Ecologically-based Management of Rodent Pests

ECOLOGICALLV-BASED
MANAGEMENT
OF RODENT PESTS

Edited by: Grant R. Singleton, Lyn A. Hinds, Herwig Leirs and Zhibin Zhang

Australian Centre for International Agricultural Research

Canberra 1999
The Australian Centre for International Agricultural Research (ACIAR) was established in June
1982 by an Act of the Australian Parliament. Its primary mandate is to help identify agricultural
problems in developing countries and to commission collaborative research between Australian
and developing country researchers in fields where Australia has special competence.

Where trade names are used this constitutes neither endorsement of nor discrimination against
any product by the Centre.

ACIAR MONOGRAPH SERIES

This peer-reviewed series contains the results of original research supported by ACIAR, or
deemed relevant to ACIAR's research objectives. The series is distributed internationally, with
an emphasis on the Third World

Australian Centre for International Agricultural Research GPO Box Canberra, ACT 2601.

Singleton, C.R., Hinds, L.A., Leirs, H. and Zhang, Z.ed. 1999. Ecologically-based management of
rodent ACIAR Monograph No. 59, 494p.

ISBN 1 86320 262 5

Editing and design by Arawang Communication Croup, Canberra
Printed by Brown Prior Anderson, Melbourne, Australia
 page

Author Contact Details 8

Abbreviations 12
List of Species 13
Preface 15

1. Ecologically-based Management of Rodent Pests-Re-evaluating 17

Our Approach to an Old Problem
Grant R. Singleton, Herwig Leirs, Lyn A. Hinds and Zhibin Zhang

Section 1 Basic Research - the Foundation for Sound Management 31

2. Current Paradigms of Rodent Population Dynamics- 33

What Are We Missing?
Charles J. Krebs

3. The Behaviour and Ecology of Rattus norvegicus: from Opportunism to 49

Kamikaze Tendencies
David W. Macdonald, Fiona Mathews and Manuel Berdoy

4. Models for Predicting Plagues of House Mice (Mus domesticus) 81

in Australia
Roger P. Pech, Greg M. Hood, Grant R. Singleton, Elizabeth Salmon,
Robert I. Forrester and Peter R. Brown

5. Rodent-Ecosystem Relationships: a Review 113

Chris R. Dickman

6. The Role of Rodents in Emerging Human Disease: Examples 134

from the Hantaviruses and Arenaviruses
James N. Mills

5
Ecologically-based Rodent Management

Section 2 Methods of Management 161

7. Rodenticides Their Role in Rodent Pest Management in 163

Tropical Agriculture
Alan P. Buckle

8. Physical Control of Rats in Developing Countries 178

Grant R. Singleton, Sudarmaji, Jumanta, Tran Quang Tan and
Nguyen Quy Hung

9. Ecological Management of Brandt's Vole (Microtus brandti) in 199

Inner Mongolia, China
Wenqin Zhong, Mengjun Wang and Xinrong Wan

10. Biological Control of Rodents- the Case for Fertility Control Using 215
Immunocontraception
Usa K. Chambers, Ma/co/m A. Lawson and Lyn A. Hinds

11. Urban Rodent Control Programs for the 21st Century 243
Bruce A. Colvin and William B. Jackson

Section 3 Case Studies in Asia and Africa 259

12. Rodent Pest Management in Agricultural Ecosystems in China 261

Zhibin Zhang, Anguo Chen, Zhendong Ning and Xiuqing Huang

13. Rodent Pest Management in the Qinghai-Tibet Alpine Meadow 285

Ecosystem
Naichang Fan, Wenyang Zhou, Wanhong Wei, Quanye Wang and
Yongjin Jiang

14. Ecologically-Based Population Management of the Rice-Field 305

Rat in Indonesia
Luke K.P. Leung, Grant R. Singleton, Sudarmaji and Rahmini

15. Population Ecology and Management of Rodent Pests in 319

the Mekong River Delta, Vietnam
Peter R. Brown, Nguyen Quy Hung. Nguyen Manh Hung and
Monica van Wensveen

16. Rodent Management in Thailand 338

Puangtong Boonsong, Sermsakdi Hongnark, Kornkaew Suasa-ard,
Yuvaluk Khoprasert, Prasarttong Promkerd, Greangsak Hamarit,
Piyanee Nookarn and Thomas Jakel

6
 Contents

17. Farmer Participatory Research on Rat Management in Cambodia 358

Gary C. Jahn, Mak Solieng, Peter G. Cox, and Chhom Nel

18. Rodents in Agriculture in the Lao PDR a Problem with 372

an Unknown Future
John M. Schiller, Bounneuang Douang Boupha and Onechanh Bounnaphol

19. Populations of African Rodents: Models and the Real World 388
Herwig Leirs

20. Ecophysiology and Chronobiology Applied to Rodent Pest 409

Management in Semi-arid Agricultural Areas in Sub-Saharan
West Africa
Bruno Sicard, Wamian Diarra and Howard M. Cooper

21. The Rodent Problem in Madagascar: Agricultural Pest and Threat to 441
Human Health
Jean-Marc Duplantier and Daniel Rakotondravony

22. Rodent Pest Management in East Africa-an Ecological Approach 460

Rhodes H. Makundi, Nicho/as O. Oguge and Patrick S. Mwanjabe

23. Ecologically-based Rodent Management in Developing Countries: 477

Where to Now?
Herwig Leirs, Grant R. Singleton and Lyn A. Hinds

Index 485

7
Berdoy. Manue!, Wildlife Conservation Research
Unit, Department of Zoology, and Department of
Veterinary Services, clo University Laboratory of
Physiology, University of Oxford, South Parks Road,
Oxford, OXl 3PT, UK.
Email: manuel.berdoyVet.ox.ac.uk
Boonsong, Puangtong, Agricultural Zoology
Research Group, Division of Entomology and
Zoology, Department of Agriculture, PO Box 9-34,
Bangkok 10900, Thailand.
Email: agrizoobkkl.loxinfo.co.th
Fax: +662 9405396
Bounnaphol, Onechanh, Agriculture and Forestry
Service, Luang Prabang, Lao PDR.
Fax: +856 71 212635
Brown, Peter R, CSIRO Wildlife and Ecology, GPO
Box 284, Canberra, ACT 2601, Australia.
Email: peter.brown@dwe.csiro.au
Fax: +61 26242 1505
Buckle, Alan P., Zeneca Agrochemicals, Fernhurst,
Haslemere, Surrey GU27 3JE, UK.
Email: Alan.Buckle@agukzeneca.com
Fax: +44 14286555668
Chambers, Lisa K., Vertebrate Biocontrol
Cooperative Research Centre, CSIRO Wildlife and
Ecology, GPO Box 284, Canberra, ACT 2601,
Australia.
Email: lisa.chambers@dwe.csiro.au
Fax: +612 6242 1505
Chen, Anguo, Changsha Institute of Agriculture
Modernization, Chinese Academy of Science,
\...ll'HI!Plld, Hunan 410125, P.R China.
Fax: +86 731 4612685
8
Chhom, 'Je!, Cambodia-IRRI-Australia Project
(ClAP), PO Box 1, Phnom Penh, Cambodia.
Email.: IRRI -CAMBODIACGIAR.ORG
Fax: +85523211728
Colvin, Bruce A., 16 Temple Road, Lynnfield, MA
01940, USA.
Email: bacolvin@juno.com
Fax: +1 781 334 5782
Cooper, Howard M., INSERc'vf (Institut National
pour la Sante et la Recherche Medicale), U37!,
Cerveau et Vision, Bron, France.
Email: cooper@albert.lyon151.inserm.fr
Fax: +33472913461
Cox, Peter G., Cambodia-IRH.l-Australia Project
(ClAP), PO Box 1, Phnom Penh, Cambodia.
Email: Peter.Cox@bigpond.com.kh
Fax: +85523211728
Diarra, Wamian, CNRST (Centre National de la
Recherche Scientifique et Technologique), BP 3052,
Bamako, Mali.
Dickman, Chris R., Institute of Wildlife Research,
School of Biological Sciences, University of Sydney,
New South Wales 2006, Australia.
Email: cdickman@bio.usyd.edu.au
Fax: + 612 93514119
Douang Boupha, Bounneuang, Department of
Agriculture and Extension, Vientiane, Lao PDR.
Email: j.m.schiller@cgiaLorg
Fax: +856 21414373
Duplantier, Jean-Marc, Programme RAMSE, IRD
(Institut de Recherche pour le Developpement, ex
ORSTOM), BP 434, Antananarivo, Madagascar.
Email: duplantier@ramse.ird.mg
Fax: +261202240451

Fan, Naichang, Department of Biology, Zhejiang
Normal University, Jinhua, Zhejiang 321004, P.R.
China.
Forrester, Robert I., CSIRO Mathematical and
Information Sciences, GPO Box 664, Canberra, ACT
2601, Australia.
Email: bobJorrester@cmis.csiro.au
Fax: 61 26216 7111
Hamarit, Greangsak, Agricultural Zoology Research
Group, Division of Entomology and Zoology,
Department of Agriculture, PO Box 9-34, Bangkok
10900, Thailand.
Email: agrizoo@bkkl.loxinio.co.th
Fax: +66 2 9405396
Hinds, Lyn A., CSIRO Wildlife and Ecology, GPO
Box 284, Canberra, ACT 2601, Australia.
Email: lyn.hinds@dwe.csiro.au
Fax: +61 26242 1505
Hongnark, Sermsakdi, Agricultural Zoology
Research Group, Division of Entomology and
Zoology, Department of Agriculture, PO Box 9-34,
Bangkok 10900, Thailand.
Email: agrizoo@bkkl.loxinfo.co.th
Fax: +66 2 9405396
Hood, Greg M., CSIRO Wildlife and Ecology, GPO
Box 284, Canberra, ACT 2601, Australia.
Email: greg.hood@dwe.csiro.au
Fax: +61 262421505
Huang, Xiuqing, Guandong Institute of Plant
Protection, Chinese Academy of Agricultural
Science, Guangzhou, Guangdong 510640, PR China.
Fax: +862087561757
Hung, Nguyen Manh, Institute of Agricultural
Science of South Vietnam, Ho Chi Minh City,
Vietnam.
Email: ias.hung@hcm.vnn.vn
Fax: +84 8 829 7650
Hung, Nguyen Quy, Institute of Agricultural Science
of South Vietnam, Ho Chi Minh Vietnam.
Email: ias.hung@hcm.vnn.vn
Fax: +84 8 829 7650
Author Contact Details
Jackson, William B., Department of Biological
Sciences, Bowling Green State University, Bowling
Green, OH 43403, USA.
Fax: +1419 372 2024
Jahn, Gary c., Cambodia-IRRI-Australia Project
(ClAP), PO Box 1, Phnom Penh, Cambodia.
Email: G.JAHN@cgiar.org
Fax: +85523211728
Jakel, Thomas, German Technical Cooperation
(GTZ, Deutsche Gesellschaft fur Technische
Zusammenarbeit), 65726 Eschborn, Germany.
Email: tojack@uni-hohenheim.de
Jiang, Yongjin, Northwest Plateau Institute of
Biology, Chinese Academy of Sciences, Xining
810001, P.R. China.
Fax: +869716143282
Jumanta, Research Institute for Rice, West Java,
h,donesia.
Email: sudarrnaji@vision.net.id
Fax: +62260520157
Khoprasert, Yuvaluk, Agricultural Zoology
Research Group, Division of Entomology and
Zoology, Department of Agriculture, PO Box 9-34,
Bangkok 10900, TI1ailand.
Email: agrizoo@bkkLloxinfo.co.th
Fax: +6629405396
Krebs, Charles, J. Department of Zoology, University
ofBritishCoiumbia, Vancouver, RC V6T lZ4,
Canada.
Email: krebs@zoology.ubc.ca
Fax: +1 6048222416
Lawson, Malcolm A., Pest Animal Control
Cooperative Research Centre, Microbiology
Department, University of Western Australia, QEII
Medical Centre, Nedlands, WA 6009, Australia.
Email: mlawson@cyllene.uwa.edu.au
Fax: +61893462912
Leirs, Herwig, Danish Pest Infestation Laboratory,
Skovbrynet 14, DK-2800 Lyngby, Denmark.
Email: h.leirs@ssLdk
Fax: +4545 931155
9
Ecologica ..y-based Rodent Management
Leung, Luke K.-P., CSIRO Wildlife and Ecology,
GPO Box 284, Canberra, ACT 2601, Australia.
Email: lukeleung@telstra.easymail.com.au
Macdonald, David W., Wildlife Conservation
Research Unit, Department of Zoology, University of
Oxford, South Parks Road, Oxford, OX1 3PT, UK.
Fax: +44 1865 310447
Mak, Solieng, Carnbodia-IRRI-Australia Project
(ClAP), PO Box 1, Phnom Penh, Cambodia.
Email: msolieng@bigpond.com.kh
Fax: +85523211728
Makundi, Rhodes H., Rodent Research Unit,
Sokoine University of Agriculture, PO Box 3110,
Morogoro, Tanzania.
Email: rmakundi@suanet.ac.tz
Fax: +255563748
Mathews, Fiona, Wildlife Conservation Research
Cnit, Department of Zoology, University of Oxford,
South Parks Road, Oxford, OXl 3PT, UK.
Email: fmathews@ermine.ox.ac.uk
Fax: +44 1865310447
Mills, James N., Special Pathogens Branch, Division
of Viral and Rickettsial Diseases, National Center for
Infectious Diseases, Centers for Disease Control and
Prevention (Mailstop G-14), 1600 Clifton Road,
Atlanta, Georgia 30333, USA.
Email: jumO@cdc.gov
Fax: +14046391118
Mwanjabe, Patrick 5., Rodent Control Centre, PO
Box 3047, Morogoro, Tanzania.
Email: Rodent@suanet.ac.tz
Ning, Zhendong, Shanxi Institute of Plant
Protection, Chinese Academy of Agricultural
Science, Taiyuan, Shanxi 030031, P.R China.
Fax: +860351 7127683
Nookam, Piyanee, Agricultural Zoology Research
Group, Division of Entomology and Zoology,
Department of Agriculture, PO Box 9-34, Bangkok
10900, Ihailand.
Email: agrizoo@bkkl.loxinfo.co.th
Fax: +66 2 405396
10
Oguge, Nicholas 0., Department of Zoology,
Kenyatta University, PO Box 43844, Nairobi, Kenya.
Email: nogugel@excite.comor
noguge@africaOnline.co.ke
Pech, Roger P., CSIRO Wildlife and Ecology, GPO
Box 284, Canberra, ACT 2601, Australia.
Email: roger.pech@dwe.csiro.au
Fax: +61 26242 1505
Promkerd, Prasarttong, Agricultural Zoology
Research Group, Division of Entomology and
Zoology, Department of Agriculture, PO Box 9-34,
Bangkok 10900, Thailand.
Email: agrizoo@bkk1.loxinfo.co.th
Fax: +66 2 9405396
Rahmini, Research Institute for Rice, Departemen
Pertanian, JI. Raya No. 9, Sukamandi, Subang 41256,
Indonesia.
Email: sudarmaji@Vision.net.id
Fax: +62260 520157
Rakotondravony, Daniel, Departement de Biologie
Anirnale, Universite d' Antananarivo, BP 906,
Antananarivo, M'ld<llgasc<lr.
Salmon, Elizabeth, CSIRO Plant Industry, GPO Box
1600, Canberra, ACT 2601, Australia.
Email: Libby.5almon@pi.csiro.au
Fax: +61 262465000
Schiller, John M., Lao-IRRl Project, PO Box 4195,
Vientiane, Lao PDR
Email: j.m.schiller@cgiar.org
Fax: +856 21 414373
Sicard, Bruno, Laboratoire de Mammalogie, !RD-
ORSTOM (Institut de Recherche pour le
Developpement, French Scientific Research Institute
for Development through Cooperation), BP 84,
Bamako, Mali.
Email: sicard@ird.ml
Fax: +223227588
Singleton, Grant R, CSIRO Wildlife and Ecology,
GPO Box 284, Canberra, ACT 2601, Australia.
Email: grant.singleton@dwe.csiro.au
Fax: +61 262421505

Suasa-ard, Kornkaew, Agricultural Zoology
Research Group, Division of Entomology and
Zoology, Department of Agriculture, PO Box 9-34,
Bangkok 10900, Thailand.
Email: agrizoo@bkkl.loxinfo.co.th
Fax: +66 2 9405396
Sudarmaji, Research Institute for Rice, Departemen
Pertanian, JI. Raya No. 9, Sukamandi, Subang 41256,
Indonesia.
Email: sudarmaji@vision.net.id
Fax: +62 260520157
Tan, Tran Quang, Nationallnstitute for Plant
Protection, Hanoi, Vietnam.
Email: tuathn.vnn.vn
Fax: +84 4 836 3563
van Wensveen, Monica, CSIRO Wildlife and
Ecology, GPO Box 284, Canberra, ACT 2601,
Australia.
Emai]: Monicavw@dwe.csiro.au
Fax: +61 262421505
Wan, Xinrong, Institute of Zoology, Chinese
Academy of Sciences, Beijing 100080, P.R. China.
Email: xwan@public.east.cn.net
Fax: +86 10 62555689
Wang, Mengjun, Institute of Zoology, Chinese
Academy of Sciences, Beijing 100080, P.R. China.
Fax: +86 10 62555689
Author Contact Details
Wang, Quanye, Northwest Plateau Institute of
Biology, Chinese Academy of Sciences, Xining
810001, P.R. China.
Wei, Wanhong, Northwest Plateau Institute of
Biology, Chinese Academy of Sciences, Xining
810001, P.R. China.
Fax: +86971 6143282
Zhang, Zhibin, National Key Laboratory of
Integrated Pest Management on Insects and Rodents,
Institute of Zoology, Chinese Academy of Sciences,
Zhongguancun Lu #19, Haidian District, Beijing
100080, People's Republic of China.
Email: zhangzb@Panda.ioz.ac.cn
Fax: +8610 62556418
Zhong, Wenqin, Institute of Zoology, Chinese
Academy of Sciences, Beijing 100080, People's
Republic of China.
Email: zhongwq@panda.ioz.ac.cn
Fax: +861062555689
Zhou, Wenyang, Northwest Plateau Institute of
Biology, Chinese Academy of Sciences, Xining
810001, P.R. China.
Fax: +869716143282
11
ACIAR Australian Centre for IPM integrated pest management
International Agricultural IRD- Institut de Recherche pour le
Research ORSTOM Developpement; the French
AHF haemorrhagic fever Scientific Research Institute for
AZRG Zoology Research Development through
Group (Thailand) Cooperation
ClAP Cambodia-IRRI-Australia Project IRRI International Rice Research
CRS Catholic Relief Service Institute
CSIRO Commonwealth Scientific and LaoPDR Lao People's Democratic Republic
Industrial Research Organisation MCMV murine cytomegalovirus
(Australia) MIA Murrumbidgee Irrigation Area
EBRM ecologically-based rodent (Australia)
management PICA Predict, Inform, Control, Assess
ECC endogenous circadian clock (strategy)
ECTV ectromelia virus RPM rodent pest management
EWS early wet season (crop) SNV Sin Nombre virus
GMO genetically modified organism TBS trap-barrier system
GTZ Deutsche Gesellschaft fUr TBS+TC TBS plus trap crop
Technische Z usammenarbeit TBW total body water
(German Technical Cooperation) VVIC viral-vectored
HFRS haemorrhagic fever with renal immunocontraception
syndrome ZP zona pellucida (glycoproteins)
HPS hantavirus pulmonary syndrome

12
Species name Common Name Species name Common Name

Acomys cahirinus spiny mouse Hystrix cristata crested porcupine

Apodemus agrarius striped field mouse Hystrix indica Indian crested
Apodemus flavicolfis yellow-necked field porcupine
mouse
Apodemus sylvaticus wood mouse; long- Uomys salvin/ Salvin's spiny pocket
tailed field mouse mouse
Arvicanthis niloticus unstriped grass rat;
Nile grass rat Mastomys coucha multimammate rat
Mastomys erythroleucus multimammate rat
Band/cota bengaJensis lesser bandicoot rat Mastomys huberti multimammate rat
Bandicota indica large bandicoot rat Mastomys nataJensis multimammate rat
Bandicota savilei Marmota himaJayana Himalayan marmot
Bolomys obscurus dark field mouse Meriones unguiculatus Mongolian gerbil;
Brachyuromys ramiroh/tra clawedjird
Micromys minutus harvest mouse
CaJomys callosus "Iaucha grande" Microtus brandti Brandt's vole
Calomys musculinus corn mouse Microtus califomicus California vole
Castor canadensis North American Microtus fortis oriental vole
beaver Microtus mandarinus brown vole
Castor fiber Eurasian beaver Microtus oeconomus root vole
CiteJ/us dauricus Daure ground squirrel Microtus pennsyJvanicus meadow vole
Clethrionomys glareolus bank vole Mus caroli rice mouse
Clethrionomys rufocanus red backed vole Mus cervicolor ryukyu mouse
Cricetomys gambianus African giant Mus domesticus house mouse
pouched rat Mus musculus house mouse
Cricetulus barabensis striped hamster Muscardinus aveJ/anarius dormouse
Cricetulus Jongicaudatus lesser long-tailed Myocastor coypus coypu; nutria
hamster Myospalax baileyi plateau zokor
Cricetulus triton rat-like hamster Myospalax fontanieri Chinese zokor
Cynomys Judovicianus plains prairie dog
Nesokia indica short-tailed
Dipodomys panamintinus Panamint kangaroo bandicoot rat
rat Nesomys rufus
Notomys alexis spinifex hopping
Geomys bursarius eastern American mouse
pocket gopher;
plains pocket gopher Ochotona cansus Gansu pika
Gerbil/us nigeriae Nigerian gerbil Ochotona curzoniae plateau pika; black-
lipped pika
Ochotona daurica Daurian pika

13
Ecologically-based Rodent Management

Species name Common Name Species name Common Name

Oligoryzomys /ongicaudatus long-tailed pygmy rice Rattus rattus diardii Malaysian house rat
rat Rattus tanezumi (formerly Philippine rice-field
Ondatra zibethicus muskrat Rattus rattus mindanensis) rat
Onychomys spp. grasshopper mice Rattus tiomanicus Malayan wood rat
Perognathus parvus Great Basin pocket Rattus villosissimus long haired rat
mouse Rhabdomys pumilio
Peromyscus boym
Peromyscus maniculatus deer mouse Sigmodon a/stoni Alston's cotton rat
Peromyscus truei big eared cliff Sigmodon hispidus cotton rat; Hispid
mouse; Pinyon cotton rat
mouse Solomys spp. tree rats
Pitymys irene Spa/ax ehrenbergi blind mole-rat
Pseudomys sandy inland mouse Suncus murinus common shrew
hermannsburgensis
Tachyoryctes splendens African mole rat
Rattus argentiventer rice-field rat Tatera indica
Rattus bowers; Bower's rat Taterillus graciliS gerbil
Rattus colletti dusky rat Taterillus petteri gerbil
Rattus exu/ans Polynesian rat Taterillus pygargus gerbil
Rattus flav{pectus buff breasted rat Thomomys bottae western American
Rattus germaini pocket gopher; valley
Rattus koratensis Sladen's rat pocket gopher
Rattus losea lesser rice-field rat Thomomys talpoides northern pocket
Rattus nitidus Himalayan rat gopher
Rattus norvegicus Norway rat; brown rat Thryonomys spp. cane rat; cutting
Rattus rattoides Synonym for lesser grass rat
rice-field rat and
Turkestan rat Xerus erythropus ground squirrel
Rattus rattus black rat; house rat;
roof rat Zygodontomys brevicauda cane mouse

14
 HE SEED FOR this book was sown in Morogoro, Tanzania, in 1996,

T following the strong ecological theme that emerged at an international

workshop: Rodent Biology and Integrated Pest Management in Africa.
Herwig Leirs and Grant Singleton were encouraged that the theme of
ecologically-based rodent research came through strongly as the future
direction for rodent management in developing countries in both Africa and
Asia. The opportunity to germinate the seed arose in 1997 when Zhibin Zhang
approached Grant Singleton and Lyn Hinds to co-convene an international
conference on rodent biology and management. The focus would be broader
than the Morogoro workshop and it was obvious that to augment the charm
and appeal of Beijing in early autumn, an impressive line-up of international
speakers would be required to attract participants to the conference. The
Australian Centre for International Agricultural Research (ACIAR), the
Chinese Academy of Sciences and the Commonwealth Scientific and Industrial
Research Organisation (CSIRO) Wildlife and Ecology each pledged support for
the conference. This led to a successful recruiting drive with all the speakers we
approached accepting an invitation to present a paper at the First International
Conference on Rodent Biology and Management held in Beijing in October
1998.
In January 1998, the editors approached ACTAR with the concept of a book
on ecologically-based management that would bring together leading
researchers of the basic biology of rodents and those charged with developing
and implementing management strategies for rodent pests, especially in
developing countries.
The book consists primarily of a selection of papers presented at the Beijing
conference and comprises three sections. Section 1 sets the scene with
contributions from leading small mammal biologists interested in theory and
current paradigms of rodent biology and management. Section 2 covers state-
of-the-art technologies of the different approaches to management of rodent
pests-rodenticides, physical control, urban management and biological
control. Section 3 describes regional case studies of rodent pest problems and
progress wi th their management for a selection of developing countries in Asia
and Africa.

15
Ecologically-based Rodent Management

Internationally, there have been two previous books of note on rodent pest
management: one edited by Ishwar Prakash, published in 1988, the other edited
by Alan Buckle and Robert Smith, published in 1994. Both provided a good mix
of papers on the principles and practices of rodent pest management, and are
compulsory reading for students and practitioners of rodent biology and
management. Our book differs from these two books in providing a
considerably stronger emphasis on (i) ecologically-based management, (H)
recent developments in innovative approaches to biological control, and (iii) the
problems, progress and challenges of rodent pest management in developing
countries. One important element missing in our book, and in the previous two
books, is a substantial contribution on rodent management in Central and South
America. We hope that this void is filled in the near future. In the interim, we
hope that our book is of interest and practical value to researchers in that region
of the world.
This has been a challenging project with more than half of the contributing
authors not having English as their native language. We thank these authors for
their perseverance in the face of obvious frustration in not being able to write in
Bahasa, Cantonese, Flemish, French, Kiswahili, Lao, Mandarin, Thai,
Vietnamese etc. We commend them for their responsiveness to our requests for
many points of clarification and in keeping to a tight schedule.
All chapters were refereed by two people and then edited. We thank fellow
authors for their contributions to the reviewing and editing process as well as
David Spratt, Abigail Smith, Wang Zuwang, Lam Yuet Ming, David
Freudenberger, Alison Mills, Christopher Hardy and Geoffrey Shellam. The
support and enthusiasm of John Copland and Peter Lynch have ensured that the
seed of an idea developed into a bountiful crop-a crop which it is hoped will be
eyed despairingly by rodents in our ecologically-led quest to battle their impact
on our lives.

Grant Singleton
LynHinds
Herwig Leirs
Zhibin Zhang
March 1999

16
Grant R. Singleton, Herwig leirs, lyn A. Hinds and Zhibin Zhang

Abstract

Rodent pest management has gone through a period of stagnation mainly because
there has been too little research effort to understand the biology, behaviour and
habitat use of the species we are attempting to manage. There is a growing demand,
particularly in developing countries, for rodent control strategies that either have
less reliance on chemical rodenticides or can better target their use. Similar
concerns exist with the control of insect and weed pests. This has led to the
development of the concept of ecologically-based pest management (EBPM) which
builds on the progress made with integrated pest management (lPM). We analyse
this idea for rodent pests and provide examples where research on the basic biology
and ecology of rodent pests has provided management strategies that are more
sustainable and environmentally benign. The theme of ecologically-based rodent
management (EBRM) was foremost in our minds when we invited people to
contribute to this book. The other significant considerations were a focus on rodent
pest management in developing countries and the importance of marrying basic and
applied research on rodents. If in developing countries we can foster the importance
of population ecology and an emphasis on management directed at the agro-
ecosystem level, then we are confident that the next decade will see rapid advances
in rodent pest management.

Keywords

Rodent management; IPM; rodent ecology; ecologically-based rodent management

17
Ecologically-based Rodent Management
INTRODUCTION
HE GENESIS of this book was a
T common concern on the lack of
progress in rodent pest
management over the past 20 years in both
developing countries and elsewhere. This
has occurred despite the advent in the 1970s
of sophisticated chemical rodenticides and
effective strategies for their use (see Buckle
1988; Buckle and Smith 1994).
We contend that rodent pest
management has gone through a period of
stagnation for four primary reasons. First,
there has been too great an emphasis on how
to develop, use, compare and market
rodenticides, with particular attention on
commensal rodents in industrialised
countries. In developing countries, on the
other hand, the lack of a critical approach to
the use of rodenticides for particular species
has in some instances led to an unreasonable
aversion to rodenticide use. Second, the
development of rodent control strategies
generally has been based on short-term
experiments where immediate declines in
rodent numbers were seen as a success,
without much consideration of long-term
consequences or ecosystem effects. Third,
field studies have rarely progressed beyond
alpha-level, descriptive population studies
(see Krebs, Chapter 2). Fourth, the
recommended management protocols have
been too prescriptive. They rarely take into
account the particular characteristics of the
pest species or of the socioeconomic
constraints of the end-users of the
technology.
What has been lacking is a solid
understanding of the biology, behaviour and
18
habitat use of the respective species we are
attempting to manage. Armed with such
knowledge we will be able to focus on
disentangling the major factors that limit the
growth of pest populations. This requires
experimental field studies conducted at an
appropriate scale and for an appropriate
length of time. Recently there has been some
progress in the assessment of rodent
management methods using replicated,
manipulative field studies based on our
understanding of the ecology of the pest
species (e.g. Singleton and Chambers 1996;
Brown et a1. 1998; White et a1. 1998; Fan et aI.,
Chapter 13), but there is still much to be
done.
In the interim, there has been a marked
attrition in the number of wildlife
researchers working on rodent pests. Ishwar
Prakash (1988) noted this trend in his
introduction to the pioneering book Rodent
Pest Management.
It is also felt that this work ... will trigger more
research effort for the benefit of mankind, ...
(which) it appears has dampened during the
last few years.
Unfortunately, his plea did not arrest this
trend.
Since 1993 there has been encouraging
evidence of an increase in the number of
young wildlife researchers interested in the
biology and management of rodent pests in
developing countries. This has been due
primarily to funding support provided by
the Australian Centre for International
Agricultural Research in Southeast Asia, the
European Union, Belgium and Denmark in
eastern Africa and ORSTOM (French
Scientific Research Institute for
Development through Cooperation) in
Western Africa. We are pleased that some of
 Re-evaluating Our Approach to an Old Problem
these researchers have been able to
contribute to this book.
China, through necessity, also has seen a
marked increase in research effort on rodent
pests. Rodent problems increased in severity
in the 1980s resulting in rodent control being
listed as one of the top three priorities for the
national plant protection program in 1985.
Since 1985, rodent control has been listed in
three successive national five-year-plans
(1985-1990;1991-1995;1996-2000). There
are now approximately 100 scientists with
the Chinese Academy of Sciences, Ministry
of Agriculture and universities working on
rodent control. Many of these are young
scientists, who received their degree in
biology or post-graduate qualifications in
the 1990s.
In this opening chapter we will set the
scene with a brief overview of the
magnitude of the impact of rodent pests, the
concept of ecologically-based management
and the aims and structure of the book.
RODENT PESTS -STILL A PROBLEM
The quest to control the depredations of
rodents, especially in agricultural systems,
has been ongoing for thousands of years.
Aristotle (384-322 BC) recounts
The rate of propagation of field mice in country
places, and the destruction that they cause, are
beyond all telling.
Although the last 50 years in particular have
provided good progress with rodent pest
management, rural people in many
countries still rank rodents in the top three of
their most important pests. Of particular
concern are the losses caused in developing
countries where rodents are literally
competing with humans for food.
A meeting on rodent pest management in
Southeast Asia was held in early 1998 at the
International Rice Research Institute (IRRI)
in the Philippines. Reports of present-day
rodent problems were presented for
Australia, Cambodia, East Africa, Indonesia,
Lao People's Democratic Republic (PDR),
Malaysia, Philippines, Thailand and
Vietnam; the accounts were impressive in
their extent and impact. Rodent problems
ranged from eruptive populations of mice in
south-eastern Australia and rats in the
uplands of Lao PDR, to the chronic problems
that occur annually in the rice fields of most
Southeast Asian countries.
There were two telling commentaries
from the meeting in the Philippines, which
place in context the impact of rodent pests in
developing countries. One reported that
although rodents were not the most
important pre-harvest pest to Laotian
farmers, they were the pest they felt they had
the least control over. The other presented
losses caused by rodents in Cambodia not in
monetary terms but in how much rice could
have been available for annual human
consumption if not for rat depredations. If
we apply this line of reasoning to Indonesia
where rats cause annual pre-harvest losses
of approximately 17%, then rats consume
enough rice annually to feed more than 25
million Indonesians for a year. In countries
such as Indonesia, rice provides 50-60% of
the daily energy requirements for people.
In some cases, the 'official' national level
of annual pre-harvest losses caused by
rodents is not high. For example, 3-5% losses
are reported in Malaysia (Singleton and
Petch 1994) and 1-3% in the Philippines
(Sumangil1990; Wilma Cuaterno, April
1998, pers. comm.). However, when detailed
19
Ecologically-based Rodent Management
damage assessment is conducted, the
damage caused by rats generally is more
severe. For example, Buckle (1994b) reported
a conservative loss estimate of 7.3% in the
entire Penang State of Malaysia. Also, both
in the Philippines and Malaysia, the patchy
nature of rodent damage often results in
farmers losing more than 60% of their crop,
which means that rodents are still a
significant national problem (Lam 1990). In
other places, rodent damage may vary
widely with limited damage in most years,
and the most extreme losses of more than
80% of the harvest in outbreak years (e.g.
Boonaphol and Schiller 1996). In countries
that live at the brink of subsistence, such
figures are a constant threat to food security.
This book contains detailed accounts of
the magnitude and importance of the impact
of rodent pests, particularly in agricultural
systems. This information in itself is
important because it provides a spotlight on
rodent problems that generally have a lower
profile than insect, weed and disease
impacts on agricultural crops. The latter
group of problems has a higher profile for
two reasons. One is that, in developing
countries, there are many entomologists,
botanists and plant pathologists who are
able to identify, quantify and sell the need
for research, education, extension and action
in their respective fields. In comparison,
there are few rodent biologists; most of these
have an entomological training and there is a
poor infrastructure for research on rodent
pests.
The second reason is that farmers have a
stronger identity with rodents than other
pests. Rodents are perceived as 'intelligent'
pests, which learn to counter whichever
control measures farmers use. Over the
20
centuries, farmers have learned to accept the
depredations caused by rats. A common
response is,
for every eight rows of rice we sow for our
family, we sow two for the rats.
Unfortunately, with the increasing
human population and the shortage of food
in developing countries, this level of loss can
no longer be tolerated.
Clearly, rodents are still an important
problem, and this is without consideration of
the losses they cause post-harvest, and the
role they play as reservoirs for debilitating
diseases of humans and their livestock.
IPM, RODENTICIDES AND
ECOLOGICALLY-BASED MANAGEMENT
Integrated pest management (rPM) is simply
the integration of a range of management
practices that together provide more
effective management of a pest species than
if they are used separately. IPM was
developed with the aim of promoting
methods for managing insect pests and plant
diseases that were least disruptive to the
ecology of agricultural systems (Smith and
van den Bosch 1967).
Ecologically-based pest management
In 1996, a review of pest management of
insects and weeds by the Board on
Agriculture of the National Research
Council (NRC) of the United States of
America, highlighted that the practice of
IPM has generally not been consistent with
the underlying philosophy of IPM. They
contend that there has been too much focus
on pest scouting and precise application of
pesticides. They argue that there is a need to
refocus objectives from pest control to pest
 Re-evaluating Our Approach to an Old Problem
management and this requires greater
emphasis on ecological research and a
systems approach (National Research
Council 1996). This extension and refocusing
of the ecological aspects of IPM led the NRC
to develop a concept termed' ecologically-
based pest management' (EBPM). The
fundamental goals of EBPM are threefold.
One is to minimise adverse effects on non-
target species and the environment. The
second is to develop an approach that is
economic for end-users, particularly
farmers, in both the developed and
developing world. The third is to establish
an approach that is durable.
The development of IPM for rodents has
followed a similar path to IPM for insects.
The primary foci have been the development
of simple monitoring systems to decide
whether or not to instigate a baiting
campaign, and the development of effective
patterns of use for particular rodenticides.
Generally, the focus in rodent control has
been mostly to achieve a visible increase in
mortality, without appropriate attention to
other demographic processes or ecological
compensation mechanisms. There have been
attempts to develop rodent IPM based on an
understanding of the habitat use and
population dynamics of rodent pests (see
Wood and Liau 1984 a,b; Redhead and
Singleton 1989; Whisson 1996; Brown et a1.
White et a1. 1998) or the use of
biological control (e.g. Lenton 1980;
Singleton and Chambers 1996), but with the
possible exception of Rattus tiomanicus in oil
palm plantations (Wood and Liau 1984a,b),
these have not been adopted successfully
over a large area. The progress of rodent IPM
in Southeast Asia and Australia has been
reviewed by Singleton (1997).
Also, biological control needs to be
viewed in the context of ecologically-based
management of pests because often it is
limited in its specificity and This is
supported by a review of one of the success
stories of biological control, the weevil-
Cyrtobagous saiviniae, for controlling the
floating fern salvinia (Salvinia molesta).
Following its establishment in South Asia in
1939, salvinia was spread by man to
Southeast Asia and Australasia. It severely
disrupts the lives of people by forming
dense mats a metre thick, choking slow
moving waterways, rice fields and lakes (see
Thomas and Room 1986 for details). Efforts
to develop biological control were thwarted
initially because the fern was incorrectly
identified, resulting in the testing of the
wrong herbivores. In 1978, salvinia was
found in Brazil where it is relatively rare.
Field studies identified three potential
herbivores and one of these, C. saiviniae, was
released into a lake in northern Queensland
and destroyed 30,000 t of salvinia within a
year (Room et a1. 1981).
When tested in other waterways the
weevil was not a success. Subsequently, a
combination of ecological and laboratory
studies revealed that, if the level of nitrogen
was too low in the fern, the weevil
population declined. Nitrogen was added to
waterways which increased the weevil '
population, until it eventually reached a
critical density at which the damage it
caused to the plant resulted in a sufficient
increase in nitrogen in the plant itself for the
weevil population to be self-sustaining
(Room 1990). This was an unexpected result
because higher levels of nitrogen generally
make weed problems worse. The salvinia
story highlights how taxonomic and
21
Ecologically-based Rodent Management
ecological research provided a strong basis
for a successful systems approach for pest
management.
Ecologically-based rodent
management
For rodents, an ecological basis for control
was suggested many years ago (Hansson
and Nilsson 1975; see also Redhead and
Singleton 1988) but the implementation of
those early ideas has been largely
overlooked. One success was the eradication
of coypu (Myocastor coypus), an introduced
rodent pest, in Britain in the 1980s. After
several decades of unsuccessful control, a
new strategy was developed based on a
long-term population dynamics study and
biological simulations. A complete solution
of the problem was obtained in less than six
years through integrating knowledge about
the animal's biology and behaviour with a
well-organised control scheme with
attractive incentives for trappers (Gosling
and Baker 1989). There are other good
examples in the rodent literature which
illustrate the importance of ecological,
taxonomic and behavioural studies for
developing effective strategies for managing
rodent pests. We provide some further
examples later in this chapter, with more
detailed case studies provided in the
ensuing chapters (Macdonald et aI., Chapter
3; Leung et aI., Chapter 14).
The advantages of viewing biological
control of rodents as part of an integrated
ecologically-based approach to rodent
management rather than a single panacea
for control has been reviewed by Singleton
and Brown (1999). For simplicity, we
propose that this strategy be termed
'ecologically-based rodent management'
22
(EBRM). The contributions by Pech et al.
(Chapter 4) and Hinds et al. (Chapter 10)
further portray the advantage of having a
strong ecological understanding of the
biology of both the rodent pest and the
disease agent when developing techniques
for biological control. In this instance, the
focus is on developing fertility control of
house mice. Without a multi-disciplinary
approach, the requisite knowledge of
reproductive biology, social behaviour
patterns and population dynamics of the
wild house mouse could not be consolidated
to allow full development of a product
which can then be tested for efficacy.
Rodenticide-based control strategies
have a clear need for a good biological basis
to build upon. Toxicity of active ingredients
and bait palatability are obvious factors
which have been studied under laboratory
conditions for many decades (see e.g. Buckle
1994a; Johnson and Prescott 1994). Less
common, but equally important, is a proper
understanding of how poisons can be
delivered. For example, rodenticides in
Hawaiian macadamia orchards were
commonly distributed by broadcasting on
the ground. Recently, population and
behavioural studies of the black rat, Rattus
rattus, revealed that those rats which
damage the nuts forage only in the trees.
This information led to placement of bait
stations in trees leading to more efficient use
of rodenticides for controlling damage
(Tobin et al. 1997).
In China, chemical rodenticides, mostly
anticoagulants, are still the routine weapons
for controlling rodents in farmland and
grassland. However, such rodent control
campaigns in the absence of a sound
ecological knowledge of the pest species
 Re-evaluating Our Approach to an Old Problem
have generally only achieved short periods
(6-9 months) of respite from the ravages of
the rodents. In the rice fields of southern
China the effects have been even shorter
(Huang and Feng 1998). Indeed, many
studies (Liang 1982; Liang et al. 1984; Zhang
1996; Huang and Feng 1998; Qi et al. 1998)
have shown that the response of rodent
populations after chemical control is non-
linear. Killing some individuals may reduce
the population numbers initially, but the
remaining animals compensate with better
survival and better breeding performance.
For example, following an 88% reduction in
a popUlation of the Mongolian gerbil
(Meriones unguiculatus), the body mass at
first pregnancy was reduced from 58 g to 35-
50 g (Wang et al. 1998).
In Malaysia, populations of the Malayan
wood rat (R. tiomanicus) also showed a rapid
population response after control, with a full
recovery in population density occurring
over 12-18 months. In this case, knowledge
of the population dynamics and factors
limiting population growth resulted in an
effective management program of rats in oil
palm plantations. Management consisted of
an intensive baiting campaign followed by
recurrent placement of baits every six
months (see Wood and Liau 1984a).
Re-invasion is another factor resulting in
populations returning quickly to pre-control
densities (e.g. Guruprasad 1992). This is
particularly a problem in developing
countries where farmers often manage their
own rodent problems on small plots of land
(0.25-2 ha) at different times to their
neighbours. The land use patterns on these
small holdings also generally result in a
patchy landscape. We therefore need
ecological studies to examine the relative
demographic importance of each patch and
the timing and rates of movements by rats
between patches (Singleton and Petch 1994).
This metapopulation approach to rodent
control is achieving more attention (see
Smith 1994), but appropriate field studies of
the spatial dynamics of rodent populations
in agro-ecosystems in developing countries
(e.g. Leirs et al. 1997b) are few.
Ethology in rodent pest management
The development of resistance by rodent
pest species to first and second generation
anticoagulants explicitly necessitated an
integrated approach to rodent management,
where use of one poison type was
complemented or alternated with the use of
other poison types, physical control
methods, exclusion, or other control
measures (Greaves 1994). Here again, more
attention was paid to short-term, and indeed
often urgently needed, quick solutions like
changing to a stronger poison. Much less
effort has been directed towards preventing
the development of, or containing the
geographical distribution of, resistance. So-
called 'behavioural resistance', where
rodents refuse to eat the poisonous baits,
poses other challenges. In the Birmingham
restaurant area, house mice were impossible
to control until detailed studies revealed that
they had difficulties in digesting starch and
were therefore unlikely to eat grain-based
baits; changing to fish baits solved the
problem quickly (Humphries et al. 1996).
The Chinese zokor (Myospalax fontanieri)
provides another practical example of the
importance of understanding rodent
behaviour in developing effective
management. In the farmland of Northwest
Loess Plateau, the zokor, which lives
23
Ecologicallybased Rodent Management
underground, shows a cautious response to
chemical baits. Less than 70% of a zokor
population can be killed by using the best
possible baiting technique for this species:
setting baits in their underground tunnels
(Zou et al. 1998). Further improvement in
this kill rate depends on a better
understanding of the behavioural aspects of
feeding for this species, particularly in
overcoming its neophobic response to baits
(Zhang and Wan 1997) or perhaps whether
they show social learning of food
preferences (see Galef 1994; Berdoy 1994 for
reviews).
A good ecological basis to management
strategies can help to provide excellent
rodent damage control without interfering
with rodent demography. Wood mice
(Apodemus sylvaticus) in Germany can be
lured away from sugar beet seeds during the
short period after sowing when they are
prone to rodent damage by providing an
attractive, unpoisoned alternative food in
the periphery of the fields (Pelz 1989). As all
the above examples show, however,
solutions are often specific and require a
detailed knowledge of the biology, ecology
and behaviour of the pest species. Obtaining
such knowledge is a laborious yet rewarding
task that will allow the development of new
damage control strategies.
Further examples of the benefits of
combining know ledge of the ecology and
ethology of rodent species for developing
better integrated control are provided by
Santini (1994) for three European species of
rodents in agriculture and forestry, and
Buckle et al. (1997) for the Malayan wood rat
in oil palm plantations.
24
RE-EMERGENCE OF POPULATION
ECOLOGY OF RODENT PESTS
The current book builds on the strong
ecological theme that emerged at an
international workshop on rodent biology
and integrated pest management in Africa,
held in Morogoro, Tanzania, in 1996 (for
published proceedings see Belgian Journal
of Zoology Volume 127, Supplement). Africa
is an economically poor continent and
control strategies which rely primarily on
rodenticides are unrealistic. This has
sparked interest in a more integrated
ecological approach to rodent pest
management. One of the conclusions of the
workshop was, however, that such strategies
cannot materialise without the availability of
population data from long-term studies
(more than three years) (Leirs 1997). In West
Africa, much iniormation was collected by
Hubert and co-workers in the 1970s (e.g.
Hubert 1982), while in East Africa it is only
in the past few years that long-term
ecological studies have begun to provide
insights into the main factors driving rodent
population dynamics (Leirs et a1. 1996,
1997a). Building on these insights, the focus
has now switched to experimental field
studies.
The workshop in Morogoro formulated
recommendations, many of which are
relevant to the present book (Leirs 1997a).
The key recommendations are as follows:
The taxonomy of many pest rodents must
be clarified so that control actions can
target the correct species.
Life-history studies and physiological
comparisons between these species are
imperative.
 Re-evaluating Our Approach to an Old Problem
.,.,. Experimental ecological studies, properly
designed with appropriate controls, must
be set up to evaluate management
strategies and, in the first place, test our
hypotheses (or, rather, unsubstantiated
beliefs) about rodent population
dynamics.
.,.,. Poisons in this framework are not
considered as something to avoid, but as
only one of the possible approaches which
should be used more effectively and
integrated with other approaches.
The development of the concept of EBPM
is important, because it builds on the solid
foundations developed by IPM. In effect,
EBPM is refocusing IPM towards
understanding the population biology of the
pest and the agro-ecosystem in which it
lives. From the viewpoint of a population
ecologist, one wonders what all the fuss is
about; EBPM is self-evident. However, when
one moves into applied wildlife
management, especially of rodents, then the
need to sell a concept such as ecologically-
based management of rodent pests becomes
a reality (Singleton and Brown 1999).
Unfortunately, too often there is a divide
between practitioners, who are more
concerned with the details of how to apply
specific control technologies, and wildlife
researchers who focus on understanding the
theory and the context of the problem
(Sinclair 1991). We have provided a mix of
pure (Section 1 and parts of 2) and applied
(Sections 2 and 3) rodent biology in this book
in an attempt to bridge this divide.
AIMS AND STRUCTURE OF THE BOOK
This book has four broad aims:
Ill> to raise the profile of the importance of
basic research for developing effective,
applied management of rodent pests;
Ill> to argue the need for an ecologically-based
approach to rodent pest management;
.,.,. to raise the profile of rodent pest
management in developing countries; and
Ill> to spark interest in prospective students in
a challenging but rewarding field of
endeavour.
The book begins with a section on theory
and current paradigms of rodent biology
and management.
This section includes contributions from
leading small mammal ecologists. Krebs
(Chapter 2) provides a thought-provoking
paper on the different phases of small
mammal ecology and concomitant shifts in
research paradigms. Macdonald and
coworkers (Chapter 3) present the results of
a series of novel studies used to disentangle
the interesting social behaviour of Norway
rats. Dickman (Chapter 5) examines, at the
ecosystem level, the positive role rodents
play as 'ecosystem engineers' through their
impact on the chemical and structural
attributes of the environment. Mills in his
chapter on arenaviruses and hantaviruses
(Chapter 6), and Pech and his coworkers
through their synthesis of models for
predicting mouse plagues in Australia
(Chapter 4), both provide a different
perspective of the need for strongly focused
population studies of rodents.
25
Ecologically-based Rodent Management
One common theme is addressed by all
authors-the importance of basic research
for developing effective management of
rodents.
The second section covers broad methods
of management-rodenticides, physical
control and biological control. This section
provides overviews on the state-of-the-art
technologies for fertility control (Chambers
et al., Chapter 10), chemical control (Buckle,
Chapter 7) and the control of rodent pests in
urban environments (Colvin and Jackson,
Chapter 11). Reviews are provided also on
physical methods of control, particularly in
rice agro-ecosystems in developing
countries (Singleton and coworkers, Chapter
8) and on the ecological management of
Brandt's vole in the grassland of Inner
Mongolia (Zhong and coworkers, Chapter
9). The common theme for this section is
ecologically-based pest management.
In a conscious effort to ensure the book is
relevant to developing countries, regional
case studies of rodent problems and the
progress with associated research are
provided for Asia and Africa in Section 3.
This section has contributions from selected
countries edited by G.R. Singleton and Z.
Zhang (Asia-contributions from
Cambodia, China, Indonesia, Lao POR,
Thailand and Vietnam) and H. Leirs
(Africa -contributions from Burkina Faso,
Kenya, Madagascar, Mali and Tanzania).
The information on the biology and
management of rodent pests in developing
countries, and the infrastructure for research
and extension, varies considerably. In some
countries, such as Cambodia and Lao POR,
the problem is only just being defined and it
is still not known which species cause the
major problems in the different agro-
26
ecosystems (see contribution by Schiller et
al., Chapter 18). The contributions in this
section comprise a mix of biological studies
aimed directly at management, and general
overviews of rodent problems and how they
are currently being managed in various
developing countries.
In seeking contributions for this book we
were heartened by the enthusiasm that it
generated from researchers across the
spectrum of pure and applied research. We
received no 'knock backs' from contributors
we targeted. Indeed, we had to limit the
contributions that were on offer. What
pleasantly surprised us was the strong
interest by 'pure' scientists in hoping their
work would not only be of heuristic value.
They were keen for their findings to be
accessible to researchers in developing
countries because they felt their research
could make a significant contribution to
tackling the problem of rodent pests in these
regions. So perhaps Oenis Chitty is indeed
correct in stating "pure and applied science
differ mainly in aims, not methods". If this
book acts as a catalyst for pure and applied
scientists to work together towards a
common aim of reducing the impact of
rodent pests in agricultural ecosystems of
developed and developing countries, then
we will be more than satisfied with our toil.
REFERENCES
Berdoy, M. 1994. Ma king decisions in the wild:
constraints, conflicts and communication in
foraging rats. In: Galef, B.G., Mainardi, M.
and Valsecchi, P., ed., Behavioural aspects of
feeding: basic and applied research in
mammals. Switzerland, Harwood Academic
Publishers, 289-313.
 Re-evaluating Our Approach to an Old Problem
Boonaphol, O. and Schiller, J.M. 1996. Rodent
problems and research priorities in Laos.
Paper for ACIAR Project Planning Meeting
on Management of Rodent Pests in Southeast
Asia, 10-12 January 1996, UPM Malaysia,
14p.
Brown, P.R, Singleton, G.R, Dunn, S.C and
Jones, D.A 1998. The management of house
mice in agricultural landscapes using farm
managment practices: an Australian perspec-
tive. In: Baker, RO. and Crabb, AC, ed.,
Proceedings of the 18th Vertebrate Pest
Conference, Costa Mesa, California, USA, 2-5
March 1998. Davis, University of California,
156-159.
Buckle, A.P. 1988. Integrated management of rice
rats in Indonesia. FAO (Food and Agriculture
Organization of the United Nations) Plant
Protection Bulletin, 36, 111-118.
Buckle, AP. 1994a. Rodent control methods:
chemical. In: Buckle, AP. and Smith, RH.,
ed., Rodent pests and their controL Walling-
ford, UK, CAB International,
127-160.
Buckle, A.P. 1994b. Damage assessment and
damage surveys. In: Buckle, AP. and Smith,
RH., ed., Rodent pests and their control.
Wallingford, UK, CAB International, 219-
248.
Buckle, AP. and Smith, R.H. (ed.) 1994. Rodent
and their control. Walling ford, UK,
CAB International,405p.
Buckle, AP., Chia, T.H., Fenn, M.G.P. and
Visvalingam, M. 1997. Ranging behaviour
and habitat utilisation of the Malayan wood
rat (Rattus tiomanicus) in an oil palm planta-
tion in Johore, Malaysia. Crop Protection, 16,
467-473.
Chitty, D. 1996. Do lemmings commit suicide?
Beautiful hypotheses and ugly facts. New
York, Oxford University Press, 268p.
Galef, E.G. Jr 1994. Olfactory communication
about foods among rats: a review of recent
findings. In Galef, E.G., Mainardi, M. and
Valsecchi, P., ed., Behavioural aspects of
feeding: basic and applied research in
mammals. Switzerland, Harwood Academic
Publishers, 83-101.
Greaves, J.H. 1994. Resistance to anticoagulant
rodenticides. In: Buckle, AP. and Smith, RH.,
ed., Rodent pests and their control. Walling-
ford, UK, CAB International,
197-217.
Gosling, L.M. and Baker, S.J. 1989. The eradica-
tion of muskra ts and coypu from Britain.
Biological Journal of the Linnean Society, 38,
39-51.
Guruprasad, E.K. 1992. I~einfestation of Bandi-
cota bengalensis (Gray) in irrigated field
habitat. In: Borrecco, J.E. and Marsh, RE., ed.,
Proceedings of 15th Vertebrate Pest Confer-
ence,3-5 March 1992, Newport Beach,
California. Davis, University of California,
277-279.
Hansson, L. and Nilsson, B. (eds) 1975. Biocon-
trol of rodents. Ecological bulletins, No 19,
Swedish Natural Science Research CounciL
306p.
Huang, X.Q. and Feng, Z.Y. 1998. Ecology and
management strategies for Rattus losea. In:
Zhibin Zhang and Zuwang Wang, ed.,
Ecology and management strategies of
rodent pests in agriculture. Beijing, China
Ocean Press, 178-194 (in Chinese).
Hubert, B. 1982. Dynamique des populations de
deux especes de rongeurs du senegal, Masto-
mys erythroleucus et Taterillus gracilis (Roden-
tia, Muridae et Gerbillidae). I. Etude
demographique. Mammalia, 46,137-166.
Humphries, RE., Sibly, RM. and Meehan, AI'.
1996. The characteristics of 'behavioural
resistance' and bait avoidance in house mice
in the UK. Proceedings of Brighton Crop
Protection Conference-Pests and Diseases,
157-164.
Johnson, RA. and Prescott, CV. 1994. The
laboratory evaluation of rodenticides. In:
Buckle, AP. and Smith, RH., ed., Rodent
pests and their controL Wallingford, UK,
CAB International, 161-179.
Lam, Y.M. 1990. Cultural control of rice field rats.
In: Quick, G.R, ed., Rodents and rice. Report
and proceedings of an expert panel meeting
on rice rodent control, 10-14 Sept. 1990. Los
Banos, Philippines, IRRI (International Rice
Research Institute), 65-72.
27
Ecologically-based Rodent Management
Leirs, H. 1997. Preface. Rodent biology and
integrated pest management in Africa.
Proceedings of the International Workshop
held in Morogoro, Tanzania, 21-25 October
1996. Belgian Journal of Zoology, 127(suppL),
3-5.
Leirs,H.,Stenseth,N.C, Nkhols,J.D., Hines,J.E.,
Verhagen,Rand W.N.1997a.
Seasonality and non-linear density-depend-
ence in the dynamics of African Mastomys
rats. Nature, 389,176-180.
Leirs,H., Verhagen, R, Sabuni, CA., Mwanjabe,
P. and Verheyen,W.N. 1997b. Spatial dynam-
ics of Mastomys natalensis in a field-fallow
mosaic in Tanzania. Belgian Journal of
Zoology, 127(suppL), 29-38.
Leirs,H., Verhagen,R, Verheyen, W.,
Mwanjabe,P. and Mbise, T. 1996. Forecasting
rodent outbreaks in Africa: an ecological basis
for Mustomys control in Tanzania. Journal of
Applied 33,937-943.
Lenton, GM. 1980. Biological control of rats by
owls in oil palm and other plantations.
Biotrop Publication No. 12,87-94.
Liang, J.R 1982. Population recovery of Chinese
zokor (Myospalax fantanieri) and plateau pika
(Oci1atona curzaniae). In: Wuping Xia, cd.,
Alpine Gansu, Gansu People's
Publishing Press, 93-100 (in Chinese).
Liang, J.R, Zhou, L., Wang, Z.W. and Song, RY.
1984. Mathematical model of population
recovery of Chinese Zokor (Myospalux fOl1 tani-
eri) and plateau pika (Ochotona curzoniae).
Acta Ecologica Sinica, 4,1-11 (in Chinese).
National Research Council 1996. Ecologically
based pest management: new solutions for a
new century. Washington, D.C, National
Academy Press, 144p.
Pelz, H.-J. 1989. Ecological aspects of sugar beet
seeds by Apodemus sylvaticus. In: Putrnan, RJ.,
ed., Mammals as pests. London, Chapman
and Hall, 34-48.
Prakash, I. (cd.) 1988. Rodent pest management.
Boca Raton, CRC Press, 48Op.
Qi,G'x', Yao, W.L., Wang,J. and Yang, B. 1998.
Research on population dynamics and
control strategies for rodents in cities and
towns of southern China. Acta Therio-Iogica
Sinica, 18, 226-230 (in Chinese).
28
Redhead, T.D. and Singleton, G.R 1988. The
PICA Strategy for the prevention of losses
caused by plagues of A,ius domesticus in rural
Australia. EPPO (European Plant Protection
Organization) Bulletin, 18, 237-248.
Room, P.M. 1990. Ecology of a simple plant-
herbivore system: biological control of
Salvinia. Trends in Ecology and Evolution,S,
74-79.
Room, P.M., Harley, K.L., Forno, 1. W. and Sands
D.P.A. 1981. Successful biological control of
the floating weed salvinia. Nature, 294, 79-80.
Santini, L. 1994. Knowledge of feeding strategies
as the basis for integrated control of wild
rodents in agriculture and forestry. In: Galef,
B.G., Mainardi, M. and Valsecchi, P., ed.,
Behavioural aspects of feeding: basic and
applied research in mammals. Switzerland,
Harwood Academic Publishers, 357-368.
Sindair, A.RE. 1991. Science and the practice of
wildlife management. Journal of Wildlife
Management, 55,767-773.
Singleton, G.R 1997. Integrated management of
rodents: a Southeast Asian and Australian
perspective. Belgian Journal of Zoology, 127,
157-169.
Singleton, G.R and Brown, P.R. 1999. Manage-
ment of mouse plagues in Australia: integra-
tion of population ecology, bio-control and
best farm practice. In: Cowan, D.P. and Feare,
c.J., ed., Advances in vertebrate pest manage-
ment. Fiirth, Filander - Verlag, 189-203.
Singleton, G.R. and Chambers, L.K. 1996. A large
scale manipulative field study of the effect of
Capillaria hepatica on wild mouse populations
in southern Australia. International Journal
for Parasitology, 26,383-98.
Singleton, G.R. and I'etch, D.A.1994. A review of
the biology and of rodent pests
in Southeast Asia. Canberra, ACIAR Techni-
cal Reports No. 30, 65p.
Smith, R.H. 1994. Rodent control methods: non-
chemical and non-lethal chemical. In: Buckle,
A.I'. and Smith, R.H., ed., Rodent pests and
their controL Wallingford, UK, CAB Interna-
tional, 109-125.
 Re-evaluating Our Approach to an Old Problem
Smith, RP. and van den Bosch, R 1967.
Integrated control. In: Kilgore, W.W and
Doutt, RT., ed., Pest control: biological,
physical, and selected chemical methods.
New York, Academic Press, 295-340.
Sumangil, rp. 1990. Control of rice field rats in the
Philippines. In: Quick, C.R, ed., Rodents and
rice. Report and proceedings of an expert
panel meeting on rice rodent control, 10-14
Sept. 1990. Los BaflOs, Philippines, IRRl
(International Rice Research Institute), 35-47.
Thomas, P.A. and Room, P.M. 1986. Taxonomy
and control of Salvinia malesta. Nature, 320,
581-584.
Tobin, M. E., Sugihara, RT. and Koehler, A.E.
1997. Bait placement and acceptance by rats in
macadamia orchards. Crop Protection, 16,
507-510.
Wang, M.J., Zhong, W.Q. and Wan, X.R 1998.
Ecology and management strategies for
Mongolia gerbils (Meriones unguiculatus). In:
Zhibin Zhang and Zuwang Wang, ed.,
Ecology and management strategies of
rodent pests in agriculture. Beijing, China
Ocean Press, 225-240 (in Chinese).
White, J., Horskins, K. and Wilson, J. 1998. The
control of rodent damage in Australian
macadamia orchards by manipulation of
adjacent non-crop habitats. Crop Protection,
17,353-357.
Whisson, D. 1996. The effect of two agricultural
techniques on popu la lions of the canefie Id rat
(Rattus sordidus) in sugarcane crops of north
Queensland. Wildlife Research, 23,589-604.
Wood, B.J. and Liau, 55. 1984a. A long term
study of Rattus tiomanicus populations in an
oil palm plantation in Johore, Malaysia. n.
Recovery from control and economic aspects.
Journal of Applied Ecology, 21,465-472.
Wood, B.}. and Liau, 55. 1984b, A long term
study of Rattus tiomanicus populations in an
oil palm plantation in Johore, Ylalaysia. Ill.
Bionomics and natural regulation. Journal of
Applied Ecology, 21, 473-495.
Zhang, Z.B. 1996. Techniques and strategies for
rodent control by contraception. In: Zuwang
Wang and Zhibin Zhang, ed., Theory and
practice of rodent pest management. Beijing,
Science Press, 367-368.
Zhang, Z.B. and Wan, Y.L. 1997. Ten years'
review on rodent pest management in China.
In: Nou, 0.5, Lou, Z.P., Wan, Y.L. and Tang,
H.Y., ed., Research on agricultural biology
and sustainable development of agriculture.
Beijing, Science Press, 146-151 (in Chinese).
Zou, B., Ning, Z.D., Wang, T.t. and Chang, W.Y.
1998. Ecology and management strategies of
Chinese zokor (Myospa/ax jontanieri). In:
Zhlbin Zhang and Zuwang Wang, ed.,
Ecology and management strategies of
rodent pests in agriculture. Beijing, China
Ocean Press, 41-64 (in Chinese).
29
 Section 1

Basic Research -
the Foundation for Sound Management

Pure and applied science differ mainly in aims, not methods

Chitty 1.996
Charles J. Krebs

Abstract

Rodent population studies have played a key role in developing our understanding of
population dynamics. The proximal stimulus to this understanding is to alleviate
problems of rodent pests in agriculture and disease transmission to humans.

Ideas about rodent population dynamics have gone through three phases. In the
1930s there were almost no quantitative data, and population control was believed
to be caused by biotic agents that operated in a density-dependent manner. By the
1950s a new paradigm of social control of numbers emerged with emphasis on
physiological stress and social aggression within populations. By the 1970s a
synthesis of sorts had emerged suggesting that multiple factors caused population
changes. Experimental manipulation of field populations in the 1960s enlarged our
outlook on the complexities of rodent populations, and the emergence of modelling
and rigorous statistical analyses of survival and reproduction in the 1980s and
1990s has shown again that rodents have been the Drosophila of population
ecology. But as precision has increased over time, generality and simplicity have
declined to near extinction.
What is missing and what do we need to do in the next 20 years? Experimentation
is the key to understanding, and no study should be undertaken without a clear set
of experimental predictions. The era of alpha-level descriptive population studies
should be over. We need large-scale, extensive studies coupled with short-term
experimental studies. Rodents are good candidates for studies of spatial dynamics,
a strongly emerging subdiscipline in ecology. Also, rodent management should focus
on the factors limiting populations and use an experimental approach. The era of
pest eradication via killing alone should be over and we need to be smarter in
developing our management options. The development of genetiC resistance to
anticoagulants and chemical poisons is a call to the ecologists of the 21 st century to
think more clearly about how we might outwit rodent pests. The accumulated
knowledge of the phYSiology, behaviour, and genetics of rodents needs to be
integrated into our management options. There is much to be done both to
understand and to outsmart these clever mammals.

Keywords

Population regulation, population limitation, food, predation, social behaviour,

rodents. pest management

33
Ecologically-based Rodent Management
INTRODUCTION
OPULATION DYNAMICS is
P
without question the most highly
developed of the sub disciplines of
ecology. From abstract mathematical models
to field experiments, ecologists have made
progress over the last 50 years in analysing
population changes in many species. In
particular, rodents have been model
organisms for studies of population
dynamics for three reasons. First, they are
conveniently short-lived so that a scientist or
a postgraduate student can accomplish
something within the constraints of a 3-4
year time window. Second, they are
ubiquitous, occur in abundance nearly
everywhere, and are relatively cheap to
study, and are often of economic importance
(Singleton et al., Chapter 1). Third, they do
interesting things such as have population
outbreaks that occur frequently enough that
even politicians think that something must
be done about them, at least when they are
superabundant. All these features have
combined to produce a very large literature
on rodent population dynamics that is
somewhat overwhelming to the novice. It is
important therefore to step back and ask
what we have accomplished with these
studies, how useful it has been for pest
control, and what is to be done next. This
book brings together ecologists,
physiologists, and ethologists with a
common interest in rodent biology and thus
provides an ideal time to address these
larger issues for rodents.
After a historical overview I will
summarise the three current paradigms of
rodent population dynamics, assess their
34
strengths and weaknesses, and suggest some
paths for future growth.
WHAT ARE THE PROBLEMS?
Ecological questions are complex and one
thing we have learned is to ask very specific
questions about populations so that we can
answer them clearly. Three major questions
have formed the focus of population
dynamics (Krebs 1994, p. 322; Krebs 1995):
... What stops population growth?
... What limits average abundance?
... What constrains geographical
distributions?
To find out what stops population
growth, we must compare a growing
population to one that is not growing, and
the usual approach is to look for some
factors causing negative feedback in the
form of density dependence. The second
question is very broad and is answered by
the use of the comparative approach in
which a high-density population is
compared with a low-density population to
see what factors are associated with the
observed differences in density. In both
these cases an experimental approach is
useful to answering the question most
quickly and avoiding spurious correlations
(Underwood 1997).
Most academic rodent ecologists have
addressed the first question-the problem
of regulation (Berryman 1986; Sinclair 1989),
and this has engendered much discussion
about density dependence in natural
populations. Fewer ecologists have worked
on the second question-limitation of
numbers, and yet this is the critical question
for pest management. In a simple world, the

same ecological factors would limit and
regulate a population, but this has never
been found in the real world. Limitation
often comes from habitat factors that
students of regulation seldom consider, as
we shall see. In a sense these two aspects of
population dynamics correspond to the two
statistical concepts of the mean and the
variance of a set of measurements. We shall
be repeating history to complain, as do many
statisticians, that scientists are often
preoccupied with the mean and tend to
forget about the variance.
The question of what constrains
geographic distributions has fallen out of
favour until fairly recently when the
consequences of global warming on north-
south geographical distribution boundaries
became a hot topic of worry. It is an
important issue that I cannot deal with here,
and there has been much discussion of the
consequences of these biological invasions
(Ehrlich 1989; Ruesink et al. 1995; Vitousek
et al. 1996).
HISTORICAL OVERVIEW
Population dynamics has gone through
three phases during the last 75 years. They
have overlapped little in time but have
phased into one another, with an abundance
of outliers of the 'flat-earth' society type that
bedevils ecology in general.
Phase I
The first phase began with the debate in the
1920s and 1930s about the role of biotic and
abiotic factors in population regulation. The
champions were A.J. Nicholson (1933) for
the biotic school and a variety of opponents
for the abiotic school (e.g. Thompson 1929;
Rodent Population Dynamics
Uvarov 1931). The winners were the
Nicholsonians with their focus on regulation
via density-dependent processes, in which
the main agents were predators, parasites,
diseases, and food shortage. The habitat was
nowhere to be seen, and weather was noise
for population dynamics. Most of this early
discussion was about insect populations,
and rodents were not a part of the
discussions. This was an age of data-free
ecology, and the arguments were typically
theoretical in the bad sense of this word with
no experiments on natural populations
available. I have referred to the Nicholsonian
world-view as the density-dependent
paradigm (Krebs 1995).
It is important to remember that from the
start all ecologists implicitly believed that a
population can be identified, that
community interactions are all direct and
easily definable, and that population
processes are repeatable in space and in
time. These are three gigantic leaps of faith
that came back later to challenge simplistic
models.
Phase 11
The second phase of population dynamics
began in the 1940s when ecologists began to
realise that social processes could affect
births, deaths and movements. Among the
leaders of this phase were David E. Davis
and John Christian in the United States and
Dennis Chitty in England (Christian 1950;
Chitty 1952; Davis 1987). Rodents were the
key to this new phase, which built partly on
the earlier recognition by some
ornithologists that territoriality could
regulate the breeding density of some bird
species. Attention turned in this phase to
studying the physiological and behavioural
35
Ecologically-based Rodent Management

impacts of individuals on one another. One be regulated by the conventional

of the early striking experiments was done
on rats in Baltimore by Davis and Christian
(1956,1958) who showed that one could
reduce the population of rats in a city block
by adding rats to the population (Figure 1), a
completely counterintuitive result for the
1950s. Social strife for breeding space in
rodents became a hot topic, and John
Calhoun suggested crowded mice and rats
as potential models for people in cities
(Calhoun 1949). Much of this early work was
done on house mice and rats in enclosures,
and one of the dominant themes of criticism
was that these enclosures were very high
density, artificial environments and of little
relevance to what went on in natural
populations.
Social regulation of population size arose
as an alternative explanation of population
changes in populations that did not seem to
Nicholsonian predators, parasites, or food
shortages (Chitty 1960). These studies
interfaced well with emerging work in
ethology and behavioural ecology, which
indicated the complex social structure of
many mammal populations, and the interest
population geneticists began to show in the
dynamics of natural populations (Ford 1975).
There was, among many ecologists,
considerable scepticism that social processes,
in contrast to the extrinsic factors of
predators, food supplies and parasites, might
explain changes in numbers. A series of
elegant experiments on bird populations (e.g.
Watson and Moss 1970; Moss and Watson
1980) helped to convince some sceptics, and
parallel work on rodents (e.g. Krebs et a1.
1969; Tamarin and Krebs 1969; Gaines and
Krebs 1971) strongly supported the concept
of social limitation of population density.

130

120

110
<Il
N
'w 100

90

80

70

60
0 5 10 15 20 25 30
Week
Figure 1.
Introduction experiments of Norway rats (Rattus norveg;cus) Into two city blocks in Baltimore in .1954.
Adding rats to a stationary population did not increase numbers but caused them to drop (after Davis and
Christian .1956).

36

Phase III
By 1970 nearly all the ideas about population
regulation and limitation were on the table for
consideration and a synthesis began by
suggesting that everyone might be correct, that
multiple factors could be involved in both
regulation and limitation (Lidicker 1973, 1988).
Two developments accompanied this phase of
population studies. First, experimental testing
of hypotheses in field situations became the
norm in ecology. Second, mathematical
models began to be applied to specific
questions about rodent systems in order to
explore assumptions with rigour (e.g. Stenseth
1978, 1981b). The question then became how to
articulate multiple factor hypotheses within
the paradigm of experimental ecology. All
ecologists are happy to conclude that the
world is a complex multivariate system, but
almost all agree that we must abstract from
this complexity to some order to make
progress.
Many multiple-factor hypotheses suffer
from three deficiencies. Excessive complexity
is the first lethal deficiency. A good example
occurs with many flow chart models of
population processes. Batzli (1992), for
example, lists 22 hypotheses for rodent
population cycles and gives a complex flow
chart to illustrate some of the
interrelationships involved. Limited
predictability is a second problem with
multiple-factor hypotheses. It does us no
service to tell managers that we cannot predict
anything about their potential pest problems
because the world is complex. Third, many
multiple-factor hypotheses are impossible to
test experimentally. Without an experimental
approach rodent ecology will make little
progress.
Rodent Population Dynamics
The solution to these problems is fairly
straightforward. We should encourage
multifactor models of limited complexity,
quantitative predictability, and feasible
experimental tests. Note that there are two
distinct types of multi-factor models of
population limitation.
Several independent factors limit
average abundance
The key point in this alternative is that
the several factors that affect abundance are
independent in a statistical sense. In practice
this means hypothetically that if you change
factor A and double numbers, and change
factor B and triple numbers, you expect that
if you change both factor A and factor B at
the same time you will change numbers by
the simple multiple (2 x 3) or 6 times.
Several interacting factors limit average
abundance
This is the most complex alternative
hypothesis since it postulates a statistical
interaction between some factors. In practice
you would recognise an interactive
explanation by the fact that changing factor
A and factor B at the same time does not
result in their joint effect being predictable.
In the above example, changing factor A and
factor B might change numbers much less
than 6 times, or much more than 6 times. If
this hypothesis applies to your rodent
population, interest centres on exactly how
the ecological interaction of factor A and
factor B operates mechanistically.
A straw poll among rodent ecologists
would probably find most of them
supporting multi-factor hypotheses of
regulation and limitation. If this turns out to
be the most frequent model for rodents, it
37
Ecologically-based Rodent Management
raises the multifactor dilemma that it is
difficult to deal with more than three factors
in any realistic modeL There are two
possible solutions to this dilemma. First, we
can hope that all factors operate
independently (hypothesis 1 above), so that
if we have four or five significant factors for
a particular herbivore, the factors do not
interact. Second, we can hope that for
systems with interactions only two or at
most three factors show interactive effects
(hypothesis 2).
The recent history of rodent population
studies has been a history of reduced
generality, increased precision, and
decreased simplicity. Philosophers would be
appalled at this, but ecologists should be
happy to see us move away from superficial
generality and simplicity. The touchstone of
our progress must be the management of
rodent pests, and we must try to answer this
important question:
how much have our ivory tower studies of
rodents in the laboratory and in the field helped
us to solve problems of rodent pests?
THREE CURRENT PARADIGMS
There are three current paradigms that
represent the dominant focus of work today
on small rodent populations.
The food paradigm
The food paradigm states that both the
quantity and the quality of food supplies
regulate rodent population density. Food
supplies also limit the average density of
populations, and outbreaks of rodents are
caused by changes in their food supplies.
The most important thing you need to know,
under this paradigm, is what do your
38
rodents eat and how much of it is out there
in their habitat. These are themselves
complex issues since diets change seasonally
and may be affected by an individual's sex
and age and also by changes in plant
productivity from year to year and season to
season. A test of the food paradigm is done
most easily by supplementing food supplies
artificially, although these experiments
themselves can be called into question if the
food given is not adequate nutritionally.
The food paradigm cannot be tested as a
unit and needs to be applied to specific cases
to make predictions that can be falsified. For
example, the average abundance of a rodent
pest might be higher where more food is
available. Ecologists often pyramid
hypotheses about food supplies. A recent
example is the hypothesis about Lyme
disease in eastern United States of America
(Ostfeld 1997; Iones et a1. 1998): food
supplies in the form of acorns from oak trees
are postulated to limit the average
abundance of deer mice (Peromyscus
maniculatus), trigger outbreaks of these mice
(when acorn crops are heavy), and regulate
density through starvation. Boutin (1990)
concluded in his review of feeding
experiments that, by adding food to
terrestrial herbivore populations, one could
increase density two to three-fold but not
more so that clearly for some populations
l
food limits density over some restricted
range only. Ecologists tend to despair when
their favourite explanation does not apply to
all species in all situations. We should be
more modest in our aims. Food is clearly one
of the dominant ecological factors limiting
and regulating rodent populations, and the
question is which populations and under
exactly what conditions.

The predator paradigm
Many things eat rodents and some ecologists
look to these trophic links to explain
regulation and limitation of populations.
The predator paradigm states that mortality
caused by predation regulates rodent
populations, that generalist predators limit
the average density of populations below the
limits that might be set by food supplies, and
that outbreaks of rodents are caused by
predator control activities, artificial or
natural. The most important thing you need
to know, under this paradigm, is who eats
whom in your community. Since this can
vary seasonally, and predators are often
selective for sex and age groups, obtaining
this information with quantitative rigour is
not easy.
Paul Errington presented the most
serious challenge to the predator paradigm
more than 50 years ago by suggesting that
predators consumed only the doomed
surplus from rodent populations (Errington
1946). This question has been restated more
recently as the question of whether
predation mortality is additive or
compensatory (e.g. Bartmann et al. 1992).
Errington suggested that it was often
compensatory. This question can be
answered directly by removing predators or
indirectly by showing what fraction of
mortality is due to predation kills. There are
considerable problems with inferring
predation limitation from predator kills
alone. If territoriality causes dispersal
movements, or parasites cause debilitation,
or food shortage causes poor condition,
predators may be the executioners rather
than the primary cause of population
changes (Murray et al. 1997).
Rodent Population Dynamics
The usual argument against predation as
a regulating factor has been that rodents
have such a high rate of reproduction, that it
is impossible for predators to kill enough of
them (e.g. Chitty 1938, 1996; but see
Korpimaki and Norrdahl1998). It is
certainly correct that sufficient numerical
and functional responses must be present for
predation to be a potential regulator of
rodent populations (Hanski and Korpimaki
1995). From a practical viewpoint the key is
to manipulate predator numbers. For
example, to see if they could reduce crop
damage by house mice in Australia, Kay et
al. (1994) provided perches in agricultural
crops for raptors. The important point is not
to be convinced that predators are limiting
or regulating numbers just because
predators kill many rodents. It is convenient
politically to show lots of dead rodents to
our political masters, but scientifically
dubious to infer from these piles of dead
bodies that predators are helping to alleviate
pest problems.
The social paradigm
The social structure of a rodent population
can affect its ecology. The social paradigm
states that social interactions between
individuals can lead to changes in
physiology and behaviour that reduce
births, and increase deaths, and thereby
regulate populations. In particular,
territoriality may limit the average density
of rodent populations. Outbreaks of rodents
are postulated in this paradigm to be caused
by changes in the social environment (e.g.
Krebs et al. 1995). The social paradigm is the
least popular of the three paradigms under
which population ecologists operate. This is
usually because ecologists assume that the
39
Ecologically-based Rodent Management
social environment is primarily determined
by habitat which is highly correlated with
food supplies. Thus, for example, food
supplies determine territory size and
territory size limits population density. The
problem is that other factors may influence
social behaviour as well, and thus the
linkage of habitat to social processes can be
very loose.
Practical problems of rat and mouse
control had highlighted already by the 1940s
that killing of rats and mice often did not
result in control, especially when the pests
were at high density (Chitty 1954, p. 6; Elton
1954). Achieving controls in rat populations
has typically involved intensive large-scale
campaigns of killing rats either directly or by
poisoning (see Singleton et aL Chapter 8).
Only recently has the possibility of using
other methods of control like parasites
(Singleton and McCallum 1990) or
immunocontraception (Caughley et aL 1992;
Chambers et a1. 1997) been able to be
explored.
The social paradigm has highlighted the
role of immigration in local population
dynamics. Removal experiments on rodents
and other small mammals have illustrated
the difficulties of controlling rodents by
increasing mortality. Figure 2 illustrates one
of the first experimental field removal
studies on voles. In spite of very high and
continuous mortality imposed by removals,
the vole population continued to maintain
high density and grow via immigration.
Sullivan and Sullivan (1986) obtained a
similar result for snowshoe hares. After a
series of laboratory and field studies it
became clear to ecologists that pest species
with high turnover (high reproduction, high
mortality, short generation times) are most
40
sensitive to reductions in k"Cundity rather
than increases in mortality rates (Figure 3)
(Stenseth 1981a; Lebreton and Clobert 1991).
The fence effect (Krebs et a1. 1969) is one
example of an experimental result that was
completely unanticipated by the food or the
predator paradigms (Krebs 1996). If fencing
a vole population without altering the food
supply or the predator fauna could produce
a 3-4--fold increase in population density,
what role are immigration and emigration
playing in population regulation? Lidicker
(1962) had raised this question long ago but
few rodent workers have responded to
analyse this phenomenon (Ostfeld 1994).
Unfortunately if you are interested in pest
control you do not wish to find a procedure
that will increase rodent density! My point is
that surprise results that are unexpected
under the conventional wisdom can result
from ignoring social processes in rodent
populations.
I do not wish to argue the merits of the
social paradigm here. The important point
for those interested in pest control is
whether or not it suggests any kinds of
manipulations that could reduce pest
numbers. To date the major contribution of
the social paradigm to rodent pest
management has been to show that dispersal
and social structure can render useless
simple forms of control via mortality
(e.g. Sullivan and Sullivan 1986).
OPTIMAL POPULATION STUDIES
Given these three paradigms, what ought we
to be doing in rodent population studies?
We can start by asking what an ideal world
of population data would look like. It would
have four components.
 Rodent Population Dynamics

700

600

500
cO
.r:
Cii 400
D-

.
Cl)
c
Q)
300
0

200

100

90

C 60
Q)
e
Q)
0...
30

0
Nov. Feb. May Aug . Nov. Feb. May Aug.

1962 1963 1964

Figure 2 .
A removal experiment on the California vole (Microtus californicus). All adult voles were removed every two
weeks from the removal area of 0.8 ha. From November 1962 to July 1963 an average of 62% of the
population was removed every two weeks with little impact on population growth because of immigration
(after Krebs 1966).

1.00
c
0
.~
0.75
ii2
0(1j
o.~
-.r::: 0.50
Figure 3 . 0-
>,3:
_0
Relative sensitivity of the population growth
rate to survival after weaning and to
:~ rn
'Vi 0.25
fecundity for mammal populations. The c
Q)
(f)
shaded area is the zone occupied by many
0.00
rodent pests (modified after Lebreton and
0.0 0.5 1.0 1.5 2.0
Clobert 1991).
Generation time

41
Ecologically-based Rodent Management
Time scale
We would like to have data covering at least
10 of the population events shown by the
species. If we are studying an annual cycle of
rice rats in Indonesia, we would like 10 years
of detailed data to show the kind of variation
we might expect in the system. ,-"'-V1l)l'ol::>l"
like to think that they can completely sample
the range of behaviours of populations in a
few years. We should be more modest.
Spatial scale
We would like data from many populations
spread over the geographic range of the
species. The spatial resolution of these data
would depend on the covariation among
sites in a given neighbourhood. There are so
few data of this type available for small
rodents that it is a necessary part of future
work. In a few cases we have these data-
house mouse outbreaks in Australia (Mutze
1990), Clethrionomys rufocanus on Hokkaido
(Stenseth et a1. 1996). In particular pest
control problems the spatial scale may seem
to be irrelevant, but it is not if we remember
that the local spatial scale can also be critical
(Stenseth 1981a). The concern about
dispersal and population structure has
focused attention on the need to find out
what a local population is and how
extensively we need to manipulate
populations to solve pest problems (Lidicker
1995). I think it is fair to conclude that
virtually all field studies of small rodents to
date have been done on too small a spatial
scale.
Individual scale
We need to understand the mechanisms
behind population changes, and we can
42
obtain this understanding only by having
detailed data on individuals. This point is
too rarely recognised in pest control studies.
The critical data needed on individuals
depends on the mechanisms proposed to
explain the dynamics. If you are concerned
about the role of barn owls as predators
causing population changes, YOLlmust
measure the difference in rodent numbers
between places with and places without
barn owls. If you think infanticide reduces
early juvenile survival, you must obtain data
on the frequency of infanticidal intrusions in
different populations (Wolff and Cicirello
1989).
Community scale
Most rodent studies are single-species
population studies but we should consider
that it may be more fruitful to analyse
interactions between species in the
community as potential influences on
dynamics. We typically think only of
predators but should consider parasites and
diseases as well (Saitoh and Takahashi 1998).
In most pest rodent studies, competition for
resources between species is presumed to be
minimal and single-species interactions are
paramount so that these community
interactions can be ignored. Generalist
predators are perhaps the most common
factor operating on small mammals in which
community interactions, including indirect
effects (Menge 1997), need to be considered.
WHAT DO WE HAVE ALREADY?
Given this ideal world, we should take stock
of what we have already accomplished and
then move on to what we are lacking. Three
strengths stand out.

.. Population ecologists are fortunate in
having a set of good quantitative methods
for dealing with the arithmetic of
population change. From the Leslie matrix
to metapopulation models, there is
quantitative rigour in abundance. The
importance of this is not always
appreciated by population ecologists, yet it
is one of the great intellectual achievements
of this century. We can use this arithmetic
to balance the books. If we know the birth
rates and death rates of a population (as
well as immigration and emigration) we
can compute exactly the rate of population
increase or decrease. We need to use this
more often to check on our estimates of
these parameters (e.g. Haydonet a1.1999).
For many rodent pests, control through
increasing mortality is the only option
available. For these cases quantitative
demographic models can estimate the
mortality required to reduce a population a
specified amount in order to plan an
optimal control program.
.. Second, we have a set of good paradigms
for analysing population regulation and
limitation. I have outlined these above, and
others can be articulated. The importance
of being able to articulate clear, testable
hypotheses is underappreciated in ecology
(Platt 1964; Undenvood 1997). Prediction,
absolutely essential for scientific
respectability, is almost unknown in
population ecology (Peters 1991).
.. Third, we have good field methods for
estimating population parameters to feed
into quantitative models and into statistical
analysis of our experiments. Population
estimation methods have been extensively
improved (Pollock et a1. 1990), elegant
Rodent Population Dynamics
methods for analysing survival rates are
available (Lebreton et a1. 1993), and
statistical methods for analysing
reproductive changes and separating
immigration from births are being
developed (Nichols and Pollock 1990;
Nichols et a1. 1994). We have the
demographic tools to understand rodent
populations with a level of precision that
was not available 25 years ago.
WHAT ARE WE LACKING?
I address here six problems that I think are
central to future studies on rodent
populations. They are not in any particular
order of importance, since some are more
relevant than others to particular situations.
Good methods for spatial dynamics
One of the contributions of the social
paradigm to rodent population dynamics
has been the stress on the importance of
immigration and emigration for
understanding population changes. But we
still lack good methods for studying the
spatial aspects of populations. Radio-
telemetry has made it possible to get some
data on individual movements, but we are
rarely able to do it on a scale that would be
sufficient to get a broad picture of landscape
dynamics. We know too little about how we
should structure our studies of spatial
processes. Should we have many small
trapping grids or a few very large grids?
How large an area should we attempt to
study? What fraction of movements that we
can document are genetically effective (i.e.
the immigrant individual survives, breeds
and leaves offspring rather than dies after
immigrating)? We have much to learn about
43
Ecologically-based Rodent Management
just how to study spatial dynamics
successfully in rodents, yet spatial processes
underlie all of the problems of pest
management. If we can reduce rats on one
rice farm, will the neighbouring farms be
affected or not? Much empirical work needs
to be done on these questions. We can model
pest populations as metapopulations in
space but if we do not know the linkage
parameters for these popuiations, our
models will not be very usefuL
Long-term experiments on limitation
There are no long term experiments on
population limitation in any rodent species.
If we feed a population for two years, we
often get a population increase (Boutin
1990). What happens if we continue this
experiment for 10 years? Is the system in
equilibrium after two years so that we will
learn nothing more from the longer study?
There are numerous examples in ecology of
short-term effects that were not sustained or
even were reversed in the longer term
(Norby et a1. 1992; Wilsey 1996). There are
also many examples from pest control in
which initial encouraging results were
followed by failures (DOT resistance,
anticoagulants). The message is to be
cautious about long-term conclusions.
Good interplay of models and
field studies
Many ecologists have lamented the lack of
interaction between field ecologists and
modellers (e.g. Kareiva 1989). There are
signs that this is finally breaking down
(Stenseth and Saitoh 1998) but I think it is a
failure on both sides that holds back
progress. Models can help us to explore the
44
logical consequences of assumptions we
make in field experiments, and provide a
quantitative estimation of the anticipated
effect sizes. I think it is particularly
important that rodent pest control studies
incorporate both adequate controls and
modelling studies as part of their overall
approach.
Methods for evaluating weather-
driven hypotheses
Climatic change is the wave of the future
and we should be more concerned that our
understanding of rodent systems will be
transient and modified by weather changes.
Hypotheses about weather-driven events
are difficult to test. Post-hoc correlational
studies are useful but inconclusive. They test
more the cleverness of the statistician than
the reality of the biological cause. We need to
state weather hypotheses clearly so they can
be tested next year, not last year, and we
need to abide by the simple rules of
experimental falsification when our
predictions faiL Ad hoc explanations are
available by the shipload for ecological
systems, and we should not get in the habit
of using them to bailout our failures of
understanding. The exact mechanisms by
which weather acts on populations need to
be determined, since we need to know
whether births or deaths are driving the
change.
Economic and environmental
analyses of pest control alternatives
This is not my area of specialty but I would
like to think that we should aim in pest
control work to achieve the best gain for the
least cost-both environmental and

economic. If we cannot achieve this, e.g.
because the lowest economic cost method
produces the highest environmental
damage, we need to state this clearly so that
the public can make an informed decision
about alternatives.
Strategies for analysing the pest
community of crops
In viewing rodent pests as single-species
populations we overlook the broader
strategy of looking at the whole community
of pests of a particular crop. If the pests are
truly independent, we can work on them one
by one. But community interactions have
ways of producing surprises via indirect
effects (Holt 1987; Menge 1995), and we
should be preemptive in looking for these
possibilities.
CONCLUSION
The ivory tower of basic research studies on
rodents has contributed little to the practical
successes of rodent pest management, either
short or long term. Much more insight has
flowed in the opposite direction, and our
understanding of rodent dynamics has been
greatly improved by the practical studies of
rodent pest control. What basic ecology can
contribute to pest management is in the
methods of study needed. The need for clear
hypotheses, rigorous experimental tests
based on good knowledge of natural history,
a sceptical view of existing ideas, and the
need to measure our successes and
failures-all of these features of good
science should be part and parcel of rodent
management.
The major deficiencies of rodent
population studies as we move into the new
Rodent Population Dynamics
millennium are three. We need to apply the
insights of theoretical ecology, behavioural
ecology, physiology, and genetics to rodent
pest problems. A promising start in this
direction is immunocontraception,
(Chambers et al. 1997; Chambers et al.,
Chapter 10). We need studies of tropical
species in varied tropical environments,
since much of our knowledge of rodent
ecology comes from the Temperate Zone (c.f.
Leirs et al. 1996). Finally, we need more
studies of parasites and diseases in field
populations. Conventional wisdom suggests
that they are of little impact on highly fecund
rodents, but their potential for biological
control is largely untested (d. Singleton and
McCallum 1990). There are many
experiments waiting to be done and much
promising modelling ahead with the goal of
understanding population processes in
rodents and at the same time alleviating the
suffering caused by rodent pests around the
world.
REFERENCES
Bartmann, RM., White, G.c. and Carpenter, L.H.
1992. Compensatory mortality in a Colorado
mule deer population. Wildlife Monographs,
121,1-39.
Batzii, G.D. 1992. Dynamics of small mammal
populations: a review. In: McCullough, D.R
and Barrett, RH., ed., Wildlife 2001: popula-
tions. New York, Elsevier Applied Science,
831-850.
Berryman, A.A. 1986. Stabilization or regulation:
what it all means! Decologia, 86,140-143.
Boutin, S. 1990. Food supplementation experi-
ments with terrestrial vertebrates: patterns,
problems, and the future. Canadian Journal
of Zoology, 68, 203-220.
Calhoun, J.B. 1949. A method for self-control of
population growth among mammals living
in the wild. Science, 109,333-335.
45
Ecologically-based Rodent Management
Caughley, G., Pech, Rand Grice, D. 1992. Effect
of fertility control on a population's produc-
tivity. Wildlife Research, 19, 623-627.
Chambers, L., Singleton, G. and Hood, G. 1997.
Immunocontraception as a potential control
method of wild rodent populations. Belgian
Journal of Zoology, 127, 145-156.
Chitty, D. 1938. A laboratory study of pellet
formation in the short-eared owl (Asio
flammeus). Proceedings of the Zoological
Society of London Series A, 108,267-287.
Chitty, D. 1952. Mortality among voles (Microtus
agrestis) at Lake Vyrnwy, Montgomeryshire
in 1936-9. Philosophical Transactions Royal
of London, 236, 505-552.
Chitty, D. 1954. Control of rats and mice, vo!. 1.
Rats. Oxford, Clarendon Press, 305p.
Chitty, D. 1960. Population processes in the vole
and their relevance to general theory.
Canadian Journal of Zoology, 38, 99-113.
Chitty, D. 1996. Do lemmings commit suicide?
Beautiful hypotheses and ugly facts. New
York, Oxford Cniversity Press.
Christian, J.J. 1950. The adreno-pituitary system
and population cycles in mammals. Journal of
Mammalogy, 31, 247-259.
Davis, D.E. 1987. Early behavioral research on
populations. American Zoologist, 27, 825-
837.
Davis, D.E. and Christian, J.J.1956. Changes in
N orwa y ra t popula tions induced by the intro-
duction of rats. Journal of Wildlife Manage-
ment, 20, 378-383.
Davis, D.E. and Christian, J.J. 1958. Population
consequences of a sustained program
for Norway rats. Ecology, 39, 217-222.
Ehrlich, P.R 1989. Attributes of invaders and the
invading processes: vertebrates. In: Drake,
J.A., ed., Biological invasions: a global
perspective. New York, John Wiley and Sons,
315-328.
Elton, C 1954. Research on rodent control by the
Bureau of Animal Population September
1939 to July 1947. In: Chitty, D., ed., Control of
rats and mice, vol. 1. Rats. Oxford, Clarendon
Press, 1-24.
Errington, P.L.1946. Predation and vertebrate
populations. Quarterly Review ofBiology,21,
144-177,221-245.
46
Ford, E.B. 1975. Ecological genetics, 4th edition.
London, Chapman and Hall.
Gaines, MS. and Krebs, CJ. 1971. Genetic
changes in fluctuating vole populations.
Evolution, 25, 702-723.
Hanski, Land Korpimaki, E. 1995. Microtine
rodent dynamics in northern Europe: param-
eterized models for the predator-prey intera-
tion. Ecology, 76, 840-850.
Haydon, D.T., Gillis, E.A. and Krebs, CJ. 1999.
Biases in the estimation of the demographic
parameters of a snowshoe hare population.
Journal of Animal Ecology, 68,501-512.
Holt, RD, 1987. Population dynamics and evolu-
tionary processes: the manifold roles of
habitat selection. Evolutionary Ecology, 1,
331-347.
lones, CC., Ostfeld, RS., Richard, M,p., Schau-
ber,E.M.and Wolff,J.o.1998. Chain reactions
linking acorns to gypsy moth outbreaks and
Lyme disease risk. Science, 279, 1023-1026.
Kareiva, P. 1989. Renewing the dialogue between
theory and experiments in population
ecology. In: Roughgarden, J., May, RM. and
Levin, S.A., ed., Perspectives in ecological
theory. New Jersey, Princeton University
Press, 68-88.
Kay, B.J., Twigg, L.E., Korn, T.]. and Nico!, H.I.
1994. The use of artificial perches to increase
predation on house mice (Mus domesticus) by
raptors. Wildlife Research, 21, 95--106.
Korpimiiki, E. and Norrdahl, K. 1998. Experi-
mental reduction of predators reverses the
crash phase of small-rodent cycles. Ecology,
79,2448-2455.
Krebs, C]. 1966. Demographic changes in fluctu-
ating populations of Microtus californicus.
Ecological Monographs, 36, 239-273,
Krebs, C]. 1994. Ecology: the experimental
analysis of distribution and abundance. New
York, Harper Collins.
Krebs, CJ. 1995. Two paradigms of population
regulation. Wildlife Research, 22, 1-10.
Krebs, CJ. 1996. Population cycles revisited.
Journal of Mammalogy, 77, 8-24.
Krebs, CJ., Chitty, D., Singleton, G.R and
Boonstra, R 1995. Can changes in social
behaviour help to explain house mouse
plagues in Australia? Oikos, 73, 429-434.

Krebs, C]., Keller, B.L. and Tamarin, RH. 1969.
Microtus population biology: demographic
changes in fluctuating populations of M.
ochrogaster and M. pennsylvanicus in southern
Indiana. Ecology, 50, 587-607.
Lebreton, J.D.,Pradel, Rand Clobert, J.1993. The
statistical analysis of survival in animal
populations. Trends in Ecology and Evolu-
tion, 8, 91-94.
Lebreton, J.-D. and Clobert, J. 1991. Bird popula-
tion dynamics, management, and conserva-
tion: the role of mathematical modelling. In:
Perrins, CM., Lebreton, l-D. and Hirons, G.,
ed., Bird population studies: relevance to
conservation and management. Oxford,
Oxford University Press, 105-125.
Leirs, H., Verhagen, R, Verheyen, W.,
Mwanjabe, P. and Mbise, T. 1996. Forecasting
rodent outbreaks in Africa: an ecological basis
for Mastomys control in Tanzania. Journal of
Applied Ecology, 33, 937-943.
Lidicker, W.A.J. 1988. Solving the enigma of
microtine cycles". Journal of Mamma!ogy,
If
69,225-235.
Lidicker, W.Z.J. 1962. Emigration as a possible
mechanism permitting the regulation of
population density below carrying capacity.
American Naturalist, 96, 29-33.
Lidicker, W.Z.J. 1973. Regulation of numbers in
an island population of the California vole, a
problem in community dynamics. Ecological
Monographs, 43, 271-302.
Lidicker, W.Z.J. 1995. The landscape concept:
something old, something new. In: Lidicker,
W.Z.J., ed., Landscape approaches in
mammalian ecology and conservation.
Minneapolis, University of Minnesota Press,
3-19.
Menge, B.A. 1995. Indirect effects in marine
rocky intertidal interaction webs: patterns
and importance. Ecological Monographs, 65,
21-74.
Menge, B.A. 1997. Detection of direct versus
indirect effects: were experiments long
enough? American Naturalist, 149, 801-823.
Moss, Rand Watson, A. 1980. Inherent changes
in the aggressive behaviour of a fluctuating
red grouse Lagopus scoticus popula-
tion. Ardea, 68, 113-119.
Rodent Population Dynamics
Murray, D.L., Cary,].R and Keith, L.B. 1997.
Interacti ve effects of sublethal nematodes and
nutritional status on snowshoe hare vulnera-
bility to predation. Journal of Animal
Ecology, 66,250-264.
Mutze, G.J. 1990. Mouse plagues in South
Australian cereal-growing areas. I. Occur-
rence and distribution of plagues. Australian
Wildlife Research, 16,677-683.
Nichols, rD., Hines,] .E., Pollock, K.H., Hinz, RL.
and Link, W.A. 1994. Estimating breeding
proportions and testing hypotheses about
costs of reprod uction with capture-recapture
data. Ecology, 75, 2052-2065.
Nichols, J.o. and Pollock, KH. 1990. Estimation
of recruibnent from immigration versus in
situ reproduction using Pollock's robust
design. Ecology, 71,21-27.
Nicholson, A.J. 1933. The balance of animal
populations. Journal of Animal Ecology, 2,
132-178.
Norby, RI., Gunderson, CA., Wullschleger,
S.D.,O'Neill,E.G.andMcCracken,M.K.1992.
Productivity and compensatory responses of
yellow-poplar trees in elevated CO2, Nature,
357,322-324.
Ostfeld, RS. 1994. The fence effect reconsidered.
Oikos, 70,340-348.
Ostfeld, RS. 1997. The ecology of Lyme-disease
risk. American Scientist, 85, 338-346.
Peters, RH. 1991. A critique for ecology.
Cambridge, Cambridge University Press.
Platt, J.R 1964. Strong inference. Science, 146,
347-353.
Pollock, KH., Nichols, 1.0., Brownie, C. and
Hines, J.B. 1990. Statistical inference for
capture-recapture experiments. Wildlife
Monographs, 107, 1-97.
Ruesink/ J.L., Parker, I.M., Groom, M.J. and
Kareiva, P.M. 1995. Reducing the risks of
nonindigenous species introductions.
BioScience, 45, 465-477.
Saitoh, T. and Takahashi, K 1998. The role of vole
populations in prevalence of the parasite
(Echinococcus multilocularis) in foxes.
Researches on Population Ecology, 40, 97-
105.
47
Ecologically-based Rodent Management
Sinclair, A.RE. 1989. Population regulation in
animals. In: Cherrett, J.M., ed., Ecological
concepts. Oxford, Blackwell Scientific, 197-
24l.
Singleton, G.R and McCallum, H.!. 1990. The
potential of Capillaria hepatica to control
mouse plagues. Parasitology Today, 6, 190-
193.
Stenseth, N.C 1978. Demographic strategies in
fluctuating populations of small rodents.
Oecologia, 33, 149-172.
Stenseth, N.C 1981a. How to control pest
species: application of models from the
theory of island biogeography in formulating
pest control strategies. Journal of Applied
Ecology, 18, 773-794.
Stenseth, N.C 1981b. On Chitty's theory for
fluctuating populations: the importance of
genetic polymorphism in the generation of
regular density cycles. Journal of Theoretical
Biology, 90, 9-36.
Stenseth, N.C, Bjornstad, O.N. and Saitoh, T.
1996. A gradient from stable to cyclic popula-
tions of Clethrionomys rufocanus in Hokkaido,
Japan. Proceedings of the Royal Society of
London Series B, 263, 1117-1126.
Stenseth, N.C and Saitoh, T. 1998. The popula-
tion ecology of the vole Clethrionomys rufoca-
nus: a preface. Researches on Population
Ecology, 40,1-3.
Sullivan, T.P. and Sullivan, D.S. 1986. Resiliency
of snowshoe hares to population reduction.
Journal of Applied Ecology, 23, 795-807.
48
Tamarin, RH. and Krebs, CJ. 1969. Microtus
population biology. H. Genetic changes at the
transferrin locus in fluctuating populations of
two vole species. Evolution, 23, 183-21l.
Thompson, W.R 1929. On natural control.
Parasitology, 21, 269-28l.
Underwood, A.J. 1997. Experiments in ecology:
their logical design and interpretation using
analysis of variance. Cambridge, Cambridge
University Press.
Uvarov, B.P. 1931. Insects and climate. Transac-
tions of the Entomological Society of London,
79,1-247.
Vitousek, P.M., D' Antonio, CM., Loope, L.L. and
Westbrooks, R 1996. Biological invasions as
global environmental change. American
Scientist, 84, 468-478.
Watson, A. and Moss, R 1970. Dominance,
spacing behaviour and aggression in relation
to population limitation in vertebrates. In:
Watson, A., ed., Animal populations in
relation to their food resources. Oxford,
Blackwell,167-220.
Wilsey, B.J. 1996. Plant responses to elevated
atmospheric CO2 among terrestrial biomes.
Oikos, 76,201-206.
Wolff, J.O. and Cicirello, D.M. 1989. Field
evidence for sexual selection and resource
competition infanticide in white-footed mice.
Animal Behaviour, 38, 637-642.
 3. The Behaviour and Ecology of
Rattus norvegicus: from Opportunism to
Kamikaze Tendencies

David w. Macdonald, Fiona Mathews and Manuel Berdoy

Abstract

While rat population management is clearly possible in the absence of a knowledge

of rat biology, we aim to show in this review how control is likely to prove more
effective if woven into a robust framework of understanding. For example, rats have
flexible population dynamics and can delay the onset of fertility in times of food
shortage . Detailed observations have demonstrated the presence of stable, near-
linear dominance hierarchies, where male social status tends to be age-related. We
discuss how the success of poisoning strategies are crucially dependent on the
foraging decisions which are made against this background of dominance
hierarchies and competition for mates. Feeding patterns also can be altered in order
to avoid predators. Rats are notoriously neophobic, and poor bait uptake is one of
the main reasons that control strategies fail. Our review highlights the importance of
social cues and Toxoplasma gondii infection in the modulation of neophobic
responses. The potential impact of ill-planned rat control operations on the spread of
zoonotic diseases is also considered. Finally, we discuss the development of
behavioural and physiological resistance by rats in the face of continued poisoning
pressure, and the apparent evolution of a new type of resistance which benefits the
rat even in the absence of poison.

Keywords

Rattus norvegicus, rats, behaviour, society, neophobia, resistance , disease

49
Ecologically-based Rodent Management
INTRODUCTION
E
XPLOSIVE DEMOGRAPHY,
adaptable ecology and
opportunistic behaviour are
capacities that cause rats to rank high
amongst those mammals that have most
affected the course of human history. Today
rats exact an immense toll on society
worldwide, whether through the costs of
prophylactic or remedial control, or through
disease transmission and damage to crops
and stored food. Throughout Southeast
Asia, for example, pre-harvest damage
caused by the rice-field rat, Rattus
argentiventer, is reckoned to reduce crop
yields by 17%, a figure which translates into
the squandered rice requirements of in the
order of 20 million people (Singleton 1997).
Such losses may also have indirect
environmental costs; lower yields forcing
larger areas into production and accelerating
the cultivation of wilderness with
consequent threats to biodiversity. More
directly, rats threaten the survival of
endemic fauna on a number of islands, from
the Galapagos to Guam (e.g. Amarasekare
1993; Robertson et al. 1994; Cree et al. 1995);
alien species are, second only to habitat loss,
the greatest contemporary force for
extinction. Finally, the enormity of the threat
posed to humanity by rat-borne disease is
heightened in the context of our huge
populations, rapid transportation and
antibiotic resistance. Amongst the emerging
infectious diseases, the viral haemorrhagic
fevers and Lassa fevers add to the already
lengthy list of blights which can be
transmitted by rodents. The 200,000 cases of
haemorrhagic fever with renal syndrome
50
diagnosed annually in Asia is doubtless just
the tip of an epizootiological iceberg. These
threats are not confined to the developing
world: the 1973 wave of deaths through
hantavirus with pulmonary syndrome
amongst healthy young Americans led to the
discovery of hantavirus in the deer mouse,
Peromyscus maniculatus (Childs et al. 1987).
Hantavirus infection has also been
discovered in Norway rats in rural Britain
(Webster and Macdonald 1995a).
Amongst the diversity of problems
caused by rodents, the Norway rat (Rattus
norvegicus), ranks high amongst the
miscreant species. Originally from Southeast
Asia, the Norway rat's versatility rivals
mankind's and our two species have, in
company, spread around the globe.
Trawling bibliographic indices reveals that
some 24,000 technical publications refer
annually to R. norvegicus (Berdoy and
Macdonald 1991). This stunning total is,
however, neither a fitting tribute to the
fascination of its adaptability nor
recognition of the enormity of its pest status;
rather it stems largely from the utility of its
domestic form as a model for studies
ranging from biochemistry to experimental
psychology. Publications on wild-type
Norway rats largely concern toxicological
studies of candidate poisons, while the
behaviour and ecology of the species in the
wild -which is where it actually does
damage-account for scarcely a handful of
those 24,000 publications annually. Indeed,
as we shall show, while a little is known of
the ecology of wild rats in farmscapes and a
few cities in a smattering of developed
countries, they are perversely unstudied
where they impact the most. Most startling
 The Behaviour and Ecology of Rattus norvegicus
of all, effectively nothing is known of the
biology of the sewer-dwelling rat.
Our objective in this chapter is to review
the behavioural ecology of wild rats, largely
within the context of our own team's
findings. Our contention is that while rat
control is manifestly possible in the absence
of much knowledge of rat biology, it is likely
to be much more effective if woven into a
robust framework of understanding. This
proposition rests on the oft-proven wisdom
of the maxim: 'know thyn enemy'. In
particular, our aim is to reveal that
seemingly disjunct, and perhaps even
rarefied, research topics, such as dispersal,
social status, feeding patterns and disease
transmission are actually inextricably linked
in formulating a biological basis for rat
management.
POPULATIONS, DEMOGRAPHY
AND DISPERSAL
Questions about the population biology of
Norway rats were at the forefront of
mammalian ecology in post-war years,
thanks to the pioneering work by Elton and
Chitty (Chitty 1954) and Davies (1949).
Considering the enormity of the rat's
agronomic and public health impact, it is
remarkable that the momentum of these
early investigations was soon dissipated.
Numerous studies on farmland
(Errington 1935; Aisenstadt 1945; Emlen et
al. 1948; Zap le tal 1964; Hartley and Bishop
1979; Brodie 1981; Huson and Rennison
1981; Homolka 1983) in differing temperate
climates indicate that hedges and fields are
generally a marginal habitat for rats, except
when crops are available as food. Middleton
(1954) noted that all rat infestations in
hedgerows at his study site in Berkshire
(United Kingdom) were either short-lived or
were associated with rat colonies in corn-
ricks or field-barns. He suggested that
scattered field colonies were themselves
ephemeral, but were probably the main
reservoirs from which infestation of farm
buildings occurred in the autumn and
winter.
Colonists tend to be rats that are
approaching, or have recently achieved,
sexual maturity (ZapletaI1964). Telle (1966)
found that most colonists weighed between
160-250 g, while Farhang-Azad and
Southwick (1979) reported a mean weight of
190 g for 26 rats collected from newly
recolonised burrows. Both concluded that
emigrants are mainly young animals, but
neither reported the sex of new colonists.
However, Bishop and Hartley (1976) found
approximately twice as many 'new' adult
males as females entering their hedgerow
population. Similarly, Calhoun (1962)
reported that more males than females were
ejected, or at least departed, from more
socially stable colonies. Kendall (1984) and
Leslie et al. (1952) found that the sex ratio of
rats in environments where they breed
tended to be biased towards females.
Rat population dynamics, and thus the
success of control operations, are intimately
linked with the availability of food supplies.
In the simplest case, bait uptake is most
likely when other sources of food are
unavailable. However, we have found that
food supply, and other environmental
factors, also produce concomitant changes in
behavioural and reproductive ecology (D.W.
Macdonald and M.G.P. Fenn, unpublished
data). These changes in turn may be of
fundamental importance to rat population
51
Ecologically-based Rodent Management

size and to control efforts. We live-trapped changes in breeding activity-pregnant

and radio-tracked rats in three contrasting
United Kingdom farmland habitats, all of
which were surrounded by winter wheat:
(i) a resource-rich agricultural tip; (ii) a
woodland where grain was intermittently
provided for pheasants; and (iii) an adjacent,
resource-poor stream bank. Several salient
findings emerge. First, and unsurprisingly,
more rats occurred at the farm rubbish tip,
where food was most abundant, and fewest
along a stream where it was most scarce
(Figure 1). Second, numbers varied at each
site: in general, there was a cycle in rat
abundance corresponding to seasonal
females were generally captured from
March until October, and peak numbers of
juveniles were found between November
and December (see also Davies and Hall
1951; Farhang-Azad and Southwick 1979,
who report a bimodal pregnancy rate, with
highs in spring and late summer). There is
also evidence that reproduction ceases in
cold winters (Leslie et a1. 1952; Andrews et
a1. 1972; Lattanzio and Chapman 1980).
However, in our study (D.W. Macdonald
and M.G.P. Fenn, unpublished data)
breeding activity was not simply a function
of season but also depended on food

100
tip
.A. woodland

stream

E
Cl
::::J
tll
0
<Jl
T!i

\.
60
'0
ID
.0
E
::::J
Z

20

Jun Dec Jun Dec Jun

1985 1986 1987

Figure 1.
Variation in rat abundance across time in three habitats with different food resources. The figure Illustrates
that more rats were captured in the resource-rich environment of the tip than in the woodland (moderate
food availability) or stream (poor availability of resources). Cyclical fluctuations in abundance were also
more marked in the resource-rich environment (D.W. Macdonald and M.G.P. Fenn, unpublished data).

52
 The Behaviour and Ecology of Rattus norvegicus

availability. In the woodland site, winter
breeding was stimulated by the provision of
grain in January for the pheasants, with peak
numbers of juvenile rats occurring in the
population approximately three months
later, in March and April.
Fertility in both sexes was affected by
season. Using perforation as a measure of
attairunent of sexual maturity, the median
weight of females at perforation at the
rubbish-tip site was higher in the summer
sample than the winter one. Fertility in
category and therefore were wllikely to have
access to bait points positioned outside their
home ranges. Within the home range, rats
regularly retreat to rest sites (Orians and
Pearson 1979; Galef 1988), which in arable
areas tend to be located in hedge bottoms
and banks (Brodie 1981).
360
ro
0 280
U
If)

males was also modulated by season and If)

Q)
UJ
food supply, with evidence for delayed ~ 200
onset of sexual maturity and facultative E
Cl
'm
cessation of spermatogenesis where food 5:
c
ro
120
was scarce. For example, males achieved a '6
Q)
greater weight before their testes became ~

scrotal in winter than was the case in 40

summer. Similarly, those animals living in
the relatively poor environment of the
stream delayed sexual maturity until they
had achieved a greater weight than their
counterparts in more productive habitats
(Figure 2) (D.w. Macdonald and M.G.P.
Fenn, lU1published data). Indeed, it was not
lU1usual for males weighing more than 300 g
to have abdominal testes [the median weight
of rats with scrotal testes reported elsewhere
was 136 g in rural rats (Davies 1949) and 145
g in rats living around Baltimore zoo
(Farhang-Azad and Southwick 1979)].
Clearly, the control of food availability may
have an important role in maintaining rat
populations at manageable levels.
Movements
Farmland rats may either occupy stable
home ranges or travel widely as transients.
Hartley and Bishop (1979) estimated that
three-quarters of rats fell in the former
Tip Wood Stream
Figure 2.
Median weight at which testes became scrotal in
winter and in summer at sites which differed in
resource availability (tip = resource rich, wood =
moderate, stream = poor). The figure illustrates
that males achieved sexual maturity at lower
weights in summer than in winter, and also in
resource-rich environments (p < 0.05 for seasonal
differences, p < 0.001 for resource differences)
(D. W. Macdonald and M.G.P Fenn, unpublished
data).
We radio-tracked rats in and around
farms (Ferm et a1. 1987; Macdonald and Felm
1995). Males ranged widely through the
fields when crop cover was available to
provide protection from predators [mean
linear home range = 678 111, standard
deviation (SD) = 535]. However, their ranges
contracted after harvest (90 m, SD = 28.2).
This effect appeared not to apply to females,

53
Ecologically-based Rodent Management
which generally had smaller home ranges. In
addition to farming activity, the absolute
availability of food also influenced home
range size.
Females at the Wytham tip, where food
was plentiful, had linear ranges averaging
only 85 m whereas those based in the
relatively poor environment of the stream
had a mean home range of 428 m. Similar
results were observed for males (see also
Taylor 1978; Hardy and Taylor 1979).
Clearly, home range is likely to vary greatly
with circumstances.
Social system
The seminal study of rat society was
undertaken by Calhoun (1962) working with
rats in an outdoor enclosure. He concluded
that social pressures were important in the
movement of rats and, specifically, that
subordinate males were forced to migrate
into less favourable sites, leading to an
unbalanced sex ratio (as low as 0.38) in good
habitats. Similar reports of unbalanced sex
ratios in areas suitable for breeding have also
been made by others (e.g. Leslie et al. 1952).
This socially-induced migration away from
good sites may explain the transient
infestation of fields and corn-ricks often
reported during the summer. Social
dominance confers feeding priority (Smith et
al. 1991a), greater reproductive access and
success (Gartner et al. 1981; McClintok et al.
1982; Adams and Boice 1983, 1989) and
reduces susceptibility to adverse conditions
(Barnett 1955, 1958a; Christian and Davies
1964; Boice 1969).
We radio-tracked rats at our three
farmland study sites (D.W. Macdonald and
M.G.P. Fenn, unpublished data), and
various other farm situations (Fenn et al.
54
1987; Fenn and Macdonald 1987). In
summary, where resources were abundant
and clumped, rats formed multi-male, multi-
female groups which we suspect defended a
clan territory. In areas with scattered or
sparse resources, male movements were
consistent with exclusive male ranges with
access to several females. Insofar as any
spatial exclusivity would affect access to
baiting stations, the spatial organisation of
rat populations is clearly important to
poisoning operations. On one farm we found
that almost all radio-tracked rats visited a
particular grain mill at some stage; when
poison was placed there an estimated 95% of
the population was killed (Fenn et al. 1987).
The sex ratio of the populations along the
stream was constant, and equal, throughout
the year. In contrast, in the wood (around the
pheasant hoppers) the sex ratio was
significantly skewed: only approximately
one fifth of rats captured between January
and June were male [X2 = 5.828, degrees of
freedon (dJ.) = 1, P = 0.016 from January-
March; X2 = 11.765, dJ. = 1, P = 0.0006 from
April-June]. We conclude that where groups
developed around reliable food supplies,
resident males excluded transient incomers,
whereas these were numerous in passage
through the poorly resourced stream habitat.
In order to examine further the
consequences of ecological variation on
social structure, we observed several
colonies of wild Norway rats housed in large
outdoor enclosures (c. 266 m 2). In an effort to
bring the field into the laboratory, we
provided rats with problems which they
would encounter in the wild, including
sufficient space, a diverse and multi-
generational social environment and a
dispersed food supply. Nocturnal
 The Behaviour and Ecology of Rattus norvegicus
observations showed that rat colonies were
organised into steady, near-linear
dominance hierarchies where male social
status tended to be age-related (Berdoy et a1.
1995a). Larger rats stood a better chance of
winning contests, particularly when
interacting with unfamiliar individuals,
(Figure 3a). But social status within colonies
became fixed: the results of earlier
encounters appeared to determine the
outcome of future ones, and dominant
individuals thus maintained their social
status long after initial body weight
asymmetries with younger individuals had
disappeared or even been reversed (Figure
3b). In stable groups, therefore, age is often a
better predictor of social status than is
with the alpha male being smaller than
many of its subordinates (see also Calhoun
1962). Such 'settled dominance' (Berdoy et
a1. 1995a) may be found in a variety of other
species (e.g. American bison, Rutberg 1983,
but see Wolff 1988; red deer, Thouless and
Guiness 1986; blue-footed booby chicks,
Drummond and Osorno 1992; see also
Huntingford and Turner 1987, for discussion
of endocrine consequences of fighting). In
rats it may explain why, despite
considerable attention, the evidence that
dominance is correlated to body weight has
been conflicting-e.g. Barnett (1958a),
Boreman and Price (1972), Nott (1988) and
Smith et al. (1994) found a positive
relationship whilst Baenninger (1966, 1970),
Boice (1972), Price et al. (1976) and Sridhara
et al. (1980) did not.
Why do large subordinates seem to
accept the status quo rather than challenging
smaller dominant individuals? We suspect
that the costs of escalated aggression are
great relative to the value of the contested
resource. Whilst adult male rats were
obviously in competition for food and mates,
even the lowest subordinates could achieve
access to feeding sites by adapting their
feeding patterns accordingly (Berdoy and
Macdonald 1991; Berdoy 1994). Moreover,
dominant males could not monopolise
access to oestrous females (Berdoy et a1.
1995b).
In one colony, observations showed that
whilst the alpha male enjoyed the highest
number of copulations, there was no
straightforward correlation between social
status and mating success, with some
subordinates achieving more copulations
than individuals far higher in the dominance
hierarchy. In another colony, the dominant
male actually had less mating success than
his two immediate subordinates.
This state of affairs is likely to be a
consequence of the rat's mating system and
may even be encouraged by the female's
behaviour. Observations in the naturalistic
environment of our enclosure colonies
showed that mating was achieved by a type
of scramble competition. Males pursued the
oestrous female wherever she moved
outside her burrow. Females were typically
in oestrus for one night only, during which
they were assiduously followed by a string
of two to three males (up to seven), and
mated repeatedly with several of them.
Importantly, males had little time to interact
with each other during the pursuit of
oestrous females because they would lose a
mating opportunity, and as a consequence
even the most dominant males could not
monopolise access to the female.
The scramble amongst males to mate
with a female was so intense that it was hard
to detect whether females displayed any
55
Ecologically-based Rodent Management

Figure 3. (a)
The effect of body weight
asymmetry on social
40
Heavier rat won
Heavier rat lost
dominance in 132 pairs of ns
adult male rats. There was a
~
Ul
30
* **
significantly increasing ns
<ll
effect of weight on the (ij
E
chance of winning contests -'='
:J
as the weight/age "0
rn
asymmetries between '0 20
~
contestants became larger ca
=
(G 20.17; d.f. 4; = '0
0-

=
p 0.0003). Asterisks show ID
.0
the statistical significance E
:J 10
of weight asymmetry on the z
likelihood of winning a
contest within each weight
asymmetry class (ns non =
=
significant, * p < 0.05, 0
** =p< 0.01, *** = 1-5 6-10 11-15 16-23 24-37
P < 0.001, using binomial
one-tailed test). Body asymmetry (%)

Figure 3.(b)
The effect of age asymmetry
on social dominance in 132
40
Older rat won
Older rat lost

pairs of adult male rats. There

was a significantly increasing ~
Ul * * *** *** ***
30
effect of age on the chance of <ll
(ij
winning contests as the E
-'='
weight/age asymmetries :J
"0
between contestants rn
'0 20
=
became larger (G 12.48; ~
=
d.t. 4; p= 0.014). Asterisks ca
0-

show the statistical sign if- '0

icance of weight asymmetry ID
.0
on the likelihood of winning a E
:J
10
contest within each weight
z
asymmetry class (ns = non
=
significant, * p < 0.05,
** =p< 0.01, *** =p< 0
0.001, using binomial 1-5 6-10 11-15 16-23 24-37
two-tailed test).
Body asymmetry (%)

56
 The Behaviour and Ecology of Rattus norvegicus
mate selection or whether they mated
promiscuously from choice.
To test these possibilities we devised an
apparatus which exploited the sexual
dimorphism in size amongst rats; male rats
were housed in cubicles around a central
arena in which a female was placed. A
circular passage permitted the female access
to each male's cubicle, but was too narrow
for the male to pass (M. Berdoy et al.,
unpublished data; C. Rolland et al.,
unpublished data). Observations in thls
'inverta-brothel' enabled us to see that
females clearly exercised mate choice.
Liberated from the constraints of male-male
competitive behaviour, females formed
enduring bonds with a single male, but
nevertheless selected to mate promiscuously
(although to a lesser extent) with a small
number of the males available to them.
Social odours
There is a huge literature on olfactory
communication amongst rats (e.g. see
reviews in Brown and Macdonald 1985).
Following observations of the males'
assiduous sniffing of females in our
enclosures, and their pursuit of those in
oestrus, we began to question the significance
of haunch odour. The scent of rat haunches
had at least 22 volatile components, the
proportions of which varied greatly between
individuals. Although no single compound
emerged as diagnostic, principal component
analysis revealed exclusive categories in the
odour profiles of oestrous females and
dioestrous females, with the first axis
explaining 80% of the variation. Haunch
odours could, therefore, be separated along
biologically meaningful lines, suggesting that
they had the potential to signal useful
information to rats (Natynczuk and
Macdonald 1994a,b).
In order to receive an olfactory signal,
rats must smell the encoding odour. We
have investigated 'sniffing behaviour' in a
group of 62 adult rats (Natynczuk and
Macdonald 1994a,b). Male rats sniffed
frequently at the various body zones of
females but there were significant
differences in the proportions of sniffs
directed to the various body zones
(F = 38.21, 32, 28 d.f., P = 0.005). The highest
proportion of sniffs (23%) were to the
females' haunches, and this behaviour was
seen whether or not the female rat was
related or in oestrus. However, the genital
region of non-related oestrous females was
investigated significantly more frequently
(29% of all sniffs) than the genital region of
any other group (X2 = 4.642, 1 d.f., P = 0.031
for difference in proportions of sniffs for
non-related oestrous and non-oestrous
females). Males tended to sniff the haunch
after sniffing the forequarters, as part of a
sequence of sniffing along the female's body
from forequarters to hindquarters. Histology
of the skin sebaceous glands indicated that
the secretory activity of glands in the
haunch, but not those in the forequarters,
changed during oestrus (Natynczuk and
Macdonald 1994a,b; Natynczuk et al. 1995).
A male rat will therefore smell a
discontinuity or gradient in scent along the
body of an oestrous female but not a
dioestrous female. Rats can thus judge a
female's reproductive status by calibrating
the odour of her haunch against that of her
forequarters.
The self-calibration model has four
advantages: (a) it obviates the need for
learning or inheriting responses to a
57
Ecologically-based Rodent Management
pharmacopoeia of scents; (b) it minimises
the difficulty posed to the recipient by
variation, in quality and quantity, between
the odours of different signallers; (c) social
odours are inevitably mixed with variable
and copious background smells, and even
one individual's scent may vary over only a
few days; and (d) it could work not only
with complex mixes of chemicals, but also
with single compounds. Self-calibration
relies on assessment of the degree of
difference between odours for the
transmission of a signal and therefore might
apply to a variety of phenomena detailed in
the literature (Natynczuk et al. 1995). It is
currently unclear whether a strategy
employing pheromones could be used to
disrupt rats mating systems and hence
control population size.
FEEDING BEHAVIOUR
Foraging decisions must be carried out
against a background of dominance
hierarchies and competition for mates. A
foraging rat must make decisions about
when to feed, how to structure foraging
bouts within its period of activity, and what
to eat. Clearly an understanding of foraging
behaviour is crucial to the development of
more effective baiting systems: neophobia,
food preferences and conditioned aversions
associated with illness may all limit bait
uptakes. As Quy et a1. (1992a,b) and Cowan
et a1. (1994) point out, rat control operations
often fail because of poor bait uptake.
Feeding decisions
Most wild rat populations are essentially
nocturnal (Calhoun 1962; Taylor et a1. 1991;
Berdoy 1994) and therefore face two related
58
problems. First, there is significant seasonal
variation in the period for which they must
fast during daylight hours. Second, and
correspondingly, they must satisfy their
energy requirements during nights of
variable duration. The shorter the night, the
more likely there is to be a conflict of interest
between feeding and mating, which is also at
its peak during the shortest summer nights.
We attempted to determine how these
conflicts are resolved.
Long-term monitoring of feeding activity
in our colony of wild rats, using a purpose-
built, continuously recording telemetric
device (Berdoy and Evans 1990) showed that
the adjustment in feeding intensity to
maintain nocturnality (which rose by 65%
between December and June as nights grew
shorter) was not uniform (Berdoy 1994). Rats
mainly compensated for the seasonal
reduction in feeding time by increasing
feeding activity during the last quarter of the
night (Figure 4). As a result, the distribution
of feeding activity through the night
gradually changed from being roughly
constant in winter (and comparable to that
obtained in the laboratory) to being clearly
skewed in summer, with a sharp peak of
activity before sunrise. Feeding during the
earlier part of the night remained at the same
intensity as in winter. Since food availability
was kept constant throughout the year why
did the timing of feeding activity change so
drastically between winter and summer?
First, during the first half of the night, the
rats were generally busy exploring the
enclosure and engaging in social activities.
In males, the disruption of feeding was
related to the intensity of mating effort
which could result in little or no feeding at
the beginning of the night. Such disturbance
 The Behaviour and Ecology of Rattus norveg;cus

Sunset Sunrise
16

14

12
?:-
Ui 10
c
QJ
C
Ol
8
c
'6
QJ
QJ 6
LL

Time of day (quarters)

Figure 4.
Nonuniform increase in feeding activity in response to shorter summer nights. To control for the 50%
reduction in night length between summer and winter, the data for day and night periods have been divided
into quarters (Oay =01-04; Night =N1-N4) . The significantly greater increase in feeding intensity during
the last quarter of the night ( N4, thick arrow) causes an increasingly skewed feeding distribution during
short summer nights. Note also the greater proportion of dawn feeding (01) during summer.

was grea test during the summe r when
oestrous females were mos t n umerous.
Whereas the feeding activity of an oestrous
female was principally disrup ted for that
night only, the presence of an oestrous
female affec ted the feeding activity of each
courting male in th.e colony.
Second, ra ts may have foraged mainly
during the last portion of the night to time
their feeding activity in anti cipa tion of
subsequent energy needs (Le Magnen 1985).
Preswl1ably rats needed to gather reserves
that allowed them to last until the following
night, particularly during periods of long
daylight hours. Analyses showed that 90%
of the monthly changes in the timing of
activity in females, which were less
disrup ted by social liv ing than were males,
could be explained in terms of seasonal
varia tion in night length.
Predator-altered behaviour
Whilst radio-tracking wild rats on farms
around Oxfordshire, we identified an.
intriguing pop ulation of diurnal rats (FetUl
and Macdonald 1995). The ra ts occupied a
rnidden on one of five study fa rms.
Coincidentally, a particularly large number
of signs of the red fox (Vulpes vulpes) were
also discovered in the vicinity of the midden,
but were more or less absent from the other
four sites, at which rats were typically
nocturnal. We had previously fOLmd that

59
Ecologically-based Rodent Management
wild-caught rats kept in enclosures modified
their foraging behaviour in order to avoid
areas scented with fox urine, whereas their
activities were unaffected by rabbit urine
(Berdoy and Macdonald 1991; see Vernet-
Maury et a1. 1968, 1984 for studies on
odorants inducing stress in laboratory
rats-some found in fox faeces). We
therefore hypothesised that the rats at the
midden were diurnally active in order to
avoid fox predation. Establishing a direct
effect of predation, however, required
careful study to eliminate two other equally
plausible explanations. First, the activity
observed could have been that of low-
ranking animals, forced to be diurnal to
escape competition with dominants (a
phenomenon observed in our enclosures -
Berdoy 1994). Second, diurnal behaviour
might have been facilitated by the lack of
human disturbance at the site (see Taylor
1975).
Infra-red photo-electric cells revealed
that the rats were active only when the foxes
were absent. Seasonal use of the mid den by
foxes meant that the risk of predation to rats
was greatest in summer, and corres-
pondingly, the rats were most strictly
diurnal in summer. The next step was to go
beyond the correlational nature of the
existing data. A fox-free enclosure was built
near the midden and stocked with rats from
the diurnal population. Rats in the enclosure
should have been able to detect the absence
of foxes though the lack of odour from fox
urine and faeces. If rats were diurnally active
to avoid fox predation, then in the absence of
foxes we expected them to revert to
nocturnal behaviour. This was exactly what
happened (Fenn and Macdonald 1995). In
contrast, disturbance alone, such as human
60
and traffic activity at night, does not appear
to interrupt the foraging activities of urban
rats (Takahashi and Lore 1980).
Interestingly, foxes visited the midden
primarily to scavenge on farm waste rather
than to feed on rats. Thus the presence of one
prey type (scavenge) increased the mortality
risk to another prey type (rats).
Meal patterns
Rats feed in bouts, so the adjustment of food
intake to calorific expenditure must be
achieved through variation in the size and
the frequency of meals. For example, females
forage in many short visits, whereas males
use fewer, longer ones (Inglis et a1. 1996). We
therefore examined how the structure of
feeding itself is influenced by variable night
length and social pressures (8erdoy 1991,
1994). The feeding of rats in the
laboratory tend to show a positive
relationship between the size of a meal and
the following inter-meal interval (referred to
as a 'post-prandial correlation'), but not
between the size of the meal and the interval
separating it from the previous one (a 'pre-
prandial correlation') (Figure 5) (see Le
Magnen 1985 for a review).
Colloquially, this means that rats decide
on the size of a meal on the basis of how
hungry they expect to be, rather than how
hungry they are (Snowdon and Wampler
1974; Le Magnen 1985). A post-prandial
correlation in an individual's feeding pattern
is generally thought to indicate that the
individual has good control of the onset and
termination of its meals. On the other hand,
an animal that is less able to control the onset
of feeding than its termination is more likely
to exhibit pre-prandial correlations. These
principles have been useful in elucidating the
 The Behaviour and Ecology of Rattus norvegicus
physiology of feeding in the laboratory, but
they also provided a tool for our investiga tion
of the complex feeding patterns of rats wIder
natural circumstances. Following Slater
(1981), we tested the hypothesis tha t in
p redictable env ironments (such as in our
enclosure) individuals would be m ore likely
to exhibit p ost-prandial than pre-p randial
correlations. Subordinates, on the other hand,
which m ay be less likely to control the onset
of feeding due to disturbances by d Olll.i nant
individuals, w ould effectively be Living in an
lUlcertain environment and should be more
likely to exhibit pre-prandial correlations
than would d ominants. If so, subordina te and
d ominant rats might respond differently to
baiting programs.
I
Meal Pause Post-prandial
co rrelation
Pause Pre-prandial
correlation
time
Figure 5 .
Types of feeding pattern wh ich reflect a food
intake regulated from meal to meal. The upper
part of the figure illustrates a positive correlation
between the size of meal and the foll owing pause.
The lower part of the figure shows a positive
correlation bet ween size of meal and the
preceding pause.
We analysed the feed ing pa tterns of 12
adult ra ts (Berd oy 1991, 1993, 1994; Brunton
1995). First, the intervals between meals in the
colony of ra ts a ppeared, on the whole, to be
dependant on the size of the preceding meal
(post-prandial correlation). Thus, knowledge
of the length of time for which a ra t h ad n ot
ea ten was, counter-inhLitively, a p oor
predictor of the size of the next meal. Meal
size appeared instead to be a reflection of the
rat's anticipated subsequent en ergy need .
Second, the consequences of agonistic and
mating behaviour w ere also reflected in the
values of these relationships. As predicted,
do minant males, 'who benefi ted from a
grea ter freed om of access to the feeding sites,
and females (who were less disrupted by
agonistic behaviour an d matings than males),
tended to regula te their feed in g m ore closely
than their subordinates or male cOlmterparts
(see also Sla ter 1981 ). Moreover, p ost-
p randial correlations were m ore apparent in
w inter than in Slim mer, supp orting the idea
th a t they were less able to regulate theiT mea ls
when matin g was a t its peak an d activity was
skewed tow a rds a pre-d awn peak.
The extent to which d ie tar y d ecisions
d epend upon th e fine structure of feeding,
and the consequences of in divid ual
varia hon - le t alone seasonal varia tion - on
d iet selection a re rarely cons id ered in the
literature on rat nutrition. Yet th ese
d ifferences not only cons tihlte the basis on
which na tural selection opera tes, but th ey
may also relate to the w ay in which dieta ry
d ecisions are made (Tempel et a!. 1985;
Hayne et al. 1986). They therefore affect the
ca pacity for d iet selection in gen eral, and for
avers ion lear ning in particular.
Neophobia
Thom pson's (1948) ea rly work clearly
d em onstra ted neophobia in wild ra ts in
response to both new food and to lighting
used to observe his expe rim ental site (see
also Barnett 1958a; Cm,van 1977; Miller and
61
Ecologically-based Rodent Management
Holzmann 1981; Beck et al. 1988 and review
in Royet 1983). To overcome neophobia, pre-
baiting is now an integral part of most rat-
poisoning operations. The function of
neophobia remains unproven, and it may
have more to do with a general timidity than
with food selection. However, while
neophobia is likely to incur the cost of
avoiding potentially harmless foods, it may
facilitate the rat's ability to associate eating a
novel food with adverse effects sometime
later (Rozin and Kalat 1972; Nachman and
Hartley 1975; Rescorla 1980: ability to
associate illness with toxic food). Given the
wide range of food qualities which a rat may
encounter, and the intensity of poisoning
pressure to which it is subjected, neophobia
is generally assumed to be of adaptive
significance (Rozin 1976). Indeed, neophobia
was found to be absent amongst rats of the
Hawea and Breaksea Islands, New Zealand
(Taylor and Thomas 1993) which were
historically uninhabited. In general, non-
commensal rodent species also only show
weak neophobia (Cow an 1977; Brammer et
al. 1988). The transition from feeding from
familiar to novel food after a period of
neophobia is usually a gradual process
(Chitty 1954; Barnett 1975). Wild rats may
initially only sample a small amount of the
food with amounts taken gradually
increasing (Thompson 1948; Shepard and
Inglis 1987; but see Beck et al. 1988).
It is difficult to quantify neophobia in rats
in the field, because of the large number of
potential confounding variables. We
therefore designed a series of experiments to
measure neophobic responses in wild-caught
rats. The first of these housed rats in large
outdoor colonies (Berdoy 1994). The rats
avoided familiar foods (on which they had
62
been fed ad libitum for two years) in a
familiar feeding box when the grain was
scented with natural oils. This reaction was
still significant after 40 days. EVt-'!l more
extreme neophobia was noted when a novel
food was presented. Inglis et al. (1996)
undertook a similar study, but his rats were
housed in concrete arenas and did not have
lower intakes of novel foods, or of unfamiliar
foods (foods which had previously been
familiar, but had not been offered to the rats
since they were caught). However, the rats
did make more visits to the new foods,
consuming less per trip. In contrast, the
introduction of a novel food bowl, even
though it contained familiar food, elicited
strong neophobic responses, reducing food
intake to 5% of its previous level. This
neophobia, while diminished, was still
evident after five days (Cowan 1976 reports
similar results for the black rat, Rattus rattus,
see also Galef and Heiber 1976; Galef 1988).
We investigated whether rats of different
provenance displayed different levels of
neophobia. In particular, we compared the
behaviour of rats which had recently
survived intensive poisoning on a Welsh
farm with those which had been in the
protected milieu of our enclosure for two
years (Brunton and Macdonald 1996). The
enclosure rats were descended from the
survivors of a poisoning treatment in
Hampshire two years previously.
Hampshire rats had proven notoriously
difficult to poison (Richards 1981; Greaves et
al. 1982a,b; Quy et al. 1992b), prompting the
hypothesis that they might be unusually
neophobic. We designed three trials to test
this. First we tested the effect of an
unfamiliar odour on a familiar object (this
involved handling the food container
 The Behaviour and Ecology of Rattus norvegicus

without gloves). Next, we tested the effect of 35

Enclosure rats
replacing a familiar food container with a
Wild farm rats
novel container while food and odour 30
remained constant. Finally, we tested the
25
effect of placing a novel food in a familiar (j)
0.
container. In each case a familiar bowl, E
co 20
Ul
containing fa miliar food, was present .r:
u
co
throughout. The latency to feed, when (j) 15
0
confronted with each form of novelty, was ~
0
10
allocated an ordinal score to indicate the
d egree of neophobia. The swnmed results 5
from the three experiments were then used
to create a 'neophobia index'. A score of 0 0
0 2 3 4 5 6
indicated no neophobia and 6 indicated Neophobia index
ex trem e neophobia .
The surv ivors fro m the readily controlled Figure 6.
Percentage of wild farm rats (from populations
We lsh popula tions we re significantly more
considered normal to control; n = 30) and
neop hobic than enclosure rats descended
enclosure rats (descended from Hampshire rats;
from the problematic Hampsh.ire n = 30) at each index of neophobia, where a score
population. Nonetheless, within both the of 0 indicated no neophobia and 6 extreme
Ha m pshire and the Welsh groups were neophobia. Overall, the farm rats were
some ind ivid uals tha t showed no neophobia significantly more neophobic (M ann Whitney U

(8% and 10%, respectively), and others that
showed ex treme neophobia (an index of 6:
4% and 10%, respectively) (Figure 6).
Perh.aps the Welsh farm rats were a
sample of the m os t neophobic individuals
from the origina l p opula tion, the less
neophobic rats having been selected out by
poisoning. The ra ts born in the enclosure,
a lthough d escended from the survivors of a
poison treatment, had not themselves been
subject to selection by poisoning and we
h ypothesise that they contained the full
range of neophobic phenotypes. Thus, the
differences in neophobia between the two
groups is likely to be due to the presence of a
higher proportion of s trongly neophobic rats
in the wild-caught population (Richter 1953;
Barnett 1958b; MitcheU 1976; Cowan 1977).
The failure of poison trea tments in
t est Z = - 3 .0; p = 0.009).
Hampshire is therefore not explicable by a
fund amental difference in the inh erited level
of neophobia. In fact, like Quy et al. (1992a),
we concluded that the difficu lty in
controlling the Hampshire ra ts was m ost
likely due to the stable and ab wld ant source
of food. Various studi.es sugges t tha t the
stronges t neophobic reaction in wild rats is
to new objects in a fan"liliar en vironment
(Shorten 1954; Covvan and Barnett 1975;
CmNan 1976, 1977; WaUace and Barnett
1990); evid ence is acclllmlating that
neophobia is related to the stability of the
environment (e. g. little neophobia in land-
fill site rats-Boice and Boice 1968) although
unfamiliar areas with new objects are rea dily
explored (Barnett and Spencer 1951; Cowan
and Barnett 1975; Cowan 1977).

63
Ecologically-based Rodent Management
The interplay of conditioned aversion
and innate neophobia is extremely complex.
Indeed, neophobia may actually assist in the
development of conditioned aversion, by
making it more likely that a rat will
remember eating a novel food, and thereby
associate it with any adverse consequences.
In the experiments with rats from Wales and
Hampshire, the temporal patterns of feeding
during the night were related to the rat's
degree of neophobia (Berdoy 1994; Brunton
1995). Individuals that displayed no
neophobia were more inclined than most to
restrict their feeding to the end of the night.
These associations are still largely
mysterious. We can conclude, however, that
subordinates may be constrained to eating
shortly before dawn (Berdoy and
Macdonald 1991) and may be more likely to
try new foods than dominant individuals
(Nott 1988). There is also evidence that the
strength of taste aversions may be
modulated by circadian rhythms (e.g.
Infurna et a1. 1979).
Social factors in food avoidance
Field studies have shown that rat colonies
living in the same area can show substantial
differences in food preferences not
explicable by availability (Gandolfi and
Parisi 1973). Subsequent laboratory studies
have demonstrated that information about
both the toxicity (the 'poisoned partner
effect'-Bond 1984) and palatability of food
can be transferred from mother to offspring,
including via the mother's milk (Galef and
Clark 1972; Hepper 1990). Diet preference
can also be socially induced in adult
laboratory rats (Galef and Wigmore 1983;
Posadas-Andrews and Roper 1983). Thus,
rats faced with novel foods are influenced by
64
the olfactory information emanating from .
another rat which has eaten at the novel food
types (the 'demonstrator effect' -Ludvigson
et a1. 1985; Galef 1988, 1994). However, the
response of the rats receiving these cues
differs according to the state of the
'demonstrator', with a dead animal
producing no effects on food selection. The
position of the cues is also important:
residues on the anterior of the rat, but not the
posterior, will elicit a response in the
'observer'. These results make functional
sense since the contextual clues which
induce preference are those which
corroborate the circumstantial evidence of
what the' demonstrators' have eaten.
We investigated the demonstrator effect
in naturalistic circumstances by
manipulating the olfactory information
emanating from some individuals in one of
two enclosure colonies and evaluating the
consequences on the diet selection of other
wild colony members (Berdoy 1994). In
order to produce individuals that exhibited
the characteristics of having eaten
cinnamon-scented food, about a dozen
drops of cinnamon oil in suspension in a
sugary solution were deposited on the neck,
shoulders and near the mouth of 11 adult
rats (informants) in one colony but not in the
other. In both colonies, the rats were
presented with two novel food types:
peppermint and cinnamon-scented wheat
grain, presented in two pairs of familiar
feeding boxes. Each pair of boxes thus
provided the rats with a choice between the
two food types (which previous experiments
had shown to be equally palatable to rats).
To eliminate the possibility that food
selection might be caused by neophobia to a
new site rather than to the new food, the
 The Behaviour and Ecology of Rattus norvegicus

scented grain was placed in four familiar
feeding boxes, previously containing non-
scented wheat grain. In addition, rats could
feed on non-scented grain in four other
familiar feedin g boxes distributed
thro ughout the enclosure. The tin1ing of
visits to all feeding sites, as well as the
iden tity of selected individua ls within th e
colony, was recorded contin uously using an
au tomatic recording system (Berdoy an d
Evans 1990).
Before the trea tmen t of the informants,
rats in both colonies were highly and equally
neophobic to both types of scented gra in
(Figure 7).

100

Experimental colony
c Control colony
Ql 80
ro
Ql
c
.~
Cl
"0
60
Ql
C
Ql
U
(/)

c 40
0
E
co
c
c
0
~
0
20

0
'I en
>.
C\I
>.
.8 co
en 0 en
>.
I'- en
>.
C\I
C') en
>.
co
.8 "0 .8 "0co co
.8 "0 .8 "0co
"0
co C') co
I !:::-
C\I
!:::- ~ C\I !:!2.
(/)
>. ~ (/)
(/) co >. (/) (/)
>. 0 co >. >.
co 0 co co
0 0 0

Before experiment After start of experimental feeding

Figure 7.
Change in preference for feeding sites in two colonies of wild rats, as expressed by
the proportion of cinnamon-scented grain eaten. In the control colony, the
introduct ion of cinnamon and peppermint-scented grain was associated with no
change in relative use of feeding sites. However, in the experimental colony which
had 'informants ' (cinnamon-scented animals) introduced t o the colony before the
introduction of the scented grain, there was a preference for the cinnamon over
peppermint flavoured feeds. This remained statistically significant for t wo weeks
(p= 0.004, permutation test).

65
Ecologically-based Rodent Management
However, following the tainting of the
informants in one colony, their companions
showed a significant preference for cinnamon
over peppermint, whereas no such effect was
observed in the control enclosure. The
informants themselves were equally reluctant
to eat both types of grain at first.
Subsequently, however, the informants too
began to show a preference for cinnamon,
perhaps as they in turn were influenced by
the odours on the rats they had beguiled into
eating cinnamon. It is noteworthy that,
although the informants reduced neophobia
towards cinnamon, nonetheless both scented
grains were largely eschewed by the colonies
which showed a neophobic response that was
significant for at least 40 days. Indeed, some
individuals avoided the scented grains
completely. Furthermore, the switch to eating
novel foods did not always occur gradually.
For example, one dominant male, after
avoiding the novel foods for the first five
nights, ate the cinnamon-scented grain
almost exclusively (90%) on the sixth night.
This new diet selection was associated with a
night of intense mating effort from the male,
with no feeding during the first half of the
night, and a characteristic pre-dawn peak of
feeding activity.
These results draw attention to a number
of other parameters which are likely to
influence diet selection in the wild:
neophobia, the level of information, site
preference, competition and the timing of
feeding (Berdoy 1994). Thus while much
remains to be discovered, our experiments
point to the costs and benefits of sociality in
decision making. And for a mammal under
such heavy poisoning pressure as the rat,
these decisions are of immediate practical
relevance.
66
RAT-BORNE DISEASE
Parasite-altered behaviour
The problems of controlling rats, many of
which revolve around neophobia, might
seem largely separate from their role as
disease vectors. However, we have revealed
an intimate link between the two, at least
with regard to infection wi th the protozoan
Toxoplasma gondii (Webster 1994a; Webster et
al. 1994, 1995; Berdoy et al. 1995b,c; Webster
and Berdoy 1997). Infection with Toxoplasma
alters rat behaviour to increase their
susceptibility to predation by domestic cats,
the parasite's definitive host (Hutchison et
al. 1969). These behavioural changes also
incidentally increase the likelihood of
poisoning. Toxoplasma infection of rats is
widespread. In a study of English farms, we
found a mean prevalence of 35% among
rural rats (Webster 1994a). Earlier studies
had, by contrast, indicated low prevalence
levels of between 0-10% (Lainson 1957;
Jackson et al. 1986). The disease is of medical
and economic importance: human
toxoplasmosis in the United States of
America reputedly accounts for more
congenital abnormalities than rubella,
syphilis and herpes combined (Schmidt and
Roberts 1989). In sheep, the annual loss of
lambs due to the infection is an estimated
100,000 in England alone (Berverley 1976).
Yet despite the importance of rats in the diet
of cats, and hence the subsequent
opportunities for direct and indirect
transmission of Toxoplasma to humans, wild
rats have until recently been dismissed as
unimportant to the dissemination of this
disease. The discovery by Webster (1994a)
that the infection can be perpetuated even in
the absence of cats (the final hosts) (d.
 The Behaviour and Ecology of Rattus norvegicus
Wallace 1981) by congenital transmission,
means that rats represent an important
wildlife intermediate-host reservoir for
toxoplasmosis.
In evolutionary terms, a parasite's goal is
to increase the chances of infective s tages
meeting their host species. One ada ptive
route to this goal is host-behaviour
manipulation. An obvious way of facili tating
Toxoplasma transmission from an inJected rat
to a cat is to enhance the rat's activity levels:
cats are attracted to mov ing objects and
show little interest in stationary ones (Hubel
and Weisel1962; Leyhausen 1979). A study
of the activities of 140 rats revealed tha t
those infected with Toxoplasma were
signi ficantly m ore active than uninfected
rats (consis te nt with findings in lab rnice-
Hutchison e t a1. 1980; H ay et a1. 1983a,b).
This was true both fo r ra ts that had acquired
Toxop lasma as adults and for those which
had acq uired it congenitally. In contrast, rats
harbouring parasites w ith direct life cycles,
such as Cryptosporidium parvurn, d id n ot
exhibit any altered activity (Webster 1994b;
Webster e t a1. 1994). To xoplas ma-infec ted
wild rats were also generally less neophobic
of novel foods (see also Stretch et al. 1960a,b)
(Figure 8) and, a t farm steads wh ere th e
majority of the rat p opulation could be
sampled, were trapped more quickly than
uninfected rats. Furthermore, they we re
more curious: when one of us s tood in the
enclosure it turned out th at rats infected
with Toxop lasma approached m ore closely
than non-infected animals (Berdoy et al.
1995b). We fur ther investigated whether
Toxoplasma-inJected and non-infected rats
differed in their reac tion to potential
predation by cats. It is widely recognised
that rats, including n aive laboratory animals,
show an innate avoidance of ca t odom
(Vernet-Maury et a1. 1984; Blanchard e t a1.
1990; Berdoy and Macdonald 1991; Klein e t
a1. 1994). In our outdoor enclosures, we
presented adult rats with. areas containing
one of four distinct odours: th e rat's own
sm ell (own straw bedding), neu tra l smell
(wa ter), rabbit odour (rabbit mine) and ca t
odour (ca t mine). As expected, non-infected
rats showed a healthy avoidance of ca t-
scented areas, visiting them significan tl y less
than other sites. However, ill accord ance
with the manipulation hypo thesis, infected
rats however were Significantly less averse
a nd showed no overall avo idance of areas
with signs of cat presence. A proportion
even ap peared to show a significan t
preference for the areas scen ted with odoms
of predators (M. Berdoy et aI., unpublished
data).
80
70
60
"0 50
Ql
U
Ql
40
~
~
0
30
20
10
0
o 2 3 4 5 6
Neophobia index
Figure 8.
Prevalence of Toxoplasma-positive rats in each
neophobia index category, where a score of 0
indicated no neophobia and 6 extreme neophobia
(n = 36). The figure illustrates t hat infected rats
were less neophobic t han noninfected
= =
individuals (F1 34 5.0, P 0.03).
67
Ecologically-based Rodent Management
Alterations induced by T. gondii infection
were confined to the predator's odour, as
both types of rats behaved similarly with
respect to areas containing other odours.
These kamikaze rats were seemingly neither
debilitated nor deranged in other respects.
For example, in scramble competition for
mates they secured just as many copulations
as did lminfected rats, and were of equal
social status (Berdoy et a1. 1995b) (cf. Rau
1983,1984; Freeland 1981: infection with
Trichinella spiralis and Heligmosomoides
polygyrus, respectively, prevent dominance
in mice). It seems, therefore, that Toxoplasma
affects specific behavioural traits likely to
make rats more susceptible to predation by
the feUd definitive host. According to the
saying, curiosity mayor may not kill the cat
...but it is likely to kill the rat!
This kamikaze tendency raises intriguing
evolutionary questions. Diminished caution
of infected rats, a well known for
their fearfulness of novel stimuli, could
arguably enhance the likelihood of their
being preyed on by cats. However,
diminished neophobia also seems certain to
make wild rats more prone to poisoning by
man. If a poisoned rat were more easily
caught by cats, the parasite would be
disadvantaged if the cats then succumbed to
secondary poisoning. Although
comprehensive vermin control programs
are a very recent development in
evolutionary terms, it is reasonable to
assume that they have already begun to
affect the ecology of Toxoplasma
transmission. It is also interesting to
consider the antiquity of this parasite-
altered behaviour; does it have its origins in
the predator-prey arms race between wild
felids and rats? Has the adaptation been
68
enhanced by the intensity of the interaction
between rats and domestic cats during the
last 4,000 years of shared peridomestic
existence? Is the same behavioural
modification found in other murine
rodents?
Rural rat diseases
We investigated the prevalence of zoonoses
in approximately 600 wild Norway rats
captured on farms in southern England
(Table 1). Thirteen zoonotic and ten non-
zoonotic species of parasite were identified,
many of which had never previously been
recorded, or even investigated, in the United
Kingdom (Webster and Macdonald 1995a).
In addition to Toxoplasma gondii, another
protozoan, Cryptosporidium parVUnl, which
causes enteritis and enterocolitis in humans
and other mammals (Perryman 1990) was
found to be widespread (Webster and
Macdonald 1995b; previously reported in
Japan-Iseki 1986). The link between
domestic cats and wild rats in the
transmission of human disease was
continued by the discovery of the rickettsian
parasite, Coxiella bumettii, the causative
agent for Q-fever (Webster et a1. 1995b).
Q-fever outbreaks occur sporadically
(Marrie 1990a,b) and no common source has
been identified. However, cases in humans
are often linked with infection in domestic
cats. In contrast, dogs, which occupy a
similar peri-domestic niche, are rarely
carriers of the parasite (Marrie et a1. 1985;
Baldelli et a1. 1992). An obvious difference
between the two, which might contribute to
their radically different likelihood of
acquiring C. burnctti, is the much more
active predatory behaviour of cats.
 The Beh aviour and Ecology of Rattus norvegicus

Perhaps the most d ramatic find of our unintended) billion-pound experimental

sur veys was the p resence, for the first tim e in demonstration of evolution in action. Ra ts
Europe, of hantavirus an tibodies in wild have two main defences against poisoning
animals (Webster and Macdonald 1995a). campaigns: increasingly efficient
Hantaan viruses cause a group of illnesses in behavioural adaptations, and physiological
humans collectively referred to as hantaan resistance to warfarin (Boyle 1960; Greaves
fe ver or haemorrhagic fever with renal 1985) and now, at least incipiently, resistance
syndrome. Although a harmless, persistent to second generation anticoagulants
infection in rodents, the disease may be (Greaves et a1. 1982a; Gill et a1. 1992).
extremely serious in humans. The discovery
of hantavirus infection reaffirms the need to Table 1.
monitor the disease status of wild rats. Prevalence of zoonoses and zoonotic agents in
Hantavirus transmission is strongly approximately 600 wild brown rats from farms in
southern England.
associated with intra specific wounding
(Glass et a1. 1988). It is therefore possible that Zoonoses/Zoonotic Agents % Infected Rats
ill-planned control operations which disrupt
Ectoparasites
the social hierarchy of rat groups, but fail to Reas 100
achieve eradication, could actually increase Mites 67
the prevalence of the disease (see Swinton et Lice 38
Ticks 0
a1. 1997). Our surveys also have
Helminths
demonstrated the importance of confirming Capillaria spp. 23
rat disease status in different environments: Hymenolepis diminuta 22
Leptospira and Salmonella are usually Toxocara cati 15
Hymenolepis nana 11
considered to be highly prevalent among
Taenia taen iaeformis 11
wild rats (e.g. Waitkins 1991; ChomeI1992),
Bacteria
yet the former was relatively rare (14% Leptospira spp. 14
prevalence) and the latter absent among rats Yersinia enterocolitica 11
studied on British farms (Webster et a1. Listeria spp . 11
Pasteurella spp. 6
1995a,c; see Nakashima et a1. 1978). Other
Pseudomonas spp. 4
possible reservoirs of these diseases must Borrelia burgdorferi 0
therefore be investigated. Salmonella spp . 0
Protozoa
Cryptosporidillm parvllm 63
RAT CONTROL AND THE EVOLUTION Toxoplasma gondii 35
OF RESISTANCE Babesia spp. 0
Sarcocystis spp. 0
Despite prolonged and intensive efforts to Rickettsia
eradicate the No rway ra t, the species Coxiella burnetti 34

remains a major pest. A ttempts to con trol rat Virus

Hantavirus 4
populations by poisoning tell a story of Cowpox o
attacks and counter-attacks, currently
unfolding before our eyes in a (largely

69
Ecologically-based Rodent Management
Behavioural resistance
As mentioned above, many rats react to
novel stimuli with extreme caution. Poison
avoidance is also enhanced by their ability to
associate the metabolic consequences of a
food, long after it was ingested, and their
capacity to interpret cues from other group
members about the safety of foods. With
these phenomena in mind, and observing
that some rat populations were
exceptionally difficult to poison (indeed they
seemed not to eat the poison), the idea arose
that they might have evolved enhanced
behavioural resistance. A notable instance
was the case of rats in Hampshire (Brunton
et a1. 1993, see also Greaves et al. 1982b). To
confirm that the failure of anticoagulant
poisons in Hampshire was not attributable
to physiological resistance, we organised a
control campaign with a non-anticoagulant
poison, calciferol, to which there is no
physiological resistance. Although some rats
succumbed, at least 20-50% survived,
despite repeated and intimate access to the
poisoned bait. Clearly, they were not eating
lethal doses. Three, not necessarily
exclusive, factors might have explained this:
(a) genetically enhanced neophobia;
(b) experience; and (c) the stability of the
environment. Our enclosure trials allowed
us to discount the genetic explanation for
hyper-neophobia: as discussed previously,
the Hampshire rats were actually less
neophobic than were survivors from a
poisoning program in Wales. The evidence
instead suggested that the abundant supply
of alternative food, and the stable
environment, were more likely explanations
of the rats' reluctance to eat bait (Brunton et
al. 1993). The farms in Hampshire, on which
70
the 'resistant' rats were found, tended to be
large and surrounded by fields of cereals,
and corn is often stored loose in buildings.
This provides an abundant and relatively
predictable alternative food supply. Rats
living in this environment can therefore
'afford' to be neophobic. By contrast, in mid-
Wales, where rat control by poisoning is
widely held to be successful, farms tend to
be smaller with more livestock and less
stored grain. In these less predictable
environments, rats would need to be more
opportunistic.
Physiological resistance
The introduction of the anticoagulant
poisons, such as warfarin, marked a
breakthrough in rat control. Due to their
slow action, there is a reduced likelihood of
aversion learning because the animal has
already ingested a lethal dose by the time the
symptoms of toxicosis develop
(e.g. Nachman and Hartley 1975).
Warfarin, the most commonly used
rodenticide for the last thirty years, is an
anticoagulant poison which affects the
biogenesis of blood clotting factors,
ultimately causing death by many small
internal haemorrhages. The production of
several blood clotting factors is driven by a
series of reactions where vitamin K is
cyclically oxidised and reduced (Bell and
Caldwell1973; MacNicoll1988). A simple
vitamin K deficiency or substances
disrupting the vitamin K cycle-such as
warfarin -lead to the production of lower
levels of blood clotting factors and therefore
to poor coagulation and an increased risk of
haemorrhage.
The prompt appearance of resistance to
warfarin is a compelling example of the
 The Behaviour and Ecology of Rattus norvegicus
power of natural selection: by 1972, less than
two decades after its introduction in
England, warfarin resistance was reported in
12 areas in the United Kingdom (Greaves
and Rennison 1973).
The costs of physiological resistance
The new benefits enjoyed by resistant rats
involve new costs. In Welsh populations,
where the mechanisms of resistance have
been best studied, the altered enzyme
present in warfarin-resistant individuals,
whilst less affected by warfarin, is also less
efficient at producing vitamin K (Greaves
and Ayres 1969). This results in a natural
vitamin K deficiency (Hermodsen et al. 1969)
and ultimately a reduced coagulating
activity in resistant individuals. Resistant
rats are in a constant state of vitamin K
deficiency and require a greater than normal
amount of vitamin K in their diet in order to
retain a normal clotting activity.
Homozygous resistant rats need even more
dietary vitamin K than heterozygous
resistant rats (Hermodsen et al. 1969; see also
Greaves and Ayres 1973; Martin 1973). The
costs of resistance are substantial,
particularly in homo zygotes, and can lead to
selective deaths during the first few weeks
after birth (Bishop et al. 1977). Even when
sub-lethal, resistance may also affect other
components of fitness such as growth, social
status and reproduction (Smith et al. 1991b).
These findings are of particular
theoretical and practical relevance because
the fitness costs incurred by resistant
animals may affect the rate of spread and
maintenance of the resistance allele in wild
populations. Balanced polymorphism in
wild rat populations regularly controlled
with warfarin suggests the existence of a
selective advantage of heterozygotes
(Greaves et al. 1977). This is consistent with
the interpretation that susceptible rats face
reduced fitness due to the effects of the
poison, and homozygous resistant rats have
reduced fitness due to vitamin K deficiency.
Heterozygotes, on the other hand, enjoy an
efficient compromise: they are more
resistant to warfarin than susceptible rats,
while being less subject to vitamin K
deficiency than homozygous resistant rats.
In practical terms, the lower fitness of
resistant genotypes (heterozygotes
included) in the absence of poison (Partridge
1979) advocates a temporary relaxation of
the use of warfarin to reduce the frequency
of resistant individuals (Smith and Greaves
1987).
A new resistance
There may be a new strain of resistance
amongst rats caught in southern England
(Smith et al. 1993). We found that
heterozygous resistance to warfarin
poisoning was much higher than expected
amongst 173 rats taken into captivity from
two poplations which had resisted
poisoning in the wild (71 % and 76% were
heterozygous resistant, versus expected
values of 50% and 49%; p < 0.001 in both
cases) (Gill et al. 1992). Moreover, while
warfarin-resistant rats normally have lower
body weights, these animals had slightly
higher than average weights (Smith et al.
1993,1994). These results suggest that these
resistant rats, instead of suffering from the
expected costs of resistance, benefited from a
similar if not higher fitness than the
susceptible genotypes, even in the absence of
poison. Apart from its theoretical interest,
these results have profound management
71
Ecologicallybased Rodent Management
implications. Resistance could lie dormant, if
not spread, amongst rat populations living
in warfarin-free environments, ready to
counteract renewed poisoning efforts
(Berdoy and Smith 1993). Nothing is yet
known about the physiological basis of this
potentially fascinating new development in
the chemical arms race between humans and
rats, but our results suggest that rats may
currently be ahead in that race.
CONCLUSION
A full understanding of the biology of target
pest species is vital to the design of effective
management strategies. Nowhere, perhaps,
is this better illustrated than in the case of the
Norway rat, R. norvegicus. Not only are rats
extremely widespread and adaptable, but
they are also notoriously difficult to control.
We have demonstrated that knowledge of
the social organisation and movements of
rats can assist in more effective placing of
bait. Recognition of the complex feeding
behaviour of rats is also of crucial
importance to the success of poisoning
strategies: neophobia, food preferences and
aversive conditioning can all limit bait
uptake. The biological approach to pest
management can also reveal the need for
strategies which may at first appear counter-
intuitive. For example, the relaxation of
poisoning pressure may actually improve
the success of rat control in the long term, by
reducing the frequency of resistant
individuals in the population. Finally, a
scientific framework will assist in the proper
evaluation of the economic and public health
significance of pest control operations. For
example, examination of data on rat
population dynamics and behaviour under
72
conditions of high food availability suggests
that new strategies are needed to
complement lethal control operations.
Similarly, we have demonstrated that rats
may be less important than generally
thought in the transmission of certain
zoonoses (for example Salmonella sp. and
Leptospira sp.), but that consideration must
now be given to their role in the
transmission of diseases such as Q-fever.
REFERENCES
Adams, N. and Boice, R. 1983. A longitudinal
study of dominance in an outdoor colony of
domestic rats. Journal of Comparative
Psychology, 97, 24-33.
Adams, N. and Boice, R. 1989. Development of
dominance in domestic rats in laboratory and
semi-natural environments, Behaviour
Proceedings, 19, 127-142.
Aisenstadt, D.S. 1945. Habitat distribution and
ecology of the brown rat (Rattus norvegicus,
Berkenh,) in a northwestern district of the
European part of the USSR. Zoo1. Zh, Mosk.,
24,182-189,
Amarasekare, P. 1993. Potential impact of
mammalian pest predators on endemic forest
birds of western Mauria Kea, Hawaii. Conser-
vation Biology, 7, 316-324,
Andrews, RV., Belknap, RW., Southard, S.,
Lorincz, M. and Hess, S. 1972. Physiological,
demographic and pathological changes in
wild Norway rat populations over an annual
cycle. Comparative Biochemistry and Physi-
ology, 41(A), 149-165.
Baenninger, L.P. 1966. The reliability of
dominance orders in rats. Animal Behaviour,
14,367-371.
Baenninger, L.P. 1970. Social dominance orders
in the rat: 'spontaneous', food, and water
competition. Journal of Comparative and
Physiological Psychology, 71, 202-209.
 The Behaviour and Ecology of Rattus norvegicus
Baldelli, R, Cimmino, C and Pasquinelli, M.
1992. Dog-transmitted zoonoses: a seriologi-
cal survey in the province of Bologna. Annali
del Istituto Superiore di Sanita (Roma), 28,
493-496.
Barnett, S.A. 1955. Competition among wild rats.
Nature, 175, 126-127.
Barnett, S.A. 1958a. An analysis of social behav-
iour in wild rats. Proceedings of the Zoologi-
cal Society of London, 130, 107-152.
Barnett, S.A. 1958b. Experiments on 'neophobia'
in wild and laboratory rats. British Journal of
Psychology, 49,195-201.
Barnett, S.A. 1975. The rat: a study of behaviour.
London, University of Chicago Press.
Barnett, S.A. and Spencer, M.M. 1951. Feeding,
social behaviour and interspecific competi-
tion in wild rats. Behaviour, 3, 229-242.
Beck, M., Hitchcock, CL. and Galef, B.G. Jr. 1988.
Diet sampling by wild Norway rats offered
several unfamiliar foods. Animal Learning
and Behaviour, 16,224-230.
Bell, RG. and Caldwell, P.T. 1973. Mechanism of
warfarin-resistance. Warfarin and the metab-
olism of vitamin K1. Biochemistry, 12, 1759-
1762.
Berdoy, M. 1991. Feeding behaviour of wild rats,
Rattus norvegicus: social and genetic aspects.
PhD Thesis, Oxford University.
Berdoy, M. 1993. Defining bouts of behaviour: a
three-process model. Animal Behaviour, 46,
387-396.
Berdoy, M. 1994. Making decisions in the wild:
constraints, conflicts and communication in
foraging rats. In: Gale' B.G. Jf. and Valsechhi,
P., ed., Behavioural aspects of feeding: Ettore
Majorana International Life Science Series,
Vo112. Chur, Switzerland, Harwood
Academic Publishers, 289-313.
Berdoy, M. and Evans, S. 1990. An automatic
recording system for identifying individual
small animals. Animal Behaviour, 39, 998-
1000.
Berdoy, M. and Macdonald, D.W. 1991. Factors
affecting feeding in wild rats. Acta Oecolog-
ica, 12, 261-279.
Berdoy, M. and Smith, P. 1993. Arms race and rat
race: adaptations against poisoning in the
brown rat. Revue d'Ecologie (Terre & Vie), 48,
215-228.
Berdoy, M., Smith, P. and Macdonald, D.W.
1995a. Stability of social status in wild rats:
age and the role of settled dominance. Behav-
iour, 132, 193-212.
Berdoy, M., Webster, J.P. and Macdonald, D.W.
1995b. Parasite-altered behaviour: is the effect
of Toxoplasma gondii on Rattus norvegicus
specific? Parasitology, 111,403-409.
Berdoy, M., Webster, J.P. and Macdonald, D.W.
1995c. The manipulation of rat behaviour by
Toxoplasma gondii. Mammalia, 59, 605-613.
Berverley, J.K.A. 1976. Toxoplasmosis in
animals. Veterinary Record, 99,123-127.
Bishop, J.A. and Hartley, D.J. 1976. The size and
age structure of rural populations of Rattus
norvegicus containing individuals resistant to
the anticoagulant poison Warfarin. Journal of
Animal Ecology, 45, 623-646.
Bishop, J.A., Hartley, D.J. and Partridge, G.G.
1977. The population dynamics of genetically
determined resistance to warfarin in Rattus
norvegicus from mid-Wales. Heredity, 39,
389-398.
Blanchard, RJ., Blanchard, D.C, Rodgers, J. and
Weiss, S.M. 1990. The characterisation and
modelling of antipredator defensive behav-
iour. Neuroscience and Biochemical Reviews,
14, 463-472.
Boice, R 1969. Dominance and survival in
stressed wild and domesticated populations.
Proceedings of the American Psychologists'
Association, 7j'1h Annual Convention, 187-
188.
Boice, R 1972. Some behavioural tests of domes-
tication in Norway rats. Behaviour, 42, 198-
231.
Boice, Rand Boice, C 1968. Trapping Norway
rats in a land fill. Journal of Science Laborato-
ries, Dension University, 49,1-4.
Bond, N.W. 1984. The poisoned partner effect:
some parametric considerations. Animal
Learning and Behaviour, 12,89-96.
Boreman, J. and Price, E. 1972. Social dominance
in wild and domestic Norway rats (Rattus
norvegicus). Animal Behaviour, 29, 534-542.
73
Ecologically-based Rodent Management
Boyle, C.M. 1960. Case of apparent resistance of
Rattus norvegicus Berkenhout, to anticoagu-
lant poisons. Nature, 188,517.
Brammer, G., Barnett, S.A and MarpIes, T.G.
1988. Responses to novelty by the Australian
swamp rat, Rattus putreolus. Australian
Mammology, 11,63-66.
Brodie, J. 1981. Norway rats (Rattus norvegicus)
on cereal stubble. Journal of Zoology,
London, 195,542-546.
Brown, RE. and Macdonald, D.W. 1985. Social
odours in mammals. Oxford and New York,
Clarendon Press,506p.
Brunton,C.P.AI995. Neophobia and its effect on
the macro-structure and micro-structure of
feeding in wild brown rats (Rattus norvegicus).
Journal of Zoology (London), 235, 223-236.
Brunton, C.P.A and Macdonald, D.W. 1996.
Measuring the neophobia of individuals in
different populations of wild brown rats.
Journal of Wildlife Research, 1,7-14.
Brunton, C.F.A, Macdonald, D.W. and Buckle,
AP. 1993. Behavioural resistance towards
poison baits in brown rats Rattus norvegicus.
Applied Animal Behaviour Science, 38,159-
174.
Calhoun, J.B. 1962. The ecology and sociology of
the Norway rat. Public Health Service Publi-
cation No. 1008, 288p.
Chitty, D. 1954. The control of rats and mice.
Oxford, Clarendon, 1041 p.
Childs, J.E., Glass, G.E., Korch, G.W. and Leduc,
J.W. 1987. Prospective seroepidemiology of
hantaviruses and population dynamics of
small mammal communities of Baltimore,
Maryland (USA). American Journal of Tropi-
cal Medicine and Hygiene, 37, 648-662.
Chomel, B.B. 1992. Zoonoses of house pets other
than dogs, cats and birds. Pediatric Infectious
Diseases Journal, 11,479-487.
Christian, J.J. and Davies, D.E.1964. Endocrines,
behaviour and population. Science, 146,
1550-1560.
Cowan, D.P., Bull, D.S., Inglis, I.R, Quy, RJ. and
Smith, P. 1994. Enhancing rodenticide
performance by understanding rodent
behaviour. In: BCPC Conference-Pests and
74
Diseases. Thornton Health, UK, BCPC
(British Council of Pest Control) Publications,
1039-1046.
Cowan, P.E. 1976. The new object reaction of
Rattus rattus L.: the relative importance of
various cues. Behavioural Biology, 16, 31-44.
Cowan, P.E. 1977. Neophobia and neophilia:
new object and new place reactions of three
Rattus species. Journal of Comparative and
Physiological Psychology, 91, 63-71.
Cowan, r.E. and Barnett, S.A 1975. The new-
object and new-place reactions of Rattus rattus
L. Zoological Journal of the Linnean Society,
56,219-234.
Cree, A, Daugherty, c.H. and Hay, J.M. 1995.
Reproduction of a rare New Zealand
the Tautara sphenodon punctatus on
and rat-inhabited islands. Conservation
Biology, 9, 373-383.
Davies, D.E. 1949. The weight of wild brown rats
at sexual maturity. Journal of Mammalogy,
39,125-130.
Davies, D.E. and Hall, 0.1951. The seasonal
reproductive condition of the female Norway
(brown) rat in Baltimore, Maryland. PhYSio-
logical Zoology, 24, 9-20.
Drummond, H. and Osorno, J.L. 1992. Training
siblings to be submissive losers: dominance
between booby nestlings. Animal behaviour,
44,881-893.
Emlen, J.T., Stokes, AW. and Winsor, c.P. 1948.
The rate of recovery of decimated popula-
tions of brown rats in nature. Ecology, 29,
133-145.
Errington, P.L. 1935. Wintering of field-living
Norway rats in south-central Wisconsin.
Ecology, 16, 122-123.
Farhang-Azad, A and Southwick, c.H. 1979.
Population ecology of Norway rats in the
Baltimore Zoo and Druid Hill Park, Balti-
more, Maryland. Annals of Zoology, 15, 1-42.
Penn, M.G.P. and Macdonald, D.W.1987. The
contribution of field studies to stored product
rodent control. In: Lawson, T.J., ed., Stored
product pest control. Thornton Heath, UK,
BCPC (British Council of Pest Control) Publi-
cations, 107-113.
 The Behaviour and Ecology of Rattus nDrvegicus
Fenn, M.G.P. and Macdonald, D.W. 1995. Use of
middens by red foxes: risk reverses rhythms
of rats. Journal of Mammalogy, 76, 130-136.
Fenn, M.G.P., Tew, T.E. and Macdonald, D.W.
1987. Rat movements and control on an
Oxfordshire farm. Journal of Zoology
(London), 213, 745-749.
Freeland, W.J.1981. Parasitism and behavioural
dominance among male mice. Science, 213,
461-462.
Galef, B.G. Jr. 1988. Communication of informa-
tion concerning distant diets in a social,
foraging species: Rattus norvegi-
cus.ln: Zentall, T.R and GaleE, B.G., ed.,
Social learning: psychological and biological
perspectives. Hillside, New Jersey, Hove and
London, Lawrence Erlbaum Associates, 119-
139.
Galef, B.G. Jr. 1994. Olfactory communications
about foods among rats: a review of recent
findings. In: Galef, B.G. Jr. and Valsechhi, P.,
ed., Behavioural aspects of feeding. Ettore
Majorana International Life Science
Vo112. Chur, Switzerland, Harwood
Academic Publishers, 83-103
Gale, B.G. Jr. and Clark, M.M.1972. Mother's
milk and adult presence: two factors deter-
mining initial dietary selection by weanling
rats. Journal of Comparative and Physiologi-
calPsychology, 78, 220-225.
Galef, B.G. Jr. and Heiber, L. 1976. The role of
residual cues in the determination of feeding
site selection and exploration patterns of
domestic rats. Journal of Comparative and
Physiological Psychology, 90, 727-739.
Gale, B.G.Jr. and Wigmore, S.W.1983 Transfer
of information concerning distant foods: a
laboratory investigation of the "information
centre" hypothesis. Animal Behaviour, 31,
748-758.
Gandolfi, G. and Parisi, V.1973. Ethological
aspects of predation of rats, Rattus norvegicus
(Berkenhout) on bivalves Unio pictorium and
Cerastoderma lamarcki. Bolletino Zoologico,
40,69-74.
Gartner, K. Wankel, B. and Gaudszuhn, D. 1981.
The hierarchy in copulatory competition and
its correla tion with pa ternity in grouped male
laboratory rats. Zeitschrift Tierpsychologie,
56, 243-254.
Gill, J.E., Kerins, G.M. and MacNicoll, A.D. 1992.
Inheritance of low grade brodifacoum resist-
ance in the Norway rat. Journal of Wildlife
Management, 56, 809-816.
Glass, G.E., Childs, J.E., Korch, G. W. and Le Duc,
J.W. 1988. Association of intraspecific
wounding with hantaviral infection in wild
rats (Rattus norvegicus). Epidemiology of
Infection, 101, 459-472.
Greaves, J.H. 1985. The present state of resistance
to anticoagulants. Acta Zoologica Fennica,
173,159-162.
Greaves, J.H. and Ayres, P.B. 1969. Linkages
between genes for coat colour and resistance
to warfarin in Rattus norvegicus. Nature, 224,
284-285.
Greaves, J.H. and Ayres, P. 1973. Warfarin-resist-
ance and vitamin K requirement in the rat.
Laboratory Animals, 7, 141-148.
Greaves, J.H., Redfern, R, A yres, P.B. and Gill,
J.E. 1977. Warfarin-resistance: a balanced
polymorphism in the Norway rat. Genetics
Research, Cambridge, 30,257-263.
Greaves, J.H. and Rennison, B.D. 1973. Popula-
tion aspects of warfarin-resistance in the
brown rat, Rattus norvegicus. Mammal
Review,3,27-29.
Greaves, J.H., Shepherd, D.S. and Gill, J.E. 1982a.
An investigation of difenacoum resistance in
Norway rat populations in Hampshire.
Annals of Applied Biology, lOO, 581-587.
Greaves, J.H., Shepherd, D.5. and Quy, R 1982b.
Field trials of second-generation anti-coagu-
lants difenacoum-resistant Norway
rat populations. Journal of Hygiene, 89, 295-
301.
Hardy, A.R. and Taylor, K.D. 1979. Radiotrack-
ing of Rattus l10rvegicus on farms. In: Amlaner,
CJ., Macdonald, D. W., ed., A handbook on
biotelemetry and radiotracking. Oxford,
PergamonPress, 657-665.
75
Ecologically-based Rodent Management
Hartley, DJ. and Bishop, J.A 1979. Home range
and movement in populations of Rattus
norvegicus polymorphic for Warfarin resist-
ance. Biological Journal of the Linnean
SOciety, 12, 19-43.
Hay, L Aitkin, P.P., Hutchison, W.M. and
Graham, DJ. 1983a. The effect of congenital
and adult acquired Toxoplasma infections on
activity and responsiveness to novel stimula-
tion in mice. Annals of Tropical Medicine and
Parasitology, 77,483-495.
Hay, J., Hutchison, W.M., Aitkin, P.P. and
Graham, D.I.1983b. The effect of congenital
and ad uIt acquired Toxoplasma infections on
the motor performance of mice. Annals of
Tropical Medicine and Parasitology, 77,261-
277.
Hayne, H., Rovee-Collier, C and Gargano, D.
1986. Ambient temperature effects on
energetics relations in growing chicks. Physi-
ologyand Behaviour, 37, 203-212.
Hepper, P.G.1990. Foetal olfaction. In: Macdon-
aId, D.W., Muller-Schwarze, D. and Natync-
zuk, S.E., ed., Chemical signals in vertebrates.
Oxford, Oxford University Press, 282-287.
Hermodsen, M.A, Suttie, J.W. and Link, KP.
1969. Warfarin metabolism and vitamin K
requirement in the warfarin-resistant rat.
American Journal of Physiology, 217,1316-
1319.
Homolka, M. 1983. On the problem of the
exanthropic occurrence of Rattus norvegicus.
Folia Zoologica, 32, 203-21l.
Hubel, nH. and Wiesel, T.N. 1962. Receptive
fields, binocular interaction and functional
architecture in the cat's visual cortex. Journal
of Physiology, 160, 106-154.
Huntingford, F. and Turner, A 1987. Animal
conflict. London, Chapman and Hall, 448p.
Huson, L.W. and Rennison, BD. 1981. Seasonal
variability of Norway rat (Rattus norvegicus)
infestation of agricultural premises. Journal
of Zoology, London, 194,257-289.
Hutchison, W.M., Dunachie, J.F., Siim, J.CHR
and Work, K 1969. The life cycle of
Toxoplasma gondii. British MedicalJournal, 4,
806.
76
Hutchison, W.M., Bradley, M., Cheyne, W.M.,
Wells, B.W.P. and Hay, J. 1980. Behavioural
abnormalities in Toxoplasma-infected mice.
Annals of Tropical Medicine and Parasitol-
ogy, 74,507-510.
Infurna, RK., Steinert, P. A, Freda, 15. and
Spear, N.E. 1979. Sucrose preference and LiCI
illness-induced aversion as a function of drug
dose and phase of the illumination cycle.
Physiology and Behaviour, 22, 955-961.
Inglis, LR, Shepherd, D.5., Smith, 1'., Haynes,
P.J., Bull, D5., Cowan, np. and Whitehead,
n 1996. Foraging behaviour of wild rats
(Rattus norvegicus) towards new foods and
bait containers. Applied Animal Behaviour
Science,47,175-190.
Iseki, M. 1986. Two species of Cryptosporidium
naturally infecting house rats, Rattus norvegi-
cus. Japanese Journal of Parasitology, 35, 251-
256.
Jackson, M.H., Hutchinson, W.M. and Siim, rC
1986. Toxoplasmosis in a wild rodent popula-
tion of central Scotland and a possible expla-
nation of the mode of transmission. Journal of
Zoology, 209, 549-557.
Kendall, P.B. 1984. Seasonal changes of sex ratio
in Norway rat (Rattus populations
in Wales. Journal of Zoology, London, 203,
208-211.
Klein, S.L., Lambert, KG., Durr, D., Schaefer, T.
and Waring, RE. 1994. Influence of environ-
mental enrichment and sex on predator stress
response in rats. Physiology and Behaviour,
56,292-297.
Lainson, R 1957. The demonstration of
Toxoplasma in animals, with particular refer-
ence to members of the Mustelidae. Transac-
tions of the Royal Society of Tropical
Medicine and Hygiene, 51, 111-117.
Lattanzio, RM. and Chapman, J.A 1980. Repro-
duction and physiological cycles in an island
population of Norway rats. Bulletin of the
Chicago Academy of Science, 12, 1-68.
Le J. 1985. Hunger. Cambridge,
Cambridge University Press, 157p.
 The Behaviour and Ecology of Rattusnorvegicus
Leslie, P.H., Venables, U.M. and Venables, L.S.
1952. The fertility and population structure of
the brown rat (Rattus norvcgicus) in corn ricks
and some other habitats. Proceedings of the
Zoological Society, London, 122, 187-238.
Leyhausen,P.1979.Catbehaviour:thepredatory
and sodal behaviour of domestic and wild
cats. Garland Series in Ethology. New York,
Garland STPM Press.
Ludvigson, H.W., Mathis, D.A. and Choquette,
K.A. 1985. Different odours in rats from
and small rewards. Animal and
Behaviour, 13,315-320.
Macdonald, D.W. and Brown, R.E. 1985. The
smell of success. New Scientist, 106, 10-14.
Macdonald, D.W. and Fenn, M.P.G. 1995. Rat
ranges in arable areas. Journal of Zoology
(London), 236, 349-353.
MacNicoll, A.D.1988. The role of altered vitamin
K metabolism in anticoagulant resistance in
rodents. In: Suttie,I.W., ed., Current advances
in vitamin K research. Amsterdam, Elsevier
Science, 407-417.
Marrie, T.J. 1990a. Q-fever-a review. Canadian
Veterinary Journal, 3, 555-564.
Marrie, T.J. 1990b. Epidemiology of Q-fever. In:
Marrie, T., ed., Q-fever: the disease, Vol. l.
Boca Raton, CRC Press, 49-70.
Marrie, T.J., Van Buren, I. and Fraser, J. 1985.
Seroepidemiology of Q-fever among domes-
tic animals in Nova Scotia. American Journal
of Public Health, 75, 763-766.
Martin, A.D. 1973. Vitamin K requirement and
anticoagulant response in the warfarin resist-
ant rat. Biochemical Society Transactions, 1,
1206-1208.
McClintok, M.K., Anisko, J.J. and Adler, N.T.
1982. Group mating among Norway rats II.
The social dynamics of copulation: competi-
tion, cooperation and mate choice. Animal
Behaviour, 30, 410-425.
Middleton, A.D. 1954. Rural rat control. In:
Chitty, D., ed., Control of rats and mice.
Oxford, Clarendon Press, 414-448.
Miller, R.R. and Holzmann, A.D. 1981. Neopho-
bia: generality and function. Behavioural and
Neural Biology, 33, 17-44.
Mitchell, P. 1976. Experiments on neophobia in
wild and laboratory rats; a reevaluation.
Journal of Comparative and Physiological
Psychology,90,190-197.
Nachman, M. and Hartley, P.L. 1975. Role of
illness in producing taste aversion in rats: a
comparison of several rodenticides. Journal
of Comparative and PhYSiological Psychol-
ogy,89,1010-1018.
Nakashima, R.S., Wetzler, T.F. and Nordin, }.P.
1978. Some microbial threats posed by Seattle
rats to the community's health. Journal of
Environmental Health, 40, 264-267.
Natynczuk, S.E. and Macdonald, D.W.1994a.
Scent, sex, and the self calibrating rat. Journal
of Chemical Ecology, 20, 1843-1857.
Natynczuk, S.E. and Macdonald, D.W. 1994b.
Analysis of rat haunch odour using the
dynamic solvent effect and principal compo-
nent analysis. Journal of Chemical Ecology,
20,1859-1866.
Natynczuk, S.E., Macdonald, D.W. and Tatter-
saIl, F.H. 1995. Morphology and chemistry of
brown rat, Rattus preputial and
clitoral glands. Journal of Chemical Ecology,
21,297-260
Nott, H.M.R. 1988. Dominance and feeding
behaviour in the brown rat. PhD thesis,
University of Reading.
Orians, G.H. and Pears on, N.E. 1979. On the
theory of central place foraging. In: Horn,
D.}., Mitchell, R. and Stair, G.R., ed., AnalysiS
of ecological systems. Columbus, Ohio State
University Press, 155-177.
Partridge, G.G. 1979. Relative fitness of
genotypes in a population of Rattus
polymorphic for warfarin resistance. Hered-
ity, 43, 239-246.
Perryman, L.E.1990. Cryptosporidium in rodents.
In: Dubey, I.P., Speer, CA. and R., ed.,
Cryptosporidiosis of man and animals. Boca
Raton, CRC Press, 125-132.
Posadas-Andrews, A. and Roper, T.]. 1983. Social
transmission of food preferences in adult rats.
Animal Behaviour, 31, 265-271.
Price, E.O., Belanger, P.L. and Duncan, R.A. 1976.
Competitive dominance of wild and domes-
tic Norway rats (Rattus Animal
Behaviour, 24, 589-599.
77
Ecologically-based Rodent Management
Quy, RJ., Cowan, D.P., Haynes, P., Inglis, LR
and Swinney, T. 1992a. The influence of
stored food on the effectiveness of farm rat
control. In: BCPC Conference-Pests and
Diseases. Thornton Health, UK, BCPC
(British Council of Pest Control) Publications,
291-300.
Quy, RJ., Shepard, D.s. and Inglis, LR 1992b.
Bait avoidance and effectiveness of anti-
coagulant rodenticides against warfarin and
difenacoum-resis tant popula tions of Norway
rats (Rattus norvegicus). Crop Protection, 11,
14-20.
Rau, M.E. 1983. Establishment and maintenance
of behavioural dominance in male mice
infected with Trichinella spiralis. Parasitology,
86,319-322.
Rau, M.E. 1984. Loss of behavioural dominance
in male mice infected with Trichinella spiralis.
Parasitology, 88, 371-373.
Rescorla, RA 1980. Pavlovian second-order
conditioning: studies in associative learning.
Hillside, Lawrence Earlbaum Associates.
Richards, C.C.J. 1981. Field trials ofbromadi-
olone against infestations of warfarin-resist-
ant Rattus norvegicus. Journal of Hygiene, 86,
363-367.
Richter, c.P. 1953. Experimentally produced
behaviour reactions to food poisoning in wild
and domesticated rats. Annals of the New
York Academy of Science, 56, 225-239.
Robertson, H.A, Hay, H.J., Saul, E.K. and
McCormack, C.V. 1994. Recovery of the
kakeroi: an endangered forest bird of the
Cook Islands. Conservation Biology, 8, 1078-
1086.
Rozin, P. 1976. The selection of food by rats,
humans and other animals. Advances in the
Study of Behaviour, 6, 21-76.
Rozin, P. and Kala t, J. 1972. Specific hungers and
poison avoidance as adaptive specializations
of learning. Psychology Review, 78,459-486.
Royet, J.P. 1983. Les aspects comportementaux
de l' aversion conitionee et de la neophobia.
L'annee Biologique, 22, 113-167.
Rutberg, AT. 1983. Factors influencing
dominance status in American bison cows
(Bison bison). Zeitschrift Tierpsychologie, 63,
206-212.
78
Schmidt, C.D. and Roberts, L.S. 1989. Founda-
tions of parasitology, 4th edn. London, Times
Mirror /Mosby, College Publishing, 750p.
Shepherd, D.s. and Inglis, LR 1987. Feeding
behaviour, social-interactions and poison bait
consumption by a family group of wild rats
living in semi-natural conditions. In: Lawson,
T.J., ed., Stored products pest control. Thorn-
ton Health, UK, BCPC (British Council of Pest
Control) Publications, 97-105.
Shorten, M. 1954. The reaction of the brown rat
towards changes in its environment. In:
Chitty, D., ed., The control of rats and mice, 2.
Oxford, Clarendon Press, 307-334.
Singleton, C.R 1997. Integrated management of
rodents: a Southeast Asian and Australian
perspective. Belgian Journal of Zoology, 127,
157-169.
Slater, P.J.B. 1981. Individual differences in
animal behaviour. Perspectives in Ethology,
4,35-49.
Smith, P., Berdoy, M., Smith, RH. and Macdon-
aId, D.W. 1993. A new aspect of resistance in
wild rats: benefits in the absence of poison.
Functional Ecology, 7, 190-194.
Smith, P., Berdoy, M. and Smith, RH. 1994.
Body-weight and social dominance in antico-
agulant-resistant rats. Crop Protection, 13,
311-316.
Smith, P., Smith, RH. and Sibly, RM. 1991a.
Pulsed baiting: laboratory evidence for
behavioural exclusion in wild rats. In:
Fleurat-Lessard, F. and Ducom, P, ed.,
Proceedings of the 5th International Confer-
ence on Stored Product Protection, Bordeaux,
France, September 1990, 1517-1526.
Smith, P., Townsend, M.C. and Smith, RH.
1991b. A cost of resistance in the brown rat:
reduced growth rate in warfarin-resistant
lines. Functional Ecology, 5, 441-447.
Smith, RH. and Creaves, J.H. 1987. Resistance to
anticoagulant rodenticides: the problem and
its management. In: Donahaye, E. and
Navarro, S., ed., Proceedings of the 4th Inter-
national Conference on Stored-Product
Protection, Tel Aviv, Israel, September 1986.
Bet Dagon, Israel, Agricultural Research
Organisation, 302-315.
 The Behaviour and Ecology of Rattus norvegicus
Snowdon, CT. and Wampler, RS. 1974. Effects
of lateral hypothalmic lesions and vagotomy
on meal patterns in rats. Journal of Compara-
tive and Physiological Psychology, 87, 399-
409.
Sridhara, S., Narasimham, AV. and Krishna-
moorthy, RV. 1980. Aggressive interactions
among wild and domestic rodents. Proceed-
ings of the Indian Academy of Science
(Animal Science), 89, 351-357.
Stretch, RG.A., Leytham, G.W.H. and Kershaw,
W.E. 1960a. The effect of acute schistosomia-
sis upon discrimination learning and activity
in mice. Annals of Tropical Yledicine and
Parasitology, 54, 476-486.
Stretch, RG.A, Leytham, G.W.H. and Kershaw,
W.E. 1960b. The effect of acute schistosomia-
sis upon learning in rats under different levels
of motivation. Annals of Tropical Medicine
and Parasitology, 54, 487-492.
SwintonJ., Tuyttens, F., Macdonald, D.W.,
Nokes, D.T., Cheeseman, CL. and Clifton
Hadley, R1997. A comparisonoffertility and
lethal control on bovine tuberculosis in
badgers: the impact of pertubation. Transac-
tions of the Royal Philosophical Society Series
B, 352, 619-631.
Takahashi, L.K. and Lore, RV. 1980. Foraging
and food hoarding of wild Rattus l10rvegicus in
an urban environment. Behavioural and
Neural Biology, 29, 527-531.
Taylor, K.D. 1975. An automatic device for
monitoring small mammal traffic on
runways. Journal of Zoology (London), 176,
274-277.
Taylor, K.D. 1978. Range of movement and activ-
ity of common rats (Rattus norvegicus) on
agricultural land. Journal of Applied
Ecology, 15, 663-677.
Taylor, K.D., Fenn, M.P.G. and Macdonald, D.W.
1991. Common rat. In: Corbett, G.B. and
Harris, S.H., ed., Handbook of British
mammals. Oxford, Blackwell Scientific Publi-
cations, 248-255.
Taylor, RH. and Thomas, B.W. 1993. Rats eradi-
cated from Breaksea Island (170HA), Fiord-
land, New Zealand. Biological Conservation,
65,191-198.
Telle, H.J. 1966. Beitrag zur Kenntis der
Verhaltensweise von Ratten, vergleichen
dargestelt bei Rattus l10rvegicus und Rattus
rattus. Angewandte Zoologie, 53, 129-196
(MAFF Tolworth translation, Rodent
Research reprint 4489).
Tempel, D.L., Shor-Posner, G., Dwyer, D. and
Leibowitz, S.F. 1985. Nocturnal patterns of
macronutrient intake in freely feeding and
food deprived rats. American Journal of
Physiology, 256, R541-R548.
Thompson, H.V.1948. Studies of the behaviour
of the common brown rat: 1. Watching
marked rats taking plain and poisoned bait.
Bulletin of Animal Behaviour, 6, 2-40.
Thouless, CR and Guiness, F.E.1986. Conflict
between red deer hinds: the winner always
wins. Animal Behaviour, 34, 1166-1171.
Vernet-Maury, R, Le Magnen, J. and Chanel, J.
1968. Comportement emotif chez le rat: influ-
ence de l' odeur d'tm predateur et d' un non-
predateur. Paris, CR Acadamie Scientifique,
267,331-334,
Vernet-Maury, E., Polak, E.H. and Demael, A.
1984. Structure-activity relationship of stress-
inducing odomants in the rat. Journal of
Chemical Ecology, 10, 1007-1017.
Waitkins, S.A. 1991. Rats as a source of
leptospirosis-Weil's disease. Sorex Techni-
cal Paper No. 4. Cheshire, Sorex Ltd.
Wallace, G.D. 1981. Re: 'association of cats and
toxoplasmOSiS'. American Journal of Epide-
miology, 113, 198-201.
Wallace, RJ. and Barnett, S.A. 1990. Avoidance of
new objects by the black rat, Rattus rattus,
after object presentation and change. Interna-
tional Journal of Comparative Psychology, 3,
253-265.
Webster,J.P. 1994a. Prevalence and transmission
of Toxoplasmagondii in wild brown rats, Rattus
norvegicus. Parasitology, 108,407-411.
Webster, J.P.1994b. The effect of Toxoplasma
gondii and other parasites on activity levels in
wild and hybrid Rattus norvegicu5. Parasitol-
ogy, 110, 583-589.
Webster J. and Berdoy, M. 1997. Toxoplasma
gondii in Rattus nomegicus: characteristics and
evolution of parasite-altered behaviour. In:
79
Ecologically-based Rodent Management
Holland, S., ed., Modern perspectives on
zoonoses. Dublin, Royal Irish Academy, 115-
121.
Webster, J.P., Brunton, c.P.A. and Macdonald,
D.W. 1994. Effect of Toxoplasma gondii on
neophobic behaviour in wild brown rats,
Rattus Parasitology, 109, 37-43.
Webster, J.P., Ellis, W.A. and Macdonald, D.W.
1995a. Prevalence of Leptospira in wild brown
rat (Rattus norvegicus) populations in the UK.
Epidemiology and Infection, 114, 195-201.
Webster, J.P., Ellis, W.A. and Macdonald, D.W.
1995b. Prevalence of Leptospira and other
zoonoses in wild brown rats on UK farms.
Mammalia, 59, 615-622.
Webster, J.P., Uoyd, G. and Macdonald, D.W.
1995c. Q-fever (Coxiella humetti) reservoir in
wild brown rat (Rattus norvegicus) popula-
tions in the UK. Parasitology, 110,31-35.
80
Webster, lP. and Macdonald, D.W. 1995a.
Parasites of wild brown rats (Rattus norvegi-
cus) on UK farms. Parasitology, 111, 247-255.
Webster, J.P. and Macdonald, D.W. 1995b.
Cryptosporidiosis reservoir in wild brown
rats (Rattus norvegicus); first report in the UK.
Epidemiology and Infection, 115,207-209.
Wolff, J,O. 1988. Maternal investment and sex-
ratio adjustment in American bison calves.
Behavioural Ecology and Sociobiology, 23,
127-133
Zapletal, M. 1964. On the occurrence of the
brown rat (Rattus nOIVegicus, Berk.) under
natural conditions in Czechoslovakia. Zool.
Listy, 13, 125-134.
Roger P. Pech, Greg M. Hood, Grant R. Singleton, Elizabeth Salmon,
Robert I. Forrester and Peter R. Brown

Abstract

In this chapter, the main features of current models for predicting the dynamics of
house mice (Mus domesticus) populations are reviewed and an assessment made
of their data requirements and their ability to contribute to the effective management
of mice in Australia. In addition , recent progress with quantifying aspects of the
dynamics of mouse populations in the MalJee region of Victoria is described.
Robust predictive models are required for the effective management of mice
because plagues (massive eruptions of mice) occur at irregular intervals and farmers
require early warning to implement control techniques and prevent economic losses.
Nine published models have been produced for the plague-prone regions of southern
and eastern Australia. Two of these aim to predict the occurrence of plagues at a
regional level and five predict changes in the abundance of mice at a district, or
local, scale. However, none of the current models for southern Australia include
estimates of the numerical response of mouse populations . This limits their value
for assessing the relative merits of control programs and new control techniques .
A model of the numerical response of mice was developed using observations of
th e abundance of mice over a 15-year period in the Victorian Mallee region. Rates of
increase per 40 days were calculated from the smoothed abundance data and
related to (i) estimates of food availability from cereal crops and grazed pasture and
(ii) a density-dependent factor representing the effects of predation , disease and
intrinsic regulatory processes such as dispersal and social organisation. Although
the model represents reasonably welJ the main features of plagues, the strong
seasonal variation in the modelled food supply is not matched by changes in the
abundance, or rate of increase of mice, in non-plague years. These seasonal effects
are likely to be more important in future models for survival and fecundity rates.

Keywords

Mus domesticus, population dynamics, predictive models, rate of increase , density-

dependence, predator-regulation , disease, management

81
Ecologically-based Rodent Management
INTRODUCTION
T HE INTRODUCE D house mouse
(Mus domes ticus ) is a major pest in
the grain-growing regions of
south ern and eastern Australia. A wide
range of climate, so ils and farming regimes
results in regional differences in the
population dynamics of mice and in the
opportunities for their management. For
exa mple, the Darling Downs in Queensland
experiences rela tively frequen t outbreaks in
contrast to the wheat belt of New South
Wales, Victoria and South A ustralia w here
occasional severe eruptions are interspersed
with long periods of low abw1dance of mke.
1. Darling Downs Western
Australia
2. Macquarie Valley
3. Murrumbidgee
Irrigation Area (MIA)
4. Mallee
5. Turretfield
Grain production
areas prone to
mouse plagues
Figure 1.
Grai n production areas of Australia that are prone to mouse plagues. The numbered locations correspond to
the 'district' models listed in Box 1.
82
The different dynamics imply that the m ajor
regulating factors are not lmiform between
regions.
There are a t least nine models, with
varying degrees of predicti ve ability, which
describe the development of mouse plagues
in Australia (Figure 1, Box 1). They include
two broad-scale, regional models that rely
on environmenta l or p roduction da ta, five
district and sm all-scale models th at describe
in detail th e processes of plague formation,
and two simplified 'process' models
focusing on one or more mechanisms tha t
might influence the ra te of change of mouse
ablmdance. The main fea tures of these
models are reviewed and their usefulness for
predicting mouse plagues is compared.
Northern
Territory
Queensland
,~------------~---,,,
South
Australia
308
o
Tasmania\]
\:,
408
1400E I 150 0E I

The model for predicting the onset,
magnitude and duration of eruptions of mice
in the cereal production areas of the
Victorian Mallee is based on a IS-year data
set that has tracked four major outbreaks of
mice. In this paper we assess the usefulness
of simple food-resource models for
predicting the numerical response of mice
during the period from 1983 to 1997.
Potential areas for future development of the
model for the Victorian Mallee are discussed,
particularly in relation to assessing
management options including the use of
fertility controL
REVIEW OF CURRENT MODELS FOR
MOUSE PLAGUES
Broad-scale models
New South Wales (NSW) and
Victoria regional model
Saunders and Ciles (1977) reviewed the
historical records of mouse plagues from
1900 to 1970 for northern NSW, southern
NSW and northern Victoria. Despite the
difficulties in using consistent definitions
for plagues and droughts, they found a
strong positive correlation between
eruptions of mice and the occurrence of a
severe drought one or two years earlier. The
model was not tested statistically but it can
provide very early warning of a potential
mouse plague. Predictions could be verified
later by monitoring mouse populations
(Saunders 1986). Saunders and Giles
suggested several processes that might be
responsible for the observed correlation.
Few data are available to test their
'pathogen' hypothesis (see Singleton 1985).
Models for Predicting Plagues in Australia
Depending on time lags in the responses of
predator populations, the suggested
mechanism of a drought-induced reduction
in predation appears to be the converse of
the predation-regulation model (below) and
the explanation offered by Newsome and
Corbett (1975) for a delayed response of
mice to favourable conditions in central
Australia. The emphasis on drought as a
causal factor also has been criticised by
Redhead (1982) on the grounds that it is
confounded with the effects of drought-
breaking winter-spring rainfalL
In its present form, the NSW and Victoria
regional model is probably too vague to
provide a prediction which can be the basis
of management actions. The rainfall data on
which the model is based are readily
available but the most appropriate definition
of a severe drought may require some
clarification. Saunders (1986) did not specify
when mouse populations should be
monitored to verify the long-term prediction
from the modeL If coupled with additional
surveys, the model's predictions, in the form
of conditional probabilities suggested by
Hone (1980), might constitute the first stage
of a plague warning system.
South Australia regional model
Mutze (1989) documented the distribution
and frequency of mouse plagues in South
Australia from 1900 to 1984. These data were
used first by Veitch and Anderson (1985),
and then Mutze et al. (1990), to examine the
relationships between the occurrence of
mouse plagues and a range of independent
variables including monthly rainfall, soil
type, temperature indices and grain
production.
83
Ecologically-based Rodent Management

Box 1.
Summary of the main features of the current models for mouse plagues. The events leading to a plague
are listed In approximate chronological order for each category of model. However, not all models
include each of the steps. The locations of the regional and district models are shown in Figure 1 .

(a) Regional models to predict t he occurrence of plagues for:

New South Wales and Victoria (NSW & V) (Saunders and Giles 19 77)
South Australia (SA ) (Mutze et al. 1990)
Sequence of even ts:
1. A low-yield crop (~ drought) two years prior to the plague. (SA )
2. A severe drought (leading to a reduced regulatory effect of disease and/or predation) followed
by one or two years with good winter-spring rain. (NSW &V)
3. A special sequence of ra infall events in the 12 months preced ing the plague. (SA)
4. High-yield crop in the summer prior to a plague in autumn . (SA, NSW &V)
(b) District models used to predict of the abundance of mice for:
Turretfield (T) (Newsome 1969a)
Murrumbidgee Irrigation Area (MIA) (Redhead 1982)
Victorian Mallee (VM) (Singleton 1989)
Darling Downs (DD) (Cantrill 1992 )
Macquarie Valley (MqV) (Twigg and Kay 1 994)
Sequence of event s:
1. Good autum n rain two years prior to plague to extend the breeding season into winter. (MIA)
2. Mice disperse from refuges to other favo urab le areas duri ng t he summer breeding season one
year prior t o the pl ague. (MIA)
3. High autumn-winter rainfa ll in t he year precedi ng th e pl ague . (VM)
4. High abundance of fem ale mice in a wide range of habitats in the spring prior to t he plague
year. (MIA)
5. Correct sequence of rainfall eve nts to provide bu rrowin g and nesting sites in cracking soils .
(T, MqV, DD)
6. Favoura ble clim atic conditi on s over the summer (a high-yield summer crop) in th e plague year.
(T, MIA , VM. MqV. DD)

(c) Process models for the regulation of mouse populations:

Predator-regulation (P-R) (Sinclair et al. 1990)
Regulation by disease (R-O) (McCallum and Singleton 1989; Shellam 1994)
Conditions th at prevent a plague:
1. Aggregation of predators in habitats with high mouse nu mbers (e.g. irrigated crops) due to low
abundance of mice in surrounding areas. (P-R)
2. Density-dependent increase in the prevalence of lethal or steril ising path ogens . (R-O)
Conditions th at all ow a pl ague :
1. Wi despread favourable cli matic cond itions re su lting in dispersed populations of prey and
predators. (P-R)
2. A long delay in a density-dependent increase in th e preva lence of pathogens or the prevalence is
independent of dens ity. (R-O)

84

The probability of a plague in the (following)
autumn was found to depend on (i) the
difference between grain production in the
current year and that two years earlier, (ii)
the difference between the November and
the October rainfall for the current year, and
(iii) the autumn rainfall in the current year.
The model accounts for 41 % of the variation
in plague occurrence. A plausible
mechanism by which each of these factors
could affect mouse populations has been
suggested. The requirement in the Turretfield
model (see below) for mid-summer rains was
not supported even when the data set was
restricted to sites with predominantly red-
brown earths. However, there was a
significant contribution from autumn rain in
the year immediately preceding a plague
which matches, to some extent, the revised
model for the Victorian Mallee (below). In
contrast to the MIA model (below), autumn
rains from earlier years appeared to have no
effect.
If projected estimates of crop yield are
available, the South Australia regional model
can be used towards the end of winter to
predict the likelihood of a mouse plague in
the following autumn. This may provide
adequate time for preventative control
measures to be implemented if, for example,
farmers have access to registered in-crop
rodenticides. The data requirements are the
autumn (March, April and May) rainfall, the
average November-October rainfall, the
harvest yield from two years before and the
preliminary estimate for the current year.
These data provide the three variables for
the model. The preliminary harvest
estimates are subjective judgements made
by district agronomists based on (i) autumn
rainfall which determines sowing time, (ii)
Models for Predicting Plagues in Australia
total (over-winter) rainfall during the
growing season [Cornish et al. (1980), as
reported in Mutze et al. (1990), found that
total rainfall from April to October accounts
for 80% of variation in crop yield], and (iii)
the amount of rainfall in September and
October to 'finish-off' crops. The model can
be used to predict plague probabilities for
each locality and a high probability is taken
as an indicator that additional evidence,
such as more detailed crop and rainfall
records, should be examined.
The model's predictions were compared
with the results of an intensive trapping
program from 1980 to 1990 (Mutze 1991).
Plagues were predicted for 1980,1984 and
1985 but not in other years. The only plagues
that occurred in South Australia during this
period were in 1980 and 1984; the failure of
mice to respond to generally favourable
conditions in 1984-85 was due probably to a
late break in a period of low rainfall during
the winter of 1985.
District and small-scale models
Turretfield model
Newsome (1969a,b, 1970, 1971) conducted
the first extensive Australian study of the
dynamics of wild mouse populations at
Turretfield in South Australia (Figure 1).
Mice were permanent residents of small
patches of favourable habitat in the
landscape; reedbeds in the case of this study
area. Mice colonised crops in early summer
but could not over-winter there due to
waterlogging of the clayey soils. According
to the modeL a plague of mice occurs as a
direct result of an unusual sequence of
events: (i) good winter rains to provide an
adequate food supply through to the
85
Ecologically-based Rodent Management
following autumn and to keep the sub-soil
moist over summer; (ii) a hot summer to
crack the soil allowing mice access to nesting
sites in the moist sub-soil; and (iii) mid-
summer rain to allow mice to burrow and
breed throughout summer. The data suggest
it is possible for mice to increase to plague
numbers in three to five months. This
conclusion was supported by an experiment
in which free-fed mice reached densities
well in excess of those observed in plagues.
The model was later modified by
Newsome and Corbett (1975) to take into
account the effects of predation. In the
revised version, predators can delay by one
year the build-up of mouse populations
generated by a pulse of favourable
conditions. This is consistent with the
predation-regulation model (below) proposed
by Sinclair et al. (1990). However the
suggestion by Newsome and Corbett that
predators may be responsible for the failure
of mouse plagues to persist under
favourable conditions is contrary to the
predation-regulation model.
Two features distinguish this model from
some of the later models. Firstly, the
relationship between the availability of
breeding sites for mice and soil moisture
over summer restricts the applicability of the
results to areas of red-brown earths
interspersed with patches of heavy cracking
soils. Mutze et al. (1990) argued that the
TurretJield model was relevant to only these
fairly restricted parts of the South Australian
wheatlands. Secondly, there is an apparent
anomaly in the lead-time (the time from the
first events triggering a plague to the plague
itself): less than six months for the area
characterised by Turretfield and up to two
years elsewhere. Despite the ease for mice to
86
burrow in sandy soils, plagues do not
appear to develop immediately in areas with
these soil types whenever good winter rains
are followed by mid-summer rains.
The apparent ability of mice at
Turretfield to reach plague densities within
five months implies little prospect for long-
term forecasting of outbreaks. However, the
model demonstrated the role of reedbeds as
the source of mice that colonise crops, and
for these land systems, the strategy of
selectively targeting minor refuge habitats
may be effective if control was conducted
routinely.
Murrumbidgee Irrigation Area (MIA)
model
A 'triphasic' model of mouse plagues was
developed by Redhead (1982) during an
intensive four-year study in the MIA in New
South Wales (Figure 1). The model is
complex and includes both intrinsic (e.g.
spacing behaviour) and extrinsic (e.g. food
quality) factors that influence mouse
population dynamics (Redhead et al. 1985;
Redhead and Singleton 1988a). According to
this model, the plague trigger (phase 1) is
above average autumn rains two years prior
to the outbreak, which extends the breeding
season into winter in refuge habitats by
providing high quality food. The effect of
food quality was experimentally
demonstrated by Bomford (1987a,b,c) and
Bomford and Redhead (1987). There is high
productivity of mice in the following
summer (phase 2) and mice disperse from
the refuges into other areas made favourable
by the earlier autumn rains. At the start of
the breeding season immediately prior to the
outbreak (phase 3), there is an abnormally
high abundance of females in a wide range

of habitat types ('induced-donor' habitats).
Provided no unusual factors intervene to
impede breeding, a plague will develop over
summer.
Redhead (1982, 1987) used a numerical
simulation model (SIMAD) to show that
between-year differences in the mean litter
size and the observed size of the initial
population could explain the variation in the
increase period for each of the three phases.
However the data set for litter sizes is
limited and the model does not allow for
changes in litter size during a breeding
season. The results of the model emphasise
the importance of between-year variability
in breeding performance in the MIA
compared to the Darling Downs model
(below) where the population is assumed to
increase at the same rate each year.
In the MIA model, the plague trigger
occurs in the autumn two years prior to a
plague. In comparing this to the NSW and
Victoria regional model, Redhead (1982)
observed that there appeared to be a
relationship between the residual mass [the
accumulated difference between the long-
term average and the actual monthly rainfall
(Poley 1957)1increasing through winter and
spring and a plague two years later. Neither
this relationship nor the prior-drought
hypothesis of Saunders and Giles (1977) was
tested statistically, but if the relationship
suggested by Redhead is true, then the
importance of a prior drought is
questionable. The residual mass will
increase with above average rains breaking a
drought (invariably in winter) or if there are
simply above average winter-spring rains.
The extended build-up period (phases 1
and 2) prior to the plague year is in contrast
to other models where mice can reach plague
Models for Predicting Plagues in Australia
densities in one season. The difference may
lie in the need for mice to colonise 'induced-
donor' habitats in phase 2, which depends
ultimately on the mix of crops and year-
round refuge habitats in the landscape. In
both the preckltion-regulation model and the
MIA model, mice are held in a low-density
state by spatial, density-dependent
processes. However in the predation-
regulation model, phases 1 and 2 were simply
classed as the predator-regulated state and
not necessarily as essential precursors to
phase 3, the outbreak state.
In 1983-84, Boonstra and Redhead (1994)
tested the hypotheses relating to phases 1 and
2 in the triphasic model. Specifically, these
were that a tight social structure during the
extended breeding season in phase 1 should
result in high dispersal rates for mice,
presumably into the 'induced-donor' habitats
outside refuge areas, and that there should be
a disproportionate abundance of female mice
at the end of phase 2. The weather conditions
prior to this study included a severe drought
in 1982 and above-average rainfall in the
autumn of 1983, both of which have been
considered important precursors to a plague
(Saunders and Giles 1977; Redhead 1982). The
data on fecundity rates, dispersal rates, sex
ratios and testosterone levels showed that (i)
it was unlikely that social organisation had
modified dispersal, (ii) the expected changes
in breeding performance in phase 2 did not
occur, and (Hi) there was no bias towards
females in the sex ratio at the end of phase 2.
The conclusions were that, for irrigated rice
crops, plagues could develop much faster
than previously envisaged by Redhead (1982)
and the <12-month time frame for the
increase in mouse abundance in 1984 was
similar to that suggested in other models.
87
Ecologically-based Rodent Management
However, the reasons for the lack of a
subsequent plague are not clear and several
hypotheses were proposed by Boonstra and
Redhead (1994). The most likely explanations
appear to be associated with the drought that
affected the surrounding dryland farms in
1984. Non-irrigated areas may have acted as a
sink for dispersing mice. Alternatively,
mouse populations may have been regulated
by predation, Le. by highly mobile raptors
moving away from dryer areas to concentrate
around irrigated crops, as in the predation-
regulation model proposed by Sinclair et al.
(1990). During an earlier mouse plague in this
area in 1979-80, Davey and Fullagar (1986)
noted a large increase in the abundance of
three species of mouse-eating raptors, the
Australian kestrel (Faleo cenchroides), the
brown falcon (Fa lea berigora) and the black-
shouldered kite (Elanus notatus).
The relationship between the
development of mouse plagues and the
timing of management actions has been
examined in detail by Redhead and
Singleton (1988b) and Singleton and
Redhead (1989) and lead to their PICA
(Predict, Inform, Control, Assess) strategy.
Their analysis demonstrated that the current
lack of registered in-crop rodenticides has
forced a series of time delays into the ability
of farmers to respond to plague warnings.
The delays result from the processes
necessary to obtain permission to use
rodenticides. However, if suitable control
techniques can be developed and be ready
on demand, predictions with a time frame of
less than 12 months may be sufficient for
farmers to control mice effectively.
88
Victorian Mallee model
A long-term demographic study of mice
beginning in 1983 in northwest Victoria
(Figure 1) has restructured the MIA model for
the Mallee wheatlands (Singleton 1989). In
the Victorian MalIee, the development of a
mouse plague may occur in the breeding
season immediately following high autumn
or winter rainfall, potentially more rapidly
after a 'trigger' than in the MIA, but because
of the different soil characteristics (Singleton
1989; Singleton and Redhead 1989) plague
development may be more sensitive to the
sequence of rainfall events. For example, a
plague in 1988 was less severe than expected
despite an increase in mouse numbers
following initial favourable conditions. As in
the Turretfield and MIA models, landscape
heterogeneity and the role of refuge habitats
appears to be important in the population
dynamics of mice in the Victorian MalIee.
For example, Singleton (1989) reported
significant temporal differences between
habitats in the increase phase of the 1984
plague.
The time taken for mice to reach plague
proportions in the Victorian Mallee in 1984
was similar to that observed by Newsome
(1969a) at Turretfield. Although the 1984
plague occurred 15 months after a drought, in
accordance with the NSW and Victoria regional
model, no data were collected to validate the
suggestion of Saunders and Giles (1977) that
predator regulation was the process
responsible for this time However, the
prevalence of macroparasites was
independent of mouse densities over a 2.5
year study prior to, and during, the plague
(Singleton 1987). The data suggest that the
macroparasites recorded by Singleton were
unlikely to regulate mouse numbers.

Based on the Victorian Mallee model, a
series of trapping surveys following high
autumn or winter rainfall could be used to
verify a prediction of conditions conducive
for a mouse plague. A suitable protocol
would be as follows.
Ill- (i) Trap in the second week in September to
determine Ca) the start of breeding, (b) litter
size [litters are larger in the build up to
plagues (Singleton and Redhead 1990a)],
and (c) the size of the breeding popula tion.
The data should be collected from
i 'donor I refuge' (e.g. fenceIines) and crop
habitats. If the data from September show
no breeding and low populations, then
there is a low probability of a plague and no
further monitoring is required until the
following September. Conversely, an early
start to breeding, large litters and a large
breeding population are conditions
favouring a plague and further monitoring
is required (step ii).
Ill- (ii) Trap in November to determine the
percentage of females breeding and the
litter size. High numbers at this time
indicate a high chance of a plague next
autumn; moderate numbers may be a
precursor to a plague 18 months away.
Ill- (iii) If data from November confirm that a
plague is expected in the next six months, a
third trapping session may be required to
detect the rapid build-up in mouse
numbers over the January - February
March period.
Ill- (iv) Moderate numbers in November (and
no plague within six months) indicate that
trapping the following September is
important.
Models for Predicting Plagues in Australia
The application of the PICA management
strategy (Singleton and Redhead 1989) is
similar in the Mallee and the MIA, except
that plague development may be more rapid
in the former. The timing of control
operations may be more constrained in the
Mallee than the MIA and this may be a
problem with future applications of a
biocontrol agent such as Capillaria hepatica
(see below).
Darling Downs model
The model is based on demographic data
collected over 12 years from a standard set of
trapping sites in the central Darling Downs
in Queensland (Figure 1) (Cantrill 1992).
Pooled data from all major habitat types
(crops, verges, fallow etc.) show a regular
annual cycle in mouse abundance with no
clear separation into plague and non-plague
years. There are minimal between-year
differences in (1) the rate of increase of the
mouse population over summer and
autumn, (1i) the onset of the main breeding
season, and (iii) the months of maximum
population denSity. The most important
factor that can interrupt the cycle of
abundance and affect the peak density of
mice in May and June is the duration of the
low abundance phase (defined as less than
1% trapping success) the previous spring. In
one year, a sequence of flooding rains,
probably increasing nestling mortality,
delayed the annual build-up in mouse
numbers. A second source of between-year
variation is a density-dependent decline in
the over-wintering population.
Trapping data are required from fixed
trapping sites for two, preferably three,
periods of the year to generate predictions
with the Darling Downs model. These are (i) in
89
Ecologically-based Rodent Management
Mayor June-to test the prediction from the
previous year and to predict the size of the
mouse population at the onset of the main
breeding season in spring, (ii) in
September-to determine the size of the
initial spring breeding population, and (Hi)
in the period from October to December-to
determine the starting time for the increase
phase in mouse abundance. Rainfall data
from spring and early summer can be used if
data from the third trapping period are not
available. The model is based on an
established trapping protocol and its
applicability to data collected elsewhere, or
for a different mix of land uses, is unknown.
The model can provide forecasts 12, 8 and
5 months in advance of the peak mouse
abundance in May. This translates into
warning of high mouse numbers 9,5 and 2
months in advance of the time when damage
due to mice is usually reported. The model
also suggests that any farming practices,
such as minimum tillage, that enhance over-
winter survival of mice may lead to a
relatively large population at the onset of
breeding the following spring. This would
generate a forecast for high mouse numbers
the following year with the result that
farmers could be locked into annual mouse
control.
Macquarie Valley model
In contrast to the 12-24 month warning-time
recommended by Redhead and Singleton
(1988a), Twigg and Kay (1994) suggested
that decisions by fanners for managing pests
would be based on relatively short-term
predictions of the order of three months.
They developed a series of models, based on
linear multiple regression analysis, for
irrigated summer crops such as soybeans,
90
sorghum, cotton and maize, in the
Macquarie valley of New South Wales
(Figure 1). Soil types in the irrigation area are
self-mulching clays to deep red clays that
crack on drying to produce refuge sites for
mice similar to those in patches of black
cracking soils at the Turretfield site studied
by Newsome (l969a).
Data were collected over three years that
included two floods. The best models
explained 68% of the variance in the index of
mouse abundance and 53% of the variance in
the abundance with seasonal trends
removed. Significant coefficients were found
for the mean daily range in temperature for
each month, the mean minimum
temperature, the mean maximum
temperature and the total monthly rainfall in
the one or two months prior to trapping.
Twigg and Kay (1994) also estimated the size
of the seed bank, soil moisture and soil
hardness, the number of crack." in the soil
and their size distribution, and indices of the
structure and biomass of the vegetation.
They found that the component of the seed
bank from barnyard grass (Echinochloa crus-
galli), rye grass (Lolium rigidum) and wild
oats (Avena jatua) explained 70% of the
variance in mouse abundance with small
additional improvements contributed by
indices of soil cracking and vegetation.
However, the regression based on total seed
bank was less satisfactory. The vegetation
data provided a link between climatic
variables and mouse abundance but a direct
effect of rainfall and temperature on
recruitment was not supported by the
models. In the model for recruitment, the
proportion of adult females lactating or
pregnant depended on the seed bank and
the distance to the closest summer crop.

The model for the abundance of mice in
irrigated summer crops provides a useful,
short-term predictive tool based on readily
accessible climatic data. Twigg and Kay
(1994) suggested that it is desirable to use
routine surveys of mouse abundance to
support their model's predictions, however
a survey protocol suitable for use by farmers
was not specified. The data from this study
support the conclusions from elsewhere in
Australia that the management of mice, or
their food supply, in refuge habitats such as
roadside verges and fencelines should help
to reduce the damage caused by mice.
Simplified process models
Predation-regulation model
The model proposed by Sinclair et al. (1990)
focuses on the third of the major extrinsic
mechanisms-food supply, shelter,
predation and disease - which may be
responsible for regulating mouse
populations. It is based on data collected
over two years on a summer-irrigated cereal
farm in the MIA (Figure 1). Although mice
increased in abundance each year over the
spring-summer period, years were divided
into plague or non-plague categories. In
non-plague years, raptors-primarily
black-shouldered kites, Australian kestrels,
brown falcons and brown goshawks
(Accipter fasciatus) and possibly
mammalian predators (foxes - Vulpes vulpes
and feral cats - Felis catus) could have been
responsible for regulating mice at low
densities despite the apparently favourable
environmental conditions for mice on the
farm. Also a second density-dependent
factor, the effects of a pathogen thought to be
Actinobacillus moniliformis was implicated
Models for Predicting Plagues in Australia
although the data are extremely limited. In
the one year when a plague occurred on the
farm, mice were abundant over a wide
region due to exceptionally favourable
conditions over the preceding months, and
the limited number of predators on the
study area failed to regulate mice. Sinclair et
al. (1990) suggested that the plague on the
study farm was caused by a combination of
wide dispersal of predators and enhanced
breeding performance of mice which
together allowed mice to escape predator-
regulation.
The analysis of Sinclair et al. (1990) can be
generalised to the following multi-state
model, which is analogous to a model for
rabbit plagues in semi-arid Australia (Pech et
aL 1992,PechetaL 1995,Pechand Hood 1998)
and is relevant to regions where predation
can be a major mortality factor for mice.
... (i) There is a plague trigger (a period of
high rainfall) which provides a finite
amount of food that ultimately is
exhausted by mice. The time to depletion of
the pulse of food is a function of the size of
the trigger, i.e. the duration of the
favourable conditions. A large pulse could
extend over two years, a moderate pulse
might last only one year, and a smaller one
still should not result in an outbreak,
merely a small seasonal increase in the
abundance of mice. It should be possible to
predict the occurrence of the pulses of food
as accurately as the long-range weather
forecasts.
... (ii) The triggerresults in high reproduction
by mice that can result in an escape from
predator-regulation. If there are many
predators in an area and no trigger or only
a small trigger, then only a minor increase
91
Ecologically-based Rodent Management
in mice should occur. If there are few
predators, then an outbreak occurs
immediately following a trigger. The
model predicts that if a trigger is not
followed by an outbreak, then predators
should have been present in large
numbers.
~ (iii) The trigger generates a finite amount of
food in excess of that normally present. If,
and only if, mice escape predator-
regulation can they reach a high density
state determined by the new food supply.
The abundant food eventually disappears,
or a drought arrives, which resets food to a
low level and the system collapses back to a
situation where predators can take over
again.
The predation-regulation model is based on
a very limited data set and is best viewed as
an hypothesis for more extensive testing.
Although the model has little to offer for
predicting the likelihood of a plague, it does
identify a key process that can be affected by
management. No action should be taken
which would deplete raptor populations,
and mouse control techniques should be
used to ensure that winter populations of
mice stay low, i.e. within the range where
regulation by predators is possible.
Pathogen-regulation model
Information on the prevalence and
distribution of pathogens in wild mouse
populations in south-eastern Australia is
summarised in Redhead (1982), Singleton
(1987), Singleton et al. (1991), Singleton et al.
(1993) and Smith et al. (1993). There is some
field-based evidence that endemic
pathogens may be responsible for regulating
mouse populations (Sinclair et al. 1990),
however most effort has been directed
92
toward a search for organisms which could
be introduced into mouse populations either
as conventional biological control agents
(Singleton and Spratt 1990) or as vectors
engineered to induce infertility in infected
mice (Singleton and Redhead 1990b; Shellam
1994). Although the conditions for a
candidate biocontrol agent are exacting
(Spratt 1990), pathogens have considerable
appeal because of their likely host-
specificity, the absence of toxic residues,
potential economic advantages and their
possible compatibility with current
management practices and land use. In
addition, Australia has the advantage that
local populations of M. domesticus lack some
of the mouse-specific parasites found
overseas (Singleton and Redhead 1990a).
Following a major review, Barker and
Singleton (1987) concluded that the liver
nematode Capillaria hepatica showed promise
as a biological control agent for mice.
Laboratory studies had established the
conditions for successful transmission
between mice (Spratt and Singleton 1986,
1987) and had shown that infection
depressed the productivity of female mice to
a level that may be sufficient to prevent
plagues (Singleton and Spratt 1986). As well,
surveys demonstrated that mice in plague-
prone areas had no prior exposure to the
parasite despite it being recorded at a range
of sites in coastal south-eastern Australia
(Singleton et al. 1991). McCallum and
Singleton (1989) and Singleton and
McCallum (1990) modelled the likely impact
of C. hepatica on mouse population dynamics
and concluded that it had the potential to
significantly reduce the density of mice
below infection-free levels. However, it was
not clear whether time delays in the host-

pathogen cycle might negate its use for
tactical release (time- and area-limited and
with 'rapid' effect) (McCallum 1993). The
effectiveness of C hepatica as a biocontrol
agent was tested in pen experiments (Barker
et al. 1991) and two large-scale, replicated
field experiments-the first in the Darling
Downs in 1992-93 and the second in the
Victorian Mallee in 1993-94 (Figure 1). The
field sites were chosen to be representative
of the farming systems in each region.
Three releases were conducted in the
Darling Downs: (i) low density, non-
breeding mouse populations in winter, (H)
low density, breeding mouse populations in
summer, and (Hi) high density, non-
breeding mouse populations in winter
(Singleton et al. 1995). Embryonated and
unembryonated eggs were released on four
sites using a combination of baits and direct
infection of mice (oral). The treated sites
were interspersed with three control sites.
The parasite appeared to persist for <5
months after the first release in winter of
1992, for at least 2-4 months with low
prevalence after the release in the summer of
1993, and persistence was uncertain after the
winter release in 1993 because the
abundance of mice declined to levels where
it was not possible to trap adequate samples.
Following all three releases, little or no
significant difference between treated and
untreated sites was detected in the age
structure of mouse populations or their
abundance, breeding performance or
survivaL An intensive trapping protocol
failed to detect any transfer of the parasite
beyond the experimentally infected areas.
The minimal impact of C hepatica was
attributed to inadequate dosage rates and
adverse climatic conditions leading to
Models for Predicting Plagues in Australia
periods of low densities of mice, poor
survival rates for mice, little fidelity in the
use of burrows and loss of access by mice to
eggs deposited in the cracking clay soils.
Four treated and three untreated
cereal! sheep farms were used for the
experiment in the Victorian Mallee
(Singleton and Chambers 1996). The parasite
was released in September 1993, two months
prior to a period of sustained increase in the
abundance of mice. About 60,000 mice, or
10% of the populations on the treated sites,
were dosed with embryonated and
unembryonated eggs. The results from the
increase phase in mouse density (November
1993 to mid-1994) showed that C hepatica
persisted for approximately eight months
but with little or no effect on the 28-day
survival rate and the fecundity or the
abundance of mice, but the treatment
appeared to temporarily delay the increase
in mouse density during the early part of the
breeding season. The causes for the poor
performance of C hepatica as a biocontrol
agent are uncertain but probably include
low survival of eggs during hot, dry weather
and delays in the life cycle of the parasite
that prevent it from regulating rapidly
increasing mouse populations.
A pathogen-regulation model based on
C. hepatica is still in the early stages of
development despite intensive efforts to
collect epidemiological data in laboratory,
pen and field experiments. To be
implemented, the timing and conditions for
release of the parasite will need to be
specified, along with the information
required to determine those conditions. This
will determine the extent to which other
control techniques might be needed to form
an integrated control strategy for mice. In
93
Ecologically-based Rodent Management
addition, the model should specify when
and what information is required to monitor
the performance of the technique.
Although the early promise ofbiocontrol
using C. hepatica has not been realised, a
recent feasibility study suggests that virally-
vectored immunocontraception may
provide a viable alternative (Chambers et a1.
1997). The fertility of laboratory mice has
been impaired successfully using ectromelia
virus as a model (Jackson et a1. 1998) for the
development of a genetically engineered
irnmunocontraceptive strain of mouse
cytomegalovirus (Shellam 1994).
Comparison of current models
There is a wealth of data, expert knowledge
and experience, published and unpublished,
relating to the formation of mouse plagues in
Australia. The resulting models,
summarised in Box 1, contain elements of
the key extrinsic factors likely to affect mice:
food availability, access to shelter and nest
sites, predation, disease and landscape
heterogeneity. In most cases food
availability is estimated from climatic data
although the South Australian regional model
can include agronomic estimates of crop
yields if they are available. No model
includes a protocol for collecting direct
estimates of the amount of food accessible to
mice. The models for areas with heavy soils,
Turretfield, Darling Downs and Macquarie
Valley, include the effect of rainfall, or lack of
rain at critical times, on the abundance of
nesting and shelter sites in sub-surface
cracks. In these and in other areas with
lighter soils, secure access to refuge sites can
depend on the frequency of disturbance
caused by the prevailing system of crop
rotations and periods of fallow. For this
94
reason the recent trend to conservation
farming, with emphasis on stubble retention
and minimum tillage, may exacerbate
problems with mice (Brown et a1. 1998;
Singleton and Brown 1998).
Predation and disease, two factors that
are likely to be density-dependent, have
been measured directly or their influence
inferred as a possible explanation of mouse
population dynamics. However the
predation-regulation model (Sinclair et a1. 1990)
is the only example where data on predation
have been analysed within the framework of
predator-prey theory. Supporting evidence
from Davey and Fullagar (1986), Kay et a1.
(1994) and Twigg and Kay (1994) reinforce
the need for field experiments to determine
the circumstances under which predation
can regulate mouse populations. For
example, Sinclair et a1. (1990) hypothesised
that predator-regulation may be effective
only when predation pressure is
concentrated on localised patches of the
landscape.
Although Smith et a1. (1993) and
Singleton et at. (1993) identified many
endemic diseases of mice in south-eastern
Australia there is little evidence that any of
these play a significant role in regulating
mouse populations, with the possible
exception of a mouse parvovirus. The effects
of a disease, attributed to Actinobacillus by
SincIair et a1. (1990), was similarly equivocaL
The pathogenic agent was probably Strepto-
bacillus moniliformis which has been recorded
at low prevalence in fluctuating mouse
populations in the Darling Downs (TayIor et
a1. 1994). With current technology there
appears to be little scope for large-scale
manipulative experiments to test for any
regulatory effects of an endemic disease. The

alternative strategy of using translocated
pathogenic agents, or engineered sterilising
agents, offers the prospect of well-designed
experimental trials.
The influence of landscape heterogeneity
is inherent in all the published models and
this factor is thought to interact with most, if
not all, potential regulatory processes.
Different habitat types provide temporal
asynchrony in food supply, refugia during
drought or when there is disturbance from
farming activities, and may result in either
concentrated or dispersed prey populations
for raptors and terrestrial predators. At this
stage there is very limited information on
dispersal and other movements between
habitats by mice (see for example, Newsome
et a1. 1982; Boonstra and Redhead 1994;
Krebs et a1. 1995b) on which to base
spatially-explicit models for mouse plagues.
Krebs et a1. (1995a) reviewed the case for
the intrinsic regulation of mouse
populations. They urged an analogy with
models for the cyclic changes in populations
of some rodent species in the Northern
Hemisphere and proposed two variants of
the Chitty hypothesis, with low, increasing,
high and declining populations of mice
characterised by differences in the ratio of
aggressive and docile phenotypes. There are
at least two problems with this view, one
practical and the other philosophical. Figure
2b illustrates the difficulty of obtaining
sample sizes sufficient to extract the simplest
demographic parameters - abundance and
rate of increase - during periods of low
mouse numbers. Even if the two
phenotypes, or their effects, can be
distinguished in the laboratory experiments
suggested by Krebs et a1. (1995a), the
resources required to measure aspects of the
Models for Predicting Plagues in Australia
social system of low-density mouse
populations in the field are unlikely to be
realised. In the two hypotheses proposed by
Krebs et a1. (1995a), social organisation acts
as a filter on the response of mice to extrinsic
factors. All the suggested social differences
appear plausible (docile versus "a,err,'"";;",,,
behaviour, weak versus strong territorial
behaviour, and open versus closed social
systems) but the causal factors in population
dynamiCS may be still extrinsic. It is
probably more important to begin by testing
rigorously the most parsimonious
hypotheses: Boonstra and Redhead (1994)
were unable to find evidence to support
aspects of intrinsic regulation in the MIA
model (Redhead 1982), and some of the
extrinsic mechanisms underlying the current
models appear amenable to direct
experimental manipulation.
Practical application of
current models
At present three models (the South Australia
regional model and the models for the Darling
Downs and Macquarie Valley) and possibly a
fourth (the NSW and Victoria regional model)
could be used unambiguously by people
other than those who produced them. Most
models imply that reliable forecasts can be
provided only for the medium (less than 12
months) or short term (less than 6 months).
This imposes some constraints on the timing
of effective control options, especially those
relying on biocontrol agents. All models
have been developed using location-specific
data and, to date, none have directly
addressed the extent of the geographic range
of their predictions. Some models - for
example, the two regional models - should
apply across state boundaries to districts
95
Ecologically-based Rodent Management
with similar soils, climate and agricultural
practices. Other models are limited by
attributes such as soil type or cropping
regime. With the exception of the Darling
Downs model, none of the existing models
relate the density of mice, or the effect of
controls, to the extent of damage caused by
mice. This is an essential requirement for
future models of mouse plagues.
A prerequisite for the application of a
model is that the data and the methods used
to generate a prediction are clearly specified.
For most models the protocols for data
collection have not been specified in
sufficient detaiL In future such protocols will
need to be designed and tested for areas
other than the localities where the models
were developed.
One approach to improving the value of
existing models is to include their predictive
capability with other expert knowledge in
'decision-support systems' (Norton 1988).
An example is the computer-based expert
system that can provide short-, medium-
and long-term predictions based on the
Darling Downs model (Cantrill1992). A
decision support system, MOUSER, is under
development for the Victorian Mallee
(Brown et al. 1998). This software provides
advice on how to achieve effective mouse
control by modifying farming activities. The
aim is to include interactive models for
plarming mouse-control campaigns in future
editions of MOUSER. However, all the
existing regional and district models focus
on predicting the occurrence of plagues or
the abundance of mice rather than their rates
of increase. This means that none of the
models are capable of assessing the impacts
of control programs in an interactive way.
96
MODIFIED MODEL FOR MOUSE
PLAGUES IN THE VICTORIAN MALLEE
Background and data
The aim of the following analysis is to
construct a quantitative model, based on the
conceptual model of Singleton (1989), to
explain the observed rates of increase of
mice over the last two decades in the
Victorian Mallee. The ultimate purpose is to
use quantitative models to predict when
mouse plagues will occur and to assess the
effectiveness of a range of control
techniques, including fertility controt for
managing mice in the Mallee region.
Mice were live-trapped at intervals of
approximately six weeks from 1983 to 1997
in several habitats either on the Mallee
Research Station or on nearby farms in
northwest Victoria (35.085, 142.02E). The
region has a Mediterranean-type climate
(Figure 2a) and is used for cropping and
livestock production. Details of the trapping
protocol are given in Singleton (1989).
Although most trapping sessions were
conducted over three nights and all mice
were tagged, the number of recaptures was
generally <15%. As well, the trapping grids
were moved periodically to cater for the
rotation of crops and pasture so that
capture-mark-recapture techniques did not
provide the best estimates of abundance.
Instead, the number of captures, adjusted to
take into account trap saturation, was used
as an index of abundance. This required
transforming the proportion, P, of traps
catching mice per night (the frequency of
captures) to the number of animals that
would have been caught per trap if the traps
were capable of multiple captures (an index
of the density of mice). Then the adjusted
 Models for Predicting Plagues in Australia

120 (a)

>,E 80
:cS
.....
c:=
OCll
:if: 1: 40
'Cij
....
o
1.0
15-0
c: OJ
0=
'tt;:::
o (/) 05
,
Cl. Cl.
e~
0.. .....
0.0

1.0
Cl.
e(.) ><
Cil
OJ c:
-ll 0,5
..c'-
S
0.0

(d)
3000

2000

1000

o
83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98
Year
Rgure2.
(a) Monthly rainfall (mm) at the Mallee Research Station (MRS), Victoria. (b) Smoothed abundance of mice
(-), indexed by the proportion of traps filled at the MRS and nearby farms. Data for 1983-85 are from
Singleton (1989), for 1993-94 from Singleton and Chambers (1996) and for 1993-97 from Singleton and
Brown (1998). The data are shown for each night of trapping with the size of the symbols Co) scaled
according to the trapping effort (minimum = 10, maximum = 355, median = 242 trap-nights). (c) Modelled
food from wheat crops (equation 1). with the mid-summer amplitude proportional to the total rainfall for the
winter/spring period. (d) The estimated total seed biomass (kg/ha) from annual grasses and medic using
the model GrassGro (Moore et al. 1997). Parameter settings for the GrassGro model are shown in Table 1.

97
Ecologically-based Rodent Management
density, N, is -In(l- p) (Caughley 1977). We
would expect this density to be affected
proportionally by various factors and
therefore the regression model would be
additive on In(N) = In[-ln(l- p)].
The data for the proportion, p, are
somewhat noisy (Figure 2b) and small
fluctuations, particularly during periods of
very low-density, are unlikely to be
important in the overall population
dynamics. Therefore the data were
smoothed in a generalised linear modelling
framework by fitting a spline in time as an
explanatory variable, with the
complementary log-log link function. The
model was fitted in S-PLUS (1997) using a
smoothing spline with 20,30, 50 and 70
degrees of freedom (d.f.). The residual
deviances (and the residual dJ.) obtained
were 9067 (515), 5921 (505),4123 (485), and
3276 (465), respectively. There was a
substantial reduction in the residual
deviance for a model with 30 d.f. compared
to 20 d.f. which was reflected in the better
tracking of the mouse plague peaks in the
data. Fitting splines with more than 30 dJ.
did not lead to a marked improvement in
the fit to the main peaks and started to
generate spurious peaks in the periods
between major plagues. The inclusion of
seasonal (sine and cosine) terms did not
improve the model which is not surprising
since mouse plagues do not occur on a
regular annual cycle. The slope of the
trajectory for In(N) is the instantaneous rate
of increase of the mouse population. Using
S-PLUS, the estimated first derivatives were
obtained from the fitted model at 40.55 day
intervals (1/9 th of a year). This
approximates the mean interval between
trapping sessions for mice.
98
During 1984 and 1985, the diet of mice
was measured at a cereal farm not far
(approximately 25 km) from the Mallee
Research Station (Tann et a1. 1991). The
findings of this study showed that the main
dietary components were monocotyledon
seed (primarily wheat- Triticum aestivum
and some grasses such as brome - Bromus
spp., barley - Hordeum leporinum, wild
oats - Avena fatua and ryegrass - Lolium
rigidum) and dicotyledon seed such as
medic - Medicago spp. and Chenopodaceae
species. The proportion of the major
component, monocotyledon seed, showed
strong seasonal variation, rising from a
minimum around July in mid-winter to a
peak at harvest six months later. Post-
harvest, the consumption of monocotyledon
seed declined to a low point in the following
winter. Bomford (1987a) found similar
changes in the diet of mice occupying a
wheatfield on an irrigated cereal farm in
central New South Wales. In the Mallee, the
exception to this pattern was a brief increase
in wheat consumption at the time of sowing
in the early winter of 1985, presumably
because mice were unearthing the sown
grain. No direct measurements of the
availability of food items for mice were
reported by Tann et a1. (1991) and other data
for the period from 1983 to 1997 are very
limited. Consequently existing models were
used to estimate the relative availability of
monocotyledon and dicotyledon seed.
French and Schultz (1984) used data for 61
sites in South Australia from 1964-1975 to
establish a linear relationship between wheat
yield and the rainfall summed over the period
from April to October, RkO" They found that
65% of the variation in wheat yield was
explained by the rainfall data. Cornish et a1.

(1980), using a similar model, accounted for
80% of the variation in wheat yield and Seif
and Pedersen (1978) used spring rainfall to
account for 86% of the variation in yield in
central New South Wales. The data set used
by French and Schultz (1984) included areas
with climate and soils similar to the Victorian
Mallee region and the validity of
extrapolating from their results was tested by
comparing April to October rainfall with crop
vields from the Mallee Research Station for
"
the period 1984-1997. The regression
accounts for 69% of the variance in the
reported values 3) indicating that, as
in the French and Schultz model, RA _O also
provides an appropriate index of the wheat
harvest in mid to late December at the Mallee
Research Station. This index does not provide
estimates of seed or crop biomass during the
period of crop establishment and maturation.
However, some measurements of spilled
grain have been made immediately post-
harvest and at irregular intervals for up to
four months (G.R. Singleton, unpublished
data). These data show a steady decline in the
quantity of spilled grain post-harvest with
little or none remaining on the ground by mid
to late autumn. Based on this information,
and the observed seasonal changes in
monocotyledon seed component in the diet of
mice (Tann et a1. 1991; Bomford 1987a), the
within-year variation in the food from wheat
crops was modelled as a sinusoidal function,
[1 + cos(2:n:(T /9)]/2, where T is the time in
ninths of a year (i.e. intervals of 40.55 days)
since the first of January. Then an index of the
amount of food from crops is:
Models for Predicting Plagues in Australia
where the rainfall has been expressed as a
proportion of the maximum observed value
of 348 mm for the winter / spring period. The
values of this index for the years 1983 to 1998
are shown in Figure 2c. A transformed
version of W that helps to distinguish
between average and bumper crops was used
later for modelling the rate of increase of the
mouse population.
3 Yield = 0.0087 Rain 0.1648
Ft = 0.69
<?
.<= 2
2:..
"0
a;
's;:,
Ol
<ll
.<=
~
o ~--~----~----~----~--~
100 150 200 250 300 350
Rain (mm from April to October)
Figure 3.
Comparison of the summer wheat yields
(tonnes/ha) at the Mallee Research Station with
total rainfall (mm) over the preceding period from
April to October. French and Schultz (1984)
established a linear relationship between April to
October rainfall and wheat yield using data from
South Australia.
Th.e model, GrassGro (Moore et a1. 1997),
was used to estimate the biomass of ripe and
unripe seed,S, produced by annual grasses
and medic (Figure 2d). It requires detailed
information on stocking regimes, soil type
and climate (Table 1). Although areas
accessible to mice included grazed pastures
and ungrazed grass along fencelines and
roadside verges, only the estimates for
grazed pasture were used. This is the
predominant non-crop habitat in the
Victorian Mallee and there are relatively
99
Ecologically-based Rodent Management
minor differences in the modelled total seed
biomass for the two types of pasture. The
seed biomass data for a grazed pasture,
including medic (Paraggio) and barley grass,
were estimated for the period from 1st
January 1980 to 31st May 1998. The 1st
January 1980 was used as a starting time to
remove any distor tion from initial values
and runs were made over 10-year intervals
to ensure that annual grass seed banks were
not exhausted after droughts and to mimic
pasture sowing after cropping. Each run had
a two year overlap with the next.
Table 1.
Data requirements for using the GrassGro
model (Moore et a l. 1997) for the Mallee
Research Station (MRS) at Walpeup (35.085,
142.02E) in the northwest of Victoria .
Attribute Units/description
Rainfall daily rai nfall in mm record ed at
MRS
Temperature daily maximum and minimum
temperature
1965-95: MRS
1996-97: Ouyen Post Office
(35.0rS , 142.32E)a
Terrain gently sloping
Soil sandy loam , fertility rating of 0.8,
soil moisture budget parameters
(including thickness of each soil
layer, volumetric water content of
each laye r, drainage and
evaporation rates)
Pasture barley grass , medic (Parragio)
species
Livestock 2.0 ewes/ ha managed for wool
and meat production
a Temperature data for 1996-97 were extrapolated
from records at the Ouyen meteorological station
using the 1965-95 records to generate a
temperaturedifference profile with the MRS .
100
Estimates of seed biomass were then
recorded for each lO-year simulation
excluding the first two yea rs to reduce the
effec t of 'reseeding' the p asture on the seed
available to mice. Finally, seed biomass, S,
for each 40-day time step was determined by
interpolation from the nearest weekly
estimates from the GrassGro model.
Modified Victorian Mallee model
Many of the existing models for predicting
mouse plagues invoke a complex set of
fac tors to explain the population dynamics
of mice (see, for example, the MIA rnodel). In
this case, as a first step, we assessed how
much of the variance in the rate of increase
of mice can be expla ined by the m.ost
obvious factor, food ava ilability. Then we
determined whether inclusion of a density-
dependent fac tor improved the predictive
ability of the model. This additional factor is
a surrogate for several processes that m ay
modify the rate of increase when the denSity
of m ice is high. Examples include disease
and cha nges in mating behaviour, socia l
organisa tion or dispersal. The approach is
similar to the generalised fo rm of the
predation-regulation model in that the
dynamics of the mouse population are
assumed to be driven p rimarily by annua l
pulses of food whose dura tion and
m agnitude is determined by the weather.
However p redator-regulation at low m ouse
d ensities, e.g. through the aggregation of
raptors, was assumed to be less important in
the croplands of the Victorian Mallee than in
the localised irr igated areas studied by
Sinclair et al. (1990).
An Ivlev model was used for the
relationship between food availability and
 Models for Predicting Plagues in Australia

the rate of increase, r, for mice. 1his has the variables, food from wheat crops and
general form: pasture and the density of mice, were tested
singly then in combination. The overall
r = -a + q1- exp(-d.F)] (2)
goodness of fit of each model was estimated
using VENSIM@ (1997) to calculate the sum
where F is an index of food biomass which is
seed from pasture and / or wheat in the case of of the squared errors between the observed
rates of increase (from the smoothed data
mice. In equation (2), a is the maximum rate of
series) and the predictions of each model for
decrease, the maximum rate of increase is
the period 1983 to 1989. 1his statistic was
rmax = (c - a), and the demographic efficiency,
used as a guide in model selection. A second
d, is a measure of the ability of the mouse
criterion was the ability to predict the rates
population to respond to scarce resources.
This form for the rate of increase has been of increase and the trajectory of mouse
abundance for the later period from 1991 to
used to model the response of several species
1994. The optimal values, with 95%
to widely fluctuating food supply in
confidence limits, were estimated for the
climatically-variable parts of Australia; e.g.
parameters in equations (2) and (3), although
kangaroos-Macropus spp. (Bayliss 1987;
Caughley 1987; Cairns and Grigg 1993), feral for some models the maximum rate of
increase and the maximum rate of decrease
pigs-Sus scrota (Caley 1993; Choquenot and
were calculated directly from the smoothed
Dexter 1996) and rabbits-Oryctolagus
trajectory of mouse abundance and used as
cuniculus (Choquenot 1992; Pech and Hood
1998). The model applies when changes in the fixed parameters. The estimates from the
wheat model (equation 1, Figure 2c) do not
abundance of a population are determined
necessarily indicate their relative availability
primarily by the availability of food.
or value for mice. Comparison of Figures 2b
When food is scarce r is negative, i.e. the
and 2c suggest that the high wheat yields in
population declines, and when food is
1984,1987,1993 and 1996 correspond, with a
readily available r is positive but ultimately
limited to a maximum value rmax' by the smalllag, to mouse plagues. The outbreaks
!
appear to be a response to the difference
species' reproductive capacity. Density-
between normal years, e.g. 1988-92, and the
dependence was included as a linear
four high-yield years. The sensitivity of the
additive factor, in a similar way to that
rate of increase to this difference can be
suggested by Caughley and Krebs (1983).
With the additional factor, the parameters a explored by transforming the amplitude of
and (c a) in equation (2) no longer retain the wheat yield index to (R AO /348)U, where
the scaling exponent u is estimated by fitting
their demographic interpretation so the
the models to the data. Then the model of the
model was simplified to:
food available from wheat crops is:
r = a + c.exp( -d.F) + gN (3)

where g is a measure of the strength of the

density-dependence.
Models were developed through an If, for example u 1, then equation (4)
iterative process where the explanatory indicates that mouse population dynamics

101
Ecologically-based Rodent Management
are sensitive to changes in high values of W,
whereas u "'" 1 implies no transformation is
necessary (Figure 4).
The results of fitting the models are
summarised in Table 2 and illustrated in
Figures 5 and 6. It is apparent in Figure 2b
that in non-plague years there is too much
variability in the field data to detect the
expected seasonal variations in the mouse
population, e.g. the annual spring decline
in abundance (Singleton 1989). The absence
of these small seasonal variations carries
through to the observed rates of increase
(Figures Sa-d). However, in all years,
seasonal effects are quite pronounced in the
models for food supply (Figures 2c,d) and
it was necessary to suppress much of this
within-year variability in fitting the models
for the numerical response. The result is
that the models including wheat, (i), (ii)
and (iv) in Table 2, have scaling exponents,
u, of 6.6,9.4, and 11.3, respectively; i.e. all
these models require a transformation of W
similar to that shown in Figure 4b.
The corresponding numerical response
functions show a rapid change from negative
to positive r at low values of the wheat index,
saturating at rmax above this threshold
(Figures 6a, b and d). Even with these
transformations, all of the food-only models
were a poor fit to the observed r in non-
plague years. Fixing the maximum rate of
decrease in model (i) resulted in a reasonable
fit to the post-plague declines in mouse
abundance (Figure Sa) but an overall poorer
fit compared to model (ii) where only the
maximum rate of increase, r max' was fixed.
Conversely, model (ii) did not match periods
with large negative rates of increase but
provided a clearer distinction between
plague and non-plague years (Figure 5b).
102
Model (iii) was based on the pasture seed
biomass which showed strong seasonal
variation and, apart from 1984 and 1993, was
mostly in the range 0-1000 kg/ha. Since this
was also the part of the fitted numerical
response with a positive slope (Figure 6c),
the predicted r showed a corresponding
pattern that was not reflected in the observed
rates of increase of mice (Figure Sc). A trans-
formation similar to that for wheat was tried
unsuccessfully with the seed biomass model.
The best fit to the observed rates of
increase was obtained by combining a
density-dependent effect with the wheat-
based food index [model (iv) in Table 2]. This
also resulted in improvements in predicting
the trajectory of mouse abundance.
However, the upper 95% confidence limit for
the density-dependent parameter, g, could
not be estimated and there was only a
relatively small improvement in the sum of
squared errors compared to model (ii).
The results in Figure 5d show that the
model matched the high rates of increase
that generated the mouse plagues in 1984,
1987 and 1993, but not the latter part of the
period of high r leading to the plague in
1997. The major density-dependent
contributions were during plague years
(Figure 5d) and, compared to the wheat-only
model (ii), produced a slightly better fit
during periods of rapid decline in mouse
numbers. All the models that include wheat
(i, ii and iv) were optimised with high values
of the scaling exponent, u.
In each case this was balanced to some
extent by high values in the demographic
efficiency, d, in the product dW. An
independent estimate of at least one of these
two parameters is required to resolve this
difficulty with the Ivlev model.
 Models for Predicting Plagues in Australia

(a) u = 1

1.0

0.8

~
~ 0.6
ro
~
g 0.4 .
u..
0.2 .

0.0

(b) u= 10

1.0

0.8
~
~
0.6
'"::>
~
to
150 0.4
u..

Figure 4.
The effect of the parameter, u, In equation (4) in modifying the index of food availability from wheat crops.
=
Time, in units of 1/9th of a year, is measured from the 1 st of January each year and rainfall is in mm. (a) U
1, which Is the same food index used for Figure 2c. (b) U = 10, showing the effect of u 1 in suppressing
the food availability for low rainfall, <200-250 mm, and greatly exaggerating the relative availability above
this threshold.

103
Ecologically-based Rodent Management

-1

ID
fIJ
<U -1
ID
....

.5
"-
0 1
~
0:::
0

-1

-1 o i
~
-0.5 ~
ID
ID o
0
c:
<U 0
-oq;
c:_
J\V
,/;lO
\VfIJ -2
.... Cl
00
x=
ID -4
1'J
.E
-6
83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98

Figure 5.
(a}-(d) The observed (.) and modelled (-) rates of increase per 40 days. (e) The observed (0) and modelled
(-) index of abundance of mice. For (a)-(d) the predicted trajectories correspond to the models (i}-(iv)
listed in Table 2. The density-dependent contribution to f(- -) is shown in Cd). For (e), the abundance of mice
was predicted using model (iv) from Table 2.

104
 Models for Predicting Plagues in Australia

(a) Model i 1.0

~!
'-
ID 0.5
(/)
C1l
~ 0.0

t
(.l
.:
'0 -0.5
<Il
n;
a:: -1.0
I I I I
-1.5
0 0.1 0.2 0.3 0.4 0.5
Wheat availability

(b) Model 11 1.0

'-
ID
(/)
{"(I
~
(.l
0.5

0.0
-/
.~
'0 -0.5 -
<Il
n;
a:: -1.0 -
-1.5 I I I I
0 0.1 0.2 0.3 0.4 0.5
Wheat availability
1.0
(c) Model III

/
'-
ID 0.5 I-
rJ)
C1l
~ 0.0
(.l

V
.:
'0 -0.5
<Il
n;
Cl: -1.0

-1.5 '----- I I I
0 1000 2000 3000 4000
Seed availability
1.0
(d) Model Iv '-
ID 0.5
(/)
C1l
~
(.1
.~
0.0 V
'0 -0.5 I-
<Il
n;
a:: -1.0 I-
-1.5 I I I I
0 0.1 0.2 0.3 0.4 0.5
Wheat availability

Flgure 6.
The estimated numerical response functions corresponding to models (I)-(Iv) In Table 2. Food is from (a)
wheat crops, with,max and Sw fixed at the observed values, (b) wheat crops, with only,max fixed, (c) pasture
seed In kg/ha, with rmax and Sw fixed, and (d) wheat crops with the density of mice set at 0.001 In the
denslty-dependent term in model (Iv).

105
Ecologically-based Rodent Management

Table 2.
Results from fitting models for the numerical response of mice. The food biomass, F, in equation (2) was
replaced by an index ofthe food available from cereal crops, W, (equation 4) for models (i) and (ii), and seed
biomass from grazed pasture,S, in kg/ha (model iii). The models' parameters have corresponding
subscripts. In model (iv), the rate of increase is determined by equation (4) for W(replacing Fin equation 3)
and the index of mouse abundance, N. Parameters were optimised using VENSIM (1997) to fit the models
to the data for the rate of increase per 40 days for the period from 1983 to 1989. SSE is the sum of the
squared errors in fitting each model. For the optimised values, 95% confidence limits are given in square
brackets unless they could not be estimated [NE]

(i) Food from wheat crops : r = - a w + cJl - exp(- dw- W)]. with the maximum rate of increase
(rmax= Cw - awl and maximum rate of decrease (-a w) fixed at the observed values .
a w = 1.34 (fixed)
Cw = 2.19 (fixed)
d w = 48.0 [NE - 77 .0]
u = 6.6 [NE - 7.5]
SSE = 34.7
(ii) Food from wheat crops: r = -a w + cJl - exp(-dw- W)]. with the maximum rate of increase
(rmax = Cw - awl fixed at the observed value .
a w = 0.24 [0.026 - 0.45]
Cw = rmax + a w = 0.84 + a w (fixed)
d w= 23.7 [4.8 - 61.3]
u = 9.4 [7.3 - 13.6]
SSE = 10.3
(iii) Pasture seed biomass : r = - as + c s [l - exp(-ds. Sl]. with the maximum rate of increase
(rmax = Cw - awl and maximum rate of decrease (- awl fixed at the observed values.
r-
as = 1.34 (fixed)
Cs = 2.19 (fixedl
d s = 0.0039 [NE]
SSE = 35.5
(iv) Wheat and density: r = aN + c N exp(-dN. W) + gN
aN = 0.22 [0.11 - 0.41]
cN = 0.74 [0.37 - 0 .94]
dN = 102.5 [19 .7 - NE]
u = 11.3 [8.8 - 16.7]
g = -0.30 [-12 .6 - NE]
SSE = 9 .1
In add ition, the estima tion procedure is Some of the main d emographic features
quite sensitive to small changes in the are represented reasonably w ell. However,
constan ts which, together with the wide the results clearly demonstra te tha t the post-
confidence limits, sugges ts that we are plague d eclines are not well modelled nor is
searchi.ng a very fla t parameter space in there an obv ious explanation for the 12-
attempting to fit th e models. month difference be tween the predicted
In Figure Se, the observed d ensities of p lague in 1996 and the observed ou tbreak ill
mice for the period for 1983 to 1997 a re 1997.
compared to the predicted trajectories using
model (iv) in Table 2.

106

Future directions for the Victorian
Mallee model
The modified Victorian Mallee model
represents some progress towards a
quantitative model for assessing the
effectiveness of mouse control operations in
southern Australia. However, any
assessments will be subject to the proviso
that the model for the numerical response is
still appropriate after a mouse population
has been reduced by a control technique.
Alternatively, in the case of immuno-
contraception, estimates would be required
of the proportional change in r resulting
from an imposed level of infertility
(Chambers et al. 1997, 1999).
An obvious shortcoming of the modified
model is the lack of detail on seasonal
changes in demographic parameters.
Improvements in this area are likely in future
models which will treat fecundity and
survival rates separately. Examples of this
approach include the density-dependent,
stochastic model of Leirs et al. (1997) for the
multi-mammate rat Mastomys nataiensis in
Africa and the model used by Pech et al.
(1997) to assess the value of fertility control
for the management of foxes in Australia.
There is some evidence that the onset of
breeding by mice depends on seasonal
environmental triggers (e.g. Olsen 1981;
Bomford 1987a,b,c; Bomford and Redhead
1987; Tann et al. 1991) and the results of
model (iv) in Table 2 suggest that density-
dependent factors are likely to be important
in terminating plagues, presumably through
decreased survival rates of mice. In addition,
other factors could be included to fine-tune
the model. For example, the ultimate
abundance of mice in a plague can be strongly
Models for Predicting Plagues in Australia
influenced by rainfall events in summer and
the autumn-winter period (Singleton 1989;
Boonstra and Redhead 1994).
It is likely that the strong seasonal
variation in food from wheat crops and
pasture will play an important role in future
age-structured models for both fecundity
and survival. Field data are required to
validate the pasture and wheat crop models,
especially the relationship between seed and
crop biomass and the amount of food
available to mice. In addition, manipulative
experiments should be used to confirm the
dependence of the numerical response on
food supply. Although it may be difficult to
manipulate factors like predation and
disease, there may be opportunities to use
future mouse control campaigns to test the
importance of the density-dependence
implied by the optimal model.
ACKNOWLEDGMENT
Part of this work was included in an
unpublished review prepared by R. Pech for
the Grains Research and Development
Corporation in 1991.
REFERENCES
Barker, S.c. and Singleton, G.R. 1987. Conclu-
sion. In: Singleton, C.R., ed., Capillaria hepatica
(Nematoda) as a potential control agent of
house mice. Proceedings of a CSIRO
workshop,June 1987. Division of Wildlife
and Rangelands Research, Technical
Memorandum No. 28, Commonwealth Scien-
tific and Ind ustiral Research Organisation
(CSIRO), Melbourne.
Barker, S.c., Singleton, C.R. and Spratt, D.M.
1991. Can the nematode Capillaria hepatica
regulate abundance in wild house mice?
Results of enclosure experiments in south-
eastern Australia. Parasitology, 103,439-449.
107
Ecologically-based Rodent Management
Bayliss, P.1987. Kangaroo dynamics. In: Caugh-
Iey, G., Shepherd, N. and Short, J., ed.,
roos: their ecology and management in the
sheep rangelands of Australia. Cambridge,
Cambridge University Press, 119-134.
Bomford, M. 1987a. Food and reproduction of
wild house mice. 1. Diet a nd breeding seasons
in various habitats on irriga ted cerea 1farms in
New South Wales. Australian Wildlife
Research, 14, 183-196.
Bomford, M. 1987b. Food and reproduction of
wild house mice. n. A field experiment to
examine the effect of food availability and
food quality on breeding in spring. Austral-
ian Wildlife Research, 14, 197-206.
Bomford, M. 1987c. Food and reproduction of
wild house mice. III. Experiments on the
breeding performance of caged house mice
fed rice-based diets. Australian Wildlife
Research,14,207-218.
Bomford, M. and Redhead, T. 1987. A field exper-
iment to examine the effects of food quality
and population density on reproduction of
wild house mice. Oikos, 48, 304-311.
Boonstra, R and Redhead, T.D. 1994. Population
dynamics of an outbreak population of house
mice (Mus domesticus) in the irrigated rice-
growing area of Australia. Wildlife Research,
21,543-598.
Brown, PR, Singleton, G.R and Dunn, C. 1998.
Best farm management practices for mouse
control: testing of recommendations. In: The
future of vertebrate pest direc-
tion for the third millennium. Proceedings of
the 11 th Australian Vertebrate Pest Confer-
ence, 3-8 May, Bunbury, Western Australia,
99-104.
Cairns, S.c. and Grigg, G.c. 1993. Population
dynamics of red kangaroos Macropus ru/us in
relation to rainfall in the South Australian
pastoral zone. Journal of Applied Ecology,30,
444-458.
Caley, P. 1993. Population dynamics of feral pigs
(Sus scrota) in a tropica 1riverine habitat
complex. Wildlife Research, 20, 625-636.
108
Cantrill, S. 1992. The population dynamicS of the
house mouse (Mus domesticus) in a dual crop
agricultural ecosystem. PhD thesis, School of
Life Science, Queensland University of
Technology.
Caughley, G.c. 1977. Analysis of vertebrate
populations. Chicester, John Wiley & Sons.
Caughley, G.C.1987. Ecological relationships. In:
Caughley, G., Shepherd, N. and Short, J., ed.,
Kangaroos: their ecology and management in
the sheep rangelands of Australia.
Cambridge, Cambridge University Press,
159-187.
Caughley, G.c. and Krebs, c.J. 1983. Are big
mammals simply little mammals writ large?
Oecologia,59,7-17.
Chambers, L.K., Singleton, G.R. and Hinds, L.A.
1999. Fertility control of wild mouse popula-
tions: the effects of hormonal competence and
an imposed level of sterility. Wildlife
Research, 26, 579-591.
Chambers, L.K., Singleton, GR and Hood, G.M.
1997. Immunocontraception as a potential
control method of wild rodent populations.
Belgian Journal of Zoology, 127, 145-156.
Choquenot, D. 1992. The outsiders: competition
between introduced herbivores and domestic
stock in rang eland grazing systems. In:
Australian rangelands in a changing environ-
ment. Proceedings of the ;nh Biennial Confer-
ence of the Australian Rangeland SOciety, 5-8
October 1992, Cobar, NSW. Cobar, Australian
Rangeland Society, 106-116.
Choquenot, D. and Dexter, N.1996. Spatial varia-
tion in food limitation: the effects of foraging
constraints on the distribution and
abundance of feral pigs in the rangelands. In:
FIoyd, RH., Sheppard, A.W. and De Barro,
P.J., ed., Frontiers of popUlation ecology.
Melbourne, CSIRO Publishing, 531-546.
Cornish, E.A., R.J. and Hill, G.W. 1980.
Yield trends in the wheat belt of South
Australia from 1896 to 1964. Agricultural
Record Ko. 7. Adelaide, Government Printer.
Davey, c.c. and Fullagar, P.J. 1986. Changes in
the abundance and distribution of raptors
during a house mouse plague. Corella, 10,52-
54.

Foley, lC 1957. Droughts in Australia. Review of
records from the earliest years of settlement
to 1955. Melbourne, Commonwealth Bureau
of Meteorology, Bulletin No. 43.
French, RJ. and Schultz,J.E.1984. Water use
efficiency of wheat in a Mediterranean-type
environment. 1. The relation between yield,
water use and climate. Australian Journal of
Agricultural Research, 35,743-64.
Hone, J. 1980. Probabilities of house mouse (Mus
musculus) plagues and their use in control.
Australian Wildlife Research, 7,417-420.
Jackson, RJ., Maguire, DJ, Hinds, L.A and
Ramshaw, LA 1998. Infertility in mice
ind uced by a recombinant ectromelia virus
expressing mouse zona pellucida glycopro-
tein 3. Biology of Reproduction, 58,152-159.
Kay, B.J., Twigg, L.E., Korn, T.J. and Nicol, HJ.
1994. The use of artificial perches to increase
predation on house mice (Mus domesticus) by
raptors. Wildlife Research, 21, 95-106.
Krebs, CJ., Chitty, D., Singleton, G. and
Boons tra, R 1995a. Can changes in the social
behaviour help to explain house mouse
plagues in Australia? Oikos, 73, 429-434.
Krebs, CL Kenney, AI. and Singleton, G.R
1995b. Movements of feral house mice in
agricultural landscapes. AustralianJournal of
Zoology, 43, 293-302.
Leirs, H.,Stenseth, N.C, Nichols, J.D., Hines, J.E.,
Verhagen, R. and Verheyen, W. 1997.
Stochastic seasonality and non-linear
density-dependent factors regulate popula-
tion size in an African rodent. Nature, 389,
176-180.
McCallum, H.!. 1993. Evaluation of a nematode
(Capillaria hepatica Bancroft, 1893) as a control
agent for populations of house mice (Mus
musculus domestiells Schwartz and Schwartz,
1943). Revue Scientifique et Techique OIE
(Office International des Epizooties), 12,83-
94.
McCallum, H.!. and Singleton, G.R 1989. Models
to assess the potential of Capillaria hcpatica to
control population outbreaks of house mice.
Parasitology, 98,425-437.
Moore, A.D., DonneIly, J.R and Freer, M. 1997.
GRAZPLAN: decision support systems for
Australia grazing enterprises. m. Pasture
Models for Predicting Plagues in Australia
growth and soil moisture submodels, and the
GrassGro DSS. Agricultural Systems, 55, 535-
582.
Mutze, G.J. 1989. Mouse plagues in South
Australian cereal-growing areas. 1. Occur-
rence and distribution of plagues from 1900 to
1984. Australian Wildlife Research, 16, 677-
683.
Mutze, G.J. 1991. Mouse plagues in South
Australian cereal-growing areas. Ill. Changes
in mouse abundance during plague and non-
plague years, and the role of refugia. Wildlife
Research,18,593-604.
Mutze, G.]., Veitch, L.G. and Miller, RB.1990.
Mouse plagues in South Australian cereal-
growing areas. n. An empirical model for
prediction of plagues. Australian Wildlife
Research, 17, 313-324.
Newsome, AE. 1969a. A population study of
house-mice temporarily inhabiting a South
Australian wheatfield. Journal of Animal
Ecology, 38, 341-359.
Newsome, AE. 1969b. A population study of
house-mice permanently inhabiting a reed-
bed in South Australia. Journal of Animal
Ecology, 38, 361-377.
Newsome, AE. 1970. An experimental attempt
to produce a mouse plague. Journal of
Animal Ecology, 39, 299-311.
Newsome, AE.1971. The ecology of house-mice
in cereal haystacks. Journal of Animal
Ecology, 40, 1-15.
Newsome, AE. and Corbett, L.K 1975. VI.
Outbreaks of rodents in semi-arid and arid
Australia: causes, preventions, and evolu-
tionary considerations. In: Prakash, 1. and
Ghosh, P.K, ed., Rodents in desert environ-
ments, The Hague, W.Junk,117-153.
Newsome, AE., Cowan, P.E. and Ives, P.M. 1982.
Homing by wild house-mice displaced with
or without the opportunity to see. Australian
Wildlife Research, 9, 421-6.
Norton, G.A 1988. Philosophy, concepts and
techniques. In: Nortol1, G.A and Pech, RP.,
ed., Vertebrate pest management in Australia:
a decision analysis/ systems analysis
approach. CSIRO Division of Wildlife and
Ecology, Project Report No. 5, 1-17.
109
Ecologically-based Rodent Management
Olsen, P. 1981. The stimulating effect of a phyto-
hormone, gibberellic acid, on reproduction of
Mus musculus. Australian Wildlife Research,
8,321-325.
Pech, RP. and Hood, G.M. 1998. Foxes, rabbits,
alternative prey and rabbit calicivirus
disease: consequences of a new biological
control agent for an outbreaking species in
Australia. Journal of Applied Ecology, 35,
434--453.
Pech, R, Hood, G.M., McIlroy, J. and Saunders,
G. 1997. Can foxes be controlled by reducing
their fecundity? Reproduction Fertility and
Development, 9, 41-50.
Pech, RP., Sinclair, AE. and Newsome, AE.
1995. Predation models for primary and
secondary prey. Wildlife Research, 22, 55--64.
Pech, RP., Sinclair, A.E., Newsome, A.E. and
Catling, P.e. 1992. Limits to predator regula-
tion of rabbits in Australia: evidence from
predator-removal experiments. Oecologia,
89,102-112.
Redhead, T.D. 1982. Reproduction, growth and
population dynamics of house mice in
irrigated and non-irrigated cereal farms in
New South Wales. PhD thesis, Department of
Zoology, Australian National University,
Canberra.
Redhead, T.D. 1987. Computer simulations of
mouse plagues: plague formation and
prevention. In: Working Papers of the 8th
Australian Vertebrate Pest Control Confer-
ence, Coolangatta, Queensland, 186-194.
Redhead, T.D., Enright, N. and Newsome, A.E.
1985. Causes and predictions of outbreaks of
Mus musculus in irrigated and non-irrigated
cereal farms. Acta Zoologica Fennica, 173,
123-127.
Redhead, T.D. and Singleton, G.R 1988a. The
PICA Strategy for the prevention of losses
caused by plagues of Mus domesticus in rural
Australia. EPPO Bulletin, 18,237-248.
Redhead, T.D., and Singleton, G.R. 1988b. An
examination of the "PICA" strategy for the
prevention oflosses caused by plagues of the
house mouse (Mus domesticus) in rural
Australia. In: Korton, G.A. and Pech, RP., ed.,
Vertebrate pest management in Australia:
110
decision analysis/ systems analysis approach.
CSIRO Division of Wildlife and Ecology,
Project Report No. 5, Melbourne, 18-37.
Saunders, G.R 1986. Plagues of the house mouse
in south eastern Australia. In: Salmon, T.P.,
ed., Proceedings of the twelfth vertebrate pest
conference. Davis, University of California,
173-176.
Saunders, G.R and Giles, J.R 1977. A relation-
ship between plagues of the house mouse,
Mus musculus (Rodentia: Muridae) and
prolonged periods of dry weather in south-
eastern Australia. Australian Wildlife
Research, 4, 241-7.
SeH, E. and Pedersen, D.G. 1978. Effect of rainfall
on the grain yield of spring wheat with an
application to the analysis of adaptation.
Australian Journal of Agricultural Research,
29,1107-1115.
Shell am, G.R 1994. The potential of murine
cytomegalovirus as a viral vector for
immunocontraception. Reproduction Fertil-
ity and Development, 6,401-409.
Sinclair, ARE., Olsen, P.D. and Redhead, T.D.
1990. Can predators regulate small mammal
poplIlations? Evidence from house mouse
outbreaks in Australia. Oikos, 59, 382-392.
Singleton, G.R 1985. Population dynamics of
Mus musculus and its parasites in mallee
wheatlands in Victoria during and after a
drought. Australian Wildlife Research, 12,
437-445.
Singleton, G.R. 1987. Parasites of mice in the
Victorian manee wheatlands. In: Singleton,
G.R, ed., Capillaria hepatica (Nematoda) as a
potential control agent of house mice. CSIRO
Division of Wildlife and Rangelands
Research, Technical Memorandum No. 28,
Melbourne, 12-14.
Singleton, G.R 1989. Population dynamics of an
outbreak of house mOllse (Mus domesticus) in
the mallee wheatlands of Australia-
hypothesis of plague formation. Journal of
Zoology (London), 219, 495-515.
Singleton, G.R and Brown, P.R 1998. Manage-
ment of mouse plagues in Australia: integra-
tion of population ecology, biocontrol and

best farm practice. In: Cowan, D.P. and Feare,
CJ., ed., Advances in vertebrate pest manage-
ment. Furth, Filander Verlag, 189-203.
Singleton, G.R and Chambers, L.K. 1996. A
manipulative field experiment to examine the
effect of Capillaria hepatiea (Nematoda) on
wild mouse populations in southern
Australia. International Journal for Parasitol-
ogy, 26, 383-398.
Singleton, G.R, Chambers, L.K. and Spratt, D.M.
1995. An experimental field study to examine
whether Capillaria hepatica (Nematoda) can
limit house mouse populations in eastern
Australia. Wildlife Research, 22, 31-53.
Singleton, C.R and McCallum, H.!. 1990. The
potential of Capillaria hepatiea to control
mouse plagues. Parasitology Today, 6, 190-
193.
Singleton, GR, McCallum, H.!. and Spratt, D.M.
1990. Fertility control of house mice using an
hepatic nematode. Proceedings of the "Fertil-
ity Control of Wildlife" Conference,
Melbourne, 21-23 November, 1990.
Singleton, G.R and Redhead, T. 1989. House
mouse plagues. In: Noble, le. and Bradstock,
RA., ed., Mediterranean landscapes in
Australia: maliee ecosystems and their
management. Melbourne, CSIRO, 418-33.
Singleton, G.R. and Redhead, T.D. 1990a. Struc-
ture and biology of house mouse populations
that plague irregularly; an evolutionary
perspective. Biological Journal of the Linnean
Society, 41, 285-300.
Singleton, G.R and Redhead, T.o. 1990b. Future
prospects for biological control of rodents
using micro- and macro-parasites. In: Quick,
C.R, ed., Rodents and rice. Report and
proceedings of an expert panel on rice rodent
control, Los Banos, Sept. 10-14,1990, IRRI,
Philippines, 75-82.
Singleton, GR, Smith, A.L, Shell am, G.R,
Fitzgerald, N. and Muller, W.J. 1993. Preva-
lence of viral antibodies and helminths in
field populations of house mice (Mus domesti-
eus) in southeastern Australia. Epidemiology
and Infection, 110,399-417.
Models for Predicting Plagues in Australia
Singleton, G.R and Spratt, D.M. 1986. The effects
of Capillaria hepatica (Nematoda) on natality
and survival to weaning in BALB/ c mice.
Australian Journal of Zoology, 34, 677-681.
Singleton, C.R and Spratt, D.M. 1990. Biological
control of mice-the challenge, progress and
prospects. In: Noble, J.C, Joss, P.J. and Tones,
C.K., ed., The mallee lands: a conservation
perspective. Proceedings of the National
Mallee Conference, Adelaide, April 1989.
Melbourne, CSIRO, 238-242.
Singleton, G.R., Spratt, D.M., Barker, S.e. and
Hodgson, P.F. 1991. The geographic distribu-
tion and host range of Capillaria hepatica
(Bancroft) (Nematoda) in Australia. Interna-
tional Journal of Parasitology, 21, 945-957.
Smith, A.L., Singleton, G.R, Hansen, GM. and
Shellam, G. 1993. A serologic survey for
viruses and Mycoplasma pulmonis among wild
house mice (Mus domestieus) in southeastern
Australia. Journal of Wildlife Diseases, 29,
219-229.
S-PLUS 4 Guide to Statistics. 1997. Seattle, Data
Analysis Products Division, Mathsoft.
Spratt, D.M. 1990. The role of helminths in the
biological control of mammals. International
Journal for Parasitology, 20, 543-550.
Spratt, D.M. and Singleton, GR. 1986. Studies of
the life cycle, infectivity and clinical effects of
Capillaria hepatica (Bancroft, 1893)
(Nematoda) in mice (Mus musculus). Austral-
ian Journal of Zoology, 34, 663-675.
Spratt, D.M. and Singleton, G.R 1987. Experi-
mental embryonation and survival of eggs of
Capillaria hepatica (Nematoda) under mouse
burrow conditions in cereal-growing soils.
Australian Journal of Zoology, 35, 337-341.
Tann, CR., Singleton, G.R and Coman, B. C 1991.
Diet of the house mouse (Mus domestieus) in the
malIee wheatlands of north-western Australia.
Wildlife Research, 18, 1-12.
Taylor, J.D., Stephens, CP., Duncan, RG. and
Singleton, C.R. 1994. Polyarthritis in wild
mice (Mus musculus) caused by Streptobacillus
moniliformis. Australian Veterinary Journal,
71,143-145.
111
Ecologically-based Rodent Management
Twigg, L.E. and Kay, B.J. 1994. The effects of
microhabitat and weather on house mouse
(Mus domesticus) numbers and the implica-
tions for management. Journal of Applied
Ecology 31, 651-663.
112
Veitch, L.G. and Anderson, R.B. 1985, Mouse
plagues in South Australia. CSIRO, Division
of Mathematics and Statistics, Consulting
Report No. SAC 85/17.
VENSIM@DSSReferenceSupplementVersion
3.0.1997. Ventana Systems, Inc., USA.
Chris R. Dickman

Abstract

Because of their ability to use agricultural production and their role in spreading
disease in humans, rodents are often viewed as having negative impacts in modified
and natural ecosystems. Some species, such as the black rat, have been further
implicated in the extinctions of many species of insular land birds, small mammals
and invertebrates. In this review, I focus on the interactions of rodents with chemical
and structural attributes of the environment, using the concept of 'ecosystem
engineering' as a framework. I also discuss the direct and indirect impacts of
rodents on food resources .
Many rodents alter the structure of their environment by surface tunnelling,
construction of leaf or stick nests, arranging pebbles around burrow entrances, or
stripping bark from trees. These activities provide living space or resource
opportunities for other organisms, and represent examples of simple allogenic
engineering. In more complex examples, digging, nest-building and other activities
modify the environment more extensively and modulate resource flows to other
organisms. Burrowing rodents such as pocket gophers, prairie dogs and mole-rats
alter soil structure and microtopography, nutrient cycling and water flows over local or
regional areas, and have dramatic effects on the growth and species composition of
plant communities. Nest structures that divert resource flows also represent complex
allogenic engineering. For example, beaver impoundments affect nutrient cycles and
water flow, and consequently the species richness of aquatic invertebrates, fish and
riparian vegetation at local and catchment scales. Rodents also engineer local
environments biotically by dispersing seeds and the storage organs of geophytes, as
well as the spores of hypogeal fungi that form mycorrhizal associations with plants .
Some species probably also play a minor role as pollinators. Rodents, finally, have
diverse and often pervasive effects on their food resources; there is much evidence of
positive and negative effects on growth form, standing crop and the species
composition and physical structure of plant communities
Rodents therefore contribute importantly to ecosystem function, and may have value
as indicators of environmental change . Management of rodent pests will need to move
away from the broadly destructive current approach of chemical warfare toward
ecologically-based solutions that sustainably control only the target species .

Keywords

Ecosystem engineering, environment, rodent, facilitation, predation, nests , burrows

113
Ecologically-based Rodent Management
INTRODUCTION
LTHOUGH SOME 1,800 species of
A modem rodents have been
described (Corbet and Hill 1991),
few have been well-studied and the majority
remains poorly known. Not surprisingly,
most knowledge has been obtained on
that impact on humans by exploiting
agricultural production or by spreading
diseases (Chitty and Southern 1954; Twigg
1978), or are useful in laboratory research
(Barnett 1975). Different species of rodents,
especially Rattus spp., have been implicated
also in the demise of island vertebrate faunas
(Atkinson 1985, 1996), and have often been
subject to intensive control to achieve
conservation objectives. Effective
management of rodent pests remains an
elusive but important goal in many parts of
the world, and for different reasons. As
discussed by various authors in this book,
solutions may lie more with ecologically-
based management than with simple one-
factor approaches that have been used
previously.
Despite the often negative effects of
rodents in natural and modified
ecosystems, many species have been shown
to contribute to ecosystem function and to
have value as indicators of environmental
change. For example, microtine rodents are
important at times in the cycling of carbon,
nitrogen and other elements (Inouye et al.
1987a, Huntly 1991), while beavers cause
alteration of hydrological regimes (Naiman
et al. 1988). Such species have been termed
'ecosystem engineers' (Jones et al. 1994).
Other species may be important as
114
pollinators or vectors of fungal spores (Tory
et al. 1997). Both microtine and sciurid
rodents have been used as indicators of
industrial pollution (Kostelecka-Myrcha et
al. 1981; Lepage and Parker 1988), while
some murids have been used to indicate the
severity of impact wrought by defoliants
used in chemical warfare (Sokolov et al.
1994; Evgenjeva and Fadeeva 1996). Several
further species also may be sensitive
barometers of climatic change (Frey 1992;
Bright and Morris 1996).
The range of interactions of rodents with
the environment is not well appreciated,
perhaps because the interactions are
diverse, often complex, or not apparent in
studies carried out in small study areas or
for short periods. However, such an
appreciation is likely to be important for
successful management of rodent
and essential if management is to be
ecologically-based.
In the present paper, I present a selective
review of rodent-ecosystem relationships,
focusing on the impacts of rodents on the
physical, chemical and biotic environments
and the consequences of these impacts for
other biota. Little attention is given to
competitive relationships among rodents or
to rodents as prey, because reviews of these
topics are available elsewhere (Sinclair
1989; Brown and Harney 1993; Dickman
and Doncaster, submitted for publication).
Where possible, studies that demonstrate
interactions experimentally have been
emphasised, because these are most likely
to identify the nature and magnitude of any
interactions that occur. The concept of
'ecosystem engineering' is used to provide
a framework for much of the review.

ECOSYSTEM ENGINEERING
The term'ecosystem engineering' was
introduced by Jones et a1. (1994) and refined
by Jones et a1. (1997). It may be defined as
follows: "Physical ecosystem engineers are
organisms that directly or indirectly control
the availability of resources to other
organisms by causing physical state changes
in biotic or abiotic materials. Physical
ecosystem engineering by organisms is the
physical modification, maintenance or
creation of habitats. The ecological effects of
engineering on other species occur because
the physical state changes directly or
indirectly control resources used by these
other species" (Jones et al. 1997, p. 1947).
Engineers were divided into two broad
groups by Jones et a1. (1994, 1997). Autogenic
engineers change the environment by their
own physical structures; an example would be
the shed limbs of trees that modulate
microclimate and microhabitat for other
organisms on the forest floor. In contrast,
allogenic engineers change the environment by
transforming living or abiotic materials from
one state to another by mechanical or other
means. An example would be the construction
of burrows by one species that could be used
by others. Rodents could be expected to be
allogenic engineers.
In both their papers, Jones et a1. (1994,
1997) drew a distinction between physical
ecosystem engineering and other ecological
processes such as pollination, dispersal,
competitive and trophic interactions,
including the utilisation of living or dead
tissue by consumers or decomposers. In the
present paper, however, I include the former
two of these processes under the term 'biotic
engineering'. Justification for this approach is
Rodent Ecosystem Relationships
given in Figure 1; this approach also provides
consistency in this review. Trophic
interactions do not fit an engineering
paradigm, and the effects of rodents on plant
and invertebrate prey species are discussed
separately below.
ALLOGENIC ENGINEERING
Simple cases: changes in physical
state (Figure 1a)
Rodent burrows are obvious and widespread
examples of allogenic engineering. Simple
burrows are made by most species at some
time in their life cycle, and vary in size,
orientation, depth and substrate
characteristics. Among Australian desert
rodents, Pseudomys hermannsburgensis and
Notomys alexis dig deep, vertical burrows in
summer to avoid high daily temperatures, but
occupy shallow surface burrows in other
seasons when temperatures, and daily
variations in temperature, are less extreme
(CR. Dickman, personal observation). Other
species, such as Rattus colletti, barely modify
cracks in the soil (Madsen and Shine 1999) Of,
like Rattus villosissimus, may construct
complex networks of tunnels under
favourable conditions (Predavec and
Dickman 1994). The burrows of many species
have been described in the literature (e.g.
Kemper 1981; Bronner 1992), with overviews
provided by Reichman and Smith (1990),
Meadows and Meadows (1991) and Hansell
(1993).
The major resource created by burrows is
living space for other organisms. Other
rodents, lizards, snakes and many species of
invertebrates make opportunistic use of
burrows (Kiviat 1978; Skinner and Smithers
1990). In arid Australia, several species of
115
Ecologically-based Rodent Management
dasyurid marsupials make extensive use of
abandoned rodent burrows, being unable to
dig burrows themselves (Dickman and Read
1992; Dickman 1996). In one study, the
burrowing activity itself, in reducing
compaction of soil, was shown to have the
additional effect of promoting germination
of seeds of an iridaceous geophyte
(Contreras and Gutierrez 1991).
Nests provide another example of
allogenic engineering. Simple constructs,
such as the cup-shaped grass nests of
Micromys minutus, may take hours or days to
build and last for the duration of one
breeding season (Harris and Trout 1991);
more complex structures of sticks and other
detritus, engineered by Leporillus spp. and
Neotoma spp., often last for generations
(Copley 1988). Nests are made from a variety
of living and non-living materials, and are
sometimes decorated with pebbles or other
materials (Anstee et al. 1997) for reasons that
remain unclear. As with burrows, nests
provide living space for other species of
vertebrates and invertebrates. Such
exploitation is usually opportunistic.
However, blind, wingless earwigs of the
genus Hemimerus are found primarily in the
nests of Cricetomys gambianus, and may be
obligatelyassociated (Knight 1984).
Two, more subtle examples of allogenic
engineering may be cited. The first involves
shallow scrapes created in surface soil by
foraging rodents that provide sites for
accumulation of seeds (McNaught 1994, see
also below). The second involves bark-
stripping of trees by Sciurus spp.,
Sundasciurus spp.and other squirrels
(Med way 1983). De-barking facilitates access
of fungal pathogens to vascular tissues
(Abbott et al. 1977), while dead trees provide
116
nesting, roosting and shelter sites for several
species of birds, bats and other arboreal
mammals (Corbet and Harris 1991;
MacKinnon et al. 1996). It is likely that
rodent-induced damage to plants provides
opportunities for exploitation by a broad
range of organisms, but few relevant studies
have been carried out to confirm this (for a
general discussion, see Karban and Myers
1989).
Complex cases: state changes that
modulate resource flow (Figure 1b)
Continual and intensive burrowing activity
by rodents may provide temporary living
space for other organisms, but it also affects
nutrient cycling, water flow, soil structure
and microtopography. Such effects have been
studied in detail in several species of fossorial
and terrestrial rodents, especially North
American geomyids, or pocket gophers,
prairie dogs and Old World mole-rats.
The digging activities of pocket gophers
(70-350 g) produce small piles of fresh surface
soil that may, over extended periods,
accumulate into large mounds termed mima
mounds (Inouye et al. 1997). In some habitats,
digging activity can cast over 15,000 kg of
soil/ha/year onto the surface, and mima
mounds of 25-50 m in diameter and 2 m in
height may be common (Beuchner 1942; Ross
et al. 1968). Some 50-100 mima mounds have
been recorded per hectare in some areas, with
higher densities occurring usually in
disturbed prairie and agricultural landscapes
(Mielke 1977). The mounds may consist
entirely of topsoil, or soil with gravel and
pebbles 50-60 mm diameter; in some
locations the presence of soil horizons within
mounds suggests a long period of
stabilisation (Cox and Gakahu 1986).
 Rodent Ecosystem Relationships

Comparisons of soils from mow1ds and
lmdisturbed inter-mou nd areas have show n
differences in texture, organic content,
water-h olding capacity and nutrient status
(Mielke 1977; H obbs an d Hobbs 1987;
Inouye et a l. 1987b; H untly and Inouye
1988). These differences in turn promote
heterogeneity in plant sp ecies com position
an d grow th responses. In shortgrass pra irie,
the bur rowing ac tivities of Thomornys bottae
may kill stand in g vegeta tion bu t p rovide
op p ortunities for establishm ent of
herbaceolls perennia l dicots (M artin sen et al.
1990). In serpentine grassland, moun ds of
T. bottae are invaded by d ifferen t species of
p lants d epending on preva il i.ng rainfall
conditions (Hobbs and Mooney 1991 ); the
timing and intensity of soil disturbance m ay
also be important (Moloney et aJ. 1992).
Fin ally, in tallgrass prairie, the mOlmds of
Geornys bursarius have complex effects on
both vegeta tion and faw1a. MOlmds brea k
the prairie canop y and provide recruitm en t
sites for d icot seedlings, often in creasing
local plant diversity (H artnett and Keeler
1995). MOlmds also a ttrac t some herbivores
such as grasshoppers, but m ay eitl1er attract
or repel mammalian herbivores sllch as the
meadow vole Microtus pennsylvanicus
(Whittaker et al. 1991; cf. Klaas et aJ. 1998) . If
mounds alter local patterns of herbivory,
this is likely to produce further effects on

Allogenic or biotic engineering

Resource Flows
1
!
(a) (b)

State 1 State 2 = Resource State 1

t
State 2 J
i
Organism
1
Organism

Figure 1.
Conceptual models of allogenic and biotic engineering, as applied to rodents (after Jones et al. 1994,
1997). In the simplest case, (a), living or non-living raw materials are transformed by animal activity from
state 1 to state 2. The point of modulation is shown by opposing arrow heads. In allogenic engineering,
state 2 is a new engineered resource such as a burrow that usually can be used immediately. In biotic
engineering, state 2 is an activated but incipient resource such as a pollinated flower or dispersed seed or
spore that may be structurally no different from the state 1 condition .
In the more complex case, (b), the product s of state 2 modulate the flow of one or more resources to other
species. Such modulation may be rapid if state 2 resources have been engineered allogenically, but slow if
engineering has been biotic and is cont ingent on growth of plant or fungal tissue. Jones et al. (1994, 1997)
discussed additional types of allogenic and autogenic engineering, but these do not appear relevant to
rodents. 'Biotic engineering' is used for the first time here.

117
Ecologically-based Rodent Management
plant community structure and
heterogeneity, perhaps promoting species
richness over time (Klaas et al. 1998).
Like their smaller counterparts, prairie
dogs (1 kg) also modulate resource flows to
other species by digging. Research on the
best-studied species, Cynomys ludovicianus,
shows that colonies develop on deep,
productive soils where flooding is unlikely,
and range in size from tens to hundreds of
hectares (Dahlsted et al. 1981; Hoogland
1994). Up to 300 burrows may occur per
hectare, with soil mounds 1-2 m diameter
surrounding each burrow entrance (Whicker
and Detling 1988). Digging affects soil
structure and compaction, increases
drainage and, with grazing by prairie dogs,
the cycling of nitrogen and other nutrients
(Coppock et al. 1983). Although grazing and
engineering effects have not been
disentangled in studies of C. ludovicianus,
both probably contribute to extensive
patteming of plant communities within
prairie dog colonies. In mixed-grass prairie,
Coppock et al. (1983) showed that grasses
decreased in biomass with colony age
whereas forbs and dwarf shrubs increased;
nitrogen in graminoid shoots also peaked in
long-established colonies. The modified
habitats produced by prairie dog
excavations favour increased local
abundances and diversity of open-plain
birds but decreased species richness of small
mammals (Agnew et al. 1986). Interestingly,
colony sites also contain higher densities of
soil nematodes than undisturbed areas
(Ingham and Detling 1984), perhaps
reflecting greater ease of establishment in
loosened soiL
Burrowing and tunnelling activities by
fossorial rodents such as mole-rats displace
118
large volumes of soil and often result in the
creation of surface mounds. These surficial
structures resemble the mounds of pocket
gophers and prairie dogs in size and
composition, and have usually similar
effects on nutrient status, water flow and
organic content (Jarvis and Sale 1971; Cox
and Gakahu 1985; Cox et al. 1987). Cox and
Gakahu (1985) showed that coverage of
forbs and shrubs on mima mounds of
Tachyoryctes splendens was more than double
that on inter-mound plots, whereas coverage
of grass and Acacia trees was much reduced.
These authors also noted a correlation
between the activity areas of mole-rats and a
fungus-gardening termite-Odontotermes
sp., and suggested that termites
preferentially use the rich organic deposits
in mole-rat nest chambers to establish
fungus gardens. A wide range of
invertebrates has been documented using
the nest mounds of the blind mole-rat Spalax
ehrenbergi (Heth 1991). However, it is not
clear here whether mound use represents a
simple case of allogenic engineering, or a
more complex case where mounds modulate
food or other resources that sustain the
invertebrate communities. Further examples
of fossorial or semi-fossorial rodents
modulating resource flow for other species
by their burrowing activities occur within
the Microtinae, Octodontidae and
Heteromyidae Chew and Whitford
1992; Contreras and Gutierrez 1991; G6mez-
Garcia et a1. 1995; Borghi and Giannoni
1997). A useful review is provided by
Huntly and Reichman (1994).
Nest structures that divert resource flow
represent a further class of examples of
complex allogenic engineering. Beaver dams
are the most conspicuous examples of such

structures; similar but less extensive nests
are made by muskrats Ondatra zibethicus and
occasionally by Myocastor coypus (Ebenhard
1988).
Beaver dams are constructed of young
and mature trees that the animals cut
themselves, as well as sediments and other
debris. The North American beaver, Castor
canadensis, builds some 2-16 dams per
kilometre of stream, with small dams
containing 4-18 m 3 and larger dams> 100 m 3
of wood (Naiman et al. 1986,1988). The
major effect of dams is to alter the stream
channel by impounding water, creating
patch bodies (sensu Johnston and Naiman
1987) of water, sediment, aerobic soil
beneath the pond and anaerobic soil in
deeper strata. The surrounding riparian
zone is also affected by damming, with
stream widths sometimes increased by an
order of magnitude from their original
condition (Naiman et al. 1988). Because of
the changed hydrological regime and the
additional effects of beaver herbivory, patch
bodies show dramatically different fluxes of
carbon, nitrogen and energy compared with
unaltered streams. Impoundments usually
have relatively low inputs of carbon, but
high standing stocks and outputs (Naiman
et al. 1986); significant fluxes arise from
release of methane (Naiman et al. 1991;
Yavitt et al. 1992). Impoundments have been
shown further to enhance accumulation of
nitrogen in sediment by 9-44 fold compared
with undisturbed streams (Francis et aL
1985). The effects of impoundment on pH,
dissolved oxygen, fluxes of energy, other
nutrients and ions have been much studied
for C. canadensis in many parts of its range
Wilde et aL 1950; Hodkinson 1975;
and Naiman 1991; Naiman et al. 1994)
Rodent Ecosystem Relationships
and, to a lesser extent, for the related Castor
fiber in Europe (Cirmo and Driscoll1993;
Macdonald et al. 1995).
The physical structure of beaver dams,
and particularly the effects of dams on
resource flows, have important
consequences for aquatic and terrestrial
animals and riparian vegetation. In the short
term (years) impoundments may kill
streamside trees and provide nest or roost
sites for volant vertebrates following
formation of hollows. In the longer term
(decades to millenia), impoundments are
likely to be colonised by wetland plants and
follow successional pathways that may lead
to meadows, bogs or wetlands (Figure 2).
The relative roles of beaver engineering and
other physical processes such as erosion,
sedimentation and fire in directing
particular pathways remain unclear, but
likely differ between regions (Naiman et al.
1988,1994; Johnston 1995).
Damming produces a shift from lotic (fast
flowing) to more lentic (still-water)
conditions, especially in higher order
streams. Among aquatic invertebrates, this
shift favours collector and predator species
such as tubificid worms, clams and
dragonflies over shredder and scraper
species such as blackflies, scraping mayflies
and net-spinning caddisflies (McDowell and
Naiman 1986). However, lotic taxa may still
be represented highly on the dam walls,
perhaps because the dam acts as a net that
traps drifting lotic fauna (Clifford et al.
1993). Among fishes, lotic taxa give way
similarly to still-water specialists in beaver
impoundments. Species richness and
composition differ in dammed headwater
and lower-order streams and vary also with
age of the impoundment (Keast and Fox
119
Ecologically-based Rodent Management

Beaver activity Sedimentation, erosion, hydrology

Po~ O Emergent

/
wetland

~o/0 - 0 ~
o \ o Bog

St"am
t
" / 0 0ldpoo
, !
f":\
\.V
Fi~ 0 Forested
wetland
Meadow

1-100 years 50-200 years 200-2000 years

Reversible Irreversible
succession succession

Figure 2.
Potential effects of beaver ( Castor canadensis) on vegetation and landscape patterns, based on work by
R.J. Naiman and colleagues in the boreal forests of northern Minnesota (after Naiman et a!. 1988).

1990; Hagglund and Sjoberg 1999; Snodgrass
and Meffe 1998).
Descriptive and experimental studies
have suggested further that beaver ponds
act as reproductive source populations for
fish whereas adjacent streams act as sinks
(Schlosser 1995). If so, beaver dams may be
seen as important components of fish
metapopulations at catchment or larger
spatial scales.
The engineering activities of beavers
may, finally, have subtle indirect effects on
terrestrial invertebrates. Martinsen et a1.
(1998) have shown recently that resprout
growth from beaver-cut cotton wood trees
(Populus fremontii and Populus angustifolia) is
attractive to a specialist leaf beetle,
Chrysomela confluens. The beetles sequester
phenolic glycosides from the cottonwood
leaves and use them as a means of predator
defense. Martinsen et a1. (1998) asserted
further that habitat mosaics created by
beaver activity increase the diversity of
arthropods and perhaps higher vertebrates
as well, but provided no evidence in support
of this claim.
A final class of examples of complex
allogenic engineering is the surface digging
activity of rodents that results in
accumulation of organic material and
diversion of water flow . Gutterman (1982)
showed that the diggings of Indian crested
porcupines, Hystrix indica, accumulate seeds
and other organic matter, and provide

120

microhabitats favourable for the
germination and establishment of certain
species of plants. Diggings are more suitable
for germination in protected than exposed
habitats, apparently because they allow run-
off of rainfall for longer periods (Gutterman
and Herr 1981; see also Yair and Rutin 1981).
Steinberger and Whitford (1983) presented
similar findings from their work on the
surface digging activities of desert
heteromyids.
Studies of larger mammals such as brush-
tailed bettongs (Bettongia penicillata) and
grizzly bears (Ursus arctos horribilis) indicate
that surface digging activity can
dramatically decrease soil water repellency
and enhance levels of mineral nitrogen
(Garkaklis et al. 1998; Tardiff and Stanford
1998). Such effects might be predicted also
from the digging activity of larger rodents,
but do not appear yet to have been
documented.
BIOTIC ENGINEERING
Dispersal of seeds and spores
Although movements of seeds or spores
from one place to another constitute biotic
engineering as defined here, the
phenomenon is ecologically more relevant
after growth of the embryonic tissue has
become sufficient to modulate resource flow
to other organisms. Movement of seeds by
rodents is well established. In some species,
such as tropical squirrels, seeds are ingested
and later excreted elsewhere in the animals'
home ranges (Em mons 1992; MacKinnon et
al. 1996). In many other species, seeds are
collected and cached, or hoarded, for later
consumption (GumeIl1983; Reichman and
Price 1993). Seeds often survive caching to
Rodent Ecosystem Relationships
germinate and become established (Vander
Wall 1990), but the role of rodents as
dispersal agents remains poorly known. In
one particularly instructive recent study,
Vander Wall (1997) showed that some 80%
of pifion pine (PilIUS monophylla) seeds,
placed experimentally on the ground
beneath trees, were gathered by rodents.
Radioactively labelled seeds were mostly
cached, either in scatter-hoards or larders, at
distances up to 38.6 m from the source. Over
a third of caches occurred beneath shrubs;
these appeared to favour establishment, and
served as nurse plants for young pines.
Vander Wall (1997) demonstrated seed
caching by four species of rodents in
captivity-Peromyscus maniculatus,
Peromyscus truei, Perognathus parous and
Dipodomys panamintinus-and inferred that
these were the main seed dispersers in his
field site too.
Fossorial rodents have also been
demonstrated to move the storage organs of
geophytic plants, often concentrating them
within mounds or burrow systems (Galil
1967; G6mez-Garcia et al. 1995). Sprouting of
storage organs at their new locations
suggests that rodent-induced dispersal can
be effective (Borghi and Giannoni 1997).
Dispersal of fungal spores by rodents has
received relatively little attention. Many
species eat the fruiting bodies of fungi (e.g.
Maser et al. 1978; Claridge and May 1994;
Tory et aL 1997), but it has not always been
shown that ingested spores remain viable.
However, spores usually remain structurally
intact following passage through rodent
guts, and Claridge et al. (1992) showed that
spores recovered from faeces of another
mammal, Potorous tridactylus, developed
ectomycorrhizae on the roots of two species
121
Ecologically-based Rodent Management
of Eucalyptus. Importantly, the fungi
ingested by many species of rodents are
hypogeal and form mycorrhizal associations
with the roots of trees and other vascular
plants, thus potentially assisting plant
growth. Future research should seek to
clarify the extent to which rodents disperse
viable spores, and also quantify their
contribution to regeneration and
development of forest environments
(Reddell et al. 1997; Tory et aL 1997).
Pollination
Bats and primates that visit flowers for food
are often effective pollinators, especially in
tropical and arid habitats (Fleming and Sosa
1994). The effectiveness of rodents as
pollinators, however, is less clear. Many
species visit flowers and could transfer
pollen that has lodged in the fur (Recher
1981). Examples include arboreal
such as dormice-Mllscardinus avellanarius
(Bright and Morris 1996), tree-rats-Solomys
spp. (D. Fisher, pers. comm.) and desert
rodents in the genus Pseudomys
(CR. Dickman, persona] observation). Few
studies have shown that rodents carry
significant loads of pollen between flowers
(Lumer 1980; Wiens et al. 1983; Van Tets
1997) and none has yet distinguished the
relative importance of rodents as pollinators
compared with other taxa (Carthew and
Goldingay 1997). As Flerning and Sosa
(1994) point out, the genetic effects of even
the more conspicuous mammalian
pollinators and frugivores on plant
populations have been rarely investigated;
there is much scope for new research.
122
TROPHIC IMPACTS OF RODENTS
Rodents take a very broad range of plant and
animal foods, so their potential effects on
prey species and communities could be
pervasive. Some species of rodents specialise
in taking only one or two prey taxa (e.g. the
heteromyid Liomys salvini specialises
seasonally on seeds of Enterolobium
cyclocarpum, a Central American leguminous
tree; Janzen 1981), whereas others are
broadly omnivorous (e.g. many species of
Australian desert rodents; Murray et al.
1999). The direct impacts of rodent predation
have a long history of study, especially with
respect to effects on crops and other
vegetation, but indirect impacts have been
recognised increaSingly in recent work. This
is a vast topic that can only be treated
superficially here.
The best estimates of rodent impact on
food resources are from agro-ecosystems in
different parts of the world (e.g. Buckle and
Smith 1994; Singleton and Petch 1994; other
chapters in this book). In these simplified
environments, rodents can reach
extraordinary densities (e.g. >3,OOO/ha for
Mus domesticus; Caughley et aL 1998) by
eating one or a very few types of food, and
cause great damage to crops. Both native
and introduced species of rodents can
become pests, and achieve higher densities
in crop systems than in the natural
environment. Very high densities may be
achieved transiently by rodents in
unmodified environments, often following
drought-breaking rains (e.g. 1,200/ha for
R. villosissimus; Palmer 1886), but impacts on
food resources under these conditions have
been little-studied (Batzli and Pitelka 1970;
Noy-Meir 1988).

In natural or little modified environ-
ments, rodents may have local or broad-
scale effects on vegetation. Below ground,
herbivory often modifies plant community
structure, reducing the standing crop but
increasing local species richness (Andersen
1987; Huntly and Reichman 1994). Selective
foraging on individual plant species may
benefit certain life-history stages such as
seeds or small bulbs by reducing
intraspecific competition (Contreras and
Gutierrez 1991), but can also depress plant
biomass and flower production (Reichman
and Smith 1991) or even result in local plant
extinction (Cantor and Whitham 1989).
Above ground, rodent herbivory (including
frugivory and granivory) has even more
dramatic effects on vegetation. Selective
foraging may again deplete favoured species
in local areas, and alter trajectories of plant
succession Uohnston and Naiman 1990).
Generalist foraging has been shown to have
pervasive effects on life form, growth,
allocation of nutrients and energy stores
within plants, as well as on the physical
structure and species composition of plant
communities (Batzli and Pitelka 1970; Brown
et a1. 1979; Brown and Heske 1990; Holland
et a1. 1992; Jefferies et a1. 1994; but d. Gibson
et a1. 1990). Although this topic is too broad
to discuss fully here, the effects and
mechanisms by which herbivores affect
plant communities have been reviewed by
Crawley (1983) and Huntly (1991), and the
induction of plant defenses has been
reviewed by Karban and Myers (1989).
Short-term feed backs and longer-term
coevolution between herbivorous rodents
and plants also have been discussed in detail
elsewhere (Crawley 1983; Coley and Barone
1996; Pastor et a1. 1997).
Rodent Ecosystem Relationships
Perhaps because the impacts of rodents
on vegetation are often obvious and
economically relevant, the effects of rodents
on other food groups have been seldom
studied. However, limited experimental
evidence suggests that high density
populations of omnivorous species may
deplete the local richness of epigeal
invertebrates (Figure 3). On Boullanger
Island, Western Australia, invertebrate
species richness increased on average by 3%
on plots from which M. domestic us had been
removed, in contrast to a decrease of 18% on
control plots (Figure 3a). Increases occurred
primarily in beetle and spider species, which
the mice ingested (CR. Dickman, personal
observation). In urban woodland in the
United Kingdom, invertebrate species
richness increased similarly by 83% on plots
from which Apodemus sylvaticus had been
removed, compared with only a 32% increase
on control plots (Figure 3b). Increases
occurred in species of beetles, spiders and
snails-taxa found commonly in the diet of
urbanA. sylvaticlIs (CR. Dickman, personal
observation). Primarily insectivorous
rodents such as grasshopper mice
(Onychomys spp.) likely affect individual
species and communities of invertebrates at
times also, but evidence is lacking.
In circumstances when omnivorous
rodents have been introduced to new
environments, they have sometimes had
dramatic effects on populations of
invertebrates and small vertebrates. On Lord
Howe Island, for example, an endemic
phasmid, Dryococelus austraIis, disappeared
following establishment of Rattus rattus,
while numbers of two species of island snails
were severely depressed (Smithers et aL
1977).
123
Ecologically-based Rodent Management

60

(/)
(/) 50
Cl>
c
.<::
u
.'"
(/)
40
Cl>
0
~ 30
(/)

Cl>
.~
Bi:::J 20
E
:::J
0 10

0
Apri l May June July August

Figure 3.
Effects of rodent removal on species richness of invertebrates. (a) Mus domesticus was removed from
trapping plots on Boullanger Island, Western Australia, and invertebrates sampled by pitfall trapping before
and after removal in both t he removal and control sites ( n = 3 control, 3 removal plots, means shown
standard error (SE); mean before/after ratios of species richness differed significantly bet ween control and
removal treatments, P < 0.0 5) .

50

(/)
(/)
Cl>
c
.<::
40 Control
Rem oval

.'"u
(/)
.~
30
u
Cl>
0-
(/)

Cl>
> 20
~
:::J
E
:::J
0 10

0
May June July August

(b) Apodemus sylvaticus was removed from trapping plots in urban woodland in Oxford, United Kingdom,
and invertebrates sampled in the same manner as in (a) (n = 3 control, 3 removal plots, means shown SE;
mean before/after ratios of species richness differed significantly between control and removal treatments,
P <0.01) . Methodological details are given in Dickman (1988), Dickman and Doncaster (1989; also
unpublished data).

124
 Rodent Ecosystem Relationships

Extinctions and range contractions of Finally, while most research has

many other species of large invertebrates
and small vertebrates, including seabirds
and flightless birds, have occurred on
islands off the coast of New Zealand and
throughout the Pacific following
introductions of R. rattus, Rattus exulans and
Rattus norvegicus (Steadman 1989; King
1990). Although rats appear to have been the
only obvious threat introduced to some
islands, in many cases their impa ct is
difficult to distinguish from the effects
wrought by other introduced species and by
habitat change.
It remains equivoca l also whether rat
impacts were caused by predation,
competition, introduction of diseases or
other processes, although direct predation
has been implicated by most authors
(Smithers et al. 1977; King 1990; Atkinson
1996).
evaluated the direct trophic impacts of
rodents, some recent work indicates that
rodent foraging may have far-reaching
indirect effec ts. In Californian grassland,
Batzli and Pitelka (1970) showed that forb
and grass cover increased in plots that
excluded the herbivorous meadow vole,
Microtus californicus, as com pared with cover
levels in control plots. An indirect effect of
the vole exclusion w as a dramatically
increased abundance of the pillbug
(Annadillidium vulgm'e) within two years; this
species was apparently favoured by the
d enser vegetation or increased food
resources that it contained. In analogous
experiments, removal of M. domesticus from
plots on Bou llanger Island resulted in a 24%
i.ncrease in li tter depth within just three
months, compared with a 16% decrease in
litter d epth on control plots over the same
period (Figure 4).

2.5
Control
T

2 Removal
E
~
.c
Cl. 1.5
Ql
<J
Cii
>-
.!!1
~
:.:J
05

0
April May June July August

Figure 4.
Effects of removal of Mus domesticus on depth of the leaf litter layer on Boullanger Island, Western Australia
(n = 3 control, 3 removal plots, means shown standard error; before/after ratios of mean litter depth
differed significantly between control and removal treatments, P < 0.01). Further details are given in
Dickman (1988).

125
Ecologically-based Rodent Management
Capture rates of the skinks (Ctenotus fallens
and Morethia lineoocellata) increased in the
Mus-removal plots by up to 35% (CR.
Dickman, personal observation), presumably
because of the increased shelter afforded by
the deep leaf litter or the more diverse food
resources that were available (Figure 3a).
Over larger periods, rodent foraging can
indirectly facilitate other taxa. In the
Chihuahuan Desert, Thompson et al. (1991)
demonstrated that the foraging activities and
abundance of granivorous birds declined
markedly in rodent exclusion plots over a
period of 10 years. In the absence of rodents,
especially kangaroo rats (Dipodomys spp.),
litter accumulated and concealed seeds on the
soil surface from the view of the visually
foraging birds. In control plots by contrast,
rodent foraging activities created areas of
bare soil and trails, hence exposing seeds and
facilitating access by birds. Brown and Heske
(1990) considered kangaroo rats in the
Chihuahuan Desert to be a keystone guild in
recognition of their major direct and indirect
effects on biological diversity and
biogeochernical processes. More complex
webs of direct and indirect effects of rodent
foraging are suspected (e.g. Klaas et al. 1998),
and will require much ingenuity to study and
understand.
CONCLUSIONS
This review shows that rodents interact
extensively with their physical, chemical
and biotic environments, and that their
activities have complex but often beneficial
effects on other organisms across a broad
range of spatial and temporal scales. This
should not be surprising, because of the
great species richness, abundance and
126
ubiquity of rodents in terrestrial
environments.
With respect to the management of
rodent pests, the review also allows the
conclusion to be drawn that we must be
more clever and more focused in our
approaches to rodent control. In many
regions, broad-scale application of poison
remains the favoured control method
(Buckle and Smith 1994; Singleton and Petch
1994). In the 1993 mouse plague in south-
eastern Australia, for example, some 350,000
ha of cropland in South AQstralia alone were
baited with the poison strychnine (Caughley
et al. 1994). Such broad-scale campaigns may
reduce the numbers of the target pest, but
very likely decimate populations of non-
target species (Dickman 1993), including
those with potentially positive effects on the
physical and biotic environment. In the
wheat-growing areas of New South Wales,
native rodents have been virtually
eliminated by introduced species, changes in
land use, and perhaps also by agrochemicals
that are used to maintain the
(Dickman 1993). It is clear that management
of rodent pests will need to eschew its
damaging reliance on chemical warfare and
embrace sustainable, ecologically-based
solutions. Heartening moves in this
direction include fertility control (Chambers
et al., Chapter 10), mortality control via
predators or parasites that target pest taxa
(Buckle and Smith 1994), and physical
barrier methods that limit access of pests to
crop areas (Singleton et al., Chapter 8).
ACKNOWLEDGMENTS
I thank G. Singleton and L. Hinds for the
opportunity to prepare this review, and

Z. Zhang and other members of the Beijing
organising committee for hosting the
conference on rodent biology and
management. R. Baxter, c.-L. Beh, R. Shine,
N. Shchipanov and G. Sjoberg helped to find
elusive references or kindly provided reprints
of the manuscripts, E. Gerzabek and G.
McNaught assisted with preparation of this
manuscript, and P. Brown, C. McKechnie and
G. Singleton provided critical comments. I am
grateful also to the Australian Research
Council for funding my work on rodents, and
to Commonwealth Scientific and Industrial
Research Organisation (CSIRO) Wildlife and
Ecology for funding my attendance at the
Beijing conference. This paper has benefited
from discussions with numerous conference
participants.
REFERENCES
Abbott, RJ., Bevercombe, C.P. and Rayner,
AD.M. 1977. Sooty bark disease of sycamore
and the grey squirrel. Transactions of the
British Mycological Society, 69, 507-508.
Agnew, W., Uresk, O.W. and Hansen, RM.1986.
Flora and fauna associated with prairie dog
colonies and adjacent ungrazed mixed-grass
prairie in western South Dakota. Journal of
Range Management, 39, 135-139.
Andersen, D.C.1987. Below-groundherbivoryin
natural communities: a review emphasizing
fossorial animals. Quarterly Review of
Biology, 62,261-286.
Anstee, S.D., Roberts, }.O. and O'Shea, J.E.1997.
Social structure and patterns of movement of
the western pebble-mound mouse,
Pseudomys chapmani, at Marandoo, Western
Australia. Wildlife Research, 24, 295-305.
Atkinson, LAE. 1985. The spread of commensal
species of Rattus to oceanic islands and their
effects on island avifaunas. In: Moors, P.J., ed.,
Conservation of island birds. Cambridge,
u.K., International Council for Bird Preserva-
tion, 35-81.
Rodent Ecosystem Relationships
Atkinson, LAE. 1996. Introductions of wildlife as
a cause of species extinctions. Wildlife
Biology, 2, 135-141.
Barnett, S.A 1975. The rat. Chicago, U.5.A,
University of Chicago Press 318p.
t
Batzli, G.O. and Pitelka, F.A 1970. Influence of
meadow mouse populations on California
grassland. Ecology, 51, 1027-1039.
Beuchner, H.K. 1942. Interrelationships between
the pocket gopher and land use. Journal of
Mammalogy, 23, 346-348.
Borghi, c.E. and Giannoni, S.M. 1997. Dispersal
of geophytes by mole-voles in the Spanish
Pyrenees. Journal of Mammalogy, 78, 550-
555.
Bright, P.W. and Morris, P.A 1996. Why are
dormice rare? A case study in conservation
biology. Mammal Review, 26, 157-187.
Bronner, GN.1992. Burrow system characteris-
tics of seven small mammal species (Mamma-
lia: Insectivora; Rodentia; Carnivora).
Koedoe,35,125-128.
Brown, J.H. and Harney, B.A 1993. Population
and community ecology of heteromyid
rodents in temperate habitats. In: Cenoways,
H.H. and Brown, J.H., ed., Biology of the
Heteromyidae. Provo, U.5.A., The American
Society of Mammalogists, 618-651.
Brown, J.H. and Heske, KJ. 1990. Control of a
desert-grassland transition by a keystone
rodent guild. Science, 250,1705-1707.
Brown J.H., Reichman, O.J. and Davidson, D.W.
t
1979. Granivory in desert ecosystems. Annual
Review of Ecology and Systematics, 10,201-
227.
Buckle, AP. and Smith, RH. (ed.) 1994. Rodent
pests and their control. Wallingford, U .K.,
CAB International, 405p.
Cantor L.F. and Whitham, T.C.1989. Importance
t
of below ground herbivory: pocket gophers
may limit aspen to rock outcrop refugia.
Ecology, 70, 962-970.
Carthew, S.M. and Goldingay, RL. 1997. Non-
flying mammals as pollinators. Trends in
Ecology and Evolution, 12, 104-108.
Caughley, J., Bomford, M., Parker, B., Sindair, R,
Griffiths, J. and Kelly, D. 1998. Managing
vertebrate pests: rodents. Canberra,
127
Ecologically-based Rodent Management
Australia, Bureau of Resource Sciences and
Grains Research and Development Corpora-
tion, 130p.
Caughley, J., Monamy, V. and Heiden, K. 1994.
Impact of the 1993 mouse plague. Canberra,
A ustra lia, Grains Research and Development
Corporation, 73p.
Chew, RM. and Whitford, W.G. 1992. A long-
term positive effect of kangaroo rats
(Dipodomys spectabilis) on creosotebushes
(Larrea tridentata). Journal of Arid Environ-
ments, 22, 375-386.
Chitty, D. and Southern, H.N. (ed.) 1954. Control
of rats and mice, 3 volumes. Oxford, U.K.,
Clarendon Press, 305,532, 225p.
Cirmo, CP. and Driscoll, CT. 1993. Beaver pond
geochemistry-acid neutralising capacity
generation in a headwater wetland.
Wetlands, 13,277-292.
Claridge,A.W. and May, T.W.1994.Mycophagy
among Australian mammals. Australian
Journal of Ecology, 19,251-275.
Claridge, A.W" Tanton, M.T., Seebeck, J.H.,
Cork, and Cunningham, RB. 1992. Estab-
lishment of ectomycorrhizae on the roots of
two species of Eucalyptus from fungal spores
contained in the faeces of the long-nosed
potoroo (Potorous tridactylus). Australian
Journal of Ecology, 17,207-217.
Clifford, H.P., Wiley, G.M. and Casey, RJ. 1993.
Macroinvertebrates of a beaver-altered boreal
stream of Alberta, Canada, with special refer-
ence to the fauna on the dams. Canadian
Journal of Zoology, 71, 1439-1447.
Coley, P.D. and Barone, J.A. 1996. Herbivory and
plant defenses in tropical forests. Annual
Review of Ecology and Systematics, 27, 305-
335.
Contreras, L.C and Gutierrez, J.R 1991. Effects
of the subterranean herbivorous rodent
SpaZacopus cyanus on herbaceous vegetation
in arid coastal Chile. Oecologia, 87, 106-109.
Copley, P. 1988. The stick-nest rats of Australia.
Adelaide, Australia, Department of Environ-
ment and Plalming, South Australia, 277p.
Coppock, D.L., Detling, J.K., Ellis, J.E. and Dyer,
M.I. 1983. Plant-herbivore interactions in a
North American mixed-grass prairie I. Effects
of black-tailed prairie dogs on intraseasonal
128
aboveground plant biomass and nutrient
dynamics and plant species diversity.
Oecologia, 56, 1-9.
Corbet, G.B. and Harris, S., (ed.) 1991. The
handbook of British mammals. Oxford, U.K.,
Blackwell Scientific Publications, 588p.
Corbet, G.B. and Hill, J.E. 1991. A world list of
mammalian species. London, U.K., "\ratural
History Museum Publications and Oxford
University Press, 243p.
Cox, G.W. and Gakahu, CG. 1985. Mima mound
microtopography and vegetation pattern in
Kenyan savannas. Journal of Tropical
Ecology, 1,23-36.
Cox, G.W. and Gakahu, CG. 1986. A latitudinal
test of the fossorial rodent hypothesis of
mima mound origin. Zeitschrift fill Geomor-
phologie, 30,485-501.
Cox, G.W., Lovegrove, B.G. and Siegfried, W.R
1987. The small stone content of mima-like
mounds in the South African Cape region:
implications for mound origin. Catena, 14,
165-176.
Crawley, M.J. 1983. Herbivory: the dynamics of
animal-plant interactions. Oxford, U.K.,
Blackwell Scientific Publications, 437p.
Dahlsted, K.J., Sather-Blair, 5., Worcester, B.K.
and Klukas, R 1981. Application of remote
sensing to prairie dog management. Journal
of Range Management, 34, 218-223.
Dickman, CR 1988. Body size, prey size, and
community structure in insectivorous
mammals. Ecology, 69, 569-580.
Dickman, CR 1993. The biology and manage-
ment of native rodents of the arid zone in
New South Wales. Sydney, Australia, New
South Wales National Parks and Wildlife
Service, 149p.
Dickman, CR 1996. Vagrants in the desert.
Nature Australia, 25, 54-62.
Dickman, CR and Doncaster, CP. 1989. The
ecology of small mammals in urban habitats.
n. Demography and dispersal. Journal of
Animal Ecology, 58, 119-127.
Dickman, CR and Read, D.G. 1992. The biology
and management of dasyurids of the arid
zone in New South Wales. Sydney, Australia,
New South Wales National Parks and
Wildlife Service, 112p.

Ebenhard, T. 1988. Introduced birds and
mammals and their ecological effects.
Viltrevy, 13, 1-107.
Emmons, L.H. 1992. The roles of small mammals
in tropical rainforest. In: Ismail,
Mohamed, M. and Omar, S., ed., Forest
biology and conservation in Borneo. Kota
Kinabalu, Sabah, Center for Borneo Studies,
512-513.
Evgenjeva, T.P. and Fadeeva, E.O.1996. The
distant consequence of defoliant effect on
cytogenetic indexes of cornea cells in some
rodents from South Vietnam. In: Sokolov,
V.E. and Shilova, S.A, ed., Long-term biolog-
ical consequences of the war in South
Vietnam. Moscow, Russia, Tropical Depart-
ment, Severtsov Institute of Ecology and
Evolution, Russian Academy of Sciences,
188-200.
Flemin g, T .H. and Sosa, V.J. 1994. Effects of
nectarivorous and frugivorous mammals on
reproductive success of plants. Journal of
Mammalogy, 75, 845-851.
Francis, M.M., Naiman, RJ. and Melillo, J.M.
1985. Nitrogen fixation in subarctic streams
influenced by beaver (Castor canadensis).
Hydrobiologia, 121, 193-202.
Frey, J.K. 1992. Response of a mammalian faunal
element to climatic changes. Journal of
Mammalogy, 73, 43-50.
Galil, J. 1967. On the dispersal of the bulbs of
Oxalis cernua Thunb, by mole-rats (Spa lax
ehrenbergi Nehring). Journal of Ecology, 55,
787-792.
Garkaklis, M.L Bradley, J.S. and Wooller, RD.
1998. The effects of woylie (Bettongia penidl-
lata) foraging on soil water repellency and
water infiltration in heavy textured soils in
southwestern Australia. Australian Journal
of Ecology, 23, 492-496.
Gibson, DJ., Freeman, CC and Hulbert, L.C
1990. Effects of small mammal and inverte-
brate herbivory on plant species richness and
abundance in tallgrass prairie. Oecologia, 84,
169-175.
G6mez-Garcia, D., Borghi, CE. and Giannoni,
S.M. 1995. Vegetation changes on subalpine
plant communities induced by pine vole
mounds. Vegetatio, 117, 61-67.
Rodent Ecosystem Relationships
Gurnell, J. 1983. Squirrel numbers and the
abundance of tree seeds. Mammal Review,
13,133-148.
Gutterman, Y. 1982. Observations on the feeding
habits of the Indian crested porcupine
(Hystrix indica) and the distribution of some
hemicryptophytes and geophytes in the
N egev Desert highlands. Journal of Arid
Environments, 5,261-268.
Gutterman, Y. and Herr,~. 1981. Influences of
porcupine (Hystrix indica) activity on the
slopes of the northern Negev mountains-
germination and vegetation renewal in
different geomorphological types and slope
directions. Oecologia, 51, 332-334.
Hiigglund, A. and Sjoberg, G. 1999. Effects of
beaver dams on the fish fauna of forest
streams. Forest Ecology and Management,
115,259-266.
Hansell, M.H. 1993. The ecological impact of
animal nests and burrows. Functional
Ecology, 7,5-12.
Harris, S. and Trout, R.C 1991. Genus Micromys.
In: Corbet, G.B. and Harris, S., ed., The
handbook of British mammals. Oxford, U.K.,
Blackwell Scientific Publications, 233-239.
Hartnett, D.C and Keeler, K.H. 1995. Population
processes. In: Joern, A and Keeler, K.H., ed.,
The changing prairie: North American grass-
lands. New York, C.s.A, Oxford University
Press, 82-99.
Heth, G. 1991. The environmental impact of
subterranean mole rats (spaZax ehrenbergi) and
their burrows. Symposia of the Zoological
Society of London, 63, 265-280.
Hobbs, RJ. and Hobbs, V.J. 1987. Gophers and
grassland: a model of vegetation response to
patchy soil disturbance. Vegetatio, 69, 141-
146.
Hobbs, R.J. and Mooney, H.A. 1991. Effects of
rainfall variability and gopher disturbance on
serpentine annual grassland dynamics.
Ecology, 72, 59-68.
Hodkinson, I.D. 1975. Energy flow and organic
matter decomposition in an abandoned
beaver pond ecosystem. Oecologia, 21, 131-
139.
129
Ecologically-based Rodent Management
Holland, E.A., Parton, W.]., Detling, J.K and
Coppock, D.L. 1992. Physiological responses
of plant populations to herbivory and their
consequences for ecosystem nutrient flow.
American Naturalist, 140,685-706.
Hoogland, J.L. 1994. The black-tailed prairie
dog: social life of a burrowing mammal.
Chicago, U.S.A., University of Chicago Press,
576p.
Huntly,N.1991. Herbivores and the dynamics of
communities and ecosystems. Annual
Review of Ecology and Systematics, 22, 477-
503.
Huntly, N. and Inouye, R 1988. Pocket gophers
in ecosystems: patterns and mechanisms.
Bioscience, 38, 786-793.
Huntly, N. and Reichman, O.J. 1994. Effects of
subterranean mammalian herbivores on
vegetation. Journal of Mammalogy, 75, 852-
859.
Ingham, RE. and Detling, J.K. 1984. Plant-
herbivore interactions in a North American
mixed-grass prairie Ill. Soil nematode
populations and root biomass on Cynomys
ludovicianus colonies and adjacent uncolo-
nized areas. Oecologia, 63, 307-313.
Inouye, RS., Buntly, N.J. and Tilman, D.1987a.
Response of Microtus pennsylvanicus to
vegetation fertilized with various nutrients,
with particular emphasis on sodium and
nitrogen concentrations in plant tissues.
Holarctic Ecology, 10, 110--113.
Inouye, RS., Huntly, N.J., Tilman, D. and Tester,
J.R 1987b. Pocket gophers (Geomys bursarius),
vegetation, and soil nitrogen along a succes-
sional sere in east central Minnesota. Oecolo-
72,178-184.
Inouye, RS., Huntly, N. and Wasley, G.A.1997.
Effects of pocket gophers (Geomys bursarius)
on microtopographic variation. Journal of
Mammalogy, 78, 1144-1148.
Janzen, D.H. 1981. Patterns of herbivory in a
tropical decid uous forest. Biotropica, 13, 271-
282.
Jarvis, J.U.M. and Sale, J.B. 1971. Burrowing and
burrow patterns of East African mole-rats
Tachyoryctes, Heliophobius, and Heterocephalus.
Journal of Zoology, London, 163, 451-479.
130
Jefferies, RL., Klein, D.R. and Shaver, G.R 1994.
Vertebrate herbivores and northern plant
communities: reciprocal influences and
responses. Oikos, 71, 193-206.
Johnston, CA. 1995. Effects of animals on
landscape pattern. In: Hansson, L., ed.,
Mosaic landscapes and ecological processes.
London, U.K., Chapman and Hall, 57-80.
Johnston, CA. and Naiman, RI. 1987. Boundary
dynamiCS at the aquatic -terrestrial interface:
the influence of beaver and geomorphology.
Landscape Ecology, 1, 47-57.
Johnston, CA. and Naiman, RJ. 1990. Browse
selection by beaver: effects on riparian forest
composition. Canadian Journal of Forest
Research, 20,1036-1043.
Jones, CG., Lawton, J.H. and Shachak, M. 1994.
Organisms as ecosystem engineers. Oikos, 69,
373-386.
lones, CG., Lawton, J.H. and Shachak, M. 1997.
Positive and negative effects of organisms as
physical ecosystem engineers. Ecology, 78,
1946-1957.
Karban, Rand Myers, J.H. 1989. Induced plant
responses to herbivory. Annual Review of
Ecology and Systematics, 20, 331-348.
Keast, A. and Fox, M.G. 1990. Fish community
structure, spatial distribution and feeding
ecology in a beaver pond. Environmental
Biology of Fishes, 27, 201-214.
Kemper, C1981. Description of Pseudomys novae-
hollandiae burrows located with radioiso-
topes. Australian Mammalogy, 4, 141-143.
King, CM. (ed.) 1990. The handbook of New
Zealand mammals. Auckland, New Zealand,
Oxford University Press, 600p.
Kiviat, E. 1978. Vertebrate use of muskrat lodges
and burrows. Estuaries, 1, 196-200.
Klaas, B.A., Danielson, B.J. and Moloney, KA.
1998. Influence of pocket gophers on meadow
voles in a tallgrass prairie. Journal of
Mammalogy, 79, 942-952.
Knight, M.H. 1984. The ecophysiology of the
African giant rat Cricetomys gambianus
(Waterhouse). Pretoria, South Africa, Univer-
sity of Pretoria, unpublished MSc. disserta-
tion.

Kostelecka-Myrcha, A., Myrcha, A. and
Gutkowska, H.E. 1981. Hematological and
morphophysiological indices of some
rodents under conditions of varying indus-
trial pressure. Polish Ecological Studies, 7,
145-153.
Lepage,P. and Parker, G.H.1988. Copper, nickel,
and iron levels in peIage of red squirrels
living near the ore smelters at Sudbury,
Ontario, Canada. Canadian Journal of
Zoology, 66, 1631-1637.
Lumer, C 1980. Rodent pollination of Blakea
(Melastomataceae) in a Costa Rican cloud
forest. Brittonia, 32, 512-517.
Macdonald, nw., Tattersall, F.H, Brown, E.D.
and Balharry, D. 1995. Reintroducing the
European beaver to Britain: nostalgic
meddling or restoring biodiversity? Mammal
I~eview, 25,161-200.
MacKinnon, K, Hatta, G., Halim, Hand Manga-
lik, A. 1996. The ecology of Kalimantan. Hong
Kong Hong Kong, Periplus Editions,
802p.
Madsen, T. and Shine, R 1999. Rainfall and rats:
climatically-driven dynamics of a tropical
rodent population. Australian Journal of
Ecology, 24, 80-89.
Martinsen, G.D., Cushman, J.H and Whitham,
T.G. 1990. Impact of pocket gopher distur-
bance on plant species diversity in a short-
grass prairie community. Oecoiogia, 83,132-
138.
Martinsen, G.D., Driebe, E.M. and Whitham,
T.G. 1998. Indirect interactions mediated by
changing plant chemistry: beaver browsing
benefits beetles. Ecology, 79, 192-200.
Maser, C, Trappe, J.M. and Nassbaum, RA.
1978. Fungal-small mamma 1interrelation-
ships with emphasis on Oregon coniferous
forests. Ecology, 54, 799-809.
McDowell, D.M. and Naiman, RJ. 1986. Struc-
ture and function of a benthic invertebrate
stream community as influenced by beaver
(Castor canadensis). Oecologia, 68, 481-489.
McNaught,G.H 1994. The foraging behaviour of
an Australian desert rodent, Notomys alexis
(Family Muridae). Sydney, Australia,
University of Sydney, unpublished Honours
dissertation.
Rodent Ecosystem Relationships
Meadows, P.S. and Meadows, A. (ed.) 1991. The
environmental impact of burrowing animals
and animal burrows. Oxford, U.K, Claren-
don Press.
Medway, Lord 1983. The wild mammals of
Malaya (Peninsular Malaysia) and Singapore.
Kuala Lumpur, Malaysia, Oxford University
Press, 128p.
Mielke, H.W. 1977. Mound building by pocket
gophers (Geomyidae): their impact on soils
and vegetation in North America. Journal of
Biogeography,4,171-180.
Moloney, KA., Levin, S.A., Chiariello, N.R and
Butt~l, L. 1992. Pattern and scale in a serpen-
tine grassland. Theoretical Population
Biology, 41, 257-276.
Murray, B.R, Dickman, CR, Watts, CH.S. and
Morton, S.H. 1999. Dietary ecology of Austral-
ian desert rodents. Wildlife Research, 26, 421-
437.
Naiman,RJ.,Johnston, CA. and Kelly,J.C1988.
Alteration of North American streams by
beaver. Bioscience, 38, 753-762.
Naiman, RJ., Manning, T. and Johnston, CA.
1991. Beaver population fluctuations and
tropospheric methane emissions in boreal
wetlands. Biogeochemistry, 12, 1-15.
Naiman, RJ., Melillo, J.M. and Hobbie, J.E. 1986.
Ecosystem alteration of boreal forest streams
by beaver (Castor canadell"is). Ecology, 67,
1254-1269.
Naiman, RJ., Pinay, G., Johnston, CA. and
Pastor, J. 1994. Beaver influences on the long-
term biogeochemical characteristics of boreal
forest drainage networks. Ecology, 75, 905-
921.
Noy-Meir, 1.1988. Dominant grasses replaced by
ruderal forbs in a vole year in undergrazed
mediterranean grasslands in Israel. Journal of
Biogeography, 15,579-587.
Palmer, E. 1886. Notes on a great visitation ofrats in
the north and north-western plain country of
Queensland, in 1869 and 1870. Proceedings of
the Royal Society of Queensland, 1885, 193-198.
Pastor, J., Moen, Rand Cohen, Y. 1997. Spatial
heterogeneities, carrying capacity, and
feedbacks in animal-landscape interactions.
Journal of Mammalogy, 78, 1040-1052.
131
Ecologically-based Rodent Management
Pinay, G. and Naiman, RI. 1991. Short-term
hydrologic variations and nitrogen dynamics
in beaver created meadows. Archiv fur
Hydrobiologie, 123, 187-205.
Predavec, M. and Dickman, C.R 1994. Popula-
tion dynamics and habitat use of the long-
haired rat (Rattus villosissimus) in south-
western Queensland. Wildlife Research, 21,
1-10.
Recher, H.F. 1981. Nectar-feeding and its evolu-
tion among Australian vertebrates. In: Keast,
A., ed., Ecological biogeography of Australia.
The Hague, Netherlands, Dr W. Junk, 1637-
1648.
Reddell, P., Spain, A.V. and Hopkins, M. 1997.
Dispersal of spores of mycorrhizal fungi in
scats of native mammals in tropical forests of
north eastern Australia. Biotropica, 29, 184-
192.
Reichman, O.J. and Price, M.V. 1993. Ecological
aspects of heteromyid foraging. In:
Genoways, H.H. and Brown, J.H., ed.,
Biology of the Heteromyidae. Provo, USA.,
The American Society of Mammalogists, 539-
574.
Reichman, O.J. and Smith, S.c. 1990. Burrows
and burrowing behavior by mammals.
Current Mammalogy, 2, 197-244.
Reichman, O.J. and Smith, S.C.1991. Responses
to simulated leaf and root herb ivory by a
biennial, Tragopogon dubius. Ecology, 72,116-
124.
Ross, B.A., Tester, J.R and Breckenridge, W.J.
1968. Ecology of mima-type mounds in north-
western Minnesota. Ecology, 49, 172-177.
Schlosser, LJ. 1995. Dispersal, boundary
processes, and trophic-level interactions in
streams adjacent to beaver ponds. Ecology,
76,908-925.
Sinclair, A.R.E. 1989. Population regulation in
animals. In: Cherrett, J.M., ed., Ecological
concepts. Oxford, U.K., Blackwell Scientific
Publications, 197-241.
Singleton, G.R. and Petch, D.A.1994. A review of
the biology and management of rodent pests
in southeast Asia. Canberra, Australia,
Australian Centre for International Agricul-
tural Research, 6Sp.
132
Skinner, J.D. and Smithers, R.H.N. 1990. The
mammals of the southern African subregion.
Pretoria, South Africa, University of Pretoria,
769p.
Smithers,C., McAlpine, D.,Colman, P. and Gray,
M.1977. Island invertebrates. In: Smith, N.,
ed.,. Lord Howe Island. Sydney, Australia,
The Australian Museum, 23-26.
Snodgrass, J.W. and Meffe, GK. 1998. Influence
of beavers on stream fish assemblages: effects
of pond age and watershed position.
79, 928-942.
Sokolov, V.E., Shilova, S.A. and Shchipanov,
N.A. 1994. Peculiarities of small mammals
populations as criteria for estimating anthro-
pogenic impacts on tropical ecosystems.
International Journal of Ecology and
Environmental Sciences, 20, 373-386.
Steadman, D. W. 1989. Extinction of birds in
eastern Polynesia: a review of the record, and
comparisons with other Pacific Island
groups. Journal of Archaeological Science, 16,
177-205.
Steinberger, Y. and Whitford, W.G. 1983. The
contribution of rodents to decomposition
processes in a desert ecosystem. Journal of
Arid Environments, 6, 177-181.
Tardiff, S.B. and Stanford, J.A.199B. Grizzly bear
digging: effects on subalpine meadow plants
in relation to mineral nitrogen availability.
Ecology, 79, 2219-2228.
Thompson, D. B., Brown, J.H. and Spencer, W.D.
1991. Indirect facilitation of granivorous birds
by desert rodents: experimental evidence
from foraging patterns. Ecology, 72, 852-863.
Tory, M.K., May, T.W., Keane, P.J. and Bennett,
A.F. 1997. Mycophagy in small mammals: a
comparison of the occurrence and diversity of
hypogeal fungi in the diet of the long-nosed
potoroo PotOtOllS tridactylus and the bush rat
Rattus fuscipes from southwestern Victoria,
Australia. Australian Journal of Ecology, 22,
460--470.
Twigg, G.!. 1978. The role of rodents in plague
dissemination: a worldwide review.
Mammal Review, 8, 77-110.
Vander Wall, S.B. 1990. Food hoarding in
animals. Chicago, USA., University of
Chicago Press, 44Sp.

Vander Wall, S.B. 1997. Dispersal of singleleaf
pifion pine (Pinus monophylla) by seed-
caching rodents. Journal of Mammalogy, 78,
181-19l.
Van Tets, LG. 1997. Extraction of nutrients from
Protea pollen by African rodents. Belgian
Journal of Zoology, 127, 59-65.
Whicker, AD. and Detling, J.K. 1988. Ecological
consequences of dog disturbances.
Bioscience, 38, 778-785.
Whittaker,J.C, List, E., Tester,J.R. and Christian,
D.P. 1991. Factors influencing meadow vole,
Microtus pennsylvanicus, distribution in
Minnesota. Canadian Field-Naturalist, lOS,
403-405.
Wiens, D., Rourke, J.P., Casper, B.B., Rickard,
E.A, La Pine, T.R., Peterson, CJ. and
Channing, A 1983. Non-flying mammal
pollination of southern African proteas; a
non-coevolved system. Annals of the
Missouri Botanical Gardens, 70, 1-31.
Rodent Ecosystem Relationships
Wilde, S.A, Youngberg, CT. and Hovind, J.H.
1950. Changes in composition of ground
water, soil fertility, and forest growth
produced by the construction and removal of
beaver dams. Journal of Wildlife Manage-
ment, 14, 123-128.
Yair, A and Rutin, J. 1981. Some aspects of the
regional variation in the amount of available
sediment produced by isopods and porcu-
pi.nes, northern Negev, Israel. Earth Surface
Processes and Landforms,6, 221-234.
Yavitt, J.B., Angell, L.L., Fahey, n., Cirmo, CP.
and Driscoll, CT. 1992. Methane fluxes,
concentrations, and production i.n two
Adirondack beaver impoundments. Limnol-
ogy and Oceanography, 37, 1057-1066.
133
 6. The Role of Rodents in Emerging Human
Disease: Examples from the
Hantaviruses and Arenaviruses

James N. Mills

Abstract

Becau se of the severity and the dramatic nature of the diseases they caus e, the
rodent-borne hae morrh agic fever viruses recently have received considerable
attenti on from ecologi sts and he alth scientists . During the past five years,
re searchers have identified at least 25 'new ' hantaviruses and aren aviru ses, all
associated with murid rodents, and coevolutionary theory suggests th at many
add itional virus-host associations await discovery. Basic research on t he eco logy of
hantavi rus and are navirus rese rvoir spec ies is providing information of practical
im portance fo r reservoir control and disease prevention. Stu di es of rese rvoir
geograph ic distribution and hab itat associations help define potenti al disease-
endem ic are as and more prec isely identify the spacial variation in rel ative ri sk to
hu mans. Cross-sectional and longitudin al studies of reservoir popul ations help
define mechan isms of viral tran smiss ion and identify the rel ationship between
environment, rese rvoir popul ations, and human disease. Integrated results from a
variety of reservoir studies can be combined with data from satell ite im ages to
provide models t hat can help scientists pred ict specifi c times and place s of
increased ris k to human pop ulation s. Biologists and pest co ntrol specialists who
work wi t h reservoir speci es may be at increased risk of infection with ro dent-borne
viru ses unless appropriate safety guideli nes are followed.

Keywords

Ro dents, infectious disease, rodent-borne disease, zoonoses, haemo rrhagic fever,

arenavirus , hantavirus

134
 The Role of Rodents in Emerging Human Disease

INTRODUCTION and the prevention of human disease; (4)

discuss the implications of our increasing
awareness of severe rodent-borne diseases
s THE chapters in this volume and the understanding of their transmission

A amply illustrate, rodents-as

pests -are responsible for
considerable economic loss, through
patterns in terms of the safety measures
recommended for mammalogists and
vertebrate pest control specialists working in
damage to crops, food stores and human disease-endemic areas; and (5) provide an
property. Rodents, as carriers of diseases up-to-date list of the rodent-borne
transmitted to humans (rodent-borne haemorrhagic fever viruses, their reservoir
zoonoses), are also responsible for hosts, and the host distributions, so that
considerable economic loss in terms of rodent biologists can assess their particular
decreased worker productivity and health- risk and take appropriate precautions.
care costs. The most important negative
impact of rodent-borne diseases, the loss of EMERGING INFECTIOUS DISEASES
human health and lives, can be assigned no
In 1967 the Surgeon General of the United
price tag.
States, William H. Stewart, declared that it
Although the specific theme of this
was time to "close the book on infectious
chapter is the application of basic research
diseases" and start paying more attention to
on rodent biology and ecology toward
chronic ailments. In 1998, Surgeon General
public health goals, the goals of pest
David Satcher, speaking to the United States
management personnel and public health
Congress, addressed the "continuing threat
practitioners are similar. Whether to prevent
of emerging infectious diseases", singling
economic loss or to prevent disease, we seek
out infectious disease as the number one
to control rodent popuiations, prevent their
killer worldwide. Even in the United States,
access to human food and other products,
according to Dr. Satcher, the death rate from
prevent their access to our dwellings, and
infectious diseases, excluding acquired
minimise their contact with humans.
immune deficiency syndrome (AIDS), rose
Accomplishment of these objectives is
by 22% between 1980 and 1992. What factors
facilitated by a thorough understanding of
are responsible for this stark difference in
the biology and ecology of the target species.
perspective coming from two surgeons
In this chapter, I will (1) introduce the extent
general? What are emerging infectious
and severity of the human disease problem
diseases? Why are they suddenly becoming
caused by the rodent-borne haemorrhagic
important?
fever viruses; (2) use examples from the
The term' emerging infectious disease'
hantaviruses and arena viruses to illustrate
applies to two groups of illnesses: those
the broad nature and potential for expansion
caused by previously unknown agents that
of the rodent-borne disease problem; (3)
are being recognised at an increasing rate,
illustrate how a structured basic research
such as AIDS, Lyme disease and hantavirus
program can provide data that may assist in
pulmonary syndrome, and those that
the management of rodent host populations

135
Ecologically-based Rodent Management
represent the re-emergence of previously
described diseases in drug-resistant or more
virulent forms, such as tuberculosis, malaria,
and the illnesses due to Escherichia coli
0157:H7 (Institute of Medicine, 1992). The
recent awareness of emerging infectious
diseases is due to several factors. The
development of antibiotic resistance is one.
Others include rapid transportation, that
quickly brings victims of remotely acquired
diseases into heavily populated cities,
improved diagnosis, and physician
education. Another important factor is our
rapidly expanding population with the
resulting incursion of humans into remote,
natural habitats where previously unknown
diseases have existed for many years in
cycles involving wild-animal hosts. Some of
these diseases of wild animals, the zoonotic
diseases, can be transmitted to humans.
Rodents, because of their tremendous
diversity, social behaviour, opportunistic life
history, high reproductive potential,
periodically high population densities, and
peridomestic affinities, are among the most
important natural reservoirs for zoonotic
diseases. A review published in 1995
described approximately 60 zoonotic
diseases, or groups of associated diseases,
for which rodents serve as hosts for the
etiologic agent (Hugh-Jones et aL 1995). The
rodent-borne haemorrhagic fevers represent
one such group.
Because of our phylogenetic relatedness
to them, other mammals are the most likely
animals with which humans may be
expected to share pathogens. Rodents are the
most diverse group of mammals, comprising
nearly half of the 4,600 species in the class. It
is likely that only a small proportion of the
organisms infecting rodents would cause
136
disease in humans. Nevertheless, the
potential number of agents is vast. The
disease agents that infect the house mouse
(Mus musculus) have been relatively well
studied because of its use as a laboratory
animaL Although it does not claim to be
comprehensive, a recent report (Committee
on Infectious Diseases of Mice and Rats,
1991) provides a discussion of about 40
bacteria, viruses, and parasites that cause
disease in M. musculus. These agents are
described mostly from populations of mice
that have not been outside of the relatively
sterile environment of the laboratory for
many generations. Thus, the potential for
pathogen diversity may be much greater in
populations of sylvatic species. It is certainly
not unreasonable to suspect that every
species of wild rodent might harbour as
many, if not more, potential disease agents.
In the natural host, coevolution likely has
lead to a relatively benign relationship
between host and pathogen. The death or
severe illness of the host is rarely to the
evolutionary benefit of a parasite. When
other animals (including humans) come into
contact with rodents, the possibility of a
cross-species infection occurs. The response
of the human immune system to these novel
agents is unpredictable. In many cases the
pathogen will be cleared rapidly by the
immune response and no disease occurs.
Some pathogens may be pre-adapted to hide
from the human immune system (e.g.
human immunodeficiency virus). Others
might elicit a strong immune response, but
that response might, itself, result in severe
pathology to the host. Such is the case with
the rodent-borne haemorrhagic fever
viruses.
 The Role of Rodents in Emerging Human Disease

THE RODENT-BoRNE cause the South American hemorrhagic

HAEMORRHAGIC FEVERS fevers in the New World, and Lassa fever in
the Old World. These diseases are the cause of
The magnitude of the potential for human
significant morbidity and mortality. There
disease involving rodent-borne is
may be 200,000 cases of HFRS each year in
largely unknown. However, one group for
Asia, primarily in China and Korea (McKee et
which considerable progress has been made
al. 1991). Lassa virus is responsible for as
in recent years is the rodent-borne
many as 300,000 human infections each year
haemorrhagic fever viruses. An awareness
in West Africa (McCormick et aL 1987).
of the distribution of these viruses and their
Secondly, each virus in both families is
disease potential is becoming increasingly
usually associated with a specific rodent host,
important to rodent biologists and
of the family Muridae, in which it establishes
management personnel. In this chapter I will
a chronic, p~rsistent infection that involves
provide brief descriptions of the most
the sporadic or persistent shedding of large
important of the known the
quantities of infectious virus into the
recognised diseases caused by each, and
environment in urine, faeces or saliva. These
their general distributions. Additional
characteristics of the infection are key to the
details concerning their importance for
transmission of the virus, both from rodent to
wildlife biologists (Childs et al. 1995),
rodent, and from rodent to human. Humans
identification and distribution of the
are believed to be infected most frequently
reservoir species (Mills and Childs 1999),
via the inhalation of infectiOll..<; aerosols of
and epidemiology of the diseases (Peters et
rodent excreta or secreta. Other likely but less
al. 1996; Peters et al. 1999; Enria et al. 1999)
frequent routes of infection include direct
can be found in other sources.
contact of broken skin or mucous membranes
The rodent-borne viral haemorrhagic
with contaminated rodent fluids or fomites,
fevers are caused by two groups of viruses,
ingestion of contaminated food, or the bite of
the hantaviruses and the arena viruses.
an infected rodent. Transmission within host
Although they are both negative-stranded,
populations may be by a variety of horizontal
enveloped ribonucleic add (RNA) viruses,
and vertical mechanisms, but evidence from
the two groups are not closely related
field studies indicates that horizontal
taxonomically. The hantaviruses constitute
transmission, and perhaps, specifically
the genus Hantavirus, within the family
aggressive encounters between adult male
Bunyaviridae; the arenaviruses constitute the
rodents, may be an important mechanism
family Arenaviridae. Nevertheless, these two
(Glass et al. 1988; Mills et al. 1992, 1997b).
groups of viruses share several important
characteristics. Both cause severe
HAEMORRHAGIC FEVER WITH
haemorrhagic fever in humans. The
RENAL SYNDROME
hantaviruses cause hantavirus pulmonary
syndrome (HPS) in the New World and HFRS is the term applied collectively to a
haemorrhagic fever with renal syndrome suite of diseases of varying severity caused
(HFRS) in the Old World. The arenaviruses by hantaviruses in Asia and Europe. The

137
Ecologically-based Rodent Management
diseases are characterised by fever, chills,
myalgia, and varying degrees of
haemorrhage and renal compromise with
1% to 15% mortality. The viruses are carried
by rodent hosts of the murid subfamilies
Murinae (Old World rats and mice) and
Arvicolinae (voles), and the distributions of
the diseases generally coincide with the
distributions of the host species (Mills and
Childs 1999). A summary of the viruses, the
known diseases, and approximate
distribution of the reservoirs is provided
(Appendix 1).
The prototype hantavirus, Hantaan
virus, gained worldwide attention during
the Korean conflict, when over 3,000 United
Nations troops contracted a severe form of
HFRS, then referred to as Korean
haemorrhagic fever (KHF). However, the
disease, which has variously been referred
to as epidemic hemorrhagic fever,
hemorrhagic nephrosonep hritis, Churilov's
disease, and Songo fever, has been
recognised for many years in Asia (McKee
et a1. 1991; Peters et a1. 1999). A Chinese
medical text from 960 AD may describe
compatible symptoms. The disease was
noted by Soviet scientists as early as 1913,
and outbreaks continued to be described by
the Soviets as well as among Japanese
troops in Manchuria during the 1930s. The
etiologic agent of KHF was not described
until the late 1970s when Ho Wang Lee
isolated a virus from the lungs of the striped
field mouse (Apodemus agrarius) captured on
the banks of the Hantaan River near the
border between North and South Korea (Lee
et a1. 1978). Currently, Hantaan virus is
responsible for perhaps 200,000 cases of
HFRS each year, in China, Korea, and the
Russian Far East (McKee et a1. 1991).
138
Epidemics are seasonal with an autumn
peak, coinciding with the maximum
agricultural activity and probably maximum
host population density. Cases occur
predominantly among adult men in rural
habitats; many cases are among farmers,
forest workers and soldiers in the field
(McKee et a1. 1991; Peters et a1. 1999).
Seoul virus, which is found nearly
worldwide in association with its
cosmopolitan host, the Norway rat (Rattus
norvegicus), is responsible for a relatively
mild form of HFRS (Lee et a1. 1980).
Although Seoul virus has been detected in
rats throughout most of the range of the
species, most confirmed cases of HFRS
caused by Seoul virus have been restricted to
Korea, Russia, and China. Reasons for the
apparent lack of disease in other parts of the
world are unknown, but may include
inadeguate case finding. A search in one
United States city revealed three suspected
cases (Glass et a1. 1994).
Dobrava virus, hosted by the yellow-
necked field mouse (Apodemus jlavicollis), is
responsible for a severe form of HFRS in the
Balkans. Dobrava virus may be associated
with A. agrarius in the Baltic region (Plyusnin
et a1. 1997; Peters et a1. 1999).
Although several hantaviruses are hosted
by arvicoline rodents in Asia and Europe,
only one is known to be associated with
human disease. Puumala virus, carried by
the bank vole (Clethrionomys glareolus), is the
etiologic agent for a mild form of HFRS
called nephropathia epidemica (NE). NE is
endemic to Scandinavia, western Europe,
and European Russia. Several additional
hantaviruses, some only recently
discovered, are associated with murine and
arvicoline rodents in Asia and Europe
 The Role of Rodents in Emerging Human Disease
(Appendix 1). These viruses have not been
definitively associated with human disease,
but extensive studies are lacking.
HANTAVIRUS PULMONARY SYNDROME
HPS is a New World hantavirus disease
characterised by a flu-like prodrome involving
fever, myalgia, and malaise, which rapidly
progresses to cardiopulmonary compromise
that may end in death in about 50% of cases in
spite of aggressive hospital care.
In early 1993, only a single
autochthonous hantavirus was known from
the New World: Prospect Hill virus (Lee et
a1. 1982), which is associated with the
meadow vole, Microtus pennsylvanicus. It still
has not been associated with any human
disease. The seemingly sudden appearance
of HPS in the spring of 1993 led to the
isolation of Sin Nombre virus (SNV; Elliott et
a1. 1994) and its association with the deer
mouse (Peromyscus maniculatus; Nichol et al.
1993; Childs et al. 1994). Armed with a
specific case definition and the molecular
and serologic tools and reagents necessary to
detect SNV, physicians and mammalogists
quickly discovered that HPS was a pan-
American disease, and that numerous
species of New World sigmodontine and
arvicoline rodents serve as hosts for a
plethora of hantaviruses. Currently,
approximately 20 viruses have been
described in association with about as many
host species occurring from Canada to
Patagonia (Appendix 1 and Figure 1). About
half of these viruses are known human
pathogens. All of those viruses responsible
for HPS are hosted by rodents of the murid
subfamily Sigmodontinae. In addition to the
United States and Canada, HPS has now
been documented in Argentina, Chile,
Paraguay, Uruguay, Brazil, and Bolivia, and
hantavirus or hantavirus antibody has been
demonstrated in rodents from Peru,
Venezuela, Costa Rica and Mexico.
THE SOUTH AMERICAN HAEMORRHAGIC
FEVERS
The South American haemorrhagic fevers
currently consist of four recognisLu diseases
that are clinically similar. 1hey are
characterised by an insidious prodrome of
fever, malaise, muscle aches, and retroorbital
headache, which may be followed by
hypotension, conjunctival injection, petechiae
on the throat, cllest, and axillary area,
dizziness, and tremors. Severe cases
demonstrate bleeding from gums and mucous
membranes, shock, coma, and convulsions
(Enria et a1. 1999). Mortality may be 10% to
33%. With a single possible exception, the New
World arenaviruses for which the reservoir is
known are all associated with sigmodontine
rodents (Appendix 2).
The first arenaviral haemorrhagic fever to
be recognised and, to date, the best studied
is Argentine haemorrhagic fever (AHF). The
disease was described in 1953 (Arribalzaga
1955) and the etiologic agent, Junin virus,
was described in 1958 (Parodi et al. 1958).
Several hundred to over a thousand cases of
AHF were confirmed each year on the
central Argentine Pampa from the discovery
of the disease until the recent development
of a vaccine.
Cases are predominantly in adult men
from rural areas, and epidemics are
seasonal, occurring in the fall-coinciding
with the harvest of principal crops (corn and
soybeans) and maximum densities of the
139
Ecologically-based Rodent Management

Laguna Negra

Figure 1.
Geographic locations of the principal currently recognised New World hantavlruses (after Mills and Chllds
1998).

140
 The Role of Rodents in Emerging Human Disease

principal reservoir, the corn mouse (Calomys between November and January (Manzione
musculinus). The introduction of an effective et a1. 1998).
treatment using immune plasma decreased Finally, an arenaviral haemorrhagic fever
the mortality from 15-30% to less than 1%, caused by Sabia virus is known from a single
and the recent use of a highly efficacious naturally acquired case near Sao PauIo,
vaccine in the AHF-endemic area has Brazil, in 1990 (Coimbra et a1. 1994). Nothing
resulted in a substantial reduction in the is known about the reservoir, or the potential
numbers of reported cases (Maiztegui et a1. endemic area.
1998).
Bolivian hemorrhagic fever (BHF), OLD WORLD ARENAVIRAL
caused by Machupo virus, was described HAEMORRHAGIC FEVERS
following several clusters of cases in 1959.
Lassa fever, which is endemic to West
The 2,000-3,000 cases of naturally acquired
Africa, is the only recognised arenaviraI
BHF have all been from the Beni Department
haemorrhagic fever in the Old World.
of north-western Bolivia. Sporadic cases
Although the magnitude and geographic
predominantly involve adult males from
extent of the cases are poorly known, Lassa
rural environments (Kilgore et a1. 1995), but
virus probably causes 100,000 to 300,000
several large outbreaks have been associated
cases and 5,000 deaths annually (McCormick
with high densities of the reservoir, Calomys
et a1. 1987). The virus has been isolated from
callosus, in and around villages. Unlike its
humans or rodents in Nigeria, Sierra Leone,
congener, C. musculinus (which is strictly
Guinea, and Liberia, but serologic surveys
associated with grassland and agricultural
show that Lassa or Lassa-like viruses are
habitats), C. callosus can be found in close
present in at least 10 other African countries
association with human dwellings. An
(Peters et a1. 1996; Appendix 1). Two or more
outbreak in the town of San Joaqufn in 1963-
species of the Mastomys natalensis species
1964 ended abruptly after two weeks of
complex appear to serve as the reservoir for
continuous trapping in homes, during which
Lassa virus. At least eight species of
3,000 C. callosus were captured (Kuns 1965),
Mastomys occur in Africa south of the
and an ongoing program of rodent trapping
Sahara, and their distribution and
in villages in the BHF-endemic area may be,
relationships are poorly understood
at least in part, responsible for the scarcity of
(Robbins and Van Der Straeten 1989). A 32-
cases since 1974 (PAHO 1982).
chromosome species, Mastomys huberti, has
Venezuelan haemorrhagic fever,
been described as being found in dwellings,
described in 1989 (Salas et a1. 1991), is caused
while a 38-chromosome species, Mastomys
by Guanarito virus, an arenavirus hosted by
erythroleucus, was found in the surrounding
the cane mouse, Zygodontomys brevicauda.
bush areas; both species were frequently
Recognised cases have been restricted to
infected with Lassa virus with a prevalence
rural areas of southern Portuguesa and
of about 30% (McCormick et a1. 1987).
northern Barinas states. The highest risk of
Lymphocytic choriomeningitis (LCM),
disease is among adult male farm workers,
caused by the arenavirus lymphocytic
and the greatest numbers of cases occur

141
Ecologically-based Rodent Management
choriomeningitis virus (LCMV), is
associated with the house mouse
(M. musculus) throughout much of its
worldwide range. LCMV usually produces a
syndrome of fever and myalgia (sometimes
complicated by meningitis), which is rarely
serious, but infections during pregnancy
have been associated with serious, even fatal
complications to neonates (Peters et a1. 1996).
LCM is not considered a viral haemorrhagic
fever and will not be discussed further.
Nevertheless the disease may be much more
common than is diagnosed, and biologists
and pest control practitioners should be
aware of the risk. Detailed reviews have
been published (Jahrling and Peters 1992;
Peters et al. 1996; Enria et a1. 1999).
PoTENTIAL VIRUS DIVERSITY
The rate of discovery of new haemorrhagic
fever viruses has increased almost
exponentially in recent years. In the
Americas, for instance, the number of
known autochthonous hantaviruses has
increased from one, in 1993, to over 20 in
1998 (Figure 1). The rate of discovery of new
hantaviruses and arenaviruses is not
slowing, and theoretical considerations
suggest that we may recognise only a small
proportion of the potential diversity. In
general, each hantavirus and arenavirus
appears to be associated with a single
species of murid rodent host. The
hantaviruses are associated with the
Murinae, the Sigmodontinae, and the
Arvicolinae; the arena viruses with the
Murinae and the Sigmodontinae. Further-
more, the phylogenetic relationships among
the viruses (with some exceptions for the
Arenaviridae) are generally mirrored by the
142
phylogenetic relationships among the
rodent hosts (Bowen et a1. 1997; Mills et al.
1997a; Schmaljohn and Hjelle 1997). This
pattern suggests that there was an ancestral
hantavirus and arena virus associated with
an ancestral mu rid rodent, before the
subfamiliallineages diverged, over 20
million years ago, and that the viruses have
been co-speciating and co-evolving along
with their rodent hosts since that time.
Implicitly, the maximum potential number
of hantaviruses and arenaviruses would be
one for each of the 143 of
Arvicolinae, 529 species of Murinae, and 423
species of Sigmodontinae (numbers of
species from Musser and Carleton 1993).
Indeed, some species (e.g. Sigmodon alstoni,
Sigmodon hispidus, Bolomys obscurus;
Appendix 1) are known to host an
arenavirus and a hantavirus. Nevertheless, it
is unlikely that this maximum number of
viral species will be found. Virus extinctions
are very likely to have occurred in some
murid lineages. For instance, some well-
studied species (e.g. M. musculus, and
C. musculinus) appear not to be associated
with a hantavirus. It is also likely, however,
that some trans-species 'host jumping' may
have occurred over time (Bowen et al. 1997;
Morzunov et al. 1998), thus further
increasing the potential diversity of viruses.
ECOLOGICAL STUDIES OF
RESERVOIR SPECIES
Just as control of pest populations for
economic reasons depends upon an
understanding of the biology and ecology of
pest species, the control or prevention of
rodent-borne disease largely depends upon
understanding the biology and ecology of
 The Role of Rodents in Emerging Human Disease
the host. Several basic research studies of the
ecology of virus reservoir species during the
past 12 years have provided information
that potentially can be very useful for risk
assessment and directed intervention in
disease control.
In an earlier paper (Mills and Childs
1998) we reviewed some of these studies and
outlined a series of directed goals toward the
understanding of reservoir ecology as it
relates to human disease. After initial
identification of the reservoir host, these
goals include (a) determining the potential
disease-endemic area by identifying the
geographic distribution of the host, and the
range of infection by the pathogen within the
host distribution; (b) more precisely defining
relative human risk by determining the
distribution of the host and pathogen among
the distinct habitats on a regional scale; (c)
investigating potential mechanisms of
transmission of the pathogen within host
populations; (d) conducting long-term
prospective studies to elucidate the temporal
patterns of infection in host populations; and
(e) integrating data from reservoir studies
toward the development of a predictive
model that would allow the early
identification of specific times, places, and
conditions that may lead to increased rodent
populations, or increased infection in rodent
populations that can cause elevated risk of
human disease. Although specifically
directed at understanding rodent ecology in
relation to human disease, many of these
goals (especially a, b, d, and e) are applicable
to studies of economic pests. In this section,
the above-listed goals are used as a
structural basis, while providing examples
from studies and theoretical problems
specific to the hantaviruses and
arenaviruses, to illustrate the value of basic
research toward the practical goal of
preventing and controlling human disease
caused by rodent-borne pathogens.
Defining disease-endemlc areas
One of the most basic pieces of information
for designing and directing a prevention
program for any disease is a precise
knowledge of the geographic area where a
disease may occur (the potential endemic
area). Prevention efforts, such as public
education and reservoir control, must be
directed throughout this area, while efforts
outside the area represent wasted time and
money. For any rodent-borne disease, the
geographic distribution of the reservoir
defines the maximum potential endemic
area of the disease. For many rodent species
in North America, the distributional ranges
are precisely known and are available in the
literature (Hall and Kelson 1959). Following
the identification of the deer mouse as the
reservoir for SNV (Childs et a1. 1994),
scientists consulted the published
distribution of P. maniculatus in North
America (Carleton 1989), and realised that
the potential endemic area for HPS caused
by SNV could encompass most of the North
American continent. Education of
physicians and increased surveillance soon
confirmed that sporadic cases of HPS
occurred throughout the range of the deer
mouse in the United States. For other rodent
species, in less extensively studied parts of
the world, these distributions are poorly
defined. This is the case with several
important hantavirus and arenavirus
reservoir species in South America. For
example, the published distribution for
C. musculinlls, reservoir of Junin virus,
143
Ecologically-based Rodent Management
includes central and northern Argentina
(Red ford and Eisenberg 1992). Nevertheless,
during recent ecological investigations of
Laguna Negra virus on the Chaco of
Paraguay (Yahnke et al. 1998), C. musculinus
was frequently captured. A concerted effort,
as well as multi-disciplinary studies that
include health scientists, ecologists, and
systematists, will be essential to define
accurately the distributional ranges of
important reservoir species.
For some host-virus systems (e.g.
P. maniculatus and SNV), the host appears to
be infected throughout its geographic range.
In other cases, the distribution of the virus
may include only a small portion of the
range of the host. In those cases, the
identification of the geographic area in
which the host and the pathogen both occur
provides a more precise definition of the
potential disease-endemic area. While
C. musculinus occurs throughout central and
northern Argentina (and apparently western
Paraguay), the AHF-endemic area occupies
only a very limited region of the central
Argentine Pampa (Maiztegui et al. 1986).
Limited searches for Junin virus in
C. musculinus populations outside the
endemic area have been unsuccessful (Mills
et al. 1991). Calomys laucha also occurs from
central Argentina through south-eastern
Bolivia, western Paraguay, and west-central
Brazil (Musser and Carleton 1993), yet
Laguna Negra virus apparently occurs only
on the Chaco of Paraguay Gohnson et aL
1997, Mills and Childs 1998). Populations of
C. laucha in Paraguay and central Argentina
appear to be disjunct, and it is possible that
the populations of C. laucha in Argentina are
genetically distinct and will not support
infection with Laguna Negra virus.
144
Populations of C. muscuIinus appear to be
continuous across the boundary of an
expanding AHF-endemic area (Maiztegui et
aL 1986). The spatially restricted but
expanding distribution of Junfn virus within
a continuous host population suggests
recent introduction of Junfn virus or recent
genetic changes in the virus, host, or both
populations. Reasons for the lack of
coincidence in host and virus distributions
are likely to be diverse and involve host
genetics, geographiC boundaries, and local
extinctions in subpopulations. The
elucidation of these factors, which will
require collaboration among ecologists,
virologists, geneticists, and systematists, will
contribute to our understanding of host-
virus coevolution, the relationships among
rodent species, and the properties of host
systems that are required to support long-
term maintenance of viral symbionts. On the
practical side, these studies will allow the
precise definition of the geographic areas in
which humans are at risk for specific
diseases.
Defining habitat associations
In addition to the large geographic patterns
discussed above, host and pathogen
populations may vary on regional or local
scales. Many species of rodents demonstrate
distinct local habitat preferences, which may
have practical implications for disease
transmission as well as reservoir
management. The risk of human disease
may be more precisely defined by describing
differences in host distribution, population
densities, and prevalence of infection among
the distinct habitats represented in a local
area. Even for species that are considered
 The Role of Rodents in Emerging Human Disease

opportunists or generalists, habitat studies working, or pursuing recreational activities

may yield useful inform ation.
The deer mouse is considered a habitat
generalist and has been reported as
occurring in nearly every dry-land habitat in
North America (Burt and G rossenheider
1976). A habitat study conducted in the
south-wes tern United States confirmed that
P. maniculatus occupied all of the major
habitats represented (Mills et al. 1997b).
Nevertheless, the preva lence of infection
with SNV varied significantly among
habitats, being lowest at the altitudinal and
climatic extremes (desert and alpine tu.ndra)
and highest at the middle altitude habitats
such as chaparral, grassland, and pmon-
juniper woodland. The last habitat is where
most H PS cases in th e south-western United
States ha ve occurred. Similar results were
demonstrated by a study in Nevada and
California (Boone et al. 1998). Results such as
these can be applied by p ublic hea lth
scientists to d efine more precisely the
relative risk of disease to humans living,
in various habitats.
The corn mouse, C. musculinus, has
historically been considered a denizen of
corn fields, as its common name implies.
Junin virus has been thought to be
transmitted to farmers working in those crop
field s during the mechanised harvesting
process (Carballal et al. 1988). Recently
however, habitat studies conducted in the
AHF-endemic area demonstrated a distinct
preference by C. musculinus for the relatively
stable, weedy border habitats (fence lines
and roadsides) adjacent to the crop fields
(Figure 2).
These res ul ts suggest a need to reconsider
the possible places and m echanisms of
transmission ofJwlin virus to humans. They
also suggest a specific intervention
mechanism for d ec reasing the inciden ce of
AHF: periodically burning or cutting the
weedy border habitats to eliminate the
preferred habitat for the reservoir host.
Habitat studies conducted for other

100 ,------------------------------------------------,

(f)
80
~
:::J
li
Cll
0 60
a
>-
u
c
Q)
40
:::J
0-
Q)

Lt 20

0
2 3 4 5 6 7 8 9 10 11 12
Row Number
Figure 2.
Cumulative numbers of captures within each of 12 rows of traps of a 12 by 12 t rapping grid located in crop
fields and adjacent roadside habitat in central Argentina, March 1998 to August 1990. Rows 1 and 2 are
roadside habitat; rows 3 through 12 are in crop fields.

145
Ecologically-based Rodent Management
reservoir species might suggest similar
approaches.
Identifying mechanisms of
transmission
Basic field studies of reservoir demography
have provided important clues to the
specific mechanisms of virus transmission
within host populations. Field studies with
SNV (Mills et aL 1997b) and Black Creek
Canal virus (Glass et aL 1998) have shown a
J-shaped curve of antibody prevalence with
host age. This pattern suggests that the
young of infected females are born with
maternal antibody, which is lost within a
few weeks. Infection is then acquired by
some horizontal mechanism later in life.
Field data have also demonstrated a positive
correlation between scars and the prevalence
of infection (Glass et al. 1988). The more
aggressive males may have a much higher
prevalence of infection than females (Mills et
aL 1992; Mills et al. 1997b). Thus, an
important, specific mechanism of virus
transfer within reservoir populations may be
aggressive encounters among adult male
animals. Laboratory studies have indicated
that lymphocytic choriomeningitis and
Lassa viruses are maintained by vertical
transmission mechanisms (Childs and Peters
1993). However preliminary field data have
demonstrated an age-associated acquisition
of antibody in Mastomys populations in
Guinea (A.R. Dembyand 10 others,
unpublished data). This and similar
disparities between laboratory and field
results for Junfn virus (Mills et aL 1992) may
indicate that laboratory results may not
always be applicable to natural field
conditions and that field studies are
146
important for testing predictions based on
laboratory results.
Long-term studies
Perhaps the greatest amount of useful
information about reservoir populations and
host-virus dynamics is achieved through the
use of longitudinal mark-recapture studies
(Mills et al. 1999b). These studies involve the
establishment of multiple pennanent trapping
plots, which are operated at defined intervals
(usually monthly for disease studies) for
several consecutive nights. Captured rodents
are measured, sampled (e.g. blood and oral
swab), identified with a permanent mark or
number, and released at the site of capture. In
subsequent trapping events, animals are
repeatedly captured, measured, and sampled.
In this way, changes in community structure
and population densities - as well as
individual growth rates, movement,
reproductive condition, and infection status
-are measured over time. Simultaneous
monitoring of environmental variables, such
as temperature, rainfall, and growth and
cover by vegetation, provides clues
concerning environmental changes that are
related to changes in reservoir populations
and, subsequently, changes in risk of human
disease.
Mark-recapture studies have helped to
elucidate the temporal population dynamics
of host virus infection for the reservoirs of
Seoul and Prospect Hill viruses (Childs et aL
1987), Puumala virus (Niklasson et aL 1995),
Junin virus (Mills et al. 1992), and SNV
(Douglass et al. 1996; Mills et aL 1999a). The
cited studies and others in progress are
helping to elucidate the associations among
environmental conditions, rodent
population densities, prevalence of infection
 The Role of Rodents in Emerging Human Disease

in reservoir populations, and human disease
risk.
The incidence of several rodent-borne
diseases is related to changes in density of
reservoir populations. Large year-to-year
fluctuations in population density are
characteristic of rodent populations of many
species. Northern Hemisphere arvicolines,
such as the reservoir for Puumala virus
(c. glareolus), undergo regular population
cycles with a periodicity of 3-4 years,
although the causes for the cycles are still
unclear (Krebs and Myers 1974; Niklasson et
al. 1995). The year-to-year incidence of HFRS
caused by Puumala virus was shown to be
correlated with the density of C. glareolus in
Russia and Scandinavia (Niklasson et al.
1995).
Regular population cycles are not known
in rodents from the Northern Hemisphere
tropics or anywhere in the Southern
Hemisphere. However, periodic, dramatic
increases in the density of some rodent
populations do occur. These population
irruptions are generally associated with
unusual climatic conditions, which result in
abundant food supplies and ideal or
prolonged conditions for reproduction. A
three-year longitudinal study of
C. musculinus in Argentina demonstrated a
clear positive association between reservoir
population density and the magnitude of
AHF epidemics (Figure 3).
The associated environmental conditions
were a relatively benign winter, followed by
a wet summer, which apparently resulted in

8~---------------------------------------'200

Corn mouse density

... AHF cases

150
u..
:::c
<C
15
(/)
(l)
(/)
<tI
100 U
U
(l)

'E0
D
50

1988-1990
Figure 3.
Mean numbers of captures of Calomys musculinus per 100 trap nights (trap success) and numbers of cases
of Argentine haemorrhagic fever (AHF) in central Argentina, March 1998 to August 1990. Reprinted with
permission from Mills et a!. (1992).

147
Ecologically-based Rodent Management
unusually lush vegetation and abundant
food supplies (Mills et al. 1992).
The outbreak of HPS in the south-western
United States in 1993 was preceded by an El
Nifio Southern Oscillation (ENSO) event,
which resulted in unusually warm winters
and high rainfall in affected areas. It has been
hypothesised that the HPS outbreak was a
direct result of increases in rodent
populations and increases in prevalence of
SNV infection among high-density reservoir
populations (Parmenter et al. 1993). Although
intuitively attractive, no longitudinal
monitoring of rodent populations and
infection status was in place in the area of the
outbreak, so this hypothesis cannot be
confirmed. Subsequent to the outbreak,
however, the Centers for Disease Control and
Prevention, in collaboration with several local
universities, initiated a series of longitudinal
mark-recapture studies in the south-western
United States (Mills et al. 1999b). These
studies were in place to document
environmental changes and associated
increases in reservoir populations in some
areas of the south-west in response to an
ENSO event in 1997/1998 (T.L. Yates, K.D.
Abbott, CH. Calisher and M.L. Morrison,
unpublished data). These increases in
reservoir populations have been associated
with increased numbers of HPS cases in the
south-western United States. As of August
1998, there have been about 14 cases in the
four-state area of Arizona, Colorado, New
Mexico, and Utah, in comparison to 2, 2, and
4, for the same time periods in 1995, 1996, and
1997, respectively (Centers for Disease
Control and Prevention, unpublished data).
A recent outbreak of HPS in southern
Chile was apparently preceded by a
dramatic increase in local populations of the
148
reservoir of Andes virus (Oligoryzomys
longicaudatus). Causes for the rodent
irruption are unclear, but may be related to
an unusually benign winter or to the
flowering of a local species of bamboo, an
event that may occur only every 40 years and
that provides abundant food for the
granivorous 0. longicaudatus (Munia et al.
1996; Toro et al. 1998). Longitudinal studies
are currently being planned and initiated in
Chile and Argentina to follow the
environmental variables associated with
changes in population density and infection
status in O. longicaudatus populations and
the relation of these variables to human
disease.
An important key to being able to predict
the relative risk of diseases to humans is
understanding the conditions that lead to
increased virus transmission and increased
prevalence of infection in host populations.
Infection appears to be associated with
behavioural events involving the
interactions of individual rodents. Given the
pattern of horizontal transmission
demonstrated for many hantaviruses and
arenaviruses, it might be predicted that
increasing population densities should
result in increased rodent-to-rodent contact
and a higher prevalence of infection in host
populations. In fact, however, investigators
are frequently unable to show a correlation
between rodent population density and
prevalence of infection in rodent
populations (Mills et al. 1992; Douglass et al.
1996; Bond et al. 1998; Boone et al. 1998). The
problem with these approaches may be in
seeking an instantaneous, linear relationship
between density and antibody prevalence.
In strongly seasonal environments, the
effects of seasonal reproduction and
 The Role of Rodents in Emerging Human Disease
horizontal transmission of virus may result
in an alternation of peaks in population
density and prevalence of infection. In
Sweden, the population density of
C. glareolus was highest in autumn
(Niklasson et al. 1995), while the prevalence
of antibody to Puumala virus in these
populations was highest in the spring, and
correlated with vole population density the
previous fall. A similar pattern of alternating
peaks in density and antibody prevalence
was observed for populations of Akodon
azarae infected with Pergamino virus in
Argentina-rodent density was highest in
the fall, antibody prevalence was highest in
the spring (Schmidt et al. 1998). This pattern
has also been observed for P. maniculatus
populations infected with SNV in Colorado
(Calisher et al. 1999), and Peromyscus boylii
infected with a Sin Nombre-like virus in
Arizona (Abbott et aL 1999). This delayed-
density-dependent prevalence of infection
may be typical for viruses transmitted by
horizontal mechanisms in seasonal
environments. Autumn populations display
peak densities because of the culmination of
the spring/ summer reproductive effort; yet
the population consists primarily of young
of the year that have not yet been infected, or
are only recently infected and do not yet
have detectable antibody. The cessation of
reproduction and over-winter mortality
results in a popUlation nadir in the spring,
but at that time the population consists
exclusively of older adults, which are more
likely to be infected. In an autumn during
which particularly high population levels
occur, crowding would presumably lead to
more intraspecific contacts, more virus
transmission events, and a proportionally
higher antibody prevalence the following
spring (Niklasson et al. 1995; Mills et al.
1999a). Research leading to an under-
standing of the conditions that lead to
increased virus transmission and prevalence
of infection in host populations will improve
the ability of public health scientists and
modellers to predict increases in the risk of
human disease.
Predictive models of disease risk
Perhaps the most important practical
application of studies of reservoir
populations is to integrate the data from
these studies into a predictive model that
would allow public health practitioners to
identify specific times and places where
conditions may pose a threat to the public
health. Such a model (Figure 4) assumes that
the risk of human disease is related to rodent
population density and prevalence of
infection; rodent populations are affected by
the quality of the biotic environment (e.g.
habitat quality and food supply); and the
abiotic environment (e.g. edaphic factors
and weather) influences rodent populations
both directly (e.g. direct effects of cold
temperatures on survival) and indirectly
(through their effect on habitat quality and
food supply; Mills and Childs 1998). It is not
possible to have scientists continuously
measuring rodent populations and
environmental variables wherever rodent-
borne diseases occur. However, this may not
be necessary.
Recent studies using satellite imaging
and geographic information systems have
demonstrated that remotely monitored
vegetation indices can help predict the
changing risk of human disease in sites as far
away as East Africa (Linthicum et al. 1987),
and the south-western United States (Boone
149
Ecologically-based Rodent Management
et a1. 1998; Cheek et a1. 1998). The success of
these mathematical models will depend
upon the accuracy of the parameter
estimates, and the accuracy of the estimates,
in turn, will depend upon data collected by
investigators conducting basic field and
laboratory research into the ecology and
biology of the rodent reservoirs.
e.m,,, s,"". .. c,;~" ~
~ ' ' 't' " ---... t
Insects
.J
Rodent Populations
t Human Disease
Figure 4.
Simplified schematic model of relationships
among ecosystem components within an endemic
area for a rodent-borne human disease. Remote
sensors (satellites) may be used for detecting
changes in the ecosystem components which
may lead to increased risk of disease. From Mills
and Childs (1998).
THE RISK TO RODENT BIOLOGISTS
In the United States, the sudden realisation
that wild rodents are the reservoir for
potentially lethal disease has resulted in
significant changes in the way many rodent
biologists conduct their research and
teaching. Mammalogy classes avoid the
handling of sigmodontine rodents by
students, the establishment of laboratory
colonies from wild captured individuals of
known reservoir species is strictly
controlled, and researchers who handle
reservoir rodents are prudent to follow
safety guidelines.
150
The chance of contracting HPS by
handling New World sigmodontine rodents
appears to be low. Nevertheless, the disease
is sufficiently severe to warrant strict safety
measures for the general public (CDC 1993),
and there is evidence that wildlife biologists
are at increased risk. Among the first 100
cases of HPS in the United States, three were
in wildlife biologists, and although a recent
serosurvey of over 1,000 American
mammalogists demonstrated that the risk of
infection with SNV appeared to be low (less
than 1%), risk increased with the munber of
Peromyscus that investigators had handled
during their careers (Armstrong et al. 1994).
A study of Fimush mammalogists showed a
more striking relationship. Although no
mammalogist with less than five years of
experience had antibody to Puumala virus,
40% of those who had trapped voles for
more than 10 years had antibody (Bnunrner-
Korvenkontio et al. 1982). These results
suggest that Puumala virus is more easily
transmitted to humans than is SNV
(although because of the high mortality of
SNV, about half of those infected would not
be available for sampling). Fortunately,
neplu'opathia epidemica is a relatively mild
disease. The murine- and sigmodontine-
associated HFRS, HPS, Lassa fever, and
South American haemorrhagic fevers can be
much more severe and can lead to fatalities
in 15-50% of cases. Relatively simple safety
precautions will minimise the risk of
infection to biologists and are highly
recommended for all researchers handling
all known viral haemorrhagic fever reservoir
species.
Standard precautions have been
promulgated for investigators conducting
field studies, which may involve handling
 The Role of Rodents in Emerging Human Disease
reservoir species for haemorrhagic fever
viruses. These guidelines, which were
developed during field studies of Junin virus
in Argentina and the sigmodontine
hantaviruses in the Americas, have been
published in English (Mills et aL 1995a,b)
and in Spanish (Mills et al. 1998). Briefly,
investigators should wear rubber gloves
when handling traps containing captured
animals, and the traps should be handled in
a manner that will prevent or minimise
contact with rodent excretions or secretions
and inhalation of potentially infectious
aerosols of these materials. If captured
rodents are transported, the traps containing
them should be placed in airtight plastic
bags. These bags subsequently should be
opened and the rodents handled only in an
isolated outdoor area by personnel wearing
protective equipment (latex gloves, gowns
or overalls, respirators fitted with high-
efficiency particulate air filters, and
goggles). Handling rodents outdoors is
preferred in order to take advantage of the
disinfectant properties of natural ultraviolet
light and the rapid dilution of aerosols in
open circulating air. Rodents should be
anesthetised before handling to prevent
bites and production of aerosols, and the use
of sharp instruments such as needles and
scalpels should be avoided when possible.
Instruments, working surfaces, and traps
should be decontaminated using an
appropriate disinfectant (e.g. 5% hospital
strength Lysol, or lOO/" household bleach in
water), and contaminated gloves, disposable
gowns, and wastes should be autoclaved or
burned. Rodent carcasses kept for museum
specimens can be decontaminated by fixing
in 10% formalin for at least 48 hours.
Pest control workers should be alert to
the possibility of inhalation of infectious
aerosols when working in closed struchlres,
which may be infested by hantavirus or
arenavirus reservoir species. The doors and
windows of such structures should be
opened, and the building allowed to air out
for at least 30 minutes before beginning
work. Clean-up of these structures should be
conducted so as to avoid the creation of
aerosols. Nesting materials or contaminated
areas should be wetted down with
disinfectant, and floors should be mopped,
not swept (CDC 1993).
Hantavirus infection in laboratory rodent
colonies has resulted in extensive outbreaks
of human disease (Kulagin et al. 1962).
Precautions when initiating laboratory
colonies from wild rodents that are known
reservoir species for hantaviruses or
arenaviruses should include quarantine as
defined cohorts and serologic screening
upon capture, and again after 30 days (Mills
et al. 1995b).
The rodent-borne haemorrhagic fever
viruses are only one example of many
zoonotic agents that are likely to be hosted
by rodents. In this chapter they have been
used as an example to illustrate the potential
diversity of rodent-borne disease agents.
Although the diseases they cause are
dramatic, the risk to researchers can be
minimised by the adherence to relatively
simple safety guidelines. Researchers
studying these agents have amassed a large
amount of new data during the last 5-6
years. Continued basic research into the
ecology of these host-virus systems
promises to provide useful models for
understanding host-pathogen coevolution,
transmission processes in natural
151
Ecologically-based Rodent Management
populations, and the relationship of
environmental factors to host populations,
pathogen transmission patterns, and human
disease.
ACKNOWLEDGMENTS
Thanks to Barbara Ellis for graphics, and to
Barbara Ellis, Jamie Childs, Tom Ksiazek,
and c.J. Peters for helpful comments on the
manuscript.
REFERENCES
Abbott, K., Ksiazek, T.G. and Mills, J.N. 1999.
Long-term hantavirus persistence in rodent
populations in central Arizona. Emerging
Infectious Diseases, 5, 102-112.
Armstrong, L.R., Khabbaz, R.P. and Childs, J.E.
1994. Occupational exposure to hantavirus in
mammalogists and rodent workers. Ameri-
can Journal of Tropical Medicine and
Hygiene, 51, 94 (Abstract).
Arribalzaga, R.A. 1955. Una nueva enfermedad
epidemica a germen desconocido: hiperter-
mia nefrotoxica, leucopenic a y enantematica.
Dia Medico, 27, 1204-1210.
Bond,CW.,Irvine, B., Alterson, H.M., Van Horn,
R. and Douglass, R.J. 1998. Longitudinal
incidence of hantavirus infection in deer
mice. Fourth International Conference on
HFRS and Hantaviruses, March 5-71998,
Atlanta, Georgia (Abstract).
Boone, J.D., Otteson, E.W., McGwire, K.C,
Villard, P., Rowe, J.E. and St. Jeor, S.C 1998.
Ecology and demographics of hantavirus
infections in rodent populations in the
Walker River basin of Nevada and California.
American Journal of Tropical Medicine and
Hygiene, 58, 445-451.
Bowen, M.D., Peters, and Nichol, S.T. 1997.
Phylogenetic analysis of the Arenaviridae:
patterns of virus evolution and evidence for
cospeciation between arenaviruses and their
rodent hosts. Molecular Phylogenetics and
Evolution, 8,301-316.
152
Brummer-Korvenkontio, M., Henttonen, H. and
Vaheri, A. 1982. Haemorrhagic fever with
renal syndrome in Finland: ecology and virol-
ogy of nephropathia epidemica. Scandana-
vian Journal of Infectious Diseases, Suppl. 36,
88-91.
Burt, W.H. and Grossenheider, R.P. 1976. A field
guide to the mammals, North America north
of Mexico, third edition. Boston, Houghton
Mifflin Company.
Calisher, Sweeney, W., Mills, ].N., and
Beaty, B.J. 1999. Natural history of Sin
Nombre virus in western Colorado. Emerg-
ing Infectious Diseases, 5, 126-134.
Carballal, Videla, CM. and Merani, M.S.
1988. of Argentine haemor-
rhagic fever. European Journal of Epidemiol-
ogy,4,259-274.
Carleton, M.D. 1989. Systematics and evolution.
In: Kirkland, G.L. and Layne, J.N. ed.,
Advances in the study of PeromysClIs (Roden-
tia). Lubbock, Texas Tech University Press, 7-
14l.
CDC (Centers for Disease Control and Preven-
tion) 1993. Hantavirus infection
southwestern United States: interim recom-
menda tions for risk reduction. Morbidity and
Mortality Weekly Report, 42, 113.
Cheek, J., Bryan, R. and Glass, G. 1998.
Geographic distribution of highrisk, HPS
areas in the U.S. Southwest. The Fourth Inter-
national Conference on HFRS and Hantavi-
ruses, March 5-7,1998. Atlanta, Georgia,
USA (Abstract).
Childs, J.E., Glass, e.W. and Le Duc,
J.W. 1987. Prospective seroepidemiology of
hantaviruses and population dynamics of
small mammal communities of Baltimore,
Maryland. American Journal of Tropical
Medicine and Hygiene, 37, 648-662.
Childs, J.E., Ksiazek, T.G., Spiropoulou, CF.,
Krebs, J.W., Morzunov, 5., Maupin, e.0.,
Rollin, P.E., Sarisky, J., R.E., Frey,
J.K., Peters, CJ. and Nichol, S.T. 1994.
Serologic and genetic identification of
Peromyscus maniClllatus as the primary rodent
reservoir for a new hantavirus in the south-
western United States. Journal of Infectious
Diseases, 169, 1271-1280.
 The Role of Rodents in Emerging Human Disease
Childs, J.E., Mills, J.N. and Glass, G.E. 1995.
Rodentborne hemorrhagic fever viruses: a
special risk for mammalogists? Journal of
Mammology, 76, 664-680.
Childs, J.E. and Peters, CJ.1993. Ecology and
epidemiology of arenaviruses and their hosts.
In: 5alvato, MS., ed., The Arenaviridae. New
York, Plenum Press, 331-384.
Coimbra, T.L.M., Nassar, E.5., Burattini, M.N., de
Souza, L.T.M., Ferreira, LB., Rocco, LM.,
Travassos da Rosa, A.P., Vasconcelos, P.F.C.,
Pinheiro, F.P., Le Duc, J.W., Rico Hesse, R,
Gonziilez, J.P., Jahrling, P.B. and Tesh, RB.
1994. New arena virus isolated in Brazil.
Lancet, 343,391-392.
Committee on Infectious Diseases of Mice and
Rats 1991. Infectious diseases of mice and
rats. Washington, D.C, National Academy
Press, 397p.
Douglass, RJ., Van Horn, R, Coffin, K and
Zanto,S.N. 1996. Hantavirus in Montana deer
mouse populations: preliminary results.
Journal of Wildlife Diseases, 32,527-530.
EIliott, L.H., Ksiazek, T.G., RoIlin, P.E., Spiropou-
Iou, CF., Morzunov, 5., Monroe, M.,
Goldsmith, CS., Humphrey, CD., Zaki, S.R,
Krebs, J.W., Maupin, G., Gage, K, Childs, J.E.,
Nichol, S.T. and Peters, Cr 1994. Isolation of
the causative agent of hantavirus pulmonary
syndrome. American Journal of Tropical
Medicine and Hygiene, 51,102-108.
Enria, DA, Bowen, M.D., Mills,J.N., Shieh, G.J.,
Bausch, D. and Peters, CJ. 1999. Arenavi-
ruses. Chapter 111. In: Guerrant, RL.,
Walker, D.H. and Weller, P.F. ed., Tropical
infectious diseases, principles, pathogens,
and practice. New York, W.B. Saunders,
1191-1212.
Glass, G.E., Childs, J.E., Korch, C.W. and Le Duc,
J.W. 1988. Association of intraspecific
wounding with hantaviral infection in wild
rats (Rattus norvegicus). Epidemiology and
Infection, 101,459-472.
Glass, G.E., Livingstone, W., Mills, IN., Hlady,
W.J., Fine, J.B., Rollin, P.E., Ksiazek, T.G.,
Peters, Cj. and Childs,J.E.1998. Black Creek
Canal virus infection in Sigmodon hispidus in
southern Florida. American Journal of Tropi-
cal Medicine and Hygiene, 59, 699-703.
Glass, G.E., Watson,AJ., Le Duc,J.W. and
Childs, J.E. 1994. Domestic cases of hemor-
rhagic fever with renal syndrome in the
United States. Nephron, 68,48-51.
Hall, E.R and Kelson, KR 1959. The mammals of
North America. New York, Ronald Press.
Hugh-Jones, M.E., Hubbert, W.T. and Hagstad,
H.V. 1995. Zoonoses: recognition, control,
and prevention. Ames, Iowa State University
Press, 369p.
Institute of Medicine 1992. Emerging infections:
microbial threats to health in the United
States. Washington D.C, National Academy
Press, 294p.
Jahrling, P.B. and Peters, CJ. 1992. Lymphocytic
choriomeningitis virus: a neglected pathogen
of man. Archives of Pathology and Labora-
tory Medicine, 116,486-488.
Johnson, AM., Bowen, M.D., Ksiazek, T.G.,
Williams, RJ., Bryan, RT., Mills, J.N., Peters,
CJ., and Nicho!, S.T. 1997. Laguna Negra
virus associated with lIPS in western
Paraguay and Bolivia. Virology,238,115-127.
Kilgore, P.E., Peters, CJ., Mills, J.N., Rollin, P.E.,
Armstrong, L., Khan, A.S. and Ksiazek, T.G.
1995. Prospects for the control of Bolivian
hemorrhagic fever. Emerging Infectious
Diseases, I, 97-99.
Krebs, CJ. and Myers, J.H. 1974. Population
cycles in small mammals. Advances in
Ecological Research, 8, 267-399.
Kulagin, S.M., Fedorova, N.I. and Ketiladze, E.S.
1962. Laboratory outbreak ofhaemorrhagic
fever with renal syndrome. Journal of Micro-
biology, Epidemiology and Immunology
(RUSSian), 33,121-126.
Kuns, M.L. 1965. Epidemiology of Machupo
virus infection. n. Ecological and control
studies ofhemorrhagic fever. American
Journal of Tropical Medicine and Hygiene,
14,813-816.
Lee, H.W., Lee, P.W. and Johnson, K.M. 1978.
Isolation of the etiologic agent of Korean
hemorrhagic fever. Journal of Infectious
Diseases, 137,298-308.
Lee, H.W., Park, D.H, Baek, L.J., Choi, KS.,
Whang, Y.N. and Woo, M.S.1980. Korean
hemorrhagic fever patients in urban areas of
Seoul. Korean Journal of Virology, 10, 16.
153
Ecologically-based Rodent Management
Lee, P.-W., Amyx, H.L, Cajdusek, D.C, Yanagi-
hara, RT., Coldgaber, D. and Cibbs, CJ.1982.
New haemorrhagic fever with renal
syndrome related virus in indigenous wild
rodents in United States. Lancet, 2,1405.
Linthicum, KJ., Bailey, CL., Davies, F.C. and
Tucker, C]. 1987. Detection of Rift Valley
fever viral activity in Kenya by satellite
remote sensing imagery. Science, 235,1656-
1659.
Maiztegui, J.l, Briggiler, A, Enria, D. and Feuil-
lade, M.R. 1986. Progressive extension of the
endemic area and changing incidence of
Argentine hemorrhagic fever. Medical Micro-
biology and Immunology, 175, 149-152.
Maiztegui, J.L, McKee, KT., Jr., Barrera Oro, J.C.,
Harrison, L.H., Cibbs, P.H., Feuillade, M.R,
Enria, D.A., Briggiler, AM., Levis, S.C,
Ambrosio, AM., Halsey, N.A and Peters, CJ.
1998. Protective efficacy of a live attenuated
vaccine against Argentine hemorrhagic fever.
Journal of Infectious Diseases, 177,277-283.
Manzione, N., Salas, RA, Paredes, H., Codoy,
0., Rojas, L., Araoz, F., Fulhorst, CF.,
Ksiazek, T.C., Mills, ].N., Ellis, B.A., Peters,
C}. and Tesh, RB.1998. Venezuelanhemor-
rhagic fever: clinical and epidemiological
studies of 165 cases. Clinical Infectious
Diseases, 26, 308-313.
McCormick,J.B., Webb, P.A, Krebs,T.W.,
Johnson, KM. and Smith, E.S. 1987. A
prospective study of the epidemiology and
ecology of Lassa Fever. Journal of Infectious
Diseases, 155,437-444.
McKee, KT., Le Duc, J.W. and Peters, CJ.1991.
Hantaviruses. In: Belshe, RB. ed., Textbook of
human Virology. St. Louis, MO, Mosby Year
Book,615-632.
Mills, J.N., Bowen, M.D. and Nichol, S.T. 1997a.
African arenaviruses-coevolution between
virus and murid host? Belgian Journal of
Zoology, 127,19-28.
Mills, ].N. and Childs, J.E. 1998. Ecologic studies
of rodent reservoirs: their relevance for
human health. Emerging Infectious Diseases,
4,529-537.
154
Mills,J.N. and Childs,J.E.1999. Rodent-borne
hemorrhagic fever viruses. In: Williams, E.S.
and Barker, 1. ed., Infectious diseases of wild
mammals. Ames, Iowa State University Press
(in press).
Mills, J.N., Childs, J.E., Ksiazek, T.G., Peters, Cr
and Velleca, W.M. 1995a. Methods for
trapping and sampling small mammals for
virologic testing. Atlanta, United States
Department of Health and Human Services.
Mills, J.N., Childs, lE., Ksiazek, T.C., Peters, C].
and Velleca, W.M. 1998. Metodos para
trampeo y muestreo de pequenos mamfferos
para estudios virol6gicos,
OPS/HPS/HCT98.104 edn. Washington,
D.C, Organizaci6n Panamericana de la
Salud.
Mills, J.N., Ellis, B.A, McKee, KT., Calder6n,
C.E., Maiztegui, J.I., Nelson, C.O., Ksiazek,
T.G., Peters, CJ. and Childs,J.E.1992.Alongi-
tudinal study ofJU11in virus activity in the
rodent reservoir of Argentine hemorrhagic
fever. American Joumal ofTropical Medicine
and Hygiene, 47,749-763.
Mills, J.N., Ellis, B.A, McKee, K.T.J., Ksiazek,
T.C., Oro, J.C., Maiztegui, J.L, Calderon, C.E.,
Peters, Cl. and Childs, J.E. 1991. Junin virus
activity in rodents from endemic and nonen-
demic loci in central Argentina. American
Journal of Tropical Medicine and Hygiene,
44,589-597.
Mills, J.N., Ksiazek, T.C., Ellis, B.A., Rollin, P.E.,
Kichol, S.T., Yates, T.L, Cannon, W.L, Levy,
CE., Engelthaler, D.M., Davis, T., Tanda,
D.T., Frampton, W., Nichols, CR., Peters, CJ.
and Childs, rE. 1997b. Patterns of association
with host and habitat: antibody reactive with
Sin Nombre virus in small mammals in the
major biotic communities of the southwest-
ern United States. American Journal of Tropi-
cal Medicine and Hygiene, 56, 273-284.
Mills, J.N., Ksiazek, Peters, C]. and Childs,
J.E. 1999a. Long-term studies of hantavirus
reservoir populations in the southwestern
United States: a synthesis. Emerging Infec-
tious Diseases, 5, 135-142.
 The Role of Rodents in Emerging Human Disease
Mills,J.N., Yates, T.L., Childs, J.E., Parmenter,
RR, Ksiazek, T.G., Rollin, P.E. and Peters,
CJ. 1995b. Guidelines for working with
rodents potentially infected with hantavirus.
Journal of Mammology, 76, 716-722.
Mills, J.N., Yates, T.L., Ksiazek, T.G., Peters, CJ.
and Childs, J.E. 1999b. Long-term studies of
hantavirus reservoir populations in the
southwestern United States: rationale, poten-
tial, and methodology. Emerging Infectious
Lnseases,5,95-101.
Morzunov, S.P., Rowe, J.E., Ksiazek, T.G., Peters,
CL St Jeof, S.C, and Nichol, S.T. 1998.
Genetic analysis of the diversity and origin of
hantaviruses in Peromyscus leucopus mice in
North America. Journal of Virolology, 72, 57-
64.
Murua, R, Gonzales, L.E., Gonzalez, M. and
Jofn', Y.C 1996. Efectos del florecimiento del
arbusto Chusquea quila Kunth (Poaceae) sobre
la demograffa de poblaciones de roedores de
los bosques templados frios del sur Chileno.
Boletin de la Sociedad de Biologia,
Concepci6n, Chile, 67, 37--42.
Musser, G.G. and Carleton, M.D. 1993. Family
Muridae. In: Wilson, D.E. and Reeder, D.M.
ed., Mammal species of the world, a
taxonomic and geographic reference, 2nd ed.
Washington, nc, Smithsonian Institution,
501-755.
Nichol, S.T., Spiropoulou, CF., Morzunov,
Rollin, P.E., Ksiazek, T.G., Feldmann, H.,
Sanchez, A, Childs, lE., Zaki, S., and Peters,
CJ. 1993. Genetic identification of a hantavi-
rus associated with an outbreak of acute
respiratory illness. Science, 262, 914-917.
Niklasson, B., Hornfeldt, B., Lundkvist, A,
Bjorsten, S. and Le Duc, J. 1995. Temporal
dynamics of Puumala virus antibody preva-
lence in voles and of nephropathia epidemica
incidence in humans. American Journal of
Tropical Medicine and Hygiene, 53,134-140.
PAHO (Pan American Health Organization)
1982. Bolivian hemorrhagic fever. Epidemio-
logical Bulletin Pan American Health Organi-
zation, 3, 1516.
Parmenter, RR, Brunt, J.W., Moore, DJ. and
Ernest, S. 1993. The hantavirus epidemic in
the southwest: rodent population dynamics
and the implications for transmission of
hantavirus-assodated adult respiratory
distress syndrome (HARDS) in the four
corners region. No. 41, University of New
Mexico. Sevilleta LTER Publication.
Parodi, AS., Greenway, D.J., Ruggiero, H.R,
Rivero, E., Frigerio, M.J., Mettler, N., Garzon,
F., Boxaca, M., de, G.L.B. and Nota, R 1958.
Sobre la etiologia del brote epidemica de
Junin. Dia Medico, 30, 2300-2302.
Peters, CJ., Buchmeier, M., Rollin, P.E. and
Ksiazek, T.G. 1996. Arenaviruses. In: Fields,
B.N., Knipe, D.M. and Howley, P.M. ed.,
Virology. Philadelphia, Lippincott-Raven,
1521-1551.
Peters, CJ., Mills, J.N., Spiropoulou, CF., Zaki,
S.R and Rollin, P.E. 1999. Hantaviruses.
Chapter 113. In: Guerrant, RL., Walker, D.H.
and Weller, P.F. ed., Tropical infectious
diseases, principles, pathogens, and practice.
New York, W.B. Saunders, 1217-1235.
Plyusnin,A, Valpalahti, 0., Vasilenko, V.,
Henttonen, H. and Vaheri, A 1997. Dobrava
hantavirus in Estonia: does the virus exist
throughout Europe? Lancet, 349,1369.
Redford, KH. and Eisenberg, J.F. 1992.
Mammals of the neotropics, the southern
cone. Chicago, The University of Chicago
Press.
Robbins, CB. and Van DerStraeten, E. 1989.
Comments on the systematics of Mastomys
Thomas 1915 with the description of a new
West African species. Senckenbergiana
Biologica, 69, 114.
Salas, R, de Manzione, N., Tesh, RB., Rico Hesse,
R, Shope, RE., Betancourt, A., Godoy, 0.,
Bruzual, R, Pacheco, M.E., Ramos, B., Taibo,
M.E., Tamayo, J.G., Jaimes, E., Vasquez, C,
Araoz, F. and Querales, J. 1991. Venezuelan
haemorrhagic fever a severe multisystem
illness caused by a newly recognized arenavi-
rus. Lancet, 338,1033-1036.
Schmaljohn, CS. and Hjelle, B. 1997. Hantavi-
ruses: a global disease problem. Emerging
Infectious Diseases, 3, 95-104.
Schmidt, K, Ksiazek, T.G. and Mills, J.N. 1998.
Ecology and biologic characteristics of
Argentine rodents with antibody to
155
Ecologically-based Rodent Management
hantavirus. The Fourth International Confer-
ence on HFRS and Hantaviruses, March 5-7,
1998, Atlanta, Georgia, USA (Abstract).
Toro, J., Vega, J.D., Khan, AS., Mills, J.N.,
Padula,P., Terry, W., Yadon,Z., Valderrama,
R, Ellis, B.A, Pavletic, c., Cerda, R, Zaki, S.,
Wun- Ju, S., Meyer, R, Tapia, M., Mansilla, c.,
Baro, M., Vergara, J.A., Concha, M., Calder6n,
G., Enria, D., Peters, c.J. and Ksiazek, T.G.
1998. An outbreak of hantavirus pulmonary
syndrome, Chile, 1997. Emerging Infectious
Diseases, 4, 687-694.
156
Wilson, D.E. and Reeder D.M. 1993. Mammal
species of the world, a taxonomic and
geographic reference. Washington, D.C.,
Smithsonian Institution Press.
Yahnke, c.J., Meserve, P.L., Ksiazek, T.e. and
Mills, J.N. 1998. Prevalence of hantavirus
antibody in wild populations of Calomys
laucha in the central Paraguayan Chaco. The
Fourth International Conference on HFRS
and Hantaviruses, March 5-7,1998. Atlanta,
Georgia USA, (Abstract).
 Appendix 1. Currently recognised hantaviruses and the diseases they produce, the small mammal host species and host distribution.
Nomenclature and distributions from Wilson and Reeder (1993).

Host subfamily Reservoir Virus Distribution of reservoir

Murinae Apodemus agrarius Hantaan HFRS C. Europe, S to Thrace, Caucasus, and Tien Mtns; Amur River
through Korea, to E. Xizang and E. Yunnan, W. Sichuan, Fujiau ,
Taiwan .
A. flavicollis DObrava England, Wales; NW Spain, France, Denmark, S. Scandinavia
through European Russia, Italy, Balkans, Syria, Lebanon, Israel ;
------IINetherlands
Bandicota indica Thai not known Sri Lanka, India, Nepal, Burma, S. China, Taiwan, Thailand,
Laos, Vietnam ; introduced to Malay Peninsula and Java
Rattus norvegicus Nearly worldwide
Arvicolinae Clethrionomys glareolus Puumala not known France and Scandinavia to Lake Baikal, S to N Spain, N Italy,
Balkans, W Turkey, N Kazakhstan ; Britain, SW Ireland
-4
C. rufocanus not-named E7. ,.~I Scandinavi a through Siberia to Kamchatka, S to Ural Mtns, Altai =-
ID
~----_~I
Mtns, Mongolia, Transbaikal, N. China, Korea, N. Japan ____ ::a
o
Lemmus sibericus Topografov not known Palearctic from White Sea, W Russia, to Chukotski Peninsula,

Microtus arvalis
NE Siberia, Kamchatka; Nearctic from W Alaska E to Baffin
Island, Hudson Bay, S in Rocky Mtns to C. British Columbia
Spain through Europe to Black Sea and Kirov region, Russi a;
o
-
iD

::a
o

-=
Cl..
ID
~_______ ,Orkney Islands, Guernsey, and Yeu (France)
en
M. rossiaemeridionalis Tula not known From Finland E to Urals, S to Caucasus, thr()ugh Ukraine Eto
Rumania, Bulgaria , S. Yugoslavia, N Greece, NW Turkey =
1'1'1
M. californicus IslaVista ___~~""," SW Oregon through California, USA, to N Baja California, Mexico 3
M. fortis Khabarovsk not known Transbaikal and Amur Region S though Nei Mongol and E China
...
ID
1!9,
to lower Yangtze Valley and Fujian =
(IQ
M. ochrogaster Bloodland EC Alberta to S Manitoba, Canada S10 N Oklahoma and :z:
c:::
Lake ~______..., trkansas E to C Tennessee and W Virginia, USA 3
I
=
~
en
....U1 ID
I
en
....... ID
.... Appendix 1. (Cont'd) Currently recognised hantaviruses and the diseases they produce , the small mammal host species and host distribution. rI1
n
~ Nomenclature and distributions from Wilson and Reeder (1993). o
o
OQ
Host subfamily Reservoir Virus Disease Distribution of reservoir n'
III
Arvicolinae M. pennsy/vanicus Prospect Hill not known C Alaska to Labrador, Newfoundland, Prince Edwards Island; S in -<
I
(cont'd) Rocky Mtns to New Mexico , Great Plains to N Kansas, er
III
Appalachians to N Georgia, USA III
ID
Cl.
Sigmodontinae Akodon azarae Pergamino not known NE Argentina, S Bolivia , Paraguay. Uruguay, S Brazil
~
o

..=
B%mys obscurus Maciel not known S Uruguay and EC Argentina Cl.
ID
I[ - -
Ca/omys /aucha Laguna Negra HPS N Argentina and Uruguay, SE Bolivia. W Paraguay, WC Brazil
O/igoryzomys chacoensis Bermejo not kn own W Paraguay, SE Bolivia, WC Brazil , N Argentina
==
III
[ O. flavescens Lechiguanas HPS SE Brazil , Uruguay, Argentina =
III
OQ
[-~--- - . O. /ongicaudatus Ande s HPS Andes of Chile and Argentina ID
:I
[
[
O. /ongicaudatus?
O. microtis
Oran
Rio Mamore
HPS
not known
Andes of Chile and Argentina
C Brazil , contiguous lowlands of Peru , Bolivia, Argentina
..=
ID

Oryzomys pa/ustris Bayou HPS SE USA

Peromyscus /eucopus New York HPS C and E USA into S and SE Canada, S to Yucatan Peninsula,
Mexico
P. manicu/atus Sin Nombre HPS Alaska across N Canada, S through USA to S Baja California and
NC Oaxaca, Mexico
Reithrodontomys mega/otis EIMoro not known SC British Columbia and SE Alberta , Canada, Wand NC USA, S
Canyon to N Baja California, and interior Mexico to C Oaxaca
R. mexicanus Rio Segundo not known S Tamaulipas and WC Michoacan, Mexico S to Panama ; Andes
of Columbia, Ecuador
Sigmodon a/stoni Cano not known NE Colombia , Nand E Venezuela, Guyana , Surinam, and N Brazil
Delgadito
S. hispidus BlackCreek HPS SE USA, interior Mexico to C Panama, N Colombia and
Canal N Venezuela
Unknown Juquitiba HPS (Human cases from Brazil)
Non-rodent Suncus murinus (insectivore) Thotopalayam not known Afghanistan . Pakistan , India, Sri Lanka, Nepal, Bhutan, Burma,
China , Taiwan. Japan, Indomalayan region; introduced to coastal
E Africa, Madagascar. Comores, Mauritius, Reunion &
coastal Arabia.
 Appendix 2. Currently recognised arenaviruses and the diseases they produce, small mammal host species and host distributions.
Nomenclature and distributions from Wilson and Reeder (1993).

Host subfamily Reservoir Virus Disease Distribution of reservoir

Murinae Arvicanthus sp. Ippy Not known S Mauritania, Senegal, Gambia , E through Sierra Leone , Ivory
Coast, Ghana , Burkina Faso , Togo, Benin, Nigeria, Niger,
:1 Chad , Sudan, Egypt, to Ethiopia ; S through N Zaire , Uganda,
S Burundi , Kenya, S Somalia & Tanzania, to E Zambia
Mastomys natalensis Mopeia Not known S Africa as far north as Angola, S Zaire , and Tanzania
Mastomys spp . Lassa Lassa fever Africa south of the Sahara
Mus musculus Lymphocytic LCM Most of world in association with humans
choriomeningiti s
Praomys sp . Mobala Not known C Nigeria through Cameroon Republic and Central African
Republic, S. Sudan, Zaire, N Angola, Uganda, Rwanda,
Kenya , south through E Tanzania to Nand E Zambia
I
-I
Sigmodontinae Bolomys obscurus Oliveros Not known S Uruguay and EC Argentina
~ ::r
111
Calomys cal/osus Machupo Bolivian N Argentina , E Bolivia , W Paraguay, WC to EC Brazil ::u
0
hemorrhagic
I
.
I
- ----
C. cal/osus Latino
fever
Not known N Argentina, E Bolivia , W Paraguay, WC to EC Brazil
----'
=:=J -
ii"
0
::u
0

-==
Cl.
C. musculinus Junin Argentine Nand C Argentina , E Paraguay 111
hemorrhagic
I II
fever
Neacomys guianae Amapari Not known Guianas , S Venezuela, N Brazil
:=J ...,
Neotoma albigula Whitewater Not known SE California to S Colorado to W Texas, USA, south to :I
Arroyo Michoacan & W Hidalgo, Mexico
...
111
'!S,
Oryzomys buccinatus? Parana Not known E Paraguay and NE Argentina =
IrQ
:z:
O. albigularis Pichinde Not known N & W Venezuela, E Panama , Andes of Colombia & Ecuador
to N Peru
=
:I
III
Oryzomys sp.? Flexal Not known Not known =
l1:li
S. alstoni Pirital Not known NE Colombia , Nand E Venezuela , Guyana, Surinam , N Brazil iij'
....
U1
111
III
III
U) 111
 Appendix 2. (Cont'd) Currently recognised arenaviruses and the diseases they produce, small mammal host species and host distributions. ....,
!""'
n
Nomenclature and distributions from Wilson and Reeder (1 993 ). 2-
C)
IrQ
Host subfamily Reservoir Virus Disease Distribution of reservoir t=j.
I
Sigmodontinae S. hispidus Tam iami Not known SE USA, Mexico to C Panama, N Colombia and N Venezuela ~
0-
Zygodontomys brevicauda Guanarito Venezuelan S Costa Rica through Panama, Colombia , Venezuela, I
III
hemorrhagic Guianas, to N Brazil; including Trinidad & Tobago and smaller 111
Cl.
fever islands adjacent Panama & Venezuela
~
C)
Unknown Sabia Unnamed (Human cases from Sao Paulo State, Brazil)

-==
Cl.
111
Non-rodent Artibeus (bats)? Tacaribe Not known (Isolates from bats on Trinidad and Tobago)

==
I
I
IrQ
111
3

-=
111
 Section 2

Methods of Management
 7. Rodenticides - Their Role in
Rodent Pest Management in
Tropical Agriculture

Alan P. Buckle

Abstract

Rodents are serious pests of tropical agriculture. Most crops are attacked ,
particularly those grown for food by smallholders in the tropics. Globally , princi pal
pest species include Sigmodon hispidus, Arvicanthis ni/oticus, Mastomys
nata/ensis, Meriones spp., Bandicota spp., Rattus argentiventer and Microtus spp.
Crop protection specialists usually recommend control programs based on
integrated pest management (IPM) technologies involving the use of rodenticides in
combin ation with various techniques of habitat manipulation . However, few proper
IPM schemes have been developed and implemented on a wide-scal e and long-term
basis. Rodenticides are much used by growers. Acute compounds , such as zinc
phosph ide, are popular with smallholders because they are ch eap but are rarely very
effect ive. First generation anticoagulants (e.g. warfarin) are potentially effective, but
only where th eir use is well managed because of the need for freq uent applications
of ba it in rel at ively large quantities . Baits contain ing t he potent second generation
compounds (e .g. brodifacoum and flocoumafe n) are likely t o be the most effective
because of th e small amounts of bait and labour needed when they are appli ed, but
questions remain about th eir potential to have adverse environmental impacts in
agro-ecosystems . Rodentici des will be im porta nt in rod ent pest manage ment in
tropi cal agriculture for t he foresee able futu re but much remains to be done to
optimise their use. Improved decision-making methods , the wider assessment of
non-target hazard, synergies between rodenticides and other rat man agement
technologies and more sustain ab le exten sion programs are all areas req uiring
development. Unfortun ately, few agenci es now seem willing to expend effort on such
research, although novel techniques to replace rodenti cide s still seem a long way off.

Keywords:

Rodents, rodenticides , rodent control, anticoagulants, resistance, integrated pest

management , rice , sugar cane, oil palm , tropical crops

163
Ecologically-based Rodent Management
THE RODENT PESTS OF
TROPICAL AGRICULTURE
EW TROPICAL crops are free from
F rodent attack. Among common
crops, perhaps only mature stands
of rubber (Hevea brasiliensis) and some crops
grown for fibre, such as sisal (Agave sisalana),
are immune from damage by these
ubiquitous pests. Crops grown in tropical
agro-ecosystems for food, such as cereals
(rice, wheat, maize, millet, barley and
sorghum), roots, fruit, legumes and
vegetables are particularly susceptible to
rodent depredation. Also, crops cultivated
on an industrial scale in plantations, such as
sugarcane, coconut, cocoa and oil palm are
commonly attacked. The extent of losses in
these agro-ecosystems is highly variable.
Two damage models may be recognised. In
the first, if left unchecked by some form of
management practice, rodent populations
reach the carrying capacity of the standing
crop they infest. This is frequently very high
due to the abundant rodent food and cover
that the crops offer. Economically significant
losses in the region of 5-25% are often
inflicted (Wood 1994). This type of damage
may be overlooked both by farmers and crop
protection specialists and becomes apparent
only when carefully planned damage
assessment programs are implemented over
large crop areas (e.g. Posamentier 1989;
Salvioni 1991). Within this model, patterns
of the supply of irrigation water and
subsequent harvesting sometimes
concentrate rodent populations from a wide
area into relatively small tracts of crop land
at the end of the season and some farmers
then suffer very heavy losses. The second
164
pattern of damage is one in which certain
overriding climatological or demographic
phenomena create specific conditions for
rodent populations to reach extraordinary,
or plague, levels. At such times crops may be
totally devastated. The development of very
high populations of Mastomys natalensis after
unseasonal rains in East African crop lands is
an example of this type of episode
(Mwanjabe and Leirs 1997). Another is the
very high populations of rodents that occur
in some parts of Southeast Asia coincident
with the irregular flowering of bamboo
forests (Singleton and Petch 1994).
The number of tropical rodent pest
species involved is very large and appears to
present a bewildering challenge to those
attempting to develop sound management
strategies. However, global rodent pest
problems were classified following work by
the Expert Consultation of the Organisation
for Economic Cooperation and Develop-
ment, Food and Agriculture Organization
and the World Health Organisation into
seven key components of global significance
(Drummond 1978) and this still provides a
useful framework. Six of these problems are
to be found in tropical and sub-tropical,
food-crop, smallholder agriculture (Table 1).
The seventh is the cosmopolitan problem of
rodent damage to stored products, mainly
by Rattus norvegicus and Rattus rattus.
The purpose of this chapter is to review
some of the learnings obtained from a
number of research and development
projects aimed at introducing management
strategies for these pests of tropical
agriculture. The majority of these projects
were broadly based investigations including
the assessment of damage levels, studies of
rodent biology and the development and
 Th e Role of Rodenticides

implementation of rodent management w ide range of nationa l and international

methods. In rela tion to the latter, many
studies were based on the use of
roden ticides, altho ugh a n umber of
subsid iary techniques were frequently
incorporated to p rovide elements of
integrated pest management (rPM).
INTEGRATED PEST MANAGEMENT AND
THE USE OF RODENTICIDES
Few who devise and evaluate rodent
management strategies fail to advocate
integrated approaches as the most reliable,
long-term so lutions to rodent prob lems (see
Richards and Buckle 1987; Mwanjabe and
Leirs 1997, among many oth ers). A review of
the principles of rodent rPM was recently
provided by Singleton (1997). This ana lysis
indicates that strong rPM programs must be
environmentally sound, cost-effective,
sustainable, capable of application over
large areas and recognisably advantageous,
both for growers who implement them and
politicians who support and fund them.
However, after many years of work by a
agencies very few schemes currently operate
to fulfil th ese criteria (Leirs 1997).
All too often those w ho conduct rodent
control programs pay only li p service to rPM
ideas and rely almost so lely on roden ticides .
There are many reasons for this but
paramount is the fact that, although
potentially effective, many of the techniques
that co mp lement rodenticides in rPM are
labour-intensive and their impact is not
immediately obvious to th ose w ho must
invest scarce resources to implement them;
in effect they do not satisfy Singleton'S
criteria. The control of rice-field rats in
Southeast Asia through hab itat
manipulation is a case in point.
[The following is based mainly on work
with Rattus argentiventer (Lam 1978, 1990)
but may be relevant to other rice rat species
in Asia, such as Rattus flavipectus, Rattus losea
and Rattus rattus l1lindanensis, and also
elsewhere.] Some of the conditions of rice
cultivation that exacerbate rat problems
have been long understood (Buckle et
a1.1985; Lam 1990; Leung et aI., Chapter 14).

Table 1.
The world's major rodent pests of agriculture (from Drummond 1978) * .

Rodent pest species involved Area affected Crops attacked

Sigmodon hispidus Centra l and Latin America ri ce, sugar , cotton

Arvicanthis niloticus , su b-Saharan Africa food crops
Mastomys (Praomys) natalensis
Meriones spp. North Africa , Midd le East cerea ls
Bandicota bengalensis Ind ian sub-continent , sugar, cereals, food crops
Southeast Asi a
Rattus argentiventer South ea st Asi a rice (oil palm)

Rattus rattus , Rattus norvegicus, Oceanic isl ands coconuts, food crops
Rattus exulans

* For various reasons certain regions and pests were omitted in th is analysis. However. a com pl ete list of global
rodent pest probl ems of open-field agriculture would certainly also include those caused by Rattus fiavipectus
in southern China and Indochi na, Microtus spp. across the Holarcti c and Mus musculus in mainland Australia.

165
Ecologically-based Rodent Management
Rats choose to build nests for breeding
almost exclusively in rice-field bunds that
are more than about 300 mm wide and 150
mm above water level. They breed primarily
during the reproductive stages of the rice
plants and asynchronous planting allows
prolonged breeding by permitting rats to
move from harvested fields to others nearby
where rice is still at an appropriate stage for
reproduction. Weedy rice fields (Drost and
Moody 1982), as well as overgrown,
uncultivated areas either in or nearby rice
fields provide refugia for rats and
supplementary sources of food. Habitat
manipulation measures to overcome these
problems are obvious; a reduction in bund
size, synchronous sowing/transplanting
and clean rice field cultivation practices, but
all are almost impossible to implement on a
wide scale because of other, overriding
agronomic and socioeconomic factors.
In contrast, rodenticides have a high
potential to contribute useful elements
within rodent IPM strategies (Singleton
1997). Of particular importance is their
relatively low cost, both in terms of the price
of baits in relation to the value of the crop to
be protected and the labour needed to apply
them. Therefore, rodenticides seem likely to
remain central to rodent management
strategies for some time to come.
RODENTICIDES AND THEIR USE IN
TROPICAL AGRICULTURE
The types of rodenticides, the techniques
used in their application and some of their
advantages and disadvantages were
reviewed recently in a general account by
Buckle (1994). A discussion of them is given
here in relation, particularly, to their
application in tropical agriculture.
166
Acute rodenticides
The fast-acting, acute rodenticides are still
much used by tropical smallholders,
although zinc phosphide is now the only
specific rodenticide in this class that remains
widely available. In the absence of
alternatives, growers frequently apply as
rodenticides other compounds with high
mammalian toxicities (e.g. certain organo-
chlorine and organo-phosphide insecticides)
contrary to the regulatory approval of the
compounds concerned.
The benefits of the acute compounds
mainly lie in their ready availability, low
cost and rapid action. They are favoured by
tropical farmers because their effects are
apparent almost immediately after
application. To be set against these
advantages are their disadvantages. They
are sold as concentrates and before use must
be mixed with bait bases, usually cereals, to
the desired concentrations. Tropical
smallholders are ill-equipped to do this
safely and accurately and often, cereals of
sufficiently high quality to provide attractive
baits are scarce. Acute rodenticides are sold
as powder concentrates and are particularly
prone to adulteration during manufacture
and distribution. These characteristics result
in baits of very dubious quality. Even when
they are properly made, acute rodenticide
baits have the drawback of eliciting 'bait
shyness'. This is where the onset of
symptoms of poisoning in sub-lethally
dosed animals is so rapid that rodents are
able to relate them to the novel food (the
bait) which has caused them. Bait shy
rodents are those that will avoid contact
with the poisoned bait when it is applied in
future. The likelihood of this occurring may

be reduced, but not eliminated, by the use of
'pre-baiting'. In this, the bait base later to be
used in the poisoning campaign is first
offered without poison for several days.
Rodents slowly overcome their suspicion of
the novel food (neophobia) and eventually
feed consistently. Only then is the acute
poison introduced. The use of pre-baiting to
overcome neophobia and reduce bait
shyness is time-consuming, poorly
understood by smallholders and rarely
practised.
Probably the best results that can be
anticipated with the use of zinc phosphide
baits, under practical conditions, were
demonstrated by Rennison (1976) on farms
in the United Kingdom. Zinc phosphide
baits, at 2.5% concentration, were applied by
trained and experienced rodent control
operators. Pre-treatment population
assessment was done by census baiting and
this provided a form of pre-baiting. An
average level of control of 84% of R.
Ilorvegicus was achieved. Few good studies
have been conducted on the efficacy of acute
rodenticides in tropical agriculture and it is
unlikely that this level of success is ever
achieved. Most studies have suffered from a
lack of replication, plot sizes that are too
small and with insufficient separation
between plots different treatments,
poor (or no) statistical analysis and, often, a
lack of detailed explanation of the methods
employed (see Chia et a1. 1990 for a
discussion of field trial methodology). These
failings are common among field studies of
rodenticides and it is not surprising,
therefore, that highly variable results have
been obtained (West et al. 1975; Lam 1977;
Mathur 1997). In spite of the shortcomings of
zinc phosphide, Adhikarya and Posamentier
The Role of Rodenlicides
(1987) used manufactured zinc phosphide
bait cakes in a successful large-scale rodent
control campaign in cereals in Bangladesh.
The recommended concentration of zinc
phosphide for field use varies from 1% to
5%. Zinc baits are generally unpalatable to
rodents and a compromise between the
active ingredient concentration used and the
quantity of bait likely to be eaten must be
reached with the objective of administering
the maximum quantity of the active
ingredient. The preferred concentration is
probably 2-2.5% (MAFF 1976). The bait
bases used are locally available cereals. They
may be soaked overnight in water before the
zinc phosphide is added and this is thought
to enhance uptake (MAFF 1976) but reduces
the stability of bait. The baits are placed in
small piles of 20-50 g at intervals of 5-20 m
on bunds in rice fields or, in other crops,
wherever rodents are active (Lam 1977;
Mwanjabe and Leirs 1997). The rate of
application may be varied, both by the
weight of bait used and the distance between
bait points, in order to accommodate
different pest infestation densities.
Undoubtedly, a few days of pre-baiting with
the cereal to be used later as the carrier for
the active ingredient will enhance
effectiveness.
First generation anticoagulants
The archetypal first generation
anticoagulant rodenticide is warfarin. After
its introduction in the early 1950s, a number
of other compounds were developed,
including pival, coumachlor, coumatetralyl,
and the indandiones diphacinone and
chlorophacinone. However, with the
possible exception of coumatetralyl (e.g.
Greaves and Ayres 1969; Buckle et a1. 1982),
167
Ecologically-based Rodent Ma nagement

there is littl e evidence tha t these com polUlds
differ m uch from each o ther in their efficacy.
All these compound s are m os t po tent when
adm inistered in sm all daily doses. H owever,
their most ad vantageous common fea ture is
their chronic mode of action, which m eans
tha t ba it shyness d oes not arise. These novel
fea tures required th e deve lopmen t of a
d ifferent means of quan tifying the p o tency
o f th e firs t genera tion an ticoagulants. This
was done in terms of the number o f days of
consump ti on of field stren g th ba its required
to obtain a given mortality percen tile an d
resulted in th e expression 'leth al feeding
period ' (LFP) .
Warfa rin was first developed for use
against the No rway rat and it is particularly
effec tive aga inst th at species (Tab le 2). Used
against Norway rats in com mensal
situa tions and in anim al husband ry
(pig/ p oultry sheds, d air ies, bee f-rearin g
units) and other fa rm buildings (mills and
granaries) the virtual elimina ti on of N orway
ra t infes ta ti ons was p ossible fo r th e firs t
time. However, o ther species are less
susceptible to it and am ong the least
susceptible are some importan t pests o f
tropi cal agriculture, such as Mn stol'l1Ys
natnlensis, Meriones spp., Bnndicota spp .,
R. argentiventer and R. rattus. Greaves (1 985)
gave da ta for 'natural resistance' to warfarin
for 11 rod ent species, of which l"line were
pests of agriculture (Table 2). This shows
tha t for only three sp ecies (R . norvegicus,
Sigmodon hispidus and Arvicanthis niloticus) is
th e LFP 99 less than 14 da ys .
It is a reasona ble conclusion that wa rfar in
(and the other similar compOlUlds) is
un likely to be as effec ti ve when used in
agriculture as it is in commensa l situations if
more than two weeks of contin uous no-
choice feeding is required to d eliver an
LFP99

Table 2.
'Natural resistance' to warfarin of key rodent pest species as indicated by the number of days of no-choice
feeding on 250 ppm warfarin baits to achieve lethal feeding period (LFP)50 and LFP99 percentiles (from
Greaves 1985)

Rodent species Feeding period (days)

LFP so LFP gg

Nesokia indica 1.9 3797. 0

Acomys caharinus 5.4 239.3
Mus musculus 4.8 29. 5
Mastomys natalensis 4. 8 26 .0
Bandicota indica 1.4 25.0
Rattus rattus 3.6 21.0
Tatera indica 5.8 19.2
Rattus argentiventer 3.2 15.5
Sigmodon hispidus 3.7 8 .1
Arvicanthis niloticus 3.8 6.0
Rattus norvegicus 1.7 5.8

168

Even against susceptible species, the
effective use of the first generation
anticoagulants requires that baits are
available for consumption by rodents, more
or less continuously, for several weeks.
Baiting programs were developed, primarily
in the Philippines, for use in tropical
agriculture with this requirement in mind
(Hoque and Olvida 1987; Sumangil1990).
Baiting stations were put out at a density of
two to five per hectare and supplied with
about 150 g of bait. The bait used was
generally whole or broken rice grains treated
with anticoagulant powder concentrates and
oil as a sticker. The bait stations were
checked at frequent intervals (at least
weekly) and the bait replenished. More bait
and baiting stations were put out at sites
where complete takes were encountered and
baiting continued until takes of bait ceased
or the crop was harvested. This technique
came to be called 'sustained baiting' and its
development, extension to smallholder
groups and practical application on a
nationwide basis is chronicled in the reports
of the Rodent Research Centre, at Los Banos,
through the mid and late 1970s. This
technique remains the only practicable
method of application of loose baits
containing the warfarin-like compounds in
tropical agriculture.
The sustained baiting technique was
adapted for use with wax-block baits
containing warfarin in oil palm plantations
in Malaysia (Wood 1969). In this practice, a
single 15 g block was placed in the weeded
circles of each palm. The baits were checked
at four-day intervals and replenished where
they were taken. Baiting continued until the
requirement to replenish baits declined to a
predetermined percentage of bait
The Role of Rodenticides
placements, normally 20%. An important
advantage of this system was that the use of
wax blocks removed the need for fabricated
bait stations to protect the bait.
All applications of rodenticides in
agriculture are more cost effective, and their
effectiveness more long lasting, when large
crop areas are treated simultaneously. Thus,
if large numbers of smallholders are
mobilised to conduct baiting campaigns, the
effort required by each farmer is minimised,
the quantities of bait used are small and the
duration of baiting is short (e.g. Buckle
1988). However, the sustained baiting
method can be employed successfully by
single smallholders in small plots, but
almost continuous baiting may be needed.
This creates a 'sink' into which are drawn
rodents from a wide area. Clearly, this
benefits more farmers than the one
conducting baiting and may not be
sustainable because its cost falls so
inequitably, both in terms of effort and
money. Using such a system, Sumangil
(1990) used 44 kg of bait per treated hectare
on small farms, during a 12-week rice
growing season, where rats were numerous.
Second generation anticoagulants
Resistance to the first generation
anticoagulants led to the development of a
further series of compounds of greater
potency that were effective against resistant
rodents. These include difenacoum,
bromadiolone, brodifacourn, flocournafen
and difethialone. A third generation of
compounds is occasionally referred to in
some publications. The last three
compounds differ from the first two in being
more potent but none differs sufficiently
169
 Ecologically-based Rodent Management
from any other to be considered in a class
apart.
Early tests of brodifacoum focused on
the objective of obtaining a degree of
effectiveness against resistant animals that
was equivalent to that of warfarin when
used against fully susceptible ones. Very
low concentrations in baits (5 to 20 ppm) fed
over several days were sufficient to achieve
this objective (Redfem et al. 1976).
\ However, it was soon observed that 50 ppm
brodifacoum baits were effective at
providing very high levels of kilL against
both susceptible and resistant rodents,
when rodents fed for only one day on small
amounts of bait (see, for example, Buckle et
al. 1982, for R. argentiventer). However this
benefit could not be readily realised as a
practical advantage because the delayed
effects of brodifacoum, as an anticoagulant,
meant that given free access to bait, rodents
consume much more before they die than
actually needed to kill them. This resulted in
the development of a technique called
'pulsed baiting' in which relatively small
quantities of bait are put out at intervals
between which there is a period in which
bait is virtually absent; allowing rodents
that have consumed a lethal dose to die
before a subsequent application (see Buckle
et al. 1984; Dubock 1984). The principle
. practical benefit to arise from the use of
pulsed baiting in agriculture is that the
quantity of bait used is substantially
reduced. Successful campaigns have been
conducted in which application rates as low
as one to two kilograms of bait per hectare
have been used (Buckle 1988). To the
advantage of a reduction in the cost of bait
and labour required to transport and apply
it is added a reduction in the amount of
170
active ingredient that enters the
environment. The use of this technique,
with wax-block baits containing one of the
potent second generation anticoagulants,
provides the most practical and cost-
effective method of rodent control using
rodenticides currently available.
Anticoagulant resistance
Resistance to anticoagulants is uncommon in
tropical agriculture. There seems to be a
relationship between the time taken for
anticoagulant resistance to develop and the
degree of selection pressure applied (i.e. the
frequency of use of the anticoagulants and
the proportion of the pest population
exposed). In the tropics, only in oil palm
plantations in Malaysia has this pressure
been such that widespread resistance has
developed to the first generation
anticoagulants (Lam 1984; Wood and Chung
1990). In the United Kingdom, where
resistance has arguably reached its current
extreme, nowhere are resistant rodent
populations impossible to control with
available techniques, although there is a cost
in terms of the need to use the more potent
compounds, sometimes for periods longer
than normal (Greaves 1994) and in greater
quantities. This perspective is not intended
to generate complacency. When
anticoagulants are used in tropical
agriculture it is essential to establish
susceptibility baselines and to monitor pest
populations for subsequent changes in
susceptibility. Published guidelines set out
how this should be done (EPPO 1995). These
baseline studies would also provide
preliminary performance data on active
ingredients and the baits that contain them.

Decision-making
Rodent pest problems in tropical crops are
rarely uniform, either in time or space. If a
rodenticide (or any other control measure) is
to be used cost-effectively, a process is
required by which to decide when and
where to apply it. Frequently in tropical
smallholder agriculture this decision is
made on the basis of subjective judgement,
either by individuals or small groups of
growers, and is often made too late. It is well
established that cost-effective rodent
management is most likely when efforts are
co-ordinated over substantial crop areas.
Surveillance and forecasting systems have
been devised to assist decision-makers in
these circumstances, based either on
information on pest density or on
meteorological observations.
Surveillance systems based on pest
density have been worked out for sugarcane,
oil palm and rice. In sugarcane, the 'Hawaii
trapping index' (Hampson 1984) is widely
used to determine the need for rodenticide
applications. Snap-trap lines are set and
rodent population density, expressed as an
index of trapping success, is used as a
decision-making tool. The pitfalls of this
technique were pointed out by Hampson
(1984) but no better method has been
devised in spite of the great economic
importance of the crop and the significance
of rodent damage as a constraint to
production in some areas.
The assessment of rodent damage can be
used as another indirect index of rodent
density and, if the relationship between
damage and crop loss is understood,
provides additional data on the latter
important parameter. However, much work
The Role of Rodenticides
remains to be done in the majority of crops
on the relationships between rodent
population density, damage levels and crop
losses. An aid to decision-making in this
context is the establishment of the economic
injury level, determined as follows (Dolbeer
1981):
T% = 100(Y/bX) (1)
where
T = economic injury level;
Y = cost of control;
X = value of potential crop loss;
b = constant representing the proportion
of potential loss saved by control.
Khoo (1980) proposed a systematic
damage sampling scheme for use in oil palm
plantations in which the percentage of palms
with fresh damage to fruit bunches provided
a criterion dictating the need for the
application of pulsed baiting with second
generation anticoagulants. Buckle (1988)
conducted large-scale pilot trials of an
integrated rice rat management scheme
which involved farmers undertaking
frequent monitoring of the percentage of rice
hills with rat damage as a trigger for the
need for control action. This parameter is
related to, and more easily assessed than, the
percentage of rat-damaged rice tillers. The
advantages of these methods are that
assessments may be conducted by the
growers themselves, the data obtained
reflects the level of rat damage/yield loss
and that it is possible to target decisions so
that applications may be made, when
necessary, to land parcels of moderate size
(e.g. 50 to 100 ha). A disadvantage is that
sampling is relatively labour intensive.
Climatic factors are of limited importance
as determinants of pest population densities
in seasonal irrigated crops (e.g. lowland rice)
171
Ecologically-based Rodent Management
and in perennial crops (e.g. oil palm) and
decision-making is then best founded on
measures of pest population.<;. However, in
rain-fed crops early-warning systems based
on rainfall data may be useful in predicting
rodent pest outbreaks. Mwanjabe and Leirs
(1997) used such a system, combined with
damage assessments, to predict outbreaks of
M. tlatalctlsis in Tanzanian maize fields. An
advantage of this approach is that useful
information is obtained from established
national meteorological monitoring systems.
Although substantial work is required on a
case-by-case basis to validate the predictive
accuracy of such methodology.
The mechanisms briefly described have
been developed in order to target rodent
control efforts more efficiently but decision-
making is always likely to carry some
uncertainty. Pest populations may be low
until a susceptible crop stage is reached and
then there may be a rapid influx from
neighbouring infested habitats. This makes
it necessary to monitor pest populations in
quite large areas and not exclUSively in
crop lands. All these systems require a
degree of coordination from some central
agency, such as a national surveillance
network or a plantation crop protection
advisor. Only the method developed for use
in sugarcane is widely adopted and none of
those designed for tropical smallholders has
yet gained general acceptance. The challenge
remains to crop protection researchers and
vertebrate pest managers, therefore, to
develop and implement practical and
effective systems for monitoring rodent pest
populations to allow timely and appropriate
management actions to be taken in tropical
smallholder agriculture. Whether direct
rodent population density measures or
172
indirect gauges, such as damage
assessments, are used as the decision-
making tools, a reasonable understanding of
the dynamic relationships between rodent
populations, damage and yield loss is
needed.
ASSESSMENT OF NON-TARGET
HAZARDS
In the 'good IPM check-list' provided by
Singleton (1997) it is probably the need for
schemes to be environmentally sound that
has caused a degree of reluctance among
crop protection researchers to use methods
based on roden !icides in tropical agriculture.
Of course, the requirement for crop
protection practices to inflict no unnecessary
harm on the environment is of the highest
importance but refusal to work with
rodenticides has now reached the level of
'chemophobia' in some quarters. In the
opening chapter of this book a 20-year hiatus
is mentioned during which little progress
has been made in rodent management
in developing countries. This is partly
attributed to too much an emphasis on
rodenticides. However, the contention that
very little innovative work on rodenticides
has been done in tropical agriculture for the
past 10 years is borne out by a study of the
reference list of this chapter and very few
projects receiving funding from
international agencies have included any
substantial element concerned with
improving rodenticide applications.
Methods for the assessment of the
potential for rodenticide applications to harm
the environment are well established
(Edwards et al. 1988; Brown 1994). Generally,
rodenticide applications pose negligible risks

to soil and aquatic systems because of the
nature of the compounds and their methods
of use. This is particularly the case with
anticoagulants. Baits are discrete, used at low
rates of application, and carry low
concentrations of, usually, insoluble active
ingredients, which are bound readily to soil
particles and do not move into plants.
However, by their nature, all rodenticides are
potent vertebrate toxicants. Their principle
risks lie in the potential for non-target
animals to consume directly baits laid for
rodents (primary poisoning) and for
scavengers and predators themselves to be
poisoned when consuming the bodies of
contaminated rodents (secondary poisoning).
Those few extensive field studies that have
been performed to quantify these potential
effects (Tongtavee et a1. 1987; Hoque and
Olvida 1988) have shown that pulsed baiting
with wax blocks containing second
generation anticoagulants poses few risks to
wildlife populations in Southeast Asian rice
fields, but more studies are needed both in
rice and in other agro-ecosystems.
All rodent management techniques have
the potential to affect the environment
adversely and this is not restricted only to
those methods based on rodenticides. For
example, the habitat modification methods
often recommended in rice fields (removal
of weedy land patches, lowering of bunds,
increasing field size, periodic deep flooding
of growing areas and extensive synchronous
planting) would have a significant
detrimental effect on a very wide range of
non-target taxa that rely on these remnant
habitats as their only footholds in an
otherwise ecologically barren rice
monoculture. Such potential impact needs to
be weighed against the possible effect of
The Role of Rodenticides
occasional rodenticide use on a limited
number of predatory and scavenging
species, but such thinking is seldom done.
EXTENSION
The fact that smallholders are the most likely
agency by which rodent control measures,
particularly rodenticides, are to be applied is
often overlooked by those developing
management techniques (Posamentier 1997).
Conflicting pressures on smallholders' time
and money, their uncertain perception of the
importance of the pest problem being
addressed and many other socioeconomic
factors jeopardise the sustainability of
otherwise well-designed and cost-effective
schemes. Adhikarya and Posamentier (1987)
undertook a 'knowledge, attitude and
practice' (KAP) survey to establish first base-
line information on rodent control practice
and perceptions among smallholder cereal
growers in Bangladesh. Armed with this
information they designed a multi-media
campaign to modify beliefs and stimulate
action. This substantial program met with
considerable initial success but its long-term
impact is uncertain.
It is tempting to look for successful
models of extension of sustainable rodent
control programs and attempt to draw
lessons from them. A search for such models
in current smallholder tropical agriculture is
largely fruitless (Leirs 1997). However, oil
palm plantations in Malaysia have long
benefited from well organised control
programs based on anticoagulant baiting.
These programs are founded on a base of
long-term research on the biology and
control of the pest funded by those with
most to gain from its results, the plantation
173
Ecologically-based Rodent Management
sector agri-businesses. As a result, Rattus
tiomanicus is arguably the best understood
rodent pest of tropical agriculture (see Wood
1984; Wood and Liau 1984 a,b). Within an
estate, or estate group, rat management
decisions are made by a single person or
small team, on the basis of well-understood
economic criteria. Resources are usually
available to conduct control operations as
and when necessary. Rodenticide
applications are made by trained workers,
with no other distracting tasks on the day of
application, and baits are applied over
extensive areas with reasonable expectation,
therefore, that the investment will be
rewarded. The situation in smallholder
cropping could not be more different.
Several agencies may be responsible for
decisions, including government crop
protection, surveillance and extension
services, farmer groups and individual
growers; all with their own inertia and
affected by different motivational factors.
The financial implications of action or
inaction are poorly understood and money
is rarely available when it is needed. Work is
done by poorly-trained smallholders, with
conflicting time constraints, and the
treatments are too often made on small areas
with little chance of return on investment
from higher crop yields. In some respects
this is an unequal comparison however. Oil
palm is a perennial crop and lends itself to
rodent pest management because of the
constancy of conditions within crop fields.
Whereas, smallholder systems based on a
mosaic of crop types and pest problems
present much more difficult conditions.
Nevertheless, until some of the problems
mentioned above are overcome the current
poor status of rodent pest management in
174
tropical agriculture mentioned by Leirs
(1997) and in the first chapter of this book
will remain.
CONCLUSIONS
In the medium and long-term we look
forward to the introduction of novel
technologies for rodent pest management
The beginnings of some of these are
described elsewhere in this book. Those
engaged in their development must keep
sight of the reasons for the past failure of
what were considered to be well-designed
crop protection systems but which proved to
be impractical or unsustainable (Singleton
1997). Presently, however, there is an urgent
need for improved rodent pest management
in many smallholder agro-ecosystems in
order to alleviate immediate hardship. For
the time being these are best founded on
IPM principles, with rodenticides used as an
important element. However, more work is
still needed. Decision-making systems are
required to help hard-pressed crop
protections workers to determine when and
where management programs are needed.
More extensive field studies of the non-
target hazards of rodenticides are required
so that objective data are available in order
to dispel fears, if these prove to be
unwarranted, of unacceptable adverse
environmental impacts. Also, the
development is needed of innovative
extension technologies to motivate
smallholder farming communities and to
make well-designed rodent pest
management programs sustainable.

REFERENCES
Adhikarya, R. and Posamentier, H. 1987.
Motivating farmers for action; how multi-
media campaigns can help. Deutsche Gesells-
chaft fUr Technische Zusammenarbeit (GTZ)
GmbH, Eschborn, Germany, Schriftenreihe,
185,209p.
Brown, R.A 1994. Assessing the environmental
impact of rodenticides. In: Buckle, AP. and
Smith, R.H., ed., Rodent pests and their
control. Wallingford, UK, CAB International,
Wallingford,363-380.
Buckle, AP. 1988. Integrated management of rice
rats in Indonesia. FAO (Food and Agriculture
Organization of the United Nations) Plant
Protection Bulletin, 36, 111-118.
Buckle, AP. 1994. Rodent control methods:
chemical. In: Buckle, AP. and Smith, R.H.,
ed., Rodent pests and their control. Walling-
ford, UK, CAB International, 127-160.
Buckle, AP., Rowe, F.P. and Abdul Rahman, H.
1982. Laboratory evaluation of the anticoagu-
lants coumatetralyl and brodifacoum against
Rattus argentiventer in peninsular Malaysia.
Tropical Pest Management, 28,126-130.
Buckle, AP., Rowe, F.P. and Abdul Rahman, H.
1984. Field trials of warfarin and brodifacoum
wax block baits for the control of the rice field
rat, Rattus argentiventer, in peninsular Malay-
sia. Tropical Pest Management, 30, 51-58.
Buckle, AP., Yong, Y.C and Rowe, F.P. 1985.
Damage by rats to rice in South-East Asia
with special reference to an integrated
management scheme proposed for peninsu-
lar Malaysia. Acta Zoologica Fennica, 173,
139-144.
Chia, T.H., Buckle, AP., Visvalingam, M. and
Fenn, M.G.P. 1990. Methods for the evalua-
tion of rodenticides in oil palm plantations.
In: Proceedings of the Third International
Conference on Plant Protection in the Tropics,
20-23 March 1990, Genting Highlands,
Malaysia. Malaysian Plant Protection Society,
19-26.
Dolbeer, R.A 1981. Cost-benefits determination
of black bird damage for corn fields. Wildlife
Society Bulletin, 9, 45-51.
ne Role of Rodenticides
Drost, D.e. and Moody, K. 1982. Rat damage in
weed control experiments in rainfed trans-
planted rice. Tropical Pest Management, 28,
295-299.
Drummond, D.C 1978. The major agricultural
rodent problems of developing countries.
Paper presented to the OECD /FAO /WHO
Expert Consultation on Rodent Problems,
Control and Research, Paris, France, 2-5 May
1978,34p.
Dubock, AC 1984. Pulsed baiting-a new
technique for high potency, slow acting
rodenticides. In: Dubock, A.C, ed., Proceed-
ings of the Conference on the Organisation
and Practice of Vertebrate Pest Control,
Elvetham Hall, UK, 30 August to 3 September
1982,105-142.
EPPO (European Plant Protection Organization)
1995. Guideline for the evaluation of resist-
ance to plant protection products. Testing
rodents for resistance to anticoagulant roden-
ticides. EPPO Bulletin, 25, 575-593.
Edwards, P.J., Brown, R.A, Coulson, J.M. and
Buckle, AP. 1988. Field methods for studying
the non-target hazard of rodenticides. In:
Greaves, M.P., Smith, B.D. and Greig-Smith,
P.W., ed., Field methods for the study of
environmental effects of pesticides. BCPC
Monograph No.40. Thornton Heath, UK,
British Crop Protection Council, 77-88.
Greaves, J.H. 1985. The present status of resist-
ance to anticoagulants. Acta Zoologica
Fennica, 173, 159-162.
Greaves, J.H. 1994. Resistance to anticoagulant
rodenticides. In: Buckle, AP. and Smith, R.H.,
ed., Rodent pests and their control. Walling-
ford, UK, CAB International, 197-217.
Greaves,J.H. and Ayres, P. 1969. Some rodenti-
cidal properties of coumatetralyl. Journal of
Hygiene, Cambridge, 67, 311-315.
Hampson, S.J. 1984. A review of rodent damage
to sugar cane with criteria for the use of
rodenticides. In: Dubock, AC, ed., Proceed-
ings of the Conference on the Organisation
and Practice of Vertebrate Pest Control,
Elvetham Hall, UK, 30 August to 3 September
1982,227-251.
175
Ecologically-based Rodent Management
Hoque, M.M. and Olvida, J.L. 1987. Comparison
of baiting methods for ricefield rats in the
Philippines. In: Richards, CG.J.R and Ku,
T.Y., ed., Control of mammal pests. London,
Taylor and Francis, 237-248.
Hoque, M.M. and Olvida, J.L. 1988. Efficacy and
environmental impact of flocoumafen
(Storm) wax block baits used for rice field rat
control in the Philippines. In: Crabb, A.C and
Marsh, RE., ed., Proceedings of the
Thirteenth Vertebrate Pest Conference, 1-3
March 1988, Monterey, USA, 75-81.
Khoo, CK. 1980. 'Pulsed baiting' with brodifa-
coum baits to control oil palm rats in Malay-
sia. Selangor Planters Associations, Annual
Report/Journal, 198,42--46.
Lam, Y.M. 1977. Zinc phosphide-an effective
rodenticide for the rice field rat, Rattus argen-
tiventer. Malaysian Agricultural Journal, 51,
228--237.
Lam, Y.M. 1978. The rice field rat. Information
Paper No.2. Malaysian Agriculture Research
and Development Institute, 37p.
Lam, Y.M. 1984. Further evidence of resistance to
warfarin in Rattus rattus diardii. MARDI
(Malaysian Agricultural Research and Devel-
opment Institute) Research Bulletin, 12, 373-
379.
Lam, Y.M.1990. Cultural control of rice field rats.
In: Quick, GR, ed., Rodents and rice. Report
of an expert panel meeting on rice rodent
control, 10-14 September 1990, Los Banos,
Philippines, 65-72.
Leirs, H. 1997. Preface. In: Leirs, H. and Schock-
aert, E., ed., Proceedings of the International
Workshop on Rodent Biology and Integrated
Pest Management in Africa, 21-25 October
1996, Morogoro, Tanzania. Belgian Journal of
Zoology, 127 (Supplement 1), 3--5.
MAFF (Ministry of Agriculture, Fisheries and
Food) 1976. Control of rats and mice-refer-
ence manual for pest control personnel.
United Kingdom, MAFF, 97p.
Mathur, R. 1'.1997. Effectiveness of various
rodent control measures in cereal crops and
plantations in India. In: Leirs, H. and Schock-
aert, E., ed., Proceedings of the International
Workshop on Rodent Biology and Integrated
176
Pest Management in Africa, 21-25 October
1996, Morogoro, Tanzania. Belgian Journal of
Zoology, 127 (supplement 1), 137-144.
Mwanjabe, P.S. and Leirs, H. 1997. An early
warning system for II'M -based rodent control
in smallholder farming systems in Tanzania.
In: Leirs, H. and Schockaert, E., ed., Proceed-
ings of the International Workshop on Rodent
Biology and Integrated Pest Management in
Africa, 21-25 October 1996, Morogoro,
Tanzania. Belgian Journal of Zoology, 127
(supplement 1),49-58.
Posamentier, H. 1989. Rodents in agriculture-
review of findings in Bangladesh. Sonder-
publikation der GTZ No.16. Deutsche
Gesellscha ft fiir Technische Zusammenarbeit
(GTZ) GmbH, Eschborn, Germany, 108p.
Posamentier, H. 1997. Communication in
national rodent management programmes.
In: Leirs, H. and. Schockaert, E., ed., Proceed-
ingsofthe International Workshop on Rodent
Biology and Integrated Pest Management in
Africa, 21-25 October] 996, Morogoro,
Tanzania. Belgian Journal of Zoology, 127
(supplement 1),171-180.
Redfern, R, Gill, J.E. and Hadler, M.H. 1976.
Laboratory evaluation of WBA 8119 as a
rodenticide against warfarin-resistant and
non-resistant rats and mice. Journal of
Hygiene, Cambridge, 77, 419--426.
Rennison, B.D. 1976. A comparative field trial,
conducted without pre-baiting, of the roden-
ticideszinc phosphide, thallium sulphate and
gophacide against Rattus l1oroegicus. Journal
of Hygiene, Cambridge, 77,55-62.
Richards, CG.J.R. and Buckle, A.P. 1987.
Towards integrated rodent pest management
at the village level. In: Richards, CGJ.R. and
Ku, T.Y., ed., Control of mammal pests.
London, Taylor and Francis, 293-312.
Salvioni, M. 1991. Evaluation ofrat damage in
rice fields in Madagascar. Tropical Pest
Management, 37, 175-178.
Singleton, GR.1997. Integrated management of
rodents: a Southeast Asia and Australian
perspective. In: Leirs, H. and Schockaert, E.,
cd., Proceedings of the International
Workshop on Rodent Biology and Integrated

Pest Management in Africa, 21-25 October
1996, Morogoro, Tanzania. Belgian Journal of
Zoology, 127 (supplement 1),157-169.
Singleton, C ,R. and Petch, D.A. 1994. A review of
the biology and management of rodent pests
in Southeast Asia. A ustralian Centre for Inter-
national Agriculture Research Technical
Reports, 30, 65p.
Sumangil, J.P. 1990. Control of ricefield rats in the
Philippines. In: Quick, G.R, ed., Rodents and
rice. Report of an expert panel meeting on rice
rodent control, 10-14 September 1990, Los
Banos, Philippines, 35--47.
Tongtavee, K., Hongnark, 5., Atrchawakom, T.,
Hongsbhanich, N., Brown, R.A. and
Richards, C.C.J.R.1987, The safety and
efficacy ofbrodifacoum (Klerat) wax blocks
and zinc phosphide for rodent control in
Thailand. In: Richards, c.G.J.R. and Ku, T.Y.,
ed., Control of mammal pests. London,
Taylor and Francis, 173-180.
West, RR, Fall, M.W. and Libay, J.L. 1975. Field
trial of multiple baiting with zinc phosphide
to protect growing rice from damage by rats
(Rattus rattus mindanensis).In: Proceedings
3rd Annual Scientific Meeting. Crop Protec-
tion Society of the Philippines, 143-147.
The Role of Rodenticides
Wood, B.]. 1969. Population studies on the
Malaysian wood rat (Rattus tiomanicus) in oil
palms, demonstrating an effective new
control method and assessing some older
ones. Kuala Lumpur, Planter, 45, 510--525.
Wood, Bl1984, A long term stud y of Rattus
tiomanicus populations in an oil palm planta-
tion in Johore, Malaysia. 1. Study methods
and population size without controL Journal
of Applied Ecology, 21, 445-464.
Wood, BJ 1994. Rodents in agriculture and
forestry. In: Buckle, A.P. and Smith, RH., ed.,
Rodent pests and their controL Wallingford,
UK, CAB International,45-83.
Wood, RJ. and Chung, C.F. 1990. Warfarin resist-
ance of Rattus tiomanicus in oil palms in
Malaysia and the associated increase in Rattus
diardii. In: Davis, L.R. and Marsh, RE., ed.,
Proceedings of the Fourteenth Vertebrate
Pest Conference, 6-8 March 1990, Sacra-
mento, USA, 129-134.
Wood, B.J. and Liau, SS. 1984a. A long term
study of Rattus tiomanicus populations in an
oil palm plantation in Johore, Malaysia. n.
Recovery from control and economic aspects.
Journal of Applied Ecology, 21, 465--472.
Wood, B.J. and Liau, 5.5. 1984b. A long term
study of Rattus tiomanicus populations in an
oil palm plantation in Johore, Malaysia. Ill.
Bionomics and natural regulation. Journal of
Applied Ecology, 21, 465--472.
177
Grant R. Singlet on , Sudarmaji, Jumanta , Tran Quang Tan and
Nguyen Quy Hung

Abstract

Digging, t rap ping, f looding, netti ng, rat drives and physic al barriers are t he norm fo r
ro dent control in rice fi elds in most developi ng countries. We provid e a bri ef overvi ew
of phys ical methods of contro l ai med at reducing pre-h arvest damage by rodents,
then cons ider in detai l th e use of trap-barrier system s . An important cata lyst for
adopti on of physic al co ntro l in South east Asi a is th e use of bounti es fo r each rat
captured. In Aust ra li a, uses of bounti es to cont rol verteb rate pests have been
si ngularl y unsuccessful. Differing socioeco nomics an d more inte nse trappi ng may
provi de better results in develop ing countries . The re is a scarcity of good dat a to
assess whet he r bounties based on physic al act ions of co ntrol are effective. In
contrast, experimental field studi es support the strategic use of trap-ba rrier
systems (TBS) using early crops ('trap crops ') as a lure to rodents . Experimental
studies in West Java , Indonesi a, and the Mekong and Red River De ltas of Vietna m,
indicate th at TBS plus trap crops (TBS+ TC) are cost-effective in most seasons. Yield
increases of up to 1 t/ha have been recorded up to 200 m from a TB S+ TC. The need
to invest money into traps and fences, which protect neighbouring crops , requires a
community-based approach for rodent management. An untested reco mmendation
is that one TBS+ TC (25 x 25 m) would be sufficient for every 15 ha of rice crop.
Although we require more detailed knowledge of the population ecology and biology
of rodent pest species , what we already know has had an important influence on the
development of management strategies incorporating physical methods.

Keywords

Rice-field rat, physical control, tra~barrier system, bounty, rodent management

178
 Physical Control of Rats in Developing Countries
INTRODUCTION
N DEVELOPING COUNTRIES,
I physical methods of control are
probably the most commonly used
approaches by farmers to combat rodent
pests. This is simply because they generally
cannot afford, or do not have ready access to,
chemical rodenticides, fumigants, nest boxes
for birds of prey, or other forms of rodent
control.
Physical methods have been long
recognised as effective for reducing the
impact of rodents in post-harvest stores and
in intensive animal production units where
they damage structures and foul foods
(Jenson 1965; Brooks and Rowe 1979;
Meehan 1984). Actions include mechanical
proofing inside and outside buildings or
ships, physical barriers preventing access to
an area and various means of trapping.
Nevertheless, post-harvest food loss to
rodents remains a substantial problem in
tropical and sub-tropical regions (Morley
and Humphries 1976; Elias and Fall 1988;
Prakash and Mathur 1988). Post-harvest
losses and impacts on livestock production
will not be considered further in this
chapter. Instead we refer interested readers
to review articles and leaflets on rodent
management in large food stores (Meyer
1994), pig production units (Brown and
Singleton 1997), and small to medium-sized
food stores and food processing units in
developing countries (Posamentier and van
Elsen 1984; Bell 1998).
This chapter will focus on physical
methods aimed at reducing pre-harvest
damage by rodent pests. We will provide a
brief overview of physical methods of
control used in developing countries, then
consider in detail the use of trap-barrier
systems. The latter will cover historical
innovations in the use of the technique, its
efficacy across different rice agro-ecosytems,
benefit-cost analyses, strengths and
weaknesses of the approach, research needs,
and its likely role in ecological and
sustainable management of rodent pests at a
village and district level.
PHYSICAL METHODS-GENERAL
Many inventive techniques have been
developed by farmers in developing
countries to catch or kill rats or to deflect
them from their crops. These include the
methods outlined in Box 1.
Other methods are more peculiar to
particular regions and countries. These
range from placing offerings in the corner of
crops to a particular god, to catching large
male rats, sewing their anus closed and
letting them go again. Farmers believe that
'sewn rats' will become aggressive through
an inability to use their bowels and therefore
scare neighbouring rats away. This latter
technique is inhumane and there is no
evidence that it is effective.
The efficacy of the techniques described
in Box 1 for controlling rodent populations is
rarely assessed. Many are inappropriate
given the risk they present to humans. For
example, in their desperation to protect their
crops from rodents, some farmers redirect
mains-power so that it flows through wire
suspended centimetres above a flood-
irrigated rice crop. The wire is strung around
the margins of the crop, killing any rat that
comes in contact with it. This method has
179
Ecologically-based Rodent Management

Box 1.
Methods promoted by farmers in developing countries to catch or kill rats or to deflect t hem from
their crops
-
Various snare and live-traps , usually made of bamboo , that garrotte a rat or break its back (see
Mathur 199 7 ; Schil ler et aI. , Chapter 18 ).
Bamboo tubes-simply offer cover for rats and either they get stuck or t hey are caught alive a ~
empt ied into a bag.
Digging of burrows to kil l rats in situ ; occasi onally dogs are used to locate burrows or to he lp hu nt
rats flushed fro m burrows (e.g. Posamenti er and van Elsen 1984).
Rat drives or battues-where rats are driven from cover and herded towards nets
(Singleton and Petch 1994).
Stalking at night wit h a kerosene light and a net at t he end of a long handle-in Co Dung village of
Hai Duong province in Vietnam , farmers apply th is method f rom 1900- 2200 hrs at specific ti mes of
the year and each farmer catches from 5-15 kg of rats per night.
Electrocution--electrical wire is strung the length of a rice crop about 10- 50 mm above a flooded
paddy; wet rats that make contact with the wire are quickly killed. As indicated below, this method
presents an unacceptably high risk to human health.
Physical barriers-these usua lly consist of plastic or meta l sheeting and are placed aro und or along
the borders of crops or arou nd areas where grai n is stored (e.g. Lam 1988 ).
Physical barriers plus traps-l ive-multi ple-capture traps are inserted intermittent ly at the base of a -
physical barrier. The t raps are placed against small holes in the barrier. Rats enter the t raps,
attracted to t he developing crop or stored food that is on th e other side of the barrier (e.g. Lam et al.
1990; Singleton et al. 1998).
Meta l rat guards-sheets of metal are wrapped aroun d t he trunk of a tree , higher th an 1 m from t he
ground , to prevent rodents from climbing trees to access fruits . The design of the guards depends on
the climbing habits of t he rodent species ; some are fl at against t he tree, whilst others are conical or
circul ar met al sleeves, flush with the t runk of t he tree but projecting outwards at, or less th an, 9 0
fro m t he t runk (e. g. Posamentier and van Elsen 1984) .
Scaring devices - wh ite cloth or plastic is attached to a bamboo pole approximately 1 .2 to 1.5 m
high. The white material flapping in the wind supposedly mimics the flight of owls and therefore
frightens rats away from the immediate vicinity. These 'scare-owls' are erected in ripening crops
where rat damage is evident.

been observed by one of us (G. Singleton) in BOUNTY SYSTEMS

the Philippines and Vietnam.
In southern Luzon, Philippines, 11
human fatalities were reported in the late
1980s (Quick and Manaligod 1990). In Thai
Binh province in the Mekong River Delta,
three people were killed in 1997.
Bounty schemes in general
In developing coun tries, management
actions are often poorly coordinated. This
results in rats q uickly re invading a reas
where control has been cond ucted.
Sometimes governments introduce a bounty
system as an incentive for w idespread
concurrent control. Inherent weaknesses of
bounty system s are that they require rats to

180
 Physical Control of Rats in Developing Countries
be caught and they are generally invoked
once densities are already high. This leads to
two major problems. The first is that
bounties promote inefficient reliance on
physical methods of rodent control such as
live-trapping, digging and rat drives,
replacing management programs based on
the use of rodenticides, better farm hygiene,
habitat manipulation and! or changes in
farm management practices. The second is
that bounties encourage a crisis
management mentality~acting when rat
numbers are high, rather than the more
appropriate use of early tactical
management (see Redhead and Singleton
1988; Brown et a1. 1998). Often the rationale
for invoking a bounty system is more to do
with political expediency rather than
developing an effective, community-based
management strategy. Governments have to
be seen to be doing something to help
farmers in their fight to save their crops from
the ravages of high density rodent
populations. The collection of tens of
thousands of rat bodies has a strong visual
effect, providing a sense of satisfaction to
farming communities that they have waged
a good fight against their perennial enemy.
Bounty schemes have been around for
hundreds of years and have been adopted in
many countries. In Australia, bounties were
first introduced in 1830 for the tails of
unregistered dogs in metropolitan Sydney.
Since then, bounties have been used for both
introduced (e.g. foxes and wild horses) and
native species (e.g. dingoes, species of
wallaby, Tasmanian tiger) (Breckwoldt
1988). In Australia, as elsewhere, there is no
compelling evidence that bounty schemes
have been successful in achieving their
management aim.
A recent review of bounty schemes by
Hassall and Associates (1998) identified the
following reasons for their failure.
~ Fraud~schemes are abused by people
they are supposed to serve.
~ Harvesting mentality-bounties are seen
as an ongoing source of income rather than
a control measure.
~ Inefficiency of control-financial
incentives promote management systems
which provide bodies of animals; as
discussed above, there are generally more
efficient methods for control.
~ Compensatory growth by pest
populations-unless more than 50% of a
pest population is removed by a bounty,
then at best, the pest population will
maintain numbers through enhanced
survival, higher rates of immigration from
uncontrolled areas and better reproductive
performance (Caughley 1977; Hone et a1.
1980).
~ Inadequate benchmarks for success-few
programs have appropriate success criteria
and so they continue from one campaign to
the next vvith the sole criteria being that
they caught many animals last time
through imposing a bounty.
This review primarily considered the
appropriateness of bounties in Australia. It
concluded that bounties were not a cost-
effective system for managing vertebrate
pests.
Bounty schemes for rodents
Rodents have all the life history
characteristics that suggest they would not
be the appropriate target for a bounty
181
Ecologically-based Rodent Management
scheme. They are highly fecund, can
produce a litter every three weeks, are
extremely mobile and are widely distributed
across a landscape. Moreover, most rat
drives or bounty systems are conducted
once rats have already become a significant
problem. Often then it is too late to protect
the ripening crop.
The issue of compensatory growth of
poputations, therefore, is particularly
important when considering the potential
effectiveness of bounties for controlling
populations of rodent pests. In the case of
Norway rats, Rattus norvegicus, in urban
environments in the United States of
America, populations which have been
reduced to 10-25% of their pre-treatment
level, double their population size within 2-
4 months and are back to >75% of pre-
treatment level by 6-8 months (Emlem et a1.
1948). Similarly, trapping high numbers of
muskrats (Ondatra zibethicus) in Germany,
had little impact on the dynamics of their
abundance. Indeed, it was estimated that
annual loss due to trapping was less than the
number of naturally surplus individuals in a
population (Halle and Pelz 1990).
Perhaps the implementation of bounty
schemes in developing countries may hold
greater promise. In these countries, the
density of people per hectare is up to two
orders of magnitude higher and individual
holdings are measured in fractions of a
hectare rather than thousands of hectares.
In Lao People's Democratic Republic
(POR), the rat bounty is around 70 kip per rat
tail (4,000 kip to US$l). In Indonesia, in West
Java, the rat bounty is 50 rupiah for the head
of a rat (9,000 rupiah to US$l). In Vietnam, in
the Red River Delta, the going price during a
bounty season (bounties are not available in
182
all years) is 200 dong for a rat tail (14,000
dong to US$l). Bounty schemes have been
also implemented in Cambodia and the
Philippines.
In 1991 in Luang Prabang province in
northern Lao PDR, a sparsely populated
region by Asian standards, over 600,000 rat
tails were collected in just 2-3 months (see
Singleton and Petch 1994 for details). The
bounty scheme stopped because the money
ran out. These figures are impressive and it
may have been a successful campaign. The
officials that one of us (G. Singleton) spoke to
were certainly impressed by the number of
rats they caught and had little doubt about
its success. However, there was no
quantitative assessment of whether there
was a substantial impact on pre-harvest
losses caused by rats. In that year it was still
common for growers to report losses of
greater than 50% to their crops (WaIter
Roder, pers. comm.).
In August 1998, a rat bounty of 50 rupiah
per rat was instigated in four adjoining
villages in West Java, Indonesia. Over
164,000 rats were collected from 1,790 ha in
less than a month. In one village of 230 ha, an
average of 222 rats were caught per ha. The
bounty was instigated during the land
preparation for a third rice crop for 1998. A
third crop is unusual for West Java and the
mass action against rats was activated to
guard against rat to the newly sown
crop. The action seemed to be successful,
although there was no control site for
comparison and no quantification of crop
damage. Nevertheless farmers were satisfied
with the outcome.
More impressive still were the numbers
of rats caught under a bounty system in
Vietnam. In 1997, 22 provinces applied a rat

bOlUlty schem e for specific times of the year
and 55 million rats were killed . The
combined cost for the provincial
governmen ts involved was approxim ately
62 billion dong (see Table 1).
In 1988, in the first two months of the
yea r, 8.5 m illi on rats were killed throughou t
Vietnam lUld er the bOlUlty system. In the
one province ofVinh Ph uc, over 5 million rat
ta ils were returned from January-September
1998-the bOlU1ty season closed in October.
This is in a province where the hu man
population is around 1.1 m illion.
Regardless of the theore tical evidence
that suggests bounties may be an inefficient
means of controlling rat pop ulations,
digging, trapp ing, flooding, fumigation, and
ra t drives are the n orm for rodent con trol in
rice fields in most developing cOlUltries (see
Ja hn et aI., Chap ter 17 and Schiller et aI.,
Chap ter 18) . Unfor tw1ately, there is a
scarci ty of good data to assess whether these
phys ica l actions of control a re effective or
not. In regions such as Wes t Ja va , the
intensity of physical activities directed at
controlling rats is high. There, some people
get paid a levy on the munber of rats they
Table 1.
Number of rats returned for bounty payments In three northern and three southern provinces of Vietnam for
the first five months of 1997. (Source: Ministry of Agriculture and Rural Development, Vietnam.)
Province Area rice damaged (ha)
Red River Delta (North)
Hai Duong 4 139
Hanoi 10000
Vinh Phuc 67 29
Mekong River Delta (South)
Long An 3500
Quang Ngai 4752
Sac Lieu 2990
Physical Control of Rats in Developing Countries
catch, however most are locked into
conducting nightly con trol campaigns
during the generative sta ge of the rice crop
beca use they can ill afford to lose much of
thei r potential ha rvest to rats. These
intensive physica l ac tivities and bolUl ty
sch emes elsewhere need to be assessed
against specific criteria of success. Apart
from a simple benefit-cost analysis, it is
important also to take into account w hether
the time, effort and resources could have
been more effectively marshalled for an
alternati ve strategy of roden t control. Such a
strategy that ma y even centre on a
coordinated, restricted, bolUlty season th at
shifts focus to earlier tactical interven tion.
PHYSICAL CONTROL AS AN ADJUNCT TO
RODENTICIDE BAITS
Knowledge from both the population ecology
and feeding behaviour of rats indicates that
the best time to use rodenticide baits in and
arolUld rice crops is at maximum tillering.
This coincides with the onset of breeding and
with the final weeks of a 2-6 month fallow
period when food quality and quantity have
been low.
Vietnamese dong paid for bounty
3363257 672651400
650000 130000000
9008700 1801 740000
4600000 100000000
180225 36015000
550000 9000000 ~
183
Ecologically-based Rodent Management
Hence the rat population would be at a
relatively low density and bait acceptance
would be high. Once panicle initiation
begins, rats show low acceptance of baits
(Buckle 1988). In India, local traps then
become a useful control measure together
with fumigation and weed control
(Mathur 1997).
TRAP-BARRIER SYSTEMS
In developing countries, a common method
for protecting a crop from invading rodents
is to use plastic fences to deflect rats and
mice away from the crop. If the rats are
successfully kept out they are generally
deflected into neighbouring crops. The net
effect is that crop losses in a village are rarely
reduced. In the 1980s, Lam (1988) developed
a variation of the drift fence and pitfall
method commonly used for trapping small
mammals. The variation consisted of placing
a plastic fence along the margin of a rice crop
and placing small holes in the fence just
above the irrigation water. Adjacent to each
hole is a multiple-capture cage trap
suspended on bamboo above the water level
(on the crop side of the fence). A mud
mound provides access to the hole and
thence to the trap. The dimensions of the
fence and trap are shown in Figure 1.
This fence plus trap method has been
variably described as the 'environmentally
friendly system', the 'active barrier system',
the 'plastic fences and multi-capture trap'
and the 'trap-barrier system' (TBS). The
trap-barrier system or TBS is now the
commonly accepted description used in
most Southeast Asian countries and is what
we will use in this chapter.
184
The TBS was first developed to protect
crops in areas where rat damage was high
(e.g. crops adjacent to abandoned
agricultural land, early planted crops). In
Malaysia, a TBS that extended for 5 km was
used successfully to protect reclaimed
cropping lands that were planted out of
synchrony. The most rats caught in one
night was 6,872, with 44,101 rats caught in
nine weeks. Subsequent studies in Malaysia
(Lam et al. 1990) and the Philippines
(Singleton et al. 1994) focused on the use of
small rectangular TBSs (0.25 ha to 4 ha).
Again, promising results were obtained
when rat densities and crop losses in
surrounding areas were high. However,
benefit-cost analyses indicated that losses
would have to be greater than 30% for the
TBS method to be cost-effective on a regular
basis (Singleton et a1. 1994; Lam Yuet Ming,
pers. comm.).
More promising results were obtained
when the TBS was used to protect a crop that
was locally attractive to rats, e.g. late-
harvested rice crops or vegetable crops
maturing after the rice crops had been
harvested (see Lam and Mooi 1994). This led
to the development of a second generation
TBS, consisting of an early or late planted
'trap crop' within the TBS which lures
rodents to the traps. The expectation was
that rats from the surrounding areas would
be drawn to the trap crop and then enter the
traps. The TBS plus trap crop (TBS+ TC)
would then provide a halo of protection to
the neighbouring rice crops.
 Physical Control of Rats in Developing Countries

(a) -

Bank ~ HIIIo--- Plastic Fence

Rice Trap
Crop ~ Water Bank
Rat Trap -----
~~~
\ - - 25m

Bamboo

Bank

Earth Mou nd Rat Trap

(c)

Cb)

Figure 1.
(a) Schematic diagram showing the design of the trap-barrier system plus tra p crop of rice (lBS+ TC) of rice .
(b) TBS and lC in Sukamandi; West Java. (c) TBS in position .

185
Ecologically-based Rodent Management
Experimental field studies in different
agro-ecosystems
Most of the early claims of the successful use of
a TBS for controlling rats could not be
substantiated because there were no
appropriate control sites or replication of trials.
Economic data for evaluating the benefit--cost
ratio of a TBS were lacking also. It was as
recent as 1993 that the first replicated and
controlled study was conducted (Singleton et
al. 1994). The results from that study indicated
that the benefits of using a TBS were at best
equivocal. These results switched the focus to
the concept of a T8S+Te, first suggested by
Lam (1988) but which again had not been
properly evaluated.
Beginning in 1995, controlled studies of
the cost-effectiveness of a TBS+Te were
conducted in irrigated lowland rice crops in
West Java, Indonesia. The trap crop was rice
transplanted three weeks earlier than the
surrounding rice crops. The results from the
1995 dry season and the 1995/96 wet season
were extremely promising with benefit--cost
ratios in the vicinity of 20:1 (Singleton et al.
1998). Subsequent studies conducted in
different geo-climatic zones in West Java
(1996-1997) and in the Mekong and Red
River Deltas in Vietnam (1997-1998), have
followed a similar experimental design
(after Singleton et al. 1998), allowing
comparisons of the robustness of the efficacy
of the second generation TBS. The main
variations in experimental design were the
size of the TBS and lack of replicates in the
Vietnamese studies (Tables 2 and 3).
The findings from these experimental
studies are summarised in Tables 2-6. The
main inferences that can be drawn from
these studies are as follows.
186
~ The TBS+Te generally provides a halo of
protection to surrounding crops within 200
m of the fence. The protection is stronger
the closer the crop is to the TBS.
~ The halo of protection provided by a TBS
varies markedly between seasons. In West
Java, protection extended toaminumum of
200 m in two of the three dry seasons, but
was less pronounced beyond 5 m in the wet
seasons. In this climatic zone, the TBS+Te
is generally more cost effective during the
dry season rice crop when rat densities are
generally at least an order of magnitude
higher than in the wet season and their
impact on rice crops is greatest.
~ Yield increases to surrounding crops are
generally 0.3 to 1.0 t/ha.
~ The relative benefit--costs are higher if rat
densities are higher, however the
relationship between rat density and yield
loss does not appear to be linear. Rice crops
are able to partially compensate moderate
tiller damage by rats if it occurs prior to
maximum tillering (see Singleton et al.
1998 for further details).
~ In West Java, the optimum size of a TBS+ Te
is in the range of 20 x 20 m to 50 x 50 m.
When a 10 x 20 m early trap crop was
employed, there was a net loss to farmers.
... The comparative performance of the
TBS+ Te across the different agro-climatic
regions indicates that the technique is likely
to be effective in a wide range of rice agro-
ecosystems. The positive reports from
Malaysia (e.g. Lam and Mooi 1994), where
it was first trialled, adds credence to this
observation.
 Physical Control of Rats in Developing Countries

In Vietnam particularly, and Indonesia in arOlmd 3 kg per ha or 45 kg if the halo of

1995- 96, the yield increases a t the treatment
sites appeared high given the relatively low
number of rats caught. G iven that rats weigh
around 165-200 g and consume about 20-
25% of their body weight per day, then an
ind ividual rat would take about 30 days to
consume 1.5 kg of rice. Yet each ra t
represented a reduction in damage of
protection to the sllrrolmding crop extended
to 15 ha. The number of ra ts caught dming
the TBS studies in Indonesia in the dry
season in 1997, and in Vie tnam in the
summer season in 1997, provide more
convincing cases for the realised increases in
yield (Table 2).

Table 2.
Overview of when rats were caught in 'trap-barrier system (TBS ) plus trap crop' in Indonesia during 1995-
1997. Note the different sizes of TBS. See Singleton et al. 1 998 for methods.

Size TBS(m) Season ....ftIQI Timing of rat captures Total

.5:! rats
Q. Tlllering- Flowering- Harvest caught
QI
a: Booting Heading
Site: West Java, Sukamandi
2 500 m 2 (50 x 50) Dry season 1 63 82 40 185
1995 2 28 45 17 90
Proportion of total rats
33.1% 46.2% 20 .7%
Wet Season 1 96 11 10 117
1995/96 2 42 4 9 55
Proportion of total rats
80.2% 8 .7% 11.1%
200 m 2 (10 x 20) Dry Season 1 96 11 29 136
1996 2 16 27 26 69
Proportion of total rats
54.6% 18 .5% 26 .8%
Wet Season 1 15 6 7 28
1996/97 2 50 4 8 62
Proportion of total rats
72.2% 11.1% 16.7%
2 500 m 2 (50 x 50) Dry Season 1 75 514 117 706
1997 2 43 441 364 848
900 m 2 (30 x 30) 1 65 202 66 333
2 11 86 54 151
400 m2 (20 x 20) 1 46 248 108 402
2 24 85 56 165
Proportion of total rats
10 .1% 60.5% 29.4%

187
Ecologically-based Rodent Management
Singleton et al. (1998) proposed three
factors that together may explain the
apparent disparity between the number of
rats caught and the resulting increase in
yield on the treatment sites. Firstly, each rat
is likely to have damaged many tillers
during the generative stage, compounding
the loss in yield. The earlier these rats are
removed the greater the resulting increase in
yield. Secondly, the removal of rats leads to
substantially fewer females breeding in the
vicinity of each TB5-an important
consideration given that breeding
commences during the maximum tillering
stage, the average litter size is around 10 and
the first litter is weaned prior to harvest.
Thirdly, rats in live-capture traps provide an
early visual cue to farmers to begin other
rodent control activities, leading to more
effective rodent control activities on the TBS
plots relative to the control plots. Typically
in West Java, farmers wait until there is
obvious rat damage to the maturing crop
before embarking on intensive rodent
control activities.
Economics of a second
generation TBS
Cost of a trap-barrier system for trapping
rats in rice crops
The cost of the materials for a 25 x 25 m TBS
with 10 cage traps (allowing for two
replacement traps during a cropping
season), and the labour costs required to
construct a TBS, varies markedly between
countries. In April 1998, the relative costs for
materials were: Indonesia-US$44.75 but
should last for four seasons, therefore the
cost is US$l1.40 (114,250 rupiah) per season;
Malaysia-US$800, should last four seasons,
188
therefore the cost is US$200 per season;
Vietnam-US$80 (1,016 million dong), the
traps last for minimum of two seasons but
not the fence, so the average cost over two
seasons is US$50. In Vietnam, this cost can be
discounted because the used plastic is
adapted for other purposes and the live rats
are often sold to the local market for meat.
The traps are the most expensive items of
a TBS. In Indonesia, they constitute about
60% of the cost. Traps also are easily
removed. It is not uncommon for traps to
disappear overnight, especially when the
system is trialled for the first time in a
district. Generally, however, peer group
pressure at the village level quickly puts a
stop to traps being stolen or 'borrowed'.
Staff at the Research Institute for Rice in
Indonesia have been experimenting with
ways of reducing the cost of traps. The most
promising development is the recycling of
18-20 litre tins which previously held
cooking oil or biscuits. They are about a
quarter of the price of a standard cage trap,
yet they catch about 90 rats for every 100
caught in a standard trap (Table 7). These
recycled traps provide the added benefit of
the possible development of a village-based
industry for their manufacture.
Adoption rate of TBS+ TC technology
The benefit-cost ratio of a TBS+TC varies
from a gain of 20 times the initial investment
in a TBS to a net cost when rat densities are
low (Table 6). High benefit-cost ratios are
only meaningful at the village level, because
they only occur if there is a halo of protection
extending 150 to 200 m from the TBS.
 Physical Control of Rats in Developing Countries

Table 3.
Overview of when rats were caught in 'trap-barrier system (TBS) plus trap crop' in Vietnam during 1997.
Note the different sizes of TBS. Methods were based on Singleton et al. 1998.

Size TBS(m) Season Replicate Timing of rat capt ures Total

Tlllerlng- Flowerlng- Harvest rats
Booting Heading caught

Site: Red River Delta

360 m 2 Spring 1997 Ha Bac 17 34 13 64
Proportion of total rats
26.6% 53 .1% 20.3%
(12 x 30) Summer 1997 Ha Bac 40 76 18 134
Proportion of total rats
29.9% 56.7% 13.4%
Summer 1998 Vinh Phuc 119 54 16 189

-
(12 x 30)
Proportion of total rats
63.0% 28.5% 8.5%
Site: Mekong Delta - Tra Vinh
1000 m 2 Summer 1 (Chien) 184 79 40 303
(30 x 30 m) -Autumn 1997 2 (Cheng) 228 88 67 383
3 148 154 21 323
4 182 87 42 311
5 151 127 34 312
Proportio n of total rats
54.7% 32 .8% 12.5%
Autumn- Spring 1 (Cheng) 72 67 46 185
1997 2 106 89 50 245
3 118 81 62 261
4 105 112 72 289
Proportion of total rats
40.9% 35.6% 23.5%
Site: Mekong Delta - Ho Chi Minh
1 000 m 2 Winter-Spring 1 482 35 5 196 1033
1997 2 529 194 4 72 7
3 551 266 5 822
Proportion of total rat s
60. 5% 31.6% 7.9%

189
Ecologically-based Rodent Management

Table 4.
Effect of the trap-barrier system (TBS) plus trap crop on rice yields (kg/ha) at various distances from the
TBS, in Indonesia. These estimates were based on the weight (water content approximately 14%) of
= =
unhu"ed rice harvested from 10 m 2 quad rats (Repl replicate; nth sample from north of TBS; sth =sample
=
from south of TBS; se standard error of mean yield estimates for the control plots).

Rice yield (kg/ha) Control

Site: West Java
Srn SOm 100 m 1SOm 200 m Mean se
Dry Season 1995
Replicate 1 5600 4750 3500 4750 2313 98.7
Replicate 2 5600 3900 3650 4100 4638 74.7
Mean 5600 4325 3575 4425 3475
Yield relative to control (%) +61% +24% +3% +27%
Wet Season 1995/96
Repll nth 6230 5930 5760 5860 5660 5736 37.6
Repl 15th 5990 6070 5920 5690 5560 5498 44 .5
Repl 2 nth 6630 5620 5560 5490 5780 4736 48.9
Repl 25th 6250 5590 5670 5430 5670 5210 48.5
Mean 6275 5803 5728 5618 5668 5295 88.0
Yield relative to control (%) +19% +10% +8% +6% +7%
Dry Season 1996
Repl 1 nth 4 608 4536 4549 4501 4604 4768 32.8
Repl15th 449 5 4593 4576 4575 4539 4705 25.4
Repl 2 nth 4525 4501 4593 4437 45 10 4646 28 .7
Repl 2 5th 4600 4694 4558 4605 4549 4667 48 .0
Mean 4 557 4581 4 569 4529 4550 469 7 19. 2
Yield re lative to co ntrol (%) -3% - 2% - 3% - 3% - 3%
Wet Season 1996/97
Repll nth 7 312 7 166 7 317 716 5 7 316 7087 12 .9
Repl 1 5th 7 30 1 7201 7112 7135 7 165 7148 52.5
Repl 2 nth 6627 661 5 6634 6580 6761 7 317 35.9
Repl 25th 6580 6782 6622 6611 6639 7273 38.4
Mean 6955 6941 692 1 6873 6970 7206 27.4
Yield relative to control (%) - 3% - 3% -4% - 5% -3%
Dry Season 1997 (50 x 50 m on ly)
Repll nth 5200 5400 5800 5400 5300 4100 114.0
Repl 15th 5000 5000 4900 4900 5000 4000 70 .7
Repl 2 nth 4800 4 700 4500 4600 4400 3920 58.3
Repl 25th 4300 4200 4200 4300 4350 3980 86.0
Mean 4825 4825 4850 4800 4762 4000 41.7
Yield relative to control (%) +21% +21% +21% +20% +19%

190
 Physical Control of Rats in Developing Countries

Table 5.
Effect of the trap-barrier system (TBS) plus trap crop on rice yields (kg/ha) at various distances from the
TBS , in Vietnam. These estimates were based on the weight (water content approximately 14%) of unhulled
rice harvested from 10 m 2 quad rats (se = standard error of mean yield estimates for the control plots ).

Mean Rice yield (kg/ha) a Control

Srn SOm lOOm lS0m 200 m Mean se

Site: Red River Delta
Spring 1997 Yield relative to 5269 5236 5028 4886
control (%) +8% +7% +3%
Summer 1997 Yield relative to 3941 3888 3736 3605
control (%) +9% +8% +4%
Site: Mekong Delta
Summer-Autumn Site 1 (Chien ) 3100 3200 3000 320 0 2700 2817
1997 Yield relative to
control (%) +10% +14% +6% +14% -4%
Site 2 (Cheng) 3 200 3 20 0 3150 3000 3600 28 1 7
Yield rel ative to
control (%) +14% +14% +12% +6% +28%
Winter-Spring (Cheng) 4960 4640 4410 4660 45 20 4 256 4 3 .9
1997/ 98 Yi eld rel ative to
co ntrol (%) +17% +9% +4% +9% +6%

a The mean rice yie lds for each distance from the TBS were from two measurements , except in winter- spring
1997 / 98 when there were six measurements .

Table 6.
The effect of a trap-barrier syst em (TBS) plus trap crop on mean yield increases up to 200 m from the TBS
and the associated benefit-cost ratios, in the Red River and Mekong River Deltas, Vietnam, and West Java,
Indonesia. Costs were calculated from material costs of the TBS and labour costs associat ed with building
the fence and the daily clearing of rats from traps. Benefits were based simply on the increase in yield
rel ative to an untreated site. The dimensions of the respective TBS, the rat densit y during t he growing
season and the t iming of rat damage to t illers, provides context for the variation in benefit-cost ratios .

Year and season Dimensions Rat density Timing of main tiller Mean yield Benefit-cost
of TBS damage increase ratio
(t/ha)
Vietnam
Red River Delta
Spring 1997
Summer 1997
Mekong River Delta
12 x 30 m
12 x 30 m
Low
Low
Flowering to harvest
Flowering to harvest
0.3
0. 3
-
Summer 1997 33 x 33 m
Site 1 Medium No data
Site 2 Medium No data
Winter 1997 33 x 33 m Low/ Med Throughout

191
Ecologically-based Rodent Management

Table 6 . (Cont'd)
The effect of a trap-barrier system (TBS) plus trap crop on mean yield increases up to 200 m from the TBS
and the associated benefit--cost ratios, in the Red River and Mekong River Deltas, Vietnam, and West Java,
Indonesia. Costs were calculated from material costs of the TBS and labour costs associated with building
the fence and the daily clearing of rats from traps. Benefits were based simply on the increase in yield
relative to an untreated site. The dimensions of the respective TBS, the rat density during the growing
season and the timing of rat damage to tillers, provides context for the variation in benefit--cost ratios .

Year and season Dimensions Rat densit y Timing of main tiller Mean yield Beneflt--cost
of TBS damage increase ratio

INDONESIA
West Java
1995 Dry 50 x 50m Very high After booting
1995/96 Wet 50 x 50 m Low Maximum tillering 0.5 7:1
1996 Dry 20 x 10 m Medium Transplanting and -0.1 Net cost
tillering
1996/ 97 Wet 20 x 10 m Low Low damage - 0 .2 Net cost
1997 Dry 50 x 50m Med/ high Maxi mum tillering to 0.8 14:1
harvest (all crops)
30 x 30 m 0 .5 10:1
20 x 20 m 24: 1

Table 7.
Comparison of the efficacy of different trap designs in a trap-barrier system (TBS). See Singleton et al.
(1998) for description of the 'standard trap'. Trap designs 11 to IV are modifications of a recycled 18 litre tin
of vegetable oil (350 x 230 x 230 mm). The comparison was conducted in rice crops at Sukamandi, West
Java, during the 1998 dry season. The rice crops were two weeks old and the traps were set for three weeks
(May 18- June 3). There were three sample plots spaced 500 m apart. Each TBS was 50 x 100 m with eight
traps per plot. One of each trap type was placed in random order along the two 100 m sides of the TBS (SE
= standard error).
Trap type Replicate Rats captured Total Mean SE Cost
( Rupiah)
I (standard trap) 1 51 31 7 105 .7 40.18 30000
2 184
3 82
11 (wire mesh back) 1 22 193 64.3 25.46 6000
2 110
3 61
III (wire mesh front and back) 1 50 277 92 .3 32.41 8 000
2 156
3 71
IV (entrance only wi re mesh) 1 24 69 23 .0 7.81 4 000
2 36
9

192
 Physical Control of Rats in Developing Countries
In developing countries in Asia, this is well
beyond the area of crop owned by an
individual family. However, the results have
been sufficiently promising to have the
governments of both Indonesia and Vietnam
express strong support for the
implementation and adoption of this simple
technology. For example, in the Mekong
River Delta the concept of a TBS+TC was
only first tested in early 1997, yet by May
1998 there were more than 100 TBSs
established in five provinces. In Indonesia,
the field trials on the TBS were initially
conducted on a research farm (440 ha) and
then on a commercial seed farm (1,000 ha
with farmers share-farming areas of up to 5
ha). Following our trials, large TBS+TC (50 x
50 m or 100 x 100 m) were established and
both institutions have been pleased with the
returns for their outlay. At the research farm
there was just one TBS+TC in 1996/97 and it
caught over 26,500 rats. The next year there
were three TBS+ TCs and over 48,000 rats
were caught. In 1998, all the plant variety
trials on the research farm were conducted
within a TBS, and there were more than five
other large TBS+TCs.
In Malaysia, the country of its origin, the
TBS is generally only used in areas that have
acute rat problems (e.g. previously
abandoned fields or asynchrony of cropping
at borders of districts with different
irrigation schedules) or high value crops
(e.g. research farms).
When to use a 185+le?
Effective and efficient pest control strategies
generally have a monitoring protocol that
determines whether particular control
actions need to be implemented. These
protocols are based on preventing a pest
population from reaching a density above
which they cause unacceptable economic
hardship to growers. This is referred to as
the economic injury level (ElL). To prevent a
species reaching its ElL, a lower population
threshold is identified at which appropriate
control actions are implemented.
This threshold level is relatively easy to
define for actions that have a rapid impact
on the pest population, such as the use of
chemical rodenticides (Buckle 1988). This is
not the case for the use of a TBS+Te. In this
situation, the decision point is at land
preparation, to enable the trap crop to be
planted three weeks in advance of the main
crop. By comparison, the decision of
whether to use chemical rodenticides is
made just before maximum tillering of the
rice crop (around day 40-45 post
transplanting).
An informed decision of whether or not
to use a TB5+TC requires a population
model that enables reasonable accurate
forecasts of rodent population densities for
the forthcoming cropping season. These
models have been developed for some
regions for mouse plague management in
Australia Pech et al., Chapter 4),
however such models in Southeast Asia are
lacking, underlining the need for sound
ecological studies of the principal rodent
pest species in rice farming systems.
Effective decision analysis on the use of
TBS+ TC therefore relies on the development
of an ecologically-based management
system for rodent pests.
Weaknesses of the TB5+ le
In weighing up the potential of the
TBS+ TC, an economic benefit-cost analysis
193
Ecologically-based Rodent Management

is one of a number of considerations. O thers rodenticides, provides governments with

include those listed in Box 2.
Whether these points are minor or major
w ill depend on the socioeconomic context of
the end-users and on the effectiveness and
thoroughness of the extension campaign.
Moreover, govenunents ha ve shown
through th e implementation of bounty
systems that they are prepared to invest in
management of rodent pests. This raises the
possib ility of govern ment subsid ies for the
TBS+ TC at village or regional levels.
Subsidising the cost of the materials for a
TBS+ TC would be much cheaper than
funding a bounty system and grain
production is likely also to be higher under a
TBS+TC pest management system.
The exciting potential of the TBS+ TC
acting as a platform for an integrated
strategy for managing rodent pests, and
therefore lessening the reliance on chemical
another op tion for investing funds into
rodent management.
Moving to village-level management
The impressive cost- benefit ratio for the
TBC +TC needs to be viewed in the context
that these were experimen tal studies. The
cha llenge is to transfer this teclulology
readily and effectively to rice farmers. An
im porta nt consid era tion is th e average size
of fam ily holdings in Southeast Asia, which
is 0.5 to 1.5 ha. A TBS which encloses 0.25 ha
could provide protection to neighbouring
farmers without them outlaying money for
materials, providing the labour required to
maintain the TBS or taking the concomitant
risks associated with planting an early trap
crop. Therefore the TBS + TC will be most
effective if it is part of a community-based
approach to rodent pest management.

High init ial cost-many farmi ng fam ilies in Southeast Asia do not have the disposable income to
invest in pest management methods.
High labou r involvement-the traps need to be checked every day, although stoppers (e.g. clump of
straw) can be placed in the opening of the traps on days when no labour is available.
Strong vigi lance on maintenance-the fence needs to be checked da ily for evidence of rats going
through or under the fence: weed growth needs to be controlled near the fence.
Early trap crop attracts avian and insect pests-this needs to be facto red into a
benefit-<:ost analysis.
Mechan ics of growing an arly crop-the main difficu lty is the avai lability of sufficient water three
weeks in advance of th e general irrigation schedule to maintai n firstly a rice nursery and then th e
transpl anted trap crop. An earlier matu ring va ri et y of ri ce may help overcome th is problem.
Non-target captures-am phibians and repti les are caught in th e traps. The experiment al protoco l
requ ires these species be re leased. Whether farmers wou ld re lease all of these species
is problematical.
Humaneness-protocols have been developed (see Singleton et al. 1998) which include the use of
carbon monoxide from the exhaust of motor cycles or automobiles fo r ki ll ing rats. Th e adopti on of
recom mended methods will depend on the operator but he/she should be encouraged to ki ll t he rats
humanely.
Envi ro nmenta l cont am in ation- proper disposal and recyc li ng of th e pl astic fe nces are required .

194
 Physical Control of Rats in Developing Countries
We have determined the ideal size range
for a TBS for farmers (G.R. Singleton and
Sudarmaji, unpublished data), but not the
optimum spatial distribution of these in the
landscape. Although there was much
variation between seasons in the extent of
the halo of protection provided to crops by a
TBS+ Te we recommend that a 25 x 25 m
TBS would significantly reduce rat damage
in the surrounding 10-20 ha of rice crop.
Therefore, at a village level we that
one TB5+ TC would be sufficient for every 15
ha of rice crop. This recommendation has not
been tested.
The spatial distribution of physical
methods for controlling rat numbers is an
important issue given the ability rats have to
re-colonise areas where their densities have
been reduced. In rice fields, rats move
hundreds of metres in a night, especially
once the developing crop reaches the
booting stage (Singleton et al. 1994; P.
Brown, pers. comm.). To reduce this ability
of rats to compensate for control activities,
management needs to be approached
initially at the village level and then at the
district level. A good extension program
with strong grower participation is
fundamental for a community-based control
campaign to be successful (FAO 1997).
At the village level, the spatial
distribution and number of TBS sites will not
simply be determined by the area of land
under rice production. Important
considerations will be how rat populations
respond to:
... the heterogeneity of the habitat (the
seasonal dynamics in habitats where rats
can take safe refuge and/ or breed);
... the degree of asynchronous planting of rice
crops; and
... the variety of other crops grown in the area.
This information requires detailed
studies of the population ecology and
behaviour of Rattus argentiventer and good
documentation of farming practices. There
are some data available on the first two dot
points. For instance, banks along the
margins of rice fields and the banks of the
major irrigation canals provide important
habitats for rats to take refuge in during non-
breeding seasons, and for rats to nest and
breed in after the crop reaches the maximum
tillering stage (see Leung et al., Chapter 14).
Also, the breeding season of R. argentiventer
is linked to the reproductive stage of the rice
crop (Lam 1983; Murakami et al. 1990).
Therefore, asynchronous planting of
neighbouring crops will extend the breeding
season of rats. Although we require more
detailed knowledge of the population
ecology and biology of R. argentiventer, what
we already know has had an important
influence on the development of
management strategies for this species. Our
efforts to manage this species would be
considerably strengthened if we had a better
understanding of the processes that
influenced whether a rat did or did not enter
a trap of a TBS. Towards this end, we need to
develop a better awareness of the
behavioural responses of rats to a TBS+TC
and of the factors that may influence this
response.
CONCWDING REMARKS
In closing, the biggest hurdle facing the
successful use of physical methods for
managing rodent pests is the ability of
195
Ecologically-based Rodent Management

rodent populations to compensate for
reductions in population size through
immigration, increased survival and/ or
better breeding performance. The early
studies of Oavis (1953) clearly demonstrated
the ability of rat populations to recover to
original levels following poisoning
operations. Similarly, H. Leirs (pers. comm.)
has shown that a 50% reduction in a
Mastomys natalensis population, through the
use of chemical rodenticides, has little
impact on the yield loss of crops. However,
sustained harvesting of rats from a
population can lead to the collapse of that
population, presumably because of a decline
in the age structure of the breeding
population (Oavis and Christian 1958).
Together, these studies indicate that one-off
uses of physical control, especially when
rodent densities are high, may have little to
no impact on rat populations. In contrast,
sustained use of physical control methods
over an appropriate spatial scale may be
both cost effective and environmentally
sustainable.
Two methods which warrant further
study are the use of TBS+ TC and the targeting
of bounty seasons at appropriate times of the
year. The timing of the latter needs to be
dictated by our understanding of the
population biology of the rat rather than the
phenology of the crop. For both methods,
success will revolve around coordinated,
synchronised actions at a village or district
level and their ability to be adopted as part of
an integrated approach to rodent
management Singleton 1997;
Leung et al., Chapter 14, for discussion of
other actions).
How the use of physical barriers plus
traps has evolved in our endeavours to
manage the rice-field rat highlights the
imperative of having sound ecological
studies in progress before embarking on
broad scale management programs of a
rodent pest (Leirs et a1. 1996; Singleton 1997).
Further population studies of rodent pests
are planned for Indonesia, Vietnam and Lao
PORI and they will complement our
progress towards optimising the use of trap
barrier systems and trap crops.
ACKNOWLEDGMENTS
The TBS studies in Indonesia and Vietnam
were part of a multi-country study on the
management of rodent pests in Southeast
Asia, funded by the Australian Centre for
International Agricultural Research (Project
numbers ASl/9420 and ASl/9679). We
acknowledge the efforts and commitment of
the support staff from the Research Institute
for Rice, Indonesia, and the Institute of
Animal Sciences and National Institute of
Plant Protection, Vietnam. We also thank
Monica van Wensveen, Ms Rahmini, Ir
Rochman, Nguyen Manh Hung, Nguyen
Viet Quoc, Nguyen Phu Tuan, Lam Yuet
Ming and Luke Leung for their support,
input and ideas. We greatly appreciated the
comments of Oavid Freudenberger and Alan
Buckle on an earlier draft of this chapter.

196
 Physical Control of Rats in Developing Countries
REFERENCES
Bell, A 1998. Integrated rodent management in
post-harvest systems. Eschborn, Germany,
Deutsche Gesellschaft fUr Technische Zusam-
menarbeit (GTZ), leaflet, 8p.
Breckwoldt, R 1988. A very elegant animal, the
dingo. North Ryde, Australia, Angus and
Robertson, 283p.
Brooks, J.E. and Rowe, F.P. 1979. Commensal
rodent control. Geneva, World Health Organ-
isation, WHO/VBC/79.726, 109p.
Brown, P.R and Singleton, G.R 1997. Review of
rodent management for pig production units.
Canberra, CSIRO Wildlife and Ecology,
Report to the Pig Research and Development
Corporation, 45p.
Brown, P.R, Singleton, G.R, Dunn, S.C and
Jones, D.A 1998. The management of house
mice in agricultural landscapes using farm
management practices: an Australian
perspective. In: Baker, RO. and Crabb, AC,
ed., Proceedings of the 18th Vertebrate Pest
Conference, Santa Clara, California, USA, 2-5
March 1998. Davis, University of California,
156-159.
Buckle, AP. 1988. Integrated management of rice
rats in Indonesia. FAO (Food and Agriculture
Organization of theUnited Nations) Plant
Protection Bulletin, 36, 111-118.
Caughley, G.C 1977. Analysis of vertebrate
populations. London, John Wiley and Sons,
234p.
Davis, D.E. 1953. The characteristics of rat
populations. Quarterly Review of Biology, 28,
373-40l.
Davis, D.E. and Christian, J.J. 1958. Population
consequences of a sustained yield program
for Norway rats. Ecology, 39, 217-222.
Elias, D.J. and Fall, M.W. 1988. The rodent
problem in Latin America. In: Prakash, I., ed.,
Rodent pest management. Boca Raton, CRC,
13-28.
Emlem,J.T.,Stokes,AW. and Winsor, CD. 1948.
The rate of recovery of decimated popula-
tions of brown rats in nature. Ecology, 29,
133-145.
FAO 1997. Community based IPM case studies.
Food and Agriculture Organization Inter-
country Programme for the Development of
Integrated Pest Management in Rice in South
and Southeast Asia. FAO, Jakarta, 36p.
Halle, S. and Pelz, H.-J. 1990. On the efficiency of
muskrat (Ondatra zibethicus) control from
trapping data ascertained in Bremen.
Zietschrift fur angewandte Zoologie, 77, 205-
218.
Hassall and Associates Pty Ltd 1998. Economic
evaluation of the role of bounties in pest
management. Unpublished report for the
Bureau of Rural Sciences, Canberra.
Hone,]., O'Grady,J. and Pedersen, H. 1980.
Decisions in the control of feral pig damage.
NSW Agriculture Bulletin, 5.
Jenson, AG. 1965. Proofing of buildings against
rats and mice. Ministry of Agriculture, Fisher-
ies and Food, Technical Bulletin No. 12,
London, Her Majesty's Stationery Office, 18p.
Lam, Y.M. 1983. Reproduction in the rice field
rat, Rattus argcntiventrr. Malayan Nature
Journal, 36, 249-282.
Lam, Y.M. 1988. Rice as a trap crop for the rice
field rat in Malaysia. In: Crabb, AC and
Marsh, RE., ed., Proceedings of the 13th
Vertebrate Pest Conference, 123-128.
Lam, Y.M. and Mooi, K.C 1994. TBS-an
environmentally friendly system for the
control of rodent pests of agriculture.
Proceedings of the 4th International Confer-
ence on Plant Protection in the Tropics, 26-31
March, Kuala Lumpur, Malaysia, 159-160.
Lam, Y.M., Supaad, M.A, Chang, P.M.,
Mohamed, MS., Goh, CS. and Radzi, H.
1990. An innovative approach for protecting
rice against severe rat depredation. Proceed-
ings of the 3rd International Conference on
Plant Protection in the Tropics, Genting
Highlands, Malaysia, 20-23 March 1990, 1-
23.
Leirs, H., Verhagen, R, Verheyen, W.,
Mwanjabe, P. and Mbise, P. 1996. Forecasting
rodent outbreaks in Africa: an ecological basis
for Mastomys control in Tanzania. Journal of
Applied Ecology, 33, 937-943.
197
Ecologically-based Rodent Management
Mathur, RP. 1997. Effectiveness of various
rodent control measures in cereal crops and
plantations in India. Belgium Journal of
Zoology, 127, 137-144.
Meehan, A.P. 1984. Rats and mice. Their biology
and control. Felcourt, East Grinstead, West
Sussex, Rentokil Ltd, 383p. ,
Meyer, A.N. 1994. Rodent control in practice:
food stores. In: Buckle, A.P. and Smith, R.H.,
ed., Rodent pests and their control.
Cambridge, Cambridge University Press,
273-290.
Morley, G.E. and Humphries, J.R.O. 1976.
Rodent damage to farm and village storage.
In: Hop, H.S., Morley, G.E.J. and Humphries,
].RD., ed., Rodent damage to growing crops
and to farm and village storage in tropical and
subtropical regions. Southampton, United
Kingdom, Centre for Overseas Pest Research
and Tropical Products Institute, 61-87.
Murakami, 0., Priyono, J. and Triastiani, H.1990.
Population management of the ricefield rat in
Indonesia. In: Quick, G.R, ed., Rodents and
rice-report and proceedings of an expert
panel meeting on rice rodent control. Manila,
Philippines, International Rice Research
Institute, 49-54.
Posamentier, H. and van Elsen, A. 1984. Rodent
pests-their biology and control in Bangla-
desh. Dhaka, Bangladesh, Department of
Agriculture and Extension, 111p.
Prakash, I. and Mathur, RP. 1988. Rodent
problems in Asia. In: Prakash, r., ed., Rodent
pest management. Boca Raton, CRC,67-84.
198
Quick, G.R and Manaligod, H.T. 1990. Some
aspects of physical control of rice rodents. In:
Quick, G.R., ed., Rodents and rice-report
and proceedings of an expert panel meeting
on rice rodent control. Manila, Philippines,
International Rice Research Institute, 31-34.
Redhead, T.D. and Singleton, G.R 1988. The
PICA Strategy for the prevention of losses
caused by plagues of Mus domesticus in rural
Australia. EPPO Bulletin, 18,237-248.
Singleton, G.R 1997. Integrated management of
rodents: a Southeast Asian and Australian
perspective. Belgian Journal of Zoology, 127,
157-169.
Singleton, G.R., Chambers, L.K. and Quick, G.R.
1994. Assessment of the IRRl active barrier
system (ABS) for rodent control. Canberra,
Final Report to Australian Centre for Interna-
tional Agricultural Research, SOp.
Singleton, G.R and Petch, D.A 1994. A review of
the biology and management of rodent pests
in Southeast Asia. Canberra, Australian
Centre for International Agricultural
Research, Technical Reports, 30, 65p.
Singleton, G.R., Sudarmaji and Suriapermana, S.
1998. An experimental field to evaluate
a trap-barrier system and fumigation for
controlling the rice field rat, lVittus argentiv-
enter, in rice crops in West Java. Crop Protec-
tion, 17, 55-64.
 9. Ecological Management of Brandt's
Vole (Microtus brandtll in
Inner Mongolia, China

Wenqin Zhong, Mengjun Wang and Xinrong Wan

Abstract

Bra ndt's vole (Microtus brandti) is a serious rode nt pest in t he grasslands of Inner
Mongol ia. Poiso n baiting is t he traditiona l approach to cont rolling Brandt's vole in
this region. Although this sharply reduces the dens ity of voles, the remain ing
res ident animals have high reproductive and surviva l rates , leading to rapid recovery
of the population. Poison baiting also causes other problems such as environmental
pollution and secondary poisoning of natural predators . Therefore, a new non-
pol luting, economically efficient technique , offering effective long-term control, is
urgently required for the management of Brandt's vole. In this paper, we investigate
the potential of an ecological strategy in managing Brandt's vole in the grassland of
Inner Mongolia.
We found the main factor facilitating infestation by Brandt's vole in the grassland
is overgrazing by livestock. The density and height of vegetation strongly influence
the habitat selection of Brandt's vole. With heavier grazing pressure by livestock, the
height and cover of vegetation are reduced and the plant composition is changed,
resulting in high quality food and shelter for voles and consequent increases in their
population densities. The density of voles was low where the height of vegetation
was >190 mm but high where vegetation height was 30- 130 mm. Fencing and
pasture management could be used in some areas to reduce vole problems by
increasing the height of grass.
In Inner Mongolia, local herdsmen fence pasture in June to increase herbage for
harvesting in autumn or for grazing by livestock in winter. The grasses grow slowly
and poorly under this traditional fencing management because extensive grazing
suppresses the germination and growth of grasses that would otherwise occur due
to high rainfall and warm temperatures in May. These grasslands then become more
vulnerable to vole infestation . A new fencing management of removing livestock from
pastures in the middle of May, rather than later in June, was examined using a series
of experimental exclosures in Taipus Qi , Inner Mongolia, from 1986-1990. The
densities of Brandt's vole populations in the experimental exclosures were reduced
sharp ly and the grass yields increased greatly compared to the exclosures under the
traditional fencing management. Over the next three years, the average
investment:income ratio was 1:7 . This new ecological management is very
prom ising for solving the Brandt's vole problem in t he Inner Mongolian grass lands.

Keywords
Bra ndt's vole, ove rgrazing, eco logica l ma nageme nt, Inner Mo ngoli a
199
Ecologically-based Rodent Management
INTRODUCTION
RANDT'S VOLE (Microtus brandti)
B is a major rodent pest in the
grasslands of Inner Mongolia,
especially in the typical steppe of the Xilin
Gele region. It can cause problems in an area
of up to 20 million ha or 75% of this region.
The populations of Brandt's vole fluctuate
remarkably, with positive feedback indicated
by 2-3 years of low density after a period of
high density. Each vole can eat 40 g of fresh
plant material per day and in high-density
years, with up to 1,384 individuals/ha, 15-
44% of grass production can be consumed by
voles (Zhang and Zhong 1981). The diet of
Brandt's voles is similar to that of sheep and
cattle (Wang et al. 1992), so it is an important
competitor of livestock.
Brandi's voles live in social groups and
dig complex burrow systems with up to
approximately 5,616 holes/ha in high-
density areas (Zhong et al. 1985b). Because
voles excavate large amounts of soil, they
accelerate erosion and desertification in
grass lands.
Poison baiting is the traditional approach
to controlling Brandi's vole in the Inner
Mongolian grasslands. Although this
sharply reduces the density of voles, the
remaining resident animals have high
survival rates because there are more
resources available to them (Dong et aL
1991). The pregnancy rate increases and
voles become mature earlier, leading to the
rapid recovery of the Brandt's vole
population (Yang et al. 1979; Zhou et al.
1992). Its relative abundance in the rodent
community recovers within four months
(Hou et aL 1993). Chemical control is applied
200
nearly every year but pest problems also
occur each year in a 'pest outbreak -+
chemical control -+ pest outbreak' cycle.
This approach is not only expensive and fails
to provide a long-term solution, but it also
results in heavy environmental
contamination. A new non-polluting,
economically efficient technique, offering
effective long-term control, is urgently
required for the management of Brandt's
vole. Major advances will depend on
ecological studies (Barnett 1988).
The habitat of rodents is determined by
many attributes of the plant community,
such as the distribution and spatial pattern
of plants, the proportions of edible and
inedible plant species, and changes in
phenology, biomass, cover and species
composition. All of these influence the
composition and dynamics of rodent
communities. On the other hand, the
animals' activities also change the
vegetation.
We have investigated the number of
species and the diversity and heterogeneity
of plant and rodent communities in five
types of steppe in the grassland of Inner
Mongolia (Zhou et al. 1982). The results
showed that the rodent and plant
communities are closely connected (Table
1). The diverSity index of the rodent
community is correlated with the diversity
index of the plant community (1' = 0.972,
P < 0.01) and that the heterogeneity index
of the rodent community is also correlated
with that of plants (r 0.905, p < 0.05). Also,
we have investigated the plant-rodent
community in different of the same
type of grassland (Wang et al. 1997; Table
2). The changes in the a-diversity index of
the plant community and rodent
 Ecologica l Management of Bra ndt's Vole in China

Table 1.
Species number, diversity and heterogeneity of plant and rodent communities in five types of steppe.
H is the ShannonW iener index: H = -'LP/In P" where p/ = is the proportion of species i in the community.
J is an index of evenness: J = Hl ln 5, where 5 is the number of species in the community (Zhou et al. 1982).

Type Rodent community Plant community of steppe

No. of H J No. of H J
species specie s

1 7 0 .9 611 0 .49 39 39 1 .744 3 0.4761

2 6 1. 2841 0 .7167 37 2 .2265 0 .6 166
3 9 1.4898 0 .6780 40 2 .498 1 0 .6 772
4 8 1.6348 0 .7862 26 2. 5621 0 .7864
5 5 0. 8 106 0. 50 3 7 26 1.8 158 0 .5573

Table 2.
Comparisons of the a-diversity index between plant and rodent communities along a grazing gradient in two
types of steppe. The impacts of grazing were classified as: I =no grazing, 11 =moderately degraded pasture ,
and III = hea vily degraded pasture. (From Wang et al. 1997.)

Season Community Aneurolepidlum chinense Season Community 5tipa grandis

steppe steppe

11 I11 I 11 II I
Spring Rodent 0.2095 0. 1515 0 .0 Spring Rodent 0.0 0 .3909 0 .4224
Plant 1.20 15 1.1638 1.0860 Plant 0 .97 98 1 .0 289 1.04 10
Autum n Rodent 0 .37 29 0 .1682 0.1283 Autu mn Rodent 0 .0 0 .3195 0.3280
Plant 1 .39 29 1. 2290 1.1621 Plant 1.2238 1. 2253 1.3524

commlmity show the same trend . The
d iversity of p lan ts reflects food ava ilability
for roden ts, heterogeneity is a measure of
the distribution of food, and heigh t and
cover are importa n t spatial factors to which
roden ts are sensitive. The result is th at
changes in any of these p lant comm unity
indices corresp ond to ch anges in the
abu nd ance and d istribution of roden ts.
In Itm.er Mongolia, the number of
livestock has been increasing year by year.
The number of livestock has increased 2.64
times in recent decades, with correspond ing
land d egrad ation reaching 27.5% of a ll the
Itlner Mongolian grasslan d area (Ren et al.
1989). The pressure of grazing has resulted
in degeneration and desertification of large
areas of grassland as well as increasing
salinity of soils. Th e plant community has
responded with an increase in the
proportion of dicoty led ons and a general
decrease in the production and biomass of
vege tation. The success ional changes in the
plant community have been matched by
changes in th e composition of the rodent
community.
Fig ure 1 summarises the sequence of
changes in the p lant-rodent community in
degraded grassland in Itm.er Mongolia
(Zhong et al. 1985a).

201
Ecologically-based Rodent Management

(a) Plant community Rodent community

Aneurolepidium chinense Cricetulus barabensis
Stipa krylovii Citellus dauricus
Ochotona daurica
Lightly over-grazed

Artemisia frigida
Aneurolepidium chinense
Cleistogenes squarrosa
l Ochotona daurica
Cricetulus barabensis
Citellus dauricus

Heavily over-grazed

l
Artemisia frigida Microtus brandti
Potentilla acaulis
Cleistogenes squarrosa
Excessively over-grazed

(b)
Planta annua

Plant community
l Meriones unguiculatus

Rodent community
SUpa krylovii Citellus dauricus
Artemisia frigida Ochotona daurica
Cleistogenes squarrosa
Lightly over-grazed

Ochotona daurica

l
Slipa krylovii
Artemisia frigida Citellus dauricus
Cleistogenes squarrosa

Heavily over-grazed

Heteropappus alta/cus
Artemisia frigida
Salsola collina
l Microtus brandti

Excessively over-grazed

Planta annua
l Meriones unguiculatus

Figure 1.
The impact of grazing by livestock on the plant-rodent communities at (a) Xilin Hot (4325'N, 11641'E;
annual precipitation 350-400 mm) and Cb) the Kelulun River (492S'N, 11642'E; annual precipitation
250-300 mm) in the typical steppe of Inner Mongolia (from Zhong et al. 1985a).

202
 Ecological Management of Brandt's Vole in China
In degraded grasslands, the biomass of
families in the Compositae, Rosaceae and
Chenopodiaceae increase, all of which are
the preferred food of the Daurian pika
(Ochotona dauriea) (Zhong et a1. 1983). The
vegetation of this stage also provides
suitable cover and food for storage for the
Daurian pika (Wang et a1. 1998). With
heavier grazing pressure, the height and
cover of vegetation is reduced, the
availability of food and shelter becomes
unsuitable for the pika, its abundance
decreases and its dominant position is taken
by Brandfs vole. This degenerative stage is
the preferred habitat of Brandt's vole and
populations increase quickly and reach high
densities (Zhong et a1. 1985b). With
excessive grazing, soil becomes susceptible
to erosion and Planta annua becomes the
main component of the plant community.
Brandt's vole is displaced by the Mongolian
gerbil (Meriones unguieulatus) and eruptions
of this species can occur (Xia and Zhong
1966; Zhong et a1. 1983).
Degradation of grassland is facilitated by
the digging and feeding activities of Brandt's
vole. Groups of voles occupy complex
burrow They often excavate soil
when they repair their burrows, especially
when constructing 3-4 storerooms per
burrow system in autumn. As each
storeroom is about 1.1 m long and 120 mm
high, large volumes of new soil are
mounded on the soil surface beside burrow
entrances. Burrow systems can have up to 36
holes covering some 14 m 2, of which 9 m 2
can be covered by fresh soil (Zhang and
Zhong 1981). Around the burrow systems,
the production of fine-grazing grasses such
as Aneurolepidium ehinense, Stipa grandis and
Cleistogenes squarrosa decreases by 20%, 86%
and 83%, respectively, and Artemisia scoparia,
ArtemisiaJrigida, Carex duriuseula and Keoleria
eristata increase to 60% of total plant
production. The result is that Brandt's vole
facilitates the degeneration of pasture for
livestock.
HABITAT SELECTION OF BRANDT'S VOLE
There are no recorded cases of rodent pest
problems in natural grassland or in areas
subject to low grazing pressure. Outbreaks
of Brandi's vole usually occur only in
degraded pasture.
In natural pasture, the location of water
sources, such as rivers, influences the
grazing pressure of livestock. In one study
(Zhong et a!. 1985a) pasture within 3 km of a
river suffered excessive grazing pressure.
The area 3-8 km from the river showed
moderate degradation, while the area 8-13
km from the river showed slight
degradation. In the Kelulun River region, the
production of herbs was 36"/0 or 53/.) lower,
and the density of Brandi's vole 3.6 or 0.9
times higher, in a heavily degraded area
compared to a moderately degraded area
(Zhong et a1.1985a; Table 3).
Other research has reported a similar
relationship between the densi ty of Brandt's
vole and the condition of the plant
community (Uu 1979). A further
investigation was conducted in a livestock
resting site that was no longer used (Zhong et
a1.1985b). Table 4 shows that in May 1974 low
plant cover was associated with a high
density of Brandt's vole, indexed by the
number of holes/ha, whereas in August an
increase in the height and cover of plants,
mostly through rapid growth of Aneuro-
lepidium ehinense and Allium anisopodium,
203
Ecologically-based Rodent Management

Table 3.
The relationship between the density of Brandt's vole and the utilisation of the pasture in a terrace of the
Kelulun River (Zhong et aI.1985a) . The abundance of holes with signs of recent animal activity was used as
an index of the density of voles. (Holes were covered with soil and checked 24 hours later. Re-opened holes
were classified as active). The botanists' standard definitions were used for the degree of degradation.

Distance from State of the pasture Area of Density of Brandt's vole

river (km) investigation (active holes/ha)
(ha)
June 1974 July 1975

1-3 Heavy degeneration 1.5 8.67 883.33

4-8 Medium degeneration 5.3 1.89 471.41
>20 Slight degeneration 4 .0 0.0 0.0

Table 4.
Influence of vegetative change on the density of Brandt's vole at a resting site recently abandoned by
livestock (Zhong et al. 1985b).

Abandoned resting site Normal grazing site

vegetation Vegetation

Height (cm) Cover Voles active Height (cm) Cover Voles active
(%) (holes/ ha) (%) (holes/ha)
May <5 972 10 204
Augu st 40-50 95 o 15- 20 30 84

A. bidenta tul11 and A. tenuiss il1111l1l after high
summer rainfall, corresponded to a sharp
decrease in the density of holes. Apparently,
the low vegeta tive cover in May was due to
grazing by voles, but by August voles would
have dispersed to avoid high, dense
vegetation.
Another stud y was carried out in 1982
by Zhong et al. (1985b). The vegeta tion and
the den si ty of holes used by Brandt's vo le
were monitored in a fenced area (A ) that
excluded livestock but allowed free access
by roden ts. The exclosure was compared to
an area with normal grazing by livestock
(B) and to a livestock resting site (C) . The
results are shown in Table 5. The density of
Brandt's vole was inversely related to the
h eight and cover of vegeta tion. The sites
were ranked A > B > C fo r plant biomass
and A < B < C for the rela tive density of
vo les. Parts of si tes A and B were only 120
m apart, well within the ra nge of
m ovements of Brandt's vole, but the
relative d ensity of voles reached 5,616
ho les/ ha in C (Zhong e t al. 1985b) and the
densi ty in B decreased by 33% to 84%. The
distance from A to B was about 120 m, and
from B to C about 500 m.
The biomass of A. cIT i ll ell se, which is
Brandt's vole's preferred food (Wang et al.
1992), was higher in A than Band C. This
suggests that food is not the main factor
influencing habitat selection by Brandt's
vole. More important factors appear to be

204
 Ecological Managem ent of Brandt's Vole in China

Table 5.
Comparisons of the density of Brandt's vole, indexed by the number of burrow holes/ha, and vegetation
condition in areas with different use by livestock (Zhong et al. 1985b). Data are given as the mean (X) and
the standard error (SE). n is the sample size.

Study site Density of voles Vegetation

No. of Height Cover Biomass

species/m 2 (cm) (%) (g/m2)

n=6 n = 20 n=5 n=5 n=5

X SE X SE X SE

A. exclosure 700 196 16.05 0.47 30 - 35 57 190 17

B. normal grazing 2063 345 14.35 0 .51 13.6 - 16.4 46 102.9 4.3
C. resting site 3661 406 10.25 0 .51 5.4 - 7 .4 33 53 11
Hest A<C* A>C* A>C* A>C*
*( p < 0.01) C<B# C<B* C<B* C<B*
#(p < 0.05) A<B* A>B# A>B* A>B*

the heigh t and cover of vegeta tion, with
voles preferring areas with sparse, low
vegetation. It is likely that high, dense
vegeta tion hinders the socia l behavior of
Brandt's vole such as comm unication for
feeding, matin g and coop erative defense
against p redators (Xin ron g Wan,
unp ublished d ata).
Although these studies indica te that
both the height and cover of vegeta tion
influence vole's habita t selection, it is not
yet clear w hich one is the m ost res tricti ve
fac tor. A study of the rela tionsh ip between
vegeta tion condition and the density of
voles was ca rried in 1998. We ch ose 18 sites
at ran dom in Taip us Q i, Inner Mongolia ,
where we investiga ted th e d ensity of
Brandt's vole and measured cover and
height of vegetation at each site (Table 6).
There was a significant negative correlation
between the height of vegetation and the
denSity of Brandt's vole (I' = -0.636, P <
0.01), but there was no significant
relationship between cover and the density
of voles (I' = -0.128, P > 0.05). These data
suggest that there is an inverse relationship
between the density of voles and the height
of vegeta tion in areas where vegetation
cover is in the ran ge 28-75% .
Fenced areas we re used to limit access by
livestock in sum mer to allow recovery of
pasture and the conservation of forage for
win ter grazing. Beca use the enclosures
should imp rove the condition of vegetation,
they should i.nfluence the denSity of Brandt's
vole. This h ypothesis was tested usin g five
large enclosures in the Xilin Geluo and Zhe
Limo region in 1987, with each enclosure
m ore than 130 ha (Zhong et a1. 1992). Table 7
in d icates the con dition of the vegetation and
the density of voles in side and outside the
exclosures d uring the course of the
experiment.
The height of vegetation is significantly,
negatively correlated with the density of
Brandt's vole (I' = -0.708, 11 = 9, P < 0.01) but
there is no obvious relationship between
plant cover and the density of voles

205
Ecologically-based Rodent Management

Table 6.
Comparison of the density of Brandt's vole and the vegetation conditions in its habitat. Data are given as
the mean (X) and the standard error (SE).

Height of vegetation (cm) Cover (%) Density of Brandt's vole

(X:t SE) (X:t SE) (Animals/ha)
(X:t SE)
12.92 0 .8 9 36.40 2 .11 308.31 21.36
13.82 0 .95 52.00 2.55 339.36 23.31
10.95 0.42 40 .00 2.24 396.85 30.14
11.38 0.61 53.00 1.22 419.20 13.29
16.52 0 .76 34.40 2.20 193.80 12 .87
19.92 1.29 70.00 3 .54 92.64 10.86
1 1.99 0.64 29 .20 1.48 233.60 21 .37
1 2.84 0 .69 6 1.00 2.33 528.88 24.22
68.84 1.91 74.60 3.26 2.40 1.17
8 .9 2 0 .58 36.00 2.45 306.38 12.46
3.62 0 .29 46 .00 1.87 464.56 47 .00
7.32 0 .47 30.00 1.58 4 0 7.36 26.37
4 .68 0.70 41.00 2 .45 577 .60 23 .90
11.68 0 .37 28.20 1.11 220.65 21. 79
7 .72 0.68 61.00 1.87 191.04 17.93
9.60 0 .28 29.20 1.52 239.4 2 16.93
7.53 0 .36 54.50 1.89 585.44 38.38
11.48 1.10 66.00 3.67 416.20 35.04

(r = -0.304, P > 0.05). This res ult is supported
by the data in Table 8 which show the rate of
change in the height of herbs (e.g. A chinense)
and the percentage chan ges in the density of
voles.
There is a significant coefficient of
correlation between these two variables of-
0.917 indicating a very significant correlative
relationship. The conclusion is that the
density and height of vegetation in the habitat
of Brandt's vole strongly influences its social
behaviour; i.e. an increased height of
vegetation d ecreases the fitness of Brandt's
vole so that the pop ulation density of the vole
decreases sharply. This suggests that fencing
and pastme managemen t could be used in
some areas to reduce problems caused by
Brandt's vole. As a guide, the density of voles
was low where the height of vegetation was
>190 mm but high where vegetation height
was 30-130 mm.
The suppressive effect of vegetation
growth on the abund ance of Brandt's vole
takes effect from spring to autumn. H owever
the most important factor influencing the
survival of voles in winter is their store of
food (Zhong 1996). As Afrigida is the largest
component (44-71 %) of stored food (Zhou et
al. 1988), pasture management such as
fencing that diminishes production of
A f rigida, will also reduce the over-winter
survival of voles.

206
 Table 7.
The density of Brandt's vole and the condition of vegetation inside and outside livestock enclosures (Zhong et a!. 1992). Dat a are given as the
mean t he standard error, n is the sample size, and I and 0 indicate measurements inside and outside the exclosures, respectively.

Exclosure 11 Cover of vegetation (%) 11 Height of vegetation (cm ) 11 Densit y (voles/ha)

n 0 n 0 n 0
A June 5 36.40 2.11 36.00 2.45 5 12.92 0.89 8 .92 0.58 6 308 .31 21.36 306.38 12.46
r9'I
September 5 52.00 2.55 46.00 1.87 5 13.82 0.95 3.62 0.29 5 339.36 23.31 464.56 47 .00 n
Cl
B June 5 40.00 2.24 30.00 1.58 5 10.95 0.42 7 .32 0.47 6 396.85 30.14 407.36 26.37 Cl
IrQ
(;.
September 5 53.00 1.22 41.00 2.45 5 11.38 0 .61 4 .68 0.70 5 (419.20 13.29 57 7.60 23 .90 Cl!
C June 5 34.40 2 .20 28.20 1.11 5 16.52 0 .76 11.68 0 .37 6 1 93 .80 12.87 220 .65 21.79
=:
Cl!
September 5 70.00 3.54 61.00 1.87 5 19.92 1.29 7.72 0.68 5 92.64 10.86 191.04 1 7.93 ::::I
Cl!
D June 10 29.20 1.48 29.20 1.52 10 11.99 0 .64 9 .60 0.28 10 2 33.60 21.37 239.42 16.93 IrQ
ell
September 10 61.00 2.33 54.50 1.89 10 12.84 0 .69 7.53 0 .36 10 528.88 24.22 585.44 38.38 :3
...
ell
::::I
E August 5 74.60 3 .26 66.00 3.67 5 68 .84 1.91 11.48 1.10 4 2.40 1.17 416 .20 35.04

-
Cl

...
Cl:!
Cl!
::::I
Cl.
"t
III
<
2.
ell
::::I
(")
N ::::I'
Cl ::::I
..... Cl!
Ecologically-based Rodent Management
Table 8.
The proportional change in the height of herbs and percentage decease of t he vole population density in the
exclosure in comparison to outside. From Zhong et al. 1992. Calculated from Table 7.
Exclosure A B
Ti me June Sept ember June September
Times change in height of herbs
1.4 5 3.82 1.50
-0 .6 3 26 .95 2 .58
ECOLOGICAL MANAGEMENT OF
BRANDT'S VOLE
In Inner Mongo lia, local herdsmen fence off
pasture in June to increase herbage for
harvesting in autumn or for grazing by
livestock in winter. However damage is still
caused by Brandt's vole in some areas
because fencing is not properly maintained
to exclude livestock. High rainfall and warm
tempera tures in May can promote early
pasture grow th th at may be useful in
suppressing vole populations. The effec t of
removing livestock from pastures in the
middle of May, rather than later in Jt.me, was
examined using a series of experimenta l
exclosures in Taipus Q i, from 1986-1990
(Zhong et al. 1991). Exclosure I was fenced
from the middle of May, the control
exclosure was fe nced fro m JW1e (according
to the local custom) and a third area (Ill)
remained w1fenced throughout the
experiment. Both fenced areas were opened
to grazing from 3 Sep tember through to May
(exclosure I) and June (exclosure 11) of the
following year. Sheep were stocked a t 1.62
animals / ha during the grazing period .
There were no diffe rences between
exclosures I, 11 and III in the vegeta tion and
the density of Brandt's vole p rior to th e
experiment.
208
C 0
June September June September August
2.43 1 .41 2 .58 1 .25 1.71 6 .00
% change in vole den sity
27.42 1 2 .17 51.51 2 .43 9.66 99.42
In 1987 rainfall was 289.59 mm, 30%
below average. In au tumn, the height, cover
and biom ass of vegetation were higher in
exclosure 11 than in exclosure I , and in the
period from spring to autumn the rate of
increase of the vole pop ulations was 0. 62 in
exclosure I compared to 1.99 in exclosure 11
and 2.11 in exclosure III (Table 9). The
autwru1 denSity of voles was 4.6 times
higher in exclosure 11 than I. The experiment
demonstra ted tha t exclusion of livestock
from pasture half a month earlier in mid -
May increased biomass of grass by 40% and
decreased the density of Brandt's vole by
78%. Comparing ex closure II and Ill, the
increased grazing on the latter site resulted
in the height, cover and biomass of
vegetation being reduced by 70%, 12% and
55%, respectively, but there was no
sign ificant difference in the density of
Brandt's vole. There appea red to be a
significant threshold between 120 and 160
lrun in the sup pressive effect of vegetation
heigh t on the abundance of voles, and the
traditional time of fencing in June was too
late to allow vegetation to exceed this
threshold.
The study w as continued in 1988 and
1989 (Table 10). The year of 1988 was a wet
year, with 339.8 mm of rainfall from May to
August, 17.33% more than normal. The
 Ecological Management of Brandt's Vole in China

height, cover an d biomass in the exclosme I
w ere 95%, 10% an d 80% higher tha n that
outside exclosme, respecti vely. The density
of Brandt's vole insid e the exclosure was
2.24 times lower inside than outside the
exclosure; the difference was significant
(p < 0.01).
In 1989, 158.7 m m rain fell d uring the
period from May to August, which was 45%
less than average and the height, cover and
biomass of vegetation was 35%, 22% and
139% higher in the exclosure than in the
grazing area , respectively. Because of the
aridity, the height of the main herb layer was
about 144.2 mm, less th an the critical value of
160 mm. The density of Brandt's vole in the
exclosure was 16.85 anima ls/ ha, not
significantly different from the density of
33.74 animals/ha in the grazed area (t = 1.68,
p> 0.05), and both were under the threshold
for causin g damage. Therefore, although no
damage occmred in 1989, the data fro m the
exclosure experiment in that year and from
other years with different climatic conditions
demonstrated the simultaneous effects of
producing more g rass and suppressing
Brandt's vole.
After the ecological measmements, many
colonies of Brandt's vole disappeared, their
complex burrow systems were abandoned,
and herbs grew quickly around these areas
fo rming patches different from the
smrolmding area . These pa tches, or
'mosaics', were classified as two types:
mosaic I where a burrow system was
abandoned dming Mayor June in the current
year and mosaic IT where the bmrow system
was abandoned in the previous year. Herbs
growing in m osa ics were big and tall (see
Table 11). Vegetation cover in mosaics I and
1I was 1.1 and 1.07 times greater than in the
non-burrow areas, respectively, and the

Table 9 .
The impact of different grazing t reatments on vegetation indices and the density of the Brandt's vole
population. The experiment was conducted from spring t o autumn in 1987 using t hree experimental areas:
( I) livestock excluded from mid-May, (11) livestock excluded from June, and (Il l) free access to livestock
(no fencing). Data are given as the mean the standard error and n is the sample size. The rate of increase
of the vole population over the period from spring to autumn is calculated as r = In (N t+ 1 / Nt)where N t is the
density at time t.

.. c 'ii Vegetation in autumn Density of Brandt 's vole/ha

III

.. .
III
III
E :.
l'IS
l'IS
2!
l'IS III Co)

I-
2! < Height (cm) Cover (%) Above-ground
biomass (g/m 2)
Spring Autumn
-..
.5
Cl
III
l'IS
a::
93 16.96 0 .43 73 .50 1.07 342.44 16.42 63.00 5.48 116.95 36.32 0.62
n = 10 n -= 10 n -= 10 n-= 10 n -= 10
11 99 12.84 0 .69 61.00 2 .33 245 .28 12 .12 73.41 8.41 535.50 25.43 1.99
n -= 10 n -= 10 n = 10 n-=8 n-=8
III 80 7 .53 0.36 54.50 1.89 157 .90 6.00 69.39 12.82 571.10 38.32 2.11
n -= 10 n-= 10 n -= 10 n-=8 n=8

209
 m
....
~

Cl
('I
o
0'
IJQ
Table 10. n'
11/
The impact of different grazing treatments on vegetation indices and the density of Brandt's vole population during 1998 and 1989. Data are
given as the mean the standard error. n is the sample size and the rate of increase is calculated as in Table 9. 9'
er
11/
III
Year Study Vegetation in autumn 11 Density of Brandt's vole/ha ID
Cl.
site ::u
n Height (cm) Cover (%) Biomass (gjm 2 ) n Spring Autumn Rate of increase o

-
Cl.
ID
1988 exclosure 30 27.59 0 .77 87 .70 1.37 785.87 24.31 10 77.41 4 .40 38.04 9 .80 - 0 .71 ::::I
=
unfenced 15 14.12 0 .64 80 .13 2.76 436.56 26 .22 10 99 .9 7 5 .15 123.42 12.46 0.21 =11/:
::::I
1989 exclosure 30 14.42 0.62 58.67 1.24 129.23 10.08 10 9 .02 2 .08 16.85 3.97 0 .62 11/
IJQ
ID
unfenced 15 10.64 0.44 47.67 1.61 54.00 5.74 10 14.00 1.64 33.74 9.25 0 .88
:I

Table 11.
-
ID
::::I

Vegetation indices in mosaics I and 11 (see text) and in an area with no vole burrows. Data are given as the mean the standard error and n is the
sample size.

Type 11 Cover of 11 Chenopodium aibum Aneurolepidium chinense Biomass of other

vegetation plants
(%) Height (cm) Biomass (gjm 2 ) Height (cm) Biomass (gjm 2 )

Mosaic I 97 .00 1.22 85.52 1.71 910.00 85.58 45 .72 2 .7 6 350.00 41 .76 530.00 53.76
(n = 5)
Mosaic 11 94.00 1.00 49.40 2.77 94.00 33.65 46 .04 1.04 970.00 87 .31 280.20 89.96
(n = 5 ) L-- ---''--

No burrows 87.70 1.37 0 '- '-I~ -_. 0 _. 2 7 .59 0 .77 121.87 11.72 664.00 26. 4 5
(n = 30)
 Ecological Management of Brandt's Vole in China
biomass of A. chinense in mosaics I and 11 was
2.87 times and 7.96 times that of the biomass
in the non-burrow area, respectively.
Productivity of a mosaic could exceed the
productivity of the surrounding area in three
months, with a considerable increase in the
biomass of fine grazing grass such as
A. chinense in the mosaics.
The ecological management of the
vegetation was economically efficient with
an increase of 530 kg/ha of dry matter
produced in 1987 and an investment:income
ratio of 1:7.1988 was a wet year, with 1,753
kg/ha more dry herbs harvested, and the
investment:income ratio was 1:8.8 (Zhong et
a1. 1991). By comparison, 1989 was the worst
drought year in 29 years. Productivity was
increased by 123 kg/ha at harvest that year
and the investment:income ratio equaled
1:2.7. Over the three years from 1987-89, the
average investment:income ratio was 1:7. In
summary, good ecological management not
only stopped the damage caused by Brandt's
vole but also enhanced the productivity of
the grassland.
DISCUSSION
Heavy grazing pressure by livestock causes
degradation of grassland in Inner Mongolia,
with different rodent pests occurring at each
Heavy grazing by livestock
Figure 2.
Schematic diagram showing the Interactions of livestock and rodents In grasslands. Overgrazlng by
livestock can degrade grasslands to a level that will be maintained Indefinitely by rodent pests.
different degenerative stage. At the same
time, the rodents' devouring and digging
activities aggravate the degradation of the
grassland, i.e. there is positive feedback
between damage caused by rodents and the
degeneration of grassland that can lead
ultimately to desertification (Figure 2).
Coordinated management of livestock,
grasses and rodents is required to break this
vicious cycle.
The main aims of an ecological approach
to controlling rodents in grasslands are
economic benefits, minimal or no use of
rodenticides to avoid chemical
contamination of the environment, and a
long-term solution brought about by
decreasing the carrying capacity of rodents.
To achieve this, we have studied the
ecological management of both M. brandti
and O. daurica Simultaneously (Zhong 1996)
from 1991 to 1996.
Detailed ecological studies of pest species
are needed before ecological management is
applied: the life history of a pest species
must be understood for identifying weak
links. For example, stored food is the most
important fador influencing the survival of
rodents during the long and cold winters in
the grassland of Inner Mongolia. Decreasing
the available storage food for rodents could
... D'O"'dai" of ,,,,,.tand I
Rodern pe," -----.-J
211
Ecologically-based Rodent Management

remarkably suppress the growth of rodent
populations (Zhong 1996).
We used exclosures to adjust the pressure
and duration of grazing by livestock. This
resulted in the recovery of degraded
grassland and a decline of the density of
Brandt's vole. This result could be achieved
in other ways. The cost of building fences
could be saved by moving herds of livestock
alternately between different areas based on
well designed grazing plans, effectively
employing 'formless exclosures'. As well,
irrigation and the use of fertiliser are
effective means of accelerating the recovery
of degraded grassland (Zhong 1996). Rapid
recovery of vegetation can quickly suppress
Brandt's vole, preventing future
degradation with economic savings in the
long term.
The need for livestock to have access to
water results in uneven grazing pressure
and degradation of grassland near rivers.
Grazing pressure near rivers can be red uced
by digging wells to change the pattern of
herding livestock and make full use of grass
resources distant from rivers.
Dicotyledons are the main component of
rodents' stored food for winter. Fencing to
exclude livestock can decrease the biomass
and proportion of dicotyledons in the plant
community. In addition, we sowed
monocotyledon seeds to increase the
proportion of fine-grazing grass in fenced
areas, with encouraging results (Zhong
1996). Another practice is to plough the
grassland. Roots of A. chinense and S. grandis
are extensive and easily cut to promote
tillering. The result is an increase in the
biomass and proportion of these fine-
grazing grasses in the plant community
(Zhong 1996).
Future research will be directed towards
the strategy of ecological management at the
landscape level. Myllymiiki (1979) reviewed
the landscape characteristics that could
create conditions favourable for rodent
plagues, and others that tend to prevent
outbreaks. Some references suggest that
characteristic changes of landscape would
influence movement, competitive
interaction, predation etc. of mammals in
this habitat (Lidicker 1995).
Our approach to ecological management
is not simply habitat modification. The main
features of habitat modification are removal
of basic life needs (food and water) from
rodents and rodent proofing (Fitzwater
1988) using techniques such as clearing
weeds quickly after crops are harvested,
spraying 2,4-0 herbicides to reduce pocket
gopher's (Tho1rloys talpoides) favourite food
and hence decrease their activity (Keith et aL
1959), planting buffer crops which pests
prefer and setting up physical barriers.
These techniques can eliminate rodent pests
but may have deleterious effects on other
vertebrate species that share the same
habitat (Howard 1988), and they are usually
expensive. Ecological management comes
from systematic view, based on ecological
studies, using natural forces to target weak
links in the life history of a pest species. The
aim is to take careful consideration of
relationships with other species and the
environment, while using existing
equipment to save money and exploring
new ways to enhance production.
Although we have achieved effective
ways to manage Brandt's vole and the
Oaurian pika Simultaneously (Zhong 1996),
many improvements are still demanded.

212
 Ecological Management of Brandt's Vole in China
Future study will continue to apply
ecological principles to pest management.
ACKNOWLEDGMENTS
We are very grateful to Dr Roger Pech for his
tremendous help in improving the English
writing of this paper and for his valuable
comments for revising the original
manuscript. This work is supported by the
national rodent key project (96-005-01-06)
and the Chinese Academy of Sciences key
projects (KZ951-Bl-106, KZ952-S1-107 and
STZ-1-05).
REFERENCES
Barnett, S.A. 1988. Ecology and economics of
rodent pest the need for
research. In: Prakash, 1., ed., Rodent pest
management. Boca Ratan, Florida, CRC
Press, 459-464.
Dong, W.H., Hou, X.x., Zhang, P.L., Zhou, Y.L.,
Yang, Y.P. and Xue X.P. 1991. 5tudy on
characteristics of population of Brandt's vole
after rodent control. Acta Ecologica Sinica, 11,
274-279 (in Chinese).
Fitzwater, W.D. 1988. Non-lethal methods in
rodent control. In: Prakash, 1., ed., Rodent
pest management. Boca Raton, Florida, CRC
Press, 421-426.
Hou, X.x., Dong, WH., Zhou, Y.L., Yang, Y.P.,
Zhang, Y.X. and Xue, X.P. 1993. A study on
succession of rodent community in grassland
after chemical control. Chinese Journal of
Vector Biology and Control, 4, 271-274 (in
Chinese).
Howard, W. E. 1988. Areas of further research.
In: Prakash, I., ed., Rodent pest management.
Boca Raton, Florida, CRC Press, 451-458.
Keith, J.O., Hansen, R, M. and Ward, A. L. 1959.
Effect of 2,4-0 on abundance and foods of
pocket gophers. Journal of Wildlife Manage-
ment,23,137-145.
Udicker, W.Z. 1995. Landscape approaches in
mammalian ecology and conservation.
Minneapolis and London, University of
Minnesota Press.
Uu, S.R. 1979. Basic study on relation between
Brandt' 5 vole and vegetation in Xinlin Guole
region, Inner Mongolia. Chinese Grassland,
2,27-31 (in Chinese).
Myllymaki, A. 1979. Importance of small
mammals as pests in agriculture and stored
products. In: 5toddart, D.M., ed., Ecology of
small mammals. London, Chapman and Hall,
239-279.
Ren,J.Z.,Ge, W.H. and Zhang, Z.H. 1989. The
future of production of grassland. In: Grass-
land research group of the Chinese Academy
of Sciences and the Agricultural Academy of
Sciences, ed., Grassland science and develop-
ment of China. Beijing, Science Press, 6-9 (in
Chinese).
Wang, G.M,Zhou, Q.Q., Zhong, W.Q. and Wang,
G.H. 1992. Food habits of Brandt's vole
(Microtus brandti). Acta Theriologica Sinica, 2,
57-64 (in Chinese).
Wang, M.J., Zhong, W.Q., Wang, G.M. and
Wang, G.H.1997. Cooperative of
plant-rodent communities in steppe on
grazing gradient. Research on Grassland
Ecosystem, 5,32-42 (in Chinese).
Wang, M.J., Zhong, W.Q., Wan, X.R. and Wang,
G.H. 1998. Habitat selection on dispersal of
Daurian pika (Ochotona daurica). Acta Zoolog-
ica Sinica, 44, 398-405 (in Chinese).
Xia, W.P. and Zhong, W.Q. 1966. The successions
and interactions of the rodent and plant
communities of abandoned fields in desert
steppe at Changan-Aobao, Inner Mongolia.
Acta Zoologica Sinica, 18, 199-208 (in
Chinese).
Yang,H.F.,Zhou,Q.Q., Wang,S.Q.,Zhong, W.Q.
and Liu, B.Q. 1979. Investigations on popula-
tion self-regulation in Brandl's vole-the
relationships between population density,
adrenal weight and gonadal weight. Acta
Zoologica Sinica,25,154-168 (in Chinese).
Zhang, J. and Zhong, W.Q. 1981. On the colonial
structure of Brandt's vole in burrow units.
Acta Theriologica Sinica, 1,51-56 (in
Chinese).
213
Ecologically-based Rodent Management
Zhong, W.Q. 1996. Strategy of ecological
management on rodent pest of grassland and
research on pollution-free technology.
Unpublished report to Chinese Academy of
Science, 1-9.
Zhong, W.Q., Zhou, Q.Q. and Sun, CL. 1983.
Study on the food consumption of Daurian
pika. Acta Ecologica Sinica, 3, 269-276 (in
Chinese).
Zhong, W.Q., Zhou, Q.Q. and Sun, CL. 1985a.
The basic characteristics of the rodent pests
on the pasture in Inner Mongolia and the
ecological strategies of control. Acta Therio-
logica Sinica,5, 241-249 (in Chinese).
Zhong, W.Q., Zhou, Q.Q. and Sun, CL. 1985b.
The vegetation and habitat selection by the
Brandt's vole (Microtus bralldti) in Inner
Mongolia steppe. Research on Grassland
Ecosystem,I,147-152.
Zhong, W.Q., Zhou, Q.Q., Sun, CL., Wang,GH.,
Zhou, P.Y., Liu, W.Z. and Jia, Y.P.1991. The
design for the ecological management of
Brandt's vole pest and its application. Acta
Theriologica Sinica, 11,204-212 (in Chinese).
214
Zhong, W.Q.,Zhou,Q.Q., Wang,G.H., Sun, CL.,
Zhou,P.Y. and Liu, W.Z. 1992. Ecological
effects of the measures for raising herbage in
enclosure on Srandt's vole population.
Research on Grassland Ecosystem, 4,199-203
(in Chinese). -
Zhou, Q.Q., Zhong, W.Q. and Sun, CL. 1982.
Study on species diversity of rodent commu-
nities in Baiyinxile typical steppe, Inner
Mongolia. Acta Theriologica Sinica, 2, 90-94
(in Chinese).
Zhou, Q.Q., Zhong, W.Q. and Wang, G.M. 1988.
Food store of Brandi's vole (Microtus brandti)
and its ecological importance. Symposium of
Asian-Pacific Mammalogy (Abstract), 77 (in
Chinese).
Zhou, Q.Q., Zhong, W.Q. and Wang, G.H. 1992.
Density factor in the regulation of Brandi's
vole population. Acta Theriologica Sinica, 12,
49-56 (in Chinese).
10. Biological Control of Rodents-
the Case for Fertility Control Using
Immunocontraception

Usa K. Chambers, Malcolm A. Lawson and Lyn A. Hinds

Abstract

Managing rodent pests on a broad scale using lethal methods is not an appropriate
long-term strategy given th e ir extraordinary breeding capacity and high mobility.
Moreover, environme ntal , an ima l welfare an d ethical concerns regarding the use of
poisons and trapping has decrease d th e acceptance of mortality methods in recent
times. Another reason for avoiding lethality is that it may promote a strong selective
pressure for resistance to the lethal agent, be it a disease or a chemical. The
additi on of fertility control , specifically immunocontraception, to the armoury
currently available to control ro dent pests, is discussed in this chapter. Fertility
control aims to re duce a specific population size by reducing the number of young
produced and recruited into the population.

Existing fertility control techniques (e.g. steroids, synthetic hormones) are flawed
because they require repeated administration to maintain the sterility level of th e
population, they have undesirable physiological and behavioural side effects and
they are not specific to the target animal . Delivery of these sterilising agents is
logistically difficult, time-consuming and expensive and therefore they are not
suitable for controlling field populations of rodent pests that are often widespread
and cryptic in their habits. An immunocontraceptive vaccine , either distributed in bait
or disseminated in a species-specific viral vector, is a new tool that could be used to
reduce the productivity of pest populations. The various components of this
approach and 'proof of concept' laboratory experiments conducted in house mice in
Australia are described. It must be recognised that to critically evaluate the efficacy
of a viral-vectored immunocontraceptive agent requires a multi-disciplinary approach
with a strong ecological and epidemiological focus . Only then can the impact of this
control technique be assessed at the population level.

Keywords

Immunocontraception, biological control , rodents, viral-vectored, fertility control ,

genetically manipulated organism , murine cytomegalovirus , zona pellucida ,
ectromelia virus , reproductive antigens

215
Ecologically-based Rodent Management
INTRODUCTION
ODENTS HAVE gained the
R reputation as one of the most
persistent and ubiquitous
vertebrate pests affecting human
populations. They cause economic problems
because of the damage they inflict in
agricultural systems (e.g. Caughley et al.
1994), environmental problems due to the
chemicals used for their control (e.g.
Saunders and Cooper 1981; Singleton and
Redhead 1989), social problems associated
with their close proximity to human
habitation (e.g. Beckmann 1988) and health
problems as carriers of zoonoses (Childs et
al. 1994; Gratz 1994; also see Mills, Chapter
6). Of increasing concern is the impact of
introduced rodents on the conservation of
native wildlife (e.g. Wace 1986; Moors et al.
1992; Key et al. 1994; also see Dickman,
Chapter 5). Rats have been reported as the
major pest in rice crops in Southeast Asia
(Geddes 1992; Singleton and Petch 1994),
and cause significant problems in Africa
(Leirs et al. 1997; also see Makundi et al.,
Chapter 22), Australia (Singleton and
Redhead 1989; Caughley et al. 1994), China
(see Zhang et al., Chapter 12) and elsewhere
(Prakash 1988a; also see Buckle, Chapter 7).
Many species that become pests do so
because of their reproductive potential. They
often have several large litters in each
breeding period, show early onset of sexual
maturity and have a short life expectancy
(Tyndale-Biscoe 1994). Rodent pests
typically show these life history traits-for
example, one breeding pair of house mice
(Mus domesticus) is theoretically capable of
producing over 600 offspring in six months
216
and the average life expectancy in a field
population is four to six months (Singleton
1989). A post-partum oestrus allows females
to produce a litter every 21 days
(Whittingham and Wood 1983). Therefore,
curbing the reproductive potential of
rodents may be a more appropriate control
tactic than increasing their mortality
(Singleton 1994; also see Krebs, Chapter 2).
Increasing community interest in
environmental and animal welfare issues
associated with conventional pest control
techniques, such as poisoning and trapping,
has focused interest on developing non-
lethal, non-toxic alternatives (Bomford
1990). One such strategy is to focus on
reducing reproduction, rather than
increasing the mortality of the pest species.
This is commonly referred to as fertility
control.
In this chapter, we will examine why
fertility control is theoretically superior in
many respects to conventional methods of
rodent control that rely on increasing
mortality. We discuss the various methods
of fertility control currently available for
red ucing rodent pest populations and then
focus on immunocontraception, a relatively
new approach to the problem of controlling
wild pest populations.
Effective pest control requires a thorough
understanding of the biology and
population dynamics of the pest species
(Howard 1967). Specifically, for effective
fertility control, a reduction is required not
only in the reproductive potential of the
species, but also in the final population size
(Bomford 1990; Bomford and O'Brien 1997)
and in the potential damage inflicted
(Bomford and O'Brien 1997). Thus, we
emphasise in this chapter the importance of

an ecological framework for considering the
use of immunocontracep tion and ferti lity
control in general. We also stress that
although the general princip le of fertility
control is similar for all mammals, the
particular approach may be different for
each pest and needs to consider the
ecological and behavioural features of each
species (Cowan and Tyndale-Biscoe 1997).
CONVENTIONAL METHODS OF CONTROL
Control of rodent pests currently relies on
increasing their mortality. For large-scale
control in agricultural systems, this typically
involves the use of rodenticides such as
anticoagulants and acute poisons (see
Meehan 1984; Prakash 1988a for reviews). In
small-scale domestic control, both
rodenticides and traps are often employed.
These methods are easily applied by farmers
or householders and there is usually an
immediate effect on population size and
Table 1.
Advantages and disadvantages of rodenticides for the control of rodent pests (after Singleton and Redhead
1989; Bomford 1990; Chambers et al.1997 unless indicated otherwise) .
Advantages
Immediate effect on population numbers and
damage
Permanent control method; re moves animals
fo r t he whole of their expected life span
Cost effective for short-term control and
reduct ion in damage caused
Biological Control of Rodents
da mage caused by rodents (Table 1)
(Bomford 1990).
However, these conventional con trol
methods are obviously not always effective
in the long term (Table 1) as rodent pests are
still a major problem. This may be because
lethal methods are often used inefficiently as
an ad hoc control approach when rodent
populations have already reached high
densities. Another major factor is the high
reproductive capacity of the pest and the
ability to re-invade treated areas from
surrounding untreated sites (e.g. Emlen et al.
1948; Twigg et al. 1991). Also, because these
methods are often labour intensive, they are
rarely applied in areas with inaccessible
terrain. The expense of poison-baiting large
areas long-term can also be prohibitive,
particularly if damage to crops is not
reduced. For example, during a mouse
plague in southern Australia in 1993, the cost
of one bait application to 46,000 ha was
approximately A$319,500 (Kearns 1993).
Disadvantages
Development of bait shyness if sublethal dose ingested
(prakash 1988b)
Non-target deaths due to primary and secondary
poisoning-not species-specific
Inhumane
May pollute the environment with pOison residues
Potenti al re-invasion of treated areas by rodents from
neighbouring untreated sites
Ineffect ive over the long-term for highly fecund or mobile
species (e.g. rodents) (Caugh ley 1977, 1985)
Expensive to apply over large areas long-term
217
Ecologically-based Rodent Management
Fertility control has the potential to
overcome some of the inadequacies of
conventional control techniques and a
naturally disseminating immuno-
contraceptive would reduce the need for
manual delivery of the control agent.
FERTILITY CONTROL AS AN ALTERNATIVE
TO CONVENTIONAL METHODS
Fertility control has been suggested as a
more appropriate control strategy than
enhancing mortality under the following
circumstances:
~ for species with high fecundity (Caughley
et al. 1992; Tyndale-Biscoe 1994);
~ for species with high natural mortality
rates and a rapid population turnover
(Stenseth 1981; Bomford 1990; Hone 1992;
Barlow 1994; Barlow et a1. 1997);
~ when a more humane method of
population control is desired (Marsh and
Howard 1973; Hutchins et al. 1982;
Hutchins and Wemmer 1987);
~ when the effects of sterilisation exceed any
increases in juvenile or adult survival due
to a lowering of birth rates (Sinclair 1997);
and
~ for preventing or reducing population
growth after some other technique has
reduced numbers, particularly in long-
lived (Bomford 1990; Barlow 1994).
The last point emphasises one of the main
differences between these two control
strategies-increasing mortality has an
immediate effect on population numbers
and damage, while reducing fertility has a
delayed response until natural mortality
218
reduces population size (Barlow et a1. 1997).
If sterile individuals inflict as much damage
as fertile individuals, sterility is of little
practical value to agriculturalists. Thus, in
some instances, fertility control may need to
be used in conjunction with another control
method.
It has been suggested that the presence of
a given number of sterile individuals in the
population exerts a greater, more sustained
biocontrol pressure than if the same number
of animals were simply removed from the
population (Howard 1967). Sterile
individuals fail to contribute to the next
generation as well as competing for space,
food and social order. 111is in turn reduces the
reproductive success and survival of fertile
individuals and continues the suppression of
breeding in subordinates if dominants are
sterilised (Howard 1967). Therefore, fertility
control could be used as a long-term strategy
for slowing a population'S growth rate and
hence maintaining numbers at this lower
level. Modelling the relative impact of culling
versus sterilisation on populations with
density-independent or exponential growth
rates supports this argument (Bomford 1990).
However, for populations with de~<;ity
dependent or logistic growth rates, the
relative efficiency of sterilisation will depend
on the nature of the density-dependent
regulation. Populations with density-
dependent mortality appear to be reduced by
sterilisation more quickly than those with
density-dependent recruitment (Barlow et al.
1997).
GENERAL AIMS OF FERTILITY CONTROL
Fertility control aims to reduce population
size by reducing the number of young

produced and recruited into the population.
This can be achieved by temporary,
permanent or partial sterilisation.
A successful fertility control method
therefore needs to (after Bomford 1990;
Bomford and O'Brien 1997):
~ cause temporary or permanent sterility
leading to reduced recruitment in the
population;
~ be deliverable in a way that allows an
adequate proportion of the target
population to be treated, particularly for
widespread and abundant species in areas
with poor access;
~ reduce the target population sufficiently to
reduce damage caused by the pest species
to an acceptable level (Braysher 1993);
~ produce minimal side effects to the target
species (e.g. behavioural changes,
interference with social structure);
~ be target-specific;
~ be environmentally benign (Marsh and
Howard 1973); and
~ be cost effective compared with
conventional methods of control.
In the following section, we explore the
various options available for fertility control
of rodents and examine how well each of
these satisfy the criteria for a suitable fertility
control agent for controlling wild
populations.
OPTIONS FOR FERTILITY CONTROL -
EXISTING TECHNOLOGIES
Many techniques have been developed for
managing or controlling the fertility of
individual animals in captivity or in
Biological Control of Rodents
confined areas that are not subject to
immigration. These methods include
surgical sterilisation or castration, use of
chemical sterilants, agonists that block the
function of natural hormones, and inhibitors
of lactation (Table 2). Most of these
approaches are expensive and time-
consuming to apply, often have undesirable
side effects (e.g. chemosterilants can induce
gastrointestinal problems, abnormal growth
and dysfunction of the gonads), and affect
non-target species. Many disrupt gonadal
function and sexual behaviour. Further,
their applicability and effectiveness for free-
ranging populations is low due to the
difficulties of delivering the sterilising agent
on a broad scale and sustaining the
inhibition of reproduction.
IMMUNOCONTRACEPTION FOR
CONTROLLING PEST POPULATIONS-
THE CONCEPT
Immunocontraception uses the body's
immune system to induce immune
responses (circulating antibodies or cellular
immune effector cells) against reproductive
cells or proteins essential to successful
gametogenesis, fertilisation or implantation,
leading to infertility. The feasibility of
immunocontraception was directly
demonstrated when Baskin (1932) injected
women with human sperm and no
conceptions occurred during the one-year
follow-up period.
Ideally, the immunocontraceptive
prevents pregnancy but does not disrupt
endocrine function (i.e. renders the animal
infertile but not impotent) and therefore
reprod uctive I social behaviour is unaffected.
Animals continue to occupy territory,
219
N
...,n
N
Q 2-
o
Table 2. (IQ

Summary of potential techniques for fertility control of pest populations and assessment of their relevance for managing rodents . Sources: n'
Dj

Singleton and Spratt 1986; Spratt and Singleton 1986; Marsh 1988; Vickery et al. 1989; Bomford 1990; Sankai et al. 1991; Gao and Short 1993;
~
Tyndale-Biscoe 1994,1997a; Marks et al. 1996; Becker and Katz 1997; Jochle 1997. r:::t'
Dj
, (11
~
Technique for fertility Major advantages Major disadvantages Efficacy for rodent pest populations c.
control ~
Current Future o
c.
Surgery Castration and ovariectomy
Permanent
One treatment only, therefore costs
Expensive and invasive
Leads to behavioural changes
Very low Very low
-==
~

==
Dj

Dj
(IQ
recouped over time ~
3
~sectomy

(Permanent
and tubal ligation

IUiPensive and invasive

~
I
Very low Very low

-=
~

~
One treatment only, therefore costs [' mpractical for high density field
recouped over time populations

'I
" I ~ , ~ ~ J ~ _ _ ., _ _ _ _" _ , _ ._ _ ._ ~ __

Disease For example, Capillaria hepatica Field tested, but Low

Natural parasite of rodents Insufficient level of persistence in ineffective at low
Australian situation densities

Complex life cycle

Chemicals IAgonist and antagonists of GnRH a (disrupt natural hormone functions)
Both sexes infertile [Costly to administer
jrcow, b", ""te<ted
Low, but untested

Not permanent, may only reach a

I proportion of the population
I
[Side effects (dose dependent)
j
, INot appropriate for promiscuous species
~onadotrophin releasing hormone
= immunostimulatory complexes,
- ------ - ---------- -- - - -------- - _.. -
 Table 2. (Cont'd)
Summary of potential techniques for fertility control of pest populations and assessment of their relevance for managing rodents. Sources:
Singleton and Spratt 1986; Spratt and Singleton 1986; Marsh 1988; Vickery et a!. 1989; Bomford 1990; Sankai et a!. 1991; Gao and Short 1993;
Tyndale-Biscoe 1994, 1997a; Marks et al. 1996; Becker and Katz 1997; Jochle 1997.

Technique for fertility Major advantages Major disadvantages Efficacy for rodent pest populations
control
Current Future

Chemicals Synthetic steroids, anti-steroids, anti-steroid receptor Low, but untested Low, but untested
(e.g. Diethylstilbestrol, RU486)
Low cost Side effects (dose dependent)
Bait or implant Must be administered regularly
Non-target effects
Prolactin inhibitors (affect lactation and/or gestation Low Moderate
(e.g. Bromocriptine, Cabergoline)
Oral delivery - Not permanent
Low cost May not be ethically acceptab le as
starves young or aborts foetuses
Must be regularly administered
Immunocontraception Disseminating vector, non-<iisseminating vector, Moderate but Expected high
synthetic delivery systems (e.g. ISCOMs b, microspheres) untested
Long term reduction in fertility Not yet available
Species-specific , humane, cost effective May need to repeat application l:1l:I
Could be reversible Includes use of genetically modified c'
C'
organisms I1Q
n'
III
a GnRH = gonadotrophin releasing hormone
b ISCOMs = immunostimulatory complexes.
.....=
(")
c
c

-
c
:::a
c
N
....
N ..=
Cl.
I'D

In
Ecologically-based Rodent Management

maintain social status and may suppress the
fecundity of subordinates. Such an approach
is potentially species-specific, considered
humane and could be cost effective in the
long term (Tyndale-Biscoe 1994).
Unlike vaccines directed against
infectious diseases, immunocontracep tive
vaccines are directed agains t 'self' proteins
that would not normally be recognised as
foreign (Alexander and Bialy 1994; Tones
1994; Dunbar 1997). There fore, th e 'self
antigen to be used in the vaccine must be
presented in a 'foreign' or 'non-self' form to
eli cit an immune response. In 1987, a new
approach to fer tility control was
conceived-the concept th a t viruses could
be used to d eliver immunocontracep tives
(Tyndale-Biscoe 1994) (see Figure 1). This
could be achieved by deliverin g the
imrm mocontraceptive vaccines th ro ugh th e
agency of a virus or other con tagious agen ts
th at spread na turally through th e target pest
pop ulation . Similarly, a non-disseminating
agent in baits co uld be used to provoke an
appropriate immune response.
Since 1992, this approach has been under
developmen t at the Cooperative Research
Centre for Biological Control of Vertebrate
Pest Populations (Vertebrate Biocontrol
CRC) and its successor the Pes t Animal
Con trol CRC based in Canberra, Australia.
The Centre's mission is "to contrib ute to the
better management of Australia's
biod iversity by limiting growth of vertebrate
pest popu lations through fertility control".

Insert DNA in
fW
DNA
- mouse virus

Sperm or egg proteins

Fertilisation blocked
by antibodies Infect host

Host cells produce

anti-sperm/egg
antibodies

Figure 1.
The concept of viral-vectored immunocontraception. Genes encoding a reproductive protein(s) are
incorporated into the genetic structure of a species-specific virus. This virus infects the host, expressing
the reproductive protein(s) as well as viral proteins on the surface of infected cells. The host 's immune
system produces anti bodies against the reproductive protein(s ), as well as the vi rus, and these spread to
the reproductive tract where they bind to either the egg or the sperm and block fe rtilisation. Redrawn with
permission from the Vertebrate Biocontrol CRC.

222

IMMUNOCONTRACEPTION -
ITS COMPONENTS
The choice of reproductive antigen(s)
Fertility control agents may be of two types:
anti-gonadal or anti-gametic. By targeting
the gametes there is likely to be less
disruption of other reproductive functions,
but sustained immune responses may be
more difficult to achieve because the gamete
proteins are produced in small quantities at
specific sites and are not highly
immunogenic (Alexander and Bialy 1994).
Initial studies by many groups have focused
on sperm proteins as candidate antigens.
The view was that male antigens might be
able to induce significant immlme responses
in the female reproductive tract because they
were not expected to be auto-antigens in
females. The potential of sperm proteins,
such as SP-1 0 and testis-specific lactate
dehydrogenase (LDH-C4) has been explored
in humans, baboons and pigs (Goldberg and
Shelton 1986; Herr et al. 1990a,b) and PH-20
in guinea pigs (Primakoff et a1. 1988). Initial
research in the Vertebrate Biocontrol CRC
also focused on sperm antigens (Bradley
1994; Holland and Jackson 1994; Tyndale-
Biscoe 1994). However, after several sperm
antigens had been tested by direct
immunisation without effects on the fertility
of female rabbits or foxes (Bradley et a1. 1997;
Hardy et al. 1997; Holland et al. 1997),
attention turned to the female gamete
antigens, specifically the zona pellucida
proteins forming the extracellular coat of the
oocyte.
In the mouse, the zona pelluCida
comprises three non-covalently linked
glycoproteins, ZPA, ZPB and ZPC, which
are expressed by the growing oocytes in the
Biological Control of Rodents
ovary. ZPC is the receptor for sperm binding
at the time of fertilisation (Florman and
Wassarman 1985; Rosiere and Wassarman
1992). Passive immunisation with
monoclonal antibodies to ZPC inhibit
fertilisation in vivo (East et a1. 1984, 1985),
and active immunisation with synthetic
pep tides which include aB-cell epitope of
ZPC also induce infertility for periods from
0-8 months (Millar et a1. 1989; Lou et al.
1995). For these reasons, mouse ZPC (also
known as ZP3) was the first candidate
antigen to be tested in a mouse viral-
vectored system.
Many of the zona pellucida proteins and
sperm proteins show high identity between
species (Harris eta1. 1994; Bradley eta!. 1997;
Holland et al. 1997; Jackson et a1. 1998).
Therefore a key challenge is to identify or
engineer the antigen to be species-specific.
This may be achievable using specific
peptides or epitopes. The difficulty then
becomes whether such small pep tides have
the ability to block fertility. The use of
epitopes alone or in combination with
immunomodulatory molecules (such as
cytokines or T-cell help epitopes) to enhance
the species-specific immune responsiveness
to these antigens are currently being
investigated (Dalum et al. 1997; Ramsay and
Ramshaw 1997).
Delivery of the immunocontraceptives
The delivery of an anti-fertility vaccine to
populations of wild animals over large areas
poses a number of unique problems. It is
essential to consider the distribution of the
species under study, whether large-scale or
localised control is desired, and any possible
consequences for non-target species. Another
factor is the genetic heterogeneity of the
223
Ecologically-based Rodent Management
wildlife population, which is certain to
generate significant individual variability in
the immune responses to a vaccine (Klein
1979). Effective application of any vaccine
requires that a high level of immunity can be
achieved amongst individuals exposed to the
vaccine (Alexander and Bialy 1994). As
mentioned previously, it may therefore be
necessary for the antigen(s) to be presented in
conjunction with other highly immunogenic
carrier proteins (e.g. cytokines and
immunomodulatory molecules) to maintain a
contraceptive level of immunity. In addition,
multiple antigenic detenninants could be
included within a vaccine to stimulate a
broad range of immune responses.
The three main delivery systems under
development are (i) non-disseminating
genetically modified organisms (GM Os) in
baits, (ii) synthetic delivery systems and (iii)
disseminating GMOs such as viruses or
bacteria. For many rodent pests, particularly
those that are native species, bait delivery
may be the method of choice for political,
social, economic and ecological reasons.
Non-dlsseminating agents
Non-replicating GMOs, such as
attenuated Salmonella, are currently being
developed and tested (Bradley 1994; Bradley
et al. 1997). Selected mutant strains of
Salmonella have the advantage that they are
avirulent without decreasing their
immunogenicity and they are not infective.
Furthermore, the introduction of a 'suicide'
plasmid into this system would have the
added advantage of degrading the foreign
deoxyribonucleic acid (DNA) and would
make it more acceptable because the bait-
delivered product would contain no foreign
224
genetic material (Knudsen et al. 1995; Tedin
et al. 1995).
Various gram bacteria
(Escherichia coli, SalltlOnella typhimurium,
Vibrio cholerae, Klebsiella pneumoniae and
Actinobacillus) can be engineered to carry a
gene (PhiX174 geneE) which, when induced,
causes lysis and release of the cytoplasmic
contents of the bacteria. This process
produces a non-living vaccine delivery
system. These bacteria can also be
engineered to carry other genes (e.g.
encoding reproductive proteins). After lysis,
the' ghost' bacteria contain only membrane-
associated recombinant antigen. Bacterial
ghosts are cheap to produce, can be stored
for long periods and can contain multiple
antigenic determinants that are present in a
highly immunostimulatory environment
(Szostak et al. 1996). Such features make
bacterial ghosts an attractive delivery system
for immunocontraceptive antigens.
However, it remains to be seen whether
these preparations produce immunity to
reproductive antigens after oral delivery.
Synthetic delivery systems
Synthetic delivery systems for antigens
include ISCOMs (immunostimulatory
complexes-e.g. Quil A, cholesterol,
phospholipid constructs), microspheres
(polylactide-coglycolide polyphosphazenes),
and liposome emulsions (Davis 1996).
The current high costs of production
mean these systems are only suitable for
human and companion animal vaccination
and not for broad-scale application to a
wildlife population. Nevertheless, the per
unit production cost will decrease as these
systems become more popular and
production teclmology improves.
 Biological Control of Rodents

Disseminating GMOs VIRAL-VECTORED

These currently have the greatest
theoretical potential for use as vectors for
immunocontraceptive agents. A viral vector
could potentially overcome problems
associated with the distribution of an
immunocontraceptive to control wild
populations (Bomford 1990). The
advantages of a viral-vectored
immunocontraceptive agent over a bait-
delivered immunocontraceptive agent are
summarised in Table 3. Clearly the selection
of a viral vector requires careful
consideration of its properties. The current
vector of choice for delivery of a mouse
immunocontraceptive is mouse
cytomegalovirus and it possesses most of the
essential and desirable characteristics
required (Table 4), although additional
research is required to confirm some of its
features.
IMMUNOCONTRACEPTION (VVIC) -
LABORATORY PROGRESS
Ectromelia
Ectromelia virus (ECTV: family Poxviridae,
genus Orthopoxvirus) causes the disease
known as mousepox and is a pathogen of
laboratory mice (Fenner and Buller 1997). It
is closely related to vaccinia virus and was
investigated as a useful model system for the
development of viral-vectored immuno-
contraception (VVIC). A recombinant
ectromelia virus, with a thymidine kinase
negative phenotype, expressing ZPC was
constructed and then used to infect female
inbred laboratory mice of the BALB/ c strain,
which are highly susceptible to mousepox
Qackson et al. 1998). Fertility was assessed
by pairing females with males from three
weeks post-infection and monitoring for

Table 3.
Viralvectored versus baitdelivered immunocontraceptives (after Bomford 1990; Shell am 1994;
Chambers et al. 1997)

Advantages of a vlral-dellvered Immunocontraceptlve

A replicating virus may induce a stronger immune response and greater immunological memory.

An infective agent can potentially spread a reproductive protein rapidly through a population.

A self-perpetuating, infectious agent is ultimately cheaper than baits which must be manually applied.

A viral vector is a species-specific carrier.

Overcomes problems associated with bait aversion or bait shyness.

Overcomes the precise timing necessary for bait delivery relative to the target animal's breeding cycle.

Reduces wastage associated with inadvertent multiple baiting of some individuals.

Advantages of a balt-dellvered Immunocontraceptlve

More acceptable to the public than the use of a disseminating genetically modifiad organism.

Easier regulation of control activities-can be readily withheld or withdrawn from use.

225
Ecologically-based Rodent Management

Table 4.
Essential and desirable properties for a virus which will act as a vector of an immunocontraceptive agent for
the biological control of rodents (after Shellam 1994). Does murine cytomegalovirus (MCMV) meet these
requirements?

Essential properties MCMV

Species-specific and naturally infects target
species
Readily transmitted in target species
Insertion of foreign gene is stable and does
not affect viral growth or transmission
Stimulates long lived immune response and
immunological memory
Recombinant virus can be introduced and
maintained in the presence of existing
immunity
Panel of isolates available
Epidemiology of infection understood and
site of viral growth known
Approval by regulatory authorities likely
Native murids will be tested to verify this
Seroprevalence >90% in wild mice (Smith et al
1993)
Insertion sites identified (Manning and Mocarski
1988); recombinant constructed with beta-
galactosidase gene. More research required on
effects on transmission
?
? Wild mice have been found with up to four
strains; infection with multiple strains can be
achieved in the laboratory (Booth et al 1993).
Epidemiology of this needs to be examined in
wild mice
..;
..; Virus perSists in submaxillary gland Weak
knowledge of epidemiology outside laboratory
Already in Australia

Desirable properties MCMV

Virus is already in the country (see Smith et al. 1993)

Virus establishes persistent and non-lethal V
latent infection
Good local immunoglobulinA response which ..;
does not interfere with transmission
Mechanism for any genetically determined Ability of subsequent infections to stimulate
host resistance is known immune response not known
Genetically determined host resistance does ..;
not interfere with infection or transmission
Mechanism of transmission known ..; Close contact; sexual and via saliva
Virus is sexually transmitted ..; Enhances species-specificity
Knowledge of the epidemiology of infection ?
and transmission of natural virus variants
A DNA rather than an RNA virus (greater ..;
genetic stability)

226

evidence of pregnancy and birth of litters.
Two major experiments were conducted,
one to assess the immediate effects on
fertility and the second to test the duration of
the effects.
The immediate effects on fertility were a
reduction in the number of litters produced
by females infected with ECTV-ZPC
compared to uninfected controls or females
infected with recombinant ectromelia virus
(ECTV-602) expressing a non-reproductive
marker protein, LacZ (Table 5). The effects on
fertility were long term, with mice infected
with ECTV-ZPC infertile for periods of 5-9
months while those infected with ECTV-602
remained fertile. Mice became fertile as the
anti-ZPC antibodies in the serum decreased,
but when they were re-infected with the
recombinant virus, antibody titres to ZPC
increased and the animals returned to an
infertile state (Jackson et a1. 1998). Therefore,
this study provided the first demonstration
of VVIC in laboratory mice.
Examination of the ovaries of i.nfertile
females revealed two possible mechanisms
for infertility. Half of the animals showed
disruption in folliculogenesis, with an
absence of mature follicles and oocytes as
Table 5.
Infertility in BALB/c mice infected with either recomblnant ectromelia virus expressing zona pellucida
glycoprotein C (ECTV-ZPC) or recombinant ECTV expressing a nonreproductive marker protein (ECTV-602)
compared with uninfected controls (after Jackson et al. 1998), SE standard error.
Ectromelia virus No. of mice with
Infection litters/total mice
None 10/10
ECTV-ZPC 4/13
Biological Control of Rodents
well as large clusters of luteinised cells
(Jackson et a1. 1998). There was no
observable oophoritis. The remaining
animals showed normal ovarian
development of follicles and ovulation;
antibody localisation studies indicated
binding of ZPC antibodies to these oocytes,
suggesting that after ovulation, sperm
would not be able to bind and result in
fertilisation (Jacks on et a1. 1998).
Murine cytomegalovirus
Ectromelia virus is not present naturally in
the Australian environment and therefore,
for ethical, political and social reasons, is not
an ideal candidate for release as a viral
vector of an immunocontraceptive agent.
Moreover, its lethality would select for
resistance more rapidly than a non-lethal
agent. Other research is being conducted
using murine cytomegalovirus (MCMV)
which is highly prevalent in Australian
mouse populations and possesses the
desirable properties of a vector (Table 4)
(Singleton et a1. 1993; Smith et a1. 1993;
Shell am 1994). This large DNA virus (230 kb,
~200 genes) is a member of the Betaherpes-
virinae sub-family of the Herpesviridae. It
=
No. of Implantations Utter size
(mean :t SE) (mean:t SE)
All animals Animals with All animals
litters
9.5 0.8 9.5 0.8 6.6 0.8 6.6 0.8
8.5 0.9 6.8 1.1 7.3 0.7 5.8 1.0
2.5 0.7 0.8 0.4 1.8 0.3 0.5 0.2
227
Ecologically-based Rodent Management
shows strict species-specificity (Hudson
1994) and establishes a persistent infection in
the salivary gland with latent infection
apparently associated with ubiquitous
elements such as macrophages (Koffron et
a1. 1995; Pollock et a1. 1997), rather than with
organ-specific cells (e.g. hepatocytes).
Infection requires close contact and the virus
is believed to be transmitted via secretions
such as saliva and sexual secretions (see
Shellam 1994).
Recombinant MCMV has been
constructed by inserting the mouse ZPC
gene into the immediate early 2 (ie2) gene.
The effects on fertility of infecting with
recombinant MCMV expressing either ZPC
or a non-reproductive marker gene (LacZ)
were assessed for several different inbred
strains of mice (BALB / c, A/I, C57BL/6,
ARC/s) with varying susceptibility to
infection with MCMV (Grundy et a1. 1981;
Allan and Shellam 1984). Recombinant
MCMV-ZPC induced a long-lasting, high-
titred antibody response to ZPC in all mice
tested. The fertility of uninfected controls
was also determined. BALB / c females
(n 9) infected with recombinant MCMV-
ZPC produced no litters for 200 days after
infection, while the uninfected controls and
the MCMV-LacZ infected group produced
approximately 250-350 pups during the
same period (Figure 2). The response was
similar in A/J females although the overall
prod uctivity of this strain was lower.
Contraceptive effects of lesser magnitude
were observed in C57BL/6 and ARC/s
strains. The ovaries of recombinant MCMV-
ZPC infected females showed histological
changes but the mechanism of infertility
remains under investigation.
228
These results (M. Lawson et al.,
unpublished data) demonstrate that
recombinant MCMV expressing the ZPC
gene can elicit an immunocontraceptive
response in mice. This response occurred in
the absence of high levels of replication of the
recombinant virus, since very low levels of
virus were found in the salivary glands of
mice relative to controls. Research is
continuing on ways to enhance this response
in less susceptible strains of mice as well as to
demonstrate the transmissibility and
competitiveness of the recombinant virus
when confronted with prior MCMV infection
in wild outbred mice (see next section).
EPIDEMIOLOGICAL CONSIDERATIONS
Manipulating the genetic structure of a virus
to incorporate an immunocontraceptive
antigen may effect its transmissibility,
persistence and species-specificity. Thus, it is
important to examine the epidemiological
consequences of such a manipulation from
both an ecological and a viral engineering
perspective. The key questions that need to
be addressed are:
.... What is the transmission rate of the wild-
type virus and the recombinant sterilising
virus? Do they differ? If so, why?
.... Do the characteristics of viral infection
such as the immune response and site of
replication differ between the wild-type
and recombinant virus?
.... What is the threshold population size
required to maintain the viral infection at a
specified prevalence? What influences
this?
 Biological Control of Rodents

~ Can a recombinant strain of the virus
establish and generate an irnmw1e
response in a rodent population that may
have a pre-existing infection with the wild-
type virus? Is the order of infection
important?
~ What is the persistence of the virus in the
environment?
Many of these questions are difficult to
test in wild populations, particularly for the
200 .------------------------------.
(a) BALB/c
recombinant viruses where thorough testing
w1der contained conditions is required
before release into a field population. A
crucial experiment will be to examine if the
impact of the sterilising, recombinant virus
on breeding affects the transmissibility of the
virus.
Experiments will be conducted to
address these questions using large
(2 m x 2 m) cages to house a simulated
'population' of mice. These cages are
200 .------------------------------.
(b) AlJ

150 150

100 100

50 50

<Jl
.t:::
o o
to
:0
Q)
o 20 40 60 80 100 120 o 20 40 60 80 100 120
>
~ Days after first introduction of male
:J
E
:J
u 250 .------------------------------. 500 .------------------------------.
(c) C57BU6 (d) ARC/s

200 400

150 300

100 200

50 100

o

L-.
L __ _ ~~ ____L __ _ ~_ __ L_ _~L_~
o 20 40 60 80 100 120 o 20 40 60 80 100 120

Days after first introduction of male

Figure 2.
Cumulative births in different strains of mice infected with either recombinant murine cytomegalovirus-
zona pelucida glycoprotein C (MCMV-ZPC) ( 'f' ), or recombinant MCMV-LacZ (a non-reproductive marker
protein gene ( ) compared with uninfected controls ( e ). Groups of nine females were infected with 2 x 104
pfu (plaque forming units) of tissue culture-derived virus 21 days prior to the introduction of males to the
breeding cages. Each cage contained three females and one male. Groups were checked for births several
times per week.

229
Ecologically-based Rodent Management
internally complex so that mice can avoid
each other within the cage if required.
Uninfected mice will be released into this
cage, a number of mice infected by
intraperitoneal inoculation and the spread of
the virus monitored. Further studies of
MCMV in wild populations will still be
necessary to assess the relevance of these
cage results. The use of a virus strain
expressing an innocuous, non-reproductive
marker gene would be useful in this instance
but awaits regulatory approvaL
A complementary approach is the use of
epidemiological models to predict the likely
behaviour of a sterilising virus in a field
population. The choice of viral vector for an
immunocontraceptive has several
epidemiological consequences. In
polyoestrous species-where the sterilising
virus is assumed to be sexually transmitted,
persists in the infected host and does not
disrupt gonadal function-the recombinant
virus will have a selective advantage over
the native strain. This is because the more
frequent return to oestrus in sterilised
females may provide more opportunities for
transmission (Barlow 1994; Tyndale-Biscoe
1995; Barlow et a1. 1997). If gonadal function
is disrupted, the animals may not show
normal mating behaviour and this may
reduce transmission of the virus. The
promiscuity of males and their persistent
infection with MCMV will then be critical for
transmission to susceptible females.
A virus that is sexually transmitted
increases the chances of the VVIC agent
contacting only the target pest population
compared with a contagious or insect-borne
virus. However, spatial modelling suggests
that a sexually transmitted virus would be at
a disadvantage when compared with a virus
230
spread by insect vectors, as the requirement
for close contact could potentially limit the
rate of spread of the virus (G.M. Hood,
unpublished data). These considerations
need to be balanced when determining
which viral vector is most appropriate in
each pest control situation.
ECOLOGICAL IMPLICATIONS
Immunocontraception can only be judged to
be successful for rodent control if it reduces
population size and damage as a
consequence of reducing reproduction
(Bomford 1990; Braysher 1993). For each
species, there is a population threshold
below which the damage inflicted is
tolerable. The objective is not to eradicate the
pest species, an impossible task in most
instances, but to reduce the pest species to
below this 'tolerable level'.
Populations have inherent regulatory
mechanisms preventing over-population
which counteract an innate ability to
produce surplus offspring (Howard 1967;
Sinclair 1989). If a population to be
controlled is already at high density,
density-dependent mortality and dispersal
are probably already high amongst juveniles
and therefore, sterilisation will simply
prevent birth of young that would otherwise
die or disperse without breeding. Sterility
rates must be sufficiently high to lower
recruitment to the adult population if
sterilisation is to reduce population size
(Bomford 1990). This emphasises the
importance of gaining some understanding
of the factors regulating populations and
how these are affected by fertility control.
Fertility control may interact with other
popUlation processes to enhance the overall

effect on population numbers. For example,
if a pest species is prevented from increasing
its reproductive rate, predation may be able
to maintain their population at a low level
(Sinclair 1997). This may apply to house
mouse populations in Australia in which
avian predators are capable of regulating the
population when mouse densities remain
low but can not maintain this regulation
when mouse densities increase to high levels
(Sinclair et al. 1990).
Several ecological questions need to be
addressed when assessing the potential of a
particular immunocontraceptive agent:
... What proportion of a wild pest population
needs to be sterilised to significantly
reduce growth rate and population size?
And can a delivery system be developed
that will reach the required proportion of
the population?
... Is the maintenance of social structure
important and does it affect the efficiency
of the immunocontraceptive?
... Is compensation a likely factor that could
reduce the efficacy of an
immunocontraceptive?
A further question is whether density-
dependence plays a role in modifying the
efficacy of a given sterility level. Will the
proportion of sterilised individuals need to
be increased in a high-density compared
with a low-density population to have the
same impact? Is compensation density-
dependent? Some of the implications for
density-dependent regulation on the
applicability of sterilisation to control
populations have been discussed in a
previous section.
Biological Control of Rodents
Level of sterility required
Modelling
Mathematical models have been used to
estimate the level of fertility control required
to produce a significant reduction in
population size. Knipling and McGuire
(1972) modelled the effects of permanent
sterilisation of females or both sexes versus
killing similar numbers in a rat population.
Their model predicted that by sterilising
90% of both sexes, this had a greater effect
than killing 90% of rats. However, they
assumed that if 90% of males don't breed,
90% of females would not breed. In a species
such as rodents with a promiscuous mating
system, this is an invalid assumption
(Kennelly et al. 1972; Pennycuik et al. 1978).
There was also no allowance for
compensatory changes in immigration and
dispersal.
N. Stenseth et al. (unpublished data) have
modelled empirical field data from
populations of the multi-mammate rat
Mastomys natalensis, in eastern Africa. The
model simulated a permanent decrease in
reproductive rate of this species and found
that long-term reductions in population
density were attained if between 50 and 75%
of females were sterile.
A simple demographic model using Iife-
history information obtained from
laboratory, enclosure and field studies was
used to examine the proportion of mice to be
sterilised to produce a significant decrease in
population size in enclosure populations
(Chambers et al. 1997). This simulation was a
useful precursor to a manipulative
experiment (described below) assisting with
f
experimental design and indicating the
types of data that needed to be obtained. The
231
Ecologically-based Rodent Management
model examined two levels of sterilisation
(67% and 75% of females) and compared the
outcome against a non-sterilised population.
The simulation found that both the 67% and
75% levels of sterility were sufficient to
reduce population size and growth rate,
relative to the unsterilised population
(Figure 3).
However, models often overestimate the
effectiveness of an immunocontraceptive as
they are generally based on higher levels of
fertility control than can be practically
achieved in the field. They also tend to
ignore or underestimate factors that may
reduce the effects of fertility controt such as
compensatory changes in behaviour,
survival or fecundity (Bomford 1990). The
latter applies to all of the models described
above.
Manipulative experiments
Manipulative experiments can be used to
examine empirically the sterility level
suggested from mathematical models.
Experiments involving surgical sterilisation
allow the degree and nature of sterilisation
required to reduce population size to be
examined (Kennelly and Converse 1997). For
example, when females in populations of
house mice housed in outdoor enclosures
(Figure 4) were surgically sterilised at a level
of 67%, this significantly reduced population
size. Over 18 weeks, populations were
reduced from a mean abundance of 221 mice
in two control populations to 104 mice in
four sterilised populations (Chambers et aL
1999).
Kennelly et al. (1972) sterilised 85% of
males in a Norway rat population and found
no effect on population size when compared
with an unsterilised population, confirming
232
that the development of an immuno-
contraceptive for males in a species with a
promiscuous mating strategy is not effective.
Social structure
Many studies of wild mammals have shown
that reproductive success is closely linked to
an animal's rank in the social hierarchy.
Lower ranking animals either do not breed
or fail to rear their young to independence
(Wasser and Barash 1983; Abbott 1988).
Caughley et al. (1992) highlighted the need
to have some understanding of the social
structure and mating system of the species to
be controlled by fertility control. They
showed via modelling that for most
scenarios, sterilisation would reduce
population growth, irrespective of mating
system or social structnre. However, where
the sterilisation of a single dominant female
releases subordinates from breeding
suppression, sterilisation actually enhanced
the overall productivity of the population.
This emphasises the need to sterilise
individuals without compromising their
social position (Chambers et aL 1997). This
would maintain the breeding performance
of subordinates at a low levet preventing
compensation by these individuals for the
reduction in population growth (Caughley
et al. 1992; Barlow 1994; Tyndale-Biscoe
1994; Cowan and Tyndale-Biscoe 1997).
The importance of maintaining hormonal
competence in surgically sterilised females
in reducing the overall productivity of
populations has been examined for mice
housed in near-natural outdoor enclosures
(Chambers et al. 1999). Female mice were
either ovariectomised (hormonally
incompetent) or tnbally ligated (hormonally
competent) at the rate of 67% per population
 Biological Control of Rodents

300 ,----------------------------------------------------------------,

Non-sterilised (control)
250 67% sterilised (8/12)
75% sterilised (9/12)

. 200
Cii
"S
n.
o
~ 150
:0
Cl!
n.
n.
~ 100
f-

50

o
o 15 30 45 60 75 90 105 120 135
Day

Figure 3.
Demographic model predicting the trappable population of mice housed in outdoor enclosures after 0%,67%
and 75% offemales have been sterilised (see Chambers et al. 1997 for details ofthe model) . Each plot is
the mean ( standard deviation) of 10 runs of the model. F11 to F14 indicates when F1 generation litters
(those produced by the founding population of mice) will enter the trappable population. F2\ indicates when
the first litter of the F2 generation (produced by the F1 1 litter) enters the trappable population (adapted
from Chambers et al . 1997).

(a) (b)

Figure 4.
Outdoor enclosures used for manipulative experiments examining the effectiveness of fertility control to
reduce mouse population abundance and rate of increase. Each enclosure is 15 m x 15 m in area and is
protected from predators by wire mesh fencing. Mice are prevented from burrowing into or out of the
enclosures by metal fences that are buried to a depth of 800 mm. Food and water are provided ad libit um .
(a) Ground-level view; (b) Aerial view.

233
Ecologically-based Rodent Management
and compared with unsterilised populations
(n 2 enclosures per treatment) after 18
weeks. The mean ( standard error)
abundance was 126 ( 17) for the tubally
ligated populations and was 81 ( 12) for the
ovariectomised populations. When these
were compared with the control populations
(mean 221 with standard error 26) using
analysis of variance, there was no significant
difference between the two methods of
sterilisation. Thus for house mice, it appears
that the maintenance of hormonal
competence in sterilised females is not
important for fertility control to be effective
in reducing population size.
If the maintenance of social structure is
important, this has consequences for the type
of immunocontraceptive antigen to be used
(Chambers et a1. 1997). Some of these antigens
are known to cause oophoritis or ovarian
dysfunction that may affect the release of
hormones controlling reproduction and
social position (Skinner et a1. 1984;
Kirkpatrick et a1. 1992; Rhim et a1. 1992).
Compensation
If fertility is reduced, the average population
size is also reduced, but only under certain
conditions. If juvenile or adult survival
improves with lower fertility or territoriality
limits populations, the effects of lower birth
rate will not change population size unless
such reduction exceeds the effects of these
processes (Sinclair 1997). If sterile
individuals have increased survival, they
may cause more damage than non-sterile
animals (Bomford 1990).
As was discussed earlier, modelling the
effects of fertility control on wildlife
populations often ignores the effects of
compensation. Therefore it is important to
234
measure this in manipulative experiments to
allow predictions of the efficacy of
immunocontraception to be improved.
PUBLIC ACCEPTANCE OF GMOs
There are some that take the view that VVIC
will never be adopted as a control strategy
for wild pest populations such as the house
mouse in Australia because GMOs will not
be accepted by the public. However, it is
important in this debate to weigh up the
risks of VVIC (Table 6) against the inherent
risks in conventional control methods (Table
1). It is also imperative to separate the
perceived and real risks of VVIC and balance
this against the benefits gained in reducing
the damage caused by the pests (Chambers
et a1. 1997). The strategy adopted by the
Vertebrate Pest Animal Control eRC is that
research should proceed incrementally with
public discussion at each step so that its
potential use can be weighed again.',t the
risks (Tyndale-Biscoe 1997b).
The risks of WIC are discussed in detail
in Tyndale-Biscoe (1994, 1995), Guynn (1997)
and Williams (1997) and are summarised in
Table 6. Most relate to the issues of public
acceptability of GMOs and maintaining the
species-specificity of the recombinant virus.
How species-specificity is achieved will
depend on the target animal, the ecosystem,
the delivery system, local non-target species
and, in general, the aims of the fertility
control program (Stohr and Meslin 1997).
The important question to address is if a
WIC encounters a non-target species, will
this cause infertility even though no
productive infection occurs? Species-
specificity operates at three levels: the viral
vector, the reproductive protein and social

factors governing the spread of the VVIC
agent (Tyndale-Biscoe 1994). Ideally, all of
these levels of specificity should be satisfied.
Public acceptability will be heavily
influenced by the media's interpretation of
this technology (Williams 1997) as well as by
international debate and agreement on its
safety (Oogjes 1997; Stohr and Meslin 1997;
Williams 1997).
Apart from the issues associated with the
use of a GMO, public acceptability also
encompasses animal welfare issues.
Although it is generally agreed by animal
Table 6
Risks and benefits of viral-vectored immunocontraception (WIC).
Risks
Public concerns about genetically modified
organisms (Regal 1986; Molak and Stara 1987;
Siddhanti 1987)
Possibility of non-target species infection (national)
and infection of target species in another' ountry
where it may be a desirable part of the fauna
(international) (Tyndale-Biscoe 1995)
Possibility of pathogens broadening their host range
after genetic modification (Regal 1986; Kurtz 1987;
Tiedje et al. 1989)
Potential for behavioural / hormonal disruptions to
cause ill effects in sterilised individuals; other
animal welfare/ ethical issues such as potential
mortality in utero (Guynn 1997)
Irretrievable once released
Virus may infect laboratory colonies
VVIC may select for animals with poor immune
systems, therefore favouring immunodeficient
animals and thus increasing their susceptibility to
pathogens (Guynn 1997; Nettles 1997)
Legal implications with respect to federal and state
registration requirements (Guynn 1997).
BiC)logical Control of Rodents
welfare groups that immunocontraception is
a more acceptable form of control than the
current lethal methods (Oogjes 1997; Singer
1997), there are other biological issues that
need to be considered. For example, Guynn
(1997) expressed concern over sterilised
females experiencing an abnormal number
of oestrous cycles and thus expending more
energy. The use of long-term field trials
should give some indication of behavioural
changes experienced by sterilised and non-
sterilised individ uals. For example, WiUiams
and Twigg (1996) found during the first year
Benefits
Environmentally benign
More humane than conventional methods of control;
supported by animal welfare groups (Oogjes 1997;
Singer 1997)
Species-specific
Can be used in terrain where pest species would be
inaccessible to instigate conventional control
methods
More appropriate for highly fecund pest species as
it targets reproduction (Bomford 1990; Tyndale-
Biscoe 1994, 1995)
May be active long- or short-term and therefore have
the potential to be a flexible tool for population
management
Presence of sterile individuals in the population may
exert a much greater biological control pressure
than if the same number of fertile animals were
removed (Howard 1967)
Self-disseminating 'release and forget' strategy
235
Ecologically-based Rodent Management
of a sterility trial on wild rabbit populations
that sterilised females had higher
survivorship and body weights than
unsterilised females.
Conclusion
Ecologically-based pest management
requires the application of a suite of
strategies to manage pest species. New
approaches, such as fertility control, will
become one of these strategies and thus must
not be seen as a replacement for
conventional methods of control. Where
damage mitigation is the objective for
population reduction, the short-term use of
lethal approaches may still be appropriate.
However, it is the prolonged use of such
techniques that should be discouraged.
Fertility control techniques that are
currently available are not logistically
appropriate for wildlife populations. They
are expensive and! or invasive, require
repeated dosing to maintain sterility levels
and often have side effects that lead to
behavioural changes. They are also difficult
to administer on a broad-scale. Immuno-
contraception, and in particular viral-
vectored immunocontraception, aims to
overcome many of these shortcomings by
being a naturally disseminating, species-
specific fertility agent. However, one
disadvantage of this method is the public
acceptance of the use of a GMO. Therefore, it
is important that the risks of new methods of
control, including GMOs, are fully assessed
through experimental trials and public
debate. These risks should be viewed in the
context of control methods which are
currently available-i.e. non-specific, fatal
poisons. This may mean that these new
methods will not be available for broad scale
236
release for at least another decade. However,
the potential rewards of this technology will
be well worth the longer-term investment.
REFERENCES
Abbott, D.H. 1988. Natural suppression of fertil-
ity. Symposium of the Zoological Society of
London, 60, 7-28.
Alexander, N.J. and Bialy, G. 1994. Contraceptive
vaccine development. Reproduction, Fertility
and Development, 6, 273-280.
Allan, J.E. and Shellam, G.R. 1984. Genetic
control of murine cytomegalovirus infection:
virus titres in resistant and susceptible strains
of mice. Archives of Virology, 81, 139-150.
Barlow, N.D. 1994. Predicting the effect of a novel
vertebrate biocontrol agent: a model for viral-
vectored immunocontraception of New
Zealand possums. Journal of Applied
Ecology, 31, 454-462.
Barlow, N.D., Kean, J.M. and Briggs, CJ. 1997.
Modelling the relative efficacy of and
sterilisation for controlling populations.
Wildlife Research, 24, 129-142.
Baskin, M.}. 1932. Temporary sterilization by the
injection of human spermatozoa. A prelimi-
nary report. American Journal of Obstetrics
and Gynaecology, 24, 892-897.
Becker, S.E. and Katz, L.S. 1997. Gonadotrophin
releasing hormone (GnRH) or active
immunization against GnRH to control fertil-
ity in wildlife. In: Kreeger, ed" Contra-
ception in wildlife management. Technical
Bulletin No. 1853, United States Department
of Agriculture, Animal and Plant Health
Inspection Service, 11-19.
Beckmann, R. 1988. Mice on the farm. Rural
Research, 138, 23-27.
Bomford, M. 1990. A role for fertility control in
wildlife management? Bulletin No. 7.,
Canberra, Bureau of Rural Resources, SOp.
Bomford, M. and O'Srien, P. 1997. Potential use
of contraception for managing wildlife pests
in Australia. In: ed. Contracep-
tion in wildlife Technical Bulle-
tin No. 1853, United Department of
Agriculture, Animal and Plant Health Inspec-
tion Service, 205-214.

Booth ,T.W.M., Scalzo, A.A., Carello, c., Lyons,
P.A., Farrell, H.E., Singleton, G.R. and
Shellam, G.R. 1993. Molecular and biological
characterisation of new strains of murine
cytomegalovirus isolated from wild mice.
Archives of Virology, 132,209-220.
Bradley, M.P.1994. Experimental strategies for
the development of an immunocontraceptive
vaccine for the European red fox, Vulpes
vu/pes. Reproduction, Fertility and Develop-
ment, 6, 307-317.
Bradley, M.P., Hinds, L.A. and Bird, P.H. 1997. A
bait-delivered immunocontraceptive vaccine
for the European red fox (Vu/pes vulpes) by the
year 2002? Reproduction, Fertility and Devel-
opment,9,111-116.
Braysher, M. 1993. Managing vertebrate pests;
principles and strategies. Canberra, Bureau of
Resource Sciences, 58p.
Caughley, G. 1977. Analysis of vertebrate
populations. London, John Wiley and Sons,
234p.
Caughley, G. 1985. Harvesting of wildlife: past,
present and future. In; Beasom, S.L. and
Roberson, S.F., ed., Game harvest manage-
ment. Kingsville, Texas, Caesar Kleberg
Wildlife Research Institute, 3-14.
Caughley, G., Pech, R. and Grice, D. 1992. Effect
of fertility control on a population's produc-
tivity. Wildlife Research, 19,623-627.
Caughley,J.,Monamy, V. and Heiden,K.1994.
Impact of the 1993 mouse plague. GRDC
Occasional Paper Series No.7. Canberra,
Grains Research and Development Corpora-
tion,73p.
Chambers, L.K., Singleton, G.R., and Hinds, L.A.
1999. Fertility control of wild mouse popula-
tions: the effects of hormonal competence and
an imposed level of sterility. Wildlife
Research, 26, 579-591.
Chambers, L. K., Singleton, G.R. and Hood, G.M.
1997. Immunocontraception as a potential
control method of wild rodent populations.
Belgian Journal of Zoology, 127, 145-156.
Childs, I.E., Ksiazek, T.G., Rollin, P.E., Krebs,
J.W., Zaki, S., Nichol, S.T., Peters, CJ. and
Glass, G.E. 1994. Hantaviruses and their
rodent reservoirs in the United States. In:
Halverson, W.s. and Crabb, A.C, ed.,
Biological Control of Rodents
Proceedings of the 16th Vertebrate Pest
Conference, Santa Clara, California, 1-3
March 1994. Davis, University of California,
188-191.
Cowan, P.E. and Tyndale-Biscoe, CH. 1997.
Australian and New Zealand mammal
species considered to be pests or problems.
Reproduction, Fertility and Development, 9,
27-36.
Dalum, I., Jensen, M.R., Gregorius, K.,
Thomasen, CM.,Elsner,I.I. and Mouritsen, S.
1997. Induction of cross-reactive antibodies
against a self protein by immunization with a
modified self protein containing a foreign T
helper epitope. Molecular Immunology, 34,
1113-1120.
Davis, S.S. 1996. Vaccine delivery systems:
particulate delivery systems. Vaccine, 14,
672-80.
Dunbar, B.S. 1997. Contraception in domestic
and wild animal populations using zona
pellucida immunogens. In: Kreeger, T.J., ed.,
Contraception in wildlife management.
Technical Bulletin No. 1853, United States
Department of Agriculture, Animal and Plant
Health Inspection Service, 1-9.
East, LJ., Gulyas, B.]. and Dean, J. 1985.
Monoclonal antibodies to the murine zona
pellucida protein with sperm receptor activ-
ity: effects on fertilization and early develop-
ment. Developmental Biology, 109, 268-73.
East, I.J., Mattison, D.R. and Dean, J. 1984.
Monoclonal antibodies to the major protein of
murine zona pellucida: effects on fertilization
and early development. Developmental
Biology, 104,49-56.
Emlen, J.T., Stokes, A.W. and Winsor, CD. 1948.
The rate of recovery of decimated popula-
Hons of brown rats in nature. Ecology, 29,
133-145.
Fenner,F. and Buller, R.M.L.1997. Mousepox. In:
Nathanson, N., Ahmed, R., Gonzalez-
Scarano, F., Griffin, D.E., Holmes, K.,
Murphy, F.A. and Robinson, H.L., Viral
pathogenesis. Philadelphia, Lippincott-
Raven Publishers, 535-553.
Florman, H.M. and Wassarman,P.M.1985. 0-
linked oligosaccharides of mouse egg ZP3
237
Ecologically-based Rodent Management
account for its sperm receptor activity. Cell, 4,
313-324.
Gao, Y. and Short, R.V. 1993. The control of
rodent populations. Oxford Reviews of
Reproductive Biology, 15, 265-310.
Geddes,A.M.W.1992. The relative importance of
pre-harvest crop pests in Indonesia.
Chatham, UK, National Resources Institute
Bulletin, 47p.
Goldberg, E. and Shelton, J. 1986. Immuno-
suppression of fertility by LDH-C4. In:
Talwar, G.P., ed., Immunological approaches
to contraception and promotion of fertility.
New York, Plenum Publishing, 219p.
Gratz, N.G. 1994. Rodents as carriers of disease.
In: Buckle, A.P. and Smith, R.H., ed., Rodent
pests and their control. Wallingford, UK,
CAB International, 85-108.
Grundy (Chalmer), J.E., Mackenzie, JS. and
Stanley, N.F. 1981. Influence ofH-2 and non-
H-2 genes on resistance to murine cytomega-
lovirus infection. Infection and Immunity, 32,
277-286.
Guynn, D.C Jr. 1997. Contraception in wildlife
managment: reality or illusion? In: Kreeger,
T.J., ed., Contraception in wildlife manage-
ment. Technical Bulletin No. 1853, United
States Department of Agriculture, Animal
and Plant Health Inspection Service, 241-246.
Hardy, CM., Clarke, H.G., Nixon, B., Grigg, J.A.,
Hinds, L.A. and Holland, M.K 1997. Exami-
nation of the immunocontraceptive potential
of recombinant rabbit fertilin subunits in
rabbit. Biology of Reproduction, 57, 879-886.
Harris, J.D., Hibler, D.W., Fontenot, G.K, Hsu,
KT., Yurewicz, E.C and Sacco, A.G. 1994.
Cloning and characterization of zona pelluc-
ida genes and cDNAs from a variety of
mammalian species: the ZP A, ZPB and ZPC
gene families. DNA Sequence, 4, 361-393.
Herr, J.C, Flickinger, CJ., Homyk, M. and Klotz,
KJ.E. 1990a. Biochemical and morphological
characteristics of the intra-acrosomal antigen
SP-lO from human sperm. Biology of Repro-
duction,42,181-193.
238
Herr, J.c., Wright, M., John, E., Foster, J., Kays, T.
and Flickinger, c.J. 1990b. Identification of
human acrosomal antigen SP-lO in primates
and pigs. Biology of Reproduction, 42, 377-
382.
Holland, M.K, Andrews, J., Clarke, H., Walton,
C and Hinds, L.A. 1997. Selection of antigens
for use in a virus-vectored immunocontra-
ceptive vaccine: PH-20 as a case study. Repro-
duction, Fertility and Development, 9, 117-
124.
Holland, M.K. and Jackson, R.J. 1994. Virus-
vectored immunocontraception for control of
wild rabbits: identification of target antigens
and construction of recombinant viruses.
Reproduction, Fertility and Development, 6,
631-642.
Hone, J. 1992. Rate of increase and fertility
control. Journal of Applied Ecology, 29,695-
698.
Howard, W.E. 1967. Biocontrol and chemosteri-
lants. In: Kilgore, W.W. and Doutt, R.L., ed.,
Pest control-biological, physical and
selected chemical methods. New York,
Academic Press, 343-383.
Hudson, J.B. 1994. Mouse cytomegalovirus
(murid herpesvirus 1). In: Osterhaus,
A.D.M.E., ed., Virus infections of rodents and
lagomorphs. Amsterdam, Netherlands,
Elsevier Science, 85-117.
Hutchins, M., Stevens, V. and Atkins, N. 1982.
Introduced species and the issue of animal
welfare. International Journal of Studies on
Animal Problems, 3, 318-336.
Hutchins, M. and Wemmer, C 1987. Wildlife
conservation and animal rights: are they
compatible? In: Fox, M.W.and Michley, L.D.,
ed., Advances in animal welfare science.
Boston, Martinus Nijoff Publishers, 111-137.
Jackson, R.J., Maguire, DJ, Hinds, L.A. and
Ramshaw, LA. 1998. Infertility in mice
induced by a recombinant ectromelia virus
expressing mouse zona pellucida glycopro-
tein. Biology of Reproduction, 58,152-159.
Jochle, W. 1997. Prolactin in canine and feline
reproduction. Reproduction in Domestic
Animals,32,183-193.
Jones, W.R. 1994. Gamete immunology. Human
Reproduction, 9, 828-841.

Kearns, R 1993. The cost of the mouse plague.
Unpublished discussion paper presented to
the mouse plague forum, 12 August 1993.
Horsham, Victorian Institute of Dryland
Agriculture.
Kennelly, J.J. and Converse, KA 1997. Surgical
sterilization: an underutilized procedure for
evaluating the merits of induced sterility. In:
Kreeger, T.J., ed., Contraception in wildlife
management. Technical Bulletin No. 1853,
United States Department of Agriculture,
Animal and Plant Health Inspection Service,
21-28.
Kennelly, J.J., Johns, RE. and Garrison, M.V.
1972. Influence of sterile males on fecundity
of a rat colony. Journal of Wildlife Manage-
ment, 36, 161-165.
Key, G., Wilson, E. and Connor,J.1994. Present
status of Rattus norvegicus on Santa Cruz
Island, Galapagos, Ecuador. In: Halverson,
W.s. and Crabb, Ae., ed., Proceedings of the
16th Vertebrate Pest Conference, Santa Clara,
California, 1-3 March 1994. Davis, University
of California, 118-123.
Kirkpatrick, J.F., Liu, LM.K., Turner, J.W. Jr,
Naugle, Rand Keiper, R 1992. Long-term
effects of porcine zonae pellucidae immuno-
contraception on ovarian function in feral
horses (Equus caballus). Journal of Reproduc-
tion and Fertility, 94, 437-444.
Klein, J. 1979. The major histocompatibility
complex of the mouse. Science, 203, 516-521.
Knipling, E.F. and McGuire, J.u. 1972. Potential
role of sterilization for suppressing rat
populations: a theoretical appraisal. US
Department of Agriculture, Agricultural
Research Service Technical Bulletin, 1455, 1-
27.
Knudsen, S., Saadbye, P., Hansen, L.H., Collier,
A, Jacobsen, RL., Schlundt, J. and
Karlstreom,O.H. 1995. Development and
testing of improved suicide functions for
biological containment of bacteria. Applied
and Environmental Microbiology, 61, 985-
991.
Koffron, AJ., Mueller, K.H., Kaufman, D.B.,
Stuart, EP., Patterson, B. and Abecassis, M.I.
1995. Direct evidence using in situ polymer-
ase chain reaction that the endothelial cell and
Biological Control of Rodents
T-lymphocyte harbor latent murine cytomeg-
alovirus. Scandinavian Journal of Infectious
Diseases, 99, 61-62.
Kurtz, J.C 1987. Assessing the risks associated
with biotechnology: the regulatory problem.
In: Lave, L.B., ed., Risk assessment and
management. New York, Plenum Press, 93-
98.
Leirs, H., Verhagen, R, Sabuni, e.A, Wanjabe,
P.M. and Verheyen, W.N. 1997. Spatial
dynamics of Mastomys natalensis in a field-
fallow mosaic in Tanzania. Belgian Journal of
Zoology, 127,29-38.
Lou, Y., Ang, J., Thai, H., McElveen, F. and Tung,
K.S.K. 1995. A zona pellucida 3 peptide
vaccine induces antibodies and reversible
infertility without ovarian pathology. Journal
of Immunology, 155,2715--2720.
Manning, W.e. and Mocarski, E.S. 1988. Inser-
tional mutagenesis of the murine cytomega-
lovirus genome: one prominent a gene (ie2) is
dispensable for growth. Virology, 167,477-
484.
Marks, CA, Nijk, M., Gigliotti, F., Busana, F.
and Short, R.V.1996. Preliminary field assess-
ment of a cabergoline baiting campaign for
reproductive control of the red fox (Vu/pes
vu/pes). Wildlife Research, 23,161-168.
Marsh, RE. 1988. Chemosterilants for rodent
controL In: Prakash, L, ed., Rodent pest
management. Boca Raton, Florida, CRC
Press, 353-368.
Marsh,RE.andHoward, W.E.1973.Prospectsof
chemosterilants and genetic control of
rodents. Bulletin of the World Health Organi-
sation, 48, 309-316.
Meehan, AP. 1984. Rats and mice. Their biology
and control. Sussex, Rentokil Limited, 383p.
Millar, S.E., Chamow, S.M., Baur, A.W., Oliver,
e.,Robey,F. and Dean,J.1989. Vaccination
with a synthetic zona pellucida peptide
produces long-term contraception in female
mice. Science, 246, 935-938.
Molak, V. and Stara, J. 1987. Risk assessment of
deliberate release of genetically engineered
microorganisms. In: Lave, L.B., ed., Risk
assessment and management. New York,
Plenum Press, 83-91.
239
Ecologically-based Rodent Management
Moors, P.}., Atkinson, AE. and Sherley, C.H.
1992. Reducing the rat threat to island birds.
Bird Conservation International, 2, 93-114.
Nettles, V.F. 1997. Potential consequences and
problems with wildlife contraceptives.
Reproduction, Fertility and Development, 9,
137-143.
Oogjes, C. 1997. Ethical aspects and dilemmas of
fertility control of unwanted wildlife: an
animal welfarist's perspective. Reproduction,
Fertility and Development, 9, 163-167.
Pennycuik, P.R, Johnston, P.C., Lidicker, W.Z. Jr
and Westwood, N.H. 1978. Introduction of a
male sterile allele (tW2) into a population of
house mice housed in a large outdoor enclo-
sure. Australian Journal of Zoo [ogy, 26, 69-
81.
Pollock, J.L., Presti, RM., Paetzold, S. and Virgin,
H.W. 1997. Latent murine cytomegalovirus
infection in macrophages. Virology, 227, 168-
179.
Prakash, I. 1988a. Rodent pest management.
Boca Raton, Florida, CRC Press, 480p.
Prakash,1. 1988b. Bait shyness and poison
aversion. In: Prakash, 1., ed., Rodent pest
management. Boca Raton, Florida, CRC
Press, 321-329.
Primakoff,P., Lathrop, W., Woolman, L.,Cowan,
A and Myles, D. 1988. Fully effective contra-
ception in male and female guinea pigs
immunized with the sperm protein PH-20.
Nature, 335, 543-546.
Ramsay, AJ. and Ramshaw, LA 1997. Cytokine
enhancement of immune responses impor-
tant for immunocontraception. Reproduc-
tion, Fertility and Development, 9, 91-97.
Regal, P.J. 1986. Models of genetically
engirleered organisms and their ecological
impact. In: Mooney, H.A and Drake, rA, ed.,
Ecology of biological invasion of North
America and Hawaii. New York, Springer-
Verlag, 111-129.
Rhim, SH., MilIar, S.E., Robey, F., Luo, AM.,
Lou, Y.H., Yule, T., Allen, P., Dean, J., and
Tung, K.S.K. 1992. Autoimmune disease of
the ovary induced by a ZP3 peptide from the
mouse zona pellucida. Journal of Clinical
Investigations, 89, 28-35.
240
Rosiere, T.K. and Wassarman, P.M. 1992. Identi-
fication of a region of mouse zona pellucida
glycoprotein mZP3 that possesses sperm
receptor activity. Developmental Biology,
154,309-317.
Sankai, T., Endo, T., Kanayama, K., Sakuma, Y,
Umezu, M. and Masakai, J. 1991. Antiproges-
terone compound RU486 administration to
terminate pregnancy in dogs and cats. Journal
of Veterinary Medical Science, 53, 1069-1070.
Saunders, G. and Cooper, K. 1981. Pesticide
contamination of birds in association with a
mouse plague. Emu, 82, 227-229.
Shellam, C.R 1994. The potential of murine
cytomegalovirus as a viral vector for
immunocontraception. Reproduction, Fertil-
ity and Development, 6, 401--409.
Siddhanti, S.K. 1987. Methodological approach
to the study of risk policy decision-making:
the case of deliberate release of genetically
engineered organisms. In: Lave, L.B., ed.,
Risk assessment and management. New
York, Plenum Press, 99-106.
Sinclair, ARE. 1989. Population regulation in
animals. In: Cherrett, J.M., ed., Ecological
concepts. Oxford, UK, Blackwell Scientific
Publications, 197-241.
Sindair, A.R.E. 1997. Fertility control of mammal
pests and the conservation of endangered
marsupials. Reproduction, Fertility and
Development, 9, 1-16.
Sinclair, ARE, Olsen, P.D. and Redhead, T.D.
1990. Can predators regulate small mammal
populations? Evidence from house mouse
outbreaks in Australia. Oikos, 59, 382-392.
Singer, P. 1997. Neither human nor natural:
ethics and feral animals. Reproduction, Fertil-
ity and Development, 9, 157-62.
Singleton, C.R 1989. Population dynamics of an
outbreak of house mice (Mus in
the mallee wheatlands of Australia
hypothesis of plague formation. Journal of
Zoology, London, 219, 495-515.
Singleton, G.R 1994. The prospects and associ-
ated challenges for the biological control of
rodents. In: Halverson, W.S. and Crabb, AC.,
ed., Proceedings of the 16th Vertebrate Pest

Conference, Santa Clara, California, 1-3
March 1994. Davis, University of California,
301-306.
Singleton, G.R. and Petch, D.A1994. A review of
the biology and management of rodent pests
in Southeast Asia. AClAR Technical Reports
No. 30. Canberra, Australian Centre for Inter-
national Agricultural Research, 65p.
Singleton, G.R and Redhead, T.D. 1989. House
mouse plagues. In: Noble, J.C and Bradstock,
RA., ed., Mediterranean landscapes in
Australia: mallee ecosystems and their
management. Melbourne, Commonwealth
Scientific and Industrial Research Organisa-
tion (CSIRO), 418-433.
Singleton, G.R, Smith, AL., Shellam, G.R,
Fitzgerald, N. and Muller, W.J. 1993. Preva-
lence of viral antibodies and helminths in
field populations of house mice (Mus domesti-
eus) in southeastern Australia. Epidemiology
and Infection, 110,399-417.
Singleton, G.R and Spratt, D .M. 1986. The effects
of Capillaria hepatiea (Nematoda) on natality
and survival to weaning in BALB / c mice.
Australian Journal of Zoology, 34,677-681.
Skinner, S.M., Mills, T., Kirchik, H.J. and Dunbar,
B.s. 1984. Immunization with zona pellucida
proteins results in abnormal ovarian follicu-
lar differentiation and inhibition of gonado-
tropin induced steroid secretion.
Endocrinology, 115,2418-2432.
Smith, AL., Singleton, G.R., Hansen, G.M. and
Shellam, G. 1993. A serologic survey for
viruses and Mycoplasma pulmonis among wild
house mice (Mus domesticus) in southeastern
Australia. Journal of Wildlife Diseases, 29,
219-229.
Spratt, D.M. and Singleton, G.R. 1986. Studies on
the life cycle, infectivity and clinical effects of
Capillaria hepatica (Bancroft) (Nematoda) in
mice,Mus musculus. Australian Journal of
Zoology, 34, 663-675.
Stenseth, N.C 1981. How to control pest species:
application of models from the theory of
island biogeography in formulating pest
control strategies. Journal of Applied
Ecology, 18, 773-794.
Biological Control of Rodents
Stohr, K. and Meslin, F.X.1997. Zoonoses and
fertility control in Wildlife-requirements for
vaccines. Reproduction, Fertility and Devel-
opment, 9,149-155.
Szostak M.P., Hensel, A, Eko, F.O., Klei, R,
Auer, T., Mader, H., Halsberger, A., Bunka,S.,
Wanner, G. and Lubitz, W. 1996. Bacterial
ghosts: non-living candidate vaccines.
Journal of Biotechnology, 44, 161-170.
Tedin, K., Witte, A., Rei singer, G., Lubitz, W. and
Bleasi, U. 1995. Evaluation of the E. coli ribos-
omal rrnB PI promoter and phage-derived
lysis genes for the use in a biological contain-
ment system: a concept study. Journal of
Biotechnology, 39, 137-148.
Tiedje, J.M., Colwell, RK., Grossman, Y.L.,
Hodson, RE., Lenski, RE., Mack, RN. and
Regal P.J, 1989. The planned introduction of
genetical] y engineered organisms: ecological
considerations and recommendations.
Ecology, 70,298-315.
Twigg, L.E., Singleton, GR and Kay, B.J. 1991.
Evaluation of bromadiolone against house
mouse (Mus domesticus) populations in
irrigated soybean crops. I. Efficacy of controL
Wildlife Research, 18, 265-274.
Tyndale-Biscoe, CH. 1994. Virus-vectored
immunocontraception of feral mammals.
Reproduction, Fertility and Development, 6,
281-287.
Tyndale-Biscoe, CH. 1995. Vermin and viruses:
risks and benefits of viral-vectored immunos-
terilisation. Search, 26, 239-244.
Tyndale-Biscoe, CH. 1997a. Immunosteriliza-
tion for wild rabbits: the options. In: Kreeger,
T.J., ed., Contraception in wildlife manage-
ment. Technical Bulletin ).,[0.1853, United
States Department of Agriculture, Animal
and Plant Health Inspection Service, 223-234.
Tyndale-Biscoe, CH. 1997b. Summing up the
conference. Reproduction, Fertility and
Development, 9,183-186.
Vickery, B.H., McRae, GI., Goodpasture,J.C and
Sanders, L.M. 1989. Use of potent LHRH
analogues for chronic contraception and
pregnancy termination in dogs. Journal of
Reproduction and Fertility Supplement, 39,
175-187.
241
Ecologically-based Rodent Management
Wace,N.M.1986. The rat problem on oceanic
islands-research is needed. Oryx, 20, 79-86.
Wasser, S.K. and Barash, D.P. 1983. Reproduc-
tive suppression among female mammals:
implications for biomedicine and sexual
selection theory. Quarterly Review of
Biology, 58, 513-538.
Whittingham, D.G. and Wood, M.J.1983. Repro-
ductive physiology. In: Foster, H.L., Small,
J.D. and Fox, I.G., ed., The mouse in biomedi-
cal research, volume HI - normative biology,
immunology and husbandry. New York,
AcadernicPress,137-164.
242
Williams, c.K. 1997. Development and use of
virus-vectored immunocontraception.
Reproduction, Fertility and Development, 9,
169-178.
Williams, c.K. and Twigg, L.E. 1996. Responses
of wild rabbit populations to imposed steril-
ity. In: Floyd, R.B., Sheppard, AW. and De
Barro, P.J., ed., Frontiers of population
ecology. Melbourne, Commonwealth Scien-
tific and Industrial Research Organisation
(CSIRO),547-560.
Bruce A. Colvin and William B. Jackson

Abstract

Urban rodent control in the 21 st century must focus on a program approach that is
both strategic and comprehensive (i.e. proactive rather than reactive). This includes
an integrated pest management approach that incorporates long-term planning,
scheduling, data management and mapping capabilities . It also must include
greater partnership among municipal agencies, private pest control companies and
community groups. Central to program success will be coordination, communication
and accountability among all program participants. Cost-effectiveness will be
achievable but predicated on effective administrative management, training, and
understanding of the ecological and political complexities of urban environments.
Greater focus on sanitation enforcement, infrastructure maintenance and
construction will be essentral for long-term removal of causal factors . The long-term
goal must be an effective and sustainable program.

Keywords

Urban, integrated pest management (IPM), rodent control, commensal rodent,

san itation, i nfrastructu re

243
Ecologically-based Rodent Management
INTRODUCTION
of urban rodent
T
HE PRINCIPLES
control have been well researched
and described in this century.
However, substantial problems continue to
persist and grow in many metropolitan
areas, as well as in small municipalities.
Although the science and technology have
been established (Jackson 1982; Frantz and
Davis 1991), the failure appears most evident
at the point of implementation. Success is
not predicated on a control tool, but rather
on coordinated efforts supported by
technical leadership at the local level.
Telle (1969) in Germany and Myllymaki
(1969) in Finland described the principle of
establishing a 'rat-free town'. The goal was
to have less than 1% of the premises showing
signs of rat activity. Drummond (1970) in
England stressed the idea and importance of
a program approach for managing urban rat
populations (Drummond et al. 1972, 1977;
Drummond 1985). Examples of successful
and coordinated programs are few, but
include Budapest, Hungary (Gacs et al. 1977;
Bajomi and Sasvari 1986), Kuwait (AI Sanei
et al. 1986), and Denmark (P. Weile 1998,
pers. comm.). All of these have included an
emphasis on sanitation and environmental
management.
The primary difference between urban
and agricultural rodent control is that the
urban environment is relatively diverse and
stable, requiring consistent application of
control measures, since food and structural
resources are consistently available. On the
other hand, in open agricultural settings, the
environment is relatively homogeneous and
subject to disruptions, and resources used by
244
rodents and control efforts tend to be
seasonally timed. Habitat manipulation is
readily feasible in urban areas for limiting
growth of rodent populations and to prevent
over-dependency on rodenticide. In
contrast, with agricultural situations there
may be limited opportunity for habitat
manipulation and thus greater emphasis
frequently is given to trapping and
rodenticide use. The basic ecological and
organisational principles of rodent control
programs, however, are transferable
between urban and agricultural
environments.
Urban rodent control in the United States
(US) typically is implemented in a limited or
disjointed fashion, rather than being
comprehensive or coordinated. Programs
commonly are reactive rather than
proactive. Reasons for this include limited
funding, training, political and teclmical
support, and organisation (Howard 1984).
The political and scientific interest in urban
rodent control in the US currently is low,
although the need appears to be great.
Commensal rodents are those described
literally as 'feeding at our table'. They
include species such as Norway rat (Rattus
norvegicus), roof rat (Rattus rattus), and
house mouse (Mus musculus). In Asia, they
also can include species such as the lesser
bandicoot rat (Bandicota bengalensis) and
Polynesian rat (Rattus exulans). Commensal
rodents have been associated with a variety
of diseases, contamination and destruction
of stored foods, structural damage, and
other aspects of environmental deterioration
(Gratz 1994; Lund 1994). They frequently
display remarkable adaptive behaviour in
urban environments.

Urban infrastructure is ageing,
congestion is increasing, and urban habitat is
expanding worldwide. These factors
accentuate the growing need for effective
rodent control programs, for public health,
economic and aesthetic reasons.
Additionally, expectations of urban
residents and businesses for quality-of-life
improvements and effective public health
management will continue to grow. The
result will be more sanitation problems to be
managed in more densely populated urban
centres, and these urban environments will
increasingly require re-design and
construction to support human population
levels and business economies.
The purpose of this chapter is to describe
strategies and issues for implementing
urban rodent control programs, while
considering both ecological and
administrative components. This includes a
historic US perspective with transition to
future opportunities worldwide.
HISTORICAL BACKGROUND
The 'Modern Rodent Control Era' began
with the advent of World War I1, as urban
destruction and concern for food supplies
(both quality and quantity) forced attention
to rodent control. This included the need for
more effective rodenticides and better
understanding of rodent biology. (The term
'ecology' had not yet been applied to this
issue.) Federal funds in the US supported
related research and development projects.
Initial efforts focused at the School of
Hygiene and Public Health of the Johns
Hopkins University in Baltimore, Maryland.
The rodenticide ANTU was brought forth
there, and Compound 1080 resulted from a
massive synthesis and evaluation program
Urban Rodent Control Programs
by the federal government. These
rodenticide baits were far more effective
than the arsenic and phosphorus baits in use
at that time. (It was not until the next decade
that warfarin would leap into the
marketplace as a 'miracle' rodenticide,
following the 'miracle' insecticide, DDT.)
The 1940s and early 19505 atJohns
Hopkins University were times of active
research on basic elements of rat behaviour,
later serving as the foundation of control
approaches and future urban programs
worldwide. Calhoun (1948,1963) performed
detailed studies of the sociology of the
Norway rat, providing new research
approaches for studying the social behavior
of semi-confined rodent populations. Basic
studies of Norway rats in the residential
blocks of Baltimore followed by Davis and
others (Davis et a1. 1948; Emlen et a1. 1948,
1949; Davis and Fales 1949,1950; Davis
1953). Those studies documented for the first
time the reproduction, movement, and
many other life history components (now
referenced as 'ecology') of urban rat
populations. Parallel studies on aspects of
commensal rodent biology also occurred in
the United Kingdom during the same time
era (Chitty and Southern 1954).
From the fundamental studies at Johns
Hopkins University came the notion of
relating a logistical growth curve (sigmoid
curve) to changes in rodent populations
(Figure 1). Such a mathematical expression
could then be used to predict population
growth and as a tool for understanding how
to strategically manage rodent populations.
By lowering the carrying capacity of the
environment, the rate of population increase
could be dampened and population declines
achieved (Davis and Jackson 1981).
245
Ecologically-based Rodent Management
t
<D
N
'iij
C
0
ia
"5 (b)
0..
0
0..
Time ---to--
Figure 1.
Sigmoid curves depicting the rate of change and
growth of rodent populations over time, (a)
without environmental (sanitation) management
and (b) with environmental management to
reduce carrying capacity. The two dashed lines on
each graph show the passing of the same amount
of time on the horizontal axis, while on the vertical
axis there is a dramatic increase in the population
size during the second time period. The most
economic and effective strategy Is to manage
populations at the low end of the sigmoid curve
under reduced carrying capacity (b).
Populations then could be more
effectively 'managed' at the low end of the
population growth curve. Baltimore officials
cooperated and set up a demonstration area
in which no rodenticides would be used, but
intensive enforcement of environmental
246
standards would be substituted. Specific
regulations were passed and sanitation
police were hired. Outdoor toilets, board
fences, old cars and other rubbish were
removed. Proper storage and handling of
refuse was enforced. With this regimen, the
outside rat population was virtually
eliminated from this residential area.
However, the program was not maintained
because of the need for intensive (i.e.
expensive) site management and the lack of
political and personal will to maintain
environmental standards. Yet, the logistical
model as a management tool had been
demonstrated successfully and, to one
degree or another, became the basis of future
management efforts in the US and elsewhere
(Figure 2).
During World War II, the US Public
Health Service set up programs in the
Communicable Disease Center (CDq in
Atlanta, Georgia, for training personnel in
the control of various disease vectors met in
military activities. In post-war years, such
training programs expanded to include state
and municipal personnel. Equipment was
provided for state as well.
Handbooks covering a wide variety of
vector and sanitation topics were developed,
and many remain as the prime resource
documents available today (Pratt and
Johnson 1975; Pratt and Brown 1976; Pratt et
a1. 1976; Scott and Borom 1976; Davis et aL
1977). However, the involvement of CDC in
technical support for urban rodent control
programs faded by the early 1980s and
personnel once active have retired.
Universities were slow to pick up the
challenge of urban rodent ecology and
controL The Johns Hopkins program faded
when the core staff left by the late 1950s.
 Urban Rodent Control Programs

160
Poison Poison Poison

140

120
t II
tt 1ISa'. . ,
Poison

IJ)
100
e
'0
Q;
.0
80 I
E
:::l
Z
60 f-

40

20

1943 1944 1945 1946 1947 1948 1949 1950

Figure 2.
Changes in Norway rat abundance in Baltimore following poisoning and sanitation Improvements (Jackson
1998). The population Increase during late 1948 was attributable to a strike by garbage collectors that
temporarily Increased food resources (carrying capacity) for rats.

Most mammalian researchers considered programs. Designations such as 'economic

urban pest rodents not 'worthy' of basic
research and the urban environment a
distasteful setting for studying mammalian
ecology. This left entomologists as the source
of much information for pest control
curricula, and they commonly did not want
to bother with rodents or approached them
as 'furry cockroaches'.
Two universities, Bowling Green State
University in Ohio and the University of
California at Davis, developed effective
vertebrate pest curricula in the 1960s and
19705. This included undergraduate and
graduate instruction and research
biology' were typical, and research
continued the kinds of efforts first begun at
Johns Hopkins University. This included
basic behaviour but also, by the 1970s and
1980s, a focus on genetic resistance to
anticoagulant rodenticide, secondary
poisoning hazards, and development of
new rodenticides, traps and bait stations.
Many personnel currently providing
leadership on vertebrate pest management
in the US came from these two university
programs. However, by the late 1980s these
programs also began to fade and ended as
faculty retired.

247
Ecologically-based Rodent Management
The Federal government, through the
US Public Health Service, supported
research on genetic resistance to warfarin
(first-generation anticoagulant
rodenticide) during the 1970s (Jacks on et
al. 1988). During that same era and into the
early 1980s, $12.8 to $15.0 million dollars
annually were provided for state and local
efforts specifically for implementing urban
rodent control programs. These programs
involved a standardised approach of
environmental management, including
target areas, systematic surveys,
education, sanitation improvements and
baiting. The goal was to progressively shift
blocks of premises to a maintenance
condition with subsequent monitoring for
re-infestation. This federal program
assisted more than 100 communities
(Jacks on 1984, 1998). However, in the early
1980s, specific designation of federal
funding for urban rodent control ended,
and subsequently many local programs
were greatly reduced and eventually
became less structured.
The US Fish and Wildlife Service, through
regional and state programs in the 1960s-
1980s, stimulated and supported efforts at
universities as related to vertebrate pests.
This included some technical support on
urban rodents, although most of their effort
was directed at predators, birds and field
rodents. In the mid 1980s, these programs
were transferred by the US Congress to the
US Department of Agriculture (USDA). With
that change, control of urban rodents was
specifically excluded from the USDA scope of
services, although rodent infestation of stored
grain and croplands remain within their area
of concern.
248
Today in the US, technical support for
urban rodent control is very limited. Active
research programs do not exist on the federal,
state or university level. Most States and
municipalities have limited knowledge and
skill in urban rodent control, resulting in
limited effectiveness when programs are
implemented. Urban (and suburban) rodent
control could be described as a composite of
work by householders, licensed pest control
operators and municipal agencies
(Kaukeinen 1994), but without defined
coordination. Most efforts by local health
departments are reactive, in response to
citizen complaints about infestations or rat
bites. Their efforts typically involve use of
anticoagulant baits within a limited area (e.g.
at the site of a complaint or perhaps a few city
blocks).
THE BOSTON MODEL
The basic research of the 1940s-1950s, the
program approaches described in the 1960s-
1970s, and the technical advances of the
1970s-1980s were combined in 1990 in
Boston, Massachusetts. The unique
opportunity to establish a truly
comprehensive and properly designed
rodent control program had arisen from the
start of an ll-billion-dollar highway
construction project funded by the Federal
Highway Administration (Colvin et al.
1990). This involved reconstruction of the
urban infrastructure along a seven-mile
route, including utility systems and an 8-10
lane highway to be built underground.
Key to the program design was a
centralised approach with well-defined
responsibilities and firm accountability. The
primary management function was
performed by personnel (biologists) skilled

in technical aspects of rodent control, yet
also with contract management, public
relations, engineering, scheduling, and
computer-based mapping and data
management skills. The second component
of the program was the municipal functions,
performed by the Inspectional Services
Department, the Code Enforcement Police,
the Water and Sewer Commission, and the
Public Works Department. The third
component involved pest control contractors
who performed poison baiting, trapping and
monitoring. The fourth component was
public participation, championed by
community leaders and organisations. These
various components were integrated to
maximise the skills and participation of each
group within the total program.
Work tasks assigned to each municipal
agency were based on their existing scope of
services. For example, the Inspectional
Services Department was assigned
standardised surveys of premises,
enforcement of State Sanitary Code and
public education. The Code Enforcement
Police dealt with violations of City sanitation
ordinances. The Water and Sewer
Commission assisted by cleaning catch
basins and providing access to sewer
manholes. The Public Works Department
helped by making infrastructure repairs and
maintaining trash receptacles in public
areas. Most of these municipal tasks focused
on environmental change to reduce rodent
habitat.
The program was an integrated pest
management (IPM) approach covering
about a seven-square-mile area. It was
tailored to the need of each neighborhood
based on surveys of sanitary conditions and
rodent activity at surface and subsurface
Urban Rodent Control Programs
levels. Where problems were chronic, IPM
methods and monitoring were applied more
intensely. The program included an
extensive public outreach and education
campaign involving community meetings,
diverse literature, videos, door-to-door
contact and school presentations. The
education campaign recognised cultural
differences among neighbourhoods, and
literature was prepared in multiple
languages.
Sanitation was given heightened
attention on more than 10,000 premises in
the project area. The Code Enforcement
Police performed ticketing daily, and the
Inspectional Services Department cited
property owners to hearings and court for
chronic sanitary code violations. In locations
highly susceptible to infestation, residents
and businesses were given refuse containers
after signing a contract to maintain and use
them. Property owners and businesses were
held responsible for maintaining their
property in an acceptable condition,
including hiring their own pest control
contractor if needed. City personnel
conducted baseline and periodic surveys on
private properties, and these surveys helped
ensure maintenance of environmental
improvements. The objectives were to
reduce the environmental carrying capacity
for rodents and also to prevent their
dispersal.
Trapping and poisoning were performed
on both surface and subsurface levels in all
public areas, and as a supportive measure on
some private properties. Engineering
drawings and information were used to
'three-dimensionally' dissect the
infrastructure. About 1,500 sewer and other
types of utility manholes were baited using
249
Ecologically-based Rodent Management
pulsed-baiting methods on a seasonal basis
(Colvin et a1. 1998). Census and monitoring
tools were used extensively to detect and
closely monitor for low levels of activity
(non-toxic census baits, tracking tiles, night-
time visual surveys and tracking with snow
cover). Night-time observations were
essential for targeting control resources and
identifying sanitary problems that only
appeared after dark (e.g. plastic rubbish
bags left out on sidewalks).
Administrative elements were
emphasised as part of the IPM program.
Contract specifications for both pest control
contractors and municipal agencies were
developed (Colvin et al. 1992). This provided
a basis for accountability and performance.
Program elements were centrally scheduled,
to maximise timing and geographical
distribution of resources. Data management
and mapping using a geographic
information system was important for
efficiency of operations (Von Wahlde and
Colvin 1994); this included tracking of
events over the entire project area, historic
patterns of problems and poison placements
in utility systems. Standardised data sheets
were used to collect information for program
planning and analysis. Data sheets also
helped ensure that personnel were
performing their assigned tasks. Cost
containment was predicated on maintaining
a proactive rather than haphazard or
reactive approach.
An important administrative element
that characterised the dynamics of the
program was a referral system. All members
of the program team were cross-trained to
identify program issues, even though the
particular issue might involve a task
assigned to another team member (agency or
250
contractor). If a pest control contractor
observed a sanitary problem on private
property, the contractor could refer that
observation through the centralised
program office to the city agency responsible
for enforcement. Similarly, observations by a
city agency of a rat infestation needing
treatment could be referred to a pest control
contractor assigned to the particular
geographic region. All referrals were tracked
on a database to ensure they were followed
through.
The program also entailed habitat
modification beyond basic rodent-proofing
of structures (i.e. sealing of holes and entry
points). Landscaping within the project area
was evaluated and factors conducive to
rodent activity were identified (Colvin et al.
1996). Rodent-proofing principles were then
incorporated into design specifications for
final surface restoration and landscaping.
For example, rat abundance in landscaped
areas was associated with the amount of
coverage by shrubbery. Plots with massed
needled evergreens were most susceptible to
infestation, in contrast to plots with broad-
leafed evergreens and deciduous shrubs.
Numerous property owners and businesses
re-designed their existing landscaping,
reducing densely planted needled-
evergreens and giving greater emphasis to
stone mulch, shrubbery spacing, refuse
containers, maintenance, and trees and
shrubs that did not produce excessive fruit.
Varieties that grow in vase-like, rather than
mounded, shapes provided more openness
within landscaped areas.
Results of the control program can be
characterised in several ways. Brick sewers
with pipe diameter <61 cm in residential
areas had the highest subsurface rat activity;

up to 38% of those manholes were active at
the program start (baseline). Bait
consumption and the number of active
sewer holes were 96% and 87% below
baseline, respectively, when seasonal baiting
was last initiated in 1997. The Code
Enforcement Police in 1996 were issuing 67"'/0
fewer tickets for sanitation violations than in
1991, even with intensified efforts. About
375 properties were identified with rat
activity during baseline surveys; by 1997,
control efforts were needed on about ten of
those properties 4% of those originally
identified). Another 25-30 properties
required close monitoring to detect potential
re-infestation. Referrals by the program's
pest control contractors to the City for
resolution of sanitary problems and rat
activity on private property declined from a
high of 153 in 1993 to 13 in 1995 and 20 in
1996 (87% reduction). Visual night-time
surveys for rat activity declined from a
localised average event of 104 sightings per
hour (range 19-500) to incidental
observations within the project area (>99%
reduction). These 'real world' statistics
collectively demonstrated that the program
approach was effective.
Throughout the program, phone calls
from the public concerning complaints
about rodents and sanitation remained
relatively constant, independent of the
obviously positive environmental changes
that were occurring. However, the
magnitude of problems identified by the
public grew less as the program continued.
The relatively constant input from the public
was a positive event, since it demonstrated
sustained participation by the communities.
This was achieved through the consistent
presence of program staff in the
Urban Rodent Control Programs
neighbourhoods, timely and effective
response to public concerns, involvement of
community leaders, and repetitive positive
feedback to political entities by program
managers and the public. Neighbourhood
'cleanup days' evolved with the
participation of residents and businesses,
and awards and recognition were given. The
pattern of success, involvement, and
commitment broke the public scepticism
that commonly degrades the opportunity to
sustain an urban rodent control program.
The Boston model had several elements
that proved crucial for success. Foremost
was a true partnership among City agencies,
community groups, and contracted
management and pest control firms. The
model demonstrated that a blend of
municipal and privatised functions worked
well and could succeed, but only with
technical leadership, open sharing of
information, consistent communication
among team members, and trust. It had to be
one team focused on an IPM strategy, with
excellent diversity of skills and
accountability on the part of each team
member. Weekly meetings within various
program groups and quarterly team
meetings enhanced training and
communication.
FUTURE OPPORTUNITIES
The future of urban rodent control is not
limited by science and technology; it is
limited by politics and bureaucracies.
Personnel management, budgets, contracts,
news media, legal proceedings and political
agendas were not facto red together when
Davis and others performed the original
ecological research in Baltimore, but they
251
Ecologically-based Rodent Management
must be today for the science of rodent
control to be implemented. When Davis
(1972) summarised rodent control in context
of future strategies, he expressed frustration
with the political impedance of urban rat
control and use of short-term solutions. He
reaffirmed the need to focus on basic
biological principles (i.e. reducing carrying
capacity) and the need for competent
administrators.
Urban programs need to be consistently
and strategically managed rather than
politically cyclic in their implementation and
focus. Lethal measures need to be intensive
rather than simply cropping populations
and spurring higher reproductive rates that
occur with lower competition. In other
words, many urban programs function
today on the steep slope of the sigmoid
growth curve (Figure 1). A temporary
lowering of the number of animals is
achieved by a punctuated control effort, but
subsequently a sudden population rebound
occurs. This sudden perturbation of rodents
often is misinterpreted as a 'new' population
of colonists. Whereas in reality, the
population may never have been effectively
controlled to start with and has simply
responded reproductively as the sigmoid
curve predicts.
There must be a commitment to long-
term management of the urban environment
and an organisational structure to achieve
closure of issues day-by-day. This type of
preventative approach is actually a more
cost-effective (economical) approach long-
term than chronically reacting to crises and
public complaints. Of course, any program
also must have the capability to respond
quickly to sudden problems and
252
emergencies (e.g. disease, rat bites or
localised outbreaks).
Rats need to be viewed as an 'indicator
species' of environmental quality (or
degradation), and programs need to focus
on causal factors for species success rather
than simply being reactive and poison
dependent. The goal must be to manage
populations at the low end of the sigmoid
growth curve by reducing carrying capacity
and giving greater emphasis to surveillance
monitoring and sanitation controls. A
behavioural shift from rat hunting to
environmental management and monitoring
is needed. This represents an ecologically-
based strategy.
The kind of organisational management
and rPM methods demonstrated in the 1990s
in Boston should become the foundation of
program implementation in the 21 st century.
An effective program will include
centralised leadership, partnerships among
participants, sound definition of work scope,
assigned responsibilities, mapping and data
management capabilities, and education of
policy makers. A program approach should
be comprehensive in scope and structure,
and inclusive in terms of participants
(Figure 3).
An emphasis on the engineering and
structural maintenance of urban
environments, to reduce pest habitat and
permanently lower carrying capacity, is
critically needed as part of public health
management. Rodent control should be
incorporated into both urban planning
(design) and urban maintenance if a truly
pro active program is desired. Municipalities
also should require rodent control for major
construction projects, since by their nature
they create rodent habitat during the

Code enforcement/Legal
"
.._--1.....
Sewer department ......
/
Neighbourhood services
FIgure 3.
A flow chart that Illustrates a centralised, inclusive and organised approach to managing an urban rodent
control program.
construction phase. Biologists need to
become more familiar with design
engineering and urban infrastructure, and
work with engineers and architects.
Basic research and stewardship on the
use of refuse containers, recycling programs,
and refuse management and containment
should be a priority. A shift from focusing
primarily on rodenticide and traps (reactive
approach) to sanitation management
(proactive approach) is needed in urban
environments. This includes use of rodent-
proof containers (rubbish bins, dumpsters,
refuse compactors and grease containers),
public education campaigns, reduction of
refuse volume, and uniform refuse
containers and pickup times. Foremost,
there must be effective and enforceable
sanitation laws and practices established.
For example, keeping lids securely closed on
containers, not using plastic bags alone for
refuse storage, avoiding placement of refuse
outside overnight the day before collection,
and timing necessary night-time collections
to minimise the number of hours that refuse
is exposed.
Urban Rodent Control Programs
Pest control contractors
......._--1..... Public works department
Community groups
Inclusion of subsurface environments
(sewers), currently ignored by most
municipalities, must become a fundamental
part of urban rodent control programs
(Figure 4). Subsurface populations function
as reservoirs that can chronically re-infest
surface areas and lelevate' disease
organisms. Effective training on subsurface
baiting and control is needed worldwide
and will have to be incorporated into
comprehensive training necessary for the
success of any program.
Use of rodenticide in the future will have
to be more carefully planned and
implemented than typically occurs in most
cities today. The spread of genetic resistance,
including the recent involvement of second-
generation anticoagulant compounds
(MacNicoll et a1. 1996), presents a significant
concern for the future of urban rodent
control and public health management.
Rodenticide, rather than habitat
management, too often is the primary
approach used today by municipalities. This
provides short-term resolution but may
result in long-term problems.
253
Ecologically-based Rodent Management

administrators, computer scientists,

Material
Brick
property managers, architects and social

t
Pipe size
Concrete, Clay, PVC

~
No action
workers must be made part of the immediate
or extended team to maximise program
effectiveness.
A rodent control program should have an

<61cm ~ 'action plan' that defines program elements

and implementation. The action plan should
> 61 cm
describe: program organisation, staff
responsibilities, training, legal requirements
1
Environment
NoL and enforcement capabilities, contracted
services, schedules, sanitation management,
infrastructure design and maintenance,
mapping and data management, lethal
Residential Mixed residentiaV Commercial measures (including subsurface), public
Commercial Industrial education and outreach, community
organising, surveys and monitoring, and
administrative elements.

1
Bait: Primary
1
Bait: Secondary
1
No action
Figure 4.
An example of how to strategically Implement
and prioritise a sewer baiting campaign (Colvin et
al. 1998).
Excessive use of rodenticide, while
allowing sanitary problems to remain,
presents a condition for rapid reproduction
by survivors and the perpetuation of genetic
qualities favourable to resistance. The
strategy for lowering the potential for
resistance must be to focus on
environmental management and the wise
use of rodenticide.
Diversity of skills should be sought when
establishing a rodent control team. For
example, a creative approach for public
education may require the involvement of
public relations and marketing experts.
Teachers, lawyers, engineers, contract
Whether in a developed or developing
nation, the ecological and organisational
principles associated with urban rodent
control are largely the same. The difference
may be the extent of program resources
available; however, it can not be assumed
today that a developed nation automatically
has an advantage regarding program
implementation. Large, developed cities can
provide more habitat, have greater
bureaucracy, but have no better technical
knowledge than found in smaller cities or
cities in developing nations. The beginning
point for all is the establishment of qualified
staffing and training, political and
budgetary commitment, enforceable
sanitation laws, and defined responsibilities
and program goals. Implementation of the
!PM plan follows, encompassing surveys,
public education, sanitation programs,
baiting/ trapping, structural improvements,
community involvement, scheduling and
monitoring.

254

Technicalleadership is a serious constraint
for urban rodent control today, and this needs
to be overcome through involvement of
universities and federal and state agencies.
Field personnel on the municipal level need
to base their efforts on facts rather than
myths. To assist them, they need technical
support that is accessible, knowledgeable and
practical. In that regard, research biologists
must help resolve any technical gaps
concerning control methods and local
ecological factors. They also must learn to
partner directly with municipal personnel
and translate the science of rodent control for
the 'real world', so programs can be better
designed, implemented and sustained.
The history of urban rodent control and
the 'lessons learned' in recent years present a
'road map' for future success. However,
without political support and effective
administration, implementation of urban
rodent control programs will continue to be
limited and public health and economic
impacts will result. The ecological and
political arenas of the urban environment
are complex and interrelated, and urban
rodent control programs can only be
implemented effectively when both of those
subjects are mastered.
REFERENCES
Al Senei, K.S., Zaghhloul, T. and Balba, M. 1986.
Organisation of the rodent control project in
Kuwait. In: Richards, e.G.J. and Ku, T.Y., cd.,
Control of mammal pests. London, Taylor
and Francis, 143-150.
Bajomi, D. and Sasvari, K. 1986. Results of eight
years' examination of the habitats of residual
urban rat populations after eradication. In:
Salmon, T.P., ed., Proceedings of the 12th
Vertebrate Pest Conference. Davis, Univer-
sity of California, 66-74.
Urban Rodent Control Programs
Calhoun, J.B. 1948. Mortality and movement of
brown rats (Rattus norvegicus) in artificially
super-saturated populations. Journal of
Wildlife Management, 12,167-172.
Calhoun, J.B. 1963. The ecology and sociology of
the Norway rat. Publ. No. 0-676-768. Dept. of
Health, Education and Welfare. Bethesda,
Maryland, US Public Health Service, 288p.
Chitty, D. and Southern, H.N. 1954. Control of
rats and mice. Volumes 1 and 2-rats, 532p.;
Volume 3-rruce, 225p. Oxford, Clarendon
Press.
Colvin, B.A, Ashton, AD., McCartney, W.e. and
Jackson, W.B. 1990. Planning rodent control
for Boston's Central Artery /Tunnel Project.
In: Davis, L.R and Marsh, RE., ed., Proceed-
ings of the 14th Vertebrate Pest Conference.
Davis, University of California, 65-69.
Colvin, B.A, De Gregorio, Rand Fleetwood, e.
1996. Norway rat infestation of urban
landscaping and preventative design criteria.
In: Timm, RM. and Crabb, Ae., ed., Proceed-
ings of the 17th Vertebrate Pest Conference.
Davis, University of California, 165-171.
Colvin, B.A, Meininger, e.A and Grealy, M.J.
1992. Administrative procedures and
contracts for vertebrate pest programs. In:
Borrecco, J.E. and Marsh, RE., ed., Proceed-
ings of the 15th Vertebrate Pest Conference.
Davis, University of California, 236-240.
Colvin, B.A, Swift, T.B. and Fothergill, F.E. 1998.
Control of Norway rats in sewer and utility
systems using pulsed baiting methods. In:
Baker, R.O. and Crabb, Ae., ed., Proceedings
of the 18th Vertebrate Pest Conference. Davis,
University of California, 247-253.
Davis, D.E. 1953. The characteristics of rat
populations. Quarterly Review ofBiology,28,
373-401.
Davis, D.E. 1972. Rodent control strategy. Pest
control strategies for the future. Division of
Biology and Agriculture, National Research
Council, National Academy of Sciences,
Washington, D.e., 157-171.
Davis, D.E., Emlen, J.T. and Stoke, A.W. 1948.
Studies on home range of the brown rat.
Journal of Mammalogy, 29, 207-225.
255
Ecologically-based Rodent Management
Davis, D.E. and Fales, W.T. 1949. The distribu-
tion of rats in Baltimore. American Journal of
Hygiene, 49, 247-254.
Davis, D.E. and Fales, W.T. 1950. The rat popula-
tion of Baltimore, 1949. American Journal of
Hygiene, 52, 143-146.
Davis, D.E. and Jackson, W.B. 1981. Rat control.
In: Coaker, T.H., ed., Advances in applied
biology, Vol. 6. New York, Academic Press,
221-277.
Davis, H., Casta, A. and Schatz, G. 1977. Urban
rat surveys. Pub!. No. 77-8344. Dept. Health,
Education, and Welfare. Atlanta, US Public
Health Service, 22p.
Drummond, D.C 1970. Rat free towns: the strat-
egy of area control. Royal Society of Health,
90,131-134.
Drummond, D.C 1985. Developing and
monitoring urban rodent control
programmes. Acta Zoologica Fennica, 173,
145-148.
Drummond, D.C, Taylor, E.}. and Bond, M.1977.
Urban rat control: further experimental
studies at Folkstone. Environmental Health,
85,265-267.
Drummond, D.C, Taylor, E.]., Bond, M. and
Greaves, J.H. 1972. Urban rodent control: an
experimental study. Monograph Series.
London, Association of Public Health Inspec-
tors,34p.
Emlen, J.T., Stokes, A.W. and Davis D.E. 1949.
Methods for estimating populations of brown
rats in urban habitats. Ecology, 30, 430-44l.
Emlen, J.T., Stokes, A.W. and Winsor, CP. 1948.
The rate of recovery of decimated popula-
tions of brown rats in nature. Ecology, 29,
133-145.
Frantz, S.C and Davis, D.E. 1991. Bionomics and
integrated pest management of commensal
rodents. In: Gorham, J.R, ed., Ecology and
management of food industry pests (FDA
Tech. Bull. 4). Arlington, VA, Association of
Official Analytical Chemists, 243-313.
Gacs, E, Herczeg, T., Papacsi, Land Elek, S.
1977. Methods of monitoring rat infestations,
and approval schemes for city-wide deratiza-
tion operations. EPPO (European Plant
Protection Organization) Bulletin, 7,533-539.
256
Gratz, N.G. 1994. Rodents as carriers of disease.
In: Buckle, A.P and Smith, A.H., ed., Rodent
pests and their controL Wallingford, UK,
CAB International, 85-108.
Howard, W.E. 1984. An effective organisation is
essential for successful rodent controL In:
Dubock, A.C, ed., Proceedings of a confer-
ence on Organisation and Practice of Verte-
brate Pest ControL Hampshire, UK, ICI, 437-
442.
Jackson, W.B. 1982. Norway rats and allies. In:
Chapman, J.A. and Feldhamer, G.A., ed.,
Wild mammals of North America: biology,
management, economics. Baltimore, MD, The
Johns Hopkins University Press, 1077-1088.
Jackson, W.B. 1984. Urban rodent control in the
United States. In: Dubock, A.C, ed., Proceed-
ings of a conference on Organisation and
Practice of Vertebrate Pest Control.
Hampshire, UK, lCI, 61-71.
Jackson, W.B. 1998. Ecology of pest rodents in the
urban environment. In: Ambasht, RS., ed.,
Modern trends in ecology and environment.
Leiden, The Netherlands, Backhuys Publish-
ers, 101-113.
Jackson, W.B., Ashton, A.D. and Delventhal, K.
1988. Overview of anticoagulant usage and
resistance. In: Suttie, J.W., ed., Current
advances in vitamin K research. Amsterdam,
The Netherlands, Elsevier Science Publish-
ing,381-388.
Kaukeinen, D. 1994. Rodent control in practice:
householders, pest control operators and
municipal authorities. In: Buckle, A.P. and
Smith, A.H., ed., Rodent pests and their
control. Wallingford, UK, CAB International,
249-27l.
Lund, M. 1994. Commensal rodents. In: Buckle,
A.P. and Smith, A.H., ed., Rodent pests and
their control. Wallingford, UK, CAB Interna-
tional,23-43.
MacNicoll, A.D., Kerins, G.M., Dennis, NJ and
Gill, J.E. 1996. The distribution and signifi-
cance of anticoagulant-resistant Norway rats
(Rilttus norvegicus) in England and Wales,
1988-95. In: Timm, RM. and Crabb, A.c., ed.,
Proceedings of the 17th Vertebrate Pest
Conference. Davis, University of California,
179-185.

Myllymaki, A. 1969. An early approach to a rat-
free town in Finland. Schriftenreihe des
Vereins fUr Wasser-Boden und Lufthygiene,
Berlin-Dahlem, 32, 161-166.
Pratt, H.D., Bjornson, B.F. and Littig, K.S. 1976.
Control of domestic rats and mice. PubL No.
76-8141. Dept. Health, Education, and
Welfare. Atlanta, US Public Health Service,
47p.
Pratt, H.D. and Brown, RZ. 1976. Biological
factors in domestic rodent control. Publ. No.
76-8144. Dept. Health, Education, and
Welfare. Atlanta, US Public Health Service,
30p.
PraJ:t, H.D. and Johnson, W.H. 1975. Sanitation in
the control of insects and rodents of public
health importance. Pub!. No. 76-8138. Dept.
of Health, Education, and Welfare. Atlanta,
US Public Health Service, 42p.
Urban Rodent Control Programs
Scott, H.C. and Horom, M.R. 1976. Rodent-borne
disease control through rodent stoppage.
Pub!. No. 97-537. Dept. Health, Education,
and Welfare. Atlanta, US Public health
Service, 33p.
Telle, H.J. 1969. 12-Jahre grossraumige Ratten-
bekampfung in Niedersachsen-kritischer
Ruck-und Ausblick. Schriftenreihe des
Vereins fUr Wasser-Boden und Lufthygiene,
Herlin-Dahlem,32,131-143.
Von Wahlde, M. and Colvin, RA. 1994. Using
geographic information systems for tracking
an urban rodent control program. In: Halver-
ston, w.s. and Crabb, AC., ed., Proceedings
ofthe 16th Vertebrate Pest Conference. Davis,
University of California, 327-334.
257
 Section 3

Case Studies in
Asia and Africa
Zhibin Zhang, Anguo Chen, Zhendong Ning and Xiuqing Huang

Abstract

Rodent pests are a serious problem of agricultural production in China. Abnormal

climate patterns of recent decades , with more severe droughts and warmer winters,
have allowed rodents to become increasingly abundant. In the early 1980s, there
was a widespread outbreak of rodents in the agricultural areas of China. Since
1986, rodent management has been listed three successive times as a national
five-year-plan key project (1986-1990; 1991-1995; 1996-2000). These key
projects aim to (1) collect long-term population data on major rodents and establish
forecasting models, (2) understand population recovery and community succession
after large scale management, (3) develop effective control techniques and
strategies, and (4) set up demonstration areas to assist local governments to
launch a large-scale rodent control campaign . In this paper, a brief introduction to
the results of our national key projects in four agricultural regions is given. The
agricultural regions and their main pest species are (i) the North China Plain, rat-like
hamster (Cricetulus triton) , (ii) the Northwest Loess Plateau, Chinese zokor
(Myospalax fontanien) , (iii) the Dongting Lake region along the Yangtze River,
oriental vole (Microtus fortis) and (iv) the Pearl River Delta, Rattus rattoides. The
problems and future challenges of rodent pest management in agricultural
ecosystems are also discussed .

Keywords

Rodent pest management, agricultural ecosystem, China

261
Ecologically-based Rodent Management
INTRODUCTION
N THE WORLD scale, China has
O the largest population (1.2 billion
people), but one of the smallest
average family holding of arable land (less
than 0.4 ha). Grain production is listed as the
top priority by the central government in
China. In most areas of the countryside,
farmers not only depend on grain for food,
but also for earning money by selling grain
to the market or to the government.
There are four key regions of grain
production in China (Figure 1). The first is
the North China Plain, which belongs to the
Qinghai
Study sites in
agricultural ecosystems
... Study sites in
grassland ecosystems
Figure 1.
Long term study sites for rodent ecology and management in China.
262
middle and lower regions of the Yellow
River and regions of the Huai-He and Hai-
He Rivers. Wheat, corn, peanuts and green
beans are the major crops. The second is the
South China Plain, located in the middle and
lower regions of the Yangtze and Pearl
Rivers. Rice is the major crop in this region.
The third is the Northeast China Plain,
which is formed by the flood plains of the
Hei-Long-Jiang River, Nen-Jiang River,
Song-Hua-Jiang River and Liao-He River.
Corn, wheat and soybean are the major
crops in this region. The fourth is the
Northwest Loess Plateau, located in the
upper reaches of the Yellow River, where
corn and wheat are the major crops.
A. Inner Mongolia
o
 Rodent Pest Management in China

Although the annual grain production in
China is now greater than 500 million t, in
the remote areas there are still several
million people who are short of grain. The
government is planning to increase
production by 25 million t every year during
1996-2000 by introducing new agricultural
technologies. Plant protection, through
management of diseases, insects, weeds and
rodents is listed as the top priority in
realising this goal. Since the 1980s, rodent
problems have become more and more
serious. Changes in climate nationally have
resulted in more severe droughts and
warmer winters (Wang and Ye 1992) which
are two important factors influencing rodent
abundance. In 1997, the Yellow River, called
About 44% of infested areas were treated
using rodenticides, and grain losses were
reduced by approximately 7.5 million t
(Table 1). However, since early 1993, input
by governments on rodent control has been
much reduced due to changes in
government infrastructure and policy.
Farmers have become responsible for their
own rodent control with much less
coordination by government.
Table 1.
The area of arable land infested by, and treated
for, rodents in China from 1980 to 1993
(from Zhao 1996).
Year Infested area Treated area
(million ha) (million ha)

the mother-river of China, was without 3.3 1.3

water flow for more than 100 days. Usually 3.3
the river flows continually. This is indicative 2.5
of the severity of the droughts that China 9.9
(especially North China) is facing. 8.7
In 1982, large-scale eruptions of rodent 14.5
populations occurred in the farmlands of 17.5
China. The area of infested arable land was 17.3
greater than 27 million ha, about 27% of the 26.7 11.7
total arable land in China. The a1U1Ual grain 1989 21.3 14.7
loss caused by rodents was over 15 million t. 1990 21.3 6 .7
In one year, over 140,000 people contracted 1991 20.3 13.9
rodent-borne diseases (mostly epidemic 1992 13.3
haemorrhagic fever, leptospirosis and 5.0
endemic typhus) (Wang 1996). 317.9 140.3

In 1983, the State COLU1cil of the Chinese
Central Government issued an urgent
document calling for local governments to
launch a movement on rodent control in
farmland and grassland. Rodent control was
listed in the top three priorities for the plant
protection program. Consequently, the
governments of different levels put much
effort and resources into rodent control.
Note: data for 1980-82 were based on surveys in
18 provinces; data for 1983-87 in 20-24
provinces; data for 1988-1991 in 27 provinces;
22 1
data. for 1992 in 26 provinces; data for 1993 in
provinces.
The magnitude of the rodent outbreaks in
the arable land of China in the early 1980s
also led to an increased effort in rodent
research. Since 1985, rodent control has been

263
Ecologically-based Rodent Management

listed in three successive national five-year-
plan projects (1985-1990; 1991-1995; 1996-
2000) by the central government. There are
approximately 100 scientists with the
Chillese Academy of Sciences, Ministry of
Agriculture and universities working on
rodent control. Long-term study sites
located in key regions of grain production
were selected (Figure 1) according to level of
the rodent infestation. Table 2 gives details
of the locations of the study sites and the
major rodent pest species in each. No sites
were on the Northeast China Plain. Two
study sites withill the grassland ecosystem
(Figure 1) were selected and the findings
from research conducted at these sites are
reported elsewhere in this book (Fan et al,
Chapter 13; Zhong et al., Chapter 9). In 1986,
another study site, which does not belong to
the key regions of grain production but with
local heavy rodent infestation, was selected
in the south part of the ltmer Mongolia
Plateau. This region is a mixture of crop land
and grassland. Mongolian gerbils (Meriones
unguiculatus) cause huge damage to crops
(mostly cereals and potato) in this region.
Since 1985, population surveys,
assessment of rodent damage, and control
techniques and strategies, have been
extensively studied by well-trained scientists.
Scientific staff at each of the study sites
provide teclmical extension and advice for
instigating local rodent control campaigns.lt1
this chapter, case studies are described from
four major agricultural regions, focusing on
the achievements of rodent control based on
biological and ecological knowledge of the
target rodent pests.

Table 2.
Long-term study sites of the three successive national five-year-plan projects on rodent ecology and
management in key regions in China.

Key regions Study sites Major pest species

North China Plain Hebei Province Cricetulus triton

Shandong Province Cricetulus barabensis
Northwes""t""'L-o-e-
ss-;;p:;-Ola~t-e-
a-u -..;;;;;;....- Shanxi Province

Shaanxi Province Microtus mandarinus

Cricetulus triton
------
South China Plain Hunan Province Rattus norvegicus
(Yangtze River Region) (Dongting Lake Region) Microtus fortis
Sichuan Province Rattus nitidus
Rattus norvegicus
Guandong Province Rattus rattoides
Bandicota indica
Inner Mongolian Plateau Inner Mongolia ----------- Meriones-------~----~~~
unguiculatus

264

CASE STUDIES OF RODENT
MANAGEMENT IN THE MAJOR
AGRICULTURAL REGIONS
Hamster management in the North
China Plain
Agricultural systems and environments
In the North China Plain, wheat is
planted in autumn (October) and harvested
the next summer in June. Summer corn is
planted immediately after the wheat harvest
and is harvested in October. Small areas of
the arable land are not planted in October.
Instead, they are planted the next spring
with crops such as cotton, peanuts, soybean
and spring corn. Therefore there are three
sowing seasons (April, June and October) as
well as two harvest seasons aune and
October).
The climate in this region belongs to the
warm-temperate zone. It is very cold in
winter and very hot in summer. The annual
rainfall is approximately 400-500 mm, with
80% of the annual rainfall occurring in
summer aune, July and August).
Ploughing and irrigation are two major
agricultural activities in the sowing seasons.
Other common activities in the fields include
spreading of chemical insecticides and
fertilisers, and the clearing of weeds. Most of
these activities have a negative effect on
rodent populations in this region. Ploughing
and irrigation destroy their burrow systems,
and sometimes kill the juveniles directly.
The spatial distribution of burrow holes of
the rat-like hamster is clearly affected by
periods of intensive agricultural activities;
most of the burrows are constructed in the
banks or in non-irrigated or non-ploughed
wastelands (small patches of trees and
Rodent Pest Management in China
grasses) during this time (Zhang et a1.
1997c).
In 1986 in Raoyang County, Hebei
Province, the impact of winter irrigation on
the habitat and distribution of the burrows
of the rat-like hamster (Cricetulus triton) was
examined. The density of burrows of the
hamster in the wasteland was 67.7 holes/ha,
while the density in irrigated farmland was
35.6 holes/ha (Zhang et al. 1997c). Winter
irrigation of wheat is therefore crucial in
reducing the over-wintering hamster
population.
Reproduction patterns and
population dynamics
The climate, crop plantation system and
agricultural activities determine the seasonal
reproduction and population dynamics of
rat-like hamsters. Rat-like hamsters usually
begin to reproduce in early March and finish
breeding by the end of August or early
September. The adult female hamsters that
survive the winter produce three litters and
their young of that season may in turn
produce 1-2 litters. Females born in early
autumn do not breed until the next spring.
The litter size of the rat-like hamster ranges
from 2-22 with an average of 9 or 10. The
reproductive performance of hamsters is
considerably affected by population density.
In the peak year of 1996 (average trap
success 7.7%) the average litter size of the
rat-like hamster was 9.1 with a pregnancy
rate of 24.8%. In the trough year of 1998
(average trap success of 1.2%), the average
litter size was 9.9, with a pregnancy rate of
39.2% (Zhang et a1. 1998).
For the rat-like hamster, there are two
periods of high density within a year; one in
spring and one in autumn. The autumn peak
265
Ecologically-based Rodent Management

(e.g. in 1986) is usually larger than the spring
peak (Figure 2). Heavy rains and high
temperatures in summer cause low
population densities by increasing the
mortality rate of the hamsters. In 1986, the
monthly mortality rates of rat-like hamsters
were 0.19, 0.14 and 0.42 in spring, autumn
and summer, respectively (Zhang et al.
1992). The high summer temperatures
(possibly together with the interaction of
heavy rain or changes in the photoperiod)
result in a ID-day longer mean interval
between pregnancies for hamsters compared
to spring (Zhang et a!. 1991).
Since the early 1980s, the winter has
become warmer in the North China Plain
and the seasonal population patterns have
begun to change, with the magnitude of the
spring peak similar to, or higher than, the
autumn peak. Taking 1986 and 1994 as
examples, the average air temperature in
January in 1986 and 1994 was -3.66C and-
2.1 QC respectively; the hamster population
density in January was 0.07% and 3.01%,
respectively; and the mortality from the
previous October to April was 0.83 and 0.56,
respectively. This resulted in a higher spring
peak than autumn peak in 1994 (Figure 2).
Damage and assessment
The rat-like hamster has a mean body
mass of 120 g and is principally a seed-eater;
70% of the food carried in its cheek pouches
is composed of crop seeds, 15% stems, roots,
flowers and leaves of crops, and 15% insects
(Wang et al. 1991). The main damage caused

25

20

Cl)
Cl)
15 -:- ~::: 1- - - - - -
Ql

8::J
Cl)

c..
f! 10
I-

Jan Feb Mar Apr May Jun Jul Aug Sep Qct Nov Dec
Month

Figure 2.
Seasonal dynamics of the rat-like hamster population in :1.986 and :1.994 in Raoyang County, Hebel
Province (from Zhang et al. 1998).

266

by the rat-like hamsters occurs to seeds
during the sowing periods and to mature
crops during the harvest periods. The rat-
like hamster stores about 200 g of seed in
both spring and summer, and much more in
autumn ranging from 0.6-20.0 kg (n = 40)
per hamster.
Peanuts, the favourite food of the rat-like
hamster, suffer high levels of damage. In
1986, two enclosure experiments were
conducted in Raoyang County, Hebei
Province, on the depredations of the rat-like
hamster to peanut crops. The first study was
conducted in a 0.26 ha enclosure. At a
capture success of 22/", the rat-like hamsters
consumed 14.8% of the peanut crop. The
second study was conducted in a 2 ha plot.
At a capture success of 24%, the rat-like
hamsters consumed 19.6% of the peanut
crop (Zhang et al. 1998).
It is difficult to assign damage to
particular species when the rodent
population consists of a mix of several
species. By employing multiple regression
statistics, Wang et aL (1996) developed a
model for estimating the damage to peanut
crops by mixed populations of rodents
during the sowing period (May) and the
harvest period (September) in Shandong
Province as below.
LS% =0.167 + 0.155D 51 + 0.260D 52 +
0.086D53 (r = 0.981, F = 87.48, df 3,10,
P < 0.001) (1)
where LS% is the peanut loss at sowing
caused by the striped hamster (D 51 ), the rat-
like hamster (D S2 ) and the striped field
mouse, Apodemus agrarius (D 53)' DS indicates
the trap success (%) of the respective rodent
populations in spring.
Rodent Pest Management in China
LA% = 1.226 + 0.176DA1 + 0.456DA2 +
O.llODAY r = 0.959, F =22.98, df = 3,6,
P < 0.001) (2)
where LA% is the peanut loss at harvest
caused by the striped hamster (DAI)' rat-like
hamster (D A) and striped field mouse (DA3)'
DA indicates the trap succesS (%) of the
respective rodent populations in autumn.
Population forecasting for management
Since the most serious damage occurs in
autumn, and rodent control often begins in
early spring, there is a strong need to be able
to forecast the popUlation of hamsters in
autumn. Based on the monitoring data from
12 years in Raoyang County and Guan
County, a short-term forecasting model for
the rat-like hamster was established (Zhang
et al. 1998):
YA = 0.98 + 8.66X4 ( r = 0.96, P < 0.01 )
when X4 s 2.33 (3)
YA = 18.43 2.14 X4 (r = 0.89, P < 0.05 )
when X4 > 2.33 (4)
Using this model, where Y A is the
maximum trap success (%) in autumn
(September, October or November), X4 is the
trap success (%) in April.
Although the autumn trap success of the
rat-like hamster is highly correlated with its
April trap success, the correlation is non-
linear. There is a strong positive association
when the April trap success is s 2.3 and a
strong negative association when the April
trap success is > 2.3 (Figure 3).
Based on the above forecasting models,
we successfully predicted the changes in
population dynamics of the rat-like hamster
in Raoyang County and Guan County in
1996 and 1997 (Table 3). The accuracy of
prediction is defined as:
A [1 I P 0 I / M] x 100% (5)
267
Ecologically-based Rodent Management

where A is the accuracy of prediction, Pis rodent campaign is necessary to reduce

the predicted trap success, 0 is the observed grain losses. We have observed that farmers
trap success, and M is the maximum trap pay less attention to damage when autumn
success ever observed . The accuracy of trap success of the rat-like hamster is less
prediction in 1996 for Raoyang COlU1ty and than 5%. This trap success is equivalent to
Guan COlU1ty was 87% and 97.9%, approxima tely 3-4% loss of the peanut crop.
respectively. In 1997 the accuracy of
Control techniques and strategies
prediction for the same two cow1ties was
95.2% and 90.7%, respectively. The ra t-like hamster is an important
rodent pest in the North China Plain and
25 farmers have developed several traditional
+ methods of pest management. In autumn
~ after harvest, farmers dig the rodent
<f)
<f)
20
Q)
u
u burrows to recover the grains stored by the
:>
-- --- --- --+ - --
<f)
0. 15 --- - -- --- ----- --- rat-like hamster. Some farmers are extremely
g
E
++ proficient at digging hamster burrows,
:>
.~ 10 - - - - - - - - - - - -+ - - - - - - - - - - - recovering 100 kg of grain within 1- 2 weeks .
++
'"c
E + The va lue of ploughing and irrigation is
5 - - - - - - - - - - - - - - - - - - ~- - - - - - - -- - - also well known by farmers for red ucing
S +
hamster populations. It is common for
o ~._ ____ I
~~ I I
L -_ _ _ _ _ _ L -_ _ _ _ _ _~_ _ _ _~
farmers to directly kill hamsters during
024 6 8
Trap success in April (%) ploughing and irrigation.
During 1994-98, the negative impact of
Figure 3.
The relationship between the autumn maximum
irrigation on the hamster population was
trap success (%) and trap success in April (spring) clearly demonstrated in Shunyi District,
in Hebei Province from 1984-1995. 8eijing. Shunyi used to irrigate its farmland
in the traditional way, plunping water
Forecasting is an important step in directly into farmland. This method wasted
deciding whether to lalU1ch a rodent control a lot of grOlU1dwater. In 1994, Shunyi
campaign in spring in this region. If the swapped over from the old ditch irrigation
predicted autumn trap success is over 5%, a to spray irrigation. The rodent community

Table 3_
Prediction of the population abundance of the rat-like hamster in Raoyang County and Guan County, Hebei
Province, in 1996 and 1997 using the forecasting models described in the text (see also Zhang et al. 1998).
A is the accuracy of prediction, P is the predicted trap success, and 0 is the observed trap success. The
maximum trap success ever observed for this cropping syst em was 23.1

Place 1996 1997

P o A% P o A%
Raoyang 7.73 8.21 97.9 7.68 5 .5 2 90.7
Guan 12.0 9 .0 87 .0 9.4 8.3 95.2

268

changed suddenly from a striped field
mouse dominant community into a rat-like
hamster dominant one. The outbreak of rat-
like hamster populations in Shunyi District
occurred for several years after this change
in irrigation system, and has resulted in
tremendous losses in crop production.
In 1978, rural China experienced a reform
from a community-based system to a family-
based system in which each family rents a
small area of land, and they decide what they
plant. This has led to a patchy landscape
consisting of a mosaic of different crops. This
heterogeneous cropping system provides a
favourable environment for hamsters by
increasing their survival and breeding
performance. Small plots of land in a patchy
landscape are particularly vulnerable to
attack by hamsters, especially those plots
growing oil crops. Chemical control has little
impact on rodent populations in such small
plots of land because recolonisation by rats
from the surrounding environment soon
counteracts any local reductions in
population density. In order to solve this
problem, a new multiple-capture physical
trap was invented (Zhang et al. 1996). The
pitfall trap was designed with a magnetic
trigger on its lid (Figure 4). One trap was set
in each of two corn fields in Guan County in
the summer of 1995. In eight days, both traps
caught 20 hamsters. During the experiment,
the capture success in snap-back traps of the
rat-like hamster, striped hamster (Cricetulus
barabel1sis) and house mouse (Mus musculus)
was 21 %, 1% and 0.3%, respectively.
Fresh wheat containing 0.005%
bromadiolone (see Guo et a1. 1997) was used
in a rodent control campaign in Beijing. In
1997, more than 333,000 ha of farmland were
treated with bromadiolone, achieving
Rodent Pest Management in China
approximately a 92% kill rate (based on pre-
versus post-treatment indices of abundance)
in early spring. The capture success of
hamster was maintained at a level less than
5/c, through the year, however the
population recovered to its original density
the next spring, and another chemical
control campaign was launched in 1998
(Xihong Guo, pers. comm.).
In addition, a male chemosterilant, 1-2%
a-chlorohydrin, has been tested for
controlling the rat-like hamster (Zhang et a1.
1997a,b). The chemosterilant was tested in
Guan County, where bromadiolone only
kills 70-80% of hamsters, enabling the
populations to recover quickly from
poisoning campaigns (Zhibin Zhang,
unpublished data). Zhang (1995,1996)
suggested that a strategy of combining
fertility control with chemical mortality
might delay the recovery of hamster
populations post-poisoning; male hamsters
that do not die following the ingestion of
bromadiolone would become sterile from
the a-chlorohydrin.
In July of 1993 and in April of 1995, an
experimental farmland site was treated with
a wheat bait of 0.005% bromadiolone and 1%
a-chlorohydrin. Similar surrounding
farmland was selected as an untreated area.
Ten days after treatment, 75% and 78%
mortality were achieved in 1993 and 1995,
respectively, on the experimental site. In
1994, no control measure was taken.
Mortality control combined with
sterilisation achieved good results. During
1995, the capture success of the rat-like
hamsters on the experimental site was less
than 5%, while eruptions of hamster
populations occurred on the untreated site
(Zhang et a1. 1998; Figure 5).
269
Ecologically-based Rodent Management
Figure 4.
A new multiple pitfall trap with a magnetic trigger.
Recommended management strategies
and research priorities
In the North China Plain, there are two
strategies recommended for farmers to
manage the rat-like hamster:
~ If the hamster is causing localised
problems within a farming system then the
pitfall trap is recommended. One trap
should be placed in the middle of a crop (in
this region the average farm size is 0.25 to
0.5 ha).
~ If the hamster is causing problems over a
large area then broad-scale application of a
wheat bait of 0.005% bromadiolone and 1%
a-chlorohydrin is recommended.
270
Further research is required on the
efficacy ofbromadiolone with and without
a-chlorohydrin, and on the social acceptance
of using male sterility baits. Other research
has beglill on Chinese herbs for enhancing
the effect of anticoagulants. In particular, is it
possible to increase the susceptibility of
hamster populations to bromadiolone and
can the time until death be reduced
(currently around 10 days for hamsters)?
Zokor management in the Northwest
Loess Plateau
Agricultural system and environment
The Northwest Loess Plateau is one of the
most undeveloped areas in China due to its
harsh climate. This region has very limited
 Rodent Pest Management in China

25

Treated
20
Untreated

~ 15
ID
ID

8
::::J
ID
a. 10
<tl
F

_L-~-L--"-_~_m I I
A M J J A S
1993 1994 1995

Figure 5.
Field trials of the effect of a-chlorohydrin combined with bromadiolone on the rat-like hamster population
In Guan County, Hebel Province, from 1993-1995.

rainfalL Corn and wheat are commonly Damage and assessment

planted in the upper Yellow River regions.
The grain production is seriously affected by
rodents, particularly the Chinese zokor
(Myospalax fontanieri). Rodent control is very
important for reducing grain losses and
increaSing the income of farmers in this
region.
Reproduction patterns and
population dynamics
The Chinese zokor is a large rodent (300-
650 g) that lives underground. The breeding
season is from March to June, and adult
females reproduce only once a year. The
litter size ranges from 1-6, averaging 3.0
0.4 (Zou et al. 1998). There is one annual
peak in zokor numbers in June or July.
The burrowing activity of the Chinese
zokor causes significant damage to crops.
Zokors also take crop seedlings or seeds to
their burrows for food. Winter wheat
sustains high damage because zokor
populations are short of other food sources
during winter. Serious damage occurs to
corn and sunflower seedlings in spring
when zokors begin to breed and more
extensive digging and burrowing take place.
Zokors also horde food in autumn for over-
wintering.
The degree of damage caused by zokors
can be divided into five classes based on the
proportion of seeds I seedlings eaten (Ning et
a1. 1994):

271
Ecologically-based Rodent Management
.. nodamage;
.. proportion of seeds / seedlings eaten is 1-
25%;
.. proportion of seeds/ seedlings eaten is 25-
50%;
.. proportion of seeds / seedlings eaten is 50-
75%; and
.. proportion of seeds / seedlings eaten is 75-
100%.
The regression models between crop
losses and density of zokor were established
as follows (Ning et al. 1994; Chang et al.
1998):
Lw = -0.4462 + 0.9748X6 (r = 0.9625,
df = 7, P < 0.01) (6)
where LW is the proportion of wheat eaten
by zokors (%) and X6 is the density of the
zokors in June (individuals/ha); and
Le = -1.6427 + 1.1913X9 (r = 0.9657,
df = 5, P < 0.01) (7)
where Le is the proportion of corn eaten by
zokors (%) and X9 is the density of zokors in
September (individuals/ha).
Control techniques and strategies
Because zokors live underground and do
not readily take baits or enter physical traps,
routine techniques like rodenticides and
physical trapping are inefficient for control.
A fumigant, aluminium phosphide, is
commonly used to kill zokors in this region.
The burrow tunnel system is very complex,
therefore it usually requires about 10 pieces
of the fumigant (3.3 g) to effectively fumigate
a burrow. The kill rate is approximately 80%
(based on pre- versus post-treatment indices
of active burrows) if conducted by experts,
but is generally less than 70% when applied
272
by farmers during a rodent control
campaign.
A new technique was invented in 1986,
called the 'explosive paper tube' (EPT)
which is specialised for controlling
underground rodents (Liu et al. 1991). The
EPT contains explosives of dinitrodiazo-
phenol or nickel nitrohydrazino, and is
triggered via a battery. The EPT is 20-30 mm
long, and its diameter is 3 mm (Figure 6).
The EPT is waterproof, easy and safe to
carry. EPT is also cost effective and available
commercially in this region.
The kill rate with EPTs is over 95% for
large-scale rodent control campaigns -
much better than using rodenticides. The
procedure for setting an EPT is simple and
requires four steps: (1) select an active
burrow tunnel; (2) dig into the tunnel to
place an EPT there, bury it; (3) block the
tunnel with a soil ball loosely; (4) connect the
two wires with a small battery (1.5 V),
making a trigger which will be touched by
the falling soil ball. When a zokor pushes the
soil ball which blocks its tunnel, the ball will
touch the trigger, and the EPT will explode
under the zokor body and kill it (Zou et al.
1998).
Recommended management strategies
and research priorities
The EPT has been successfully
commercialised and proven very efficient
for zokor control in this region. However,
the trigger system is still complex, and thus
needs to be improved further. A trigger
system is required that does not re-use the
battery and the connecting wires. This
would make it easier to set the EPT in the
field.

Figure 6.
The explosive paper tube (EPT) for controlling zokors.
The EPT occasionally only injures the
zokor, and has potential risk to people,
especially children. From the view of
humaneness and public safety, alternative
control techniques need to be considered for
replacing the EPT for zokor control. Two
research priorities are suggested. The first is
to improve the acceptance of rodenticide bait
by adding an attractant. This requires
detailed study of zokor behaviour and their
chemical communication. The second is to
find an ecological management strategy for
zokor control. The Chinese zokor's favourite
food are the roots of grass and crops. The
clearing of weeds using herbicides in non-
crop lands has shown potential for reducing
zokor populations (Zhengdong Ning,
Rodent Pest Management in China
Z
""c:
~
o
".c:c:
N'"
~
0..
unpublished data). Also, planting toxic
herbs in wastelands (non-crop habitats),
banks and crop lands could be effective in
reducing the favourite grasses of zokors and
in poisoning zokors when they eat the roots
of the herbs.
Vole management in the South China
Plain (Yangtze River region)
Agricultural system and environment
Dongting Lake is a very large lake located
in the middle of the Yangtze River. It plays
an important role in regulating floods of the
Yangtze River during the rainy season. The
oriental vole (Microtus fortis) is a rodent
species well adapted to the environment of
273
Ecologically-based Rodent Management

Dongting Lake (Chen et a1. 1995; Wu et a1. et a1. 1995, 1996, 1998; Wu et al. 1996; Guo et a1.
1996,1998; Guo et a1. 1997). The voles 1997).
migrate back and forth between the beaches
The seasonal population patterns of the
and islands of Dongting Lake and the
oriental vole in rice fields and in the grass
surrounding rice fields as the floodwaters
habitats on the islands and near the beaches
rise and recede (see below).
of the lake are affected greatly by their
migration between lake habitats and the rice
Reproduction patterns and population
fields . There is a population peak in summer
dynamics
in the rice fields following the migration of
The adult oriental vole weighs 59.5 11.3 g the voles. After October, voles begin to
(n = 378) for females, and 77.5 15.0 g return to the beaches and ephemeral islands,
(n = 415) for males. Although the oriental vole and then their density in rice fields is low.
can breed throughout the year at Dongting
Lake, its breeding is much affected by Population forecasting models
flooding. Unlike most rodent species, the
As the invasion of oriental voles into rice
pregnancy rate is high prior to spring. The
fields during the flooding season causes
pregnancy rate from February to April is 64%.
huge losses to rice production, it is necessary
Even in deep winter, the pregnancy rate is still
to predict these population changes. Chen et
maintained at 23.5-35.3% (Chen et a1. 1998).
a1. (1998) found that the trap success (%) of
This reproductive strategy is clearly adapted
voles in farmland during the flooding season
to the flooding cycles of the Yangtze River.
is mainly determined by the duration of
The beach and island habitat with grass in the
breeding of the vole population living in
lake is the optimal habitat for the herbivore
beaches in the non-flooding season and the
voles in winter. During the flooding season
rainfall in March. The dmation of breeding
these habitats are flooded. Just prior to
in lake beaches is strongly correlated with
flooding, oriental voles migrate in large
the pe riod when the water level of Dongting
swarms to the surrounding farmland where
Lake is below 27.5 m. The regression model
their breeding is greatly reduced (Table 4).
was established as follows:
During September and October, the flooding
waters recede and the voles retun\ to the Y = 0.0394X 1 - 0.0048X 2 - 5.02 (R = 0.957,
beach and island habitats for the winter (Chen df = 9, F = 49.23, P < 0.0001) (8)

Table 4.
Reproduction of oriental voles in flooding and non-flooding seasons (from Chen et al. 1998)_

Habitat Lake beaches Rice fields

Duration November to May May to October
Season Non-flooding season Flooding season
--------'
Number of females 185 280
Pregnancy rate (%) 51.4 20.4
Litter size 5.06 0.15 5.37 0.25

274
 Rodent Pest Management in China

where Y is the trap success of voles in rice prevent immigration of voles from lake
fields during the next flooding season, XI is beaches to the rice fields. Pots which were
the duration of breeding in lake beaches and 0.8 m deep and 0.3 m diameter were buried
X2 is the rainfall in March. In 1994 and 1996, between two fences. The pots were located
this model was used with good accuracy for every 50 m along the fences. When the
predicting the trap success of voles in rice flooding season approached, the large
fields during the next flooding season (Chen swarm of voles was channelled into the pots
et al. 1998). en route to the surrounding rice fields. From
1981-1987, along the west bank of Dongting
Damage and assessment
Lake (which covers two state farms and
In 1986, in the Yueyang County, 5,213 ha eight towns), 1,588 t of voles were captured
of rice fields were damaged by voles with along a total of 231 km of fence (Table 5).
grain losses of 918 t. More than 50% of the
surrolmding trees were seriously chewed by Table 5.
voles (Chen et al. 1998). The quantity (t) of voles captured by burying deep
pots in the dyke along the west bank of Dongting
The oriental vole also spreads a serious
Lake from 1981 to 1988 (from Chen et aI.1998).
rodent-borne disease, leptospirosis, to
farmers working in rice fields . In 1979,527 Year Barrier line (m) Voles captured et)
people on one state farm were infected by 3050 11.00
leptospirosis and 214 of them were sent to 17300 159.00
hospital. 1983 43200 58.55
The regression model between rice loss 40700 106.75
(L%) and the trap success is established by 1985 35450 240.50
Wang et al. (1997): 1986 42160 511.00
L = 0.0674X 1 + 0.0307X 2 - 0.1627 (R = 0.83, 1987 47200 501.2
df = 2,10, F = 11.07, P < 0.01) (9) 1988 2850 49.00
where XI is the trap success of oriental voles, Total 231910 1637
and X2 is the trap success of the striped field
mouse. Chen et al. (1998) later improved this
technique by enclosing the banks of the dyke
Control techniques and strategies with a 0.5 m high brick wall at the top of the
Based on the habits of migration of the dyke, with a 80 mm overhang. This physical
oriental voles between beaches of the lake structure prevented the voles from entering
and rice fields during the flooding and non- the rice fields (Figure 7). The damage of this
flooding seasons, a new method for rodent species has been well controlled since the
control was invented in 1981 by the local construction of the rodent-proof wall . This is
farmers of Jingpen State Farm. They buried an example of successfully controlling rodents
deep pots between fixed fences that were based on understanding their ecology, and
erected along the dyke surrounding the without using chemical rodenticides.
Dongting Lake. The plate fence was 500 mm
high, and buried 50-100 mm into the soil to

275
Ecologically-based Rodent Management
Recommended management strategies
and research priorities
The modified dyke-barrier is an efficient
method for controlling voles in this region.
This system also satisfies the demand for
flood management, and is readily
incorporated in the flood prevention program
in the Dongting Lake region when the dyke
needs repair. Therefore, it is important to
maintain the dyke with the modified physical
barrier system for vole control. Future study
should focus on another pest species, the
Norway rat (Rattus norvegicus), which causes
damage in both fields and houses.
Rat management in the South China
Plain (Pearl River Delta)
Agricultural system and environment
The climate in the Pearl River Delta is
subtropical with an annual rainfall of 1500-
Figure 7.
The modified dyke-barrier system for controlling oriental voles.
276
2000 mm. Rice is the main crop in this region
and it is planted twice a year. Rattus rattoides
and Bandicota indica are two major rodent
pests in the rice fields. In this region, the
amount of arable land is decreasing because
of industrialisation and this has created
more wastelands (non-cultivated lands). In
some areas, B. indica is becoming more
abundant (He 1998).
Reproduction patterns and
population dynamics
R. rattoides is a rodent of medium size.
The adult body weight ranges from 100-200
g. It breeds through all seasons of the year,
with two pregnancy peaks, one i.n June and
the other in October. The pregnancy rates in
January and December are very low, less
than 0.6%. The average pregnancy rate
ranged from 35.4-54.6% during 1987-1991
(Huang et al. 1994a). The litter size ranges
from 2 to 14, averaging 6.78 0.10.

R. rattoides displays seasonal movements
between different crop fields. During the
growing seasons for rice (April to July;
August to November) they invade rice
fields. After the harvest of rice they migrate
into orange and banana plantations, where
they over-winter.
In winter, the average densities of rats in
orange and banana plantations are 13.6%
and 13.7%, respectively, while the average
densities in rice fields surrounding orange
and banana plantations are only 6.2% and
7.1 %, respectively (Feng et a1. 1990a).
Population recovery of R. rattoides after
chemical control
The population recovery of R. rattoides
after chemical control has been well studied
in 1987 by Feng et a1. (1990b). The experi-
mental area of each treatment was 27 ha
without replicates. Twenty days after use of
diphacinone rodenticides in the middle of
January, February, March or April, the kill
rates (based on pre- versus post-treatment
abundance indices) of rats were 78.5%,
91.2%,94.8% and 70%, respectively. In the
experimental area treated in mid-January,
the rat population had recovered to or
surpassed its original level by early July. In
the other experimental areas treated either in
the middle of February, March or April, the
rat populations also had recovered to their
original levels by July (Figure 8). Therefore,
even if the kill rate during the first half-year
was over 90%, populations of R. rattoides
recovered so rapidly that another chemical
control campaign was necessary to protect
the autumn rice crop. Rodenticides were
applied in August to protect the autumn
crop, however heavy rain, strong winds and
thick ground weed cover resulted in a much
Rodent Pest Management in China
lower kill rate than in the first half-year.
Following an August baiting campaign,
populations of rats recovered to the original
level, or even higher, by November (Figure
8) (Feng et a1. 1990b).
Damage and assessment
In the Pearl River Delta, rice, orange,
bananas and vegetable crops suffer great
losses from R. rattoides. Rice that ripens early
suffers more damage than rice that ripens
later. The levels of rat damage to early,
medium and late ripening rice are 5.3%,1.5%
and 0.6%, respectively. Huang et a1. (1990a)
reported that an adult R. rattoides could
cause losses amounting to 3,150 g of rice in
one year. Huang et a1. (1990b,c) established a
regression model between infestation rate of
rice (L, %) and the trap success of R. rattoides
(X, %) in 1987:
L = -D.27 + 0.29X (d.f. 3, r =0.998) (10)
For assessing damage caused by a
mixture of populations of several rodent
species, Feng et al. (1995) established two
regression models between the loss rate (%)
of rice and rodent densities in 1992:
L1 = -0.0990 + 0.3367X1 + 1.4578X2 +
0.0361X3 (R 0.976, d.f. = 5, F 73.22) (11)
-D.4250 + 0.3781X1 + 1.4523X2 +
0.6639X3 (R = 0.904, d.f. = 5, F 122.24) (12)
where L1 and L2 represent the loss rate
(%) of early ripening rice and late ripening
rice andX1, and X3 are the trap success
(%) of R. rattoides, B. indica and the house
mouse, respectively.
Control techniques and strategies
Since the 19805, coumatetralyl and
diphacinone have been widely used for
controlling R. rattoides. Two separate
chemical control campaigns are needed
277
Ecologically-based Rodent Management

25

20

--
-+- Kill in Jan. & Aug.

Kill in Feb.

--
~ Kill in Mar. & Aug.

~ 15 Kill in April
'"'"
ID
0
0
::J
(/)
a.
~ 10 --------------
I-

o
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Month

Rgure8.
Population recovery dynamics of Rattus rattoides after chemical control (from Feng et al. 1990b).

every year in this region in order to reduce
rat damage to rice because of the rapid
recovery of the rodent populations after
chemical control. Therefore ecologically-
based management is urgently needed as an
alternative method.
R. rattoides depends much upon the
ground vegetation cover in the orange
groves, banana plantations, wastelands and
banks of rice fields. When the ground
vegetation cover in orange or banana
plantations is over 60% with dry biomass of
766 g/ m 2, the number of active burrow holes
of R. rattoides is 48 4.7 holes/lOO m
(estimated using a line transect method);
when the ground vegetation cover in these
plantations is less than 20% with dry
biomass of 369.5 g/m2, the number of active
burrow holes of R. rattoides is only 2.4 0.3
holes/WO m (Feng et al. 1996). Therefore, the
clearing of weeds in the orange groves,
banana plantations, wastelands and banks of
rice fields is important for reducing the
density of rat populations. Huang et al.
(1994b) demonstrated that the density of
R. rattoides was reduced from 52.0 5.5
holes/lOO m to 5.0 1.1 holesllOO m after
the clearing of weeds in rice fields and in the
optimal habitats surrounding the rice fields.

278

Feng et al. (1996) examined whether
planting some economic fruit trees with
thick branches and leaves such as lychee,
mango and longan in the wastelands or at
the edge of rice fields could greatly reduce
the weed cover, and thence reduce the rat
density. After planting such evergreen fruit
trees, weed biomass was reduced by 62.5-
79.5% and the rat density was reduced by
77.3-89.1% (Table 6).
Recommended management strategies
and research priorities
This study clearly indicates that chemical
control is not a solution for sustainable
management of rodent pests in this region.
We recommend the strategy of combining
chemical control with ecological
management. In regions with high rat
density, rodenticides should be used in
February or March, followed by clearing of
weeds in the wastelands and / or fruit tree
plantations, and modifying these habitats by
planting trees with thick leaves, such as
lychee, mango or longan. The latter would
be a more promising method of
management because it not only reduces the
population denSity of rats by reducing weed
cover and their damage to crops, but also
Table 6.
Changes in weed biomass and rat density (active holesj100 m) after planting fruit trees on the river dyke and
waste hill lands in Pear River Delta. Control plots were not planted with any fruit trees (from Feng et a!. 1996).
Habitat Fruit tree type Plots
River dyke Lychee 10
Orange 10
Control 10
Hill lands Lychee 10
Orange 10
Control 10
Rodent Pest Management in China
provides additional economic income for the
farmers.
PROBLEMS AND POSSIBLE SOLUTIONS
Since the 1980s, China has achieved
promising advances in rodent pest
management in agricultural systems (Zhang
and Wan 1997). Firstly, acute poisons were
replaced with anticoagulants. This alleviated
the environmental pollution and secondary
poisoning of nahlral predators and
increased public safety.
Secondly, population ecology has been
considered more than before as a basis for
developing strategies for rodent pest
management. Prediction of population
increases and data on damage assessment
have been listed as important aspects for the
development of cost-effective and
environmentally sensitive rodent control.
The concept of ecological management is
becoming much more accepted by people,
even by those who were strong proponents
of pure chemical control. For some of the
major rodent pest species, reliable prediction
models and sound damage assessment
models have been established. These
provide important information on when,
Weeds blomass Rat density
(gjm2) (active holesj100 m)
299.3 21.5 36.3 4.4
354.1 27.9 39.9 4.1
945.2 58.2 175.6 12.0
120.6 35.9 14.7 3.2
187.3 29.6 23.5 2 .7
589.4 37 .3 135.2 19.3
279
Ecologically-based Rodent Management
where and how to manage rodents before
launching a control campaign.
Thirdly, some new techniques for
managing target rodent pests have been
developed and proven effective. For example,
the EPTs for managing zokors in the
Northwest Loess Plateau, the multiple
magnet-triggered traps for managing rat-like
hamsters in the North Plain, the dy ke-barrier
system for managing oriental voles at
Dongting Lake and habitat modification for
managing rats in the Pearl River Delta. These
advances depend heavily on understanding
the behaviour of the target species, in
particular how they respond to and use their
environment. Therefore, ecologically-based
rodent management must focus on detailed
research of the biology, ecology and
behaviour of the target species as well as the
surrounding environment, instead of looking
for a popular generic recipe applicable for
managing all rodent pest species.
Despite this promising progress, rodent
control in China still faces many problems.
One problem is that the role of government in
rodent control has been recently reduced
under the new policy of relieving the
economic burden on farmers. China used to
manage rodent problems in farmland by
launching state-level or provincial-level
rodent control campaigns, with strict
coordination of rodenticide use, baiting
methods and public education. Farmers paid
part of the cost for rodent control on their
own land under the organised by
government. Without the coordination by
government, rodent control by farmers is
conducted sporadically and not concurrently.
As indicated in this chapter, chemical control
with a kill rate of less than 90% or with a
higher kill rate but only in a small area is not
280
effective. In some instances the problem
worsens follmving application of chemical
control. Therefore, government involvement
-through training farmers and coordinating
the timing of their control actions-needs to
occur for there to be effective rodent control
in agricultural systems.
A second problem is that farmers have
strong reservations about the effectiveness
of anticoagulants. Farmers seldom buy
anticoagulants in markets because these
chemicals kill rodents too slowly. The
resistance of rodents to anticoagulants could
be another reason for their poor acceptance.
This would be likely if they have been used
in the same region for many years. Although
public education is necessary, it is also
important to improve the present
anticoagulants to give a shorter kill time, and
make them more acceptable to farmers.
A third problem is that population
recovery by rodents after chemical control is
too fast to achieve sustainable control. Many
studies have indicated that the response of
rodent populations after chemical control is
non-linear (Liang 1982; Liang et al. 1984;
Zhang 1996; Huang and Feng 1998; Qi et al.
1998). Killing some individuals may reduce
the population numbers initially, but the
remaining animals have less competition for
food and nesting sites, and less social stress.
Therefore, the surviving animals have
higher productivity and higher survival
rates than untreated populations. Re-
invasion is another factor resulting in
populations returning quickly to pre-control
densities. This is illustrated by the results of
a field experiment on Mongolian gerbils
(Meriones unguiculatus). When 88% of the
population was removed, the body mass of
pregnant females was reduced from 58 g to
 Rodent Pest Management in China

35-50 g (Wanget al. 1998). Dong et al. (1991)
reported that, comparing with an untreated
area, the litter size and pregnancy rate of
Brandt's vole (Microtus brandti) increased
after 75-83% population was reduced by
using warfarin in Inner Mongolia. This
compensation in fecundity after chemical
control resulted in the population returning
quickly to its pre-poisoning density. In the
Pearl River Delta, chemical control, even
with >90% kill rate, only was effective for
less than six months (Huang and Feng 1998).
To overcome this problem of population
compensation, it is important to follow up
chemical control with other control methods
such as ecological or physical control.
Fertility control, as a sustainable and
environmentally benign control technique
(see Chambers et al., Chapter 10), is a
promising alternative control method. By
using mathematical models, Zhang (1995,
1996) demonstrated that fertility control has
an extra effect in keeping rodent populations
at a lower level, mostly due to the mating
interference by sterilised males or females
(Figure 9). Mating interference is largest
when there is a mating system involving one
male with one female, or one male with
multiple females. Other mating systems,
such as polygamy, decrease the effect of
mating interference. The greater the number
of females per male, the lower the level of
interference. Experimental studies are
urgently needed to test this hypothesis.
Rodent control in China is now facing
new challenges. The first challenge is the
likely escalation of rodent problems in the
coming century. For example, climate
change, especially warmer winters and
heavy droughts in North China, has been

0.8

(j)
> 0.6

---
~
c::
0
~
n=1
:sa. n=5
0 0.4
a. -e- n=10
-+- n=oo

0.2

o ~---'------'-------'... _ _~I ... _ _~--

10 20 30 40 50 60 70 80 90
Proportion of population which is sterile (%)

Figure 9.
The effect on the rodent population of mating Interference caused by fertility control; n is the number of
female partners per male (from Zhang 1995).

281
Ecologically-based Rodent Management
implicated in causing serious rodent
problems. The second challenge comes from
the changes in the agricultural system with
new technologies being adopted for
ploughing, planting and irrigation. Many
studies have shown that traditional deep
ploughing, canal irrigation, and plantations
over a large area are important factors in
limiting the carrying capacity of rodent
populations (Zhao 1996; Zhang et aL 1997a).
In Shanxi Province, it was estimated that
14.9 million Daure ground squirrels were
killed from 1985-1987 following canal
irrigation of 467,000 ha of arable land (Zhao
1996). In an experiment in Zhang Bei County
of Hebei Province, the density of ground
squirrels in an arable land area of 20 ha was
reduced from 1.05 individuals/ha to 0.3
individual/ha, 10 days after irrigation. In
another study, in June of 1964 in Inner
Mongolia, the rodent density in ploughed
wheat lands, wastelands and banks was 2,
84, and 312 individuals/ha, respectively,
which indicated how ploughing affects the
rodent population. Unfortunately, with the
introduction of new agricultural technology
like minimum tillage and drip irrigation or
spray irrigation, as well as more diversified
and patchy plantations, the negative effect
on rodents of traditional agricultural
systems is diminishing. The outbreaks of rat-
like hamsters and community changes of
rodent species in Shunyi District, Beijing,
appear to be related to the adoption of these
new farming systems. New efforts are
needed to deal with these new challenges.
ACKNOWLEDGMENTS
We are very grateful to Dr. Grant Singleton,
Dr. Roger Pech and Dr. Lyn Hinds for their
282
tremendous help in improving the English
writing and for their critical comments in
revising the original manuscript. This work
is supported by the national rodent key
project (96-005-01-06), the Chinese Academy
of Sciences key projects (KZ951-B1-106,
KZ952-S1-107 and STZ-I-05) and the
National Natural Science Foundation key
projects (39730090 and 39825105).
REFERENCES
Chang, W.Y., Ning,Z.DandZou. B.1998.
Ecology and management of Chinese zokor.
In: Zhang, Z.B. and Wang, Z.W., ed., Ecology
and management strategies for rodent pests
in agriculture. Beijing, China Ocean Press, 93-
113 (in Chinese).
Chen, A.G. 1996. Ecological characteristics and
control techniques of rodent pests in the
southern agricultural area. In: Wang, Z.W.
and Zhang, Z.8., ed., Theory and practice of
rodent pest management. Beijing, Sciences
Press, 247-312 (in Chinese).
Chen, A.G., Guo, c., Wang, Y., Wu, Z.J., Li, B. and
Zhang, M.W.1995. Study on the population
characteristics of Microtus fortis and the
damage they cause. In: Zhang, J., ed., Studies
of mammalogy in China. Beijing, China
Forestry Publishing House, 31-38 (in
Chinese).
Chen, A.G., Guo, c., Wang, Y., Zhang, M.W. and
Li, B. 1998. Ecology and management of
oriental voles. In: Zhang, Z.B. and Wang,
Z.W., ed., Ecology and management strate-
gies for rodent pests in agriculture. Beijing,
China Ocean Press, 130-152 (in Chinese).
Dong, W.H., Hou, X.X., Zhang, P.L., Zhou, Y.L.,
Yang, Y.P., and Xue, X.P. 1991. A study on the
population demography of Brandi's vole
after rodent control. Acta Ecologica Sinica, 11,
274--279 (in Chinese).
Feng, Z.Y., Huang, X.Q., Chen, M.L. and Shuai,
Y.H. 1990a. The spatial and temporal dynam-
ics, and the reasons for worsening rat
problems in recent years. Ecological Science,
1, 78-83 (in Chinese).

Feng Z.Y., Huang, X.Q., He, M. and Jiang, Q.L.
1996. The ecological management of rodents
in the Pearl River Delta. In: Zhang Z.L., ed.,
Proceedings of integrated pest management
in China. Beijing, China Agricultural Science
Press, 1017-1021 (in Chinese).
Feng, Z.Y., Huang, X.Q. and Yan, SX 1995. The
multi-threshold of rodent control in the Pearl
River Delta. Journal of Zhongshan Univer-
sity, 3, 70-74 (in Chinese).
Feng, Z.Y., Shuai, YH., Huang, X.Q. and Chen,
M.L. 1990b. The population recovery of
Rattus rattoides after chemical control and
optimal control time. Ecological Science, 1,
74-77 (in Chinese).
Guo, c., Wang, Y., Chen, A.G., Li, B., Zhang,
M.W. and Wu, Z.J. 1997. A study on Microtus
fortis in the Dongting Lake area. Acta Therio-
logica Sinica, 17,279-286 (in Chinese).
Guo, X.H., Zhang, Z.B., Wang, S.Q., Cao, X.P.,
Wang, F.s., Hao, S.s. andZhang,J.X.1997.
Laboratory effect of bromadiolone on rat-like
hamsters. Chinese Journal of Hygiene Pest
Control, 13,269-270 (in Chinese).
He, M. 1998. Ecology and management of Bandi-
cota indica.In: Zhang, Z.B. and Wang, Z. W.,
ed., Ecology and management strategies for
rodent pest in agriculture. Beijing, China
Ocean Press, 192-208 (in Chinese).
Huang, X.Q. and Feng, Z.Y. 1998. Ecology and
management strategies of Rattus raUiodes. In:
Zhang, Z.B. and Wang, Z.W., ed., Ecology
and management strategies for rodent pests
in agriculture. Beijing, China Ocean Press,
175-191 (in Chinese).
Huang, X.Q., Feng, Z.Y., Chen, M.L. and Yan,
S.X. 1990a. A study on damage by Rattus
rattoides to rice. Ecological Science, 1, 70-73 (in
Chinese).
Huang, X.Q., Feng, Z.Y., Chen, M.L. and Yan,
S.X. 1990b. A study on multi-threshold of
rodent control in rice-fields. Plant Protection,
1,48-49 (in Chinese).
Huang, X.Q., Feng, Z.Y., Chen, M.L. and Yan,
S.X. 1990c. The relationship between the loss
rate and infested rate of rice. Ecological
Science, 1, 70-73 (in Chinese).
Rodent Pest Management in China
Huang, X.Q., Feng, Z.Y., Chen, M.L. and Yan,
S.X.1994a. The reproductive characteristics of
Rattus rattoides. Acta Theriologica Sinica, 14,
74-78 (in Chinese).
Huang, X.Q., Feng, Z.Y., Chen, M.L. and Yan,
S.X. 1994b. The integrated control of Rattus
rattoides. In: Zhang, Z.L., ed., Integrated pest
control in agriculture in Chi na. Beijing, China
Science and Technology Press,326-330 (in
Chinese).
Liang, J.R 1982. Population recovery of Chinese
zokor (Myospalax fontanien) and plateau pika
(Ochotona curzoniae). In: Xia, W.P, ed., Alpine
ecosystems. Gansu, Gansu People's Publish-
ing Press, 93-100 (in Chinese).
Liang, J.R, Zhou, L., Wang, Z.W. and Song, RY.
1984. Mathematical model of population
recovery of Chinese zokor (Myospalax fontani-
eri) and plateau pika (Ochotona curzoniae).
Acta Ecologica Sinica,4, 1-11 (in Chinese).
Liu, S., Ning, Z.D. and Zou, B.1991. Report on the
research using the LB-explosive-tube to kill
zokors and voles. Journal of Shaanxi Normal
University, 19 (Supplement), 81-86 (in
Chinese).
Ning, Z.D., Zou, B.,Chang, W.Y.andWang, T.L.
1994. 'The damage and density threshold for
Chinese zokor control in Yuanxi County,
Northwest Loess Plateau, China. Journal of
Shanxi Agriculture, 14, 104-107 (in Chinese).
Qi, GX, Yao, W.L., Wang, J. and Yang, B. 1998.
Research on population numbers resuming
and control strategies for rodents in cities and
towns of southern China. Acta Theriologica
Sinica, 18,226-230 (in Chinese).
Wang, ex. 1996. The rodent-borne diseases in
China. In: Wang, Z.W. and Zhang Z.B., ed.,
Theory and practice of rodent pest manage-
ment. Beijing, Science Press, 38-52 (in
Chinese).
Wang, M.]., Zhong, W.Q. and Wan, X.R 1998.
Ecology and management strategies of
Mongolian gerbils (Meriones unguiculatus). In:
Zhang,Z.B. and Wang,Z.W., ed., Ecology
and management strategies for rodent pests
in agriculture. Beijing, China Ocean Press,
221-238 (in Chinese).
283
Ecologically-based Rodent Management
Wang, S.Q., Yang, H.F., Hao, 5.5. and Xu, T.Q.
1991. The food habits and intake of the rat-like
hamster. Zoologica Sinica, 8, 53-70 (in
Chinese).
Wang, S.W. and Ye, D.Z. 1992. Long-term fluctu-
ation of climate. In: Climate variability-
Proceedings of the International Workshop
on Climate Variability, July 13-171992,
Beijing, China, 23-32.
Wang, Y., Guo, c., Li, B., Wu, Z.J. and Chen, A.G.
1997. The multi-threshold of rodent control.
Modern Agriculture Research, 18, 185-187 (in
Chinese).
Wang, Y.s., Lu, H.Q., Su, C.D. and Wang, Y.Z.
1996. The multi-threshold for mixtures of
populations of several rodent
Chinese Journal of Zoology, 31, 19-22 (in
Chinese).
Wu, L., Zhang, M.W. and Li., B. 1998. A study on
the food composition of Microtus fortis in the
Dongting Lake area. Acta Theriologica Sinica,
18,282-291 (in Chinese).
WU,Z.J.,Chen,A.G.,Li, B.,Guo,c., Wang, Y.and
Zhang, M.W. 1996. A study on the breeding
characteristics of the oriental vole (Microtus
fortis) in the Dongting Lake area. Acta Therio-
logica Sinica, 16, 142-150 (in Chinese).
Zhang,Z.B.1995. Theecologicalfundamentalsof
fertility control on rodents. Acta Theriologica
Sinica, 15,229-234 (in Chinese).
Zhang, Z.B. 1996. The techniques and strategies
of rodent control by contraception. In: Wang,
Z.W. and Zhang Z.B., ed., Theory and
practice of rodent pest management. Beijing,
Sciences Press, 367-378 (in Chinese).
Zhang, Z.B. and Wan, Y.L. 1997. Ten years'
review on rodent pest management in China.
In: Niu, D.s, ed., Research on agricultural
biology and sustainable development of
agriculture. Beijing, Sciences Press, 146-151
(in Chinese).
284
Zhang, Z.B., Wang, S.Q., Hao, 5.5., Cao, X.P. and
Wang, p.s. 1997a. Effect of alpha-chlorohy-
drin on male white rats in laboratory. Acta
Zoologica Sinica, 43, 232-233 (in Chinese).
Zhang,Z.B., Wang,F.S.,Hao,S.S.and Wang,S.Q.
1996. A report on the trial of a new multiple
trap. Plant Protection, 22, 46-48 (in Chinese).
Zhang, Z.B., Wang, S.Q., Hao, 5.5., Wang, P.5.
and Cao, X.P. 1997b. The sterility effect of
alpha-chlorohydrin on male rat-like hamster,
Cricetulus triton. Acta Theriologica Sinica, 17,
232-233 (in Chinese).
Zhang, Z.B., Yang, H.F., Wang S.Q., Hao 5.5.,
Wang P.5., Cao, S.P., J.X.1998.
Ecology and management the rat-like
hamster. In: Zhang, Z.B. and Wang, Z.W., ed.,
Ecology and management strategies for
rodent pests in agriculture. Eeijing, China
Ocean Press, 1-19 (in Chinese).
Zhang,Z.B.,Zhu,J, and Yang, H.F.1991. The
estimation on the population reproductive
parameters of rat-like hamsters, Zoological
Research, 12, 253-258 (in Chinese).
Zhang, Z.E., Zhu,J. and Yang, H.F. 1992. The
estimation on the mortality of
rat-like hamsters. Acta Zoologica Sinica, 38,
151-155 (in Chinese).
Zhang, Z.E., Zhu, J., Yang, H.F., Wang, S.Q.,
Hao, S.S., Wang, F.S. and Cao, X.P. 1997c.
Seasonal and spatial dynamics of burrow
holes ofthe rat-like hamsters. Chinese Journal
of Zoology, 32, 32-33 (in Chinese).
Zhao, G.Z. 1996. Rodent pest control and plant
protection. In: Wang, Z.W. and Zhang Z.B.,
ed., Theory and practice of rodent pest
management. Beijing, Sciences Press, 19-37
(in Chinese).
Zou, E" Ning, Z.D., Wang, T.L. and Chang, W.Y.
1998. Ecology and management strategies for
the Chinese zokor (Myospalax fontanieri). In:
Zhang, Z.B. and Wang, Z.W., ed., Ecology
and management strategies for rodent pests
in agriculture. Beijing, China Ocean Press, 41-
64 (in Chinese).
13. Rodent Pest Management in the
Qinghai-Tibet Alpine Meadow
Ecosystem

Naichang Fan, Wenyang Zhou, Wanhong Wei, Quanye Wang and

Yongjin Jiang

Abstract

The available area of the natural grasslands of the Qinghai-Tibet Plateau is about
1.4 million km 2 . As a result of inappropriate reclamation and over-grazing in the past
decades, serious degeneration of up to 0.71 million km 2 of the grasslands has
occurred. Of this area, 0.37 million km 2 has been damaged by rodents and about
40,000 km 2 of black sandy soil has been formed due to rodent infestation. The
plateau pika (Ochotona curzoniae) and the plateau zokor (Myospalax baileY/) are the
two dominant rodent species .
Rodent control is essential for reversing the heavy degeneration of the grassland
so that it can be used again for grazing. Beginning in the 1960s, more than ten types
of rodenticide have been used for controlling rodents in the Qinghai-Tibet Plateau . In
order to reduce the risk of rodenticides to predators and to improve baiting
efficiency, a baiting machine was invented which puts baits into the rodents'
underground tunnels, based on the invading behaviour of zokors . Both the baiting
and killing efficiencies, as well as the safety advantages of using the baiting
machine, are greater than the traditional, manual method of ground baiting.
Since the mid-1980s, studies have shifted to developing a sustainable strategy for
managing pika and zokor damage by understanding their ecology and interaction
with the grazing activities in the region. A demonstration area of 200 ha was set up
in a heavily degenerated region with black sandy soil. An integrated management
program, which included use of a baiting machine , seeding, fencing, control of
weeds and control of grazing intensity, was implemented. The vegetation and the
productivity of the grassland were increased shortly after treatment began. An
increase of about 648.4 t of dried grasses was observed in the area during the next
three years.

Keywords

Alpine meadow, Qinghai-Tibet Plateau, plateau pika (Ochotona curzoniae) , plateau

zokor (Myospalax baileY/), integrated pest management

285
Ecologically-based Rodent Management
INTRODUCTION
HE AREA of natural grassland in
T the Qinghai-Tibet Plateau is about
1.4 million km2, with alpine
meadow being the most widespread
vegetation type. This area is an important
base for animal husbandry. As a result of
inappropriate reclamation and overgrazing
in past decades, the grass lands have been
seriously degenerated. These degenerated
grasslands comprise about 0.71 million
km2, of which 0.37 million km2 is infested
with rodents. The main pest rodents are
plateau pika (Ochotona curzoniae), plateau
zokor (Myospalax baileyi), Ochotona daurica,
Pitymys irene and Marmota himalayana.
Plateau pikas and plateau zokors are the
dominant rodents and their feeding and
burrowing activities damage grasslands.
About 40,000 km 2 of black sandy soil has
been formed as a result of rodent
infestation. In the grasslands of the
Qinghai-Tibet Plateau, the average
densities of the plateau pika and plateau
zokor are more than 4.29 individuals/ha
and about 1.07 individuals/ha,
respectively. These rodents compete with
livestock for food resources. They consume
about 0.15 billion t of fresh grass every year,
which is equal to the total food intake of
0.15 billion sheep. Rodents also dig and
destroy vegetation causing many serious
problems such as soil erosion, and
reductions in livestock carrying capacity
and ecosystem biodiversity.
Zinc phosphate, a rodenticide, was first
used for rodent control in the Qinghai-Tibet
area in 1958. During the early 1960s, the area
of grassland treated with zinc phosphate
286
was more than 333 km2 in southern Qinghai.
From 1964 to 1965, more than 26,667 km2 in
20 counties was treated using both zinc
phosphate and '1080' (fluoroacetate). The
area of rodent infestation was reduced from
54,000 km2 in the 1960s to 38,130 km2 in 1990.
Cumulatively, more than 208,000 km2 of the
infested area was treated with rodenticides
during this period.
However, zinc phosphate and 1080 also
caused many serious social and
environmental problems. Both are acute
poisons that have secondary poisoning
effects, and are unsafe for non-target species
including humans. With the appearance of
anticoagulants such as diphacinone,
diphacinone-Na, gophacide, difenacoum,
bromadiolone and brodifacoum, use of the
acute poisons was no longer permitted. A
new type of rodenticide, botulin C, was also
found to be very effective in killing plateau
pika and plateau zokor in the grasslands.
The killing rate with botulin C was up to
98%, with less environmental pollution and
no secondary poisoning effects on other
animals (Shen 1987).
Although anticoagulants and botulin C
are effective in reducing pika and zokor
damage initially, the populations of these
rodents recover rapidly after treatment
(Liang 1982). Since the rnid-1980s, studies
have shifted towards developing a
sustainable strategy for managing pika and
zokor damage by understanding their
ecology and interaction with the grazing
activities in this region. In this chapter, the
major achievements of these studies are
reported and future research priorities are
discussed.
 Rodent Pest Management in the Qinghai-Tibet Plateau

ECOLOGICAL ASPECTS OF THE PLATEAU
PIKA AND PLATEAU ZOKOR
Plateau pika
Habitat
Pikas mainly inhabit the plateau steppe,
steppe meadow, plateau meadow, alpine
meadow and alpine desert steppe at an
elevation of 3,100-5,100 m above sea level.
They prefer open habitats and avoid dense
shrub or thick vegetation (Shi 1983). Table 1
shows that the number of pika burrows
decreases with increased vegetation cover
and height.
Burrow systems
The burrow systems of the pika comprise
two types. One type is the simple or
temporary burrow that is shallow and short.
It is mainly used in summer and usually has
two or three openings, but may only have
one opening to the surface. The other type is
the complex burrow system which occupies
areas of 21-162.14 m 2. The average length of
the tunnels is 13 m, with a maximum length
of up to 20 m. The average depth of the
burrows is 0.33 m, but they may be up to 0.6
m deep. The burrow system has many
branches which are connected to each other
to form a complex network, sometimes with
two layers. There are usually five or six
openings, although some have 13 openings.
The diameter of the openings is about 8-12
cm. There is one nest in the burrow system,
which is located about 0.45 m below the
surface (Xiao et al. 1981).
Feeding behaviour and activity rhythms
The plateau pika is a herbivore and
responds variably to different plants and
plant parts. Enclosure experiments indicate
that of 31 plant species in the natural habitat,
pikas feed on 23 grass species (mainly
belonging to the Gramineae), as well as
species belonging to the Cyperaceae and
Leguminosae. An adult pika consumes (on
average) 77.3 g of fresh grass per day, which
is about 50% of its body weight, whereas a
375 kg cow consumes 18 kg of fresh grass per
day, which is only 4.8% of its body weight.
The food intake of 56 adult plateau pikas
equals the food intake of one Tibetan sheep
(Pi 1973).

Table 1.
The relationship between the cover and height of ground vegetation and the number of active burrow holes
of plateau pika (Shi 1983).

Habitat typea 1 2 3 4
~
No. of plots 9 41 3 8
Vegetation cover (%) 84.22 4.09 96.93 0.84 66.67 8.28 90.63 2.90
Vegetation height (cm) 9.22 2.57 69.29 4.25 2.00 0.00 85.00 6.61
No. of pika holes 24.67 6.74 3.93 1.54 43.33 8.82 0.00 0.00

a Key to habitat types: 1 = natural grassland; 2 = original grassland which was cultivated for several years, then
abandoned due to deterioration; 3 = grassland which was ploughed but then considered unsuitable for
cultivation; 4 =grass sown in deteriorated grassland. Values are Mean standard error of the mean.

287
Ecologically-based Rodent Management
Feeding comprises 63-78% of the total
activity of pikas. When feeding, pikas look
around frequently and have a special
feeding pattern termed 'pecking'-after
feeding for a moment they look around or
move a short distance, then feed again.
Feeding frequency is 5.7 1.3 pecking bouts
per minute (Bian et a1. 1994; Fan and Zhang
1996).
The plateau pika is a diurnal animaL Its
above-ground activities appear to have two
peaks (forenoon and afternoon), and these
change according to the different seasons. In
October the first activity peak occurs at 09:00
h and the second at 18:00 h, while in July the
activity peaks are at 08:00 hand 19:00 h,
respectively.
Life span
Based on observations of 401 ear-tagged
individuals over three years, the average life
span of a plateau pika is 119.9 days, with the
longest life span recorded being 957 days
(Wang and Dai 1989).
Home range and territory
Plateau pikas live in family groups and
show territorial behaviour. Before the
reproductive period (March), the average
home range is 1,262.5 m 2, while in the
breeding season (April) the home range
expands to 2,308 m 2 . Plateau pikas protect
their territories all year. In the breeding
season, males and females form pairs to
establish new families and new home ranges
and protect this territory. During the period
of oestrus, the female holds no territory, but
the male monopolises the female and drives
away other invading males (Wang and Dai
1990).
288
Reproduction and sex ratio
Each year an adult female can produce
3-5 litters with 1-9 individuals per litter. In
the Kuaierma region (3726' N, 9842' E), a
total of 1,529 plateau pikas were captured
from 1964 to 1965. Males made up 46.5% of
the total capture and were less abundant
than females (X2 = 7.35, P < 0.01). The
average litter size was 4.55 0.95
individuals. In the Duofudun region (3515'
N, 101 43' E), a total of 777 pikas was
captured. Males made up 53.16% of the total
capture. The average litter size was 4.68
1.29 individuals (Shi et al. 1978).
Plateau zokor
Habitat
The plateau zokor mainly inhabits the
alpine meadow, steppe meadow, alpine
shrub, farmland, banks and wasteland and is
widely distributed in Qinghai, southern
Gansu and western Sichuan. Their choice of
habitat is characterised by moist soil and
degenerated grassland with many weeds.
Burrow system
The plateau zokor constructs a very
complicated burrow that consists of one or
two main nests, feeding and transportation
tunnels, food store holes and blind endings.
Feeding tunnels, which are formed during
feeding activities, are 60-100 mm deep and
70-120 mm in diameter. The transportation
tunnels are about 200 mm deep, and are
smooth, large and stable paths from the
main nest to the feeding tunnels, with some
holes for storing food built nearby. One or
two vertical tunnels connect the transport-
ation tunnels to the main nest, which is
0.5-2 m deep, about 150-290 mm in
 Rodent Pest Management in the Qinghai-Tibet Plateau

diameter, and often filled with dry and soft
grass. The nests of females are commonly
deeper than those of males. The holes for
food storage and a defecation site are usually
near the nest (Fan and Gu 1981).
Feeding habits
Plateau zokors mainly feed on roots,
rhizomes and other underground parts of
weeds. They also frequently pull parts of
plant stems into the burrow as food or nest
material. Only the rhizomes and green
leaves of Gramineae spp. are consumed.
Potentilla anserina and Aiania tenuifolia are
also important food resources for the zokor
in Kobresia flllmiIis meadows. Zokors store
root tubers of P. anserina and subterranean
stems of A. tenuifolia mainly as food over
winter (Fan et al. 1988).
Activity rhythms
Although plateau zokors live in the
subterranean environment, they exhibit
circadian rhythms in the dark burrow
systems. Based on a study of 80 marked
animals, two peaks of digging and feeding
activity occurred in summer and autumn.
These were from 15:00 h to 22:00 h (making
up 65.3% of the total day's activities) and
from 0:00 h to 7:00 h (making up 21.6% of the
day's activities). In winter, the activity
frequency is low and the daily activities are .
mainly limited to near the nest at 12:00-22:00
(making up 79.7% of the total day's
activities) (Zhou and Dou 1990).
Home range
The home range of zokors changes
between seasons. In the reproductive season
(spring), the home range of males
(499.0 390.9 m 2) is larger than that of the
female (21.0 11.8 m 2). In the other seasons,
there are no significant differences in the
home ranges of males (156.5 45.4 m 2) and
females (162.1 153.9 m 2) (t =0.332, P > 0.05)
(Zhou and Dou 1990).
Reproduction
The plateau zokor has one litter of 1-5
individuals (2.91 1.08) every year. About
38.4% of litters are all males (Zheng 1980).
The reproductive period is from March to
July with a lactation period of about 50 days
(Zhang et al. 1995). A radio-telemetry study
indicated that adult males and females never
live together, even when females are in
oestrus. Mating occurs at the intersection of
two tunnels of a male and a female burrow
system. An analysis of 20 burrow systems of
zokors showed that, in the mating period,
the male digs a few long tunnels to intercept
the female burrow systems, and usually
these tunnels have two branches in order to
increase the chance of meeting a female. The
situation where two males meet each other
has never been observed. A male may mate
with several females and a female also may
mate with a few males. Therefore the mating
system of zokor is probably promiscuous. In
the Fengxiakou area of Menyuan County,
the sex ratio of the adult male to female is
1:1.67. The mosaic distribution of male and
female home ranges increases the chance of
meeting with each other (Zhou and Dou
1990).
Invading behaviour
Field studies showed that a burrow
system was often occupied rapidly by its
neighbours if the host zokor was removed
(Fan et al. 1990). This observation indicates
that zokors tend to invade any vacant
territory. This invading behaviour was

289
Ecologically-based Rodent Management
further studied in the laboratory, using two
large boxes connected by a wooden tunnel.
Each box contained one zokor and was filled
with damp soil so that the zokor could build
its burrow system. After the burrow systems
in both boxes were built, one zokor was
removed, and the remaining animal's
behaviour was monitored. In six
experiments, the remaining zokor invaded
the other burrow system rapidly, usually
within 2-30 hours. In four of the six
experiments, the burrow tunnels but not the
nests were occupied; in the other two cases,
both the tunnels and nests were occupied
(Fan et al. 1990).
Based on this observation, a baiting
machine, which digs tunnels under the
grassland and puts baits in the tunnels, was
invented. It was found that zokors used
some parts of these artificial tunnels. The
longest section used by zokors was 32 m and
the shortest section was 0.3 m. Most of the
artificial tunnels created by the baiting
machine were incorporated into the zokors'
natural burrow systems (Fan and Gu 1981;
Fan et al. 1990).
Aggressive behaviour
The behaviour patterns of the plateau
zokor are classified as follows: sleeping and
resting, feeding, carrying food, digging, self-
grooming, moving, exploring, approaching,
contacting, attacking, escaping and
retreating. In the reproductive period, the
duration of digging, approaching,
contacting, attacking, escaping and
retreating increases and mutual tolerance is
higher than that in the non-reproductive
period. Mutual tolerance between zokors of
the same sex is lower than that between
males and females (Wei et al. 1996).
290
Digging behaviour
The zokor is a fossorial animal. Its front
claws are very stocky and strong. The
digging process is divided into seven steps:
excavating, digging up, kicking, pushing,
humping up soil, feeding and modifying the
tunnel (Wang et al. 1994). By digging, zokors
build burrow systems for foraging, hoarding
food, reproduction and avoiding predators.
One zokor was observed to push 1,023.82 kg
of dry soil to the ground surface (30.8% in
the grass-greening period, 6.3% in the grass-
growing period and 62.9% in the grass-
withering period) and form 242.1 mounds
per year. The mounds covered about
22.53 m 2 of the grassland (Table 2). In the
grass-greening period (April-June) which
corresponds to the reproductive period of
the zokor, the level of digging is medium.
The level of digging is lowest in the grass-
growing period (July-August) because the
zokor can feed partly on green plants above
ground. The level of digging is highest in the
grass-withering period (September-
November) because dispersal and hoarding
activities of the zokor are high (Wang and
Fan 1987; Fan and Gu 1981).
 Table 2.
The number of and area covered by mounds of the plateau zokor from April to November (Wang et al. 1994).

Month Stage of grass No. of zokors Mounds/zokor Mound volume Area covered by Dried weight of soil
growth (m 3 ) each mound (m 2 ) for each zokor (kg)
-1L- ----IL-
-:Jr-
-------'''---- _.L....-

April ~l Greening JC 29 _._L 36.00 4.42 Il

2.33 li
87 .60
May
=[ Greening JC 25 ~C 28.98 ~C 7.42 =:le 2.61 :JC 113.77
:::!Cl
Cl

-
.. .. "'Ir- Cl..
-l[~- --~r-
June IL- Greening 22 23.96 11
9 .16 If" 2.58 '.
I.
113.40 Cl)

==
July ~C Growing
.JC
27 =:lC 10.06 ::JC 6.48 :JC 0 .90
=ll:. . 2.94

-
"'V
August Growing lr- - 33 -;1 4 .03 1I 5.34 Il 0.32 ~C 8.68
Cl)
III

September
=
Withering JC 33 =:lC 29.97 ::JC 8.75 :=JC 3.02 =:JC 157.80
==
October d
Withering li 36 -.r 95 .04 lL 8.05 JL 9 .09
~r--

_L 403.31
\11
==
\11
(IQ
November I[ Withering JC 35 ~C 14.04 ~C 10.14 ::JC 1.68 -=:JC 116.32 Cl)

lr- I, :3
1I -11
-
-~r
Total 242.08 1I 22.53 I 1023.82
Cl)

==

-=-
==
Cl)

e
==
(IQ
=-
\11
T
:!

-
er
Cl)

~
.... -
"'V
iii
Cl)
\11
C
Ecologically-based Rodent Management

INTERACTIONS BETWEEN RODENT sheep over this period (2.7 million sheep in
DAMAGE AND GRASSLAND 1959 to 1.4 million sheep in 1984) (Table 4) .
DEGENERATION Up to 1991, there was a t leas t 547.6 km 2 of
black sandy soil created by rodent activity
The Qinghai- Tibet alpine meadow following grassland degradation, with the
ecosystem is very vulnerable to disturbances area of grassland available for winter and
that reduce the coverage and height of the spring use being only 5,873.6 km2- 45.5% of
grass vegetation, such as over-grazing by the tota l area . In theory, this area has a
livestock and cultivation. Weeds, as well as carrying capacity of on ly 1.03 million sheep,
rodents, will invade degenerated grass land. but is actually grazed by 1.30 million sheep
Jing et a1. (1991) reported that the number of (Ma 1991). Therefore, in the first year after
zokor mounds increased with increased management strategies have been
grazing intensity (r = 0.795, rO.05= 0.707, implemented (e.g. chemical control and re-
p < 0.05) (Table 3). seeding or fencing), livestock grazing should
be prevented, with on ly low-level grazin g
Table 3 . permitted in the second year; this will
The density of zokors under different grazing preven t further degenera tion of the
intensities (Jing et al. 1991). grassland through rodent infes tation.
Grazing intensity Zokor density
A study by Shi (1983) showed that the
Plot a
( sheepjhaj yr) (moundsjO.25 ha) population densities of the p lateau pi ka and
the p la teau zokor in abandoned, cultiva ted
1 5.30 46 2
grassland was much higher than that in
2 4.43 198
origin al grass land (Table 5) . C ultivation and
3 3.55 164
overgrazing by humans created the suitable
4 2 .68 4
habitats for pika and zokor.
5 1.80 152
In summary, over-grazing by livestock
6 6.07 362 and the inapprop riate recla mation of the
7 3.12 270 grasslands for o ther purposes are the major
8 2.12 68 reasons fo r grassland degenera tion. In the
a Pl ot s 1 to 5 are in grasslands on the slope of a presence of invasion by weeds and roden ts,
mounta in; plots 6 to 8 are in grassland on the the d egenerated grassland w ill con tinue to
pl ain.
be changed as follows: over-grazin g by
livestock lead s to grassland degenera tion
In Tianjun County of Qinghai in 1959, the which results in rodent infestation and
area of a vailable grasslan d was 14,664 km 2, further grassland degeneration. H u m an
but decreased to 12,901 km 2 by 1984 because activities, especiall y cultivation and
of over-grazing and rodent infesta tion. It is livestock gra zing, play an important role in
estima ted that the livestock carrying this vicious circle.
capacity was decreased by 1.33 million

292
 Rodent Pest Management in the Qinghai-Tibet Plateau

Table 4.
Comparison of the area of different grassland types, the yield of fresh grasses and the livestock carrying
capacity between 1959 and 1984 (Ma 1991).

Grassland type Grass yield Area of grassland Carrying capacity

(kg/ha) (km 2 ) (sheep/ha)

Steppe 1959 2052 .0 2284.67 1.45

1984 1467.3 2156 .20 1 .01
Meadow 1 9 59 1731.0 12303.07 1 .40
1984 1278.3 7368.60 0 .88
Shrub 1959 3154.5 384 .60 2.17
1984 2049.3 382.53 1.42
Swamp 1959 29 80 .5 3560.00 1.5 5
1984 19 51. 8 2882 .53 0.46
Degenerated grassland a 19 59 1 79 5 .5 105.80 1.15
19 84 586 .7 2484.20 0. 33

a 'Note th at t he area of degenerated grassland has increased greatly during th is period .

Table 5.
The community composition and population density (individuals/0.25 ha) of small mammals in various
grassland types (Shi 1983).

Species Common name Artificial Semi-artificial Secondary Wasteland

grassland grassland grassland
Microtus Root vole 70.1 (91.3)a 54 .3 (85 .9) 0 (0 .0 ) 0 .0 (0 .0)
oeconomus
Cricetulus Lesser long- 1.1 (1.4) 0 .6 (1.0) 0 .7 (1. 1) 7 .5 (69 .5)
longicaudatus ta iled hamster
Ochotona Gansu pika 0 .0 (0.0) 5.6 (8 .8) 0 .2 (0 .3) 0. 0 (0.0)
cansus
Myospalax Plateau zoko r 5.6 (7 .3) 2. 7 (4.3) 11.1 (17 .8) 0.9 (8 .3)
baileyi
Ochotona Plateau pika 0.0 (0.0) 0 .0 (0 .0) 50 .6 (80 .8) 2.4 (22 .2)
curzoniae
Total 76.8 (100.0) 63.2 (100.0) 62.6 (100.0) 1 0.8 (100.0)

aFi gures in brackets re present th e percentage of eac h species in each habitat.

293
Ecologically-based Rodent Management

POPULATION DYNAMICS AND meadow, which is not their primal"y habitat,

THEIR PREDICTION is only 18.99 individuals /ha.
A long-term study has shown that pika
Plateau pika populations change significantly between
years (Figure 1). The factors causing these
The total available grassland area of
fluctuations are not clear, but populations
Menyuan, Qilian, Haiyian, Gangcha and
appeared to be oppressed after heavy snows
Tianjun cOlmties in Qinghai Province is
in 1982 and 1988 (Zhou and Wei 1994). In a
about 36,800 km 2, of which up to 12,000 km 2
study conducted from 1985 to 1988, using
(32.7%) is occupied by plateau pikas (Wang
ear-tagged animals, 57% of the animals that
et al. 1996) (Table 6).
reproduced in the population were those
which were born in the first litter of the
Table 6.
previous year and 26% were from the second
The population densities and area occupied by
plateau pikas and plateau zokors in the useable litter of the previous year. Most pikas from
grassland in some counties of Qinghai Province the third and fourth litters died before
(Wang et al. 1996). winter. Therefore, the survival rate of the
first litter not only decides the population
Species Density Area occupied Percentage
(ha) (km2 ) (%) numbers for the present year, but is also an
important determinant of the population
Pika <20 3872 10.52
density of pika in the following year (Wang
20-70 5633 15.31
and Smith 1988). Field observations also
>70 2527 6 .87
indicated that if the reproductive period was
Total 12 032 32.70
extended, the adult females had more litters
Zokor <20 660 1.79 but the mortality rate of the young pika was
20-70 433 1.18 higher. Therefore, the population density
>70 3370 9.16 was lower in the next year (Wang and Smith
Total 4463 12.13 1988).

According to a study by Liu et al. (1980) at Plateau zokor

the Haibei Alpine Meadow Ecosystem
Research Station of the Chinese Academy of
Sciences, plateau pikas prefer habitats such as
banks, terraces, foothills and gentle slopes in
the K. hum.ilis meadow and the steppe
meadow. The highest population density of
the plateau pika was fOlmd in wasteland with
an average density of 74.64 individuals/ha.
The second highest density of plateau pikas
(70.36 individuals/ha) was found on the
banks and gentle slopes. The average
population density in the Kobresia capillifolia
In the total area of available grassland in
Mengyuan, Qilian, Haiyian, Gangcha and
Tianjun counties of Qinghai province,
12.13% was inhabited by zokors. Highest
densities of zokors occurred over an area of
3,370 km 2 (Wang et al. 1996) (Table 6).
One long-term study at Haibei station
from 1976 to 1992 showed that the zokor
population was relatively stable until 1987
when it declined coincident with large sca le
rodent control (Figure 2) (Zhou and Wei
1994). One explanation for the stable

294
 Rodent Pest Management in the Qinghai-Tibet Plateau

numbers is that the effects of climate and Plateau pika

natural enemies are minimal (Wang and Fan
1987).
The population density of zokors in
autumn is closely related to the numbers of
reproductive females in spring. Density
increases after reproduction and is about
1.56 times higher in autumn than in spring.
From autumn to the next spring, the
population decreases. The population of
zokors in spring or autumn can be predicted
by using the following regression models
(Zhang et a1.1991):
Y Autumn 4.864 + 0.9672XSprmg .
(r 0.9823> r 0.01 = 0.824, P < 0.01) (1)
where is the autumn density of the
present year, and XSpring is the spring density
in the present year; and
Yspr1'ng 1.426 + 0.7664XA utuml1
(r 0.8930 > r 0.01 = 0.824, P < 0.01) (2)
Pikas cause a lot of damage to the grass
vegetation, mainly by feeding, gnawing
grass roots and burrowing. The damage
level to the grassland is closely related to the
population density of pikas. The infested
area of the grassland escalates with
increasing rodent density and high grazing
intensity by livestock. Some climatic factors
may also influence the level of rodent
damage to the grassland (Liang and Xiao
1978; Xiao et a!. 1981).
Liu et al. (1980) reported that the
population density of the plateau pikas was
positively correlated to grassland damage.
The relationship between pika density and
grass damage can be used to calculate the
economic threshold for each habitat using
the formula:
Y=a +blogX (3)

where Y Spring is the spring density of the where Y = % of grass damage and
present year, and XAutumn is the autumn X = pika/ha, and solving for Y = 0
density of the previous year. (Table 7).

DAMAGE ASSESSMENT AND

Plateau zokor
ECONOMIC THRESHOLD
Plateau zokors destroy grasslands by
According to the definition of Headley
feeding and digging. Digging mainly takes
(1972), 'economic threshold' is the
place 40-160 mm underground, which
population density of a pest when the cost
destroys most plant root systems (90.7% of
for its control equals the minimal value of
the total underground biomass). The
the product that is lost because of the pest.
original vegetation of grasslands disappears
Therefore, the costs of rodent control should
and seriously degenerated grasslands or
be less than the losses occurring in the
even large areas of black sandy soil are
absence of control, and the benefits should
formed. Field studies indicate that, with
be maximised. Further, when the cost of pest
increases in zokor density, the total output of
control in a unit area equals the economic
grasses decreases while the total output of
loss due to rodent damage, the pest density
weeds increases (Fan et a1. 1988) (Table 8).
is at the economic threshold where rodent
control should be undertaken.

295
Ecologically-based Rodent Management

350
300
250
~ 200
-Ui
c
Q)
0 150
100
50
0
(Q C\J
r--
Q)
co
Q)

Year

Figure 1.
Population densities ofthe plateau pika (individuals/hectare) (Zhou and Wei
1994).

30

25

20
.~
Cl)
c 15
Q)
0
10

0
(Q C\J M ~ ~ (Q r-- co Q) 0 C\J M
r-- co co co co co co co co Q) Q) Q) Q)
Q) Q) Q) Q) Q) Q) Q) Q) Q) Q) Q) Q) Q)

Year
Figure 2.
Population densities of the plateau zokor (individuals/hectare) (Zhou and Wei
1994).
Table 7.
The relationship between the density of pika and the grass damage used to define economic thresholds
(Y = percentage of grass damage, X = pika/ha) (Liu et al. 1980).

Habitat y= a + blogX Economic threshold (pika/ha)

All areas Y = -115.46 + 120.681ogX 9.05

First terrace y= -104.42 + 194 .731ogX 5.95
Second terrace y= -70 .53 + 101 .871ogX 4.92
Gentle slope Y = -63.54 + 88.151ogX 5.26
Hills Y = - 40.13 + 63 .121ogX 4_32

296
 Table 8.
The relationship between the density of plateau lokor, the damage to grass and the yield of fresh vegetation types (Fan et al. 1988).

Density rank Average density1 Damage 2 Area of mound 3 Grass yleld 4 Sedge yieldS Weed yieldS Total yield 7
(zokors/ha) (zokors/ha) (%) (m 2 ) (kg/ha) (kg/ha) (kg/ha) (kg/ha)
-...lL..- ---lL ..Ja.. ..JL .JL
0 I 0 _ .J[ _ 0 0 _J 3020 I~ 2360 11 8212 .. :. 13592
:::cl
1-10 -::JC 6.7 -=:JC 13 ::::JC 116 =:J[ 2797 J[ 2158 J[ 6229 J[ 11184 o

-=
Cl.
I1 ID
14.5 !r I
11-20 t ,. 18 212 2277 1206 II 6750 I 10233
=:JC =:JC =:JC I[ JC ..J[ J[
-=
21-40 30.2 45 481 1046 666 5483 7195 ."
ID
41-70
----,.----
62.9
=:Jr-
64 1076 JL 690
--,-
75 . . JI 7180 - .1 7945
In

3:
>70 91.6 84 1489 229 0 1 1 4450 4679 I
I
Coefficient of correlation r2,1 = 0.982 r3,1 = 0.999 r4 ,1 = -0.937 r5,1 = -0.905 r6,1 = -0.634 r7,1 = -0.900 IJQ
ID
p < 0.01 p < 0.01 p < 0.01 P < 0.05 p> 0.05 p < 0.05 :3
Notes:
-==
ID

-e
(1) Numbers in superscript in the column headings relate to the correlation coefficients at the bottom of the table. (2) Significant levels of correlation :
ro.os = 0.811; rO.Ol = 0 .917.
=-
ID

=
IJQ
=-
I
T'
=!

-
er
ID

~
...., -
."
iii
ID
I
I::
Ecologically-based Rodent Management
Zokors not only gnaw grass roots but also
push large amOLmts of poor quality soil to
the ground surface, forming mounds of
different sizes (Table 2). These mounds
cover the original vegeta tion and the result is
degradation of the grassland. The mound
volume of the zokor varies with sex, age and
seasons. Generally, the mOLmd volume of
males is larger than that of females and the
mounds of adults are larger than those of the
young. The annual losses of grassland due to
damage by feeding, hoarding and covering
of vegetation by mounds under different
zokor densities (Tao et al. 1990) are listed in
Table 9.
In the Haibei region of Qinghai Province,
K. hUlllilis is the dominant plant species in
the alpine m eadow. In 1987, the optimal
grazing intensi ty was 3.29 sheep/ha. The
maximum productivity of the grass was
3,554.4 kg / ha. The price of each sheep was
110 RMBl / sheep. So the p rice of grass
(Yu RMB /kg) is:
Y, = 110 x 3.29 /3554.4 = 0.102 (4)
The cost of di p hacinone-Na baits Llsed to
control plateau zokor was 3.105 H.MB/ha
and the achieved killing rate was 74.4%. In
theory, when the ki lling rate nears 100%, the
rodent damage would be reduced to zero
1RMB = renminbi y uan. As of May 1999,
8.14 RMB = US$l.
Table 9.
Vegetation damage for different densities of zokors over one year (Tao et al. j,990).
Density 1 3 4 6
(zokor/ha)
Dry grass losses 14.75 35.55 39.85 50.80 75.75
(kg/ha)
298
and, at this point, the cost of control (Y2 ,
RMB/ha) using baits is:
Y2 = 3.105/74.4% = 4.173 (5)
The amount of grass (Yo' kg/ha) which
equals the cost of control should be:
Yo = Y/O.102 = 4.173/0.102 = 40.92 (6)
The correlation model between the grass
losses (Y, kg/ha) and density of zokors (X,
zokors/ha) is:
Y = 1l.2X - 5.61 (7)
If Y = Yo = 40.92, then the economic
threshold Xo = 4.16. That is, when the
population density of zokors reaches the
level of 4.16 zokor /ha, control by baiting
should be undertaken if significant
economic losses are to be avoided.
INTEGRATED PEST MANAGEMENT
(lPM)
An experiment that aimed to find a
sustainable solution to rodent damage in the
Qingha i-Tibet alpine mea dow ecosystem
was conducted from 1984 to 1989 in the
Haibei region. The study area, located at
Panpo, Menyuan County, Qinghai Province,
was serioLlsly infested by plateau pikas and
plateau zokors. From 1984 to 1986, the
above-ground mow1d density of the plateau
zokor was 2,683 mounds/ha, and the active
7 11 13 14 19 24
132.69 152.81 153.66 235 .67 266.69
 Rodent Pest Management in the Qinghai-Tibet Plateau
burrow density of pla teau pika was 752
holes/ ha. The seriously infes ted area m ade
up 53.35% of the to tal grassland area . The
vegetation types are m ainly K. humilis
mead ow, and Dasiphoia j i'uticosa slu'ub and
swamp meadow, and the soil types are
platea u m ead ow soil, swa mp soil and
plateau shr ub m eadow soil. A series of
measures, includ ing chemical control,
sowing, fencin g, control of grazin g in tensity
and weed control, were carried out in a
demonstration area of 200 ha.
The baiting machine
In spring 1987, 0.075% diphacinone-Na and
0.6% gophacide grain baits were deli vered
into simulated w1derground tW1l1els with
the aid of the baiting m achine. About 19 g of
bait was dropped every 5.5 m in each lin e.
The in terval between lines was 10 m. The
baiting machine delivered the baits at the
speed of about 3.3 ha / hr, abou t 20 times
fas ter than m anual baiting. The killing rates
with diphacin one-Na and gop hac ide ba its
were 85.1% and 77.3%, respectively, for
zokors and 96.9% and 99.8%, respec tively,
for plateau p ikas Ging et al. 1991) (Table 10),
Table 1 0.
Comparison of the efficiency of using the baiting machine and manual baiting methods for t he control of
plateau zokors and plateau pikas (Jing et al. 1991).
Baiting method Bait
Zokor (active mounds)
Baiting mach ine 0 .075% diphacinone-Na
Baiting machine 0.6% gophacide
Manual baiting 0 .075% diphacinoneNa
calculated 10 days after baiting and based on
the numbers of ac tive burrow holes before
and a fter trea tment. The zokor killing ra te
was significantly higher th an tha t obtained
by tra dition al manual baiting on the ground
(Table 10) (Fan et al. 1986; Jin g et a l. 1987).
After half a year, the con h'ol effic iency of
0.075% diphacinone-Na grain baits was re-
assessed . Of 107 zokor burrow sys tems
checked, only 2 were occupied. The control
efficiency for zokors rema ined high (Jin g et
a l. 1991) and indica ted tha t ba iting delivered
by the machine was effective for long
p eriods in controlling zoko rs.
Management by sowing and fencing
Sowing and fencing were implemen ted afte r
chemical control. Major grass species sow n
were Elymus nutas, Elymus sibiricus and
Pllccillnelia ten Ll iflora. Two years later, th e
average heigh t of the grass had in creased
from 99 mm to 860 mm and the average
gro und coverage increased from 35% to 90%.
The grass vegetation formed two layers; an
upper layer of sown grasses an d a lower
layer of weeds. Grasses (Gramineae spp.)
became dominan t and th eir yield increased .
Killing rate (%)
Pika (active burrows)
85.1 96.9
77.3 99.8
69 .3
299
Ecologicallybased Rodent Management
In 1987, grasses, sedges and weeds made up
18.9%,19.3% and 61.8% of the total above-
ground biomass, respectively. In 1988,
grasses, sedges and weeds made up 35.9%,
17.7% and 46.4% of the total above-ground
biomass, respectively (Jing et al. 1991) (Table
11). The total biomass of the grass in the
demonstration area increased by 1.9 times at
the end of the first year and by 9.1 times at
the end of the second year. After three
consecutive years the population densi ties of
the plateau zokors and the plateau pikas
were still at low levels (Jing et al. 1991).
Weed control using herbicide
The degenerated grassland is dominated by
weed species and is the optimal habitat for
zokors. Therefore, weed control using
herbicide will help destroy the habitat of
zokors, and also has the added benefit
allowing greater growth of foraging grasses
by reducing competition from weeds. In
June 1987, 2,4-dichlorophenoxyacetic acid
was used to kill weeds as they began to
germinate. A month later, the yield of
grasses and sedges had increased and in
correlation with the concentration of
herbicide applied. At the optimal
concentration of the herbicide (0.75 kg/ha),
the yield of weeds decreased by 68.8/" while
the yield of the forage grasses increased by
47%. In autumn, the zokor density was
greatly decreased. At the highest herbicide
concentration (2.25 kg/ha), the density of
zokor mounds decreased from 323 to
mounds/ha (Jing et al. 1991) (Table 12).
Economic impact of IPM
By implementing IPM from 1987 to 1989,
rodent damage was reduced and the ground
300
vegetation and productivity of the grassland
increased greatly in the experimental area.
The weight of dry grasses was only 84.7
g/m2 in 1987 when the IPM experiment
commenced, but it increased to 406.52 gl m 2
by 1989. In the experimental area (200 ha),
the total yield of dry grasses increased from
80.6 t (which could feed 122 sheep) in 1987 to
728.9 t (which could feed 1,103 sheep) in
1989 (Jing et al. 1991).
The total investment in this IPM,
including machine baiting, fencing and
sowing was 40,170 RMB, and the income in
the three consecutive years of 1987,1988 and
1989 was 227,710 RMB. Therefore, the ratio
of benefit to cost of the IPM in this region
was 5.7 (Table 13) (Jing et al. 1991).
RECOMMENDATIONS AND FUTURE
RESEARCH PRIORITIES
In Qinghai Province, the total area of black
sandy soils caused by rodents is more than
13,000 km 2. If the IPM strategy
demonstrated in the experimental area could
be extended successfully to regions of the
Qinghai-Tibet alpine meadow ecosystem
with heavy rodent infestation, the
degenerated grassland could recover.
Indeed the grass production would be over
0.103 billion t and this could feed 0.156
billion sheep. However, there are several
important pointers for implementing this
IPM strategy:
.... For highly degenerated grassland with
heavy pika and zokor problems, chemical
control using baiting machines should be
considered first, with weed control and
sowing of grasses following immediately.
Grazing by livestock should not be
permitted in the first year, and limited
 Table 11.
Changes in above-ground biomass after sowing or fencing following chemical rodent control (Jing et al. 1991).

Treatments Above-ground dry blomass (g/m2)

Pre-treatment 1987 Post treatment: 1988

Grasses Sedges Weeds Total Grasses Sedges Weeds Total

Chemical control + semifenced 33.02 33.57 107 .54 174.13 97 .68 48 .17 126.02 271.9
Chemical control + sowing + semi-fenced 50.81 33 .98 107 .53 192.32 389.36 48.16 126.03 563.6
Untreated 11.82 32.61 74.54 118.97 29 .26 17.68 253.73 300.7

Table 12.
Grass yields after weed control with different concentrations of herbicide and the effect on numbers of new m ounds created by zokors (Jing et al.
1991).

Herbicide concentration Above-ground dry weight (kg/ha) Zokor density (mounds/O.25 ha) :::a
Q

-=
~
(kg/ha) ID
Grasses Sedges Weeds Pre-treatment (Apr-May) Post-treatment (Sep-Oct)
JL JL

-=
0 .0 448 793 I~ 1206 483 376 "'Cl
.c ID
III
0 .3 75 757 JC 995 JC 980 457 164
3:::
0.75 1640 972 377 377 46 III

1.5 1462 JC 503 JC 401 4 39 9 III

IrQ
ID
2 .25 1161 310 331 323 o 3

Table 13. -==

ID

Comparison of the costs and benefits of integrated pest management in the experimental area located at Panpo, Menyuan County, Qinghai
Province, from 1987 to 1989 (Jing et al. 1991). -e
:::r
ID

Cost (RMB 8 ) In 1987 Benefit (RMB) =

IrQ
:::r
III
Machine baiting Fencing Sowing 1987 1988 1989 T"
:!
Motor power Toxic bait Iron net Structure Seed Ploughing
-
Cl"
ID

w
Cl
1-6
1296 418 20000

a RMB = renminbi yuan . As of May 1999. 8.14 RMB = US$1.

3516 12000 2940 1209 117 157 109344

-"'Cl
iii
ID
III
C
Ecologically-based Rodent Management
grazing should be permitted only after the
grassland is fully recovered.
.. For lightly degenerated grassland with
lesser pika and zokor problems, grazing
should be stopped immediately and
followed weed control and sowing.
Grazing should not permitted again until
the grassland is fully recovered.
.. For abandoned, non-cultivated grassland,
chemical control using the baiting machine
followed by weed control and grass
sowing must occur to achieve recovery of
the grassland.
.. In the alpine meadow ecosystem, one
polecat can kill 1,554 plateau pikas or 471
plateau zokors each year (Zheng et aL 1983i
WeiandZhou 1997). Therefore, the natural
enemies of the pikas and zokors such as
weasels, polecats, foxes and eagles should
fully protected.
Because the plateau zokor was found
recently to have some medical value in
healing rheumatism (like tiger bone) (Zhang
and Zhang 1997), it is now believed to be of
economic importance in this region. Thus,
future studies will focus on finding a
solution to the conflict between this newly
recognised benefit versus the damage
caused by zokors through managing their
population at a level suitable for good
harvest output but causing little damage to
the grassland. For pika management,
biological control using parasites and! or
contraceptive vaccines that potentially have
minimal non-target environmental impacts
will be considered as research priorities in
the future.
302
ACKNOWLEDGMENTS
We thank Professor Zhibin Zhang of the
Institute of Zoology, the Chinese Academy
of Science for revising the manuscript. We
also wish to thank Dr Lyn Hinds of CSIRO
Wildlife and Ecology, Australia, for her great
help in improving the English in our
manuscript. This research is supported by
the key project of the Chinese Academy of
Science (No. KZ951-B1-106-2) and by the
National Rodent Key Project
(No. 96-016-01-06).
REFERENCES
Bian, J.H., Fan, N.C., Jing, Z.c. and Zhang, D.C.
1994. Foraging and behavioural adaptation of
rodents to their habitat. In: Cheng, Y.S. ed"
Proceedings of the 60 th Anniversary of the
Founding of China Zoologica Society. Beijing,
Science Press, 403-408 (in Chinese),
Fan, N.C. and S.Q. 1981. The structure of the
tunnel system of the Chinese zokor. Acta
Theriologica Sinica, 1,67-72 (in Chinese).
Fan,N.C.Jing,Z.c., Wang,Q.Y.andZhou, W.Y.
1986. Studies on the effect of the bromadi-
alone in killing the pika and the zokor. Acta
Theriologica Sinica, 6, 211-217 (in Chinese).
Fan,N.C.,Jing,Z.C.andZhou, W.Y.1990.On the
invading behaviour of the plateau zokor and
a new approach to its controL Acta Theriolog-
ica Sinica, 10, 114-120 (in Chinese).
Fan,N.C., Wang,Q.Y.,Zhou, W.Y.andJing,Z.C.
1988. Relationship between the density of
plateau zokors and the severity of their
damage to vegetation. In: Xia, W.P., ed"
Proceedings of the International Symposium
of Alpine Meadow Ecosystem. Beijing,
Science Press, 109-116 (in Chinese).
Fan,N.C. and Zhang, D.C.1996.Foragingbehav-
iour of Ochotona curzoniae and Ochotona
daurica and their adaptation to their habitats.
Acta Theriologica Sinica, 16,48-53 (in
Chinese).
 Rodent Pest Management in the Qinghai-Tibet Plateau
HeadIey, J.c. 1972. Defining the economic
threshold. In: Metcalf, R.L, ed., Pest control
strategies for the future. Washington, D.C.,
National Academy of Sciences, 100-108.
Jing, Z.c., Fan, N.C., Zhou, W.Y. and Bian, J.H.
1991. Integra ted management of rodent pests
in the grassland at Panpo area. Chinese
Journal of Applied Ecology, 2, 32-38 (in
Chinese).
Jing, Liu, L.G., Zhou, W.Y. and Fan, N.C.
1987. A screening experiment on new bai ts
and their potential for rodent control. Chinese
Journal of Rodent Control, 3, 165-168 (in
Chinese).
Liang, J.R. 1982. On the restoration of population
density of the plateau pika and the common
Chinese zokor after control. In: Xia, W.P., ed.,
Alpine meadow ecosystem. Gansu, People
Publishing House, 93-100 (in Chinese).
Liang, J.R. and Xiao, Y.F. 1978. Relationship
between the numbers of Chinese zokors and
plateau pikas, and their effects on
vegetation. In: Xia W.P., ed., Contributions to
rodent control and rodent biology. Beijing,
Science Press, 3, 118-124 (in Chinese).
Liu, J.K., Zhang, YZ. and Xin, G.W. 1980.
Relationship between the numbers of plateau
pika and the level of their damage. Acta
Zoologica Sinica, 26, 278-285 (in Chinese).
Ma, S.Z. 1991. A preliminary study on the
dynamics of the natural grassland and suita-
ble methods of exploitation. Journal of
Qinghai Environment, 28, 26-29 (in Chinese).
Pi, N.L. 1973. Study on the feeding habits of the
plateau pika. In: Xi a, W.P., ed., Contributions
to rodent control and rodent biology. Beijing,
Science Press, 1,91-102 (in Chinese).
Shen,S.Y.1987. A preliminary study on the effect
of the botulin model C on killing pikas. Acta
Theriologica Sinica, 7, 147-153 (in Chinese).
Shi, Y.Z.1983. On the influences of grassland
vegetation on the density of plateau pika.
Acta Theriologica Sinica, 3, 181-187 (in
Chinese).
Shi, Y.Z, Fan, N.C, Wang, X.G. and He, X.Q.1978.
Study on the population age and l"~JlV'LIWL
tion of plateau pika. In: Xia, W.P., ed., Contri-
butions to rodent control and rodent biology.
Beijing, Science Press, 3,104-118 (in Chinese).
Tao, Y.D, Fan, N.C. and Jing, Z.C.1990. Assessing
the damage of Myaspalax baileyi to grassland
and the economic threshold. Grassland of
China, 2, 103-106 (in Chinese).
Wang, Q.Y and Fan,N.C.1987. Studies on the
digging activities and exploration of the
plateau zokor. Acta Theriologica Sinica, 7,
283-290 (in Chinese).
Wang, Q.Y, Jing, Z.c. and Fan, N.C. 1996. The
dynamics and the integrated management of
pest rodents in the alpine meadow. In: Wang,
Z.W. and Zhang, Z.6., ed., Theory and
practice of rodent pest management. Beijing,
Science Press, 206-228 (in Chinese).
Wang, Q.Y, Zhou, W.Y., Zhang, Y.M. and Fan,
N.C. 1994. Observations on the burrowing
behaviour of the plateau zokor (Myospalax
baileyi). Acta Theriologica Sinica, 14,203-208.
Wang, X.G. and Dai, KH.1989. Natural life span
of the plateau pika (Ochatona curzoniae). Acta
Theriologica Sinca, 9, 56-62 (in Chinese).
Wang,X.G. and Dai,KH.1990. A study on the
breeding area and the territorial behaviour in
plateau pika (Ochatana curzaniae). Acta Theri-
ologica Sinca, 10,203-209 (in Chinese).
Wang, X.G. and Smith, A.T. 1988. The mortality
rate of the plateau in winter. Acta Theri-
ologica Sinca, 8,152-156 (in Chinese).
Wei, W.H. and Zhou, W.Y 1997. A natural
enemy of rodents and its protection in the
Haibei alpine meadow ecosystem. Journal of
Qinghai Environment, 7, 145-149 (in
Chinese).
Wei, W.H., Zhou, W.Y., Wang, Q.Y and Fan,
N.C. 1996. Comparison of the behaviour of
the plateau zokor between the reproductive
and non-reproductive periods. Acta Therio-
logica Sinica, 16, 194-201 (in Chinese).
Xiao, Y.E, Liang, J.R and Sha, Q. 1981. The distri-
bution of the plateau pika and its effect on
grass Kobresia pygmaea in the Tianjun and
Kangyang regions. In: Xi a, W.P., ed., Contri-
butions to rodent control and rodent biology.
Beijing, Science Press, 4, 114-124 (in Chinese).
Zhang, D.e, Zhou, W.Y., Wei, W.H. and Wang,
Q.Y. 1995. Study on the reproduction and
reproductive behaviour of the plateau zokor.
303
Ecologically-based Rodent Management
In: Cheng, Y.s., ed., Studies on mammal
biology in China. Beijing, China Forestry
Publishing House, 174-179 (in Chinese).
Zhang, Y.M., Zhou, W.Y., Fan, N.C. and Zhang,
D.e. 1991. Population dynamics and
tion of the plateau zokor. In: Liu, J.K. and
Wang, Z.W., ed., Alpine meadow ecosystem.
Beijing, Science Press, 175-179 (in Chinese).
Zhang, X.F. and Zhang, B.5. 1997. Studies on the
chemical constituents of bone oil in Myaspa/ax
baileyi. Acta Theriologica Sinica, 17, 155-157
(in Chinese).
Zheng, S.W. 1980. Studies on the reproduction in
the mole rat (Myaspa/ax fontanieri). Acta
Zoologica Sinica, 1,465-477 (in Chinese).
304
Zheng, S.W., Zeng, J.X. and Cui, RX. 1983. On the
ecology and energy dynamics of the masked
polecat (Mustela eversmanni) in Haibei,
QinghaiProvince. Acta Therio]ogica Sinica,3,
35-46 (in Chinese).
Zhou, W.Y and Dou, F.M. 1990. Studies on the
activity and home range of the plateau zokor.
Acta TheriologicaSinica, 10,31-39 (in
Chinese).
Zhou, W.Y. and Wei, W.H. 1994. Study on the
population dynamic of polecats and its effec-
tive factors. Acta Biologica Plateau Sinica, 12,
161-171 (in Chinese).
 14. Ecologically-Based Population
Management of the Rice-Field Rat in
Indonesia

Luke K.P. Leung, Grant R. Singleton, Sudarmaji and Rahmini

Abstract

The rice-field rat, Rattus argentiventer, occurs throughout most of Southeast Asia and
is one of the most economically important pre-harvest pests in rice crops. Based on a
capture-recapture study in an irrigated lowland rice agro-ecosystem, populations are
limited by the availability of nest sites and food . We recommend the following
management strategies : (1) minimise the number of banks to reduce the availability of
nest sites ; (2) maintain/retain fallow to limit populations; (3) synchronise crops to
minimise breeding period; and (4) time the application of mortality control at the early
to mid-tillering stage when population density is low, individuals are generally in poor
condition and not breeding, and thus populations are least able to compensate for the
imposed mortality. Active burrow counts are recommended for assessing population
size for management purposes. Live-trapping is recommended for demographic
studies . Decision analysis identified strengths in current management practices in
West Java , as well as key gaps in our scientific knowledge for developing effective
management. Future research priorities are as follows: (1) evaluate the impact of
secondary food sources on rat populations ; (2) identify cues used by rats to trigger
mating; (3) examine the spatial dynamics of crop asynchrony and its effect on the
movements of rats; (4) determine age-specific survival for targeting control; (5)
develop standardised methods for assessment of yield loss caused by rats; and (6)
develop appropriate decision models for use by growers.

Keywords

Rattus argentiventer, rice-field rat, population dynamics, pest management, rice

305
Ecologically-based Rodent Management
INTRODUCTION
HE RICE-FIELD rat, Rattus
T argentiventer, is one of the most
economically important pre-
harvest pests in rice agro-ecosystems in
Southeast Asia (Buckle et al. 1985; Geddes
1992; Singleton and Petch 1994). It occurs
throughout most of Southeast Asia,
including parts of Burma, Cambodia,
Vietnam, Thailand, Malaysia, Indonesia and
the Philippines (Corbet and Hill 1992). It is
the predominant rodent in most rice
agro-ecosystems. Other rodent (e.g.
Rattus tanezumi) become dominant only on
islands where the rice-field rat is absent
(Wood 1994).
The island of Java has long been
Indonesia's rice bowL In 1996, it produced 28
million tonnes, accounting for about 56% of
national production. Over half of the rice
grown in Java is in irrigated lowland fields
(about 3 million ha), a quarter in rainfed
lowland fields and the rest in dry, upland
fields. The lowland systems produced 96%
of rice production in Java in 1996 (Bureau of
Statistics, Indonesia).
Rice agro-ecosystems have expanded to
increase production. Raffles (1817) estimated
that only one-eighth (12%) of Java was
cultivated. This had increased to about 18%
by 1870, to 50% by 1920 (Booth 1988), and to
64% by 1997. Further expansion is limited by
the availability of suitable land (RePPProT
1989). Recent increases in rice production
have been achieved mainly through the use
of varieties with short growing seasons,
increased cropping intensity, and improved
water supply. Further increases in rice
production may be achieved through
306
reducing pn~-harvest losses to rodents
(Singleton and Petch 1994). This underpins
the purpose of reviewing ecologically-based
management of the rice-field rat in this
chapter.
Tropical rice agro-ecosystems are one of
the most complex and stable agro-
ecosystems in the world. Despite its
economic significance, the ecology of rice
agro-ecosystems is under-studied (Anwar et
a1. 1984). We conducted a bibliographic
search (August 1998) on this topic and found
that studies have focused primarily on rice
and insect pests.
Rice is the staple food in Asia. By 2025
Asia's production of rice will have to
increase by 70% to meet the needs of four
billion people (Lampe 1993). Although many
methods are used to control rodent pests in
Indonesia, pre-harvest damage levels have
not been reduced in recent times below).
Tolerance of rodent damage to rice crops
will diminish under the intense pressure to
increase production. This chapter aims to
provide an ecologically-based approach to
integrating rodent control methods. We
begin by reviewing the economic
significance of the rice-field rat and
describing the rice agro-ecosystem of West
Java. We then review what is known about
the likely key factors that limit the
population size of the rice-field rat in rice
agro-ecosystems, drawing heavily on the
findings of a study by LX-P. Leung and
Sudarmaji (unpublished data) on the
population ecology and habitat use of the
rice-field rat in rice fields in West Java. In the
final section, we discuss decision-support
models for the management of the rice-field
rat.
 Management of the Rice-field Rat in Indonesia
ECONOMIC SIGNIFICANCE
In Indonesia, the rice-field rat has been
ranked as the most important pest (non-
weed) in Indonesia since 1983 (Table 1). Data
collected by the forecasting centres of the
Directorate of Food Crop Protection in
Indonesia indicate that rats cause pre-
harvest losses of around 17% per year to rice
crops in Indonesia (Geddes 1992).
Rodent damage to rice crops is a chronic
problem in Indonesia, although some years
experience higher damage than others
(Table 2). In West Java, rodent damage
varies greatly among villages and sub-
districts, and even among farms within a
village-possibly because of the small size of
average holdings 2 ha) and different levels
of rodent control effort among growers.
Rodent control also lacks coordination. The
patchy distribution of rodent damage has a
significant impact on individual farmers and
villages (Singleton and Petch 1994). On
occasions, parts of some provinces suffer
total crop loss to rats (Table 3).
IRRIGATED LOWLAND RICE
AGRo-ECOSYSTEM
This chapter draws heavily on research
conducted in lowland irrigated rice fields of
West Java. The main rice producing area in
Java is located on the north coast of West
Java, and consists of the neighbouring
districts of Karawang, Subang and
Indramayu. The irrigated area in these
districts is 271,000 ha (Kartaatrnadja
et a1. 1997).
Climate
Each year, the Southeast Asian region has
distinct dry and wet seasons. In West Java,
the dry season is from May to October and
the wet from November to April. Around
75% of the annual rainfall occurs in the wet
season. In general, 1\-vo crops of rice are
grown each year in the irrigated lowland rice
agro-ecosystem: a wet season crop (first
crop) from mid-October to February; and a
dry season crop (second crop) from April to
July I August. The field is fallow over the
later part of the dry season when irrigation
water is in short supply. Vegetables and a
third crop of rice may be grown in some
fields where water is still available. A brief
fallow may occur between the wet and dry
season crops because of a limited supply of
labour for land-preparation and
transplanting. The timing of crops varies
among years because the planting of the first
crop is dependent upon the onset of the wet
season.
Phenology of rice
The main growth stages of rice, based on a
120-day variety of rice (lR 64) are: tillering
(55 days long); reproductive (35 days); milky
ripe grain (10 days; abbreviated as milky);
and ripening (30 days) (after Reissig et a1.
1986). The reproductive and ripening stages
are collectively termed the generative stage.
There are often different crop stages in a
given area, because crops are planted over a
period of time depending on the availability
of irrigation water and labour for planting.
Planting of the first crop (wet season) is
usually more synchronised than the second
crop.
307
Ecologically-based Rodent Management

Table 1.
Ranking of economically important, non-weed pests in rice crops in Indonesia.

Pest Ranking of decreasing order of economic significance

1983-1985 1986-1990 1991-1994 1995-1997

Rice-field rat 1 1 1 1
Brown plant hopper 2 4 2 3
Rice stem-borer 4 2 3 2
Rice leaf folder 3 3 Not ranked Not ranked

Source: Forecasting Center for Pest and Diseases, Jatisari , West Java.

Table 2.
Incidence of rat damage and damage intensity of rice in Indonesia during the period
1980-1996.

Year Incidence of rodent damage

to rice crops

Area damaged (ha) Damage intensity (%)

1981 198546 14.5
1982 194144 17.6
1983 186989 20.1
1984 186036 15.9
1985 179765 17 .5
1986 119602 15.6
1987 80865 13.6
1988 100 1 71 15.7
1989 95175 17 .7
1990 76 140 17.2
1991 77 114 18.7
1992 85512 19 .5
1993 No data No data
1994 86694 21.9
1995 103109 20.8
1996 103109 20 .8
Average 119819 17.8
Source:
Bureau of Statistics , Govern ment of Indonesi a.

308
 Management of the Rice-field Rat in Indonesia

Table 3.
Rodent damage and crop loss to rats for lowland rice in 1995.

Province Damage area (ha) Mean damage Intensity Area of total crop loss
(%) (ha)

DKI Jakarta 35 10.37 0

Jawa Barat 29006 16.08 241
Jawa Tengah 11282 14.32 662
D.1. Yogyakarta 2138 11.28 0
Jawa Timur 4493 22.18 485
D.1. Aceh 5755 17 .90 0
Sumatera Utara 878 16.00 10
Sumatera Barat 1073 20.20 25
Riau 700 20.50 12
Jambi 597 24 .10 105
Sumatera Selatan 2380 22 .50 217
Bengkulu 949 12.70 0
Lampung 1473 18.60 70
Bali 212 25.40 0
NTB 550 15.90 0
NTI 45 5.30 0
Kalimantan Barat 1476 32.40 337
Kalimantan Tengah 2781 23.70 446
Kalimantan Selatan 140 20.20 0
Kalimantan Timur 1385 23.10 19
Sulawesi Utara 612 28.80 110
Sulawesi Tengah 9815 14.60 158
Sulawesi Selatan 23362 32.40 2179
Sulawesi Tenggara 1 9 72 24.80 149
Source: Bureau of Stati sti cs. Government of Indonesia.

Sympatric species korros, and Colu ber radiatus. Based on o ur

As well as d omes tic species, seven other
vertebrate sp ecies are commonl y fo und in
the lowland irrigated rice agro-ecosystem in
West Java: the house shrew ( SUI1C II S
1I111rinus ), the large band icoo t rat (Ba lldicota
indica), a skink-Mabuia IIIl1/ tijasciata, a frog
- Rana crytra/a, and two snakes - Ptyas
trapping d ata, the large bandicoot ra t occurs
a t rela tively low densities compared to the
rice-fie ld ra t.
Avian preda to rs are rare in the lowland
rice agro-ecosystem in Java . O ther p reda tors
are sighted occasion ally s uch as the Javan
mongoose (Herpestes javan icus), the fishjng

309
Ecologically-based Rodent Management
cat (Felis viverrina), the small Indian civet
(Viverricula malaccensis), and many
unidentified species of snakes also occur
(L.K.-P. Leung, personal observation).
The native house rat (Rattus rattus diardii)
is a commensal species and is occasionally
found in the rice fields. The Polynesian rat
(Rattus exulans) and the rice mouse (Mus
caroli) occur in other rice agro-ecosystems.
Food
Very little is known of the diet of the rice-
field rat. Stomach content analysis reveals
that they consume crabs, snails, insects, rice
and other plant material, which commonly
occur in the rice ago-ecosystem (Rahmini,
unpublished data). Rice at the ripening stage
is the most important food; fragments of rice
grains were found in 100% of stomachs
examined (n = 50) and constituted more than
50% of the stomach content by volume
(L.K-P. Leung, unpublished data).
Goot (1951) observed that rats in captivity
survived for only four or five days when fed
exclusively on weeds, rice plants at the
tillering stage, crabs, snails, or insects;
whereas rats survived for several months
when fed on starch food such as rice grains,
maize, soya beans, peanuts, and sweet
potatoes. Murakami (1990) found that
captive rats survived for more than two
weeks when fed exclusively on rice plants at
the ripening stage. In contrast, survival was
poor when they were fed rice plants at the
tillering and reproductive stages. The results
of these experiments need to be interpreted
cautiously because rats in the field have a
choice of food. Nevertheless, these two
studies indicate that rice at the ripening
stage is the most nutritious food for the rice-
field rat in a rice agro-ecosystem.
310
In mixed-crop systems in West Java,
relative damage to crops by the rice-field rat
indicates it prefers rice to maize, peanut,
and, soya bean (5. Suriapermana, personal
communication). For this group of crops,
litter size is highest in rats living in rice
(Goot 1951).
Habitat
The original habitat of the rice-field rat is
grassland, where its breeding is seasonal
(Harrison 1951, 1955). In West Java, the
irrigated lowland rice agro-ecosystem
consists of a continuous tract of rice fields
and a network of roads, streams, irrigation
channels and drainage lines. 'Islands' of
villages (kampongs) are scattered across the
landscape. Sugarcane, peanuts and other
crops are grown on 'islands' of elevated
ground. Populations of the rice-field rat are
present in villages and elevated crop lands
only when rice crops are not at the ripening
stage (Goot 1951).
Earth banks along margins of paddies are
an important habitat for the rice-field rat in
this ecosystem because when the paddies are
flooded, burrows in earth banks are the
primary source of shelter. Banks are
important also for breeding females because
young are born and reared almost
exclusively in burrows (L.K.-P. Leung and
Sudarmaji, unpublished data). Rat burrows
are found in the substrate of paddy fields
only after harvest when the fields are
drained and not waterlogged. After harvest,
rats also construct nests underneath piles of
rice straw left in the fields.
 Management of the Rice-field Rat in Indonesia
Breeding
Lam (1980, 1983) conducted detailed and
extensive studies of the reproduction of the
rice-field rat in Malaysia and has shown that
a single breeding season occurs where one
crop is grown per year, and that two
breeding seasons occur if two crops are
grown per year. He also has shown that the
breeding seasons correspond closely with
the reproductive and ripening stages of the
rice crop, and suggests that nutritional
factors, particularly the presence of rice at
the reproductive stage, trigger reproduction
in the rice-field rat. Tristiani et a1. (1998)
obtained similar findings in populations of
the rice-field rat in West Java.
L.K.-P. Leung and Sudarmaji
(unpublished data) have determined from
autopsy of females that mating begins just
prior to maximum tillering. At this stage, the
testes are at their largest in nearly all males,
indicating peak breeding condition. The first
litters of young are born during the booting
stage. Post-partum mating occurs (Lam
1983) and a second litter can be born during
the ripening stage and a third shortly after
harvest (L.K.-P. Leung and Sudarmaji,
unpublished data).
The onset of mating at the maximum
tiller number stage allows females to make
full use of the ripening crop for meeting the
high energy demands of raising young. The
ability of the rice-field rat to synchronise
reproduction with the phenology of the rice
crop is possibly the key to its success in the
rice agro-ecosystem. This ability was
possibly selected in its original grassland
habitat. Thus the rice-field rat may be
evolutionarily pre-adapted to invade rice
agro-ecosystem because of the similar
phenology of rice and other species of the
Graminaceae.
Female rice-field rats can conceive at the
age of 45 days (Lam 1983), and can
potentially produce up to three litters per
cropping season. The first litter of the season
generally does not breed because of the
rapid decline in quality of the food supply
within two weeks of harvest. However, the
availability of high quality food only needs
to be extended by 3-4 weeks for these
females to breed and successfully rear
young. Asynchronous crops will therefore
extend the breeding season of rats and allow
the first litters of the cropping season to
breed and successfully raise young.
Therefore the number of breeding females
would increase exponentially. This
contention is supported by studies of the
breeding status of different age classes of
female rats collected in an asynchronous
cropping area in West Java in 1996: from a
sample of 410 adult females caught
immediately post-harvest, 56% of those
pregnant were less than two months of age
(Rahmini, unpublished data).
Asynchronous cropping is typical of fast-
growing varieties of rice. Traditional
varieties in Southeast Asia are
photosensitive and mature at a particular
time of the year, even when crops are
planted over a period of weeks (Grist 1975).
Synchronous maturation was possibly
selected as a defensive mechanism against
rodents or other pests.
Monitoring pest populations
Accurate assessment of rodent population
density assists the grower to properly
schedule control efforts. Also, reliable
estimation of population density is a
311
Ecologically-based Rodent Management
prerequisite for research into factors
determining distribution and abundance of
the pest. Management decisions are only as
accurate as the sampling methods
employed.
The reliability of a relative abundance
index is based on its linear relationship with
the population size (Caughley 1977). Two
relative indices (active burrow counts and
catch index) have been developed and
calibrated against directly enumerated
population sizes of the rice-field rat in
lowland rice agro-ecosystem (LK-P. Leung
and Sudarmaji, unpublished data). We are
aware of only two other studies that have
achieved this (Parer and Wood 1986; Moller
et a1. 1997).
LK-P. Leung and Sudarmaji
(unpublished data) recommend active
burrow counts for assessing population size
of the rice-field rat for management
purposes because this method is simple and
reliable, and is not affected by the growth
stage of rice crops. The catch index, although
less accurate than active burrow counts, is
recommended for population studies
because animals caught in live traps can be
measured to estimate other demographic
characteristics such as sex ratio, survival and
age structure.
LIKELY KEY FACTORS LIMITING
POPULATION SIZE
In WestJava, there appear to be three
principal factors limiting the population size
of the rice-field rat. However, there is a
dearth of long-term studies and an absence
of manipulative studies, so these
conclusions are preliminary.
312
Farm management practices
Agricultural ecosystems undergo major
fluctuations in biomass and disturbance. In
West Java, the major disturbances occur
during land preparation and at harvest. For
example, land preparation for each rice crop
involves concurrent, widespread cultivation
of fields and substantial reformation of
banks. These actions have major impacts on
the distribution and abundance of the rice-
field rat. Added to this is the rapid switching
of rice crops from sub-optimal to optimal rat
habitats associated with the growing and
ripening phases of the crop.
Farm management practices by growers
therefore have major impacts on rat
populations. They influence the temporal
and availability to rats of shelter,
nesting and the quality and quantity of
food. They also impose direct mortality on
rats through physical actions such as
hunting, trapping, fumigating etc.(see
Singleton et al., Chapter 8).
Availability of nest sites
In the irrigated lowland rice agro-ecosystem,
the availability of nest sites is a key limiting
factor for populations of the rice-field rat
(LK-P. Leung and Sudarmaji, unpublished
data). The abundance of rats is lower in the
middle than along the margin of crops,
where nest sites are available in adjacent
banks.
In West Java, earth banks of primary and
secondary irrigation channels form a major
network in lowland rice fields. These are
prime nesting sites for rats and as a
consequence large populations of rats are
associated with these structures (L.K-P.
Leung and Sudarmaji, unpublished data).
 Management of the Rice-field Rat in Indonesia

Ra t burrow s a re occasion ally fOlmd in Table 4.

blmds (sm all banks) in p addies. Farm ers, Total counts of rats present in irrigated lowland
rice fields in West Java after harvest of the dry
how ever, lI sually cons truct blmds to a
season crops, August-September, 1998.
dimension tha t is too small for ra ts to build a
burrow system . Site Area (ha) Number Number of
of rats rats per ha

Food supply Pabuaran 230 51000 222

Popula tion size a nd body condition d ecline
d uring the dry season fa llow and bo th a re a t
theil" m inimum sh ortly a fter fallo w, a rollnd
the ea rly to mid-tiUer i.ng stage. This is
p ossibly due to diminish ed availability of
food during the la ter p art of the dry season
fallow w hen bo th the abundan ce of
invertebra tes (Wolda 1978) and p lant
growth are reduced.
A fter fa llow, the rice-field ra t persists in
low numbers in r ice field s dming the
tillering s tage of the fi rs t crop. Goot (1951 )
observed tha t a sm all number of ra ts
rem ained in the field when the ir burrow s
w ere n o t disturbed by land p rep ara tion.
Popula tion grow th is closely associa ted with
the ripening of the crop when quality food is
abw1dant. Th e process of population growth
a t the local level con sists of breeding and
subsequent recruitment of juveniles into the
p opulation as w ell as immigra tion of rats
a ttrac ted to the ripening crop s (Lam e t a1.
1990; L.K.-P. Leung and Suda rmaji,
un publishe d d a ta). M aximum densities of
popula tion are reach ed sh ortl y a fter harves t.
Direct enumeration of ra ts fro m fumiga ting
burrow s and disturb in g s tra w p iles indica tes
tha t d ensities range from 120 to 240 ra ts per
h ec ta re (L.K .-P. Leung and Sud arm aji,
w1p ubLish ed d ata). Simila r d ensities h ave
been obtained by co unting all ra ts caught in
la rge a reas during an e radica tio n ca mpaign
(Table 4).
Patokbeust 130 17157 131
Binong 516 64844 125
Pagaden 911 30644 34
Source: S. Suriapermana, unpublished data.
ECOLOGICALLY-BASED POPULATION
MANAGEMENT PRACTICES FOR THE
RICE-FIELD RAT
Popula tion ecologists o ften rem ark tha t their
studies and theories 'prov ide in sight' into
p est m anagem ent. To be useful, ho w ever,
these id eas m us t be p a rt of a d ecision-theory
fram ework. The challenge is to combine the
current scientific kno wledge on the biology
and m an agem ent of the rod e nt sp ecies we
wi sh to control with the social, econ omic and
p olitica l factors tha t influen ce the ad optio n
of managem ent actions by fa rmers. Th e
simple res triction s tha t cultural and
religious beliefs place on some managem ent
action s for agricultural pests (see N orton
and Heon g 1988) emphasises the impo rtance
of d e termining ho w farm ers are likely to
pe rce ive and react to a rod ent pest problem
and to recommended m an agem ent actions.
Fram ew orks for d eveloping a 'decision
analysis/system s ana lysis' approach to
vertebrate p est managem ent h ave been
d eveloped (N orton and Pech 1988; Braysher
1993) an d h ave been applied to rod ent p est
p roblem s in agricultma l systems (Brow n e t
al. 1998).

313
Ecologically-based Rodent Management
We have applied our knowledge of the
ecology of the rice-field rat, reviewed in this
chapter, to develop an ecologically-based
appraisal of the appropriateness of different
management actions (Table 5). This
appraisal was developed through
consultation with scientists and agricultural
extension staff. In some cases the scientific
knowledge was too weak to critically
evaluate the likely efficacy or economics of a
particular management practice. Therefore,
the decision-analysis presented in Table 5
includes a number of 'best guesses' and
simply provides a working model. Among
the 15 practices examined, seven practices
are not supported by any scientific evidence
and will need to be critically tested by field
trials.
Based on this decision analysis approach,
we reco111mend an integration of four
management strategies for appraisal in
replicated field trials prior to their
implementation.
~ First, the number of banks in rice fields
should be kept to the minimum to limit the
availability of nest sites. The flooding of
rice fields serves to deter not only rats but
also weeds and other pests. Banks are the
only nesting sites in irrigated rice fields,
and cannot all be eliminated because they
are used as roads and for managing water
levels for irrigation. Better planning and
technology may reduce the number of
banks. On the experimental farm of the
International Rice Research Institute
(IRRI), the number of banks was
minimised by converting open drain lines
to underground pipes, and rodent damage
was reduced as a consequence (Mark Bell,
IRRI, personal communication). Also,
314
concreting the surface of banks will
prevent rats from building burrows.
However, both underground pipes and
concreting are costly and are rarely
appropriate for developing countries.
~ Second, planting of crops and harvesting
should be synchronised over a large area so
as to shorten the period that ripening rice is
available to rats and hence reduce their
breeding season. In West Java the main
constraints to synchronised planting are
water and labour supply. However, it
appearsthatwatersupplyschedulescanbe
modified to reduce asynchrony of planting
of crops (5. Suriapermana, personal
communication).
~ Third, fallow over the dry season should be
maintained to reduce rat population size.
This also reduces numbers of some insect
pests (Grist 1975). In some regions in
Southeast Asia there is pressure to boost
crop production by growing three crops
per year. This is already beingpracticed in
the Mekong Delta in Vietnam, and rodent
damage appears to have increased as a
consequence (Singleton and Petch 1994). In
1998, massive rodent control campaigns
were necessary for protecting a third rice
crop trialled in parts of Java and Bali (5.
Suriapermana, personal communication).
~ Fourth, if mortality control is used, it is best
applied at the early to mid-tillering stage
after the dry season fallow when
populations densities are low and animals
are weak. If rat densities are significantly
reduced over a large area at this time,
populations would have little immediate
capacity to compensate for their reduction
in numbers. This timing of action is
 Table 5.
Decision analysis of practices for managing populations of the rice-field rat in West Java, Indonesia. Eight parameters were considered. The table
also includes scientific basis and priority given to each practice. No analysis was conducted for empty matrix cells. The analysis took place during
a workshop at the International Rice Research Institute in 1998. (Timing: Ip = land preparation; sb = seed bed; tp = transplanting; b = booting;
m = milky; r = ripening; h = harvest; f = fallow. Suitability of practices: / = yes; X =no; ? = unknown; N/A = not applicable.)

Management 1 ~ ,. .le
~>- :I
Cl)
GI E III
practice .2
~ :IUi E ~i GI- III 'S ;
!>-u! .Cl
u
>-
t0
E III
0
I:
.!la
g GI El!
I: GI
I:
ii .!i.
III 0 iI:
1= ,If' 0
e:e t::::J ~.e "C
u Cl) 8 u"
III
Cl) U GI 11 D.
LII III ">
I:
Cl) LII
;g
LII 11
" !I
'L...
Routine
Field sanitation Ip to b / ~ / / If / I~ / village /
Synchronous seeding and planting sb,tp / 11
/ / 11
/ / village /
11
TBSa by farmer group crop
,"""-
/ 11 / ? 11 / 11 / village /
Reduce bund size within rice fields Ip / / / / / village / IUf
11 11 11 l flllSfl

hig~ ==
Dj
Encourage natural enemies of rats all ? 11 ? / 11 / Il ? district / ? [ -, =
Dj

Plastic barrier for nurseries sb / 11

/ / 11
? I[ / village N/ A for
r
high
(IQ
ID
:3
Ft7"m igation
Digging burrows
Hunting at night
IP,b,m,r l1
Ip,sb,r
Ip,sb J---=!
/
/
JL..! Il
11
IL.!
?
Jr
]I
/
/
/
11
11
11
X
X
/
Il
If
11
/
X
X
village
village
village I
?
X
"
J[
against
?

? JL...
moderate
moderate

10~
---=
ID

Cl

::::r"
- --- -'rI,
Trapping with net >h / 11
X 11 / 11 / ![ ? village
L
? lr low ] ID
::Ill!
n
Apply If high densities are forecast
Do not plant rice as the third crop
Remove rice straw after harvest
-JI
f
11
/
/
11
/
/
11 /
/
11
I~
/
X
/
X
farmer
village
N/ A
/
for
for
[
[
high

higt i J
J
]
~

;:
-
iD'

Monitor abundance of rats in field

Apply rodenticide
Ip to m
f, tp
/
/
11
Il
I[
/
?
11
Il
If
/
/
I[
11
/
X 11
/
X
village
village
?
? 1I
for
?
?b

moderate
-,

-=
::Ill!
Dj

Pump water down burrow entrance >dry f / I[ ? 11

/ 11
/ I,
I.
? village X .1 ? moderate =
Cl.

I
Cl

....Utw BTrap-barrier system =

ID
en
bNeed economic threshold data. Dj'
Ecologically-based Rodent Management
consistent with the recommended use of
chemical rodenticides. After the rice crop
reaches the booting stage, rodenticides
become less efficacious because the baits
are less attractive to rats (Buckle et al. 1979).
How OUR ECOLOGICAL KNOWLEDGE
CAN INFLUENCE CURRENT
MANAGEMENT PRACTICES
The previous section highlighted manage-
ment strategies that need to be implemented
or maintained (e.g. fallow) for successful rat
management. Our ecological knowledge of
rodent populations can also be used to
modify existing actions so that they are more
efficient and effective.
In West Java, much energy and effort is
applied each year in digging and fumigating
rat burrows. This is the most common form
of pre-harvest control of rodents in rice
fields because it requires little capital
compared to rodenticides. However, these
actions generally are conducted during the
ripening stage of the crop-about five weeks
too late for maximal effect. In many parts of
Java, 'gropyokan' is a tradition, with people
joining together just after the harvest to kill
rats by digging and fumigating burrows in
rice fields. Mortality control at this stage is
not an efficient use of resources and labour
because many of the rats they kill would
have died anyway during the fallow period.
Indeed, removing high numbers of rats early
in the fallow period may result in better
survival of remaining rats than if no control
was implemented because of reduced
competition for food and shelter. If more rats
were to survive through until the
commencement of the next breeding season,
316
then post-harvest control activities may be
counter-productive.
In West Java, the one exception to
applying mortality control at early to mid-
tillering would be when the fallow period is
short and rat densities are high. In this
situation there is a high risk of significant rat
damage when the next rice crop is
transplanted.
WHERE To FROM HERE?
An important output from a decision
analysis process is clear identification of the
key gaps in our scientific knowledge for
developing effective management of a
particular rodent pest. Table 6 summarises
the gaps in our knowledge of the ecology
and biology of the rice-field rat and the
priority for obtaining this information to
strengthen our management of the species.
Our major shortcomings include a lack of
understanding of factors that influence age-
specific survival and inter-year fluctuations
in the amplitude of populations, and
knowing little of where rats live and what
they survive upon during the fallow periods.
Our understanding of the ecology of the
rice-field rat in West Java has enabled
identification of optimal timing, location and
scale of actions and whether they are
consistent with goals of sustainable
agriculture, minimal environmental impact
and humaneness. Through know ledge of the
socioeconomic status of the farming
communities in the region we have assessed
also the likely impact and practicalities of the
recommended management actions. An
important next step is to closely liaise with
growers to determine which management
actions they are willing to adopt, what
 Management of the Rice-field Rat in Indonesia
modifications they would require before an
action would be adopted, and to ascertain
whether there are other actions they would
like to include in an integrated management
program. Once the management practices
have been modified and approved by the
growers we will need to assess the impact of
integrating these management actions on
populations of the rice-field rat. This would
require a village-level study involving close
cooperation with growers and a replicated,
controlled, experimental design. The
assessment of the success of the study will
not be measured by the number of rats
caught or killed, but by the reduction in
damage caused by the rice-field rat,
compared to the cost of implementing the
management practices. For this approach to
be successful and to be sustained by farmers,
the management actions need to be
reviewed at least annually in consultation
with growers.
In summary, an ecological approach has
provided the tools and building blocks for
developing an integrated management
approach. Developing field projects to
evaluate the approach in close liaison with
growers at the village or district level
provides the necessary furnishings and
quality control for developing an effective
and operational management approach.
ACKNOWLEDGMENT
The population studies in Indonesia were
part of a multi-country study on the
management of rodent pests in Southeast
Asia, funded by the Australian Centre for
International Agricultural Research (Project
ASlj9420).
REFERENCES
Anwar, J" Damanik, S.J" Hisyam, N. and
Whitten, AJ. 1984. Ekologi ekasistem Sumat-
era. Yogyakarta, Gadjah Mada University
Press.
Booth, A 1988. Agricultural development in
Indonesia. Sydney, AlIen and Unwin.
Braysher, M. 1993. Managing vertebrate pests:
principles and strategies. Canberra, Austral-
ian Government Publishing Service, 58p.
Brown, P.R., Singleton, G.R., Dunn, S.c. and
Jones, D.A. 1998. The management of house
mice in agricultural landscapes using farm
management an Australian
perspective. In: Baker, KO. and Crabb, AC.,
ed., Proceedings of the 18th Vertebrate Pest
Conference Santa Clara, California, USA, 2-5
March 1998. Davis, University of California,
156-159.
Buckle,AP., Rowe, F.P. and Yong, Y.C.1979.
Field trials of two warfarin baiting methods
against the rice field rat (Rattus argentiventer)
in Peninsular Malaysia. Malaysian Agricul-
turalJournal, 52, 78-95.
Buckle,A.P., Yong, Y.C. and Rahman, A 1985.
Damage by rats to rice in South-east Asia
with special reference to an integrated
management scheme proposed for Peninsu-
lar Malaysia. Acta Zoologica Fennica, 173,
139-144.
Caughley, G. 1977. Analysis of vertebrate
populations. London, John Wiley and Sons.
Corbet, G.B. and Hill, I.E. 1992. The mammals of
the Indomalayan region. Oxford, Oxford
University Press, 488p.
Geddes, AM.W. 1992. The relative importance
of pre-harvest crop pests in Indonesia. Kent,
UK, Natural Resources Institute, 70p.
Goot, P. van Der 1951. Over levenswijze en
bestrijding van sawah ratten in het laagland
van Java. Landbouw, 23,123-275.
Grist, D.H. 1975. Rice. Tropical Agriculture
Series, Fifth EditionNew York, Longman,
548p.
317
Ecologically-based Rodent Management
Harrison, J.L. 1951. Reproduction in rats of the
sub genus Rattus. Zoological Society of
London Proceedings, 121,673-694.
Harrison, J.L. 1955. Data on the reproduction of
some Malayan mammals. Zoological Society
of London Proceedings, 125, 445-460.
Kartaatmadja, S., Soejitno, J. and Wardana, I.P.
1997. Pest management practices of rice
farmers in West Java, Indonesia. In: Heong,
KL. and Escalada, M.M., ed., Pest manage-
ment of rice farmers in Asia. Manila, Philip-
pines, International Rice Research Institute,
87-98.
Lam, Y.M. 1980. Reproductive behavior of the
rice field rat, Rattus argentiventer and impli-
cations for its control. Proceedings National
Rice Conference, 243-257.
Lam, Y.M. 1983. Reproduction in the rice field
rat, Rattus argentiventer. Malaysian Nature
Journal, 36, 249-282.
Lam, Y.M., Supaad, M.A., Chang, P.M.,
Mohamed, M.s., Goh, e.S. and Radzi, H.
1990. An innovative approach for protecting
rice against severe rat depredation. Proceed-
ings of the 3rd International Conference on
Plant Protection in the Tropics, 1-23.
Lampe, K 1993. Will Asia starve? In: Food
comes first for Asia. Parkville, Australia,
Crawford Fund for International Agricul-
tural Research, 59-71.
Moller, H., Clapperton, B.K and Fletcher, D.J.
1997. Density of rabbits (Oryctolagus cunicu-
lus L.) in the Mackenzie Basin, South Island,
New Zealand. New Zealand Journal of
Ecology, 21, 161-167.
Murakami, 0., Priyono, J. and Triastiani, H.
1990. Population management of the
ricefield rat in Indonesia. In: Quick, G.R, ed.,
Rodents and rice. Manila, Philippines, Inter-
national Rice Research Institute, 49-60.
Norton, G.A. and Heong, Kt. 1988. An
approach to improving pest management:
rice in Malaysia. Crop Protection, 7, 84-90.
318
Norton, G.A. and Pech. RP. 1988. Vertebrate
pest management. In: Australia: decision
analysis I systems analysis approach. CSIRO
(Commonwealth Scientific and Industrial
Research Organisation) Division of Wildlife
and Ecology, Project Report No. 5.
Melbourne, CSIRO Publications, 67p.
Parer, 1. and Wood, D.H.1986. Further observa-
tions of the use of warren entrances as an
index of the number of rabbits, Oryctologus
cuniculus. Australian Wildlife Research, 13,
331-332.
Raffles, T.S. 1817. The history of Java. Oxford,
Oxford University Press.
Reissig, W.H., Heinrichs, E.A., Litsinger, J.A.,
Moody, K, Fiedler, L., Mew, T.W. and
Barrion, A.T. 1986. Illustrated guide to
integrated pest management in rice in tropi-
cal Asia. Los Banos, Philippines, Interna-
tional Rice Research Institute, 411p.
RePPProT 1989. Regional physical planning
programme for Indonesia, review of phase 1
results: Java and Bali. Jakarta, Directorate
General of Settlement Preparation, Ministry
of Transmigration (Jakarta) and Natural
Resources Institute Overseas Development
Adminstration (London).
Singleton, G.R and Petch, D.A. 1994. A review
of the biology and management of rodent
pests in Southeast Asia. Canberra, Austral-
ian Centre for Interna tional Agricultural
Research, Technical Report No. 30, 65p.
Tristiani, H. Priyono,}. and Murakami, 0. 1998.
Seasonal changes in the population density
and reproduction of the ricefield rat, Rattus
argentiventer (Rodentia: Muridae), in West
Java. Mammalia, 62, 227-239.
Wolda, H. 1978. Fluctuation in abundance of
tropical insects. American Naturalist, 112,
1017-1045.
Wood, B.}. 1994. Rodents in agriculture and
forestry. In: Buckle, A.P. and Smith, R.H., ed.,
Rodent pests and their control. Wallingford,
UK, CAB International, 45-83.
 15. Population Ecology and Management of
Rodent Pests in the Mekong River
Delta, Vietnam

Peter R. Brown , Nguyen Quy Hung, Nguyen Manh Hung and

Monica van Wensveen

Abstract

Rodent pests are a growing problem in rice agro-ecosystems of Vietnam. However,

little is known about which rodent species are responsible for losses to crop
production, let alone how best to man age their impact. A survey of rat species in nine
provinces in the Mekong River Delta found that the dominant rodent species found in
rice ecosystems were t he ri ce-field rat, Rattus argentiventer (60%) and the lesser
rice field rat, Rattus losea (15%). Ten other species accounted for the remaining
2 5% of the population, and were unlikely to cause significant damage to pre-harvest
rice. The breeding patterns of th e two main rodent species and the relative
popul ation dynamics of rod ents in different habitats were obtained from live-trapping
studies (ca ptu re-mark-rel ease) in a range of representative habitats based in and
around th e rice growing regions of Long An, Ki en Giang and Tra Vinh provinces. Traps
we re set in ri ce crops (one and two rice crops per year) , ch an nel banks, melaleuca
forest, undistu rbed grassland , and coconut and banana plantations . Supplementary
kill trapping was cond ucte d to determine the breeding status (percentage of adult
fem ales breeding, litter size and embryo development) of the rats and to confirm
their taxonomy.
Our focus on the ecology of the key rodent pest species has helped to define a
range of potential management practices that are considered to be environmentally
sustai nable, economi cally feasible and socially acceptable. These practices are
divided into routin e actions that can be conducted all the time and preventative
actions if high rat numbers are forecast .

Keywords

Species composition, habitat use, breeding, rice crops , ecology, Rattus

argentiventer, Rattus losea

319
Ecologically-based Rodent Management
INTRODUCTION
R ATS ARE THE number one pre-
harvest pest of rice in many
cOlmtries in Southeast Asia
(Geddes 1992; Singleton and Petch 1994). In
Vietnam, rodents are one of the three most
important problems faced by the
agricultural sector (Huynh 1987) and the
level and intensity of damage has increased
since about 1992 (Table 1). In the Mekong
River Delta, there were 10,125 ha of
damaged rice recorded in Long An, Dong
Thap and Kien Giang provinces in the
1991/92 winter-spring season. In the
winter-spring season of 1992/93, the area
damaged increased to 44,000 ha over ten
provinces. Crop losses were estimated at
300,000-400,000 tonnes of rough rice.
Damage was recorded in Long An Province
to over 10,000 ha with 10- 30% losses, and
4,000 ha with 50-100% losses; ill Ha Tien
(Kien Giang Province), 800 ha were
damaged with 80% losses. In 1996, the area
damaged by rats increased to 130,000 ha
over most provinces of the Mekong Delta.
Rats also cause damage to other crops such
as corn and potato in the suburbs of Ho Chi
Minh City,in coastal regions (Binh Thuill1
Table 1.
Area damaged by rats (ha) in the Mekong River Delta and other parts of Vietnam, 1992-1997
(adapted from Hung et al. 1998).
Year Mekong River Delta
1992 18640
1993 107481
1994 134616
1995 74408
1996 130777
1997 129512
320
Province) and in highland regions (Dong
Nai Province). The factors that have lead to
increased losses include more intensive
farming and a general increase from two to
three crops plill1ted per year (Singleton and
Petch 1994).
The factors that lead to increases in rat
numbers and the importance of various
habitats for breeding and shelter have not
been addressed in Vieb1am. The principal
pest species is thought to be Rattus
argentiventer, however little is known about
the taxonomy or the population ecology of
the species of rodents which inhabit rice
ecosystems in Vietnam. In contrast, rodent
species that are hosts for humill1 plague i.n
Vietnam have been studied extensively
(Gratz 1988; SW1tsov et al. 1997). Population
studies of R. argentiventer have been
conducted in Malaysia (Wood 1971; Lam
1980,1983; Buckle 1990), Indonesia
(Murakami et al. 1990; Leung et aI., Chapter
14) and the Philippines (FaH 1977), but these
results may not be appropriate for the
Mekong River Delta which experiences
annual floods. Furthermore, the mosaic
pattern of habitats that exist in the Mekong
Delta may be favourable or tmfavourable for
rats.
Other areas Total area damaged In Vietnam
18640
107481
134616
18849 93257
130723 261500
245488 375000
 Management of Rodent Pests in the Mekong River Delta, Vietnam
This chapter aims to provide an
ecologically-based approach to the
management of rodent pests in the Mekong
River Delta of Vietnam. With a good
understanding of the species composition,
biology and behaviour of pest species it
should be possible to devise management
actions that are sustainable
(environmentally and culturally) and could
be combined with integrated pest
management programs that are in place for
insects, weeds and plant diseases (Singleton
1997). To provide some initial insight into
the ecology of rodent pests in rice agro-
ecosystems in the Mekong River Delta, data
were collected from 1994 to 1998 on
(i) the composition of rat species and (ii) the
population dynamics, habitat use and
breeding of the main rodent species, the
CAMBODIA
Minh Hai
Figure 1.
Provinces of the Mekong River Delta of southern Vietnam. Shown are locations where the composition of
rat species was determined ( A ) and where the population dynamics of rats were assessed ( . ).
rice-field rat (R . argentiventer) and the lesser
rice-field rat (Rattu s losea). Because none of
this work has been published, we will begin
by considering the methods adopted. We
will then present the results of this study and
suggest some preliminary management
recommendations. As the data are limited,
further studies are necessary to refine these
strategies. Also highlighted are areas where
critical information is lacking and further
research is required.
METHODS
Study sites
The provinces of Long An, Kien Giang and
Tra Vinh are situated in the Mekong River
Delta of Southern Vietnam (Figure 1).
South China Sea
N
t
0 km 100
321
Ecologically-based Rodent Management
The annual rainfall in the region is 2,000-
2,500 mm, which falls predominantly in the
wet season (April to November). The
topography is generally flat and some areas
are regularly flooded during October and
November, when the river systems
overflow. The average temperatures range
from 22-32C in the dry season and 25-30C
in the wet season. There is an extensive
network of channels and canals running
through the delta delivering water for
irrigation of rice crops. The width of the
channels ranges in size from 1 m (tertiary
channel), 2-5 m (secondary channel) and
>5 m (primary channel).
The main rice crops grown at each study
site were improved variety rice crops (90 day
duration such as IR-54404, OM-1490 and
OM-1037) and traditionaL local variety rice
(160-180 days duration). The first crop of
improved variety rice was sown when flood
waters subsided in December and was
harvested in March (dry season crop), then
the second improved variety crop was sown
soon after, in late March, and was harvested
in early July (wet season crop). The second
crop was planted in the same paddies as the
first. Once the second crop was harvested,
the ground was left fallow until December.
The traditional, local variety was planted in
mid-July and harvested any time from mid-
December to February (depending on
conditions). The crop stubble of the
traditional variety was left until the
following season. There can be some overlap
of the improved variety rice crops and the
traditional variety rice crops.
Rat species composition
The composition of rat species from nine
provinces was determined from six
322
sampling occasions from November 1995 to
July 1997. Fifty rats were collected at each
sampling occasion (except in March 1996
when 100 rats were caught) from rice fields
and from a 100 m length of a channel bank.
Rats were collected by live-capture wire
traps (200 x 100 x 100 mm) and from digging
burrows until the required number of rats
were obtained. It is not known whether this
sampling procedure may cause bias towards
some species. There have been no published
studies that consider this bias for rodents in
Southeast Asia. On subsequent visits to the
sites, rats were collected from the same
general area or within approximately 100 m
of the area used previously. Rats were
identified to species following van Peenen et
a1. (1969), Lekagul and McNeely (1977) and
Tien (1985a,b) using external features and
skeletal dimensions. Data are presented as
percentages.
Population dynamics
Table 2 describes the trapping schedules for
capture-mark-release studies and for
breeding studies. At Long An, the rats
collected were not identified to species, nor
were they assessed for breeding condition.
Live-trapping was conducted using
hand-made, single-capture traps (100 x 100 x
200 mm) baited with dried fish. The
abundance of rats was pooled for each
month, and was expressed as the number of
rats caught per 100 trap-nights (trap
success). On its first capture within a trap
session, each rat was identified to species
(based on external features, using van
Peenen et a1. 1969, Lekagul and McNeely
1977, and Tien 1985a,bi it was not possible to
identify some animals because of
taxonomical problems) and was marked
 Management of Rodent Pests in the Mekong River Delta, Vietnam
using a numbered brass ear tag (Hauptner,
Cermany). Each rat was sexed and assessed
for breeding condition, weighed ( 1 g), and
tail length (if intact), hind foot length, ear
length and head-body length were
measured ( 1 mm). Each rat was released at
the point of capture.
The minimum weight for an adult female
classification was based on the lowest
weight at which a rat was pregnant
(determined by palpation) or lactating. Any
rat lighter than this was considered juvenile
or sub-adult. Palpation generally detects
embryos from the second trimester, and so
willtmderestimate breeding performance.
Breeding
At Kien Ciang, kill samples were taken of
rats from various habitats from captures in
trap-barrier systems (see Singleton et al.
Table 2.
Summary of trapping conducted at Long An, Kien Giang and Tra Vinh provinces for capture-mark-release
and breeding studies.
Study site Trapping regime schedule
(No. traps per trap line)
Capture-Mark-Release
Long An 50 traps, 1 night per week
Kien Giang 35 traps, 2 nights per 2 weeks
Tra Vinh 35 traps, 3 nights per 4 weeks
Kien Giang 50 rats each month
1998 and Chapter 8 for description), from
live-capture traps, digging burrows and
catching by hand with nets. Females were
dissected to determine the condition of the
uterus, number of embryos, size of embryos
( 1 mm) and number of uterine scars. Rats
were considered pregnant if the uterus
contained visible embryos.
RESULTS AND DISCUSSION
Species composition
Twelve species of rodents were recorded
from nine provinces (Table 3). Overall, the
most common species was R. argentiventer
(61 %) followed by R. losea (15%) and Rattus
koratensis (7.2%). R. losea was the most
common species in Kien Ciang. The Mu s
genus was likely to include M. caroli and
M . musculus.
Duration Habitats and number of trap lines
Aug 94-Dec 96 Improved variety rice (1)
Grassland (1)
Cassava field (1)
Melaleuca forest (1)
Oct 97-May 98 Traditional variety rice (3)
Improved variety rice (3)
Melaleuca forest (3)
Grassland (3)
Secondary channel (3)
May 97-May 98 Improved variety rice (5)
Primary channel bank (1)
Banana plantation (1)
Coconut plantation (1)
Breeding
Aug 97 -May 98 Various
323
Ecologically-based Rodent Management
According to Sung (1999), there are 64
species of rodents belonging to 27 genera
and 7 families in Vietnam. The species
identified by Sung (1999) are generally
similar to those that were found in our
samples. In the agricultural fields of the
Mekong River Delta, Sung (1999) lists
R. argentiventer and M. caroli as common
species with Rattus flavipectus, Rattus exulans,
Rattus nitidus, Mus musculus, R. koratensis,
R. losea and Bandicota indica found primarily
around settlements. Other species of rodents
identified by Sung from the zoo graphical
zone of the Mekong River Delta were
Bandicota bengalensis, Rattus germailli and
Rattus norvegicus, but Rattus rattus was not
listed. We did not capture Mus cervicolor
which was a species identified by Sung
(1999) as being present in the Mekong River
Delta. Rodent species can be morpho-
metrically similar but genetically distinct
species (e.g. Mastomys spp., Granjon et al.
1997). Because of the diversity of species
present, it can be easy to misidentify
animals, particularly juveniles. Therefore,
more work needs to be done to understand
the taxonomy of these species in the Mekong
River Delta.
Population dynamics at Long An
The highest abundance of rats occurred in
September, when the local variety of rice
was in the vegetative growth stage (Figure
2). The lowest abundance of rats occurred in
June-when the second improved variety
crop was at the maximum tillering stage,
and in November and December-at the end
of the flooding period. The proportion of rats
caught in improved variety rice crop
habitats compared to other habitats was
324
similar throughout the year. There were no
data on breeding from Long An, so we
caffilot determine whether increases in rat
numbers were due to immigration or
reproduction.
Population dynamics at Kien Giang
Eight species of rodents were identified from
449 captures. The capture rates of
R. argentiventer (43.7%) and R. losea (45.2%)
were similar. The next highest capture rate
was R. flavipectus (5.1 %). Seven captures
were not identified to species (1.6%), with
the six other species accounting for 4.4% of
captures.
Both R. argentiventer and R. losea were
more abundant in the improved rice variety
habitat than in other habitats (Figure 3). The
relative proportion of each species was
similar within these habitats, except R. losea
was more abundant in improved variety rice
in May 1998.
Breeding of adult female R. argentiventer
and R. losea was intermittent, although both
species tended to be in breeding condition in
similar proportions over time (Figure 3)
(November: X2I =0.11, P > 0.05; February:
X\ = 1.03, P > 0.05; March: X2I =0.16, P > 0.05;
May: X2I =0.14, P > 0.05). No breeding was
evident in October 1997, December 1997 (very
little trapping occurred because of flooding)
or April 1998. The only breeding that occurred
in January 1998 was in the melaleuca forest
(n 1). Breeding did not occur in the
vegetative growth stage of the improved
variety rice crop, but tended to occur in the
latter two-thirds of crop growth from late
tillering to harvest (a period of two months).
Therefore breeding appeared to be linked to
the presence of high quality food.
 Table 3.
Composition of rat species (%) from nine different provinces in the Mekong Delta (1995-1997). Rats were collected by live-trapping and digging
of burrows in rice crops and along channel banks. In each province, fifty rats were collected (except in March 1996 when 100 rats were collected)
from six sampling occasions from November 1995 to July 1997.

Species Ben Tre Tlen Glang Long An Can Tho Dong Thap Klen Glang Minh Hal An Glang Vlnh Long

Rattus argentiventer
_.J 50.2
1
72.0 63.4 67 .8 69.7 28.4 57.3 68.6 75.4
Rattuslosea I, 9.8 ,1 5.3 11 8 .9 11 10.5 11.0 45.9 18.7 12.8 8.9 I!=
:::11
I
Rattus koratensis 17.3 1.3 '1 8.3 4.3 5.5 6.7 11 9.0 11 4.0 d 8 .5 Ir::
:I
Rattus germaini

Rattus rattus
11

11
8.5
1.8
Jl
11
2.0
4.2
Jl
!f
3.0
2.1
H
-f
1.0
3.3
H 2.7
1.0
]I
I:
2.0
0.2
]I
I[
2.7
2.3
]I
j[
2.4
2.8
Ji
j[
2.8
1.2
--
CD
:::11

Cl

::IQ
Cl
Rattus flavipectus 11 1.3 11 1.3 11 2.3 11 2.3 11 1.8 11 0.7 11 0.2 11 0.4 11 0.0 Q.
CD
:::11
2.0 1.7 1.5 2.7 1.3 1I 2.3 3.3 0.8 1-
Rattus nitidus 11 11 I1 1I 11 1.6 11
"a
CD
Rattus exulans 11 1.8 11 3.3 11 2.5 11 1.3 11 0.7 11 1.8 11 0.3 11 2.4 11 0.8 I='
III

Rattus norvegicus

Bandicota indica
11

Il
2.5
3.0
Jl
If
2.7
4.3
JI
If
1.3
4.5
Jl
If
1.5
4.5
2.3
4.0
J
11
3.4
2.6
~I
11
3.8
0.2
]I
11
1.2
0 .6
Jl
11
1.2
0.4 I;
:a:

If
0.0 0.3 2.2 0.7 0.0 1.2 1~
Bandicota bengalensis 11
11 11 JI .l 0.2 . _ 1 !I 1.6
IL
0 .0
Mus sp. 11 1.7 II 1.5 If 0.0 lr 0.0 0.0 !f 5.9 , f 1.0 H 1.6 lr 0 .0
::IQ
:::C.
...CD
Cl
CD
;::;
I

<

I;-
iD'
w
N
UI
Ecologically-based Rodent Management

35

30

25
---
----
Improved variety (2 crops/year)
Non-rice habitats
Flooding

en
1
0 20
Cll
u
c
<1l
"0
c 15
:::J
.0
<i

10

0
Improved variety
Local variety

Jan Feb Mar Apr May Jun Jul Aug Sep Qct Nov Dec
Month
Figure 2.
Mean abundance of rats (per 100 trap nights; standard error) per month in improved variety rice (two
crops per year) and all other habitats combined, Long An, 1994-1997.

The average weight of adult breeding Population dynamics at Tra Vinh

females for R. argentiventer was 114.6 g ( 3.2
standard error [SE]; range = 62-200 g; n = 63)
and for R. losea it was 111.0 g ( 3.4 SE; range =
60-200 g; n = 69).
The proportion of adult females in
breeding condition was similar in improved
variety rice habitat and other habitats
(X 2 = 0.69, degrees of freedom [dJ.] = 1,
F> 0.05; X2 = 1.09, d.f. = 1, P > 0.05 for
R. argentiventer and R. losea, respectively).
The rate of recapture of tagged rats was very
low both within trips (0.7%) and between
trips (0.2%).
Four species of rodents were identified from
384 captures. The most common species
trapped was R. argentiventer (61.7%). R. losea
(20.3%). R. koratensis (10.2%) and R. germaini
(4.2%) made up the other species captured,
with 3.6% of captures not identified to
species.
Both R. argen tiventer and R. losea were
more abtmdant in the rice habitat than in
other habitats (coconut, banana and charmel
bank combined) (Figure 4). No rats were
caught in the rice habitat from November
1997 to May 1998, but some rats were caught
in other habitats in February, March and
May (:::;0.3% trap success).

326
 Management of Rodent Pests in the Mekong River Delta, Vietnam

(a) Improved variety rice

(5.1 0) (3.6) (-.-) (0. 0) (54, 44) (14.1 2) (9 , 6) (3, 15)
16 100

14

80
12

60

40

20

Flooding

O :~~~~~~;::~~;'II~:i====:r"iiiiilliiiiill"""Ii;;~.
Improved variety
Local variety
~ _ _- L_ _ _ _~_ _~_ _L -_ _ _ _~_ _ _ _- L_ _ _ _- L_ _ _ _~_ _ _ _~

Oet Nov Dee Jan Feb Mar Apr May

1997 1998
Monlh

_ R. argentiventer R.losea - . . - R. argentiventer ____ R losea

(b) All other habitats

16 -
(-. -) (1,0) (0,0) ( t. l ) (2,1) (3,0) (7 . 9) (2,6)
- tOO
14 -
- 80
12 -

- 60

- 40

4 -
- 20

2 -
Not

-- - L,--- o
Improved variety

Local variety
trapped

c======~
a.
- ----- -

Oet Nov Dee Jan Feb Mar Apr May

1997 t 998
Month

R. argenliventer R.losea - . . - R. argentiventer ____ R. losea

Figure 3.
Abundance of Rattus argentiventer and Rattus losea ( number caught per 100 t rap nights) and proportion of
adult females breeding (lactating or pregnant) in (a) improved variety rice and ( b) other habitats in Ha Tien,
Kien Giang, October 1997 to May 1998. Numbers in brackets refer to the number of adult female rats
caught for each species R. argentiventer, R. losea). No t rapping occurred in December 1997 in improved
variety rice or in Octo ber 1997 in other habitats. (Unpublished data G.R. Singlet on and N.Q. Hung)

327
Ecologically-based Rodent Management

(a) Rice habitats

(15, 21) (-.- ) (- ,-) (11 ,9) (24 ,27) (14,10) (-,-) (-.-) (-,- ) (-.-) (-,-) (-,-) (-.- )
18 100

Ul
Ul
Ql
16
14

12

-
Abundance (trap success)

--*-
R. argentiventer
R./osea
Other species
80

60
0>
c
'0
Ql
~
.0

-
U
u 10 Ul
Ql
~
Ul
<ii
0. 8 % adult females breeding E
ca 40 ~
t= R. argentiventer "S
6 "0
ca
c=J R. /osea
~
4 20
0

Not
2 trapped
0 0
Improved variety
Local variety

May Jun Jul Aug Sep Oct Nov Oec Jan Feb Mar Apr May
1997 1998
Other habitats
(b) (10, 2) (-.-) (-,-) (4 ,1) (4,1) (10,5) (0 ,0) (0,0) (0,0) (1 ,1) (0,1) (0,0) (2,0)
10 100

Flooding
8 80 0>
C
'0
Ql
Ul ~
Ul
Ql 6 60 .0
U Ul
u Ql
~
Ul
<ii
0. E
ca 4 40 ~
t=

2
20
""
~
"0
ca
~
0

Not
trapped
0 0
Improved variety
Local variety

May Jun Jul Aug Sep Oct Nov Oec Jan Feb Mar Apr May
1997 1998

Figure 4 .
Abundance of Rattus argentiventer, Rattus losea and other species (number caught per 100 trap nights)
and proportion of adult females breeding (lactating or pregnant) in (a) rice habitats and (b) other habitats
in Tra Vinh, May 1997 to May 1998. Numbers in brackets refer to the number of adult R. argentiventer and
R. losea females caught, respectively. No trapping occurred in June or July 1997.

328
 Management of Rodent Pests in the Mekong River Delta, Vietnam
Only one adult female R. argentiventer
was found pregnant (by palpation) during
May 1998 in the rice habitat (6.7% of adult
females) (weight = 174 g). No other adult
females were in breeding condition for any
other period or habitat. No rats were re-
caught within or between trips.
Population ecology of rats in Mekong
River Delta
The dynamics of rat populations were
different for Long An, Kien Giang and Ira
Vinh. In Long An and Ira Vinh, the
abundance of rats was highest during
flooding, whereas in Kien Giang, the highest
abundance was during the early stages of the
reproductive phase of the crop. In Kien
Giang, the abundance of rats in March 1998
(milky /harvesting stage) was lower than
expected considering the breeding that
occurred in February (assuming the young
were trappable). Unfortunately, no trapping
was conducted there during the period of
flooding.
Few rats were caught in non-rice habitats
in either Ira Vinh or Kien Giang. Ihis
suggested that rat populations were
building up within the rice fields rather than
in adjacent non-crop habitats. Breeding
occurred during the reproductive phase of
the crop in Kien Giang, but a comparison
carmot be made with Ira Vinh where only
one rat in breeding condition was caught.
In Ira Vinh, the population abundance of
rats remained very low after flooding. Ihese
low numbers could be attributed directly to
the floodwaters (a) drowning rats, (b)
making food scarce, and/or (c) making it
difficult for rats to find shelter. Furthermore,
low numbers could be indirectly attributed
to the waters driving rats to patches of high
ground where farmers focused their control
campaigns. When floodwaters subsided and
new crops were planted, rat populations
were at such low densities that they could
not recover. For three months after the
floodwaters had subsided, the only rats
present in Ira Vinh were found in a banana
plantation. How long does it take for rats to
recolonise the rice habitats after such a
flooding event? From our data, it seems that
the effect of flooding on rat populations in
Long An or Kien Giang are not as severe as
found in Tra Vinh. Ira Vinh is situated
toward the edge of the delta, and the area is
more prone to flooding events than the other
provinces that have a higher elevation and
where floodwaters can subside more
rapidly. We would not expect the effect of
flooding on rat populations to be the same
each year because the severity and duration
of flooding differs between years.
A radio-telemetry study is required to
gain an understanding of the habitat use of
rats during flooding, and to determine if rats
breed when the local traditional variety rice
is in the reproductive phase of growth.
Furthermore, when flooding occurs it is
important that farmers know where they can
concentrate their rat control activities.
In Kien Giang, there were few rats caught
in April (from transplanting to maximum
tillering). We would expect that populations
would increase during April as new rats
enter the trappable population from births in
February and March. Are the low numbers
attributable to low survival rates and poor
recruitment of young in March and April
1998, or had these rats emigrated to other
areas? Another factor that complicates our
interpretation is that the catchability of rats
is low, or the animals are very trap shy.
329
Ecologically-based Rodent Management
The population dynamics of rats in
Indonesia follows a general pattern of
increasing abundance during the fallow
period after harvesting as young enter the
trappable population (Murakami et al. 1990;
Leung et al., Chapter 14). The difference in
Vietnam is that there is no fallow period
behveen crops. Subsequent crops are sown
within a few days of harvest. Lelmg et al.
(Chapter 14) found that the factors limiting
rat populations in Indonesia were food
quality, availability of nesting sites and
human activities such as land preparation
and rodent control activities. If rice is
planted soon after harvest, then high quality
food is available to the rats sooner. We
therefore expect that rat populations in
Vietnam would have a higher survival rate
in the period behveen breeding seasons.
The low recapture rates of rats fotmd in
the Mekong River Delta are a problem for
live-trapping studies. Wood (1971) obtained
recapture rates in Indonesia of 14% within a
trapping period and 6% behveen trapping
periods for R. argentiventer. Our estimates
were both <1 %. Population parameters are
more difficult to estimate when recapture
rates are low (Krebs et al. 1994). It is not
known whether R. argentiventer or R. losea in
Vietnam are trap-shy or are transient animals
moving through a trapping area. To improve
population parameters, the recapture rate of
rats needs to be enhanced. Research is
required to examine better methods for
trapping rats and to compare results with
other areas such as Indonesia (LK-P. Letmg
and Sudarmaji, unpublished data).
One critical population parameter that we
have not been able to gain sufficient
information on is the survival rates of rats (we
have been restricted by low recapture rates).
330
How is survival influenced by crop stage and
flooding? Survival rates of rats could be
estimated using static life tables rather than
by following individuals over time. This
approach requires a comprehensive set of
data to follow cohorts through time and an
accurate method for ageing rats using eye
lens weight (Murakami et aL 1992).
Breeding at Kien Glang
The only breeding (embryos in the uterus)
evident from kill trapping was in February
and March 1998, where 100% of adult female
R. argentiventer were pregnant. The mean
number of embryos was 11.4 ( 0.4 SE, range
6-18, n = 65). The mean weight of pregnant
females was 91.1 g (5.1 range = 29-183, n
= 65). These rats were significantly smaller
than live-captured pregnant or lactating
R. argentiventer (t =3.87, dJ. =126, P < 0.001).
The minimum weight of pregnant females by
kill-trapping was half that from live-trapping.
This difference is interesting, because it
suggests there could be a bias in the collection
methods (rats caught in burrows versus rats
caught in live-capture traps). Therefore, more
work is required to tmderstand this bias and
to look at improvements to trap design and
trapping procedures.
At harvest in October 1997, many adult
females had uterine scars. The mean number
of scars for R. argentiventer was 10.8 ( 0.5 SE,
range = 6-17, n = 39), which was not
significantly different to the mean number of
embryos (t = 0.93, dJ. = 102, P > 0.05). The
litter size was significantly higher for rats
with more sets of uterine scars (one-way
analysis of variance-ANOVA; F 28.4,
d.f. = 2,36, P < 0.001). The mean number of
scars for rats with one set was 8.0 ( 0.4 SE,
n = 16), hvo sets was 12.1 ( 0.7 SE, n = 15)
 Management of Rodent Pests in the Mekong River Delta, Vietnam
and for three sets was 14.0 ( 0.6 SE, n = 8).
This is evidence to show that R. argentiventer
can have up to three litters during a single
breeding season and that the size of the litter
increases with each litter.
The proportion of rats in breeding
condition collected by kill-trapping in March
1998 was higher than that found by live-
capture trapping (X2 I = 4.42, P < 0.05),
whereas there was no difference in February
(X 2I = 0.64, P > 0.05).
Breeding was evident only during the
reproductive stage of the rice crop. The
relative percentages of R. argel1tiventer adult
females in breeding condition during the
different crop stages were: the vegetative
growth stage, 0% (0/22 rats); tillering stage,
9% (1/11 rats); flowering stage, 100% (5/5
rats); and at harvest, 76% (57/76 rats).
Breeding of rats in Kien Giang was linked
to the development of the improved variety
rice crop. No breeding occurred in the
vegetative stage of growth, but breeding was
initiated at some point prior to maximum
tilleTing stage to take advantage of high
quality food during the reproductive and
ripening stages of rice development. This is
generally the case with R. argentiventer in
other regions such as Indonesia (Murakami et
al. 1990; L.K.-P. Leung and Sudarmaji,
unpublished data) and Malaysia (Wood 1971;
Lam 1980, 1983; Buckle 1990). The discovery
that breeding by rats commenced prior to
maximum tillering of rice crops led to an
important re-evaluation of what triggers
breeding by rats (Leung and Sudarmaji,
unpublished data). Although we have limited
data, we hypothesise that breeding is linked
with the rice crop stage, and that if there are
three rice crops grown per year (2 x improved
variety and 1 x local traditional variety), then
there will be three distinct breeding seasons
per year. Furthermore, if the crops are not
grown in synchrony (planting over a period
of >2 weeks in an area), then the duration of
time in which high quality food is available is
prolonged. Therefore, we would expect that
the breeding season would be prolonged,
resulting in a higher numbers of rats.
Ecologically-based population
management practices for rats
in the Mekong River Delta
Rats have always been part of the rice-
cropping ecosystem in Southeast Asia;
R. argentiventer in particular is believed to
have evolved from a grassland existence
(Lekagul and McNeely 1977). The reason why
rats have become a major pest of rice crops in
the Mekong River Delta is thought to be due
to the increased amount of cropping
occurring (three crops per year instead of one
or two; Singleton and Petch 1994). Other
factors may include the increased awareness
of the problem of rats and the mosaic of
favourable habitats for rats.
We have applied our knowledge of the
ecology of rodents in the Mekong River
Delta reviewed in this chapter, to develop an
ecologically-based appraisal of the
appropriateness of different management
actions (Table 4). These actions were
developed by the Institute of Agricultural
Sciences in Ho Chi Minh City, and were
discussed by rodent scientists at an annual
meeting of the Australian Centre for
International Agricultural Research
(ACIAR) funded project on the management
of rodent pests in Southeast Asia in April
1998 at the International Rice Research
Institute (IRRt Los Banos, the Philippines).
331
 ~
IN n
IN 0
t.J Table 4.
0'
(IQ
Decision analysis of recommended best practices for managing Rattus argent/venter and Rattus losea in rice agro-ecosystems of the Mekong
= = = =
River Delta, Vietnam. (Timing: Ip land preparation; sb seed bed; tp transplanting; b booting; Suitability of actions: ./ yes; X no; = = n'
I

=
? unknown). ~.
=-
I
Management actions 11 Parameters for ecologically-based pest management III
ID
c:a.
l~ QI

..ca I: ::11:1
:is :?;o 0
III
:::J 0

......
III 11 u c:a.
i.
-
QI
G) U "S. :is oS ID
"QI)
:is E QI
u
I:
QI-
....
III
0
=
-ca
I: "a E
E~
-
0 u I: III 'C
E Ui
III I: III I: QI ca ~
G) 0
'C :I:
j::
If
0
U
w
:?;o
ca
O'C
... ~
:::J
0
QI
....III CL I
=
">

~L
I: 0 I
13 w u u (IQ

JLJ 0
I/)
I/)
11 W ID
:3
Routine actions
Field sanitation and dyke management
Synchronous seeding and planting
Ip>b
sb, tp
./
7
./
7 labour
./
./?
./
./
./
?
village
village
?
./
I' high
medium
-
ID
=

-- --~-

Reduce bund size within rice fields Ip ./ ./ 7 ./ ./ ./

JI
village 1I J- high
Keep water level high in the field Dry crop 7 ./ ]I ./ ./ ./ village n/a high
Encourage natural enemies of rats in rice ecosystems all 7 ?
JI ./ ./ 7 district ./ high
Linear TBSa to halt movement into paddies
Management of rats on high ground
< sb, tp
Flood
?
./
?
./
r ./
./
7,/
./
./
./
7
village
./
./
medium
high
1
Locate and destroy rat burrows by fumigating and digging < sb, tp ./ ./ ./ ./ ./ village 7 high
Establish TBSa with trap crops Each crop ? ./ ./ ?./ ./ ? ./ medium
Actions if high numbers forecast
- --- --------
Rat drive using nylon nets Flood ./ ./ 1 ./ ./ ./ village ./ high
Apply chemical bait in fields and villages < sb, tp ./ 7 ./ X X village X low
-------- _I

Burn rice straw after harvest Dry crop ./ ./ II ./ X X village X low

Drive tractor through fallow fields (high terrain)b Flood 7 ? I ./ ? X village X low
aTBS = trap plus barrier system
bPlant Protection Departments perform this activity.
 Management of Rodent Pests In the Mekong River Delta, Vietnam
There are two types of actions; routine
actions that can be conducted all the time
and actions that can be conducted if high rat
numbers are forecast. For each action, eight
parameters were considered: the timing of
implementation, the feasibility, whether it is
economical, socially acceptable,
environmentally friendly, ecologically
sustainable, the scale of adoption and
whether it has an ecosystem focus. We also
considered priority for implementation of
each practice.
Ecologically-based pest management is a
new paradigm for pest management
(National Research Council 1996). It
promotes the use of information on the
biology and ecology of the pest species to
formulate management actions. This
approach has been used for house mice in
Australia (Singleton 1997; Singleton and
Brown 1999) and for R. argentiventer in
Indonesia (Leung et al., Chapter 14). Our
current knowledge of the biology and
ecology of rats in southern Vietnam has
allowed Table 4 to be formulated.
Manipulative field experiments are now
required to examine the effectiveness of
these actions. Until these actions have been
critically tested, they remain a 'best guess' of
what limits rodent populations in the rice
agro-ecosystem of southern Vietnam.
These actions were designed for the rice-
growing areas of southern Vietnam. Some of
the actions could be appropriate for other
areas of Southeast Asia, particularly
northern Vietnam, where many of the
farming practices are similar, although
major flooding events are rare.
A major hurdle to the success of these
actions is being able to forecast when rat
numbers are likely to cause damage. These
monitoring systems could be operated by
farmers themselves or by government
officers. The amount of time farmers require
to implement these actions also needs to be
investigated.
WHERE TO FROM HERE?
The challenge ahead is whether these
actions will be readily adopted at the village
and! or district level. This is an essential
requirement for mobile animals such as
rodents, which can readily reinvade small
areas following a reduction in rodent
densities. In Vietnam, the level of interaction
between researchers (Institute for
Agricultural Sciences) and provincial
extension staff (Plant Protection
Departments) is very good, which bodes
well for the success of implementation of
actions by farmers.
Given our current understanding of the
ecology of rodent species in the Mekong
River Delta, we are able to determine the
likelihood of success of a range of
management practices on rodent
populations even though we do not
currently have sufficient scientific evidence
(Table 5). We identify where we have
sufficient data and which actions have a low,
medium or high chance of success in
reducing rat populations. We lack
experimental data for many recommended
actions, however, we can draw upon
knowledge gained from studies conducted
in Indonesia (Leung et al., Chapter 14). As
our understanding improves, the likelihood
for success will also change. This is an area
for further research.
333
Ecologically-based Rodent Management

Table 5.
Strength of ecologically-based knowledge for the management of rats in the Mekong River Delta. If we have
sufficient data on the ecology and population biology of rats, then 'Yes' appears in the table, if we lack
sufficient data, then ' No' appears in the table. The likely level of success in reducing rat populations using
each recommended action is based on our current knowledge (Iow, medium or high).

Sufficient data Ukellhood of sucess

Routine actions

Field sanitation and dyke management No High

Synchronous seeding and planting No High
Reduce bund size within rice fields Yes High
Keep water level high in paddies No High
Encourage natural enemies of rats in rice ecosystems _ _IL_ _ No Low
Linear TBS a to halt movement into paddies No Low
Management of rats on high ground during flood ing No Medium
Locate and destroy rat burrows by fumigating and digging No Medium
Establ ish TBsa with t rap crops Yes High

Actions if high numbers are forecast

Rat drive using nylon net s No Low/ Medium

Apply chemica l bait in fields and villages No Medium
Burn rice straw after harvest No Medium
Dri ve tractor th ro ugh fallow fi elds (h igh terrain) No Low
aTBS = trap- barrier system

An important ou tput from a dec ision
analysis process is clearer identifica tion of
the key gaps in our scien tific kn owledge for
developing effective management of a
particular rodent p est. To further our
understanding of the ecology of the major
roden t species, we have listed a series of
ecological pa rameters and indica ted the
amow1t of informa tion that is known about
each in Table 6 (follo wing Single ton and
Petch 1994). We have a good understanding
of the abundance, habitat use and breeding
characteristics of R. argentiventer and R.losea,
but we lack specific informa tion for a range
of oth er ecological parameters.
The ecological approach used here to
critically examine management p rac tices has
allowed us to develop an integ rated
management strategy. Developing field
projects to evalu ate these recommenda tions
in close association with fa rmers at the
village or district level (as pointed out by
Leung et a1., Chapter 14) will p rovide the
necessary feedback for developing an
effec tive m anagement strategy.

334
 Management of Rodent Pests in the Mekong River Delta, Vietnam

Table 6.
Summary of the extent of information available on various ecological parameters of the major rodent pest
species in southern Vietnam (- =no information; * =anecdotal reports; * * = restricted to a single sample or
survey; * * * = restricted to one or two growing seasons or a longterm data set not calibrated against other
measures).

Parameter Rattus argentiventer Rattus losea Other species

Abundance *** *** *

Habitat use ** * *** *
Dispersal

Breeding ** **
Survival

Age structure

Diet

Predator/ prey

Disease

Taxonomic status ** ** **
Species interaction * *
Crop damage *
Postharvest damage

CONCLUSIONS Suggested areas for fu rther resea rch

Although we h ave collected data for a
limited period of time, we have been able to
formulate proposals for the management of
rodent pests in southern Vietnam. Current
ev idence suggests that R. argen tiventer in
Vietnam behaves in a similar fa shion to
R. argentiventer in other Southeast Asian
countries. However, the interaction w ith
R. losea requires further study. We sugges t
specific hypotheses about the likely
population ecology of R. argentiventer and
R. losea in the Mekong River Delta. These
hypotheses need to be tes ted when a t leas t
two yea rs d ata, but preferably three or more
years of data , have been collected .
include:
~ Conduct a radio-telemetry study to
examine the habitat use of rats during
flooding p eriods.
~ Conduct a trap catchability study to
examine the best trapping m ethod
available and to test it over different crop
stages and breeding and non-breeding
stages.
~ Estim ate survival rates of rats over
different crop stages and d uring flooding.
~ Conduct manipulative field experiments
to examine the effectiveness of actions to
limit rat da mage.
335
Ecologically-based Rodent Management
~ Develop a monitoring system to enable
forecasting of high rat numbers.
The information gained from this
research will help establish a better
understanding of the population ecology
and habitat use of rodent
ACKNOWLEDGMENTS
The authors are grateful for the efforts of
field staff who collected the data: Nguyen
Viet Quoc, Ngo Dinh Hoang, Le Thien
Thang, Doan Nguyen Thach, Vo Van Tarn,
Tran Van Nhan and Phan Duc Son; and to La
Pham Lan for assistance with translation.
Thanks to Professor Bien and staff at the
Institute of Agricultural Sciences, Ho Chi
Minh City, who have made FRB and MvW
welcome whenever they have visited. We
also thank all the farmers who have allowed
us access to their land for research purposes.
We thank Grant Singleton, Herwig Leirs and
Alison Mills for critically reviewing this
manuscript. This study was part of an
ACIAR-funded project on the Management
of Rodent Pests in Vietnam (ASl/9679). The
CSIRO animal ethics approval number was
94/95-28.
REFERENCES
Buckle, A.P. 1990. Recent developments in rice
rat control. In: Prakash, I., ed., Rodent pest
management. Boca Raton, Florida, CRC
Press, 87-96.
Fall, M.W. 1977. Rodents in tropical rice. Techni-
cal Bulletin No. 36. Los BaI'i.os, College of
Agriculture, University of the Philippines,
39p.
Geddes, A. W.M. 1992. The relative importance of
pre-harvest crop pests in Indonesia.
Chatham, UK, Natural Resource Institute
Bulletin, 47, 7Op.
336
Granjon, L., Duplanter, J.M., Catalan, J. and
Britton-Davidian, J. 1997. Systematics of the
genus Mastomys (Thomas, 1915) (Rodentia:
Muridae): a review. Belgian Journal of
Zoology, 127, 7-18.
Gratz, N.G. 1988. Rodent and human disease: a
global appreciation. In: Prakash, I., ed.,
Rodent pest management. Boca Raton,
Florida, CI~C Press, 101-169.
Hung, N.Q., Hung, N.M. and Sang, ND. 1998.
The rice field rat in Vietnam: integrated
management (Bien, P.V., ed.). Ho Chi Minh
City, Agriculture Publications, Mp.
Huynh, N.V. 1987. Common vertebrate pests of
deepwater rice in the Mekong Delta of
Vietnam. Proceedings of the 1987 Interna-
tional Deepwater Rice Workshop. Manila,
International Rice Research Institute, 599-
603.
Krebs, CJ., Singleton, G.R. and Kenney, A.I.
1994. Six reasons why feral house mouse
populations might have low recapture rates.
Wildlife Research, 21, 559-567.
Lam, Y.M. 1980. Reproductive behaviour of the
rice field rat, Rattus argentiventer and implica-
tions for its control. In: Proceedings of the
National Rice Conference, Malaysia,243-257.
Lam, YM. 1983. Reproduction in the rice field
rat, Rattus argentiventer. Malayan Nature
Journal, 36, 249-282.
Lekagul, B. and McNeely, J.A. 1977. Mammals of
Thailand. Bangkok, Association for the
Conservation of Wildlife, 758p.
Murakami, 0., Priyno, J. and Tritiani, H. 1990.
Population management of the ricefield ra t in
Indonesia. In: Quick, G.R., ed., Rodents and
rice. Los Banos, International Rice Research
Institute, 49-54.
Murakami, 0., Kirana, V.L.T., Priyno, J. and
Tristiani, H. 1992. Tikus sawah: final report
Indonesia-Japan joint programme on food
crop protection project (ATA-162) Phase n.
BOgOf, Indonesia, Directorate of Food Crops
Protection,I01p.
National Research Council 1996. Ecologically
based pest management: new solutions for a
new century. Washington, D.C, National
Academy Press, 144p.
 Management of Rodent Pests in the Mekong River Delta, Vietnam
Singleton, G.R. 1997. Integrated management of
rodents: a Southeast Asian and Australian
perspective. Belgian Journal of Zoology, 127,
157-169.
Singleton, G.R and Brown, P.R 1999. Manage-
ment of mouse plagues in Australia: integra-
tion of population ecology, bio-control and
best farm practice. In: Cowan, D.P. and Feare,
CJ., ed., Advances in vertebrate pest manage-
ment. Zoological Library Vol. 7, Fiirth, Filan-
der Verlag, 189-203.
Singleton, G.R and Petch, D.A 1994. A review of
the biology and management of rodent pests
in Southeast Asia. ACIAR Technical Reports
No. 30. Canberra, ACIAR (Australian Centre
for International Agricultural Research), 65p.
Singleton, G.R., Sudarmaji and Siapermana, S.
1998. An experimental field study to evaluate
a trap-barrier system and fumigation for
controlling the rice field rat, Rattus argentiv-
enter, in rice crops in West Java. Crop Protec-
tion, 17, 55--64.
Sung,C.V.1999. The rodent diversity in Vietnam.
In: Zhang, Z., Hinds, L., Singleton, G. and
Wang, Z., ed., Rodent Biology and Manage-
ment. Abstracts of papers presented at the
International Conference on Rodent Biology
and Management, October 5-9,1998, Beijing,
China. ACrAR Technical Reports No. 45.
Canberra, ACIAR, 130p.
Suntsov, V.v., Ly, T.V.H., Suntsova, N.!. and
Gratz, N.G. 1997. Plague foci in Viet Nam:
zoological and parasitological aspects. Bulle-
tin of the World Health Organisation, 75, 117-
123.
Tien, D.V.1985a. Identification of rats (Rodentia:
Muridae) in Vietnam. Part L Bulletin of
Biology, 7(1), 9-11.
Tien, D.V. 1985b. Identificationofrats (Rodentia:
Muridae) in Vietnam. Part n. Bulletin of
Biology, 7(2), 5-7
van Peenen, P.F., Ryan, RF. and Light, RH. 1969.
Preliminary identification manual for
mammals of South Vietnam. Washington DC,
United States Natural Museum, 31Op.
Wood, B.J. 1971. Investigations of rats in
ricefields demonstrating an effective control
method giving substantial yield increase.
PANS,17,180-193.
337
Puangtong Boonsong, Sermsakdi Hongnark, Kornkaew Suasa-ard, Yuvaluk
Khoprasert, Prasarttong Promkerd, Greangsak Hamarit, Piyanee Nookarn
and Thomas Jakel

Abstract

Thailand is an agricultural country where rice and other agricultural products

contribute a substantial part to the gross domestic product . Rodents cause
problems both in agriculture and as reservoirs of human diseases. Some basic data
on the agricultural situation in Thailand and the damage inflicted by rodents as well
as an overview of past and present efforts in rodent control are presented . The
Ministry of Agriculture and Cooperatives is responsible for rodent problems in
agriculture. In the Department of Agriculture, the Agricultural Zoology Research
Group conducts research on rodent problems . It devises methods for rodent control
and is responsible tor the transfer of technical know-how to extension programs .
Some aspects of the group's research are highlighted, including the use of the
endemic parasitic protozoan Sarcocystis singaporensis as a potential biocontrol
agent. As the Thai government aims to substantially reduce the use of pesticides,
integrated pest management concepts, including new strategies in rodent control ,
are being pursued in pilot areas.

Keywords

Thailand, rodent management, rodent research, integrated pest management,

rodent borne diseases, biological control, Sarcocystis singaporensis, ecology

338
 Rodent Management in Thailand

INTRODUCTION fruit crops and vegetables are also of

economic significance in the region. The
southern region has several sizeable coastal
HAILAND IS A tropical country plains and a mountain chain running along

T located in Southeast Asia

surrounded by Cambodia, Lao
People's Democratic Republic, Myanmar
its western coast. This region has mostly
sandy loam soil suitable for fruit trees and
tree crops, especially rubber and oil palm.
and Malaysia. The peninsula runs down to
the Indian Ocean, and receives the south- MAIN CROPS, AREAS INVOLVED AND
west monsoon from mid-May to October. It CONTRIBUTION TO THE ECONOMY
covers a land area of 513,178 square
Thailand cultivates about 26,523,836 ha or
kilometres and extends about 1,620
51.7% of its total land area (Anonymous
kilometres from north to south and 775
1996). Irrigated land comprises about 15.3%
kilometres from east to west. There are three
of the agricultural area. Of this area, rice
seasons: cool from November to February,
comprises about 52%, field crops 25%, fruit
summer from March to May and rainy
trees and tree crops 16% and idle land 2.5%.
season from June-October. The average
The remaining land is grassland, housing
minimum temperature is 20C and average
and other areas. Rice has a farm value of
maximum temperature is 37C. Annual
nearly US$3 billion. Other major field crops
rainfall averages from 1,000-2,000 mm,
are cassava, corn, sugarcane, oil crops,
varying greatly from place to place and year
perennial trees such as rubber, and fruit
to year.
trees (Table 1).
Thailand is divided into four regions. The
The importance of agriculture to the Thai
northern region is mostly mountain highlands
economy can be measured by its
where many rivers originate and run down
contribution to the gross domestic product
to the central plain. In the north, agriculture
(GOP). Agriculture comprises 16% of the
is mostly limited to the fertile valleys of the
GOP, industry 24%, and commerce and
Chao Phraya River tributaries. Fruit trees,
service sectors the remaining 60%. Apart
forest trees and vegetable crops are the main
from its contribution to the GOP, agriculture
sources of income in the region. The north-
boosts the national economy through wealth
eastern region has a flat rolling terrain called
distribution and provides gainful
the Khorat Plateau. Much of the land has
employment for approximately 64% of the
poor soil fertility and little water. Large
Thai population.
areas are flooded during the rainy season
but are very dry during the rest of the year.
Provinces along the Mae Kong River use this CROP DAMAGE BY RODENTS
water for agriculture with rice being
cultivated mainly for home consumption. Major pest species
Fruit trees and rubber are being promoted to Although there are about 33 murid species in
help green the area. The central plain is the southeastern end of the Asian continent
regarded as the rice bowl of Thailand. Corn, (Corbet and Hill 1992 ), less than half of these

339
Ecologically-based Rodent Management

are considered pests in Thailand. The two
main ecotypes of rodents fOlmd in temperate
zones, those occurring in grassland and
woodland (Wood 1994), also largely apply to
the situation in Thailand. There are pests of
field crops and those of forestry and
orchards. Additionally, cosmopolitan
species like Rattus norvegicus are also
prevalent. Table 2 lists the key pest species of
various crops as observed by the
Agricultural Zoology Research Group
(AZRG) of the Department of Agriculture
during field surveys (Ratanaworabhan 1971;
Suasa-ard et al. 1987; Khoprasert et al. 1990;
Hongnark et al. 1994).
Damage in lowland
Rodent problems in lowland occur mostly in
rice and field crops in the central regions of
Thailand. Because the Chao Phaya River and
Tha Chin River run through the area and
70% of the region is irrigated, farmers can
cultivate throughout the year. In Suphan
Buri, Nakhon Pathom and Pathum Thani,
rice varieties are cultivated which allow five
harvests every two years; alternatively, field
crops (soybean, mungbean, baby corn etc.)
are grown after harvesting the major rice
crop. When food is available all year, rodents
can breed throughout the whole period
(Boonsong et al. 1984a).
The importance of the problem of
damage to rice by rodents in Thailand
previously led to the introduction of a
nationwide control scheme by the Thai-
German Rodent Control Project (see below).
At that time (1976-77), damage assessment
in rice was performed in central, southern,
northern, and north-eastern Thailand
according to established methods (Weis
1981).

Table 1.
Area planted and estimated farm value of principal crops in 1995/96 in Thailand (rate of exchange
1 US$ = 40 baht).

Crop Area (million ha) Farm value (US$ million)

Rice 10.14 2622.1

Rubber 1.82 1064.0
Fruits 0.60 659 .1
Vegetables 0.30 527.1
Coconut 0 .38 78.0
Oil palm 0.10 115.6
Soy bean 0.30 83.5
Sugar cane 1.00 559.4
Maize 1.33 420.7
Cassava 1.26 426.0
Mungbean 0 .35 69.5

Source:
Agricultural Statistics of Thailand, Crop year 1995/ 1996. Office of Agricultural Economics . Ministry of Agriculture
and Cooperatives

340

--- - - - - - -- -
 Rodent Management in Thailand

In each of the four regions, three provinces
were randomly selected and three districts in
each province inspected. In each district,
percentage damage was measured on eight
plots (30 m x 30 m each) two weeks before
harvest of the wet-season rice. Figure 1 shows
that, on average, about 18% of the rice was
damaged in the central plams which translates
to losses of approximately US$300 million. A
more recent survey (199~93) by the AZRG,
employing the same methods in the same
areas, showed that the situation in rice had
improved (Figure 1) (Hongnark et al. 1993)
although an average of 1.5% damage still
equates to losses of about US$35 million.
Whether this reduction of the problem in rice
can be entirely attributed to the control scheme
(see below) or is in part due to other factors
such as natural fluctuations in rodent
populations is not known. Certainly,
awareness of control measures among farmers
has increased substantially. Currently,
problems with rodents in rice appear to be
moderate. It should be noted that Rnttus
argentiventer which is considered the most
serious rice field pest besides Bandicota indica in
Thailand (Wood 1994) was observed rarely
during recent surveys of AZRG. Instead,
Rnttus losea seemed to be more abundant.
In oil palm plantations, losses caused by
rats vary considerably both between years
and between plantations. Damage to mature
palms generally ranged from 6-36%
(Boonsong et a1. 1987). Rodents infesting
older plantations are climbing species which
prefer ripe oil palm fruits. Younger oil palms
are attacked by ground-dwelling species
(Table 2). Although conspicuous damage
was patchy, trapping showed that rats are
well spread (Wood 1987). The density of
Rattus tiomanicus was reported to range from
about 125-625 rats/ha in Malaysia (Wood
and Liau 1984), and the situation appears to
be similar in Thailand (Wood 1987).

Table 2.
Major rodent pest species in Thailand and the crops/areas they affect.

Species Rice fields 011 palms Fruit trees Storage and

and field (mango,longan, houses
crops macadamla etc.)

Bandicota indica J J (young palms, < 3 years)

Bandicota savilei J
Rattus argentiventer J J (young palms, < 3 years)
Rattus bowersi J (young palms, < 3 years)
Rattus exulans
Rattus losea J J
Rattus norvegicus
Rattus rattus J J
Rattus tiomanicus J
Mus caroli J J
Mus cervicolor J J
Mus musculus

341
Ecologically-based Rodent Management

(B . indica, R. losea, Mus cervicolor, Mu s caro/i)

20
are a problem in wheat (damage 6.4%,

1990-1993 Hongnark et al. 1994). In the north, these
18
1976-1977 species-and additionally Rattus mttus-
16 occur in barley plantations which mainly
serve the brewing industry. Various plots in
14
a field of 800 ha were damaged by 0.4-17%
Q)
12 during harvest (Artchawakom e t a1. 1986).
Dl
<1l
E 10
<1l
-0
0~
RODENTS AS CARRIERS OF DISEASE
8
A recent outbreak of leptospirosis in a rural
6
area in north-eastern Thailand killed 107
4 people be tween October-December 1997
while a total of 2,236 had to be treated for
2 leptospirosis during that year (Chokviva t
1998). The incident was broadly covered in
0
s N NE c AV the media and it dre\".1 fresh a ttention to the
rodent problem. Although it is not clea r
Figure 1. which rodent sp ecies actually transmitted
Average pre-harvest damage (%) in rice in
the disease, this incident emphasises the
southern (5), northern (N), north-eastern (NE) ,
and central (C) Thailand, and the average (AV) of
need for a be tter knowledge of the
the four regions from 1990-19 93 compared with epidemiology of rodent-borne diseases in
1997-1977 (Hongnark et al. 1993). this region. During 1986-1988, only 466 cases
of leptospirosis w ere reported. This record is
believed to underes tim.ate the real incid ence
Damage in upland
of the disease beca use laboratory facilities
Upland is d efined as those a reas that are 600 w ere not appropriate for screening of la rge
m above sea level. In Thailand, most upland numbers of blood sa mples from all parts of
is in the north, with some in the north- Thailand. Frequently, leptospirosis is
eastern and southern regions. In these areas, incorrectly diagnosed as influenza or a virus
rice, whea t, barley, te mperate fruits (apple, infection (SilapapochakuI1992). As a
p ear, strawberry, macadamia e tc.), consequence of the recent outbreak of
vegetables (cabbage, carrot, onion, broccoli lep tospirosis, the Environ mental Health
etc.) and coffee are cultivated. In upland, rice Burea u (Departme nt of Hea lth) started an
damaged by rodents appears to be generally extension program (three years, 1998-2000)
low. Birds such as spotted munia (Lonchura to monitor and control leptospirosis in
pUl1ctulatl7), sharp -taiJed munia (Lonchura north-eastern Thailand. The activ ities
striata) and Pegu sparrow (Passerflaveolus ) include (i) training of officials of the
are more problema tic (Hongnark e t a1. 1984). De partment of H ealth w orking in regional
In the north-east, four species of rodents hea lth centres in appropriate protection

342

against leptospirosis, (ii) training of local
technicians to work on rodent population
estimation in urban areas, and (Hi) training
in correct and efficient use of rodenticides.
Also, universities in Thailand (i.e. Mahidol)
started to focus on diseases transmitted by
rodents.
B. indica in Thailand and R. norvegicus
were found to be infected with species of the
hantavirus family (Schmaljohn and Hjelle
1997), members of which are responsible for
an increasing global health problem. The
Seoul virus carried by Norway rats is known
to induce haemorrhagic fever with renal
syndrome in humans, a condition which can
be potentially fatal (Schmaljohn and HjeUe
1997). Sera collected from wild rodents
(B. indica, M. cervicoior, R. losea and R. rattus)
caught in rural areas of Chiang Rai Province,
northern Thailand, were positive using an
immunofluorescent antibody test
(Lietmeyer 1988).
These two examples of rodent-borne
diseases are especially relevant to Thailand
and form only a part of the long list of
pathogens in rodents that can affect humans.
This subject has been dealt with in detail
previously (Gratz 1994; Singleton and Petch
1994; Schrag and Wiener 1995). It must be
noted that humans in Thailand do not only
have close contact with rodents (be it in the
rice field or inside houses), but additionally,
rodents form a part of the diet of the rural
population (see also below). Of 142 and 123
farmers who were surveyed for their rodent
control practice in two different regions in
northern Thailand, 79% and 55%,
respectively, reported consuming rodents
regularly (Boonsong et al. 1994). In
particular, B. indica is highly regarded as a
meat source.
Rodent Management in Thailand
CURRENT CONTROL METHODS AND
EXTENT OF ADOPTION BY FARMERS
Since the Thai-German Rodent Control
Project was set up in 1975, systematic,
preventative rodent control methods have
been recommended for use in rice and field
crops (Weis 1981). After the project ceased in
1979, the Department of Agricultural
Extension continued promoting the methods
in all regions by establishing the Rodent
Control Campaign Project (1988-1993). The
objective of the project was to decrease
rodent populations in agricultural areas and
to maintain low populations by training
2,500 extension officers at the subdistrict
level to have a good knowledge of rodent
control which they could transfer to 250,000
farmers. The operating area included rice
fields in 40 provinces, field crops in 12
provinces, and oil palm and cacao in 12
provinces. The total area was 0.864 million
hectares with extension plots (where
rodenticides were provided through
governmental funds) of 16,000 hectares in 40
provinces. After the first year, the Division
of Project and Programme Evaluation, Office
of Agricultural Economics, undertook an
evaluation of practical adoption by farmers
(Anonymous 1989). It revealed that most
farmers accepted systematic and
preventative rodent control. The survey
evaluation showed that all farmers in
extension plots were highly accepting of
mechanical or physical control techniques,
but 33% did not accept use of chronic
rodenticides. In extension plots, 67% of
farmers accepted use of both acute and
chronic rodenticides. Acceptance of these
poisons in service areas (those areas that had
to buy rodenticides on the private market)
343
Ecologically-based Rodent Management
was less than 10%. After this initial
evaluation was completed, the Department
of Agricultural Extension continued the
Rodent Control Campaign Project for five
years until 1993.
In rice and field crops, the control method
consists of two steps (Weis 1981); knock
down of the rat population and subsequent
maintenance at low density.
~ Knockdown step
Chemical control with zinc phosphide or
trapping, digging, blanketing or drives.
Blanketing or drives are conducted by
groups of people who circle an area (about
0.24 ha), cut the vegetation and herd the rats
into a small area (2-4 square metres) before
they are caught or clubbed.
~ Maintenance of population at low density
Mechanical or physical methods as
mentioned above and chemical control using
chronic rodenticides (anticoagulants) such
as coumatetralyl, brodifacoum, flocoumafen
etc.
After the Rodent Control Campaign
Project had ended in 1993, further
campaigns were organised from 1995 to
1997 in an area of 864,000 ha near the Kong
River in the north-eastern region to control
rodent invasions from Lao People's
Democratic Republic. Governmental service
continued also for every province in that
free rodenticides were provided if extension
officers had spotted a rodent problem. The
AZRG further monitored the control
practices of farmers to obtain a realistic view
of the degree of adoption of the publicised
methods. They interviewed farmers
growing soybeans (after the rice crop was
harvested) in the north, north-eastern and
central regions. This revealed that about
344
90% of farmers used zinc phosphide to
control rodents when damage was
conspicuous (Boonsong et a1. 1994, 1995,
1996). Farmers used germinated rice (paddy
soaked in water for three nights) or broken
rice as bait. The proportion of zinc
phosphide mixed with the bait was
generally 2-3 times higher than the
recommended dose indicating an overuse of
acute poison. When farmers observed bait-
shyness of rats, they switched to a more
attractive bait type such as fresh fish, field
crabs (Somanniathelphusa spp.) or golden
apple snails (Pomacea canaliculata) caught in
rice fields. In the dry season, when crabs
and snails are not abundant, they used
mechanical control techniques such as
shooting, digging or trapping. As
mentioned earlier, many farmers consume
rats; in such areas mechanical control over
chemical methods were preferred. Farmers
also learned that during the booting stage to
harvest of rice, most rodents do not take
poisoned bait due to the presence of the
more attractive rice crop. At that time,
farmers usually dug out rat holes instead of
applying poison. The surveys further
showed that farmers growing rice and field
crops did not like to use anticoagulant
rodenticides for various reasons. First, the
price of anticoagulants was higher than that
of zinc phosphide. Second, anticoagulants
were sold only in big towns and were not as
readily available as zinc phosphide, and
third, the effect of chronic anticoagulants
was considered too slow.
In conclusion, in the long term, the rodent
control scheme has been only partially
adopted by farmers in rice fields and field
crops. There is a reliance on the use of acute
rodenticides despite their limitations

(Prakash 1988), with the only alternative
being traditional, mechanical control.
Control is usually only considered when the
problem is obvious. Preventative measures
during periods when rodent numbers are
low are the exception rather than the rule,
and this situation prevails today.
Rats cause extensive damage to oil palm
estates in southern Thailand. In large oil
palm estates (>30 hectares) the farm
managers generally follow Malaysian plant
protection technology from the Palm Oil
Research Institution of Malaysia (PORlM).
PORIM recommends that control with
anticoagulant rodenticides should
commence when 5% of the oil palm fruits
show fresh damage, and control should be
repeated over large areas every six months.
During field visits by AZRG to oil palm
plantations, it was observed that second-
generation anticoagulants like flocournafen
occasionally led to extensive secondary
poisoning of predators like barn owls (Tyto
alba) (AZRG, unpublished observation). In
small holdings, most farmers are not
interested in rodent control. When the price
of fresh fruit is low (less than 2 baht per
kilogram), oil palms grow in natural
conditions without the use of fertiliser or
rodent pest control.
Other control approaches like habitat
manipulation and protection of known
predators of rodents are regularly proposed
to farmers by the Department of
Agricultural Extension during extension
activities. For instance, farmers employ the
former by regularly clearing excessive
vegetation on dykes. Measures such as
reduction of dyke size are also considered
when new fields are designed. Individual
farmers reported that they especially take
Rodent Management in Thailand
care of predators like birds of prey or snakes,
however, this does not appear to be a
common view. Thus to date, an integrated
rodent control approach, though pursued by
individuals, is not occurring on a broad
scale.
RODENT RESEARCH BY THE AZRG
Rodent research in agriculture is primarily
the responsibility of the Department of
Agriculture, especially the Agricultural
Zoology Research Group (AZRG) of the
Division of Entomology and Zoology.
Research activities focus on various species,
such as rodents, bats, birds, crabs, various
snails and slugs which are injurious to
plants. During the last 20 years, research has
been conducted on the following topics:
~ Species iden tification and density
estimation of rodents in economic crops
such as rice, maize, soybean, mungbean, oil
palm and longan.
~ Life history of key pest species.
~ Ecology: seasonal variations in rodent
density in economic crops.
~ Crop damage and loss assessment in rice,
oil palm, soybean, maize etc.
~ Chemical control: efficacy of rodenticides
in the laboratory and in the field.
~ Integrated pest management: combined
application of rodenticides, mechanical
control and cultural practices.
Following are some examples of research
on the population ecology of pest rodents in
Thailand.
The home range length (or maximum
diameter of the home range) of rodents in
345
Ecologically-based Rodent Management
rice fields has been studied at the Rice
Research Station of Pathum Thani using
eosin stain (Khoprasert et al. 1977) and
mark-release captures in Prachin Buri
Province (Somsook et al. 1983). It was found
that the maximum radius moved within a
week was 90 m and 100 m for B. indica and
Bandicota savilei, respectively. R. argentiventer
moved a maximum radius of 50 m and
R. losea 46 m.
The long-distance movement of rodents
in rice fields has been studied in an area
located in Bang Plama District, Suphan Buri
Province (about 100 km north-west of
Bangkok) (Boonsong et al. 1984b). Two crops
of rice per year were grown with the major
crop planted in July and harvested in
October and the second crop planted in
February and harvested in May. A total of
1,253 rodents were caught in monthly field
trips during January to October 1984. They
were ear-tagged and released in an area of 8
km2 . The catch consisted of four species,
R. argentiventer (51.3%), R. losea (18'Yo),
B. indica (20.4%) and B. savilei (10.3%). Only
six marked rats were recaptured in the
release area; three B. indica, two B. savilei and
one R. losea. This low number was in part
explained by the extreme trap-shyness of
tagged rice-field rats; additionally, farmers
conducted intensive rodent control
campaigns during the study period. Two
great bandicoot rats (B. indica) were re-
trapped 63 and 95 days after release. They
had moved about 1 km from the area where
rice had already been harvested to the area
where rice would be harvested in the next 2
weeks. Two lesser bandicoot rats (B. savilei)
were recaptured after 70 days when they
had moved about 2 km from a rice crop at
tillering stage to a harvesting area. The
346
single recaptured R. losea had moved about 1
km within 40 days from rice at the seedling
stage to a harvesting area. In conclusion, it
appeared that rodents moved towards areas
where rice was being harvested.
In the Central plain, Somsook et al. (1983)
studied the population dynamics of the
lesser rice-field rat, R. from March 1982
to March 1984 in rice fields in Prachin Buri
Province (about 300 km east of Bangkok). In
this area, floating rice is grown once a year.
Rice is planted in June and harvested in
December. The rice stubble is left in the field
until the following February. During the
rainy season, the study area was flooded up
to 1.5-2 m from August to October 1983. In
1984, the water level in the rainy season was
lower with a maximum of 0.50 m. Rodents
were trapped for four nights of each month
(800 trap nights). When the water level was
high, live-traps were placed on polystyrene
sheets. Trapping revealed that B. indica,
R. argentiventer and R. losea were present
with R. losea being the dominant species
(90%). The population of R. losea showed a
clear cyclic pattern with numbers increasing
towards the harvest period of rice (Figure 2).
Quality and availability of food are certainly
major factors influencing the breeding of
rodents (Singleton and Petch 1994). It
appears that the water level in the rice fields
also influenced the population as the
numbers of rats were considerably higher in
the second wet season (Figure 2) when the
vvaterlevel vvaslo>ver.Interestingly,
R. argentiventer could not be trapped in the
area when the water was high (1982), but
was present during a low water level (1983)
indicating that the rats had probably moved
out of the area during flooding. Breeding of
R. losea commenced in September, and most
 Rodent Management in Thailand

pregnant females were caught in November
and December.
The population dynamics of mice in corn
fields were studied in the northern region in
Tak Fa District, Nakhon Sawan (240 km
north of Bangkok) from May 1986 to
September 1991 (Boonsong et al. 1991). In
this area, corn was the principal crop which
suffered extensive damage from rodents.
Mark-release trapping was used to study
the movements and to estimate population
density of rodents. A total of 2,200 rodents
were caught on 37 occasions. Most of the
rodents were mice (96.6%) with 71.8% M.
cervicolor and 24.8% M. caroli. Figure 3 shows
the population dynamics of M. ceruicolor.
Similar to the situation in rats, fluctuations in
mouse populations reflected the changes in
agricultural and climatic conditions. During
the dry season (November-May)

500 28

Transplant-Harvest Transplant-Harvest 24
400

(f)

e
(5
cl
300 -
-----
No. of rats
Rainfall (mm)
20

16 E
.
~
z 12 '(ij
200 0:

o
M A M J J A SON D J F M A M J J A SON D J F M
1982 1983 1984
Year

Figure 2.
Population dynamics ofthe lesser rice-field rat, Rattus losea, in floating rice In the central plains ofThailand
from March :1982 to March :1984 (Somsook et al. :1983), showing the numbers of rats trapped in 800 trap
nights per month and the monthly average rainfall.

347
Ecologically-based Rodent Management

populations increased to 40-90 mice/ha,
while in the rainy season (May--October)
only 12-19 mice/ha were found. Pregnant
females were recorded in the dry as well as
in the wet season (Figure 3) indicating that
M. cervicolor reproduced most of the year
except February to May. The survival period
of this species in the field was about five
months. A home range of 300-400 m 2 was
recorded for M. cervicolor.
Some recommendations for control
campaigns in Thailand have been derived
from the population studies outlined above.
Usually, knockdown of rodent populations
starts in the dry season after harvest when
less food is available and rats readly accept
rodenticide bait containing zinc phosphide.
This is also the time when drives and
blanketing are conducted. Once the
population has decreased in the wet season,
chrome poisons and mechanical control are
used until the booting stage of the various
crops.

140 50

120 - ISeed-Harvest

40
--e- No. of mice
100 __ % pregnant females
g'l
30 ii.i
~ 80 E
'E .if'
'E
as
c:
~
20 0..
"#
40

10
20

MJJASONDJFMAMJJASONDJFMAMJJASONDJFMAM
1986 1987 1988 1989
Year

Figure 3.
Population fluctuations of the fawn-coloured mouse, Mus cervic%r, in corn fields In northern Thailand from
May 1986 to May 1989 (Boonsong et al. 1991), showing numbers of mice trapped during 800 trap nights
per month and the percentage of pregnant females.

348

BIOLOGICAL CONTROL OF RODENTS
USING SARCOCYSTIS SINGAPORENSIS
Since 1993, the AZRG and the Department of
Parasitology, Hohenheim University,
Germany, have cooperated in the framework
of a GTZ (Deutsche Gesellschaft fUr
Technische Zusammenarbeit-German
Technical Cooperation) project to develop a
biological method for rodent control using
the apicomplexan protozoan Sarcocystis
singaporensis which naturally occurs in rats in
Southeast Asia. This section provides some
background information on the biology of
this parasite and considers its possible
application against rodent pest species.
Biology and host range of Sarcocystis
singaporensis
S. singaporensis was discovered by Zaman
and Colley (1975) at a time when the obligate
two-host life cycle of the sarcosporidia had
been recognised (Rommel and Heydom
1972) and research on this group of parasites
was intensive. The original material
(sporocysts) was obtained from faeces of a
reticulated python (Python reticulatus) sold at
a butcher's shop in Singapore (Zaman and
Colley 1975). Shortly after discovery, Zaman
(1976) investigated the intermediate host
range of this species and found the parasite
to be highly host-specific. Only laboratory
rats (R. norvegicus) were susceptible to
infection through the oral route using
sporocysts containing sporozoites, the stage
infective for the intermediate host. Other
animals, like mice (Mus musculus), dogs,
cats, chickens, and a rhesus monkey were
not susceptible to infection and showed no
clinical signs of disease (Zaman 1976; Beaver
and Maleckar 1981). Infection of rats by
Rodent Management in Thailand
sporozoites is usually followed by two
rounds of asexual multiplication inside
endothelial cells of various organs, a process
by which merozoites are formed (Brehm and
Frank 1980). About one month after
infection, merozoites eventually invade the
muscles to form characteristic cysts (so-
called 'sarcocysts') in the striated muscles
which contain a third stage, the bradyzoite.
Bradyzoites are infective for pythons once
the snake preys on rodents.
Subsequently, the definitive and
intermediate host range of S.singaporensis
was studied in more detail using numerous
snake and rodent species from various parts
of the world (HiHner and Frank 1984; Hafner
1987; Jakel et aL 1996, 1997b). These studies
confirmed the reticulated python in
Southeast Asia as the natural definitive host
and most suitab le to infection with respect to
the quantity and quality of sporocysts that
developed in the snake's intestine. It appears
that S. singaporensis also occurs in Australia.
Morphologically similar sarcocysts were
found in Rattus fuscipes (Rzepczyk and
Scholtyseck 1976). Aspidites meianochephaius,
the Australian black-headed python, was
found experimentally to be a suitable
definitive host (Hafner 1987). The natural
definitive host in Australia still remains to be
determined. Among boid snakes outside the
Australasian region, only Python sebae, the
African rock python, could be
experimentally infected. However, numbers
of sporocysts shed with faeces were low and
morphological anomalies of sporocysts
occurred indicating that these snakes do not
provide optimum conditions for the
parasite's development (Hafner 1987).
Therefore, it seems unlikely that
S. singaporensis can survive elsewhere, even
349
Ecologically-based Rodent Management
if other python species and rats were
present.
Among rodents, Rattus spp. and
Bandicota spp. were suitable intermediate
hosts (Hafner and Frank 1984). Additionally,
Nesokia indica, the short-tailed bandicoot rat,
was highly susceptible to infection (Jakel et
al. 1996).
Pathogenic effects of
S. singaporensis in rodents
Zaman (1976) was the first to recognise the
pathogenic potential of S. singaporensis. He
observed that infection of laboratory rats
resulted in acute disease, and death, beyond
a particular inoculation dose. Wood (1985)
made similar observations on infected
Malayan wood rats (R. tiomanicus) in the
laboratory. This was important because it
indicated that there existed a parasite with a
potential to control wild rats. Up-to-date
data on the pathogenicity of parasites in
wild rodents are scarce.
The stage responsible for disease in
rodents is the merozoite which develops
inside endothelial cells. Subclinical
infections (which probably prevail in the
Table 3.
Dose-dependent mortality of wild Norway rats (Rattus norvegicus) of different geographic origin after
infection with Sarcocystis singaporensisa (original data).
Origin of rats Inoculation dose (Number of sporocysts)
1x104 2 X 104
Thailand n.d. b 0/10 (O)C
Egypt 0/6 10/10
Germany 0/6 14/18
a a parasite isolate from Thailand was used
b n.d. = not determined
C numbers of rats that died within 16 days/numbers of rats inoculated. Numbers in parenthesis indicate those
rats which were naturally infected.
350
wild) are characterised by two distinct peaks
of merozoite development in the rat-one
occurs around day 6 post infection (p.i.), the
other around day 16 p.i. (Brehm and Frank
1980). After inoculation of a lethal quantity
of sporocysts, numbers of merozoites
increase enormously around day 11 p.i.,
especially in the lungs. This induces a fatal
pneumonia (Jakel et al. 1996). The factors
responsible for the pathology are not fully
understood . Mechanical destruction of
endothelial cells due to massive
development of merozoites seems one likely
cause. Furthermore, it has been
demonstrated that tumour necrosis factor
released by macrophages upon encounter
with parasite-antigen is able to kill
cultivated cells (Fayer et al. 1988).
The project has determined the degree of
pathogenicity of S. singaporensis in wild
Norway rats (R . norvegicus) from Southeast
Asia (Thailand), North Africa (Egypt) and
Europe (Germany) . The parasite appears to
be more virulent in hosts occurring outside
its natural distribution range. Rats outside
Southeast Asia can be killed with about one
tenth of the inoculation dose (Table 3).
5X 104 1X 105 2 X 105
2 (1)/10 (2) 7 (0)/10 (1) 10 (7)/10 (7)
4/4 n.d. n.d.
3/3 n.d. n.d.
 Rodent Management in Thailand

R. norvegicus is the most resistant species in (R. norvegicus) from Germany, infection had
Thailand. Other species like Rattus exuians, no effect on fertility as the number of
R. argentiventer and R. tiomanicus become progeny of infected females was similar to
moribund at much lower sporocyst doses. non-infected controls (T. Jake!, unpublished
Bandicoot rats (Bandicota spp.) appear to be observation).
particularly susceptible to infection as
1200 , - - - - - - - - - - - - - ,
indicated by massive development of
sarcocysts. Adult bandicoot rats usually do
Wild
not survive an inoculation with 8 x 104 to
iD 1000 Wistar
1 X 105 sporocysts (AZRG, unpublished o
~

~ F-344
observation). Q)
:::J
Whether the bradyzoite (the chronic (JJ
:2 800
stage inside muscles of the rat) can cause Q)
u
(JJ
considerable pathologic effects is equivocal. :::J
E
Intriguingly, the number of parasites which E 600
develop in striated muscles can be extremely ~
Cl

high (up to a billion per gram muscle), IDa.

especially in wild rats (Figure 4), often (JJ 400
2
is
without causing any apparent signs of N
>-
"0
disease. On the other hand, we have <1l

observed considerable numbers of m 200

laboratory rats as well as wild orway rats

(R. norvegicus) that become anorexic, and
show rough fur and slow movements during
chronic infection. It has been dem onstra ted
that Sarcocystis infection renders rod ents
prone to predation (Hoogenboom and
Dijksh'a 1987). Impaired mobility due to
high sarcocyst n umbers in muscle tissue
may be an important contributing fa ctor.
There are controversial observations
concerning the impact of S. singaporensis
infection on rodent fecundity. During
experiments performed in the last 10 years at
the Department of Parasitology, H ohenheim
University, Germany, it was observed that
sub-lethally infected female laboratory rats
(Wistar and F-344 strains) either did not
become pregnant or aborted litters.
However, in a recent experiment with
laboratory-bred wild Norway rats
i ______
o '--__ L-.O___ _ -
400 1,000 4,000 10,000
Dose of sporocysts
Figure 4.
Density of bradyzoites of Sarcocystis
singaporensis in the muscles of wild and
laboratory strains of Rattus norvegicus (F-344,
Wistar) eight weeks after inoculation with va rious
sporocyst doses (400-10,000). Bradyzoites were
obtained by tryptic digestion of muscles; values
are the mean standard deviation of 6-8 rats per
treatment.
In conclusion, S. singaporensis infection in
rodents induces almost 100% mortality once
above a threshold level of infection (Table 3).
The effect of S. singaporensis infection on the
fertility of rodents is less well founded.
When there is a reduction in fertility, the
effect may be influenced by the sex and age

351
Ecologically-based Rodent Management

of the host, its genetic background, or status
of the immune system.
S. singaporensis as a potential
biological control agent
We have examined whether this parasite
could be an effective tactical tool for rodent
control by artificially disseminating food
pellets containing sporocysts among rodents
in the field.
Evidence that the parasite increases the
mortality of rodents under natural
conditions was provided during a field
experiment performed in Egypt Gakel et al.
1996). Infection of a small population of roof
rats (Rattus rattus frugivorous) with food-
pellets containing a lethal amount of
sporocysts killed 73% of the rats when
compared to the effect of a placebo. A
characteristic time-course of activity of
rodents artificially infected in the field is
presented in Figure 5.
Recent field experiments in Thailand
(plots up to 4 ha) showed that S. singaporensiS'
is highly effective against R. norvegicus and
B. indica. Parasite-induced mortality ranged
between 60% and 80% Gakel et at. 1997a;
Jakel et al., 1999). Importantly, the latter
results indicate that S. singaporensis can be
used as a biocontrol agent inside its natural
distribution range in Southeast Asia despite
the fact that the parasite frequently occurs in
rodents in this region (O'Donoghue et al.
1987; Jakel et al. 1997b). This conforms with
the prospects previously outlined by Wood
(1985).

120,---------------------------------.

100
.l!l
'2
::::i 80
~
~
e
.... 60
.0
~
~
:;;;: 40
13

20

O~~--~~--L--L--L__L _ _L _ _ L_ _L_~

o 2 4 6 8 10 12 14 16 18 20 22
Days

Figure 5.
Representative measurement of the activity of wild rats ( Rattus norvegicus) after infection
=
with Sarcocystis singaporensls (day 0 day of infection) (modified from Jikel et al. 1996,
1.997a). The activity of rats (1.0-1.00 animals) is expressed as the consumption of plain
bait or the number of footprints on tracking plates. Note that activity declines around day
10, indicating the onset of parasite-induced mortality among rats. Usually, dead rats can
be seen In the field 1.0-1.6 days post infection. At that time, merozoites of S. slngaporensls
synchronously leave their host cells in the lungs inducing a fatal pneumonia in their hosts.

352

Laboratory experiments showed that
natural infections usually do not provide
protection against lethal sporocyst doses
(Table 3). The reasons behind this remain to
be fully determined, however, analysis of the
T-cell response after infection revealed that
low numbers of sporocysts do not induce the
formation of memory CD4 T-cells (Jiikel et al.
1998) which are usually responsible for the
persistence of immunological memory in the
rat (Bell et al. 1998). Assuming that the high
numbers of bradyzoites developing in the
muscles (Figure 4) are proportional to those
of the preceding stage (the merozoite which
can cause extensive pathology), these data
indicate that rats can tolerate substantial
numbers of multiplying parasites before they
show signs of disease. A certain degree of
infection is tolerated, and may even suppress
immune function (Gill et al. 1988). However,
rats can become resistant to acute infection if
they ingest a sublethal but high number of
sporocysts (Jakel et al. 1996); accordingly, in
this case numbers of memory CD4 T-cells
were significantly increased (Jakel et al. 1998).
With regard to rodent control, these
observations indicate that low numbers of
parasites in the environment are no obstacles
to control measures using the parasite.
However/ a proper bait formulation is needed
which prevents unintended immunisations
during field application.
From research into practice
Current research activities of the project focus
on three major topics: (i) development of a
parasite-bait which is highly palatable and
preserves viability of sporocysts, (ii)
developing conditions for mass-production
of the parasite, and (iii) defining an
application scheme which could complement
Rodent Management in Thailand
or become integrated with other practices of
rodent controL
Developing a suitable parasite-bait is the
most difficult part and will be the key factor
in a possible commercial exploitation of the
method. Ouring our field experiments, a bait
was used which had been developed by the
Bayer AG, Monheim, Germany. It consisted
of a wheat paste with a high oil content and
was sweetened with sugar. The parasite-
inoculated mixture was superior to other
food items and especially attractive to Rattus
spp.and B. indica from Thailand.
Unfortunately, a high oil content seems to
have negative effects on the viability of
sporocysts, and therefore other parasite-bait
formulations are currently under
investigation. The project aims to achieve a
storage stability period of at least three
months under local conditions (high
ambient temperature, high humidity).
Sporocysts can be mass-produced in
reticulated pythons (Jiikel et al. 1996),
therefore the project is constructing a pilot
production unit in cooperation with Thai
private industry to rear and keep pythons. In
Thailand, snake farms are found all over the
country, serving as tourist attractions or
providing products for the leather industry.
We have observed experimentally that
pythons (2-3 m length) shed up to 4 x 109
sporocysts in the faeces after a single
infection without showing any signs of
disease. The animals can be reinfected six to
eight times per year. In the snake's intestine,
immune reactions against the parasite are
not apparent. A lethal dose for a single rat
ranges between 2 x 105 sporocysts (inside the
natural distribution range of the parasite;
Table 3) to 2 X 104 sporocysts (outside the
natural distribution range; Jakel et al. 1996).
353
Ecologically-based Rodent Management
Therefore a single infection of a snake can
yield material to kill about 2 x 104 to 2 x 105
rats.
S. singaporensis could be a new tactical
tool in rodent control, with its application
being similar to chemical rodenticides, and
complementing other non-chemical
approaches (McCallum 1996; Chambers et
a1. 1997; Singleton et aL 1998). Field studies
are planned to determine the effectiveness
and acceptance of the method at the farmers'
leveL As well as an easy-to-use design of a
parasite-bait, a low price will be crucial for
its success on the rodenticide market.
SYNOPSIS AND FuTURE CONCEPTS
As outlined previously, rodent management
in Thailand mainly relies on the use of
chemical rodenticides and mechanical
methods, an approach which was developed
in the mid-seventies. This approach has also
contributed to a substantial reduction in the
rodent problem, especially in rice, because
extension programs were continuously
conducted to reach the farmer. However,
rodent damage to agriculture continues,
notably that by mice to various field crops or
climbing rat species to oil palm. Effective
control against rats is lacking in crop stores
of smalllandholders or in urban areas.
Recent outbreaks of leptospirosis among
humans in Thailand indicate that research
and practical control measures are necessary
to restrict the spreading of rodent-borne
diseases.
According to the 8th National Economic
and Social Development Plan (1997-2001), it
is the policy of the Ministry of Agriculture
and Cooperatives that the use of pesticides
should be reduced as much as possible.
354
Therefore, efforts are underway to implement
an integrated pest management (IPM)
strategy in selected pilot areas in the country.
In 1997, the Department of Agricultural
Extension started an IPM program against
plant diseases and insect pests including 100
demonstration plots (80 ha each) in various
parts of the country. Currently, strategies for
rodent control management included in the
curricula of the education program mainly
focus on chemical approaches, with
environmentally friendly techniques only
playing a minor role at present. The AZRG is
testing new methods in biological control
using the parasitic protozoan Sarcocystis
singaporensis and mechanical approaches like
the trap -barrier system. It is planned to
regularly apply these techniques in
demonstration plots of the above mentioned
IPM program. The future will show if these
methods are accepted by farmers in Thailand
and can be promoted at a larger scale.
REFERENCES
Anonymous 1989. Evaluation of adoption by
farmers of rodent control practices in the
Rodent Control Campaign Project. Division
of Project and Program Evaluation Office of
Agricultural Economics. Ministry of Agricul-
ture and Cooperatives, 59p. (in Thai).
Anonymous 1996. Agricultural data at provin-
ciallevel. Division of Planning, Department
of Agricultural Extension, 160p. (in Thai).
Artchawakom, T., Hongnark, S., Khoprasert, Y.
and Chanyapate, e. 1986. Study on rodent
pests and damage appraisal of barley.
Research Annual Report, Entomology and
Zoology Division, Department of Agricul-
ture, 1-4 (in Thai).
Beaver, P.e. and Maleckar, J.R. 1981. Sarcocystis
singaporensis, Sarcocystis villivillosi sp. n., and
Sarcocystis zamani sp. n.: development,

morphology and persistence in the laboratory
rat, Rattus norvegicus. The Journal of Parasi-
tology, 67, 241-256.
Bell, E.B., Sparshott, S.M. and Bunce, C 1998.
CD4+ T-cell memory, CD45R subsets and the
persistence of antigen-a unifying concept.
Immunology Today, 19,60-64.
Boonsong, P., Chanyapate, C, Artchawakom, T.
and Somsook, S. 1984a. Observation of repro-
ductivity of rice pest rats in Amphor Banglan
Nakornpathom. Research Annual Report,
Entomology and Zoology Division, Depart-
ment of Agriculture, 19p. (in Thai).
Boonsong, P., Chaowattanawong, P., Dhamma-
bamrung, N., Hongnark, 5., Artchawakom,
1., Sianglew, P. and Sarakun, N. 1987. Assess-
ment of damage caused by rats in oil palm.
Research Annual Report. Entomology and
Zoology Division, Department of Agricul-
ture, IIp. (in Thai).
Boonsong, P., Hongnark, 5., Niyomvit, L. and
Khoprasert, Y. 1994. Survey on rodenticides
used by farmers in soybean fields in 1994.
Research Annual Report. Entomology and
Zoology Division, Department of Agricul-
ture, 136-158 (in Thai).
Boongsong, P., Puangprakhone, K., Khoprasert,
Y., Artchawakom, T., Chanyapate, C. and
Chamkrachang, W.1991. Population
dynamic study of rodents in maize. Research
Annual Report, Entomology and Zoology
Division, Department of Agriculture (in
Thai),154-167.
Boonsong, P., Somsook, 5., Artchawakom, T.,
Seehabutr, V., Suwanachai, C and
Tongtavee, K. 1984b. Rat movement in rice
fields. Research Annual Report, Entomology
and Zoology Division, Department of
Agriculture, 1-10 (in Thai).
Boonsong, T., Niyomvit, L., Hongnark, 5.,
Khoprasert, Y. and Nookarn, P. 1995. Survey
on rodenticides used by farmers in soybean
fields in the lower north of Thailand.
Research Annual Report. Entomology and
Zoology Division, Department of Agricul-
ture, 91-133 (in Thai).
Boonsong, T., Niyomvit, L., Hongnark, 5.,
Nookarn, P. and Chanyapate, C.1996. Survey
on rodenticides used by farmers in soybean
Rodent Management in Thailand
fields in the northeast of Thailand. Abstract in
Research Annual Report, Entomology and
Zoology Division, Department of Agricul-
ture, 47-50 (in Thai).
Brehm, H. and Frank, W. 1980. Der Entwicklung-
skreislauf von Sarcocystis singaporensis
Zaman und Colley, 1976 im End- und
Zwischenwirt. Zeitschrift fUr
Parasitenkunde, 62, 15-30.
Chambers, LX, Singleton, G.R. and Hood, G.M.
1997. Immunocontraception as a potential
control method of wild rodent populations.
Belgian Journal of Zoology, 127 (Supp!. I),
145-156.
Chokvivat, W. 1998. Leptospirosis. Department
of Communicable Disease Control, Ministry
of Health. Bangkok, Cooperative Press, 104p.
Corbet, G.B. and Hill, J.E.1992. The mammals of
lndomalayan region: a systematic review.
Natural History Museum Publications.
Oxford, Oxford University Press, 448p.
Fayer, R., Andrews, C. and Dubey, J.P. 1988.
Lysates of Sarcocystis cruzi bradyzoites stimu-
late RA W 264.7 macrophages to produce
tumor necrosis factor (cachectin). The Journal
of Parasitology, 74, 660-664.
Gill,H.s.,Charleston, W.A.G.and Moriarty, K.M.
1988. Immunosuppression in Sarcocystis
muris-infected mice: evidence for suppres-
sion of antibody and cell-mediated responses
to a heterologous antigen. Immunology and
Cell Biology, 66, 209-214.
Gratz, N.G. 1994. Rodents as carriers of disease.
In: Buckle, AP. and Smith, R.H., ed., Rodent
pests and their control. Wallingford, UK,
CAB International, 85-108.
Hafner, U. 1987. Z ystenbildende Coccidien mit
Nager ISchlange-Syklen unter besonderer
Berucksichtigung der Wirtsspezifitat der
Gattung Sarcocystis. Dissertation, Univer-
sitat Hohenheim, Stuttgart.
Hafner, u. and Frank, W. 1984. Host specificity
and host range of the genus Sarcocystis in
three snake-rodent life cycles. Zentralblatt
fUr Bakteriologie, Mikrobiologie und
Hygiene, Originale A, 256, 296-299.
Hongnark, 5., Boonsong, P., Khoprasert, Y.,
Tippayaruk, 5., Tangjittrong, A and Jairin, J.
1994. Study on rodent pests and damage
355
Ecologically-based Rodent Management
apprisal of wheat. Research Annual Report,
Entomology and Zoology Division, Depart-
ment of Agriculture, 75-81 (in Thai).
Hongnark, S., Boonsong, T., Tongtavee, K.,
Chanyapate, C, Suasa-ard, K. and Kaewta, T.
1993. Damage apprisal of rice due to rodent
in Thailand. Research Annual Report,
Entomology and Zoology Division, Depart-
ment of Agriculture, 10-18 (in Thai).
Hongnark, S., Sudto, P., Wongraj, C, Artchawa-
kom, T., Somsook, S. and Tongtavee, K. 1984.
Study on bird pests, damage appraisal and
their control in upland rice. Research Annual
Report, Entomology and Zoology Division,
Department of Agriculture, 9p. (in Thai).
Hoogenboom, l. and Dijkstra, C 1987. Sarcocystis
cernae: a parasite increasing the risk of preda-
tion of its intermediate host, Microtus arvalis.
Oecologia, 74, 86-92.
Jake!, T., Burgstaller, H. and Frank, W. 1996.
Sarcocystis singaporensis: studies on host
specificity, pathogenicity, and potential use
as a biocontrol agent of wild rats. The Journal
of ParaSitology, 82, 280-287.
Jakel, T., Khoprasert, Y., Baumann, C, Kliemt, D.,
Sangchai, J.and Hongnark, S. 1997a. Artificial
Sarcocystis infection in controlling commen-
sal rodents. Abstract, 7th International Theri-
ological Congress, September 6-11,
Acapu1co, Mexico.
Jlikel, T., Khoprasert, Y., Sorger, 1., Kliemt, D.,
Seehabutr, V., Suasa-ard, K. and Hongnark, S.
1997b. Sarcosporidiasis in rodents from
Thailand. Journal of Wildlife Diseases, 33,
860-867.
Jakel, T., Sangchai, J. and Khoprasert, Y. 1998.
The role of immunological memory in Sarco-
cystis infection. In: Ortega-Mora, L.M.,
Gomez-Bautista, M., Bueno-Pereira, J. and
Tenter, A.M., ed., Proceedings of the COST-
8201998 Annual Workshop: vaccines against
animal coccidioses. Toledo, Spain, European
Commission, 50.
Jake!, T., Khoprasert, Y., S., Archer-
Baumann, C, Suasa-ard, K., Promkerd, P.,
Kliemt, D., Boonsong, P and Hongnark, S.
1999. Biological control pf rodents using
Sarcocystissingaporensis. InternationalJournal
for Parasitology (in
356
Khoprasert, Y., Artchawakom, T., Sehabutr, V.
and Boonsong, P. 1990. Survey on rat species
and damage assessment in longan. Research
Annual Report, Entomology and Zoology
Division, Department of Agriculture, 62-68
(in Thai).
Khoprasert, Y., Suasa-ard, K., Chanyapate, C,
Wongraj. C, Suwanachai, C and Thitipawat,
K 1977. Movement study of rodents in the
rice fields. Research Annual Report, Entomol-
ogy and Zoology Division, Department of
Agriculture, 2p. (in Thai).
Lietmeyer, K. 1988. Epizootiology of hantavi-
ruses in northern Thailand. Masters Thesis of
Public Health, Yale University.
McCallum, H. 1996. Immuncontraception for
wildlife population control. Trends in
Ecology and Evolution, 11,491-493.
O'Donoghue, J.P., Watts, CH.S. and Dixon, BR
1987. Ultrastructure of Sarcocystis spp. (Proto-
zoa apicomplexa) in rodents from North
Sulawesi and West Java, Indonesia. Journal of
Wildlife Diseases, 23, 225-232.
Prakash, I. 1988. Bait shyness and poison
aversion. In: Prakash, I., ed., Rodent pest
management. Boca Raton, CRC Press, 321-
329.
Ratanaworabhan, S. 1971. Rodent pests and their
control. Journal of Agricultural Science, 4, 45-
57 (in Thai).
Rommel, M. and Heydorn, AO. 1972. Beitrage
zum Lebenszyklus der Sarcosporidien, Ill.
Isospora hominis (Railliet und Lucet, 1891)
Wenyon, 1923,eine Dauerformder
Sarcosporidien des Rindes und des
Schweinsl. Berliner und MUncher Tierarztli-
che Wochenschrift, 85,143-145.
Rzepczyk, C and Scholtyseck, E. 1976. Light and
electron microscope studies on the Sarcocystis
of Rattus fuscipes, an Australian rat. Zeitschrift
rur Parasitenkunde, 50,137-150.
Schmaljohn, C and Hjelle, B. 1997. Hantaviruses:
a global disease problem. Emerging Infec-
tious Diseases, 3, 95-104.
Schrag, S.J. and Wiener, P. 1995. Emerging infec-
tious disease: what are the relative roles of
ecology and evolution? Trends in Ecology
and Evolution, lO,319-324.

Silapapochakul, K 1992. Acute PUO. In: Infec-
tious Disease Society of Thailand: current
therapy of common infectious desease.
Bangkok, Thailand, Medical Media, 66-78.
Singleton, G.R. and Petch, D.A. 1994. A review of
the biology and management of rodent pests
in Southeast Asia. Canberra, Australian
Centre for Agricultural Research, Technical
Report No. 30, 65p.
Singleton, G.R., Sudarmaji and Suriapermana, S.
1998. An experimental field study to evaluate
a trap-barrier system and fumigation for
controlling the rice field rat, Rattus argentiv-
enter, in rice crops in West Java. Crop Protec-
tion, 17, 55-64.
Somsook, S., Hongnark, S., Suwanachai, c.,
Kaewta, T., Artchawakom, T. and Tongtavee,
K 1983. Study on population dynamics of
Rattus losea (Swinhoe) in rice fields. Research
Annual Report. Entomology and Zoology
Division, Department of Agriculture, lOp. (in
Thai).
Suasa-ard, K, Hongnark, S., Khoprasert, Y.,
Sawanachai, S., Bamrungsuk, S. and Keawta,
T. 1987. Survey on rat species and damage
assessment in sugar cane fields. Research
Annual Report, Entomology and Zoology
Division, Department of Agriculture, 20p. (in
Thai).
Rodent Management in Thailand
Weis, N. 1981. Rodent pests and their control.
Eschborn, Germany, GTZ (Deutche Gesells-
chaft fi.ir Technische Zusammenarbeit), 206p.
Wood, B.J. 1985. Biological control of verte-
brates-a review, and assessment of
prospects for Malaysia. Journal of Plant
Protection in the Tropics, 2, 67-97.
Wood, B.J. 1987. Oil palm pest control in
Thailand. Thailand Oil Palm Research and
Development Project THA/84/007, FAO.
Kuala Lumpur, Sime Darby Services, 82p.
Wood, RJ. 1994. Rodents in agriculture and
forestry. In: Buckle, A.P. and Smith, R.H., ed.,
Rodent pests and their control. Wallingford,
UK, CAB International, 45-83.
Wood, RJ. and Liau, 5.5.1984. A long term study
of R. tiomanicus populations in an oil palm
plantation in Johore, Malaysia. n. Recovery
from controt and economic aspects. Journal
of Applied Ecology, 21, 465-472.
Zaman, V. 1976. Host range of Sarcocystis orienta-
lis. Southeast Asian Journal of Tropical
Medicine and Public Health, 7, 112.
Zaman, V. and Colley, F.C. 1975. Light and
electron microscopic observations of the life
cycle of Sarcocystis orientalis sp. n. in the rat
(Rattus norvegicus) and the Malaysian reticu-
lated python (Python reticulatus). Zeitschrift
fUr Parasitenkunde, 47,169-185.
357
Gary C. Jahn, Mak Solieng, Peter G. Cox, and Chhorn Nel

Abstract

In Cambodia , rats destroy an estimated average 0 .1% of the total rice production
area annually. The impact of rat damage can be great on individual farmers and their
families . Due to the small-scale, subsistence nature of Cambodian rice farming, and
because of poor food distribution, rat outbreaks destroy savings and create food
shortages. An outbreak in 1996 destroyed rice sufficient to feed over 50,000 people
for a year. Typically, farmers' rat management efforts have poor success . To improve
rat management at a village leve l, we initiated farmer participatory research (FPR) in
April 1998 and began meeting with farmers in nine villages in the Svay Teap district
of Svay Rieng Province in south-eastern Cambodia , bordering Vietnam. The
objectives of the FPR were to identify weaknesses in current farmer practices to
manage rats , compare community-based to individual farmer-based rat
management, and test the effectiveness of early wet season trap crops as a means
of reducing rat populations in the wet season.

Keywords:

Rats, rice, farmer participatory research, Cambodia, crop loss

358

OVERVIEW OF RODENT DAMAGE IN
LOWLAND RICE
ARMERS REPORT two to six kinds
F of rats or bandicoots in their
villages. Their descriptions coincide
with accepted rodent taxonomy. Collection
of rats by hunting and trapping confirm the
presence of three species in or near rice
fields: Rattus argentiventer, Bandicota indica
and Rattus exulans. Based on farmers'
descriptions, Rattus rattus, Rattus koratensis
and Rattus losea are probably present in
villages as well (Leung 1998).
In Cambodia, rats are the only pests that
attack all stages of the rice crop in each of the
major rice ecosystems: rainfed lowland,
irrigated lowland, upland and deepwater
rice (Jahn et a1. 1997a). The area of rice
production is 86% rainfed lowland, 8%
irrigated lowland and 6% deepwater and
upland rice Oavier 1997). Among lowland
rice farmers (n 1265),27% reported wet
season rat problems and 46% reported dry
season rat problems. However, only 13% of
farmers actively attempted to manage rats.
Farmers often regard rat infestations, like the
weather, as a force that cannot be controlled.
In both seasons, rodenticides are the most
common form of rat management (Jahn et a1.
1997b).
Annual pre-harvest rice losses due to rat
damage average US$439,000 (Table 1), a
substantial figure in a country where the
average annual per capita income is U5$200
(Anon. 1996). These crop loss figures are
from only 8 of 20 provinces. They represent
significant under-estimates because rat
damage in fields can easily go undetected or
unreported and estimates of post-harvest
Rat Management in Cambodia
damage are not included. Nevertheless,
some observations can be made:
.... the total rice area damaged by rodents in
Cambodia is estimated to average
approximately 2,000 ha/year -this is
about 0.1 % of the total rice area (2.3 million
ha in 1996);
.... the level of rat damage varies greatly from
year to year; and
.... some provinces are more affected than
others e.g. Svay Rieng compared with Siem
Reap (Table 1).
Estimated average losses suggest that rats
may not be a particularly important
constraint to national rice production. For
example, in 1996, yield losses from rats
represented 0.3% of national production and
only 4% ofSvay Rieng's total paddy
production. However, national statistics do
not convey the fact that, during an outbreak,
hundreds of farmers lose their entire crop,
sending them into a cycle of poverty from
which few escape. To recover losses, a
farmer must borrow money at very high
interest rates. This means that the farmer
cannot afford essential inputs, such as
fertiliser or additional labour, in the
following season. This leads to weaker
crops, greater susceptibility to pests and
ultimately lower yields, necessitating further
borrowings.
Another aspect of crop loss is the poor
distribution of food in Cambodia. From 1995
to 1997 Cambodia produced a rice surplus
for the first time since 1969, but hundreds of
villages suffered food shortages. The rat
outbreaks of 1996 destroyed at least 12,600 t
of paddy (Table I)-equivalent to enough
rice to feed more than 50,000 people for a
359
Ecologically-based Rodent Management

year. Many of Cambodia's rice farmers are
self-sufficient, producing just enough food
for their own families. When these families
lose their crops to rats, they generally do not
have the resources to purchase rice, and food
shortages result. While rats appear to have a
relatively minor impact on national rice
production, they can have a major impact on
individual fields (ClAP 1998). Occasionally
the total loss of some rice crops is reported in
most Cambodian districts. In certain
provinces, e.g. in Svay Rieng (Figure 1),
entire crops are destroyed by rats every year.

o 50 100 km
I I

t
THAILAND

Figure 1.
Map of Cambodia.

360

Because more dry season rice is now
produced, rice is available all year round in
more places. This provides a source of food to
sustain and increase rat populations,
potentially promoting rat problems in the
future. The trend in the yield loss data (Table
1) is consistent with this projection. The
different timing of rice crops in Cambodia and
Vietnam may also contribute to increasing rat
numbers and population migrations
following the next available crop, or
flood waters (G.R. Singleton, pers. comm.).
Farmers in Svay Rieng near the Vietnam
border reported that rat populations were
very high just after the end of the Khmer
Rouge period in 1979. This may have been
associated with poor maintenance of fields
and loss of control of rat populations during
that time. Altematively, when the war ended
and rice cultivation was extremely low, rats
from the vast, weE..'<iy, uncultivated areas may
have congregated in the cultivated fields.
Farmers (n = 50) in Svay Rieng reported
high rat activity during 1995-97, which is
consistent with data in Table 1. These
farmers saw rats as a significant pest
problem and made traps or bought poisons
to control them. The poison commonly
available is zinc phosphide imported from
Thailand or Vietnam. Chemical analysis by
the Division of Agricultural Toxic
Substances of the Department of Agriculture
of Thailand revealed that 'purple powder',
said by vendors to come from Vietnam,
contained no zinc phosphide. A product
from Thailand, labelled "80% zinc
phosphide" was found to contain 12% zinc
phosphide. The instructions on the label,
however, were appropriate for 12% zinc
phosphide. Rodenticides are sold in
Cambodia without any labels, or with labels
Rat Management in Cambodia
in Thai, Vietnamese or Chinese, but not in
Khmer (the language of Cambodia). Chronic
rodenticides, used in Thailand and Vietnam,
are not commonly available. This may be
due to their high price.
We have seen examples of active rat
fences with traps, known as trap-barrier
systems (TBSs), in Prey Veng (Figure 1)
around early wet season (EWS) crops in high
risk situations. The Cambodia-IRRI
Australia Project (ClAP), a collaborative
research project between the Cambodian
Department of Agronomy and the
Intemational Rice Research Institute (IRRI)
with funding from the Australian
government, has used TBSs in farming
systems demonstrations throughout
Cambodia and on research stations since
1990. ClAP produced an instructional video
in Khmer about the TBS in 1993. The video
tape has been broadcast nationally several
times every year since then. of the
video were distributed to provincial
agriculture offices free of and it is
occasionally seen being played in rural
restaurants around the country. Infonnation
about TBSs is also coming from Vietnam. In
addition, some farmers learn about the TBS
through integrated pest management (IPM)
farmer field schools or from non-
governmental organisations (NGO).
Although they are eaten in Cambodia
(and are quite tasty), rats do not appear to
form an appreciable part of the diet in rural
areas of Cambodia, where rats are
sometimes considered a snack for men, but
inappropriate food for women. Naturally,
during famines and food shortages, rats (and
just about any other animals) are eaten.
Some Cambodians have made a business of
catching rats and selling them to Vietnam.
361
tAl m
CO)
~
N C
Table 1.
C
1!9.
CO)
Lowland rice production losses due to rat damage in Cambodia from 1990 to 1996 (- = no information available). I

-<
Province 'I 1990 1991 11 1992 1993 1994 1995 1996 Total Average I~
I
CfI
ID
Area Damaged (ha) I
Q.
J ::a
cQ.
Kampong Thom
il 161 118 76 68 456 1965 2844
1 474 I
Siem Reap I'
!
11

1
1
1
L
181
L l
11
103 I
- ~ I
284 142
.J

J ..=
ID

Battambang 592 [ 98 [ 193

Ir 883 294 ] ==
... 'JII
I
11 11 11 I1 =
r
" :r I l I
Kandal
--- - -
452 [ 1
L_._---1L
72 1 ] 524 262
J
(IQ
ID
W-
..=
~-

r r 11 1i 1 :3
Prey Veng
lL [ 64 86 511 125 786 197
J ID
1
Svay Rieng 1I 93 164 236 472 230 786 4902 6883 983
1
Takeo
-I
If [ e-
56
11
II
183
1~
-.~
123 362
H 121
Kampong Cham 1
375 56 580 1011 337
._ J
Total "l 1512 543 535 1052 384 1856 7695 13577 1 940

Average yield (t/ha) 1.30 1.35 1.40 1.45 1.50 1.60 1.64
--
Est. production loss (t) 1 965 733 749 1525 576 2969 12619 21136 3019

Value of production loss (US$ ) 285818 106618 108945 221818 83782 431855 1835490 3074327 439190

Note: (a) price of rice is 400 riel/kg; (b) exchange rate US$l = 2750 riel (1997 ).
Source: Department of Agronomy, Ministry of Agriculture, Forestry and Fisheries, Royal Government of Cambodia.

FARMER PARTICIPATORY RESEARCH ON
RAT MANAGEMENT
In April 1998, ClAP joined with the Catholic
Relief Service (CRS)-an international
NCO, to improve rat management in Svay
Rieng through education and research. CRS
was already using action research to work
with farmers on issues that they saw as
important. We refer to this approach as
'farmer participatory research' (FPR). By
involving farmers directly in the research
process, we aimed to develop more
appropriate strategies more quickly (Cox et
aL 1997). We assessed current rodent
management techniques, identified obvious
weaknesses that could be improved
immediately, and identified areas that
required investigation to see if
improvements could be made.
Farmers' practices
The farmers were familiar with TBSs, but did
not use them because they were too
expensive. Some farmers in neighbouring
provinces were using the TBS on EWS crops,
which are more valuable and more
susceptible to rat damage than traditional
wet season crops. In these cases, the value of
the crop may have been high enough to
justify the cost (Table 2). In the wet season,
farmers usually grow traditional
Cambodian rice varieties that take five to
seven months to mature. These varieties will
be at the booting stage, a stage particularly
vulnerable to rat damage, in October or
November. EWS crops are usually modern
IR varieties (i.e. varieties developed by
IRRI), which take only 110 to 130 days to
mature. Planted at the beginning of the wet
season, the EWS crops are at booting stage in
Rat Management in Cambodia
June or July, when other fields have not been
transplanted or are still tillering. The wet
season farmers in Svay Teap (a district of
Svay Rieng Province) were trying to control
rats, but with limited success. Some farmers
in the area were already experimenting with
EWS crops. None of the farmers had ever
heard of using an EWS crop with a TBS as a
trap crop for rats.
The last rat control campaign was a
government-sponsored rat hunt about 15
years ago, when villagers were paid for rat
tails. Some farmers dug out the rat burrows
in their own fields and a few set home-made
traps. A number of farmers used rat bait but
complained that it was not very effective.
Zinc phosphide was the only rodenticide
used and no pre-baiting occurred. Baits were
only left out overnight and collected in the
morning to prevent poisoning of domestic
animals. Farmers had no clear idea of when
to apply poison baits or the dosage of poison
to use. Only a few farmers built traps, but
these were not very effective (G.c. Jahn,
personal observation).
Most of the farmers in Svay Teap thought
that rats were indigenous populations that
lived in the forest, sisal, bunds and weedy
areas during the dry season. They thought
that rats did not migrate very far. In contrast,
farmers in Kampong Ra (another district of
Svay Rieng) and Takeo thought that rats
migrated from Vietnam. If the rat
populations in Svay Teap were indigenous,
we reasoned that hunting rats before the wet
season, and using trap crops in the EWS
would help reduce the rat population in the
wet season. ClAP and CRS paid for TBSs on
EWS trap crops in three villages: Bot Slok,
Veal and Chrok Metes (Figure 2).
363
 Table 2.
w 1"1"1
Cost of trap-barrier system (TBS) materials relative to the value of different rice crops. C"')

~ o
0'
Crop Variety Field size (ha) Yield (kg) Value of rice (rlel/kg) Value of harvest (rlel) Cost of T85 (rlel) ~.
C"')
III
Wet season Traditional 0.7 500 290 145000 113500 ~
eT
III
Early wet season Modern 0 .7 990 500 495000 113500 III
ID
Cl.
~
o

..=
Cl.
ID

3:::

t
III

1----1
=III
IrQ
ID
1 km

..=
Vietnam 3
scale ID
N

Community rat management

Individual rat management
Usual farmer practices

Tnottor
Toeung

Figure 2.
Map of Chrok Metes Commune (Svay Teap District, Svay Rieng Province, Cambodia). Trap crops and other communal practices were tested in
the vii ages of Veal, Chrok Metes and Bot Slok. Individual farmers were trained in rat management in Kampot Skir, Toul Ampil and Thlok. Usual
rat management practices were monitored in Dombok Chour, Kampot Pros and Prey Top.

Rat hlmts were conducted in each of
these villages before the TBS was erected
(Figure 3). The villagers constructed the
traps and put up the fences after receiving
instructions (Figures 4 and 5). Later in the
season, improved baiting and trapping
methods were taught to these villagers as a
group, encouraging synchronous,
community-based rat control. Farmers in all
of these villages monitored rat populations
and damage to crops. EWS yields were
weighed, adjusted to 14% moisture and
converted to t/ha . Technicians from CRS
assessed grain damage.
RESUL1S AND DISCUSSION
Rat hlmts were very popular with the
villagers, although the number of rats
L-______________~. .~~__~__________~____~~__~~~__~~~__________~O"
Figure 3.
Rice farmers prepare for a night rat hunt in Svay Rieng Provi nce, Cambodia
Rat Management in Cambodia
(unidentified species) captured on these
hunts varied from none to 48, depending on
the village. The differences in the number of
rats captured may have been due to the
number of rats living in the village, as
indicated by the rat counts from the rice
bunds later in the season (Table 3). The trap
crops caught few rats (Table 3). The higher
number of rats caught in bunds throughout
Bot Slok suggests that the plastic barrier may
have kept rats out of the field, although rats
did not enter the traps. If so, then the TBS
protects the EWS crop from rats, but does
not actually serve as a means of reducing rat
populations in the wet season. Besides rats,
the traps caught other animals (Table 3). The
villagers ate all these animals.
x
o
U
Q)
Q)
0..
.9o
.L
365
Ecologically-based Rodent Management

Figure 4.
Villagers in Svay Rieng Province, Cambodia, building a rat fence with traps around an early
wet season crop.

Figure 5.
Khmer villagers building rat traps.

366
 Table 3 .
Numbers of rats (and other animals) captured in three villages by early wet season (EWS) trap cropsa, organised rat huntingb, and individual
digging of bunds c .

Village Size of EWS TBS trap crop >.

.Q
....\'Cl
trap crop (m)
Rat Snake Frog Crab Bird
_
.
",

~'Q.O
QI C
.l!l
ea::a
'Q.O
'C
QI
== 'Q.O",,,,
'0 :; .:.: c 3: ~
o 'Q.O:;:;
c:i.!c .'S,.c
o \'Cl >. 'la"'1:::S
.. .Q
z~.s Z~.Q
- :; c
O:'C.Q_

Bot Slok 50 x 50 9 5 1 9 39 (night) 8 (night) 52

50 (day) 40 (day)
Veal 85 x 75 2 4 10 1 30 (day) 10 (day)
L 0
Chrok Metes 40 x 35 7 4 3 1 25 (day) o (day) ,r 0

a The table indicates the number of rats captured by trap crops from June to September 1998.

b Villagers hunted rats before the beginning of the wet season on a single day in April 1998. In Bot Slok, rats were also hunted the night before the
daytime hunt.
C Villagers dug and killed rats in the bunds of rice fields other than the EWS crop from July to September 1998.

-=
:::a
I

3:
I
I
(IQ
CD
3

-==
CD

(")
I
3
w
=-
o
Cl.
(7)
I
"""
Ecologically-based Rodent Management
The EWS crops fared poorly, yielding an
average of 0.6 t/ha. Average yields (n 239
crops) in this area are 2.0 t/ha for early
maturing rice varieties in the dry season, and
0.7 t/ha for late maturing varieties in the wet
season (including fields with zero yield due
to drought or pests) (Jahn et a1. 1996). Yields
were reduced by drought, weeds, and rice
bugs (Leptocorisa oratorius). On average, 40%
of the grains were empty (primarily due to
rice bug feeding), an additional 27% of the
grains were partially filled or had rice bug
damage. On average, only a third of the
grains were intact and undamaged. The
scarcity of water allowed a variety of weeds
to take over EWS fields. The weeds gave rice
bugs a place to breed and feed until the rice
reached the milk stage, which rice bugs
prefer. The EWS crop was the only available
milk stage crop in the village, so rice bugs
were concentrated there at higher than usual
levels. The EWS crop itself ripened unevenly
due to the puddling of water in parts of an
otherwise dry field. This asynchronous
development raised the effective density of
rice bugs in the fields as they moved from
one patch of milk stage rice to another. An
attempt to save the EWS crop in Bot Slok
from rice bugs with insecticide was
unsuccessful. The farmers said that they
would not try to grow an EWS crop again.
There were advantages and disadvantages
of each of the rat management teclmiques
(Table 4). Traps and baits are both used
already (but techniques could be improved);
rat fences are not used in the wet season
because they are too expensive. From
discussions with farmers we found that the
cost of a rat fence is prohibitive, based on
private costs and returns. In fact, the cost of
the fence may exceed the value of the crop in
368
some cases in the wet season. A TBS, like
plain trapping, requires continuous
supervision, both to check its integrity as well
as to remove any rats caught. We saw
remains of a rat fence in one village-the
technology had been tried already,
considered and discarded. Other problems
encountered included: poor quality poisons;
the need to time baiting operations to avoid
poisoning domestic animals; the need to
monitor traps regularly; damage to plastic
fences from wind and cattle; the requirement
for supervision of the TBS; and the danger of
theft of the traps and plastic.
We can improve the effectiveness of
baiting and trapping by paying attention to
the quality of materials and training in
techniques. But we see little that we can do
to improve the cost-effectiveness of plastic
fences - the major problem is the high cost
in relation to the value of the losses
associated with rat damage. The traps were
effective for catching B. indica, but less
effective for R. argentiventer. The latter was
the only species captured in rice bunds.
The justification for early rat
management (when populations are low)
suggests the use of a trap crop. Although it
may be technically feasible to plant a crop
out of season just to catch rats, this would
need to be done at a community level.
However, in Svay Teap, villagers were
reluctant to participate in communal
activities. pointed out that during the
Khmer Rouge period tlwy were forced to
produce rice communally with terrible
results.
The TBS is used most extensively in
Cambodia by dry season farmers growing
irrigated rice. Some farmers are currently
experimenting during the EWS with modern
 Table 4_
Comparison of technologies used in the rat management farmer participatory research in Svay Teap District, Svay Rieng Province_

Baiting Traps Hunts Trap-Barrier system

Initial Cost Medium Medium Low High

Labour requirements Low Low Medium High
Effectiveness Low Medium High Low
- --- ---
Durability Days Years Unknown A season
Reliability No Yes Yes No
-
Economies of scale No No No Yes (cost! t depends on
field size and shape)
Provides information about No (7) Yes Yes Yes
rat populations
Requires continuous No 'IYes No Yes
monitoring
Environmental hazard Yes No No Yes
Gender impact Primarily men's work Primarily men 's work - Men, women and children Men and women
Health impact Unknown No No No
Compatibility with farming
system
No (Livestock)
r es IYes No (Livestock)

Complexity
Adaptability
---
Medium
Farmers modified the bait
stations
ILow
Novel trap designs were
we~cePted ~
Low
Adapted to include dogs in
I the hunt __
High
Difficult to modify, but
possible use as fish fences
!

_l -=
=
III

:i:
III
III
!IQ
Relative disadvantage Some rat spp . are bait-shy; Traps do not target rats Non-target species are also High cost; high complexity; ID
3

-=
presence of cattle reduces which attack rice crops; caught (although some of need for continuous ID
effectiveness of pre-baiting theft these may be used as food monitoring; incompatible
e.g. frogs, snakes) with cattle in the farming :;
system ; theft (")
--- -- - - - - III

eN
en
CD
Popularity Individuals Ilndividuals BasiS for social interaction Initial interest, but this
dissipated l 3
C"
0
Cl.
Dj'
Ecologically-based Rodent Management
rice varieties to produce a short duration
crop before transplanting traditional
varieties in the normal wet season.
Normally, such early-maturing IR varieties
are grown in the dry season along receding
rivers or with irrigation. The EWS crop
provides food at a time of expected food
shortages Gust before the rains), and it
overcomes a persistent problem with a drop
in the germination rate of traditional seeds
stored for a long period (from one dry
season to the next) (Mak 1998). These
modern rice varieties are potentially of high
value because they are higher yielding than
traditional varieties and produce rice at a
time when the food supply is low. EWS
crops are scattered and at particular risk
from rat damage; perhaps because they are
out of season. If a TBS is to be used at all in
the Cambodian wet season, it would be with
these EWS crops. However, even here the
value of the crop does not appear to justify
investment in a rat fence. Exceptions are
situations where the potential yield is
relatively high and the risk of rat damage is
also very high. Under these circumstances,
the benefit-to-cost ratio of a TBS plus trap
crop was high in West Java, Indonesia
(Singleton et al. 1998). The value of a TBS
depends on: its cost and the number of
seasons it can be used; the value of the crop it
protects; the severity of the risk of rat
damage; and the social mechanisms
available for valuing externalities and
incorporating these into private decisions.
In Bot Slok, burrow digging killed more
rats than the TBS, but in the other two
villages farmers did not notice any burrows
in bunds. Due to its proximity to the forest,
scrub, and uncultivated land, Bot Slok
probably has greater rat problems.
370
Before the farmer participatory research
(FPR), farmers were mainly concerned with
reducing rat numbers during an outbreak.
After the FPR, farmers said they were
actively committed to preventing rat
population increases. Farmers also began to
see the usefulness of community rat control,
and the need to organise for more successful
rat management. Farmers made an effort to
participate in all aspects of the FPR, however
a facilitator was required to organise and
motivate farmers.
Farmers preferred working in small
groups, which were more efficient because
they could share ideas and consult each
other. Farmers working in small groups
expressed confidence in the process
than did those in groups.
Use of participatory methods meant that
we were able to find out more about the
nature of the rat problem in Svay Teap, and
the constraints to its solution, much more
quickly than if we had simply imposed pre-
defined experimental trials. The design of
appropriate procedures for rat management
in Cambodia cannot be separated from the
social and economic circumstances of the
problem-owners. This is best incorporated in
the research process through the active
participation of farmers.
ACKNOWLEDGMENTS
We thank the Commonwealth Scientific and
Industrial Research Organisation (CSIRO),
the Australian Centre for International
Agricultural Research (ACIAR) and IRRI for
information on rat management. We are
grateful to Mr. Tuy Samram, Mr. Numa
Shams and the rest of the CRS staff for
assistance in facilitation and funding of this

project. AusAID (Australian Agency for
International Development) funded the
ClAP contributions to the farmer
participatory research. We are indebted to
Dr. Luke Leung for his contribution to our
understanding of the rodent pests of
Cambodia. Finally, we appreciate the
participation of Ms. Kul Saram, Mr. Sons
Chheng, Mr. Pech Saron, Ms. Chao Sophal
and the hundreds of other rice farmers
involved in this research.
REFERENCES
Anon. 1996. Vitalsigns.AsiaweekMay31996,60.
ClAP (Cambodia-IRRI-Australia Project) 1998.
Annual research report 1997. Phnom Penh,
ClAP.
Cox, P.G., MacLeod, N.D. and Shulman, AD.
1997. Putting sustainability into practice in
agricultural research for development. In:
Stowell, P.A, Ison, ARL., Holloway, RJ.,
Jackson, S. and McRobb, 5., ed., Systems for
sustainability: people, organizations and
environments. London, Plenum Press, 33-38.
Jahn, G.c., Pheng, 5., Kiev, B. and Poll C. 1996.
Farmers' pest management and rice produc-
tion practices in Cambodian lowland rice.
Baseline Survey Report No. 6. Phnom Penh,
Cambodia-IRRI-Australia Project.
Rat Management in Cambodia
Jahn, G.c., Kiev, B., Pheng, S. and Pol, C.1997a.
Pest management in rice. In: Nesbitt, H.J., ed.,
Hice production in Cambodia. Manila, Inter-
national IDce Hesearch Institute, 83-91.
Jahn, G.c., Pheng, S" Kiev, B. and Poll C. 1997b.
Pest management practices of lowland rice
farmers in Cambodia. In: Heong, K.L. and
Escalada, M.M., ed., Pest management of rice
farmers in Asia. Manila, International IDee
Research Institute, 35-52.
Javier, E. 1997. Rice ecosystems and varieties. In:
Nesbitt, H.]., ed., Rice production in Cambo-
dia. Manila, International IDce Research Insti-
tute, 39-81.
Leung, L.K.P. 1998. A review of the management
of rodent pests in Cambodian lowland rice
fields. A consultancy report for Cambodia-
IRRl-Australia Project. Canberra, CSIRO
Wildlife and Ecology.
Mak, S. 1998. Rainfed lowland rice and agricul-
tural change in Cambodia. PhD Dissertation,
School of Agriculture and Rural Develop-
ment, University of Western Sydney,
Hawkesbury, Australia.
Singleton, G .R, Sudarmaji and Suriapermana, S.
1998. An experimental field study to evaluate
a trap-barrier system and fumigation for
controlling the rice field rat, Rattus argelltiv-
enter, in rice crops in West Java. Crop Protec-
tion, 17,55-64.
371
 18. Rodents in Agriculture in the
Lao PDR - a Problem with an
Unknown Future

John M. Schiller, Bounneuang Douang Boupha and Onechanh Bounnaphol

Abstract

Rice accounts for more than 80% of the cultivated land area in the Lao People 's
Democratic Republic (PDR). Rodent problems are mainly (but not exclusively)
associated with rice cultivation. Rainfed lowland rice accounts for about 70% of the
area and 76% of production. Rainfed upland rice accounts for about 21% and 14% of
the area and production, respectively. Smallholder producers in the main rainfed
lowland rice-growing areas of the Mekong River Valley generally do not rate rodents
as a major pest problem and consistently rank rodents very low among potential
production constraints. Conventional trapping techniques are generally capable of
giving satisfactory control. In the upland environment, however, smallholder
producers regard rodents as their most important pest, and the rodent problem
second only to weeds as the overall most important constraint to production . It is
also the production constraint over which they have least control. The severity of the
problem varies with locality and between seasons. Complete loss of upland rice
crops on a localised basis , with famine conditions resulting, is not unusual.
Conventional trapping techniques do not give adequate control in the uplands. Often
areas of lowland cultivation in the narrow valleys of the more mountainous regions
can also be devastated by the movement of rodents from adjacent upland areas.
Official policy is to actively discourage the use of rodenticides as a means of rodent
control in both the upland and lowland environments. However, there is increasing
use of uncontrolled imports of rodenticides, particularly in the upland environment.
Little is currently known about the species and ecology of rodents in the uplands of
the Lao PDR.

Keywords

Rodents, rice, Lao PDR

372

INTRODUCTION
GRICULTURE IS THE principle
A economic sector in the Lao People's
Democratic Republic (POR),
accounting for about 52% of real output.
About 80% of the population live in rural
areas and are engaged in agriculture. Rice is
the most important crop, contributing about
60% of total agricultural production and
accounting for about 80% of the cultivated
area. More than 90% of rice is grown under
rainfed conditions during the annual wet
season. The rainfed lowland ecosystem
accounts for about 70% of the area and 76% of
production; the rainfed upland environment
accounts for about 21% of the area and 14% of
production. Less than 10% of total rice
production is traded. National policy is aimed
at achieving a greater level of rice self-
sufficiency by the year 2000 by raising total
production by about 25% to approximately 2.1
million t.
Because of the national importance of rice,
most information available on rodents relates
to their effects on rice cultivation (Lao-IRRI
1992,1996; Khotsimuang et al. 1995), however
most of this information is of a secondary
nature. There have been few attempts to
understand and quantify the significance of
the problem, and little research done to
develop appropriate management strategies.
Singleton and Petch (1994) documented some
of the perceptions and data on the rodent
problem in the upland rice environment. Data
on rodents related to the rainfed lowland and
irrigated environments have been collected as
part of general surveys of pests and weed
infestation problems (Khotsimuang et a1. 1995;
Lao-IRRI 1996; Rapusas et a1. 1997).
Rodents in Agriculture in the Lao PDR
This chapter reviews the magnitude of
the rodent problem in the different agro-
ecosystems of Lao POR, and what we know
about the biology and management of
rodent pests in the upland environment
where they are an important constraint to
rice production. It also discusses future
issues that are likely to influence the
dynamics of rodent populations in Lao FOR.
RODENTS IN THE RAINFED
LOWLAND ENVIRONMENT
More than 70% of the area under rainfed
lowland rice cultivation is in provinces
adjacent to the Mekong River in the central
and southern agricultural regions of the
country. Rodents are present throughout
this area. However, a 1993 survey of farmer
perceptions of production constraints, in
nine districts of seven provinces in the
Mekong River Valley, indicated that in most
districts rodents were not regarded as a
significant production constraint
(Khotsimuang et a1. 1995). In a ranking of 11
potential production constraints, rodent
damage was ranked among the three least
important factors in seven of the nine
districts; in the remaining districts, rodents
were never ranked higher than seventh in
relative importance (Figure 1). In 1994, a
separate survey considered farmer
observations of various pests in areas of
lowland cultivation in Vientiane
Municipality and the provinces of
Savannakhet and Champassak in central
and southern Lao POR. In only one district
(Nasaythong) of Vi entia ne Municipality did
a significant number of farmers (300;,,) report
observations of rodents as a pest (Table 1).
373
Ecologically-based Rodent Management

Vientiane Province (Thourakhom District)

Credit
Crabs and snails
Flooding
Varieties
Rodents
Labor
Diseases
Soil fertility
Weeds
Drought
Insects

o 50 100 150 200 250 300

Relative importance
Savannakhet Province (Khanthabouly District)
Flooding
Credit
Rodents
Diseases
Varieties
Labor
Soil fertility
Weeds
Insects
Crabs and snails
Drought

o 50 100 150 200 250

Relative importance

Champassak Province (Phonthong District)

Flooding
Credit
Rodents
Varieties
Labor
Crabs and snails
Soil fertility
Diseases
Weeds
Insects
Drought

50 100 150 200 250

Relative importance

Figure 1.
Farmer perception of the relative importance of different production constraints in the rainfed lowland
environment in selected provinces of Lao People's Democratic Republic (Khotsimuang et al. 1995).

374

- -- -- - -----
 Table 1.
Pests that attack rice plants as reported by respondents in a survey of rainfed lowland environment (Raspusas et al. 1997).

Pests 1I Provinces (Percentage of farmers reporting)

Vientlane Savannakhet Champassak Total

Mun
11
iI
I'

ca "O.D "O.D ~1Ir- QI

E
...
.c
E >-
e::
ca
J:I
>-
e::
ca
::I
0
"O.D
e::
>-
ca
)( QI
..:
as
1/1
::I
.8
ca ca ~ :; .c 0 >- QI QI E >- e:: QI 1/1
E
z )( a. 0 :a:: .c e::
ca e:: ~ ca :; 8. e::
ca
as
c- o
E J:I ~ .c .!!! e:: .c 1/1
"O.D
e:: 0
"O.D
e:: ca "O.D
e:: ca
~
..:
1:> "C .8>- E >
~
E
..: ca
ca
....::I
"a
::I
0
ca
....::I
>-
ca
(1)
QI
)( z
III
ca ca
(1)
.!
>a.
e 0
!XI
ca
(1)
as
.c
(,)
ca
(1)
as
.c
(,)
=ca
a.
e::
ca
(1)

Leaf feeders
- - - --- . - -- -
Armyworm 0 0 0 3 0 7 0 0 0 8 2 0 0 0 0 20
Cutworm / worm 5 11 6 3 0 5 3 0 0 5 3 7 0 3 4 55
Caseworm 0 0 1 0 0 5 2 0 0 0 1 0 5 0 0 14
Leaffolder 9 6 4 7 0 5 1 0 0 13 3 17 3 0 0 68
--_. _.- - _._----
:::u
Whorl maggot 0 0 0 0 0 0 0 3 0 0 0 0 1 0 0 4 0

-
Cl.
ID
Rice skipper 3 0 0 1 0 0 0 0 0 1 0 0 0 0 0 5 ::::I
I II
Grasshopper/ locust 20 23 19 12 0 23 16 4 1 8 0 3 0 0 0 129 ::::I

Thrips

Brown planthopper
. __ ._- 0
0
0
3
5
3
1
2
0
0
4
2
10
8
1
3
0
0
3
3
9
0 19
0 3
0
6
2
10
14
52
59
.
>
IJQ
C;.
=
Whitebacked planthopper 0 6 0 0 0 0 0 0 0 0 0 0 2 1 7 16 .=
;:;
ID
Green leafhopper + zigzag leafhopper

Stem feeders

Stemborers
- - ----_.
0

0
2

7 25
2

4
1 0

0
1

23 15
0

25
0 0

8 30
0 0

2
0

6
3

3
0

3
0

2
9

173
-
::::I

::r
ID
r-
III
II....- 0
w Gall midge 0 10 9 15 0 0 0 4 8 22 3 1 5 0 20 97 ."
..... - - .---- Cl
U'I ---
:::u
w Table 1. (Cont'd) I ,.,
~ Pests that attack rice plants as reported by respondents in a survey of rainfed lowland environment (Raspusas et al. 1997). ~
Cl
!IQ
Pests il Provinces ( Percentage of farmers reporting) I g.
Vlentiane ~I Savannakhet Champassak Total I~er
Mun ! ::
ID
GI Cl..
~
.I::
"QI)
c I "QI)
c 11 11 I1 E ::ell
_
... ~ I ~"QI) ~ ,:t/.:::II _

E
~

~
i... 11.1g::
"S
C
Cl Q) GI
<
E
GI
C Q)
~.8
en E
~
:::11
Z < Q. Cl :IC: .I:: ~ ~u ~ "S 8. ~:i Cl -
i!'
~
~
"tI
i!'
~
i i!'
Q)
ien oS i~:i
CI"C
.8 E
~
fa
... E
~
31
,:t/.
IQ
C
==
III
:::11 :::11 :::11 ~ _ ~ ~ .!! ... Cl ~ .I:: ~ .I:: ~ ~ :::11
....J 0 ....J Cl) >< z Cl) > Q. ID Cl) (,) Cl) (,) Q. Cl) III

Grain feeders 3
!IQ

Rice bugs

Mole crickets
1

0
1

3
14

3
4

1
0

0
28

0 4
7 24

2
2

0
8

0
1

0
6

0
0

0
2

0
2

0
100

13
-~

Ants/ termites 0 0 0 0 0 2 0 0 1 0 0 0 0 0 0 3

Other pests

Rats 1 0 3 4 0 14 9 0 0 0 0 0 1 0 0 32

Crabs 0 0 3 0 il 0 6 6 0 5 20 8 1 0 2 8 59

Birds 0 0 1 1 0 1 3 0 0 0 0 1 0 0 0 7

Diseases (blast, sheath bl ight, 0 0 0 0 0 0 0 0 0 0 2 1 0 0 1 4

false smut)

However, the level of losses was not
indicated and most farmers reported that
they were able to manage the problem
(Rapusas et al. 1997). The main rodent
species encolmtered in the lowland
enviromnent have been identified as Rattus
argentiventer, Rattus norvegicus, l\attus
exuians and Bandicota indica.
In areas of raWed lowland rice
cultivation in the narrow valleys of much of
northern Lao PDR, rodents can cause a
significant, but as yet unquantified, level of
damage. These areas are usually adjacent to
Figure 2.
Lowland rice field adjacent to forest regrowth and areas of upland cropping which harbour endemic rodent
populations.
Rodents in Agriculture in the Lao PDR
forest regrowth and areas of upland
cropping, which harbour high endemjc
rodent populations (Figure 2). As the
lowland rice crops ripen, the rodents move
from these forested areas into the rice crops.
Early maturing rice crops in this
environment are particularly targeted. The
rodent species responsible for this damage
are believed to be the same as those
responsible for losses in upland rice crops,
however the species have yet to be
identified.
377
Ecologically-based Rodent Management

RODENTS IN THE IRRIGATED Provinces which have reported

ENVIRONMENT significant rodent problems in the uplands
are shown in Figure 6. These problems have
A 1995 survey of farmer perceptions of
been further highlighted in recent area-
production constraints under irrigated
specific surveys. McLaren (1996) reported
conditions in nine administrative districts
rodents to be the most consistently reported
extending over five provinces and Vientiane
pest among production constraints (Table 2)
Municipality, in central and southern Lao
in parts of Luang Namtha Province
PDR, found that rodents were not listed as a
adjoining the Chinese border. (Birds and
significant production constraint in any of
wild pigs were also often a cause of damage
the districts surveyed (Figure 3). The highest
to upland crops.) In the 1998 wet season,
ranking in three districts was seventh in the
parts of Houaphanh Province, on the border
list of 11 constraints; in two districts rodents
with Vietnam, reported losses exceeding
were ranked as the least important
30% of the upland rice crop due to a
production problem (Lao-IRIU 1996). As in
'population explosion' of rodents and
the rainfed lowland environment,
subsequent damage to the main wet season
smallholder farmers generally reported that
crop; more than 1,000 ha was totally
they could adequately manage the rodent
destroyed (M. Phouthavong, 14 September
problem encountered under irrigated
1998, pers. comm.). Agricultural officials
conditions. As in the rainfed lowland
from this province report that there have
environment, small areas of dry season
been four such outbreaks over the period
irrigated rice crops in the narrow in
1968 to 1998; all were apparently associated
northern Lao PDR can be targeted by
with the wet season flowering and fruiting
rodents from adjacent upland areas; losses
of particular of bamboo, as discussed
under these conditions can be substantial.
below. Unusually dry years are also
associated with higher levels of rodent
RODENTS IN THE UPLAND damage in upland areas.
ENVIRONMENT The damage by rodents to upland crops is
In the early 1990s, surveys undertaken of not confined to rice, however reports of such
farmer perceptions of production constraints damage most often relate to rice. This is
to upland rice cultivation in several because of its significance in the uplands,
provinces of the northern agricultural region where it accounts for more than 75% of
of the country indicated that rodents were cultivated area, and more than 80% of calorie
regarded as the most important pest in the intake of people living in this environment.
uplands (Lao-IRRI 1992). Smallholder The rodent problem is regarded by upland
upland rice farmers rated rodents as being farmers as the production constraint over
second only to weeds as the overall most which they have least control.
important constraint to upland rice
cultivation (Figure 4). The nature of the
upland environment lends itself to high
endemic rodent populations (Figure 5).

378
 Rodents in Agriculture in the Lao PDR

Vientiane Province (Phonehong District)

Rodents
Crabs and snails
Credit
Diseases
Varieties
Drainage system
Soil fertility
Lack of labour
Weeds
Insects
Lack of water

o 50 100 150 200 250 300

Relative importance
Savannakhet Province (Saybouly District)

Rodents
Drainage system
Varieties
Credit
Diseases
Lack of labour
Lack of water
Crabs and snails
Soil fertility
Weeds
Insects

o 50 100 150 200 250 300

Relative importance
Champassak Province (Champassak District)

Crabs and snails

Varieties
Soil fertility
Drainage system
Rodents
Diseases
Lack of water
Weeds
Insects
Lack of labour
Credit

o 50 100 150 200 250 300

Relative importance

Figure 3.
Farmer perception of the relative importance of different production constraints in the irrigated
environment in selected provinces of Lao People's Democratic Republic (La(}-IRRI1996).

379
Ecologically-based Rodent Management

Weeds
Rodents
Insufficient rainfall
Land availability
Insects
Labor
Soil fertility
Erosion
Domestic animals
Wild animals
Disease
Suitable varieties

o 10 20 30 40 50 60 70 80 90
Respondents (%)

Figure 4.
Farmer perception of major constraints to upland rice production (Lao-IRRI1992).

Figure 5.
Typical appearance of the upland environment, the nature of which lends itself to high endemic rodent
populations.

380
 Rodents in Agriculture in the Lao PDR

Table 2.
Main production constraints to rice production in villages of Luang Namtha District of Luang Namtha
Province (McLaren 1996).

Village/Ethnicity Total Land area (ha) Main production constraints

families
Lowland Upland

Namthung/Thai Leu 175 135 0 Low prices

Lack of knowledge
Lack of seed
Ban Phoung/Thai Dam 166 109 5 Low prices
Rodents
Lack of land
Namngen/mainly Thai Dam 383 192 15 Lack of land
Pests (snails, crabs,birds, rodents)
Low prices
Houaidam/ Khamu Ou 60 6 29 Low fertility
Damage by livestock
Low prices
Nateuil/ Khamu Ou 83 2 30 Weeds
Rodents
Low fertility
Tintok/ Khamu-Hmong 37 1 32 Birds, rodents
Weeds
Low fertility
Hat-Ngao/Hmong 87 13 82 Lack of land
Weeds
Diseases
Lakkhammai / lkaw 37 4 32 Lack of land
Lack of seed
Damage by pigs
Namkhon/ Lao Huay 28 11 20 Damage by pigs
Rodents
Birds
Namke/ Lao Huay 21 6 15 Weeds
Diseases
Birds, rodents
Xuanya/ Lao Huay 14 8 2.5 Flooding
Damage by livestock
Low prices
Total 481 262.5

381
Ecologically-based Rodent Management

Vietnam

Thailand

Rainfed upland rice Savannakhei

as a proportion of
total rice production

D <20%

--
D 20--40%

40-60%

60-80%

>80%

Figure 6.
Provinces of the Lao People's Democratic Republic which have reported significant rodent problems in the
upland environment (.)

382

CHARACTERISTICS OF THE UPLAND
RODENT PROBLEM
Although farmers in the uplands of Lao PDR
have a good knowledge of the rodent
problem in local terms, little has been done
to characterise and understand the problem
in a scientific sense. Grain losses due to
rodent damage in the uplands are a chronic
annual problem. Annual losses have not
been properly quantified but are believed to
account for at least 15% of the rice harvest.
As almost all upland farmers have an annual
rice shortage, the losses due to rodents can
significantly aggravate the chronically poor
nutritional status that is a feature of most
upland households.
At irregular intervals, conditions that
favour massive eruptions of the upland
rodent population can result in localised
losses in excess of 50% of the rice crop.
Occasionally farmers lose 100% of their crop.
For example, in 1991, several villages in the
northern province of Luang Prabang
reported the total loss of their wet season
rice crops (Singleton and Petch 1994). In
almost all instances, these population
eruptions are associated with the mid-wet
season flowering and fruiting of certain
species of bamboo. The bamboo is found in
areas of regenerated forest where it can be a
sub-climax vegetation form following a
period of 'slash-and-burn', generally rice-
based, shifting cultivation. Although the
species of bamboo associa ted with these
rodent population explosions are known in
local terms, a botanical characterisation has
yet to be made. Flowering and fruiting of the
bamboo is reported to take place in July
(following the opening wet season rains in
May); fruiting is followed by the death of the
Rodents in Agriculture in the Lao PDR
bamboo. Increased rodent activity is soon
observed and rapidly builds up during
August when the bamboo fruit provides a
regular food supply for the rodent
population. Following the fruiting of the
bamboo, the rodents then move into upland
rice crops and other upland annual crops.
The harvesting of the earliest maturing
upland rice varieties commences in early
September. This usually corresponds to a
period of heavy wet season rain and yields
of these early varieties can be lower than for
later maturing varieties. The early varieties
are usually grown to help meet immediate
household rice needs in situations where a
chronic rice deficit is usual in the period
prior to the harvest of the main wet season
rice crop. The damage to rice crops does not
only occur during the period approaching
harvest. Damage is sometimes reported
prior to heading, when rodents move from
nearby forested areas. In particularly serious
cases, a rice crop can be destroyed within a
couple of nights.
In the uplands, the damage to crops is
associated with two main types of rodents,
both of which are well known to upland
farmers but which have yet to be classified.
The most serious damage is caused by large
numbers of mice-sized rodents weighing
between 65-80 g. However, the movements
of these 'nuu khii' is usually in association
with a smaller number of larger-sized
rodents, possibly of another species.
Columns of 'nuu khii' are reported to move
into crops, following individual or a small
numbers of the larger animals. Singleton and
Petch (1994) have suggested that the larger
rodent might be the rice-field rat, Rnttus
argentiventer, and the smaller one possibly a
species of Mus (M. caroli or M. cervicolor).
383
Ecologically-based Rodent Management
This needs confirmation with taxonomic
studies.
The rodent problem in the uplands of Lao
PDR appears to be shared in the uplands of
the neighbouring countries of Vietnam,
Myanmar and Thailand. Only in Thailand is
there some knowledge of the biology of the
rodent species living in this system (see
Boonsong et al., Chapter 16).
RODENT CONTROL MEASURES IN THE
LAo PDR
Lowland rainfed and irrigated
environments
Trapping techniques developed by Lao
farmers, in combination with a number of
cultural practices, are usually employed in
most areas of lowland rice cultivation
(lowland rainfed and irrigated). These
techniques are generally regarded as being
effective and are carried out throughout the
year. Often the trapped animals are eaten or
sold in local markets. There is relatively little
use of rodenticides, although rodenticides
are often available in local markets. If
rodenticides are purchased and used, it is
usually to protect early maturing rice
varieties, which can be targeted by both
birds and rodents. A 1994 survey of pesticide
use by lowland rice farmers in 11 provinces
in the north, central and southern regions of
Lao PDR indicated that less than 2% of
respondents were using rodenticides at that
time (Rapusas et aL 1997). The only
rodenticide reported as being used in the
lowlands was zinc phosphide. Until about
1990, the main source of rodenticides was
Russia. More recently, the Japanese
government has provided rodenticides on a
regular annual basis under that country's
384
development assistance program, based on
requests from the Lao government. It is
generally recognised that the level of
rodenticide use has recently increased from
the levels indicated by the 1994 survey.
Much of this increase has taken place in
lowland areas in the north of the country,
using rodenticides originating from China;
the active ingredients of these rodenticides
have not yet been determined.
Upland environment
As in the lowland environment, upland
farmers have developed a number of control
techniques based on the use of various types
of single-capture traps, snares and pitfall
traps to help control the rodent population.
Occasionally, trained dogs are also used to
hunt and kill rodents. These techniques are
used all year round but with increased
intensity as the upland rice crops approach
maturity. The effectiveness of the techniques
is probably limited, as they usually target
only the rice crop, whereas the rodents often
move from adjoining forested and fallow
areas into the rice crop. In those years when
upland rodent populations erupt,
conventional trapping and catching
techniques have little impact. Following
significant crop losses due to rodents in 1991
in the northern Lao province of Luang
Prabang (Singleton and Petch 1994), rodent
control committees were established at the
provincial and district levels. These
committees had several functions: to
estimate rodent damage, to conduct
eradication campaigns, to encourage and
train farmers in the use of traditional and
new control methods, to evaluate new
methods of control, and to develop incentive
and awareness campaigns. During the 1991

outbreak, a bounty on rodent tails was
offered as an incentive for both reducing the
rodent population, and to provide financial
support for households who lost part of their
rice crop. The effectiveness of this type of
program has been difficult to assess as it has
been implemented when rodent populations
are already at a high level, and when much
of the damage has already been done. In
Southeast Asia, the implementation of
bounty systems is a typical response by
provincial governments to severe rodent
problems. Lao POR, however, cannot learn
much from its neighbours regarding the
impact of such bounty schemes, as their
effectiveness has not been assessed (see
Singleton et al., Chapter 8).
The chronic nature of the rodent problem
in the uplands provides a ready market for
rodenticides. Despite an official government
policy discouraging their use, the use of
rodenticides in the uplands is increasing.
Most are originating from China, from
which there is easy access to the northern
provinces of Lao POR for an increasing
number of agricultural inputs. No statistical
information is yet available on rodenticide
use in the uplands, however their increasing
availability in local markets reflects an
increasing demand. The active ingredients in
these rodenticides is unknown. Apart from
the danger to non-target animals (the cat
population has been eliminated from most
northern provinces, due to their
consumption of poisoned rodents), there is
increasing concern about their impact on
human health, with increasing reports of
human fatalities due to accidental poisoning.
Rodents in Agriculture in the Lao PDR
FUTURE MANAGEMENT OF THE RODENT
PROBLEM IN LAo PDR
A number of factors will influence both the
direction of future research on the rodent
problem in Lao POR and the development of
effective management control strategies;
significant among these are the following.
Changes in upland agriculture
The current system of slash-and-burn
shifting cultivation that predominates in the
uplands favours the maintenance of high
endemic rodent populations and makes it
difficult to implement effective control not
based on the use of rodenticides. National
policy for the future development of the
uplands of Lao POR is based on a move from
these cultivation practices to more
ecologically sustainable agro-forestry and
forestry-based systems. This change may
facilitate easier management of the endemic
rodent populations in the uplands.
However, it is also recognised that
significant changes in agricultural practices
may bring changes in rodent behaviour.
Monitoring of the rodent population ecology
will therefore be an important component of
monitoring the impact of the proposed
changes to traditional agricultural practices.
A reduction in the area cropped to rice in
the uplands is also proposed, from about
164,000 ha in 1997 to 58,000 ha in the year
2000. It is now recognised that, in absence of
demonstrated alternative agricultural
technologies for the uplands, a reduction of
this magnitude will be difficult to achieve.
Nevertheless, a gradual but continued
reduction in the area under upland rice
cultivation can be expected. The cessation of
'shifting cultivation' practices, combined
385
Ecologically-based Rodent Management
with the allocation and management of the
uplands by villages and individual
households in these villages, aim to provide
a basis for more responsible and sustainable
forms of land use in the uplands. Land
allocation in key northern provinces under
the new national guidelines commenced in
1996. The official policy is for approximately
100,000 of the 300,000 households believed
to be dependent on shifting cultivation to be
allocated land for the adoption of more
sustainable forms of land use by the year
2000. The community-based agricultural
systems that are aimed to be the cornerstone
of this development may provide a basis for
better management of rodent populations in
the uplands. Future rodent research should
focus on these systems.
Pesticide registration and
distribution control
Almost all pesticides used in Lao PDR are
imported. Only small quantities of botanical
pesticides are produced locally. Authority
for the import of pesticides is with the
Department of Agriculture and Extension,
within the Ministry of Agriculture and
Forestry. In the late 1980s and early 1990s,
pesticide imports were often in the form of
development assistance. As discussed earler,
some rodenticides (mainly zinc phosphide)
are still being supplied by the Japanese
government under its development
assistance program. Some of the pesticides
observed in local Lao markets are
theoretically not marketed in developing
countries by producers. The open borders
with Thailand, Vietnam and China mean
that not all pesticides sold throughout Lao
PDR are approved imports (Rapusas et al.
1997). Potent rodenticides originating from
386
China have become readily available in
markets of provinces in the north of Lao
PDR without appropriate import approval.
Attempts are also being made to introduce
other types of rodenticides that are regarded
as potentially pathogenic and dangerous to
humans and their livestock into the Lao
market. A proper evaluation system capable
of alerting Lao officials to the potential
dangers of approving the import of certain
types of pesticides has yet to be put in place.
It can be expected that attempts will
continue to be made to seek approval for the
import and sale of pesticides that are
currently banned in more developed
countries. Farmer education on the potential
dangers from the abuse of pesticides,
including rodenticides, is also needed as a
matter of urgency.
Extension services and agricultural
technologies for the uplands
The extension services of the Lao PDR are in
an early stage of development. In the upland
environment, few technologies have been
demonstrated to be capable of meeting the
national objective of ecological sustainability
while also meeting the food and income
needs of upland farmers. The
interdependence of the development of the
extension services and the availability of
appropriate agricultural technologies for the
uplands is recognised and being reflected in
research planning for the uplands. The
development of community-based rodent
management programs will need to be
undertaken within the context of an effective
extension service.

PRIORITIES FOR FUTURE RODENT
RESEARCH IN THE LAo PDR
Although it is recognised that smallholders
in both the upland and lowland
environments have a good knowledge and
understanding of the rodent problem in
their respective environments, little has been
done to characterise the problem in a
systematic way as a basis for the
development of better rodent management
strategies. There is little in the way of rodent
expertise in the research and extension fields
in the Lao PDR. In addition to meeting the
need for training of Lao scientists and
extension workers, a number of areas of
short and medium term research priority
have been identified:
~ Identification of the different rodent
species in all production environments.
~ Quantification of economic losses due to
rodents for the major food crops.
~ Characterisation of existing control
measures and assessment of their
effectiveness.
~ Studies of the population dynamics and
habitat use of rodents in the upland
environment.
~ Characterisation of current use and
possible abuse of rodenticides, and
formulation of recommendations for better
control over their registration, import,
distribution and use.
~ Development of community-level rodent
management strategies for the upland
environment.
Rodents in Agriculture in the Lao PDR
Research initiatives in several of the
above areas are scheduled to be initiated in
1999.
ACKNOWLEDGMENT
Some of the data referred to in this report
was collected with support provided
through the Swiss Agency for Development
and Cooperation to the Lao National Rice
Program through the International Rice
Research Institute.
REFERENCES
Khotsimuang, S., Schiller, J.M. and Moody, K.
1995. Weeds as a production constraint in the
rainfed lowland rice environment of the Lao
POR. Proceedings of 15th Asian-Pacific
Weed Science Society Conference, Tsukuba,
Japan, 444-454.
Lao-IRRI (International Rice Research Institute)
1992. National rice research program 1991.
Annual Technical Report, 156p.
Lao-IRRI 1996. National rice research program
1996. Annual Technical Report, 296p.
McLaren, A. 1996. Integrated rural development
project, Luang Namtha Province ALA/LAO
93/94: report on a production systems study,
Luang Namtha Province, 107p.
Rapusas, H.R., Schiller, J.M. and Sengsoulivong,
V. 1997. Pest management practices of rice
farmers in the rainfed lowland environment
of the Lao POR. In: Heong, K.L. and Escalada,
M.M., ed., Pest management of rice farmers in
Asia. Manila, Philippines International Rice
Research Institute, 99-114.
Singleton, C.R. and Petch, O.A. 1994. A review of
the biology and management of rodent pests
in Southeast Asia. ACIAR (Australian Centre
for International Agricultural Research)
Technical Report No. 30, 65p.
387
Herwig Leirs

Abstract

The eruptive nature of African rodent populations has stimulated the development of
several models, mainly to explain and forecast outbreaks. Regression models , built
on the observed relation between rainfall events and rodent numbers, are quite
reliabl e but do not provide biological explanations or allow simulations . Conceptual
models combine various pieces of ecological information into an integrated
rep resentation of the species' population ecology, but they lack a numerical
component; this makes it difficult to evaluate the relative importance of the different
elements and the models cannot be used in practice. Writing these models
mathematically allows for simulations , but realistic results are obtained only if
environmental stochasticity can be included . Evaluation of a stochastic simulation
model for Mastomys nataiensis in Tanzania shows that this stochasticity is also the
major drawback of this model type for use in forec asting. In conclusion , the
regressive models are the best ones for prediction while stochastic population
dynamics models are more appropriate for simulation. Both model types have their
place in ecologically based rodent management.

Keywords

Afri ca, rod ents, models, forecasting, simulation

388

INTRODUCTION
N AFRICA, RODENTS are serious
I pests in fields and in human dwellings,
both in the countryside and in cities.
They cause serious damage to crops before
and after harvest, damage installations and
are reservoirs or vectors of serious infectious
diseases 1988a). Rodent damage in
the field can occur during all crop stages, but
often it is most serious at planting time and
just before harvesting. African field rodent
populations are characterised by irregular
population explosions which often occur
over large geographical areas (Fiedler 1988a;
Leirs et al. 1996). Such outbreaks are not
cyclic and therefore happen mostly
unexpectedly. Damage during such
outbreaks is enormous and may significantly
worsen the already bad food security
conditions in the affected countries. This is a
direct disaster for the subsistence farmers
involved, but also has national and
sometimes even international political
consequences. Panic-stricken authorities
may quickly initiate control operations, often
with very poor results and always too late.
There is thus a very clear need to predict and,
if possible, prevent such outbreaks (Taylor
1968; Shuyler 1977; Mwanjabe 1990;
Mwanjabe and Sirima 1993).
Forecasting changes in irregularly
fluctuating rodent populations relies on
observed correlations with other
environmental parameters and an
understanding of which mechanisms are
active in these populations. That information
can be formalised in some way to a set of
rules which is called a model (Stenseth 1977).
The word model, however, covers a variety
Populations of African Rodents
of such sets of rules which differ in the
nature of input and output, the complexity
of the rules that are incorporated and the
language that is being used. The last can
vary from vague concephlal schemes to
detailed mathematical expressions.
In this chapter, I will present an overview
of the different types of models that are used
to predict or simulate in rodent
populations. Although much of the
following may apply to population models
in general, there will be a strong focus on
models that were developed for African
rodent populations. I will use examples from
our own work in Tanzania to illustrate the
possible practical uses of different model
types.
MODEL TYPES
As with all science, the current knowledge
about rodent populations is based on a large
collection of studies which describe or
investigate different of populations
at various levels of detail. For African
rodents, that source material is rather
limited. '\1ultimammate rats, species
belonging to the genus Mastomys, are the
most important rodent pests in Africa, and
undoubtedly the best studied group of
African rodents. Nevertheless, Leirs (1995)
found only 115 references, including
unpu bUshed reports, with a direct rela tion to
Mastomys population ecology. Even so, an
overview is difficult because many of the
original studies are particular and spread
out over a wide geographical and ecological
scale. There are only few examples where
researchers have collected information on
different aspects from the same African
rodent population over several years
389
Ecologically-based Rodent Management
(Hubert and Adam 1985; Sicard et al. 1994;
Leirs et al. 1996). In order to develop a more
holistic view of the population biology of
these animals, there is a need to formalise
that combined knowledge in models.
Models are then simplified
representations of reality - known to be
different from reality, but claiming to be
reasonably good at simulating some
particular aspects of interest. The
representation is formalised as a set of rules
which can be simple or complex, depending
on the model. The common line of thinking
behind all models that I will discuss here is
that, given a certain population state and
information about the environment, they
predict the state of the population at a later
time. How useful such predictions are, how
long beforehand they can be made and how
accurate they are is different for different
models. Based on the kind of data used to
formulate the model, the language it uses for
expressing its set of rules and the kind of
information needed to feed into the model or
expected to come out of it, one can
differentiate between several types of
models.
Regression models
Regression models are based on observed
co-occurrences between certain
environmental conditions and changes in
the rodent populations. Properties of such
models in a rodent management context
have been discussed before (Stenseth 1977).
They do not build on any biological concepts
about the processes that affect rodent
population dynamics, except a belief that
unusual rodent numbers must be related to
unusual environmental events. Often, there
is a traditional local common knowledge of
390
such relationships, but rarely is this
knowledge explicit enough to be of practical
use. The relation between outbreaks of
African rodents and rainfall had been
suggested for many years (see, for example,
Harris 1937) but only in the 1980s was it
pointed out that many rodent outbreaks
were preceded by abundant rainfall at the
end of a dry spell of several years. With that
information, the 1986-87 outbreak in Sudan
was successfully predicted (Fiedler 1988b).
While the recognition of the relation
between high rodent numbers and the end of
a dry period basically is a regression model,
it was not expressed mathematically. That
was done by Leirs et al. (1996) who
constructed a logistic regression curve to
show the relation between rodent outbreaks
in East Africa and rainfall during the early
months of the rainy season (Figure 1). Such a
mathematical approach is superior to a
purely verbal description because it presents
the observed relation in a less subjective way
and allows an assessment of the associated
errors by calculating a probability level.
Both the Fiedler and the Leirs models
were based on records of outbreaks, often
collated from the grey literature, and then
relating them to a plausible environmental
factor, rainfall. In both cases, there was some
biological explanation to support the model.
Fiedler (1988b) hypothetised that during dry
years, vegetation growth would be limited
and rodent populations would decrease, but
so would also predators and competitors.
Upon the return of the rains, an abundant
vegetation regrowth would occur and
rodent populations would react much faster
to that than predators or large herbivores,
allowing an uncontrolled explosion of
rodent populations.
 Populations of African Rodents

77
. ...
75 566862 7151
'--~.'--------------'--
63

49 _
1------_ 65 53 585750
(I
o /If; _ _--___Gi. . ._ . - - .

100 200 300 400 500 600

December-January rainfall (mm)

Figure 1.
Logistic regression curve of rodent outbreaks probability (y axis) In Tanzania between 1.947 and 1.977 on
rainfall early In the wet season (in December-January) in Tabora, central Tanzania (x axis). Numbers on the
figure represent years with or without outbreaks (redrawn from Leirs et a!. 1.996).

A similar hypothesis was used, and
supported by data, to explain outbreaks of
insect populations in Africa Oanssen 1993).
The regression model used by Leirs et al.
(1996) for Mastol1l1lS l1alalcnsis was
biologically supported by data showing that
unusually abundant rainfall during the first
peak of the wet season initiated faster
maturation and early reproduction,
resulting in an additional generation (Leirs
et al. 1993). A proximate mechanism for the
effects of rainfall on reproduction, through
the presence of germinating grasses, was
documented (Linn 1991; Firquet et al. 1996).
~evertheless, it should be stressed that the
models themselves do not rely on biological
mechanisms. To some extent, this can be
considered an advantage since it means that
time-consuming biological studies are not
necessary to develop the model, and even if
the biological hypothesis turns out to be
wrong, the model would still retain its value.
However, the lack of biological background
in a model makes it impossible to judge its
generality. As Leirs et al. (1996) pointl'd out,
Fiedler's (1988b) model was developl'd for
the semi-arid region of Sudan and it d id not
explain earlier outbreaks in south-eastern
Africa. Yet, that does not change the fact that
the model works for the Sudan region,
neither does it prove that Fiedler's
explanation for his model would be invalid.
Conversely, the Leirs et al. (1996) regression
model can only be used for regions with a
bimodal rainy season and is therefore not
useful in, for example, the Sudan region.

391
Ecologically-based Rodent Management

Even though regression models can be populations of holarctic rodents (Krebs

accurate (i.e. they predict correctly), they are 1996).
not always very precise (Le. they do not give
Many papers report on particular aspects
a detailed prediction). Indeed, the model
of African rodent biology and several of
proposed by Leirs et al. (1996) provides a
them point towards the importance of
binary answer only: an outbreak is
specific factors in population dynamics.
predicted, yes or no, but there is no
Here, however, I will briefly discuss only the
information about the extent of the outbreak.
few papers which explicitly combined
In order to make models which could
several studies into a conceptual model.
predict the magnitude of an outbreak on, for
Taylor and Green (1972), in an unpublished
example, a scale from 1 to 5, one would need
report, stressed the importance of extended
more detailed information about past
rainy seasons. They hypothesised that in
outbreaks. Unfortunately, old reports rarely
normal years, cereals would be harvested at
give reliable information about an
the end of the rainy season and weeds would
outbreak's magnitude and if they do the
die off because of drought; rodents would
methodology used differs betw:en '
stop breeding because of food and cover
outbreaks since they were monitored by
limitation and their populations would
different people, sometimes in different
remain low. When rainfall at the end of the
areas and with different technical means.
breeding season is prolonged and heavy,
harvest is delayed and weeds start
Conceptual models regrowing. This increases food supply and
reproduction continues much longer,
Several authors have tried to organise
resulting in outbreaks of populations. By
their extensive knowledge of particular
contrast, in Tanzania, abnormally heavy
African rodent systems into a schematic
rainfall early in the season is the key factor
model, combining the different underlying
for populations of M. natalensis, according to
biological concepts that they observed in
Leirs et aL (1996). Unusual early rainfall
their studies. Based on the background work
induces faster maturation and the birth of an
that was carried out before, these conceptual
early generation which will reproduce
models focus on the autecology of the
during the normal breeding season; the
rodents only or use a community approach
presence of this extra generation increases
including competitors, predators and
the production of young so much that an
diseases as explicit factors. A common factor
outbreak follows. Leirs et al. (1996)
for all African rodent models is that climatic
presented their conceptual model in a
factors, mainly rainfall, always play a major
flowchart diagram (Figure 2). A more
role. This corresponds to some models .
complex model was developed for
developed for another eruptive species, the
populations of Mastomys erythroleucus and
house mouse (Mus domesticus), in Australia
Taterillus gracilis in Senegal (Hubert and
(see Pech et al., Chapter 4). There is much
Adam 1983).
less attention paid to intrinsic factors that are
thought to be important in fluctuating

392
 Populations of African Rodents

NO

YES high
densities
expected

damage at damage IS
planting time possible

NO

high
densities
expected

Figure 2.
Flowchart diagram redrawn from Leirs et al. (1996) as a conceptual model to predict rodent outbreaks in
Tanzania. The key factor is rainfall during the 'vuli' season, i.e. the first part ofthe rainy season, while rainfall
during the 'masika' season, the second half of the rainy season, is less important.

Aga in, abundant ra infall is the central predator d en siti es are low, whj ch in itself is a
fac tor and the resulting imp ro ved food result of a lon ger pe riod with low rodent
quantity and quality increase reproduction d ensities. High rodent densities are
and surviva l. An o utbreak occurs when subsequently reduced mainly by parasites

393
Ecologically-based Rodent Management

and d iseases, together with o ther factors
(Figure 3). One of the most elabora te
conceptual models, or rather se t of models,
was developed for A rvicnn tliis niloticus in
Wes t A frica (Sicard et aI., Cha pter 20, and
references therein) . Based on a very
ex tensive know ledge of the anima ls' bi ology
(with detailed information on physiology,
behaviour a nd population fluctuations),
they describe severa l models for different
habitats and conditi on s. Their work is,
h owever, an excellent example of h ow
complex conceptual models can become.

_ Predators
_ Rodents I

Figure 3.
Schematic conceptual model of population dynamics of two Sahelian rodents in Senegal, redrawn from
Hubert and Adam 1983.

394

Demographic models
Changes in population size are the net
result of natality and mortality (and, in open
populations, immigration and emigration).
If one can express these processes as
demographic rates, population size at time
t+ 1 can be calculated from population size at
time t. The problem is, of course, that these
demographic processes are complex and
affected by a multitude of factors. For
example, rodents may die because of
starvation, diseases or predation and the
chances of this happening are themselves a
function of other environmental factors, age
of the individual and the rodent population
density itself. It is exactly this complexity
that one attempts to structure into
conceptual models. Whereas most
conceptual models stop at recognising that
there is an effect of particular factors,
demographic models assign values to
demographic rates. These values are
preferably estimates based on field data but
if these are not available, theoretical
approximations can be derived from the
literature or they can be guessed by trial and
error (e.g. French 1975; Poulet 1985).
French (1975) developed an age-
structured model for Mastomys sp. where the
timing of reproduction was synchronised
with rainfall, and both natality and mortality
were age-specific. This model was used for
simulations to investigate the effects of
different rainfall regimes, competition and
predation. Unfortunately, the model was
published in a medical journal and remained
largely overlooked by biologists. It is worth
noting that the model was developed in an
attempt to understand the biology of
Mllstomys species in villages in West Africa,
Populations of African Rodents
but the demographic parameter values were
obtained from a published study on a
population in a national park in Uganda.
Later it was recognised that these animals
belonged to a different Mastomys species and
the model was updated with data from
Sierra Leone (French 1985).
Hubert et al. (1978) estimated death rates
and counted litter sizes in M. erythroleucus
and T. gracilis during and after the 1975-76
outbreak in Senegal. They constructed a
model with these demographic values and
included some age-structure data (animals
-,-
start breeding at an age of three months).
Number and age structure of animals in 1976
was chosen as a starting population and the
model was then run with the estimated
demographic parameters for 15 months
afterwards. The model performed well for
that period, but since it used observed
demographic rates rather than model them
as response rates, it could not be used for
actual predictions.
Poulet (1985) used data from an eight-
year study on Tllterillus pygargus in northern
Senegal. He developed a demographic
model, on the basis of the reproductive and
mortality parameters obtained during that
study period. He claimed that predation was
a major driving force in mortality, but it is
unclear whether and how that information
was integrated in the model. He reported a
linear relation between rainfall and li tter size
and used that to calculate natality for years
when no field data were available. Using the
model, he calculated the offspring produced
in different cohorts and compared that with
the observed numbers in order to assess
cohort-specific mortality. He also simulated
a number of different values for mortality
395
Ecologically-based Rodent Management
rates in order to see how that could affect a
population.
Surprisingly, none of these demographic
models was used to attempt simulations
beyond the period for which they were
formulated. The simulations that were
carried out, however, contributed to a better
understanding of the relative importance of
different processes in the population
dynamics of those species.
Leirs et al. (1997) developed a
demographic model that differed in several
respects to those considered above.
Statistical modelling of their 1986-89
capture-mark-recapture data from a
population of M. natalensis in Morogoro,
Tanzania, showed that monthly survival of
subadults and adults was affected not only
by rainfall in preceeding months but also by
density. The nature of these relations was
not the same for subadults and adults.
Moreover, maturation, the growth process
from subadults to adults, was also
dependent on rainfall and density.
Therefore, they estimated these parameters,
as well as natality, for different combin-
ations of rainfall and density Box 1) and
used this information as a set of rules to
quantify a simple population dynamics
model with three functional age groups
(juveniles, subadults and adults). Rainfall
and density were the only factors affecting
parameter values and they did so in a simple
non-linear way: below some rainfall or
density threshold, the parameter would
have one value, above that threshold, it
would have another value. Thus, the model
did not rely on empirical estimates for each
time step, but used rainfall and density as
state variables to determine the parameter
values. This modelling of the demographic
396
rates makes this model much more general
than the above ones by Hubert et al. (1978)
and Poulet (1985). The major advantage is
that one can use rainfall to incorporate
environmental stochasticity in the Morogoro
Mastomys model. This is particularly
important since it is obvious that rodents
live in a stochastic, not a deterministic,
environment. Leirs et aL (1997) primarily
investigated the properties of the dynamics
created by their model but they also
compared the predictions of their model
with actual observed values in an
independent data set.
THE MOROGORO MASTOMYS
DEMOGRAPHIC MODEL
We continue with the Morogoro Mastomys
demographic model (Leirs et aL 1997) to
discuss the problems that are associated
with the development of such a model and
its use in practice. As mentioned above, the
foundation must be a plausible concept of
how demography is affected by other
factors. This requires a sufficient knowledge
of the life history of the rodent species.
The basic question is to decide which
environmental factors are likely to have
major impacts on demographic processes
and how to include them in a model. Many
different factors may play a role in this
respect, but in order to make a model
workable the key factors should be
identified. In the case of M. natalensis, it was
clear that rainfall in preceding months was
the most important factor in the timing of the
reproductive season (e.g. Leirs et aL 1989;
Telford 1989 in Morogoro, and many others
elsewhere as reviewed in Leirs 1995).
 IB~x 1
I
J
I
For the calculation of the survival and maturation rates, we used data from a capture-mark-recapture study on a 1 ha grid in fallow land in
Morogoro, Tanzania . The data were collected in monthly capture sessions of three days each between March 1987 and February 1989 and
were analysed in MSSURVIV (Hines 1994) as follows. We designed several models in which survival , maturation and capture rates could vary
freely between months. were fixed during the whole period or were fixed during periods of months with similar rainfall and/or rodent density
characteristics. These models were tested statistically to verify how well they represented the real data. We used the Akaike Information
Criterion to select the model that gave the best representation of reality and at the same time used only a limited number of parameters. This
wa s the model in which capture probability varied between months, while maturation and survival rates were the same for months which had
similar properties with regard to density as well as precipitation (more information can be found in Leirs et al. 1997).

Based on this selected model. we estimated parameter values for survival of subadult and adult females and maturation probabilities for
subadult females in si x different categories of months. defined by a combination of rainfall and density properties (see table below). From
rem ova l trappings in the same area, we calculated the net reproductive rate per female for each of these categories by multiplying the mean
litter size with the mean proportion of pregnant individuals among adult females in such months. These values are listed below and were used
as parameter va lues in the population dynamics model that is described in the text of this chapter. We assumed for simplicity that male and
female survival rates were equal (even though this is biologically unlikely).

."
Cl
Density (N/ha) "Cl
C
I
> 1 50 < 150 > 150 < 150 > 150 < 150 I DI
!:!:
Cl
=
1 Subad ult su rvi val
FSubad ult maturation
1 Adu lt survival
0.629 0.02
110.000 0.015
0.583 0.066
0 .513 0.053
0.062 0 .037
0 .650 0 .078
0.682 0 .051
0 .683 0 .112
0 .513 0 .074 0 .602 0.092
0 .678 0 .059
0 .155 0 .111
0.505 0 .074
0 .595 0 .146
1 0.0
0.858 0 .099
_ _I
1

1
--
III
Cl

:I=-
::::!.
[ Net reproductive rate lr 1.29 - 5-: 32 ' : 0'.30 Il 6.64 I. 4.69 Ir 5 .82 1 n
DI
=
::11:1
Cl

-=
Cl..
(1)

~
..... I II
Ecologically-based Rodent Management
We also knew that growth and
maturation of subadult animals was
stimulated by rainfall, probably through an
effect of growing grass (Leirs et al. 1990,
1994; Firquet et al. 1996). Therefore, it was
clear that the model should include rainfall
as one of the factors which would have an
effect on demographic rates. As a second
factor, based on general ecological theory
rather than data, we selected density as an
integrator of many extrinsic and intrinsic
factors such as ULC,,;;;ao,;;;, predation, or social
suppression of maturation and
reproduction. The choice for these two
factors was confirmed to be appropriate by
comparing different statistical models where
one or both of the factors were included
(Leirs et al. 1997).
A second important question is how the
population is structured. It is indeed highly
likely that different parts of the population
are affected in a different way by factors like
predation or disease (e.g. Dickman et al.
1991). We also knew that reproductive
maturation was linked to size rather than
age and that individual growth was affected
by rainfall (Leirs et aL 1990). For our
Mastomys population, we used a very simple
structure with three age groups, each with
their own survival probabilities: juveniles
(young animals from birth until they enter
the trappable population); subadults (older
animals that have not yet reached sexual
maturity) and adults (animals that had
reached sexual maturity).
It should be stressed that construction of
a demographic model is only possible when
there is enough biological information about
the species. For African rodent populations
other than M. natalensis in Morogoro, the
necessary intensive and long-term studies,
398
both descriptive and experimental, are
available only for Mastomys, Taterillus and
Arvicanthis species in West Africa, not
surprisingly the same species for which
conceptual models were developed (see
above). To a lesser extent, it may be possible
to construct population models for A1astomys
spp. in southern Africa, where information
from different authors can be collated (e.g.
Sheppe 1972; Sheppe and Haas 1981;
Chidumayo 1984; Bronner et al. 1988). In
other places, or for other species, sporadic
information may be available, but needs to
complemented by generalisations from
other studies before a demographic model
can be designed.
Once the demographic model is
designed, it needs to be parameterised,
meaning that one has to make explicit the
rules of how the demographic rates change
with varying environmental conditions.
Indeed, it is not enough to know that, for
example, rainfall has an effect on sexual
maturation, but one must also know how
high the maturation rate is under given
rainfall conditions. Although an
experimental approach may provide
information about the nature of these
relations, the actual estimates can only come
from extensive long-term descriptive studies
(see also Krebs, Chapter 2).
We have not yet touched on any aspects
of use of space or community ecology which
may affect population ecology. Community
factors, like predation, competition or
disease can often be hidden in overall
density-dependent variations in
demography (e.g. Hansson and Henttonen
1988). Use of space is clearly very important
in population dynamics (Lidicker 1975) but
in the Mastomys Morogoro model, it has so

far been neglected by assuming that
immigration and emigration are
insignificant or, more likely, match each
other in the modelled population.
Testing the demographic model
The accuracy and reliability of a model in
practice is dependent on several possible
sources of error. The model design itself, and
the underlying biological concepts, are
fundamental and if these are faulty, good
results cannot be expected of a model. The
only solution is to continuously try to
improve our scientific understanding but
that problem is common to all biology.
Looking at sources of error which are more
typical of simulation/prediction models,
two major problems can be identified: the
quality of the model input data (the starting
situation) and the quality of the parameter
estimates.
In order to evaluate the model design, we
used the 1986-1989 data from Morogoro,
Tanzania-the same data that were used to
develop the model (Leirs et al. 1997), but also
the data collected between 1994 and 1998 on
the same site, and thus independent of the
data used for model development. This
second data set comprised 14,862 captures of
4,636 individual M. natalensis during 52;400
trapping nights on a 3 ha grid in a robust
capture-mark-recapture set-up with
monthly sessions of three consecutive
trapping nights. The study site consisted of a
mixture of fallow land and small maize
fields, providing typical Mastomys habitat.
Population size estimates were obtained in
program CAPTURE, using the jackknife
estimator M(h) which allows for individual
heterogeneity in capture probability (Otis
1978). For the model runs, we determined
Populations of African Rodents
the starting population composition
(number of adults and subadults) in
December of the study period and used the
actual rainfall that was observed during the
period of the model run. Each model was
run for a 12 months simulation. Predicted
population sizes were compared with actual
estimates.
The runs show predicted and observed
patterns which are largely parallel, although
the values are sometimes quite different
(Figure 4). A notable exception is the most
recent year, 1998, where population size
after December 1997 did not show the
predicted increase. The realistic results that
were obtained provide good support for the
underlying concept on which the model is
built, but it also raises questions as to what
happened in 1997-1998. It should be
mentioned that apart from the starting point
in November 1986, no outbreak peaks were
observed in the periods used to make
Figure 4.
We investigated the model's sensitivity to
changes in initial values by assuming a
December starting population with 0
juveniles, 0 adults and a number of
subadults varying between 20 and 300
animals of each sex. The model was run 10
times, each time with another starting
number of subadults. The monthly rainfall
values were chosen randomly among values
observed in that month between 1970-1997
but were identical for each run. Figure 5
shows the expected large variation in the
first months but a convergence of the
different curves after the first year.
Interestingly, however, the curves do not
simply run parallel as one intuitively would
expect.
399
Ecologically-based Rodent Management

900 900
November 1986 1987

ID ID
N
<iD
600 N
<iD
600
c::
0
~
'S
0..
0
Il. 300 300

o -~~~~--~~~~--~~~

N D J F M A M J J A S 0 J F M A M J J A SON

900 ,----------------------------, 900

December 1994 December 1995

ID ID
N
<(fj 600 N
<iD
600
c:: c::
0 0
'~
~
'S 'S
Cl. 0..
0 0
Il. 300 Il. 300

J F M A M J J A SON J F M A M J J A SON

900 ,-----------------------------, 900

December 1996 December 1997

gJ 600 gJ 600
<(fj
<iD
c:: c
<2 <2
Cil Cil
'S 'S
Cl. 0..
o 300 o

~kC
Il. Il.

o ,---L . .............. .:
J F M A M J J A SON D J F M A M J J A S 0
Month Month

Figure 4.
Model-predlcted (circles) and actual estimated population (squares) sizes for a 1 ha area. Twelve-month
slmulatlons from November or December with starting values equal to actual estimated values for these
months. Simulations were run with actual rainfall data (see text for further details).

400
 Populations of African Rodents

It is noteworthy that smaller starting parameter value, with a confidence interval

populations may, after some months, result
in higher population numbers than those
with a larger starting population.
The numerical values for the
demographic parameters in the models
(such like survival rates) are estimated from
empirical data. This means that they are
prone to statistical errors and rather than
really knowing the exact value of the
parameter, we obtain a probability
distribution (normal distribution) of the
around a point estimate. In order to verify
how sensitive the model is to the uncertainty
associated with the parameter estimates, we
ran the model 100 times with identical
starting conditions and monthly rainfall
values, but the demographic parameter
values were, at each step, chosen from a
normal distribution around the point
estimate.

900

750

600
ill
'00
c
0
.~

S
0.
0 450
0.
-0
(I)
U
'5
~
a..
300

150

6 12 18 24
Months since start

Figure 5.
Twenty-four month model runs with 10 different starting values between 20 and 300 subadults of each sex.
All parameter and rainfall values equal between runs (see text for full explanation).

401.
Ecologicallybased Rodent Management

The resulting curves (Figure 6) show Predictions under environmental

relatively minor variation during the first
months but this variation accumulates and
becomes very large after longer periods.
Thus, while quality of input data seems to
primarily affect the first months of a
prediction run, the quality of estimates is
more important for longer term predictions.
stochasticity
An inevitable problem for practical
predictions is that the rainfall values for
future months are unknown at the time of
prediction (unlike the above 'a posteriori
predictions'). A possible approach is to try
out the several possible values that rainfall
can have in the coming months and compare
the model results under these circumstances.
I tried this out as follows.

900,--- ---------------------------------~-- -------------------~

750

600
re
'00
c:
0
.~

'5
<i
0 450
<i
-0

.~
-0
~
Cl.
300

150

6 12 18 24
Months since start

Figure 6.
Twentyfour month model runs (n = 100) with demographic parameter values sampled normally from around
point estimate and its standard error. Sampling varies between each run, but all rainfall and starting values
are equal between runs.

402
 Populations of African Rodents

Starting values were chosen again from the AFRICAN RODENT MODELS AND
observed populations in December in our ECOLOGICALLY-BASED RODENT
study. Rainfall for the twelve months of each MANAGEMENT
run were 'bootstrapped' from rainfall data
Stenseth (1977) expressed a clear preference
obtained for that particular month in the
for regression models as predictive tools.
period 1970-1997; that is, for each month of
Although our knowledge of rodent biology
the run, and independently from the values
has increased since then and modern
for the other months, we chose a value at
computing facilities allow an easier use of
random from the 27 values that we had for
numerical models, his opinion may still
that month. The model was run 100 times,
hold. Regression models have several
each time with a different random seed,
important advantages. The ones that were
resulting in 100 different rainfall series of 12
developed for African rodent outbreaks are
months. We then compared the distribution
fairly simple with a single factor only,
of model outcome values for each month
rainfall, and a binary response, outbreak or
with the estimated population size for that
not (Fiedler 1988b; Leirs et al. 1996). The
month.
binary response can be associated with a
In most months, but not always, the
certain probability, but the model does not
observed values fall within the 95% range of
indicate how serious the outbreak is
predicted values (Figure 7). Unfortunately,
expected to be. This makes such models
these intervals are often so large that they do
intuitive to understand and easy to use.
not have any practical value at all. Moreover,
However, simplicity is not a typical
in order to predict real outbreaks, which are
characteristic of regression models (e.g. see
known to be related to unusual rainfall
Pech et al., Chapter 4) and although
events, it may be necessary to actually look
multivariate regression models may have a
at the model results which fall beyond the
considerably better fit to reality, they are
95% range. Comparing these wide
more difficult to understand intuitively. The
confidence intervals with the relatively
major advantage of regressive models is that
much better results that were obtained when
they attempt to give a fair representation of
the actual rainfall data were used (Figure 4),
observed reality without the need, or risk, of
shows how important the problem of the
having to explain the relative importance of
stochasticity is for practical use of this
different, sometimes hidden, mechanisms in
demographic model. In order to obtain more
the system. At the same time, this is a major
practical results, it will be necessary to use a
disadvantage since these models help little
set of rainfall data which resembles the
in understanding the underlying biology. As
coming rainfall events more closely. This
a consequence, regression models can only
requires the prediction of these values
be used within the observed data space since
themselves through separate climatological
any simulation beyond the limits of the
models.
empirical data would require the acceptance
of a plausible mechanism.

403
Ecologicallybased Rodent Management

~
600
[)AI~Arnb.~r 1994 600 December 1995
[g, 500 ~ 500
c
l!! l!!
'# 400 '# 400 T
o
C1>
"0
c
300 ~ 300 II -

j:: -~,!JJJlill
<ll
@ 200
'0
ID
:2 100

o
o 1 2 3 4 5 6 7 8 9 10 11 o 1 2 3 4 5 6 7 8 9 10 11
Month after prediction Month after prediction

600 600
December 1996 December 1997
[g, 500 [g, 500
c c T
~ ~
'# 400 ~ 400
o
C1>
o
C1>
"0 300 "0 300
c
T c
<ll <ll
@ 200 @ 200
'0 '0
ID ID
:2 100 :2 100

o
o 1 2 3 4 5 6 7 8 9 10 11 o 1 2 3 4 5 6 7 8 9 10 11
Month after prediction Month after prediction

Figure 7.
Actual estimated population sizes (circles) versus median and 95o/..range of model-predicted values
(diamonds) for each month In 100 runs. Twelve-month simulatlons starting in December with starting
values equal to actual estimated values for these months.

It is also impossible to use such models to
evaluate the effects of pest control
interactions, since that would require some
knowledge about how, rather than when,
poputations react to changing
environmental conditions.
The conceptual models, on the contrary,
focus exactly on how the environment or the
population itself may affect densities. They
attempt to explain a complex by
including a variety of different explanations.
This complexity makes the models
sometimes difficult to understand (see, for
example, 3 or some of the models by
Sicard et al., Chapter 20). A common
problem with all these models is that they are
imprecise in their definitions (e.g. they talk
about 'dry' or 'wet', 'low density' and 'high
density') and nearly always miss a
quantitative expression of the mechanisms
that they include. For example, Leirs et aL
(1996) recognise that there is some density-
dependent mechanism, but fail to mention at
what densities that should become active and

404

how large its effect would be. Likewise,
Hubert and Adam (1983) incorporated a
time-lagged response of predators in their
model, but were not able to say how much
time was needed for that time-lag and how
quickly effective densities of predators
would be reached. Taylor and Green (1972)
discussed the importance of field sanitation,
but did not give a quantitative relation
between the amount of weeds on a field and
the response in the rodent population. These
shortcomings make these conceptual models
difficult to use in practical applications,
although they may be very useful to
structure an otherwise complex set of
biological findings and theories.
The demographic models that were
mentioned above, do allow a quantitative
use. Since they are build on concepts about
mechanisms that play a role in population
dynamics, they can be used for simulations.
The main problem with these models is the
parameterisation of the demographic
processes. In order to obtain a reliable model,
and given that the underlying concepts are
biologically all right, the parameter estimates
need to be rather precise. Yet, even with large
data sets, some estimates may have large
confidence intervals (e.g. Leirs et a1. 1997).
Another problem is the determinism in these
models, meant here as the absence of
environmental stochasticity in the model
system. The model developed by French
(1975) simulated populations of rodents over
several years, but assumed that each year, the
same values for natality and survival would
apply. This is obviously not true, and
therefore such models cannot give realistic
predictions. Nevertheless, they may very
well simulate what the effect would be of
rodent management applications.
Populations of African Rodents
Including stochasticity in demographic
models makes them considerably more
complex and less easy to understand for non-
specialists. Also, the interpretation of the
results becomes more difficult since the
stochasticity will cause a different result
every time the model is run, even though all
other parameters are the same. This means
that instead of a single model outcome, a
probability distribution of possible results is
obtained. The practical use of stochastic
demographic models will therefore, maybe
paradoxically, depend on methods to reduce
the amount of stochasticity.
In ecologically-based rodent
management, models can play a big role as
forecasting tools, indicating when problems
can be expected and allowing people to
organise control campaigns in a timely
manner. Outbreaks that come as a surprise
are a major concern for agriculture in Africa
(e.g. see Makundi et al., Chapter 22). For all
practical purposes of forecasting, the
regression model seems to be the most
simple and reliable one. As part of a national
strategy, an early warning system for
Tanzania can be based on the rainfall data
that are being collected by the usual
meteorological network. Applying the
regression model to these data can be done at
a central laboratory, with simple computing
facilities. For more localised outbreaks,
rainfall should also be collected locally; data
interpretation, however, is less
straightforward because it will require a
comparison of the actual rainfall data with
the usual rainfall data for that place; this may
require input from the central laboratory.
Models can be very helpful to structure
and integrate our knowledge about rodent
populations. Conceptual models are useful
405
- - - - - - - - - _ ...._ - _ . _ - - _ . _ - - -
Ecologicallybased Rodent Management
to get a basic understanding, but have
limited practical use. For simulation
purposes, demographic models are the most
appropriate. Simulation itself is not an
element of ecologically-based rodent
management, but it may be very
instrumental in evaluating what the effects
could be of certain interventions in the
population or how the population would
react under specific conditions. Apart from
advanced computer skills, their use requires
detailed biological knowledge, a careful
design of the simulations and a permanent
critical evaluation of the output quality. It
can be very attractive to explore population
dynamics with these models, but because of
their complexity, simulations may
sometimes have side effects that do not
correspond to the user's intentions. The
application of models is
therefore not to be locally but they
should remain a tool for the scientists who
are designing Only when enough
experience is with the model, and an
intelligent user interface is provided, can
such demographic models become more
useful at the local level.
CONCLUSIONS
There has been, and will continue to be, a
small group of scientists with good first-
hand knowledge of African pest rodent
populations, but if we want to make that
knowledge more accessible and less
subjective, we must attempt to structure it
and formulate it in the form of models. In
order to be useful in practice, these models
must be quantitative. For outbreak
prediction purposes, regression models have
the best properties and these can be easily
406
applied. More complex, quantitative
demographic models are very useful for
simulation purposes, but stochasticity
hampers their application for actual
forecasting. Both model types are available
for African rodent systems, but they have
been constructed based on data from a few
localities only. More and longer data series
from other sites are needed to evaluate and
improve the models' general validity.
REFERENCES
Bronner, G., Rautenbach, LL. and Meester, J.
1988. Environmental influence on reproduc-
tion in the Natal multimammate mouse
Mastomys natalensis (ASmith, 1834). South
African Journal of Wildlife Research, 18, 142-
148.
Chidumayo, E.N. 1984. Observations on popula-
tions of multimammate mice at Livingstone,
Zambia. Mammalia, 48, 363-376.
Dickman, CR., Predavec, M. and Lynam, AJ.
1991. Differential predation of size and sex
classes of mice by the barn owl, Tyro alba.
Oikos, 62, 67-76.
Fiedler, L.A 1988a. Rodent problems in Africa.
In: Prakash, 1., ed., Rodent pest management.
Hoca Ra ton, CRC Press Inc., 35-65.
Fiedler, L.A. 1988b. Rodent pest problems and
management in eastern Africa. FAO (Food
and Agriculture Organization of the United
Nations) Plant Protection Bulletin, 36, 125-
134.
Firquet, E., Leirs, H. and Bronner, G. 1996.
Germinating grasses and reproductive
seasonality of MIISt0ll11jS (Rodentia,
Muridae). Mammalia, 60, 775-779.
French, N.R. 1975. Evaluation of demographic
parameters of native rodent populations and
implications for control. Bulletin ofthe World
Health Organisation, 52, 677-689.
French, N.R.1985 Test of model with field data
for a population of Mllstomys huberti. Acta
Zoologica Fennica, 173, 119.

Hansson, L. and Henttonen, R 1988. Rodent
dynamics as community processes. Trends in
Ecology and Evolution, 3,195-200.
Hines,J. 1994. MS-SURVIV Users' ManuaL
Laurel, MD, Patuxent Environmental Science
Center.
Harris, W.V.1937. The grey field mouse. The East
African Agricultural Journal, 2, 315-318.
Hubert, R, Adam, F. and Poulet, AR 1978.
Modeling of the population cycles of two
rodents in Senegal. Bulletin of the Carnegie
Museum of Natural History, 6, 88-91.
Hubert, B. and Adam, F. 1983. The regulation of
the population dynamics of two Sahelian
rodents in a hypothesis. Annales de
la Musee Royale \' Afrique Central, Zoolo-
gie, 237, 193-201.
Hubert, B. and Adam, F. 1985. Outbreaks of
Mastomys erythroleucus and Taterillus gracilis
in the Sahelo-Sudanian zone in SenegaL Acta
Zoologica Fennica, 173, 113-117.
Janssen, J.AM. 1993. Soil nutrient availability in
a primary outbreak area of the African
armyworm, Spodoptera exempta (Lepidoptera,
Noctuidae), in relation to drought intensity
and outbreak development in Kenya. Bulletin
of Entomological Research, 83, 579-593.
Krebs, CJ. 1996. Population cycles revisited.
Journal of Mammalogy, 77, 8-24.
Leirs, H. 1995. Population ecology of Mastomys
natalensis (Smith, 1834). Implications for
rodent control in Africa, Agricultural
Editions. Brussels, Belgian Administration
for Development Cooperation, 35, 268p.
Leirs, H., Stenseth,N.C, Nichols,J,D.,Hines,J.E.,
Verhagen, Rand Verheyen,W. 1997. Stochas-
tic seasonality and nonlinear density-
dependent factors reguLate population size in
an African rodent. Nature, 389, 176-180.
Leirs, H., Stuyck, J., Verhagen, Rand
Verheyen,W. 1990. Seasonal variation in
growth of Mastomys natalensis (Rodentia:
Muridae) in Morogoro, Tanzania. African
Journal of Ecology, 28, 298-306.
Leirs, H., Verhagen, Rand Verheyen,W. 1993.
Productivity of different generations in a
population of Mastomys natalensis rats in
Tanzania. Oikos, 68, 53-60.
Populations of African Rodents
Leirs, H., Verhagen, Rand Verheyen, W.1994.
The basis of reproductive seasonality in
Mastomys rats (Rodentia: Muridae) in Tanza-
nia. Journal of Tropical Ecology, 10, 55-66.
Leirs, R, Verhagen, R, Verheyen, W.,
Mwanjabe, P. and Mbise, T. 1996. Forecasting
rodent outbreaks in Africa: an ecological basis
for Mastomys control in Tanzania. Journal of
Applied Ecology, 33, 937-943.
Leirs, R, Verheyen, W., Mimiels, M., Verhagen r
Rand Stuyck, J. 1989. The relation between
rainfall and the breeding season of Mastomys
natalensis (Smith, 1834) in Morogoro, Tanza-
nia. Annales de la Societe Royale Zoologique
de Belgique, 119,59-64.
Lidicker, W.Z. 1975. The role of dispersal in the
demography of small mammals. In: Golley,
E, Petrusewicz, K. and Ryskowski, L, ed.,
Small mammals: their productivity and
population dynamics. Cambridge,
Cambridge University Press, 103-128.
Linn, LJ. 1991. Influence of 6-methoxybenzoxa-
zolinone and green vegetation on reproduc-
tion of the multimammate rat Mastomys
coucha. South African Journal of Wildlife
Research, 21, 33-37.
Mwanjabe, PS. 1990. Outbreak of Mastomys
natalensis in Tanzania. African Small
Mammal Newsletter, 11, 1.
Mwanjabe, PS. and Siri!l1a, F.B.1993. Large scale
rodent control in Tanzania: present status. In:
Machang'u, RS., ed., Economic importance-
and control of rodents in Tanzania.
Workshop proceedings, 6-8 July 1992,
Morogoro,134-142.
Otis, D.L., Burnham, K.P, White, G.C and
Anderson, D.R. 1978. Statistical inference
from capture data on closed animal popula-
tions. Wildlife Monographs, 62, 1-135.
Poulet, AR 1985. The ecological basis of
forecasting rodent outbreaks in a Sahelian
agrosystem. Acta Zoologica Fennica, 173,
107-111.
Sheppe, W. 1972. The annual cycle of small
mammal populations on a Zambian flood-
plain. Journal of Mammalogy, 53, 445-460.
407
Ecologically-based Rodent Management
Sheppe, W. and Haas, P. 1981. The annual cycle
of small mammal populations along the
Chobe River, Botswana. Mammalia, 45,157-
176.
Shuyler, H.R. 1977. FAO's need for rodent
ecology, population dynamics and forecast-
ing data. European Plant Protection Organi-
sation (EPPO) Bulletin, 7, 297-302.
Sicard, B., Maurel, D., Fuminier, F. and Boissin,J.
1994. Climate, trophic factors, and breeding
patterns of the Nile grass rat (Arvicanthis
niloticus solatus): a 5-year study in the sahelian
region of Burkina Faso (formerly Upper
Volta). Canadian Journal of Zoology, 72, 201-
214.
408
Stenseth, N.C. 1977. Forecasting of rodent
outbreaks: models and the real world.
European Plant Protection Organisation
Bulletin, 7, 303-315.
Taylor, K.D. 1968. An outbreak of rats in agricul-
tural areas of Kenya in 1962. East African
Agricultural and Forestry Journal, 34, 66-77.
Taylor, K.D. and Green, M.G. 1972. An ecological
study of the rodent pests of cereals in the
Kenya Highlands. Unpublished report,
54+42p.
Telford, S.R., Jr. 1989. Population biology of the
multimammate rat, Praomys (Mastomys)
nataiensisatMorogoro, Tanzania, 1981-1985.
Bulletin of the Florida State Museum Biologi-
cal Sciences, 34,249-288.
 20. Ecophysiology and Chronobiology Applied
to Rodent Pest Management in Semi-arid
Agricultural Areas in Sub-Saharan
West Africa

Bruno Sicard, Wamian Diarra and Howard M. Cooper

Abstract

A better understanding of rodent population dynamics and strategies for surviving in

various habitats can be gained from combined studies of genetics, ecology,
ecophysiology and chronobiology. Each of these disciplines can contribute
complementary information to improve rodent pest management (RPM) .

Field studies of the ecophysiology of several species (Arvicanthis niloticus and

Mastomys huberti living in wet habitats; Arvicanthis niloticus in easily flooded
habitats; Mastomys erythroleucus and Taterillus gracilis in semi-arid habitats ;
Taterillus petteri in arid habitats) reveal a set of phase-relat ionships between the
annual cycles of reproduction, mobility and metabolism, and tile annual cycle of
climate and trophic conditions in rodent habitats, defined here as the 'vital-cycle ' .
Improvement of RPM spatial and temporal strategies directed to control mortality,
mobility and reproduction should thus be adapted to the vital-cycle according to
species and habitat. Analysis of the results of field and laboratory studies in
A. niloticus living in easily flooded habitats , allowed an understanding of the
mechanisms involved in regulation of the reproductive cycle and development of
models of the reproduction-dependent outbreak of A. niloticus in 1987 in Burkina
Faso. We also describe data related to other species and discuss the advantages of
integrating a chronobiological approach to the study of population mobility and
mobility-dependent outbreaks.

Keywords

Tropical rodents, rodent pest management, reproduction, water metabolism,

mobility, rodent outbreaks, rodent habitats, rodent population dynamics, circannual
and circadian rhythms, vital-cycle

409
Ecologically-based Rodent Management
INTRODUCTION
ROPICAL RODENTS HAVE been
T studied in several Sahelian-
Sudanese countries of West Africa
in order to improve rodent pest management
(RPM). Long-term investigations conducted
by the French Scientific Research Institute for
Development through Cooperation
(ORSTOM), in the Ivory Coast, Senegal,
Burkina Faso, Guinea, Niger and Mali have
contributed significantly to the current
knowledge of systematics, evolution,
ecology, and more recently, ecophysiology
and chronobiology of Sahelian-Sudanese
rodents. We present here the
ecophysiological and chronobiological
approaches to RPM developed by Institut de
Recherche pour le Developpement (lRD-
ORSTOM, Mammal Laboratory) in Burkina
Faso (1984-1991) and Mali (1992-1997). After
presenting an analysis of the relationships
between RPM approaches using genetics,
ecology, ecophysiology and chronobiology,
we will provide some general information on
methods and discuss the constraints of the
ecophysiological and chronobiological
approaches. Three aspects of the applicatil)Jl
of our research on RPM will then be
described: (i) the characterisation of rodent
habitats; (ii) the identification of temporal
and spatial strategies which are specifically
determined according to the 'vital-cycle' of
the pest species, and (iii) the forecasting of
rodent outbreaks. Lastly, some ideas will be
presented on the prospects for improvement
of RPM through the development of a
chronobiological approach to rodent
population mobility.
410
INTER-RELATIONS BETWEEN
SYSTEMATICS, ECOLOGY,
ECO-PHYSIOLOGY AND CHRONOBIOLOGY
ApPLIED TO RPM
Sahelian-Sudanese rodents often show a
chromosomal polymorphism related to the
presence of heterochromatin and various
chromosomal re arrangements (Arvicallthis,
Ducroz et a1. 1997; Mastomys, Granjon et a1.
1997; Taterillus, Sicard et a1. 1988b; Acomys,
Sicard and Trani er 1996 and Volobouev et a1.
1996; Gerbillus, Maddalena et a1. 1988 and
Volobouev et a1. 1988). In addition to its
fundamental interest, knowledge of the
systematics of Sahelian-Sudanese rodents is
crucial because it allows updating of
identification keys used by practitioners of
agricultural development, and because
biological mechanisms revealed through
ecological, ecophysiological or
chronobiological approaches of RPM are
often species-specific or even population-
specific (Sicard 1995a; Sicard et a1. 1995).
Ecology applied to RPM aims to better
understand species-specific demographic
strategies and establishes correlations between
habitat characteristics and life history traits.
This approach also seeks to understand causes
and mechanisms of population cycles to more
accurately predict outbreaks. Outbreaks may
have specific causes, or may merely represent
particularly marked annual cycles of
abundance. Research undertaken to predict
outbreaks involved the study of factors
determining rodent population dynamics, and
modelling of the role of various internal
(reproduction, mobility, mortality) and
external (climate, predation, competition,
resources, diseases) factors (Hubert et a1. 1978;
Poulet 1980; Hubert and Adam 1985).
 Rodent Pest Management in SubSaharan West Africa
To fully understand the adaptative
significance of primary physiological
functions (reproduction, metabolism and
mobility) it is necessary to understand how
homeostasis relates to ecology in the natural
environment. The field of ecophysiology has
emerged to answer this question with
particular reference to species adaptation in
extreme environments. Our comparative
ecophysiological investigations in several
Sahelian-Sudanese rodents show that
adaptation to variability in arid and semi-
arid environments involves precise phase
relationships between primary physiological
functions and seasons (Sicard and Papillon
1996). This ecophysiological approach is
thus complementary to the ecological
approach since primary physiological
functions are the main internal factors
involved in species' life history and in
rodent population dynamics (Figure 1).
The majority of biological processes are
expressed in the form of interdependent
cycles: biochemical, anatomical,
physiological, behavioural or ecological.
Certain cycles referred to as endogenous
I physiological functions according to seasons
biological rhythms' are controlled by a
neuronal 'pacemaker' (a complex neural
network oscillating with its own specific
period; Morin 1994) involving species-
specific mechanisms.
Since chronobiology aims to understand
the mechanisms regulating such rhythms, it
is interesting to approach the study of a
complex cyclic phenomenon, such as rodent
population dynamics, through the concepts
ofchronobiology. Indeed, to develop models
it is necessary to determine whether
agonistic or antagonistic relationships
between internal (reproduction, mobility,
metabolism and mortality) and external
(climate, resources, chemical signals,
competition, predation, diseases) factors
controlling population dynamics involve
mere cycles or bona fide rhythms.
The endogenous circadian clock (ECC) in
mammals is located in the suprachiasmatic
nucleus (SCN) of the hypo thalamus. The
ECC oscillates with its own species-specific
period which approximates 24 hours. The
period is synchronised to the 24-hour day-
night light cycle by photic information
conveyed from the retina to the SCN via the
retino-hypothalamic pathway (Cooper et aL
1993). The ECC is also subject to seasonal
influences by two brain structures (the
intergeniculate leaflet and the pineal gland)
which integrate seasonal changes in
daylength via direct or indirect connections
with the retina (Moore 1973; Moller and
Pevet 1994; Attar et a1. 1995; Negroni et al.
1995). Because the day-night cycle and
annual changes in daylength are precise,
universal and predictable, these photic
cycles are the environmental cues most used
by living organisms to synchronise primary
(Bronson 1988).
Nevertheless, particularly in small tropical
mammals and depending on the species,
many non-photic factors act on primary
physiological functions. Examples include
temperature (Vivien-Roels and Pevet 1983;
Ouarour et a1. 1991), relative humidity
(Haldar and Saxena 1988), food and water
(Bronson 1989), chemical signals like
6-methoxy-2-benzoxazolinone secreted by
plants during germination (Neal and Alibhai
1991), social factors like pheromones or rodent
density, elements of the landscape structure
(Delattre et al. 1992), and/ or some events in
the environment (Cutrera et al. 1994).
411
Ecologically-based Rodent Management

Forecasting of rodent outbreaks

Knowledge
Rodent population of rodent
dynamics habitats

:E Reproduction Mobility Mortality

c-
a: Metabolism
...
.E
t/)
.!!!
Cl
Q)

~ Species-specific phase relationships between et ho-

u; physiological daily rhythms (Le. sleep, locomotion, food
.g intake, exits of burrows etc.) and the day-night cycle.
'(3
Q)
Q,
t/) ...
en
Q) Species-specific phase relationships between etho-
'(3 physiological functions (reproduction, metabolism and
Q)
Q, mobility) and seasons.
VI
'0
c:: Integration of biological activities In relation
0
.~ to dally and seasonal variability of the enVIronment
u
~
'EQ)
!!

t
Water
redistribution

Figure 1.
Ecology, ecophysiology and chronobiology applied to rodent pest management (RPM) (ECC = endogenous
circadian clock; see text for full explanation) .

412
 Rodent Pest Management in Sub-Saharan West Africa
The degree to which the ECC is directly
or indirectly implicated in the physiological
response to these factors is not fully known.
Nevertheless, the ECC, synchronised by
photic and/or non-photic factors, controls
many daily and seasonal activities (Buijs et
al. 1992; Pevet 1992). A specific example of
the complexity of these interactions was
recently observed in the inner delta of the
Niger River (Mali) in which an increase in
relative humidity associated with the first
rain dramatically modified the normal daily
activity pattern of Arvical1 this. The animals
emerged to eat winged termites during their
first occlusion in the early afternoon of the
following day (Sieard, unpublished results).
This change in diet has important
physiological consequences for Arvicanthis
and illustrates the adaptive significance of
the role of non-photic climatic factors which
can potentially affect the ECC via induced
locomotor activity (Mrosovsky 1996). More
generally, the importance of
chronobiological rhythms for a species is
illustrated by key events which occur at
precise times of the day-night cycle (sleep,
locomotion, food intake, exit from the
burrow etc) or at precise periods of the year
(reproduction, aestivation, dispersal etc). In
arid environments the role of the ECC is
particularly important because it allows
certain behavioural or physiological
functions (dispersal, reproduction) to
anticipate the seasonal occurrence of
predictable favourable or unfavourable
conditions (Sicard and Fuminier 1996).
Because the cyclic nature of primary
physiological functions determined by the
ECC is more or less well adapted according
to predation and competition, it may be
expected that the latter factors sculpt the
evolution of the ECC via natural selection.
Thus, factors acting on individuals, which
arise from various social and environmental
levels and potentially involving the ECC,
introduce feedback into the regulation of
rodent population dynamics (Figure 1).
Phenomena acting on ind ividuals also act
on populations and phenomena acting on
populations are inevitably perceived, at
least, by certain individuals. Ecology allows
integration of individual data by taking into
account the crucial adaptive role of
individual variability. Ecophysiology and
chronobiology allow us to understand the
mechanisms involved in the response of
individuals to external factors. The majority
of these mechanisms are species-specific or
population-specific, thus a genetic approach
also seems necessary. All these approaches
are thus complementary and necessary for
modelling rodent population dynamics.
METHODOLOGICAL CONSIDERATIONS OF
ECOPHYSIOLOGICAL AND CHRONo-
BIOLOGICAL ApPROACHES TO RPM
Our ecophysiological and chronobiological
approaches applied to RPM use, as models,
wild animals living in their natural
envirorunent and aim to identify species-
specific and non species-specific
mechanisms inducing the phase
relationships which link external and
internal factors regulating rodent population
dynamics (Figure 1). These approaches
include three phases: long-term field
monitoring, laboratory and terrarium
experiments, and modelling of the results.
413
Ecologically-based Rodent Management
Long-term field monitoring
Long-term field monitoring was carried out
in Burkina Faso (1984-1991) and Mali (1992-
1997) with the goal to determine the 'vital-
cycle' of the monitored rodent populations.
The vital-cycle is defined here as a set of
phase relationships between annual cycles of
primary physiological functions and annual
cycles of climate and trophic parameters.
Daylength was calculated according to
the latitude of the study region. Rains and
seasonal changes in temperature and air
humidity were recorded from our own
meteorological stations (see Sicard 1987).
The capture, mark and release method was
used. Grids made up of 10 lines of 10 traps,
each separated by 10 or 20 metres, were
established approximately every 30 days, for
7 to 10 days, during many years, in many
habitats. Therefore, it was possible to
calculate various abundance indices,
indicators of reproduction (sex ratio,
percentages of young and sexually active
adults, state of testes and uterus) and
mobility (home range size, home range
overlap, displacement of activity centres)
(Meunier and Solaris 1979; Gautun and
Sicard 1986; Sicard 1987; Gautun et aL 1989).
In addition, trapping networks were
placed elsewhere in each habitat to obtain
data on physiological parameters. Diet was
determined from analysis of stomach
contents (Sicard 1987) and reproduction was
estimated from numbers of embryos in
pregnant females. Reproductive onset and
offset were determined from an analysis of
oestrous cycles using vaginal smears
(Kyelem and Sicard 1994), seasonal changes
in blood sexual steroids using radio-
immunoassays (Maurel et aL 1981) and
414
seasonal changes in the gonadotrophin-
releasing hormone activity of the
gonadotrophic areas of the brain (which
regulate reproduction), using
immunohistochemical methods (Fuminier
1994). Seasonal changes in water metabolism
and aestivation period were determined
from analysis of total body water balance
and of water turnover expressed as a
percentage of total body water (WT as
% TBW; Sicard et a1. 1985), and from analysis
of the activity of the vasopressinergic system
of the brain (which regulates water intake
and related behaviours) using
immunohistological methods (Fuminier et
aL 1993).
Experimental study of causal
relationships
The aim of our experimental studies was to
determine whether correlations derived
from field studies were causally related. We
present here the results of our study on the
regulation of reproduction in Arvican th is
niloticus which examined the effects of all
possible combinations of the various factors
that regulate the reproductive cycle. We also
present results obtained in other species and
results obtained on the regulation of water
metabolism. While reproduction and water
metabolism are only expressed at the
seasonal level, mobility is both expressed at
daily (circadian rhythm of activity) and
seasonal (dispersal and non-dispersal
periods) levels. The experimental study of
mobility is thus more complex and potential
future approaches are discussed at the end
of this chapter.
 Rodent Pest Management in Sub-Saharan West Africa

Modelling for RPM improvement in the Sahelian-Sudanese region, human

demographic growth and climate
Vital-cycles, as defined here (Sicard and
aridification encourage man to occupy easily
Papill;n 1996), are species-specific and
flooded low zones and to develop
habitat dependent. allow a better
permanent crops in the vicinity of
understanding of species' adaptations to
habitations in certain urban and rural areas.
environmental variability and
A sufficient knowledge of pest rodent
determination of favourable and
distribution within the continuum of the
unfavourable periods (Le. temporal and
agro-ecosystem would allow prediction of
spatial strategies) for rodent control.
the consequences of these environmental
Examples presented here concern A. niloticus
modifications on the evolution of the rodent
and Mastomys huberti living in wet habitats;
problem. Ecological and biogeographic
A. niloticus in easily flooded habitats;
studies in the Sahelian-Sudanese region
M. erythroleucus and Taterillus gracilis in
were designed to characterise rodent
semi-arid habitats; and Tatcrillus petteri in
habitats according to several criteria such as
arid habitats.
the type of soil and vegetation (Hubert et al.
Experimental studies allow simulations
1977), the importance and type of human
of both typical and atypical climato-trophic
activity (Sicard et al. 1995) or the landscape
conditions derived from bio-climatological
struct~re (Papillon and Sicard 1995a).
analysis. Comparisons between field and
Further, ecophysiological and
laboratory results then allow modelling of
chronobiological studies sought to
the regulation of primary physiological
understand how cyclic variations in food
functions involved in the vital-cycle.
and water resources allow the definition of
Examples presented here concern regulation
different habitats, and drive adaptive
of reproduction in A. niloticus during typical
mechanisms that govern rodent distribution.
and atypical years, allowing models to be
Four categories of habitats can be
developed of the reproduction-dependent
distinguished based on cyclic variations of
outbreak of A. lliiotiCllS in 1986-1987 in
food and water resources which depend on
Burkina Faso.
numerous factors (rains, altitude, slope,
soils, vegetation, agronomical practices etc;
VITAL-CYCLES, KNOWLEDGE OF RODENT see Papillon and Sicard 1995b and Figure 2):
HABITATS AND RPM IMPROVEMENT
~ In wet but non-flood able habitats (villages
Certain such as M. erythroleucus
and permanent cultivations located at the
occupy many habitats whereas other species
bottom of slopes; i.e. in the first belt of the
like T. petteri occupy only specific habitats
agricultural system named 50joro by the
(Sicard 1992). Analysis of the distribution of
Bambaras population), rodents find
pest species would facilitate the definition of
abundant water and food resources
priorities for research and RPM, at both
throughout the year.
national and regional levels and help to
answer the question: which are the most ~ In flood able habitats (rice growing and
important species and habitats? Nowadays natural low areas), rodents have abundant

415
Ecologically.based Rodent Management
food all year long but must face a period of
flood during rains.
~ In semi-arid habitats (sandy covered
areas and large open fields located mid-
slope), rodents must face a water-
restricted food period during the dry and
hot season.
~ In arid habitats (granite islets, dunes and
higher parts of the hydrographic system),
rodents have rich and varied foods only
during rains and must face a water- and
protein-restricted food period during the
remainder of the year.
Rodent distribution depends on
numerous etho-ecophysiological factors.
Rodents adjust the openings and the depth
of their burrows so that the burrow
atmosphere is saturated and temperature is
close to thermal neutrality. This behaviour,
in addition to a mainly nocturnal activity
rhythm, allows rodents to avoid the rigours
of the Sahelian-Sudanese climate (Sicard
1992). The soil of the most arid Sahelian-
Sudanese habitats contains an enormous
quantity of seeds (Grouzis 1988) from which
rodents make important food reserves.
Nevertheless, water and food are the main
factors limiting survival of Sahelian-
Sudanese rodents and water metabolism (i.e.
the aptitude of rodents to save water by
reducing water losses) is a key factor in
species distribution (Sicard 1987,1992;
Sicard and Fuminier 1994). Water, water
redistribution and food availability are
highly variable throughout seasons and
years, and are dependent on many climatic
and non-climatic factors.
Sahelian and Sudanese climates are often
regarded as sub-desert climates, a term that
misleadingly suggests that rains appear
416
randomly in time and space. Indeed, despite
an important variability in annual rainfall
(200-400 mm in the Sahel versus 500-1000
mm in the Sudan), rains almost always occur
between June and September in the
Sahelian-Sudanese region. Thus, from a
chronobiological point of view, the rainy
season is thus more predictable in Sahelian-
Sudanese regions than in equatorial or
temperate regions. Indeed, in the latter, the
abundance and temporal pattern of rainfall
is of low predictive value in the absence of a
well marked annual rainy season. In
contrast, according to rainfall, the Sahelian-
Sudanese climate is characteri ed by three
well-defined seasons (the dry and hot season
from March to May, the rainy season from
June to September, and the dry and cool
season from December to January) delimited
by less well defined transitional periods
(Figure These seasons are marked by a
succession of potential synchronisers that
include:
1. A maximal rate of increase in temperature
near the end of February;
2. A maximal rate of increase in daylength
near the vernal equinox at the end of
March;
3. A maximal rate of increase in air humidity
near mid-May (related to a seasonal
weakening of the harmattan wind);
4. The arrival of rains in June;
5. A simultaneous decrease in day length, air
humidity and rains near mid-September;
and
6. A decrease in temperature near the end of
November.
 Rodent Pest Management in Sub-Saharan West Africa

(a)

Rains
(!::::,.<150mm/month)

D J F M A M J J A S o N
Dry cool Dry hot
Wet hot season
,---,s:..:e:..::a:.::
s.=.:
on,,-------, TP season TP

(b)
Arid Water- and protein-restricted foods FTP

(f)
Sub-arid Water-restricted foods

:0
-!l!. Easily flooded

Wet Favourable trophic period (FTP)

D J F M A M J J A S o N
Time of the year

(c)
Arid habitats

600

E
<D
"0
-~ Easily flooded
~ habitats

Wild and wet Inhabited by humans Wild and dry

Figure 2.
Climate, trophic resources and rodent habitats. (a) Sahelian-Sudanese climate (!l = range of annual
variations of climatic parameters; TP = transitional periods). The numbers 1-6 indicate the chronological
order of appearance of potential synchronisers (see text). (b) Variability of trophic resources according to
rodent habitats. (c) Diversity of rodent habitats.

417
Ecologically-based Rodent Management
Thus, the Sahelian-Sudanese climate
provides reliable synchronisers that can
influence the ECC of mammals for
anticipatory regulation of physiological
functions (Sicard 1987, 1992). The animal's
capacity to predict external factors and to
adjust in advance its physiology represents
the adaptive advantage of endogenous
rhythms. AnalYSis of the relationship
between a species' distribution according to
habitat (Sicard 1992) and reproductive
timing (Sicard 1999) demonstrates that
anticipatory adaptive capacity increases in
increasingly arid environments.
ECOPHYSIOLOGY, KNOWLEDGE OF
RODENT VITAL-CYCLES AND
RPM IMPROVEMENT
Life history strategies, vital-cycles
and RPM improvement
The "r- and K-selection theory" of Pianka
(1970) is one of the most used models for
describing demographic strategies
developed by species to adapt to their
environment. Other theories have been
proposed (Steams 1976, 1989; Sibly and
Callow 1985; Southwood 1988) and
investigations by Perrin (1989) indicate that
in small mammals, temporally-dynamic
selection, adversity-selection and bet-
hedging theories may be more appropriate
than the dualistic r- and K-selection theory
to establish correlations between life history
patterns and habitat characteristics
(stability, predictability etc.). Knowledge of
life history traits is important for RPM
because pests with different life history
patterns require different control strategies
(Conway 1981; Stenseth 1981; Sullivan
1987). Poulet (1982), Hubert (1982) and
418
Sicard (1992) have shown that life history
traits of Sahelian-Sudanese rodents fit more
or less well to r- and K-selection theory.
Muridae have a more rapid sexual
maturation, a shorter generation time,
larger litter sizes and a lower investment per
young, than Gerbillidae. Thus, Muridae are
more r-strategists than Gerbillidae which
conversely are more K-strategists.
Nevertheless, it is not always possible to
differentiate Muridae and Gerbillidae on the
basis of their longevity and survival. In
addition, certain species such as Gerbillus
nigeriae (which is apparently a K-strategist)
are able to reach extremely high densities of
the outbreak type (Sicard 1987).
The vital-cycle (see earlier definition)
which characterises physiological cycles
according to species, habitat and season
provides fundamental insight to species'
adaptation, crucial for defining the temporal
schedule of RPM actions. Three types of
RPM actions can affect population
dynamics: actions targeting mobility,
reproduction, and/ or mortality (see text
above and Figure 1). Our main focus in this
section is to propose how to build strategies
for scheduling RPM actions as a function of
the vital-cycle rather than to discuss the
possible modes of action.
Actions targeting mobility
Figure 3 summarises previous results
indicating how home range size, home range
overlap and displacements of activity
centres allowed the description of an annual
cycle of rodent population mobility (Sicard
1987,1992; Papillon and Sicard 1995b; Sicard
and Papillon 1996).
 Rodent Pest Management in Sub-Saharan West Africa

Home range Home range Displacement of

overlap (HRO) size (HRS) activity centres (DAC)

SGS High Low (higher in male) Low

DP Low (decreasing) High High

MAXIMAL during
pre-aestivation
SDS Low Low
MINIMAL during
aestivation

RP High (increasing) High High

Figure 3_
Annual cycle of mobility In rodents. Changes In home range size, home range overlap and displacement of
activity centres between successive trapping sessions allow determination of the four stages of the annual
cycle of mobility.

In its most complete expression, this cycle
includes four phases:
~ a sedentary phase while the population is
in a grouped state;
~ a dispersal phase;
~ a sedentary phase while the population is
in a dispersed state;
~ a regrouping phase.
Sahelian-Sudanese human populations
have developed some RPM strategies that
consist either of attracting rodents away
from habitations and crops or protecting
them using sound, physical or chemical
barriers (Diarra 1996). These methods, which
are aimed at affecting rodent mobility, are
probably most effective during those phases
of the annual cycle when animals are most
likely to be actively mobile, i.e. during the
dispersal and re grouping phases. The
hypothesis that knowledge of the annual
cycle of rodent mobility improves the timing
and efficiency of these RPM methods will be
tested in the near future.
Actions targeting reproduction
Strategies affecting reproduction are not
yet Widely used in West Africa, although

419
Ecologically-based Rodent Management
this alternative to an actions targeting
mortality or mobility could be very useful in
certain cases. Factors influencing rodent
population dynamics are so tightly
integrated that an action to
permanently decrease reproduction would
probably elicit powerful compensatory
mechanisms. Our suggestion is that since
avoidance of compen.qatory mechanisms is
preferable, a discontinuous, rather than a
continuous, action aimed at inhibiting
reproduction would be more effective. If this
assumption-currently under
experimentation in the field-is correct, it is
necessary to define the appropriate period
for action. Mechanisms that regulate
reproduction result in young being born at
the most favourable time for their
development. The first half of the birth
period, which is often much more
favourable than the second half (Bronson
1989), is thus an appropriate period for an
action targeting reproduction. Therefore, for
a bait-delivered immunocontraceptive
control to be an effective management
strategy (see Chambers et al., Chapter 10), a
high proportion of the breeding population
would need to be sterilised at the beginning
of the breeding season.
Actions targeting mortality
Poisoning is the most frequently used
technique for rodent control in West Africa.
During the dispersal and regrouping phases,
spatial mobility of animals increases
according to the landscape structure and
certain environmental events. Since rodents
cover long distances, the probability that an
individual will encounter the treated area is
low. This also carries an increased risk in the
event that animals ingest non-lethal
420
amounts of rodenticide and develop
resistance. Conversely, when rodent
densities are low, animals are sedentary and
non reproductive, and we suggest that the
use of anticoagulants is advisable, even in
some r-strategist species. It is thus
interesting to take into consideration the
annual cycle of rodent mobility to determine
favourable or unfavourable periods for
chemical control.
Cases studies in Sub-Saharan
agriculture
Mastomys huberti and Arvicanthis
niloticus living in wet habitats
Figure 4 schematically summarises data
obtained for populations of M. huberti and
A. niloticus in wet habitats where a rich diet
and water is available throughout the year
(Sicard 1987). Investigations of the effects of
water restriction on water balance show that
neither species shows mechanisms to reduce
water losses (Gautun et al. 1989; Sicard 1992;
Fuminier 1994; Sicard et al. 1994; Sicard and
Papillon 1996). Analysis of home range size
(HRS) shows that locomotor activity is
important throughout the year with no
seasonal variations (HRS '" 400 m 2). Analysis
of seasonal displacement of activity centres
(DAC) in consecutive months and analysis
of percentage of home range overlap (HRO)
indicates that displacements of populations
always remain low (DAC .. 33 111;
HRO .. 72%). In spite of the fact that both
species breed all year in wet habitats (except
for a short period of August-September in
A. niloticus) outbreaks never occurred,
whereas in other habitats these species
showed reproduction-dependant outbreaks
(in 1987 in Burkina Faso and in 1994 in Mali
 Rodent Pest Management in Sub-Saharan West Africa
for A. niloticus). Comparative analysis of life
history traits clearly shows that poputations
living in wet habitats are less r-strategists
than those living in easily flooded habitats.
Since wet habitats are stable environments,
species tend to shift their demographic
strategy in the K direction of the r-K axis.
Thus, a typical characteristic of the life cycle
of populations living in wet habitats is the
absence of a period of physiological
imbalance that could offer a window of
opportunity for implementation of RPM.
These features of the vital-cycles in wet
habitats suggest the following strategies
(Figure 4). First, we propose permanent
actions targeting mobility for the two
species. Second, since A. niloticus has a short
period of sexual rest an action targeting
reproduction should be focused on the
period from October to March, whereas in
M. huberti (where reproduction is
continuous) this action can be achieved
during any continuous six-month period of
the year. Finally, due to the K demographic
trends and the absence of a dispersal phase
in these species, a continuous action
targeting mortality is possible or could be
used alternately with actions targeting
reproduction.
Arvicanthls nllotlcus living In easily
flooded habitats
Figure 5 schematically summarises data
obtained in populations of A. niloticus in
easily flooded habitats. This species has a
rich diet all year, but faces a period of
flooding during rains that can be regarded as
an unfavourable period (Sicard 1987). In
easily flooded habitats, as in wet habitats,
water balance in A. niloticuB is always in
equilibrium since water losses are easily
compensated by water gains (Gautun et al.
1989;Sicard 1992; Fuminier 1994; Sicard et al.
1994; Sicard and Papillon 1996). Observed
differences in water turnover between
populations in wet or easily flooded habitats
are partly related to the fact that locomotor
activity is more important in the latter
habitat during the dispersal phase. In easily
flooded habitats, animals are sedentary from
October to April (HRS "" 390 m 2; DAC '" 31 m;
HRO "" 65%), and display an increase in
mobility from May to September
(HRS"" 1160 m 2; DAC "" 60 m). Initially
(May-August) animals show a dispersal
phase marked by a decrease in overlapping
home ranges (HRO "" 23%), followed in
September by a regrouping phase marked
by an important increase in overlapping
home ranges (HRO "" 75%). Animals breed
from October to the end of April. Thus, the
dispersal period starts only when animals
are in sexual rest and, conversely, sexual
activity starts only when animals are
sedentary. Although the role of non-trophic
synchronisers (NTS in Figure 5) in
reproduction has previously been
characterised (Sicard and Fuminier 1996),
possible effects on mobility (i.e. induction of
onset and offset of dispersal and regrouping
phases) remains to be determined. In
addition to the direct effects of non-trophic
factors, an indirect effect of sexual steroids
on central mechanisms involved in the
regulation of mobility cannot be excluded.
The vital-cycle in easily flooded habitats
(Figure 5) suggests focusing actions on
mobility during the rainy season and actions
on reproduction from October to February.
These proposals may be an alternative to
chemical control which is not advisable since
A. niloticlIs is a typical r-strategist in easily
421
Ecologically-based Rodent Management

(a) Action on
mobility

Action on
mortality

Action on
[
reproduction

Continuous reproduction

Sedentary population in a grouped state

High water turnover

Equilibrated water balance

(b) Action on
mobility

Action on
mortality

Action on
reproduction

ca
Reproduction
-
ro
"0 Sedentary population in a grouped state
<ii
u
"0,
o High water turnover
"0
"Ui
>.
.r::: Equilibrated water balance
a.
o
u
llJ

Figure 4.
Vital-cycle and rodent pest management (RPM) strategies in wet habitats. Strategies for RPM (upper
panels) may be deduced from an analysis ofthe vltal-cycles (lower panels) of (a) Mastomys huberti and (b)
Arvicanthls nllot/cus (redrawn from Sicard and Papillon 1996).

422
 Rodent Pest Management in Sub-Saharan West Africa

flooded habitats. Nevertheless, limited
actions targeting mortality may be effective
during two periods: (i) the period of low
rodent densities, decreased reproduction,
and prior to the dispersal phase; and (ii)
during the regrouping phase at the
beginning of reproduction (Figure 5).
Mastomys erythro/eucus and Taterillus
gracilis in semi-arid habitats
Figure 6 summarises data obtained for
populations of M. enjthroleucus and T. gracilis
in semi-arid habitats (Sicard 1987, 1992;
Gautm\ et al. 1989; Fmninier 1994; Sicard and
Papillon 1996). In this habitat these species
have a rich diet from June to the end of
January and a water-restricted diet during the
rest of the year. Seasonal variations in water
turnover are strongly correlated with diet in
the two species, but water balance is in
equilibrium throughout the year.
In M. erythroleucus, analysis of mobility
and reproduction shows that during periods
of food availability, animals are sexually
active and the population is in a grouped
sedentary state (HRS = 530 m 2; DAC = 26 m;
HRO = 84%).

Action on
(+ )
mobility
Action on
(+ ) (+ )
mortality
Action on
reproduction (+ ) (+ )

NTS

Reproduction

C1l
Cii SGS
"0
Cii
.S! High water turnover
Cl
0
(5
w
>-
.r:
Cl.
0
U
LU

-- ..
Figure 5.
Vital-cycle and rodent pest management (RPM) strategies in easily flooded habitats. Strategies for RPM
(upper panel) may be deduced from analysis of the vital-cycle (lower panel) of Arvicanthis niloticus (OP =
dispersal phase; NTS = non-trophic synchronisers-daylength, temperature, humidity etc.; RP = regrouping
phase; SGS = sedentary phase of individuals while the population is in a grouped state). Arrows indicate
that NTS: (i) induce the cessation of reproduction and trigger the DP before the flooding period; (ii) trigger
the RP then the SGS near the end of the flooding period; and (iii) trigger reproduction in October (redrawn
from Sicard et al. 1995; Sicard and Papillon 1996

423
Ecologically-based Rodent Management

(a) Action on
(+)
mobility
Action on
(+ ) (+)
mortality
Action on
reproduction (+ )

f _ _L,
Reproduction

SGS

High water turnover

M RAINS o N
Abundant food resources

(b) Action on
(+ )
mobility
Action on
mortality ~-(_+_
) ~I IIIIII I~--------(+-)--------~
Action on
(+)
reproduction

NTS

Reproduction
C1l
1ii
Cl
SGS SGS
<ii
u
Cl
.Q High water turnover
.Q
(fJ
>.
.r:: Equilibrated water bal ance
a.
o
u
W
M RAINS o N
Abundant food resources
L'-----'
Figure 6.
Vital-cycle and rodent pest management (RPM ) strategies in semi-arid habitats. Strategies for RPM (upper
panels) may be deduced from analysis ofthe vital-cycles (lower panels) of (a) Mastomys erythroleucus and
= = =
(b) Taterillus gracilis (DP dispersal phase; NTS non-trophic synchronisers; RP regrouping phase; SGS
= sedentary phase of individuals while the population is in a grouped state). Food restriction induces a
decrease in water turnover, the cessation of reproduction and triggers DP (February). NTS trigger RP in
May. Changes in diet (Le. period of abundant food resources ) induce an increase in water turnover followed
by the SGS phase (June-July) and subsequently trigger reproduction (August) (redrawn from Sicard et al.
1995; Sicard and Papillon 1996).

424
 Rodent Pest Management in Sub-Saharan West Africa
During the period of restricted foods,
when animals are sexually inactive, mobility
and displacements increase (HRS '"' 1,200
m 2; DAC "" 71 m). From February to the end
of April animals undergo a dispersal phase
(HRO"" 22%), while in May they undergo a
regrouping phase (HRO "" 77%).
Experiments have confirmed that water
restriction induces the decrease in water loss
and the cessation of reproduction (Sicard
1992). We are currently investigating
whether (1) this sequence of events (increase
in water turnover, sedentarisation,
reproductive onset) are linked by causal
relationships and (2) the triggering of the
regrouping phase is due to non-trophic
synchronisers or to the vasopressinergic
system (which is involved in physiological
and behavioural regulation of water
metabolism; Fuminier et a1. 1993; Fuminier
1994). Although chemical control is usually
not recommended for r-strategists such as
M. erythroleucus, this species' vital-cycle
suggests a precise chronological schedule for
complementary (or alternative) actions
targeting mobility, reproduction and
mortality (Figure 6a).
In T. gracilis, as in M. erythroleucus,
restricted food induces a decrease in water
turnover which in turn triggers the cessation
of sexual activity and the start of the
dispersal phase (HRS '"'1,250 m 2; DAC "" 78
m; HRO "" 32%). In contrast, the sedentary
phase (HRO "" 92%) and sexual activity of
T. gracilis start prior to favourable conditions
and the increase in water turnover. Our
experimental results indicate that the
cessation of reproduction is mainly
controlled by trophic resources whereas the
anticipatory reproductive onset is mainly
controlled by daylength and temperature
(review in Sicard et a1. 1996). The non-
trophic synchronisers triggering regrouping
and sedentary phases remain to be
determined. The involvement of different
synchronisers in the reproductive strategies
of T. gracilis and M. erythroleucus is
significant. Due to the high fecundity of
M. erythroleucus (litter size", 10), a short
period of reproduction is sufficient to ensure
reproductive success. In contrast, the low
fecundity of T. gracilis (litter size"" 3)
requires a longer period of reproduction to
reach similar productivity. The onset of
reproduction before rains illustrates the
adaptive significance of a physiological
function that depends on seasonal factors
through an endogenous rhythm. This type of
adaptation is probably just as important as
metabolic adaptations for the maintenance
of species in semi-arid habitats. Although
chemical control is usually recommended
for K-strategists such as T. gracilis, the vital-
cycle of this species suggests that such action
is inappropriate during the dispersal phase.
In addition, a precise chronological schedule
for complementary (or alternative) actions
targeting mobility, reproduction and
mortality is suggested in Figure 6b.
Taterillus petter/llving In arid habitats
Figure 7 summarises data obtained for
populations of T. petteri in arid habitats
(Sicard 1987, 1992; Sicard et a1. 1988b;
Gautun et a1. 1989; Fuminier et a1. 1993;
Fuminier 1994; Sicard and Fuminier 1994;
Sicard and Papillon 1996). These animals
have a rich diet only during the rainy period.
Field monitoring and laboratory results
indicate that, as for many other species,
restricted food induces a decrease in water
turnover (WT "" 44% TBW during rains
425
Ecologically-based Rodent Management

versus 32% TBW during cool season), which
in turn induces the cessation of sexual
activity and the start of the dispersal phase.
Contrary to other species, T. petteri presents
a water imbalance phase (WIP) in February
(WIP = -3% TBW per day). As in other
species, the sedentary phase while the
population is in a grouped state (SGS)
coincides with the rich diet period and is
followed by a dispersal phase (DP), but both
phases are more marked in males: SGS (HRS
= 800 m 2 in males versus 500 m 2 in females;
DAC = 25 m in males versus 17 m in females;
HRO = 79%); DP (HRS= 1,600 m 2; DAC = 65
m; HRO = 10%) from October to December.
In contrast to other species, the dispersal and
regrouping phases are separated by a long
sedentary phase while the population is in a
dispersed state (SDS) from January to May,
whi.ch probably ensures a better spatial
distribution of individuals during the long
period of restricted food. This SDS phase
comprises two stages. During pre-
aestivation (PE in Figure 7) animals build a
complex burrow and must increase the size
of their home range to find foods for
provisions (HRS = 2,100 m 2 while DAC = 0 m
and HRO = 0%).

Action on (+ ) (+ )
mobility

Action on
mortality (-) '---_ _ _( +_) _ _ _ --'11 (-)
Action on
reproduction (+ )

NTS

j ,...--_---.-_+
------------t 1_ G-N_D-=,...J....I-GS-T ....
. . . . . ._
1 _ 1 _ _ B_irt_hs_ _...I

~ ~__~~4_A_e_s_tiv_a_ti_on__~
C\l
co
"0
SGS
c;;
""o
Ol

"0
Equilibrated water balance
"(ji
>.
.<::
Cl.
o High water turnover
u
W

M RAINS

Figure 7.
Vital-cycles and rodent pest management (RPM) strategies in arid habitats. Strategies for RPM (upper
panel) may be deduced from analysis of the vital-cycles of Taterillus petteri (lower panel)(DP = dispersal
phase; GND = pituitary gonadotrophic activity; GST = gestation; NTS = non-trophic synchronisers ; PE =
= =
pre-aestivat ion; RP regrou ping phase; SGS sedentary phase of individuals while the population is in
a grouped state; WIP = water imbalance phase) (redrawn from Sicard 1 992; Sicard and Fuminier 1994;
Sicard et al. 1995; Sicard and Papillon 1 996) . See text for details.

426
 Rodent Pest Management in Sub-Saharan West Africa
During aestivation, animals remain in
their burrow for periods of several weeks
during which they only eat accumulated
provisions and show daily periods of torpor
(HRS "'" 0 m 2; DAC .. 0 m; HRO '" 0%). The
end of aestivation is marked by a short
regrouping phase on the grassy sandy
hollows located between dunes (HRS .. 1,400
m 2; DAC .. 55 m; HRO '" 75%).
Pituitary gonadotrophic activity starts in
the middle of aestivation, so that sexual
coupling is possible during either
aestivation or the regrouping phase. Young
thus appear at the first rains. We have
studied several causal relationships between
these various phenomena. The dispersal
phase is extremely short because it is rapidly
followed by the pre-aestivation period. We
propose that decreases in temperature
trigger pre-aestivation (see NTS in Figure 7).
During pre-aestivation, water losses
decrease (WT", 20% TBW) in spite of the
increase in home range size and the energy
required to build the aestivation burrow and
collect provisions. This indicates that
mechanisms for reducing water losses are
fully active, although a temporary water
imbalance phase occurs in February.
Considering the time between two
recaptures using the tritiated water method,
the water imbalance phase lasts
approximately five days, and indicates that
the water deficit is about 15% of TBW. We
have previously proposed that a water
imbalance phase is the internal trigger for
aestivation (Sicard and Fuminier 1994).
During aestivation, water balance is restored
because animals become inactive and water
requirements are markedly reduced. In
T. petteri, daylength exerts a gonadal-
stimulating effect throughout the year,
which is thwarted by many factors such as
water restriction (Sicard and Fuminier 1994).
This explains why animals are sexually
inactive when they enter aestivation. We
suggest that (1) animals must undergo a
short period of water imbalance to enable a
resurgence of gonadotrophic activity via
changes in daylength and (2) the renewal of
sexual activity is involved in the cessation of
aestivation (see NTS in Figure 7).
The vital-cycle of T. petteri in arid habitats
suggests focusing management actions on
mobility, reproduction and mortality (Figure
7). Actions targeting mobility are
appropriate during the dispersal phase, but
also during the pre-aestivation phase
because animals undergo a water imbalance
phase and are therefore physiologically
fragile. Although aestivation would
theoretically be the ideal time to act on
reproduction, access to the animals in their
burrows is not feasible; thus actions
targeting reproduction are only appropriate
from June to August (the first half of the
reproductive period). Although chemical
control is usually recommended for K-
strategists such as T. petteri, the vital-cycle of
this species indicates that chemical control
only should be implemented during pre-
aestivation and/or during the rainy season.
REGULAtION OF REPRooucTIoN AND
PREDIcTIoN OF REPRODUCTION-
DEPENDENT OUTBREAKS
Knowledge of the mechanisms involved in
rodent population dynamics is very useful
for RPM. Indeed, the extent of damage
caused by rodents depends on annual
variations in rodent population densities.
Rodent outbreaks often have a major impact
427
Ecologica"y-based Rodent Management
on human health and agriculture in
Sahelian-Sudanese regions. Forecasting of
outbreaks mainly depends on an
understanding of their causes and several
mechanisms have been proposed to explain
them. For example, in Senegal, successive
years with favourable rainy seasons allow
the populations of certain rodents to reach a
'pre-outbreak' level that can induce an
outbreak if the following year is favourable
(Poulet 1980; Hubert and Adam 1985).
Analyses by Fiedler (1988a,b) indicate that in
Sahelian Africa, rodent outbreaks are
preceded by several years of prolonged
drought, followed by years with normal or
high rainfall. The increase in rodent density
would be partially due to the effects of the
prolonged drought on the numbers of
rodent predators and competitors. It may
also be that during prolonged drought the
ground becomes fertilised because of the
death of many animals, which would
support vegetation growth in a year with a
normal rainfall (Mutze 1991). In Tanzania,
where there are two annual rainy periods,
the risk of occurrence of Mastomys outbreaks
is related to the aridity of preceding years
and to the rainfall pattern of the current year
(Leirs 1995). In temperate regions, outbreaks
of Microtus always occur in certain types of
habitats, from which they propagate, in
subsequent years according to the landscape
structure. At the end of this propagation
period, they always reappear in the same
types of habitats in a regular pattern every
six or seven years (Delattre et a1. 1996). By
contrast, in Sahelian Africa, rodent
outbreaks occur irregularly, often on a
regional scale, and simultaneously in one or
more species of the genera Arvicanthis,
Mastomys, Taterillus or Gerbillus (Sicard
428
1995b). These examples show that different
mechanisms can lead to outbreaks.
Whatever the initial trigger of an outbreak, it
must act more or less directly on
reproduction, mobility and/ or mortality.
Mortality is an ecological function broader
and more difficult to study than
reproduction or mobility, which are more
specific physiological functions. Below are
the results relating our research on the
regulatory mechanisms of reproduction in
the main pest rodents of the Sahelian-
Sudanese region using A. niloticus living in
easily flooded habitats as an example.
Laboratory and field results have enabled us
to understand the mechanisms leading to the
outbreak of A. niloticus in Burkina Faso in
1987. Analyses of the results and their
prospects for modelling reproduction-
dependent outbreaks are also presented.
Regulation of reproduction in
A. ni/oficus populations living in
easily flooded habitats
A. niloticus is undoubtedly one of the most
important pest rodents of the Sahelian-
Sudanese region. Reproductive patterns of
this species have been extensively studied.
In Uganda, where the climate is constant
with rainfall distributed throughout the
year, reproduction of A. niloticus is
continuous (Neal1981). In Kenya, where the
climate undergoes slight seasonal variations
with two rainy periods, reproduction of A.
niloticus becomes seasonal (Taylor and
Green 1976; Neal1981). In Tanzania (Packer
1983), Ethiopia (Muller 1977), Senegal
(Poulet 1982) and Sudan (Ghobrial and
Hodieb 1982), where the climate shows
differentiated dry and rainy seasons,
reproduction begins more or less near the
 Rodent Pest Management in Sub-Saharan West Africa
end of the rainy season and ends late in the
dry season. Our field studies in Burkina Faso
and Mali show that the reproductive pattern
of A. niloticus also depends on habitat
variability (Sicard et a1. 1994). Indeed,
reproduction of A. niloticus is almost
continuous in stable wet habitats, whereas
reproduction stops from the end of the dry
season to the end of the rainy season in
easily flooded habitats.
Many factors can be involved in the
regulation of reproduction in tropical
rodents. They include daylength (Happold
1983; Khammar and Brudieux 1986,1987),
air humidity (Muller 1977; Haldar and
Saxena 1988), temperature (Vivien-Roels
and Pevet 1983), length of the dry season
(Packer 1983), rainfall via increased water
intake (Yahr and Kessler 1975; Beatley 1976;
Christian 1979), rainfall via variation in the
quality and the quantity of foods (Delany
and Happold 1979), rainfall via a triggering
effect of substances found in germinating
plants (Negus and Berger 1977; Sanders et a1.
1981; Alibhai 1986; Daya et a1. 1990; Linn
1991; Neal and Alibahai 1991). We have
studied the effects of many of these factors
on reproduction, both experimentally and in
animal populations monitored in the field, in
Mali and Burkina Faso (Gautun and Sicard
1985;Sicard eta1.1988a,1992,1993,1994i
Kyelem and Sicard 1996; Papillon et a1.
1996a,b; Sicard and Fuminier 1996; Sicard
and Papillon 1996). In studies over the past
fifteen years, A. niloticus populations in
easily flooded habitats are those which more
frequently show outbreaks. Figure 8a
summarises the combined effects of
daylength, temperature, relative humidity
and food availability on the development
and regression of sexual activity in
A. niloticus originating from easily flooded
habitats and includes the following findings:
.. 6-methoxy-2-benzoxazoli.none does not
influence reproduction.
.. Water restriction or poor foods prevent
development of sexual activity in sexually
inactive animals and conversely induce
regression of sexual activity in sexually
active animals (situation 1 in Figure 8a).
.. Long days exert a gonadal-inhibitory effect
that fully counteracts any gonadal-
stimulating effects of rich foods and
unrestricted water, but only when animals
are under high temperature and low
humidity (situation 2).
.. Gonadal-stimulating effects of rich foods
and unrestricted water never fully
counteract the gonadal-inhibiting effects,
except when animals are in situation 4
(situation 3).
.. Rich foods and unrestricted water
combined with high humidity and low
temperature induce the strongest gonadal
stimulation (situation 4).
From these results, and from
comparisons between experimental and
natural conditions, it is possible to construct
a scenario of gonadal-stimulating and
gonadal-inhibiting effects of climatic and
trophic synchronisers in easily flooded
habitats throughout the year (Figure 8b).
Under natural conditions and during years
with a typical climate, only situations 2 and 3
exist. Situations 4 and 1 correspond to
atypical climatic conditions. Situation 1,
which would correspond to a massive
collapse of resources in eaSily flooded
habitats, is more improbable than situation
429
Ecologically-based Rodent Management

4, which would correspond to additional
rainfall when temperatures are low
(December-January) or to an exceptional
decline in temperature during the rainy
season (August). There is a strong inhibitory
coincidence near the end of March and a
strong stimulatory coincidence near the end
of September (Figure 8b).
Comparisons between experimental and
field results allow a clearer understanding of
the regulation of reproduction in the field. In
Figure 9a, the reproductive onset appears
near the end of September, when the gonadal-
stimulating effects exerted by water and food,
air humidity (and possibly rains) are no
longer thwarted by the gonadal-inhibiting
effect exerted by long days. Once
reproduction has started, animals become
insensitive to gonadal-inhibiting factors
(refractory phase: RPl in Figure 9). Near the
end of March, the three strongly gonadal-
inhibiting conditions (long days, high
temperature and low humidity) concur and
this coincidence induces the end of RPl, the

(a) (b)
Cl
.
Gonadal 1i
development Eoo
c 0~
._
-!..t3
C1l C1l
-0-
C1l
c
o
Cl
t Stimulatory coincidence

Gonadal
Inhibitory coincidence
!
regression
I
Daylength
Air humidity
Temperature

Food and RF &

water UW

6methoxy2benzoxazolinone has no effect

T ::J
(2) (3)

Rare environmental
situation
Improbable
environmental Typical climatotroph ic
situation ' - - - - - conditions in natural
environment

Figure 8.
Regulation of reproduction in Arvlcanthis niloticus living in easily flooded habitats. (a) Results of
experimental studies on the combined effects of potential synchronisers on development and regression of
gonads. Potential synchronisers include daylength (long or short), air humidity (Iow and high), temperature
(Iow and high), 6-methoxy-2-benzoxazolinone found in germinating plants, foods and water (PF = poor foods;
RF =rich foods; RW =restricted water; UW =unrestricted water). Comparisons between experimental and
natural conditions (situations 1,2,3 and 4). (b) Scenario of gonadal-stimulating and gonadal-inhibiting
effects of climatic and trophic synchronisers in easily flooded habitats ( LT = low temperature). See text for
details.

430
 Rodent Pest Management in Sub-Saharan West Africa

appearance of a phase of insensitivity (Pevet 1987). Refractory phases have an

towards gonadal-stimulating factors
(refractory phase: RP2 in Figure 9a), and the
regression of sexual activity. The cessation of
RP2 is not possible before the gonadal-
stimulating effects exerted by food and high
humidity are expressed in the absence of
gonadal-inhibitory effects of long days. Since
investigations by Turek (1972), refractory
phases have been invoked to explain seasonal
variations in the sensitivity of individuals to
the key factors regulating reproduction
obvious physiological significance: a seasonal
physiological activity, once started (or
stopped), cannot be inhibited (or stimulated)
before a certain delay.
The reproduction-dependent outbreak
of A. niloficus in 1986-1987
Analysis of the rainfall records shows that in
1986 there was an atypical climate, not only in
Burkina Faso where we studied an outbreak of
A. niloticus and G. nigeriae, but also in the entire

Unusual
rains
0>
:
:0
i: w
.~ 0
ca ti
'0$
ca
rn '"
c:
0
'0 CJ
>-
0
'
0 0>
.0 c:
ca fJ
...J
'3
E 0~
.~
w-
(J
(0,$
'0
ca
c:
0
CJ
Qi
'0
0
~

rn'"
'0
'0
Qi
u::

~ ~____~(~
S_itu_a_tio_n_3~)____~ @iij[J ~____--,(,---S_itu,-,a_tio_n_3~)____~

Typical climatic and trophic conditions in natural Unusual climatic and trophic conditions in natural
environment during years with usual rainfall environment during the year 1986

(a) (b)
Figure 9,
Reproductive cycle and reproduction-dependent outbreak of Arvicanthis niloticus living in easily flooded
habitats. (a) Modelling of regulatory mechanisms of reproduction during a year with typical rainfall, Arrows
indicate the gonadal-stimulatory and gonadal-inhibitory coincidences. RP1 = refractory phase toward
gonadal-inhibiting factors; RP2 = refractory phase toward gonadal-stimulating factors (situations 2 and 3
are similar to experimental conditions in Figure 8 a). ( b) Modelling of regulatory mechanisms of reproduction
during the year 1 986 with an atypical rainfall (see text for details),

431
Ecologically-based Rodent Management
Sahelian-Sudanese region. In Burkina Faso,
early rains occurred during all of January and
part of February, resulting in a total of
approximately 80 mm. These rains,
accompanied by an unusual increase in
relative humidity, allowed a transient growth
of the vegetation, particularly in extreme wet
and arid habitats. Indeed, the ground was not
completely drained in January, and the rains
caused a sufficient streaming for water to
converge towards the lowest wet areas. In arid
habitats, the ground is covered by gramin-
aceous seeds that can germinate after small
amounts of rainfall. This atypical climatic
situation in January-February 1987
corresponded exactly to the experimental
situation 4 that is described in Figure 8b and
which is the most gonadal-stimulating
situation studied. As shown in Figure 9b, a
strong, additional gonadal-stimulating
coincidence curtailed the refractory phase RP1,
reinforced reproductive activity and started a
refractory phase towards the gonadal-
inhibiting factors that usually cause cessation
of reproduction (unusual RP1 in Figure 9b ). As
indicated from field results, sexual activity of
A. niloticus did not cease during the rainy
season of the year 1986 and reproduction
continued during the entire year, resulting in a
five-month lengthening of the breeding
period. Rodent density was particularly high
in August 1986 and in February 1987 reached
levels of the outbreak type (approximately 20
times higher than during typical years).
In the case of G. nigeriae, January rains in
1986-1987 prevented the water imbalance
phase from occurring and induced the
cessation of the refractory period to the
gonadal-stimulating effect of long days (see
Figure 7). Hence, reproduction started three
432
months in advance and resulted in an
outbreak.
Thus, the experimental study of gonadal-
inhibiting and gonadal-stimulating effects of
climato-trophic factors associated with the
descriptive study of the reproductive cycle
in the field allows a better understanding of
reproductive regulation during typical years
as well as how atypical climate conditions
combine to remodel reproduction.
Modelling reproductlon-dependent
outbreaks
The preceding analyses provide a
demonstration of how the timing and
duration of reproduction depend on the
order in which gonadal-stimulating and
gonadal-inhibiting effects occur. When a set
of stimulating (stimulatory coincidence) or
inhibiting (inhibitory coincidence) effects
occurs, it induces: (i) the cessation of a
refractory phase to stimulating (or to
inhibiting) factors; (ii) the development (or
the regression) of sexual activity; and (iii) a
refractory phase to inhibiting (or
stimulating) factors (gonadal-stimulating
coincidences-GSC; gonadal-inhibiting
coincidences-GIC; refractory phase-RP;
Figure 10). During years with a typical
rainfall, the chronicle of GSC and GIC is
typical and induces a typical breeding
pattern. In contrast, during years with an
atypical rainfall the chronicle of GSC or GIC
becomes unusual and may induce an
atypical breeding pattern, which in turn can
induce a reproduction-dependent outbreak.
These chronicles depend on the chronicle of
environmental synchronisers during typical
and atypical years, and the sensitivity of the
endogenous clock to synchronisers. Thus,
predictive research should take into account:
 Rodent Pest Management in Sub-Saharan West Africa

(1) species-specific influences of and often population specific. For example,

synchrorusers on reproduction; (2) induction
of refractory phases; (3) a climatic approach
aiming to identify the most frequent
disturbances of the Sahelian-Sudanese
climate; and (4) modelling of the impacts of
these disturbances on the chronicle of
synchrorusers.
CHRONOBIOLOGY ApPLIED TO RPM
AND FUTURE RESEARCH PROSPECTS
Improving the systematics of
Sahelian-Sudanese rodents
Regulatory mechanisms of reproduction and
endogenous rhythms are species-specific
A. l1iloticus , which live in easily flooded
habitats, show a period of sexual inactivity
during the rains and are more responsive to
photoperiodic changes than A. niloticus
which live in wet habitats and reproduce
continually (Sicard et al. 1992). Further
genetic studies are needed to determine
whether these two populations of A. niloticus
(Burkina Faso) constitute two different
species. Indeed, two species of Arvicanthis
(A. niloticLls Desmarest 1822 and A. ansorgei
Thomas 1910; Ducroz 1998) have recently
been shown to be sympatric in Mali and
probably also in northern Burkina Faso
(Sicard, unpublished data).

METEOROLOGY PHYSIOLOGY BIOMETEOROLOGY

Analysis of the chronicle Experimental Main disturbances of climate

of synchronizers - studies and trophiC environment

~ / t
Chronicle of the gonadalstimulating and
Chronicle of the gonadal-stimulating and gonadal-inhibiting factors during
gonadal-inhibiting factors during unusual situations
usual years
NEUROSCIENCES
t t
Chronicle of the gonadal -stimulating and Chronicle of additional GSC and GIC
gonadal-inhibiting coincidences during unusual situations

Study of the influences of

t
daily rhythms and seasonal Identification of climatiC and trophiC
functions on reproduction situatIOns that Induce a
(via refractory phases ?) reproduction-dependant outbreak

Field monitoring of outbreaks

(if possible)

ECOLOGY AND ECOPHYSIOLOGY

Figure 10.
Modelling of reproduction-dependent outbreaks (GSC =gonadal-stimulating coincidence; GIC =gonadal-
inhibiting coincidence; RP1 =refractory phase toward gonadal-inhibiting factors; RP2 =refractory phase
toward gonadal-stimulating factors).

433
Ecologicallybased Rodent Management
In the past few years, the genus Mastomys
has increased from six to nine species
(Lavrenchenko et al. 1998) and the genus
Arvichanthis from one to nine species
(Ducroz et al. 1997). Several new species of
Gerbil/us and Taterillus have recently been
found in the Inner Delta of the Niger River.
As shown by these examples, a large array of
species, currently concealed as morpho-
logically similar forms, remains to be
described. Advancing this problem of
systematics at a finer level requires several
complementary tools (morphometry, cyto-
genetics, electrophoresis, gene sequencing).
Improving the classification of rodent
habitats
Genetic and biogeographical approaches
should be combined to determine internal
(chromosomal rearrangements) and external
(habitat structure) factors influencing
population isolation. These approaches
should provide a better understanding of the
distribution of Sahelian-Sudanese rodents
and identify areas (e.g. the Inner Delta of the
Niger River) which generate biodiversity.
Knowledge of these zones is important if
future RPM programs are to be integrated on
a more widespread regional scale covering
several Sahelian countries.
Improving the concept of vital-cycles
The hypotheses raised earlier in this chapter
show that a better characterisation of the
vital-cycle requires further insight into the
regulation of rodent mobility by external
factors (i.e. climato-trophic) and internal
factors (i.e. water metabolism and sexual
steroids). The characterisation of the vital-
cycle also necessitates a finer description of
434
)
/
the annual cycle of mobility. Indeed,
traditional trapping methods are generally
not sufficient to take into account the entire
geographical range of seasonal movements
of individuals and populations.
Radiotelemetry techniques would allow
more complete characterisation of these
seasonal movements, which may occur from
one habitat to another in the same agro-
ecosystem. The description of the annual
cycle of mobility also needs to consider the
effects of the landscape structure on the
activity of rodents during the dispersal and
regrouping periods. Knowledge of
ecophysiology, chronobiology and
landscape ecology is thus fundamental for
improving our understanding of the concept
of vital-cycles.
Forecasting mobility-dependent
outbreaks
The analyses presented above indicate that
there are at least three types of outbreaks:
reproduction-dependent outbreaks (ROO),
mobility-dependent outbreaks (MBDO) and
mortality-dependent outbreaks (MROO). In
the Sahelian-Sudanese region, rodent
outbreaks often occur on a regional scale and
concern only certain species. This suggests
that triggering of outbreaks directly or
indirectly involves climato-trophic factors,
and that the perception of these factors
depends on species-specific mechanisms.
The examples presented above show that
additional rains in January are able to trigger
a ROO in certain (G. nigeriae and A. niloticus),
but not all, species.
Rainfall was abundant in 1986 (more than
400 mm) after a period of prolonged drought
(annual rainfalls lower than 200 mm). It is
thus possible that prolonged drought could
 Rodent Pest Management In Sub-Saharan West Africa
have played a role via previously evoked
factors. The observed RDO may have been
due not only to a lengthening of
reproduction but also to an increase in the
productivity of trophic resources. It is also
possible that these RDO were partly
mortality- and! or mobility-dependent. We
believe that the key factors regulating
population dynamics are so closely
integrated that it is unlikely that one of them
alone can trigger an outbreak. Thus, our
distinction between ROOf MBDO and
MRDO is not exclusive.
Mechanisms analysed in examples
described earlier in this chapter are species-
or population-specific. Because these
mechanisms are different depending on
whether A. niloticus live in easily flooded
habitats or in wet habitats, the different
reproductive cycles of these populations are
differentially affected by the January rains
(only the populations in easily flooded
habitats were affected). A. niloticus did not
show an outbreak in wet habitats in 1986-
1987, probably because reproduction is
typically continuous in this type of habitat.
For this reason, we think that in wet habitats
rodent populations face stronger intra- and
inter-specific competition than in easily
flooded habitats. Consequently, even
r-strategists like Arvicanthis or Mastomys,
would not be able to achieve a RDO in wet
habitat. This does not exclude other
phenomena from triggering either MRoo or
MBDO.
Our experimental studies in
M. erythroleucus and T.gracilis show that
regulatory mechanisms of reproduction are
not sensitive to early rainfall (Sicard 1995b),
which explains why reproduction-
dependent outbreaks were not observed in
1986-1987. This does not exclude other types
of atypical rainfall patterns from triggering a
RDO in these species.
By definition, MRDO result from a
decrease in mortality. The latter can be due
to natural selection (Hubert et a1. 1978) or to
a collapse of predation and! or competition
(Fiedler 1988a,b). Competitors and predators
of rodents are not very species-specific.
Thus, when a MRDO relates only to certain
species it is probably selection-dependent,
whereas if it relates to the majority of species
it is more likely to be predation-dependent
or competition-dependent. Chronobiology
takes into account the fact that different
species do not react in the same way to
disturbances in the environment because
they perceive these disturbances differently.
The specificity of outbreak mechanisms
explains, independently of the fact that
species are r- or K-strategists, why a given
phenomenon evoked to account for an
outbreak in a given species does not
necessarily explain the outbreak of another
species.
We have initiated study of the
mechanisms affecting mobility and
determination of MBDOs in order to address
the question of how synchronisers induce
seasonal changes in mobility (at the
population level) via their effects on the
daily rhythms (activity and behaviours) of
individuals. Indeed, mobility shows daily
and seasonal variations. Therefore it is
necessary to continuously record rodent
activity (not only at the beginning and at the
end of the experiments).
435
Ecologically-based Rodent Management
It is also necessary to understand
relationships between the daily expression
of mobility
which are the key components of the daily
rhythm of activity of an individual?
and the seasonal expressions of mobility
which are the components of the daily rhythm
of activity which change during the annual
and regrouping phases of the
population?
Tt is probable that the factors triggering
outbreaks act in synergy. Modelling of the
mechanisms of outbreaks thus requires
comprehensive knowledge of the various
factors that regulate the population
dynamics of rodents in different habitats.
ACKNOWLEDGMENTS
Research supported by the Department of
Milieux et Activites Agricoles de l'IRO-
ORSTOM (Institut de Recherche pour le
Oeveloppement) and grants from HFSP
(RG-68/95B) and BIOMED (BMH4
CT972327).
REFERENCES
Alibhai, S.K. 1986. Reproductive response of
Gerbil/us harwoodii to 6-\1BOA in the Kora
National Reserve, Kenya. Journal of Tropical
Ecology,2,377-379.
Attar, A, Merrouche, C, Kyelem, M., Sicard, 80
and Cooper, H.M. 1995. Organization of the
SCN and IGL in a nocturnal rodent (Taterillus
petted): a neuroanatomical viral tracing and
neuropeptide study. Biological Rhythms
Research,26,363.
Beatley, rC 1976. Rainfall and fluctuating plant
populations in relation to distributions and
numbers of desert rodents in southern
Nevada. Oecologia, 24, 21-42.
436
Bronson, F.H. 1988. Genes, photoperiod and
latitude. Reproduction, Nutrition and Devel-
opment, 28,335-347.
Bronson, F.H. 1989. Mammalian reproductive
biology. Chicago, University of Chicago
Press, 325p.
Buijs, RM., Jouvet, M., Pevet, P., Valatx, J.L. and
Wirz-Justice, A 1992. Neurobiology of circa-
dian and seasonal rhythms: animal and clini-
cal studies. In: Conference Philippe Laudat
1991, INSERM (Institut National pour la
Sante et la Recherche Medicale), 176-229.
Christian, D.P. 1979. Comparative demography
of three Namib desert rodents: responses to
provision of supplementary water. Journal of
Mammalogy, 60, 679-690.
Conway, G. 1981. Man versus pests. In: May,
RM., ed., Theoretical ecology: principles and
applications (2nd edition). Oxford, B1ackwell,
356-386.
Cooper, H.M., Herbin, M. and Nevo, E. 1993.
Ocular regression conceals adaptive progres-
sion of the visual system in a blind subterra-
nean mammaL Nature, 361, 156~159.
Cutrera, RA, Ourour, A and Pevet, P. 1994.
Specific destruction of the serotonergic affer-
ents to the suprachiasmatic nuclei prevents
triazolam-induced phase advances of
hamster activity rhythms. Behavioural Brain
Research,62,21-28.
Oaya, 5., 80, Pangerl, A, Troiani, M.E.
and Reiter, RJ. 1990. Effect of 6-methoxy-2-
benzoxazolinone on the activities of rat pineal
N-acetyltransferase and hydroxyindole-O-
methyltransferase and on melatonin produc-
tion. Journal of Pineal Research, 8, 57-66.
Delany,M.J. and Happold, o.Co.1979. Ecology of
African mammals. London, Longman, 434p.
Delattre, P., Giraudoux, 1'., Saudry, J., Musard,
P., Toussaint, M., Truchetet, D., Stah!. P. and
Poule,M.L.1992. Land use patterns and types
of common vole (Microtus arvalis) population
kinetics. Agriculture Ecosystems and
Environment, 39, 153-169.
Delattre, P., Giraudoux, 1'., Baudry,J., Quere, J.P.
and Fichet, E. 1996. Relationship between
landscape structure and common vole
(Microtus arvalis) population kinetics.
Landscape Ecology, 11,279-288.
 Rodent Pest Management in SubSaharan West Africa
Diarra, W. 1996. Savoir traditiOlUlel sur les
rongeurs et la lutte anti-rongeurs: application
au contrOle des populations de rongeurs
nuisibles. Mali, ORSTOM Laboratoire de
Mammalogie, 56p.
Ducroz, J.-F. 1998. Contribution aux approches
cytogenetiques et moleculaires al' etude
systematique et evolutive des genres de
rongeurs murides de la" division" Arvican-
this. These de Doctorat, Paris, Museum
National d'Histoire Naturelle.
Ducroz, J.-F., Granjon, L., Chevret, P., Duplan-
tier, J.M., Lombard, M. and Volobouev, V.
1997. Characterization of two distinctive
species of Arvicanthis (Rodentia: Muridae) in
West Africa: cytogenetic, molecular and
reproductive evidence. Journal of Zoology,
241,709-723.
Fiedler, L.A 1988a. Rodent problems in Africa.
In: Prakash, I., ed., Rodent pest management.
Boca Raton, CRC Press Inc., 35-65.
Fiedler, L.A. 1988b. Rodent pest problems in
Eastern Africa. FAO (Food and Agriculture
Organization of the United Nations) Plant
Protection Bulletin, 36(3), 125-134.
Fuminier, F. 1994. Influences des disponibilites en
eau et de la temperature sur le contr6le
photoperiodique de la reproduction chez une
espece pullu lante de rongeur sahelien (Aroican-
this niloUcus). These Doctorat, Montpellier,
University of Montpellier-Il, 143p.
Fuminier, F., Sicard, B., BOissin-Agasse, L. and
Boissin, J. 1993. Seasonal changes in the
hypothalamic vasopressinergic system of a
wild Sahelian rodent (Taterillus petteri). Cell
and Tissue Research, 271, 309-316.
Gautun, J.e. and Sicard, B. 1985. Record de fertil-
ite de A1astomys erythroleucus au Burkina Faso
(ex. Haute Volta). Mammalia, 49, 579.
Gautun, J.e. and Sicard, B. 1986. Conditions
climatiques et evolution des densites des
populations de rongeurs saheliens nuisibles
aux cultures. In: Rijks, D. and Mathys, G., ed.,
L' agrometeorologie et la protection des
cultures dans les zones semi-arides. Geneve,
OMM,147-166.
Gautun, J.e.,Sicard, B. and Mai'zy,J.1989. Spatial
distribution of a rodent population in
Guinean savannah of Ivory Coast. Mamma-
lia, 53, 480.
Ghobrial, L.I. and Hodieb, AS.K. 1982. Seasonal
variations in the breeding of the Nile rat
(Arvicanthis niloticus). Mammalia, 46,319-
333.
Granjon, L., Duplantier, J.M., Catalan, J. and
Britton-Davidian, J. 1997. Evolutionary
systematics in the genus Mastomys (Thomas
1915) (Rodentia, Muridae). Belgian Journal of
Zoology, 127,7-18.
Grouzis, M. 1988. Structure, productivite et
dynamique des systemes ecologiques
saheliens (Mare d'Oursi, Burkina Faso). Paris,
ORSTOM,
Haldar, C. and Saxena, N. 1988. Pineal gland and
humidity effects on testicular function of the
Indian palm squirrel (Funambulus pennantI).
Journal of Pineal Research,S, 411-418.
Happold, D.C.D. 1983. Rodents and lagomorphs.
In: Bourliere F., ed., Tropical savannas.
Amsterdam, Elsevier, 363-400.
Hubert, B. 1982. Dynamique des populations de
deux especes de rongeurs du Senegal,Masto-
mys erythroleucus et Taterillus gracilis (Roden-
tia, Muridae et Gerbillidae). 1. Etude
demographique. Mammalia, 46,137-166.
Hubert, B. and Adam, F. 1985. Outbreaks of
Mastomys erythroleucus and Taterillus gracilis
in the Sahelo-Sudanian zone in Senegal: an
hypothesis. Acta Zoologica Fennica, 173,113-
117.
Hubert, 8., Adam, F. and Poulet, AR. 1978.
Modeling of the population cycles of two
species in Senegal. Bulletin of the Carnegie
Museum of Natural History, 6, 88-91.
Hubert, B., Leprun, J.e. and Poulet, A.R. 1977.
Importance ecologique des facteurs
ectaphiques dans la repartition spatiale de
quelques rongeurs au Senegal. Mammalia,
41,35-59.
Khammar, F. and Brudieux, R. 1986. Variations
saisonnieres de I'activite testiculaire du rat
des sables (Psammomys obesus). In: Assen-
macher 1. and BoissinJ., ed., Endocrine
437
Ecologically.based Rodent Management
regulation as adaptative mechanisms to the
environment. Paris, CNRS (Centre National
de la Recherche Scientifique), 49-55.
Khammar, F. and Brudieux, R. 1987. Seasonal
changes in testicular contents and plasma
concentration of androgens in the desert
gerbil (Gerbillus gerbillus). Journal of Repro-
duction and Fertility, 80, 589-494.
Kyelem,M. and Sicard, B.1994. Ecoregulationde
l' activite ovarienne des rongeurs soudano-
saheliens; consequences sur la modelisation
des risques de pullulation . .'\Juisibles Pest
Pragas,2,149-159.
Kyelem, M. and Sicard, B. 1996. Population
heterogeneity in regulation of reproduction
in Mastomys erythroleucus (Rodentia,
Muridae). Mammalia, 60, 796.
Lavrachencko, LA., Likhnova, a.p., Baskevitch,
M.L and Afework, B. 1988. Systematic and
distribution of Mastomys (Muridae, Rodentia)
from Ethiopia, with the description of a new
species. Zeitschrift fur Siiugetierkunde, 63,
37-51.
Leirs, H. 1995. Population ecology of Mastomys
natalensis (Smith 1834): implication for rodent
control in Africa (agricultural edition).
Brussels, Belgian Administration for Devel-
opment Cooperation, 35, 268p.
Linn, LJ. 1991. Influence of 6-methoxybenzoxa-
zolinone and green vegetation on reproduc-
tion of the multimammate rat Mastomys
coucha. South African Journal of Wildlife
Research, 21, 33-37.
Maddalena, T., Sicard, B., Tranier, M. and
Gautun, J.C 1988. Note sur la presence de
Gerbil/us henleyi (De Win ton, 1903) au Burkina
Paso. Mammalia, 52, 282-284.
Maurel, D., Laurent, A.M. and Boissin, J. 1981.
Short-term variations of plasma testosterone
concentrations in the european badger (Me/es
meTes). Journal of Reproduction and Fertility,
61,53-58.
Meunier, M. and Solaris, A. 1979. Estimation des
densites de populations apartir des captures-
recaptures: application au campagnol des
champs.Mammalia,43,11-24.
Moller, M. and Pe vet, P. 1994. Advances in pineal
research. London, John Libbey, 320p.
438
Moore, RY. 1973. Retinohypothalamic projec-
tions in mammals: a comparative study. Brain
Research, 49, 403-409.
Morin, LP. 1994. The circadian visual system.
Brain Research Review, 19, 102-127.
Mrosovsky, N. 1996. Locomotor activity and
non-photic influences on circadian docks.
Biological Review, 71, 343-372.
Muller, J.M. 1977. Populationsokologie von
Arvicanthis abyssinicus in der Grassteppe des
Semien Mountains National Park (AHhio-
pen). Zeitschrift fUr Saiigetierkunde, 42, 145-
172.
Mutze, G .J. 1991. Mouse plagues in South
Australian cereal growing areas. Ill. Changes
in mouse abundance during plague and non-
plague years, and the role of refugia. Wildlife
Research, 18,593-604.
Neal, B.R 1981. Reproductive biology of the
unstriped grass rat,Arvicanthis, in East Mrica.
Zeitschrift fUr Saiigetierkunde, 46, 174-189.
Neal, B.R and Alibhai, S.K. 1991. Reproductive
response of Tatera leucogaster (Rodentia) to
supplemental food and 6-methoxybenzoxa-
zolinone in Zimbabwe. Journal of Zoology,
223,469-473.
Negroni, }., Attar, A., Merrouche, C, Bennet, N.,
Nevo, E., Sicard, B. and Cooper, HM. 1995.
Pseudorabies virus differentially infects the
visual system of normal and of blind subter-
ranean mammals. American Society of
Neurosciences,21,178.
Negus, N.C and Berger, P.J. 1977. Experimental
triggering of reproduction in a natural
population of Microtus montanus. Science,
196,1230-1231.
Ouarour, A., Kirsch, Rand Pevet, P. 1991. Effects
of temperature, steroids and castration on
daily torpor in the Djungurian hamster
(Phodopus sungorus). Journal of Comparative
Physiology B, 168,477--482.
Packer, C 1983. Demographic changes in a
colony of Nile grass rats (Arvicanthis niloticus)
in Tanzania. Journal of Mammalogy, 64, 159-
161.
Papillon,Y., Kyelem, M. and Sicard, B. 1996a.
Adaptation of reproductive function in
Arvicanthis niloticus (Rodentia, Muridae).
Mammalia, 60, 796.
 Rodent Pest Management in Sub-Saharan West Africa
Papillon, Y., Kyelem, M., Fuminier, F. and Sicard,
B. 1996b. Relationships between reproduc-
tion and other vital functions in Gerbillus
nigeriae (Rodentia, Muridae). Mammalia, 60,
797.
Papillon,Y. and Sicard, B. 1995a. Biogeography a
tool to control rodent in soudano-sahelina
region. European Journal of Plant Pathology,
101,0545 (Abstract).
Papillon,Y. and Sicard, B. 1995b. Comprendre la
repartition spatiale des rongeurs nuisibles
soudano-saheliens. Sahel Integrated Pest
Management,2,11-17.
Perrin, M. 1989. Alternative life-history styles of
small mammals. In: Bruton M.N., ed., Alter-
native life-history styles of animals.
Dordrecht, Kluwer Academic Publishers,
209-242.
Pevet, P. 1987. Environmental control of the
annual reproductive cycle in mammals. In:
Pevet P., ed., Comparative physiology of
environmental adaptations. Strasbourg,
Karger Basel, 82-100.
Pevet, P. 1992. Physiological role of neuropep-
tides in the mammalian pineal gland. In:
Foldes, A. and Reiter, R.J., ed., Advances in
pineal research. London, John Libbey, 6, 17-
36.
Pianka, E.R. 1970. On r- and K-selection. The
American Naturalist, 104,592-597.
Poulet, A.R. 1980. The 1975-1976 rodent
outbreak in a northern Senegal irrigated
farmland. Biotrop Special Publication, 12,
123-138.
Poulet, A.R. 1982. PuUulation de rongeurs dans
la Sahel. Mecanismes et determinisme du
cycle d'abondance de Taterillus pygargus et
d' A rvicanthis nilotieus (Rongeurs, Gerbillide
et Muride) dans le Sahel du Sem!gal de 1975 a
1977. These d' etat, Paris, ORSTOM, 276p.
Sanders, E.H., Gardner, P.D., Berger, P.J. and
Negus, N.C. 1981. 6-methoxybenzoxa-
zolinone: a plant derivative that stimulates
reproduction in Mierotu5 montanus. Science,
214,67-69.
Sibly, R. and Callow, P. 1985. Classification of
habitats by selection presure: a synthesis of
life-cycle and r /K theory. In: Sibly, R.M. and
Smith, R.H., ed., Behavioural ecology.
Oxford, Blackwell Scientific Publications, 5-
90.
Sicard, B. 1987. Mecanismes ecologiques et
physiologiques de regulation des variations
regulieres et irregulieres d'abondance des
rongeurssaheliens. Thesed'etat,Montpellier,
Universite de Montpellier-II, 303p.
Sicard, B. 1992. Influences de l' aridite sur la
biologie des rongeurs soudano-saheliens. In:
LeFloc'h, E., Grouzis, M. and Bille, J.c., ed.,
L'aridite une contrainte au developpement:
caracterisation, reponses biologiques et
strategie des societes. Paris, ORSTOM
(Collection Didactiques), 309-333.
Sicard, B. 1995a. Le probleme rongeurs nuisibles
soudano-saheliens. Sahel Intergrated Pest
Management, 1, 17-22.
Sicard, B. 1995b. Climate variations and
outbreaks in rodent populations in the
Soudano-Sahelian region. European Journal
of Plant Pathology, 101,0830 (Abstract).
Sicard, B. 1999. Photoperiodisme chez les
rongeurs soudano-saheliens. In: Buisson, B.,
ed., Communications du 28eme congres du
Groupe d'Etude des Rythmes Biologique.
Saint Etienne, Universite Jean Monnet (in
press).
Sicard, B. and Fuminier, F. 1994. Relations entre
les variations saisonnieres du metabolisme
hydrique, I'estivation et la reproduciton chez
Gerbillus nigeriaeet Taterillus petteri (Rodentia,
Gerbillidae). Comptes Rendus de l'Academie
des Sciences, 317, 231-238.
Sicard, B. and Fuminier, F. 1996. Environmental
cues and seasonal breeding patterns in
Sahelian rodents. Mammalia, 60, 667-675.
Sicard, B., Fuminier, F., Maurel, D. and Boissin, J.
1993. Temperature and water conditions
mediate the effects of daylength on the breed-
ing of a Sahelian rodent, Arviccmthis
niiotiells. Biology of Reproduction, 49, 716-
722.
Sicard, B., Kyelem, M., Papillon, Y., Diarra, W.
and Keita, M. 1995. Rongeurs nuisibles
soudano-saheliens. Paris, John Libbey, 46p.
Sicard, 5., Maurel, D., Fuminier, F. and Boissin, J.
1992. Circadian rhythm of photosensitivity
and the adaptation of reproductive function
439
Ecologically-based Rodent Management
to the environment in two populations of
Arvicanthis niloticus from Mali and Burkina
Faso. Journal of Reproduction and Fertility,
95,159-167.
Sicard, B., Maurel, D., Fuminier, F. and Boissin, J.
1994. Climate, trophic factors and breeding
patterns of the Nile grassrat (Arvicanthis
ni/oticus solatus); a 5-year study in the
Sahelian region of Burkina Faso (formerly
Upper Volta). Canadian Journal of Zoology,
72,201-226.
Sicard, B., MaureI, D., Gautun, J.C and Boissin, J.
1988a. Activation ou inhibition testiculaire
par la photoperiode chez sept especes de
Rongeurs soudano-saheliens: premiere
demonstration d'une photogonado-sensibil-
ite. Compte Rendus de I' Academie des
Sciences, 307, 11-16.
Sicard, B., Navaras, M., Jacquart, T., Lachiver, F.
and Croset, H. 1985. Metabolisme hydrique
de deux populations de Mus musculus domes-
ticus (Rutty) etMus spretus (Lataste) soumises
adivers regimes hydrigues. Comptes Rendus
de l'Academie des Sciences, 300, 699-704.
Sicard, B. andPapillon, Y.1996. Waterredistribu-
tion and the life cycle of Soudano-Sahelian
rodents. Mammalia, 60, 607-617.
Sicard, B. and Tranier, M. 1996. Caracteres et
repartition de three phenotypes d'Acomys au
Burkina Faso. Mammalia, 60,1-18.
Sicard, B., Tranier,M. and Gautun,J.C 1988b. Un
rongeur nouveau du Burkina Faso (ex. Haute
Volta): Taterillus petteri sp. novo (Rodentia,
Gerbillidae). Mammalia, 521 187-198.
Southwood, T.R.E. 1988. Tactics 1 strategies and
templets. Oikos, 52, 3-18.
Steams, S.C 1976. Life history tactics: a review of
the ideas. Quarterly Review of Mammalogy,
51,3-47.
440
Stearns, S.e.1989. Trade-offs in life history
evolution. Functional Ecology, 3, 259-268.
Stenseth, N.e. 1981. How to control pest species:
application of models from the theory of
island biogeography in formulation of pest
control strategies. Journal of Applied
Ecology, 18, 773-794.
Sullivan, T.P.1987. Understanding the resiliency
of small mammals to population reduction:
poison or population dynamics? In: Richards,
e.G.J. and Ku, T.Y., ed., Control of mammal
pests. London, Taylor and Francis, 69-82.
Taylor, KO. and Green, M.G. 1976. The influence
of rainfall on diet and reproduction in four
African rodents species. Journal of Zoology,
180,367-389.
Turek, F.W. 1972. Circadian involvement in
termination of the refractory period in two
sparrows. Science, 178, 1112-1113.
Vivien-Roels, B. and Pevet, P. 1983. The pineal
gland and the synchronization of reproduc-
tive cycles with variations of climatic condi-
tions, with special references to temperature.
In: Reiter, R.J., ed., Pineal research review.
New York, R. Liss Inc., 91-143.
Volobouev, V., Gautun, J.C, Sicard, B. and
Tranier,M.1996. Thechromosomecomple-
ment of Acomys spp. (Rodentia, Muridae)
from Oursi (Burkina the ancestral
karyotype of the Cahirinus dimidiatus group?
Chromosome Research, 4, 526-530.
Volobouev, V., Viegas-Pequignot, E., Petter, E,
Gautun, J.e., Sicard, B. and Dutrillaux, B.
1988. Complex chromosomal polymorphism
in Gerbillus nigeriae (Rodentia, Muridae).
Journal of Mammalogy, 69, 131-134.
Yahr, P. and Kessler, S. 1975. Suppression of
reproduction in water-deprived Mongolian
gerbils (Meriones unguiculatus). Biology of
Reproduction, 12, 249-254.
 21. The Rodent Problem in Madagascar:
Agricultural Pest and Threat to
Human Health

Jean-Marc Duplantier and Daniel Rakotondravony

Abstract

In Madagascar the rodent problem is linked to one species, the black rat (Rattus
rattus). This chapter will describe its population dynamics in agro-ecosystems and
its impact in agricultural crops , in stored grain, on human health and on the endemic
rodent community. The black rat has spread absolutely everywhere: from sea level to
more than 2, 000 m- in houses, fields and also in the forests . It represents more
than 95% of rodent catches in the fields and inside houses. Reproduction of rats
living in fields stops during the cold season when their maximum annual abundance
is observed. Irrigated rice crops suffer the greatest damage with losses estimated at
2.5% of the harvest. Rodent damage is also important for pluvial rice and to a lesser
degree for cassava , sweet potatoes and tomatoes . Damage to cacao and sugar
cane are important only in the small, poorly-maintained personal plantations. Plague
is undeniably the most important disease linked with rodents in Madagascar. It is
endemic to the centre of the island in rural areas located above 800 m and its
prevalence is increasing. Rodent control in Madagascar is extremely complex
because of the economic difficulties facing the country and because the black rat
has displayed such successful colonisation in absolutely all habitats.

Keywords

Black rat , conservation biology , Madagascar, plague , rice fields , rodent control,
rodent damage

441
Ecologically-based Rodent Management
INTRODUCTION
N MADAGASCAR, the rodent
I problem is clearly linked to one
species, the black rat (Rattus rattus),
and concerns both agriculture and public
health, as well as conservation biology.
Eruptions of rat populations were reported
in 1916,1932 and 1965 (Rakotomanana 1965;
Ravoavy 1966; in Zehrer 1999). After the last
eruption, black rats were declared a public
calamity by the Malagasy state, and
agricultural pest management of rodents
became state-controlled. With respect to
public health, the major disease problem is
undeniably the plague (caused by the
bacillus Yersinia pestis). It spread from the
seaports of Madagascar at the end of the last
century and during the past 70 years
permanent rural foci have existed in the
central part of the island, above 800 m.
Another major issue is that Madagascar's
endemic rodents are threatened with
extinction as a result of competition with the
black rat in addition to habitat loss. This
chapter will examine how this situation
arose, and describe the impact of the black
rat in agricultural fields, in stored grain, on
human health and on the endemic rodent
community in Madagascar.
Madagascar is the fourth largest island in
the world (after Greenland, New Guinea and
Borneo), with a surface area of 587,000
square km. It is 1,600 km long from north to
south and 580 km at its widest point. It has
been separated from the African continent
for 160 million years. The minimum distance
to Africa is now 300 km. The first human
442
settlements occurred approximately 2,000
years ago. Significant immigration occurred
only 1,000 years ago. The human settlers are
of both Asiatic (Indonesian) and African
origin. A mountain range, 2,800 m high,
divides the island into two from north to
south (Figure 1). The east side of the country
is more abrupt than the west. The effect of
the monsoon and the trade winds on this
relief determines the different climates and
different types of vegetation within the
island. Human activities have significantly
changed the vegetation and now the
different climatic regions can be
characterised by their agricultural
landscapes.
The east coast of Madagascar has no dry
season and rainfall ranges from 2,000 to
3,000 mm per year. The original type of
vegetation is rainforest. This type of forest is
in rapid regression, due mostly to 'slash-
and-burn' agriculture. The centre of the
island, called the highlands, has a high
altitude, tropical climate characterised by a
cold and dry season from May to October
and a hot, rainy season from November to
April. Before human settlement, it was a
mosaic of forests-savannah. Today, the
landscape is totally modified by people,
dominated by rice-growing in the valleys
and by dry farming on the slopes (mainly
corn and cassava). The primary forests are
extremely rare, but there are plantations of
pine and eucalyptus. The drier, west coast is
less populated, and is a mixture of
cultivation and pasture. The south is semi-
arid with a 10 month dry season, dominated
by spiny bush. It is a cattle-rearing region
with some small patches of cultivated land.
 Rodent Problems in Madagascar

i
A
'" - \ plague areas

-- <200 m

200- 500 m

-
500-1000 m

>1000 m

0 120 km

Figure 1.
Map of Madagascar: relief, localisation of plague foci, main towns and study sites.

443
Ecologically-based Rodent Management
THE RODENT COMMUNITY IN
MADAGASCAR
At least 23 species of rodents exist today in
Madagascar (Rakotondravony and
Randrianjafy 1998). Like all animal and plant
groups on the island, this order is
characterised by a high rate of endemism.
All the endemic rodents belong to the same
sub-family, Nesomyinae, which is divided
into 8 genera and 20 species. They live
almost exclusively in the forest and their
current distribution is limited to the
remaining primary forests. Brachyuromys
ramirohitra and Nesomys rufus are captured
on farmland but only where this is close to a
forest (Rakotondravony and Randrianjafy
1998). The destruction of primary forests in
Madagascar, well illustrated by Green and
Sussman (1990), threatens the extinction of a
great number of animal species. It is
particularly the case for endemic rodents of
the Nesomyinae sub-family: not only do all
of them live in the forests and will therefore
suffer from habitat loss, but the
fragmentation of these forests makes it easier
for the highly competitive black rat to
penetrate into the forest. Today, it is
common to encounter this species in the
primary forests (Goodman 1995).
In Madagascar, populated areas are the
exclusive domain of three introduced
species, the black rat (R. rattus), the Norway
rat (Rattus norvegicus) and the house mouse
(Mus musculus). R. norvegicus has the most
restricted distribution-it is only found in
the seaports and the big cities. It has been
located in the seaports since the 1930s, but
the date of its spread to the highland cities is
unknown (Brygoo 1966). At the beginning of
the 1980s, it represented more than 80% of
444
the catches in the capital, Antananarivo
(Rakotondravony 1992), which is situated in
the centre of the island, and today it has
reached nearly 95% (Duplantier et al.,
unpublished data). Recently, we trapped
some individuals in rice fields on the east
coast, several km from the cities.
The house mouse is commonly found in
houses but is less numerous than the black
rat. It is also found in rice fields, savannas
and swamp edges, but in low numbers. The
date and method of its settlement of
Madagascar is not known.
The black rat could have come to
Madagascar with the first immigrants
approximately 2,000 years ago, however its
presence is confirmed only from the 11th
century from excavations of an Islamic
archaeological site in the north of
Madagascar (Rakotozafy 1996; Radimilahy
1997). The shrew Suncus murinus, native to
Southeast Asia, must have settled in a
similar way. Today, the shrew is found all
over the island, but it is less abundant than
the black rat. The black rat has spread
everywhere. It can be found from sea level to
more than 2,000 m-in houses, fields and
also in the forests. In the highlands and in
the middle-west, R. rattus represents more
than 98% of rodent catches both inside and
outside buildings (Duplantier et al.,
unpublished data). Ratsimanosika (1995)
quoted the same figure from the east coast
coconut plantations. In the fields of the
coastal regions of Tulear (south-west) and
Tamatave (east), Rafanomezana (1998a)
found 97% of rodents to be R. rattus. In the
natural forest of Andranomay, in the
highlands, R. rattus represented two thirds
of captures from 1981 to 1982 and the

endemic genus Eliurus only one third
(Rakotondravony 1992).
Important studies have been undertaken
recently on the endemic rodents which are
the most threatened order of mammals in
Madagascar (Goodman 1995). However, due
to their restricted distribution and scarcity,
they are of minor importance to agriculture
and probably also to human health.
POPULATION BIOLOGY OF 'rHE
BLACK RAT
It is a paradox that the black rat is the
number one problem of agriculture and
public health but little research has been
conducted on this species. Different survey
programs have been conducted by the
Department of Plant Protection (Ministry of
Agriculture), but most of the results are still
unpublished except for some data on
reproduction (Rafanomezana 1999). In fact,
only three studies provide accurate data
through at least one annual cycle: first, the
study conducted at Andranomay (Figure 1)
in the 1980s by D. Rakotondravony (1981,
1992) which was mostly carried out in the
forest but also in slash-and-burn agriculture
areas. This was followed up by a monthly
survey in the fields of this locality which was
conducted in 1996-97 (Rafanomezana 1999).
Secondly, a monthly survey of less than a
year was conducted in the Lake Alaotra
region (Figure I), which is one of the most
important rice growing regions of the
country (Salvioni 1989). Finally, Duplantier
et aL (unpublished data) conducted a
monthly two-year survey in the villages and
fields of the middle-west region around the
city of Mandoto (Figure 1).
Rodent Problems in Madagascar
Reproduction of the black rat
At the edge of the forests and within fields at
Andranomay, Rakotondravony (1992)
monitored reproduction in the black rat
over two years (1981-1982) (Figure 2).
Reproduction begins before the rainy season
(November to April) with the maximum
number of pregnant females occurring in the
middle of this period. These data have been
confirmed recently: in 1996-1997 there was
an interruption to reproduction in the fields
from May to August, with maximum
reproduction in January (Rafanomezana
1999).
In Mandoto, reproduction of rats living in
houses does not seem to be linked with
season (Figure 3; Rahelinirina and Duplantier
1999). However, among the rats trapped
outside, no reproduction occurs between July
and December-thereafter breeding
increases until May and ceases abruptly in
June. In this region, reproduction seems to be
linked with the harvest of crops rather than
rainfall. The first rice harvest in the valleys
takes place in December-January, then, from
February till May-June, there follows one
after another the harvest of corn, cassava,
peanuts and pluvial rice on the hills. A second
rice harvest can take place in the valleys in
May, at the beginning of the dry and cold
season (Handschumacher et al. 1999).
In Andranomay, as in Mandoto, these
data are confirmed by the age structure of
the rat populations: there are no young less
than 50 g from May to September in
Andranomay, whereas in Mandoto they
are least abundant from September to
March.
445
Ecologically-based Rodent Management

100

90
<Il
Q)
Cii 80
E
$ 70
=:
::J
"0
60
~
Q)
Cii 50
E
$
40
C
ell
c
Cl 30
~
(L
20
0~

10

0
J F M A M J J A S 0 N D J F M A M J J A S 0 N D
1981 1982

Figure 2.
Percentage of pregnant female black rats among adult females trapped monthly in Andranomay forest in
1981-1982 (Rakotondravony 1992).

80

70 Inside
<Il
Q) _______ Outside
Cii 60
E
2
"S 50
"0

~Q)
Cii 40
E
2
C 30
ell
C
Cl

/
Q)

a:: 20
~
0

10

0
Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul
1996 1997
Figure 3.
Percentage of pregnant female black rats among adult females trapped monthly inside and outside houses
in villages of the middle-west region around Mandoto (Rahelinlrina and Duplantler 1999).

446

In the Lake Alaotra fields, reproduction is
concentrated between March and June and
seems to be linked with the availability of
food in the fields (Salvioni 1989). However,
in the presence of a permanent source of
food and a shelter, the rats can breed
throughout the year.
Differences between these localities are
probably linked with the different
landscapes and crop types. In the less
modified landscape of Andranomay (small
slash-and-burn agriculture areas
surrounded by forests), rainfall seems to be
the most important factor influencing the
season of reproduction. By contrast, where
irrigated fields are predominant (Mandoto
and Lake Alaotra), the seasonality of
harvests seems more important.
The aforementioned sites are in the
highlands. In coastal regions, specifically
Fenerive-Est on the East coast (Figure 1), the
minimum number of pregnant females is
observed from July to October and the
maximum from December to March. In the
Tulear region (Figure 1), on the south-west
coast, the maximum rate of pregnancy
occurs approximately over the same period
(January to March) while the minimum
occurs earlier, from May to August
(Rafanomezana 1999).
Variations in abundance with habitat
In the Mandoto region, we compared four
different habitats: houses; sisal fences
around cattle pens; dry farming on the hills;
and the irrigated fields in the valleys
(Duplantier et al., unpublished data). The
black rats were most abundant in the sisal
fences. The permanent and spiny canopy of
the fences provides a good, permanent
Rodent Problems in Madagascar
refuge and in addition, they are situated
near-or even sometimes against-houses
where supplies of food are available.
Conversely, their abundance is lowest in the
dry farming areas because these provide
food for rodents only for a short period and
the fields are often burned by fire once a
year.
In Andranomay, trap success is higher in
fields than in the forest (Figure 4;
Rakotondravony 1992). Data collected
around Lake Alaotra confirm that
environments with abundant vegetation and
near farms host a high density of rats
(Salvioni 1989).
Seasonal variation in abundance
In the fields of Mandoto, we observed a well-
marked, seasonal variation in the abundance
of rodents with a maximum in July-August
in the irrigated fields as well as in the dry
farming areas on the hills (Figure 5;
Duplantier et al., unpublished data). This
was also noted by Salvioni around Lake
Alaotra (Salvioni 1989). On the other hand,
the fluctuations in population abundance
observed in the villages, inside houses and
in the sisal fences do not show any marked
seasonal variation. In the Andranomay
forest, populations peak earlier between
February and April (Figure 4;
Rakotondravony 1992). Rafanomezana
(1999) observed a similar pattern in fields in
the same locality.
Thus, the annual maximum abundance of
rat populations occurs in the middle of the
cold and dry season on the borders of the
highlands (Mandoto and Lake Alaotra), and
earlier, at the end of the rainy season, in the
centre of the highlands (Andranomay).
447
 Rodent Problems in Madagascar

Inter-annual variation DAMAGE IN THE FIELDS

Only one long-term study has been
lmdertaken: from 1980 to 1986, in
Andranomay (Figure 6; Rakotondravony
1992). Inter-annual variation in the
populations of black rats in the forest is low,
but there is high variation in the fields, with
the maximum abundance about eight times
higher than the minimum. In these slash-
and-burn fields, harvest differs greatly from
one year to another, being especially reliant
on the distribution and the total rainfall.
However, it is important to note that at the
same time, Rakotondravony (1992) observed
similar inter-annual variations among urban
populations of black rats. During the 1980s,
their ablmdance in the capital
(Antananarivo) was double that of normal
densities in each of 1981, 1985 and 1989.
A survey conducted throughout the country
showed that 86% of the farmers considered
that they had suffered important damage
from rats during the previous year. Rodent
control was used by 82')'0 of farmers, mostly
in the rice fields (29%) and in the granaries
(27%); 41 % used rodenticides and 27% used
traps (Rafanomezana 1998b).
According to Zehrer (1998), it is the
irrigated rice that suffers the greatest
damage. From surveys conducted in the
different rice-producing regions of the
country, rodent damage affects around 2.2%
of cut stalks, which is equivalent to a loss of
2.5% of the harvest (Raobsoamanitrand-
rasana 1998). For all the country, the overall
annual losses caused by rats are estimated at
62,500 t of rice paddy or 40,000 t of
marketable rice.

4.5

4.0 Forest

3.5 Fields

<1>
0 3.0
c
Cl!
-0
c 2.5
:::J
.0
Cl!
C 2.0
<1>
-0
0
a: 1.5
1.0

0.5

0
1980 1981 1982 1983 1984 1985 1986

Figure 6.
Annual variations of black rat abundance in Andranomay from 1980- 1 986 in forest and field habitats
(Rakotondravony 1992).

449
Ecologicallybased Rodent Management
Rodent damage is also important on
pluvial rice cultivated on the hills or on
slash-and-burn areas (Zehrer 1998). Other
cereals (wheat, barley and oats) seem to be
less affected. With respect to tubers, little
damage is observed in potato crops, whereas
rodents damage cassava and sweet potatoes.
In market garden farms, the more important
damage is to tomatoes and, to a lesser
degree, to green beans, peas, cabbages and
cucumbers.
Among the fruits, the most damage
caused by rodents is to pineapples. Damage
to jackfruit and papaya has also been
observed, but only rarely to bananas.
Damage to cacao is important in the small,
poorly-maintained personal plantations, but
low in industrial plantations-a similar
situation applies to sugar cane plantations.
According to Rasolozaka (1998), ",<uHnF,~
caused by rodents is estimated at 19% of the
production in the coffee plantations of the
east coast. According to Zehrer (1998),
although the farmers complain a lot about
rat depredation and large quantities of
rodenticides are used, the damage is difficult
to evaluate. Indeed the majority of imported
rodent poisons in the 1980s were used in
these coffee plantations. Finally, in the
industrial coconut plantations of the east
coast, the losses caused by rodent damage
are estimated at 2.2% of production
(Ratsimanosika 1998).
DAMAGE TO STORED FOODS
Village stores
In rural areas, according to surveys
conducted in four of the six provinces of the
country, the damage caused by rodents is
considered by rural respondents as similar
450
to that caused by insects (37.5% versus
36.1 %). It is only in the south (Tulear region)
that the damage caused by rodents is low
compared to insect damage (24% versus
58%) (Andriantsileferintsoa 1998). Most
harvested crops are stored in heaps, except
rice which is generally held in sacks. In the
centre, the north and the west of the country,
approximately 74% of the rural respondents
assert that rodent attacks are continuous. In
the south only 21 % of farmers feel this to be
the case.
With respect to health, the abandonment
of traditional granaries by the farmers is a
problem. Traditionally, different types of
independent granaries, set apart from
human houses existed according to the
regions (huts built on stilts, cavities dug in
the earth etc.) (RasamoeI1998). Nowadays
-for security reasons and for fear of
robberies -a great number of farmers
prefer to store their cereals in their houses
and even in their bedrooms. This increases
the possibility of man-rodent contact and
thus the risk of disease transmission. This
could be one of the causes for the increase in
cases of human plague in recent years.
Intermediary stores
These are stores used by the collectors who
gather the harvest of the farmers before
selling it to retailers or big mills. These
structures are rarely rat-proof and the
hygiene is rudimentary (Andriantsile-
ferintsoa 1998). However, the duration of
storage is limited, and this reduces the
importance of damage despite the lack of
protection.

Industrial stores
In Madagascar, only the national silo
(where seeds are stored) and the warehouses
of big commercial companies are rat-proof.
According to Andriantsileferintsoa (1998),
these companies do not have any serious
problems caused by rodents. Their
warehouses are new and well-kept, and they
regularly use the services of rodent/insect
control companies that exist in the capital.
RODENTS AND HEALTH
Buck and COllrdurier (1962), as well as Ribot
and Coulanges (1982), established the first
lists of the main zoonoses in Madagascar. A
synthesis of rodent-borne zoonoses was
completed recently (DlIplantier 1999). The
most well known disease linked with
rodents is the plague, but rodents are also
involved in the transmission of viruses,
including hantaviruses. The importance of
rodents in the transmission of other diseases
in Madagascar is either poorly known (e.g.
intestinal bilharziasis), or less important (e.g.
rabies). Due to the isolation of the island and
its particular community of rodents (a single
endemic sub-family with a very restricted
distribution), a number of diseases linked
with rodents and important elsewhere in the
world are unknovvn on the island (Brygoo
1972). However, the invasion of all the
habitats by the black rat allows for rapid
propagation of any rodent-borne epidemics
throughout Madagascar.
Plague
Plague is undeniably the most important
disease linked with rodents in Madagascar
(Brygoo 1966). It reached the largest seaports
of Madagascar between 1898 and 1907,
Rodent Problems in Madagascar
during the third pandemic. During the 1920s
it reached the highlands, but disappeared
from the coastal regions. Thus, the plague
became endemic to the centre of the island in
rural areas located above 800 m and is now
on the increase (Chantea u et al. 1998b). From
1925 until the Second World War,morethan
one thousand human cases occurred
annually. A maximum of more than 3,000
cases was reached in 1935. Mass vaccination
campaigns, insect control and the arrival of
antibiotics caused an important regression in
morbidity and mortality; for more than
thirty years, an average of only about fifty
cases were reported annually. However, in
1978, the plague reappeared in the capital
after a 28-year absence. Since the end of the
1980s, in the country as a whole, the number
of human cases increased to more than 100
per year, reached 200 in 1994, and 459 in
1997. In 1991, the human plague reappeared
in the sea port of Mahajanga after an absence
of 63 years (Chanteau et a1. 1998a,b). These
official figures include only the cases
confirmed by a bacteriological test.
Serological tests indicate the number of cases
to be two to three times higher (Chanteau et
a1. 1998a).
The real situation is difficult to estimate.
On the one hand, important regions are out
of reach of health services, and on the other,
information campaigns of the national
program of plague control set up since 1991
clearly increased the number of declarations.
Despite a low number of official cases
compared with other diseases, the economic
cost of plague is very high for the Malagasy
state government, which must cover all the
medical expenses in the treatment of this
disease. Due to the slowness of the
bacteriological diagnostics (several days are
451
Ecologically-based Rodent Management
necessary), all cases suspected from clinical
examination must be immediately treated.
In addition, all persons in contact with those
suspicious cases must also undergo an
antibiotic treatment. The houses inhabited
by people with suspected cases, as well as all
the neighbouring houses, must be treated
with insecticides by the health services.
Therefore, several thousand people are
treated every year.
The plague is transmitted from one rodent
to another, and from rodents to humans
through haematophagous flea vectors. In
most of the plague fod in the world, the
reservoirs are wild rodents resistant or less
sensitive to the plague bacillus (Yersinia
pestis). generally belong to the Sciuridae
family (marmots, prairie dogs) and to the
Gerbillinae sub-family (gerbils, Meriones etc.).
The black rat is sensitive to plague and thus
cannot be, in principle, the reservoir. It is a
commensal rodent which is the link between
the wild rodents and humans and hence
provokes urban epidemics. The distinctive
feature of the Malagasy foci lies in the absence
of a wild reservoir; the black rat is involved in
all kinds of foci, urban as well as rural
(Brygoo 1966). It must also be noted that
different fleas are involved depending on the
different habitats: Xenopsylla cheopis occurs
inside houses and Synopsyllus fonquerniei, an
endemic species, occurs outside. Thus, it
seems that in Madagascar the black rat is at
the same time the reservoir and the main
victim of plague. This paradoxical and
unique situation is the result of a balance
between the mortality due to plague and the
re-colonisation abilities of the black rat, which
is spread all over the island, while the plague
occurs only in very localised outbreaks.
452
A program set up in 1996 by the Ministry
of Health (DLMT), the Institut Pasteur de
Madagascar (IPM) and the Institut de
Recherche pour le Developpement (IRD, ex
ORSTOM-the French Scientific Research
Institute for Development through
Cooperation) has made it possible to better
understand the mechanism of the Malagasy
foci. The beginning of the human plague
season in the highlands (November)
coincides with the minimum abundance of
rats outside houses and the annual maximum
abundance of fleas. We have shown that
particular habitats seem to be important for
transmission; i.e. the sisal fences around the
cattle pens, situated inside or on the of
villages. These are where the rodents and the
fleas are most abundant, and it is also where
the highest antibody seroprevalence against
Y. pestis was noted among rats. The rural
plague occurs only above 800 m, yet the black
rat is widespread throughout the island from
sea level to more than 2,000 m, and it is the
same chromosomal form (2n = 38) regardless
of the altitude. Inside houses X. cheopis is also
present independent of the altitude.
However, below 800 m, outside the houses, as
well as on the rodent fur and in their burrows,
we found almost no fleas, while an average of
two fleas per rodent was observed during the
same season above this altitude (IRD /IPM/
DLMT, unpublished data). Thus, it seems that
the distribution of the plague in Madagascar
is limited by the factors that influence the
geographical distribution of the endemic
vector, S. fonquerniei, which parasitises
exclusively rats living outside. It will be
important for plague control to monitor the
evolution of its distribution for a possible
extension of plague foci below the present
limit.

Other diseases linked to rodents in
Madagascar
Murine typhus is relatively important in
Madagascar because Mayoux and
Coulanges (1970) found positive serologies
in 20% of the rats tested in Antananarivo.
Borreliosis (or relapsing fever) is a
disease transmitted by ticks living in the
rodent burrows. It has been well known in
the west of the country since the 18th
century, but there have been no human cases
described since the 1950s (Brygoo 1972).
Leishmaniasis is unknown in
Madagascar, although infested dogs have
been imported (Brygoo 1972) despite
R. rattus being a known reservoir in Italy,
and R. rattus and R. norvegicus being
suspected reservoirs in many other countries
(WHO 1990). According to Brygoo (1972), it
was the absence of vectors that has
prevented the occurrence of this disease in
Madagascar. However in 1982, Ribot and
Coulanges reported the discovery of a
potential vector, an anthropological
phlebotome. Nevertheless, no human cases
have been reported.
The absence of leptospirosis in
Madagascar is undoubtedly a paradox, as it
is a disease which is characteristically
associated with rice fields, sugar cane
plantations and pig rearing (major farming
activities in Madagascar), and R. rattus and
R. norvegicus, as well as S. murinus (a very
widespread shrew in Madagascar), are well
known reservoirs (Faine 1987). In addition,
this disease occurs in the neighbouring
island of Reunion.
Intestinal bilharziasis is widespread in
humans, especially in the highlands, but has
not been reported in rodents (Breuil et al.
Rodent Problems in Madagascar
1982; Ribot and Coulanges 1982). Yet,
recently a Schistosoma mansoni adult worm
was found in one of ten black rats trapped in
the Antananarivo suburbs (E. Sellin, pers.
comm.). Larger surveys and more sensitive
techniques of detection (work in progress:
Ministry of Health/ Institut Pasteur /IRD-
ORSTOM) will determine if the black rat is a
wild reservoir, as is the case in the West
Indies and South America (Rey 1993).
The only known reservoir of rabies in
Madagascar is the domesticated dog. No
rodent has been found to be a rabies carrier
(Ribot and Coulanges 1982).
Until now, no human case of
haemorrhagic fever with renal syndrome,
due to hantaviruses, has been reported in
Madagascar. Since 1985, however,
antibodies of hantavirus have been reported
in rodents (H. Zeller, pers. comm.). More
extensive studies are under way (Institut
Pasteur /IRD-ORSTOM) to determine the
importance of these viruses in black rat
populations trapped in different
environments.
RODENT CONTROL IN MADAGASCAR
Organisation of rodent control
Big cities, and the administrative centres of
the provinces (with more than 100,000
inhabitants) have a Health Municipal Office,
dependent on the mayor. This department is
responsible for keeping the town clean,
monitoring epidemics and controlling
insects and rodents. In the largest cities of
the highlands, it is this department that is in
charge of plague control. The Health
Municipal Office is also in charge of control
of plague (by trapping) among rodent
453
Ecologically-based Rodent Management
populations in the capital, Antananarivo,
and since 1998, in the seaport of Mahajanga.
In small towns and villages, the
Department of Communicable Diseases
(Ministry of Health) intervenes as soon as
human plague cases are reported and
organises the control of fleas and rodents in
collaboration with the basic health care
units.
In rural areas, it is the Department of
Plant Protection (Ministry of Agriculture)
which leads the farmers in rodent control in
the fields and in and around food stores.
Activities of the Department of
Plant Protection
After the 1965 outbreak of the plague,
rodents were declared a public calamity by
the Malagasy state, which meant that the
state covered all the expenditure incurred in
rodent control. Then in 1976, a division
inside the Department was created and
dedicated to rodent controL Until the 1980s,
this division provided free rodenticides to
the farmers, who in turn provided the baits.
Zinc phosphide was first used in the
seventies and progressively replaced by
anticoagulants, mainly chlorophacinone and
coumatetralyL Plant protection technicians
determined the abundance of rodents by
trapping in different regions and crop types
and then, according to the trap success in
each particular situation, decided whether to
use rodenticides. However, this practice was
abandoned due to the deterioration of the
economic situation. Nowadays, the farmers
are in charge of rodent control with only
supervision provided by technicians of the
Department of Plant Protection.
454
In collaboration with GTZ (Deusche
Gesellschaft filr Technische Zusammen-
arbeit -the German Technical
Cooperation), the Division of Rodent
Control carries out three types of actions:
... a survey of the population dynamics in
experimental stations situated in the
different agricultural regions of the
country;
... the promotion of snap-trap use by
demonstration in experimental stations
and through selling at a low price; and
... training and communication on health
matters, cleanliness and promotion of rat-
proofing methods.
Activities of the Division of
Communicable Diseases
This division is responsible for the national
program of plague control and hence rodent
control. The agency supports the local health
structures during epidemics and usually
provides some training and information.
In collaboration with the municipal
health office, a survey of the prevalence of
plague in rat populations has been set up in
the two most important urban foci:
Antananarivo and Mahajanga. This division
has distributed two thousand live-traps to
the health centres located in the plague
endemic zones. These traps are for use in
epidemics, and thus promote this type of
rodent controL
The aim of these training and information
campaigns is to improve the hygiene within
habitations. Information posters written in
Malagasy on the plague-also highlighting
rodent control methods in plague foci-are
posted in all health centres within the plague

areas. A technical guide, which is regularly
updated and includes rodent control
methods, is given to all the doctors working
in the plague endemic zones (Division de la
peste 1990).
The particularities of rodent control in
plague foci
The control of rodent populations in plague
foci must follow certain rules. One must
always remember that the main problem is
not the rodent itself but the fleas. To avoid
proliferation of the disease, it is imperative
to kill the flea before killing the rodent.
During the plague season, rat poisoning
must not be done without prior insect
control. Live-traps are better than snap-
traps, because the latter encourages the
dispersion of the fleas immediately after the
rodent's death. For the same reason, in the
case of chemical control, one must not use
acute rodenticides but anticoagulants that
work only after a few days, thus giving time
for an insecticide to kill the fleas before the
rodent dies. Until now, anticoagulants have
been little used in Madagascar, so no
resistance problems have been reported.
However, resistance to insecticides is
multifaceted and very widespread in fleas
(Ratovonjato et al. 1998). The annual use of
preventive insect controls in some cities is
mainly responsible for this situation and has
had to be abandoned. In rural areas, the
problem is also encountered due to
resistance to the insecticide DOT, which has
been used extensively for malaria control.
All these factors indicate the need for a
regular survey of the efficiency of the
different insecticides and for a rotation in the
use of these products.
Rodent Problems in Madagascar
Improving rodent control
As described above, until recently, few
studies of black rats had been performed.
Rodent control in Madagascar took place
when damage was significant but rarely
took into account knowledge of the biology
of the black rat. Since the beginning of the
1990s, with the creation of a national
program of plague control by the Ministry of
Health and the project "Promoting
integrated protection of crops and stored
foods" by GTZ and the Department of Plant
Protection, new studies have begun with the
aim of improving rodent control.
Since 1993, twelve monitoring stations
have been established throughout the
na tional territory by the Department of Plant
Protection. For various reasons, only eight
stations have operated correctly to provide
reproductive data and two have monitored
abundance. As soon as this program
becomes fully efficient, it will provide useful
data which will enable an adequate control
schedule for each region to be established.
Two long-term shldies of the black rat are
being undertaken on the east and north-west
coasts and these will provide important
additions to our knowledge of population
dynamics, which today, is restricted to the
central part of the island.
Concerning plague control in the
highlands, the monthly survey carried out
over the past two years already allows us to
propose that rodent control should be
focused on a particular area and season. We
have demonstrated that sisal fences are the
most important refuge area for rodents, fleas
and plague. According to the maximum
annual abundance of rats and fleas and
taking into account the agricultural
455
Ecologically-based Rodent Management
activities, we consider that the best time to
practice rodent control is in May-June.
Human sero-positive results against Y. pestis
are more numerous in houses located at the
edges of the villages, where sisal fences are
more abundant and in houses where food is
stored in bedrooms. This clearly shows that
it is necessary to break the close rodent-man
contact and this could be achieved by
developing rat-proof buildings.
CONCLUSIONS:
NEEDS AND PERSPECTIVES
In Madagascar, the major public health
problem linked to rodents is the plague. The
most important reservoir is the black rat,
which is also the most important pest in the
fields and food stores. Along with the
economic difficulties in Madagascar, rodent
control is extremely difficult because it is the
only country in which the black rat has
displayed such successful colonisation in all
habitats. In spite of the recent progress in our
knowledge of the population dynamics of
the black rat, there are still important gaps in
their population ecology which must be
filled before our fight against these rodents
becomes efficient and effective. It is
necessary to obtain reliable data on the
population flow between habitats, due to,
for example, the flooding of rice fields in the
valleys and the slash-and-burn practices on
the hills and at the edges of the forests. Also,
we do not have enough long-term studies on
the population dynamics in cultivated areas.
More needs to be done with respect to
information transfer and training. For
example, some rat-proofing methods are
well known to farmers, but often they are
poorly implemented and inefficient. In rural
456
areas, better coordination of the activities of
the health services and the plant protection
services is necessary.
ACKNOWLEDGMENTS
We would like to thank the Department of
Plant Protection for providing copies of
manuscripts still 'in press' and L. Wilme for
the English translation.
REFERENCES
Andriantsileferintsoa, VD. 1998. Les degats
causes par les rats dans les denrees stockees:
resultats d'une enquete faite aMadagascar.
In: Proceedings-Rongeurs et 1utte
antimurine aMadagascar, Tamatave, 6-9
Decembre 1994. DPV / GTZ (Direction de la
Protection dex Vegetaux - Department of
Plant Protection /Deutsche Gesellschaft fUr
Technische Zusammenarbeit-German
Technical Cooperation), 171-181.
Breuil, J., Moyroud, J. and Coulanges, P. 1982.
Elements de la lutte ecologique anti-bilharzi-
ose it Madagascar. Archives de l'Institut
Pasteur de Madagascar, 50, 131-144.
Brygoo, E.R. 1966. Epidemiologie de la peste it
Madagascar. Archives de l'Institut Pasteur de
Madagascar, 35, 9-147.
Brygoo, E.R.1972. Human diseases and their
relationship to the environment. In: Battistini,
R. and Richard-Vindard, ed., Biogeography
and ecology of Mad agascar. The Hague, Junk
Publishers, 765p.
Buck, G. and Courdurier, J. 1962. Les zoonoses a
Madagascar. Revue d'Elevage et de Medecine
Veterinaire des pays tropicaux, 15,181-191.
Chanteau, 5., Rahalison, L., Duplantier, J.M.,
Rasoamanana, B., Ratsitorahina, M.,
Dromigny,J.A., Laventure, 5., Duchemin,
J.B., Boisier, P., Rabeson, D. and Roux,J.
1998a. Actualites sur la peste aMadagascar.
Medecine TropicaJe, 58 (2S), 25-31.

Chanteau, S., Ratsifasoamanana, L., Rasoa-
manana, B., Rahalison, L, Randriambelosoa,
J., Roux, J. and Rabeson, D. 1998b. Plague, a
reemerging disease in Madagascar. Emerg-
ing Infectious Diseases, 4, 101-103.
Division de la peste 1990. Note technique sur la
peste aMadagascar. Ministere de la Sante,
64p.
Duplantier, J.M. 1999. Rongeurs et sante a
Madagascar: principales maladies
concernees, particularites de la lutte dans les
foyers de peste. In: Proceedings -
Symposium national sur les rongeurs et la
lutte antimurine, Antananarivo, 1-5 Decem-
bre 1997. DPV / GTZ (Direction de la Protec-
tion dex Vegetaux - Department of Plant
Protection/Deutsche Gesellschaft fUr
Technische Zusammenarbeit -German
Technical Cooperation), 173-185.
Faine, S. 1987. Guide pour la lutte contre la
leptospirose. Geneve, World Health Organi-
sation, Publication offset No. 67, 166p.
Goodman, S.M. 1995. Rattus on Madagascar and
the dilemma of protecting the endemic rodent
fauna. Conservation Biology, 9, 450-453.
Green, G.M. and Sussman, KW. 1990. Deforesta-
tion history of the eastern rain forests of
Madagascar from satellite images. Science,
248,212-215.
Handschumacher, P., Brutus, L., Mauny, F.,
Andriantseheno, H., Rajanoarivelo, V. and
Sellin, B. 1999. Des lies dans la Grande lIe:
isolement et risque sanitaire dans le Moyen-
Ouest Malgache. In: Societes et espaces
littoraux et insulaires dans les pays tropicaux
(in press).
Mayoux, A. and Coulanges, P. 1970. Enquete sur
les rikettsioses a Madagascar: I Rickettsia
mooseri- Typhus murin. Archives de l'Insti-
tut Pasteur de Madagascar, 39,41-49.
Radimilahy, C. 1997. Mahilaka, an eleventh to
fourteenth-century Islamic port: the first
impact of urbanism in Madagascar. In:
Goodman, S.M. and Patterson, B.o., ed.,
Natural change and human impact in
Madagascar. Washington DC, Smithsonian
Institution Press, 342-377.
Rodent Problems in Madagascar
Rafanomezana, S. 1998a. Dynamique des
populations des rats: premiers resultats des
postes d' observation de Toliara et
Toamasina. In: Proceedings-Rongeurs et
lutte antimurine a Madagascar, Tamatave, 6-
9 Decembre 1994. DPV /GTZ (Direction de la
Protection dex Vegetaux - Department of
Plant Protection /Deutsche Gesellschaft rur
Technische Zusammenarbeit-German
Technical Cooperation), 105-112.
Rafanomezana, S. 1998b. Comportement et
pratiques des paysans face au probleme "rat".
In: Proceedings-Rongeurs et iutte
antimurine aMadagascar, Tamatave, 6-9
Decembre 1994. DPV /GTZ (Direction de la
Protection dex Vegetaux-Department of
Plant Protection/ Deutsche Gesellschaft fUr
Technische Zusammenarbeit - German
Technical Cooperation), 150-158.
Rafanomezana, S. 1999. Dynamique de popUla-
tion et reproduction des rats dans differents
biotopes de Madagascar. In: Proceedings-
Symposium national sur les rongeurs et la
lutte antimurine, Antananarivo,1-5 Decem-
bre 1997. DPV / GTZ (Direction de la Protec-
ti on dex Vegetaux - Department of Plan t
Protection /Deutsche Gesellschaft filr
Technische Zusammenarbeit-German
Technical Cooperation), 37-56.
Rahelinirina, S. and Duplantier, J.M. 1999.
Dynamique des populations du rat noir
(Rattus rattus) dans les foyers de peste de la
region de Mandoto. In: Proceedings -
Symposium national sur les rongeurs et la
lutte antimurine, Antananarivo, 1-5 Decem-
bre 1997. DPV / GTZ (Direction de la Protec-
tion dex Vegetaux - Department of Plant
Protection/Deutsche Gesellschaft fUr
Technische Zusammenarbeit German
Technical Cooperation), 91-107.
Rakotomanana, KM. 1965. Rapport sur
!'invasion des rats en foret, vue par un fores-
tier. Unpublished report, Departe-ment des
Eaux et Forets, Madagascar, 26p.
Rakotondravony, A.D.S. 1981. Contribution a
I' etude de la dynamique des populations de
rongeurs, application a Rattus rattus et a la
deratisation. Memoire de DEA en Sciences
biologiques appliquees, Universite d' Antan-
anarivo, 281 p.
457
Ecologically-based Rodent Management
Rakotondravony, AD.s.1992. Etude compan?e
de trois rongeurs des milieux malgaches:
Rattus norvegicus Berkenhout (1769), Rattus
rattus Unne (1757) et Eliurus sp., biologie et
dynamique des populations. These de 3
cycle, Universite d'Antan- anarivo, 281p.
Rakotondravony, AD.s. and Randrianjafy, RV.
1998. Especes introduites et endemiques de
Madagascar: description, repartition, biolo-
gie et ecologle. In: Proceedings - Rongeurs et
lutte antimurine aMadagascar, Tamatave, 6-
9 Decembre 1994. DPV /GTZ (Direction de la
Protection dex Vegetaux - Departmen t of
Plant Protection/Deutsche Gesellschaft fUr
Technische Zusammenarbei - German
Technical Cooperation), 51-76.
Rakotozafy, L.M.AI996. Etude de la constitu-
tion du regime alimentaire des habitants du
site de Mahilaka du XIO au XIV O siecle apartir
des produits de fouilles archeologiques. PhD
thesis, University of Antananarivo.
Raobsoamanitrandrasana, V. 1998. Importance
des degats en riziculture irriguee a Madagas-
car: resultats d'une enquete. In: Proceedings
- Rongeurs et lutte antimurine a Madagas-
car, Tamatave, 6-9 Decembre 1994. DPV /
GTZ (Direction de la Protection dex
Vegetaux - Department of Plant Protection/
Deutsche Gesellschaft fi.ir Technische Zusam-
menarbeit - German Technical Coopera-
tion),159-170.
Rasamoel, M. 1998. Les mesures d'hygiene pour
la protection contre les infestations et les
attaques des rats. In: Proceedings -
Rongeurs et lutte antimurine a Madagascar,
Tamatave, 6-9 Decembre 1994. DPV /GTZ
(Direction de la Protection dex Vegetaux-
Department of Plant Protection/Deutsche
Gesellschaft fur Technische Zusammenarbeit
- German Technical Cooperation), 211-218.
Rasolozaka, LR 1998. Importance economique
des degats causes par Rattus raUus dans les
polycultures de la cote Est: premiers resultats.
In: Proceedings-Rongeurs et lutte
antimurine aMadagascar, Tamatave, 6-9
Decembre 1994. DPV /GTZ (Direction de la
Protection dex Vegetaux -Department of
Plant Protection/ Deutsche Gesellschaft fUr
Technische Zusammenarbeit-German
Technical Cooperation), 191-196.
458
Ratovonjato, J., Duchemin, J.B., Duplantier, J.M.,
Laventure, S., Rabarison, P., Chanteau, S. and
Roux, J. 1998. Evaluation de la sensibiliM des
puces pestigenes malgaches aux insecticides
en milieu urbain: resultats et analyse prelimi-
naires. Archives de }'Institut Pasteur de
11adagasca~64,29-33.
Ratsimanosika, L.1995. Contribution al'etude
des Murides ravageurs des cultures, Rattus
raUus Unne(1758) dans les cocoteraies de la
Soavoanio a Sambava: dynamique des
populations, evaluation des degats,
comparaison des methodes de lutte. Memoire
de DEA en Sciences biologiques appliquees,
Universite d' Antananarivo, 48p.
Ratsimanosika, L. 1998. Les problemes de rats
dans les cocoteraies de la Soavoanio
Samba va. In: Proceedings-Rongeurs et 1utte
antimurine aMadagascar, Tamatave, 6-9
Decembre 1994, DPV /GTZ (Direction de la
Protection dex Vegetaux - Department of
Plant Protection /Deutsche Gesellschaft hir
Technische Zusammenarbeit-German
Technical Cooperation), 197-209.
Ravoavy, R 1966. Invasion murine aMadagascar
en 1965. Madagasgar, Service de la Protection
des Vegetaux, unpublished report, 70p.
Rey, L. 1993. Non-human vertebrate hosts of
Schistosoma mansoni and schistosomiasis
transmission in Brazil. Research and Review
in Parasitology, 53, 13-25.
Ribot, J.J. and Coulanges, P.1982. Les zoonoses a
Madagascar. Archives de l'Institut Pasteur de
Madagascar,50,147-166.
Salvioni, 11.1989. Les rongeurs. In: Protection
inMgree en riziculture au Lac Alaotra:
rapport d' activite 1989. Antananarivo,
11adagascar, Cooperation Suisse au Develop-
pement, 7-29.
WHO (World Health Organisation) 1990.
Control ofleishmaniases. WHO Technical
Report Series No. 793, 158p.
Zehrer, W. 1998. Aspects generaux de la lutte
antimurine aMadagascar. In: Proceedings
Rongeurs et lutte antimurine a Madagascar",
Tamatave,6-9 Decembre 1994, DPV /GTZ
(Direction de la Protection dex Vegeta ux-

Department of Plant Protection/Deutsche
Gesellschaft fUr Technische Z usammenarbeit
German Technical Cooperation), 1-19.
Zehrer, W. 1999. La grande invasion murine de
1965 a Madagascar. In: Proceedings-
Symposium national sur les rongeurs et la
Rodent Problems in Madagascar
lutte antimurine, Antananarivo, 1-5 Decem-
bre 1997. DPV / GTZ (Direction de la Protec-
tion dex Vegetaux-Department of Plant
Protection /Deutsche Gesellschaft fUr
Technische Zusammenarbeit-German
Technical Cooperation), 21-36.
459
Rhodes H. Makundi, Nicholas O. Oguge and Patrick S. Mwanjabe

Abstract

Rodents are by far the greatest vertebrate pest problem in East Africa . They are
responsible for substantial damage to food and cash crops, structures and industrial
and domestic property. More than 25 species of rodents have been recorded as
pests in agriculture, causing a wide range of damage and losses in cereals,
legumes, vegetables, root crops, cotton and sugarcane. Pest species occupy a
diversity of habitats, including cultivated fields, urban environments and domestic
areas. Other than being instrumental in crop damage, they are also reservoirs and
carriers of zoonotic diseases, which in some areas of East Africa have claimed many
victims.
The management of rodents has focused on conventional methods, mainly the
use of rodenticides as a symptomatic treatment approach. These methods are
supported by government, especially to contain outbreaks. However, conventional
control methods have remained largely ineffective.
An ecological approach for management of rodent outbreaks is not widely
practiced for lack of basic experimental data to substantiate its efficacy. Measures
that are practiced on a limited scale but have a wide scope for future management
of rodents in East Africa include various techniques of environmental manipulation
that specifically focus on altering the suitable habitats for rodents to reduce their
carrying capacity. Strategies for management of rodent populations in urban areas,
in post-harvest crop systems and in response to disease outbreaks are not well
developed. For the future, a more pragmatic approach is required, involving among
other things, better planning of urban housing schemes, sanitation and hygienic
measures; improved storage structures and practices; and ecologically focused
rodent management techniques .
Recent studies on rodent ecology in East Africa have enabled the development
of models to forecast outbreaks. These, when incorporated in development and
implementation of control activities, may assist in alleviating the damage and losses
due to rodents in the future .

Keywords

Rodents, East Africa, pest management, Mastomys natalens;s, Rattus rattus, Mus
musculus, mathematical models, ecology

460

INTRODUCTION
ORLDWIDE, RODENTS ARE
W the most important group of
mammals in terms of the
problems they create in agriculture,
horticulture, forestry and public health.
They show a wide range of adaptation,
enabling them to successfully colonise and
inhabit almost any type of habitat (De Graaf
1981). Rodents play an important role as
reservoirs and carriers of zoonotic diseases
for which some epidemics have afflicted
mankind for centuries. Indeed, in Africa,
some rodent-borne diseases constitute a
serious burden on the human population in
those areas where they are endemic (Gratz
1997).
Rodent infestations
In developing countries, the major
emphasis in agriculture has been to increase
crop production to feed an increasing
population. In addition, demands for animal
feed and raw materials for basic industrial
production have led also to increased land
usage, both extensively and intensively. This
has created habitats that are suitable for
sustaining higher densities of rodents than
was previously possible. Thus, in rural
communities in Africa, crop damage in the
field and grain losses in storage, although
not fully quantified, are undoubtedly high.
These losses reduce the amount of food
available to the population, and at times of
food scarcity, could lead to an increase in
human suffering due to malnutrition and
starvation in a continent already afflicted by
many natural disasters.
Rodent Pest Management in East Africa
Some species of rodents have developed
close association with human settlements.
Therefore, with fast growing urbanisation in
East Africa and other African countries,
these species have become a serious problem
in terms of public health, spoilage of
foodstuffs, and material and structural
damage. Unlike many developed countries,
where organised and systematic rodent
management strategies are already in place,
lack of resources and insufficient measures
for reduction of rodent populations in urban
areas often compound the rodent problems
in East Africa.
Post-harvest problems
The duration of grain storage at the village
level lasts 3-12 months during which time
rodent infestation can cause economic
losses. The main purposes of on-farm
storage are to provide enough food for the
family through to the next harvest, to
provide seed, and to sell grain when prices
are favourable. Therefore, ensuring that the
food and seed are kept safe is very important
in view of current high demands for
sufficient food for the family and to be able
to sell the surplus to obtain other essentials.
Unprotected storage structures provide
an abundant food source for certain species
of rodents. Under such conditions, numbers
can increase markedly to very high levels
within a short period, leading to severe
losses in the post-harvest period (Makundi
et a1. 1991).
Human diseases
Plague, a disease whose causal agent is
Yersinia pestis, has been known in East Africa
for many years (Roberts 1935; Davis et a1.
461
Ecologically-based Rodent Management
1968; Kilonzo and Msangi 1991) and is
endemic in several areas of Kenya, Tanzania
and Uganda. In recent years, epidemics of
plague have been experienced. The Lushoto
District, Tanzania, recorded 6,599 cases and
580 deaths until December 1996 (Kilonzo et
a1. 1997). Similarly, the Nebbi and Arua
districts in Uganda have experienced plague
outbreaks in the last 10 years (B.5. Kilonzo
1998, pers. comm.). Occurrences of other
human diseases involving rodents ha ve been
reported in other countries (Mills et a1. 1997),
but have been little studied in East Africa.
Factors influencing rodent
pest outbreaks
Although many studies have been carried
out to elucidate the biology of the pest
species, only recently have some of the
reasons for rodent population explosions
become better understood. The reasons
hypothesised for pest increases are
biological-mainly related to the
characteristics of the species themselves;
ecological-ascribed to the total
environment and associated climatic factors;
and those related to human activities-
especially agriculture, urbanisation and
modifications to the natural habitats of
rodents. Thus any attempt at successful
rodent management in East Africa must
consider all these factors.
The objective of this chapter is to review
rodent pest management strategies in East
Africa, namely Tanzania, Kenya and
Uganda, detailing:
~ rodent pest problems;
~ management strategies for rodents in
agriculture (including grain storage),
public health and urban environments,
462
with emphasis on ecological approaches;
and
Ill- measures with potential for future
ecologically-based rodent pest
management.
RODENT PESTS
Although serious arthropod pests
sometimes afflict the East African countries,
rodents are by far the greatest vertebrate
pest problem in agriculture and public
health (Fiedler 1994). They are responsible
for substantial damage to food and cash
crops and play an important role as
reservoirs and carriers of zoonotic diseases
(Fiedler 1994; Gratz 1997; Mills et a1. 1997).
Several species of rodents are pests, ranging
from the medium-sized multimammate rat
(Mastomys natalensis Smith) to the giant rat
(Cricetomys gambianus Waterhouse) and the
crested porcupine (Hystrix cristata Thomas).
Some of the pest species are found
specifically in certain geographical and
environmental conditions, while others are
widely distributed. For example, Rattus
norvegicus (Barkenhout) is restricted to
coastal sea-ports and is not found inland
(Mwanjabe 1989; Fiedler 1994), the house
mouse (Mus musculus L.) is found mostly in
urban areas and in some village dwellings
(Delany 1975; Fiedler 1994) and Rhabdomys
pumilio (Sparman) is commonly found in
grasslands lying at high elevations
(Hubbard 1972). However, the house/roof
rat (Rattus rattus L.), the multimammate rat
(M. natalensis) and the Nile rat (Arvicanthis
niloticus Desmarest), are widely distributed
over East Africa (Kingdon 1974).
More than 25 species of rodents have
been recorded as pests in agriculture,

causing a wide range of damage and losses
in many crop types, including cereals,
legumes, vegetables, root crops, cotton and
sugarcane (e.g. Hubbard 1972; Fiedler 1994).
In Kenya, about 10 species of rodents have
been recorded as the most important pests
(N.O. Oguge 1998, unpublished report).
Most of the serious rodent problems in East
Africa are caused by species belonging to the
family Muridae. They occupy a diversity of
habitats, including cultivated fields where
crop damage occurs. Other than being
instrumental in crop damage, they also
spread diseases, which in some areas of East
Africa have caused many deaths (Gratz
1997).
Some species cause damage in some areas
or in certain crops but not in others. Among
the factors that possibly influence the
occurrence and severity of rodent attack are
the following.
Farming practices
These affect the nature of the habitat, shelter
and population density of rodents. Thus
where small farms-typical of peasant
farming in East Africa-are interspersed
with fallow land, serious damage of the
crops occurs mainly at the edges of fields
(Taylor 1968). The fallow lands also provide
suitable ground for shelter and breeding
while grass and weed seeds are
supplementary food for rodents (Mwanjabe
1993). The introduction of some crops to
certain areas has been associated with
persistent rodent outbreaks. For example, in
the Lake Rukwa Valley, Tanzania, the
opening up of bush and forest for cotton and
maize farming is presumed to have
precipitated the rodent outbreaks in the mid
1960s and 1970s (Mkondya 1975).
Rodent Pest Management in East Africa
Changes in climatic factors
The influence of climatic factors, especially
rainfall patterns, on rodent populations in
East Africa are well-documented (Chapman
et al. 1959; Taylor and Green, 1976; Telford
1989; Leirs 1992, 1995). The timing and
duration of rainfall affects the vegetation,
which is the main source of food for rodents.
The availability of abundant and nutritious
food is a crucial factor for rodent
reproduction and survival. Rainfall,
therefore, is the most likely proximate
influence on the breeding season in some
localities (NeaI1977), with most of the young
born when nutritious food is plentiful (Field
1975; NeaI1984).
The Intrinsic characteristics of the
pest species
Some pest species are able to exploit large
differences in environmental conditions,
even within small geographical areas,
leading to rapid breeding, increase in
abundance and consequently, severe crop
damage. Certain species breed prolifically
under favourable conditions, making it
possible for outbreaks to occur. For example,
M. natalensis is the main species causing
severe damage to cereals and cotton in many
areas in East Africa. This species is able to
live in diverse climatic and geographic
circumstances and therefore is found widely
distributed in fallow and cultivated land
where it thrives primarily on wild plants and
crops, respectively (Kingdon 1974; De Graaf
1981). It has a wide habitat tolerance and
adaptability, which makes it the first invader
of cultivated land (Taylor and Green 1976;
De Graaf 1981). Prolific breeding and rapid
succession are additional characteristics
463
Ecologically-based Rodent Management
which make this species a serious
agricultural pest (Leirs 1992, 1995; Leirs et al.
1993). It has been observed that this species
might be present in low numbers in a
particular area and then rapidly increase,
often associated with ripening cereals
(Taylor and Green 1976). Further, studies on
its distribution have shown that it occurs in
habitats ranging from grasslands to areas
with rainfall of up to 1,800 mm annually (De
Graaf 1981). The distribution of M. natalensis
in habitats which are heterogeneous in many
aspects and its occurrence in some areas also
inhabited by other species suggests that it
must encounter inter-specific competition
(Taylor and Green 1976; K.D. Taylor 1976,
unpublished report). The fact that high
densities of this species occur relative to
others, clearly indicates that M. natalensis is
capable of taking advantage of prevailing
conditions to dominate both natural and
man-altered habitats such as cultivated farm
land.
Three rodent species are responsible for
most post-harvest crop damage. R. mttus
and M. musculus inhabit houses and storage
structures, whereas M. natalensis moves
from the fields to frequently invade rural
storage structures. R. mttus is ubiquitous in
both cities and villages, whereas
M. natalensis is a semi-synanthropic rodent
found only in the peripheries of both cities
and large villages (Mwanjabe 1989).
CROP DAMAGE AND LOSSES
Serious outbreaks of M. natalensis in
Tanzania were recorded as early as the 1930s
(Harris 1937), and in subsequent years in
various parts of the country (Chapman et aL
1959; Mkondya 1977; Mwanjabe 1990).
464
During the 1997/1998 cropping season,
rodent outbreaks were reported in several
regions in Tanzania and resulted in
widespread crop damage (Rodent Control
Centre, Morogoro, Tanzania unpublished
rodent monitoring and control reports,
1998). In Kenya, severe damage in wheat
farms was recorded in 1962 and attributed to
A. niloticus, M. natalensis, and R. pumilio
(Taylor 1968). Pre-harvest losses are
common also as a result of attack by ground
squirrels, Xerus erythropus (Key 1990a).
Rodent damage to cereals manifests itself
in four ways:
~ removal of planted seeds, which
necessitates purchase of new seed and
replanting;
~ attack on the vegetative (growing) stage;
~ damage to mature crops before harvest;
and
~ extensive damage and contamination (via
urine,droppings,hairs,diseaseorganisms)
of grain in village storage granaries.
The emphasis on increasing agricultural
production by expanding the acreage under
crop has certainly created more suitable
habitats for rodents, with a subsequent
increase in population numbers and crop
damage in areas where such problems were
previously unknown. It is also apparent that
some crops are more prone to attack than
others, with the small-grain cereals (rice and
wheat) being more susceptible to rodent
damage than maize during the growing
CAW. Massawe 1998, pers. comm.).
The time of planting also affects crop
damage since in many areas in East Africa,
sowing and crop growth coincides with
increasing rodent populations, especially of

M. natalensis (Taylor and Green 1976; Telford
1989; Mwanjabe and Sirima 1993). Studies in
some areas of East Africa have also shown
that the occurrence of high rainfall during
the short rainy season initiates aseasonal
breeding resulting in high densities of M.
natalensis at the beginning of the main rainy
season (Leirs 1992; Leirs et a!. 1989, 1996).
This coincides with the time of planting
seeds which are removed by rodents,
followed by their attack on emerging
seedlings.
Recent quantitative information on
economic losses due to rodents is generally
lacking in East Africa. However, earlier
reports (Taylor 1968) indicated 20% damage
to maize plantations, 34-100'10 loss of young
wheat in some fields and 34% loss of barley
after outbreaks of rodents in Western Kenya.
Key (1990a,b) found an average of 9.7% of
planted seeds and seedlings and 5.4% of
maize cobs damaged by striped squirrels (X.
erythropus) in southern Kenya. Some reports
in Kenya have indicated that most of the pre-
harvest damage in maize and wheat occurs
between planting and germination, with a
final loss of 2-10% (N.o. Oguge 1998,
unpublished data). In Tanzania, rodent
damage to field crops is widespread
although rodent outbreaks tend to be
sporadic. Rodent outbreaks occur in some
areas for one or a few seasons and cause
widespread losses, but these losses have not
been properly assessed and quantified. In
one study, it was shown that about 6% of the
seedlings were devoured during the
planting season, but the final crop losses or
damage were not available (J.T. Christensen
1984, unpublished data). Some districts
experience rodent attack on crops after
planting, during crop growth and during
Rodent Pest Management in East Africa
storage, increasing the level of losses
significantly (Makundi et a1. 1991). Studies
conducted in Chunya District, Lake Rukwa
Valley, Tanzania, indicated 10% damage of
maize seedlings, resulting in a 9.9% crop loss
at harvest (Myllymiiki 1989, Denmark-
Tanzania Rodent Control Project, Final
Report, unpublished). Mwanjabe and Leirs
(1997) observed crop damage of 40-80% in
Morogoro and Chunya districts but the final
crop loss levels were not provided.
In Uganda, striped squirrels were
reported to be digging up forestry nursery
seeds, and attacking cotton boIls, bean pods
and sweet potato (Kingdon 1974).
Reliable estimates of food storage losses
are unavailable since critical assessments
have not been conducted in any of the
countries in the region. However, rodents
are exceptionally wasteful feeders and
therefore, they spoil more food than they
actually consume (Hall 1970).
RODENT PEsT MANAGEMENT
STRATEGIES USED IN THE PAST
Traditional rodent control strategies in East
Africa have evolved over a long time and
were probably most suited to managing low-
density rodent populations. Thus, in areas
where there have been persistent rodent pest
problems, a variety of techniques have
evolved. In other areas (e.g. Lushoto, north-
eastern Tanzania), there are virtually no
traditional rodent control techniques,
indicating that rodent problems in such
areas are relatively new (R.H. Makundi,
personal observation).
Some methods that were used in the past
were selected to solve localised rodent
465
Ecologically-based Rodent Management
problems in certain areas. These included
the following.
Bounty schemes
These were organised to control rodents,
especially in plague outbreak areas. In
Kilimanjaro region, Tanzania, Lurz (1913)
reported a bounty scheme to control rodents
in the 1912 plague outbreak. The scheme was
introduced also to the Rukwa Valley in
Tanzania to control rodent outbreaks in the
late 1960s (Mkondya 1975). The bounty
schemes were not sustainable for two main
reasons-(i) financial resources were scarce
and (ii) they made villagers less responsible
for rodent control in the absence of payment.
In addition, villagers viewed bounty schemes
as an economic activity and, therefore, those
who participated in killing rodents were not
interested in altering the conditions that
enabled rodents to multiply. This resulted in
bounty schemes being successful in reducing
numbers of rodents temporarily, but did not
change the carrying capacity of suitable
habitats for rodents within the villages.
Burning of houses and vegetation
This was practiced in plague outbreak areas,
in which the dwellings of the victims were
burnt down while villagers with sticks and
clubs killed escaping rodents (Kilonzo 1984).
This approach was probably effective where
inhabitants built shelters which were simple
and temporary and where there was shifting
cultivation and a semi-nomadic life. This
technique probably targeted R. rattus, which
mainly inhabits human dwellings, but not
the other species that are also reservoirs of
plague. The burning of vegetation was based
on the assumption that burnt land was freed
466
from rodent infestation. However, it has
limited use in management of rodents in
agriculture over a whole season since it is
now clearly understood that some of the
major pest species invade regenerating
vegetation after fire (De Graaf 1981). Also,
this technique did not consider the dispersal
capacity of the different rodent species.
Trapping
A large proportion of the rural population in
East Africa has made use of different, locally
produced traps to control rodents. These
have enabled the reduction of rodent
numbers in some localities (Lund 1977;
Kilonzo 1984).
Poisoning
Harris (1937) reported successful control of
rodent outbreaks in maize and cotton fields
in the early 1930s in Tanzania by using
barium carbonate mixed with maize and
sorghum meals. In the 1970s, widespread
baiting with warfarin and zinc phosphide
was used to control rodent outbreaks in
Tanzania (Mkondya 1975, 1977; Fiedler
1994). Government intervention, in the form
of free supplies of rodenticide and
distribution and supervision of bait
application, enabled the reduction of crop
losses. However, with poor extension
services in most villages in rural East Africa,
this management strategy has often failed to
be implemented at the appropriate time to
reduce populations of rodents to
uneconomic levels. It follows that
management of rodents by poisoning in the
past was not based on good knowledge of
their population dynamics, which is
essential if there is to be a strong impact.

CURRENT RODENT MANAGEMENT
STRATEGIES
Use and choice of rodenticides
The use of rodenticides to control rodent
outbreaks is not widely practiced on an
individual farm basis. In Tanzania, the
government has organised control
campaigns since the mid 1970s, but in areas
where major outbreaks do not occur, farmers
do not feel the need to control rodents.
Success in the use of rodenticides
whether acute poisons or anticoagulants
has been influenced by three factors, as
outlined below:
Ill- Availability of the required rodenticides,
often influenced by available funds for
their purchase.
Ill- Acceptability of bait formulations to
rodents, often influenced by palatability
under field conditions. The availability of
other food resources for rodents in the field
may determine the level of bait
consumption. In Tanzania, a highly
acceptable bait formulation of
bromadiolone, targeted againstM.
natalensis, was developed in 1988 (A.
Myllymaki 1989, unpublished report).
Ill- The timing of bait application. This is
critical for alleviating damage. One
hypothesis that has not been widely tested
under different agro-climatic and agro-
ecological conditions, is that rodents
should be effectively controlled during the
season when the population is low and
before animals start breeding, to prevent
them from reaching harrnfullevels at the
time of planting through to harvesting.
However, with small farm holdings, which
Rodent Pest Management in East Africa
can be easily re-colonised, prophylactic
treatment when the population is low may
produce less than expected results,
especially for those species which breed
prolifically (Myllymaki 1987).
Rodenticides have commonly been used
for symptomatic treatment to reduce
damage when rodent populations are
already high. This necessitates the
application of large amounts of rodenticides.
If knowledge of rodent population dynamics
is available, this could be used to suggest
appropriate timing of prophylactic
treatment to alleviate the damage caused in
rodent outbreaks.
Zinc phosphide has been most commonly
used for controlling rodent outbreaks in
Tanzania. In Kenya, it accounts for over 80%
of all rodenticides used in rodent pest
management (N.G. Oguge 1998,
unpublished data). The choice of zinc
phosphide by farmers with little disposable
income is based on low cost and a
reasonably quick effect relative to
anticoagulants, but even this rodenticide is
not easily available to most farmers. In the
1998 rodent outbreaks in various regions
(e.g. Tanga, Pwani and Morogoro) in
Tanzania, zinc phosphide was widely used
because the Ministry of Agriculture supplied
it to farmers free (p.s. Mwanjabe 1998, pers.
comm.). Poor extension services also affect
the useofrodenticides (A.W. Massawe 1998,
pers. comm.).
Physical measures
Physical measures are used widely to control
rodents in East Africa. The measures
commonly practiced by farmers include
trapping, digging and flooding burrows
467
Ecologically-based Rodent Management
(Tachyoryctes spp., Tatera spp.), exclusion
and hunting (Hystrix spp., C. garnbianus and
Thryonomys spp.).
There are many different types of traps
used to capture rodents in different areas
within East Africa, but basically they are of
two main designs: kill and live-traps. Traps
are widely used to control rodents within
houses, storage structures and in crop fields.
Trapping methods are generally popular
among peasant farmers who lack other
resources for rodent pest management, and
in few places are used to capture rats to
supplement the diet. For rodent control to be
highly successful where there is crop
damage or threat of disease, use of traps has
to be combined with other control methods
such as environmental sanitation, proper
storage of food, and when necessary,
application of rodenticides (K.D. Taylor
1976, unpublished report). When
rodenticides are available, trapping is a less
favoured method because it is labour
intensive and is less effective in controlling
outbreaks. However, pitfall especially
those which also combine drowning in tins
or buckets half filled with water, have been
claimed by farmers to be effective during
times of rodent outbreaks.
Exclusion by rat proofing of the storage
house or structure is recognised as an
effective method to reduce post-harvest
losses in rural communities (Hall 1970). The
practice of storing grain in the ceiling or
stores built within traditional houses,
usually with walls constructed with mud, is
common in East Africa. The structural
nature of the houses makes it difficult to
protect the grain from rodent damage
because an effective barrier between the
commodity and rodents cannot be created.
468
Improving storage structures, therefore, will
be the most appropriate long-term strategy
to reduce rodent damage to crops during
storage. This should entail, first, raising the
basement of the structure to one metre above
ground level and, secondly, incorporating a
sleeve or a band of sheet metal that makes
the surface too slippery for rodents to
traverse when fitted closely to the slit.
Further, rodents can be kept away by the
removal of vegetation around the vicinity of
the storage structure, maintaining the stores
in a good state of repair and ensuring the
surrounds have minimal food residues and
other rubbish on which rodents feed.
Although rodent proofing of outdoor
storage structures is very effective in
excluding synanthropic rodents, it has not
been adopted by many rural communities.
Exclusion using rodent proof containers,
such as steel drums and clay pots, is
common in rural areas. These are used for
storage of seed and smaller quantities of
grain. The containers provide adequate
protection against M. musculus and R. rattus
for stored grain in the household.
Environmental manipulation
Habitat manipulation has been encouraged
in a few in East Africa. It assumes that
shelter and food are the main factors
affecting rodent numbers in any given
habitat. This approach also focuses on the
fact that rodents are extremely dependent on
shelter for survival. For numbers to increase
conditions must be favourable for breeding
and survival of the young to reprod uctive
stage (K.D. Taylor 1976, unpublished
report). Disruptions of the environment,
caused by harvesting and ploughing, lead to
a decrease in the shelter available for rodents

and possibly expose them to predators and
reduce their population density. For
example, Taylor and Green (1976) noted that
in Kitale, Kenya, arable fields were unstable
habitats for rodents. In Tanzania, Leirs et al.
(1997) observed a rapid decrease in rodent
abundance immediately after ploughing and
planting but densities increased again after a
few days. It has been suggested that in some
parts of East Africa, the destruction of the
natural environment has displaced many
predators of rodents, while the new
conditions created are favourable for high
rodent population density (K.o. Taylor 1976,
unpublished report).
Environmental manipulation as a rodent
control strategy in East Africa has not been
very successful because it has not been
extensive enough or incorporated as a
component of the small holder farming
system. In order for this strategy to benefit
many farmers it must not be confined to a
few individual fields, as is currently the case.
Where environmental manipulation has
been carried out, it has included one or more
of the following practices.
Grazing, regular bush clearing and grass
cutting
Areas that are regularly cleared of bushes
or that support grazing usually have a lower
carrying capacity for rodent populations
(Green and Taylor 1975), however pasture
land that is not grazed regularly can support
high populations of rodents, especially
granivorous species (A. Ililoticus, R. pumilio,
M. natalensis and Otomys angoniensis). Green
and Taylor (1975) found that cover was an
important population regulating factor for
these species in Kenya and that when cover
was removed it resulted in depletion of
Rodent Pest Management in East Africa
rodent populations. The population density
of A. niloticus is markedly affected by regular
cutting of vegetation, which reduces suitable
habitats for this species (Green and Taylor
1975). Observations in the plague outbreak
villages of Lushoto District Northeast
Tanzania, showed that clearance of bushes,
especially of the perennial Rumex
usambarensis, removed pockets of A. niloticus
populations near houses (R.H. Makundi,
personal observation). It also has been
suggested that the population of A. niloticllS
in Kampala, Uganda, dropped considerably
because the municipal council was
continuously cutting grass around the city
(De Graaf 1981).
Grazing in the fields immediately after
harvest and in the fallow land between
farms might help to destroy vegetation cover
and remove food sources for rodents. This
practice is common in many areas in East
Africa, although the common purpose is not
to control rodents, but to make use of the
stubble and crop remains as animal feed
during the dry season. However, there is a
delicate balance between vegetation cover
and soil erosion. The risks of soil erosion due
to overgrazing must always be considered in
using this approach.
Regular weeding has been reported to
affect rodent popUlation density in
cultivated fields. For example, Mwanjabe
(1993) reported that clean, weeded farms
were less severely attacked and sustained
lower rodent populations throughout the
year than unweeded farms in Chunya
District. A potential widespread
management strategy could be the
application of herbicides to control weeds
over a large area but the economic and
environmental implications do not allow
469
Ecologically-based Rodent Management
this method to be implemented on a wide
scale.
Agricultural practices and land
management strategies
In East Africa, the rains are seasonal with
one or two rain seasons in a year. This also
determines the cropping patterns, with
intensive agricultural practices being found
where the rainfall is well distributed over the
year. Over a vast area the farming system is
composed of small farms that are 0.5-2 ha,
forming a mosaic of fallow land interspersed
with cultivated areas, which is ideal for
maintaining large rodent numbers that
invade crop fields. In these areas, rodent
management strategies require changes in
land practices, producing less fallow patches
that are a refuge for rodents and from where
invasion of crops occurs. To reduce the
damage to crops by rodents, cropping needs
to be synchronised over a large area.
In general there is no common approach
to land management that is aimed at
reducing rodent damage to crops in East
Africa. Encouragement of large block farms
and clearing of headlands reduces the
density of rodents considerably. Taylor
(1968) noted that large, mechanically
cultivated monocultures were often not
highly infested with rodents, but these are
not common at the small holder farmer level.
Other practices like efficient harvesting of
cerea Is and cotton are recommended in most
rodent outbreak areas in Tanzania
(Mkondya 1977) to reduce the available food
resources for rodents.
Many farmers burn their fields in the
aftermath of the harvest or immediately
before planting. This probably changes the
habitat for a short duration, but most likely it
470
has no detrimental effect on the future
population size of rodents because burnt
areas soon have new vegetation and are
reinvaded ra pidly by species from other
areas. For example, Green and Taylor (1975)
reported an increased catch of M. natalensis
following burning in some areas in Kenya.
Presumably, more grass seed becomes
available on the ground after fires, which
probably explains why rodents are attracted
to burnt areas.
In many parts of East Africa, the
harvesting time coincides with the
beginning of the dry season. It is common
for farmers to leave the crop in the fields, a
form of temporary storage, for extended
periods to allow further drying before
threshing. This is common for
cereals, especially maize, sorghum and
millet. A crop left in the field for extended
periods is predisposed to severe attack by
rodents. Among the practices that are
encouraged are early harvesting and storage
of the crops in improved, rodent-proofed
cribs constructed for the dual purpose of
storage and in-storage drying to reduce both
rodent and insect damage of crops.
RODENT MANAGEMENT AND PUBLIC
HEALTH
Rodent infestation in urban areas
Rodent infestation in urban areas poses a
great risk to pub lie health in East Africa and,
therefore, their control is most important. In
the city of Dar es Salaam, Tanzania, > 70% of
households were reported to be infested
with commensal rodents (R. rattus, R.
norvegicus and M. musculus), with highest
infestations in commercial premises dealing
with food handling and processing {Rongo

1993). Studies carried out in the city of
Nairobi, Kenya, similarly showed that there
was a serious rodent infestation problem,
both indoors and outdoors, that required
immediate control activity (Njenga et al.
1993). These problems are also widespread
in small urban areas as pointed out by Lyimo
(1993) for Morogoro municipality in
Tanzania. The unhygienic environment
infested with rodents is ideal for
transmission of zoonotic diseases to man
and domestic animals.
The factors that are responsible for high
urban infestations by rodents and potential
management strategies are summarised in
Table 1.
Control of rodents in response to
plague outbreaks
The management of plague in East Africa
requires an understanding of the ecological,
social and cultural factors which have led to
persistence of the disease for many decades,
as well as why the conventional approaches
to plague control have been ineffective in
some areas. Ecological factors that
contribute towards the spread and
persistence of plague in Lushoto District,
Tanzania, and possibly other plague
outbreak foci in East Africa may be divided
into biotic and abiotic factors (Table 2).
RODENT OUTBREAK FORECASTING
AND MANAGEMENT
Rainfall and the nature of agricultural
activities have a significant effect on the
severity of rodent outbreaks. This is
particularly well understood for M. natalensis
which have breeding seasons that are
strongly influenced by the pattern of rainfall
Rodent Pest Management in East Africa
(Chapman et al. 1959; Taylor 1968; Telford
1989; Leirs 1992; Leirs et al. 1997). Studies in
Tanzania (Telford 1989; Leirs 1992) have
indicated that M.natalensis normally starts
breeding towards the peak of the long rains in
areas where there are two rain seasons
(usually long and short rain seasons) in a
year. However, when the short season rains
are high and extended, animals survive better
and extend their breeding season, mainly due
to an abundance of food and shelter (Leirs et
al. 1996; Mwanjabe and Leirs 1997). This
extended reproduction results in high
recruitment at the beginning of the following
breeding season before the long rain season.
Therefore rodent outbreaks will be
experienced at the time of planting and in
subsequent crop growth stages, resulting in
crop losses. The key factor in this model is the
amount of rain and the duration of the short
rains season. This indicates that early
warning systems based on rainfall data could
be developed to enable farmers to prevent
crop damage and losses. In Tanzania, such a
system was developed towards the end of
1996 (Mwanjabe and Leirs 1997). In order for
such a system to function, the rain patterns
must be well understood and there must be a
reliable surveillance of rodent population
levels. Using this model, the possible
organisation of rodent management activities
in East Africa is shown in Figure 1. With this
management strategy in place it should be
possible to control outbreaks of M. llataiensis
before crop damage occurs. The suggested
organisation of rodent management
strategies in East Africa assumes that the
respective Ministries or Departments will
provide the necessary information and
logistics to enable farmers to control the
outbreaks.
471
Ecologically-based Rodent Management
Table 1.
Factors responsible for high urban infestations by rodents and strategies for their management.
Factors responsible for high rodent Infestations
Lack of efficient collection and proper disposal of
refuse, which creates an abundant supply of food
for rodents.
Lack of well-planned housing schemes creates
suitable shelter and breeding grounds for rodents in
human dwellings.
Lack of long-term rodent management strategies
that are centrally coordinated and implemented.
Tolerance of rodent infestation by the public and
lack of awareness of rodent control measures .
Table 2.
Ecological factors contributing towards the spread and persistence of plague in Lushoto District, Tanzania.
Biotic factors
Abundance of fleas infesting rodents and houses,
with high prevalence of species which are known to
be highly infective. These include Xenopsylla
cheopis, X. brasiliensis, Ctenocephalides felis,
Pulex iritans, Dinopsyllus Iypusus, Nosopsylla spp.
and Leptopsylla aethopica (Njunwa et al. 1989;
Makundi and Kilonzo 1994).
Presence of plague reservoirs, mainly rodents.
These species are Mastomys natalensis, Rattus
rattus, Arvicanthis spp., Grammomys spp.,
Lophuromys favopunctatus and Praomys spp.
Tatera spp., Rhabdomys pumilio, Lemniscomys
spp., Otomys spp. and Aethomys spp . also have
been found to be reservoirs of the disease in other
areas of East Africa (Davis et al. 1968).
High human population pressure increases
interaction between people, wild and commensal
rodents and fleas.
472
Management strategies
Improved sanitation of both residential and
commercial premises through efficient refuse and
garbage collection and their proper disposal.
Proper planning of residential and commercial
areas to reduce potential rodent-attractive habitats.
Incorporating rodent proofing in the construction of
dwellings, warehouses and food storage structures .
Controlled and systematic rodenticide application .
Public health education, especially on the potential
for disease outbreaks involving rodents and
methods of rodent control at household and
community levels.
Abiotic factors
Changes that have occurred after deforestation and
opening the land for agricultural development (R .W.
Bell 1963, unpublished report) have contributed to
colonisation by savanna species of rodents, mainly
M. natalensis and Arvicanthis spp. which are
reservoirs of plague.
The mild temperature and moist conditions (October
-March) enable a drastic increase in abundance and
prevalence of flea vectors, which coincide with
increasing numbers of rodents and plague victims
(Njunwa et al. 1989; Makundi and Kilonzo 1994).
 Rodent Pest Management in East Africa

MONITORING

l
FIELD OBSERVATIONS

Rain patterns

! - - - . . . . Numbers of rodents
Stage of growth of crops

NO ACTION INTERVENTION

REPORTING ---001..... EARLY WARNING

Ministry of Agriculture/Health

PREPAR/.;nONS
Training manpower - farmers. the community, extension staff
Securing and distributing rodenticides
Consideration of all other control options

CONTROL ACTIVITIES
Farmers, extension staff, the community

ASSESSMENT

Figure 1.
Organisational model for management of rodent outbreaks for Eastern Africa (based on models produced by
Leirs et al. 1996 and Mwanjabe and Leirs 1997).

473
Ecologically-based Rodent Management
FUTURE RODENT MANAGEMENT
STRATEGIES IN EAST AFRICA
Success of future rodent management
requires integration of control approaches in
a manner that will reduce the risks of severe
outbreaks and their consequences both in
the agricultural and public health sectors.
The approaches required include:
.. monitoring of rain patterns and rodent
densities in the field;
.. forecasting of outbreaks to enable
necessary measures to be taken to reduce
their effects;
.. efficient inspection of premises and
monitoring of populations to determine
the presence and density of rodents before
and after implementation of control
measures;
.. rodent exclusion, especially in food storage
structures, and where possible, residential
and commercial premises;
.. improved sanitation, both in rural and
urban areas;
.. application of rodenticides to reduce
population levels either by symptomatic or
prophylactic treatments;
.. training of rodent control personnel,
especially those who are involved directly
in the extension service, and raising
community awareness on rodent pest
problems and measures for their control;
.. facilitating timely availability of
rodenticides to farmers when required;
and
.. encouraging farmers themselves to control
rodents on a routine basis, rather than
depending on external (e.g. government)
intervention.
474
REFERENCES
Chapman, B.M., Chapman, R.P. and Robertson,
D.1959. The growth and breeding of the
multimammate rat,Rattus (Mastomys)
natalensis (Smith) in Tanganyika Territory.
Proceedings of Zoological Society, London,
133,1-9.
Davis, D.H.S., Heisch, R. B., McNeill, D. and
Meyer, KP. 1968. Serological survey of
plague in rodents and other small mammals
in Kenya. Transactions of the Royal Society of
Tropical Medicine and Hygiene, 62, 838-861.
De Graaf, G. 1981. The rodents of southern
Africa. Pretoria, Buttersworth, 267p.
Delany,M.J.1975. The rodents of Uganda.
London, Trustees of the British Museum
(Natural History), 165p.
Fiedler, L.A. 1994. Rodent pest management in
eastern Africa. Rome, FAO (Food and
Agriculture Organization of the United
Nations) Plant Production and Protection
Paper No. 123,83p.
Field, A.c. 1975. Seasonal changes in reproduc-
tion, diet and body composition of two
equatorial rodents. East African Wildlife
Journal, 13, 221-235.
Gratz, N. 1997. The burden of rodent-borne
diseases in Africa south of the Sahara. Belgian
Journal of Zoology, 127, 71-84.
Green, M. and Tay tor, KD. 1975. Preliminary
experiments in habitat alteration as a means
of controlling field rodents in Kenya. In:
Hansson, L. and Nilsson, 8., ed., Biocontrol of
rodents. Ecological Bulletin No. 19. Stock-
holm, Swedish Natural Science Research
Council,175-181.
Hall, D.W. 1970. Handling and storage of food
grains in tropical and sub-tropical areas.
Rome, FAO (Food and Agricultural Organi-
zation of the United Nations), 350p.
Harris, W.V.1937. The grey fieldmouse. The East
African Agricultural Journal, 2, 315-318.
Hubbard, c.A. 1972. Observations on life histo-
ries and behaviour of some small rodents
from Tanzania. Zoologica Africana, 7, 419-
449

Key, G.E. 1990a. Pre-harvest crop losses to the
African stripped ground squirrel, Xerus eryth-
ropus, in Kenya. Tropical Pest Management,
36,223-229.
Key, G.E. 1990b. Control of the African striped
ground squirrel, Xerus erythropus, in
In: Davis, L.R and Marsh, RE., ed., Proceed-
ings of the 14th Vertebrate Pest Conference.
Davis, University of California, 99-103.
Kilonzo, B.S. 1984. Studies on the present status
of endemicity, mammalian reservoirs and
flea vectors of plague in Tanzania. PhD
Thesis, University of Oar es Salaam.
Kilonzo, B.s. and Msangi, A.S. 1991.
Mwaluko, G.M.P., Kilama, W.L.,
M.P., Murru, M. and MacPherson,
ed., Health and diseases in Tanzania. London
and New York, Harper CoIlins Academic, 98-
116.
Kilonzo, B.S., Mvena, Z.s.K., Machang'u, RS.
and Mbise, T.J.1997. Preliminaryobserva-
tions on factors responsible for long persist-
ence and continued outbreaks of plague in
Lushoto District, in Tanzania. Acta Tropica,
68,218-227.
Kingdon,J.1974. East African mammals: an atlas
of evolution in Africa, Vo!. II Part B (hares and
rodents). London and New York, Academic
Press, 343-704.
Leirs, H. 1992. Population ecology of Mastomys
natalensis (Smith, 1834). Possible implications
for rodent control in Africa. PhD Thesis,
University of Antwerp, Vt:l)';WlllL.
Leirs, H. 1995. Population ecology of Mastomys
natalensis (Smith, 1834): implications for
rodent control in Africa. Brussels, Belgian
Administration for Development Coopera-
tion,268p.
Leirs, H., Verhagen, R, Sabuni, c.A., Mwanjabe,
P. and Verheyen, W.N.1997. Spatial dynam-
ics of l'v1astormjs natalensis in a field-fallow
mosaic in Tanzania. Belgian Journal of
Zoology, 127,29-38.
Leirs, H., Verhagen, Rand Verheyen, W. 1993.
Productivity of different generations of
Mastomys l1atalensis rats in Tanzania. Oikos,
68,53-60.
Rodent Pest Management in East Africa
Leirs, H., Verhagen, R, Verheyen, W.,
Mwanjabe, p.s. and Mbise, T. 1996. Forecast-
ing rodent outbreaks in Africa: an ecological
basis for Mastomys control in Tanzania.
Journal of Applied Ecology, 33, 937-943.
Leirs, H., Verheyen, W., Michiels, M., Verhagen,
Rand Stuyck, J. 1989. The relation between
rainfall and breeding season of Mastomys
natalensis (Smith, 1834) in Morogoro, Tanza-
nia. Annales de la Societe Royale Zoologique
de Belgique, 119,59-64.
Lund, M. 1977. Report on DANIDA consultant-
ship concerning rodent problems in Chunya
District, Mbeya Region, Tanzania. Unpub-
lished communication presented to the
Ministry of Agriculture, Oar es Salaam,
November 1977.
Lurz, R 1913. Eine Pestepidemie aus Kilimands-
charo in Jahre 1992. Archiv filr Schiffsund
Tropenhygiene, 17, 593-599.
Lyimo, e.S. 1993. An overview of rodent
problems in Morogoro Urban area and their
control. In: Machang'u, RS., ed., Economic
importance and control of rodents in Tanza-
nia. Workshop Proceedings, 6-8 July 1992.
Morogoro, Sokoine University of Agricul-
ture, 146-147.
Makundi, RH. and Kilonzo, BS. 1994. Seasonal
dynamics of rodent fleas and its implication
on control strategies in Lushoto district,
north-eastern Tanzania. Journal of Applied
Entomology, 18, 165-171.
Makundi, RH., Mbise, T.J. and Kilonzo, B.S.
1991. Observations on the role of rodents in
crop losses in Tanzania and control strategies.
Beitrage zur Tropischen Landwirtschaft und
Veterinarmedizin, 4, 465-474.
Mills, J., Bowen, M. and Nichol, S. 1997. African
arenaviruses--coevo]ution between virus
and murid host. Belgian Journal of Zoology,
127,19-28.
Mkondya, C.B. 1975. An appraisal of the Chunya
District rodent outbreak areas in the Eastern
Rukwa Basin. Oar es Salaam, Tanzania,
Ministry of Agriculture Report.
Mkondya, C.B. 1977. Preliminary proposals and
hints for approaches to outbreak evaluation
and control against heavy rodent
475
Ecologically-based Rodent Management
infestations in the Shinyanga outbreak foci in
Tanzania, Dar es Salaam, Tanzania, Ministry
of Agriculture, Crop Development Division.
Mwanjabe, P.S. 1987. Distribution of synan-
thropic rodents in human dwellings in Tanza-
nia. Tanzania Journal of Science, 15, 87-96.
Mwanjabe, p.s. 1990. Outbreak of Mastomys
natalensis in Tanzania. African Small
Mammal Newsletter, 11, 1.
Mwanjabe, P.S. 1993. The role of weeds on
population dynamics of Mastomys natalensis
in Chunya (Lake Rukwa) valley. In:
Machang'u, RS., ed., Economic importance
and control of rodents in Tanzania.
Workshop Proceedings, 6-8 July 1992.
Morogoro, Sokoine University of Agricul-
ture, 34-42.
Mwanjabe, PS. and Leirs, H. 1997. An early
warning system for IPM-based rodent control
in smallholder farming systems in Tanzania.
Belgian Journal of Zoology, 127, 49-58.
Mwanjabe,P.s. andSirima, F.B.1993. Large scale
rodent control in Tanzania: present status. In:
Machang'u, RS., ed., Economic importance
and control of rodents in Tanzania.
Workshop Proceedings, 6-8 July 1992.
Morogoro, Sokoine University of Agricul-
ture, 134-142.
MyHymaki, A. 1987. Control of rodent problems
by use of rodenticides: rationale and
constraints. In: Richards, e.G.J. and Ku, T.Y.,
ed., Control of mammal pests. London,
Taylor and Francis, 83-111.
Neal, B.R 1977. Reproduction of the multimam-
mate rat, Praomys (Mastomys) natalensis
(Smith). Sondedruck aus Zeitschrift fur
Saugetierkunde, 42,
221-231.
476
Neal, B.R 1984. Relationship between feeding
habits, climate and reproduction of small
mammals in Meru Ndtional Park, Kenya.
Journal of Ecology, 22, 195-205.
Njenga, RK, Ngindu, A.M., Abdi, A., Osoro, P.
and Achola, P.s. 1993. Rodent control in
Nairobi city using first and second generation
anticoagulant rodenticides (chlorophacinone
and difenacoum). In: Machang'u, RS., ed.,
Economic importance and control of rodents
in Tanzania. Workshop Proceedings, 6-8 July
1992. Morogoro, Sokoine University of
Agriculture, 143-145.
Njunwa, KJ., Mwaiko, C.L., Kilonzo, B.S. and
Mhina, U.K., 1989. Seasonal patterns of
rodents, fleas and plague status in the
Western U sambara Mountains, Tanzania.
Medical and Veterinary Entomology, 3,17-
22.
Roberts, J.I. 1935. The endemicity of plague in
East Africa. East African Medical Journal, 12,
200-219.
Rongo, L.M.B. 1993. Problems of rodent control
in urban areas and their economic impor-
tance: the case of Oar es Salaam city. In:
Machang'u, R.S., ed., Economic importance
and control of rodents in Tanzania.
Workshop Proceedings, 6-8 July 1992.
Morogoro, Sokoine of Agricul-
ture, 123-133.
Taylor, KD. 1968. An outbreak of rats in agricul-
tural areas of Kenya in 1962. East African
Agricultural and Forestry Journal, 34, 66-77.
Taylor, KO. and Green,M.G.1976. The influence
of rainfall on diet and reproduction in four
African rodent species. Journal of Zoology,
London, 175,453-471.
Telford, S.R1989. Population biology of the
multimammate rat, Praomys (Mastomys)
natalensis at Morogoro, Tanzania, 1981-1985.
Bulletin of the Florida State Museum, Biolog-
ical Sciences, 34, 249-288.
 23. Ecologically-based Rodent Management
in Developing Countries:
Where to Now?

Herwig Leirs, Grant R. Singleton and Lyn A. Hinds

Abstract

This book has catalogued the impacts of rodents at an international scale and
establishes that sustainable and ecologically-compatible management of rodent
pests should be of equal importance to that of insect and weed pest management.
Ecologically-based rodent management (EBRM) is a holistic approach to rodent
management, but must be understood at all levels-by decision-makers, extension
staff, scientists and the end-users. Inevitably more basic and strategic research will
be required before appropriate strategies for EBRM are developed, and different
constraints will apply in different situations when considering its implementation.
Clearly EBRM must be finely targeted to specific agro-ecosystems and pest species .
New technologies, conventional and biotechnological, will be developed over the
next decades. Of critical importance will be full assessment of these before their
adoption and widespread implementation. Training of the next generation of
scientists and managers is also an essential element of EBRM because the solution
to rodent management problems is complex and must be considered as a long term
problem requiring continuing development and application .

Keywords

Ecologically-based rodent management, rodent pests, training,

developing countries

477
Ecologically-based Rodent Management
INTRODUCTION
HIS BOOK HAS achieved, for a
T subset of developing countries, an
audit of the status of rodents as
pests in the 1990s. In this chapter we
consider the progress of 'ecologically-based
rodent management' (EBRM) in developing
countries through drawing on the
contributions provided from Asia (eight
chapters, six countries) and Africa (four
chapters, three regions). We will focus
primarily on management, training and
research issues.
EsTABLISHING A PROFILE FOR
ECOLOGICALLY-BASED
RODENT MANAGEMENT
The importance of the impact of rodents is a
common theme which has emerged from the
chapters contributed from developing
countries. This recognition is felt commonly
to be overdue. The breadth of information
coalesced in this book is immense-for
example, in Indonesia, the depredations of
rodents are equivalent to the amount of rice
required to provide 25 million people 65% of
their annual dietary requirements, while the
impacts of rodents are increasing in Lao
People's Democratic Republic (agriculture),
Madagascar (conservation), Tanzania
(agriculture and health) and Vietnam
(agriculture). We hope that this cataloguing
of impacts at an international scale will assist
in establishing that sustainable and
ecologically compatible management of
rodent pests requires equivalent stature to
that of insect and weed management.
478
The inclusion of research on rodent pests
in the current national plan in China
(see Chapters 12 and 13) and the recent
development of a national rodent pest
laboratory in Indonesia (see Chapter 14),
indicates that the impacts of rodent pests are
beginning to gain appropriate status. In both
these countries there is strong recognition
for the need to develop and apply EBRM.
These programs need to be progressed so
that they become flagships for EBRM in
developing countries.
A common message to emerge from those
developing countries which have embraced
EBRM, is that basic studies of the taxonomy,
ecology and population dynamics of the
target rodents are required before
management strategies can be developed
and implemented (see Chapters 15, 18,21
and 22). The difficulty of convincing
governments in developing countries of the
need for basic research cannot be
underestimated; these countries have major
rodent problems today, whilst the food
demands of society are intensifying, leading
to increasing pressure on biologists to
deliver immediate solutions.
MOVING BEYOND MORTALITY CONTROL
Traditionally, rodent control has focused on
numbers of rodents. An intuitive and simple
response to high rodent numbers became
generally accepted: if there are too many
rats, then kill some. Methods for doing this
killing have been developed over thousands
of years and vary from simple hunting to use
of sophisticated chemical compounds. This
minimalist approach to rodent damage
management is often successfuL but it is not
sustainable. Indeed, population size is the

result of several processes-not only
mortality, but also natality, immigration and
emigration. Focusing only on mortality
neglects the other demographic processes
and possible interactions. It is a strategy
which concentrates on 'reparation' of the
problem, rather than prevention. The holistic
approach, which we have termed
'ecologically-based rodent management'
(Chapter 1; see also Singleton and Brown
1999), at least provides for some optimism
that more sustainable and persistent
strategies can be developed and that rodent
damage is not an insoluble problem.
The chapters on rodent pests in
developing countries indicated a plethora of
pre- and post-harvest problems in rural
areas and storage and disease problems in
both rural and urban areas. Several chapters
also provided examples on how aspects of
demography other than just mortality could
become important elements in the
management of rodent damage (e.g.
Chapters I, 10, 14 and 20). Another
persuasive message to emerge from the
contributions by authors from developing
countries is that there is a strong will by their
governments to adopt more ecologically-
based rodent management. However,
despite a wide interest in more integrated
approaches of rodent control (see, for
example, in Prakash 1988; Buckle and Smith
1994), the development and implementation
of the principles of what we now call EBRM
have been slow and difficult. The reasons
and constraints for this are many, and may
differ between situations. We discuss the
most important of them here.
Where to Now?
CONSTRAINTS ON DEVELOPING AND
IMPLEMENTING EBRM
Basic biological knowledge
The complex biology and behaviour of
rodents is probably the main reason why
integrated pest management in rodent
control has not evolved as fast as that for
insect or weed control. For many rodent
species, even for the common pest species,
knowledge of their population ecology is far
from adequate. In some instances even the
taxonomy of the rodent pests is not well
defined (see Chapters 15 and 18). Although a
farmer may not care which species is
destroying his crop, management strategies
based on ecological characteristics require
knowledge on which species are present,
their habitat use, breeding patterns and
population dynamics.
Sometimes, taxonomic differences are of
paramount importance to a rodent problem.
For example, within the African genus
Mastomys, there are two morphologically
similar species that can be recognised only
by the number of chromosomes. One of
them, Mastomys natalensis, is resistant to, and
a reservoir for, the plague (Yersinia pestis),
while the other, Mastomys coucha, is very
susceptible to the plague (Isaacson et al.
1981). Knowing which species are present in
an urban or semi-urban environment has
major implications for management
strategies, especially when resources are
limited, as is the case in most eastern African
rural communities.
Krebs (Chapter 2) claims that it is now
time to move on from descriptive studies of
rodent pests to experimental work. This
unfortunately is only true for a few pest
species in African and Asian countries
479
Ecologically-based Rodent Management
where sufficient knowledge is available. The
case studies presented in this book on the
population ecology of rodent species from
China, Indonesia, Malaysia, Mali, Tanzania
and Thailand indicate examples of where an
experimental approach would be timely. In
most instances, however, more basic
research is required. Information at
community or landscape level is even more
scarce, but vital for investigations on such
issues as the potential for biological control
using predators.
The development of specialised
techniques like immunocontraception
requires a detailed knowledge of the
reproductive physiology of the target
species and a thorough understanding of its
population ecology (Chapter 10). In
developing countries, such a broad
knowledge base is available for just a few
species of rodent pests.
This shortage of biological knowledge
often reflects a limited understanding of its
importance by funding organisations, the
low number of interested and adequately
trained scientists, and the poor economic
situation of many of the countries that we
discuss here. Added to this is the reality that
most of the research underpinning EBRM in
developing countries has to be done in situ,
where scientists are often working in
isolation from international colleagues
because of language problems, poor
communication infrastructure and! or
international political issues.
Ownership of rodent problems
In many countries, national rodent control
programs or services have been established,
often for good reasons (e.g. Madagascar, see
Chapter 21). Yet, however well meant or
480
successful, some of these programs
contribute to a perception that rodent
control is not the farmers' responsibility but
the government's and, at best, farmers
should participate in control actions directed
by government technicians. Moreover,
rodent problems are often considered to be a
fact of life and a common view held by
farmers is that nothing can be done about
them. As a result, there is often not a strong
commitment by farmers and other end-users
in applying strategies that may require a
long-term involvement.
Few governments in developing or
developed countries are involved in early
tactical management of rodent pests.
Instead, governments generally only become
involved when problems suddenly reach a
level at which emergency actions are
required. After this 'crisis management',
rodent pests generally again fall back to the
bottom of the political agenda. If farmers
take a lead from how governments handle
rodent pests, then it makes it harder to
convince farmers that they need to adopt
early tactical control rather than crisis
management. Unfortunately, this happens at
all levels, be it local, regional, national or
international funding agencies.
An ecologically-based approach also
requires an acceptance that activities in one
place may have a dampening effect on
rodent populations in another place. For
example, EBRM may recommend changes in
land management or in irrigation schemes
where the individual who applies the
actions is not always the one (or the only
one) who will benefit. Socioeconomic,
cultural or political aspects may then
dominate the development of EBRM

strategies (Singleton and Petch 1994; see
Chapters 8 and 14).
Access to Infonnation
EBRM is, by necessity, targeted to specific
agro-ecosystems and pest species. Solutions
are rarely generic; they cannot simply be
transplanted between places. Therefore,
large information campaigns cannot be set
up easily. On top of that, the rather complex
message of EBRM has to compete with
straightforward techniques of pesticide
application. Promotion of the latter is nearly
always driven by commercial objectives, and
therefore often is better organised and has
more financial support.
Where national authorities have a large
influence on rodent control they
will search for national solutions. These will
not always take into account the specific
local conditions that are important in EBRM.
Investment capital
Finally, it should be noted that most EBRM
strategies are, by definition, aiming for long-
term results. In countries such as Indonesia,
Tanzania and Vietnam, most farmers are
risk-aversive, low-capital entrepreneurs.
Thus, many will not invest in solutions
which immediately are more expensive,
even though they may payoff in the long
run. Therefore, EBRM ideally should only be
marginally more expensive to implement
than existing practices, even if these
practices are less effective and more
expensive than EBRM in the long term.
Where to Now?
MAINTAINING THE MOMENTUM FOR
ECOLOGICALLY-BASED
RODENT MANAGEMENT
Training the next generation of
scientists
Although we have discussed a number of
factors which are likely to constrain the
development and implementation of EBRM,
the most basic and important of them is the
lack of biological knowledge. Fortunately,
there is a growing interest among zoologists
and wildlife managers to spend the time and
energy on understanding the population
biology of rodent pests. Many authors of
chapters in this book stand proof of that. In
order to secure this development,
particularly in developing countries, there is
an urgent need for more Young
population ecologists and wildlife managers
need to be educated at academic levels
(MSc., Ph.D.) with a strong emphasiS on
pure science, but in the context of applied,
strategic research. Far too often there has
been a wide gap between basic and applied
research, leading to technologists providing
incremental improvements on methods that
are part of a classical but unimaginative
template, while pure scientists conduct their
research and gather new insights outside
agricultural systems. Today, we have a need
for scientists and managers to be adept in
both fields. The chapters in Section 1 of this
book are by scientists who have a major
focus on basic research but are aware that
their research in ethology (see Chapters 2
and 3), modelling (Chapter 4), ecosystem
dynamics (Chapter 5) and epidemiology
(Chapter 6) could make important
contributions to the management of rodent
pests. However, the findings of their
481
Ecologically-based Rodent Management
research need to be readily accessible to field
practitioners. Often this is not the case and
we are pleased that they were willing to use
this book as one forum to bridge this
communication gap.
Once young scientists have been trained
formally, they require the opportunity to
conduct strategic research. In an EBRM
context this would consist of basic biological
research under the field conditions where
rodents cause problems. This kind of
research will not quickly provide new
techniques and solutions meaning that long
term funding has to be secured.
Also, the results of strategic research
need to be transferred to practitioners. This
technical transfer should focus on new
EBRM methods that have been tested under
field conditions in replicated, experimental
studies conducted on an appropriate scale. It
is very important that at this the
science can be integrated with particular
socio-economic systems. Investigating socio-
cultural aspects and consequences of EBRM
strategies, as well as economic cost-benefit
ratios, is essential.
Short-term thinking for long-term
problems
EBRM requires a long-term approach, both
in development and in application. Much
effort will be needed to convince policy-
makers of this. On the other hand, the
impacts of rodent pests will not abate until
we have developed appropriate EBRM-
techniques. Therefore, there will be
continued pressure for urgent, short-term
solutions, particularly in developing
countries. Consequently, EBRM proponents
must be prepared to develop interim
management practices, even though they are
482
unlikely to be sustainable. An example of
this is the work of the Commonwealth
Scientific and Industrial Research
Organisation (CSIRO) Rodent Research
Group. The focus of this group is EBRM
(Singleton 1997; Singleton and Brown 1999)
and long term research (> 5 years) is
progressing to develop strategies for EBRM.
Mouse plagues have occurred in the past
few years whilst this research has been in
progress. To help combat these outbreaks,
research effort has been deflected
periodically to help develop broad-scale
aerial application of rodenticides (Brown et
al. 1997; Brown and Singleton 1998). This
balance of research focus has meant that
farmers see the research outputs of the
group as relevant to their needs, and this has
maintained their interest in the ongoing
research on EBRM.
Maintaining credibility with farmers
and other end-users
Ecologically-based rodent management will
rarely provide spectacular quantities of dead
rodents. Indeed, the concept of EBRM is to
maintain rodent densities below levels that
cause significant economic losses to
agricultural produce or significant health
problems. On the other hand, as discussed
above, EBRM will require investments in
material or labour. For these reasons,
farmers may be sceptical about applying
such methods. Therefore, a major challenge
is to convince farmers of the benefits of
EBRM.
An equally large challenge is to ensure
that new methods are only promoted when
they have proven their value. There is a risk
that enthusiastic scientists, managers of
scientists, impatient funding organisations

or politicians may promote strategies that
are not yet well tested but that are popular,
politically correct or attract research funds. If
these strategies are less effective than
announced, the loss of credibility would
make it difficult to later convince end-users
to adopt an improved version. A topical
example in Southeast Asia is the trap-barrier
system plus trap crop. This method has good
potential to be the cornerstone for
developing EBRM in rice ecosystems, but a
number of possible weaknesses have been
identified (see Chapter 8). This management
system is being strongly promoted by some
government agencies in the region, yet its
performance has not been assessed at the
village level, nor do we know whether any of
the potential weaknesses are sufficiently
major to require some modification of this
simple technology (Singleton et al. 1998).
Incorporating and integrating
different strategies
As a holistic approach, EBRM will rarely
focus on one single element in the rodent's
biology. Management strategies will
therefore implicate different techniques and
methods. Some may be slight modifications
of existing technologies, others may be more
innovative. The challenge will be to find a
balance, ensuring that focusing too much on
one technique does not compromise the
benefits of an EBRM approach. For example,
the that immunocontraception of
house mice, Mus domesticlls, if successful,
needs to be integrated into an EBPM context
(Chambers et al. 1997; Chapter 10) is a good
example of the type of thinking that is
required.
I
Where to Now?
CONCLUDING COMMENTS
The re-emergence of the importance of
population ecology and an emphasis on
management directed at the agro-ecosystem
level raises hopes that rodent pest
management will begin to match the progress
made by entomologists and botanists in
controlling insect and weed pests. We are
confident that the next decade will see rapid
advances in ecologically-based rodent pest
management. The development of this
strategy will be driven not only by a new
generation of wildlife managers who will have
stronger training in the theory and practice of
population ecology and ecosystem
management, but also by the imperative to
produce'clean and green' produce for
domestic and export markets. Therefore, by
necessity, we will have to develop more
environmentally benign and sustainable
methods for rodent
reducing our sole reliance on rodenticides as
killing agents.
Ecologically-based rodent management
provides the necessmy platform for designing
management strategies which are
environmentally safe. The combined
contributions to this book indicate strongly
that reasonable progress towards the
development of EBRM requires good cross
fertilisation between basic and applied
research on rodent pests, plus the ability to
apply the fruits of this union in a meaningful
sodo-economic context: In developing
countries in particular, advances in EBRM
could translate to significant improvement in
the human condition. If this book can influence
the basic and applied rodent biologists of
today and tomorrow to contribute towards
this outcome then we will be well satisfied.
483
Ecologically-based Rodent Management
ACKNOWLEDGMENTS
We thank Roger Pech and Peter Brown for
their thoughtful comments and discussion
on a previous draft of this chapter. We also
thank Ann, Robyn and David for their
support and understanding during the
preparation of this chapter and the editing of
this book.
REFERENCES
Brown, P.R., Singleton, G.R., Kearns, B. and
Griffiths, J.1997. Evaluation and cost-effec-
tiveness of strychnine for control of wild
house mouse (Mus dornesticlIs) populations in
Victoria. Wildlife Research,24, 159-172.
Brown, P.R. and Singleton, G.R. 1998. Efficacy of
brodifacoum to control house mice in wheat
crops in southern Australia. Crop Protection,
17,345-352.
Buckle, A.P. and Smith, R.H. (ed.) 1994. Rodent
pests and their control. Oxon, u.K., CAB
International, 405p.
Chambers, L.K., Singleton, G.R. and Hood, G.M.
1997. Immunocontraception as a potential
control method of wild rodent populations.
Belgian Journal of Zoology, 127, 145-156.
484
Isaacson, M., Arntzen, L. and Taylor, P. 1981.
Susceptibility of members of the Mastomys
natalensis species complex to experimental
infection with Yersinia pestis. Journal of Infec-
tious Diseases, 144, 80.
Prakash, 1. (ed.) 1988. Rodent pest management.
Boca Raton, Florida, CRC Press, 480p.
Singleton, G.R. 1997. Integrated management of
rodents: a Southeast Asian and Australian
perspective. Belgian Journal of Zoology, 127,
157-169.
Singleton,C.R. and Brown, P.R. 1999. Manage-
ment of mouse plagues in Australia: integra-
tion of population ecology, bio-control and
best farm practice. In: Cowan, D.P. and Feare,
CJ., ed., Advances in vertebrate pest manage-
ment. Fiirth, Filander Verlag, 189-203.
Singleton, G.R. and Petch, D.A. 1994. A review of
the biology and management of rodent pests
in Southeast Asia. Canberra, ACIAR Techni-
cal Reports, 30, 65p.
Singleton, G.R., Sudarmaji and Suriapermana, S.
1998. An experimental field study to evaluate
a trap-barrier system and fumigation for
controlling the rice field rat, Rattus an'entiv-
enter, in rice crops in West Java. Crop Protec-
tion, 17, 55-64.
Index
 Index

A Sarcocystis singaporensis, 338,

349,351-352,354
a-chlorohydrin, 269-271
viral-vectored
Acomys, 168, 410 immunocontraception, 222,
Actinobacillus moniliformis, 91 235-236
active barrier system Black Creek Canal virus, 146, 153
see trap-barrier system Black rat
Akaike information criteria, 397 See Rattus rattus
anticoagulants, 22-23, 33, 44, 163, borreliosis, 453
167-171,173,217,270,279-280, bounty schemes
286,344-345,420,454-455,467 bounty season, 183
see also rodenticides compensatory growth, 182
Apodemus agrarius, 138, 157, 267 East Africa, 19, 24, 149, 390,
Apodemus f/avicollis, 138 460-472,474
Apodemus sylvaticus, 24, 123-124 Indonesia, 182
arenaviruses, 25, 134-135, 137, 139, Laos, 182, 385
142-143, 148, 151-154, 159 Vietnam, 182
Arvicanthis niloticus, 163, 165, 168, Brachyuromys ramirohitra, 444
394, 409, 414, 420-423, 430-431, Brown rat, Norway rat
462 see Rattus norvegicus
arvicoline rodents, 138-139 burrow
assessment of damage, 164, 186-188, active burrow counts, 305, 312
191-192, 295-298

c
B Calomys callosus, 141, 159
baiting
pre-baiting, 62, 167,.363, 369
shyness to baits, 166-168, 217,
225,344,346
see also rodenticides
bamboo flowering, 164
Bandicota indica, 276
behaviour
aggreSSion, 33, 55
conditioned aversive behaviour, 58,
64
intra specific wounding, 69
neophobia, 49, 58, 61-62, 64,
66-68, 70, 72, 167
olfactory communication, 57
opportunistic, 50, 70, 115-116, 136
parasite-altered, 68
sexual,219
taste aversion, 64
Bettongia penicillata, 121, 129
biological control
Capillaria hepatica, 89,92, 107,220
Salmonella, 69, 72, 224
Calomys musculinus, 141, 147
Capillaria hepatica, 89, 92, 220
capture-mark-recapture, 319,322-323,
396, 414
Castor canadensis, 119-120, 129, 131
Castor fiber, 119
chemosterilant, 219, 269
Chitty hypothesis, 95
chronobiology
external factors
climate, 409-411, 414-416.
418,428-429,433
habitats, 409
trophic conditions, 409
trophic resources, 417, 425
rodent pest management, 409-410,
412, 422, 424, 426
circadian clock, 411-412
circadian rhythms, 64, 289, 409
Citellus dauricus, 264
C/ethrionomys glareo/us, 138, 157
Clethrionomys rufocanus, 42
coevolution
host and pathogen, 136, 143-144

487
Ecologically-based Rodent Management

commensal rodents, 18, 470, 472 D

community ecology, 127, 398
damage assessment, 20, 164,
compensation by fecundity, 40-41, 81, 279-280, 340
87,93,107,218,222,232,281,351,
damage by rodents
425 pre-harvest
Cricetomys gambianus, 116, 131, 462 Cambodia, 359
Cricetulus barabensis, 264, 269 China, 264
Cricetulus longicaudatus, 293 Indonesia, 306
Cricetulus triton, 261, 264-265 Laos, 377
chemical control, 269, 279 Madagascar, 449
chemosterilant, 269 Tanzania, 465
control strategies, 268, 270 Vietnam, 320
multiple-capture trap. 269 urban
population dynamics, 265, 269 United States of America, 244
population forecasts, 267 decision analysis, 313-314, 334
reproduction, 265, 269 demography, 24, 50, 146, 396, 398,
storing seeds, 267 479
Dipodomys panamintinus, 121
crops
disease
cacao, 343, 441, 450
human, 68, 134-135, 137-139,
cassava, 339. 441-442,445, 450
143-144, 146-152
cereals, 70, 164-167. 264, 392, habitat association, 134,
450,460,463-464,470 144-145, 149
coffee, 342, 450 reservoir ecology, 143
corn fields, 145. 269, 347-348 risk, 134, 143-146, 148-151
maize, 90, 164, 172, 310, 345, surveillance, 143
399.463-466,470 public health, 51, 72, 135, 145,
oil palm plantations, 21, 23-24, 149,245,252-253,255,442,
169-171,173,341,345 445,456,461-462,470,474
papaya, 450
pineapples, 450
potato, 264, 320, 450, 465 E
rice ecologically-based pest
deepwater, 359 management, 17, 21, 26, 236, 332
irrigated lowland, 186, ecologically-based rodent
305-307, 310, 312-313, 359 management, 17, 22, 280, 405-406,
rainfed lowland. 306, 359, 478-479
372-375, 377-378 economic injury level, 171, 193
ecophysiology, 131, 409-412, 434
rainfed upland, 373
aestivation, 413-414, 426-427
rubber. 151, 164, 339
reproduction, 33, 39-40, 52, 91,
sugar beet, 24
411
sugar cane. 163. 441, 450, 453 water metabolism, 409, 414, 416,
sugaroane, 164, 171-172, 339, 425,434
460,463 seasonal changes, 411,
sweet potato, 465 414, 435
tomatoes, 441, 450 total body water, 414
Cryptosporidium, 67-69 water turnover, 414, 421,
Cynomys ludovicianus, 118, 130 423-425

488
 Index

ecosystem engineering Geographic Information Systems, 149,

allogenic engineering, 113, 115-120 250
autogenic engineering, 117 Geomys bursarius, 117, 130
biotic engineering, 115, 117, 121 Gerbillidae, 418
Ectromelia virus, 94, 215, 225, 227 Gerbillus nigeriae, 418
El Nino Southern Oscillation event, 148 gonadal function, 219, 230, 429, 431,
emerging infectious diseases, 50, 433
135-136 grasslands
endemic rodents, 442, 444-445 carrying capacities of livestock, 286,
environmental contamination, 200 292
environmental effects on populations degradation, 292
deterministic, 396 grass losses, 298
stochastic, 107, 388, 396, 405 management, 300
environmentally friendly system over-grazing, 292
see trap-barrier system Qinghai-Tibet Plateau, 300
extension
farmer field schools, 361
farmer partiCipatory research, 358, H
363, 369-371 habitat manipulation, 163, 165, 181,
244, 345

I F
fence effect, 40
fertility control
habitat use, 17-18, 21, 132, 319,
335-336, 387, 479
Hantavirus disease
Dobrava virus, 138
see compensation by fecundity Hantaan virus, 138
delivery systems Prospect Hill virus, 146
bait-delivered, 224-225 Puumala virus, 138
disseminating, 224-225 reservoirs, 146
epidemiology of vectors, 226 Seoul virus, 138
level of infertility or sterility, 219 Sin Nombre virus, 139
immunocontraception, 234 Heligmosomoides polygyrus. 68
ecologicalconsiderations, 217 herbivory, 118-119, 123,127,
vi ral-vectored , 225 129-130, 132
mechanisms of infertility, 227 Hystrix cristata, 462
food paradigm, 38 Hystrix indica, 121, 129
foraging behaviour, 58, 60, 131
demonstrator effect, 64
poisoned partner effect, 64
post-prandial pattern, 60 immune response, 136, 222, 225-226,
pre-prandial pattern, 60 228-229
fumigation, 183-184, 198 immunocontraception, 217, 225, 234
indicator species, 252
induced-donor habitats, 87
G integrated pest management, 17, 24,
genetic resistance, 33, 247-248, 253 163, 165, 243, 249, 285, 301, 321,
genetically modified organisms 338,354,361,479
animal welfare, 235
public acceptance, 234-236
recombinant viruses, 227, 234 K
species-specificity, 215, 234, 236 K-selection, 418, 425, 427

489
Ecologically-based Rodent Management
L systematic and preventative,
Laguna Negra virus, 144, 153
land management, 470, 480
landscape ecology, 434
landscape heterogeneity, 88, 94-95
Lassa fever, 159
cases, 137
geographical distribution, 141
mortality, 150
reservoir, 141
Leptospira, 69, 72
leptospirosis, 263, 275, 342-343, 354,
453
Leslie matrix, 43
Liomys sa/vini, 122
logistic regression curve, 390
lymphocytic choriomeningitis, 138, 141,
146
M 422
management of rodents
methods and strategies
blanketing, 344, 348
bounties, 178, 181-183, 197
burning houses and
vegetation, 145,466,470
community based, 178, 181,
194-195,269,358,365,386
crisis management. 181, 480
digging, 113, 116, 118,
120-121,132,181,183,203,
211-212,268,271,289-290,
295,316,322-323,325,332,
334,344,367,370,465,467
drives, 178, 180-183, 288,
344,348
habitat manipulation, 244
hunting dogs, 384
hunts. 365
landscaping, 250
metal guard, 468
PICA strategy, 88-89, 198
public participation, 249
rodent-proofing, 250
sanitation, 243-249, 251-254,
315,332,334,405,460,468,
472,474
scaring devices, 180
snares, 384
490
343
tactical, 93, 181, 183, 352,
354,480
trapping, 43, 85, 89-90, 93,
96-98, 124, 141, 145-146,
154,171,178-179,181-184,
188,197,215-216,244,249,
254,272,305,309,312,319,
322-325,327-331,335,341,
344,347,359,365,368,372,
384,399,414,419,434,
453-454, 467-468
urban areas, 244, 248
weeding, 469
Marmota himalayana, 286
Mastomys coucha, 479
Mastomys erythroleucus, 141, 392, 409,
423-424
Mastomys huberti, 141, 409, 415, 420,
Mastomys natalensis, 107, 141, 159,
163-164, 168. 196, 231, 388, 391,
460,462,472.479
population dynamics, 164, 396
mating
mate choice, 57
scramble competition, 55, 68
Meriones spp, 163, 165, 168
Meriones unguiculatus, 23, 203, 264,
281
metapopulations, 44, 120
Microtus brandti, 199-200, 281
burrow systems, 200, 203, 209
ecological management, 211
habitat selection, 199, 203
social behaviour, 206
Microtus californicus, 41, 125
Microtus fortis, 261, 264, 273
control strategies, 275
damage, 275
population forecasts, 274
reproduction, 274
research priorities, 276
Microtus mandarinus, 264
Microtus pennsylvanicus, 117.130, 133,
139
migration of rats, 313
mobility-dependent outbreaks, 434-435
modelling

accuracy, 399
age-structured, 107, 395
compensation, 232
conceptual, 96, 389, 394, 398,
404-405
demographic, 88, 95, 101-102,
106, 395-396, 398-399, 401,
403, 405
fertility control, 83, 96, 107, 231
linear multiple regression, 90
multi-state, 91
multivariate regression models, 403
numerical simulation, 87
predator regulation, 88
predictive, 81-82, 91, 96, 100, 403
reliability, 399
simulation, 87, 100,231,388,399,
403, 406
mortality, 21, 39-40, 43, 60, 89, 91,
126, 137-138, 149-150, 168,
215-218, 230, 235, 266, 269, 294,
305, 312, 350-352, 395, 409-411,
418,420-421,423,425,427-428,
435,451-452,479
mortality-dependent outbreaks, 434-435
mouse plagues, 25, 94-96, 98, 100
multiple factor hypotheses, 37
Muridae, 131, 137, 155, 418, 463
Mus caroli, 310, 341-342
Mus cervicolor, 324, 341-342, 348
Mus domesticus, 81-82,122,124-125,
198,216,392,483
diseases, 94
forecasting outbreaks, 86
hab~atuse,85, 89, 95, 99
integrated management, 93
models, 94
population dynamics, 94
Mus musculus, 136, 159, 165, 168,
244,269,324,341,349,444,460,
462
Muscardinus avellanarius, 122
Myocastor coypus, 22, 119
Myospalax baileyi, 285-286
burrow, 288
feeding habits, 289
habitat, 288
reproductive period, 289
Myospalax tontanieri, 23, 261, 264, 271
control strategies, 271
Index
damage assessment, 271
economic threshold, 272
fumigant, 272
reproduction, 271
research priorities, 273
N
Nesokia indica, 168, 350
Nesomys rutus, 444
nests
mounds, 116-118, 121, 128-129,
132, 290-292, 298-301
non-target species, 21, 126, 219, 223,
234-235, 286
Norway rat
see Rattus norvegicus
Notomys a/exis, 115, 131
o
Ochotona curzoniae, 285-286
ecology, 287
Ochotona daurica, 203, 286
Oligoryzomys /ongicaudatus, 148
Ondatra zibethicus, 119, 182, 197
p
Pergamino virus, 149
Perognathus parvus, 121
Peromyscus boylii, 149
Peromyscus manicu/atus, 38, 50, 121,
139, 152
Peromyscus truei, 121
pest control contractors, 249-251
physical barriers, 178-179, 196, 212
behavioural responses, 195
decision analysis, 193, 316
plastic fences, 180, 184, 365
plastic fences and multi-capture
trap, 184
pitfall traps, 384, 468
Pitymys frene, 286
plague
East Africa, 461
geographic range (Africa), 462
information campaigns, 451, 454
Madagascar, 441, 443, 451
rodent control, 471
seroprevalence, 451
491
Ecologically-based Rodent Management
plant-rodent communities, 202
pollination, 115, 131, 133
population cycles, 37,147,410
population growth rate, 41
population limitation, 33, 37, 44
predator paradigm, 39
predators
Australian kestrel, 88
black-shouldered kite, 88, 91
brown falcon. 88
cat, 67-68, 310, 385
effect on behaviour, 67
fishing cat, 309
Javan mongoose, 309
red fox, 59
small Indian civet, 310
snakes, 115, 309-310, 345, 349,
353-354,369
protozoan, 66, 68, 338, 349, 354
Pseudomys, 115, 122, 127, 130
Pseudomys hermannsburgensis, 115
R reproduction, 276
rate of increase, 89, 95, 98-100, 208,
233,416
Rattus argentiventer, 50,163, 165,168,
195, 197-198, 305-306, 319-320,
325,327-328,332,335,341,359,
377, 383
breeding, 310
diet, 310
ecologically-based management, 306
farm management practices, 313
habitat use, 306
Indonesia
population dynamics, 330
mortality control, 314, 316
nest sites, 305, 312, 314
Vietnam
habitat use, 321, 329, 334
management, 331
population dynamics, 321
reproduction, 324
Rattus bowersi, 341
Rattus colletti, 115
Rattus exulans, 125, 165, 244, 310,
324-325,341,351,359,377
Rattus flavipectus, 165, 324-325
Rattus germaini, 324-325
492
Rattus koratensis, 323, 325, 359
Rattus /osea, 165, 319, 321, 325,
327-328,332,335,341,347,359
Mekong River delta
population dynamiCS, 330
Rattus nitidus, 264, 324-325
Rattus norvegicus, 36,49-50, 125, 138,
153, 157, 164-165, 168, 182, 244,
264, 276, 324-325, 340-341,
350-352.377,444,462
behavioural ecology, 51
colonists, 51
diseases, 49-50, 68-69, 72
feeding behaviour, 72
food supply, 51, 53-54, 70
mating system, 55, 58
movement, 72
reproductive ecology, 51
sex ratio, 51, 54
Rattus rattoides, 261, 264, 276, 278
chemical control, 277-278
control strategies, 277
damage, 277
research priorities, 279
Rattus rattus, 22, 62, 123, 164-165,
168, 244, 306, 310, 324-325,
341-342,352,359,441-442,460,
462, 472
damage to crops, 449-450
population biology, 447
reproduction, 445
zoonoses, 451
Rattus rattus diardii, 310
Rattus rattus mindanensis, 165
Rattus tanezumi, 306
Rattus tiomanicus, 21, 174, 341
Rattus villosissimus, 115, 132
reproduction
dioestrous, 57
social hierarchy. 232
reproduction-dependent outbreak, 427,
431-433
reproductive gametes, 223, 227
Rhabdomys pumi/io, 462, 472
rodenticides
acute, 166-167,344,455
1080,245,286
aluminium phosphide, 272
barium carbonate, 466
 Index

zinc phosphide, 167 social dominance, 56

advantages and disadvantages, 166
anticoagulants, first generation
chlorophacinone, 167, 454
coumachlor, 167
coumatetralyl, 167, 344, 454
diphacinone, 286
pival, 167
warfarin, 167
anticoagulants, second generation
brodifacoum, 163, 169-170,
286,344
bromadiolone, 169, 286
difenacoum, 169, 286
difethialone, 169
flocoumafen, 169, 344
bait stations, 169
lethal feeding period, 168
non-target effect, 173. 174
pre-baiting, 62, 167, 363, 369
pulsed baiting. 170-171, 173
resistant rodents, 169-170
sustained baiting, 169
r-selection, 418, 420, 435
5 systematics of rodents, 155, 410,
6-methoxy-2-benzoxazolinone, 429, 430
Salmonella, 69, 72, 224
Salmonella typhimurium, 224
Salvinia, 21
sanitation, 243-249, 251-254, 315,
332,334,405,460,468,472,474
Schistosoma mansoni, 453
seed dispersers, 121
sewers, 250, 253
sexual activity, 425-427, 429, 431-433
sexual contact, 226
sexual dimorphism, 57
sexual maturation, 398
sexual maturity, 51, 53, 216, 398, 418,
421
sexual secretions, 228
sexual steroids, 414, 421, 434
Sigmodon alstoni, 142, 158
Sigmodon hispidus, 142, 153, 163, 165,
168
slash and burn, shifting cultivation, 383,
385,442,445,447,449-450,456
social behaviour, 22, 25, 33, 40, 136,
219
social hierarchy, 69
social paradigm, 39-40, 43
socioeconomics, 18, 166, 173, 178,
194, 316, 482-483
soil nematodes. 118
Solomys spp., 122
South American Haemorrhagic Fevers
Argentine haemorrhagic fever, 139
Junln virus, 139
mortality, 141
Bolivian hemorrhagic fever, 141
Machupo virus, 141
reservoir, 141
reservoir, 141
Venezuelan haemorrhagic fever, 141
Guanarito virus, 141
Spalax ehrenbergi, 118, 129
spatial dynamics, 23, 33, 43-44, 305
spore dispersers, 113-114, 117,
121-122, 128, 132
Suncus murinus, 158, 309, 444
surveillance of rodent population
density, 143, 172, 174, 252, 471
sustained harvesting, 196
433-434
T
Tachyoryctes splendens, 118
Tatera indica, 168
Taterillus gracilis, 392, 409, 415,
423-424
Taterillu$ petteri, 409, 415, 425-426
Taterillus pygargus, 395
taxonomy, 24,319-320,324,359,410,
433,478-479
Thomomys bottae, 117
Thomoys talpoides, 212
Toxoplasma, 49, 66-69
behavioural effects, 66-68
congenital transmission, 67
training of pest control operators, 248
transmission of zoonoses
age-associated, 146
aggressive encounters, 137
horizontal, 137, 148
mechanisms, 146
precautions, 150

493
Ecologically-based Rodent Management

risk to rodent biologists, 135, 150 Circadian rhythms, 414

rodent to human, 137
rodent to rodent, 137, 148
seasonal, 138-139, 148-149
vertical, 137, 146
trap-barrier, 178-179, 184-185,
187-192,198,323,334,361,364,
483
trap-barrier system plus trap crop
adoption rate, 193-194
benefit-cost, 183-184, 186, 188,
191, 194
comparative performance, 186
halo of protection, 184, 186, 188,
195
optimum size, 186
timing, 191, 196
Trichinella spiralis, 68
circannual, 409
life history, 410-411, 418, 421
w
wildlife management, 25
water metabolism, 409, 414, 416, 425,
434
seasonal changes, 52, 411, 414,
435
total body water, 414
water turnover, 414, 421, 423-425
x
Xerus erythropus, 464

y
u Yersinia pestiS, 442, 452, 461, 479
uplands, 19, 372, 378, 383-386
urban rodent control, 244-246, 248,
251, 253-255 z
Ursus arctos horribilis, 121 zona pellucida proteins, 223, 227-229
zoonoses, 68-69, 72, 134,451

v Zygodontomys brevicauda, 141, 160

vital-cycles, 418, 421-422, 424, 426,

434

494