Food Microbiology 27 (2010) 1071e1077

Contents lists available at ScienceDirect

Food Microbiology
journal homepage: www.elsevier.com/locate/fm

Variation of microbial load and visual quality of ready-to-eat salads

by vegetable type, season, processor and retailer
Vittorio Caponigro a, *, Maddalena Ventura b, Ida Chiancone a, Letizia Amato a,
Eugenio Parente c, Filippo Piro a
a
Consiglio per la Ricerca e Sperimentazione in Agricoltura, Centro per l’Orticoltura, Centro di Ricerca per l’Orticoltura, via Cavalleggeri 25, 84018 Pontecagnano, Italy
b
Ortomad, via Lago Carezza 16, 84098 Pontecagnano, Italy
c
Università della Basilicata, Dipartimento di Biologia, Biotecnologie e Difesa Agro forestali. Campus di Macchia Romana, viale dell’Ateneo lucano 10, 85100, Potenza, Italy

a r t i c l e i n f o a b s t r a c t

Article history: Microbial components and visual quality were determined on 1158 consumer units of ready-to-eat salads
Received 21 May 2010 from several processors, two per each of 579 process lots, with residual shelf-life varying around a mode of
Received in revised form five days, collected over 19 months in the years 2006e2008 from retail stores of two Italian cities close to
1 July 2010
a major producing and processing area. The salads were mainly baby leaf of single species (lettuce, arugula,
Accepted 8 July 2010
Available online 16 July 2010
spinach, lamb’s lettuce), with approximately 10% of the lots made up by mixes of 2e4 species. One unit per
lot was analyzed on the day of collection and the other at the consume-by date. No Salmonella or Listeria
monocytogenes was found (detection limit: presence in 25 g). Escherichia coli was detected in 27% of the lots
Keywords:
Ready-to-eat salads
(detection limit: 5 cfu/g), with probability of occurrence and counts highest in Autumn and for lettuce and
E. coli arugula. Average visual quality was higher and other components of the microbial load were lower in
Salmonella Winter and Spring compared to Summer and Autumn (
0.6 log cfu/g of total aerobic counts,
1.3 log cfu/g
Listeria monocytogenes of coliforms,
0.6 log cfu/g of yeasts and moulds). Lactic acid bacteria were detected more frequently in
Microbial components Spring and Summer (up to 50% of the lots). The rate of increase of microbial populations during shelf life
Visual quality was not affected by the level of initial contamination. At the consume-by date total aerobic count exceeded
7.2 log cfu/g for 50% of the lots and 7.7 log cfu/g for 25%. Salads from the biggest processor and retailer
showed slightly higher visual quality scores, lower odds of E. coli occurrence and lower microbial loads.
Visual quality scores showed significant negative relationships with the levels of lactic acid bacteria,
coliforms and total viable counts.
Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction
Ready-to-eat salads (RTES) are low calorie, convenience foods
of good dietary value. RTES technology implies procurement of
raw vegetables from vertically co-ordinated farms, cutting, sorting,
washing, drying, packaging in permeable plastics and retailing
in cold chain regime (2e4  C). High levels of humidity and atmo-
sphere modification induced by permeable plastic packaging and
low storage temperature slow down vegetable leaf metabolism and
help preserve freshness and nutrient value for at least one week of
shelf-life. This micro-environment and an integrated cultivation-
processing-distribution channel allow the marketing of extremely
perishable vegetable organs like baby salads cut in very juvenile
stages (Kader et al., 1989; Rico et al., 2007; Risch, 2009).
* Corresponding author.
E-mail addresses: vittorio.caponigro@entecra.it, vcaponigro@tiscali.it (V. Caponigro).
With a high surface/weight ratio and a relatively high pH, salad
vegetables host a large microbial population, particularly bacteria,
which may contribute to the natural decay of vegetative organs
detached from the plant (Nguyen-The and Carlin, 1994; Ragaert
et al., 2007). Human pathogens like Listeria monocytogenes,
Salmonella and Escherichia coli O157-H7 may contaminate the
product during plant cultivation and processing (Brandl, 2006;
Francis et al., 1999; Franz and van Bruggen, 2008). Contamination
with enteric pathogens may occur through various routes,
including use of organic waste as fertilizer, contamination of irri-
gation waters with fecal material, direct contamination by livestock
and wild animals, hygienic problems in handling and processing
(Heaton and Jones, 2007). Micro-organisms and enteric pathogens
can be protected from sanitizers applied with washing by irregu-
larities of leaf surface, injuries to vegetable tissue, inclusion in
phyllosphere biofilms or internalization within the plant (Gòmez-
Lòpez et al., 2008). The low temperature regime and the modified
atmosphere of RTES are weak hurdles against psychrotrophic

0740-0020/$ e see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.fm.2010.07.011
1072 V. Caponigro et al. / Food Microbiology 27 (2010) 1071e1077

pathogenic micro-organisms like Listeria (Jacxsens et al., 2001;
Junttila et al., 1988; Scifò et al., 2009).
Microbiological surveys of retail RTES have considered the
occurrence of enteric bacterial pathogens, Escherichia coli, coli-
forms, total aerobic and spoilage bacteria, fungi and yeasts (Abadias
et al., 2008; Catellani et al., 2006; de Curtis et al., 2002; Fröder et al.,
2007; Gianfranceschi et al., 2003; Hagenmaier and Baker, 1998;
Loncarevic et al., 2005; Mosupye and von Holy, 1999; Pavan da
Silva et al., 2007; Pianetti et al., 2008; Sagoo et al., 2001; Tournas,
2005; Valentin-Bon et al., 2008; Wojcik-Stopczynska, 2004). Most
reported counts range between 4 and 8 log cfu/g for total aerobic
bacteria and between 0.7 (detection limit) and 6 log cfu/g for
coliforms. E. coli has been often detected at low prevalence and low
counts. Pathogens like Salmonella, L. monocytogenes , E. coli
O157:H7 have seldom been found.
Containment of microbial presence on RTES is important for
preserving produce quality, given that several microbial compo-
nents of the phyllosphere can actively contribute to spoilage of
vegetable tissues and to the growth of a biofilm matrix on the
surfaces (Carmichael et al., 1999). To minimize hazards, EU and
individual EU countries regulations have set microbiological criteria
for RTES on delivery to consumer (CNERNA-CNRS, 1996; DGHM,
2009; EC, 2005). EU criteria include limits for E. coli, as indicator
of contamination by enterobacteria and of the hygienic quality of the
production process.
Apart from the binding EU regulation, no other microbial spec-
ifications or guidelines are actually enforced in Italy and processors
tend to follow the regulations of countries with relevant markets for
Italian produce. The effort to comply with microbiological criteria is
an ongoing process, given access to the expanding industry of new
vegetable producers not yet acquainted with the strict production
rules required for RTES of good microbiological quality.
The objective of this study was to gather further information
on the microbiology of RTES marketed in Italy, sampling lots of
retail products of the most used vegetable species from a number of
processors and retailers in two cities bordering the main growing
and processing area. The sample was spread over 19 months,
to assess seasonal influence on microbial loads and occurrence of
micro-organisms of hygienic relevance. Total aerobic population
and some of its components were determined both at the collection
and at the consume-by date (CBD), to quantify the changes occur-
ring during shelf-life and to verify the impact of microbial abun-
dance on RTES visual quality.
2. Methods
The study was carried out on RTES available in retail stores of
Potenza (PZ) and Salerno (SA), which border the main producing
and processing area of baby fresh-cut salads in Italy. Two consumer
units (125-250 g) per each of 579 process lots were collected
from local retailers at approximately uniform monthly rates for 19
months in SA (from summer 2006 to winter 2008) and for the year
2007 in PZ, totalling 1000 units of 500 lots in SA and 158 units of 79
lots in PZ, where only baby lettuce products were collected, bagged
(56 units) or in trays (23 units). The SA lots included only bagged
products, 445 single species of baby leaf (175 arugula, 143 lettuce,
93 lamb’s lettuce, 34 spinach) and 55 mixes of up to four species,
including also endive, cichory and radicchio (Table 1). Overall, 20
brands from 16 processors and 9 retailers were included. The same
processors and retail chains (supermarkets and minimarkets) were
represented in the two local markets. Because of differences in
continuity of supply among processors, four of them accounted for
80% of the lots, and one for 33% (referred in the following as the
biggest processor). Likewise, one retailer accounted for 42% of the
lots, and will be referred as the biggest retailer. Arugula and lettuce
salads were mainly produced locally and were found in all months,
while spinach and lamb’s lettuce were not regularly available and
mainly or totally (spinach) grown and processed in Northern Italy.
To assess the influence of industry factors on microbiological and
visual quality of RTES with a less sparse sample, processors, brands,
and distributors were grouped in two categories, contrasting
the biggest processor and distributor with the others and RTES
marketed with the processor own brands with those produced by
the same processor for distribution under retailer brands (private
labels). Vegetable mixes were also grouped in one category.
Owing to the modest size of the RTES market it was practically
impossible to collect produce with a residual shelf-life of the same
desired length. The distribution of residual shelf-life in days was as
follows: 7 (2.4%), 6 (21.2%), 5 (34.6%), 4 (30.8%), 3 (9.4%), 0e2 (1.6%)
in SA; 7 (2.5%), 6 (7.6%), 5 (46.8%), 4 (15.2%), 3 (14.7%), 0e2 (10.2%)
in PZ. Therefore, more than 80% of the lots had a residual shelf life
between 4 and 6 days. Collected items were placed in insulated
coolers with ice packs and brought to the laboratory within at most
2 h from collection.
2.1. Microbiological and visual quality analysis
Processor, brand, retailer, origin, packaging type and volume,
weight and CBD were recorded for each lot and one unit per lot
was stored at 4  C to be observed at the CBD. Produce units were
randomly sub-sampled to collect the fraction for microbiological
analyses. The remaining produce was visually evaluated for quality
traits, including freshness, wholeness, colour, odour, crunchiness,
presence of foreign material, browning and rot. The visual evalua-
tion was performed by laboratory personnel on 1:9 point scales and
scores for individual traits were added by produce unit to synthe-
size an index of visual quality which was normalized to a decimal
scale.

Table 1
Breakdown of consumer units of RTES collected in two cities (SA and PZ) by vegetable species and processor, brand and retailer group. PZ units were all baby lettuces
(second row).

Produce Biggest processor Others Totals

Processor’s brand Private labels Processor’s brand Private labels Processor Brands Retailer Total

Biggest Others Biggest Others Biggest Others Biggest Others Biggest Others Processor’s Private Biggest Others
retailer retailer retailer retailer
Lettuce (SA) 68 50 2 8 64 94 118 168 78 208 116 170 286
Lettuce (PZ) 8 18 22 30 6 12 62 48 110 44 114 68 90 158
Arugula (SA) 102 30 2 12 64 140 134 216 114 236 94 256 350
Spinach (SA) 2 44 4 18 2 66 50 18 44 24 68
Lamb’s lettuce (SA) 20 14 4 96 52 34 152 24 162 110 76 186
Mixes (SA) 6 28 20 2 4 18 32 54 56 40 70 46 64 110
Total 14 220 132 24 78 38 254 398 390 768 350 808 478 680 1158
 V. Caponigro et al. / Food Microbiology 27 (2010) 1071e1077
Microbiological analyses included test for Salmonella sp. and L.
monocytogenes and counts of E. coli, total aerobic viable micro-
organisms (TVC), total coliforms, lactic acid bacteria (LAB), moulds
and yeasts. Leaf samples of 25 g were homogenized for 120 s in
a 1:5 dilution of 1% sterile peptone water using a stomacher (Lab
Blender 400, Sewar, UK). Filtered stomacher bags were used to
eliminate soil particles from the homogenates and ten-fold dilu-
tions series as required for plating were made in peptone buffered
water. TVC were determined by spread-plating on Plate Count Agar
(Oxoid) incubated at 30  C for 48 h; coliforms and E. coli on Coli-ID
(Biomérieux) by double layer inclusion and incubation at 37  C for
48 h; LAB on Rogosa Agar (Biomérieux) acidified to pH 5.4 with
glacial acetic acid, by double layer inclusion and incubation at 30  C
for five days; moulds and yeasts on Glucose Yeast Extract Agar with
a 10 mg/ml solution of oxytetracycline-chlorhydrate, by pour
plating and incubation at 25  C for five days. The lower limit of
sensitivity for these microbial components was 5 cfu/g.
Salmonella sp. and L. monocytogenes were determined with
the immuno-enzymatic method mini-VIDAS SLM (Biomérieux),
according to the AFNOR BIO 12/11-04/94 protocol, after enrichment
in 1% sterile peptone water, with presence in 25 g of produce as the
lower limit of sensitivity.
Analyses at the CBD were done only for the SA lots and were
limited to visual quality assessment, TVC, coliforms and E. coli.
Some lots nearing the end of shelf-life at collection were analyzed
the second time within a few days past the CBD.
2.2. Statistical analysis
Microbial counts were analyzed in log scale (log cfu/g), attrib-
uting zero to observations where no colonies were obtained at any
dilution (detection limit of 0.7 log cfu/g).
Distributions of microbial components and visual quality scores
by area, seasonal and industry factors were compared by quantile
summaries of predictions in log scale adjusted at five days of
residual shelf-life by linear regression on days to CBD and all other
factors. Tests of significance for effects of season, processor, retailer,
brand, packaging and vegetable species were similarly adjusted for
shelf-life stage including days to CBD among predictor variables.
A logistic regression was used for assessing occurrence at countable
levels (as presence/absence) and risk ratios of E. coli in relation to
the considered seasonal and industry factors. Confidence intervals
for contrasts were computed following the procedure of Hothorn
et al. (2008). Relationships between responses and the most
6
log cfu/g
4
2 SA
0
6
log cfu/g
4 days to
2 >5
0
E.coli LAB
Fig. 1. Boxplot diagrams of microbial counts on 579 consumer units of ready-to-eat salads collected from retail stores in two city areas: PZ (79 units) in 2007; SA (500 units, over 19
months in 2006e2008). Residual shelf-life ranged between 4 and 6 days for over four fifths of the units and between 0 and 7 days for the whole.
1073
relevant factors were also assessed with canonical variate biplots
(Gower and Hand, 1996).
Average shelf-life trends for TVC, coliforms, E. coli counts and
visual quality scores were estimated by a linear mixed model with
days to CBD as predictor variable, allowing random intercept and
slope effects for produce lots.
The relationship between microbial counts and visual quality
score was summarized by simple regression of visual quality scores
at CBD on log counts at 5 days to CBD, both predicted from respective
observations by multiple regression on all considered factors.
Statistical analyses were made with the R environment
(R Development Core Team, 2009), using the contributed packages
lme4 (Bates and Maechler, 2009), multcomp (Hothorn et al., 2008),
and Zelig (Imai et al., 2009) for model fitting and tests and BiplotGUI
(la Grange et al., 2009) and ggplot2 (Wickham, 2009) for graphics.
3. Results
3.1. Variation of microbial counts
Microbial counts for TVC, coliforms and E. coli showed compa-
rable distributions for the two sampling areas (Fig. 1). The range of
E. coli on the 158 positive lots (27% of total) was 0.7 (detection limit)
to 6.6 log cfu/g and the interquartile range 1.8e3.3 log cfu/g. The
TVC range was 4e8.7 log cfu/g, with an interquartile of 6e7.1 log
cfu/g. The coliforms range was 2e7.2 log cfu/g, with an interquartile
of 4.1e5.4 log cfu/g. LAB were detected in 18% of the PZ and 38% of
the SA lots with counts on positive lots ranging between detection
limit and 6 log cfu/g and above 3 log cfu/g for the higher quartile.
Yeasts were detected more frequently (94% vs 73%) and with higher
counts on the SA lots, while moulds were more prevalent (82% vs
67%), but with lower counts on the PZ lots. Average counts showed
little variation with the shelf-life stage for E. coli and fungi, but were
higher at the CBD for LAB, coliforms and TVC.
No occurrence of L. monocytogenes or Salmonella was found at
the detection limit (presence in 25 g of produce). E. coli was found
in all months except April, with varying frequency and counts
(Fig. 2). Units with countable E. coli were under 20% of the monthly
sample in eight months, around 33% in seven months and over 33%
in four months, with a peak (56%) in December of the second year.
Count distributions for microbial components are compared
with values adjusted to five days to CBD, predicted by linear
regression of the observed values on days to CBD and other factors
(month, processor, retailer, brand, vegetable species, day of residual
Area
PZ
CBD
3−5
<3
moulds yeasts coliforms TVC
1074 V. Caponigro et al. / Food Microbiology 27 (2010) 1071e1077

Fig. 2. Dot and box plots of monthly distributions of E. coli, TVC and coliforms on 500 consumer units of ready-to-eat salads collected over 19 months in the years 2006e2008 from
retail stores in the SA area. Numbers show sample sizes for TVC and coliforms, sample fraction for detectable E. coli. Horizontally jittered observations adjusted to five days of
residual shelf life by regression.

shelf-life). Values for visual quality scores are adjusted with the
same procedure.
TVC counts showed considerable variation within and between
months. Count ranges within months varied from 1.7 log cfu/g
(December of the first year) to 3.9 log (April). The widest within
month variations were due mostly to extreme values, as those from
some marginal processors found in September and November of
the first year.
Monthly ranges for counts of total coliforms varied between 2.2
and 4.2 log cfu/g, peaking in Autumn months. Coliform loads were
below 6 log cfu/g for 75% of the units in all months except November
and December of the first year and below 5 log cfu/g for 50% of the
units in 11 months.
LAB were found sporadically and were not detected in December
of the first year (Fig. 3). Frequencies of positive units by months
were generally below 50%, except for July and September of the
first year, when LAB counts up to 5.9 log cfu/g were recorded on two
thirds of the units. Counts of moulds and yeasts, with prevalence of
the latter (98%), showed monthly ranges between 2 and 7 log cfu/g
and were 5 log cfu/g for 75% of the units in 13 months and below 4
log cfu/g for 50% the units in six months.
Ranges of visual quality scores in decimal scale by month varied
between 2.2 and 7.5, with the largest range in April and smallest in
September of the second year. Most units were of good visual
quality, with median scores varying between 7.5 and 9.4. Only few
units were given scores below 6.
The biplots in Fig. 4 summarize the influence of season, vege-
table species and packaging on microbial counts and visual quality
index adjusted to five days of residual shelf life. On average, values
of the microbiological indices tended to be higher, and visual
quality scores lower, for mixes and lettuce compared to lamb’s
lettuce, for bags compared to trays (lettuces of the PZ area), and for
the warm season. Minor differences were found between proces-
sors, retailers and brand groups.
The probabilities of countable E. coli and of counts above 100 cfu/
g were about two and three times higher in Summer and Autumn,

Fig. 3. Dot and box plots of monthly distributions of counts of lactic acid bacteria (LAB) and fungi (yeasts þ moulds) and of visual quality scores on 500 consumer units of ready-to-
eat salads collected over 19 months in the years 2006e2008 from retail stores in the SA area. Numbers show sample sizes for fungi and visual quality, sample fraction for detectable
LAB. Observations adjusted to five days of residual shelf life by regression and horizontally jittered to mitigate overlapping.
 V. Caponigro et al. / Food Microbiology 27 (2010) 1071e1077 1075

Fig. 4. Canonical variate biplots of microbial and visual quality (VQI) indices of the lots of ready-to-eat salads collected in the PZ and SA areas, with 80% bagplots (multivariate
extension of the boxplot) for the three most separable vegetable species, season and package type. Observations adjusted to five days of residual shelf life by regression. Sample size:
579 units for species and season, 79 for package type. E. coli ¼ counts; pr.E.coli ¼ probability of detection.

respectively, compared to Winter and Spring (Fig. 5). E. coli presence
was also different among vegetable species, with the probability of
occurrence for lamb’s lettuce five times lower compared with the
other species and the mixes. Arugula showed the highest frequency
of E. coli positives (50%), but counts comparable to mixes, lettuce and
spinach. Frequencies of E. coli positives at low counts were higher
for the biggest processor and for processor’s brands compared
with other processors and private labels, while the biggest retailer
showed lower frequency of positives, but higher counts.
Samples of the Summer and Autumn months showed on average
significantly higher counts for all microbiological indices and lower
quality scores (
10%) compared with Winter and Spring. Differences
between Winter and Spring and between Summer and Autumn were
comparatively small: slightly higher TVC, coliforms and LAB counts,
with quality scores 4% lower in Spring; yeast and mould counts
lower by half, but LAB population four times higher in Summer.
RTES of the biggest processor and retailer averaged slightly
lower counts for all microbial indices and higher quality scores.
However, processor’s brands showed slightly higher TVC, mould
and LAB counts, on average, compared with private labels from the
same processors.
Effects of seasons, processors, retailers, vegetable species and
package type on coliform and mould counts paralleled effects
on TVC. LAB occurrence was lowest for lamb’s lettuce and similar
among other species, processors and retailers. Lamb’s lettuce
carried high counts of yeasts and moulds with lower total bacterial
loads. Spinach was characterized by high levels of TVC, coliforms
and moulds and low levels of LAB, with visual quality better than
lettuce’s on average. Arugula showed lower counts of coliforms
and LAB, higher counts of fungi and higher visual quality scores
compared with lettuce.
3.2. Shelf-life trend
TVC increased during storage at an average rate of 0.12 log cfu/g/
day (Fig. 6). TVC counts at the CBD were above 8 log cfu/g for 18% of the
sample. Coliforms showed a higher rate of increase during storage,
averaging 0.21 log cfu/g/day. At the CBD 21% of the lots showed coli-
form counts above 7 log cfu/g. E. coli counts on positive lots showed
little variation on average. Storage variation of yeasts, moulds and
LAB, estimated cross-sectionally from counts on units with different
residual shelf-life measured at the collection time, was noticeable and
positive only for LAB. Visual quality declined at an average rate of 0.33
points per day, with large variation among lots, but quality scores at
the CBD were still above 7/10 for 50% of the sample.
The relationship between microbial load at 5 days to CBD and
visual quality at CBD is statistically significant for TVC, coliforms
and LAB (Fig. 7). One log cfu/g more of TVC or coliforms or LAB at 5
days to CBD appears associated with a reduction of one half point of
the visual quality index at CBD. The correlation is stronger for
LAB (r ¼
0.58), compared to coliforms (r ¼
0.51) and to TVC
(r ¼
0.36). So, the levels of LAB, a component not always countable
on RTES, and the levels of coliforms, a component rarely absent,
appear to be, more than TVC, useful predictors of produce quality
preservation during the shelf life.
4. Discussion
Tests for Salmonella and L. monocytogenes were motivated by
their implication in disease outbreaks associated with consumption
of fresh and ready-to-eat vegetables (DeWaal et al., 2009). The
negative results suggest that levels of occurrence of these pathogens
on the sampled population of RTES may be below the sensitivity of

Fig. 5. Effect of season, vegetable species, processor, brand and retailer on the occurrence and counts of E. coli, counts of TVC, coliforms, moulds and LAB and visual quality scores
(range 0e10) of 579 consumer units of ready-to-eat salads collected in the PZ and SA areas in the years 2006e2008. Contrasts, with 95% confidence interval (bars), for values
adjusted to five days of residual shelf life by regression.
1076 V. Caponigro et al. / Food Microbiology 27 (2010) 1071e1077

6 10
8

Visual quality index (0:10)

8
5

Coliforms (log cfu/g)

7

E.coli (log cfu/g)

TVC (log cfu/g)
8
7 6 4

5 3
6 6
4 2
5 4
3 1

4 2 0
5 4 3 2 1 0 5 4 3 2 1 0 5 4 3 2 1 0 5 4 3 2 1 0
Days of residual shelf−life

Fig. 6. Individual shelf-life trends of total aerobic viable counts (TVC), coliforms, E. coli and visual quality scores for 579 batches of ready-to-eat salads collected in the PZ and SA
areas in the years 2006e2008 (grey lines) and average trend, with individual and average 95% confidence bands (black lines).

our detection method and little affected by season and other associated with flooding has been indicated as a contributing factor
considered factors. for spread and incidence of the enterotoxic E. coli in a major salad
Microbial loads and visual quality of the examined RTES varied growing area of California (Cooley et al., 2007).
considerably among lots in all conditions. Month, season and vege- E. coli is a frequently used indicator of contamination with
table species, and to a lesser degree processor, distributor, brand and enteric material, given also the ease of determination, and on this
package types, showed some influence on microbial counts and account has been chosen by the EU regulation about fresh ready-to-
produce quality, but these factors accounted for a minor proportion eat vegetables as a microbial indicator of processing quality (EC,
(less than 25%) of the differences among lots. The high residual 2005). The slightly higher prevalence of E. coli and higher TVC
heterogeneity of both microbial levels and visual quality scores, and fungi counts on processor’s brands of RTES compared to retail
after accounting for the above factors, show that other factors not private labels from the same processors shows that there are
considered in this study may have a larger impact on the overall margins for brand quality improvement.
quality of RTES marketed in Italy, linked to conditions of the culti- Yeasts predominance over moulds is in agreement with the
vation environment and also to processing operations, as delays of finding of Tournas (2005). Levels of TVC and coliforms were
refrigeration after harvesting or of processing after reception of the comparable with those of samples previously studied in Italy
vegetables in the factory. and elsewhere, but LAB levels were considerably lower. While
Ailes et al. (2008), analysing at post-harvest and processing psychrotrophic strains of LAB isolated from commercial salads
stages a large sample of fresh vegetable produce of several species have been shown capable of inhibiting pathogens like Aeromonas
(not all ready-to-eat products) from areas of southern United States hydrophila, L. monocytogenes , Staphylococcus aureus and Salmonella
and Mexico, found significant variability among species and seasons, typhimurium (Vescovo et al., 1996), LAB counts of 6 log cfu/g have
with microbial loads peaking in Autumn, as in our case, observing been considered signal of the beginning of spoilage (García-Gimeno
also that microbial concentrations, as well as the probability of E. coli and Zurera-Cosano, 1997; Ragaert et al., 2007). The significant
detection could increase during processing, particularly at steps like negative relationship found between produce visual quality and the
binning, turntable after washing, conveyor belt and packing in box, levels of TVC, coliforms and LAB confirms the association between
involving direct contact between produce items and equipment spoilage and LAB levels and the statement that RTES shelf-life is
surfaces and worker’s hands. The higher frequency E. coli in Autumn largely conditioned by the microbiological and sensory quality
may be favored by wetter conditions than in Summer coupled with levels at the processing factory gate (Sinigaglia et al., 1999).
temperatures milder than in Spring. In our growing area Autumn Quality deterioration during the residual shelf-life, at a constant
and Spring are the rainy seasons and rain is often concentrated in temperature around 3  C, was on average slower than reported in
short periods, so temporary overflowing of drains with pollution of other studies (García-Gimeno and Zurera-Cosano, 1997; Sinigaglia
fields and waters reservoirs cannot be excluded. Higher water flow et al., 1999), but microbial counts were already high at the start

8
VQI at CBD (0−10)

b = −0.55 [−0.43;−0.68] b = −0.51 [−0.59; −0.44] b = −0.54 [−0.61; −0.47]

5 r = −0.36 [−0.44; −0.28] r = −0.51 [−0.57; −0.44] r = −0.58 [−0.63; −0.52]

5.5 6.0 6.5 7.0 7.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 0 1 2 3
TVC at 5 days to CBD (log cfu/g) Coliforms at 5 days to CBD (log cfu/g) LAB at 5 days to CBD (log cfu/g)

Fig. 7. Relationships between visual quality scores at the CBD and microbial load at five days to CBD for the sample of 579 lots of ready-to-eat salads collected in the PZ and SA areas
in the years 2006e2008. Linear fit, with individual and average 95% confidence bands, and coefficients of slope and linear correlation, with 95% confidence interval.
 V. Caponigro et al. / Food Microbiology 27 (2010) 1071e1077
on one fifth of the lots, so that TVC counts exceeded 7.7 cfu/g for one
quarter of the lots at CBD.
The better average quality of RTES from the biggest processor
and retailer show that produce management and efficiency within
the production-distribution channel tend to improve with the scale
of operations. However, the unaccounted high residual variation of
microbial loads and visual quality suggests that for better microbial
containment other possible causes of variability should be inves-
tigated, at the raw leaf production and pre-processing stages, which
up to now have been little studied to this purpose.
Acknowledgements
The research was funded by the Italian Ministry of Agrifood
and Forestry Policies. The authors would like to thank Mr Panfilo
Marrollo and Ms Ilaria De Luca for help in sample collection and
laboratory work.
References
Abadias, M., Usall, J., Anguera, M., Solsona, C., Viñas, I., 2008. Microbiological quality
of fresh, minimally-processed fruit and vegetables, and sprouts from retail
establishments. Int. J. Food Microbiol. 123, 121e129.
Ailes, E.C., Leon, J.S., Jaykus, L.A., Johnston, L.M., Clayton, H.A., Blanding, S.,
Kleinbaum, D.G., Backer, L.C., Moe, C.L., 2008. Microbial concentrations on fresh
produce are affected by postharvest processing, importation, and season. J. Food
Prot. 71, 2389e2397.
Bates, D., Maechler, M., 2009. lme4: Linear mixed-effects models using S4 classes. R
Package Version 0.999375-32.
Brandl, A.M., 2006. Fitness of human enteric pathogens on plants and implications
for food safety. Ann. Rev. Phytopathol. 44, 367e392.
Carmichael, I., Harper, I.S., Coventry, M.J., Taylor, P.W.J., Wan, J., Hickey, M.W., 1999.
Bacterial colonization and biofilm development on minimally processed vege-
tables. J. Appl. Microbiol. 85, 45Se51S (Symposium Supplement).
Catellani, P., Giaccone, V., Novelli, E., 2006. Microbial flora of freshly-produced
vegetables: a preliminary report. Vet. Res. Commun. 30, 345e347.
CNERNA-CNRS, 1996. Produits de la IV gamme. In: Jouve, J.L. (Ed.), Qualité Micro-
biologique des Aliments, Maitrise et Critères. Polytechnica, Paris.
Cooley, M., Carychao, D., Crawford-Mikza, L., Jay, T., Myers, C., Rose, C., Keys, C.,
Farrar, J., Mandrell, R.E., 2007. Incidence and tracking of Escherichia coli O157:H7
in a major produce production region in California. Plos ONE 2, e1159.
de Curtis, M.L., Franceschi, O., De Castro, N., 2002. Listeria monocytogenes in vege-
tables minimally processed. Arch. Latinoam.Nutr. 52, 282e288.
DeWaal, C.S., Tian, X.A., Plunkett, D., 2009. Outbreak Alert! Washington, DC.
Analyzing Foodborne Outbreaks1998 to 2007.
DGHM, 2009. Veröffentlichte richt- und warnwerte zur beurteilung von lebensmitteln.
empfehlung der fachgruppe lebensmittel-mykrobiologie und lebensmitte-hygiene
der deutschen gesellschaft für hygiene und mikrobiologie. stand 2009.
EC, 2005. Regulation 2073/2005. EU Official Journal, 7/12/2007, L 322, p. 1e12.
Francis, J., Thomas, C., O’Beirne, D., 1999. The microbiological safety of minimally
processed vegetables. Int. J. Food Sci. Technol. 34, 1e22.
Franz, E., van Bruggen, A.H.C., 2008. Ecology of E. coli O157:H7 and Salmonella
enterica in the primary vegetable production chain. Crit. Rev. Microbiol. 34,
143e161.
Fröder, H., Martins, C.G., De Souza, K.I., Landgraf, M., Franco, B., Destro, M.T., 2007.
Minimally processed veegetable salads, microbial quality evaluation. J. Food
Prot. 70, 1277e1280.
García-Gimeno, R.M., Zurera-Cosano, G., 1997. Determination of ready-to-eat
vegetable salad shelf-life. Int. J. Food Microbiol. 36, 31e38.
Gianfranceschi, M., Gattuso, A., Tartaro, S., Aureli, P., 2003. Incidence of Listeria
monocytogenes in food and environmental samples in Italy between 1990 and
1077
1999, serotype distribution in food, environmental and clinical samples. Eur. J.
Epidemiol. 18, 1001e1006.
Gómez-López, V.M., Ragaert, P., Debevere, J., Devlieghere, F., 2008. Decontamination
methods to prolong the shelf-life of minimally processed vegetables, state-of-
the-art. Crit. Rev. Food Sci. Nutr. 48, 487e495.
Gower, J.C., Hand, D.J., 1996. Biplots. Monographs on Statistics and Applied Proba-
bility. Chapman & Hall, London.
la Grange, A.M., le Roux, N.J., Gardner-Lubbe, S., 2009. Biplotgui: interactive biplots
in r. J. Stat. Softw. 30, 1e37.
Hagenmaier, R.D., Baker, R.A., 1998. A survey of the microbial population and
ethanol content of bagged salad. J. Food Prot. 61, 357e359.
Heaton, J.C., Jones, K., 2007. Microbial examination of fruit and vegetables and the
behaviour of enteropathogens in the phyllosphere, a review. J. Appl. Microbiol.
104, 613e626.
Hothorn, T., Bretz, F., Westfall, P., 2008. Simultaneous inference in general
parametric models. Biom. J. 50, 346e363.
Imai, K., King, G., Lau, O., 2009. Zelig, Everyone’s Statistical Software. R package
version 3.4-5.
Jacxsens, L., Devlieghere, F., Debevere, J., 2001. Temperature dependence of shelf-life
as affected by microbial proliferation and sensory quality of equilibrium modi-
fied atmosphere packaged fresh produce. Postharvest Biol. Technol. 26, 59e73.
Junttila, J.R., Niemela, S.I., Hirn, J., 1988. Minimum growth temperatures of Listeria
monocytogenes and non-haemolytic listeria. J. Appl. Bacteriol. 65, 321e327.
Kader, A.A., Zagory, D., Kerbel, E.L., 1989. Modified atmosphere packaging of fruits
and vegetables. Crit. Rev. Food Sci. Nutr. 28, 1e30.
Loncarevic, S., Johannessen, G.S., Rorvik, L.M., 2005. Bacteriological quality of
organically grown leaf lettuce in Norway. Lett. Appl. Microbiol. 41, 186e189.
Mosupye, F.M., von Holy, A., 1999. Microbiological quality and safety of ready-to-eat
street-vended foods in Johannesburg, South Africa. J. Food Prot. 62, 1278e1284.
Nguyen-The, C., Carlin, F., 1994. The microbiology of minimally processed fresh
fruits and vegetables. Crit. Rev. Food Sci. Nutr. 34, 371e401.
Pavan da Silva, S.R., Frizzo Verdin, S.E., Pereira, D.C., Schatkoski, A.M., Rott, M.B.,
Corção, G., 2007. Microbiological quality of minimally processed vegetables sold
in Porto Alegre, Brazil. Braz. J. Microbiol. 38, 594e598.
Pianetti, A., Sabatini, L., Citterio, B., Pierfelici, L., Ninfali, P., Bruscolini, F., 2008.
Changes in microbial populations in ready-to-eat vegetable salads during shelf-
life. Ital. J. Food Sci. 20, 245e254.
R Development Core Team, 2009. R: A Language and Environment for Statistical
Computing. R Foundation for Statistical Computing, Vienna, Austria, ISBN
3-900051-07-0.
Ragaert, P., Devlieghere, F., Debevere, J., 2007. Role of microbiological and physio-
logical spoilage mechanisms during storage of minimally processed vegetables.
Postharvest Biol. Technol. 44, 185e194.
Rico, D., Martin-Diana, A.B., Barat, J.M., Barry-Ryan, C., 2007. Extending and
measuring the quality of fresh-cut fruit and vegetables, a review. Trends Food
Sci. Technol. 18, 373e386.
Risch, S.J., 2009. Food packaging history and innovations. J. Agric. Food Chem. 23,
8089e8092.
Sagoo, S.K., Little, C.L., Mitchell, R.T., 2001. The microbiological examination of
ready-to-eat organic vegetables from retail establishments in the United
Kingdom. Lett. Appl. Microbiol. 33, 434e439.
Scifò, G., Randazzo, C.L., Restuccia, C., Fava, G., Gaggia, C., 2009. Listeria innocua
growth in fresh cut mixed leafy salads packaged in modified atmosphere. Food
Control 20, 611e617.
Sinigaglia, M., Albenzio, M., Corbo, M.R., 1999. Influence of process operations on
shelf-life and microbial population of fresh-cut vegetables. J. Ind. Microbiol.
Biotechnol. 23, 484e488.
Tournas, V.H., 2005. Moulds and yeasts in fresh and minimally processed vegetables
and sprouts. Int. J. Food Microbiol. 99, 71e77.
Valentin-Bon, I., Jacobson, A., Monday, S.R., Feng, P.C.H., 2008. Microbiological quality
of bagged cut spinach and lettuce mixes. Appl. Environ. Microbiol. 74, 1240e1242.
Vescovo, M., Torriani, S., Orsi, C., Macchiarolo, F., Scolari, G., 1996. Application
of antimicrobial-producing lactic acid bacteria to control pathogens in ready-
to-use vegetables. J. Appl. Bacteriol. 81, 113e119.
Wickham, H., 2009. ggplot2: Elegant Graphics for Data Analysis. Springer, New York.
Wojcik-Stopczynska, B., 2004. Microbiological quality of minimally processed
vegetable salads. Rocz. Panstw. Zakl. Hig. 55, 139e145.