -2.4 ± 0.10 7.3 ± 1.62
March 56.5 ± 2.23 -2.3 ± 0.31 8.9 ± 0.86
April 60.2 ± 1.10 -2.6 ± 0.08 5.4 ± 1.23
May 54.8 ± 2.12 -1.0 ± 0.19 9.8 ± 0.8
References
Alam MGM, Tanaka A, Allinson G, Laurenson LJB, Stagnitti F, Snow LK. 2002. A comparison of trace element concentrations
in cultured and wild carp (Cyprinus carpio) of Lake Kasumigaura, Japan. Ecotoxicol Environm Safety 53:348-354
AOAC.1990. Official methods of analyses of association of analytical chemist (15th ed.). Washington DC: AOAC.
Aoki T, Takada K, Kunisaki N. 1991. On the study of proximate composition, mineral, fatty acid, free amino acid, muscle
hardness and colour difference of six species of wild and cultured fishes. Nippon Suisan Gakkaishi, 57(10):1927-
1934
Bligh EG, Dyer WJ. 1959. A rapid method of total lipid extraction and purification. Can J Biochem Physiol 37:911-917.
Caklı S, Alpbaz AG. 1997. An investigation on the meat quality of cultured and wild gilthead seabream (Sparus aurata
L. 1758) seabass and seabream culture. Problems and Prospects. October, 16-18, Verona, Italy.
Caklı S, Dincer T, Cadun A, Fırat K, Saka S. 2005. Quality characteristics of wild and cultures common dentex (Dentex
dentex, Linnaeus,1758). Arch Lebensmittelhyg 56: Article in press.
Carbonell I, Izquierdo L, Costell E. 2002. Sensory profiling of cooked gilthead sea bream (Sparus aurata): Sensory
evalution procedures and panel training. Food Sci Techn Int 8(3):169-177.
Chang-Yang P, Hsueh-Chen T. 2004.Comparison of the chemical composition between cultured and wild milkfish (Chanos
chanos). J Taiwan Fish Res 12(1):71-77
Fåhraeus-Van Ree GE, Spurrell DR. 2003. Structure of and energy reserves in the liver of wild and cultured yellowtail
flounder, Limanda ferruginea. Mar Biol 143:257-265
FAO. 2002. Fisheries Statistics. Food and Agriculture Organization of the United Nations
Gallagher ML, Paramore L, Alves D, Rulifson RA. 1998. Comparison of phospholipid and fatty acid compositioın of wild
and cultured striped bass eggs. J Fish Biol 52:1218- 1228
Iwamato M, Yamanaka H. 1986. Remarkable differences in rigor mortis between wild and cultured specimens of the Red
Sea bream Pagurus major.. Bull Jap Soc Sci Fish 52(2):275-279
Jonsson B. 1997. A review of ecological and behavioural interactions between cultured and wild Atlantic salmon. ICES
J Mar Sci 54:1031-1039
binnenwerk.indd 475 21-08-2006 12:35:58

Khalil MS, Hilmy AM, Badavi HK, Wassef EA. 1986. Proximate composition of wild and reared Gilthead Bream, Chrysophyrs
auratus. Bull Fac Sci Cairo Univ 54:1-30
Jeong BY, Jeong WG, Moon SK, Ohshima T. 2002. Preferential accumulation of fatty acids in the testis and ovary of
cultured and wild sweet smelt Plecoglossus altivelis. Comp Biochem Physiol B 131:251-259
Nettleton JA, Exler J. 1992. Nutrients in Wild and Farmed Fish and Shellfish. J Food Sci 57(2):257-260
Schubring R, Meyer C. 2002. Quality factors of terrestrial snail products as affected by the species. J Food Sci 67:3148-
3151.
binnenwerk.indd 476 21-08-2006 12:35:58

Composition and nutritional value of fishery products consumed in
Portugal
Maria Leonor Nunes1, Narcisa Bandarra1, Luisa Oliveira2, Ireneu Batista1 and Maria Antónia
Calhau2
1Department of Technological Innovation and Upgrading of Fish Products, National Research Institute
for Agriculture and Fisheries, INIAP/IPIMAR, Avenida de Brasília, 1449-006 Lisboa, Portugal
2National Health Institute Dr. Ricardo Jorge, INSA, Avenida Padre Cruz, 1649-016 Lisboa, Portugal
Abstract
Taking into account the importance of seafood products in the Portuguese diet, the aim of this
study was to determine the proximate composition, fatty acid profile, minerals, vitamins and
cholesterol levels in raw and cooked edible part of several species. All samples were purchased
in different periods of the year at retailer markets. The analyses were done in several samples
(five for the most consumed), with approx. 10 kilos each. The sampling period took place
over three years. Three cooking processes were used: boiling, frying and grilling, following
the normal household preparation. The several macro and micronutrients were determined by
using standard and well-established chemical methodologies. The results obtained allowed to
conclude that there are many nutritional benefits associated to consumption of seafood, namely
the amounts of omega-3 fatty acids and minerals.
Keywords: fish products composition, nutritional value, minerals, vitamins, fatty acids, cooking
processes
Introduction
Consumers, especially after the middle of the twentieth century, have become aware that good
health and well-being is linked to a balanced diet. However, it is still difficult to define a correct
diet because the local gastronomy, cultural habits and the tradition, among other aspects, can
strongly influence its definition. Nevertheless, the consumption of a wide diversity of foods,
the reduction of fat, particularly saturated fat, cholesterol, sodium and sugar intakes, and the
increase of fibre levels are basic guidelines.
Among this broad variety of food products, seafood can give a significant contribution to the
nutrient needs since it is a source of top-quality protein food that is low in calories, total fat,
and saturated fat when compared to other protein-rich animal foods. Thus, taking into account
that in Portugal the consumption per capita of fishery products is very high (about 60 kg/
person. year), a study on the chemical composition and nutritional values of the most relevant
products consumed over the last years was accomplished. The main objectives of this work were
(i) to complement data in international food composition tables since the information in those
tables is of little use for national purposes, as most of the fish and fish products consumed
by the Portuguese population are not included and (ii) to organise a national database with
relevant information. Another goal was to contribute to increase the scientific evidence of the
importance of fish and farmed products in a healthy diet.
binnenwerk.indd 477 21-08-2006 12:35:58

Maria Leonor Nunes, Narcisa Bandarra, Luisa Oliveira, Ireneu Batista and Maria Antónia Calhau
Raw material
Wild and farmed species (Common cockle Cerastoderma edule, grooved carpet shell Ruditapes
decussatus, common octopus Octopus vulgaris, European squid Loligo vulgaris, Norway lobster
Nephropus norvegicus, desalted Atlantic cod Gadus morhua, Atlantic salmon Salmo salar, black
scabbard fish Aphanopus carbo, blackspot seabream Pagellus bogaraveo, European eel Anguilla
anguilla, European hake Merluccius merluccius, European plaice Pleuronectes platessa, farmed
gilthead seabream Sparus aurata, farmed seabass Dicentrarchus labrax, horse mackerel Scomber
japonicus, meagre Argyrosomus regius, monkfish Lophius piscatorius, sardine Sardina pilchardus,
silver scabbardfish Lepidopus caudatus, wreck fish Polyprion americanus) were purchased in
different periods of the year at the Portuguese retailer markets. The analyses were done in
several composite samples (one to six according to the consumption), with approx. 10 kilos
each. The sampling period took place over three years.
Analyses were performed in the edible part of raw product, which was calculated according to
the usual preparation, e.g., heading, gutting or filleting, of each species. It was expressed as
percentage of the whole fish or shellfish weight as it is usually purchased. The products were
boiled in water with 1.5% (small portions) or 2% of salt (big portions), using a fish/water ratio
of 1:2. For grilling, fish was previously spiked with 1.5 or 2.0% salt (depending on the fish
size). Salt was partially removed after 15 minutes and grilling was done on an electrical grill
(2000 Watts). In the case of frying, fish was salted as for grilling, then coated with wheat flour
and deep-fried in vegetable oil. The oil was discarded after each batch frying. The analysis of
canned products was only done with fish and the covering sauce was discarded.
Methods
Energetic value is expressed as kcal and kJ, taking FAO (1989) factors into account for fat
and protein. Protein was calculated from total nitrogen content, determined by the Kjeldahl
method (AOAC 1998), using 6.25 as the conversion factor. Lipids, also designated as fat, were
determined by gravimetric method after extraction with organic solvents (petroleum ether or a
mixture of chloroform, methanol and water), according to NP 1972 (IPQ 1992) and Bligh and
Dyer method (1959), respectively. Fatty acids were analysed by gas chromatography, following
the experimental procedure developed by Lepage and Roy (1986) and modified by Cohen and
others (1988). The values are expressed as mg/100 g edible part, using the conversion factors
proposed by Weihrauch and others (1977). Water was calculated by gravimetry after drying
in an oven at the constant temperature of 105 ºC (IPQ 1991); it includes water and a small
quantity of volatile substances, which volatilise at the assay temperature. Ash was determined
by gravimetry after drying and combustion of organic matter in a muffle furnace at the constant
temperature of 500 ºC (IPQ 1988). Cholesterol was assayed by gas chromatography, after
saponification in alkaline conditions, based on the method of Naemmi and others (1995) modified
by Oehlenschläger (2000). Calcium and magnesium were measured by flame atomic absorption
spectrometry in the ash hydrochloric solution, and using lanthanum as the releasing agent
(AOAC 2000a). Chloride was determined by a volumetric method (IPQ 1988) after precipitation
of chloride with a silver nitrate solution in excess and titration of the excess with ammonium
thiocyanate in the presence ferric indicator. Copper, iron and zinc were analysed by flame atomic
absorption spectrophotometry in the ash hydrochloric solution (AOAC 2000b). Manganese was
determined in an ash nitric solution by atomic absorption spectrometry equipped by using a
graphite furnace (AOAC 2000a). Phosphorus was analysed by spectrophotometry, after reaction
binnenwerk.indd 478 21-08-2006 12:35:58

Composition and nutritional value of fishery products consumed in Portugal
of molybdovanadate reagent with the ash hydrochloric solution (AOAC 2000b). Potassium and
sodium were determined by flame photometry in the ash hydrochloric solution, using lithium
solution as an ionization buffer (AOAC 2000c). Vitamin A, expressed as mg of all-trans retinal,
was determined by high performance liquid chromatography with normal phase and fluorescence
detector (EN 12823-1) (CEN 2000a). Vitamin B1 (thiamine) was analysed by high performance
liquid chromatography with reverse phase, pre-column reaction and fluorescence detection
(EN 14122) (CEN 2003a). Vitamin B2 (riboflavin) was assayed by high performance liquid
chromatography with reverse phase and fluorescence detector (EN 14152) (CEN 2003b). Vitamin
B6 (free and phosphorilated vitamers) was determined by ion-pair high performance liquid
chromatography with fluorescence detection (EN 14164) (CEN 2002). Vitamin B12, expressed as
mg of cobalamin, was calculated by turbidimetry in microplates, using Lactobacillus delbruekii
(ATCC 7830), following an internal procedure, based on an AOAC (2000d) procedure. Vitamin D,
expressed as mg of cholecalciferol, was determined according to EN 12821 (CEN 2000b), which
includes the vitamin purification and concentration by semi-preparative high performance
liquid chromatography, with normal phase. Identification and quantification of vitamins
D3 (cholecalciferol) by high performance liquid chromatography with reverse phase and UV
detector, using the vitamin D2 (ergocalciferol) as an internal standard. Vitamin E, expressed
as mg of dl-a-tocopherol, was calculated by high performance liquid chromatography with
normal phase and fluorescence detector (EN 12822) (CEN 2000c). Folates, expressed as mg
of folic acid, were determined by turbidimetry in microplates, using Lactobacillus casei, ssp.
rhamnosus (ATCC 7469), based on EN 14131 (CEN 2003c). Niacin, expressed as mg of nicotinic
acid, was determined by high performance liquid chromatography with reverse phase, with post-
column reaction and fluorescence detection (CEN/TC 275/WG 9 N145 document). All results are
expressed on a basis of 100 g of edible part.
Chemical composition and nutritional value

As it can be seen in Table 1, all studied species are a good source of protein, which accounts
for a significant percentage of the energy content. The protein range was between 12 and 22%,
being the lowest and highest values registered for grooved carpet shell and Norway lobster,
respectively. Moreover, all of them were well balanced in respect to all of the nine essential
amino acids (data not shown) (Bandarra and others 2004). The dominant amino acid in all
species was glutamic acid, followed by alanine and lysine. In average, the relation between
essential and total amino acid contents was between 41 and 45%, values quite close to that
found in the egg protein (≅43%). In most species the limiting amino acid was methionine. The
pattern was very similar to that found in other high protein foods, such as beef and chicken.
As it has been extensively referred, fat content is strongly influenced by the species, the time
of the year and what the fish feeds on. Most of the species consumed in Portugal contained
less than 2.5% of total fat and only a small number has more than 15% (e.g., Atlantic salmon,
European eel and summer sardine). Significant differences were found not only among species
but also in the same species analysed in different seasons (specially in the case of fatty
species). In all species an inverse correlation between fat and moisture contents was found.
The proportions and amounts of saturated (SAT), monounsaturated (MUFA), polyunsaturated

(PUFA) and omega-3 (ω3) fatty acids varied considerably from one species to another, and
in farmed fish species is too much dependent on the diet used on feeding. As a rule, the
binnenwerk.indd 479 21-08-2006 12:35:58

Table 1. Nutritional data of molluscs, crustaceans and fish products consumed in Portugal
(per 100 g of edible part).
Nutritional data Common Grooved Norway

cockle carpet shell Common octopus European squid lobster
(/100 g) Raw Raw Raw Boiled Raw Grilled Raw
Energetic value 93/387 58/243 77/324 117/488 76/316 153/641 94/392

(kcal/kJ)
Edible part (%) 21 27 74 — 69 — 23
Protein (g) 14.7 11.7 15.6 23.7 15.8 32.5 20.9
Total fat (g) 3.3 0.9 1.2 1.3 0.9 1.6 0.5
16:0 (mg) 400 138 177 190 183 340 60
Total Sat (mg) 678 224 266 283 242 497 89
18:1 (mg) 58 25 41 44 17 45 67
Total MUFA (mg) 610 120 90 97 58 121 88
18:2 ω6 (mg) 6 5 5 0.9 1.7 8.8 3.9
20:5 ω3 (mg) 594 59 197 211 111 112 57
22:6 ω3 (mg) 215 55 225 239 242 417 77
Total PUFA (mg) 1195 256 560 591 369 672 155
Total PUFA ω3 (mg) 1025 190 497 526 362 559 139
Total PUFA ω6 (mg) 170 66 63 66 1.7 113 16
Cholesterol (mg) 30 44 64 105 140 n.a. 68
Calcium (mg) 79 51 13 26 18 28 72
Phosphorus (mg) 223 178 165 185 261 243 216
Magnesium (mg) 81 103 43 49 49 49 41
Iron (mg) 6 9 0.7 0.5 0.3 1 0.4
Sodium (mg) 526 244 259 178 196 1510 444
Potassium (mg) 87 78 236 164 225 73 413
Manganese (mg) 0.6 0.7 <0.02 0.04 <0.02 0.06 0.1
Copper (mg) 0.7 0.2 0.2 0.5 <0.03 0.7 2.5
Zinc (mg) 1.5 2.1 1.3 2.4 1.0 0.7 4.5
Chloride (mg) 808 347 438 258 314 2065 n.a.
Vit. A (μg) n.a. n.a. 2.7 6.7 9.6 n.a. 8.3
Vit. E (mg) n.a. n.a. 0.73 2.1 1.2 1.8 2.2
Vit. D (μg) n.a. n.a. 0 0 3.5 <0.70 n.a.
Vit. B1 (mg) n.a. n.a. 0.02 <0.02 0.07 0.04 n.a.
Vit. B2 (mg) n.a. n.a. 0.04 0.04 0.02 0.03 n.a.
Vit. B6 (mg) n.a. n.a. 0.07 0.05 0.05 0.1 0.1
Vit. B12 (μg) n.a. n.a. 1.3 1.7 1.1 n.a. 1.4
Folate (μg) n.a. n.a. 12 13 7.1 8.8 13.5
Niacine (mg) n.a. n.a. 1.3 2.5 1.0 0.63 2.2
n.a. not analysed

— not determined
binnenwerk.indd 480 21-08-2006 12:36:00

Blackspot
Desalted Atlantic cod Atlantic salmon Black scabbardfish seabream
Raw Boiled Grilled Raw Boiled Grilled Raw Fried Grilled Raw
85/355 113/472 131/547 267/1116 279/1166 315/1320 92/386 253/1059 116/487 105/438
79 — — 89 — — 58.4 — — 44.3
19 26.2 30.2 16.2 20.7 23.8 15.7 24.2 20.5 18.8
0.4 0.1 0.2 21.9 21.1 23.7 2.8 16.6 3.2 2.7
51 10.8 27.3 2688 2450 2754 301 1114 358 424
68 14.5 35.3 4291 4049 4488 478 1846 630 693
38 5 10.7 3810 2450 2822 643 4388 811 518
53 13.1 27.8 10037 7825 8747 1622 4859 1697 761
3.7 1.6 1.9 691 604 695 20.4 8117 75 24.4
35 7.5 21.4 1172 1630 1800 48 61 57 58
86 18 42 1773 2326 2594 140 281 256 490
139 30 71 5148 6591 7359 235 8611 439 779
128 27 67 4326 5623 6255 211 358 351 704
11 3.3 5.1 766 968 1104 24 8252 88 74
52 n.a. n.a. 40 n.a. n.a. 24 n.a. n.a. 36
33 121 196 12 61 68 14 n.a. n.a. 7.3
116 103 169 209 216 322 181 n.a. n.a. 256
23 31 57 23 26 40 29 n.a. n.a. 32
0.2 0.6 0.5 0.5 0.3 0.4 0.1 n.a. n.a. 0.4
1483 1228 2245 38 148 783 138 n.a. n.a. 104
36 21 40.0 301 234 408 332 n.a. n.a. 369
<0.02 0.04 0.19 <0.02 0.02 0.04 <0.02 n.a. n.a. <0.02
<0.03 1.40 0.90 0.06 0.06 0.04 <0.03 n.a. n.a. 0.04
0.8 1.1 1.2 0.5 0.8 0.9 0.5 n.a. n.a. 0.7
1990 1749 2914 46 225 1125 176 n.a. n.a. 119
3.8 2.7 1.4 33 65 70 23 n.a. n.a. 19
0.3 0.3 0.34 4.0 5.3 4.3 1.1 n.a. n.a. 0.70
4.5 <0.70 <0.70 11 11 9.2 2.1 n.a. n.a. 15
0.05 0.02 0.05 0.18 0.17 0.19 <0.02 n.a. n.a. 0.05
0.07 0.09 0.06 0.04 0.08 0.12 0.04 n.a. n.a. 0.07
0.07 0.06 0.05 0.45 0.34 0.21 0.16 n.a. n.a. 0.4
0.95 n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a.
8.1 5.7 9.9 10 8.4 10 8.3 n.a. n.a. 15
0.8 0.28 0.48 n.a. 3.0 4.4 1.8 n.a. n.a. 2.6
binnenwerk.indd 481 21-08-2006 12:36:01

Table 1. Continued.
Nutritional data European European

eel European hake plaice Farmed gilthead seabream
(/100 g) Raw Raw Boiled Fried Raw Raw Boiled Grilled
Energetic value 307/1285 74/309 119/498 164/685 96/400 151/631 202/844 174/730
(kcal/kJ)
Edible part (%) 61.4 83.6 — — 55.7 45.9 — —
Protein (g) 13.4 17.0 20.1 21.7 19.0 17.8 20.8 25.2
Total fat (g) 27.7 0.7 3.7 7.1 1.6 8.3 12.5 7.4
16:0 (mg) 5553 90 549 508 160 1114 1658 974
Total Sat (mg) 8604 143 857 778 253 1674 2339 1466
18:1 (mg) 1155 55 332 1659 257 1990 2428 1722
Total MUFA (mg) 2369 110 650 1790 416 2867 3697 2484
18:2 ω6 (mg) 121 7.3 41 3445 4.7 1143 1250 975
20:5 ω3 (mg) 4428 66 371 92 157 236 646 232
22:6 ω3 (mg) 7270 155 980 259 153 790 1289 811
Total PUFA (mg) 13296 273 1645 3861 396 2763 4318 2561
Total PUFA ω3 (mg) 12254 247 1492 388 386 1468 2733 1438
Total PUFA ω6 (mg) 1042 26.6 153 3472 10.4 1295 1584 1123
Cholesterol (mg) 26 19 28 25 85 51 116 97
Calcium (mg) 138 15 29 54 196 15 30 65
Phosphorus (mg) 246 219 230 303 173 252 290 322
Magnesium (mg) 16 26 32 43 29 28 34 35
Iron (mg) 0.5 0.5 0.5 0.7 0.4 0.4 0.8 0.5
Sodium (mg) 77 69 169 1344 112 59 168 649
Potassium (mg) 179 408 373 595 234 383 367 494
Manganese (mg) <0.02 <0.02 <0.02 0.03 <0.02 <0.02 <0.02 <0.02
Copper (mg) 0.04 <0.03 0.03 <0.03 0.07 <0.03 0.05 0.05
Zinc (mg) 2.5 0.7 0.8 0.8 0.6 0.8 0.7 1
Chloride (mg) 108 85 195 1592 160 76 228 959
Vit. A (μg) 887 2.8 5.3 4.3 0.35 11 12 8.8
Vit. E (mg) 2.4 0.24 0.45 n.a. 0.4 0.82 0.32 0.16
Vit. D (μg) 16 5.6 5.2 7.0 10 12 8.4 7.9
Vit. B1 (mg) 0.28 0.02 0.02 0.04 0.12 0.2 0.23 0.23
Vit. B2 (mg) 0.26 0.04 0.04 0.07 0.05 0.78 0.11 0.12
Vit. B6 (mg) 0.15 n.a. n.a. n.a. 0.13 0.36 0.27 0.28
Vit. B12 (μg) n.a 0.6 0.36 0.83 1.5 4.8 n.a. 4.2
Folate (μg) 9.3 27 23 28 8.5 24 n.a. 29
Niacine (mg) 1.3 1.2 1.0 1.8 0.5 5.1 2.9 4.4
n.a. not analysed

— not determined
binnenwerk.indd 482 21-08-2006 12:36:03

Farmed seabass Horse mackerel Meagre Monk fish
Raw Boiled Grilled Raw Fried Grilled Raw Raw
116/486 188/787 216/902 116/484 195/816 154/643 100/417 78/327
38.2 — — 49.5 — — 74.4 34.9

19.8 21 23 18.6 22.7 23.9 20.4 17.9
7.9 10.9 13 2.2 9.3 2.8 1.4 0.2
970 1525 1756 351 978 448 235 25
1439 2229 2634 560 1548 705 342 38
1234 1885 1905 265 1796 320 144 17
2467 3496 4223 626 2541 757 465 34
599 786 803 25 2943 37 7.9 1.2
540 653 722 128 246 145 5.7 6.2
1104 1322 1213 363 754 501 147 38
2734 3443 3470 610 4129 794 234 56
2016 2487 2500 580 1177 751 216 48
718 956 970 30 2951 42 17.7 8.3
52 87 n.a. 22 39 48 50 42
52 74 116 69 86 61 13 6.7
234 273 438 263 292 285 233 207
38 34 46 33 37 35 31 27
0.4 0.4 0.4 1.2 2 2.1 0.3 0.2
95 244 561 80 525 454 56 86
346 333 427 403 465 468 430 333
0.02 0.03 0.08 <0.02 <0.02 <0.02 <0.02 <0.02
0.10 0.06 0.10 0.08 0.09 0.11 <0.03 <0.03
1.2 1.1 1.6 1.2 1.1 0.9 0.5 0.5
150 372 862 131 762 889 63 108
36 n.a. n.a. 15 8.2 n.a. 1.3 24
0.17 0.34 0.21 0.37 0.79 0.18 0.45 0.23
5.0 5.8 6.7 4.1 2.8 4.0 16.0 0
0.26 0.18 0.21 0.15 0.13 0.14 0.07 0.04
0.08 0.13 0.18 0.15 0.17 0.16 0.14 0.02
0.49 0.14 0.13 0.36 0.30 0.34 0.29 0.05
n.a. n.a. n.a. 5.7 6.6 6.4 n.a. n.a.
9.5 9.3 n.a. 18 24 24 12 7.3
2.7 2.6 3.5 5.0 5.5 5.9 2.7 2.0
binnenwerk.indd 483 21-08-2006 12:36:04

Table 1. Continued.
Nutritional data Sardine Silver scabbardfish Wreck fish
(/100 g) Raw Grilled Canned Raw Grilled Raw
Energetic value (kcal/kJ) 187/783 198/827 211/882 123/514 123/515 137/573

Edible part (%) 56.8 — — 59.4 — 38.2
Protein (g) 17.9 24.1 24.0 20.3 20.5 17.9
Total fat (g) 10.9 9.2 12.7 2.9 4.7 6.7
16:0 (mg) 1695 1488 1995 457 731 946
Total Sat (mg) 2746 2396 3001 739 1176 1376
18:1 (mg) 980 742 4376 527 861 1840
Total MUFA (mg) 2558 2070 5582 830 1350 2999
18:2 ω6 (mg) 105 86 424 37 71 47
20:5 ω3 (mg) 1672 1288 792 192 336 127
22:6 ω3 (mg) 1169 1334 1256 424 714 923
Total PUFA (mg) 4071 3494 2806 848 1427 1527
Total PUFA ω3 (mg) 3753 3246 2308 748 1244 1415
Total PUFA ω6 (mg) 318 248 498 101 183 112
Cholesterol (mg) 28 38 n.a. 38 n.a. 47
Calcium (mg) 70 67 445 16 11 8.1
Phosphorus (mg) 296 307 637 183 234 165
Magnesium (mg) 29 35 42 26 31 27
Iron (mg) 1.7 1.9 3.0 0.4 0.6 0.4
Sodium (mg) 65 390 187 77 816 108
Potassium (mg) 404 496 369 252 352 290
Manganese (mg) <0.02 <0.02 0.21 <0.02 <0.02 <0.02
Copper (mg) <0.03 0.11 0.15 <0.03 0.03 0.04
Zinc (mg) 1.7 1.2 2.5 0.6 1 0.7
Chloride (mg) 152 740 327 109 1164 134
Vit. A (μg) 12 9 9.0 17 13 55
Vit. E (mg) 0.03 0.7 1.5 1.2 6.9 2.6
Vit. D (μg) 17 11 8.8 n.a. n.a. 6.6
Vit. B1 (mg) 0.02 0.05 <0.02 0.05 0.05 0.04
Vit. B2 (mg) 0.14 0.19 0.04 0.06 0.08 0.03
Vit. B6 (mg) 0.41 0.30 0.1 0.19 0.07 0.24
Vit. B12 (μg) 10 9.3 n.a. 2.2 2.2 0.7
Folate (μg) 24 31 21 30 26 6.0
Niacine (mg) 6.2 8.4 6.0 3.6 3.6 1.8
n.a. not analysed

— not determined
binnenwerk.indd 484 21-08-2006 12:36:04

fattiest species contained more ω3 fatty acids than the leaner; the amount of saturated fatty
acids in terms of percentage was almost constant and the most relevant ω3 fatty acids were
eicosapentaenoic (20:5ω3) and docosahexaenoic (22:6ω3). In most of the species indicated in
Table 1, PUFA were the dominant group, however there were some exceptions, namely meagre
and silver and black scabbardfish where the content of MUFA was higher than that of PUFA.
Generally, the palmitic acid was the most relevant within the saturated group, oleic acid was the
dominant monounsaturated and, among the PUFA, 20:5ω3 and 22:6ω3 presented the highest
amounts, although the latter was very often the most abundant. Regarding omega-3/omega-6
fatty acids ratio, the values were quite diverse as 1.1 in farmed gilthead seabream and 11.8
in European eel. All studied species had higher levels of PUFA and lower levels of SAT than
meat from mammals. The usual percentage of SAT, MUFA and PUFA found in beef and pork are,
approximately 40-45%, 50%, and 4-10%, respectively, whereas in chicken fall between those
of fish and beef and pork (30-35% SAT, 35-40% MONO and 25-30% PUFA).
The cholesterol levels of fish and shellfish species were between 24 and 85 mg/100 g (Table 1).
Cephalopods contained somewhat higher amounts, respectively 64 and 140 mg/100 g in
common octopus and European squid, but, lower than the levels very often referred for this
group of species. However, the presence of high contents of taurine in these species contributes
to reduce the cholesterol absorption (Militante and Lombardini 2004).
Concerning minerals, the average content in edible part was very close to 1 g/100 g. For most
fish species (Table 1) the order in terms of amounts was: potassium> chloride or phosphorus>
sodium> magnesium> calcium> iron> zinc> copper> manganese. The exception of this order
was desalted cod for which sodium and chloride were the most abundant. In the case of bivalve
molluscs and Norway lobster the order was different from that of fish species. The highest
potassium contents were registered in Norway lobster, European hake, horse mackerel and
sardine.
Fish products usually are not considered an interesting source of vitamins except vitamins A and
D, however, the levels of vitamin B are comparable to those of other foods with high protein
content and some fatty species could be a reasonable source of vitamins A and D. Within the
studied species (Table 1) the levels of vitamins B6 and B12 in sardine and horse mackerel were
considerable, respectively 0.41 and 0.36 mg/100 g and 10 and 5.7 µg/100 g. Besides, the
content of vitamins A and D in European eel, respectively 887 and 16 µg/100 g, as well as the
level of vitamin E in salmon (4 µg/100 g) were also relevant.
Effect of cooking
The type of cooking method may affect some chemical and nutritional components while other
remain unaffected. Moisture content usually decreased during the cooking process and the
size, thickness and the fish species also influenced such decrease. Consequently, the proportion
of solids increase and the amounts of certain nutrients could be higher in cooked relatively
to raw products. In the Table 1 the influence of some cooking processes in several species is
shown. Usually, frying led to a higher water loss associated to the absorption of oil resulting
in an increase of fat content. Nevertheless, the oil absorption seems to be higher when the fat
content of the product is lower. The fatty acid profile in fried products was quite influenced by
the composition of the vegetable oil used, as it can be observed, for instance, for fried hake,
where a considerable increase of MUFA and PUFA ω6 is noticeable as a result of the absorption
of oleic and linoleic acids, respectively. Relatively to minerals and vitamins there was not a
binnenwerk.indd 485 21-08-2006 12:36:05

common trend, even so the salt added before cooking has to be considered when analysing
the mineral content; some vitamins seemed to be destroyed while the content of others was
not significantly changed. Among the three cooking processes, boiling and grilling were more
satisfactory in terms of nutrient keeping.
Conclusion
There are many nutritional benefits associated with the consumption of the species studied.
Thus, most of them provide a protein with high quality, and one portion of 150 g supply 50-60%
of daily protein needs for an adult; they are low in terms of fat content, and the majority of such
fat is polyunsaturated with high levels of omega-3 fatty acids; the cholesterol content, with the
exception of cephalopods, is quite low; and the minerals and vitamins in terms of distribution
and concentration may contribute to the wholesomeness of fish as food. On the other hand,
the data obtained after the different culinary processes could help to understand the changes
occurring during cooking and fulfil a lack of information since most of the available data for
chemical and nutritional composition of fish products is concerned to uncooked products.
Among the studied species, only boiled and grilled Atlantic cod and grilled scabbard fish
presented a level of EPA (eicosapentaenoic acid) and DHA (docosahexaenoic acid) below the
recommended value for cardiovascular disease protection (ISSFAL 2004), considering portions
of 150 g.
Acknowledgements
Part of this work was supported by the project POCTI 2/2.1/MAR/1747/95 financed by Fundação
para a Ciência e Tecnologia (Portugal).
References
AOAC. 1998. Association of Official Analytical Chemists. Nitrogen (total) in Seafood (940.25). In: Official Methods of
Analysis, Association of Official Analytical Chemists. 16th Edition, 4th rev. Washington, DC.
AOAC. 2000a. Association of Official Analytical Chemists. Minerals in Infant Formula, Enteral Products and Pet Foods
AOAC Official Method 985.35 In: Official methods of analysis of AOAC International, 17th Edition. Washington,
DC. cp. 50, pp 15-17.
AOAC. 2000b. Association of Official Analytical Chemists. Determination of Lead, Cadmium, Copper, Iron and Zinc in
Foods AOAC Official Method 999.11 In: Official Methods of Analysis, Association of Official Analytical Chemists.
17th Edition. Washington, DC. cp. 9, pp 19-22.
AOAC. 2000c. Association of Official Analytical Chemists. Sodium and Potassium in Seafood AOAC Official Method
969.23 In: Official Methods of Analysis, Association of Official Analytical Chemists. 17th Edition. Washington,
DC. 35, pp 10-11.
AOAC. 2000d. Association of Official Analytical Chemists. Cobalamin (Vitamin B12 Activity) AOAC Official Method
952.20 In: Official Methods of Analysis, Association of Official Analytical Chemists. 17th Edition. Washington, DC.
cp. 45, pp 47-48.
Bandarra NM, Calhau MA, Oliveira L, Ramos M, Dias MG, Bártolo H, Faria MR, Fonseca MC, Gonçalves J, Batista I, Nunes
ML. 2004. Composição e valor nutricional dos produtos da pesca mais consumidos em Portugal. Publicações Avulsas
do IPIMAR nº 11, 103 p.
Bligh EG, Dyer W J. 1959. A rapid method of total lipid extraction and purification. Can J Biochem Physiol 37:911–917.
binnenwerk.indd 486 21-08-2006 12:36:05

FAO. 1989. Yield and nutritional value of the commercially more important fish species. FAO Fisheries Technical Papers
309, 187 p.
CEN. 2000a EN 12823-1 - Foodstuffs - Determination of vitamin A by high performance liquid chromatography - Part
1: Measurements of all-trans-retinol and 13-cis-retinol.
CEN. 2000b. EN 12821 - Foodstuffs - Determination of vitamin D by high performance liquid chromatography -
Measurement of cholecalciferol (D3) and ergocalciferol (D2).
CEN. 2000c. EN 12822 - Foodstuffs - Determination of vitamin E by high performance liquid chromatography -
Measurement of alpha-, beta-, gamma-, and delta-tocopherols.
CEN. 2002. EN 14164 - Foodstuffs - Determination of vitamin B6 by HPLC.
CEN. 2003a. EN 14122 - Foodstuffs - Determination of vitamin B1 by HPLC.
CEN. 2003b. EN 14152 - Foodstuffs - Determination of vitamin B2 by HPLC.
CEN. 2003c. EN 14131 - Foodstuffs - Determination of folate by microbiological assay.
CEN/TC 275/WG 9 N145 document - Determination of niacine. (in press)
Cohen Z, Vonshak A, Richmond A. 1988. Effect of environmental conditions on fatty acid composition of the red algae
Porphyridium cruentum: correlation to growth rate. J Phycol 24:328-332.
ISSFLA. 2004. Recommendations for intake of polyunsaturated fatty acids in healthy adults. International Society for
the Study of Fatty Acids and Lipids, 22p.
IPQ. 1988. NP 2032 – Pescado: Determinação do teor de cinza.
IPQ. 1988. NP 2929 – Pescado: Determinação do teor de cloreto.
IPQ. 1991. NP 2282 – Pescado: Determinação da humidade. Processo de referência.
IPQ. 1992. NP 1972 – Pescado: Determinação do teor de matéria gorda livre.
Lepage G, Roy CC. 1986. Direct transesterification of all classes of lipids in one-step reaction. J Lipid Res 27:114–119.
Militante JD, Lombardini JB. 2004. Dietary taurine supplementation: hypolipidemic and antiatherogenic effects. Nutr
Res 24:787-801.
Naemmi ED, Ahmad N, Al-sharrah TKM, Behbahani M. 1995. Rapid and simple method for determination of cholesterol
in processed food. J AOAC Intern 78:1522–1525.
Oehlenschläger J. 2000. Cholesterol content in edible part of marine fatty pelagic fish species and other seafood. In:
S.A. Georgakis, Editors. Proceedings of 29th WEFTA Meeting. 10–14 October 1999, Greek Society of Food Hygienists
and Technologist, Pieria, Greece. p 107–115.
Weihrauch JL, Posati LP, Anderson BA. Exler J. 1977. Lipid conversion factors for calculating fatty acids contents in
foods. JAOCS 54:36-40.
binnenwerk.indd 487 21-08-2006 12:36:05

binnenwerk.indd 488 21-08-2006 12:36:05
Polychlorinated biphenyls and organochlor pesticides in brown
shrimp (Crangon crangon) of the Belgian continental shelf
Marc Raemaekers1, Sabrine Derveaux2 and Koen Parmentier2

1FAVV – Federal Food Safety Laboratory, Braemkasteelstraat 59, B-9050 Gentbrugge, Belgium
2CLO - Gent, Departement Zeevisserij, Ankerstraat 1, B-8400, Oostende, Belgium
Abstract
This paper reports the concentrations and patterns of Polychlorinated Biphenyls (PCBs) and
organochlor pesticides in brown shrimp from the Belgian part of the southern North Sea in the
period 1994 - 2004. No significant trends could be found for any of the contaminants analysed,
when expressed on lipid weight basis. The median content of the sum of 7 marker CBs (IUPAC
Nos 28, 52, 101, 118, 138, 153 and 180) was 220 µg kg-1 of extracted lipid, i.e. 3.4 µg kg-1 of
cooked tail muscle during this 11 year period. When expressed on wet weight basis however,
a significant decreasing trend was found for the sum of 7 marker CBs. The PCB pattern did
change significantly for the CB congeners 105, 138 and 180 with respect to CB congener 101
in brown shrimp. A redistribution of CB congeners in the marine environment of the Belgian
continental shelf confirms the world-wide transportation and deposition model and seems to
indicate a steady-state of PCB input in the Belgian part of the southern North Sea. Aldrin,
endrin and trans-nonachlor could hardly be detected. HCB and the hexachloro cyclohexane
isomers α-HCH and lindane were close to their limits of quantification (LOQs). The effective ban
of pp’-DDT was illustrated since it was mainly found as its degradation products pp’-DDE and
pp’-DDD. Hexachlorocyclohexane isomers behave similar as pp’-DDE and dieldrin in the marine
environment, but might not be as persistent as the PCBs.
Keywords: monitoring, PCB, organochlor pesticides, Crangon crangon
Introduction
Polychlorinated biphenyls (PCBs) have been a major cause for concern since their discovery in
the environment. Between 1930 and 1983, companies in the US, Japan and Europe manufactured
large amounts of technical mixtures of PCBs (De Voogt and Brinkman 1989). During this period
large quantities of PCBs reached the environment through disposal, leakage, evaporation,
accidents, etc. (Pearsson 1982).
Organochlor pesticides (dieldrin, aldrin, endrin, DDT and degradation products, lindane, alpha-
HCH, hexachlorobenzene (HCB), trans-nonachlor) are subject of a parallel story. These highly
hydrophobic substances are mainly associated with sediment and fat tissue. Their widespread
use in agriculture makes them ubiquitous in the environment. In most cases, there is clear
negative gradient from the production or application site towards more remote areas. Their
introduction into the marine environment mainly originates from runoff, and rivers are the
main transport routes.
binnenwerk.indd 489 21-08-2006 12:36:05

The widespread distribution of these contaminants in the marine environment and their
persistence raised questions about the hazards posed to the marine ecosystem. This was
recognised by international marine organisations such as the Oslo and Paris Commissions
(OSPARCOM). As a result, PCBs and a number of organochlor pesticides, considered as hazardous
substances, became routinely monitored determinants in the Joint Monitoring Programme
of OSPARCOM (North Sea Task Force 1993) and in its follow up, the Joint Assessment and
Monitoring Programme (JAMP).
Recommendations for protecting the marine ecosystem imply the continuously reduction of
discharges, emissions and loss of hazardous substances, with the ultimate aim of achieving
concentrations in the marine environment close to background values for naturally occurring
substances and close to zero for xenobiotics.
Since the early eighties, PCBs and organochlor pesticides have been routinely monitored in
marine organisms and sediments. Analysis is mostly done by capillary gas chromatography (GC)
with electron-capture (ECD) or mass-spectrometric (MS) detectors.
In the past two decades, a reduction of contaminant concentrations in the marine system
similar to that achieved for emissions, discharges and losses has not been observed (OSPAR
2000a) The general absence of decreasing trends might originate from the fact that most time
series are still too short to reveal reliable information on trends, from a high natural variability
of contaminant levels, or from an insufficient sampling frequency.
Since the eighties, 10 marker CBs (UPAC Nos. 28, 31, 52, 101, 105, 118, 138, 153, 156 and 180)
and 10 organochlor pesticides (HCB, lindane, α-HCH, dieldrin, aldrin, endrin, transnonachlor, pp’-
DDT, pp’-DDD and pp’-DDE) have been monitored in marine organisms of the Belgian continental
shelf (BCS) by the laboratory of organic contaminants of the Sea Fisheries Department. This
paper will deal with the concentrations of PCBs and organochlor pesticides in brown shrimp
(Crangon crangon) sampled in the period 1994 – 2004.
Samples of brown shrimp (Crangon crangon) are annually taken in the month of September
using beam trawling at more than 20 sampling stations on the Belgian continental shelf by the
fishing vessel “O.29 Broodwinner”. Shrimps are cooked on board in seawater and stored frozen.
In the laboratory, the frozen samples are pooled and five sub samples comprising more than 100
individuals are taken. Before analysis, shrimps are peeled and the tail muscle is cut by knife into
3 to 4 pieces. The extraction of the lipids is performed using chloroform and methanol according
to the method of Bligh and Dyer (1959). The lipid extract is cleaned-up on a deactivated
aluminium oxide column, eluted with hexane. The eluate is evaporated by a nitrogen stream to 2
ml volume and fractionated on a deactivated silica column, eluted with hexane (fraction 1) and
hexane - diethyl ether (90/10, v/v) (fraction 2), respectively. The two fractions are evaporated
and an internal standard (tetrachloronaphthalene, 100 mg kg-1) is added. Quantification is
carried out by a GC-system equipped with an electron capture detector. As such, 10 marker CB
congeners (IUPAC Nos. 28, 31, 52, 101, 105, 118, 138, 153, 156 and 180) and 10 organochlor
pesticides (HCB, α-HCH, lindane, transnonachlor, dieldrin, aldrin, endrin, pp’-DDT, pp’-DDD and
pp’-DDE) are analysed. The participation at international laboratory proficiency tests organized
by QUASIMEME is an important part of the quality assurance programme. Contents of PCBs and
binnenwerk.indd 490 21-08-2006 12:36:05

Polychlorinated biphenyls and organochlor pesticides in brown shrimp
organochlor pesticides are expressed as microgram per kilogram of cooked tail muscle. Results
obtained between 1994 and 2004 are statistically evaluated and discussed. Trends were analysed
using the trend-analysis software Trend-y-tector (http://www.trendytector.nl), based on Mann-
Kendall statistics. After log-transformation of the average concentrations or concentration
ratios, tests were carried out two-sided with a significance of 10% and a power to detect a
trend of 90%. Mann–Kendall is a straightforward and robust method in detecting monotonic
(upward or downward) trends, and is largely unaffected by isolated extreme values. As such it
is recommended by the ICES Working Group on Statistical Aspects of Environmental Monitoring
(WGSAEM) (OSPAR 2000b). CB patterns were analysed with linear correlation analysis using the
Statistica (Statistica 2000) software and considered to be significant if p < 0.05.
Concentrations and trends

An overview of the concentrations of PCB and organochlor pesticides in brown shrimp from
the Belgian continental shelf is given in Table 1. The median content of the sum of 7 marker
CBs was 220 µg kg-1 of extracted lipid, i.e. 3.4 µg kg-1 of cooked tail muscle in this 11 year
period. This value is far below the Belgian threshold of 75 µg kg-1 which is applicable to fishery
products and derivatives aimed for consumption (Anon 2002).
Aldrin, endrin and trans-nonachlor could hardly be detected and never exceeded the limit
of quantification (LOQ). The LOQ was derived from a signal-to-noise ratio of 6. In the cases
where n is lower than 11, concentrations were lower than the LOQ for some years. Other
compounds like HCB and the hexachlorocyclohexane isomers α-HCH and lindane were close to
their LOQs. Furthermore, pp’-DDT was mainly found as its degradation products pp’-DDE and
pp’-DDD. The differences between the average values and the medians are in agreement with
other reported monitoring data, indicating the non-Gaussian distribution of contaminants in
the marine ecosystem.
Compounds exceeding their respective LOQs for at least 7 years, were statistically assessed by
Mann-Kendall regression analysis. At a α-level of 0.10, no negative significant trend could be
detected for the organochlor pesticides and the separate CB congeners, when expressed on a
lipid weight basis. However, as for the organochlor pesticides all trends were negative, except
for HCB showing a slightly positive, non-significant trend (p = 0.28).
When the period 1999 – 2004 was considered, no significant trends could be detected either,
except for CB138. This trend is shown in Figure 1. The higher variability with respect to the
other CB congeners might hamper the detection of significant trends.
When expressed on a wet weight basis, a significant trend could be found for the sum of 7
marker CBs (UPAC Nos. 28, 52, 101, 118, 138, 153 and 180; Figure 2). This could be explained
by the variation in the lipid content. In the period 2002 – 2004, lipid contents were slightly
lower than before, and caused a steeper decrease in the wet weight related concentrations.
However, the average lipid content in brown shrimp in the period 1994–2004 amounted to 1.58
± 0.09% (standard error) and no significant difference between the period 1994 – 2001 and the
period 2002 – 2004 could be observed for the lipid content.
binnenwerk.indd 491 21-08-2006 12:36:06

120
Concentration (μgkg-1 lipid)

100
80
60
40
20
0
1992 1994 1996 1998 2000 2002 2004 2006
Year
Figure 1. Concentrations of CB138 expressed on lipid weight in brown shrimp from the Belgian continental shelf.
The curve indicates the significant decreasing trend during the last 6 years. Intervals show standard errors.
Concentration (μgkg-1 wet weight)
8
7
6
5
4
3
2
1
0
1992 1994 1996 1998 2000 2002 2004 2006
Year
Figure 2. Mean concentrations and standard errors of sum of 7 marker CBs in brown shrimp sampled on
the Belgian continental shelf. The curved line shows the significant decreasing trend (p < 0.05) found by
Mann-Kendall analysis.
Since lipid based concentrations in biota are accepted to reflect the environmental contaminant
levels, rather than wet weight based concentrations, it was concluded that the quality of the
marine environment of the Belgian continental shelf has not significantly improved during the
last 10 years, as far the persistent organochlor pollutants discussed in this paper are considered.
Moreover, this observation is in agreement with Roose and others (2005) who were not able to
find evidence for decreasing trends of PCBs in the sediment of the same study area.
From Figures 1 and 2 it appears that the standard errors in the former period are larger than
in the latter period. This observation coincides with the improvement of the sub sampling
procedure. After collection by the fishing vessel Broodwinner, the samples coming from a
great number of stations (more than 20) on the Belgian continental shelf, are in the last 5
year period well-mixed in the laboratory, prior to taking sub samples for the extraction and
binnenwerk.indd 492 21-08-2006 12:36:06

analysis of contaminants. The mixing is carried out manually, after partial defrosting of the
cooked and frozen shrimps, in a 20 litre container. As such, the individual shrimps could be
properly randomised. 5 sub samples consisting of 10 grams each of peeled muscle tissue were
analysed annually.
Correlations and relative distributions

In the case 7 or more observations were available, i.e. for values greater than the LOQ, correlations
were calculated (Table 3). It appeared that, except for HCB, all compounds correlated positively.
This observation agrees with the slightly positive trend of HCB concentrations. pp’-DDE and
dieldrin showed significant trends most frequently. They even significantly correlated with the
sum of 7 marker CBs. This is an indication of similar behaviour in the environment and/or in the
shrimp metabolism. As such, pp’-DDE, dieldrin and the marker CBs might be characterized by a
similar degree of non-degradability. This is an indication that pp’-DDD, the other degradation
product of pp’-DDT, might be less stable in the marine environment than dieldrin and the PCBs, as
the correlation factors of the latter with pp’-DDD are not significant. The hexachlorocyclohexane
isomers correlate well with each other and with pp’-DDE and dieldrin respectively, but not
with the marker CBs. From this observation it was concluded that the hexachlorocyclohexane
isomers behave similar as pp’-DDE and dieldrin in the marine environment, but might not be
as persistent as the PCBs.
Correlation factors among individual CB congeners were calculated as well. Almost all congeners
correlated positively with each other. CB118 was the only congener that correlated significantly
and positively with all other congeners, r2 values ranging from 0.31 (CB28) to 0.82 (CB138).
CB28 and CB31 correlated significantly with each other (r2 = 0.96) and with CB52, but did not
significantly correlate with CB105, CB138 and CB153, CB156 and CB180. These low correlations
reflect the different chemical behaviour in the marine environment, and/or the shrimp
metabolism. CB28, CB31 and to a lesser extent CB52, are much more volatile than the higher
chlorinated congeners. As such these lower chlorinated congeners may disappear more rapidly
from the aqueous phase to the atmosphere, where they are transported and deposited into
colder areas, such as the arctic and Antarctic Polar Regions. As a result, persistent organochlor
pollutants disappear from areas with relative warm climatological conditions, and are enriched
in the colder regions of our planet (OSPAR 2001).
In earlier papers, CB patterns were studied in sediments from the present study area (Delbeke and
others 1990; Vyncke and others 1998; Roose and others 2005) and the various CBs significantly
correlated as well with each other. However, it seemed that the correlations in sediment
were slightly better than in brown shrimp. In order to evaluate these correlations in depth,
CB patterns were calculated as their relative concentration versus the CB101 concentration
(Figure 3). The CB patterns found in brown shrimp reflect rather well the CB patterns found in
the sediment of that area. CB153 and CB138 were most abundant in both types of samples.
Compared with sediment patterns, CB156 and CB180 were relatively more abundant in the
shrimp patterns. On the contrary, CB101 appeared to be about twice less prevalent in shrimp
than in sediment. These differences must be attributed to differences in bio-availability, uptake
and degradation metabolism in the shrimp organism. Higher chlorinated congeners would be
slower degraded than less chlorinated congeners, resulting in their enrichment in the lipid
phase of the shrimp organism.
binnenwerk.indd 493 21-08-2006 12:36:06

4.5
Relative congener ratio

4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
CB28 CB31 CB52 CB118 CB153 CB105 CB138 CB156 CB180
CB congener
Figure 3. Averaged CBx/CB101 congener ratios in brown shrimp from the Belgian continental shelf in the
period 1994-2004. Intervals indicate standard errors.
Roose and others (2005) did not find trends in CB patterns in sediment of the Belgian
continental shelf. In this study CB patterns (CBx/CB101) in brown shrimp were assessed for
significant trends by Mann-Kendall analysis. CBx/CB101 ratios for CB105, CB138 and CB180
showed positive significant (p < 0.10) trends (Figure 4). This observation clearly shows that
the CB patterns in brown shrimp have been changing during the last decade. The trend for the
CB153/CB101 ratio was positive, but insignificant. Since the CB101 concentration does not
CB105/CB101 ratio
CB138/CB101 ratio
0.8 5
0.7
0.6 4
0.5 3
0.4
0.3 2
0.2 1
0.1
0 0
1992 1994 1996 1998 2000 2002 2004 2006 1992 1994 1996 1998 2000 2002 2004 2006
Year Year
6 7
CB153/CB101 ratio
CB180/CB101 ratio
5 6
4 5
4
3
3
2 2
1 1
0 0
1992 1994 1996 1998 2000 2002 2004 2006 1992 1994 1996 1998 2000 2002 2004 2006
Year Year
Figure 4. PCB congener ratios in brown shrimp from the Belgian continental shelf. Curved lines show
significant trends.
binnenwerk.indd 494 21-08-2006 12:36:06

show a significant trend, the relative abundance of the higher chlorinated congeners tend to
increase. Since a dynamic equilibrium exists between sediment and biota, it is concluded that
the CB patterns in the marine environment of the Belgian continental shelf have changed in
the study period. It appears that lower chlorinated CB congeners were depleted and the higher
chlorinated congeners were enriched, supporting the atmospheric transportation and deposition
model. However, changing CB patterns could not be confirmed in sediment (Roose and others
2005). This may be explained by the much higher variation of CB congener concentrations
in sediment than in brown shrimp, due to both measurement uncertainty and variability of
sediment composition (grain size, organic matter, mineralogy) which is typical for the Belgian
continental shelf.
Conclusions
As lipid weight-based contaminant concentrations are most representative for the environmental
quality, it was concluded that in this 11 year period no significant decreasing trends of PCB and
organochlor pesticide concentrations were observed in brown shrimp. Concentrations of these
non-polar contaminants are very low in brown shrimp, partly due to their low lipid content
(1.6%) and therefore consumption of Belgian brown shrimps probably does not present a threat
to human health. When investigating the CB patterns and correlations between CB congeners,
it was concluded that the CB patterns in brown shrimp, and in the marine environment of the
southern North Sea in general have changed during the last decade. Assuming the atmospheric
transportation and deposition mechanism, being the largest force for redistribution of persistent
non-polar contaminants in the environment, higher chlorinated CB congeners are enriched and
lower chlorinated CB congeners are depleted in the study area. This environmental redistribution
of CB congeners would only be observable in the case CB inputs did not change in this period.
Indirectly, it can be concluded that there is an indication that the world-wide ban on the
manufacture and use of PCBs has measurable results on the Belgian continental shelf more than
20 years after its implementation.
Acknowledgements
The authors wish to thank Ides Dobbelaere, Nancy Tyberghein and Marc Van Ryckeghem who
were responsible for carrying out the analytical work in a dedicated way. Furthermore, the
authors are grateful to Patrick Roose and Kris Cooreman for their scientific and technical
supervision in the implementation of the marine environmental monitoring programme at the
Sea Fisheries Department – CLO-Gent.
References
Anon. 2002. Koninklijk besluit tot wijziging van het koninklijk besluit van 19 mei 2000 tot vaststelling van maximale
gehaltes aan dioxines en polygechloreerde bifenylen in sommige voedingsmiddelen –March 6, 2002. Belgisch
Staatsblad April 16,2002.
Bligh EG, Dyer WJ. 1959. A rapid method of total lipid extraction and purification for organic compounds. Can J
Biochem Physiol 37:911-917.
Delbeke K, Joiris CR, Bossicart M. 1990. Environ Pollut 66:325-349.
De Voogt P, Brinkman UATh. 1989. Topics in Environmental Health Vol 4. In: Kimbrough RD, Jensen AA, editors.
Amsterdam: Elsevier Science. p 3–43.
North Sea Task Force. 1993. North Sea Quality Status Report 1993. London: Oslo and Paris Commissions.
binnenwerk.indd 495 21-08-2006 12:36:06

OSPAR. 2000a. Quality Status Report 2000, Region II – Greater North Sea. London: Oslo and Paris Commissions.
OSPAR. 2000b. Trend Assessment Tools, SIME 00/5/4-E. London: Oslo and Paris Commissions.
Pearsson CR. 1982. The handbook of environmental chemistry Vol 3 Part B. In: Hutzinger O, editor. Berlin: Springer
Verlag. p 89–116.
Roose P, Raemaekers M, Cooreman K, Brinkman, UATh. 2005. Polychlorinated biphenyls in marine sediments from the
Southern North Sea and Scheldt estuary: a ten-year study of concentrations, patterns and trends. J Environ Monit
7:1-9.
Statistica. 2000. Data analysis software system, Version 5.5, http://www.statsoft.com, StatSoft Inc.
Vyncke W, Roose P, Hillewaert H, Guns M, Van Hoeyweghen P, Baeten H, Hoenig M. 1998. Trace metals and organochlorines
in sediments from the Western Scheldt (1990-1995), ASMO 98/4/Info.2-E. London: Oslo and Paris Commissions.
binnenwerk.indd 496 21-08-2006 12:36:07

Concentrations of mercury, lead and cadmium in bivalves from the
Portuguese coast
Helena Lourenço, Carmen Lima, Ana Oliveira, Susana Gonçalves, Claudia Afonso, Maria
Fernanda Martins and Maria Leonor Nunes
Abstract
The accumulation of heavy metals in bivalve tissues is due to their morphological and biological
characteristics associated to the habitat. In Portugal, these species are important in terms of
economic value and local consumption. The purpose of this work was to evaluate the levels
of some heavy metals which can be toxic for humans at elevated concentrations in bivalves
collected in 2004. Total mercury levels were lower than the maximum legal limit by EU (0.5
mg/kg wet weight). Regarding lead and cadmium levels, most samples did not exceeded the
legal maximum limits (1.5 mg/kg wet weight and 1.0 mg/kg wet weight, respectively). As
exceptions some furrow shell and Portuguese oyster samples from Tagus and Sado estuaries
could be identified.
Keywords: mercury, lead, cadmium, bivalve molluscs, production areas
Introduction
Contamination of the marine environment by heavy metals has risen in recent years due to
the global population increase and industrial development (Arellano and others 1999). This is
a serious problem due to their toxicity and their ability to accumulate in the biota (Islam and
Tanaka 2004). Metals like mercury (Hg), cadmium (Cd) or lead (Pb) seem not to participate in
any metabolic functions (Dallinger 1995; Suzuki and Suzuki 1996). These elements can provoke
changes in the central nervous, cardiovascular or respiratory systems. They may be mutagenic and
they may promote the risk of cancer, in both, animals and human beings (Vercruyse 1984).
Estuaries and near-shore marine waters can be considered a space of great importance for the
survival of a large variety of plants, animals and marine species (Castro and others 1999),
but they are very vulnerable to all kinds of influences (Blasco and others 1999). These zones
have become extensively degraded over the recent years (Chase and others 2001) since these
productive and sensitive areas are often directly and most seriously affected and exposed
due to their proximity of sources of pollution (Arellano and others 1999; Cohen and others
2001). The main sources of metals, resulting from various processes like underwater geothermal
activities, geological weathering, industrial processing of metals, use of metals and metal
components, leaching from dumps and fertilisers, atmospheric deposition, animal excretion
and the discharge of human sewage (Wright and Mason 1999). Consequently, a determination
of metal concentrations in organisms should be part of any assessment and monitoring program
in these risk zones.
binnenwerk.indd 497 21-08-2006 12:36:07

H.M. Lourenço, C. Lima, A. Oliveira, S. Gonçalves, C. Afonso, M.F. Martins and M.L. Nunes
Sedentary molluscs, like bivalves, are the animals most often used for screening of metal
contamination (Phillips 1977). They have the ability to accumulate and concentrate metals
found in their environment by several orders of magnitude above the background levels.
However, physiological changes, like life cycle, can play a role in the regulation mechanisms of
heavy metals in bivalves (Bebiano 1995).
In Portugal bivalve molluscs are much appreciated by consumers and there are many production
areas along estuaries and coastal zones. The most representative molluscs are some clam
species, like grooved carpet shell, carpet shell or European razor clam, common cockle and
mussel, oysters, furrow shell and wedge shell (Table 1)
The main aim of this study was to evaluate heavy metal contamination in whole body tissues of
several mollusc bivalves. Various estuaries and coastal zones in Portugal, exposed to different
metal load, were chosen and the concentrations of Hg, Cd and Pb were measured.
Material
Bivalves species used in this study were collected in 2004 at various production areas localised
along the Portuguese coastal zone and estuaries. Common name, scientific name, number of
analysed samples and production area are described in detail in Table 1. The collected molluscs
were immediately transferred to the laboratory at 2 ºC in isotherm conditions. They were not
depurated prior to processing. From each sampling, 20-30 bivalves were analysed. Bivalves were
Table 1. Species analysed (common and scientific name, number of samples analysed and production
area).
Common name Scientific name Samples Production area

analysed (n)
Grooved carpet shell Ruditapes decussatus 249 Ria Formosa Lagon; Alvôr Estuary;
Óbidos Lagoon
Carpet shell Venerupis Senegalensis 33 Tagus Estuary; Óbidos Lagoon; Aveiro
Estuary
Thick trough shell Spisula solida 16 Aveiro Estuary; Mouth of Douro River
Cockle Cerastoderma edule 54 Sado Estuary; Óbidos Lagoon; Aveiro
Estuary; Mouth of Lima River
Furrow shell Scrobicularia plana 29 Tagus estuary; Sado Estuary
European razor clam Solen marginatus 25 Sado Estuary; Óbidos Lagoon; Aveiro
Estuary
Mussel Mytilus edulis 105 Ria Formosa Lagoon; Mira Estuary; Tagus
Estuary; Óbidos Lagoon; Aveiro Estuary;
Mouth of Lima River
Portuguese oyster Crassostrea angulata 22 Mira Estuary; Sado Estuary; Aveiro
Estuary
Giant cupped oyster Crassostrea gigas 8 Alvôr Estuary
Wedge shell Donax trunculus 5 Sado Estuary
Smooth calista Callista chione 8 Sado Estuary
binnenwerk.indd 498 21-08-2006 12:36:07

Concentrations of mercury, lead and cadmium in bivalves from the Portuguese coast
washed to eliminate sediment and debris, using potable water and the excess water in their
mantle was drained. Stainless steel scalped blades were used to cut open the animals and to
remove the whole body tissue, which was homogenised by a food blender with stainless steel
cutters and stored in plastic bags at 5 ºC for some hours prior to analysis.
Analytical methods
Total Hg was determined by cold vapour atomic absorption spectrometry (CVAAS) (Bacharad MAS
50D) according to Lawson and Keikwood (1980). The levels of Pb and Cd were measured by flame
atomic absorption spectrometry (Varian, Spectr AA-20 with deuterium background correction)
in agreement with the procedures described by Jorhem and others (2000). All analyses were
carried out in duplicate, using the external calibration method. Tuna fish (CRM 463) and cod
muscle (CRM 422) certified standard reference materials from BCR (Bureau Communitaire de
Référence) were used to prove the accuracy of the methods. The determined values of Hg, Pb
and Cd were in good agreement with the certified values (Table 2). Detection limits (calculated
by residual standard deviation from linear regression) were: 0.01 mg kg-1 for Hg, 0.05 mg kg-1
for Pb and 0.01 mg kg-1 for Cd. All data were reported on mg kg-1 wet weight basis.
Metal concentrations in whole body tissues of bivalve molluscs studied are shown in Table 3.
Mercury levels analysed were always lower than 0.1 mg kg-1. The highest mercury concentrations
were found in carpet shell (0.08 mg kg-1) and furrow shell (0.07 mg kg-1) from Tagus Estuary
and in European razor clam (0.06 mg kg-1) from Aveiro Estuary, however, these results are far
below the legal limit of 0.5 mg kg-1 established by EU (2001). Several authors reported similar
values like Wright and Mason (1999) in the Stour estuary for mussel and cockle, de Mora and
others (2004) in different oysters species from the Gulf and Gulf of Oman and Usero and others
(1996, 1997, 2005) for striped venus (Chamelea gallina), grooved carpet shell or wedge shell
from the Atlantic coast of Southern Spain.
Some of the furrow shell samples sampled off Tagus Estuary exceeded the legal limit for lead
(1.5 mg kg-1), set for these molluscs (EU 2002). Similar Pb concentrations were also found in
furrow shell samples from the river Gualdalquivir estuary after the disaster in the Aznalcóllar
mine (Blasco and others, 1999). Highest values were observed by Ruiz and Saiz-Salinas (2000)
in the same bivalve from the Bilbao estuary, caused by the 1989-90 droughts and in Tagus
estuary by França and others (2005). Pb values registered in the other Portuguese production
areas were quite low, indicating a lower contamination by this metal in those areas.
Table 2. Mercury, lead and cadmium (mg kg-1) concentrations in certified standard reference materials
determined in the present study (n=5).
Certified standard reference material Hg Pb Cd
CRM 463 (European Com.) Certified 2.85 ± 0.16 - -

Found 2.88 ± 0.22 - -
CRM 422 (European Com.) Certified - 0.085 ± 0.015 0.017 ± 0.002
Found - 0.081 ± 0.002 0.019 ± 0.001
binnenwerk.indd 499 21-08-2006 12:36:07

Table 3. Concentrations of mercury, lead and cadmium (means ± S.D. and range in mg kg-1 wet weight) in
whole body tissues of bivalve molluscs.
Species Mercury Lead Cadmium
Grooved carpet shell 0.02 ± 0.00 0.1 ± 0.1 0.02 ± 0.01

(0.02-0.02) (< 0.1-0.2) (< 0.01-0.08)
Carpet shell 0.04 ± 0.02 0.2 ± 0.3 0.05 ± 0.02
(0.02-0.08) (< 0.1-1.3) (< 0.01-0.09)
Thick trough shell 0.02 ± 0.00 0.1 ± 0.1 0.06 ± 0.01
(0.01-0.02) (< 0.1-0.2) (0.04-0.08)
Smooth calista 0.03 ± 0.00 0.1 ± 0.1 0.06 ± 0.01
(0.03-0.03) (< 0.1-0.2) (0.04-0.07)
Furrow shell 0.04 ± 0.01 1.4 ± 1.1 0.03 ± 0.02
(0.03-0.07) (< 0.1-3.0) (0.01-0.07)
Cockle 0.02 ± 0.00 0.1 ± 0.1 0.03 ± 0.02
(0.02-0.03) (< 0.1-0.3) (0.01-0.07)
Wedge shell 0.03 ± 0.00 0.1 ± 0.1 0.04 ± 0.00
(0.03-0.04) (< 0.1-0.2) (0.03-0.05)
European razor clam 0.04 ± 0.00 0.2 ± 0.1 0.04 ± 0.02
(0.04-0.06) (<0.1-0.3) (0.02-0.08)
Mussel 0.02 ± 0.00 0.2 ± 0.1 0.13 ± 0.07
(0.02-0.03) (< 0.1-0.6) (< 0.01-0.35)
Portuguese oyster 0.04 ± 0.00 0.1 ± 0.1 0.37 ± 0.34
(0.04-0.05) (< 0.1-0.2) (0.06-1.09)
Giant cupped oyster 0.04 ± 0.00 0.1 ± 0.0 0.22 ± 0.06
(0.04-0.04) (0.1-0.1) (0.18-0.26)
The highest concentration of Cd was obtained in Sado Estuary, in Portuguese oyster samples,
exceeding the maximum legal limit of 1.0 mg kg-1 (EU 2001). Previous results obtained by
Blasco and others (1999) showed values in the range found in this work. Studies performed in
Crassostrea gigas presented different levels (Hunter and others 1995; Shulkin and others 2003).
The content of Cd in some of the mussel samples reached 0.35 mg kg-1 which is comparable
with values obtained by Mason and Wright (1999), Chase and others (2001), Zauke and others
(2003), but lower than that found by Liang and others (2004). The concentrations in the other
studied species were much lower than 1.0 mg kg-1.
Conclusions
The levels of Hg, Pb and Cd in whole body tissues of bivalves have been found to be generally
low. However, high concentrations of Pb and Cd occurred in bivalves from two areas, Tagus
estuary and Sado estuary.
binnenwerk.indd 500 21-08-2006 12:36:08

Concentrations of mercury, lead and cadmium in bivalves from the Portuguese coast
References
AOAC, 1998. Official Methods of Analysis. 16th Ed., 4th revision. Association of Official Analytical Chemistry. Washington,
D.C. (cd-rom).
Arellano JM, Ortiz JB, Capeta da Silva D, González de Canales ML, Sarasquete C, Blasco J. 1999. Levels of copper, zinc,
manganese and iron in two fish species from salt marshes of Cadiz Bay (Southwest Iberian Peninsula). Bol Instit
Esp Oceanog 15(1-4):485-352.
Bebiano MJ. 1995. Effects of pollutants in the Ria Formosa Lagoon, Portugal. Sci Total Environ 171:107-115.
Blasco J, Arias AM, Sáenz V. 1999. Heavy metals in organism of the river Gualquivir estuary: possible incidence of the
Aznalcóllar disaster. Sci Total Environ 242:249-259.
Castro H, Aguilera PA, Martinez, JL, Carrique, EL. 1999. Differentiation of clams from fishing areas an approximation
to coastal quality assessment. Environ Monit Assess 54:229-237.
Chase ME, Jones SH, Hennigar P, Sowless J, Harding GCH, Freeman K, Wells, PG, Krahforst C, Coombs K, Crawford R,
Pederson J. 2001. Gulfwatch: Monitoring spatial and temporal patterns of trace metal and organic contaminants
in the gulf of Maine (1991-1997) with the blue mussel, Mytilus edulis L. Mar Pollut Bull 42(6):491-505.
Cohen T, Hee S, Ambrose R. 2001. Trace metals in fish and invertebrates of three California coastal wetlands. Mar Pollut
Bull 42(3):232-242.
Dallinger R. 1995. Metabolism and toxicity of metals: metallothioneins and metal elimination. In: Cajaraville MP, editor.
Cell biology in environmental toxicology. Bilbao: p 171-190.
De Mora S, Fowler SW, Wyse E, Azemard S. 2004. Distribution of heavy metals in marine bivalves, fish and coastal
sediments in the Gulf and Gulf of Oman. Mar Poll Bull 49:410-424.
European Communities, 2001. Commission regulation (EC) No 466/2001 of 8 March 2001 setting maximum levels for
certain contaminants in foodstuffs. Off J Eur Comm L77:1-13.
European Communities, 2002. Commission regulation (EC) No 221/2002 of 6 February 2002 that it modifies the
Commission regulation (EC) No 466/2001 that setting maximum levels for certain contaminants in foodstuffs. Off
J Eur Comm L37:4-5.
França S, Vinagre C, Caçador I, Cabral, HN. 2005. Heavy metals concentrations in sediment, benthic invertebrates and
fish in three salt marsh areas subjected to different pollution loads in the Tagus Estuary (Portugal). Mar Poll Bull
50:998-1003.
Hunter CL, Stephenson MD, Tjeerdema RS, Crosby DG, Ichikawa GS, Goetzl JD, Paulson KS, Crane DB, Martin M, Newman
JW. 1995. Contaminants in oysters in Kaneohe bay, Hawaii. Mar Poll Bull 30(10):646-654.
Islam MD, Tanaka M. 2004. Impacts of pollution on coastal and marine ecosystems including coastal and marine
fisheries and approach for management: a review and synthesis. Mar Poll Bull 48:624-649.
Jorhem L. 2000. Determination of metals in foods by atomic absorption spectrometry after dry ashing: NMKL
collaborative study. JAOAC Int 83(5):1204-1211.
Lawwson DA, Keikwood DS. 1980. An automated method for the determination of total mercury in environmental
samples. ICES CM 1980/E:29.
Liang LN, He B, Jiang, GB, Chen, DY, Yao ZW. 2004. Evaluation of molluscs as biomonitors to investigate heavy metal
contaminations along the Chinese Bohai Sea. Sci Total Environ 324:105-113.
Phillips DJ. 1977. Biological indicator organisms monitor metal pollution. Environ Pollut 13:281-317.
Shulkin VM, Presley BJ, Kavun VIa. 2003. Metal concentration in mussel Crenomytilus grayanus and oyster Crassostrea
gigas in relation to contamination of ambient sediments. Environ Int 29:493-502.
Suzuki KT, Suzuki T. 1996. An introduction to clinical aspects of toxicity. In: Chang W, editor. Toxicology of metals.
Boca Raton: LCRC Lewis Publishers. p 333-335.
Ruiz JM, Saiz-Salinas JI. 2000. Extreme variation in the concentration of trace metals in sediments and bivalves from
the Bilbao estuary (Spain) caused by the 1989-90 drought. Mar Environ Res 49:307-317.
Usero J, González-Regalado E, Garcia I. 1996. Trace metals in the bivalve mollusc Chamelea gallina from the Atlantic
coast of southern Spain. Mar Pollut Bull 32(3):305-310.
binnenwerk.indd 501 21-08-2006 12:36:08

Usero J, González-Regalado E, Garcia I. 1997. Trace metals in the bivalve molluscs Ruditapes decussatus and Ruditapes
philippinarum from the Atlantic coast of southern Spain. Environ Int 23(3):291-298.
Usero J, Morillo, J, Garcia I. 2005. Heavy metals concentrations in molluscs from Atlantic coast of southern Spain.
Chemosphere 59:1175-1181.
Vercruysse A. 1984. Evaluation of analytical methods in biological systems. Part B: Hazardous metals in human
toxicology, Elsevier Science B. V., Amsterdam, 337p.
Wright P, Mason, CF. 1999. Spatial and seasonal variation in heavy metals in the sediments and biota of two adjacent
estuaries, the Orwell and the Stour, in Eastern England. Sci Total Environ 226:139-156.
Zauke GP, Clason B, Savinov VM, Savinova T. 2003. Heavy metals of inshore benthic invertebrates from the Barents
Sea. Sci Total Environ 306:99-110.
binnenwerk.indd 502 21-08-2006 12:36:08

Contaminant metals in cod products
Abstract
Total mercury, cadmium, lead, nickel and chromium concentrations were measured in different cod
samples. The metal contents are expressed in mg kg-1 wet weight. Mean mercury concentration
was 0.09±0.07, the maximum observed in salted and dried cod samples was 0.47. This value was
below the limit being set by EU regulation for this species (0.5). Mean cadmium levels also did
not reach the legal limit of the EU (0.05). Concerning lead levels, the EU legal limit (0.2) was
exceeded in some samples of strips of salted and dried cod samples. There are no legal limit for
nickel and chromium in the EU, however, the concentrations found in this study (about 0.06
and 0.4, respectively) did not seem to represent a risk for human consumption.
Keywords: mercury, lead, cadmium, nickel, chromium, cod
Introduction
Marine pollution by organic and inorganic chemicals has been identified as one of the most
important factor of poisoning for marine species, including fish (Al-Ghais 1995). Within these
compounds heavy metals can of utmost importance due to their capacity to be bio accumulated
in animal tissues during life time.
Accumulation of these chemicals in fish tissues depends on several endogenous factors such as
physiological condition, geographic habitat, fat content, adaptation capacity and the biotope
characteristics (Reichenbach-Klinke 1980).
Salted and dried cod products (bacalau) are of great importance in Portugal regarding their
economic value and role in the food industry. They are an important part of the national
gastronomy not only by tradition but also due to their unique sensory characteristics. Like
other fish species, cod and similar species, can accumulate some toxic heavy metals, especially
mercury (Hg) in muscle tissue, and cadmium (Cd) and lead (Pb) in liver and kidney, which can
affect the consumers health when consumed. Concerning the concentrations of nickel (Ni) and
chromium (Cr) in these products Portuguese industry needs information about the levels of
these elements.
The purpose of this work was to characterise the levels of five heavy metals – Hg, Cd, Pb, Ni
and Cr – in about 400 cod samples.
Material
In this study, about 400 cod samples (frozen, salted, salted and dried, soaked salted and dried,
and strips) were evaluated. Samples were collected from several retailers in Portugal since 2002,
binnenwerk.indd 503 21-08-2006 12:36:08

and after sampling, they were analysed within one week. At the time of metals evaluation, cod
samples were homogenised in a metal free blender and immediately analysed.
Analytical methods
Total mercury was determined by cold vapour atomic absorption spectrometry (CVAAS), using
a Bacharach Coleman Mas-50D Mercury Analyser System based on the method developed by
Hatch and Ott (1968) and described in detail by Joiris and others (1991). The accuracy was
tested using the Certified Reference Material CRM-463 (tuna fish muscle: 2.85 ± 0.16 mg kg-1
Hg dry weight) of the Community Bureau of Reference. The CRM-463, analysed in duplicate
(n=5), was in the range of the certified material (2.87 ± 0. 09 mg kg-1 Hg dry weight) with a
relative error of 0.1%.
Cadmium, lead, nickel and chromium analysis was based on the methods described by Jorhem
and others (2000). These metals were determined by atomic absorption spectrophotometric
method by flame atomisation in a Varian apparatus Spectr AA-20 with a background deuterium
correction. The accuracy for Cd and Pb was tested in an intercalibration exercise (z-score:
-0.3 for Cd and –1.1 Pb) organized by Central Science Laboratory (Metallic Contaminants –
FAPAS Series 7 Round 40). The accuracy for Cr was tested through the participating on a
proficiency study (Trace metals in fish tissue) carried out by Community Reference Laboratory
for Residues - Istituto Superiore di Sanità (Z-score: -0.72). The accuracy test for nickel is still
under determination. Detection limits (calculated by residual standard deviation from linear
regression) were: 0.01 mg kg-1 for Hg, Cd and Ni, 0.05 mg kg-1 for Pb and 0.1 mg kg-1 for Cr.
Analytical data are reported on a mg kg-1 wet weight basis.
The concentrations of three metals obtained for cod samples are summarised in Table 1.
The mercury mean level (0.09 ± 0.07 mg/kg-1 wet weight) was below concentrations reported
for the same species by other authors (Hellou and others 1992; Polak- Juszczak 1996, 1997;
Burger and Gochfeld 2005). Soaked salted and dried cod showed the highest mean level (0.11 ±
0.05 mg kg-1 wet weight) and the maximum content was verified in salted and dried cod (0.47
mg kg-1 wet weight). Nevertheless, all values found were below the EU limit for this species
(0.5 mg kg-1 wet weight) (EU 2005).
Most of the concentrations of cadmium and lead were found to be below their detection limits
and also far below the limits defined in the EU Regulation (EU 2005), with 0.05 and 0.2 mg
kg-1 wet weight, respectively. Values reported by other authors (Hellou and others 1992; Polak-
Juszczak 1996, 1997; Henry and others 2004; Burger and Gochfeld 2005) for cod products were
also below the EU legal limits.
Mean levels of nickel and chromium determined were 0.06 ± 0.06 mg/kg-1 wet weight and 0.4
± 0.4 mg kg-1 wet weight, respectively. The highest level of nickel (0.34 mg/kg-1 wet weight)
was found in a sample of salted and dried cod, however this value is lower than that reported
by Hellou and others (1992) (maximum: 0.53 mg kg-1 wet weight). The human requirement
for functional nickel is between 35 and 500 µg/day (Belitz and Grosch 1999). Regarding this,
cod can only contributed with a modest fraction to human necessity. Concerning chromium
the maximum contents (≥ 1 mg kg-1 wet weight) were verified in two samples strips of salted
binnenwerk.indd 504 21-08-2006 12:36:08

Contaminant metals in cod products
Table 1. Concentrations of mercury, nickel and chromium (means ± S.D. and range in mg kg-1 wet weight)
in cod samples.
Sample N Mercury Nickel Chromium
Frozen 149 0.07 ± 0.09

(0.01-0.39) ND ND
Salted 19 0.05 ± 0.00
(0.05) ND ND
Salted and dried 95 0.09 ± 0.07 0.06 ± 0.06 0.4 ± 0.2
(0.04-0.47) (< 0.01-0.34) (0.2-1.1)
Soaked salted and dried 82 0.11 ± 0.05 0.04 ± 0.02 0.4 ± 0.2
(0.03-0.19) (0.01-0.07) (0.2-0.8)
Strips 56 0.09 ± 0.06 0.06 ± 0.03 0.4 ± 0.3
(0.01-0.23) (0.02-0.13) (0.1-1.2)
Mean 401 0.09 ± 0.07 0.06 ± 0.06 0.4 ± 0.4
(0.04-0.06) (< 0.01-0.34) (0.1-1.2)
N – Number
ND – Not determined
and dried cod. This level was also report by Burger and Gochfeld (2005). In general, the mean
values found were similar with those reported in other studies: < 1 mg kg-1 (Hellou and others
1992) and 0.34 ± 0.27 mg kg-1 wet weight (Burger and Gochfeld 2005). The National Academy
of Sciences (NAS 1989) recommends a daily intake of chromium of 50 to 200 µg. According to
the present work, cod can be considered a good source of this metal.
Conclusion
Regarding the results of this study, commercialised cod and derivate products, do not represent
a risk for human consumption.
Acknowledgements
This work was supported by the Project 22-05-01-FDR-00008 “Vigilância, Segurança e Qualidade
Alimentar” and P.O. Mare.
References
Al-Ghais SM. 1995. Heavy metal concentrations in tissue of Sparus sarba Forskal, 1775 from the United Arab Emirate.
Bull Envirom Contam Toxicol 55:581-587.
Belitz H, Grosch W. 1999. Food chemistry. 2nd ed. Springer, Germany.
Burger J, Gochfeld M. 2005. Heavy metals in commercial fish in New Jersey. Environm Res 99:403-412.
Hatch WR, Ott WL. 1968. Determination of sub-microgram quantities of mercury by atomic absorption spectrometry.
Anal Chem 40(14):2085-2087.
Hellou J, Warren WG, Payne JF, Belkhode S, Lobel P, 1992. Heavy metals and other elements in three tissues of cod,
Gadus morhua from the Northwest Atlantic. Mar Poll Bull 24(9):452-458.
binnenwerk.indd 505 21-08-2006 12:36:09

Henry F, Amara R, Courcot L, Lacouture D, Bertho M-L. 2004. Heavy metals in four fish species from the French coast
of the Eastern English Channel and Southern Bight of the North Sea. Environm Intern 30:675-683.
Joiris CR, Holsbeek L, Bouquegneau JM, Bossicart M. 1991. Mercury contamination of the harbour porpoise Phocoena
phocoena and other cetaceans from the north sea and the Kattegat. Water Air Soil Poll 56:283-293.
Jorhem L. 2000. Determination of metals in foods by atomic absorption spectrometry after dry ashing: NMKL
collaborative samples. ICES CM 1980/E:29.
NAS (National Academy of Sciences). 1989. Recommended Dietary Allowances, Tenth Edition, NAS, Washington, DC
In: FDA (U.S. Food and Drug Administration), 1993. Guidance document for chromium in Shellfish. U.S. FDA,
Washington, DC 27p.
Polak- Juszczak L. 1996. The correlation between the concentration of heavy metals in the muscle tissue of fish and
their habitat. Bull Sea Fish Inst 1(137):35-39.
Polak- Juszczak L. 1997. The assessment of the hazard of contamination with heavy metals to consumers´ health based
on monitoring research on the quality of Baltic fish and their products. Bull Sea Fish Inst 1(140):49-57.
Reichenbach-Klinke H-H. 1980. Enfermidades de los peces. Acribia. Zaragoza. 507p.
UE. 2005. Regulation (EC) N.º 78/2005. JO L16, 19.01.2005. 43-45pp.
binnenwerk.indd 506 21-08-2006 12:36:09

Chapter 8:
Advanced methods for quality determination of
seafood
In many fields and areas of seafood research advanced modern instrumental techniques have
substituted and improved the existing traditional chemical, microbiological and sensory
methods. More methods are under development and are in a stage where implementation will
occur in the near future. The reason for this principal changes lie in the fact that instrumental
methods are less personal intensive, are objective, use less consumables and are on a long
time run cheaper.
New methods which were complete unknown in the fish sector 20 years ago are nowadays
used in almost every laboratory as NIR, electronic noses, instrumental colour analysis, NMR,
texture profile analysis, PCR based techniques and others. The development is very rapid and
fish industry is hesitating to implement these new technologies.
Time-temperature indicators (TTIs) are potential alternatives to provide use-by date information.
In the first paper the development of an intelligent packaging by selection of an adequate TTI
for fresh turbot. Effect of temperature on product spoilage and on kinetics of TTI response was
also studied.
In the other papers of this chapter information is given about instrumental quality control of
stockfisk (dried cod) and on the instrumental colour analysis of Atlantic salmon (Salmo salar)
muscle. A review about the revision of analytical methodologies used for the verification of the
production method of seafood (farmed versus captured) and a final paper about the detection
of noroviruses closes this chapter.
binnenwerk.indd 507 21-08-2006 12:36:09

binnenwerk.indd 508 21-08-2006 12:36:09
Instrumental quality control of stockfish
Abstract
Quality control of stockfish is made by trained experts, evaluating quality and flaws according to
market standards. There are however defects difficult to spot even by experts, and in such cases
the reduced quality is not revealed until after rehydration. The aim of this study was to evaluate
the applicability of instrumental methods to determine stockfish quality. The methods used
were visible/near infrared spectroscopy and X-ray measurements. Experts assessed the quality
of 30 stockfish by sensory evaluation and, identified potential defects, followed by instrumental
measurements. Therafter the fish was rehydrated. Quality defects found by visual inspection
of the rehydrated fish, was then applied as a reference for the study and interpretation of
the results from the instrumental measurements. The study shows that it is possible to apply
instrumental measurements to evaluate stockfish quality. With further work it should be possible
to develop commercial instrumentation to aid in assessment of stockfish quality.
Keywords: stockfish, VIS/NIR spectroscopy, X-rays, quality control
Introduction
In Norway drying of fish as a preservative technique has been used for over a thousand years
(Kurlansky 1997). Production of stockfish is a well known tradition, especially in the northern
parts of Norway. Stockfish from Norway is mainly produced from cod, and over the years this
product has had a significant financial contribution for the people involved in fishing and the
fishing industry. The main regions of Norwegian stockfish production are the Lofoten islands
and the coastal area of Finnmark. At present the majority of stockfish produced in Norway is
exported to other countries. The export of stockfish has a value of about 500 mill NOK per year
(Export statistics 2004). The fish is exported amongst others to Italy and Nigeria where it is
rehydrated and used in a number of different dishes.
Stockfish production in Norway takes place in the late winter / early months of spring. The
fresh fish is usually bled and gutted onboard the vessel. After landing it is cleaned and sorted
according to size. Then two and two fish of similar size are tied together and hung on outdoor
wooden wattles to dry and mature for the next two to three months. The traditional process
of drying the fish in open air enables the special quality characteristics to develop, but it also
exposes the fish to a climate allowing for negative quality issues to occur. The quality of the fish
will be affected by factors such as the temperature and humidity during drying. After drying,
the fish has lost about 80% of its weight (Di Luccia and others 2005), and the dry, light-weight
product is considered convenient with respect to transport and storage.
Along with the old tradition of producing stockfish goes the knowledge and expertise of
assessing the quality of the fish. As it has been done over the last centuries, the quality of the
fish is currently evaluated manually. The stockfish is graded in classes differentiating amongst
binnenwerk.indd 509 21-08-2006 12:36:09

several qualities from the supreme via ordinary to the lower quality range. The evaluators
conducting this control have years of training and expertise. There are however some quality
flaws that impose a significant challenge to the assessors. Some defects can not be observed
on the surface or the open areas of the fish, but are found in the closed structure of the muscle.
The most important of these “hidden defects” are the following: mucoso, muscle discoloration
due to blood, frost damage of the muscle, damage due to larvae and finally oxidation.
Mucoso is one of the most challenging defects to identify (Joensen and others 2005). When
present, it is usually observed as an alteration or degradation of the muscle structure in the
neck area, along the backbone, or in the vent area. In the stockfish the mucoso affected area
appears like a white layer of powder along the affected area. Upon rehydration, the mucoso
tissue seems like a slimy layer covering the underlying muscle, and the thickness of the layer
may vary from one to several millimetres. This slimy appearance is of course not appreciated
by the stockfish consumer.
Blood marks in the fish – due to rough handling during catch or insufficient bleeding of the
fresh fish – may in some cases appear as brown areas on the skin. In other cases however,
there will not be any indications of blood marks visible on the surface of the fish until after
rehydration.
Oxidation in stockfish from cod is mainly due to remains of the liver in the abdomen or on the
collar bone of the fish. The effect on the stockfish is yellowish colour and rancid smell. Drying
in open air makes the fish a potential housing for flies or other insects. Some fly species prefer
the drying fish as the hatching place and nutritional storage for their eggs and larvae. When
the fish gets worm eaten, the muscle is removed from the bones mainly in the neck area. The
muscle then appears as etched away with a characteristic unpleasant smell.
Due to the problems of efficiently identifying these defects, there has been an increasing
interest in finding a tool for quality verification of stockfish. Work has been done in order
to evaluate stockfish production (Joensen and others 2005) and rehydration (Santoro and
others 2001), but to our knowledge there is little done in order to facilitate instrumental
quality control of stockfish. The aim of this work was to evaluate if spectroscopy- and X-ray
measurements are viable methods for assessing the quality of stockfish.
Stockfish
30 samples of commercially produced stockfish were applied in this trial. The stockfish was
produced in the winter/spring season of 2004. The sample set was also a part of a parallel study
on the effect of the raw material quality on the quality of the stockfish (Joensen and others
2005; Joensen and others 2004). In seven of the 30 samples significant blood marks on the
skin of the fresh fish were observed; hence this blood in muscle was positively documented
prior to drying of the fish.
After instrumental measurements, the fish was rehydrated according to traditional regimes. The
stockfish was placed in two 100 litres tanks filled with water and ice. Water and ice was replaced
once a day. After four days in water, the fish was skinned, and then put back in ice-water for
binnenwerk.indd 510 21-08-2006 12:36:09

another four days. This ended the rehydration process, and the quality of the rehydrated fish
was subsequently evaluated.
Sensory evaluation of samples

The quality of the stockfish was assessed by two trained expert. During the last three years the
experts have actively participated in commercial grading of stockfish throughout the stockfish
season. The stockfish was evaluated for indications of quality defects related to mucoso, blood in
the muscle, oxidation, frost damage and worms. Appearance (colour of the skin and visible muscle
area as well as shape), smell and texture were evaluated in relation to evidence of high quality
and quality defects, as described in Table 1. Some samples had signs of more than one defect;
for instance both worm eaten and blood in the muscle. The quality of the rehydrated stockfish
was evaluated by the same two experts. The rehydrated fish could be inspected more carefully
as it had been skinned and could be opened and studied from the inside. During evaluation
of the rehydrated fish, the backbone was removed to look for evidence of mucoso. For every
sample the quality state was denoted, identifying the location of possible defects. Hence, the
quality reference for the instrumental measurements was a table with all samples and respective
observations and locations of quality defects on the stockfish and the rehydrated fish.
Spectroscopic measurements
Spectroscopic measurements were performed by use of the commercial instrument NIRS6500
(Perstorp Analytical Inc., Silver Spring, Md, U.S.A). NIRS6500 was operated in transflection
(diffuse reflection) mode, applying a fibre optic probe to register the 400 – 1100 nm spectrum.
The fibre bundle ends in a probe of size 4x4x6 cm, and hence this probe could only be applied
on the outer surface of the stockfish. Due to the size it was not possible to place it into
Table 1. Quality issues of stockfish and rehydrated stockfish. The different defects are characterised by
changes in appearance, odour and texture compared to flawless samples. The experts examined every sample
in the trial looking for evidence of defects, denoting the location of the affected area, if any.
Quality state Stockfish - sensory properties Rehydrated stockfish - sensory properties
Perfect, Very hard texture, stiff, no off odours, Firm structure of the rehydrated fish, no
no defects no discoloration or marks on the fish off odours, white-yellow muscle colour
Mucoso Visual and textural observation of Visual and textural observation of
powdery texture along backbone and alteration in the muscle structure, the
vent area, “mucoso odour” muscle appears to be covered by a
layer of mucus
Blood in muscle Brown areas on the skin and in the neck Visual observation of muscle
discoloration, brown coloured muscle
areas
Frost damage Porous texture, affected areas are bendy Porous and spongy texture
Oxidised Dark yellow muscle areas, rancid odour Dark yellow muscle areas, rancid odour
Worm eaten Unpleasant off odour, the muscle Unpleasant off odour, less muscle on the
appears to be etched off the bones bones, remaining muscle appears to
be covered by a layer of mucus, larvae
may be found
binnenwerk.indd 511 21-08-2006 12:36:09

the interior of the fish. The instrument bandwidth is 8 nm. The location for transflection
measurements is shown in Figure 1, and the fibre optic transflection probe was located directly
onto the back, on the right side, of the fish. To reduce noise, ten spectra from each measurement
location were averaged before recording to disk. Due to irregularities in the fish shape as well
as irregularities in the skin, the contact between the measurement probe and the fish was not
optimal. Measurements were however performed in a dark room in order to avoid light scatters
from the surroundings.
In NIR spectroscopy, the recorded spectrum is usually an absorbance spectrum in which the
light reflected or transmitted from the sample is related to a reference spectrum. A reference
for the transflection spectra was recorded from a 20 mm white Teflon block.
X-ray measurements
X-ray measurements were made by use of Sensor X (Marel hf., Iceland). Every stockfish was run
through the machine twice; once with the left side of the fish placed downward and once with
the gut-opening placed downward. This was done in order to facilitate see-through of the fish
from different angles. Voltage and current reading of the instrument during measurements were
35 kV and 6 mA. This setting gave good contrast in the X-ray images.
Data analysis
Data analysis of the spectroscopic recordings was made by use of the multivariate method
principal component analysis (PCA). PCA was applied to look for similarities and diversities
amongst spectra as function of the manually identified quality state of the stockfish. Data
analysis was run by use of the software program The Unscrambler version 8.5 (CAMO Process
AS, Oslo, Norway). No pre-processing of the spectral data was performed.
The X-ray images were evaluated manually for evidence of the quality defects found in the
stockfish after rehydration. The images were evaluated according to areas of high/low absorption
indicating differences in texture and presence of irregularities.
to NIRS6500
Figure 1. Sketch to illustrate the measurement location of transflection probe.
binnenwerk.indd 512 21-08-2006 12:36:10

Manual evaluation of stockfish and rehydrated stockfish

The manual quality assessment of the stockfish gave the result shown in Table 2. Here it is seen
that of the 30 stockfish seven were considered of perfect quality, whereas the rest had one
or more quality defects. Table 2 also shows the result of sensory evaluation of the rehydrated
fish. From the table it is seen that more quality defects were revealed after rehydration. As for
the six stockfish samples classified as mucoso fish, two of them did not show signs of mucoso
after rehydration and were hence wrongly classified prior to rehydration. This demonstrates the
problem of documenting the quality of stockfish (Joensen and others 2005); both the trouble
of finding the defects as well as wrong classification. This problem is mainly due to the form
and structure of the fish, making the inside of the fish inaccessible to the evaluators. The defect
most difficult to spot in the dry fish is development of mucoso. As this occurs – to a large extent
– along the backbone and in the vent area, this is hardly available to manual inspection.
Spectroscopic measurements
Visible (VIS) and near infrared (NIR) spectroscopy has formerly shown useful in quality assessment
of fish (Heia and others 2003; Nilsen and Esaiassen 2005; Zhang and Lee 1997). Its potential with
respect to speed as well as operational ease makes it an interesting candidate for instrumental
quality control of stockfish. Examples of transflection spectra recorded with the NIRS6500
instrument are shown in Figure 2. No immediate spectral characteristics could be identified as
function of quality defects. Visual evaluation of the spectra could be interpreted according to the
skin colour, whether it was dark or with a light brown colour at the area of measurement.
Principal Component Analysis was performed on the transflection spectra, applying the NIR
range from 700 to 1100 nm as the spectral basis for the analysis. The result of the PCA is given
in the score plot of Figure 3. As can be seen from the figure, the most obvious group to be
identified in the score plot is the spectra from the fish with frost damages. Spectra from fish
with blood in the muscle also separate from the other spectra, although not as distinct as the
Table 2. Quality state of stockfish and rehydrated stockfish as evaluated by sensory inspection (N=30).
Quality state Number of stockfish with given Number of rehydrated samples with
defects as identified by sensory given defects as identified by sensory
assessment assessment
Perfect, no defects 7 5
Mucoso 6a 12
Blood in muscle 7b 10
Frost damage 4 4
Oxidised 3 3
Worm eaten 5 9
a Two of the mucoso-classified samples did not show signs of mucoso after rehydration. They were thus
incorrectly classified.
b These seven samples had been selected prior to drying, and hence the classification of these stockfish
was based on this information.
binnenwerk.indd 513 21-08-2006 12:36:10

absorption
1.0
0.9
0.8
0.7
0.6
0.5
0.4 Samples
400 500 600 700 800 900 1000 1100
wavelength (nm)
Figure 2. Transflection spectra from five stockfish. The spectra are recorded on the back of the fish.
PC2 Scores
1.0
0.5
-0.5
-1.0
-1.5 PC1
-4 -2 0 2 4 6
Figure 3. PCA score plot from transflection spectra. Each sample is named according to the quality of the
stockfish. The spectra were recorded on the back of the stockfish. Ellipses illustrate the grouping of samples.
frost-group. The rest of the measurements are all in a mixed group, samples with defects such
as oxidation or mucoso in the same area of the score plot.
Considering the location for transflection measurements – on the back of the fish – the results
of the score plot can be explained. Quality defects in stockfish do generally not affect the
whole fish, but more often the flaws are found in local spots of the fish. The exceptions in
these measurements were the frost damaged fish, which seemed to have an altered structure
throughout the whole fish. From this point of view the defects having an effect in the selected
measurement location would be the frost damaged fish as well as the fish having blood marks
in the back of the fish. Mucoso which is mainly located along the backbone is too far from the
measurement area to be expected to influence on the spectral reading. The same can be said
for oxidation and worm-eaten samples of which the defects are usually located on the collar
bone or in the neck respectively.
binnenwerk.indd 514 21-08-2006 12:36:12

If the optical probe can be located adjacent to the defected area, the spectroscopic method
has the potential of differentiating amongst the high quality stockfish, and fish with blood
spots, frost damage and mucoso. Mucoso alters the structure of the muscle; hence the optical
characteristics are changed compared to that of a high quality sample. As for measuring
oxidation and the effects of worms in stockfish, the results are ambiguous. In Figure 3 it is seen
that the samples in this category are all mixed in a big group with the samples considered of
high quality, hence the analysis gives no clear distinction of these defects.
X-ray measurements
By means of X-rays it is possible to image the inside of objects. The technique is already applied
in use in quality control in foods (Zwiggelaar and others 1996), and during the last few years
the technique has been adapted for on-line detection of bone and bone remnants in fresh fish
fillets (Heia and Gregersen 2003). As some of the flaws in stockfish are found inside the fish,
X-rays were considered a viable tool for quality assessment of stockfish. Figure 4 shows x-ray
images from the neck of two stockfish, the fish being placed with the left side facing downward
during measurement. The most obvious structures to be seen in the images are the fish bones.
Comparing the left image to the right one, it is clear that the stockfish in the left absorbs
more X-rays than the fish in the right. The quality of the stockfish in the left and right images
is perfect and worm-eaten, respectively. In the worm-eaten fish most of the muscle tissue is
gone; and this accounts for the lower X-ray absorption in this sample.
Another example of X-ray measurements is given in Figure 5, displaying the mid-and tail part of
the stockfish. Here the difference in the circled area of the fish should be noted. The left image
shows a sample of perfect quality whereas the right sample had mucoso in the vent area. The
latter appears as a small area of higher X-ray absorption. The mucoso affected areas seem to
hold more humidity than the adjacent muscle, and thus accounts for more X-ray absorption.
Figure 4. X-ray images of perfect stockfish (left) and worm eaten stockfish (right). The darker area of the
left fish indicates more X-ray absorption and hence more muscle, whereas the light area (less absorption)
in the right fish implies that more of the muscle is eaten away by worms.
binnenwerk.indd 515 21-08-2006 12:36:13

Figure 5. X-ray images of perfect stockfish (left) and stockfish with mucoso in the vent area (right).
From the manual inspection of the X-ray images it was found that mucoso, frost damages and
the effect of worms could be identified by X-ray measurements. Blood spots and oxidation could
however not be detected by this methodology. For X-ray imaging to become an industrially
viable method one would have to develop image analyses for this specific application. This is
considered a complex and time consuming task.
Conclusion
Measurements and evaluation of methodology demonstrated that some of the most common
quality defects in stockfish can be measured instrumentally. X-rays imaging will reveal defects
such as mucoso, frost damage and worm eaten fish. Visible / near infrared spectroscopy can
be used to identify blood in the muscle, frost damage and most likely mucoso as well. None
of the tested techniques proved useful for assessing all of the different quality defects. Even
so, an instrumental method for identification of some of the most ordinary flaws, would be a
valuable tool for the stockfish industry; easing the work of the manual stockfish assessors. In
view of the time and costs required for instrument development and commercialisation of a
method, it is recommended that further work should be based on spectroscopic measurements.
The price of a commercial instrument based on spectroscopy is expected to be considerably
lower than a detection system based on X-ray imaging, and therefore more acceptable to the
industry as well.
Acknowledgements
The authors wish to thank Tørrfiskforum, Norway and Innovation Norway for the financing of
this project, and we also want to express our gratitude to Marel for the possibility to apply their
Sensor X in this work. This work was part of a study conducted in cooperation with colleagues
at MATFORSK and SINTEF IKT Norway, and we wish to thank them for interesting discussions
and fruitful responses during the project.
binnenwerk.indd 516 21-08-2006 12:36:13

References
Export statistics. 2004. Norwegian Seafood Export Council. 6 p.

Heia K, Esaiassen M, Nilsen H and Sigernes F. 2003. Visible spectroscopy - Evaluation of storage time of ice stored cod
and frozen hake. In: Luten JB, Oehlenschläger J, Ólafsdóttir G, editors. Quality of Fish from Catch to Consumer.
Labelling, Monitoring and Traceability. The Netherlands: Wageningen Academic Publishers. p 201-209.
Heia K, Gregersen F. 2003. New technology removes pin bones. Fiskeriforskning Info 3/2003.
Joensen S, Akse L, Bjørkevoll I, Mathisen I. 2005. Quality improvement of raw material for stockfish production – Catch
damages and consequences for stockfish quality. In Norwegian. Fiskeriforskningsrapport 2/2005.
Joensen S, Akse L, Bjørkevoll I and Mathisen I. 2004. Quality improvement of raw material for production of salted fish
– Catch damages and consequences for salt fish quality. In Norwegian. Fiskeriforskningsrapport 16/2004.
Kurlansky M. 1997. Cod. A biography of the fish that changed the world. New York: Walker Publishing Inc. 294 p.
Luccia AD, Alviti G, Lamacchia C, Faccia M, Gambacorta, Liuzzi V, Musso SS. 2005. Effects of hydration process on
water-soluble proteins of preserved cod products. Food Chem 93:385-393.
Nilsen H, Esaiassen M. 2005. Predicting sensory score of cod (Gadus morhua) from visible spectroscopy. Lebensm Wiss
Technol 38:95-99.
Santoro A, Sarli TA, Murru N, Naoop C and Cortesi ML. 2001. Stockfish rehydration: technology and controls. Ind Aliment
Italy 40:520-524, 529.
Zhang HZ and Lee TC. 1997. Rapid near-infrared spectroscopic method for the determination of free fatty acid in fish
and its application in fish quality assessment. J Agric and Food Chem 45(9):3515-3521.
Zwiggelaar R, Bull CR and Mooney MJ. 1996. X-ray simulations for imaging applications in the agricultural and food
industries. J Agr Eng Res 63(2):161-170.
binnenwerk.indd 517 21-08-2006 12:36:13

binnenwerk.indd 518 21-08-2006 12:36:13
Time temperature indicators as quality and shelf life indicators for
fresh turbot (psetta maxima)

AZTI-TECNALIA, Txatxarramendi Ugartea z/g, 48395 Sukarrieta (Bizkaia), Spain
Abstract
Time-temperature indicators (TTI) are potential alternatives to provide use-by date information.
The objective was to develop an intelligent packaging by selection of an adequate TTI for
fresh turbot. Effect of temperature on product spoilage and on kinetics of TTI response was
also studied. Preservation studies of fresh turbot obtained from aquaculture, and packaged in
extended PVC film were conducted under isothermic storage conditions (0, 5, 10 and 15 ºC),
monitoring microbial growth, sensory acceptability and kinetic response of 3 TTIs (Monitor
Mark®, Fresh Check® and Check Point®). Response of Fresh Check® and Check Point® indicated
good correlation with sensory shelf life and bacteria spoilage. Further experiments under
dynamic storage conditions are being conducted.
Keywords: turbot, time temperature indicators (TTI), microbiology, sensory analysis, shelf life,
intelligent packaging
Introduction
Quality, safety and shelf life of foods are highly dependant on storage time and temperature.
The temperature history from production, through distribution and storage to consumption
should be controlled in order to reduce food safety problems and wastes due to spoilage. This
information on temperature history, that can be provided down to the food product unit using
suitable time temperature indicators attached to the food package, allow the determination of
spent and remaining shelf life at any point of the chill chain (Taoukis and others 1999).
Usually shelf life determinations are based on assumptions of either a most probable average
temperature for the food or a worst case temperature exposure. The first approach may result
in products of unacceptable quality before the stated end of shelf life in cases of temperature
abuse above the average assumed temperature. The conservative second approach can lead
to waste of perfectly good products, since considerable fraction of the products may not be
exposed to adverse conditions (Taoukis and Labuza 1989).
Time-temperature devices can serve as dynamic or active shelf-life labels instead of, or in
conjunction with, open-date labelling that is currently being used in packaged products (Labuza
and Fu 1995).
A time temperature indicator (TTI) is a simple, inexpensive device that shows an easily
measurable, time-temperature dependent change that reflects the full or partial temperature
history and quality status of the food product to which it is attached to. The principle of
operation is a mechanical, chemical, electrochemical, enzymatic or microbiological irreversible
binnenwerk.indd 519 21-08-2006 12:36:13

change usually expressed as a visible response, in the form of a mechanical deformation, colour
development or colour movement. The rate of change is temperature dependent, increasing
at higher temperatures. The development and application of reliable TTI systems must be
approached based on kinetic principles (Taoukis and Labuza 1989).
The objective of this work, was to select an adequate time-temperature indicator for fresh
seafood that provided useful information to consumers or food producers by a smart or
intelligent packaging. However, the use of these systems to monitor the shelf life of a certain
food product needs a previous concise kinetic evaluation of their response and a comparison
with the kinetics of food spoilage. The effect of temperature on the product spoilage and on
the kinetics of TTI response was also studied.
The approach and methodology used aimed to carry out shelf life experiments for fish products
under isothermal conditions (0, 5, 10 and 15 ºC) in order to obtain kinetic data of microbial
growth kinetic, sensory acceptability and response of time-temperature indicators.
The seafood studied was turbot (Psetta maxima) obtained directly from aquaculture, and
packaged in extended PVC film individually wrapped. Microbial growth and sensory attributes
were monitored in samples stored at controlled isothermal conditions 0, 5, 10 and 15 ºC in
incubators (Ibercex V-450). The response of three types of TTIs, Monitor Mark® (3M, USA) 5 ºC
model, Fresh Check® (TempTime Corporation, USA) 6D422 model and Check Point® (Vitsab
Sweden, Sweden) M2 model, were measured at the same temperatures.
Activation of the time temperature indicators and measurement of responses

TTIs were adequately activated (following manufacturer´s instructions) just before introducing
them in the incubator chambers, and responses of 10 TTIs of each type were measured for
replicates at each time. Monitor Mark® indicator responses, X, was the distance of the blue
colour line advance in mm. Fresh Check® and Check Point® indicators responses (based on
colour changing) were measured in two different ways, quantitatively in the CIE L*a*b* colour
space using a Minolta CR-200 colorimeter (Minolta, Japan) and qualitatively by photographs
and visual evaluation of the colour changes using previously defined colour evolution scales.
Sample preparation and microbiological analysis

Fish (10 g) was transferred to a stomacher bag (Seward Medical, UK), peptone water with salt
(NaCl, 0,5%) was added and homogenized for 1 minute with a Stomacher (Lab-Blender 400,
Seward Medical, UK).
Samples (0.1 ml) of 10-fold serial dilutions of naturally contaminated fish homogenates were
spread on the surface of the appropriate media in Petri dishes for enumeration of following
parameters: (i) total aerobic viable count on Plate Count Agar (PCA, Pronadisa, Spain) and
incubated at 31 ºC for 3 days (Gilchrist and others 1977); (ii) Enterobacteriaceae, on violet
red bile glucose agar (VRBGA, Pronadisa, Spain) and incubated at 37ºC for 1 day (Manafi
2003); (iii) Pseudomonas on Agar Pseudomonas Base (Oxoid code CM 559, supplemented with
SR 103, Oxoid, Basinstoke ,UK) and incubated at 37 ºC for 2 days (Jeppesen and others
2003); (iv) Brochothrix thermosphacta on streptomycin sulphate thallous acetate cycloheximide
(acitidione) agar, (STTA, Oxoid code CM 881, supplemented with SR 151, Oxoid Basinstoke, UK)
binnenwerk.indd 520 21-08-2006 12:36:14

Time temperature indicators as quality and shelf life indicators for fresh turbot
and incubated at 20 ºC for 3 days (Gardner 1966); (v) Shewanella putrefaciens on Iron Agar
(IA, Oxoid code CM 867, Oxoid, Basinstorke, UK) and incubated at 20 ºC for 4 days (Gram and
others 1987).
Evaluation of organoleptic shelf life

The sensory evaluation was carried out according to the conditions and general methodology set
down in ISO 4121:1987. Triplicate samples were taken at regular intervals. The sensory attributes
of whole raw and cooked fish were evaluated by a trained sensory panel of 10 judges using a
6-point descriptive scale (ISO 4121:1987), 9 denoting absolutely fresh and 4 completely putrid
or spoiled. Sensory evaluation was conducted in individual booths under controlled conditions
of light, temperature and humidity. Sensory assessment of raw fish included examination of the
following quality attributes: skin, external odour, eyes and gills. The measurement of freshness of
cooked fish (odour, flavour and texture) was also evaluated. To prepare the cooked fish sample,
fish were scaled, gutted and gilled before cooking. Fish samples were cooked individually over
boiling water at 100 ºC for 30 minutes. The cooked samples were served hot to panellists.
The shelf life according to each TTI was determined when the response showed the end point of
the TTI. The observed end points for each TTI at each temperature tested are shown in Table 1.
Figures 1 and 2, show the general appearance of turbot at day 0 and at day 9 of storage at
10 ºC. The sensory quality decay can be seen. Turbot in Figure 1 shows convex pupil shape and
no slime on the skin. In contrast, in Figure 2, turbot eyes are sunken, mucus has appeared on
eyes and the skin is covered with yellow slime. The evolution of sensory scores given by the
judges for raw and cooked turbot, at four storage temperatures (0, 5, 10 and 15 ºC) is shown
in Figures 3 and 4. When samples got an unacceptable assessment, it was considered that their
shelf life was finished (see Table 1). The rejection point was established to be a score below
6, according to the ISO 4121:1987 for 6-point ordinal scales to assess the quality of specific
food products.
Table 1. Shelf life according to sensory data, microbiological results and response of TTIs in isothermal
preservation studies of turbot at four temperatures (0, 5, 10, 15 ºC). Last time (hours) with acceptable
(not exceeding the limits established for each parameter) measured values is indicated.
Tª (ºC) Sensory shelf life Microbiological limits TTI shelf life

(hours) (hours) (hours)
Raw Cooked Mesophilic Enterobacteria Pseudomonas Fresh Check Monitor

aerobia Check Point Mark
(for N=106 cfu/g) (for N=103 cfu/g) (for N=104 cfu/g)
0 226 223 360 432 264 261 399 >500

5 173 174 192 240 168 169 169 71
10 85 80 135 111 87 53 78 52
15 50 52 64 72 48 24 68 44
binnenwerk.indd 521 21-08-2006 12:36:14

Figure 1. Turbot at day 0 of storage at 10 ºC.
Figure 2. Turbot at day 9 of storage at 10 ºC.
Raw turbot
9
8
Sensory score
0ºC
7
5ºC
6 10ºC
5 15ºC
4
0 24 48 72 96 168 192 216 240 264
Storage time (hours)
Figure 3. Sensory score evolution of raw turbot during storage time (h) at 0, 5, 10 and 15 ºC.
binnenwerk.indd 522 21-08-2006 12:36:14

Time temperature indicators as quality and shelf life indicators for fresh turbot
Cooked turbot
9
Sensory score 8 0ºC
7 5ºC
10ºC
6
15ºC
5
4
0 24 48 72 96 168 192 216 240 264
Storage time (hours)
Figure 4. Sensory score evolution of cooked turbot during storage time (h) at 0, 5, 10 and 15 ºC.
For each sampling day, triplicate analyses were conducted for every microbial parameter. The
following microbial population acceptability limits were considered to determine turbot shelf
life in Table 1: total aerobic flora N=106 cfu/g, enterobacteria N=103 cfu/g, pseudomonas
N=104 cfu/g. The limits for total aerobic flora (106 cfu/g) and enterobacteria (103 cfu/g)
were determined following the Spanish regulation on fish and aquaculture products for human
consumption (RD1521/1984).
The acceptability limit for pseudomonas was established according to previous sensory studies
and studies by Gram and Huss (1996), Taoukis and others (1999), and Huss (1995). A comparison
was made between the kinetic response of TTI-s and kinetic data of food spoilage was studied.
Sensory shelf life of fresh turbot was compared with total aerobic flora and pseudomonas
population. Sensory rejection corresponded to a pseudomonads level of 104 cfu/g. The response
of two of the TTIs tested, Fresh Check® and Check Point®, indicated good correlation with
sensory shelf life and pseudomonads counts.
Conclusion
The indicator that showed the best performance was the Fresh Check®. The time to reach the
rejection colour was very close to the shelf life according to sensory analysis. The time to reach
the rejection colour for Check Point® was longer than the product shelf life at low temperatures
(0 and 5 °C).
For an effective monitoring of fish shelf life in the chill chain using the TTIs, models of
sensory quality decay, growth of spoilage microorganisms and TTI response evolution are being
developed using these experimental data obtained at isothermal storage conditions. Moreover,
further experiments under dynamic temperature conditions are being conducted for turbot in
order to validate the models and TTIs response. This work is being prepared for publication.
Acknowledgements
This work was granted by the Department of Agriculture and Fisheries and the Department of
Presidence from the Basque Government.
binnenwerk.indd 523 21-08-2006 12:36:15

References
Gardner GA. 1996. A selective medium for the enumeration of Microbacterium thermosphactum in meat and meat
products. J Appl Bacteriol. 29(3):455-460.
Gilchrist JE, Donnelly CB, Peeler JT, Campbell JE. 1977. Collaborative study comparing the spiral plate and aerobic plate
count methods. J Assoc Off Anal Chem (60):807-812.
Gram L, Huss HH. 1996. Microbiological spoilage of fish and fish products. Int J Food Microbiol 33(1):121-137.
Gram L, Trolle G, Huss HH. 1987. Detection of specific spoilage bacteria from fish stored at low (0°C) and high (20°C)
temperatures. Int J Food Microbiol (4):65-72.
Huss HH (ed). 1995. Quality and Quality Changes in Fresh Fish. FAO Fisheries technical paper No. 348. Rome, Italy:
FAO.
ISO 4121:1987. Sensory analysis - Methodology - Evaluation of food products by methods using scales. The International
Organization for Standardization. Geneva, Switzerland.
Jeppesen VF, Jeppesen C. 2003. Media for Pseudomonas spp. and related genera from food and environmental samples.
In: Corry JEL, Curtis GDW, Baird RM editors. Handbook of culture media for food microbiology, Elsevier Science.
p 345-354.
Labuza TP, Fu B. 1995. Use of time/temperature integrators, predictive microbiology, and related technologies for
assessing the extent and impact of temperature abuse on meat and poultry products. J Food Safety 15:201-227.
Manafi M. 2003. Media for detection and enumeration of total Enterobacteriaceae, coliforms and Escherichia coli from
water and foods. In: Corry JEL, Curtis GDW, Baird RM editors. Handbook of culture media for food microbiology,
Elsevier Science. pp 167-193.
RD1521/1984. Real Decreto 1521/1984 de 1 de agosto. BOE 22-8-84. Reglamentación técnico sanitaria de los
establecimientos y productos de la pesca y acuicultura con destino al consumo humano.
Taoukis PS, Labuza TP. 1989. Applicability of Time-Temperature Indicators as Shelf Life Monitors of Food Products. J
Food Sci 54(4):783-788.
Taoukis PS, Koutsoumanis K, Nychas GJE. 1999. Use of time temperature integrators and predictive modelling for shelf
life control of chilled fish under dynamic storage conditions. Int J Food Microbiol 53:21-31.
binnenwerk.indd 524 21-08-2006 12:36:15

Instrumental colour analysis of Atlantic salmon (Salmo salar L.)
muscle
Lars Helge Stien1, Asbjørn Høyem Amundsen1, Turid Mørkøre2, Simon Nesse Økland3 and
Ragnar Nortvedt1,4
1Section of Applied and Industrial Biology, Department of Biology, University of Bergen, PO Box
7800, NO-5020 Bergen, Norway
2AKVAFORSK, Institute of Aquaculture Research, PO Box 5010, NO-1432 Ås, Norway
3Bremnes Seashore AS, Øklandsvågen, NO-5430 Bremnes, Norway
4NIFES, National Institute of Nutrition and Seafood Research, PO Box 2029 Nordnes, NO-5817
Bergen, Norway
Abstract
Instrumental colour analysis of raw salmon flesh was performed by one tristimulus filter
colorimeter and two spectrophotometers. A full factorial experimental design at two levels
with three variables (1 or 5 cm thick cutlet, dark or white background, and lighting on or off)
showed that the instrumental readings were strongly influenced by sampling conditions (p <
0.001). The highest correlations between astaxanthin and instrumental readings were obtained
for the a* values of the spectrophotometers (r > 0.780) when sampling on 1 cm-thick cutlets on
a white background. The cutlets were also scanned with a flatbed scanner for colour assessment
by image analysis. The image analysis correlated well with both astaxanthin (r = 0.709) and
visual colour ranking (r = 0.951).
Keywords: tristimulus filter colorimeter, spectrophotometer, scanner, image analysis, salmon
Introduction
The colour of a food product strongly influences the consumer’s choice (Chambers and Bowers
1993; Calvo and others 2001). Up to 40% of the consumer’s decision to buy a fish product may
be determined by its colour (Robb 2001b). Flesh colour is of special importance for salmonid
products (Sigurgisladottir and others 1997; Robb 2001b; Cardinal and others 2004). Effects on
colour from feed composition, feeding regime, stress and different slaughtering and processing
techniques are therefore a highly studied subject area in salmonid research (Storebakken and
No 1992; Rørå and others 1998; Wathne and others 1998; Cardinal and others 2001; Nickell and
Springate 2001; Robb 2001a; Kiessling and others 2004). There are, however, fewer studies into
how the colour of salmonid flesh should be measured, and few comparisons between different
measuring techniques have been done. It was, however, concluded in a study by Christiansen
and others (1995) that visual colour card rating was superior to instrumental colour readings
by Minolta Chroma Meter CR200 in predicting pigment concentration in salmon flesh. In a
review by Robb (2001b) subjective rating, colour card rating, tristimulus filter colorimeter,
spectrophotometry and high pressure liquid chromatography (HPLC) are presented as methods
for colour assessment of salmonid flesh.
binnenwerk.indd 525 21-08-2006 12:36:15

Lars Helge Stien, Asbjørn Høyem Amundsen Turid Mørkøre, Simon Nesse Økland and Ragnar Nortvedt
Subjective rating may be a verbal grading of the colour into a range from “strongly dislike” to
“strongly like”. However, this being a subjective method, variations between assessors influence
both the accuracy and the precision of the final rating. It has therefore limited use as a scientific
tool, especially if the assessment is made by only one person. A more objective evaluation of
colour can be obtained by a sensory panel of trained assessors. The human eye is very good at
side-by-side comparisons of colour, even when the differences are very small (Ray 1999). This is
the underlying rationale for the development of colour cards, which enable assessors to compare
the flesh colour with a series of standards. The manufacturers of these colour cards usually
specify the ideal lighting conditions under which the visual assessment should be performed,
often light simulating daylight outdoors on a cloudy day. Standardized lighting conditions are
especially important for repeated measurements performed over a period of time, e.g. routine
monitoring in a processing plant. Colour cards give more reproducible results than subjective
rating alone (Robb 2001b).
A tristimulus filter colorimeter registers colour through three photocells or photomultipliers

(Figure 1a) so that (to a sufficient approximation) its responses are proportional throughout
the visual spectrum to some standard observer, e.g. the CIE 1931 colour-matching functions
(Wyszecki and Stiles 2000). A spectrophotometer, on the other hand, compares each wavelength
of the radiant power leaving the object with that incident on it (Figure 1b) (Wyszecki and Stiles
2000) and the tristimulus values are derived from the resulting spectrum. Spectrophotometers are
regarded as being more accurate than tristimulus colorimeters (Wyszecki and Stiles 2000).
Photocells
a) Object
X
Light Tristimulus
source filters
b) Object Photon
detector
Monochromator
Light
source
Figure 1. a. Schematic diagram of a tristimulus-filter colorimeter. The response of the upper photocell
gives the X-tristimulus value; the two other give the response of Y- and Z-tristimulus values respectively. b.
Basic design of a spectrophotometer. The wavelength of the radiant power leaving the object is measured
wavelength-by-wavelength by a photon detector, thus creating the colour spectrum of the object.
binnenwerk.indd 526 21-08-2006 12:36:16

Instrumental colour analysis of Atlantic salmon muscle
The first aim of the present study was to compare three different colorimetric instruments;
one tristimulus filter colorimeter and two spectrophotometers, focusing especially on how
the different instruments are affected by sampling conditions as cutlet thickness, background
colour and room lighting. The second aim was to investigate how visual colour rankings (side-
by-side comparison and Roche SalmoFan™-scores) relate to instrumental colour readings.
Thirdly, it was aimed to see how visual and instrumental colour assessments are related to the
astaxanthin content of salmonid muscle. A standard consumer flatbed scanner was also tried
as an alternative to relatively expensive colorimetric instruments.

Twenty Atlantic salmon, weighing on average 2.5 kg, were sampled from the same batch at the
processing line at the producer Bremnes Seashore AS (Bremnes, Norway) on 30 March 2004. The
fish were eviscerated and then stored on ice for four days until sampling. A 5 cm-thick cutlet
from directly behind the dorsal fin of each fish was first placed with its anterior part facing
down towards the glass plate and scanned on a flatbed scanner (CanoScan LiDE 80, Canon
Inc., Ohta-ku, Japan), ranked according to Roche SalmoFan™ (F.Hoffmann-La Roche Ltd, Basle,
Switzerland) by a trained researcher and then analysed by three colorimeters. These instruments
were the tristimulus filter colorimeter Minolta Chroma Meter II-CR200 (measurement area: φ
8 mm, 2° standard observer, diffuse light, 0° viewing angle, illuminant D65, from Minolta,
Osaka, Japan), the spectrophotometer Hunterlab Miniscan/XE instrument (measurement area:
φ 25 mm, 10º standard observer, 45° illuminating angle, 0° viewing angle, illuminant D65,
from Hunter Associates Laboratory Inc., Virginia, USA), and the spectrophotometer X-Rite
Ca22 Tethered Spectrophotometer (measurement area: φ 6.35 mm, 10º standard observer, 45°
illuminating angle, 0° viewing angle , illuminant D65, from X-Rite Inc., Grandville, Michigan,
USA). The instruments were calibrated to white and black (only Hunter) standard tiles. A 1
cm-thick cutlet was then cut from the anterior part of the 5 cm-thick cutlet, and analysed
once more by the colorimetric instruments. Finally the colours of a selected sub-sample of five
cutlets with apparently wide colour variation, according to the instrumental methods, were
ranked by 20 volunteers, before samples were taken from all cutlets for chemical determination
of astaxanthin.
The cutlets were scanned with a spatial resolution of 200 points per inch, equivalent to a pixel
size of 0.127x0.127 mm, with 256 grey levels in each of the three layers in the RGB- colour
model and stored in the TIFF image file format. The scanner was calibrated against the white
surface of the scanner lid and the same calibration was used throughout the experiment. The
scanner was wiped clean between each scanning using paper tissue and a fat solvent washing
liquid. A light tight box was put on top of the scanner in order to prevent light from external
sources from reaching the image sensor in the carriage. An RGB colour image consists of three
colour layers; the R layer represent the amount of red light, G the amount of green and B the
amount of blue light. Pixels representing cutlets have a bright red colour in the scanned images,
and pixels representing background have a dark greyish colour. This was used to create a binary
matrix α where all pixels representing cutlet in the image were set to 1 and pixels representing
background to 0.
{
a(x,y) = 1 Û R(x,y) > G(x,y) ∧ R(x,y) > B(x,y)
0 otherwise
(1)
binnenwerk.indd 527 21-08-2006 12:36:16

In other words, all pixels in the RGB image whose R-value was greater than both the G- and
B-values were identified as representing cutlet, all others as background. The individual RGB
values of the pixels representing cutlet were then translated to the CIE 1976 L*a*b*-colour
system (Wyszecki and Stiles 2000) and statistical colour descriptors such as mean values
and percentiles (P25, P50 and P75) were calculated. The KODAK Q-60R2 (2004:01) (Eastman
Kodak Company, Rochester, New York, USA) colour target was used to calibrate the scanner’s
RGB values to the CIE 1931 XYZ values needed to calculate the L*a*b* values (for formulae,
see Wyszecki and Stiles 2000). The estimated XYZ values from this calibration correlated well
with the known CIE 1931 XYZ-values of the colour target (Figure 2, rx = 0.998, ry = 0.998 and
rz =0.999, respectively), indicating a close match. Unfortunately, the three first cutlets were
scanned incorrectly. There are therefore only 16 images of scanned cutlets, including only four
of the selected subgroup for visual ranking (see below).
The outputs of all the colorimetric instruments were by default CIE 1976 L*a*b* colour values.
The L*, a* and b* values represent three axes in the colour model ranging from dark to
light, green to red and from blue to yellow, respectively. Hue (colour) and Chroma (colour-
intensity or saturation) values give an intuitive description of the sample colour, and are
therefore often preferred to the a* and b* values when describing the colour of a sample;
a. b.
80 80
60 60
40 40
20 20
0 0
0 20 40 60 80 0 20 40 60 80
c.
80
60
40
20
0
0 20 40 60 80
Figure 2. The relationship between the known XYZ values of the KODAK Q-60R2 colour map and estimated
XYZ values calculated from the scanner’s RGB values of the colour map. a. X, b. Y, c. Z.
binnenwerk.indd 528 21-08-2006 12:36:16

Chroma = (a*2+b*2)0.5, Hue = tan-1(b*/a*) for a*>0 and b*>0, Hue = 180+ tan-1(b*/a*) if a*
< 0 otherwise Hue = 360 - tan-1(b*/a*).
A planned full factorial experimental design at two levels (FD23) with the following three
variables was employed to test the colorimetric instruments; 1 or 5 cm thick cutlet, dark or
white background, and light on or off in the room. The room was lit by four fluorescent lamps
(Aura ultimate long life 36W 830, Aura Light AB, Sweden), which emitted a warm white, slightly
yellow, light. The colour samplings were performed to the right of the spine at the centre of
the hypaxial white muscle (Figure 3). In the case of the Minolta and the X-Rite, efforts were
made to avoid directly sampling over lipid stripes. This was not possible with the Hunter, due
to the large size of the sampling area (φ = 25 mm).
Visual assessment of the colour of all the cutlets by Roche SalmonFan™ was performed by a
trained researcher under the same lighting conditions as described above. A sub-sample of five
cutlets were selected for visual measurement of colour by a large number of volunteers (n =
20). The volunteers had no prior experience in colour scoring, and were not tested for colour
blindness. The colour rankings were performed in a neighbouring room dominated by indirect
diffuse light from the outside. There was no direct sunlight. Each volunteer first classified the
cutlets, one by one, into one of the Roche SalmoFan™ colour classes. The scale on the fan
ranges from light (20) to dark (34). Secondly, each judge lined up the cutlets on a table and
sorted them from light to dark (rank 1-5) in colour using side-by-side comparison.
Pigments were extracted using a standardised method (MET.03-28, 044 Astaxanthin og

Canthaxanthin bestemmelse ved hjelp av HPLC) at the National Institute of Nutrition and
Seafood Research (NIFES, Norway). In short: 1 g of muscle was homogenised in 3 ml distilled
water, 3 ml methanol and 9 ml chloroform. The sample was then subsequently dried, diluted
in hexane and injected into HPLC- test-tubes. Peaks were detected at 470 nm and quantified
using standards for astaxanthin and canthaxanthin (Hoffman LaRoche, Basle, Switzerland). The
pigment content ranged from a minimum of 2.07 mg kg-1 to a maximum of 6.44 mg kg-1.
The statistical analysis was performed using the SAS software package (SAS Institute Inc., Cary,
North Carolina, USA) for Windows, v.8e. Effects of sampling conditions were tested for each
individual instrument using the GLM procedure. A p-value < 0.05 was considered significant.
Figure 3. The position of the colour sampling on the salmon cutlets is indicated by a circle.
binnenwerk.indd 529 21-08-2006 12:36:17

The p-values > 0.20 are denoted as n.s. (not significant) in the text. All results are given as
mean ± standard error. The data were tested for normality by a normal probability plot (proc.
univariate). Recent studies have shown that the relationship between instrumental colour
values and astaxanthin is non-linear (Christiansen and others 1995; Bjerkeng 2000; Rønsholdt
2005). Spearman’s rank correlation coefficient (Johnson and Bhattacharyya 2000) (proc. Corr.
Spearman) was therefore used to describe the relationship between the colorimetric colour
values, the colour rankings, the Roche SalmoFan™ scores and astaxanthin content.
Sampling sensitivity
There are no significant effects of lighting conditions (light on or off in room) on the registered
colour values from either of the two spectrophotometers (Table 1). However, there is a slight
effect of lighting conditions in the case of the Minolta tristimulus colorimeter. This can be
explained by the physical design of the two first instruments compared to the latter. The Minolta
has a pen-like shape that leaves the photocells exposed to light reflections in the translucent
flesh from external light sources. Operating the Minolta in highly variable light conditions may
therefore influence the results.
Significant effects of cutlet thickness are seen on the colour values from all the three colorimetric
instruments (Table 1). Interestingly, there are differences in the colour values that are affected.
The L* values of the Hunter, for instance, are not significantly affected, while the L* values
from both the Minolta and the X-Rite are. For samplings on a white background more light was
reflected back to the Minolta and X-Rite instruments for thin cutlets than for thick, resulting in
higher L* values for thin cutlets (1 cm vs. 5 cm thick cutlets on white background, L*: Minolta
44.4 vs. 43.5 ± 0.5 p = 0.070, Hunter 51.9 vs. 51.5 ± 0.2 p = n.s, X-Rite 39.3 vs. 38.3.3 ± 0.3
p = 0.009). Although significant effects of thickness on the L* values are found, R2 is very low
Table 1. Statistical effects (GLM p-values) of sampling condition (light on or off, 1 or 5 cm thick cutlet,
and dark or white background) for each of the three colorimetric instruments. There are only significant
interactions between “Thickness” and “Background”. Other interactions are therefore not included in the
final glm-model.
Sampling conditions Colour variable Minolta Hunter X-Rite
Light L* n.s n.s n.s

Chroma 0.098 n.s n.s
Hue 0.141 n.s n.s
Thickness L* 0.018 n.s 0.019
Chroma n.s 0.117 0.039
Hue 0.005 <0.001 <0.001
Background L* n.s n.s 0.013
Chroma n.s 0.006 n.s
Hue <0.001 <0.001 0.042
Thickness*Background L* n.s 0.108 n.s
Chroma n.s <0.001 n.s
Hue 0.017 0.005 n.s
binnenwerk.indd 530 21-08-2006 12:36:17

(R2 ≤ 0.09), and in the case of a dark background the L* values are not significantly affected
by thickness in any of the instruments. This suggests that sample thickness has little or no
impact on the L* values.
Interestingly the L* values differ between instruments (over all samples L*: Hunter 51.7 ±
0.1, Minolta 44.0 ± 0.1, X-Rite 38.4 ± 0.1, p < 0.001 between all instruments). This difference
can probably be explained on the basis of the size of their respective sampling areas. The
Hunter sampling area has a diameter of 25 mm, the Minolta 8 mm and the X-Rite 6.35 mm. As
a result, the sampling area of the Hunter included myosepts (concentrated visible fat), while
this could be avoided with the two other instruments. The myosepts are white in colour and
are less translucent than the muscle tissue. This tissue will therefore reflect more light back to
the instrument than from light hitting translucent muscle tissue. The relative reflectance of the
cutlet surface was thus higher for the Hunter than for the two other instruments, leading to
higher L* values. The transparency of the muscle tissue, together with the size of the sampling
areas, is also important. A light beam emitted from the colorimetric instrument traverses the
translucent muscle tissue until it meets some reflective surface within the tissue (Figure 4).
The light reflected from this surface is less likely to hit a small than a large sampling area,
resulting in higher L* values for the largest (φ = 25 mm) than for the medium (φ = 8 mm) and
the smallest (φ = 6.35 mm) sampling areas (see above).
All three instruments’ Hue values are significantly affected by cutlet thickness (Table 1).
Interestingly, only the X-Rite’s Hue values are significantly affected by thickness when the
Colorimetic
instrument
Surface
Figure 4. Light beams emitted from the colorimetric instrument enter the translucent muscle tissue (grey
box) where they either are absorbed or reflected. Light beams that meet a reflective surface may be reflected
to the instrument for measurement, but may also be reflected outside the instrumental sampling area. The
light beams emitted by the instrument are more likely to return to a large than a small sampling area.
Light beams that meet a black background are absorbed, while those that meet a white background are
reflected.
binnenwerk.indd 531 21-08-2006 12:36:17

sampling background is white (Hue: Minolta p = n.s, Hunter p = n.s, X-Rite p < 0.001).
In contrast, all the instruments’ Hue values are significantly affected when the sampling
background is dark (Hue: Minolta p < 0.001, Hunter p < 0.001, X-Rite p < 0.001). For samplings
on a dark background thin cutlets have a less red hue (higher hue angle) than thick cutlets
for all the instruments (1 vs. 5 cm tick cutlets on dark background, Hue: Minolta 49.8 vs. 48.5
± 0.2, Hunter 43.5 vs. 42.7 ± 0.1, X-Rite 41.2 vs. 39.5 ± 0.2). This can probably be explained
by the fact that non-red wavelengths are mainly absorbed by pigments in the tissue, while the
red wavelengths are either reflected or transmitted through the tissue (hence the red colour
of salmon flesh). It is therefore mainly red light that meets the background surface, and in
the case of a dark background is absorbed. Red light is more likely to hit pigments and to be
reflected back to the instrument when traversing through a thick than a thin cutlet. As a result,
less red light is transmitted back through a thin than a thick cutlet when the background is
dark (Figure 5).
a. 50 b. 50
Relative reflectance (%)
40 40
30 30
20 20
10 10
400 500 600 700 400 500 600 700

c. 50 d. 50
Relative reflectance (%)
40 40
30 30
20 20
10 10
400 500 600 700 400 500 600 700

Wavelength (nm) Wavelength (nm)
Figure 5. Smoothed curved lines (mathemathical term: splines) (proc. tpspline, SAS) of relative reflectance
(%) over registered wave lengths by Hunter Miniscan/XE (a and b) and X-Rite Ca22 Tethered Spectrophotometer
(c and d). Dotted splines indicate 1 cm-thick cutlets and solid splines 5 cm-thick cutlets. * = significantly
different. Blue: 400-500 nm, green: 500-550, yellow: 550-600, orange: 600.650, red: 650-700. a. and c. is
when the background is dark, and b. and d. is when the background is white.
binnenwerk.indd 532 21-08-2006 12:36:18

The overall Chroma values are only significantly affected by cutlet thickness in the case of the
X-Rite spectrophotometer (Table 1). The Chroma values of both spectrophotometers, however,
are dependent on cutlet thickness in combination with background colour (samples on a
white background: Minolta p = 0.173, Hunter p < 0.001, X-Rite p = 0.021, samples on a dark
background: Minolta p = n.s, Hunter p = 0.063, X-Rite p = n.s.). The thin cutlets have higher
Chroma values than thick cutlets when the background is white (1 vs. 5 cm-thick cutlets on
white background, Chroma: Minolta 19.5 vs. 18.8 ± 0.3, Hunter 41.1 vs. 39.6 ± 0.3, X-Rite
13.8 vs. 13.1 ± 0.2). This is due to the fact that more red light is reflected back through thin
cutlets than thick cutlets, increasing colour saturation (Figure 5bd). On a dark background, only
the Hunter registered significant effects from cutlet thickness. Given the above argument of
increased colour saturation on white background when the cutlet is thin, one would expect the
opposite effect on a dark background. This is confirmed by the data, in that the Chroma values
of the Hunter, when sampling on a dark background, are higher for thick (39.9 ± 0.3) than for
thin (39.3 ± 0.3) cutlets. This contrary effect, depending on sampling background, explains
why no significant effects from cutlet thickness were found in the data-set as a whole (both
dark and white background) for the Hunter readings. The higher sensitivity of the Hunter in
comparison with the two other instruments is probably explained by the larger sampling area;
a lower proportion of the red light is returned back to the Minolta and the X-Rite (Figure 5ab
vs. 5cd). This is explained similarly as for Hue; the red light is less likely to be reflected back
to the colorimetric instrument when the sampling area is small than when it is large (Figure 4),
leading to increasing colour saturation with increasing sampling area [Chroma: Hunter (φ = 25
mm) 40.0 ± 0.1, Minolta (φ = 8 mm) 19.0 ± 0.1, X-Rite (φ = 6.35 mm) 13.5 ± 0.1].
The above discussion shows that there are differences in how the three colorimetric instruments
react to lighting conditions, cutlet thickness and sample background. However, it is not obvious
which instrument is most suitable for measuring salmon flesh colour. High sensitivity to sampling
conditions may suggest high sensitivity to colour changes in the fish muscle. On the other
hand, the high sensitivity may lead to erroneous results due to lack of standardisation during
sampling. The instrument least affected by sampling conditions is the most robust choice. The
following section discusses the instruments in terms of how they correlate with astaxanthin
content and visual colour rankings of the cutlets.
Instrumental readings vs. astaxanthin
The L* values (higher L* means more light) correlate negatively with astaxanthin, while the
Chroma (higher chroma means more saturated colour) and a* values (higher a* means more
red light) correlate positively with astaxanthin (Table 2). Higher pigment concentration means
that more of the light is absorbed and not reflected to the instrument, resulting in lower L*
values. It is particularly the green and blue wavelengths that are absorbed, while the red
wavelengths are reflected, leading to a more saturated colour and explaining the positive
correlations between astaxanthin and the Chroma and a* values. The highest correlations
between instrumental colour readings and astaxanthin are found for Croma and a* values
when thin cutlets are sampled on a white background (Table 2). The correlations for thin
cutlets on the dark background are much lower. The red light transmitted through the muscle
tissue is absorbed when it meets the dark background and is not returned to the instrument
as in the case of a white background, and as a consequence the colour readings become less
sensitive to pigment content. For thick cutlets the X-Rite’s L* values have lower correlation to
astaxanthin when sampled on a white compared to on a dark background, while there is little
binnenwerk.indd 533 21-08-2006 12:36:18

difference between backgrounds on the correlations of the Hunter colour readings (Table 2).
This is probably due to the difference in the size of the sampling areas. Light beams reflected
from the background are less likely to hit the small sampling area of the X-Rite (Figure 4) and
therefore become a random effect (the light may or may not be registered by the instrument)
introducing noise to the colour registrations. In contrast, the comparative larger sampling
area of the Hunter receives higher amounts of these beams decreasing the randomness of the
measurement. There were no significant correlations between the Hue-values from any of the
colorimetric instruments and astaxanthin (Table 2).
Many authors report the a* value to exhibit the strongest correlation to astaxanthin of the
L*a*b*-values (reviewed by Bjerkeng 2000), and spectrophotometers are generally considered
to be more accurate than tristimulus colorimeters (Wyszecki and Stiles 2000). It is therefore
as expected when the a* and Chroma values of the two spectrophotometers show the highest
correlation to astaxanthin content. The relative low correlations achieved by the Minolta are,
however, surprising (Table 2). The Minolta tristimulus colorimeter has previously shown highly
correlated relationships to astaxanthin content for fillets of Rainbow trout (Oncorhynchus
mykiss) (Storebakken and others 2004) and Arctic charr (Salvelinus alpinus L.) (Hatlen and
others 1998). One possible explanation for this discrepancy is that the colour was measured at
numerous positions over the fillet in these studies, making the total colour reading of each fish
more reliable compared to the current study where only one sampling was done per cutlet per
sampling condition. The correlations between the results from image analysis of the scanned
cutlets followed that of the instruments, except for Hue where the scanner in opposition to the
colorimetric instruments showed a significant correlation to astaxanthin (mean L* r = -0.532,
75-percentile Chroma r = 0.681, median Hue r = -0.448, 75-percentile a* r = 0.709, median b*
r = 0.607).
Visual colour assessment vs. astaxanthin
The volunteers classified the cutlets to have Roche SalmoFan™-scores of minimum 21 and
maximum 30. Median Roche SalmoFan™-score is 26. High Roche SalmoFan™-scores are strongly
related to high side-by-side ranking scores (Table 3), except for volunteer number 5 (r = 0.300).
The results from volunteer number 5 are therefore removed from the subsequent data analysis.
The results from side-by-side colour rankings have a much stronger relationship to pigment
content than the Roche SalmoFan™-scores (overall r = 0.695 vs. r = 0.413). This is agreement
with previous studies stating that the human eye has a superior ability to distinguish between
objects of slightly different colour when the object are placed side-by-side (Ray 1999). It
is, however, clear that many of the volunteers obtained a high correlation to astaxanthin
also for the Roche SalmoFan™-scores (Table 4). Volunteer 2, 13 and 20 obtained very low
correlations to astaxanthin and are therefore considered as outliers and removed from the
subsequent analysis. The overall correlations between colour ranking, Roche SalmoFan™-scores
and astaxanthin then increases to r = 0.765 and r = 0.462, respectively. A strong positive
correlation between colour card-scores and pigment content is in agreement with studies by
Christiansen and others (1995), Storebakken and others (2004) and Baker and others (2002)
who all found Roche SalmoFan™-scores or Roche Colour Card for Salmonids-scores to be very
good indicators of astaxanthin content. Other studies have, however, found little correlation
between astaxanthin and Roche SalmoFan™ -scores (Johnston and others 2004). This can
probably be explained by differences in the levels of astaxanthin content; the Atlantic salmon
used in the study by Johnston and others had a astaxanthin content between 8-12 mg kg‑1,
binnenwerk.indd 534 21-08-2006 12:36:18

Table 2. Spearman’s rank correlation coefficient between instrumental colour readings and astaxanthin
content under different sampling conditions.
Thickness Background Light Colour Minolta Hunter X-Rite
1 White On L* 0.036 -0.285 -0.152

1 White Off L* 0.075 -0.226 -0.335
1 Dark On L* -0.156 -0.501 -0.472
1 Dark Off L* -0.332 -0.480 -0.365
5 White On L* -0.377 -0.355 -0.307
5 White Off L* -0.024 -0.281 -0.374
5 Dark On L* -0.383 -0.399 -0.215
5 Dark Off L* -0.092 -0.400 -0.224
1 White On a* 0.498 0.780 0.794
1 White Off a* 0.589 0.807 0.798
1 Dark On a* 0.195 0.586 0.463
1 Dark Off a* 0.254 0.591 0.350
5 White On a* 0.361 0.743 0.565
5 White Off a* 0.329 0.735 0.614
5 Dark On a* 0.230 0.705 0.627
5 Dark Off a* 0.129 0.689 0.719
1 White On b* 0.311 0.560 0.651
1 White Off b* 0.359 0.620 0.672
1 Dark On b* 0.051 0.541 0.081
1 Dark Off b* 0.123 0.570 0.163
5 White On b* 0.116 0.653 0.296
5 White Off b* 0.317 0.662 0.460
5 Dark On b* 0.189 0.589 0.580
5 Dark Off b* 0.090 0.556 0.601
1 White On Hue -0.137 -0.185 -0.030
1 White Off Hue -0.005 -0.177 -0.126
1 Dark On Hue -0.143 -0.047 -0.361
1 Dark Off Hue -0.134 -0.067 -0.308
5 White On Hue -0.310 -0.232 -0.317
5 White Off Hue 0.023 -0.232 -0.078
5 Dark On Hue -0.203 -0.266 0.140
5 Dark Off Hue -0.048 -0.250 0.113
1 White On Chroma 0.415 0.732 0.716
1 White Off Chroma 0.441 0.776 0.779
1 Dark On Chroma 0.080 0.549 0.335
1 Dark Off Chroma 0.152 0.552 0.284
5 White On Chroma 0.147 0.698 0.389
5 White Off Chroma 0.331 0.695 0.553
5 Dark On Chroma 0.230 0.699 0.633
5 Dark Off Chroma 0.057 0.665 0.645
binnenwerk.indd 535 21-08-2006 12:36:19

Table 3. Spearman’s rank correlations between Roche SalmoFan™-scores and side-by-side colour ranking
measured on Atlantic salmon muscle.
Judge r Judge r Judge r
1 0.975 8 0.738 15 0.872

2 0.975 9 0.821 16 0.949
3 1.000 10 0.975 17 0.975
4 0.667 11 0.872 18 0.975
5 0.300 12 0.975 19 0.949
6 0.949 13 0.791 20 0.975
7 1.000 14 1.000 21 1.000
Table 4. Spearman’s rank correlations for Roche SalmoFan™-scores and side-by-side colour ranking vs.
astaxanthin measured on Atlantic salmon muscle.
Volunteer Roche Rank Judge Roche Rank Judge Roche Rank
1 0.667 0.700 8 0.580 0.900 15 0.669 0.900

2 0.359 0.400 9 0.667 0.500 16 0.738 0.900
3 0.700 0.700 10 0.667 0.700 17 0.616 0.500
4 0.564 0.500 11 0.616 0.900 18 0.872 0.800
5 12 0.975 0.900 19 0.738 0.700
6 0.580 0.800 13 -0.369 0.200 20 0.359 0.300
7 1.000 1.000 14 0.700 0.700 21 0.900 0.900
while the figure in Storebakken and others was 2-8 mg kg-1, in the current study, 2-6 mg
kg-1, and in that of Baker and others, 0-4 mg kg-1. Rønsholdt (2005), in a review of several
studies, concludes that increased carotenoid content has a non-linear relationship to colour
as measured by instrumental colorimeters; a relatively high correlation at low concentrations
(0-4 mg kg-1), some correlation at medium concentrations (4-8 mg kg-1), and no correlation
between astaxanthin and registered colour values for higher concentrations of astaxanthin (>8
mg kg-1). It is therefore as expected when the correlation between Roche SalmoFan™-scores
and astaxanthin in the study of Johnston study is low, while they are relatively high in other
studies involving fish with lower concentrations of muscle astaxanthin. Other factors that
must be taken into account are sampling conditions and the subjectivity of the visual colour
assessment. In the current study the colour assessment carried out on all the cutlets by one
person (the qualified researcher) showed no significant correlation (r = 0.158) to astaxanthin,
which demonstrates the fact that single-person colour assessment is difficult. It also illustrates
the importance of correct lighting conditions. When the trained researcher performed the colour
assessment under conditions of daylight for the five subset cutlets, the correlation increased
to r = 0.900 compared to r = 0.735 for the original assessment carried out in the experimental
room with slightly yellow fluorescent lighting. The relatively high correlation for the subset
even under poor lighting conditions vs. the correlation for the entire subset r = 0.158 shows
that it is easier to make a correct colour differentiation when sampling cutlets with relatively
binnenwerk.indd 536 21-08-2006 12:36:19

large differences in colour (the subset was selected so as to span the whole range from light to
dark) in rapid succession, compared to sampling cutlets of more similar colour over time (the
duration of the experiment).
Instrumental readings vs. visual colour assessment
The instrumental Chroma values correlate well with visual side-by-side colour ranking for all
the three colorimetric instruments when thin cutlets were assessed on a white background
(Table 5). This is the instrumental sampling condition closest to the conditions under which
the visual colour assessments of the subset were made (1 cm thick cutlet, white background
and daylight). The L* values of the Hunter generally have a strong negative correlation with
the visual colour rankings (Table 5). This is as expected, since the side-by-side colour ranking
ranges from 5 (darkest, low L* value) to 1 (lightest, high L* value). The colour values of the
Minolta and X-Rite correlate with the side-by-side colour rankings in a much less consistent way.
As pointed out above, the relatively small sampling areas of these two instruments means that
light reflected in the tissue is less likely to reach the instrumental sampling area, introducing
a random effect that will distort the samplings. This is clearly illustrated by the fact that the
Table 5. Spearman’s rank correlation coefficient between manual side-by-side colour rankings and instrumental
readings measured on Atlantic salmon muscle.
Thick-ness Back-ground Light Colour Minolta Hunter X-Rite
1 White On L* -0.529 -0.343 0.534

1 White Off L* -0.400 -0.342 0.683
1 Dark On L* -0.213 -0.603 0.364
1 Dark Off L* -0.491 -0.559 0.424
5 White On L* 0.124 -0.789 -0.527
5 White Off L* 0.020 -0.722 -0.306
5 Dark On L* -0.636 -0.300 -0.402
5 Dark Off L* 0.038 -0.287 -0.255
1 White On Hue -0.458 -0.629 0.465
1 White Off Hue -0.469 -0.714 0.107
1 Dark On Hue -0.068 -0.384 -0.049
1 Dark Off Hue 0.011 -0.277 -0.120
5 White On Hue -0.031 -0.702 0.485
5 White Off Hue -0.034 -0.639 -0.394
5 Dark On Hue 0.544 -0.560 -0.246
5 Dark Off Hue 0.312 -0.608 -0.191
1 White On Chroma 0.127 0.552 0.696
1 White Off Chroma 0.596 0.575 0.496
1 Dark On Chroma 0.120 0.256 0.078
1 Dark Off Chroma 0.274 0.272 -0.223
5 White On Chroma -0.125 0.252 -0.277
5 White Off Chroma -0.582 0.270 -0.267
5 Dark On Chroma 0.412 0.360 0.246
5 Dark Off Chroma -0.017 0.368 0.242
binnenwerk.indd 537 21-08-2006 12:36:20

trends were much clearer for these instruments when compared with astaxanthin measurements
of all the cutlets (Table 2), rather than of a small subset. This suggests that the X-Rite and
the Minolta need more samplings to obtain consistent results than does the Hunter. The image
analysis of the scanned cutlets achieved the highest correlations with visual side-by-side colour
ranking of all the instrumental readings (mean L*: r = -0.836, mean Hue: r = 0.110, mean
Chroma: r = 0.951).
Conclusions
The results of the three colorimetric instruments were highly influenced by sampling conditions.
The instruments gave different values in absolute terms, depending on sampling conditions.
Sampling conditions also influenced how the colour readings correlated with astaxanthin
content. The highest correlations (r > 0.700) between astaxanthin and instrumental colour
readings were obtained for the a* and Chroma values of the spectrophotometers when sampling
on 1 cm thick cutlets on white background. The visual colour assessments correlated well (r
> 0.700) with astaxanthin for most of the volunteers. A surprising result of this study is that
the standard flatbed scanner achieved as high, or higher, correlations with both astaxanthin
content and visual colour rankings as the dedicated colorimetric instruments. This suggests
that these relatively expensive colorimetric instruments could be replaced by a flatbed scanner
for instrumental colour analysis of fish flesh.
Acknowledgements
The Research Council of Norway financially supported the project through project grants
nos.153178/120. The practical part of the project was carried out at the National Institute of
Nutrition and Seafood Research (NIFES). We are especially grateful to the staff at NIFES who
were involved in the visual colour assessment of the cutlets. We are also grateful to Professor
Anders Kiessling at the Norwegian University of Life Sciences (UMB) for participating in the
discussion leading up to the experiment and to Kjersti Ask for performing the HPCL analysis of
the pigment content.
References
Baker RTM, Pfeiffer AM, Schöner FJ, Smith-Lemmon L. 2002. Pigmenting efficacy of astaxanthin and canthaxanthin in
fresh-water reared Atlantic salmon, Salmo salar. Animal Feed Sci Tech 99(1):97-106.
Bjerkeng B. 2000. Carotenoid pigmentation of salmonid fishes - recent progress. In: Cruz -Suárez LE, Ricque-Marie D,
Tapia-Salazar M, Olvera-Novoa MA, Civera-Cerecedo R, editors. Avances en Nutrición Acuícola V. Memorias del V
Simposium Internacional de Nutrición Acuícola. 19-22 Noviembre, 2000. Mérida, Yucatán. p 71-89.
Calvo C, Salvador A, Fiszman SM. 2001. Influence of colour intensity on the perception of colour and sweetness in
various fruit flavoured yoghurts. Eur Food Res Technol 213(2):99-103.
Cardinal M, Knockaert C, Torrissen O, Sigurgisladottir S, Mørkøre T, Thomassen M, Valle JL. 2001. Relation of smoking
parameters to the yield, colour and sensory quality of smoked Atlantic salmon (Salmo salar). Food Res Int 34:537-
550.
Cardinal M, Gunnlaugsdottir H, Bjørnevik M, Ouisse A, Vallet JLV, Leroi F. 2004. Sensory characteristics of cold-smoked
Atlantic salmon (Salmo salar) from European market and relationships with chemical, physical and microbiological
measurements. Food Res Int 37:181-193.
Chambers IVE, Bowers J. 1993. Consumer perception of sensory quality in muscle foods: sensory characteristics of flesh
influence consumer decisions. Food Technol 47:116-120.
binnenwerk.indd 538 21-08-2006 12:36:20

Christiansen R, Struksnæs G, Estermann R, Torrissen OJ. 1995. Assessment of flesh colour in Atlantic salmon, Salmo
salar L. Aquacult Res 26:311-321.
Hatlen B, Jobling M, Bjerkeng B. 1998. Relationships between carotenoid concentration and colour of fillets of Arctic
charr, Salvelinus alpinus (L.), fed astaxanthin. Aquacult Res 29(3):191-202.
Johnson RA, Bhattacharyya GK. 2000. Statistics: Principles and Methods. New York: John Wiley & Sons, Inc. 723 p.
Johnston IA, Manthri S, Bickerdike R, Dingwall A, Luijkx R, Campbell P, Nickell D, Alderson R. 2004. Growth performance,
muscle structure and flesh quality in out-of-season Atlantic salmon (Salmo salar) smolts reared under two different
photoperiod regimes. Aquaculture 237(1-4):281-300.
Kiessling A, Espe M, Ruohonen K, Mørkøre T. 2004. Texture, gaping and colour of fresh and frozen Atlantic salmon flesh
as affected by pre-slaughter iso-eugenol or CO2 anaesthesia. Aquaculture 236(1-4):645-657.
Nickell DC, Springate JRC. 2001. Pigmentation of Farmed Salmonids. In: Kestin SC, Warris PD, editors. Farmed Fish
Quality. Cornwall: Fishing News Books. p 58-75.
Ray SF. 1999. Limits of human visual perception. In: Ray SF, editor. Scientific Photography and Applied Imaging.
Oxford: Focal Press. p 7-10.
Robb DHF. 2001a. The Relationship Between Killing Methods and Quality. In: Kestin SC, Warris PD, editors. Farmed Fish
Quality. Cornwall: Fishing News Books. p 220-233.
Robb DHF. 2001b. Measurement of Fish Flesh Colour. In: Kestin SC, Warris PD, editors. Farmed Fish Quality. Cornwall:
Fishing News Books. p 298-306.
Rønsholdt B. 2005. Can carotenoid content in muscle of salmonids be predicted using simple models derived from
instrumental colour measurements? Aquacult. Res. 36(5):519-524.
Rørå AMB, Kvåle A, Mørkøre T, Rørvik K-A, Steien SH, Thomassen MS. 1998. Process yield, colour and sensory quality of
smoked Atlantic salmon (Salmo salar) in relation to raw material characteristics. Food Res Int 31(8):601-609.
Sigurgisladottir S, Torrissen O, Lie Ø, Thomassen M, Hafsteinsson H. 1997. Salmon quality: Methods to determine the
quality parameters. Rev Fish Sci 5(3):223-252.
Storebakken T, No HK. 1992. Pigmentation of rainbow trout. Aquaculture 100(1-3):209-229.
Storebakken T, Sørensen M, Bjerkeng B, Hiu S. 2004. Utilization of astaxanthin from red yeast, Xanthophyllomyces
dendrorhous, in rainbow trout, Oncorhynchus mykiss: effects of enzymatic cell wall disruption and feed extrusion
temperature. Aquaculture 236(1-4):391-403.
Wathne W, Bjerkeng B, Storebakken T, Vassvik V, Odland AB. 1998. Pigmentation of Atlantic salmon (Salmo salar) fed
astaxanthin in all meals or in alternating meals. Aquaculture 159(3-4):217-231.
Wyszecki G, Stiles WS. 2000. Color Science, Concepts and Methods, Quantitative Data and Formulae. Second edition.
New York: John Wiley and Sons, Inc. 950 p.
binnenwerk.indd 539 21-08-2006 12:36:20

binnenwerk.indd 540 21-08-2006 12:36:20
Revision of analytical methodologies to verify the production
method of fish
Iciar Martinez
SINTEF Fisheries and Aquaculture, Ltd., N-7465 Trondheim, Norway
Abstract
There are currently no officially recognized standard methods to confirm the production method
of fish. This review examines the application of some analytical methods to discriminate farmed
and wild fish, which is important for correct consumer information, food safety and fisheries
management. Morphological, genetic, lipid and protein analyses and analysis of carotenoids
have been used, mostly on salmonids. Reliable analytical methods and databases need to
be developed to identify farmed and wild fish from both traditionally cultivated and new
species.
Keywords: authenticity, traceability, food safety, farmed fish, fish feed, fish oils, fish muscle
Introduction
The EU Commission regulation No 2065/2001 of 22 October 2001 has laid down detailed rules
for the application of Council Regulation (EC) No 104/2000 as regards informing consumers
about fishery and aquaculture products. The information includes specification of the commercial
designation and scientific name, method of production of a species (“caught” or “caught in
freshwater” or “farmed” or “cultivated”) and the area in which it was caught. In the case of
cultivated species, article 5 of the regulation indicates that a reference should be made to the
country in which the product undergoes the final developmental stage.
Within the last 10-15 years, aquaculture has become a major commercial activity in the world.
European countries farming Atlantic salmon (Salmo salar) include Norway, Ireland, Scotland, but
Atlantic salmon is also farmed at least in Chile, Canada and Tasmania. Farmed and wild Atlantic
salmon have different prices, wild salmon being most expensive, followed by organically farmed
salmon and then traditionally farmed fish. The location where the fish is fished or farmed is also
important since some areas are considered to be clean and others polluted, (Foran and others
2004; Hites and others 2004; Jacobs and others 2002; Madeniian and others 2002), because
fish from some areas is protected and finally because consumers usually favor and are willing
to pay higher prices for products from certain regions.
Correct identification and labelling of fish as farmed or wild is also of relevance for the authorities
in order to protect wild and endangered stocks. For example in the case of Atlantic salmon,
farmed salmon may escape and be captured by fishermen that would consider it wild, and price
it accordingly with the consequent misinformation to customers. Mapping of the extent of
escaped salmon is important, in addition to the issue of consumer protection, to protect wild
populations: escaped salmon is considered at present a high risk regarding the contamination of
binnenwerk.indd 541 21-08-2006 12:36:20

Iciar Martinez
the genetic stocks of wild populations that may then disappear as such. Also, many populations
of wild Atlantic salmon are endangered and therefore they must not be harvested at all.
In the case of cod (Gadus morhua) the situation is similar. Some wild populations are protected
due to overexploitation and therefore, unless illegal captures have occurred, they must not be
found as available for human consumption. On the other hand, farming of cod is relatively new
and as of now not as successful as the farming of salmon: deformations, a greyish flesh colour
(Luten and others 2002) and abnormal features in the fillet (Cooper and Midling 2004) are still
common, although it is to be expected that technical improvements and a deeper knowledge
of the fish biology and physiology will help to overcome these problems in the near future.
There is however an “intermediate” type of cod: wild caught cod that is kept alive, i.e. farmed
for a period of time until it is convenient to slaughter it. Luten and others (2002) found that
consumers gave similar quality profiles to wild cod and to wild caught farmed cod, although the
composition of the wild caught farmed specimens is affected by the composition and amount
of feed received.
Genetic contamination of wild stocks with escaped farmed cod is not yet considered a risk, but
it will when farming takes over, specially if the wild stocks are still reduced to low numbers.
Morphological analyses
There are no official guidelines to differentiate farmed from wild cod or salmon based on
morphological characters, although there are some general aspects that may be used for that
purpose. Thus, morphological abnormalities, such as spinal deformities, rare in wild specimens
are known to occur frequently in many species of intensively reared fish (Daoulas and others
1991; Fraser and others 2004), including salmonids (Dabrowski and others 1990; Toften and
Jobling 1996). Farmed cod specimens also show morphological deformities.
Colour is also a variable that may be used to differentiate farmed from wild fish. The diets
of farmed Atlantic salmon are formulated to yield a nice bright pinkish-reddish colour, while
wild specimens may vary in colour according to their diet. Wild salmon is usually caught
when migrating back to the rivers for spawning and may display in higher or lower degree the
morphological characters associated to sexual maturation: deformation of the jaw, stronger
red pigmentation of the skin, and muscle depletion and low fat content, due to the effort of
migration. Farmed salmon on the hand is selected to reach a commercially interesting size (4-
5 kg) in the shortest possible time and with no display of signs of sexual maturation.
Farming of cod is relatively new compared to farming of other species and therefore the
farming conditions have not been optimized yet, which permits at present to differentiate
farmed from wild cod. Farmed Atlantic cod has a body morphology different from wild-captured
cod. The most prominent differences are the higher condition factor, larger liver and smaller
head (Gildberg 2004) as well as backbone malformations in farmed specimens. Farmed, but
seldom wild, cod often present as well unattractive black lines in the fillet due to deposition
of melanin in muscle and blood vessels. Cooper and Midling (2004) have examined the role of
tyrosinase on these melanin depositions without finding a clear answer. The flesh of farmed
cod presents a translucent greyish aspect, in contrast to the white opaque colour of the wild.
Farmed cod flesh also has higher water content and is less firm than the flesh of wild cod.
Finally the liver in farmed cod is much bigger than in wild fish (Jobling 1988). However, the
binnenwerk.indd 542 21-08-2006 12:36:20

Revision of analytical methodologies to verify the production method of fish
characteristics of farmed cod are continuously improving as farmers become more experienced
and the requirements of the species are mapped.
Part of the farming success consists in reaching a high survival rate from fertilization of eggs
until the fish reaches a commercial size. In nature, however, most of the eggs would not
reach maturity. Farming is therefore permitting the survival of fish that would otherwise be
selected for destruction, for example if they had abnormal development. It is not known yet
whether some of the malformations noticed in farmed cod are because the fish is a carrier of
genetic malformations or because the farming conditions require to be improved for example
by selecting the correct photoperiod, temperature, exercise regime and physical characteristics
of the feed and chemical composition of the diet.
Individual tagging of fish
Individual tagging of fish would be optimal to differentiate farmed from wild fish and to identify
the farm where the fish were bred. There are several types of tags commercially available
manufactured using different technologies that may carry variable amounts of information.
Unfortunately, the process of tagging itself implies extra costs to the breeder, and when the
origin of escaped fish is identified, the farmers received very high penalty fees. Therefore
implementation of this type of technology has encountered reluctance. The situation would be
reversed if fish from some particular farm or region could achieve higher market prices, which
seems to be the case for fish from some European regions, such as Scotland and France.
Genetic analyses
Genetic analyses have traditionally been based in the analysis of proteins and more recently of the
DNA (Hoelzel 1992). Doyle and others (1991) proposed that the genetic diversity of aquacultured
stocks of fish should be maintained and their genetic impact on wild stocks minimized by using
breeding programs designed to generate genetic diversity. If this policy had been followed it
would be relatively difficult to find markers for wild and farmed fish, since diversity would be
one of the selected traits in the farmed fish. However, in most breeding programs the fish are
indeed selected based on commercially interesting traits such as growth performance (Friars and
others 1995; Herbinger and others 1999) and resistance to diseases or to stress (Fevolden and
others 2003). Moreover, in the later years research has been carried out to find genetic markers
for traits of interest. Indeed, effort is being invested in mapping the whole genome of salmon
and cod with the intention of optimizing the farming of these species.
Analysis of protein isoforms also has the potential to generate genetic markers. In addition to
some of the very polymorphic enzymes commonly used for population genetic analyses (Hoelzel
1992; Manchenko 1994), many proteins, including most or all myofibrillar proteins are present
as isoforms. In particular the isoforms of the myosin light chains are very polymorphic and
they can be used to identify the species and tissue (Martinez and others 1990a, 1990b, 1991)
as well as the breeding stock (Martinez and Christiansen 1994; Martinez and others 1990c) and
the developmental-stage (Martinez and others 1991).
It is theoretically possible to differentiate wild from farmed fish based on these markers with
a relatively high probability of making correct classifications. However, this requires that the
markers should be identified and collected in a database to which the public or at least some
binnenwerk.indd 543 21-08-2006 12:36:21

Iciar Martinez
laboratories should have access. Such databases are not currently available and need to be
constructed.
Analyses of the fat content
The proportion of body fat in storage compartments in fish correlates with the total amount of
fat in the feed and the fatty acid profile of the storage fat (mainly triacylglycerols) reflects the
fatty acid composition of the diet (Henderson 1996; Jobling 1994, 2004; Jobling and others
2002, Tocher and others 2000; Watanabe 1982). Conventionally farmed Atlantic salmon has a
fat content varying between 10% to 20%, while in wild specimens it varies from 4-10%.
Diets used for intensive aquaculture have been mainly based on the use of fish meal and fish
oils. This imposes a serious risk to the world fisheries specially to pelagic species (Naylor and
others 2000) most of which have either reached sustainable limits or are already overexploited
and endangered (Sargent and Tacon 1999). Therefore, there is great interest in the development
of formulated diets where fish oil and meal is at least partially substituted by vegetable oils
(Bell and others 2002, 2003; Kaushik and others 1995; Mooney and others 2002; Nichols and
others 2002; Watabe and others 1999) and proteins (Burel and others 2000; Carter and Hauler
2000; Gomes and others 1995; Kaushik and others 1995). An additional reason to use vegetable
oils to substitute fish oils is the possibility to reduce the load of toxic dioxins (the generic
term given to polychlorinated dibenzo-p-dioxins and dibenzofurans) and PCBs (polychlorinated
biphenyls) and PCB-like compounds (Bell and others 2005). PCBs and dioxins are biomagnified
as they progress through the food chain, concentrating in the fatty tissues of land animals and
fish, predominantly in the Northern hemisphere.
Changes in the amount and composition of the feeds are reflected in the amount of storage fat
and its fatty acid profile (Mooney and others 2002; Nichols and others 2002). Thus, while the
fatty acid profile in wild fish will be dictated by the composition of the natural prey (Jobling
2004), farmed fish will reflect the vegetable and fish oils contained in the feed. Accordingly,
the fatty acid profile in the storage tissue, which is muscle in salmonids but liver in cod
(Jobling 1988) may be used as a marker to distinguish between farmed and wild fish (Aursand
and Axelson 2001; Aursand and others 1994a, 2000, 2004; Igarashi and others 2002; Bell and
others 2002, 2003; Villarreal and others 1994). Fatty acid profiling has also been proposed as
a forensic tool for conservation biologists and law enforcement officials to distinguish cultured
red drums from illegally marketed wild red drums (Villarreal and others 1994).
The fatty acid profile of the polar lipids (phospholipids), which are the main constituents of
membranes, does not resemble so closely the composition of the feed (Lie and others 1986; Dos
Santos and others 1993) and it has been successfully used to identify the species in canned
tuna (Medina and others 1997). Fatty acid profiling of several tissues has also proved useful to
differentiate among several species and stocks of Sebastes (Joensen and Grahl-Nielsen 2000,
2001, 2004).
Robin and others (2003) and Jobling (2004) have shown that a dilution model could be used to
describe the changes in the fatty acid profiles of Atlantic salmon muscle following a change in
the dietary fatty acid source and composition. Studies in cod (Morais and others 2001) indicate
that the fatty acid composition of the feed is mainly reflected in the fatty acid profile of the
liver in this species, although the fatty acid profile of muscle is also altered. The practical
binnenwerk.indd 544 21-08-2006 12:36:21

application of such a model would be the prediction of fatty acid profiles within the fillet in
salmon and the liver in cod following a dietary shift.
The fatty acid profile will also indicate the production method (wild, farmed, organic) and give
an indication of the natural-to-artificial feed ratio used. Artificial feeds usually have a higher
content of typically vegetable fatty acids (C18:2n6, C18:1n9), at the expense of the typical fish
oils (C22:6n3, C20:5n3, C20:1n9 and C22:1) which should be the main constituent for wild and
organically farmed fish. Analyses of the fat content and composition are therefore necessary to
verify claims of organically produced fish.
There are several methods suitable to estimate total fat and fatty acid profiles, including gas
chromatography (GC), nuclear magnetic resonance (NMR) and non-destructive methods such as
low field NMR and near infrared spectroscopy (NIR), but the correct classification of samples
demands the construction of databases containing information from authentic samples from
wild, conventionally and ecologically farmed specimens for each fish species.
Recently, the total amount of omega-3 fatty acids and the content of docosahexanoic acid (DHA)
has been estimated by 1H NMR spectroscopy and this method gave values comparable to those
obtained by GC or 13C MR (Aursand and others 1994b; Sacchi and others 1993). The analysis
can be carried out with a high degree of automation and with short acquisition times (2-5
min). Additional relevant information, however, such as the identification and quantification
of individual fatty acids, total saturated, mono- and polyunsaturated fatty acids, omega-3 and
omega-6 and the preferential positional distribution of 20:5, 22:5 and 22:6 in triacylglycerols
does require 13C NMR analyses (Aursand and others 1994b). 2H and 13C NMR isotope ratios have
also been used to differentiate wild from farmed Atlantic salmon (Aursand and others 2000;
Aursand and Axelson 2001) and 1H and 2H NMR spectroscopy to discriminate between essential
fatty acids of plant and animal origin (Aursand and others 1997).
Finally, an estimation of whether a fish has been farmed or is wild may also be made by measuring
its total fat content, which is usually higher in farmed fish. This estimation may be invalid if
the fish was originally farmed but escaped (it will show a low fat content) or in the case of
organically farmed fish, whose fat content and profile should resemble that of wild individuals.
Carotenoid content
Carotenoids comprise several groups of natural pigments, one of which is astaxanthin. Astaxanthin
is produced principally by plants, yeast and microalgae and it occurs in several different forms:
stereoisomers, geometric isomers, and free or esterified forms. In its natural state, astaxanthin
is usually associated with other molecules: it is often complexed with proteins, for example in
the chromophore in the blue, green, and yellow pigments of lobsters; it may simply be dissolved
in the lipid fraction of complex molecules such as egg lipoproteins, or it may actually be bound
chemically to molecules such as fatty acids to form esters (Bernhard 1990). Free unbound
astaxanthin is less stable and although it may also be found in cells its occurrence is rare. The
most common geometric configuration in both synthetic and natural astaxanthin is the most
thermodynamically stable all-E (all-trans) isomer, but while astaxanthin from natural sources
tends to occur predominantly as either the 3S,3’S or 3R,3’R stereosimers, the meso (3R,3’S) isomer
is the most abundant in synthetic astaxanthin, produced as the free (unesterified) xanthophyll and
as a 1:2:1 mixture of the three stereoisomers: 3S,3’S, 3R,3’S, and 3R,3’R (Bernhard 1990).
binnenwerk.indd 545 21-08-2006 12:36:21

Iciar Martinez
Most animals cannot synthesize carotenoids de novo, and must obtain these pigments from
their diet, i.e., plants or algae (Prache and others 2005), copepods are an exception and have
been shown to be able to synthesize astaxanthin (Andersson and others 2003). Astaxanthin,
which is responsible for coloration of the flesh of salmonids, is added to their feeds in order
to make up for the lack of a natural dietary source of the pigment and it is essential for their
proper growth and survival (Torrissen and Christiansen 1995). The major form currently being
used in fish feeds is synthetic astaxanthin (McCoy 1999). The higher abundance of the isomer
3R,3’S fed to farmed fish is reflected in its composition: as shown by Lura and Sægrov (1991)
the proportion of the (3R, 3’S)-isomer is 10 higher in the farmed than in the wild salmon. This
criterium has been satisfactorily used to identify the origin of Atlantic salmon populations
(whether wild native of colonizing from escaped farmed fish) in a Norwegian river by Sægrov
and others (1997). Astaxanthin deposition in Atlantic salmon tissues has also been shown to
be influenced by the type of dietary oil used in the feed (Bjerkeng and others 1999).
Thus, analysis of the astaxanthin content and isomers and the pattern of deposition in different
tissues can be useful to discriminate wild from farmed species, but the application is currently
limited to species whose feed may contain these pigments, such as some crustaceans, salmonids
and the breeders in some other species. If the addition of these compounds, which seem to
have a positive effect on human health (Baker and Günther 2004), was to be allowed to all
fish feeds, then these analytes could have a wider application in the identification of the
production method. Incidentally, carotenoids have been mentioned as potential tracers to verify
traceability information on meat and milk of small ruminants (Prace and others 2005).
Analyses of the protein/enzyme profiles in some tissues
As already mentioned, fish feeds are being developed that contain protein of vegetable origin
to substitute fish meal (Burel and others 2000; Carter and Hauler 2000; Gomes and others
1995; Kaushik and others 1995). The source of protein is very important because most teleost
fish species are adapted to use protein as a preferred energy source over carbohydrate, and
thus require high levels of dietary protein (30–60%) (Cowey 1995). The essential amino acid
requirements of fish correlate well with the amino acid composition of the whole animal and
to a certain extent that of the muscle tissue alone (Mambrini and Kaushik 1995; Wilson and
Cowey 1985). The use of plant proteins in feed diet formulations requires the supplementation
with amino acids, because the amino acid profiles of plant proteins do not meet the essential
amino acid requirements of fish (Krogdahl and others 1994). Also, plant ingredients contain
different antinutritional factors of different nature and at different concentrations, which may
have adverse effects in fish (Francis and others 2001; Krogdahl and others 1994; Moyano and
others 1991; Vielma and others 2000, 2002).
Martin and others (2003) used a proteomics approach to study the protein profiles of livers of
rainbow trout that had been fed diets containing different proportions of plant -soy- ingredients
and found that growth rates of fish were not altered by the dietary treatments, although
protein consumption was greater for fish fed diets with higher amounts of soy protein. Their
work led to the identification of 33 proteins including heat shock proteins, enzymes, fatty
acid binding proteins and structural proteins that were differentially expressed between the
livers of fish fed diets with lower and higher amount of soy protein. The most likely reason
for altered metabolism was considered to be the co-purification of soy protein with vegetal
antinutritional factors such as phytoestrogens, antigenic agents and other compounds such as
binnenwerk.indd 546 21-08-2006 12:36:21

phorbol esters. Similar diverse alterations in gene expression have been shown in rats fed soy
protein extracts (Iqbal and others 2002). From this work it can be concluded that a proteomics
analysis of certain tissues of the fish can be used to estimate the ingestion of artificial feeds,
an indication that the fish had been farmed and are not wild. This approach has therefore the
potential to discriminate as well between organically and intensively farmed fish. In order to
apply it systematically to discriminate farmed from wild fish, one would have to map first the
changes provoked by the actual diets used, and identify the differentially expressed proteins.
Once the altered proteins are identified, it is possible to simplify the analysis by targeting only
those proteins, so that it would be possible to design immunological tests, which are easier to
perform, and many of them can be made semi-quantitative and in a portable format.
Conclusions
There are currently no standardized analytical methods suitable to verify the production
method. Optimally, farmed fish should be tagged, preferably using micro-devices such as chips
that are easy to read. However that is not enough: the chip may be lost, not inserted or
falsified. Therefore, it is necessary to develop analytical methods to verify the information
provided by the traceability data, and suitable databases containing the profiles of authentic
samples. Parameters that need to be recorded are: lipid, protein and DNA profiles of the fish
species from different locations and of their feeds. Suitable techniques are genetic analyses,
gas chromatography, nuclear magnetic resonance, analysis of isomers for carotenoids and
proteomics. This work has only started for salmonid fish and needs to be applied also to newer
relevant species such as cod, seabream, halibut, or shellfish.
Acknowledgements
This work was performed within the Integrated Research Project SEAFOODplus, contract No
FOOD-CT-2004-506359. The financing of the work by the European Union and the Norwegian
Research Council is gratefully acknowledged, as well as the help and useful comments provided
by Dr Trine F. Galloway (Biomar, Norway).
References
Andersson M, Van Nieuwerburgh L, Snoeijs P. 2003. Pigment transfer from phytoplankton to zooplankton with emphasis
on astaxanthin production in the Baltic Sea food web. Mar Ecol-Prog Series 254:213-224.
Aursand M, Axelson D. 2001. Origin recognition of wild and farmed salmon (Norway and Scotland) using 13C NMR spectroscopy
in combination with pattern recognition techniques. In: Webb GA, Belton PS, Gil AM, Delgadillo I, editors. Magnetic
Resonance in Food Science: A View to the Future. UK: RSC-The Royal Society of Chemistry. 227-231.
Aursand M, Bleivik B, Rainuzzo JR, Jørgensen L, Mohr V. 1994a. Lipid distribution and composition of commercially
farmed Atlantic salmon (Salmo salar). J Sci Food Agric 64:239-248.
Aursand M, Rainuzzo JR, Grasdalen H. 1994b. Quantitative High Resolution 13C and 1H Nuclear Magnetic Resonance of
omega-3 fatty acids from white muscle of Atlantic salmon (Salmo salar). J Am Oil Chem Soc 70:971-981.
Aursand M, Mabon F, Martin GJ. 1997. High-resolution H-1 and H-2 NMR spectroscopy of pure essential fatty acids for
plants and animals. Magn Res Chem 35:S91-S100.
Aursand M, Mabon F, Martin GJ. 2000. Characterization of farmed and wild salmon (Salmo salar) by a combined use of
compositional and isotopic analyses. JAOCS 77:659-666.
Aursand M, Dauksas E, Schei M, Halvorsen J, Sandbakk M, Axelson D, Martinez I. 2004. Confirmation of the origin of
salmon - Fat analyses reflect the fish diet. In: Proceedings of the WEFTA Conference 2004, Rehbein H, Karl H,
Manthey-Karl M, Oehlenschläger J, Schubring R, editors, Hamburg, 158-160
binnenwerk.indd 547 21-08-2006 12:36:21

Iciar Martinez
Baker R, Gunther C. 2004. The role of carotenoids in consumer choice and the likely benefits from their inclusion into
products for human consumption. TIFST 15:484-488.
Bell JG, Henderson RJ, Tocher DR, McGhee F, Dick JR, Porter A, Smullen R, Sargent JR. 2002. Substituting fish oil with
crude palm oil in the diet of Atlantic salmon (Salmo salar) affects tissue fatty acid compositions and hepatic fatty
acid metabolism. J Nutr 132:222–230.
Bell JG, Tocher DR, Henderson RJ, Dick JR, Crampton VO. 2003. Altered fatty acid compositions in Atlantic salmon
(Salmo salar) fed diets containing linseed and rapeseed oils can be partially restored by a subsequent fish oil
finishing diet. J Nutr 133:2793–2801.
Bell JG, McGhee F, Dick JR, Tocher DR. 2005. Dioxin and dioxin-like polychlorinated biphenyls (PCBs) in Scottish
farmed salmon (Salmo salar): effects of replacement of dietary marine fish oil with vegetable oils. Aquaculture
234:305-314.
Bernhard K. 1990. Synthetic astaxanthin. The route of a carotenoid from research to commercialization. In: Krinski NI,
Mathews-Roth MM, Taylor RF editors. Carotenoids: Chemistry and Biology. New York: Plenum Press. pp. 337-364.
Bjerkeng B, Hatlen B, Wathne E. 1999. Deposition of astaxanthin in fillets of Atlantic salmon (Salmo salar) fed diets
with herring, capelin, sandeel, or Peruvian high PUFA oils. Aquaculture 180:307-319.
Burel C, Boujard T, Kaushik SJ, Boeuf G, Van der Geyten S, Mol KA, Kuhn ER, Quinsac A, Krouti M, Ribaillier D. 2000.
Potential of plant-protein sources as fish meal substitutes in diets for turbot (Psetta maxima): growth, nutrient
utilisation and thyroid status. Aquaculture 188:363–382.
Carter CG, Hauler RC. 2000. Fish meal replacement by plant meals in extruded feeds for Atlantic salmon, Salmo salar
L. Aquaculture 185:299–311.
Cooper M, Midling K. 2004. The use of anti-mouse tyrosinase antibody to investigate mechanisms of melanisation in
farmed cod. In: Proceedings of the 34th WEFTA Meeting, Lübeck, Germany, 12-15 Sept 2004. pp: 44.
Cowey CB. 1995. Protein and amino acid requirements: a critique of methods. J Appl Ichthyol 11:199-204.
Dabrowski K, El-Fiky N, Kock G, Frigg M, Wieser W. 1990. Requirement and utilisation of ascorbic acid and ascorbic
sulphate in juvenile rainbow trout. Aquaculture 91:317–337.
Daoulas C, Economou AN, Bantavas I. 1991. Osteological abnormalities in laboratory reared seabass (Dicentrarchus
labrax) fingerlings. Aquaculture 97:169–180.
Doyle RW, Schackel NL, Basiao Z, Uraiwan S, Matricia T, Talbot AJ. 1991. Selective diversification of aquaculture stocks:
a proposal for economically sustainable genetic stock conservation. Can J Fish Aquat Sci 48:148-154.
Fevolden SE, Roed KH, Fjalestad K. 2003. Title: A combined salt and confinement stress enhances mortality in rainbow
trout (Oncorhynchus mykiss) selected for high stress responsiveness. Aquaculture 216:67-76.
Foran JA, Hites RA, Carpenter DO, Hamilton MC, Mathews-Amos A, Schwager SJ. 2004. A survey of metals in tissues of
farmed Atlantic and wild Pacific salmon. Environ Toxicol Chem 23:2108-2110.
Francis G, Makkar HPS, Becker K. 2001. Antinutritional factors present in plant-derived alternate fish feed ingredients
and their effects in fish. Aquaculture 199:197–227.
Fraser MR, Anderson TA, de Nys R. 2004. Ontogenic development of the spine and spinal deformities in larval barramundi
(Lates calcarifer) culture. Aquaculture 242:697-711.
Friars GW, Bailey JK, O’Flynn FM. 1995. Applications of selection for multiple traits in cage-reared Atlantic salmon
Gildberg A. 2004. Digestive enzyme activities in starved pre-slaughter farmed and wild-captured, Atlantic cod (Gadus
morhua) . Aquaculture 238:343-353.
Gomes EF, Rema P, Gouveia, A, Teles AO. 1995. Replacement of fish meal by plant-proteins in diets for rainbow trout
(Oncorhynchus mykiss)—effect of the quality of the fish-meal based control diets on digestibility and nutrient
balance. Water Sci Technol 31:205–211.
Henderson RJ. 1996. Fatty acid metabolism in freshwater fish with particular reference to polyunsaturated fatty acids.
Archives of Animal Nutrition 49:5-22.
Herbinger CM, O’Reilly PT, Doyle RW, Wright JM, O’Flynn F. 1999. Early growth performance of Atlantic salmon full-sib
families reared in single family tanks versus in mixed family tanks. Aquaculture 173:105-116.
binnenwerk.indd 548 21-08-2006 12:36:22

Hites RA, Foran JA, Carpenter DO, Hamilton MC, Knuth BA, Schwager SJ. 2004. Global assessment of organic contaminants
in farmed salmon. Science 303:226-229.
Hoelzel AR. 1992. Molecular Genetic Analysis of Populations. A Practical Approach. Cambridge, UK: IRL Press. 315p.
Igarashi T, Aursand M, Sacchi R, Paolillo L, Nonaka M, Wada S. 2002. Determination of docosahexaenoic acid and n-3
fatty acids in refined fish oils by 1H-NMR spectroscopy: IUPAC interlaboratory study. JAOAC Int 85:1341-1354.
Iqbal MJ, Yaegashi S, Ahsan R, Lightfoot DA, Banz WJ. 2002. Differentially abundant mRNAs in rat liver in response to
diets containing soy protein isolate. Physiol. Genomics 11:219-226.
Jacobs MN, Covaci A, Schepens P. 2002. Investigation of selected persistent organic pollutants in farmed Atlantic
salmon (Salmo salar), salmon aquaculture feed, and fish oil components of the feed. Environ Sci Technol 36:2797-
2805.
Jobling M. 1988. A review of the physiological and nutritional energetics of cod, Gadus morhua L., with particular
reference to growth under farmed conditions. Aquaculture 70:1-19.
Jobling M. 1994. Fish Bioenergetics. London: Chapman Hall. 328 p.
Jobling M. 2004. Are modifications in tissue fatty acid profiles following a change in diet the result of dilution? Test
of a simple dilution model. Aquaculture 232:551-562.
Jobling M, Larsen AV, Andreassen B, Sigholt T, Olsen RL. 2002. Influence of a dietary shift on temporal changes in fat
deposition and fatty acid composition of Atlantic salmon post-smolt during the early phase of seawater rearing.
Aquaculture Res 33:875-889.
Joensen H, Grahl-Nielsen O. 2000. Discrimination of Sebastes viviparus, S. marinus, and S. mentella from Faroe Islands
by chemometry of the fatty acid profile in heart and gill tissues and in the skull oil. Comp Biochem Physiol
126B:69-79.
Joensen H, Grahl-Nielsen O. 2001. The redfish species Sebastes viviparus, Sebastes marinus and Sebastes mentella have
different composition of their tissue fatty acids. Comp Biochem Physiol 129B:73-85.
Joensen H, Grahl-Nielsen O. 2004. Stock structure of Sebastes mentella in the North Atlantic revealed by chemometry
of the fatty acid profile in heart tissue. ICES J Mar Sci 61:113-126.
Kaushik, SJ, Cravedi JP, Lalles JP, Sumpter J, Fauconneau B, Laroche M. 1995. Partial or total replacement of fish meal
by soybean protein on growth, protein utilization, potential estrogenic or antigenic effects, cholesterolemia and
flesh quality in rainbow trout, Oncorhynchus mykiss. Aquaculture 133:257-274.
Krogdahl A, Lea TB, Olli JL. 1994. Soybean proteinase inhibitors affect intestinal trypsin activities and amino acid
digestibilities in rainbow trout (Oncorhynchus mykiss). Comp Biochem Physiol 107A:215–219.
Lie Ø, Lied E, Lambertsen G. 1986. Liver retention of fat and of fatty acids in cod (Gadus morhua) fed different oils.
Aquaculture 59:187-196.
Lura H, Sægrov H. 1991. A method of separating offspring from farmed and wild Atlantic salmon (Salmo salar) based
on different ratios of optical isomers of astaxanthine. Can J Fish Aquat Sci 48:429-433.
Luten J, Kole A, Schelvis R, Veldman M, Heide M, Carlehög M, Akse L. 2002. Evaluation of wild cod versus wild caught,
farmed raised cod from Norway by Dutch consumers. Økonomisk Fiskeriforskning 12:44-60.
McCoy, M. 1999 Astaxanthin market a hard one to crack. Chem Eng News 77:15-17.
Madeniian CP, O’Connor DV, Stewart DJ, Miller MA, Masnado RG. 2002. Field estimate of net trophic transfer efficiency
of PCBs to Lake Michigan chinook salmon from their prey. Environ Sci Technol 36:5029-5033.
Mambrini M, Kaushik SJ. 1995. Indispensable amino acid requirements of fish: correspondence between quantitative
data and amino acid profiles of tissue proteins. J Appl Ichthyol-Z Angew Ichthyol 11:240-247.
Manchenko GP. 1994. Handbook of Detection of Enzymes on Electrophoretic Gels. Boca Raton, Fl: CRC Press LLC. 341 p.
Martin SAM, Vilhelmsson O, Medale F, Watt P, Kaushik S, Houlihan DF. 2003. Proteomic sensitivity to dietary manipulations
in rainbow trout. Biochim Biophys Acta 651:17-29.
Martinez I, Christiansen JS. 1994. Myofibrillar proteins in developing white muscle of the Arctic charr, Salvelinus alpinus
(L.). Comp Biochem Physiol 107B:11-20.
Martinez I, Ofstad R, Olsen RL. 1990a. Myosin isoforms in red and white muscles of some marine teleost fishes. J Muscle
Res Cell Motility 11:489-495.
binnenwerk.indd 549 21-08-2006 12:36:22

Iciar Martinez
Martinez I, Ofstad, R, Olsen RL. 1990b. Electrophoretic study of myosin isoforms in white muscles of some teleost
fishes. Comp Biochem Physiol 96B:221-227.
Martinez I, Ofstad R, Olsen RL. 1990c. Intraspecific myosin light chain polymorphism in the white muscle of herring
(Clupea harengus harengus, L.). FEBS Lett 265:23-26.
Martinez I, Christiansen JS, Ofstad R, Olsen RL. 1991. Comparison of myosin isoenzymes present in skeletal and cardiac
muscles of the Arctic charr Salvelinus alpinus (L.). Sequential expression of different myosin heavy chains during
development of the fast white skeletal muscle. Eur J Biochem 195:743-753.
Medina I, Aubourg SP, Martin RP. 1997. Species differentiation by multivariate analysis of phospholipids from canned
Atlantic tuna. J Agric Food Chem 45:2495-2499.
Mooney BD, Nichols PD, Elliot NG. 2002. Seafood, the Good Food II. Hobart, Tasmania, Australia: CSIRO Marine Research
and FRDC. 126 pp.
Morais S, Bell JG, Robertson DA, Roy WJ, Morris PC. 2001. Protein/lipid ratios in extruded diets for Atlantic cod (Gadus
morhua L.): effects on growth, feed utilisation, muscle composition and liver histology. Aquaculture 203:101-119.
Moyano FJ, Cardenete G, de la Higuera M. 1991. Nutritive and metabolic utilization of proteins with high glutamic-acid
content by the rainbow-trout (Oncorhynchus mykiss). Comp Biochem Physiol 100A:759-762.
Naylor RL, Goldburg RJ, Primavera JH, Kautsky N, Beveridge MCM, Clay J, Folke C, Lubchenco J, Mooney H, Troell M.
2000. Effect of aquaculture on world fish supplies. Nature 405:1017-1024.
Nichols PD, Mooney BD, Elliot NG. 2002. Nutritional Value of Australian Seafood II. Hobart, Tasmania, Australia: CSIRO
Marine Research and FRDC. 198 pp. ISBN 1 876 996 07 2.
Prache S, Cornu A, Berdague JL, Priolo A. 2005. Traceability of animal feeding diet in the meat and milk of small
ruminants. Small Rum Res 59:157-168.
Robin JH, Regost C, Arzel J, Kaushik SJ. 2003. Fatty acid profile of fish following a change in dietary fatty acid source:
model of fatty acid composition with a dilution hypothesis. Aquaculture 225:283-293.
Sacchi R, Medina I, Aubourg SP, Addeo F, Paolillo L. 1993. Proton nuclear magnetic resonance rapid and structure-
specific determination of n-3 polyunsaturated fatty acids in fish lipids. JAOCS 70:225-228.
Dos Santos J, Burkow IC, Jobling M. 1993. Patterns of growth and lipid deposition in cod (Gadus morhua L.) fed natural
prey and fish-based feeds. Aquaculture 110:173-189.
Sargent JR, Tacon AGJ. 1999. Development of farmed fish: a nutritionally necessary alternative to meat. Proc Nutr Soc
58:377–383.
Sægrov H, Hindar K. Kålås S, Lura H. 1997. Escaped farmed Atlantic salmon replace the original salmon stock in the
River Vosso, Western Norway. ICES J Mar Sci 54:1166-1172
Tocher DR, Bell JG, Dick JR, Henderson RJ, McGhee F, Michell D, Morris PC. 2000. Polyunsaturated fatty acid metabolism
in Atlantic salmon (Salmo salar) undergoing parr-smolt transformation and the effects of dietary linseed and
rapeseed oils. Fish Physiol Biochem 23:59-73.
Toften H, Jobling M. 1996. Development of spinal deformities in Atlantic salmon and Arctic charr fed diets supplemented
with oxytetracycline. J Fish Biol 49:668-677.
Torrissen OJ, Christiansen R. 1995 Requirements for carotenoids in fish diets. J Appl Ichthyol 11:225-230.
Vielma J, Makinen T, Ekhol P, Koskela J. 2000. Influence of dietary soy and phytase levels on performance and body
composition of large rainbow trout (Oncorhynchus mykiss) and algal availability of phosphorus load. Aquaculture
183:349-362.
Vielma J, Ruohonen K, Peisker M. 2002. Dephytinization of two soy proteins increases phosphorus protein utilization
by rainbow trout, Oncorhynchus mykiss. Aquaculture 204:145-156.
Villareal BW, Rosenblum PM, Fries LT. 1994. Fatty-acid profiles in red drum muscle. Comparison between wild and
cultured fish. Trans Am Fish Soc 123:194-203.
Watabe S, Hirayama Y, Nakaya M, Kakinuma M, Kikuchi K, Guo XF, Kanoh S, Chaen S, Ooi T. 1997. Carp expresses fast
skeletal myosin isoforms with altered motor functions and structural stabilities to compensate for changes in
environmental temperature. J Therm Biol 22:375-390.
Watanabe T. 1982. Lipid nutrition in fish. Comp Biochem Physiol 73B:3-15.
Wilson RP, Cowey CB. 1985. Amino acid composition of whole-body tissue of rainbow trout and Atlantic salmon.
Aquaculture 48:373-376.
binnenwerk.indd 550 21-08-2006 12:36:22

Detection of noroviruses: Comparison of viral extraction methods in
bivalve molluscs

Ghent University, Faculty of Bioscience Engineering, Department of Food Safety and Food
Quality, Laboratory of Food Microbiology and Food Preservation, Coupure links 653, 9000 Ghent,
Belgium
Abstract
Noroviruses (NV) are proven to be associated with outbreaks due to the consumption of bivalve
molluscs. In this study, two categories of viral extraction methods were tested: (1) methods
which directly isolated the total amount of RNA from shellfish and (2) methods firstly isolating
the virus particles from the shellfish matrix and then extracting the viral RNA. A semi-nested
RT-PCR (Booster) was used to detect NV for all tested extraction methods. The main goal was
to search a promising methodology to screen bivalve shellfish samples for the presence of
NV. The viral extraction method using TRIzol®Reagent was the most rapid, user-friendly, and
consequently most promising manner to extract NV from shellfish.
Keywords: noroviruses, Norwalk-like viruses, bivalve molluscs, shellfish, RT-PCR
Introduction
Noroviruses (NV), previously called Norwalk-like viruses or small round structured viruses (SRSV)
are a leading cause of non bacterial foodborne gastroenteritis (De Wit and others 2001; Rockx
and others 2002; Van Duynhoven and others 2005). The majority of human infecting NV strains
are comprised in genogroup I (GI) and II (GII). The symptoms are rather mild (diarrhea,
abdominal cramps, nausea, vomiting) nevertheless the economic losses are not negligible
(Anonymous 2001).
Bivalve shellfish is a naturally contaminated food vehicle provoking transmission of NV.

Most bivalves are “filter–feeders” and are therefore able to bring about the concentration of
pathogens from the growing waters when these are fecal contaminated. Growing and harvesting
waters near the shore or in estuaries can be polluted by the influx of inadequately cleaned
sewage water containing human fecal waste (Lees 2000) or by overboard dumping of fecal
waste (McDonnell and others 1997; Koopmans and others 2002).
Outbreaks related to raw shellfish (Sugieda and others 1996; Prato and others 2004), specifically
to oysters (Kohn and others 1995; LeGuyader and others 1996; McDonnell and others 1997;
Shieh and others 2000; Berg and others 2000) and to cockles (Hutson and others 2004) are
well known.
The inability to grow NV in a cell culture obliges the screening and detection of NV to be
dependent on molecular techniques. The molecular RT-PCR reaction is well documented and
widely applied (Xi and others 1992; Moe and others 1994; Vinje and others 2003, 2004).
binnenwerk.indd 551 21-08-2006 12:36:22

However this technique requires a viral extraction method on beforehand to remove inhibitors
from the food matrix such as shellfish. Moreover, the viral RNA should be concentrated into
a volume, small enough and suitable for RT-PCR. In general, two approaches of extraction
methods from foods are applied: (1) a prior separates the viruses first from the food matrix
and precipitates the virus particles using a PEG (Polyethylene glycol)-based solution with
subsequent isolation of the viral RNA or (2) a second isolates the total amount of RNA directly
from the food matrix. Many variants of the virus elution - concentration (first) approach are
reported (Atmar and others 1995; Jaykus and others 1996; Dix and others 1998; Green and
others 1998; Shieh and others 1999; Kingsley and others 2001). Differences can be observed
in the use of a neutral or alkaline buffer to elute the virus particles, the concentration of the
used PEG to precipitate the virus particles, the number of purification steps, etc. The second
approach was applied by Schwab and others (2000). It is difficult to look for the best resulting
extraction method based on published scientific data because different inoculation numbers,
molecular amplification and detection assays were used.
In this study, several viral extraction methods based on the above described principles
were tested in parallel with artificially contaminated mussels and cockles to evaluate the
manageability and efficiency of every method. This comparison was possible because the same
stool samples were used for inoculation of the shellfish samples and the clarified RNA extracts
were amplified and observed with the same appropriate semi-nested RT-PCR (Booster) using the
broadly functioning JV12Y/JV13I primer pair (i.e. detecting both GI and GII).
Norovirus strains
The virus strains, used for artificial inoculation of the shellfish samples, originated from two
stool samples connected with outbreaks of NV infections. These stool samples were kindly
provided by Dr. Vennema - the National Institute of Public Health and the Environment from The
Netherlands (Bilthoven). One stool sample contained a strain classified in genogroup I (Desert
Shield). Another stool sample contained a strain classified in genogroup II (Lordsdale).
RT-PCR: booster
A semi-nested RT-PCR protocol, named Booster was used for NV amplification and detection of
the clarified RNA resulting from a viral extraction method. The Booster protocol described by
De Medici and others (2004) was used with slight modifications. Briefly, 1x AMV/Tfl reaction
buffer (Promega, Madison, USA), 0.2 mM dNTP mix (Promega), 1 mM MgSO4 (Promega), 0.1 U/µl
AMV reverse transcriptase (Promega), 0.1 U/µl Tfl DNA polymerase (Promega), 0.1 µM of JV13I
and JV12Y and 5 µl of clarified RNA were mixed. The primer pair JV12Y/JV13I detected both
genogroup I and genogroup II NV strains (Vennema and others 2002). The second amplification
used a higher concentration of the same primers namely, 1 µM of JV12Y and JV13I. A fresh
master mix was prepared with the 10-times more concentrated primers, 1x AMV/Tfl reaction
buffer (Promega), 0.2 mM dNTP mix (Promega), 1 mM MgSO4 (Promega), 0.1 U/µl Tfl DNA
polymerase (Promega). Five µl of the reaction products obtained from the first amplification
round was added to the master mix. The PCR product comprised 327 bp.
Shellfish samples
Mussels (Mytilus edulis) and cockles (Cerastoderma edule) were bought from a local supermarket.
Mussels and cockles represented in this study the group of the bivalve shellfish. The virus
binnenwerk.indd 552 21-08-2006 12:36:22

Detection of noroviruses: Comparison of viral extraction methods in bivalve molluscs
extraction methods started from digestive glands dissected from cockles and mussels. One
mussel and cockle sample were inoculated with 50 µl of 100-fold diluted stool sample containing
GI, one mussel and cockle sample were inoculated with 50 µl of 100-fold diluted stool sample
containing GII. This amount of inoculation was estimated to correspond with around 500 to
5000 virus particles. Every extraction method took a not inoculated sample along (negative
control).
Virus extraction methods

An overview of the viral extraction methods is given in Figure 1.
Direct RNA isolation

D1: Inoculated mussels and cockles (2 g) were rocked for 20 min with 2 ml PBS/glycine (50 mM)
pH 9.5 (Acros organics, Geel, Belgium). The addition of 3 ml of TRIzol®Reagent (Invitrogen,
Paisley, Scotland) was followed by centrifugation (Sorvall SS34 rotor, Newtown, USA) for 20
minutes, 8000 x g, at 4 °C. The watery phase was taken off and stored at -20 °C. Hundred µl was
purified by the use of a RNeasy Mini kit (Qiagen, Hilden, Germany) according the instructions
of the manufacturer.
D2: Inoculated cockles and mussels (2 g) were rocked for 10 min with 1 ml of TRIzol®Reagent.
The liquid was kept separately. Another 1 ml TRIzol®Reagent was added to the sample and
rocked again for 10 min. The liquid was brought together with the previously collected liquid.
The watery phase was taken after centrifugation (8,000 x g, 20 min, 4 °C) and stored by
freezing. Hundred µl was purified by the use of a RNeasy Mini kit.
Virus particles elution – concentration

V1: The method used was based on a former published method (Atmar and others 1995).
Inoculated shellfish samples (1.5 g) were homogenized in 10 ml PBS pH 7.4 and 0.2 ml
antifoam B (Sigma Aldrich, St-Louis, USA). Six ml of chloroform–butanol (1:1, vol/vol) (VWR
international, Fontenay sous Bois, France) was added and rocked for 5 min. The samples were
allowed to settle for 15 min. After centrifugation the watery phase was taken and 6.5 ml of
24% PEG 6000 (Fluka Chemie, Buchs, Germany) – 1.2 M NaCl solution was added. These mixtures
were rocked for 1 h at 4 °C. To pellet the virus particles, a centrifugation step was necessary.
Proteinase K with a concentration of 0.2 mg/ml (Merck, Darmstadt, Germany) was added to
lyse the virus particles at 56 °C (30 min). An equal volume phenol-chloroform-isoamylalcohol
25:24:1 (Fluka Chemistry, Buchs, Germany) separated the watery phase containing the viral
RNA. Afterwards precipitation was allowed by the addition of ethanol (Merck, Leuven, Belgium).
Finally, RNA was dissolved in 100 µl RNase, DNase free water.
V2: Inoculated shellfish samples (2 g) were homogenized with 6 ml of 0.05 M glycine – 0.15
M NaCl for 20 min at 4 °C. The homogenate was centrifuged (15 min, 10 000 x g, 4 °C). The
pH of the supernatant was adjusted to 7.2-7.4 before 6% PEG – 0.3 M NaCl was added to
rock overnight. To precipitate the virus particles, a centrifugation step of 10,000 x g for 30
min at 4 °C was included. The pellet was suspended in PBS pH 7.4. Purification consisted of
chloroform:n-butanol (1:1, vol/vol) with a contact time of 5 min. After centrifuging (15 min,
10,000 x g, 4 °C), the watery phase was separated.
V2 short: Hundred µl of the watery phase was taken and purified with a RNeasy Mini kit.
binnenwerk.indd 553 21-08-2006 12:36:23

Virusparticle elution - concentration Direct viral RNA extraction
binnenwerk.indd 554
554
V1 V2 V3 D1 D2
Elution PBS pH 7.4 + antifoam B 0.05 M glycine – 0.15 M NaCl pH 9 PBS/glycine pH 9.5
Purification Chloroform – butanol (1:1)
Precipitation of virus 24% PEG 6000 – 1.2 M NaCl Adjusting pH: 7.2-7.4 + 6% PEG 6000 -1.2 M NaCl
particles
1h rocking at 4°C Over night rocking at 4°C
Purification Chloroform – butanol (1:1)
Virus particles
Second precipitation Adjusting pH: 7.2-7.4
of virus particles 2h rocking 12% PEG 6000 -1.2 M NaCl
at 4°C
Extraction of RNA Proteinase K Proteinase K/CTAB TRIzol TRIzol
Purification of RNA Phenol – chlroroform – isoamylalcohol (25:24:24) Chloroform

RNA stage
Ethanol precipitation Isopropanol precipitation
100 μl purified
100 μl of Rnase, Dnase Wash with 75% ethanol
free water V2 short
V1 V2 long
100 μl of Rnase, Dnase 100 μl purified
free water
V3 short D1 D2
V3 long
Figure 1. Tested viral extraction methods on shellfish.
21-08-2006 12:36:23
V2 long: The rest of the watery phase was subjected to a second precipitation step, by adding
12% PEG – 0.3 M NaCl and 2 h of rocking. The precipitated virus particles were lysed by the
use of the enzyme proteinase K (0.2 mg/ml) with an incubation time of 30 min at 37 °C.
Ten % CTAB (Sigma Aldrich, St-Louis, USA) and 1.2 M NaCl were added, resulting in a final
concentration of 1.25% CTAB and 0.45 M NaCl in the sample. After the incubation time of
30 min at 56 °C, phenol-chloroform-isoamylalcohol 25:24:1 was added. The RNA containing
watery phase was removed after centrifugation (Biofuge pico centrifuge, Heraeus instruments,
Osterode, Germany) (10,000 x g, 20 min, room temperature). Ethanol was used to precipitate
the RNA and an extra washing step with acetone (Merck, Leuven, Belgium) was included. Finally
the RNA was dissolved in 100 µl RNase, DNase free water.
V3: Inoculated shellfish samples (2 g) were homogenized with 6 ml of 0.05 M glycine – 0.15 M
NaCl for 20 min at 4 °C. The homogenate was centrifuged (15 min, 10,000 x g, 4 °C). The pH of
the supernatant was adjusted to 7.2-7.4 and 6% PEG – 0.3 M NaCl was added to rock overnight.
To precipitate the virus particles, a centrifugation step of 10,000 x g for 30 min at 4 °C was
included. Two ml of TRIzol®Reagent was added to the pellet and left shaking for 20 min.
V3 short: Hundred µl was taken and purified with a RNeasy Mini kit.
V3 long: The rest of the TRIzol®Reagent solution was purified. Therefore, chloroform was utilized
and mixed with the sample for 15 s. After centrifugation, the RNA from the watery phase was
precipitated with propan-2-ol (Sigma Diagnostics, St-Louis, USA) applying a contact time of 10
min. After centrifugation (Biofuge pico centrifuge, Heraeus instruments, Osterode, Germany)
(10 min, 10,000 x g, room temperature), the pellet was washed with 75% ethanol and followed
by an additional washing step with acetone. The dried pellets were dissolved in 100 µl RNase,
DNase free water.
NV are characterized by genetic heterogeneity and could be divided in at least 5 genetic

groups (Hutson and others 2004). The majority of human infecting strains are comprised in
GI and GII. In this study, the primer couple JV12Y/JV13I was used in the semi-nested RT-PCR
protocol (Booster) because of the broadly active function, enabling the detection of both GI
and GII noroviruses. These are optimized primers of the formerly JV12/JV13 primers (Vennema
and others 2002). The sensitivity of the Booster protocol, using JV12Y/JV13I, was tested with
a tenfold serial dilution of two stool samples containing a GI and a GII strain respectively. A
105 diluted stool sample contaminated with GI and a 106 diluted stool sample contaminated
with GII were detected with this semi-nested RT-PCR reaction (Baert and others, submitted for
publication). In a highly contaminated stool sample from an infected person with NV, around
106-107 virus particles per milliliter can be present (Anonymous 2001). It is estimated that
the highest detectable dilution with RT-PCR (i.e. 105 for GI/ 106 for GII) corresponds with circa
one copy of one NV RNA genome. From the observed sensitivity could be carefully stated that
the Booster protocol would be able to detect approximately 10 to 100 virus particles which is
the human infectious dose. Consequently, this semi-nested RT-PCR system would be useful to
screen shellfish samples. Therefore, the Booster assay was used to compare the viral extraction
methods.
binnenwerk.indd 555 21-08-2006 12:36:23

D1 and D2 methods enabled the detection of both GI and GII (Table 1). Figure 2a illustrates
the amplification products of the D1 method. D2 gave similar amplification products. No non-
specific products were co-amplified illustrating that these methods efficiently isolated the viral
RNA from the shellfish matrix and removed possible PCR inhibitors. D1 and D2 did not include
a precipitation step with PEG (Polyethylene glycol) like the “V- methods” and consequently
reduced the time needed to finish the extraction method. D2 used only TRIzol®Reagent.
TRIzol®Reagent is known to be a mono-phasic solution of phenol and guanidine isothiocyanate
suitable to isolate the total amount of RNA from cells and tissues. D1 eluted first the virus
particles from the shellfish followed by the TRIzol®Reagent extraction of the virus containing
elution buffer.
V1 was not able to detect uniformly GI and GII in mussel and cockle samples and needed almost
one working day to accomplish the method (Table 1). This method was derived from a widely
used extraction method, originally developed to detect Hepatitis A virus (Atmar and others
1995). Successfully NV extractions with this method were reported (LeGuyader and others
1996, 2003). However in this study not the exact reported assay was applied (Cat-Floc T was
not used) and could be a possible reason for not achieving as good results as it was reported
earlier. Additionally, shellfish samples inoculated with NV after the 24% PEG 6000 – 1.2 M NaCl
precipitation step and just before the proteinase K treatment were tested (results not shown).
Detection occurred in all of those samples. Therefore the virus loss could be situated at the level
of the virus particles elution – concentration step. It is possible that the elution buffer (PBS pH
7.4) did not capture all the present virus particles. A more widely applied elution buffer, instead
of PBS (pH 7.4), is an alkaline buffer (Lewis and others 1988; Lees and others 1995; Häfliger
and others 1997; Pina and others 1998). Other explanations could be that the percentage of PEG
is not optimal to precipitate the present virus particles or Cat-Floc T was essential in this buffer
conditions or the precipitated virus particles were not completely dissolved. It is reported that
precipitated virus particles are not easily dissolved (Dix and Jaykus 1998). In the V1 method, in
contrary to other reported viral extraction methods, the pH was not adjusted to neutral before
PEG was added. The pH was adjusted in methods V2 and V3.
Table 1. Comparison of viral extraction methods on shellfish.
Method Mussel Cockle Time estimation of the sample

preparation protocol
GI GII GI GII
D1 + + + + 2 hours
D2 + + + + 2 hours
V1 - - + + 7 hours
V2 long + + + - ON + 7 h
V2 short + + - - ON + 2 h
V3 long + + + + ON + 3 h
V3 short + + + + ON + 2 h
+: detection of amplification product with the Booster protocol

-: no detection of amplification product with the Booster protocol
ON: overnight incubation
binnenwerk.indd 556 21-08-2006 12:36:23

a. b. c. d.
1 2 3 4 5 6 7 1 2 3 1 2 3 1 2 3 4 5 6 7
Figure 2. Viral extraction methods on shellfish: D1 and V3.

The Booster amplification products were visualized on a 2% agarose gel. The size comprised 327 bp and is
indicated with an arrow.
Figure 2a: D1 method: (1) 50 bp DNA step ladder, (2) Mussel blank, (3) Mussel GI, (4) Mussel GII, (5)
Cockle blank, (6) Cockle GI, (7) Cockle GII. Figure 2b: V3 long method: (1) Cockle blank, (2) Cockle GI,
(3) Cockle GII. Figure 2c: V3 long method: (1) Mussel GI, (2) Mussel blank, (3) Mussel GII. Figure 2d: V3
short method: (1) 50 bp DNA step ladder, (2) Mussel blank, (3) Mussel GI, (4) Mussel GII, (5) Cockle GII,
(6) Cockle blank, (7) Cockle GI
V2 eluted virus particles with an alkaline buffer (0.05 M glycine-0.15 M NaCl pH 9) in contrary
to V1. Subsequently, a precipitation step using 6% PEG – 1.2M was included to concentrate the
virus particles. It was earlier reported that the elution buffer 0.05 M glycine-0.15 M NaCl pH 9
recovered enteric viruses successfully (Traore and others 1998). This glycine–saline buffer was
also applied by others (Kingsley and Richards 2001; Mullendore and others 2001). The influence
of the PEG concentration towards the recovery of virus particles was reported by Zhou and
others (1996) and stated that that 6-8% was optimal. A 6% PEG was applied for the methods
V2 and V3. After the PEG precipitation step the viruses were dissolved in PBS and purified with
chloroform-butanol instead of the former use of Freon because of the negative effect on the
environment (Atmar and others 1995). From this step, two alternatives (V2 long and V2 short)
were tested in order to isolate the viral RNA (Figure 1).
The alternative V2 short method purified 100 µl of the watery phase created by chloroform-
butanol with a RNeasy Mini kit. V2 short extracted not systematically viruses from the shellfish
samples (Table 1). A lot of non-specific products were produced. These observations could be
due to remaining inhibitory residues of the shellfish hampering the RT-PCR detection assay.
V2 long included an additional 12% PEG precipitation to concentrate the virus particles more.
It was reported that a secondary PEG precipitation needed a higher concentration of PEG and
optimization studies showed that 12% gave good results (Leggitt and others 2000). After lysis
of the virus particles with proteinase K, CTAB was supplemented in order to imply an extra
purification (Xi and others 1992). However, detection could not be observed for all tested
shellfish samples and the method was labor intensive (Table 1).
binnenwerk.indd 557 21-08-2006 12:36:24

V3 method used the same elution and concentration stages as V2. However, after the precipitation
step, the virus particles were dissolved in TRIzol®Reagent to isolate the RNA. Two alternative
methods (V3 long and V3 short) were tested.
V3 short included RNA purification of 100µl with a RNeasy Mini kit. V3 long included additional
purification steps of the samples with chloroform. The RNA was afterwards concentrated to
100 µl. Detection was possible in both alternatives (Table 1) preferring V3 long because no
non-specific products could be detected anymore (Figure 2b, 2c, 2d) and the time required for
extraction was comparable. The V3 long method was similar with an extraction method reported
by Schwab (Schwab and others 2000) for the application on delicatessen foods.
Conclusions
Several viral extraction methods on mussels and cockles were tested in this experimental setup.
On the one hand methods which directly extracted the RNA from shellfish and on the other
hand methods which first eluted the virus particles from the shellfish and then isolated the
viral RNA. In order to compare the results of every extraction method, a semi-nested RT-PCR
reaction, named Booster, was used. This broadly active Booster assay would probably be able to
detect the presence of 10 to 100 NV particles and was therefore appropriate to screen shellfish
for NV contamination. However in this study, high amounts of virus particles (103-104) were
inoculated on the shellfish samples to compare the aspects of easy handling, the influence of
elution buffers or PEG concentrations of several extraction methods.
The direct extraction method using TRIzol®Reagent enabled the detection of both GI and
GII in shellfish (mussels and cockles) and gave satisfactory results considering the criteria of
manageability and the time needed for the extraction method. This method did not include a
lot of manipulations which reduced the loss of virus particles during handling steps. Further
experiments on shellfish inoculated with lower amounts of NV should be done to test the
sensitivity of this promising extraction method. This is necessary in the view of food safety
towards NV because of the low infectious dose (10 to 100 virus particles).
References
Anonymous. 2001. Norwalk-like viruses. Public health consequences and outbreak management. Centers for Disease
Control and Prevention. MWR 50(9):1-18.
Atmar RL, Neill FH, Romalde JL, LeGuyader F, Woodley CM, Metcalf TG, Estes MK. 1995. Detection of Norwalk Virus and
Hepatitis-A Virus in shellfish tissues with the PCR. Appl Environm Microbiol 61(8):3014-3018.
Berg DE, Kohn MA, Farley TA, Mcfarland LM. 2000. Multi-state outbreaks of acute gastroenteritis traced to fecal-
contaminated oysters harvested in Louisiana. J Infect Dis 181:S381-386.
De Medici D, Croci L, Suffredini E, Toti L. 2004. Reverse transcription-booster PCR for detection of noroviruses in
shellfish. Appl Environ Microbiol 70(10):6329-6332.
De Wit MAS, Koopmans MPG, Kortbeek LM, Wannet WJB, Vinje J, Van Leusden F, Bartelds AIM, Van Duynhoven YTHF.
2001. Sensor, a population-based cohort study on gastroenteritis in the Netherlands: Incidence and etiology. Am
J Epidemiol 154(7):666-674.
Dix AB, Jaykus LA. 1998. Virion concentration method for the detection of human enteric viruses in extracts of hard-
shelled clams. J Food Prot 61(4):458-465.
Green J, Henshilwood K, Gallimore CI, Brown DWG, Lees DN. 1998. A nested reverse transcriptase PCR assay for
detection of small round-structured viruses in environmentally contaminated molluscan shellfish. Appl Environ
Microbiol 64(3):858-863.
binnenwerk.indd 558 21-08-2006 12:36:24

Häfliger D, Gilgen M, Lüthy J, Hübner P. 1997. Seminested RT-PCR systems for small round structured viruses and
detection of enteric viruses in seafood. Int J Food Microbiol 37(1):27-36.
Hutson AM, Atmar RL, Estes MK. 2004. Norovirus disease: changing epidemiology and host susceptibility factors. Trends
Microbiol 12(6):279-287.
Jaykus LA, DeLeon R, Sobsey MD. 1996. A virion concentration method for detection of human enteric viruses in oysters
by PCR and oligoprobe hybridization. Appl Environ Microbiol 62(6):2074-2080.
Kingsley DH, Richards GP. 2001. Rapid and efficient extraction method for reverse transcription-PCR detection of
hepatitis A and Norwalk-like viruses in shellfish. Appl Environ Microbiol 67(9):4152-4157.
Kohn MA, Farley TA, Ando T, Curtis M, Wilson SA, Jin Q, Monroe SS, Baron RC, Mcfarland LM, Glass RI. 1995. An outbreak
of Norwalk virus gastroenteritis associated with eating raw oysters - implications for maintaining safe oyster beds.
J Am Med Assoc 273(6):466-471.
Koopmans M, Von Bonsdorff CH, Vinje J, De Medici D, Monroe S. 2002. Foodborne viruses. FEMS Microbiol Rev 26(2):187-
205.
Lees D. 2000. Viruses and bivalve shellfish. Int J Food Microbiol 59(1-2):81-116.
Lees DN, Henshilwood K, Green J, Gallimore CI, Brown DWG. 1995. Detection of Small Round Structured Viruses in
shellfish by Reverse Transcription PCR. Appl Environ Microbiol 61(12):4418-4424.
Leggitt PR, Jaykus LA. 2000. Detection methods for human enteric viruses in representative foods. J Food Prot
63(12):1738-1744.
Le Guyader F, Neill FH, Estes MK, Monroe SS, Ando T, Atmar RL. 1996. Detection and analysis of a small round-structured
virus strain in oysters implicated in an outbreak of acute gastroenteritis. Appl Environ Microbiol 62(10):4268-
4272.
Le Guyader FS, Neill FH, Dubois E, Bon F, Loisy F, Kohli E, Pommepuy M, Atmar RL. 2003. A semiquantitative approach
to estimate Norwalk-like virus contamination of oysters implicated in an outbreak. Int J Food Microbiol 87(1-
2):107-112.
Lewis GD, Molloy SL, Greening GE, Dawson J. 1988. Influence of environmental factors on virus detection by RT-PCR
and cell culture. J Appl Microbiol 88(4):633-640.
McDonnell S, Kirkland KB, Hlady WG, Aristeguieta C, Hopkins RS, Monroe SS, Glass RI. 1997. Failure of cooking to
prevent shellfish-associated viral gastroenteritis. Arch Int Med 157(1):111-116.
Moe CL, Gentsch TJ, Ando T, Grohmann G, Monroe SS, Jiang X, Wang J, Estes MK, Seto Y, Humphrey C, Stine S, Glass
RI. 1994. Application of PCR to detect Norwalk virus in fecal specimens from outbreaks of gastroenteritis. J Clin
Microbiol 32(3):642-648.
Mullendore JL, Sobsey MD, Shieh YSC. 2001. Improved method for the recovery of hepatitis A virus from oysters. J
Virol Methods 94(1-2):25-35.
Pina S, Puig M, Lucena F, Jofre J, Girones R. 1998. Viral pollution in the environment and in shellfish: Human adenovirus
detection by PCR as an index of human viruses. Appl Environ Microbiol 64(9):3376-3382.
Prato R, Lopalco PL, Chironna M, Barbuti G, Germinario C, Quarto M. 2004. Norovirus gastroenteritis general outbreak
associated with raw shellfish consumption in South Italy. BMC Infect Dis 4(37):1-6.
Rockx B, de Wit M, Vennema H, Vinje J, De Bruin E, Van Duynhoven Y, Koopmans M. 2002. Natural history of human
Calicivirus infection: A prospective cohort study. Clin Infect Dis 35(3):246-253.
Schwab KJ, Neill FH, Fankhauser RL, Daniels NA, Monroe SS, Bergmire-Sweat DA, Estes MK, Atmar RL. 2000. Development
of methods to detect “Norwalk-like viruses” (NLVs) and Hepatitis A virus in delicatessen foods: Application to a
food-borne NLV outbreak. Appl Environ Microbiol 66(1):213-218.
Shieh YSC, Calci KR, Baric RS. 1999. A method to detect low levels of enteric viruses in contaminated oysters. Appl
Environ Microbiol 65(11):4709-4714.
Shieh YSC, Monroe SS, Fankhauser RL, Langlois GW, Burkhardt W, Baric RS. 2000. Detection of Norwalk-like virus in
shellfish implicated in illness. J Infect Dis 181:S360-S366.
Sugieda M, Nakajima K, Nakajima S. 1996. Outbreaks of Norwalk-like virus-associated gastroenteritis traced to shellfish:
Coexistence of two genotypes in one specimen. Epidemiol Infect 116(3):339-346.
binnenwerk.indd 559 21-08-2006 12:36:24

Traore O, Arnal C, Mignotte B, Maul A, Laveran H, Billaudel S, Schwartzbrod L. 1998. Reverse transcriptase PCR detection
of astrovirus, hepatitis A virus, and poliovirus in experimentally contaminated mussels: Comparison of several
extraction and concentration methods. Appl Environ Microbiol 64(8):3118-3122.
Van Duynhoven YTHP, De Jager CM, Kortbeek LM, Vennema H, Koopmans MPG, van Leusden F, Van der Poel WHM, Van
den Broek MJM. 2005. A one-year intensified study of outbreaks of gastroenteritis in The Netherlands. Epidemiol
Infect 133(1):9-21.
Vennema H, de Bruin E, Koopmans M. 2002. Rational optimization of generic primers used for Norwalk-like virus
detection by reverse transcriptase polymerase chain reaction. J Clin Virol 25(2):233-235.
Vinje J, Vennema H, Maunula L, Von Bonsdorff CH, Hoehne M, Schreier E, Richards A, Green J, Brown D, Beard SS,
Monroe SS, de Bruin E, Svensson L, Koopmans MPG. 2003. International collaborative study to compare reverse
transcriptase PCR assays for detection and genotyping of noroviruses. J Clin Microbiol 41(4):1423-1433.
Vinje J, Hamidjaja RA, Sobsey MD. 2004. Development and application of a capsid VP1 (region D) based reverse
transcription PCR assay for genotyping of genogroup I and II NV. J Virol Methods 116(2):109-117.
Xi J, Wang JX, Graham DY, Estes MK. 1992. Detection of Norwalk virus in stool by Polymerase Chain-Reaction. J Clin
Microbiol 30(10):2529-2534.
binnenwerk.indd 560 21-08-2006 12:36:24

Keyword index
A calcium 164, 166, 167, 461–466, 478–485

Aeromonas 365–368 capelin 45, 48, 49, 59–67, 151–155, 158,
alkali/acid solubilisation 427 275
allergy 309 – oil 67
amino acids 416 captured fish 252
ammonia 441, 449 Carnobacterium 403–409
anchovy 45, 48, 49, 275, 399 – divergens 396, 404, 404–409
angler 46, 48, 49 – piscicola 404
Anisakis 309–314 carp 201, 202, 205
– simplex 312 carpet shell 478, 479, 498, 500
anisidine value 61, 62 catfish 194
Antarctic krill 20 – African catfish 193–200
antimicrobial 387 – Channel catfish 43
antioxidant 17, 21, 23, 63, 72, 87, 87–93, cephalodods 287
95–102, 105–116, 127–134, 419, 420 chitosan 387–390
– activity 420, 421 chloride 113–116, 449, 478–485
– dietary fibre 95 cholesterol 29, 31, 35, 41–55, 478–485
Arctic charr 534 chromium 503–505
argentine 45, 48, 49 citrate 106–116
– greater argentine 439, 454 citric acid 115
ascorbate 87–93, 106–115, 186 clam 284–287
ascorbic acid 106 – European razor clam 498, 500
ash 415, 441, 460, 463–466 – Manila clam 460
ATP 204 – shucked 286, 287
– degradation 173 cliff-fish 119, 120, 123
authenticity 541 Clostridium sporogenes 398, 400
cobalt 461–466
B cockle 284, 478, 480, 498, 500, 552, 553
bacalhau 241 – shucked 286, 287
Bacillus subtilis 398, 400 cocktail 284, 286, 287
belly bursting 275 cod 45, 47, 48, 55, 105–112, 139, 140,
biopreservation 395, 403 149–158, 161–169, 173–183, 185–191,
bivalve molluscs 459, 460, 497, 551 232, 241, 242, 265–272, 280, 284,
breaking force 434 349–357, 359–363, 389, 399, 439–454,
bream 45, 48 478, 480, 503, 504, 542–545
brill 45 – farmed 139, 166–168, 173, 181, 185,
brining 185 189, 190, 542
Brochothrix thermosphacta 398, 400, 520 – liver 128
burbot 45 – liver oil 127, 131–135
by-products 413, 419, 420 – polar cod 45, 440
– salted and dried 241
C – wild 163–168, 185, 542
cadmium 441, 443, 448, 450, 497–500, colour 64, 108–110, 114, 187, 189,
503, 504 201–208, 243, 291–293, 318–339, 353,
binnenwerk.indd 561 21-08-2006 12:36:26

430, 443–453, 469, 471, 474, 475, 510, emulsion 59
511, 525–537 enrichment 193
composition 469 Enterobacteriaceae 520, 521, 523
condition 267 Environmental Scanning Electron Microscopy
– factor 151, 271 (ESEM) 309, 312
conger 45 EPA 71
conjugated diene hydroperoxides 61, 63, Escherichia coli 365–367, 398, 400, 401
96–101 ESEM – See: Environmental Scanning
conjugated triene hydroperoxides 61, 63, Electron Microscopy
64, 96–101 ESR – See: electron spin resonance
consumer 213–215, 220, 229, 238, extracellular matrix (ECM) 162, 168
241–249, 317, 542
– panel 244 F
cooking 459, 477 faecal coliforms 365–367
cook loss 443, 446 farming 193
copper 105–115, 163–167, 441, 443, 449, fat 445, 446, 449, 472
450, 461–466, 478, 480–485 – content 442, 444, 448, 544
crab 50 fatty acids 130, 469, 477, 478, 544, 545
– edible 46, 48 – composition 33, 36, 37, 471, 474
crustaceans 41 – free 61, 73, 414
cusk 45, 47 – profile 60
cuttle fish 54 feed 413
firmness 474
D fish
dab 45, 48 – farmed 219, 222, 225, 252, 541
defect 269, 511, 513 – feed 541
desalting 359 – oil 59–67, 71–85, 541
DHA 71 – oil flavour 65
dietary – products 41, 381
– fats 32 – products composition 477
– fibre 95, 96, 232 – proteins 427
– modulation 193, 194 fishmeal 297
dimethylamine (DMA) 187–190 flakiness 186
discoloration 119, 125 flavour 67, 72, 76, 77, 445, 449, 450, 453,
dogfish 45, 48 474
drip loss 150–153, 158 – fishy 83
– rancid 66, 83
E flounder 45, 47, 48
ECM – See: extracellular matrix fluorescence spectroscopy 119, 121, 125
EDTA 106–116 foaming capacity 429, 433, 436
eel 45, 48, 478, 479, 482 foam stability 429, 434
eelpout 45, 48, 50 focus groups 213, 216
elasticity 435 food
electron spin resonance (ESR) 127, 128, – poisoning 359
131–135 – safety 251, 541
electrophoresis 164, 276, 277 forkbeard 45, 48
emulsifier 59, 72, 73, 77, 78, 81, 84 freshness 289, 291
emulsifying capacity 428, 432, 436 frozen sea products 283
binnenwerk.indd 562 21-08-2006 12:36:26

functional properties 427 J
furrow shell 498, 500 John Dory 45, 48
juiciness 185, 186, 474
G
GADF – See: grape antioxidant dietary fibre K
gaping 151–154, 173, 174, 177, 178, 180, K-factor 152
182, 265, 266, 271, 272 K-value 208, 209
gelatinolytic activity 161, 166, 275–281 knowledge 229–238, 246
glycogen 175
grape antioxidant dietary fibre (GADF) 96, L
98 LAB – See: lactic acid bacteria
grape polyphenols 95 lactate 175
grenadier 45, 47, 48 lactic acid bacteria (LAB) 395, 404
– rough head grenadier 399 Lactobacillus farciminis 398, 400
grooved carpet shell 480, 498, 500 larvae 310
growth lead 441, 443, 448, 450, 497–500, 503,
– inhibition 387 504
– kinetic 359 lemon sole 45, 48
gurnard 45, 48 ling 45, 47, 48, 442, 445, 452, 453
– tub gurnard 289–291, 295 – blue ling 45, 47, 48
lipid 415
H – oxidation 59, 71, 72, 85, 87, 95, 101,
haddock 46–48, 389, 399, 449 105–110, 119, 120, 125, 127, 334
hake 46–48, 284, 285, 287, 310, 311, 478, lipids
482 – structured 29
– Cape hake 427, 428, 430 liposomes 18, 22, 87, 88, 93
halibut – marine 89–91
– Greenland halibut 46–48 liquid loss 142, 143
– white halibut 46, 47, 49 Listeria 356, 387
hardness 434, 435 – innocua 381–385, 389–391, 405
herring 45, 48, 49, 275, 276, 297, 298, – monocytogenes 389, 390, 391, 396,
301, 399 397, 398, 400, 404, 405, 407, 409
horse mackerel 46, 48, 49, 96, 284, 285, liver 36, 37, 175, 413, 414, 545
478, 482 lumpsucker 46
houting 46, 48
hydrolysates 419 M
mackerel 45, 48, 49, 399
I magnesium 164, 166, 167, 461–466, 478,
identification 256, 371 480–485
image analysis 525 manganese 478, 480–485
information 217, 221, 225, 229, 230, 234, marine phospholipids (MPL) 17–24, 87, 88
235 matrix metalloproteinase (MMP) 161–168,
inhibition 395 275, 280
in rigor 149–158 mayonnaise 59–67
intelligent packaging 519 meagre 284, 478, 482
iron 106, 164–167, 461–466, 478, 480– – brown meagre 469, 470, 473
485 – cultured 472, 474, 475
– wild 472, 474, 475
binnenwerk.indd 563 21-08-2006 12:36:27

megrim 45, 48 origin 213, 218, 221, 225
mercury 497–500, 503–505 ox crab 284, 286, 287
Mesophilic aerobia 521 oxidation 22, 89, 516
microbial inactivation 381 – products 63, 74, 75
microbiology 519 – stability 62, 71, 72, 80, 84, 127, 128,
minerals 459, 477 133, 334
MMA – See: monomethylamine oyster 372
MMP – See: matrix metalloproteinase – giant cupped 498, 500
moisture 113, 414, 441–445, 449, 460, – Portuguese oyster 498, 500
472
monitoring 489 P
monkfish 284, 478, 482 PBN 127
monomethylamine (MMA) 187, 188, 190 PCB 489–494, 544
Mora moro 45, 48 PCR – See: polymerase chain reaction
Morganella morganii 403 pepsin 275, 276, 278
MPL – See: marine phospholipids perch 45, 48, 50
muscle softening 161 – ocean perch 45, 48
mussel 50, 365, 366, 368, 372, 399, 498, peroxidation 89
500, 552, 553 peroxide value (PV) 61, 62, 73–75, 78–84,
– blue mussel 45, 48, 50 132, 283–287
– Mediterranean mussel 460 pH 62, 84, 106–108, 112, 116, 140–144,
– shucked 284, 286, 287 151, 152, 173–182, 185, 188, 190, 191,
201, 203, 205, 207, 209, 290, 291, 295,
N 334, 336, 428, 441, 449
nanofiltration 419–421 pharmaceuticals 17
nematodes 271 phosphates 186
new species 469 phospholipids 17, 18, 414, 544
nickel 461–466, 503–505 phosphorus 441, 443, 448, 449, 461–466,
non-destructive 119, 120, 125 478, 480–485
non-thermal process 381 Photobacterium 395
noroviruses 551 – phosphoreum 187, 190, 191, 389, 391,
northern wolffish 440 392, 395, 396, 398, 400, 401, 403
Norwalk-like viruses 551 pikeperch 46, 48, 50
Norway lobster 43, 45, 48, 478–480 plaice 46, 47, 48, 478, 482
nutraceuticals 17 – American plaice 45, 48
nutritional value 477 plasma 36
pollack 46, 47, 48
O pollock 414–417, 442, 445, 452, 453
octopus 45, 48, 50, 284–287, 478, 480 polymerase chain reaction (PCR) 372–376,
odour 64, 67, 76, 77, 291–293, 298, 353, 406, 407, 551, 555
449, 450, 474 polyunsaturated fatty acids (PUFA) 29,
– rancid 77 32, 36, 37, 71–73, 85, 87, 130, 473,
off-flavour 65, 66 479–485
oil stability 129, 131 pope 45, 48, 50
omega-3 fatty acids 17, 19, 59, 60, 71, 95, post mortem 150, 156, 180, 182, 205, 208,
231–233, 486, 545 275, 350
organic acids 387 post rigor 149–158, 173, 174, 177, 178,
organochlor pesticides 489–491 180, 183
binnenwerk.indd 564 21-08-2006 12:36:29

potassium 461–466, 479–485 salmon 45, 53, 54, 140, 194, 231–233,
pout 45, 47, 48 331, 389, 399, 403–409, 413–417, 478–
prawn 50 480, 525, 527, 529, 533, 534, 542–546
– rainbow 43 – Atlantic salmon 48
pre rigor 149–158, 173–183 – Coho salmon 43
processing aids 297 – farmed 546
production areas 497 – liver 414–416
product value 265 – trout 331
protein 415, 420, 428, 441, 443, 445, 449, – wild salmon 546
472, 479 Salmonella 400
– content 442, 444, 471 – enterica 398, 400, 401
– vegetable proteins 139 salt 186
Pseudomonas 395–401, 520, 521, 523 sandeel 45, 49
Psychrobacter 398, 400 sardine 43, 45, 49, 284, 285, 399, 478,
psychrophilic lactic acid bacteria 395 479, 484
PUFA – See: polyunsaturated fatty acids scabbardfish 284–287
pulsed light 381–383 – black scabbardfish 46, 48, 49, 478, 480
PV – See: peroxide value – silver scabbardfish 478, 484
scallop 45, 50
Q scanner 525
QIM – See: Quality Index Method Scanning Electron Microscopy (SEM) 309,
quahaug 312
– ocean quahaug 45, 50 seabass 482
quality 297 – farmed 478
– control 509 seabream 399
Quality Index Method (QIM) 289–295, 350, – black seabream 46
351, 353 – blackspot seabream 478, 480
questionnaire 217, 230 – gilthead seabream 478, 482
seasonal variation 265
R selenium 193–200, 440, 441, 443, 448
rancidity 102 SEM – See: Scanning Electron Microscopy
rapeseed oil 71–79 sensory 304, 511, 522
rapid analysis 349 – analysis 62, 64, 71, 472, 519
rat 29, 32, 35, 37 – assessment 291, 474
rays 49 – data 76
recall 251 – evaluation 59, 74, 75, 289, 290, 511,
redfish 284 521
red mullet 399 – inspection 513
retentions 459 – methods 151
rigor mortis 173, 181, 182, 201–205 – panel 66, 77, 83
risk management 349 – scores 65
roach 45, 49 – tests 444
rockfish 414–417 sequence analysis 371
roe 43, 50, 53, 54, 399 Serratia liquefaciens 396–401
ruffe 45, 48, 50 shark 49, 399, 439
– Atlantic sharpnose shark 49
S – blacktip shark 49
saithe 46, 47, 49, 119, 120, 123, 318 shear force 139–145
binnenwerk.indd 565 21-08-2006 12:36:30

shelf life 289, 293, 295, 349, 519, 520 T
shellfish 551 TAG – See: triacylglycerol
shell liquor 459 taste 248, 339, 447
Shewanella 395 TBARS – See: thiobarbituric acid reactive
– putrefaciens 156, 389, 391, 392, 395, substances
398, 400, 521 texture 140, 151–155, 174, 185, 186,
shrimp 43, 53, 54, 284, 286, 287, 399 291–293, 339, 353, 434, 445, 450, 451,
– brown shrimp 45, 489, 490, 491, 492, 453, 511
494, 495 thick trough shell 498, 500
– cold water shrimp 45, 48 thiobarbituric acid index 96, 98, 101
– Georgia white shrimp 53 thiobarbituric acid reactive substances
– Honduras pink shrimp 53 (TBARS) 107–111, 114, 115, 120, 121,
– Northern shrimp 53 124, 125, 129, 131, 335
– pink shrimp 43 tilapia 42
– red shrimp 43 time temperature indicators (TTI) 349,
– Texas brown shrimp 53 355, 356, 519–521
smelt 46, 49 TL – See: total lipids
– silver smelt 439–441, 444–454 TMA – See: trimethylamine
smooth calista 498, 500 TMA-N 287, 334
smooth venus 460 TMAO – See: trimethylamine oxide
sodium 461–466, 479–485 tocopherol 63, 64, 66, 67, 72, 73, 85,
sole 46, 47, 49 87–93, 127–135
solubility 427 total coliforms 365
Sparus auratus 42 total fat content 441, 443
spatial variation 265 total lipids (TL) 414, 471
specific spoilage organisms 156 total viable count (TVC) 150, 151, 156,
spectrophotometer 525 157, 187–191, 298–304, 450
spin trapping 127 total volatile basic nitrogen (TVB-N)
spoilage 289, 297, 304 283–287, 290, 291, 294, 295, 298, 299,
– bacteria 387 302–305, 333–335, 441, 444, 448–450
sprat 46, 49 total volatile nitrogen (TVN) 120–125
squid 43, 50, 54, 284–287, 478, 480 toxR gene 371
– flying squid 45, 48 traceability 213, 214, 217, 224, 225, 237,
stability 67 251, 252, 257, 258, 260, 541, 547
Staphylococcus trace metal 161, 163, 167, 168
– aureus 359–363, 398, 400, 401 triacylglycerol (TAG) 29–38, 63
– xylosus 396–400 trimethylamine (TMA) 187–191, 283–286,
starvation 139–145 395
– pre-slaughter 140 trimethylamine oxide (TMAO) 187–191,
stockfish 509–516 395, 441, 443, 448, 449
storage 77, 80, 95, 100, 101, 120, 123, tristimulus filter colorimeter 525
129, 141–144, 158, 178, 180, 181, 185, trout 45, 49, 399
187–191, 207, 290, 291, 295, 301–303, – rainbow trout 43, 194, 534
318, 350–352, 355, 404, 405, 440 trust 213–215, 226, 230, 234, 235
– temperature 173 TTI – See: time temperature indicators
sulphide producing bacteria 349–356 tuna 284, 285, 287, 318–320, 323, 331,
sulphur 461–466 332, 336, 340, 399
synergy 87 turbot 46, 49, 519–521
binnenwerk.indd 566 21-08-2006 12:36:31

TVB-N – See: total volatile basic nitrogen
TVC – See: total viable count
TVN – See: total volatile nitrogen
U
ultrafiltration 419–421
V
vendace 43, 50
Vibrio 372, 374
– parahaemolyticus 371–379
virulence factors 371
VIS/NIR spectroscopy 509
vitamins 477
volatiles 78, 81–84
– compounds 80
– oxidation products 71
W
water 415, 445, 446
water holding capacity (WHC) 139–141,
145, 185, 187, 189–191, 442, 446
wedge shell 498, 500
WHC – See: water holding capacity
whelk 46
whitefish 50
whiteness 186, 191, 436, 450
whiting 46, 47, 49, 399
– blue whiting 45, 47, 48
wild fish 219, 222, 225
witch 45, 49
wolffish 45, 49
wreck fish 478, 484
X
X-rays 509
Y
yield 427, 431, 442, 445, 448
yoghurt 71–85
Z
zinc 164, 166, 167, 441, 443, 448, 450,
461–466, 478, 480–485
zymography 164, 275–279
binnenwerk.indd 567 21-08-2006 12:36:31

binnenwerk.indd 568 21-08-2006 12:36:31

Seafood Research From Fish To Dish

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Seafood Research From Fish To Dish

Uploaded by

Copyright:

Available Formats

Seafood

binnenwerk.indd 1 21-08-2006 12:32:50

binnenwerk.indd 3 21-08-2006 12:32:50

The publisher is not responsible for possible

Quality, safety and processing of wild and farmed fish 

binnenwerk.indd 7 21-08-2006 12:32:50

Fiskeriforskning, Tromsø, Norway

binnenwerk.indd 8 21-08-2006 12:32:50

Chapter 1: Nutritional properties and oxidation of marine lipid 15

Marine phospholipids (MPL): Resources, applications and markets 17

Cholesterol content in seafood, data from the last decade: A review 41

Capelin oil for human consumption 59

Oxidative stability of fish oil enriched yoghurts 71

Antioxidant synergy effect between α-tocopherol and ascorbate on the autoxidation

Quality, safety and processing of wild and farmed fish 

binnenwerk.indd 9 21-08-2006 12:32:50

Pre-slaughter starvation of farmed Atlantic cod fed vegetable proteins: Effects on

Brining of farmed cod fillets: Effects on quality aspects 185

Enrichment of functional selenium in farmed African catfish (Clarias goriepinus) by

Comparison of commercial and experimental slaughter of farmed carp (Cyprinus carpio)

Consumer knowledge and interest in information about fish 229

The importance of bacalhau consumption in Portugal and a preliminary product

Traceability: Simulated recall of fish products 251

10  Seafood research from fish to dish

binnenwerk.indd 10 21-08-2006 12:32:50

Characterization of the quality of frozen sea products commercialized in Portugal 283

Development of a Quality Index Method scheme to evaluate freshness of tub gurnard

Chapter 5: Microbial quality of seafood 347

Aeromonas spp. and potential indicators in mussels in Northern Greece 365

Characterisation of Vibrio parahaemolyticus isolated from seafood products 371

Quality, safety and processing of wild and farmed fish 11

binnenwerk.indd 11 21-08-2006 12:32:51

Selection of psychotrophic bacteria active against spoilage and pathogenic micro-

Selection of non-tyramine producing Carnobacterium strains for the biopreservation of

Chapter 6: Full utilisation of the catch 411

Production and characterization of a sockeye salmon (Oncorhynchus nerka) liver meal

Evaluation of antioxidant activities in by-product hydrolysates, fractionation

Chapter 7: Nutrients and contaminants in seafood 457

Composition and nutritional value of fishery products consumed in Portugal 477

12  Seafood research from fish to dish

binnenwerk.indd 12 21-08-2006 12:32:51

Contaminant metals in cod products 503

Chapter 8: Advanced methods for quality determination of seafood 507

Instrumental quality control of stockfish 509

Revision of analytical methodologies to verify the production method of fish 541

Detection of noroviruses: Comparison of viral extraction methods in bivalve molluscs 551

Keyword index 561

Quality, safety and processing of wild and farmed fish 13

binnenwerk.indd 13 21-08-2006 12:32:51

Quality, safety and processing of wild and farmed fish 15

binnenwerk.indd 15 21-08-2006 12:32:51

Keywords: marine phospholipids, omega-3 fatty acids, nutraceuticals, pharmaceuticals,

Quality, safety and processing of wild and farmed fish 17

binnenwerk.indd 17 21-08-2006 12:32:51

Characteristics of marine phospholipids (MPL)

Hydrophobic tail Polar head group

Figure 1. Structure of a phospholipid.

18  Seafood research from fish to dish

binnenwerk.indd 18 21-08-2006 12:32:52

A Fatty acid composition

Quality, safety and processing of wild and farmed fish 19

binnenwerk.indd 19 21-08-2006 12:32:54

Natural resources of MPL

Quality, safety and processing of wild and farmed fish

Quality, safety and processing of wild and farmed fish

10 Seafood research from fish to dish

Quality, safety and processing of wild and farmed fish 11

12 Seafood research from fish to dish

Quality, safety and processing of wild and farmed fish 13

Quality, safety and processing of wild and farmed fish 15

Quality, safety and processing of wild and farmed fish 17

18 Seafood research from fish to dish

Quality, safety and processing of wild and farmed fish 19

20 Seafood research from fish to dish

Quality, safety and processing of wild and farmed fish 21

22 Seafood research from fish to dish

24 Seafood research from fish to dish

Quality, safety and processing of wild and farmed fish 25

26 Seafood research from fish to dish

Quality, safety and processing of wild and farmed fish 27

28 Seafood research from fish to dish

Quality, safety and processing of wild and farmed fish 29

30 Seafood research from fish to dish

Quality, safety and processing of wild and farmed fish 31

32 Seafood research from fish to dish

Quality, safety and processing of wild and farmed fish 33

34 Seafood research from fish to dish

Quality, safety and processing of wild and farmed fish 35

36 Seafood research from fish to dish

Quality, safety and processing of wild and farmed fish 37

38 Seafood research from fish to dish

Quality, safety and processing of wild and farmed fish 39

40 Seafood research from fish to dish