Environmental Risk Assessment of Genetically Modified Volume 1: A Case Study of Bt Maize in Kenya (Cabi Publishing)

Environ. Risk. Chap 00 Prelims

23/8/04

9:31

Page i

ENVIRONMENTAL RISK ASSESSMENT OF GENETICALLY MODIFIED ORGANISMS Volume 1. A Case Study of Bt Maize in Kenya

Environ. Risk. Chap 00 Prelims

23/8/04

9:31

Page ii

Environ. Risk. Chap 00 Prelims

23/8/04

9:31

Page iii

ENVIRONMENTAL RISK ASSESSMENT OF GENETICALLY MODIFIED ORGANISMS Volume 1. A Case Study of Bt Maize in Kenya

Edited by

A. Hilbeck Geobotanical Institute Swiss Federal Institute of Technology Zurich Switzerland and

D.A. Andow Department of Entomology University of Minnesota USA Series Editors:

A.R. Kapuscinski and P.J. Schei

CABI Publishing

Environ. Risk. Chap 00 Prelims

10/9/04

14:22

Page iv

CABI Publishing is a division of CAB International CABI Publishing CAB International Wallingford Oxfordshire OX10 8DE UK Tel: +44 (0)1491 832111 Fax: +44 (0)1491 833508 E-mail: [email protected] Website: www.cabi-publishing.org

CABI Publishing 875 Massachusetts Avenue 7th Floor Cambridge, MA 02139 USA Tel: +1 617 395 4056 Fax: +1 617 354 6875 E-mail: [email protected]

© CAB International 2004. All rights reserved. No part of this publication may be reproduced in any form or by any means, electronically, mechanically, by photocopying, recording or otherwise, without the prior permission of the copyright owners. A catalogue record for this book is available from the British Library, London, UK. Library of Congress Cataloging-in-Publication Data Environmental risk assessment of genetically modified organisms / edited by A. Hilbeck and D. Andow. p. cm. Includes bibliographical references and index. ISBN 0-85199-861-5 (alk. paper) 1. Crops--Genetic engineering--Environmental aspects. 2. Transgenic plants--Risk assessment. 3. Corn--Genetic engineering--Kenya--Case studies. I. Hilbeck, A. (Angelika) II. Andow, David Alan. III. Title. SB123.57.E59 2004 631.523--dc22 2004007981 ISBN 0 85199 861 5 Disclaimer The findings, interpretations and conclusions expressed in this publication are those of the authors and should not be attributed in any manner to the Global Environment Facility, United Nations Environment Programme, United Nations Development Programme or World Bank. These bodies do not guarantee the accuracy of the data included in this publication and accept no responsibility for any consequence of their use. Recommended citation Hilbeck, A. and Andow, D.A. (eds) (2004) Environmental Risk Assessment of Genetically Modified Organisms: Vol. 1. A Case Study of Bt Maize in Kenya. CAB International, Wallingford, UK. Typeset in 10/12pt Souvenir Light by Columns Design Ltd, Reading. Printed and bound in the UK by Biddles Ltd, King’s Lynn.

Environ. Risk. Chap 00 Prelims

23/8/04

9:31

Page v

Contents

Series Foreword

vii

Contributors

xi

Preface

xv

1

Bt Maize, Risk Assessment and the Kenya Case Study D.A. Andow and A. Hilbeck

2

The Maize Agricultural Context in Kenya L. Muhammad and E. Underwood

3

Problem Formulation and Options Assessment (PFOA) for Genetically Modified Organisms: the Kenya Case Study K.C. Nelson, G. Kibata, L. Muhammad, J.O. Okuro, F. Muyekho, M. Odindo, A. Ely and J.M. Waquil

4

Transgene Locus Structure and Expression of Bt Maize D.A. Andow, D.A. Somers, N. Amugune, F.J.L. Aragão, K. Ghosh, S. Gudu, E. Magiri, W.J. Moar, S. Njihia and E. Osir

5

Biodiversity and Non-target Impacts: a Case Study of Bt Maize in Kenya A.N.E. Birch, R. Wheatley, B. Anyango, S. Arpaia, D. Capalbo, E. Getu Degaga, E. Fontes, P. Kalama, E. Lelmen, G. Løvei, I.S. Melo, F. Muyekho, A. Ngi-Song, D. Ochieno, J. Ogwang, R. Pitelli, T. Schuler, M. Sétamou, S. Sithanantham, J. Smith, N. Van Son, J. Songa, E. Sujii, T.Q. Tan, F.-H. Wan and A. Hilbeck

6

Gene Flow and its Consequences: a Case Study of Bt Maize in Kenya J. Johnston, L. Blancas and A. Borem

1 21

57

83

117

187

v

Environ. Risk. Chap 00 Prelims

23/8/04

9:31

Page vi

vi

Contents

7

Resistance Risks and Management Associated with Bt Maize in Kenya 209 G.P. Fitt, D.A. Andow, Y. Carrière, W.J. Moar, T.H. Schuler, C. Omoto, J. Kanya, M.A. Okech, P. Arama and N.K. Maniania

8

Risk Assessment of Bt Maize in Kenya: Synthesis and Recommendations A. Hilbeck, D.A. Andow, A.N.E. Birch, G.P. Fitt, J. Johnston, K.C. Nelson, E. Osir, J. Songa, E. Underwood and R. Wheatley

Index

251

271

Environ. Risk. Chap 00 Prelims

23/8/04

9:31

Page vii

Series Foreword

The advent of genetically modified organisms (GMOs) offers new options for meeting food and agriculture needs in developing countries, but some GMOs used in agriculture can also affect biodiversity and natural ecosystems. These potential environmental risks and benefits need to be taken into account when making decisions about the use of GMOs. International trade and the unintentional trans-boundary spread of GMOs can also pose environmental risks depending on the national and regional contexts. The complex interactions that can occur between GMOs and the environment heighten the need to strengthen worldwide scientific and technical capacity for assessing and managing environmental risks of GMOs. The Scientific and Technical Advisory Panel (STAP) of the Global Environment Facility (GEF) provides strategic scientific and technical advice on GEF policies, operational strategies and programmes in a number of focal areas, including biodiversity. Its mandate covers inter alia providing a forum for integrating expertise on science and technology, and synthesising, promoting and galvanizing state of the art contributions from the scientific community. The GEF, established in 1991, helps developing countries fund projects and programmes that protect the global environment. GEF grants support projects related to biodiversity, climate change, international waters, land degradation, the ozone layer, and persistent organic pollutants. Global environmental management of GMOs and the strengthening of scientific and technical capacity1 for biosafety will require building policy

1By

‘scientific and technical capacity’ we mean ‘the ability to generate, procure and apply science and technology to identify and solve a problem or problems’ including ‘the generation and use of new knowledge and information as well as techniques to solve problems.’ (Mugabe, J. (2000) Capacity Development Initiative, Scientific and Technical Capacity Development, Needs and Priorities. GEF-UNDP Strategic Partnership, October 2000.) vii

Environ. Risk. Chap 00 Prelims

viii

23/8/04

9:31

Page viii

Series Foreword

and legislative biosafety frameworks. The latter is especially urgent for developing countries, as the Cartagena Protocol on Biosafety of the Convention on Biological Diversity makes clear. And the World Summit on Sustainable Development also identified the importance of improved knowledge transfer to developing countries on biotechnology. This point was also stressed in recent international fora such as, the Norway/UN Conference on Technology Transfer and Capacity Building and the capacity building decisions of the first meeting of the parties to the Cartagena Protocol on Biosafety. The STAP is collaborating with a number of international scientific networks to produce a series of books on scientific and technical aspects of environmental risk assessment of GMOs. This complements the projects being undertaken by the United Nations Environment Programme and the GEF to help developing countries design and implement national biosafety frameworks. The purpose of this series is to provide scientifically peer-reviewed tools that can help developing countries strengthen their own scientific and technical capacity in biosafety of GMOs. Each book in the series will examine a different case study in developing countries. The workshops and writing teams used to produce each book are also capacity building activities in themselves because they bring together scientists from the case-study country, other developing countries and developed countries to analyse and integrate the relevant science and technology into the book. This first book, a case study of Bt maize in Kenya, was written by 52 chapter co-authors, including 24 scientists from Africa as well as scientists and technical experts from Brazil, China, Vietnam, Europe and the USA. A second book, a case study of Bt cotton in Brazil, is in preparation. Each book provides methods and relevant scientific information for risk assessment, rather than drawing conclusions. Relevant organizations in each country will therefore need to conduct their own scientific risk assessments in order to inform their own biosafety decisions. This book is the outcome of a scientific partnership between the STAP and the Global Working Group on Transgenic Organisms in Integrated Pest Management and Biological Control (under auspices of the International Organization for Biological Control). An international Advisory Board provided scientific and strategic advice that led to this book and included representatives from the STAP, the Secretariat of the Convention on Biological Diversity, and numerous agricultural, environmental, academic and governmental organizations, listed in the preface. The STAP then conducted an independent, international and anonymous scientific peer review.

Environ. Risk. Chap 00 Prelims

Series Foreword

23/8/04

9:31

Page ix

ix

We hope that this book will help governments, scientists, potential users of GMOs and civil society organizations in Kenya, other parts of Africa, and other regions of the world to strengthen their understanding of the scientific knowledge and methods that are available for conducting environmental risk assessments of GMOs. We encourage readers to draw their own insights in order to help them devise and conduct robust environmental risk assessments for their own countries. Julia Carabias Chair, Scientific and Technical Advisory Panel, Global Environment Facility Mexico City, Mexico Anne R. Kapuscinski Member, Scientific and Technical Advisory Panel, Global Environment Facility St Paul, Minnesota, USA Peter J. Schei Member, Scientific and Technical Advisory Panel, Global Environment Facility Trondheim, Norway 21 June 2004

Environ. Risk. Chap 00 Prelims

23/8/04

9:31

Page x

Environ. Risk. Chap 00 Prelims

23/8/04

9:31

Page xi

Contributors

Dr Nelson Onzere Amugune, Department of Botany, University of Nairobi, Riverside Drive, Chiromo, PO Box 30197, GPO, Nairobi, Kenya. E-mail: [email protected] Dr David A. Andow, Department of Entomology, University of Minnesota, 219 Hodson Hall, 1980 Folwell Avenue, St Paul, MN 55108, USA. E-mail: [email protected] Dr Beatrice M. Anyango, Department of Botany, University of Nairobi, PO Box 30197, GPO, Nairobi, Kenya. E-mail: [email protected] Dr Francisco J.L. Aragão, EMBRAPA Recursos Genéticos e Biotecnologia (CENARGEN), PqEB W5 Norte, 70770-900 Brasília, DF, Brazil. E-mail: [email protected] Dr Peter Arama, Department of Horticulture, Maseno University, Private Bag, Maseno, Kenya. E-mail: [email protected] Dr Salvatore Arpaia, ENEA – Italian National Agency for New Technologies, Energy and Environment, S.S. 106 Jonica km 419.5, I-75026 Rotondella (MT), Italy. E-mail: [email protected] Dr A. Nick E. Birch, Host Parasite Coevolution, Scottish Crop Research Institute, Invergowrie, Dundee DD2 5DA, UK. E-mail: [email protected] Dr Lesley Blancas, Ecology and Evolutionary Biology, University of California, Irvine, 321 Steinhaus Hall, Irvine, CA 92697-2525, USA. E-mail: [email protected] Dr Aluizio Borem, Department of Agronomy, Federal University of Viçosa, Viçosa 36571-000, MG Brasil. E-mail: [email protected] Dr Deise Capalbo, EMBRAPA Environment, Caiza Postal, Rodovia Campinas Mogi Mirim, Km 127.5 Barrio Tanquinho Velho, CEP 13820-000, Jaguariuna, Brazil. E-mail: [email protected] Dr Yves Carrière, Department of Entomology, The University of Arizona, Forbes Building 410, Tucson, AZ 85721, USA. E-mail: ycarrier@ ag.arizona.edu xi

Environ. Risk. Chap 00 Prelims

xii

23/8/04

9:31

Page xii

Contributors

Mr Adrian Ely, Science and Technology Policy Research, University of Sussex, Mantell Building, Brighton BN1 9RF, UK. E-mail: [email protected] Dr Emana Getu Degaga, Entomologist, Crop Protection, Ethiopian Agricultural Research Organisation (EARO), PO Box 436, 251-2-Nazret, Ethiopia. Email: [email protected] Dr Gary P. Fitt, Long Pocket Laboratories, CSIRO Entomology, 120 Meiers Road, QLD 4068 Indooroopilly, Australia. E-mail: [email protected] Dr Eliana Fontes, EMBRAPA Genetic Resources and Biotechnology, Caixa Postal 02372, Parque Estacao Biologica, Av. W3 Norte-Final, DF 70770901 Brasilia, Brazil. E-mail: [email protected] Dr Kakoli Ghosh, FAO of the UN, Seeds and Plants Genetic Resources Service, Viale DelleTerme di Caracalla, Rome 00100, Lazio, Italy. E-mail: [email protected] Dr Samuel Gudu, Moi University, PO Box 1125, Eldoret, Kenya. E-mail: [email protected] or [email protected] Dr Angelika Hilbeck, Geobotanical Institute, Swiss Federal Institute of Technology, Zurichbergstrasse 38, CH-8044 Zurich, Switzerland. E-mail: [email protected] Dr Jill Johnston, Plant Biology, University of Minnesota, 250 Biological Sciences Center, 1445 Gortner Avenue, St Paul, MN 55108, USA. E-mail: [email protected] Dr Patrick Kalama, Kitale Field Station, Kenya Agricultural Research Institute, PO Box 450, Kitale, Kenya. E-mail: [email protected] Mr James Kanya, International Centre of Insect Physiology and Ecology, PO Box 30772-00100, GPO, Nairobi, Kenya. E-mail: [email protected] Dr Gilbert Kibata, National Agricultural Research Laboratories, Kenya Agricultural Research Institute, PO Box 14733, Nairobi, Kenya. E-mail: [email protected] Mr Elijah Lelmen, International Centre of Insect Physiology and Ecology, PO Box 30772-00100, GPO, Nairobi, Kenya. E-mail: [email protected] Dr Gabor L. Løvei, Department of Crop Protection, Danish Institute of Agricultural Science, Flakkebjerg Research Centre, DK-4200 Slagelse, Denmark. E-mail: [email protected] Dr Esther Magiri, Jomo Kenyatta University of Agriculture and Technology, PO Box 62000, Nairobi, Kenya. E-mail: [email protected] Dr Nguya K. Maniania, International Centre of Insect Physiology and Ecology, PO Box 30772–00100, GPO, Nairobi, Kenya. E-mail: [email protected] Dr Itamar Soares Melo, EMBRAPA Microbiology, Senior Research, Rua Avelina do Amaral 160, Sp.26 Campinas, 13095-130 Brazil. E-mail: [email protected] Dr William J. Moar, Department of Entomology and Plant Pathology, College of Agriculture, Auburn University, 301 Funchess Hall, Auburn, AL 36849, USA. E-mail: [email protected] Dr Lutta Muhammad, Kenya Agricultural Research Institute, National Dryland Farming Research Centre, Katumani, PO Box 1764, Machakos, 2544420828, Kenya. E-mail: [email protected]

Environ. Risk. Chap 00 Prelims

Contributors

23/8/04

9:31

Page xiii

xiii

Dr Francis Muyekho, International Centre of Insect Physiology and Ecology, Field station Mbita Point, PO Box 30772-00100, Nairobi, Kenya. E-mail: [email protected] Dr Kristen C. Nelson, Forest Resources/Fisheries, Wildlife and Conservation Biology, University of Minnesota, 115 Green Hall, 1530 Cleveland Avenue North, St Paul, MN 55108, USA. E-mail: [email protected] Dr Adele Ngi-Song, International Centre of Insect Physiology and Ecology, PO Box 30772-00100, GPO, Nairobi, Kenya. E-mail: [email protected] Dr Samuel Njihia, Kenya Agricultural Research Institute Muguga Research Centre, PO Box 30148, Nairobi, Kenya. E-mail: marcmuguga@ africaonline.co.ke Mr Dennis Wanyama Ochieno, International Centre for Insect Physiology and Ecology, PO Box 30772-00100, GPO, Nairobi, Kenya. E-mail: [email protected] Dr Maurice Oduor Odindo, Community Capacity Building Initiative (CCBI), PO Box 1244-00606, Nairobi, Kenya. E-mail: communityinitiative@ rediffmail.com Dr James Ogwang, Namulonge Agricultural Research Institute, Biocontrol of Pests and Weeds, PO Box 7084, Kampala, Uganda. E-mail: jogwang@ naro-ug.org Dr Matilda Angela Okech, International Centre of Insect Physiology and Ecology, Molecular Biology and Biochemistry Research, PO Box 3077200100, GPO, Nairobi, Kenya. E-mail: [email protected] Dr James Ouma Okuro, Embu Station, Kenya Agricultural Research Institute, PO Box 27, Embu, Kenya. E-mail: [email protected] Dr Celso Omoto, Escola Superior de Agricultura ‘Luiz de Queiroz’, Entomology, Universidade de Sao Paulo, Avenida Pádua Dias 11, Piracicaba, 13418-900, SP, Brazil. E-mail: [email protected] Dr Ellie Osir, Molecular Biology and Biotechnology Department, International Centre for Insect Physiology and Ecology, PO Box 30772-00100, GPO, Nairobi, Kenya. E-mail: [email protected] Dr Robinson A. Pitelli, Universidade Estadual Paulista – UNESP, Departamento de Biologia Aplicada à Agropecuária, Via Paulo D. Castellane, s/n, 14.884900 Jaboticabal, SP, Brazil. E-mail: [email protected] Dr Tanja H. Schuler, Division of Plant and Invertebrate Ecology, Rothamsted Research, Harpenden AL5 2JQ, UK. E-mail: [email protected] Dr Mamoudou Sétamou, Beneficial Insects Research Unit, ARS-USDA, 2413 E Highway 83 Building 200, Weslaco, TX 78596, USA. E-mail: [email protected] Dr S. Sithanantham, International Centre for Insect Physiology and Ecology, PO Box 30772-00100, GPO, Nairobi, Kenya. E-mail: [email protected] Dr Julian Smith, CAB International UK Centre (Egham), Bakeham Lane, Egham, Surrey TW20 9TY, UK. E-mail: [email protected] Dr David A. Somers, Department of Agronomy and Plant Genetics, University of Minnesota, 411 Borlaug Hall, 1991 Upper Buford Circle, St Paul, MN 55108, USA. E-mail: [email protected]

Environ. Risk. Chap 00 Prelims

xiv

23/8/04

9:31

Page xiv

Contributors

Dr Nguyen Van Son, Institute of Agricultural Genetics, Vietnamese Ministry of Agriculture and Rural Development, Head of Scientific Management, International Cooperation Division, Pham Van Dong Str., Tu Liem, Hanoi, Vietnam. E-mail: [email protected] Dr Josephine Songa, Biotechnology Centre, Kenya Agricultural Research Institute, PO Box 14733, Nairobi, Kenya. E-mail: jmsonga@ africaonline.co.ke Dr Edison Ryoiti Sujii, Biological Control, EMBRAPA Genetic Resources and Biotechnology, PqEB Final Av. W5 Norte, Caixa Postal 02372 Brasilia, DF 70 849.970, Brazil. E-mail: [email protected] Dr Tran Quang Tan, National Institute for Plant Protection, Vietnamese Ministry of Agriculture and Rural Development, Dong ngac, Tu Liem, Hanoi, Vietnam. E-mail: [email protected] Ms Evelyn Underwood, IOBC GMO Guidelines Project, Geobotanical Institute, Swiss Federal Institute of Technology, Zurichbergstrasse 38, CH-8044 Zürich, Switzerland. E-mail: [email protected] Dr Fang Hao Wan, Biological Control Institute, Chinese Academy of Agricultural Sciences, 12, Zhong-Guan-Cun, Nan-Da-Jie, 100081 Beijing, China. E-mail: [email protected] or [email protected] Dr José Magid Waquil, EMBRAPA Centro Nacional de Pesquisa de Milho e Sorgo, Rodovia MG 424 Km 65 S/N, Caixa Postal 151, Sete Lagoas 35701-970, Brazil. E-mail: [email protected] Dr R. Wheatley, Plant–Soil Interface, Scottish Crop Research Institute, Invergowrie, Dundee DD2 5DA, UK. E-mail: [email protected]

Environ. Risk. Chap 00 Prelims

23/8/04

9:31

Page xv

Preface

The Cartagena Protocol on Biosafety of Living Modified Organisms (Biosafety Protocol) under the Convention on Biodiversity (CBD) and many other international forums identify a need in both developing and developed countries for comprehensive, transparent, scientific methods for meaningful pre-release testing and post-release monitoring of transgenic plants to ensure their environmental safety and sustainable use. This need has been repeatedly expressed by both the private and public sector (CBD, 2000). For example, Chapter 16 of Agenda 21 recognizes that the maximum benefits of genetically modified crops can be achieved only if appropriate biosafety procedures are in place and the relevant capacities to implement them are acquired (UN DSD, 1999). There is wide recognition that the regulatory and scientific capacity for conducting risk assessments needs to be strengthened worldwide. Most importantly, the needs of developing countries for capacity building and policy development must be addressed. Article 22 of the Biosafety Protocol requires that parties shall cooperate in the development and/or strengthening of human resources and institutional capacities in biosafety. It is also recognized that this capacity building activity will require significant investments, as many countries do not have the capability to make independent risk assessments or to evaluate independently submitted risk assessments on biosafety. This Kenya case study is a product of the GMO Guidelines Project, ‘Development of International Scientific Biosafety Testing Guidelines for Transgenic Plants’. This Project was launched by scientists of the Global Working Group on ‘Transgenic Organisms in Integrated Pest Management and Biological Control’, under the aegis of the International Organization for Biological Control (IOBC). It is funded by the Swiss Agency for Development and Cooperation (SDC) as a part of the Swiss governments’ commitment to the Biosafety Protocol. The project is advised by a 20-member advisory board representing a wide array of agricultural, environmental and development xv

Environ. Risk. Chap 00 Prelims

xvi

23/8/04

9:31

Page xvi

Preface

organizations from around the world. The board members function both as scientific advisors and as international mediators to the policy arena and relevant decision-making processes. The Project addresses the environmental and agricultural impacts of transgenic organisms, and does not evaluate human health impacts or ethical implications. Because most of the presently available transgenic plants produce a novel gene product used in pest control, the Project has focused on these kinds of transgenic crop plants. This focus is propitious, because there is more information available on this class of transgenic organisms than any other class, plus it is possible to mobilize considerable expertise in this area. One of the aims of the Project is to improve the capacity of scientists to support environmental risk assessment of transgenic crop plants in each of their countries. To accomplish this, the project concentrates on scientist-to-scientist exchange, because these personal connections are likely to persist over time. To leverage these efforts, the project focuses on a few countries with reasonably developed scientific infrastructures, a desire to develop the scientific basis of risk assessment, and a need to do so. By strengthening the scientific capacities for risk assessment in these countries, expertise should be able to diffuse more readily to neighbouring countries. Kenya was the first focal country of the Project and work conducted in Kenya forms the basis for this book. Among the countries of sub-Saharan Africa, Kenya and South Africa have two of the better-developed scientific infrastructures related to agriculture and the environment. Kenya became the first case study in part because of interest in using Bt maize in its production system. The other main aim of the project is to develop risk assessment methods for use in developing countries. The Biosafety Protocol and the EU Directive on release of genetically modified (GM) plants specify that risk assessment should be conducted on a case-by-case basis. A case-by-case approach is necessary because there is insufficient experience available to allow aggregate analysis and assessment. Each GM plant and ecosystem must be looked at individually, because the relevant questions will differ on a case-by-case and country-by-country basis. Consequently, it is premature to propose general risk assessment guidelines, but it may be possible to develop general, robust approaches by extrapolation from detailed case studies. Hence, the project has focused on developing scientifically sound, transparent case studies to instantiate the principles of risk assessment. This book is the final product from the first case study of the project, Bt maize in Kenya. We would like to thank the International Centre for Insect Physiology and Ecology (ICIPE), and its Director, Dr Hans Herren, for hosting the workshop on which this book is based, and the Kenya Agricultural Research Institute (KARI) for their support of the workshop and project. Dr Josephine Songa (KARI) and Dr Ellie Onyango Osir (ICIPE) were instrumental in setting up the Workshop in Nairobi, Kenya. Without their assistance, we would not have been able to have the considerable Kenyan expertise at the Workshop. We would also like to thank Professor William Overholt (Chapters 2 and 5), Dr Danny Llewellyn (Chapter 4), Dr Gavin Ramsay (Chapter 6), Professor Norman Ellstrand, Dr

Environ. Risk. Chap 00 Prelims

Preface

23/8/04

9:31

Page xvii

xvii

Remy Pasquet (Chapter 6) and Professor Mike Cohen (Chapter 7) for their substantial suggestions. We thank the six reviewers associated with the STAP review process, including Drs David Bergvinson and David Hoisington (CIMMYT), Bruno Le Ru and Fritz Schultess (ICIPE) who provided important information in their signed reviews. We would also like to thank Dr Ana Cristina Brasileiro, Dr Weber Amaral and all of the members of the Advisory Board to the project, who read and commented on an early draft of this book. These Board members are Dr Ana Lucia Assad (Brazil Ministry of Science and Technology), Dr Joel Cohen (ISNAR), Dr Les E. Ehler (President, IOBC), Dr Les G. Firbank (Coordinator, UK Field Trials Programme), Dr Helmut Gaugitsch (Austrian Federal Environment Agency), Dr Hans Herren (Director General, ICIPE, Kenya), Mr Ryan Hill (Biosafety Protocol Secretariat, CBD), Dr Katharina Jenny (Swiss Development Cooperation), Dr Anne Kapuscinski (UNEP-GEF Scientific and Technical Advisory Panel), Dr Peter Kenmore (Director, FAO Global IPM Facility), Dr Chris Ngichabe (Kenya Agricultural Research Institute), Dr William Padolina (International Rice Research Institute), Dr Francois Pythoud (Swiss Agency for Environment, Forest and Landscape), Dr Maria José Sampaio (EMBRAPA), Dr Julian Smith (CABI), Dr Wilson Songa (Kenya Plant Health Inspection Service), Dr Braulio Souza de Dias (Brazil Environment Ministry), Dr Sutat Sriwatanapongse (Thailand Biotechnology Centre), Dr Hermann Waibel (University of Hannover) and Dr Jing Yuan Xia (Chinese Ministry of Agriculture). We also acknowledge the considerable efforts of the Steering Committee of the Project, without whom the Project would not exist. These members are Drs Nick Birch, B.B. Bong, Deise Capalbo, Gary Fitt, Eliana Fontes, K.L. Heong, Jill Johnston, Kristen Nelson, Ellie Osir, Allison Snow, David Somers, Josephine Songa and FangHao Wan. Most importantly, we acknowledge Evelyn Underwood, without whose help the Kenya Workshop and this book would not have been possible. The overall task of the project is large and complex, and we invite the involvement of all public sector scientists and encourage interested researchers to contact us via our website. We are presently discussing how to involve both the private sector and the non-government organizations to further improve the quality of the project’s products. The more people who become involved and engage in developing these products, the better they will become and the more widely they will be recognized. Interested public sector scientists can enrol in the project at www.gmo-guidelines.info Angelika Hilbeck Zürich, Switzerland David Andow St Paul, Minnesota, USA December 2003

Environ. Risk. Chap 00 Prelims

xviii

23/8/04

9:31

Page xviii

Preface

References CBD (2000) Cartagena Protocol in Biosafety to the Convention on Biological Diversity: text and annexes. Secretariat of the Convention on Biological Diversity, Montreal, www.biodiv.org/doc/legal/cartagena-protocol-en-pdf (accessed 1 December 2003). UN DSD (1999) United Nations Division for Sustainable Development. Agenda 21 Chapter 16: Environmentally sound management of biotechnology, www.un.org/esa/ sustdev/agenda21chapter16.htm (accessed 1 December 2003).

Environ. Risk. Chap 01

1

23/8/04

9:31

Page 1

Bt Maize, Risk Assessment and the Kenya Case Study D.A. ANDOW AND A. HILBECK Corresponding author: Dr David Andow, Department of Entomology and Center for Community Genetics, University of Minnesota, St Paul, MN 55108, USA. E-mail: [email protected]

In September 2003, the Cartagena Protocol on Biosafety went into force, calling for scientific risk assessments of genetically modified organisms (GMOs) prior to their introduction into the environment.1 The use and utility of GMOs has been hotly debated for about 15 years and one of the purposes of the Protocol is to establish the basis on which these controversial organisms will be evaluated. This book provides a detailed examination of one controversial GMO, Bt maize, in its proposed application in Kenya. We develop components of a scientific risk assessment process, which are consistent with that called for by the Protocol, and illustrate how they can be applied to the case study. In our view, risk assessment is not a decision-making process; it is an activity that supports a decision-making process. Indeed, in this book we do not even attempt a full-blown risk assessment of Bt maize in Kenya. This would take far more pages than this book could hold. Instead, through this case study, we illustrate the scientific and logical process by which risk assessment can be conducted. At its best, science is both transparent and rigorous. Science is transparent when all participants and observers have full access to the evidence, can verify the methods used to gather the evidence, and can understand the processes used to analyse and interpret the evidence. When conducted transparently, science invites and addresses criticism while making rapid progress in measured sure steps. Science is rigorous when the interpretation follows logically from the evidence and competing interpretations are proven inadequate. Indeed, it is its logical rigor and exacting methodological standards that gives science the perceived value-neutrality and predictive power that is assumed to provide the best basis for risk assessment. 1The Biosafety Protocol addresses ‘living modified organisms’ (LMOs), but for the purpose of this book, our use of GMO is equivalent to the definition of LMO in the Protocol.

© CAB International 2004. Environmental Risk Assessment of Genetically Modified Organisms: Vol. 1. A Case Study of Bt Maize in Kenya (eds A. Hilbeck and D.A. Andow)

1

Environ. Risk. Chap 01

23/8/04

9:31

Page 2

2

D.A. Andow and A. Hilbeck

In this case study, we develop transparent and rigorous scientific methodologies for assessing potential environmental risks of Bt maize in Kenya. Although human health risks are also important to assess, the expertise we mobilized on this case study allowed us to focus on the potential environmental risks. By focusing in this way, we were able to make considerable progress toward developing the needed protocols for the environmental risk assessments called for by the Biosafety Protocol. Bt maize is produced by transferring a gene that codes for a toxin from a soildwelling bacterium, Bacillus thuringiensis (Bt) by transgenesis. This chapter provides some of the scientific background to understand the Kenya Bt maize case study. We first give a brief overview of transgenesis, summarize the production of transgenic crops worldwide, describe the various kinds of Bt maize that have been grown commercially throughout the world, and give a brief overview of the complex subject of maize breeding. We then describe a commonly used model of risk assessment, discuss the evidential standards that are used typically in risk assessment, and then indicate how these will be applied throughout the case study. We finish this chapter with a description of two significant valuation issues, biodiversity and appropriate scientific controls, and then indicate the potential scope of the Kenya Bt maize case study, discussing the timeliness and geographical generality of this case study, and outlining the rest of the contents of the book.

Transgenesis and Bt Maize Use of transgenic crops worldwide The first commercial transgenic crops were planted in China during the early 1990s, primarily virus-resistant tobacco and tomato. During 1995, numerous transgenic crops were commercialized, and by 1996, the USA was planting more transgenic crops than any other country in the world (Table 1.1A). The large decrease in planting area in China during 1998 (Table 1.1A) was caused by a collapse of the international markets for transgenic tobacco and tomato. The USA, Canada, China and Argentina dominate world production of transgenic crops; indeed, about 68% of the world’s transgenic crops were planted in the USA alone during 2001. Initially, a large variety of transgenic crops were planted commercially (Table 1.1B). By 1998, however, four crops dominated – soybean, maize, cotton and oilseed rape – and these continue to dominate. The primary traits are herbicide tolerance and insect resistance (Table 1.1C). In 1999–2001, herbicide-tolerant soybeans, Bt maize, herbicide-tolerant maize, Bt cotton, herbicide-tolerant cotton and herbicide-tolerant oilseed rape accounted for over 99% of the commercial transgenic crops grown worldwide. Although Bt genes have been incorporated into broccoli, cabbage, oilseed rape, cotton, maize, aubergine, poplar, potato, soybean, tobacco and tomato, the only crops planted on significant commercial hectarage in 2001 were Bt maize and Bt cotton. After being introduced during 1995, the cropping area of

Environ. Risk. Chap 01

23/8/04

9:31

Page 3

Bt Maize and Risk Assessment

3

Table 1.1. Area of transgenic crops in the world from 1996 to 1999 (in millions of ha).

A. Country USA China Canada Argentina Australia Mexico South Africa B. Crop Soybean Maize Cotton Oilseed rape Tobacco Tomato Potato Papaya Squash C. Trait Herbicide tolerance Insect resistance Virus resistance Quality traits D. Total

1996

1997

1998

1999

2000

2001

1.5 1.1 0.1 0.1 + + 0

8.1 1.8 1.3 1.4 + + 0

20.5 + 2.8 4.3 0.1 + 0

28.7 0.3 4.0 6.7 0.1 + +

30.3 0.5 3.0 10.0 0.2 + 0.2

35.7 1.5 3.2 11.8 0.2 + 0.2

0.5 0.3 0.8 0.1 1.0 0.1 + 0 0

5.1 3.2 1.4 1.3 1.7 0.1 + 0 0

14.5 8.3 2.5 2.5 + + + 0 0

21.6 11.1 3.7 3.4 + + + 0 0

25.8 10.3 5.3 2.8 0 0 + + +

33.3 9.8 6.8 2.7 0 0 + + +

0.7 1.0 1.1 + 2.8

6.9 4.7 1.8 + 12.8

20.1 8.0 + + 27.8

31.0 11.8 + + 39.9

35.9 11.5 + + 44.2

44.8 12.0 + + 52.6

Data from James (1997, 1998, 1999, 2001, 2002). + Indicates that <100,000 ha were grown. The first commercial crops were planted in China during the early 1990s. The first commercial production in the USA was tomatoes during 1994. Several crops were first commercialized during 1995, including Bt maize. A, Area by country. During 1999, small areas of transgenic crops were also grown in Spain, France, Portugal, Romania and Ukraine. During 2000, small areas were also grown in Spain, France, Romania, Bulgaria, Germany and Uruguay. During 2001, small areas were also grown in Spain, Romania, Bulgaria, Germany, Indonesia and Uruguay. B, Area by crop. Several minor crops are not listed. C, Area by transgenic trait. This does not always sum to the worldwide total because some crops have more than one transgenic trait. D, Total area worldwide.

these transgenic crops has grown substantially in the USA (Table 1.2). By 1999, Bt maize was grown on 9.6 million ha, fully one-quarter of the total transgenic crop area worldwide. Bt cotton lagged behind substantially in total area because about five times more maize than cotton is grown in the USA. Clearly, Bt maize in the USA is one of the dominant transgenic crops in the world today.

Transgenesis The development of transgenic techniques allowed the isolation, sequencing and transformation of genes into plants, providing a novel mechanism for creating genetic diversity. Transgenesis can be used to accomplish two goals. A

Environ. Risk. Chap 01

23/8/04

9:31

Page 4

4

D.A. Andow and A. Hilbeck

Table 1.2. Area of Bt maize, Bt cotton and Bt potato; 1995–1997 and 2000–2003 for the USA only; 1998–1999 for the world, although nearly all was planted in the USA.

Bt maize Bt cotton Bt potato

1995

1996

1997

1998

1999

2000

2001

2002

2003

+ + +

0.3 0.7 3.7

2.8 0.8 10.0

6.7 na 20.0

9.6 2.2 23.0

6.3 2.2 na

5.8 2.4 na

7.7 2.0 0

9.3 2.3 0

Maize and cotton data are in million ha. Potato data are in thousand ha. Data for 1995–1999 from James (1997, 1998, 1999) and for 2000–2003 from NASS (2000, 2002, 2003). na means data were not available.

gene that controls a valuable trait can be isolated and introduced into another plant without simultaneously carrying along thousands of other genes. This can avoid potential genetic drag associated with these extra genes. In addition, genes can be introduced from species that are not sexually compatible with the recipient (horizontal gene transfer). Presently, there are two common procedures to introduce transgenes into plant cells (Birch, 1997). One makes use of the Ti-plasmid of Agrobacterium tumefaciens to transfer the gene as part of this plasmid’s DNA (referred to as TDNA). This plasmid is capable of transforming most dicotyledonous plant cells, and has seen only limited use in monocots, such as maize. The second method, and one most commonly used to transform maize, uses the biolistic method, in which a metal particle or fibre pierces the plant cell wall and carries the cloned DNA into the nucleus. The first application of this technique used tungsten particles discharged by a 22-calibre cartridge (Sanford et al., 1993), a device known as a gene gun. Biolistic transformation was the first technique used to transform monocots, such as maize. Transgenes must be inserted into totipotent cells so that they end up in the plant germ line and can be transmitted to progeny. For this reason, tissues capable of generating somatic embryos, such the cotyledons of young embryos, are routinely used for biolistic transformation. Because only a few of the cells are transformed and the transgene is usually not detectable in transformed cells in cell culture, a selectable marker transgene is often included to allow identification of cells carrying the desired transgene. Thus, both the target and marker transgene end up in these transformed cells. It is common for two or more foreign DNA molecules to combine end to end in the plant cell prior to insertion into the nuclear genome. Many transformed plants contain multiple copies of a transgene, which occur in tandem or inverted repeats. In some cases, fragments of host DNA can become interspersed between the duplicated copies of the transgene, creating complex loci (Kohli et al., 1998; Pawlowski and Somers, 1998). This will be discussed in greater detail in Andow et al., Chapter 4, this volume. Multiple transgene insertions and complex insertions at single loci are highly correlated with transgene silencing, which can lead to inactivation of the transgene and possibly other genes in the genome with sequence homology to regions of the transgenic DNA (Birch, 1997). Transgene silencing may not become detectable until several generations beyond the initial transformation event.

Environ. Risk. Chap 01

23/8/04

9:31

Bt Maize and Risk Assessment

Page 5

5

In summary, transgenesis is a new method for creating genetic diversity in organisms with the potential to incorporate novel traits into plants. It is these novel traits that are the focus of risk assessment. This case study focuses on Bt maize to ground the scientific discussion of risk assessment, so that practical approaches to risk assessment of these novel crop traits can be developed in Kenya.

Maize breeding The first step in generating a transgenic maize variety is to transform maize plant cells with the target transgene, as described above. While in cell culture, these cells are selected, usually using a selectable marker transgene, and selected cells are regenerated into whole plants. Regenerated plants are transplanted into the greenhouse so that they can be self-pollinated and/or crossed to another maize line. These progeny are selected for desirable traits, and are backcrossed into a useful inbred line or population and developed into varieties that can be released to farmers (see Lamkey, 2004, for additional details). The most restrictive step in the transformation process is regenerating a plant from cell culture. Very few maize cultivars can be used to regenerate a plant from cell culture (Armstrong et al., 1991), and most maize transformation labs use the same basic maize germplasm for plant regeneration. This germplasm is not useful for maize production. Consequently, efforts are made immediately to move the transgene into useful maize genetic backgrounds. Regenerated plants are self-pollinated, backcrossed to an elite line and crossed to a hybrid when sufficient plants and pollen are available. All of these crosses typically occur in the greenhouse. They are self-pollinated to increase the seed of the original plant that contains the transgene; they are backcrossed to transfer the transgene into elite breeding material (Fig. 1.1); they are crossed to a hybrid to produce enough grain to be able to extract enough protein to characterize the transgene product. These progeny are evaluated to determine how the transgene is inherited genetically and to verify that the intended transgenic protein product is being produced. Desirable plants are retained for additional development. Backcrossing is a breeding scheme designed to move genes from one maize line (called the donor) into another line (called the recipient) without otherwise genetically changing the recipient. Backcrossing begins by making a cross between the recipient and donor. The progeny produced from this and later generations are repeatedly crossed with the recipient. Each cross is called a backcross, and with each generation of backcrossing, the donor genome is reduced by half. The number of backcrosses determines how closely the progeny resemble the recipient. If there are five backcrosses, the progeny differ from the recipient by about 3% of the genome; if there are seven backcrosses, this difference is about 0.8%. When the recipient is already used as an elite inbred or open-pollinated variety (OPV) there is reduced need to field test for agronomic performance because the final product resembles the recipient line.

Environ. Risk. Chap 01

23/8/04

9:31

Page 6

6

D.A. Andow and A. Hilbeck

After backcrossing is completed, the newly developed transgenic line is crossed or increased to evaluate agronomic performance (Fig. 1.1). The number of locations and replications per location will vary with breeding programmes, and if performance is adequate, the transgenic inbred or OPV might be released quickly as breeder seed.

Breeders make a cross to

Transformable

make a population of maize

maize line

with a new genetic composition Transformation Transformed maize line

New population or old inbred line Backcrossing

Population or inbred has transgene

Transgene is not added

Selection and/or selfing

Selection and/or selfing Breeder seed

Parent seed

Hybrid seed

Open-pollinated seed

Fig. 1.1. Schematic outline of maize breeding seed production incorporating a transgene. The lower box represents the steps to produce seed for actual use by farmers.

Environ. Risk. Chap 01

23/8/04

9:31

Bt Maize and Risk Assessment

Page 7

7

Both hybrids and OPVs are produced and used in Kenya, and the preferred varieties for human consumption are the white maize varieties. A hybrid is a variety that is the progeny from two genetically different parents. In hybrid maize, each parent is genetically uniform and different from the other parent, so all of their hybrid progeny are similar genetically. However, progeny of the hybrid are different from the hybrid parent, so if hybrid seed is saved for future production, yields typically decline. This yield decline can be large for some hybrids (e.g. single cross hybrids in developed countries) and smaller for others. An OPV is a variety produced from a population of genetically variable individuals that mate with each other. Instead of creating genetically uniform seed so that each seed is genetically the same as another seed, an OPV is a group of genetically different individuals that together perform well in fields to produce a good maize crop. In contrast to hybrids, however, most OPVs can be saved for the next crop for multiple generations without losing much yield. OPVs with robust genetic variation lose little yield, but improved OPVs with a narrow genetic base can lose significant yield when they are saved. Most maize hybrids are single cross hybrids and are produced from breeder seed by two steps (Fig. 1.1). First, breeder seed is increased by selfing so that there is sufficient quantity of parent seed to produce the desired hybrids. The two inbred parents are then crossed to create hybrid offspring with desired traits. Typically, a Bt transgene is incorporated into a maize hybrid by backcrossing the transgene into only one of the inbred parents, usually the seed parent. Hence, a Bt maize hybrid is hemizygous for the transgene (having one copy of the transgene). If the Bt gene is backcrossed into both inbred parents, then a homozygous Bt maize hybrid can be produced. Maize OPVs can be produced in several ways. One common way is to start with a diverse, desirable pool of genes and by mass selection produce an improved population that maintains traits of interest (such as yield) for multiple generations. The Bt transgene would be incorporated into this OPV by recurrent backcrossing. However, because the OPV is genetically diverse, backcrossing will increase the frequency of the Bt gene in the population, but not lead immediately to 100% prevalence in the population. This means that considerable screening for the transgene must be done during each generation, so that the transgenic OPV ends up with a sufficiently high transgene frequency. Ideally, all plants in a Bt OPV will have at least one Bt transgene, and in general, an OPV will be a mixture of hemizygous and homozygous (having two copies of the transgene) plants. Some of these difficulties can be alleviated by using very large populations in the backcrosses. It is possible to produce a homozygous Bt OPV if the genetic basis of the variety is relatively narrow and genotypic screens are used in combination with the more standard phenotypic screens. Because it is easier to produce a Bt hemizygous hybrid than a genetically diverse Bt homozygous OPV, it is likely that some Bt hybrids will be available earlier than the more robust Bt OPVs.

Bt maize The first generation of commercial Bt maize was designed to protect maize from damage by the larvae of European corn borer, Ostrinia nubilalis

Environ. Risk. Chap 01

8

23/8/04

9:31

Page 8

D.A. Andow and A. Hilbeck

(Hübner) [Lepidoptera: Crambidae] (Ostlie et al., 1997; Rice and Pilcher, 1998), and southwestern corn borer, Diatraea grandiosella Dyar [Lepidoptera: Crambidae] (Porter et al., 2000), important pests in the US corn belt. Bt maize can in principle provide farmers with significant economic benefits. These perceived benefits have resulted in a rapid adoption by US farmers. Other moth pests of maize in the USA, especially the noctuids fall armyworm Spodoptera frugiperda (J.E. Smith), corn earworm Helicoverpa zea (Boddie), black cutworm Agrotis ipsilon (Hufnagel), armyworm Pseudaletia unipunctata (Haworth) and stalk borer Papaipema nebris (Guenée), are not effectively controlled by Bt maize. Moreover, this generation of Bt maize has no economic effect on non-moth pests. Consequently, Bt maize does not eliminate the need for farmers to remain vigilant in monitoring insect pests, and in some cases may not affect insecticide use patterns. Recently, a new kind of Bt maize was commercialized that was designed to protect maize from damage by corn rootworms, Diabrotica virgifera virgifera LeConte and Diabrotica barbari [Coleoptera: Chrysomelidae]. This new kind of Bt maize is not being considered for introduction into Kenya, because the target pests are not present in Kenya. Not all Bt maize hybrids are the same, and not all Bt cry genes are the same (this summary follows Andow, 2001). Indeed, Bt Cry toxin in Bt maize is not exactly the same as Bt Cry toxin in B. thuringiensis, the source bacterium for cry genes. B. thuringiensis is a soil-dwelling bacterium that produces large amounts of insecticidal delta-endotoxin when it sporulates into a resting stage. These endotoxins are biologically inactive protein toxins that crystallize into characteristic shapes. In the bacteria, the endotoxins are mixtures of several specific crystalline protein toxins (Cry toxins) that are classified into several numbered major classes, which are themselves subdivided into many subclasses. The microbial Bt insecticides targeting larvae of moths and butterflies typically contain toxins in the Cry1A class. Other strains of B. thuringiensis have different mixtures of Cry toxins. In addition, the bacteria also produce other toxins with insecticidal properties. The first generation of Bt maize contained Cry1Ab, Cry1Ac, Cry1F or Cry9C toxin, although virtually all of the Bt maize commercially grown today contains only Cry1Ab. The crystalline delta-endotoxins kill insects by a complex process. After ingestion, the crystals must dissolve in the insect midgut. This occurs readily when the pH of the midgut is alkaline, but occurs hardly at all under acidic conditions. In the presence of certain enzymes, the crystal releases a 130–135kDa biologically inactive protoxin of Cry1Ab or Cry1Ac. In a series of poorly understood reactions, this protoxin is processed by proteolytic enzymes to yield a 65-kDa activated toxin that can bind to receptors on the midgut epithelium. This receptor–toxin complex somehow induces pore formation in the midgut wall, lysis of the midgut, septicaemia and rapid death of the insect. The bacterium reproduces in the insect as a saprovore feeding on the cadaver, not as a pathogen feeding on living tissue. Cry1Ac is similar to Cry1Ab, but Cry9C is produced naturally as a 70.46-kDa activated protein toxin in bacteria. In Bt maize, the Bt cry gene has been modified in at least two ways. First, the gene has been truncated so that the resultant protein is not identical to the

Environ. Risk. Chap 01

23/8/04

9:31

Bt Maize and Risk Assessment

Page 9

9

inactive protoxin from the bacterium. Different companies have truncated the genes differently, resulting in several different truncated cry genes in Bt maize. Bacterial genes do not express well in plants, in part because bacterial genomes are enriched with the nucleotides A and T, while plant genomes are enriched in G and C. Consequently, the truncated cry genes are modified further to enrich their G–C content, allowing high levels of expression in plants. As discussed elsewhere (Hilbeck, 2002), these alterations may change the ecological effects of these toxins.

Risk Assessment As discussed in NRC (2002), risk is the combination of the probability of occurrence of some hazard and the harm corresponding to that hazard. This is both a highly technical and somewhat vague interpretation, involving the related ideas of hazard, occurrence and some combining process. A hazard has the potential to produce harm, injury or some other undesirable consequence. When we say a slippery road is hazardous, we do not mean that any harm or accident has occurred. We simply mean that the conditions could potentially cause an accident. In this case, the hazard is a car accident, which may or may not happen, and the harm is the adverse consequences of the accident. Likewise, when we say a transgenic crop has a hazard, this does not mean that the hazard or any harm has or will occur. Hazard identification is one of the most subjective and potentially contentious elements of risk assessment. While this case study is limited to a consideration of environmental risks, there is some ambiguity in deciding what is and is not an environmental hazard. Does this include or exclude the potential for adverse impacts on human health that are mediated by the environment (not directly by food consumption)? Does it include or exclude the potential for adverse impacts on farming practices and profitability? Is a non-specific effect on habitats or ecosystems an identifiable hazard? Is an effect on an ecological process an environmental hazard? These issues are addressed implicitly throughout this Kenya case study. Environmental risk analysis is used in decision-making processes to reduce adverse effects on the environment. Risk analysis is often divided into two components, risk assessment and risk management (NRC, 1983). Risk assessment is used to provide information into the decision-making process, and risk management is used to provide decision options for consideration. Risk assessment is the process by which risk is measured; this measurement can be quantitative or qualitative, probabilistic or deterministic. In risk management, society determines how to address the risk. This is done either by direct involvement of society or via social representatives or delegated authority. The main risk management decisions are whether to tolerate, mitigate or avoid the risk. Measures taken to restrict the use of a transgenic organism also avoid risk; projected risks are avoided. When a transgenic organism is used freely, measures taken to reduce the resulting hazards and harms mitigate risk. Measures taken to reduce perceived risk or increase the acceptability of the resulting hazards and harms increase the tolerance of risk.

Environ. Risk. Chap 01

23/8/04

9:31

Page 10

10

D.A. Andow and A. Hilbeck

Risk assessment is frequently considered to be comprised of several stages: problem identification, hazard and harm identification, effects assessment, exposure assessment and a process to combine all of these (see Box 1.1 for some definitions). Problem identification sets the scope of the assessment (see Nelson et al., Chapter 3, this volume). The identification of hazards involves the identification of potential adverse impacts on the environment and the potential causes of those impacts. Identification of harm involves identification of the adverse consequences of these impacts, and the identification of whom or what part of human society suffers the consequences. As noted above, identification of hazard and harm is a valueladen process, requiring judgements about what is an adverse effect on the environment and what constitutes harm and to whom it matters. Effects assessment is a ranking or quantification of the adverse effects and adverse consequences. Exposure assessment is an evaluation of the degree that the environmental stressor (the transgenic organism, transgene or transgene product) occurs in the environment. The final stage of combining these involves a critical process of valuation of risk. In risk assessment, the valuation of risk is framed in terms of human values, such as the usefulness of biodiversity to humans. In some risk assessment models (NRC, 1983), the valuation problem is de-emphasized (although it is still present), and estimating risk focuses on ranking or quantifying the likelihood that an adverse effect (hazard) or an adverse consequence (harm) will occur. In other risk assessment models (NRC, 1996), the valuation problem permeates the processes of hazard and harm identification, and effects and exposure assessment. In these cases, the entire process is called risk characterization. In this sense, characterizing environmental risks of transgenic organisms is still in a rudimentary state, primarily because the decision options are sometimes poorly characterized, and the hazards and harms are not concretely identified.

Box 1.1. Definitions of risk analysis terms. Exposure – the degree a transgenic organism, transgene or transgene product occurs in the environment. Harm – an adverse consequence of a hazard; this is usually framed in terms of harm to humans; harm is conditioned on the occurrence of a hazard. Hazard – an adverse effect on the environment. Risk (def. 1) – the likelihood of occurrence of a hazard; this can be estimated as the likelihood of exposure, times the conditional likelihood of the hazard given that exposure occurs; this is the standard technical definition of risk and will be used in this book. Weighted Risk (def. 2) – the weighted likelihood of hazard; this can be estimated as the likelihood of occurrence of a hazard, times the magnitude of the hazard. Weighted Risk (def. 3) – the weighted likelihood of harm; this can be estimated as the likelihood of occurrence of a hazard, times the conditional likelihood that a harm will occur, given that a hazard occurs, times the magnitude of the harm; this is the colloquial idea of risk.

Environ. Risk. Chap 01

23/8/04

9:31

Page 11

Bt Maize and Risk Assessment

11

Risks can be reduced by our actions, and risk management is an important concept used in this case study. By managing risks, it may be possible to reduce them to such an extent that we consider them insignificant. Risk management includes all of the processes aimed to influence the avoidance of risk, the perception of risk, the acceptability of risk or the mitigation of risk. Management includes regulatory approaches that restrict, exclude or prohibit a transgenic organism, and actions that reduce risk, by either limiting exposure or limiting the harm to the environment. In the formulation we use here, risk communication is a part of risk management, because risk communication is involved in establishing how people avoid, accept (or reject), perceive and mitigate risks. All of these risk management methods are discussed in more detail in later chapters.

Standards of evidence in risk assessment Broadly conceived, the techniques of risk assessment are of five general kinds (NRC, 2002): 1. Epidemiological analysis. Events of interest are observed, and the statistical relations of these events in sampled populations are analysed. This epidemiological approach has been very effective in identifying disease risks among populations such as smokers, industrial workers exposed to certain substances and persons with a specific genotype. The scientific rationale for this method is that empirical correlations provide a basis for predicting effects and may indicate cause. This method could be used to associate risks with particular transgenic plants that are intensively planted in large or specific areas. 2. Theoretical models. A theoretical model that mimics or simulates the causal interaction of elements in a complex system is used to identify likely sources of system failure. This approach is widely used to study the risk of failure in engineering contexts such as instrumentation and control design. It has also been applied in biology to develop strategies for ecosystem management. The scientific rationale for this method is that it provides the logical consequences of a set of scientific assumptions about risks. It has played a critical role in analysis of the risk of evolution of resistance to transgenic insecticidal plants (Alstad and Andow, 1995; Roush and Osmond, 1997; Gould, 1998), and could play a role in evaluating community-level non-target effects of transgenic plants (Andow, 1994). 3. Experimental studies. Controlled experiments are conducted to identify cause-and-effect relationships. Variations on this approach (often combined with statistical analysis) are used in product testing, including clinical trials for drugs and therapeutic devices. The scientific rationale for this method is that it establishes the cause or causes of a risk. Laboratory experiments can be used to establish a potential hazard (Hilbeck et al., 2000), and field experiments can be used to evaluate potential risks. With field experiments, however, it is necessary to guard against false negatives (type II statistical errors), which would lead to concluding erroneously that there is no risk when, in fact, the experimental design is incapable of detecting any risk except a very large one (Marvier, 2002; Andow, 2003).

Environ. Risk. Chap 01

23/8/04

9:31

12

Page 12

D.A. Andow and A. Hilbeck

4. Expert judgements. A group of experts use personal knowledge of a given system to estimate likely system performance under untested conditions. This is a widely used approach in risk analysis and has become the basis for risk analysis for some invasive species in the USA (Orr et al., 1993). The group is chosen to represent a range in necessary expertise with due consideration to potential conflicts of interest. The scientific rationale for this method is that the consensus of the group of experts represents a synthesis of the best-available scientific knowledge on the risk. The US Environmental Protection Agency (EPA) uses this method frequently to aid in risk assessment of transgenic plants in its SAP (Scientific Advisory Panel) process. 5. Expert regulatory judgement. Regulatory personnel use personal knowledge of a given system to estimate likely system performance under untested conditions. This is the most widely used approach in risk analysis. The scientific rationale for this method is that regulatory personnel have ready access to confidential business information and understand both current scientific knowledge and the process of risk analysis, so their judgements can be rendered with minimal delays and sufficient scientific accuracy. The first three of these approaches are generally accepted as scientifically rigorous methods of analysis. Expert judgements, whether by external experts or by regulatory experts, are less rigorous but often are acceptable. A consensus of multiple external experts is likely to be more rigorous than the expert regulatory judgements because disagreements among external experts are likely to lead to more robust risk assessments (Jasanoff, 1986). In this case study, we seek to identify how experimental studies and theoretical models can be used to support risk assessment of Bt maize in Kenya. Hence, this book is not a full risk assessment of Bt maize in Kenya, but it provides a scientific blueprint by which a full risk assessment can be accomplished.

Some Valuation Issues Biodiversity as a value The Convention on Biological Diversity (CBD, 1992) defines biological diversity broadly to include the variability among living organisms from all sources including diversity within species, between species and of ecosystems. The objectives of the Convention are the conservation of biological diversity, the sustainable use of its components and the fair and equitable sharing of the benefits arising out of the utilization of genetic resources. To maintain consistency with the definitions and objectives of the CBD, in the Kenya case study we begin to address the issues of fairness and equitability in a transparent, logical manner in Chapter 3, and address genetic conservation in Chapters 6 and 7 and species conservation in Chapter 5. Parties to the Convention are aware that conservation and sustainable use of biological diversity is of critical importance for meeting the food, health and other needs of the growing world population, for which purpose access to and sharing of both genetic resources and technologies are essential (CBD, 1992). In addition,

Environ. Risk. Chap 01

23/8/04

9:31

Page 13

Bt Maize and Risk Assessment

13

parties are conscious of the intrinsic value of biological diversity (CBD, 1992). A major challenge within a risk assessment framework is to address issues related to the intrinsic value of biological diversity. Intrinsic value means that biological diversity has value simply by existing, independent of any particular use for humans or function it may have in nature. The extension of intrinsic value to all biological diversity means that the most insignificant bacterium has a similar value as a human life, and there are poorly specified bases for prioritizing some kinds of biological diversity over others. As will be seen throughout this book, we have not yet determined how to implement this intrinsic value concept into the risk assessment process. Instead, we apply utilitarian values, taking into account local agricultural and food concerns as a basis for prioritization. A related issue is whether risks are assessed using species-specific methods or using the intact biodiversity in nature. While there is a role for assessing risks on the intact biodiversity (both genetic diversity and species diversity), in this case study, the focus has been on species-specific methods. These methods are more amenable to laboratory, greenhouse and small-scale field testing, and will by necessity play a significant role in risk assessment. However, the role of studies on intact biodiversity in the risk assessment process remains to be developed in the future.

Controls, counterfactuals and alternative futures The problem of what is an appropriate control or counterfactual can be viewed from a variety of perspectives. These perspectives, however, have quite different implications for risk assessment. From a narrow mechanistic perspective, the risks associated with a transgene should be evaluated in such a way that the effects of the transgene by itself can be assessed. In many transgenic crops, a single gene is introduced and one possible genetic comparison for risk assessment is the untransformed isogenic line (NRC, 2002). This comparison is not perfect because the transgenic plant will differ genetically from the untransformed plant in regions near the transgene. Taken to its logical extreme, however, this perspective would require that the transgenic plant be compared to a similar transgenic plant from which the transgene has been removed. This is a sound mechanistic approach toward experimental design, but does not allow for many indirect pathways by which risks could arise. This approach puts a premium on determining that the transgene itself is the exact cause of any possible environmental effect. However, risks that arise from differences in plant breeding strategies or the way the transgenic variety is used in actual production systems will not be detected. Hence, an isogenic control cannot be used as a general approach for risk assessment. Depending on the potential exposure pathway and the potential hazard, various other controls could be chosen. For example, assessment of non-target risks will require both toxicological evaluations using the purified transgene product, and whole plant evaluations (Andow and Hilbeck, 2004; Birch et al., Chapter 5, this volume). The toxicological evaluations would use the absence of the transgene product as an appropriate control. Sole reliance on

Environ. Risk. Chap 01

14

23/8/04

9:31

Page 14

D.A. Andow and A. Hilbeck

toxicological evaluations is insufficient for risk assessment of the transgenic plant. The transgene product can be metabolized to new compounds in the plant and, in some non-target species, can interact with other plant chemicals (Hilbeck, 2002) and alter the expression of other plant genes. Any of these reactions can, in turn, have effects on non-target species independently of the transgene product itself. For example, for one transgenic Bt maize event, a variety of different transgene products has been reported (AGBIOS, 2003). The concentration of some secondary plant compounds changed in some transgenic plants (lignin, Saxena and Stotzky, 2001; glycoalkaloids, Birch et al., 2002). Hence, an additional methodology is therefore necessary, which we call the ‘whole plant’ method (Andow and Hilbeck, 2004; Birch et al., Chapter 5, this volume). This method evaluates the effects of the transgenic plant, not just the transgene product, because the effect of a transgene is possibly greater than the isolated effect of the transgene product. To conduct such tests, it is necessary to have an appropriate experimental control for the transgenic plant, and to mimic exposure experimentally as it would occur in the field. We have argued that an agronomically similar variety may be a useful control in this instance (Andow and Hilbeck, 2004). The potential hazard that antibiotic resistance genes could be transferred from transgenic crops (such as Bt maize with the nptII gene) to soil bacteria is unique to transgenic crops. However, even this unique hazard can be evaluated using appropriate reference models for comparison. A laboratory model optimized so that the bacteria are readily transformed by naked DNA in the culture environment provides a worst-case scenario (Nielsen et al., 2000). The rate of occurrence in the natural environment is assumed less than that in the permissive experimental environment. The experimental transformation rate was low, which suggests that gene transfer via transformation in natural environments is correspondingly rare. Another informative comparison is the number of nptII genes in the transgenic crop plant compared to the number in the soil. In several agricultural soils, the nptII gene is already present at 105/g soil (Smalla et al., 1993), a very high rate of occurrence. This high rate of occurrence may be characteristic of only this antibiotic resistance gene, but this comparison suggests that a transgenic crop may not be a significant source for this resistance gene in nature. It should be clear from these examples that appropriate controls for assessing risk depend on the nature of the risk. Conventional plant breeding and isogenic lines will not always be an appropriate control. More broadly yet, the assessment of options and alternatives is a standard part of risk assessment (NRC, 1983). The potential use of a transgenic crop is about changing how crops are grown; this focuses on the possibilities for the future. It is natural for us as a society to hope that the future is an improvement over the present, and therefore, from a risk assessment perspective it is essential to compare the transgenic plant to present agricultural practices. Consequently, an important scientific control and economic counterfactual for risk assessment is present practice. This was the approach used in conducting the UK Farm Scale Evaluations of the environmental effects of herbicide-tolerant crops (Firbank, 2003, and associated papers).

Environ. Risk. Chap 01

23/8/04

9:31

Page 15

Bt Maize and Risk Assessment

15

Using only present conditions as a control for risk is insufficient. Doing so is tantamount to comparing one possible future to the present, as if the present were the future. We humans do not act this way. Instead, we compare alternative futures. Similarly, in risk assessment it is important to consider the various future options and alternatives, because the risks associated with each future must be compared with those associated with an alternative future. In Chapter 3, we develop an assessment process to provide some of these alternative futures to which the transgenic crop can be compared. These alternatives include conventional resistance, insecticides, intercropping, biological control and several others. A risk assessment should then use one or more of these alternative futures as a scientific control or economic counterfactual.

Scope of the Case Study Now that the Cartagena Biosafety Protocol has entered into force, it is essential that the scientific basis for risk assessment be developed so that parties to the Protocol can develop the necessary infrastructure to evaluate transgenic organisms. This book represents the efforts of public-sector scientists to bring the collective wisdom of the scientific community together to help develop a scientific risk assessment that fits the conditions of the environment and the aspirations of the Protocol. Among the developing countries, the African nations are among the lowest in terms of per capita income, and have low and variable food production, which creates an ever-present threat to food security. While genetic engineering has promised to improve food security, scepticism and uncertainty among African nations remain, in part because they recognize that many of their food problems require more than just a technical solution and existing genetically engineered plants do not address other more pressing problems (Masood, 2003). Among the African countries, Kenya became the focus for this case study for three related reasons. First, decision making on the acceptance or rejection of transgenic organisms will continue to be made by countries, so it is important to develop scientific risk assessment capacity at a similar scale of organization so that the science can be integrated with the decision-making process. Hence, rather than focus on a region, such as East Africa, or a part of a country, it was important to focus at the national level. Second, to conduct a scientific risk assessment, it is essential that there is a scientific infrastructure in the country. Investments in science have been declining throughout the world, and especially so in the developing countries of Africa. In sub-Saharan Africa, two countries have historically supported and continue to support relatively vigorous scientific infrastructures, these being Kenya and South Africa. While both have suffered significant declines during the past decade or so, both maintain a strong physical and human capital base for science. Third, as Kenyans gain expertise in transgenic organism risk assessment, Kenya could become a regional centre of expertise, with which neighbouring countries could consult as they experience the challenges of evaluating transgenic organisms.

Environ. Risk. Chap 01

23/8/04

16

9:31

Page 16

D.A. Andow and A. Hilbeck

Kenya ratified the Cartagena Biosafety Protocol in January 2002 (CBD, 2002) and is keenly interested in developing the needed expertise to implement the Protocol. In addition, Kenya has been interested in the use of Bt maize in their maize-based cropping system, and feels the need to conduct a risk assessment on Bt maize as soon as is feasible. While there are almost countless efforts by international organizations to assist Kenya in the development of regulatory, oversight and governance systems for transgenic organisms, there are virtually no such efforts that focus primarily on the scientific dimensions to the development of risk assessment procedures. The efforts recorded in this book are initial steps toward filling this scientific gap. As we point out in various places throughout this book, the case study conducted in Kenya is not a full risk assessment of Bt maize in Kenya. It is a case study of a methodology for conducting an environmental risk assessment of Bt maize. Every effort was made to have available the most recent information at the workshop so that the case study would be as accurate as possible, and the emphasis throughout is on transparency and scientific rigor in the risk assessment methods. Invariably, however, significant sources of information were unavailable at the time of the workshop, and new information became available after the workshop. When possible and appropriate, we incorporated the additional information into the chapters of this book. Some of the conclusions in this book partially reflect the limitations to access of information. In addition, some essential scientific issues, which are identified in the respective chapters, were not fully resolved during the workshop and remain to be addressed in the future. Therefore, this case study of an environmental risk assessment of Bt maize in Kenya is insufficient support for a regulatory decision on Bt maize in Kenya. A full risk assessment must update and complete the findings in this book and extend some of the methods prior to any decision-making process.

Geographic scope The methods developed in this book for risk assessment of Bt maize in Kenya can probably transcend Kenyan national borders and be used for similar cases in neighbouring countries with similar environments. As the present book is a case study, it is not yet clear how widely these methods will be applicable. In Chapter 8, we have indicated that this issue will be addressed more directly as we complete additional case studies and begin to see which methods may have broad applicability, and which may be specific to the case of Bt maize in Kenya. For example, in Kenya, no wild relatives exist with which maize could interbreed. This is completely different from the tropical maize production systems in Central America, where additional methods would need to be developed to assess risks of gene flow to wild and weedy relatives. Risk issues may also differ depending on prevalent type of agriculture – industrial style or smallholder subsistence farming – and what type of Bt maize will be introduced – OPVs or hybrids of inbreds. Therefore, while the methods developed in Kenya should apply beyond Kenyan national borders, they do not eliminate the need for case-by-case consideration of a number of aspects that could vary from region to region.

Environ. Risk. Chap 01

23/8/04

9:31

Bt Maize and Risk Assessment

Page 17

17

The results of the case study should also be able to generalize to neighbouring countries with similar environments; however, the results are probably not as generalizable as the methods, and their generalizability will be specific to the risk assessment aspect. For example, the description of transgene locus structure (Andow et al., Chapter 4, this volume) may generalize readily, while the expression of transgene product across Kenyan environments may have limited value beyond Kenyan national boundaries. The ability to generalize results beyond the particular case study will be addressed formally in future case studies.

Outline of book Chapter 2 reports on the agricultural and socio-economic context of maize production in Kenya. It describes the regional constraints to maize production and contextualizes the stemborer problem within these constraints. It also introduces possible alternative solutions to the stemborer problem, which will be taken up in Chapter 3. Chapter 3 provides a framework for specifying the problem, evaluating the utility of the transgenic plant in specific Kenyan crop production contexts and comparing it to other potential solutions to the problem. It sets the context for the environmental analyses that follow in Chapters 5–7. It defines the target agroecosystems for which Bt maize or an alternative solution is proposed, including the crop system, farming system and ecological and structural context, and the people who will be affected. Chapter 4 identifies the genetic issues relevant for risk assessment and proposes methodologies to address these issues and thereby reduce or identify the hazards arising from Bt maize in Kenya. It structures the characterization of the transgene into four components: transgene design, genotypic characterization, phenotypic characterization and transmission between generations. Some of these characteristics should be measured in the field, and others can be measured in the lab. Nutritional characterization of the harvested Bt maize is outside the scope of this chapter. Chapter 5 provides scientific procedures to determine which non-target species, structural characteristics of the biota and/or ecosystem functions should be tested prior to environmental release, and secondly, specifies scientific procedures for testing the potential impact of Bt maize on these species or functions. These procedures are primarily laboratory and greenhouse ones, and some field procedures are proposed. A remaining challenging question is: what is the role of other ‘neutral’ or ‘value unknown’ non-target species? The vast majority of species found in an agricultural field are ‘neutral’ or ‘value unknown’ species. Although this will not be taken up directly in this chapter, it was a question that attracted considerable discussion during the workshop. Chapter 6 deals with the issues around gene flow and its consequences. It examines methods for establishing the likelihood of intraspecific gene flow to other maize varieties, the possibility of subsequent geographic spread of transgenes, and the potential ecological effects resulting from gene flow.

Environ. Risk. Chap 01

18

23/8/04

9:31

Page 18

D.A. Andow and A. Hilbeck

Several types of recipient populations are considered, and the focus is on gene flow among cultivated relatives in the various regions of Kenya. The chapter then addresses whether transgenes are likely to increase in frequency due to natural and/or artificial selection and what the ecological consequences of this process might be. Chapter 7 identifies procedures to determine the risk that target pests will evolve resistance to Bt maize in Kenya and feasible management responses needed to reduce this risk. It also considers approaches for developing a practical monitoring and response system to detect resistance and to adapt management appropriately. Finally, Chapter 8 summarizes the major achievements, key findings and lessons learned during the Kenya workshop. It concludes with recommendations for completing a risk assessment of Bt maize in Kenya and for improving the scientific basis for risk assessment.

References AGBIOS (2003) Essential biosafety, 2nd edn. Merrickville, Ontario, www.essentialbiosafety.com (accessed 25 November 2003). Alstad, D.N. and Andow, D.A. (1995) Managing the evolution of insect resistance to transgenic plants. Science 268, 1894–1896. Andow, D.A. (1994) Community response to transgenic plant release: using mathematical theory to predict effects of transgenic plants. Molecular Ecology 3, 65–70. Andow, D.A. (2001) Resisting resistance to Bt corn. In: Letourneau, D.K. and Burrows, B.E. (eds) Genetically Engineered Organisms: Assessing Environmental and Human Health Effects. CRC Press, Boca Raton, Florida, pp. 99–124. Andow, D.A. (2003) Negative and positive data, statistical power, and confidence intervals. Environmental Biosafety Research 2, 75–80. Andow, D.A. and Hilbeck, A. (2004) Science-based risk assessment for non-target effects of transgenic crops. Bioscience 54, 637–649. Armstrong, C.L., Green, C.E. and Phillips, R.L. (1991) Development and availability of germplasm with high Type II culture formation response. Maize Genetics Newsletter 65, 92–93, www.agron.missouri.edu/mnl/65/146armstrong.html (accessed 25 November 2003). Birch, A.N.E., Geoghegan, I.E., Griffiths, D.W. and McNicol, J.W. (2002) The effect of genetic transformations for pest resistance on foliar solanidine-based glycoalkaloids of potato. Annals of Applied Biology 140, 143–149. Birch, R.G. (1997) Plant transformation: problems and strategies for practical application. Annual Review of Plant Physiology and Plant Molecular Biology 48, 297–326. CBD (1992) Convention on Biological Diversity: convention text, www.biodiv.org/ convention/articles.asp (accessed 25 November 2003). CBD (2002) Cartagena Protocol on Biosafety: status of ratification and entry into force, www.biodiv.org/biosafety/signinglist.asp (accessed October 2002). Firbank, L.G. (2003) Introduction. Philosophical Transactions of the Royal Society of London B 358, 1777–1778.

Environ. Risk. Chap 01

23/8/04

9:31

Bt Maize and Risk Assessment

Page 19

19

Gould, F. (1998) Sustainability of transgenic insecticidal cultivars: integrating pest genetics and ecology. Annual Review of Entomology 43, 701–726. Hilbeck, A. (2002) Transgenic host plant resistance and non-target effects. In: Letourneau, D.K. and Burrows, B.E. (eds) Genetically Engineered Organisms: Assessing Environmental and Human Health Effects. CRC Press, Boca Raton, Florida, pp. 167–185. Hilbeck, A., Meier, M.S. and Raps, A. (2000) Review on Non-target Organisms and Bt Plants. EcoStrat GmbH, Zurich. James, C. (1997) The global review of commercialized transgenic crops: 1997. ISAAA Briefs No. 5. International Service for the Acquisition of Agri-biotech Applications, Ithaca, New York. James, C. (1998) The global review of commercialized transgenic crops: 1998. ISAAA Briefs No. 8. International Service for the Acquisition of Agri-biotech Applications, Ithaca, New York. James, C. (1999) Global status of commercialized transgenic crops: 1999. ISAAA Briefs No. 12: Preview. International Service for the Acquisition of Agri-biotech Applications, Ithaca, New York. James, C. (2001) Global review of commercialized transgenic crops: 2000. ISAAA Briefs No. 23. International Service for the Acquisition of Agri-biotech Applications, Ithaca, New York. James, C. (2002) Preview: Global review of commercialized transgenic crops: 2001, ISAAA Briefs No. 24. International Service for the Acquisition of Agri-biotech Applications, Ithaca, New York. Jasanoff, S. (1986) Risk Management and Political Culture: a Comparative Study of Science in the Policy Context. Sage, New York. Kohli, A., Leech, M., Vain, P., Laurie, D.A. and Christou, P. (1998) Transgene integration in rice engineered through direct DNA transfer supports a two-phase integration mechanism mediated by the establishment of integration hot spots. Proceedings of the National Academy of Sciences 95, 7203–7208. Lamkey, K. (2004) Seed production in corn and soybean. In: Mellon, M. and Rissler, J. (eds) Achieving Zero Contamination of the Food Supply by Pharm Crops. Union of Concerned Scientists, Boston, Massachusetts. Marvier, M.A. (2002) Improving risk assessment for non-target safety of transgenic crops. Ecological Applications 12, 1119–1124. Masood, E. (2003) GM crops: a continent divided. Nature 426, 224–226. NASS (2000) Prospective Plantings, March 31, 2000. National Agricultural Statistical Service, Agricultural Statistics Board, USDA, Washington, DC. http://usda.mannlib. cornell.edu/reports/nassr/field/pcp-bbp/psp10300.pdf (accessed 25 November 2003). NASS (2002) Acreage Report Supplement. National Agricultural Statistical Service, Agricultural Statistics Board, USDA, Washington, DC. http://usda.mannlib. cornell.edu/reports/nassr/field/pcp-bba/acrg0602.pdf (accessed 10 December 2003). NASS (2003) Acreage Report Supplement. National Agricultural Statistical Service, Agricultural Statistics Board, USDA, Washington, DC. http://usda.mannlib. cornell.edu/reports/nassr/field/pcp-bba/acrg0603.pdf (accessed 10 December 2003). Nielsen, K.M., van Elsas, J.D. and Smalla, K. (2000) Transformation of Acinetobacter sp. strain BD413 (pFG4nptII) with transgenic plant DNA in soil microcosms and effects of kanamycin on selection of transformants. Applied Environmental Microbiology 66, 1237–1242.

Environ. Risk. Chap 01

20

23/8/04

9:31

Page 20

D.A. Andow and A. Hilbeck

NRC (National Research Council) (1983) Risk Assessment in the Federal Government: Managing the Process. National Academy Press, Washington, DC. NRC (1996) Understanding Risk: Informing Decisions in a Democratic Society. National Academy Press, Washington, DC. NRC (2002) Environmental Effects of Transgenic Plants: the Scope and Adequacy of Regulation. National Academy Press, Washington, DC. Orr, R.L., Cohen, S.D. and Griffin, R.L. (1993) Generic Non-indigenous Pest Risk Assessment Process (for Estimating Pest Risk Associated with the Introduction Of Non-indigenous Organisms). US Department of Agriculture, Washington, DC. Ostlie, K.R., Hutchison, W.D. and Hellmich, R.L. (1997) (eds) Bt Corn and European Corn Borer: Long-Term Success through Resistance Management. NCR Publication No. 602, University of Minnesota, St Paul, Minnesota. Pawlowski, W.P. and Somers, D.A. (1998) Transgenic DNA integrated into the oat genome is frequently interspersed by host DNA. Proceedings of the National Academy of Sciences USA 95, 12106–12110. Porter, P., Schuster, G., Morrison, W.P., Troxclair, N.N., Cronholm, G.B. and Patrick, C.D. (2000) Bt Corn Technology in Texas: a Practical View. Texas Agricultural Extension Service, B-6090, 1-00. Rice, M.E. and Pilcher, C.D. (1998) Potential benefits and limitations of transgenic Bt corn for management of the European corn borer (Lepidoptera: Crambidae). American Entomology 44, 75–78. Roush, R.T. and Osmond, G. (1997) Managing resistance to transgenic crops. In: Carozzi, N. and Koziel, M. (eds) Advances in Insect Control: the Role of Transgenic Plants. Taylor and Francis, London, pp. 271–294. Sanford, J.C., Smith, F.D. and Russell, J.A. (1993) Optimizing the biolistic process for different biological applications. Methods in Enzymology 217, 483–509. Saxena, D. and Stotzky, G. (2001) Bt corn has a higher lignin content than non-Bt corn. American Journal of Botany 88, 1704–1706. Smalla, K., van Overbeek, L.S., Pukall, R. and van Elsas, J.D. (1993) Prevalence of nptII and Tn5 in kanamycin resistant bacteria from different environments. Microbiology Ecology 13, 47–58.

Environ. Risk. Chap 02

2

23/8/04

9:32

Page 21

The Maize Agricultural Context in Kenya L. MUHAMMAD AND E. UNDERWOOD Corresponding author: Lutta Muhammad, KARI Katumani, PO Box 1764, Machakos, Kenya. E-mail: [email protected] or [email protected]

The risks associated with transgenic crops are related to the environment in which the crop is grown, and the scale and character of the farming systems in which it is included. Some potential impacts will vary according to which farmers take up the technology. It is important in precautionary risk assessment to examine the constraints to production of that crop, and consider whether the transgenic crop may have a negative effect by exacerbating the effects of any of these constraints. This chapter describes maize production in Kenya, and explains why new technologies are needed to improve maize production. One possible technique is transgenic maize containing the Bt gene to make it toxic to stemborers, and the chapters that follow will detail scientific approaches to the ecological risk assessment of Bt maize. To specify the context for this assessment, this chapter outlines the environmental and socio-economic conditions under which maize is produced in each maize agroecological zone in Kenya. The chapter further describes some of the main pest and disease problems affecting maize production in Kenya. As Bt maize in Kenya is designed to control stemborers, more detail is given on this pest problem, and various current and possible future options for controlling stemborers are described.

Maize Production in Kenya Kenya is one of the leading producers and consumers of maize in the eastern and southern Africa region. Kenya produces about 2.38 million t/year of maize (FAOSTAT, 2002). The maize production area, after expanding steadily into the semiarid areas, is now at around 1.5 million ha. This is about 2.5% of the total land area of Kenya. Over 3.5 million households and 90% of the poor in Kenya live in rural areas and depend predominantly on agricultural income (World Bank, 1995). In 2001, Kenyans consumed 3.19 million t of maize, on © CAB International 2004. Environmental Risk Assessment of Genetically Modified Organisms: Vol. 1. A Case Study of Bt Maize in Kenya (eds A. Hilbeck and D.A. Andow)

21

Environ. Risk. Chap 02

23/8/04

9:32

Page 22

22

L. Muhammad and E. Underwood

average about 40–45% of their total calorie consumption, and the Kenyan population is projected to continue growing by around 3% per year (World Bank, 1995; FAOSTAT, 2001). Unless there is a corresponding increase in total production on the available area, maize consumption will continue to grow faster than increases in maize production, obliging Kenya to import increasing quantities of maize. Kenyan researchers are therefore looking for new ways to stabilize and increase maize yields for Kenyan farmers.

Maize production areas and climate The maize production area of Kenya can be divided into six agroecological zones on the basis of growing season variables crucial to maize production, such as altitude and climate (Table 2.1 and Plate 1 – see colour frontispiece). These are: the Lowland Tropics, comprising the coastal strip and adjoining inland area, the Dry Mid-altitude and Dry Transitional zones in the southeast, the Highland Tropics, the Moist Transitional zone to the east and west of the Highland Tropics, and the Moist Mid-altitude zone around Lake Victoria (Corbett, 1998). Table 2.1. Maize agroecological production zones and associated climatic conditions (Corbett, 1998). Temperatures (°C) Mar.–Aug. Agroecological zone Lowland Tropics:a Dry Moist Dry Mid-altitudeb Dry Transitional Moist Transitionalc Highland Tropics:d Dry Moist Cool Moist Mid-altitudee Extreme water stressf aLowland

Elevation Rainfall (mm) (m) Mar.–Aug. >700 <400 700–1400 1100–1700 1200–2000 1600–2300 1600–2700 2000–2900 1200–2000 400–1100

300–550 >550 300–550 <550 >500 <550 >550 <1000 >500 <400

Mean Average Average (°C) minimum (°C) maximum (°C) 25.4 25.8 22.0 19.7 19.7 16.6 16.7 13.8 22.1 23.8

20.0 20.0 14 11 11 8 7 5 13 16

30.0 31.0 33 27 29 26 27 22 30 32

Tropics = Districts Kilifi and Kwale. Mid-altitude and Dry Transitional zones = Districts Machakos, Makueni, Kitui, Mwingi, Taita and Taveta districts, and Kerio and Baringo valleys. cMoist Transitional zone = Districts Kisii, Bomet, Nandi, Trans-Nzoia, Kapenguria, E part of Bungoma (Kimilili, Tongaren), E part of Kakamega (Lugari), Muranga, small parts of Embu and Meru. dHighland Tropics = Districts Kiambu, Kericho, Nyeri, Nakuru, Narok, Uasin Gishu. eMoist Mid-altitude zone = Districts South Nyanza, Kisumu, Siaya, Busia, W part of Bungoma (Kanduyi), W part of Kakamega (Kabras, Mumias). fThis zone includes areas to the east and north of the Dry Mid-altitude zone. It is not further considered in the chapter as it makes up only 2% of the total maize area and many characteristics are very similar to the Dry Mid-altitude zone. bDry

Environ. Risk. Chap 02

23/8/04

9:32

Page 23

The Maize Agricultural Context in Kenya

23

The Moist Transitional zone is the most important maize production zone (Table 2.2), followed by the Highland Tropics. These zones mostly lie above 1200 m and have rainfall above 550 mm/year. They contain the present highyielding areas for maize, and produce 85% of Kenya’s maize with 70% of the production area and average maize yields of 2.7–3 t/ha, up to 5 t/ha (Hassan et al., 1998a). They are the only zones that produce a surplus over the consumption of the resident population (Mills et al., 1998). The other zones make up 30% of the total maize area but produce only 15% of Kenya’s maize. The Dry Transitional and Moist Mid-altitude zones (c. 15% of production area) are regarded as of medium yield, with average maize yields of a little over 1 t/ha. The Lowland Tropics and Dry Mid-altitude zones (approximately 15% of production area) are regarded as low-yield areas, with average maize yields of around 1 t/ha, and soils low in organic matter and fertility and with a poor capacity for moisture retention. In addition, yields in these zones are highly variable from season to season and from year to year, and sometimes the crop fails completely. Half of Kenya’s arable land is semiarid or arid. Overall, rainfall is highly variable in timing, duration and intensity (Table 2.3). In most parts of the country, rainfall is bimodal; the ‘long rains’ fall between March/April and May/August, and the ‘short rains’ between October and December. In the Lowland Tropics, Moist Transitional, Highland Tropics and Moist Mid-altitude zones, most farmers regard the March rains as the most important growing season. In the Highland Tropics and Moist Transitional zones, above 1500 m, there is only one season of rains a year, and most farmers grow one maize crop annually between March and November, in from 150 to over 200 days until maturity (Table 2.3). In the Moist Mid-altitude zone, the average time to maturity is intermediate and over half the farmers attempt a second crop. In the Lowland Tropics, maize takes on average only 120 days to maturity but most farmers plant maize in only one season. Rainfall variability is highest in the Dry Mid-altitude and Dry Transitional zones, where half the farmers regard the March rains as the most important maize growing season and half regard the November rains as more important; the latter (the ‘short rains’) is often the better growing season. Nevertheless, half to three-quarters of farmers in these zones are attempting two crops per season. About 1% of the total maize production area, mainly sweetcorn, is irrigated and grown in three or four cycles throughout the year mainly in the Moist Transitional zone (J. Songa, Nairobi, January 2004, personal communication).

Farm size distribution and productivity of agricultural holdings in Kenya Smallholder farming dominates Kenyan agriculture. Individual farm holdings measuring 2 ha or less account for more than 75% of the total area of Kenya’s agricultural land, while farms within the 2.1–10 ha size category account for another 13% (Republic of Kenya 1995, 2000, 2001). The Dry Transitional, Lowland Tropics and Moist Transitional zones, in particular, are characterized by an average holding size below 2 ha, and high population density. Large-

3.5 40.9 29.6 11.4

37

424

307

118

1.44

2.91

2.76

1.21

1.03

0.47

0.03

0.10

0.79

0.41

0.25

1.11

1.73

1.50

1.08

0.83

0.99

t/ha

232

909

1234

76

162

53

9

34

46

3

6

2

51

95

94

12

35

16

79

98

98

91

67

43

from the MDBP farmer survey 1992–1993 of 1400 farmers at 75 different sites in Kenya. Yield is the average of three observations made by farmers about their maize yields: yield during the survey year (1992), yield in the preceding year (1991), and yield in a ‘normal’ year. bCalculated from yield and area in previous columns. cCollected during MDBP farmer survey 1992–1993 through retrospective questioning.

11.4

118

1.36

t/ha

% Farmers ever purchased improved seedc

24

aResults

3.2

33

% of total

Maize yield in the ‘short rains’ (1991–1992)a

9:32

Lowland Tropics Dry Midaltitude Dry Transitional Moist Transitional Highland Tropics Moist Midaltitude

000 ha

Maize yield in the ‘long rains’ (1991–1992)a

% Maize area in zone Annual maize under improved production materials (in both rains) (OPVs or b (1990–1992) hybrids) 000 t % of total (1992–1993)

23/8/04

Agroecological zone

Area of maize during ‘long rains’ (1990)

Proportion of ‘long rains’ maize area planted to maize in the ‘short rains’ (1990)

Table 2.2. Maize production and use of certified seed by farmers according to agroecological zone in Kenya (Hassan et al., 1998a,b).

Environ. Risk. Chap 02 Page 24

L. Muhammad and E. Underwood

35 50 76 40 22 60

45 122 45 1170 893 170

8 40 31 64 16 62

120 (33%) 114 (47%) 144 (20%) 181 (39%) 213 (53%) 163 (40%)

bPercentage

collected during MDBP farmer survey 1992–1993 of 1400 farmers in 75 sites in Kenya. coefficient of variation for mean total precipitation during the March–August season (the ‘long rains’), calculated based on longterm rainfall data for Kenya. cCalculated from maize production area and yield for long and short rains as in Table 2.2. dBased on time of planting and farmers’ estimates of the date when the maize crop reached maturity (i.e. was ready to harvest) for both single-cropping and double-cropping farmers. Information collected during MDBP farmer survey 1992–1993. Figures in parentheses are the percentage coefficient of variation (CV). Much of this is due to within-zone variation in altitude.

aInformation

36 52 40 27 32 32

Lowland Tropics Dry Mid-altitude Dry Transitional Moist Transitional Highland Tropics Moist Mid-altitude

9:32

99 48 46 98 89 96

April (1st week) March (3rd week) March (3rd week) March (1st week) March (2nd week) March (3rd week)

Agroecological zone

23/8/04

Interannual Maize production % Farmers variability in (000 t) (1990–1992)c Average date regarding March Mar.–Aug. % Farmers Average days when 80-mm as main precipitation double-cropping Long rains Short rains to maturity rainfall accumulates planting timea (CV %)b maizea (Mar.–Jun./Aug.) (Oct.–Dec.) and CV%d

Table 2.3. Rainfall accumulation season and variability for ‘long rains’ season, percentage of farmers regarding March rains as main planting time for maize, percentage of farmers double-cropping maize, maize production per growing season and average time to maturity of maize (Hassan et al., 1998a,b).

Environ. Risk. Chap 02 Page 25

The Maize Agricultural Context in Kenya 25

Environ. Risk. Chap 02

26

23/8/04

9:32

Page 26

L. Muhammad and E. Underwood

scale farmers with over 8 ha of land constitute less than 5% of Kenya’s rural households; however, in 1990 they were cultivating 15% of Kenya’s maize area and producing a quarter of Kenya’s maize (CBS, 1990). Only a quarter of crop production is marketed. Maize production averages about 12% of total rural household income in Kenya; only in two districts did maize production account for more than 30% of total income (ArgwingsKodeck et al., 1998). Fertilizer comprises a significant share (10–20%) of the costs of producing maize for the smallholders who use it, comparable to the costs of labour or land preparation (Wanzala et al., 2001). Transport costs are also high in Kenya and most roads are in bad condition, but most farmers now sell their maize at or very near their farms (Argwings-Kodeck et al., 1998). Commercial maize production is concentrated in the higher yielding Highland Tropics and Moist Transitional zones, in the districts Trans-Nzoia, Uasin Gishu and Nakuru, plus some parts of Narok, Kericho, Nandi and Bungoma districts, which produce 90% of the maize that is marketed in Kenya (Plate 2). Four districts (Trans-Nzoia, Uasin Gishu, Lugari and Nakuru) produce 70% of this. Kenya’s richest farmers are in maize and dairy in these districts, but smallholder farmers in these areas also successfully grow maize commercially. A comparison in 1999 found that small-scale farmers and largescale farmers with comparable yields also made a similar level of profit per bag of maize sold (Nyoro et al., 2001). Despite the differences in average yields between zones, net production (the amount of food or money farming households gain from their maize after subtracting the costs of production) is more variable between farming households within an agroecological zone than the inter-zonal variation (Karanja et al., 1998). For instance, in 1997 the top 25% of households in the Lowland Tropics and Dry Mid-altitude zone had a net maize production 12 and 13 times that of the bottom 25% of households. The net production for individual farming households is therefore strongly influenced by factors other than the average agroecological potential of their region. The average maize yield in Kenya has been rising during the last decade, as has total maize production (Pingali, 2001). However, household net production from maize has declined in the last 30 years, though the impact varies among agroecological zones (Karanja et al., 1998). Net production has declined most (by about 10%) for households in parts of the Moist Mid-altitude zone and the Highland Tropics, except for the areas close to Nairobi. This may be due to the decline in fertilizer use. In the Lowland Tropics, Dry Mid-altitude and Dry Transitional zone and parts of the Moist Mid-altitude zone, the decline was smaller, probably because fertilizer use has always been low in these regions. In the Lowland Tropics, smallholders have less of the farm area under maize than in the other zones (Table 2.4), and cassava is an important staple, also sorghum. Kenyan maize is also increasingly more expensive in comparison with maize coming from Uganda and Tanzania, despite the high transport and transaction costs of informal cross-border trade (Nyoro et al., 2001). Despite the fact that virtually all rural households in Kenya grow maize, over 60% of them are net maize buyers because they do not produce enough for their domestic consumption (World Bank, 1995; Argwings-Kodeck et al.,

Environ. Risk. Chap 02

23/8/04

9:32

Page 27

The Maize Agricultural Context in Kenya

27

Table 2.4. Size of agricultural holdings, average per cent of farm area under maize, farmer characteristics and population density per agroecological zone (Hassan et al., 1998a). Average percentage of farm area under maize-dominated cropping system in 1992–1993a

Agroecological zone Lowland Tropics Dry Mid-altitude Dry Transitional Moist Transitional Highland Tropics Moist Mid-altitude

% Farms <2 ha (very small) 60 61 80 64 54 56

Population Farm size <8 ha Farm size >8 ha density in 1992 (small and medium) (large) (pers/1000 km2) 66 64 57 46 45 38

7 31 –b 26 12 12

121 210 398 331 238 310

aInformation

collected during MDBP farmer survey 1992–1993 of 1400 farmers in 75 sites in Kenya. bNot surveyed. There are very few farms of this size in this zone.

1998). Even in the high-yielding maize areas, almost 30% of households are net buyers. Smallholder farmers consume their maize themselves or sell it on the informal market. Poor and rural consumers mill their own grain or take it to small-scale local hammer mills to obtain a coarse maize meal. The larger-scale farmers sell their maize via the formal Kenyan marketing system to the large mills, who sell refined white meal to wealthier and urban consumers. Since the cereal market reform in 1992, the consumption of the cheaper coarse maize meal has increased, and that of refined meal has declined. White maize is the dominant calorie source for all Kenyans. In some areas, cooking bananas, Irish potatoes, cassava or rice also contribute significantly but never more than maize (Argwings-Kodeck et al., 1998).

Farming systems, crops and soil management Small-scale farms in Kenya practise crop and livestock farming with little specialization. Intercropping with a wide variety of crops, often legumes, is practised by almost all small-scale farmers and over half the large-scale farmers except in the Moist Transitional and Highland zones where more intensive maize monoculture systems are more frequent (Plate 2). On average for all zones, maize-dominated cropping systems take up over half of the area on a small or medium-sized farm under 8 ha (Table 2.5) (Argwings-Kodeck et al., 1998). Despite the large seasonal and annual fluctuations in local maize prices that make it risky to move out of staple food production, farming households in the lowland zones (Lowland Tropics and Dry and Moist Mid-altitude zones) have moved land out of maize production in the decade since maize market liberalization in 1992. Only in the high-yielding areas of the Highland Tropics and Moist Transitional zone have the majority of households increased cereal production. Table 2.5 shows the main cropping systems in each agroecological zone. Many Kenyan

78

88

95

89

90

77

Lowland Tropics

Dry Mid-altitude

Dry Transitional

Moist Transitional

Highland Tropics

Moist Mid-altitude

57

65

67

34

38

32

56

48

58 85

23

49

% Farmers for whom soil erosion is a serious problem 1992–1993a

24

59

78

43

12

2

% Farmers applying basal mineral fertilizer to maize 1992–1993c

19.5

42.0

58.5

34.7

16.0

na

Average total mineral nutrients appliedd (kg nutrients /ha) 1992–1993

52

50

45

89

57

22

% Farmers using animal manure on maize 1992–1993e

cMainly

bBergvinson

28

in MDBP farmer survey 1992–1993 (Hassan et al., 1998a). et al. (2001). diammonium phosphate (DAP) or 20:20:0 NPK. Does not include mineral fertilizer applied as top dressing (mainly calcium ammonium nitrate) (Hassan et al., 1998d) (Lowlands 100% DAP; Dry Mid-altitude 33% DAP, 50% NPK; Dry Transitional 59% DAP, 22% NPK; Moist Transitional 77% DAP, 10% NPK; Highland Tropics 79% DAP, 11% NPK; Moist Mid-altitude 91% DAP, 9% NPK). dTotal NPK. Includes basal mineral fertilizer and top dressing mineral fertilizer. Often oversupply of P and undersupply of N – for nutrient ratio see footnote c (Hassan et al., 1998d). eRates and quality not recorded (Hassan et al., 1998d). fAll root crops except cassava, arrowroot, sweet potatoes and Irish potatoes (Solanum tuberosum), which are listed separately. gThese zones are not differentiated in this survey (Bergvinson et al., 2001). hNot surveyed as there are very few farms of this size in this zone. iCentral highlands = Districts of Muranga, Nyeri, Meru. Partly in Moist Transitional zone, partly in Highland Tropics. na = data not available.

50

39

16

Maize, rice, root cropsf, maize/beans, maize/cassava, horticulture, bean/cowpea, maize/cowpea, maize/rice, maize/sorghum, maize/cashew, maize/coconut gHorticulture, maize/beans, bean/cowpea, maize, root crops, sorghum/millet, banana, Irish potato, coffee gHorticulture, maize/beans, maize, root crops, Irish potato (but not Kisii), wheat (but not Kisii or Central highlandsi), banana, coffee (Kisii and Central highlands), bean/cowpea, millet/sorghum, tea, sunflowers, Pyrethrum Root crops, maize/beans, horticulture, millet/sorghum, banana, sugarcane, maize

% Maize area on slopes and very steep land 1992–1993a

12:38

—h

77

50

Farms >8 ha

Main cropping systems in descending order of total cropping areab

13/9/04

aRecorded

Farms <8 ha

Agroecological zone

% Farmers intercropping maize in 1992–1993a

Table 2.5. Cropping systems, erosion potential and severity and fertilizer input in agroecological zones.

Environ. Risk. Chap 02 Page 28

L. Muhammad and E. Underwood

Environ. Risk. Chap 02

23/8/04

9:32

Page 29

The Maize Agricultural Context in Kenya

29

smallholder farmers have successfully moved into cash crop production. The main cash crop is horticulture, which now occupies 16% of the cropped area in Kenya, but with large variation among regions (Argwing-Kodeck et al., 1998). Coffee, tea, cotton and sugarcane are also important in some regions. Many farming households in Kenya own some livestock, mainly goats or cattle. Crosses between the East African Zebu cattle and European breeds, known as ‘grade cattle’, are kept for milk production. Dairying has increased over the last decade in the central highlands north of Nairobi, and in the Dry Mid-altitude and Dry Transitional zones. This is the single most important source of regular cash income for about 40% of smallholders, and smallholders in the high-yielding areas produce 80% of the milk in Kenya (World Bank, 1995). In the Moist Mid-altitude and western parts of the Moist Transitional zone, dairy activities have remained stagnant or declined. In the areas where population density is high, cattle are now kept in enclosures called ‘bomas’ in the dry season and fed with crop residues and forage grasses such as Napier grass, which is both planted by farmers and occurs naturally as a weed in and around fields throughout the year. For many farmers, the bottom line for cattle ownership is a viable plough team. Farmers value their plough oxen highly, as farmers without their own plough and oxen generally have to plant and weed too late (Tiffen et al., 1994). However, only a quarter of farming households in the Dry Mid-altitude and Dry Transitional zones own an oxplough and team or tractor, and in the Moist Transitional zone and the Highland Tropics, less than 20% have ownership (Argwings-Kodeck et al., 1998). In the Lowland Tropics, less than 4% of households own an ox-plough or tractor, and almost all soil preparation is done manually. Many agricultural soils in Kenya show a long-term decline in carbon content. Intensive cropping and low input of nutrients mean that soil nutrients are intensively mined in most areas. Many farmers do not apply any nutrient inputs to maize at all, though their use of legumes as intercrops may add nitrogen through nitrogen fixation (Table 2.5). Use of animal manure and mineral fertilizer (Wanzala et al., 2001) is also variable among zones (Table 2.5). In the Moist Transitional and Highland Tropics zones, 60–78% of farmers are applying mineral fertilizer to maize, and over 60% of these are applying 10 or more kg/ha of mostly diammonium phosphate (Argwings-Kodeck et al., 1998; Wanzala et al., 2001). However, although maize receives more fertilizer than any other crop in Kenya (37% of the total fertilizer applied in 1997/98; Wanzala et al., 2001), fertilizer use on maize has declined since 1985. This may be due to the decline in real maize prices, which has been faster than the decline in fertilizer prices (Karanja et al., 1998). Studies showing that fertilizer application in the moist zones is still profitable (Nyoro et al., 2001) are contradicted by others suggesting that the profitability of fertilizers with the currently used varieties is not clear-cut (World Bank, 1995). In the Lowland Tropics, Moist Mid-altitude and Dry Mid-altitude zones, fertilizer use is very low, and in the first two zones, the majority of farmers do not use animal manure. The response to fertilizer in semiarid areas often does not bring a significant economic advantage to farmers. If it does not rain, the investment is lost. Fertilizer can exacerbate water stress in maize, and the soils retain the nutrients from inorganic fertilizers poorly because of their low cation-exchange capacity.

Environ. Risk. Chap 02

23/8/04

9:32

30

Page 30

L. Muhammad and E. Underwood

Maize seed use and cultivars Over half of the maize area in Kenya is planted to hybrid seed (ArgwingsKodeck et al., 1998; Pingali, 2001). In the high-yielding zones (Moist Transitional and Highland Tropics zones) almost all farmers purchase certified hybrid seed every year (but with large regional and individual differences, e.g. in Embu, 71% of farmers are recycling Makueni composite, with the main one being lack of cash; De Groote et al., 2001). In all the other zones, half to threequarters of farmers use recycled seed of improved or local cultivars. Farmers growing open-pollinated (composite) varieties (OPVs) in the lower altitude zones (Lowland Tropics, Dry and Moist Mid-altitude) tend to renew them only years after the recommended 3 years. Some farmers also regrow hybrid seed for several seasons, mostly in the Lowland Tropics and Dry Transitional zone but also in the high-yielding maize areas (Argwings-Kodeck et al., 1998; Hassan et al., 1998c; IRMA, 2002b). However, frequent crop failures in the dry zones lead to some fresh seed input regularly. Maize seed is also exchanged across the country; e.g. in the Dry Mid-altitude and Dry Transitional zone, 7% of farmers stated that they purchase and sow seed intended for food and not for planting, originating from districts in western Kenya (De Groote et al., 2001). The Kenyan Plant Health Inspection Service (KEPHIS) is responsible for cultivar testing and release, and seed quality assurance through inspection, certification and licensing (KEPHIS, 1998). Private companies such as Pioneer and Syngenta have an increasing share of the seed market, especially in Machakos and Embu, though the Kenya Seed Company still dominates (De Groote et al., 2001). However, many farmers consider the quality of seed from the Kenya Seed Company to be unreliable (World Bank, 1995; De Groote et al., 2001). The Highland Tropics and Moist Transitional zones have better seed supply systems, transportation networks and extension services than the other zones, which make it easier for farmers to use hybrid seed every year. Education level, closeness to roads, and assets that can be converted to cash or credit (such as small livestock or farm equipment) also influence hybrid adoption (Karanja et al., 1998). Almost a fifth of Kenyan farmers surveyed in 1997 stated that they have never used improved seeds (Argwings-Kodeck et al., 1998). Farmers do not buy certified seed partly because of the high price and poor availability, but also for other reasons (Hassan et al., 1998c; De Groote et al., 2001). The Lowland Tropics is the only zone where the majority of farmers have never purchased improved seed, either because they were unaware of the improved seed available, or because of lack of access and cash (Hassan et al., 1998c). Most farmers in the Dry Mid-altitude and Dry Transitional zones have purchased improved seed at some time, but they show a clear preference for local varieties. Even in the high-yielding zones, a fifth to a third of farmers are using some seed from local landraces. In the Dry Mid-altitude and Dry Transitional zones, smallholder farmers consider early maturity and drought tolerance to be more important criteria for seed selection than high yield. They also prefer local varieties because of their taste, resistance to shattering (important for pounding technique), cooking

Environ. Risk. Chap 02

23/8/04

9:32

Page 31

The Maize Agricultural Context in Kenya

31

qualities, and resistance to vertebrate and storage pests (Hassan et al., 1998c; Songa et al., 2002a). Even in farms where cash was not limiting, and improved seed was available locally, the main proportion of the maize cropped land in these zones in 1996 was under local varieties. The surveys indicate that the most favoured KARI variety Katumani Composite B, and even more KARI hybrids available at that time, did not give enough yield advantage to compensate for the superiority of local cultivars in qualities other than yield (Hassan et al., 1998c). Farmers in the Moist Mid-altitude, Moist Transitional and Highland Tropics zones are growing a wide range of both improved and local varieties, with most farmers growing more than one cultivar on the farm. In the Moist Mid-altitude zone, tolerance to low soil fertility and Striga are very important for farmers, but they also have diverse perceptions and criteria for selection. Each district has three or four different local varieties, which include both yellow and white cultivars, with only one of these being grown in more than one district (IRMA, 2002a). Farmers in the Moist Transitional and Highland Tropics zones also grow a range of local cultivars, OPVs and hybrid varieties. They keep track of new cultivars and test them before adopting them (De Groote et al., 2001). KARI have in the past released significantly more improved late-maturing varieties in these zones than medium- or early-maturing varieties for the midaltitude zones and Lowland Tropics (Hassan et al., 1998c). Local varieties have not been investigated so far in Kenya to see to what extent they comprise distinct local landraces or are more like advanced generations of released OPVs (IRMA, 2002b). They are very likely to be a mix. For instance, in the Dry Mid-altitude zone, the majority of farmers using recycled seed stated that they were using a local cultivar called Machakos Local White. This is close to the most commonly used improved open-pollinated cultivar (Katumani Composite), which was released over 25 years ago and is often replanted by farmers, and it could therefore be a mix of local landraces with genetic material from Katumani Composite (Songa et al., 2002a). Some of the OPVs were originally bred from landraces. Almost all of Kenya’s maize cultivars are white endosperm maize (FAO/CIMMYT, 1997). Kenyan consumers prefer white endosperm maize cultivars over yellow endosperm maize, which has negative associations with food aid.

Constraints on maize production A number of projects have gathered farmers’ views on the constraints they are under with regard to maize production. The most recent and thorough of these was carried out by the Insect Resistance Management for Africa (IRMA) project in 2000 (Box 2.1 on p. 47; De Groote et al., 2001). The participatory rural appraisal (PRA) asked farmers to rank their major constraints to maize production (Table 2.6), and then to rank the importance of different pests, but not diseases (Table 2.7). Farmers were also asked to rank their main criteria for selecting maize cultivars. In 1992, the Maize Data Base Project (MDBP) (Hassan et al., 1998b) asked farmers to estimate the damage caused by

Environ. Risk. Chap 02

23/8/04

9:32

Page 32

32

L. Muhammad and E. Underwood

Table 2.6. Farmers’ ranking of constraints in all phases of maize production in Kenya (De Groote et al., 2001). Agroecological zone

Ranked production constraint Rank 1

Rank 2

Rank 3

Mtwapa (Coastal Lowland Tropics)

Field pests

Cash

Katumani (Dry Mid-altitude)

Rain

Katumani (Dry Transitional)

Rank 4

Rank 5

Rank 6

Soil fertility Wildlife

Drought

–

Pests and diseases

Cost of inputs

Seed availability

Know-how

–

Rain

Know-how

Pests and diseases

Input cost

Poverty

–

Embu (Moist Transitional)

Cash

Rain

Know-how

Seed cost

Stemborers Low fertility

Kitalse (Highland Tropics)

Poor seed quality

Seed price

Fertilizer price

Low maize Cash price

Pests

Kakamega (Moist Mid-altitude)

Farm Soil fertility implements

Cash

Extension Certified (know-how) seed availability

Pests

–, no response given.

Table 2.7. Farmers’ ranking of pest problems in all phases of maize production in Kenya (De Groote et al., 2001). Agroecological zone

Ranked pest problem Rank 1

Rank 2

Rank 3

Rank 4

Rank 5

Mtwapa (Coastal Lowland Tropics)

Rodents

Stemborers

Weevils

Beetles

Storage moths

Katumani (Dry Midaltitude)

Weevils

Stemborers

Chafer grubs Termites

Katumani (Dry Transitional)

Weevils

Chafer grubs Stemborers

Termites

–

Embu (Moist Transitional)

Stemborers

Weevils

Squirrels

–

–

Kitale (Highland Tropics)

Stemborers

Weevils

Cutworms

Rodents

–

Kakamega (Moist Mid-altitude)

Striga

Weevils

Stemborers

Termites

Rodents

Squirrels

–, no response given.

problems according to area, yield loss, frequency and season. From this, the causes of the perceived yield loss can be ranked in priority for farmers (Table 2.8). Along with rainfall, farmers in the IRMA PRA ranked lack of cash, credit and tools, cost of fertilizer, lack of technical know-how and poor extension services, and problems with maize seed (high cost, poor quality and low

Environ. Risk. Chap 02

23/8/04

9:32

Page 33

The Maize Agricultural Context in Kenya

33

Table 2.8. Ranking of farmers’ perceived yield losses and productivity constraints in field production of maize in Kenya by agroecological zone (excludes losses in storage, based on Hassan et al., 1998b). Agroecological zone Rank 1

Ranked perceived severity of yield loss Rank 2

Rank 3

Rank 4

Chafer grubs, Stemborer weeds,b other pestsc

Other diseasesc Other pestsd

Lowland Tropics

Game animals

Rainsa

Dry Midaltitude

Low soil fertility

Rains

Chafer grubs, Streak virus weeds

Dry Transitional

Rains

Stemborer

Other pestsc

Moist Transitional

Low soil fertility

Streak virus Hail

Highland Tropics

Head smut Rainsa

Moist Mid-altitude

Streak virus

Striga

Rank 5

Rank 6

Stemborer

Chafer grubs, Stalk weedsb lodging

Head smut

Stemborer

Stalk lodging

Rains,a head smut

Stemborer

Hail

Stalk lodging

Weedsb

Other pestsd

Low soil fertility

Rainsa

Weeds,b chafer grubs, stemborer

aBoth

inadequate and erratic or unfavourable distribution of rainfall, depending on the zone in question. bWeeds other than Striga. cDiseases other than head smut and streak virus, such as stalk and ear rots and leaf blights. dPests other than stalk borer, chafer grubs or game animals. The most important among these were termites, cutworms, weevils, grain borers, rodents and birds.

availability) highly across all agroecological zones (Table 2.6). These factors constrain farmers’ ability to deal with other problems: the farmers in the Moist Mid-altitude zone stated that cash constraint is a major problem whose alleviation would lead to mitigation of many other constraints (IRMA, 2002a). The MDBP survey did not include consideration of input costs or access to information or credit, and therefore does not imply that these are not problems for farmers (Table 2.8), as the IRMA PRA revealed (Table 2.6). In addition, surveys were carried out with farmers in semiarid eastern Kenya in 1996 (Songa et al., 2002a), and with farmers on the coast in 1997 (Bonhof et al., 2001). All surveys found that inadequate and erratic or unfavourable distribution of rainfall is a priority problem in all agroecological zones (Tables 2.6 and 2.8). Low soil fertility was also identified as a major problem in the Dry Mid-altitude, Moist Mid-altitude and Moist Transitional zones, but not in the Highland Tropics. In the Lowland Tropics, farmers ranked low soil fertility below insect pest pressure (Bonhof et al., 2001). In the high-yielding maize zones (Moist Transitional and Highland Tropics), farmers estimated that hailstorms and stalk lodging caused high crop losses. Weeds were considered a significant constraint

Environ. Risk. Chap 02

34

23/8/04

9:32

Page 34

L. Muhammad and E. Underwood

in all zones. In the Moist Mid-altitude zone, the most important constraint is the parasitic weed Striga, ranking above all the other pest problems mentioned (Table 2.7). Farmers in all the zones considered diseases important constraints (Table 2.8). Streak virus was considered the most important pest or disease constraint in the Moist Mid-altitude zone, and as very important in the Dry Midaltitude and Moist Transitional zones (Table 2.8). The other most mentioned disease was head smut, but other fungal diseases were also mentioned. Farmers considered damage from animal pests to be of high significance in the drier areas (Lowland Tropics, Dry Mid-altitude and Dry Transitional; Tables 2.6 and 2.8). The most important insect pests mentioned were storage pests, root feeders (chafer grubs – Melolonthinae, cutworms), termites and stemborers (Tables 2.6–2.8). Damage caused by birds and vertebrates, such as the African striped ground squirrel, rats, elephants, wild pigs, porcupines or guinea fowl, were also considered an important problem in the field (Bonhof et al., 2001; Songa et al., 2002a). In the Lowland Tropics, field and storage pests were considered the most important constraint (Tables 2.6 and 2.8), above soil fertility and drought. In a survey, nearly all farmers in the Lowland Tropics clearly recognized and knew about stemborer larvae, though only 66% knew that moths and larvae were life stages of the same pest (Bonhof et al., 2001). Most farmers in this zone also considered armyworm (Spodoptera spp.) a pest in some years. Storage pests comprise mainly the grain weevil (Sitophilus zeamais), the larger grain borer (Prostephanus truncatus), the angoumois grain moth (Sitotroga cerealella) and rodents. Most maize is stored near to fields, and most of these pests are a problem both in the field and in storage. Women are mainly responsible for postharvest management (Hassan and Salasya, 1994). Better storage quality and techniques would give farmers more flexibility in when they sell their maize, enabling them to get better prices (Mugo et al., 2001a). In the moist zones (Highland Tropics, Moist Transitional and Moist Mid-altitude), insect pests ranked behind other constraints but were still ranked significant (Tables 2.6 and 2.8). The damage due to stalk lodging in the Highland Tropics and Moist Transitional zones could partly be a secondary effect of stemborer damage, combined with grey leaf spot or the stalk rots that enter damaged tissue.

Some Major Insect, Disease and Weed Pests of Maize in Kenya The next section describes some of the main pest problems as ranked by Kenyan farmers. We focus on Striga, maize streak virus, fungal pathogens and stemborers. More detail is given for stemborers, being the target pests of Bt maize. Other weeds, termites, soil macroorganisms, storage pests (including vertebrates, insects and microbes) and vertebrate pests, such as rodents, though also ranked highly by farmers, are not addressed further in this chapter. However, they are likely to be significant constraints, as they are in Ethiopia and elsewhere in Africa (Ferdu et al., 2001; Pingali, 2001). Possible non-target impacts of Bt maize on these pests are discussed in Birch et al. (Chapter 5, this volume). Some of the storage pests could pose a resistance risk on Bt maize and are discussed in Fitt et al. (Chapter 7, this volume).

Environ. Risk. Chap 02

23/8/04

9:32

Page 35

The Maize Agricultural Context in Kenya

35

Striga Striga (witchweed) species are weeds that are obligatory root parasites attacking most cereal crops in Africa. They are considered to be one of the most important biological constraints to maize production in Africa, and the problem is growing both in area and intensity. Striga exerts a potent phytotoxic effect on maize, and the majority of the Striga-induced yield loss occurs before the Striga plant emerges from the soil. The plants produce massive amounts of tiny dust-like seeds that can remain in the soil seed bank for several years. At a density of 20 plants/m2, a density which is commonly observed in farmers’ fields, millions of seeds can be added to each square meter of soil each year. As few as two or three flowering Striga plants/m2 produce sufficient seeds to maintain Striga seed numbers at damaging levels (Ransom, 2000). Striga hermonthica and Striga asiatica are the common species in Kenya. S. hermonthica attacks all of the major tropical cereal crops and sugarcane, and also grows under wild conditions. It was found to infest 39% of the maize area in the Moist Mid-altitude zone in 1992 (Hassan and Ransom, 1998). However, there is evidence that it is moving into the Moist Transitional zone (the largest maize growing area), where the area with suitable climatic conditions for infestation is much larger than the current area of infestation. Striga species are adapted to a wide range of environmental conditions in tropical Africa, occurring in areas with rainfall ranging from 400 to 1000 mm/year, so S. hermonthica could potentially spread into most of the Kenyan agroecological zones except the Highland Tropics, which is too cold and wet. S. asiatica occurs locally along the coast and in isolated areas of western Kenya, and parasitizes wild vegetation more frequently than crops. Farmers in the Lowland Tropics did not report Striga as a problem, perhaps because S. asiatica is less aggressive, but also because maize production is broken up by large areas of tree crops where Striga is rare. Smallholder farmers control Striga mainly by hand weeding, although this is too late to have any effect on that seasons’ crop, and investigations have shown that after infestation has become severe, farmers would need to weed consistently for more than three seasons before seeing an effect (Ransom and Odhiambo, 1994). Most farmers in the Moist Mid-altitude zone are also aware that adding fertilizer or manure reduces Striga infestation, but only 24% of farmers use inorganic fertilizer and 52% use manure on maize. Intercropping with a crop that does not host Striga reduces infestation on the maize, but only increases yields if the competitive effect of the intercrop is low. One promising intercrop for Striga control is Desmodium, part of the ‘push–pull’ system (see below). Farmers in the Moist Mid-altitude zone say that some of their local cultivars have better tolerance to Striga than the available commercial hybrids (De Groote et al., 2001), but this has not so far been evaluated experimentally (Hassan and Ransom, 1998). Tolerance, if it exists, may partly be due to a longer maturity time, as maize cultivars that grow more slowly tolerate Striga better than rapidly maturing maize (Ransom and Odhiambo, 1995). Some soils in western Kenya also show an inherent suppressiveness of Striga germination, which leads to much greater seed bank decline than any other management

Environ. Risk. Chap 02

23/8/04

36

9:32

Page 36

L. Muhammad and E. Underwood

practice (Odhiambo, 1998). The mechanism is associated with microbial activity because soil suppression is significantly decreased by soil fumigation (Odhiambo, 1998).

Maize streak virus Maize streak virus (MSV) is the most important and widespread viral disease of maize in Africa (Bosque-Pérez, 2000). MSV reduces yield and growth of maize and, in susceptible cultivars, yield reductions can exceed 70% if the plant is infected at an early stage. Heavy outbreaks of MSV occurred in Kenya in 1988 and 1994. In a 1995–1996 survey, MSV infection was severe in the Moist Mid-altitude zone, in the central highlands, and in the Lowland Tropics (Mwangi, 1998). MSV is persistent and transmitted by nine species of leafhopper of the genus Cicadulina (=Balclutha), of which Cicadulina mbila is the most important vector in East Africa. Cicadulina spp. are grassland species, and MSV is presumed to be an endemic African virus that spread from wild grasses to maize with its leafhopper hosts. The host transmission ability of C. mbila is a genetically variable trait, related to the permeability of the insect’s gut. Consequently, there are both virus transmitters and non-transmitters in a leafhopper population. The leafhoppers can pick up the virus quickly from infected plants when feeding on mesophyll cells. Inoculation of MSV into healthy maize takes longer as it depends on the insects reaching the phloem tissue and salivating into it. The plant hosts that harbour MSV when maize is unavailable are not well known. Some grass species may form a reservoir for maize infection, but there is also evidence that the virus types found in grasses do not infect maize or cause only mild symptoms, whereas the maize types are virulent in maize but do not infect grasses (Mesfin et al., 1992; Rybicki et al., 1998). Napier grass (Pennisetum purpureum) in Kenya was found to be a good host for C. mbila but does not host MSV (Njunguna, 1996). MSV disease epidemics are spreading along with Cicadulina in Africa, and this is probably due to increases in area and intensity of maize cultivation, including irrigated maize, increase in maize monoculture cropping, and introduction of new susceptible genotypes of maize. The most popular cultivars in Kenya are completely susceptible, but new germplasm with improved resistance to MSV has been developed by KARI and CIMMYT using conventional breeding and marker-assisted selection (Diallo and Friesen, 2001). Some KARI hybrids show only mild reactions to MSV and maintain yield under infection (Ininda et al., 2002).

Fungal pathogens Significant fungal diseases on maize in Kenya are northern leaf blight caused by Exserohilum turcicum, leaf spot caused by Phaeosphaeria maydis, southern leaf blight caused by Bipolaris maydis, grey leaf spot caused by Cercospora

Environ. Risk. Chap 02

23/8/04

9:32

Page 37

The Maize Agricultural Context in Kenya

37

zea-maydis, lowland rust caused by Puccinia polysora, common rust caused by Puccinia sorghi and head smut caused by Sphacelotheca reiliana (Mwangi, 1998; Ininda et al., 2002). There was a severe outbreak of northern leaf blight in 1999 and in a survey in 1995–1996, it was observed in all maize-producing areas of Kenya, with disease incidence reaching over 45% (Mwangi, 1998). The most popular Kenyan cultivars were found to be totally susceptible. KARI has now released new hybrids with a high level of resistance (Ininda et al., 2002). The survey also found severe disease infestations (over 65%, sometimes over 85%) from southern leaf blight, often together with leaf spot, in western Kenya, and leaf spot in the central highlands. Common rust was severe in the high-yielding maize area of Nakuru, and lowland rust was severe in the Lowland Tropics. Kenya has also experienced increasingly severe incidences of head smut (Ininda et al., 2002). Some fungal infections of maize kernels, occurring either on the plant or in storage, can produce toxic compounds that constitute a health risk for Kenyan consumers and livestock. Aspergillus flavus is widespread in Africa and produces extremely carcinogenic and hepatotoxic aflatoxins (Bankole and Adebanjo, 2003). Kenyan maize is widely contaminated. A. flavus is saprophytic in the soil, growing on a wide range of substrates, and can also grow in maize plants. It infects maize, usually through damaged or broken kernels in the developing ear. Maize stemborers, the large grain weevil Sitophilus zeamais, the false codling moth Cryptophlebia leucotreta, the sap beetle Carpophilus spp. [Coleoptera: Nitidulidae] and drought stress can damage kernels allowing A. flavus to infect (Sétamou et al., 1998; Odvody and Chilcutt, 2002). Some stemborer larvae may also transport fungal spores within the plant and from the soil to the plant. A. flavus can also be spread in storage when conditions are suitable for fungal growth (Hell et al., 2000). Insect pests of stored maize, including S. zeamais and C. leucotreta, carry fungal spores and damage kernels thereby creating a substrate for fungal growth, inoculating the substrate with spores and increasing aflatoxin levels. Fusarium verticillioides is also widely distributed in Kenyan maize, but is not considered a problem by farmers as most infections are asymptomatic and do not directly affect yields (Munkvold et al., 1997). However, Fusarium spp. also produce mycotoxins such as fumonisin, which is carcinogenic and immunotoxic in humans, livestock and other mammals. F. verticillioides has an ecology similar to A. flavus, being both saprophytic in soils and able to infect and grow in living maize plants. Fusarium spores are transmitted within and between plants by maize stemborer larvae (Ako et al., 2002; Schultess et al., 2002). Stemborer larval frass and faeces are an ideal medium for the development of the fungus. In addition, however, F. verticillioides infection actually promotes infestation by stemborers by increasing survival of older larvae in the upper stem and ear. The mechanism of this is unknown, but it is speculated that systemic F. verticillioides infection is interfering in some way with host plant insect resistance mechanisms (Ako et al., 2002; Schultess et al., 2002). Fusariuminfected plants are also more attractive to ovipositing females of Eldana saccharina stemborers. F. verticillioides produces a volatile compound that

Environ. Risk. Chap 02

23/8/04

9:32

Page 38

38

L. Muhammad and E. Underwood

is attractive to nitidulid beetles such as Carpophilus spp. (though not S. zeamais) that may facilitate transmission. The attraction of these beetles and female stemborers to Fusarium-infected plants could lead to increased infection levels of both F. verticillioides and A. flavus, and a corresponding increase in both aflatoxin and fumonisin levels in maize grain.

Maize stemborers Cereal stemborers attack maize, sorghum, sugarcane, millet, rice and wild grasses. In the case of maize, which does not produce tillers to replace the damaged stem, any stem damage can result in grain and possibly stover yield reduction. Quality can also be adversely affected. The larvae of most stemborer species feed first on the leaf whorls, and after reaching a certain size they penetrate the stem, and consume the maize stem. Later generations penetrate the tassel or ear. Larvae pupate within the stem or cobs, cutting themselves an emergence hole before they do so. Stemborer species complex Five stemborer species occur in maize in Kenya: Chilo partellus (Crambidae), Chilo orichalcociliellus (Crambidae), Sesamia calamistis (Noctuidae), Busseola fusca (Noctuidae) and Eldana saccharina Walker (Crambidae). B. fusca and C. partellus are the dominant pest species; the introduced species C. partellus across lowland and dry regions, while the native species B. fusca is generally dominant in the moist regions and highlands (Plate 3). Table 2.9 summarizes the results of several stemborer surveys in the 1990s in 269 maize fields (referred to as survey A), with mean densities, maximum densities and percentage of sampling sites occupied for each species (Zhou et al., 2001b). However, these results underestimate the higher yielding zones, due to the much lower number of sampling sites in the Moist Transitional and Highland Tropics zones. Better estimates for these areas will soon be available from another survey (referred to as survey B) that has been carried out from March 2001 to November 2003 at 871 sampling sites (B. Le Ru, Nairobi, January 2004, personal communication). B. fusca occurs in all areas of Kenya up to 2600 m except the Lowland Tropics and is the dominant maize and sorghum stemborer species in the Moist Transitional zone and Highland Tropics. In survey A, it accounted for 92% of the stemborers present in almost all of the sites. Stemborer density generally was found to be lower in this zone, but B. fusca is much larger than C. partellus and one individual causes much more damage. Survey A also found that B. fusca was relatively abundant in the southern halves of the Moist Midaltitude and western Moist Transitional zone, making up nearly a quarter of the stemborer complex at three-quarters of the sites, and at lower densities in the northern part of the Moist Mid-altitude zone; survey B found that B. fusca made up a little under two-thirds of the stemborer complex in the Moist Midaltitude zone. Stemborer densities are dynamic; different species may have

2.71 3.22 0.15 0.83

0.74

Lowland Tropics

Dry Mid-altitude, Dry Transitional

Highland Tropics, east Moist Transitional

Southwest Moist Transitional, south Moist Mid-altitude

Northwest Moist Transitional, north Moist Mid-altitude

2.86

4.21

0.83

21.64

12.80

max

90

85

30

100

100

%

0.33

0.51

1.45

0.07

–

x

1.82

1.71

5.5

0.85

–

max

72

73

85

23

–

%

0.13

0.1

0.18

0.47

0.47

x

0.57

0.27

0.37

1.67

2.00

max

Sesamia calamistis

65

67

23

96

92

%

0

0.11

–

–

–

x

0

0.32

–

–

–

max

Eldana saccharina

0

65

–

–

–

%

–

–

–

0.06

0.49

x

–

–

–

0.20

2.13

max

–

–

–

7

95

%

Chilo orichalcociliellus

9:32

x , mean density (larvae per plant); max, maximum density (larvae per plant); %, percentage of sampled sites infested; –, species not present (Zhou et al., 2001b).

x

Busseola fusca

23/8/04

Agroecological zone

Chilo partellus

Table 2.9. Stemborer distribution and density in the 1990s.

Environ. Risk. Chap 02 Page 39

The Maize Agricultural Context in Kenya 39

Environ. Risk. Chap 02

23/8/04

9:32

40

Page 40

L. Muhammad and E. Underwood

different relative densities from year to year, particularly in the mid-altitude and transitional zones (B. Overholt, Florida, 2003, personal communication; B. Le Ru, Nairobi, January 2004, personal communication). Therefore, the relative densities of B. fusca and C. partellus in the Moist Mid-altitude and eastern Moist Transitional zones may vary from year to year so that one year the first species is dominant, another year the other. B. fusca was found at very low densities in the Lowland Tropics in survey B (B. Le Ru, Nairobi, 2004, personal communication), and was found at low frequencies above 800 m in the Dry Mid-altitude zone and Dry Transitional zone in survey A, with over 85% of occupied sites above 1000 m. C. partellus is an exotic species that was first reported in Kenya in the 1950s (Nye, 1960), and has spread throughout the maize and sorghum growing areas of Kenya at elevations below 1500 m, and sometimes higher (Overholt et al., 1994a; Zhou et al., 2001b; Songa et al., 2002b). The Dry Midaltitude and Dry Transitional zones are dominated by C. partellus everywhere; survey A found a mean of more than three larvae per plant and a maximum of 21 larvae per plant in these zones, and it was also the dominant species in survey B. C. partellus, like most stemborers, has an aggregated distribution in its early life stages because eggs are laid in groups. Larvae become progressively less aggregated as they age due to mortality and dispersal (Overholt et al., 1994b). In the Lowland Tropics, C. partellus accounted for about two-thirds of the total stemborer complex in survey A and was present everywhere, and was also the dominant species in survey B. However, densities vary greatly from year to year, partly because of the rainfall variability (Mathez, 1972). As above, survey A found that C. partellus dominated in the Moist Mid-altitude and eastern Moist Transitional zones, accounting for about 60% of the total stemborer population and present in 90% of the sampling sites in both the northern and southern parts of this area, whereas survey B found that C. partellus made up a little under 40% of the stemborer complex in the Moist Mid-altitude zone. Only in the Highland Tropics is C. partellus relatively rare, but it was none the less present at low densities at 30% of the sites sampled in survey A. S. calamistis is a native stemborer species that occurs in low densities in all areas of Kenya up to 2400 m. In survey A, it made up about a sixth of the stemborer complex in most fields, except in the highlands where it was less frequent. C. orichalcociliellus is a native stemborer species that made up about a sixth of the stemborer complex in nearly all fields of the Lowland Tropics in survey A, and was also present in very low densities in the Dry Mid-altitude zone. E. saccharina is a pest of maize, sorghum and sugarcane in western Kenya, and has only been recorded in sorghum or maize in Kenya since the 1950s. In survey A, it was found only at low densities in two-thirds of the sampled fields in the southern half of the Moist Mid-altitude zone. Stemborer ecology C. partellus may be displacing some of the indigenous stemborer species in Kenya, e.g. C. orichalcociliellus on the coast (Overholt et al., 1997; Ofomata

Environ. Risk. Chap 02

23/8/04

9:32

Page 41

The Maize Agricultural Context in Kenya

41

et al., 1999, 2000) and B. fusca in the Dry Mid-altitude zone (Songa et al., 1998, 2001; however B. Le Ru, Nairobi, January 2004, personal communication, provides an alternative perspective). This may be due to its competitive superiority. C. partellus has a high reproductive rate, emerges from diapause and attacks maize earlier than the other species, and has a shorter development time (Kfir, 1997; Mbapila et al., 2002). When equal numbers of C. partellus and C. orichalcociliellus infest the same maize, sorghum or wild sorghum plant, more C. partellus successfully completed development (Ofomata, 1997). However, there is also evidence that C. orichalcociliellus has a higher survival rate in some wild grasses than C. partellus, and may therefore be able to coexist in typical Kenyan landscapes (Ofomata et al., 2000). C. partellus is most abundant in lowland areas, but temperature does not restrict its range. It occurs in Kenya in some highland areas at elevations of up to 2300 m, and diapausing larvae can survive cold temperatures (Zhou et al., 2001b). In South Africa, C. partellus has partially displaced B. fusca at 1600 m (Kfir, 1997), so species shifts in the higher altitude areas in Kenya may occur in the future. During the dry season when there are no cultivated crops in the field, mature stemborer larvae will diapause in crop residues left in or near the fields (Scheltes, 1978). In Kenya, B. fusca always diapauses, whereas in C. partellus populations, some individuals diapause in crop residues and others continue feeding on wild sorghum grasses (Mathez, 1972; Unnithan, 1987; Kfir, 1991; Mbapila, 1997). Diapause in both B. fusca and C. partellus is induced by a decline in the nutritive quality of the host plant, caused by drying and ageing (Kfir et al., 2002). C. partellus and C. orichalcociliellus can emerge from dry stalks left in the field at any time after harvest (Mathez, 1972). S. calamistis and E. saccharina are also capable of developing throughout the year. Some stemborer species are also found on wild host plants, from which they can re-infest the maize and sorghum fields (Polaszek and Khan, 1998). Adult moths actually prefer some wild and fodder grasses over maize for oviposition, although larval survival may be much lower than that on maize. For example, C. partellus larval survival on Napier grass (P. purpureum) is very low, as the grass tissues produce a sticky sap that traps the larvae (Khan et al., 2000). On the other hand, wild sorghum (Sorghum versicolor) and Sudan grass (Sorghum vulgare Pers. var. sudanense), both commercial fodder grasses, support survival of C. partellus at a rate similar to survival on maize and cultivated sorghum plants (Khan et al., 1997; Songa et al., 2002c). A recently completed 1-year survey of more than 45 different wild host grasses in all of south Kenya (and more than 7000 Lepidopteran larvae) found that almost 90% of the C. partellus outside maize was in wild sorghum (B. Le Ru, Nairobi, January 2004, personal communication), as did other surveys in the Lowland Tropics, Dry Mid-altitude and Dry Transitional zones (Ofomata et al., 1999; Songa, 1999). Wild grass populations around or in maize fields can therefore function both as a source of reinfection and as a trap crop reducing infestation (Cardwell et al., 1997). Some species may also function as effective refuges (Fitt et al., Chapter 7, this volume). However, wild host preference seems to be species specific. C. partellus was seldom found in Panicum spp., whilst almost 90% of the C. orichalcociliellus outside of maize was in these grasses. B. fusca,

Environ. Risk. Chap 02

23/8/04

9:32

Page 42

42

L. Muhammad and E. Underwood

on the other hand, was very rarely found in wild hosts, and never when the wild hosts were more than a few metres from the crop, indicating that it is specialized on maize and cultivated sorghum, and reinfects fields from diapausing larvae or other crops rather than wild refuges (B. Le Ru, Nairobi, January 2004, personal communication). Crop loss due to stemborers Past and present estimations of yield loss from maize stemborers exhibit high variability. Researchers in Kenya have measured average yield losses from artificial and natural stemborer infestations between 5% and 73% (Seshu Reddy and Walker, 1990; Ajala and Saxena, 1994; Mulaa, 1995; Gethi et al., 2001; Songa et al., 2002d). However, using measured yield losses to estimate the losses in large areas under natural conditions is often misleading, particularly in the very heterogeneous environments and management conditions in Kenya (De Groote, 2002). Farmers in all agroecological zones in Kenya estimate that the stemborers cause significant yield loss in their maize crops (Grisley, 1997; Hassan et al., 1998b). A 1992 survey asked farmers to estimate the amount of yield they lose to stemborers (Hassan et al., 1998b); however, these farmer estimates have not been checked against field measurements (De Groote, 2002). In the lower altitude and dry zones where average yields are below 1.4 t/ha (Lowland Tropics, Dry Mid-altitude zone, Dry Transitional, Moist Mid-altitude zones), farmers believed losses to stemborers to be 14–20% of their potential production. In the zones with average yields of 2.6–2.9 t/ha (Moist Transitional and Highland Tropics), farmers believed losses to be around 10–12% (De Groote, 2002). As 80% of Kenyan maize is produced in the Moist Transitional and Highland Tropics zones, losses to stemborers in these zones could be a significant share of the potential national production (IRMA, 2001). From the stemborer distribution surveys, it can be expected that most of these potential losses will be due to B. fusca and C. partellus. Survey B (B. Le Ru, Nairobi, January 2004, personal communication) found B. fusca responsible for almost all of the crop losses to stemborers in the Moist Transitional and Highland Tropics zones. In the Moist Mid-altitude zone, B. fusca caused 59% of the crop losses due to stemborers, and C. partellus 30%. C. partellus was responsible for 86%, 58% and 93% of stemborer losses respectively in the Lowland Tropics, Dry Mid-altitude and Dry Transitional zones. Across the entire country, B. fusca therefore causes greater yield losses than C. partellus; however, in terms of the number of rural households dependent on maize production (Table 2.4), the C. partellus-caused losses in the Dry Transitional and Moist Mid-altitude zones are significant. The surveys confirm that farmers perceive stemborers as a widespread pest problem in Kenya. Stemborer damage might increase if maize double-cropping increases, allowing two or more generations of stemborers on maize (which is a better food resource than wild hosts) in 1 year (Ofomata et al., 1999; Zhou et al., 2001a). Surveys in the Lowland Tropics have shown that densities of all three stemborer species increased linearly from 1992 to 1997, which could correspond

Environ. Risk. Chap 02

23/8/04

9:32

Page 43

The Maize Agricultural Context in Kenya

43

to increases in maize cropping frequency, whereas until the early 1990s, densities were more or less stable (Ofomata et al., 1999; Zhou et al., 2001a). Yield loss due to stemborer damage has been shown to be influenced by cultivar, by time of damage and by number of larvae involved in the infection (Gebre-Amlak et al., 1989; Seshu Reddy and Sum, 1991, 1992); e.g. in Ethiopia second-generation larvae caused significantly higher losses than first-generation larvae (Gebre-Amlak et al., 1989). Studies in the Lowland Tropics have found two generations of C. partellus per maize crop in some years (Mathez, 1972; Warui and Kuria, 1983; Overholt et al., 1994b). In many years, stemborer infestations may also be more severe in the second growing season (the ‘short rains’) due to the build-up of populations in the first season, particularly when the period between the first and second season is short, as it is in the Lowland Tropics (Overholt et al., 1994b). In the low-altitude areas, stemborer damage (holes in the leaves) can be noticed 2–3 weeks after planting, whereas in the high-altitude areas of the western Rift Valley and Central Provinces of Kenya, the damage is not noticeable until 4 weeks after the crop has germinated (Overholt et al., 1994b). Control strategies for stemborers Approaches to stemborer control in Kenya fall into four broad categories: chemical control (application of insecticides), biological control (use of native and/or introduced natural enemies of stemborers), cultural control (entails the use of a range of farm practices to delay or reduce insect attack) and host plant resistance (the plant offers its own resistance to insects). At present, two new strategies are being developed: host plant resistance using Bt transgenes, and cultural control using the push–pull integrated crop management system. This is the most widely used approach to control stemborers in Kenya. Bulldock (beta-cyfluthrin) is currently the most widely available and used, followed by Dipterex (tricholorfon), Thiodan (endosulfan), Ambush (permethrin) and Pymac (the residue from pyrethrum processing), but farmers use a wide range of chemicals. Most farmers are aware of chemical insecticides as a control method, but for many they are too expensive or not available (Bonhof et al., 2001). However, smallholder farmers use pesticides frequently on cash crops such as horticulture, coffee, tea or cotton. In the Dry Mid-altitude and Dry Transitional zones, Bulldock is used by 40% of farmers and is the most used control method (Songa et al., 2002b). However, farmers that said they used it actually used it only occasionally when money was available, limiting application to sections of the maize field showing signs of stemborer infestation at the vegetative stage. In the Lowland Tropics, 20–40% of farmers had used insecticides to control stemborers, and more than 90% of the farmers were aware of this control method (Bonhof et al., 2001). On the other hand, in the Moist Mid-altitude zone, very few farmers use synthetic pesticides (IRMA, 2002a). In the central highlands, farmers reported using chemical insecticides on maize to control weevils only, though they frequently used pesticides on coffee and horticultural crops (Grisley, 1997). CHEMICAL CONTROL USING SYNTHETIC PESTICIDES.

Environ. Risk. Chap 02

23/8/04

9:32

44

Page 44

L. Muhammad and E. Underwood

Local and botanical pesticides can be effective and cheap. Many farmers in Kenya apply dry soil or wood ash to the maize whorl when they notice stemborer damage. In the Dry Mid-altitude and Dry Transitional zones, wood ash was the second most commonly used stemborer control method (Songa et al., 2002a), and in the central highlands, half of the farmers put dry soil in the maize funnel, and a fifth used wood ash (Grisley, 1997). Farmers are also using dry battery cell powder, chilli pepper powder, sawdust, Mexican marigold (Tagetes minuta), pyrethrum (Chrysanthemum pyrethrum) and extracts of neem tree leaves or bark.

USE OF PEST CONTROL AGENTS MADE FROM LOCAL MATERIALS.

This method relies on reducing the stemborer populations to below seriously damaging levels using native or introduced natural enemies such as predators, parasitoids and pathogens (Birch et al., Chapter 5 this volume). The gregarious larval parasitoid, Cotesia flavipes (Cameron), was introduced at the Kenyan Coast in 1993 for the control of C. partellus, and has established successfully (Overholt et al., 1997). It and the native Cotesia sesamiae now account for 83% of the parasitized borers in southern coastal Kenya (Zhou et al., 2003). A recent study suggested that the density of C. partellus in southern coastal Kenya has declined by about 30 to 50% due to parasitism by C. flavipes (Zhou et al., 2001a). It was also released in some parts of semi-arid eastern Kenya, and preliminary results indicate successful colonization of this parasitoid on C. partellus and S. calamistis populations. It has now spread to some other parts of the country, including areas where B. fusca occurs; however, C. flavipes cannot successfully develop in B. fusca, which is mainly parasitized by C. sesamiae, unless the B. fusca larva is parasitized by both Cotesia species at the same time, in which case C. flavipes can develop at the expense of C. sesamiae (Ngi-Song et al., 2001). In other countries, such as Ethiopia (Emana et al., 2001) and Uganda (Kauma, 2000), it successfully parasitizes B. fusca. In 2003, another Asian parasitoid, Xanthopimpla stemmator (Thunberg) [Ichneumonidae, Hymenoptera], was released in several countries in eastern and southern Africa. Laboratory studies showed that it successfully parasitizes C. partellus, S. calamistis, E. saccharina and B. fusca (Gitau, 2002). Releases of the egg parasitoid Telenomus isis Polaszek [Scelionidae, Hymenoptera] from West Africa are also planned, in order for it to parasitize noctuid eggs, such as S. calamistis and B. fusca (F. Schultess, Nairobi, 2004, personal communication). Ongoing bitrophic tests are being conducted at the International Centre for Insect Physiology and Ecology (ICIPE). Parasitoids cannot locate and parasitize diapausing larvae in crop residues, therefore they rely on the stemborer populations on forage or wild grasses between cropping seasons (Mbapila and Overholt, 1997). Farmers can enhance populations of parasitoids and other natural enemies by managing wild grass populations around their fields to maintain wild stemborer species as alternate hosts (Khan et al., 1997; Songa et al., 2002b). Some areas of Kenya already have many wild refuges for natural enemies near maize fields (e.g. the Lowland Tropics), but other areas have less, e.g. the semi-arid areas (Dry Mid-altitude and Dry Transitional) (Songa, 1999; IRMA, 2002c; Songa et al., 2002e). BIOLOGICAL CONTROL.

Environ. Risk. Chap 02

23/8/04

9:32

Page 45

The Maize Agricultural Context in Kenya

45

The destruction of crop residues after harvest destroys diapausing stemborers and reduces the populations re-infesting the field during the next growing season. However, most farmers leave the stalks in the field to provide soil cover and improve soil fertility, or store them nearby for use as fodder (Bonhof et al., 2001; Songa et al., 2002b). Few farmers demonstrated knowledge of diapausing stemborers. In a survey on the coast, only 12 farmers out of 240 were aware that stemborers are carried over in crop residues, and only two of these actually destroyed crop residues in order to control stemborers (Bonhof et al., 2001). Mathez (1972) also found that most farmers on the coast left the dry stalks standing in the field until the next ploughing, and planted maize twice a year, so that most stemborer infestation came directly from the crop residues. On the other hand, crop residue destruction by shredding and ploughing in after harvest can be a very successful strategy for large-scale farmers (Kfir et al., 1989). Early planting and early maturity can reduce yield loss to stemborers (Warui and Kuria, 1983; Gebre-Amlak et al., 1989; Kumar and Saxena, 1994). However, farmers in Kenya decide when to plant on the basis of criteria other than pest management, e.g. if they own or borrow machinery or sow by hand, if the soil cannot be ploughed until the rain has softened it, the reliability and timing of rain, or if another crop is planned for that year, requiring a shorter growing season for the first crop (Hassan et al., 1998b; Songa et al., 2002a). In the Dry Mid-altitude and Dry Transitional zones, most farmers plant before the rains, but on the coast, over three-quarters of farmers plant after the onset of the rains (Bonhof et al., 2001). Stemborer infestations on maize intercropped with non-cereals can be lower than on monocropped maize (Minja, 1990). Intercropping with cassava or leguminous crops such as leucaena has been shown to reduce stemborer densities significantly compared to monoculture (Van den Berg et al., 1998; Ogol et al., 1999). In contrast, the intercropping of two host plants, such as maize and sorghum, which is a common practice of subsistence farmers, can increase the intensity of stemborer attack in maize, as sorghum has a higher attraction for ovipositioning females than does maize (Ogwaro, 1983). Maize may only remain suitable for infestation during a limited period, but sorghum produces tillers and thus always has young plant material available for infestation (Overholt et al., 1994b).

CULTURAL CONTROL STRATEGIES.

PUSH–PULL SYSTEM. This is an integrated crop management system developed by ICIPE, which addresses not only stemborer control, but also soil fertility and Striga suppression. It is based on the use of natural attractant (pull) plants adjacent to the maize plots, and repellent (push) plants intercropped with the maize (Khan et al., 1997, 2000). Molasses grass (Melinis minutiflora), the silver leaf Desmodium (Desmodium uncinatum) and greenleaf Desmodium (Desmodium intortum) are among the most effective repellents for maize stemborers. If intercropped with maize, either of these crops produces substances that repel gravid female stemborers. Napier grass (P. purpureum) and Sudan grass (S. vulgare sudanense) are the most effective pull crops. These produce a substance that attracts stemborers, increasing the amount of oviposition in the grasses and therefore

Environ. Risk. Chap 02

46

23/8/04

9:32

Page 46

L. Muhammad and E. Underwood

reducing it in the maize crop. The grasses must be harvested at the right time in order to prevent the stemborers from reaching maturity and reinfecting the maize. The Desmodium species also contribute to soil nitrogen content through nitrogen fixation, and if incorporated into the soil after harvest, will benefit soil fertility longterm. Importantly for farmers in the Moist Mid-altitude zone, Desmodium root exudates suppress Striga (Khan et al., 2002). Molasses grass and D. uncinatum grow best in areas with rainfall above 700 mm/year (Skerman et al., 1988; Skerman and Riveros, 1990). ICIPE scientists are evaluating other Desmodium species (D. intortum and D. sandwichese) for the dry zones (F. Muyekho, Mbita, January 2004, personal communication). HOST PLANT RESISTANCE BY CONVENTIONAL PLANT BREEDING. Kenyan cultivars differ in their susceptibility to yield loss from stemborers, but most of the existing commercial maize cultivars are highly susceptible (Kumar, 1988; Seshu Reddy and Sum, 1992; Gethi et al., 2001). It is likely that due to insecticide use on breeding trials, stemborer resistance was lost or overlooked (Omolo, 1983). However, new cultivars are being developed. ICIPE scientists bred an early-tomedium-maturing white maize variety with high resistance via antibiosis (Mutinda and Ajala, 1998). The IRMA project and KARI are developing maize varieties with conventional host plant resistance to stemborers (Mugo et al., 2001b). They have identified resistant germplasm from African and Mexican lines, and at the end of 2003 have the first improved varieties in early field trials (Ininda et al., 2002; IRMA, 2002c, 2003). TRANSGENIC HOST PLANT RESISTANCE. Transgenic plants expressing Bacillus thuringensis endotoxins are being developed in the IRMA project (Box 2.1) and other initiatives (Andow et al., Chapter 4, this volume). All these projects are evaluating cry genes to control the four main stemborer species. In May 2003, the IRMA project began its phase II (IRMA, 2003). CIMMYT has produced ‘secondgeneration’ events of some promising cry genes (cry1Ab and cry1Ba) that are available in subtropical elite maize lines adapted to Mexican environments (Mugo et al., 2001b). They will be backcrossed into Kenyan varieties that have been screened for resistance to biotic and abiotic stresses. IRMA plans to have Bt maize varieties available to farmers as early as 2007. It is possible also that commercial biotechnology companies might apply to introduce Bt maize varieties to Kenya in the future. A collaborative project between ICIPE and South Africa, funded by USAID, is doing biosafety research on Bt maize containing Mon-810, with a view to introduction in Kenya (E. Osir, Nairobi, November 2003, personal communication). A Bt white maize variety for human consumption was grown on 84,000 ha in South Africa in 2003 (CropBiotech Net, 2004).

Summary Maize is the most important food in Kenya, and over 3.5 million rural households are dependent on it for food and income. Despite the fact that virtually all rural households grow maize, over 60% are net maize buyers. Over

Environ. Risk. Chap 02

23/8/04

9:32

Page 47

The Maize Agricultural Context in Kenya

Box 2.1. The Insect Resistant Maize for Africa project (Mugo et al., 2002). The Insect Resistant Maize for Africa (IRMA) project was established in 1999 between KARI, the International Maize and Wheat Improvement Centre (CIMMYT) and the Syngenta Foundation for Sustainable Agriculture, who are providing funding. The project aims to provide improved food security to African farmers via improved insect resistant maize varieties, both by conventional host plant resistance and Bt transgenes. The project has established four guiding principles. IRMA should: (i) be a model of good practice (including biosafety aspects), from which other countries can learn; (ii) serve as a pilot project for public–private partnership and cooperation; (iii) employ state of the art technology and methodology; and (iv) be transparent and open with ongoing stakeholder dialogue. The project aims to have: 1. Maize inbred lines, hybrids and OPVs that combine conventional and biotechnologybased insect resistance, tested and available in Kenya (and in the future in other African countries). 2. Protocols developed and KARI and other Kenyan scientists and staff trained in all aspects of the project, including: (i) the development and evaluation of insect resistant maize cultivars at the experimental station level; (ii) the deployment and monitoring of insect resistant varieties in farmers’ fields, impact assessments and public awareness. 3. KARI laboratory, greenhouse and field facilities improved for handling the project materials. 4. Insect resistance management strategies developed and implemented in all zones of Kenya where the insect-resistant maize will be grown. 5. Documented impacts of Bt gene-based resistance in maize on non-target organisms. Research activities: 1. Plant screening: genotypes from CIMMYT and KARI, and multiple borer resistant lines from Africa and Mexico are being screened in the field in Kenya for resistance to stemborers, leaf toughness, low-nitrogen tolerance, Turcicum blight resistance, resistance to MSV, and resistance of the kernels to weevils (IRMA, 2002b). Farmers’ varieties are also being collected and characterized (IRMA, 2002b). 2. Non-target impacts: identification, quantification and characterization of arthropods in four of the five maize growing zones in Kenya for use as baseline data during monitoring and evaluation phases of the IRMA project. 3. Gene flow: experimental plots have been planted on experimental stations to look at contamination rates using yellow maize in white maize plantings. This research has been extended on-farm. Socio-economic surveys will assess the expected rate of incorporation into recycled seed. 4. Resistance management: surveys to establish the acreage under wild and alternative crop hosts of stemborers and screening of selected species for their suitability for use as refuges. The project also carries out public awareness and education: media training seminars, extension agent workshops, farmers sessions, numerous documents, press releases and media publications communicated to the general public.

47

Environ. Risk. Chap 02

48

23/8/04

9:32

Page 48

L. Muhammad and E. Underwood

three-quarters of farms are less than 2 ha, with half or more of the farm area under maize intercropped with beans, sorghum or other crops. The most important maize production areas are in the Moist Transitional and Highland Tropics zones, which produce 85% of Kenya’s maize, with 70% of the production area and average yields of nearly 3 t/ha. Large-scale farmers with over 8 ha produce a quarter of Kenya’s maize in these zones. Many smallholders in these zones are using hybrid seed and fertilizer. The other zones have average yields of around 1 t/ha, with rainfall variability leading to frequent crop failures and low soil fertility as important constraints, and farmers use very little fertilizer or manure, seed recycled from previous crops, and grow two maize crops a year on the same land. Important environmental constraints to maize production prioritized by farmers are inadequate and unreliable rainfall, low soil fertility, and in the higher yielding zones stalk lodging (may be due to hail, hybrid variety, fungal disease or insect damage). The key maize pests for most farmers in Kenya are weeds, storage pests (rodents, weevils and storage moths), root feeders (chafer grubs, e.g. Melolonthinae, termites and cutworms), stemborers, and birds and mammals as field pests (squirrels, rodents, etc.). Diseases such as MSV and various fungal pathogens can also be devastating, and some may have an impact on human health through their production of carcinogens. Some of these pest problems are increasing in Kenya. In the areas where it occurs, the parasitic weed Striga is the second most important constraint, and there is evidence that it is spreading eastwards. The stemborer C. partellus was introduced into Kenya and has been expanding its range in Kenya over the last three decades so that it now occurs in all areas of Kenya except above 2400 m where little maize is grown. Stemborer damage may also be increasing in most areas as maize double-cropping increases in response to pressures on land and resources, allowing two or more generations of stemborers in 1 year (Zhou et al., 2001b). Stemborer infestations may also increase if fertilizer use increases (Sétamou et al., 1993). Most popular cultivars (landraces, OPVs or hybrids) sown by smallholder farmers have low genetic resistance to many of these pests and pathogens, although KARI and CIMMYT are now developing new cultivars with good resistance levels. Landraces favoured by many farmers have not been examined so far for their pest tolerance; the IRMA project has started to do this (IRMA, 2002b). Smallholder farmers also do not have or know about effective and practical control methods for many of these problems. For instance, once Striga infestation has built up, it is very difficult for smallholder farmers to control with their current techniques, as most of the yield loss causing damage occurs before the plants appear, and only two flowers can restock the seed bank for years. A promising new technique of integrated crop management, the ‘push–pull’ system, may help if farmers are trained to use it and can obtain the seed material. The target pests of Bt maize are the five principal stemborer species on maize in Kenya. All farmers in Kenya ranked stemborers as important pests (Tables 2.6–2.8). In general, the most damaging species of the maize stemborer complex are B. fusca and C. partellus (Table 2.9). B. fusca makes up most of the stemborer complex in the Highland Tropics and Moist Transitional zones,

Environ. Risk. Chap 02

23/8/04

9:32

Page 49

The Maize Agricultural Context in Kenya

49

and is responsible for most of the stemborer-induced crop loss in Kenya. C. partellus is the dominant species in the Lowland Tropics, Dry Mid-altitude and Dry Transitional zones. However, stemborer densities appear to be very dynamic, in both quantities and relative densities, particularly in the midaltitude and transitional zones. Farmers most commonly use synthetic pesticides to control stemborers. Most are aware of the use of pesticides and use them on other crops, but use on maize is generally low and often restricted to the parts of the field that show visible damage. Over a third of farmers in the Lowland Tropics and Dry Midaltitude zones are using pesticides against stemborers. Use of soil, ash or other locally made pest control agents, is also common, e.g. over half the farmers in the Dry Mid-altitude and Highland Tropics zones do this. Farmers lack knowledge of cultural methods and the pest biology behind them. On the coast, almost all surveyed farmers knew about synthetic pesticides, but less than 10% knew about any other methods and less than 5% of these actually used cultural control methods (Bonhof et al., 2001). Even if farmers are aware of them, their strategies are often based on conflicting goals, e.g. with regard to crop residue management and planting time. However, some methods could be successful for large-scale farmers. Finally, many smallholder farmers do not use any methods to control stemborers. A fifth of farmers surveyed in the Dry Mid-altitude zone and half of the farmers in the Lowland Tropics were not using any method (Bonhof et al., 2001; Songa et al., 2002a). At the same time, farmers generally estimated that their cultivars have low resistance to stemborers, although their landraces have not yet been investigated to test this (Hassan et al., 1998b). Kenyan maize farmers therefore need better knowledge and new strategies to manage stemborers. This could include crop management techniques that are compatible with smallholder practices, such as the ‘push–pull’ system, and new resistant germplasm such as Bt maize. This chapter has provided an example of the kind of background information needed to contextualize an environmental risk assessment of Bt maize in Kenya. However, important information may have been overlooked, and more recent and accurate unpublished information would be available to risk assessors. For example, research on stemborer ecology has highlighted the variation and dynamic changes in distribution, abundance and pest status of the species complex. An assessment of the impact of any new agricultural technology on the stemborer problem for Kenyan farmers must consider the possible consequences of the variation and change in this ecological complex so as not to further constrain Kenyan farmers, in particular resource-poor farmers.

References Ajala, S.O. and Saxena, K.N. (1994) Interrelationship among Chilo partellus (Swinhoe) damage parameters and their contribution to grain yield reduction in maize (Zea mays L.). Applied Entomology and Zoology 29, 469–476.

Environ. Risk. Chap 02

50

23/8/04

9:32

Page 50

L. Muhammad and E. Underwood

Ako, M., Schultess, F., Gumedzoe, M.Y.D. and Cardwell, K. (2002) The effect of Fusarium verticillioides on oviposition behaviour and bionomics of lepidopteran and coleopteran pests attacking the stem and cobs of maize in West Africa. Entomologia Experimentalis et Applicata 106, 201–210. Argwings-Kodek, G., Jayne, T.S., Nyambane, G., Awuor, T. and Yamano, T. (1998) How can micro-level household information make a difference for agricultural policy making? Kenya Agricultural Marketing and Policy Analysis Project. Tegemeo Institute of Agricultural Policy and Development/Egerton University; Kenya Agricultural Research Institute; Michigan State University. Bankole, S.A. and Adebanjo, A. (2003) Mycotoxins in food in West Africa: current situation and possibilities of controlling it. African Journal of Biotechnology 2, 254–263. Bergvinson, D., Mulaa, M. and Mugo, S. (2001) Develop an insect resistance management strategy for Kenya. In: Insect Resistant Maize for Africa Project Annual Report 2000. KARI/CIMMYT IRMA Project, IRMA Project Document No. 4, Nairobi. Bonhof, M.J., Van Huis, A., Kiros, F.G. and Dibogo, N. (2001) Farmers’ perceptions of importance, control methods and natural enemies of maize stemborers at the Kenya coast. Insect Science and its Application 21, 33–42. Bosque-Pérez, N.A. (2000) Eight decades of maize streak virus research. Virus Research 71, 107–121. Cardwell, K.F., Schultess, F., Ndemah, R. and Ngoko, Z. (1997) A systems approach to assess crop health and maize yield losses due to pests and diseases in Cameroon. Agriculture, Ecosystems and Environment 65, 33–47. CBS (Central Bureau of Statistics) (1990) Crops Focus Survey. Ministry of Planning and National Development, Nairobi. Corbett, J.D. (1998) Classifying maize production zones in Kenya through multivariate cluster analysis. In: Hassan, R.M. (ed.) Maize Technology Development and Transfer: a GIS Application for Research Planning in Kenya. CAB International, Wallingford, UK, pp. 15–25. CropBiotech Net (2004) Global status of commercialized transgenic crops: 2003. CropBiotech Update Special Edition, 14 January, 2004, International Service for the Acquisition of Agri-biotech Applications, Ithaca, New York, www.isaaa.org/kc (accessed January 2004). De Groote, H. (2002) Maize yield losses from stemborers in Kenya. Insect Science and its Application 22, 89–96. De Groote, H., Bett, C., Mose, L., Mugo, S., Odendo, M., Okuro, J.O. and Wekesa, E. (2001) Assessing the demand for insect resistant maize varieties through Participatory Rural Appraisals (PRA). In: Insect Resistant Maize for Africa Project Annual Report 2000. KARI/CIMMYT IRMA Project, IRMA Project Document No. 4, Nairobi. Diallo, A.O. and Friesen, D. (2001) Developing and disseminating stress tolerant maize for sustainable food security in west, central and East Africa. In: CIMMYT-Kenya annual report. CIMMYT, Nairobi, pp. 8–13. Emana, G., Overholt, W.A. and Kairu, E. (2001) Distribution and species composition of stemborers and their natural enemies in maize and sorghum in Ethiopia. Insect Science and its Application 21, 353–359. FAOSTAT (2001) Food balance sheet for maize in Kenya for 2001. http://apps.fao.org/page/collections?subset=agriculture (accessed November 2003). FAOSTAT (2002) Agriculture Data for Kenya 1995 to 2002. http://apps.fao.org/page/ collections?subset=agriculture (accessed November 2003).

Environ. Risk. Chap 02

23/8/04

9:32

Page 51

The Maize Agricultural Context in Kenya

51

FAO/CIMMYT (1997) White Maize: a Traditional Food Grain in Developing Countries. CIMMYT/FAO, Rome. Ferdu, A., Demissew, K. and Birhane, A. (2001) Major insect pests of maize and their management: a review. In: Second National Workshop of Ethiopia, 12–16 November, 2001, pp. 89–96. www.cimmyt.org/worldwide/CIMMYT_ Regions/CIMMYT_Africa/procee_Ethiopia/14Insect.pdf (accessed November 2003). Gebre-Amlak, A., Sigvald, R. and Petersson, J. (1989) The relationship between sowing date, infestation and damage by the maize stalk borer, Busseola fusca (Noctuidae), on maize in Awassa, Ethiopia. Tropical Pest Management 35, 143–145. Gethi, M., Mugo, S., Odongo, O., Ombakho, G., Njoroge, K., Muasya, W. and Chivatsi, W. (2001) Evaluating commercial and elite experimental varieties for resistance to stemborers. In: Insect Resistant Maize for Africa Project Annual Report 2000. KARI/CIMMYT IRMA Project, IRMA Project Document No. 4, Nairobi. Gitau, C.W. (2002) Development of pupal parasitoid Xanthopimpla stemmator (Thunberg) (Hymenoptera: Ichneumonidae) in various cereal stem borers (Lepidoptera). MSc thesis, Kenyatta University, Nairobi. Grisley, W. (1997) Crop-pest yield loss: a diagnostic study in the Kenya highlands. International Journal of Pest Management 43, 137–142. Hassan, R.M. (ed.) (1998) Maize Technology Development and Transfer: a GIS Application for Research Planning in Kenya. CAB International, Wallingford, UK. Hassan, R.M. and Ransom, J.K. (1998) Determinants of the incidence and severity of Striga infestation in maize in Kenya. In: Hassan, R.M. (ed.) Maize Technology Development and Transfer: a GIS Application for Research Planning in Kenya. CAB International, Wallingford, UK, pp. 163–174 Hassan, R.M. and Salasya, P. (1994) The gender factor in maize production and technology transfer in Kenya. Proceedings of the Fourth KARI Scientific Conference. Kenya Agricultural Research Institute, Nairobi. Hassan, R.M., Corbett, J.D. and Njoroge, K. (1998a) Combining geo-referenced survey data with agroclimatic attributes to characterize maize production systems in Kenya. In: Hassan, R.M. (ed.) Maize Technology Development and Transfer: a GIS Application for Research Planning in Kenya. CAB International, Wallingford, UK, pp. 43–68. Hassan, R.M., Onyango, R. and Rutto, J.K. (1998b) Relevance of maize research in Kenya to maize production problems perceived by farmers. In: Hassan, R.M. (ed.) Maize Technology Development and Transfer: a GIS Application for Research Planning in Kenya. CAB International, Wallingford, UK, pp. 71–88. Hassan, R.M., Njoroge, K., Njore, M., Otsyula, R. and Laboso, A. (1998c) Adoption patterns and performance of improved maize in Kenya. In: Hassan, R.M. (ed.) Maize Technology Development and Transfer: a GIS Application for Research Planning in Kenya. CAB International, Wallingford, UK, pp. 107–136. Hassan, R.M., Murithi, F. and Kamau, G. (1998d) Determinants of fertilizer use and the gap between farmers’ maize yields and potential yields in Kenya. In: Hassan, R.M. (ed.) Maize Technology Development and Transfer: a GIS Application for Research Planning in Kenya. CAB International, Wallingford, UK, pp. 137–161. Hell, K., Cardwell, K.F., Sétamou, M. and Schultess, F. (2000) Influence of insect infestation on aflatoxin contamination of stored maize in four agro-ecological regions in Benin. African Entomology 8, 169–177. Ininda, J., Gethi, M., Odongo, O., Njuguna, J. and Ochieng, J.A.W. (2002) Research abstract: Coordinated ecosystem breeding in Kenya: progress, challenges and opportunities. In: Biotechnology, Breeding and Seed Systems for African Crops: Research and Product Development that Reaches Farmers. Program, Participants

Environ. Risk. Chap 02

52

23/8/04

9:32

Page 52

L. Muhammad and E. Underwood

and Abstracts. The Rockefeller Foundation, Nairobi. www.africancrops.net/ news%20kenya.htm (accessed October 2003). IRMA (2001) IRMA Updates, December 2001, Vol. 2, Issue 4. www.cimmyt.org/ ABC/InvestIn-InsectResist/htm/InvestIn-InsectResist.htm (accessed February 2003). IRMA (2002a) IRMA Updates, August 2002, Vol. 3, Issue 2. www.cimmyt.org/ ABC/InvestIn-InsectResist/htm/InvestIn-InsectResist.htm (accessed February 2003). IRMA (2002b) IRMA Updates, September 2002, Vol. 3, Issue 3. www.cimmyt.org/ABC/ InvestIn-InsectResist/htm/InvestIn-InsectResist.htm (accessed February 2003). IRMA (2002c) IRMA Updates, December 2002, Vol. 3, Issue 4. www.cimmyt.org/ABC/ InvestIn-InsectResist/htm/InvestIn-InsectResist.htm (accessed February 2003). IRMA (2003) IRMA Updates, June 2003, Vol. 4, Issues 1+2. www.cimmyt.org/ABC/ InvestIn-InsectResist/htm/InvestIn-InsectResist.htm (accessed December 2003). Karanja, D.D., Jayne, T.S. and Strasberg, P. (1998) Maize productivity and impact of market liberalization in Kenya. Kenya Agricultural Marketing and Policy Analysis Project. Tegemeo Institute of Agricultural Policy and Development/Egerton University; Kenya Agricultural Research Institute; Michigan State University. Kauma, T.M. (2000) Yield losses caused by stemborers on maize and establishment of Cotesia flavipes Cameron (Hymenoptera: Braconidae) in eastern Uganda. Unpublished MSc thesis, Makerere University. KEPHIS (1998) Annual Report 1998. Kenya Plant Health Inspectorate Service (KEPHIS), Nairobi. Kfir, R. (1991) Duration of diapause in the stemborers, Busseola fusca and Chilo partellus. Entomologia Experimentalis et Applicata 61, 265–270. Kfir, R. (1997) Competitive displacement of Busseola fusca (Lepidoptera: Noctuidae) by Chilo partellus (Lepidoptera: Pyralidae). Annual Entomological Society of America 90, 619–624. Kfir, R., Van Hamburg, H. and Van Vuuren, R. (1989) Effect of stubble treatment on the post-diapause emergence of the grain sorghum stalk borer, Chilo partellus (Swinhoe) (Lepidoptera: Pyralidae). Crop Protection 8, 289–292. Kfir, R., Overholt, W.A., Khan, Z.R. and Polaszek, A. (2002) Biology and management of economically important lepidopteran cereal stem borers in Africa. Annual Review of Entomology 47, 701–731. Khan, Z.R., Chiliswa, P., Ampong-Nyarko, K., Smart, L.E., Polaszek, A., Wandera, J. and Mulaa, M.A. (1997) Utilization of wild gramineous plants for management of cereal stem borers in Africa. Insect Science and its Application 17, 143–150. Khan, Z.R., Pickett, J.A., van den Berg, J., Wadhams, L.J. and Woodcock, C.M. (2000) Exploiting chemical ecology and species diversity: stem borer and Striga control for maize and sorghum in Africa. Pest Management Science 56, 957–962. Khan, Z.R., Hassanali, A., Overholt, W.A., Khamis, T.M., Hooper, A.M., Pickett, J.A., Wadhams, L.J. and Woodcock, C.M. (2002) Control of witchweed Striga hermonthica by intercropping with Desmodium spp., and the mechanism defined as allelopathic. Journal of Chemical Ecology 28, 1871–1885. Kumar, H. (1988) Effect of stalk damage on growth and yield of certain maize cultivars by the maize stalkborer, Chilo portellus. Insect Science and its Application 9, 303–307. Kumar, H. and Saxena, K.N. (1994) Infestation and damage on three maize cultivars by the stalk-borer Chilo partellus (Swinhoe) in relation to their yield in western Kenya. Insect Science and its Application 15, 331–335. Mathez, F.C. (1972) Chilo partellus Swinh., C. orichalcociliellus Strand (Lep., Crambidae) and Sesamia calamistis Hmps. (Lep., Noctuidae) on maize in the Coast Province of Kenya. Mitteilungen der Schweizerischen Entomologischen Gesellschaft 45, 267–289.

Environ. Risk. Chap 02

23/8/04

9:32

Page 53

The Maize Agricultural Context in Kenya

53

Mbapila, J.C. (1997) Comparative seasonal adaptation of Cotesia flavipes and Cotesia sesamiae (Hymenoptera: Braconidae) to Chilo partellus on the Kenya Coast. PhD dissertation, University of Dar-es-Salaam, Dar-es-Salaam. Mbapila, J.C. and Overholt, W.A. (1997) Oviposition, development and searching behaviour of Cotesia flavipes Cameron and Cotesia sesamiae (Cameron) (Hymenoptera: Braconidae) in relation to aestivating and non-aestivating Chilo spp. (Lepidoptera: Pyralidae) larvae in maize. African Entomology 5, 35–39. Mbapila, J.C., Overholt, W.A. and Kayumbo, H.Y. (2002) Comparative development and population growth of an exotic stemborer, Chilo partellus (Swinhoe), and an ecologically similar congener, C. orichalcociliellus (Strand) (Lepidoptera: Crambidae). Insect Science and its Application 22, 21–27. Mesfin, T., Bosque-Perez, N.A., Buddenhagen, I.W., Thottappilly, G., Olojede, S.O., Den Hollander, J. and Markham, P.G. (1992) Studies of maize streak virus isolates from grass and cereal hosts in Nigeria. Plant Disease 76, 789–795. Mills, B., Hassan, R.M. and Mwangi, P. (1998) Estimating potential benefits from research and setting research priorities for maize in Kenya. In: Hassan, R.M. (ed.) Maize Technology Development and Transfer: a GIS Application for Research Planning in Kenya. CAB International, Wallingford, UK, pp. 89–104. Minja, E.M. (1990) Management of Chilo spp. infesting cereals in Eastern Africa. Insect Science and its Application 11, 489–499. Mugo, S.N., Likhayo, P., Odongo, O. and Bergvinson, D. (2001a) Weevil resistance, a value-added trait: IRMA project germplasm development activities. In: Insect Resistant Maize for Africa Project Annual Report 2000. KARI/CIMMYT IRMA Project, IRMA Project Document No. 4, Nairobi. Mugo, S.N., Bergvinson, D. and Hoisington, D. (2001b) Options in developing stemborer-resistant maize: CIMMYT’s approaches and experiences. Insect Science and its Application 21, 409–415. Mugo, S.N., Songa, J.M., De Groote, H. and Hoisington, D. (2002) Insect Resistant Maize for Africa (IRMA) Project: An overview. Paper presented to the Syngenta Symposium, Washington, DC, 25 June, 2002, the Syngenta staff in Greensboro, North Carolina, 26 June, 2002, and to the Syngenta staff in Basel, Switzerland, 27 June, 2002. Mulaa, M.A. (1995) Evaluation of factors leading to rational pesticide use for the control of the maize stalk borer Busseola fusca in Trans Nzoia district, Kenya. PhD thesis, University of Wales, Cardiff. Munkvold, G.P., McGee, D.C. and Carlton, W.M. (1997) Importance of different pathways for maize kernel infection by Fusarium monoliforme. Phytopathology 87, 209–217. Mutinda, C.J.M. and Ajala, S.O. (1998) Advances in breeding for resistance to Chilo partellus Swinhoe in maize in Kenya. The First ARPPIS Symposium. ICIPE, Nairobi. Mwangi, S.F.M. (1998) Status of Northern Leaf Blight, Phaeosphaeria maydis Leaf Spot, Southern leaf blight, Rust, Maize Streak Virus and physiologic specialization of Exserohilum turcicum in Kenya. Dissertation submitted to the faculty of the Virginia Polytechnic Institute and State University. Blacksburg, Virginia. http://scholar.lib.vt.edu/theses/available/etd-020199–125602/ (accessed October 2003). Ngi-song, A.J., Kimani-Njogu, S. and Overholt, W.A. (2001) Multiple parasitism by Cotesia sesamiae and Cotesia flavipes on Busseola fusca. Biocontrol Science and Technology 11, 381–390.

Environ. Risk. Chap 02

54

23/8/04

9:32

Page 54

L. Muhammad and E. Underwood

Njunguna, J.G.M. (1996) Epidemiology of maize streak disease in Kenya. Unpublished PhD thesis, Ohio State University. Nye, I.W.R. (1960) The insect pests of graminaceous crops in East Africa. Colonial Research Studies 31, 1–48. Nyoro, J.K., Wanzala, M. and Awuor, T. (2001) Increasing Kenya’s competitiveness: farm level issues. Working paper No. 4. Tegemeo Agricultural Monitoring and Policy Analysis Project (TAMPA). Tegemeo Institute, Nairobi. Odhiambo, G.D. (1998) Studies on witchweed [Striga hermonthica (Del.) Benth] seed longevity and maize (Zea mays L.) productivity under different management systems in Kenya. PhD Thesis, University of Reading, p. 198. Odvody, G.N. and Chilcutt, C.F. (2002) Aflatoxin and insect response in South Texas of near-isogenic corn hybrids with Cry1Ab and Cry2Ab events. Texas A&M University Agricultural Research and Extension Center, Corpus Christi, Texas. Ofomata, V.C. (1997) Ecological interactions between Chilo orichalcociliellus and Chilo partellus on the Kenya coast. PhD thesis. Nnamdi Azikiwe University, Anambra State, Nigeria. Ofomata, V.C., Overholt, W.A., Van Huis, A., Egwuatu, R.I. and Ngi-Song, A.J. (1999) Niche overlap and interspecific association between Chilo partellus and Chilo orichalcociliellus on the Kenya coast. Entomologia Experimentalis et Applicata 93, 141–148. Ofomata, V.C., Overholt, W.A., Lux, S.A., Van Huis, A. and Egwuatu, R.I. (2000) Comparative studies on the fecundity, egg survival, larval feeding, and development of Chilo partellus and Chilo orichalcociliellus (Lepidoptera: Crambidae) on five grasses. Annals of the Entomological Society of America 93, 492–499. Ogol, C.K.P.O., Spence, J.R. and Keddie, A. (1999) Maize stemborer colonization, establishment and crop damage levels in a maize-leucaena agroforestry system in Kenya. Agriculture, Ecosystems and Environment 76, 1–15. Ogwaro, K. (1983) Intensity levels of stemborers in maize and sorghum and the effect on yield under different inter-cropping patterns. Insect Science and its Application 4, 33–37. Omolo, E.O. (1983) Screening of local and exotic maize lines for stemborer resistance with special reference to Chilo partellus. Insect Science and its Application 4, 105–108. Overholt, W.A., Ngi-Song, A.J., Kimani, S.W., Mbapila, J., Lammers, P.M. and Kioko, E. (1994a) Ecological considerations of the introduction of Cotesia flavipes Cameron (Hymenoptera: Braconidae) for biological control of Chilo partellus (Swinhoe) (Lepidoptera: Pyralidae) in Africa. Biocontrol News Information 15, 19N–24N. Overholt, W.A., Ochieng, J.O., Lammers, P. and Ogedah, K. (1994b) Distribution and sampling of Chilo partellus (Swinhoe) (Lepidoptera: Pyralidae) in maize and sorghum at the Kenyan coast. Bulletin of Entomological Research 84, 367–378. Overholt, W.A., Ngi-Song, A.J., Omwega, C.O., Kimani-Njogu, S.W., Mbapila, J., Sallam, M.N. and Ofomata, V. (1997) A review of the introduction and establishment of Cotesia flavipes Cameron (Hymenoptera: Braconidae) in East Africa for biological control of cereal stemborers. Insect Science and its Application 17, 79–88. Pingali, P.L. (ed.) (2001) CIMMYT 1999–2000 World Maize Facts and Trends. Meeting World Maize Needs: Technological Opportunities and Priorities for the Public Sector. CIMMYT, Mexico, D.F. Polaszek, A. and Khan, Z.R. (1998) Host plants. In: Polaszek, A. (ed.) African Cereal Stem Borers: Economic Importance, Taxonomy, Natural Enemies and Control. CAB International, Wallingford, UK, pp. 3–10.

Environ. Risk. Chap 02

23/8/04

9:32

Page 55

The Maize Agricultural Context in Kenya

55

Ransom, J.K. (2000) Long-term approaches for the control of Striga in cereals: field management options. Crop Protection 19, 759–763. Ransom, J.K. and Odhiambo, G.D. (1994) Long-term effects of fertility and hand weeding on Striga in maize. In: Pieterse, A.H., Verkleij, J.A.C. and ter Borg, J.S. (eds) Biology and Management of Orobanche. Proceedings of the Third International Conference on Orobanche and Related Striga Research. Royal Tropical Institute, Amsterdam, pp. 513–519. Ransom, J.K. and Odhiambo, G.D. (1995) Effect of corn (Zea mays) genotypes which vary in maturity length on Striga hermonthica parasitism. Weed Technology 9, 63–67. Republic of Kenya (1995) Agriculture Sector Review. Ministry of Agriculture, Nairobi. Republic of Kenya (2000) Statistical Abstract 1999. Central Bureau of Statistics, Government Printer, Nairobi. Republic of Kenya (2001) Agriculture Sector Review. Ministry of Agriculture, Nairobi. Rybicki, E.P., Dennis, S., Napier, G. and Hughes, F.L. (1998) Novel grass and wheat strains of Maize Streak Virus detected by DNA amplification and sequencing. Molecular Plant Pathology On-line, www.bspp.org.uk/mppol/1998/0929rybicki (accessed November 2003). Scheltes, P. (1978) The condition of the host plant during aestivation-diapause of the stalk borers Chilo partellus and Chilo orichalcociliellus (Lepidoptera: Crambidae) in Kenya. Entomologia Experimentalis et Applicata 24, 679–688. Schultess, F., Cardwell, K.F. and Gounou, S. (2002) The effect of endophytic Fusarium verticillioides on infestation of two maize varieties by lepidopterous stemborers and coleopteran grain feeders. Phytopathology 92, 120–128. Seshu Reddy, K.V. and Sum, K.O.S. (1991) Determination of economic injury of the stemborer, Chilo partellus (Swinhoe) in maize, Zea mays L. Insect Science and its Application 12, 269–274. Seshu Reddy, K.V. and Sum, K.O.S. (1992) Yield-infestation relationship and determination of economic injury level of the stem-borer, Chilo partellus (Swinhoe) on three varieties of maize, Zea mays L. Maydica 37, 371–376. Seshu Reddy, K.V. and Walker, P.T. (1990) A review of yield losses in graminaceous crops caused by Chilo spp. Insect Science and its Application 11, 563–569. Sétamou, M., Schultess, F., Bosque-Pérez, N.A. and Thomas-Odjo, A. (1993) Effect of plant nitrogen and silica on the bionomics of Sesamia calamistis Hampson (Lepidoptera: Noctuidae). Bulletin of Entomological Research 83, 405–411. Sétamou, M., Cardwell, K.F., Schultess, F. and Hell, K. (1998) Effect of insect damage to maize ears, with special reference to Mussidia nigrivenella (Lepidoptera: Pyralidae), on Aspergillus flavus (Deuteromycetes: Monoliales) infection and aflatoxin production in maize before harvest in the Republic of Benin. Journal of Economic Entomology 91, 433–438. Skerman, P.J. and Riveros, F. (1990) Tropical grasses. FAO Plant Production and Protection Series No. 23. Skerman, P.J., Cameroon, D.G. and Riveros, F. (1988) Tropical forage legumes. FAO Plant Production and Protection Series No. 2. Songa, J.M. (1999) Distribution, importance and management of stemborers (Lepidoptera) in maize production systems of semi-arid Eastern Kenya with emphasis on biological control. PhD dissertation, Kenyatta University, Nairobi. Songa, J.M., Overholt, W.A., Mueke, J.M. and Okello, R.O. (1998) Distribution of stemborer species in semi-arid eastern Kenya. In: Maize Production Technology for the Future: Challenges and Opportunities. Proceedings of the Sixth Eastern and Southern Africa Regional Maize Conference, Addis Ababa, 21–25 September, 1998. CIMMYT and EARO, Addis Ababa, pp. 117–120.

Environ. Risk. Chap 02

56

23/8/04

9:32

Page 56

L. Muhammad and E. Underwood

Songa, J.M., Overholt, W.A., Mueke, J.M. and Okello, R.O. (2001) Colonization of Cotesia flavipes (Hymenoptera: Braconidae) in stem borers in the semi-arid eastern province of Kenya. Insect Science and its Application 21, 289–295. Songa, J.M., Overholt, W.A., Mueke, J.M. and Okello, R.O. (2002a) Farmers perceptions of aspects of maize production systems and pests in semi-arid eastern Kenya: factors influencing occurrence and control of stem borers. International Journal of Pest Management 48, 1–11. Songa, J.M., Overholt, W.A., Okello, R.O. and Mueke, J.M. (2002b) Control of lepidopteran stem borers in maize by indigenous parasitoids in semi-arid areas of eastern Kenya. Biological Agriculture and Horticulture 20, 77–90. Songa, J.M., Mugo, S., Mulaa, M., Taracha, C., Bergvinson, D., Hoisington, D. and De Groote, H. (2002c) Towards development of environmentally safe, insect resistant maize varieties for food security in Kenya. Presented to the Syngenta Symposium, Washington, DC, 25 June, 2002, the Syngenta staff in Greensboro. Songa, J.M., Guofa, Z. and Overholt, W.A. (2002d) Relationships of stemborer damage and plant physical conditions to maize yield in a semi-arid zone of Eastern Kenya. Insect Science and its Application 21, 243–249. Songa, J.M., Overholt, W.A., Mueke, J.M. and Okello, R.O. (2002e) Regional distribution of lepidopteran stem borers and their parasitoids among wild grasses in the semi-arid areas of eastern Kenya. African Crop Science Journal 10, 183–194. Tiffen, M., Mortimore, M. and Gichuki, F. (1994) More People, Less Erosion: Environmental Recovery in Kenya. Wiley, Chichester, UK. Unnithan, G.C. (1987) Development and reproductive biology of the maize stem borer Busseola fusca (Lepidoptera: Noctuidae). Journal of Applied Entomology 104, 172–179. Van den Berg, J., Nur, A.F. and Polaszek, A. (1998) Cultural control. In: Polaszek, A. (ed.) Cereal Stemborers in Africa: Economic Importance, Taxonomy, Natural Enemies and Control. International Institute of Entomology. CAB International, Wallingford, UK, pp. 333–347. Wanzala, M., Jayne, T.S., Staatz, J.M., Mugera, A., Kirimi, J. and Owuor, J. (2001) Fertilizer markets and agricultural production incentives: insights from Kenya. Working Paper No.3. Tegemeo Agricultural Monitoring and Policy Analysis Project (TAMPA). Tegemeo Institute, Nairobi. Warui, C.M. and Kuria, J.N. (1983) Population incidence and control of maize stalk borers Chilo partellus (Swinh.) and C. orichalcociliellus Strand and Sesamia calamistis Hmps in Coast Province of Kenya. Insect Science and its Application 4, 11–18. World Bank (1995) Kenya Poverty Assessment. Population and Human Resources Division, Eastern Africa Department, Africa Region. Zhou, G., Baumgartner, J. and Overholt, W.A. (2001a) Impact assessment of an exotic parasitoid on stemborer (Lepidoptera) population dynamics in Kenya. Ecological Applications 11, 1554–1562. Zhou, G., Overholt, W.A. and Mochiah, M.B. (2001b) Changes in the distribution of Lepidopteran maize stemborers in Kenya from the 1950s to 1990s. Insect Science and its Application 21, 395–402. Zhou, G., Overholt, W.A. and Kimani-Njogu, S.W. (2003) Species richness and parasitism in an assemblage of parasitoids attacking maize stem borers in coastal Kenya. Ecological Entomology 28, 109–118.

Environ. Risk. Chap 03

23/8/04

3

9:32

Page 57

Problem Formulation and Options Assessment (PFOA) for Genetically Modified Organisms: the Kenya Case Study KRISTEN C. NELSON, G. KIBATA, L. MUHAMMAD, J.O. OKURO, F. MUYEKHO, M. ODINDO, A. ELY AND J.M. WAQUIL Corresponding author: Dr Kristen C. Nelson, Department of Forest Resource and Department of Fisheries, Wildlife, and Conservation Biology, University of Minnesota, 115 Green Hall, 1530 Cleveland Avenue North, St Paul, MN 55108, USA. E-mail: [email protected]

Introduction Throughout the world, countries are discussing the role genetically modified organisms (GMOs) will have in their future. Each country begins the discussion at a different starting point depending on distinct historical, economic, social and environmental factors. For some, GMOs are a new technology that should be used based on market principles – if it is a viable product, it will survive and contribute to economic growth. For others, it is a question of considering long-term risks and uncertainties before making short-term decisions – precaution is the guiding principle for these decision makers. Still others are caught in the conflict between these points of view as they make decisions regarding the introduction of GMOs to their countries. Decision makers and citizens have the right and responsibility to design their own policy and regulatory systems to address GMOs. One critical component of this discussion is the nature of environmental risk assessment. In this case, it is clear countries must create a responsive system to facilitate socially acceptable choices (Stern and Fineberg, 1996). At its core, the discussion focuses on the critical societal need that will be addressed by the GMO, i.e. what needs will be satisfied and at what risk? Societal risk requires societal reflection. Countries accept risk with most major policy decisions but they never do so lightly. A deliberative process with multi-stakeholder participation allows members of society to participate in the evaluation of critical needs and risks. A cross-section of society – farmers, consumer groups, industry, environmental © CAB International 2004. Environmental Risk Assessment of Genetically Modified Organisms: Vol. 1. A Case Study of Bt Maize in Kenya (eds A. Hilbeck and D.A. Andow)

57

Environ. Risk. Chap 03

23/8/04

9:32

Page 58

58

K.C. Nelson et al.

representatives, policy makers, etc. – must have a vehicle to express their concerns and evaluate the future alternatives for addressing basic needs. Finally, this deliberative process will be increasingly important for resource-scarce nations if public investment is involved, because a comparative reflection by a cross-section of society may be beneficial in the prioritizing and targeting of resources. Given the present uncertainties surrounding GMOs and the new discoveries about them that occur annually, the system to conduct a deliberative discussion should be flexible and able to respond to a society’s core values, concerns and needs. At the same time, the discussion is best served if it is driven by sound, scientifically guided assessment and review (CBD, 1992; Gibbons, 1999; National Research Council, 2002). A robust environmental risk assessment clearly delineates when scientific knowledge, information and analysis can effectively respond to key questions and when it cannot. In most natural resource arenas, practitioners and scholars are implementing and evaluating diverse options for societal discussion about critical issues (O’Brien, 2000; Wondolleck and Yaffee, 2000). Specific contributions within the biosafety arena have detailed potential approaches to (McLean et al., 2002; Glover et al., 2003) and evaluations of key attributes of multi-stakeholder dialogues (Irwin, 2001) at an international and national scale. Societal discussion is advocated in the setting of priorities and strategies for agricultural research and development, in the formulation of national biosafety frameworks, and in the environmental risk assessment of specific biotechnologies. A Problem Formulation and Options Assessment (PFOA) is applicable in all of these contexts, and can be employed in an iterative fashion so as to incorporate feedback from changing societal values and the scientific state of the art. In the Kenya Bt maize case study presented in this chapter, PFOA creates the context for societal dialogue concerning the potential use of a proposed technology based on a transgenic organism, i.e. Bt maize. It is a public deliberation about the transgenic organism that provides a rational, science-driven planning process by which multiple stakeholders can assess their needs, evaluate the risks related to multiple future options, and make recommendations to decision makers about policies to reduce societal risks and enhance the benefits provided by various options. Certainly the requirements set forth to accomplish a PFOA mean it is a complex process – but this should not be considered as a possible argument against its use. Most important societal decisions are complex, messy and controversial, and because they are complex, messy and controversial we should work to improve them with transparent, systematic and scientifically based discussion. By doing so, the decision-making process gains social legitimacy and society gains greater confidence in the decisions taken. A PFOA can be used to provide such a discussion, and may play a significant role in environmental risk assessment.

Relation of PFOA to Environmental Risk Assessment Practitioners and scholars have tested numerous techniques that serve as a methodological foundation for the PFOA in environmental risk assessment

Environ. Risk. Chap 03

23/8/04

9:32

Page 59

PFOA for Genetically Modified Organisms

59

(Grimble and Wellard, 1997; Kessler and Van Dorp, 1998; Schmoldt and Peterson, 1998; Biggs and Matsaert, 1999; Loevinsohn et al., 2002, to name a few). Two crucial steps in risk assessment are addressed by many of these techniques, and this PFOA Model is designed specifically to address them. The first critical step in risk assessment is problem identification (NRC, 1983, 1996). What is the scope of the problem and how is it defined? Problem identification frames the entire risk assessment. A second critical step is the identification of potential alternative solutions to the problem (NRC, 1983, 1996). The proposed action, in this case the use of Bt maize in Kenya, is never the only possible way to address the problem. Risk assessment depends entirely on an appropriate specification of alternatives (including taking no action and doing nothing), so that comparative risk can be assessed, and appropriate controls for risk assessment science can be defined and used (Andow and Hilbeck, Chapter 1, this volume). This PFOA Model is comprised of specific brainstorming, discussion and analytical components (Table 3.1). First, formulating the problem serves as the core foundation. The problem is defined as an unmet need that requires change (Goldstein, 1993). Basic human needs are most commonly identified as food, shelter and safety. Other human interests are stakeholder specific such as enhanced economic opportunity, positive social interactions, cultural richness, etc. For example, individuals have the basic need for a certain amount of calories per day or the security that their children will continue to live healthy lives as a minimum foundation for well being. Once the needs for food, shelter and safety are met, individuals can expand their interests to include numerous options for wellbeing. These interests will differ from one individual to another and from one group to another. After a problem is identified, the PFOA Model requires a comparative approach to risk assessment. The participants clarify the relative importance of this problem as compared to other problems or issues. Once the group agrees the problem is sufficiently important to merit an analysis, the range of future alternatives for solving the problem are compared in relation to their attributes, potential ability to address the problem, changes required to implement the option and potential adverse effects. The PFOA is planning for alternative futures, not for the current conditions against one option, but rather making a comparative assessment of alternative futures. After a complete analysis by a multi-stakeholder group, a recommendation is made to decision makers to continue research and development (in some cases risk assessment research) with the technology or to halt the development of the technology. A science-driven PFOA must be a deliberative process (Forester, 1999) designed to provide for social reflection and discussion about transgenic organisms. A sound deliberative process is transparent, equitable, legitimate and data driven when possible (Susskind et al., 2000). Transparency allows for the open communication of information between all parties and easily accessible reporting of decisions to the public (Hemmati, 2002). Providing an equitable PFOA process means that information from the broadest spectrum of society must be included with all stakeholders having the possibility to contribute. Civic society must perceive that there are sufficient avenues for input and consideration of diverse viewpoints and concerns. When

Environ. Risk. Chap 03

23/8/04

9:32

Page 60

60

K.C. Nelson et al.

Table 3.1. Problem formulation and options assessment process. Initiating proposal: A. Proposal to use GMO: Any PFOA for transgenic organisms will be initiated by the request or suggestion that a particular GMO would be a beneficial alternative to the way things are currently being done in a particular cropping system B. Decision by regulatory body: Is there merit to moving forward to evaluate the GMO as a possible option or is the initiating proposal premature? Yes/No PFOA process: questions to be answered by all representatives and shared in the deliberative process Step 1: Problem formulation: Formulation of problem: An unmet need that requires change

Basic human needs Food, shelter, safety

Interests A stakeholder group’s values, goals and perspectives

A. Whose problem is it? Whose problem should it be? 1. What needs of the people are not being met by the present situation? 2. What aspects of the present situation must be changed to meet the needs? Step 2: Prioritization and scale: A. Is this problem a core problem for the people identified? 1. Do the people recognize the problem as important to their lives? 2. What are the potentially competing needs of these people? 3. How do the identified needs rank in importance to these other competing needs? B. How extensive is the problem? 1. How many people are affected? 2. In what part of the country are these people located? 3. How large an area is affected by the problem? 4. How severe is the problem (local intensity)? Step 3: Problem statement: Shared understanding of the unmet need and its relative importance for a particular group of people Step 4: Recommendation by regulatory body: Do we move forward to identify options and conduct an options assessment? Option identification and assessment chart Step 5 Options Future alternatives

Step 6 Characteristics

Step 7 Changes

Step 8 Effect on the system

For problem solving

Required/anticipated

Internal External (Social, environmental, economic)

Option A Option B Option C etc. Continued

Environ. Risk. Chap 03

23/8/04

9:32

Page 61

PFOA for Genetically Modified Organisms

61

Table 3.1. Continued. Step 5: Option identification: Brainstorm possible future alternatives to solve the identified problem, transgenic organisms would be one option. This step can be completed by the multiple stakeholder group for the initial identification of options. The multi-stakeholder group can do steps 6–8 or a technical committee can develop a report that covers steps 6–8 and the multi-stakeholder group can use the document to begin their evaluation of options and modify the assessment. Step 6: Assessment of the options in relation to the problem: Assessing capability of potential solution to solve problem. 1. What are the characteristics of the ‘technology’ option? i.e. transgene, intercropping system, etc. 2. What is the range of crop production systems and what is the geographic region the option is likely to be used in or have an effect on? 3. What is the efficacy of the ‘technology’ on the target? 4. What are the costs of the technology within the crop production system? 5. What barriers to use exist, i.e. is the seed distribution system in place; can the potential solution be integrated into present production; can the farmers afford the potential solution? 6. How might the use of the option change cropping practices, such as tillage systems or pesticide use (including impacts on non-target pests)? What useful practices are reinforced by the potential? 7. What information is needed to show that the changes are likely to occur? Baseline data associated with the diversity of present practices should be used if it is available. 8. How will anticipated changes in agricultural practices affect the needs identified in steps 1 and 2? Step 7: Changes required and anticipated for a specific option: 1. What changes in farm management practices might contribute to the solution? 2. What changes in the local community might contribute to the solution? 3. What changes in government support for farmers might contribute to the solution? 4. What changes in the structure of agricultural production might contribute to the solution? 5. What other changes would likely be needed to facilitate widespread use of this option? Step 8: Adverse effects: Potential adverse consequences from this option. Potential beneficial effects can be considered ‘negative’ adverse effects. 1. How might the potential solution affect the structure of agriculture or agricultural infrastructure? 2. How might the potential solution reinforce poor agricultural practices or disrupt useful practices? 3. What are the potential adverse effects of these changes internally and externally to the production system? 4. How will its use affect other nearby crop production systems and non-agricultural environments (can its use be restricted to a particular cropping system or geographic region)? 5. Are any of these changes difficult to reverse, once they occur? Step 9: Recommendation: The multiple stakeholder group should present their problem formulation and option assessment to the appropriate decision-making body.

Environ. Risk. Chap 03

62

23/8/04

9:32

Page 62

K.C. Nelson et al.

transparency and equitable input are central to the process, the PFOA gains legitimacy in the public eye. This public legitimacy must be matched by traditional legitimacy or sanctioning by a formal political body that embeds the deliberative process. The deliberative process can be tied to a regulatory authority or legislative authority, but it must provide a means by which results from the PFOA inform government decision making and action. Finally, the foundation of PFOA is a science-driven inquiry. Questions are answered with data, impacts are assessed with valid indicators and the limits of our understanding are clearly delineated by a research agenda or procedures for taking uncertainty into account. Each country will need to develop a country-specific deliberative process that fits the particular structure and authority of the relevant decision-making bodies and implementing agencies. For many political systems in the world, the legitimating authority exists to incorporate PFOA in a legislative or regulatory context but there are debates about necessary modifications of policies and regulation for transgenic organisms (Munson, 1993; Miller, 1994; Hallerman and Kapuscinski, 1995; Sagar et al., 2000; National Research Council, 2002). For some legislative or regulatory situations, a PFOA can be incorporated into a public consultative process that is authorized by regulation or it may be added as an alternative process, supported by civic society, that informs the debate in traditional decision making bodies. The following sections document a trial run of the PFOA Model proposed for this risk assessment approach. It is not a PFOA for Bt maize as a stemborer control. It is an evaluation of the concepts and protocols for this PFOA Model. In the Kenya workshop, eight participants, representing Kenyan and international scientists, evaluated the PFOA Model by taking it through a trial run. In the end, the group summarized their findings about the PFOA content and process within the context of Kenya’s deliberation over Bt maize.

Test run of the PFOA Model in the Kenyan Case Study for Bt Maize Overall, the case study provided applied insights, allowing everyone to test the questions, modify steps in the PFOA that did not make sense, and think about how a PFOA would work within their system. As we moved through each step, we were constantly working with both process and content, answering questions but also thinking about the process through the eyes of multiple stakeholders. One of the most striking realizations was that a PFOA could not be done in a single 1-day session. In addition, to make this a scientifically driven discussion, more consideration needed to be given to how the PFOA steps linked to the other sections of environmental risk assessment as well as the timing of each step. Tables 3.1–3.7 walk the reader through each step of the PFOA Model and show how this group answered the questions using two future options for comparison. In a standard PFOA process, all participants have preparatory

Environ. Risk. Chap 03

23/8/04

9:32

Page 63

PFOA for Genetically Modified Organisms

63

Table 3.2. Kenya example: problem formulation and options assessment, steps 1–3 trial run. Step 1: Problem formulation A. Whose problem is it? Whose problem should it be? It is the problem of producers, consumers and society at large a. Food security b. Poverty 1. What needs of the people are not being met by the present situation? 2. What aspects of the present situation must be changed to meet the needs? Increase yields, reduce costs of production Step 2: Prioritization and scale A. Is this problem a core problem for the people identified? Yes. Information is available in dispersed PRA reports and the Maize Database (Muhammad and Underwood, Chapter 2, this volume) 1. Do the people recognize the problem as important to their lives? Different stakeholders would rank the problem differently but for this case the farmers are considered the direct target while society and consumers are considered the indirect target for food security. 2. What are the potentially competing needs of these people? Depends on the stakeholder. For example, a farmer may have problems with soil fertility or market access. 3. How do the identified needs rank in importance to these other competing needs? There is a need for farmers to do further matrix ranking exercises (more specific information is available in Muhammad and Underwood, Chapter 2, this volume). B. How extensive is the problem? Occurs in all major maize growing zones (see Muhammad and Underwood, Chapter 2, this volume). 1. How many people are affected? All farmers in maize growing zones, to varying extents. 2. In what part of the country are these people located? 3. How large an area is affected by the problem? 4. How severe is the problem (local intensity)? Infestation may reduce maize yields by between 13 and 46%. Step 3: Problem statement: Shared understanding of the unmet need and its relative importance for a particular group of people. Context: Kenyan maize is not competitive in regional markets, resulting in maize products that are too expensive for consumers. Problem statement: Current maize varieties are susceptible to stemborers and under high infestation suffer heavy damage reducing yields and lowering profitability.

Environ. Risk. Chap 03

10/9/04

14:26

Page 64

64

K.C. Nelson et al.

Table 3.3. Kenya example: options assessment, step 5, trial run. Step 5: Option identification: Brainstorm possible alternatives to solve the identified problem, transgenic organisms would be one option. This step can be completed by the multiple stakeholder group for the initial identification of options. The multi-stakeholder group can do Steps 6–8 or a technical committee can develop a report that covers Steps 6–8 and the multi-stakeholder group can use the document to begin their evaluation and modify the assessment. Brainstormed list of technology options for maize stemborers, no preference implied by the order listed: A. Bt maize B. Intercropping/habitat management: push–pull/crop combinations C. Classical biological control (Cotesia flavipes) D. Local technical knowledge (LTK)/indigenous technical knowledge (ITK) – chilli pepper/ash E. Classical host plant resistance alone F. Synthetic pesticides – systemic/contact G. Biopesticides – Bt alone H. Botanicals – neem, among others I. Cultural control – time of planting, removal of crop residue, etc.

materials as shared knowledge. For example, they would have access to relevant scientific findings, agronomic conditions, socio-economic studies, etc. In Kenya, this was a document prepared by L. Muhammad regarding maize production and the Bt maize GMO options (see Muhammad and Underwood, Chapter 2, this volume, for the most recent version of this document). The process is designed as a facilitated discussion. The facilitator’s task is to instruct participants about the process, maintain an open discussion with participation by all members, clarify uncertainties, summarize common agreements and keep the group on task. The discussion-based process is guided by a set of questions that require agreement on the answers. For each question, the group holds an open discussion of relevant information and then proposes answers to the questions based on the common discussion. Answers are considered final when everyone agrees they ‘can live with the answer’. (There are numerous texts and training manuals that provide suggestions for ways to support efficiency, transparency, equity, decision making and other desired qualities in a facilitated discussion (Forester, 1999; Susskind et al., 2000; Hemmati, 2002; as a beginning).) In Kenya, the initial steps 1–3 (Table 3.1) proved to be a wide-ranging discussion about the state of agriculture in Kenya, world markets and pest problems. As we moved through the questions, the group began to focus on the specific problem(s) of maize production and worked to develop a shared problem statement (Table 3.2). Throughout the rest of the PFOA steps, we used this problem statement to focus our discussion when we were diverted by tangential discussions and speculations. Brainstorming the list of potential future alternatives for addressing the problem allowed participants to put anything up on the list without a veto from other participants (Table 3.3). This opened up the process and supported creative

● ● ● ●

Ready for adoption in some zones Preferred host ‘attractant’ and repellent plants Persistent during the entire cycle Integrated technology – addresses soil fertility, two pest organisms (stemborers and Striga), using two or more extra crops. Soil, two pests, two crops

Habitat management/push–pull intercropping (Khan et al., 1997a,b, 2000)

PFOA for Genetically Modified Organisms

Continued

9:32

3. What barriers exist for the option? i.e. is the seed distribution system in place; can the potential solution be integrated into present production; can the farmers afford the potential solution? Bt maize is not yet commercially available in Kenya. Social ● Not all farmers can use it acceptability may be a barrier to adoption. This will depend on ● Farmer knowledge needed, meaning you have to scale up extension public opinion (influenced by international developments and ● New technology = adoption curves unknown health/environmental effects). Access to markets will ● Area used for ‘trap’ cannot be used for maize production be a barrier if those markets demand non-GM produce. Lack of ● Increased labour for controlling maize–Desmodium intercrop in high quality, low-cost seed maize due to limitations in the the first year distribution network may act as a barrier to adoption in remote ● Limited scientific knowledge about other ‘trap/host’ and areas of Kenya repellent species ● Cost and limited supply of Desmodium seed ● Reduction of other intercrop options (beans) ● Limitation of mechanized agriculture

23/8/04

2. What is the range of crop production systems the option is likely to be used in or have an effect on? Bt maize is expected to be effective in all Kenyan production Target small-scale crop and livestock farmers, currently ready for systems where the significant stemborer species are susceptible mid-high altitude. Not large maize producers to the line in question. For Bt hybrids, it is likely to be used in those areas where hybrid maize is already widespread, whereas adoption of Bt open-pollinated varieties (OPVs) is more likely where farmers usually buy OPVs

1. What are the attributes of the option? (characteristics) Seed-incorporated resistance to stemborer; product; pest specific; easy to adopt

Bt maize

Table 3.4. Kenya example: options assessment, step 6 attributes, Bt maize and habitat management/push–pull intercropping trial run.

Environ. Risk. Chap 03 Page 65

65

Farmers can afford the technology with some problems associated with initially establishing Desmodium. Napier grass is locally available, cheap, easy to grow; diseases may be a threat

Habitat management/push–pull intercropping (Khan et al., 1997a,b, 2000)

9:32

66

6. How might the use of the option change cropping practices, such as tillage systems or pesticide use (including impacts on non-target pests)? The control of stemborers by Bt maize may result in a decrease Reduces pesticide use. Encourages minimum tillage in subsequent in insecticide use, or a reduced need for insecticides to be cropping cycles. It is difficult to integrate this method into large-scale adopted in areas where they are not currently used commercial production systems due to limitations on mechanization

23/8/04

5. Can its use be expanded to other cropping systems or geographic regions? The Bt transgene can be expected to spread through the maize Once the farmer learns the trap crop methodology they may transfer population depending on the rate of natural and artificial this principle to other cropping systems. It could be modified for use selection and the rate of use. Farmer selection and insect by the same sector of farmers in other regions pressure will result in expansion to other maize cropping systems

4. How will its use beneficially affect internal and external (nearby) crop production systems? Internal crop production systems will benefit from the increased ● Refuge for natural enemies of maize stemborers in Napier grass yields resulting from control of stemborer populations. ● ‘Freeloader’ uses his neighbour’s trap crop to reduce stemborer Expectation that input costs of insecticide are reduced. External ● Neighbour may adapt after observing the benefit (nearby) cropping systems may benefit in subsequent seasons ● Increased source of fodder for community from some stemborer control resulting from gene flow of Bt ● Reduced weeding in second year transgenes. Neighbouring farmers may in addition be motivated ● Reduced lodging of plants to adopt the Bt technology if they witness benefits associated ● Increased soil fertility with its use ● Controls Striga ● Soil erosion cover ● Increased soil moisture

Bt maize

Table 3.4. Continued.

Environ. Risk. Chap 03 Page 66

K.C. Nelson et al.

● Multiple benefits result from complex management tool ● Increase biological diversity ● Improved human diets with complex system: crop, manure, milk, seed reserve ● Room for farmer experimentation to meet site-specific needs

9:32

10. What information is needed to demonstrate that the anticipated benefits are likely to occur? ● Demonstration plots for farmers ‘to see is to believe’ ● Training for farmers on how to work with the system ● More information on which species are refuges based on location- ● How to produce seed locally specific data ● Demonstration plots – ‘to see is to believe’ ● Impact on animals – domestic and wild ● Comparative advantage against other options: there is a lot of information on biological control and pesticides. We need more information on local technical knowledge, other local practices, crop rotation (contradictory data now), crop residue infestation ● Data to reduce uncertainty about impact of Bt maize on environmental and health issues

23/8/04

9. How might the potential solution positively affect the structure of agricultural sector? ● Agriculture system based on improved hybrids and increased ● Opportunity for local economic development: seeds for income attention to seed quality generation ● Expansion and possible diversification of the agricultural ● Increased markets responding to yield increases sector based on maize production ● More diversity in agricultural production, for example milk production

8. How will anticipated changes in agricultural practices affect the needs identified in steps 1 and 2? Any decrease in pesticide use will reduce production costs Reduce stemborer and Striga infestation resulting in better yields and (this follows from the problem statement) reduced production costs. It will increase soil fertility resulting in a more robust plant and the possibility of higher production.

7. What useful practices are reinforced by the potential option? If Bt maize leads to increases in hybrid adoption, this could be viewed as reinforcing a useful practice

Environ. Risk. Chap 03 Page 67

PFOA for Genetically Modified Organisms 67

Habitat management/push–pull intercropping

68

2. What changes in the local community might contribute to the solution? Local seed distribution sites and farmer training options will need ● Establish local seed distribution sites and farmer training options. to be established. Special emphasis can be placed on training Special emphasis can be placed on training trainers or community trainers or community resource persons. On-farm demonstrations resource persons to build confidence and ownership of the technology to enhance ● On-farm demonstrations to build confidence and ownership of the adoption. Well-run local seed banks may assist farmers in technology to enhance adoption. Well-run local seed banks for trap distribution and quality concerns and repellent species as well as crop species may assist farmers in distribution and quality concerns 3. What changes in government support for farmers might contribute to the solution? ● Government programmes will need to support farmers if the ● Need increased resources for extension or farmer-to-farmer farmer is going to benefit from reducing input costs rather than education programmes cost savings being passed on to the consumer ● Need marketing of maize and milk to respond to increased ● Variety Release Committee will need to expand to include production representatives from the Ministry of Health and identified ● Need back-stopping research to understand interactions and environmental groups validate pest control benefits ● The extension services need to be strengthened to supply appropriate information ● Seed inspection services will need to be enhanced to increase farmer confidence in the seed. Concerns exist about seed quality and truthfulness in labelling ● Government evaluation system for seeds supported by scientific data provided by public institutions ● Need a monitoring system for tracking GMO and non-GMO products

1. What changes in farm management practices might contribute to the solution? Bt maize requires an integrated package of pest and agronomic ● Increased farmer confidence and ability to manage intercropping management. For example, the use of cultural control and others systems would enhance stemborer control by Bt maize ● Need crop/livestock systems to take advantage of this technology ● Other control strategies could be added to this technology

Bt maize

Table 3.5. Kenya example: options assessment, step 7 changes, Bt maize and habitat management/push–pull intercropping trial run.

Environ. Risk. Chap 03 10/9/04 14:26 Page 68

K.C. Nelson et al.

5. What other changes would likely be needed to facilitate widespread use of this alternative? ● There are several changes: improvement of extension services ● There are several changes: improvement of extension services to to support Bt maize, training about integrating Bt maize in the support the technology, training about integrating push–pull in the system, promotion of Bt maize, seed subsidies especially for farm system, promotion of habitat management benefits, seed small farmers, regulations for seed distribution to ensure quality, production of the trap and repellent plants strengthen seed quality inspection services ● A philosophical emphasis on eco-friendly, natural, sustainable ● The general infrastructure of rural Africa will need to be improved production systems focused on habitat management to provide the best services for farmers in roads, communication and distribution options ● Adoption of Bt maize may contribute to increased household income that could be used for better storage structures and other improvements that would be multiplied across the rural sector

4. What changes in the structure of agricultural production might contribute to the solution? ● More seed companies will result in healthy competition and ● Improved cooperative systems for diverse product processing, improved services distribution and marketing ● There may need to be a segregated maize management system ● Increased support for diverse products coming from this new if the market requires it. This may be expanded to products that technology such as milk, new fodder species, etc. have used Bt maize such as milk, meat, etc. ● Local seed producers for trap and repellent systems ● Increased capacity in maize storage and distribution to respond to the increased production

Environ. Risk. Chap 03 23/8/04 9:32 Page 69

PFOA for Genetically Modified Organisms 69

Habitat management/push–pull intercropping

9:32

70

3. What are the potential adverse effects of these changes internally and externally to nearby crop production systems? ● Uncertainty about these adverse effects ● Possible reservoir for plant diseases and other insect pests (?) ● Cheaper production of Bt maize may threaten other maize and ● Link to pest scientists: what is the relationship of other crop borers, crop production systems, specifically: markets and plant diversity disease, legume pests to attractant crop? ● Efficiency of pesticide use may be reduced at the farm level because one primary crop will not be using the equipment, etc.

23/8/04

2. How might the potential solution reinforce poor agricultural practices or disrupt useful practices? ● With Bt maize, farmers could believe they do not have to take ● Limits introduction of more mechanized agriculture care of other production stresses ● It may not be practical to manage such a complex system on a ● Farmers may relax other pest control strategies thinking Bt maize large scale is the ‘miracle’ control option. For example, soil pests, disease ● Reduces the crop combinations that may be used vectors and leaf hoppers ● Link to farm management: unknown whether they would stop doing other practices or increase use of other options to protect the new production gains ● There is a potential loss of local technical knowledge with Bt maize

1. How might the potential solution affect the structure of agriculture or agricultural infrastructure? ● Reduction in the pesticide market ● May contribute to keeping farmers in small-scale production ● Increased production may initially result in uneven supply and systems. The technology does not directly translate into demand problems large-scale production ● Difficult task to separate grains if you do not want to contaminate ● Reduced market for pesticides non-Bt maize ● Demand for repellent and trap seeds will require an agricultural ● GMO labelling may be needed if demanded by consumers supply system ● There could be a concentration of seed producers, with a loss ● There will be a shift from monocultures to intercropping of small regionally specific producers ● Increase in integrated crop and livestock production, on the farm or at the community level ● Organization of cooperatives may assist in economies of scale ● Increase in demand for veterinary services

Bt maize

Table 3.6. Kenya example: options assessment, step 8 impact, Bt maize and habitat management/push–pull intercropping trial run.

Environ. Risk. Chap 03 Page 70

K.C. Nelson et al.

● Baseline data associated with the diversity of present IPM practices should be used if it is available ● There is baseline data on farmers’ opinions about the system ● On-farm trial data exists

● Unlikely, remote possibility ● If you use introduced repellent or attractant species it could become invasive

23/8/04

5. What information is currently available?

4. Are any of these changes difficult to reverse, once they occur? ● Once Bt maize is part of the system it would not be easy to reverse any impacts. You could stop seed distribution to minimize impact. East African farmers will continue to recycle seeds

● Gene flow may interfere with other crops ● Risk of resistance breakdown in Bt maize may result in higher crop damage ● Non-target species may be affected

Environ. Risk. Chap 03 9:32 Page 71

PFOA for Genetically Modified Organisms 71

Habitat management/push–pull intercropping

72

5. How might the use of the option change cropping practices, such as tillage systems or pesticide use (including impacts on non-target pests)? Links: to resistance section (see Fitt et al., Chapter 7, this volume) – do we need to manage a non-Bt maize plot on site?

9:32

4. How will its use beneficially affect internal and external (nearby) crop production systems? Links: to transgene section (see Andow et al., Chapter 4, this Link: to non-target group (see Birch et al., Chapter 5, this volume): volume): yield increases will depend on efficacy of varieties against will grain/legume pests increase with Desmodium use? various stemborer species. What is the anticipated efficacy? Link: to non-target group (see Birch et al., Chapter 5, this volume): Links: to non-target effects (see Birch et al., Chapter 5, this how will disease and vector environments be affected? volume): what are the impacts (if any) on non-target pests? Links: to gene-flow section (see Johnston et al., Chapter 6, this volume): what is the expected rate of gene flow to nearby cropping systems? Links: to resistance section (see Fitt et al., Chapter 7, this volume): what is the preference/non-preference for stemborer oviposition on Bt maize as opposed to non-Bt maize?

23/8/04

3. What barriers exist for the option, i.e. is the seed distribution system in place; can the potential solution be integrated into present production; can the farmers afford the potential solution? Links: Question to an economist – what will be the additional cost (if any) of the Bt insect-resistance trait at the farmer level?

2. What is the range of crop production systems the option is likely to be used in or have an effect on? Links: to Transgene Section (see Andow et al., Chapter 4, this volume): Expression and efficacy against stemborer species depends on the Bt construct and maize variety. We assume all other characteristics of the Bt variety are the same as the recipient germplasm. Is this true?

1. What are the attributes of the option? (characteristics)

Bt maize

Table 3.7. Kenya example: options assessment, step 6 attributes, uncertainties implying important linkages with other GMO guidelines sections, Bt maize and habitat management/push–pull intercropping trial run.

Environ. Risk. Chap 03 Page 72

K.C. Nelson et al.

Environ. Risk. Chap 03

23/8/04

9:32

Page 73

PFOA for Genetically Modified Organisms

73

thinking. In Kenya we did not use the next technique, which would be to narrow down the list by clearly defining each option as distinct and appropriate for targeting the problem. Under time pressure, we picked two options – Bt maize and the push–pull system (Box 3.1) – that were of special interest to the participants. These were the options we would focus on in the trial run. In the following sessions, we used the Option Identification and Assessment Chart (Table 3.1) to compare Bt maize and the push–pull system. Moving through step 6, we worked to clarify the technology attributes, possible barriers to its use and other issues embedded in the questions (Table 3.4). In this trial run, we did not insist on answering one question at a time but rather moved through the questions, circling back to expand on our answers as we learned new things. This was the richest time for sharing information, coming to a common understanding of variables, and identifying gaps in our understanding. In step 7, we considered the changes required in order to make each option a viable solution (Table 3.5). Given the training of the participants it was easiest to discuss the on-farm changes, but working at the macro-scale and considering factors such as the structure of agricultural support, gave everyone new insights into what it takes to develop a successful technology. This understanding of assessment at different scales carried over to step 8 as we discussed adverse effects of each technology (Table 3.6). After 2 days, we were winding down in the amount of energy we had left for discussion but this step also included the greatest degree of uncertainty. Many of our responses were speculations or educated guesses that would need to be confirmed by research finding produced by the other risk assessment sections.

Box 3.1. Push–pull system. The push–pull strategy relies on diversified plantings in and around maize fields to reduce stemborer attack, Striga infestations and erosion losses (Khan et al., 1997a, 2000). The most significant maize stemborers in Kenya, Busseola fusca and Chilo partellus, feed on many graminaceous host species (Khan et al., 1997b), but prefer Sudan grass (Sorghum vulgare sudanense) and Napier grass (Pennisetum purpureum) over maize (Khan et al., 1997b, 2000). Stemborer survival, however, is lower on these hosts, due to host plant physiology and/or increased attractiveness to parasitoids (Khan et al., 1997a). Furthermore, other plants, such as molasses grass (Melinis minutiflora) and Desmodium spp., repel ovipositing stemborers (Khan et al., 1997a,b). Thus, intercropping molasses grass or Desmodium with maize would ‘push’ the pests away from maize, and nearby Sudan grass could ‘pull’ them away. The push–pull strategy also confers protection against Striga, a parasitic weed with minute seeds, when Desmodium is used as the intercrop with maize. Desmodium is not a host for Striga, and it also exerts an allelopathic effect to kill germinating Striga seeds (Khan et al., 2000). Maize fields intercropped with Desmodium spp. had vastly lower levels of Striga than maize monocultures had, and maize yield was significantly increased relative to the monoculture (Khan et al., 2000). Finally, Desmodium are nitrogen-fixing legumes that improve soil fertility and provide early season cover, which should reduce erosion and eliminate conditions favorable for Striga.

Environ. Risk. Chap 03

23/8/04

9:32

Page 74

74

K.C. Nelson et al.

Evaluation of Kenyan PFOA Overall, the participants felt the PFOA Model was practical, encouraged open dialogue, proved to be expert driven and required very few process-modifications to be applied in Kenya. The potential weaknesses were that it is only as good as the minds in the room, that meetings can be long and tiring and that too many people participating could make it ineffective. The following findings summarize the major conclusions from the trial run of the PFOA Model in Kenya. Finding 1: PFOA is a ‘good idea’ for any agricultural technology but essential for GMOs.

A science-based PFOA provides an opportunity for multiple stakeholders to review the extent of the problem, the merits of a range of options that can address the problem, and choose to support or not support a technology based on its merits in relation to other options. As the Kenyan trial run of the PFOA Model was finishing up, one of the participants said, ‘This was a really good idea, we should do this for all of our agricultural technologies.’ Everyone in the room supported his statement and the discussion went on to identify ways a societal PFOA could become part of Kenya’s GMO oversight process. The deliberative problem identification and technology review process offers a systematic, participatory approach to making decisions that will impact the nation, its people and its environment. Participants believe this review is especially important when implementing a GMO technology that has the possibility for irreversible consequences that could negatively impact society beyond the user. Finding 2: The ‘needs’ concept proved to be particularly useful for encouraging constructive dialogue and potential agreements; however, more work is required on the distinction of ‘needs’ and ‘interests’.

A multi-stakeholder PFOA provides a technology evaluation process in which all stakeholders can contribute to the public discussion about the role of transgenic organisms in their nation. It focuses discussion on broad societal concerns rather than narrow individual interests. It requires that participants struggle to frame the risks and benefits of a technology in terms of multiple social goods. After framing the discussion in this way, participants can begin a participatory process, using the best data available to understand the contribution and impact of a particular technology. It creates the potential for planned development at the national level and encourages the exploration of potential agreements among many stakeholders. Formulating the shared problem statement (Table 3.2) made the group aware of the difficulty of directly linking a problem such as stemborer damage to basic needs for food, shelter and safety. Issues of profitability and competitiveness could not be excluded, but also blurred the distinction between ‘needs’ and ‘interests’. This is a problematic area that should be refined through further analysis and case studies. Finally, we discussed that while each option was considered separately, further analysis would need to address combinations of technologies. It would be possible for two control strategies to be complementary, e.g. the push–pull combined with Bt maize for mixed cropping systems in many parts of eastern Africa.

Environ. Risk. Chap 03

23/8/04

9:32

Page 75

PFOA for Genetically Modified Organisms

75

Finding 3: For a successful PFOA a nation should reduce uncertainty about GMOs when possible.

One of the most difficult aspects of the PFOA Model was dealing with uncertainty related to the potential effects of GMOs such as Bt maize and specific information about the agricultural conditions in the country. Once uncertainties had been identified and acknowledged, participants believed there were three creative ways to deal with them in the short-term: (i) be sure the right minds and representatives of all the stakeholders are in the room; (ii) establish a national and regional database of studies that address common questions; and (iii) understand if and when the answers will be available based on the proposed work in other sections. In the trial run, some of the uncertainty in the discussion emerged because experts in a particular discipline or representatives from particular stakeholder groups were not present to share their knowledge. We were limited by the minds in the room, which could be the case for any PFOA. To minimize this limitation, the agency convening a PFOA group should invite the key expertise required to discuss a particular GMO technology and the major stakeholders for a particular problem. To enhance learning and increase a clear understanding of the issues, it is important that expert scientists and representatives of major stakeholders have the opportunity to discuss the PFOA questions together. Since there is a ‘reasonable limit’ to the number of participants in a productive discussion (estimated at ~15 people), the PFOA questions could be sent out to a broader range of stakeholders and experts in an attempt to gather information and representative opinions about the topic under discussion. There could also be a staggered series of committee sessions during which specialists prepare responses to specific scientific questions that can be considered by the larger group. Uncertainty related to the genetic trait and environmental impacts may never be completely addressed but it would be possible to reduce it by maintaining a national and regional database of all the studies that address common questions about a genetically modified organism. Studies done in other ecosystems or agricultural systems should not replace site-specific research but they can provide some insights into the background necessary for reviewing the potential effects of a particular organism. Finally, the other components of this case study were designed to suggest ways to answer many of the questions that emerged during our discussion. For example, our discussion in Stage 6 was limited by many uncertainties during the Kenyan Workshop (Table 3.7), but we imagined that other sections were addressing them. We realized that future applications of the PFOA Model should consider how the timing of a PFOA discussion interrelates with the outcomes from the non-target, gene flow and resistance risk assessments (see Chapters 5–7, this volume). Alternatively, a PFOA could be developed as an ongoing process that receives feedback from these scientific assessments as data emerge. Finding 4: The case study trial helped identify key issues associated with the chosen scenarios and contributed to consensus building.

The PFOA Group was composed of people who had experience with insect control technologies from diverse national and professional backgrounds.

Environ. Risk. Chap 03

23/8/04

9:32

Page 76

76

K.C. Nelson et al.

However, this diversity did not deter us from building common points of agreement on matters of PFOA. Participants were encouraged to discuss the pertinent issues associated with Bt maize, freely drawing from their knowledge and experience. Each topic was considered in turn, and the group moved to the next topic only when the whole group was satisfied that the major and pertinent issues for each topic had been considered and covered. Wider representation of stakeholders would be desirable in an actual PFOA. It is possible that this diversity would lead to intractable disagreements. However, based on the experience in the Kenya workshop, it seems more likely that this diversity would prevent differences in opinion from becoming entrenched publicly and facilitate the mutual understanding of divergent values and framings of the problem at an early stage. In cases were consensus is impossible, the PFOA process will add to legitimacy of the decision-making process if dissent is recognized and reported transparently. Furthermore, since there was insufficient time for comprehensive discussions for each topic, only key points were considered. For example, in the selection of various alternative technologies/options for addressing the problem, the participants listed the various pest management options open to the farmer, but selected only two contrasting future scenarios for the trial run. One option was based on Bt maize and the other on agricultural practice, which is also in its early stages of adoption: the push–pull technology. Using the PFOA Model, strengths and weaknesses of these two approaches for stemborer management could be reviewed in detail, and the attributes and potential benefits of each technology as well as their weaknesses were discussed and documented. The availability of leading questions and topics for discussion and the stepwise arrangement of each topic (Table 3.1) made the discussions orderly, and gave the participants direction and discussions depth. Factors that could have led to dissent and controversy were noted, e.g. moral issues, laws and litigation on GM crops, farmers’ rights on genetic resources, seed ownership issues, etc. While it was clear that these were significant issues, the structure of the model enabled the group to acknowledge their validity and importance, and avoid entanglement in these issues. This aspect of the model should be maintained for future applications of PFOA. It is unlikely that the PFOA case study could be shortened to a 1-day process. The 2-day review worked well in Kenya. The group was coming together for the first time, and GMO organisms are often new and controversial technologies for a country, such as Bt maize in Kenya. Discussion takes time, as participants have to agree on the issues under discussion. The Kenya workshop, timing and content proved to be an excellent way forward, and should be used for other similar conferences. Finding 5: A multi-stakeholder group should conduct the PFOA.

Among its numerous functions, a PFOA involves processes in which the gaps in the needs of a society are identified and possible future solutions are compared. A PFOA assumes that the actual people affected are the centre of the assessment. Therefore, adequate and fair representation should be ensured during such assessments. The initial data gathering for a PFOA can take the

Environ. Risk. Chap 03

23/8/04

9:32

Page 77

PFOA for Genetically Modified Organisms

77

form of a participatory rural appraisal (PRA), participatory learning and appraisal (PLA), focus group discussions (FGD) or questionnaire interviews (QI) (there are numerous resources that provide possible methodologies, a few include Chambers et al., 1989; Nicholes, 1991; Krueger and Casey, 2000; sources of pamphlets, manuals and short books include International Institute for Environment and Development (IIED, 2003), Intermediate Technology Publications (ITDG, 2003), among others). To increase efficiency, efforts should be made to use existing data and encourage any ongoing studies to collect data that may be relevant for future PFOA discussions. Once the existing data are organized, representatives of the various stakeholders should participate in the PFOA review of the GMO. As mentioned above, an actual PFOA would ideally involve a wide group of stakeholders, including farmers, consumer groups, industry, environmental representatives, policy makers and technology scientists, etc. In the context of the GMO oversight in Kenya, participants felt national scientists should be involved through their national institutions (e.g. Kenya Agricultural Research Institute (KARI) in Kenya) to ensure that PFOAs are conducted with the relevant target groups. International institutions involved in the generation of data such as the International Centre for Insect Physiology and Ecology (ICIPE) and the International Agricultural Research Centres will also need to participate in this process. Whoever has the duty of selecting stakeholder participants for the PFOA will need to be bear in mind that the process is only likely to gain public legitimacy if all relevant stakeholders have the possibility to contribute. Finding 6: In Kenya, the PFOA should be embedded in the policy and regulatory process.

Kenyan policy makers and scientists have been actively involved in national, regional and international conferences focused on the design of GMO policies and regulations (National Council for Science and Technology, 1998; Thitai et al., 1999; National Council for Science and Technology/BIO-EARN, 2002, to name a few). The inclusion in our discussion of other Kenyans with greater experience and authority in the regulatory process would have been beneficial, and would have allowed us to better assess the options for embedding the PFOA into the Kenyan process. During the workshop, participants felt that any PFOA for a new technology based on a transgenic organism will be initiated by the suggestion that a particular technology would be a beneficial alternative to the way things are currently being done in a particular cropping system. As an initial suggestion, the participants proposed the following stages for a PFOA: ●

Stage 1: KARI would receive the GMO proposal and evaluate the general viability of the proposal in consultation with the Kenyan Plant Health Inspection Service (KEPHIS). This is only an entry review to assess whether the proposal is sufficiently well documented to merit conducting a PFOA. KARI would then send a recommendation as to whether to begin a PFOA to the National Biosafety Committee (NBC) for review. The NBC makes the decision that this technology is permissible under current national law and merits a PFOA.

Environ. Risk. Chap 03

78

23/8/04

9:32

Page 78

K.C. Nelson et al.

●

●

Stage 2: If the decision is to proceed, KARI and KEPHIS will convene a PFOA Board. Each PFOA Board will be composed of individuals with specific expertise and knowledge pertinent to the cropping system under review. Representatives of the multiple stakeholders will be asked to participate based on the social sectors impacted by the decision. Stage 3: The National Biosafety Committee would review the output of the PFOA. They are charged with the final evaluation of whether this technology fits into Kenya’s policy and national interest.

The mechanics of how to guide a PFOA for GMOs will need to be developed in the future. Another option for the entry point for a new technology is the Ministry of Agriculture, but several suggestions were made about how to conduct the process. In the PFOA Board process, an information-seeking period would be necessary to gather all the relevant data that currently exists. At the same time, there should be a consultation period, during which the PFOA questions should be sent to multiple stakeholder groups. Finally, the PFOA Board should meet for 3–4 days in a facilitated review of the data and discussion. We recognize that this suggested process may require significant modification prior to implementation. KEPHIS is granted significant legal authority over the assessment process, and no representatives of KEPHIS were in the group to express their point of view.

Discussion The PFOA is unique and necessary to develop a socially viable process that contributes to a risk evaluation of GMOs. The healthy debate it engenders provides a forum for consideration across disciplines, between policy makers and regulators, and among national stakeholders. At present, the discussion centres on purpose, civic engagement, timing, scale and information management, as well as links to decision making. Many scholars have provided the framework for many of the characteristics of this PFOA Model, and this scholarly work will inform continuing discussion (O’Brien, 2000; Susskind et al., 2000; Wondolleck and Yaffee, 2000; Irwin, 2001; McLean et al., 2002). At this point, it is clear that a PFOA is not merely backgrounding or context setting to risk assessment. It is not an elaborate socio-economic assessment but does use such data when available. It is not a tool to understand how to sell the idea of GMOs to the public. It is a multistakeholder discussion. It is a comparative evaluation of a GMO technology as a future alternative solution to a problem in comparison to other potential options. It sets the scope of risk assessment and frames the hazard identification process. It requires that participants consider multiple scales, discuss within an interdisciplinary context, include the diversity of stakeholder issues and finish with a recommendation for decision making. It should assist with priority setting at the risk assessment level and the decision-making level. As each nation designs their own PFOA, they must seriously consider how to achieve multi-stakeholder representation and input from civic society. This will be for all stages of the process: data gathering, discussion, and in regard to the

Environ. Risk. Chap 03

23/8/04

9:32

Page 79

PFOA for Genetically Modified Organisms

79

regulatory body that will receive the report for consideration and decision making. One option for public input prior to the deliberative discussion is to circulate the PFOA questions for comments. Responses from interested stakeholders could be organized as supporting documents for stakeholder representatives during the discussion. The strength of this option is that anyone can comment but there is efficiency in written responses. The weakness is that there is an unequal capacity to respond within every society. Proactive measures will need to be taken to solicit input from under-represented groups. A public comment period would also be possible after the PFOA report is prepared, for consideration by the decision-making body. Finally, to maintain legitimacy there should not be major biases in the composition of the decision-making body. The timing of a PFOA session relative to the risk assessment process needs to be considered further. A PFOA session would not be helpful at the end of risk assessment, after regulatory approval, or after a GMO crop has been released into the environment. It is helpful before risk assessment for problem formulation, and some components will be helpful at critical points during lab, field trial and/or release stages of an environmental risk assessment. Designers of PFOAs will have to address concerns that PFOA will delay or make GMO review impossible. A few options for addressing these concerns include running most of it in parallel to other risk assessment components or setting a definite ending date to prevent manipulation of the process. The PFOA Model is designed to provide the best recommendation based on available data and reasonable discussion. One suggestion was, for efficiency’s sake, a PFOA should be done at a larger spatial scale – regional or world, rather than the national level. The experience in the Kenya workshop showed us that we would lose a great deal if it were not done at a national scale – stakeholder specificity, micro-level impacts, discussion within national regulatory/goals context, to name a few. At this point in history, the nation state is the main organizing unit for society. Regional- or internationalscale discussions can assist but should not replace a PFOA deliberation at the national level. Regional or international initiatives can assist with common data management, availability and funding of studies with broad implications but they cannot replace the national level discussion and recommendations. Implicit to the PFOA is a concern about data management and analysis. The information and studies used within the process must adhere to scientific values such as reliability and validity as well as public policy values, such as efficiency and accessibility. Information and research that would be necessary prior to the PFOA will need to be balanced between existing information and key studies. Existing government data can be the foundation – e.g. PRAs, census data, farm surveys, etc. Possible studies could be life cycle analysis, livelihood analysis, limited cost–benefit assessment, and market studies, among many others (along with previously mentioned methodological resources, International Service for National Agricultural Research (ISNAR) publications may add economic assessment tools (ISNAR, 2003)). As a guide for evaluating how to invest in new studies, we have to remember that the PFOA is not a socio-economic research project. It should use available studies and expertise. It can encourage the organization of current data gathering or identify data needs to facilitate comparative analysis but it is not a separate research agenda.

Environ. Risk. Chap 03

80

23/8/04

9:32

Page 80

K.C. Nelson et al.

Once the preliminary data have been compiled, special consideration will have to be given to the management of information during the discussion. Most proposals from other natural resource arenas argue for streamlining and improving the systematic progress of data discussion (Grimble and Wellard, 1997; Kessler and Van Dorp, 1998; Schmoldt and Peterson, 1998; Biggs and Matsaert, 1999; Loevinsohn et al., 2002, to name a few). For example, generic criteria should be developed to begin the option comparison process but these criteria can be modified to make them situation specific. Finally, we reiterate that the concepts of needs and interests will require additional research before they can be implemented broadly. Finally, as the PFOA section develops in future workshops, participants should recognize that strong links to risk assessment and decision making are necessary to prevent the PFOA from being a pointless exercise. In its strongest design, a PFOA assists the risk assessment process by framing concerns and identifying critical uncertainties. It provides multi-stakeholder recommendations based on the best available data for the official GMO decision-making authority within a nation. Its best use would be to inform policy discussions by identifying the common ground among multiple stakeholders centring around the GMO technology compared to alternative future solutions for the problem.

References Biggs, S. and Matsaert, H. (1999) An actor-oriented approach for strengthening research and development capabilities in natural resource systems. Public Administration and Development 19, 231–262. CBD (1992) Convention on Biological Diversity: Convention text, www.biodiv.org/biosafety/signinglist.asp (accessed October 2003). Chambers, R., Pacey, A. and Thrupp, L.A. (1989) Farmer First: Farmer Innovation and Agricultural Research. Intermediate Technology Publications, London. Forester, J. (1999) The Deliberative Practitioner: Encouraging Participatory Planning Processes. MIT Press, Cambridge, Massachusetts. Gibbons, M. (1999) Science’s new social contract with society. Nature 402, C81–C84. Glover, D., Keeley, J., McGee, R., Newell, P., Da Costa, P., Ortega, A.R., Loureiro, M. and Lin, L.L. (2003) Public Participation in National Biosafety Frameworks: a Report for UNEP-GEF and DFID. Institute of Development Studies, Brighton, UK. Goldstein, I. (1993) Training in Organizations, 3rd edn. Brooks/Cole Publishing Company, Pacific Grove, California. Grimble, R. and Wellard, K. (1997) Stakeholder methodologies in natural resource management: a review of principles, contexts, experiences and opportunities. Agricultural Systems 55, 173–193. Hallerman, E.M. and Kapuscinski, A.R. (1995) Incorporating risk assessment and risk management into public policies on genetically modified finfish and shellfish. Aquaculture 137, 9–17. Hemmati, M. (2002) Multi-stakeholder Processes for Governance and Sustainability: Beyond the Deadlock and Conflict. Earthscan Publications Inc., London. Intermediate Technology Development Group (ITDG) (2003) Publications, www.itdgpublishing.org.uk (accessed October 2003).

Environ. Risk. Chap 03

23/8/04

9:32

Page 81

PFOA for Genetically Modified Organisms

81

International Institute for Environment and Development (IIED) (2003) Publications, www.iied.org (accessed October 2003). International Service for National Agricultural Research (ISNAR) (2003) Publications, www.isnar.cgiar.org/publications/ (accessed October 2003). Irwin, A. (2001) Constructing the scientific citizen: science and democracy in the biosciences. Public Understanding of Science 10, 1–18. Kessler, J. and Van Dorp, M. (1998) Structural adjustment and the environment: the need for an analytical methodology. Ecological Economics 27, 267–281. Khan, Z.R., Ampongo-Nyarko, K., Chiliswa, P., Hassanali, A., Kimani, S., Lwande, W., Overholt, W.A., Pickett, J.A., Smart, L.E., Wadhams, L.J. and Woodcock, C.M. (1997a) Intercropping increases parasitism of pests. Nature 388, 631–632. Khan, Z.R., Chiliswa, P., Ampong-Nyarko, K., Smart, L.E., Polaszek, A., Wandera, J. and Mulaa, M.A. (1997b) Utilisation of wild gramineous plants for management of cereal stemborers in Africa. Insect Science Application 17, 143–150. Khan, Z.R., Pickett, J.A., van den Berg, J., Wadhams, L.J. and Woodcock, C.M. (2000) Exploiting chemical ecology and species diversity: stem borer and striga control for maize and sorghum in Africa. Pest Management Science 56, 957–962. Krueger, R.A. and Casey, M.A. (2000) Focus Groups: a Practical Guide for Applied Research. Sage Publications, Thousand Oaks, California. Loevinsohn, M., Berdegué, J. and Guijt, I. (2002) Deepening the basis of rural resource management: learning processes and decision support. Agricultural Systems 73, 3–22. McLean, M.A., Frederick, R.J., Traynor, P.L., Cohen, J.I. and Komen, J. (2002) A conceptual framework for implementing biosafety: linking policy, capacity and regulation. International Service for National Agricultural Research Briefing Paper 47. Miller, H.I. (1994) A need to reinvent biotechnology regulation at the EPA. Science 266, 1815–1818. Munson, A. (1993) Genetically manipulated organisms: international policy-making and implications. International Affairs 69, 497–517. National Council for Science and Technology (1998) Regulations and Guidelines for Biosafety in Biotechnology for Kenya. NCST, 41. National Council for Science and Technology, Nairobi. National Council for Science and Technology and BIO-EARN, East Africa (2002) Safety in Biotechnology of Foods and Feeds. A Kenyan Workshop under the Bio-Earn Programme. NCST, 43. National Council for Science and Technology, Nairobi. National Research Council (NRC) (1983) Risk Assessment in the Federal Government: Managing the Process. National Academy Press, Washington, DC. National Research Council (NRC) (1996) Understanding Risk: Informing Decisions in a Democratic Society. National Academy Press, Washington, DC. National Research Council (2002) Environmental Effects of Transgenic Plants. Committee on Environmental Impacts Associated with Commercialization of Transgenic Plants. Board on Agriculture and Natural Resources: Division on Earth and Life Studies: National Research Council. National Academy Press, Washington, DC. Nicholes, P. (1991) Social Survey Methods: a Field Guide for Development Workers. Development Guidelines No. 6. Oxfam, Oxford, UK. O’Brien, M. (2000) Making Better Environmental Decisions: an Alternative to Risk Assessment, 3rd edn. MIT Press, Cambridge, Massachusetts. Sagar, A., Daemmrich, A. and Ashiya, M. (2000) The tragedy of the commoners: biotechnology and its publics. Nature Biotechnology 18, 2–4. Schmoldt, D. and Peterson, D. (1998) Analytical group decision making in natural resources: methodology and application. Forest Science 46, 62–75.

Environ. Risk. Chap 03

82

23/8/04

9:32

Page 82

K.C. Nelson et al.

Stern, P. and Fineberg, H. (eds) (1996) Understanding Risk: Informing Decisions in a Democratic Society. Committee on Risk Characterization. Commission on Behavioural and Social Sciences and Educations. National Research Council. National Academy Press, Washington, DC. Susskind, L., Levy, P.F. and Thomas-Larmer, J. (2000) Negotiating Environmental Agreements: How to Avoid Escalating Confrontation, Needless Costs, and Unnecessary Litigation. Island Press, Washington, DC. Thitai, G.N.W., Mbaratha-Rurigi, J., Gakuru, O. and Amuyunzu, P. (1999) Kenya biosafety framework. UNEP/GEF Pilot Biosafety Enabling Activity Project. National Council for Science and Technology, Nairobi. Wondolleck, J.M. and Yaffee, S.L. (2000) Making Collaboration Work: Lessons from Innovation in Natural Resource Management. Island Press, Washington, DC.

Environ. Risk. Chap 04

23/8/04

4

9:32

Page 83

Transgene Locus Structure and Expression of Bt Maize D.A. ANDOW, D.A. SOMERS, N. AMUGUNE, F.J.L. ARAGÃO, K. GHOSH, S. GUDU, E. MAGIRI, W.J. MOAR, S. NJIHIA AND E. OSIR Corresponding author: Dr D.A. Andow, Department of Entomology, University of Minnesota, 219 Hodson Hall, 1980 Folwell Avenue, St Paul, MN 55108, USA. E-mail: [email protected]

Introduction Transgene locus structure and expression are critical elements for the assessment of environmental risks of transgenic plants. A transgene locus is the physical location on a chromosome formed by integration of a delivered transgene. Consequently, a transgene locus includes the various elements of the delivered DNA, such as the target gene and its promoter, introns, exons, terminators, spacer regions; a selectable marker gene and its promoter; and possibly the backbone of the plasmid carrying these transgene elements. The transgene promoter regulates the tissue- and developmental-specificity of transgene expression (transcription), which may be modulated by the inclusion of introns and exons. The structure and genomic location of the transgene locus determines the type and quantity of transgene transcripts, which in turn determines the phenotype conferred by expression of the transgene products, which may be either transgene transcripts themselves or proteins translated from the transcripts. Simple transgene loci might produce only one gene product, but more complex transgenes can produce numerous gene products. Transgene loci range from simple integrations of the delivered DNA to complex structures of integrated multiple whole and rearranged transgene copies. Knowledge of transgene locus structure is essential for understanding transgene expression and predicting the transgene phenotype providing a solid scientific basis for risk assessment. One of the most serious and vexing concerns for risk assessment of transgenic plants is that they might produce unexpected or inadvertent gene products resulting from rearrangements of the delivered transgene DNA as it is integrated into the transgene locus, or through interactions between the elements integrated into the transgene locus and adjacent plant genomic DNA, or vice versa. Such rearrangements may cause transgene or plant gene © CAB International 2004. Environmental Risk Assessment of Genetically Modified Organisms: Vol. 1. A Case Study of Bt Maize in Kenya (eds A. Hilbeck and D.A. Andow)

83

Environ. Risk. Chap 04

84

23/8/04

9:32

Page 84

D.A. Andow et al.

expression where such expression is not expected (referred to as ectopic expression) and lead to the formation of expressed unpredicted open reading frames (ORFs). These structures have previously been reported, and it is possible that they could have adverse human health or environmental effects. In addition, other concerns stem from three sources: uncertainty about gene expression from the transgene locus, interactions between the transgene products and normal gene products in the plants (pleiotropy or epistasis), and environmental interactions with gene expression. Effective characterization of transgene locus structure provides the strongest scientific evidence for reducing the uncertainties associated with transgene locus structure. For example, one concern is that the transgene will produce some unknown gene product. This can occur via known genetic mechanisms (ectopic expression, spurious ORFs, homologous or illegitimate recombination), all of which can be assessed by characterizing the transgene locus structure. Another concern is that the transgene could block expression of an important plant gene, and this can also be assessed by characterizing the transgene locus structure. Finally, a concern that the transgene might ‘jump’ from one location to another is only a theoretical speculation at this time. Despite this fact, even this concern can be assessed by characterizing transgene locus structure. Transgene locus complexity determines the gene products that must be evaluated for risk assessment of a specific transformation event. For example, Bollgard® cotton Event 531 has two transgene loci (Monsanto, 2002). One is functional and contains single copies of the full-length cry1Ac gene, the nptII gene and the aad antibiotic resistance gene. This locus also contains an 892-bp portion of the 3 end of the cry1Ac gene fused to the 7S 3 transcriptional termination sequence. This segment of DNA is at the 5 end of the insert, is contiguous and in the reverse orientation with the full-length cry1Ac gene, and does not contain a promoter but is transcribed. The second transgene locus contains 242 bp of a portion of the 7S 3 polyadenylation sequence from the terminus of the cry1Ac gene, and is evidently not transcribed. Clearly, transgene locus structure can be complex, and it is therefore crucial that the structural information is sufficient to determine if unexpected products are directly produced from the transgene. The structure of the plasmid used to produce a transformation event provides predictive information about transgene locus structure and expression. Plasmids are rarely integrated whole and intact into a transgene locus, and the genomic regions flanking the locus may influence transgene expression. Hence, plasmid structure cannot substitute for characterization of the structure of the transgene locus and is only useful for risk assessment in comparisons of plasmid structure and transgene locus structure. Characterization of transgene locus structure is conducted using Southern DNA blot analysis, PCR and sequencing of the transgene locus. Knowledge of transgene expression is essential for environmental risk assessment. It is impossible to conduct even a superficial evaluation of environmental risks without information about transgene expression. Transgene expression is conditioned by both the transgene locus (genotype) and the environment in which the transgene is expressed. Transgene transcript

Environ. Risk. Chap 04

23/8/04

9:32

Page 85

Transgene Locus Structure and Expression of Bt Maize

85

detection using Northern (RNA) blot analysis or reverse transcriptase–PCR is a common means of characterizing transgene expression. For Bt maize, the concentration of the Bt protein, the transgene product, provides another level of transgene expression information that is more directly related to the transgene phenotype. Ultimately, however, the Bt maize phenotype is the level of protection from insect pests. Because there are many species of pests, and the effect of their feeding depends on the environmental conditions of the plant, it should be appreciated that expression can be strongly determined by the environment in addition to the transgene, and that the characterization of expression can be complex. Other issues related to gene–gene and gene–environment interaction are more problematic and are addressed in the appropriate sections of this chapter in the context of risk assessment. This chapter describes the outcome of the working group on transgene locus structure and expression. The current knowledge of transgene locus structure and expression of Bt maize lines being evaluated for introduction in Kenya is reviewed and considered. We outline a scientific strategy for characterizing transgene locus structure and expression and apply it to Bt maize. This strategy relies on general principles underlying risk assessment of transgenic organisms and on the classical distinction in genetics between genotype and phenotype, which we develop in the context of transgenes. Finally, it is based on scientific methodologies that are commonly used, widely accepted and logically rigorous.

Scientific strategy for characterizing transgene locus structure and expression Two general principles that underlie many risk assessment frameworks for transgenic organisms and are applicable to characterizing Bt maize in Kenya: 1. Case-by-case risk assessment (Annex 3 in CBD, 2000). Each transgene destined for commercial development should be evaluated individually in a risk assessment. While some aspects of the transgene may be similar to other transformation events, it should not be assumed at the beginning that the transgene is similar in all other respects as well. In addition, because transgene loci are randomly located in plant genomes and locus structures are highly variable, each transgene locus produced from a given delivered transgene should also be characterized. In Kenya, this means that the risk from each of the several possible transformation events should be assessed individually. 2. Crop, transgene and environment (Annex 3 in CBD, 2000). Risk assessment should take into account the crop, transgene and environment of introduction. This means that risks should be assessed in varieties that are adapted to Kenyan production systems. Unadapted varieties will not be used in the Kenyan context, and provide limited information about the environmental dependence of risk. While some elements of the risk assessment may be generalized across both adapted and unadapted varieties, it should not be assumed at the beginning of the risk assessment that such generalizations hold.

Environ. Risk. Chap 04

86

23/8/04

9:32

Page 86

D.A. Andow et al.

The environmental (and health) risks associated with transgenic crops all originate from the structure of the transgene locus in the plant (plant genotype) and how this genotype is expressed in relevant environments (phenotype). Several factors in the process of producing a transgenic crop can affect potential risks. Consideration of how aspects of transgene design and production can affect risks is helpful for pre-emptively reducing risk during the design and production process. Moreover, these aspects can affect the need for risk assessment, which can have profound effects on the nature of the risk assessment to follow. Once the transgenic plant is created, the structure of the transgene locus can be characterized (genotypic analysis), and the expression of the transgene evaluated (phenotypic analysis). Lastly, the intergenerational transmission of the transgene should be examined to ensure that the transgene is behaving like a normal Mendelian trait in the plant (transgene transmission). Hence, this case study is divided into the following four major sections: 1. Transgene design. Proper design and construction of transgenes can reduce some aspects of risk assessment. For example, if the transgene is designed not to contain a selectable marker gene, then analysis of the risks associated with the marker do not need to be considered. Similarly, if events are produced that lack the plasmid backbone, then risks associated with potential expression of the backbone also do not need to be considered. If the selection processes for isolating transformed cells or selecting transformed plants can select for transgene loci with simple structures, risk analysis can also be simplified. For Bt maize in Kenya, we considered specific risk contexts to develop some goals for transgene design, and from an analysis of how expression of a complex transgene locus should be characterized, we proposed additional design goals that would simplify substantially the needs for risk assessment. 2. Transgene locus structure. It is important to characterize the transgene locus structure (genotype) for several reasons relating to possible rearrangements of the transgene DNA integrated into the transgene locus and possible interactions between the transgene locus and genomic DNA flanking the locus. (i) The transgene DNA could integrate into or be adjacent to plant genes and perturb their expression either by decreasing or increasing their expression. (ii) The transgene could be expressed ectopically (expressed in an unanticipated manner) through actions from promoters in adjacent plant genes. Conversely, plant genes could be expressed ectopically via interactions of plant gene ORFs with promoter elements in the plant transgene. (iii) Transgene rearrangements during integration can create spurious ORFs. Spurious ORFs could allow the transgene to produce unintended gene products. (iv) Recombination due to repeated sequences in the transgene could result in intralocus instability or may lead to ectopic recombination and possible movement of the transgene to other locations in the plant genome. (v) The theoretical possibility that the transgene had been incorporated into a transposable element could be assessed. For Bt maize in Kenya, we propose that the transgene locus is sequenced and the sequence analysed for the possibilities listed above. 3. Transgene expression. Characterizing the phenotype is essential for risk assessment. The characterization of expression (phenotype) should address what is essential for evaluating efficacy, non-target effects, resistance

Environ. Risk. Chap 04

23/8/04

9:32

Page 87

Transgene Locus Structure and Expression of Bt Maize

87

management and gene flow effects. Characterizing expression for each of these will likely require different information. Moreover, the crop phenotype should be evaluated across genetic backgrounds, in different environments, and during varying ontogenetic states. If there is variability in expression, it is essential to characterize it fully so that the implications of the variability can be assessed. Characterizing expression should focus on the target gene or genes, the marker gene (if present), and gene–environment interactions through testing in multiple locations over multiple years in the whole plant. 4. Transgene transmission. Evaluation of transgenerational inheritance is essential to ensure that the transgene is inherited as a normal Mendelian trait. A transgene can be integrated into the plant genome but its phenotype can be altered because the genetic background changes or because the gene is silenced or enhanced. This section focuses on the transgenic plant in crosses with conspecifics (including other varieties, landraces, wild escapees and feral subpopulations). It does not include interspecific crosses, such as to wild relatives or other sexually compatible species.

Scope of assessment of Bt maize events available for use in Kenya The transgenes that were being considered for introduction into Kenya during 2002 were reviewed by the authors, recommendations offered for the design of Bt maize transgenes that would reduce environmental risks, and findings listed about the present status of knowledge about transgene locus structure, expression and transmission. Our main findings were that the public-sector transgenes are in such an early stage of development that they have not yet been fully characterized; however, the private-sector transgenes, which have been commercialized elsewhere in the world, also have not been fully characterized, or the scientific data are not publicly available. In addition, according to the limited efficacy testing done, none of the lines exhibit sufficient control efficacy against both key stemborer pests in Kenya. This is an acceptable state of knowledge for the public-sector CIMMYT (International Centre for the Improvement of Maize and Wheat) lines, which are in the early stages of development. Several different Bt maize transformation events involving five different Cry toxin genes initially were considered for use in Kenya (IRMA, 2001, 2002; E. Osir, Nairobi, 2002, personal communication). Subsequently, new transgenic maize lines were developed and the focus of the Insect Resistant Maize for Africa (IRMA) project shifted (Muhammad and Underwood, Chapter 2, this volume). Presently, one private-sector event and perhaps nine public-sector events are being considered for use in Kenya (Table 4.1). Mon-810, a private-sector event, is a part of a USAID-funded research project at International Centre for Insect Physiology and Ecology (ICIPE) (E. Osir, Nairobi, 2002, personal communication). The nine public-sector events were developed by CIMMYT and are being evaluated by the IRMA project (IRMA, 2001, 2002). The IRMA project is a formal collaboration between CIMMYT and KARI (Kenya Agricultural Research Institute), which was established during 1999 to develop public-sector Bt maize for use in Kenya. These nine events are called ‘second-generation’ events.

Environ. Risk. Chap 04

23/8/04

9:32

Page 88

88

D.A. Andow et al.

The private-sector line is available in finished yellow maize hybrids, and can be converted to the white maize favoured in Kenya. Based on conventional maize breeding techniques, it should be possible to backcross a transgene into white maize backgrounds in five or more generations of breeding (taking 2 or more years). Thus, it is possible that adapted white maize hybrids could be available soon after publication of this volume. The public-sector second-generation CIMMYT lines (Table 4.1) are in the early stages of development in the IRMA project. They project that the first Bt maize varieties will be ready for commercial use in Kenya around 2007 (David Hoisington, Mexico City, 2004, personal communication). All of the secondgeneration events were created by CIMMYT using linear transformation with only the cry gene construct (cry1Ab and cry1Ba) via co-bombardment with a bar-selectable marker construct (David Hoisington, Mexico City, 2004, personal communication). These are events 3, 6, 10, 58, 93, 127, 216, 223 and 396. It is likely that further evaluation and selection among these secondgeneration public-sector lines will be necessary before full characterization of transgene locus structure, expression and transmission is completed. The private-sector event 176 and the first-generation public-sector event 1835 are listed in Table 4.1 for comparison. Event 176 has been used in the IRMA project to evaluate the efficacy of Cry1Ab against stemborers, and the IRMA project will not incorporate it into any Kenyan maize lines (David Hoisington, Mexico City, 2004, personal communication). Event 1835 is a cry1Ba event. Efficacy testing against the major stemborer pests in Kenya had not been completed at the time of this assessment (some results are discussed below), so it is not entirely clear which lines, if any, would be useful in Kenya. More recent results on the second-generation public-sector events (David Hoisington, Mexico City, personal communication) indicate that events based on Cry1Ab and Cry1Ba provide the best control of stemborers, and efforts within the IRMA project are concentrating on these. We recognize that for all of the public-sector events, it is premature to expect that the information is available to complete a full characterization of transgene locus structure, expression and transmission. Despite this, it will be useful to illustrate what would be involved in such a characterization. We also recognize that there is considerable information on the private-sector events that is not publicly available. Some of the needed information probably exists for the two listed private-sector events, but was unavailable to us via our Kenyan collaborators. Hence, the characterization of these events is incomplete. As addressed in Chapter 1, however, we present only publicly available information in the interests of transparency and scientific rigor.

Transgene Design of Bt Maize Events The Bt maize events that are the subject of this study were produced using biolistics. Briefly, the transgene(s) are constructed into a plasmid, which is adsorbed on to microprojectiles that are shot into plant cells, the delivered DNA elutes from the microprojectiles and is integrated into the plant genome, creating

Environ. Risk. Chap 04

23/8/04

9:32

Page 89

Transgene Locus Structure and Expression of Bt Maize

89

Table 4.1. Bt maize transformation events considered in this chapter. Event

Plasmida

Source

176c

pCIB4431 and pCIB3064

Private-sector, Syngenta

Mon-810c

PV-ZMCT01 and PV-ZMGT10

Private-sector, Monsanto

1835

pCIRAD4

3

Unknown

6

Unknown

10

Unknown

58

Unknown

93

Unknown

127

Unknown

216

Unknown

223

Unknown

396

Unknown

1st-generation public-sector event (Frutos, CIRAD) 2nd-generation public-sector evente 2nd-generation public-sector evente 2nd-generation public-sector evente 2nd-generation public-sector evente 2nd-generation public-sector evente 2nd-generation public-sector evente 2nd-generation public-sector evente 2nd-generation public-sector evente 2nd-generation public-sector evente

aIdentity

Genes on introduced plasmidsb PEPC:cry1Ab POL:cry1Ab 35S:bar bacterial?:bla d E35S:cry1Ab 35S?:nptII 35S?:EPSPS 35S?:gox Ubi:cry1Ba 35S:bar Ubi:cry1Ba 35S:bar Ubi:cry1Ba 35S:bar Ubi:cry1Ba 35S:bar Ubi:cry1Ba 35S:bar Ubi:cry1Ba 35S:bar Ubi:cry1Ba 35S:bar Ubi:cry1Ab 35S:bar e Ubi:cry1Ab 35S:bar e Act:cry1Ab 35S:bar e

of plasmid used to transform maize. For the Monsanto line, insufficient information was publicly available to determine the plasmid source unambiguously. bName of genes on the plasmids. Each promoter–gene combination is listed on a separate line, listing the promoter, followed by a colon and the gene. ? indicates that the information was not available publicly. Cry toxin genes are listed before the marker genes. PEPC, promoter from phosphoenolpyruvate carboxylase; POL, pollen-specific promoter; 35S, the 35S promoter from cauliflower mosaic virus; E35S, a 35S promoter enhanced for higher expression in plants; Ubi, ubiquitin promoter; Act, actin promoter; bar, phosphinothricin acetyl transferase; bla, ampicillin resistance; nptII, neomycin phosphotransferase; EPSPS, 5enolpyruvyl shikimate-3-phosphate synthase. cAGBIOS (2003). Event 176 is not being considered for introduction into Kenya by IRMA, but is used experimentally. dIRMA (2001) (Fig. 4.1) indicate that the promoter was CaMV 35S, while AGBIOS (2003) indicates that the promoter was of bacterial origin. eDavid Hoisington, Mexico City (2004), personal communication. Event 176 was used experimentally by the IRMA project and is not being considered by them for use in Kenya. Mon-810 is part of a USAID-funded project involving South Africa and ICIPE in Kenya. Events 3–376 are public-sector second-generation events that are being evaluated by the IRMA project (IRMA, 2002; David Hoisington, Mexico City, 2004, personal communication). We expect that there has been additional selection and improvement among the public-sector second-generation events.

Environ. Risk. Chap 04

90

23/8/04

9:32

Page 90

D.A. Andow et al.

a transgene locus. After transformation, plant cells are selected, usually aided by a selectable marker gene, and the transformed cells are regenerated into whole plants. Transformed plants are selected for the target trait, and then incorporated into plant breeding programmes, where commercial varieties can be produced. A transgenic lineage derived from a single transformed cell is referred to as a transformation ‘event’. The biolistic transformation method often produces complex transgene loci that exhibit multiple copies of the transgene at each locus, as well as potentially complex rearrangements within the transgene locus (Pawlowski and Somers, 1998; Svitashev et al., 2000; Windels et al., 2001). The main strategy is to produce a large number of transformation events from a specific transgene construct and to select events that exhibit simple transgene loci that confer stable transgene expression. This is the case for the private-sector events that have been commercialized already, so it should be expected that nearly all of the information necessary for characterizing transgene locus structure, expression and transmission is available. The plasmid maps of the cassettes used for transformation are shown in the figures below. These maps are predictive of expected transgene locus structure because they indicate the relative positions of the various genetic elements that should occur in the transgene locus in the plant. Moreover, it is possible to design appropriate experimental methodologies to evaluate the structure of the transgene locus using the restriction sites in the plasmid cassettes. 1. Event 176 (Syngenta, Fig. 4.1). This is a private-sector event that is used to test the efficacy of Cry1Ab within the IRMA project (David Hoisington, Mexico City, personal communication). The IRMA project has no plan to incorporate this event into any Kenyan maize variety. The event was created using cotransformation of two plasmids, and has been used in commercial maize hybrids in a number of countries for several years. It is no longer commercially available in the USA. The reasons for its withdrawal from the USA are complex. The plasmid containing the cry genes was pCIB4431 and contained two synthetic truncated cry1Ab genes, one under the control of the PEPC promoter and the other under the POL promoter (a pollen-specific promoter), and a bla gene controlled either by a 35S (IRMA, 2002) or a bacterial (AGBIOS, 2003) promoter. The PEPC promoter was derived from the maize phosphoenolpyruvate carboxylase gene as described in Hudspeth and Grula (1989). The POL promoter was derived from a maize calcium-dependent protein kinase (CDPK), which is expressed in pollen (Estruch et al., 1994) and is 1.49 kb long. The PEPC promoter is expressed in photosynthetic (green) plant tissue, and the other is expressed in pollen. This results in a high level of expression of Cry1Ab toxin in the leaves and pollen. The CaMV terminator was cloned as the terminal sequence of the two cry1Ab genes. PEPC intron 9 was added in the 3 untranslated region to increase gene expression. Two marker genes were also incorporated in this transformation. The bla gene from Escherichia coli confers resistance to the antibiotic ampicillin. The bar gene from Streptomyces hygroscopicus codes for phosphinothricin acetyl transferase (PAT), which confers some resistance to the herbicide glufosinate (Liberty®), and was on the other plasmid, pCIB3064. The promoter for bar was the 35S promoter. The construct was developed at Ciba Seeds prior to its transformation into a Syngenta genotype.

Environ. Risk. Chap 04

23/8/04

9:32

Page 91

Transgene Locus Structure and Expression of Bt Maize

91

2. Mon-810 (Monsanto) (AGBIOS, 2003). This is a private-sector event, which is being used in a US-AID funded project involving South Africa and ICIPE in Kenya. This event was also produced by co-transformation of two plasmids, and has been used in many commercial maize hybrids in the USA for several years. Full details of the creation of this event are not available publicly (AGBIOS, 2003). The plasmid PV-ZMBK07 contained a synthetic cry1Ab under the control of the enhanced CaMV 35S promoter (E35S), which is modulated by an hsp70 intron. The E35S promoter is enhanced to allow higher levels of expression of associated genes in plants. The hsp70 intron is from a maize heat-shock protein, which also increases expression. The PVZMGT10 plasmid contained a CP4 EPSPS and gox gene. Both plasmids contained the nptII gene under the control of a bacterial promoter, and an origin of replication from a pUC plasmid (ori-pUC). More information is needed on nptII, EPSPS and their respective promoters. There is a fragment of the LacZ gene and remnants of the puC18 gene present in the plasmid. The construct terminates with the Agrobacterium tumefaciens nos terminator. 3. Event 1835 (CIRAD, Fig. 4.2). This is a first-generation public-sector event and is not being considered for use in Kenya. Like many of the secondgeneration public-sector events, it uses cry1Ba. The plasmid contained a synthetic cry1B gene fused to the maize ubiquitin Ubi-1 promoter to regulate transcription and enhance expression, and a bar gene under the control of a CaMV 35S promoter. The Ubi-1 promoter includes an exon, which is the 5 untranslated region of the maize Ubi-1 gene, and its first intron, and the cry1B coding sequence is terminated with the A. tumefaciens nos 3 untranslated region. The Ubi-1 promoter is expressed constitutively in maize. The bar gene terminates with the CaMV terminator. 4. Unspecified event (second-generation CIMMYT events). These are publicsector, second-generation events. Details about these constructs were unavailable at the time of the workshop. The events are 3, 6, 10, 58, 93 and 127. They were produced by co-transformation with a ubi:cry1Ba construct and a separate bar selectable marker construct (David Hoisington, Mexico City, personal communication). The marker was eliminated by selection on progeny for independent assortment of cry1Ba and bar. 5. Unspecified event (second-generation CIMMYT events). These are publicsector, second-generation events. Details about these constructs were unavailable at the time of the Workshop. The events are 216 and 223. They were produced by co-transformation with a ubi:cry1Ab construct and a separate bar selectable marker construct (David Hoisington, Mexico City, personal communication). The marker was eliminated by selection on progeny for independent assortment of cry1Ab and bar. 6. Unspecified event (second-generation CIMMYT events). This is a publicsector, second-generation event. Details about this construct were unavailable at the time of the Workshop. The event is 396 (Table 4.1). It was produced by co-transformation with an act:cry1Ab construct and a separate bar selectable marker construct (David Hoisington, Mexico City, personal communication). The marker was eliminated by selection on progeny for independent assortment of cry1Ab and bar.

BamHI (3297)

Kpn I (6564)

CaMV 35S promoter bar gene

pUC

PEPC promoter Hin dIII (1)

BglII (3967)

Pollen-specific promoter synthetic cryIAb gene bIa gene

CaMV terminator

Restriction sites not specified

CaMV 35S promoter

CaMV Intron 9 terminator pUC

Sal I (2790)

Nco I (2238)

CaMV terminator promoter

intron

synthetic cryIB gene

nos terminator

pUC18

Fig. 4.2. Linkage map of the plasmid used to create event 1835. Selected restriction sites are also shown (from IRMA, 2001).

bar gene Sal I (3348)

CaMV 35S promoter HindIII (3580)

pUC18

exon

5 untranslated region of maize Ubi-1 gene

Nco I (4633)

pCIRAD4

PCIRAD4 (pUC p35S bar t35S pUbi cryIB tnos)

Fig. 4.1. Linkage maps of the plasmid cassettes used to create event 176. Selected restriction sites are also shown (from IRMA, 2001).

pUC

CaMV terminator

BglII (545)

pCIB3064

synthetic cryIAb gene

EcoRI (3737)

Bam HI (8719) EcoRI (641)

pUC Eco RI (1139) Bam HI (1487)

CaMV terminator Intron 9

Kpn I (3642)

pCIB4431

SalI (4187)

92

Sal I (5584)

9:32

Nco I (5600)

23/8/04

HindIII (7845)

Environ. Risk. Chap 04 Page 92

D.A. Andow et al.

Environ. Risk. Chap 04

23/8/04

9:32

Page 93

Transgene Locus Structure and Expression of Bt Maize

93

Recommendations Many potential environmental risks can be avoided or the necessity for risk assessment eliminated by appropriate transgene design. For Bt maize in Kenya, we developed specific recommendations for transgene design to address particular environmental risks. These recommendations are not comprehensive, and we will discuss the limitations to them at the end of this chapter. 1. It is recommended that marker genes be eliminated from the transformed plants The risks associated with use of marker genes, such as antibiotic resistance genes, have been controversial. However, even use of potentially benign marker genes, such as sugars or other common biochemical products, adds steps to a risk assessment. If a marker is used and later removed from the plant genome, it will be necessary to prove this. Similarly, scientific evidence would be necessary to show that the marker is not expressed even though it is present. If it is expressed, it will be necessary to evaluate its expression, even if it had been determined to be of low risk in another plant or presumed to be safe. It is difficult to rule out the occurrence of interactions between the marker product and other plant gene products. By designing the transformation construct for eventual elimination of marker genes, risk assessment associated with the markers becomes unnecessary. 2. It is recommended that all copies of transgenes that do not express the intended product be eliminated from the transformed plants The biolistic transformation method frequently results in transgene loci with multiple copies of the whole and rearranged delivered DNA transgene. Agrobacterium-mediated transformation is also possible in maize. While this method generally produces a higher proportion of transgene loci with simpler structures compared to biolistics, at some frequency transgene loci are composed of multiple rearranged delivered DNA (T-DNA) and binary plasmid backbones. Thus for both transformation methods some transgene loci consist of whole and intact delivered DNAs. Both methods at some frequency result in more than one transgene locus per event and each locus may contain multiple transgene copies or different transgenes can be integrated into different transgene loci. Prospectively, the transformation process should be conducted to deliver only the minimum DNA sequence required to confer the transgene phenotype. If multiple delivered DNAs are integrated into a single locus, it will probably be impossible to remove them. Thus, selecting for loci that have only one copy of the whole, intact transgene and no copies of altered transgenes reduces risk assessment. This eliminates the need to search for unintended gene expression products from such transgenes. If the transgene copies are integrated into unlinked loci, they can be removed by segregation.

Environ. Risk. Chap 04

94

23/8/04

9:32

Page 94

D.A. Andow et al.

3. It is recommended to use two closely linked Bt toxin genes, each with independent action to reduce the possibility of cross-resistance in the target pests. Each gene should express at a high dose against the target pests In the resistance risk assessment and management chapter (Fitt et al., Chapter 7, this volume), there was considerable discussion about how to ensure that appropriate resistance management practices would be followed when Bt maize was released as an open-pollinated variety (OPV). Under these conditions, farmers would be expected to save their seeds to produce next year’s crop, and the transgene would segregate in such a way that mixtures of transgenic and non-transgenic plants would come to prevail in the actual working landscape. When this happens, then it will become difficult to ensure that a refuge-based resistance management strategy can be implemented. If refuge-based management is not possible, the resistance risk is increased substantially (Fitt et al., Chapter 7, this volume). One way to reduce this risk is to ensure that the transgenic Bt plants all produce a high dose of two toxins (Fitt et al., Chapter 7, this volume). These two toxins should have independent action, so that there is low likelihood of cross-resistance and must be tightly linked so that they are nearly always expressed together (Mani, 1985). Crossresistance would make the two toxins redundant and no better than one toxin. With independent action, it should require that the target insect pests have two resistance alleles at two different gene loci to be phenotypically resistant. Both toxins should express at a high dose so that all resistance alleles in the target pests are rendered effectively recessive. A resistant target pest would therefore need two copies of each resistance allele at both loci to be phenotypically resistant. If all of these assumptions hold, then phenotypically resistant insects will be rare indeed. For example, if each resistance allele were available in natural populations of the pest at 1 in 1000 (0.001), resistance phenotypes would exist in only 11012. Such a construct is less likely to suffer resistance failures than a single gene construct.

Transgene Locus Structure of the Present Bt Events in Kenya Transgene locus structure must be characterized for appropriate risk assessment because the transformation process can alter the structure of the delivered transgene. Transgene locus structure is characterized in various ways. The structural characterization includes both the fine structure of the locus itself, and the location of the locus in the plant genome. The fine structure can be characterized in a number of ways. The most common approach is to use Southern analysis to determine the presence and order of the main genetic elements of the transgene locus. The design of the strategy would be based on the complete restriction maps of the delivered plasmid DNAs. Partial restriction maps of the original plasmids of two Bt maize events are illustrated in Figs 4.1 and 4.2. Fine structure analysis is most comprehensively provided by completely sequencing the transgene locus including flanking genomic sequences. From the DNA sequence, it is possible to construct a detailed map

Environ. Risk. Chap 04

23/8/04

9:32

Page 95

Transgene Locus Structure and Expression of Bt Maize

95

of the locus that characterizes the linkage of genetic elements, restriction sites and flanking genomic sequences, some of which can be used to track the locus in subsequent generations. The location of the transgene in the maize genome can also be characterized in numerous ways with varying degrees of precision. First, the transgene can be located in the nuclear genome or in one of the organelle genomes. This can be determined readily by segregation analysis. All organelle integration events will be maternally inherited while all nuclear events will follow Mendelian inheritance. As indicated below, all second-generation public-sector Bt maize events for Kenya are nuclear transformations. Fluorescence in situ hybridization (FISH) is useful for locating the transgene locus to a specific chromosome. Linkage maps also can be established by segregation analysis with a large battery of possible markers and be used to determine which chromosome and approximately where on that chromosome the transgene locus resides. Since the maize genome has not yet been sequenced, the much more precise method of sequencing out from the transgene locus into the flanking genomic DNA can not be applied to locating the transgene locus in the maize genome. We recommend that DNA sequencing is necessary to characterize the fine structure of the transgene locus and its flanking genomic region. With sequence information, it is possible to identify rearrangements that may produce unexpected protein products, to design probes necessary to establish the presence or absence of unexpected RNA or protein products, to evaluate the risk of combinations or other rearrangements, and to evaluate the possibility that plant gene expression is disturbed or other possibilities described above. None of the other methods for characterizing transgene locus structure can enable assessment of all of these potential risks. Finding 1: Transgene locus structure has either not been characterized sufficiently or is not available for any of the Bt maize transformation events that are considered for introduction into Kenya.

Location in nuclear or organelle genomes Location of the transgene in the nuclear or organelle genome can be determined by segregation analysis. To illustrate how this can be done, we examine the first-generation CIMMYT transformations even though they will not be used in Kenya. Data provided in Pellegrischi et al. (2002) can be used to deduce if the transgene phenotype is inherited like a nuclear or organelle gene. The segregation ratios for these events are shown in Table 4.2. For a nuclear transgene, phenotypes should segregate with a 1:1 ratio (resistant to susceptible). If the transgene is in an organelle, it should be inherited as a maternal trait. If the female is the transgenic plant, then the segregation ratio should be 1:0 (resistant to susceptible), and if the male gametes come from the transgenic plant, the segregation ratio should be 0:1. The segregation data are consistent with nuclear inheritance of the transgenes (Table 4.2) for events 5207, 5601, 1835 and 7. Event 602 is not inherited like a nuclear allele. The

Environ. Risk. Chap 04

23/8/04

9:32

Page 96

96

D.A. Andow et al.

Table 4.2. Segregation ratios of first-generation public-sector events (insect resistant: susceptible). Event

Trial 1

Trial 2

Trial 3

Trial 4

P (nuclear)

5207a 5601a 1835a 602a 7a

19:17 1:15 7:9 12:13 –

28:36 17:31 6:8 0:32 6:8

24:24 23:26 12:17 0:192 12:16

65:73 190:187

0.41 0.21 0.24 <0.001 0.35

For nuclear inheritance, we expect a 1:1 segregation ratio. For maternal inheritance we expect a 1:0 or 0:1 segregation ratio, depending on whether the maternal parent is the transgenic parent or the non-transgenic parent. Data are from Pellegrischi et al. (2002). P is the probability that the transgene segregates like a nuclear allele, based on Pearson’s chi-square (analysis done here). aNone of these five first-generation events have been continued for possible use in Kenya after 2001; data are presented for illustrative purposes.

statistical analysis also shows that there was significant inter-trial variability for events 5601 and 602. Event 5601 started out with non-Mendelian inheritance but ultimately showed Mendelian inheritance. Event 602 was the opposite, showing Mendelian inheritance that switched to non-Mendelian inheritance. It is not known what caused these shifts, but gene silencing might be involved. In any case, none of these events was carried into the second-generation lines (Table 4.1). Three new second-generation lines (Table 4.1) also show Mendelian inheritance (analysis not shown). Both of the private-sector events are nuclear (AGBIOS, 2003).

Number of transgene loci in Bt maize Transformation can lead to multiple transgene integration events in a single transformed plant and these can be at one or more genetic loci and either active or inactive. The determination of the number of transgene loci should include information on non-expressing loci, because non-expressing loci tend to cause silencing of the active transgene and could express an unexpected product. Transgene silencing is stochastic in many cases creating uncertainties that need to be addressed in risk assessment. Some of these transgene insertions may not express the intended gene product, so it is necessary to use molecular (DNA or RNA) methods and not rely solely on detecting the transgene product. As mentioned above, characterization of event 531 (Bollgard® cotton) was done genetically; two loci were identified, of which one was non-functional for the target gene. Sexually propagated plants, such as maize, can be examined by coupling segregation and molecular analysis (Southern analysis). This can be a very sensitive method for determining the numbers of unlinked loci and to assure that progeny carrying only one transgene locus have been selected. For the public-sector events, there are 1–10 transgene loci in each of the first-

Environ. Risk. Chap 04

10/9/04

14:27

Page 97

Transgene Locus Structure and Expression of Bt Maize

97

generation events (IRMA, 2001). Transgene locus number was further characterized for four of the first-generation events, although it is not clear how the copy number was determined (Pellegrischi et al., 2002). Southern blots were used to determine the number of loci in the second-generation lines, but the details of these analyses are not provided (Pellegrischi et al., 2002). The reported number of transgene loci for the second-generation events is given in Table 4.3, despite this uncertainty, event 223 is the only one reported to have one transgene locus. There are two or more transgene loci of each of the plasmids integrated into the maize genome for event 176, as determined by Southern analysis (AGBIOS, 2003). Mon-810 has only one integrated transgene (EPA, 2000b; AGBIOS, 2003). For the public-sector events with multiple transgene loci, it is not yet clear how the loci segregate. At this point in the development of these lines of Bt maize, it is not necessary to have this information available. As the events are developed, independently segregating transgenes are likely to disassociate so that the final transgenic plant has only one or few loci. At that time, it would be appropriate to determine definitively transgene locus numbers.

Number of transgene copies at each locus Multiple copies of the transgene can exist at a transgene locus. When enumerating copy number, it should include non-functional copies, because they can affect expression of the functional copies. As discussed above, event 531 (Bollgard® cotton) has a transgene locus with two copies of the target cry gene. One of these cry genes is functional and produces Cry1Ac toxin, but the other gene in the locus is not functional (fusion of part of the cry gene with the terminator sequence and adjacent cotton DNA). The most accurate method for determining transgene copy number per locus is to sequence the entire locus. Other methods of transgene copy number determination such as Southern analysis provide estimates and are not definitively quantitative. For the public-sector events, Pellegrischi et al. (2002) report estimates of copy number per locus, but the methods for making these estimates are not provided (Table 4.3). None of these events has only one copy per transgene loci. They report that events 10 and 223 may have only two copies of the transgene per locus. Some events may have as many as 12 copies of the transgene per locus. As Mugo et al. (2002) report, there is still additional research to be done before the transgenic lines can be adapted into Kenyan varieties. The private-sector events have undergone considerable screening. Event 176 has two active target genes in the transgene locus (Koziel et al., 1993). Southern blot analysis of MON810 genomic DNA indicated the incorporation of a single copy of the truncated cry1Ab gene, together with the enhanced CaMV 35S (E35S) promoter and hsp70 leader sequences (AGBIOS, 2003). The NOS 3 termination signal was not integrated into the host genome but was lost through a 3 truncation of the gene cassette.

Environ. Risk. Chap 04

23/8/04

9:32

Page 98

98

D.A. Andow et al.

Table 4.3. Summary of information related to transgene locus structure for the Bt maize transformation events potentially available for use in Kenya (see the text for full justification).

Event

No. of transgene loci (copy no./locus)a

Cellular locationb

Insertion site known

Transgene sequenced in plant

Flanking regions sequenced

176c Mon-810 1835d 3 6 10 58 93 127 216 223 396

≥2 loci (2) 1 (1) 3–5 loci (14–20) 4 loci (5) 6 loci (5) 2 loci (2) 2–3 loci (7–10) 1–3 loci (10–12) 1–2 loci (3–5) 3–5 loci (8–10) 1 locus (2–3) 3 loci (9)

Chrom. Chrom. Chrom. Chrom. Chrom. Chrom. Chrom. Chrom. Chrom. Chrom. Chrom. Chrom.

No No No No No No No No No No No No

No No No No No No No No No No No No

No No No No No No No No No No No No

Additional information is presently available for the second-generation public-sector events, but this was not available at the time of the Workshop. For example, event 216 now has only one transgene locus (David Bergvinson, Mexico City, 2004, personal communication). aFor the public-sector second-generation events in this table, the number of loci and transgene copy number per locus are for an early stage of development (IRMA, 2002). Additional work has been done to simplify these events. bChrom., chromosome. cEvent 176 is not being considered for introduction into Kenya by the IRMA project, but they used it experimentally. dThis is a first generation event with ubi:cry1Ba, that is not being considered for introduction into Kenya, but is included in this table for comparative purposes. The second generation events 3–127 are also ubi:cry1Ba events.

DNA sequence of transgenes Transgene locus sequencing is recommended to provide the detailed structure of transgene rearrangements and changes to flanking genomic DNA. This is essential for evaluating the possibility that the transgene expresses some unexpected products. Sequence information would allow design of probes and PCR primers for further characterization of locus stability and tracking. Briefly, the transgene locus and flanking genomic DNA sequences or fragments thereof can be isolated by standard gene-cloning procedures or PCR-based methods. These clones or PCR products are then sequenced and contigged to reconstruct the entire locus. Transgene locus and genomic sequence should be evaluated bioinformatically to identify similarities and deviations from the delivered transgene DNA constructs, determine the integration points into the genome, evaluate the proximity to known plant genes or transposable elements and determine if new ORFs were created by the integration or rearrangements (if any) of the delivered transgene DNA or surrounding genome.

Environ. Risk. Chap 04

23/8/04

9:32

Page 99

Transgene Locus Structure and Expression of Bt Maize

99

DNA sequences of the transgene locus and the flanking regions have not been reported for any of the Bt maize events. For the public-sector events, it is premature to consider sequencing. It would be better first to simplify the events (reduce the number of transgene loci and select loci with one or a few copies of the transgene) and to ensure sufficient levels of efficacy in the lab before the events are sequenced. The sequences of the private-sector events should be available for scientific analysis. At this time, however, they were not publicly available (Table 4.3). Molecular analysis of Mon-810, however, indicated that no plasmid backbone sequences from the plasmid PV-ZMGT10 were integrated into the maize genome (AGBIOS, 2003).

Development of DNA-based monitoring tools Molecular tools should be developed based on the transgene locus sequence to monitor transgene structure in the commercial variety and its derivatives both in respect to locus stability and to allow detection or traceability of the transgene locus. These tools could be PCR primers or Southern probes. They could be used as a tag to identify a particular transgene locus and allow tracking of the event in the environment (or in contamination of non-GM crops, etc.). These tools should be developed as soon as an event is likely to be commercialized, preferably before initial field release. It will be important to determine the sensitivity and accuracy of such a tool as it is developed. These tools should be completed and verified by the time large-scale field testing commences and prior to commercial release or use. Once large-scale field testing commences, the probability that a transgene will escape into the environment increases, and monitoring tools should be ready for use at that time. Monitoring tools have not been reported for any of the Bt maize events. For the public-sector events, it is premature to consider developing such tools. Once a few simplified, efficacious events are identified, monitoring tools could be developed. Monitoring tools for the two private-sector events should be available now; however, if they exist it has not been reported publicly.

Maize variety testing While many analyses described above are likely to be conducted on the original transgenic event, some data concerning these questions should be provided for each commercial transgenic variety. For example, follow-up analysis on each commercial variety could be done with PCR primers to confirm that the transgene is still intact and in its expected location, i.e. that there was no breeding or other mix-up during variety production. In addition, follow-up analysis could be done with ELISA or Western blots to confirm that the transgene is being expressed as expected. There are no Bt maize varieties ready for use in Kenya yet, however, the need for such analysis can be planned for, so that molecular probes are established and new varieties can be evaluated and registered quickly.

Environ. Risk. Chap 04

23/8/04

9:32

100

Page 100

D.A. Andow et al.

Transgene Expression The expression of each transgene in every transgene locus needs to be evaluated. It should not be assumed that each copy of the same transgene, should multiple transgene copies be present in a single locus or multiple loci containing the same transgene exist, will be expressed in the same way. Variation in transgene expression is the general rule. For example, if a transgenic plant has two transgene loci, and at one locus there is one functional target gene, one non-functional target gene and one marker gene, and at the other there is one non-functional target gene and a non-functional marker gene, then the expression patterns of five transgenes need to be evaluated. Clearly, the fewer transgenes there are in a transgenic plant, the less work is needed to characterize the expression pattern of the transgenic plant. Transgene expression can be conducted at many levels. Consequently, it is possible to waste considerable time and effort characterizing gene expression. It is essential that expression be characterized in a way useful for risk assessment. Basically, for risk assessment, it is necessary to know what are the active transgene products, how much of each is expressed and how they perform in their intended purpose. This requires molecular, biochemical and functional characterization. At a molecular level, RNA and protein products are the expression products of a transgene. For risk assessment, this information is important, because it is necessary to measure the transgene product and ensure that experiments to assess risks use the actual product produced by the transgenic plant. Although surrogates for the actual transgene product can and have been used in environmental risk assessment of transgenic plants, evaluating the actual transgene product gives greater scientific rigor. In addition, molecular characterization is needed to prove that a particular gene is absent or produces no product, and to predict expression patterns. Biochemical characterization is needed to quantify the amount of gene product in various plant tissues as the plant develops. This information is essential to develop appropriate hazard scenarios and construct appropriate exposure scenarios during the risk assessment process. A functional characterization is essential for risk assessment because this will allow identification of potential hazards associated with environmental risk. For Bt maize, this analysis is needed so that the risk that the target pests will evolve resistance to Bt maize can be assessed and potential management interventions can be suggested to delay or prevent resistance from developing. In this section, we first assess target gene expression followed by marker gene expression in the Bt maize events that have been and may still be considered for Kenya. For the target gene, we consider the importance of molecular, biochemical and functional characterization for risk assessment.

Target gene expression Available information on target gene expression is summarized in Table 4.4. All of the transformation events express the target gene. Biochemical and functional

Environ. Risk. Chap 04

23/8/04

9:32

Page 101

Transgene Locus Structure and Expression of Bt Maize

101

Table 4.4. Summary of information related to target gene expression for the Bt maize transformation events potentially available for use in Kenya (see text).

genea

Target product

Event

Target

176d

PEPC:cry1Ab POL:cry1Ab

Present

Mon-810 1835f 3 6 10 58 93 127 216 223 396

E35S:cry1Ab Ubi:cry1Ba Ubi:cry1Ba Ubi:cry1Ba Ubi:cry1Ba Ubi:cry1Ba Ubi:cry1Ba Ubi:cry1Ba Ubi:cry1Ab Ubi:cry1Ab Act:cry1Ab

Present Present Present Present Present Present Present Present Present Present Present

Characterization Molecular

Biochemicalb

Functionalc

65 kDa and 60, 40 and 36 kDa in leavese 91 kDae Unknown Unknown Unknown Unknown Unknown Unknown Unknown Unknown Unknown Unknown

Incomplete

Tested

Incomplete Unknown Unknown Unknown Unknown Unknown Unknown Unknown Unknown Unknown Unknown

? Tested ? ? ? ? ? ? ? ? ?

Additional information is presently available for the second-generation public-sector events (3–396), but this was not available at the time of the Workshop. aAll target genes are synthetic cry genes. bThis is discussed in the biochemical characterization section below. c?, Several of these second-generation events have been tested, but the results of these tests were unavailable; available results are discussed in the functional characterization section below. dEvent 176 is not being considered for introduction into Kenya by the IRMA project, but they used it experimentally. eAGBIOS (2003). For event 176, the additional leaf proteins were detected by ELISA specific to Cry1A toxin. f This is a first-generation event with ubi:cry1Ba, that is not being considered for introduction into Kenya, but is included in this table for comparative purposes. The second-generation events 3–127 are also ubi:cry1Ba events.

characterization was incomplete for all of the events. Functional characterization is critical for determining which events might be useful for further development and for assessing resistance risk (Fitt et al., Chapter 7, this volume). Target gene expression is characterized using molecular, biochemical and functional methods. Gene–gene interaction (pleiotropy and epistasis) and gene–environment interaction are important aspects of transgene expression, so they will be treated in separate sections. In this section, we focus on describing the actual transgene product (for Bt maize, this is the protein product, not the RNA product), quantifying product concentration in various plant tissues and determining the functional significance of the produce. All of these methods can be conducted in the lab. Molecular characterization Finding 2: The target gene products of the second-generation public-sector lines have not yet been molecularly characterized. The private-sector events have been characterized molecularly.

Environ. Risk. Chap 04

23/8/04

9:32

Page 102

102

D.A. Andow et al.

Table 4.5 provides information on the promoters for the target genes. Knowing the promoter should allow predictions about the tissue specificity, ontogeny and environmental influences of expression of the transgene. These predictions, however, should be empirically tested, and would provide evidence that the transgene is expressing in a predictable manner (i.e. that unpredicted expression of the transgene was unlikely). For example, POL, the pollen-specific protein kinase promoter used in event 176, is supposed to express only in pollen. Similarly, PEPC is supposed to express only in green tissues. Hence any production in roots, the pith of the stalk or kernels should be considered unexpected and would require risk assessment. For event 176, root, pith and kernel expression is very low (AGBIOS, 2003), verifying the predictions. Because of the large number of integration events and high transgene copy number in the CIMMYT events, it is possible that each of the ubi-1 and act-1 promoters has a slightly different spectrum of activity. It will be necessary to evaluate this possibility at the appropriate time. TARGET GENE PROMOTER.

Rearrangements and mutations in the delivered DNA occurring during formation of the transgene locus may result in changes in the transgene product. The transgene product can be predicted from the sequence of the transgene locus, but either this has not been done or the information is not available publicly for all events. For the two commercially available events, event 176 and Mon-810, the size and the amino acid sequence of the toxin has

TARGET TRANSGENE PRODUCT.

Table 4.5. Summary of information related to the promoters of the target genes for the Bt maize transformation events potentially available for use in Kenya. Event 176a Mon-810 1835b 3 6 10 58 93 127 216 223 396 aEvent

Target gene promoter

Predicted expression

PEPC POL E35S Ubi-1 Ubi-1 Ubi-1 Ubi-1 Ubi-1 Ubi-1 Ubi-1 Ubi-1 Ubi-1 Act-1

Photosynthetic tissue Pollen Constitutive Constitutive Constitutive Constitutive Constitutive Constitutive Constitutive Constitutive Constitutive Constitutive Constitutive

176 is not being considered for introduction into Kenya by the IRMA project, but they used it experimentally. bThis is a first-generation event with ubi:cry1Ba, that is not being considered for introduction into Kenya, but is included in this table for comparative purposes. The second-generation events 3–127 are also ubi:cry1Ba events.

Environ. Risk. Chap 04

23/8/04

9:32

Page 103

Transgene Locus Structure and Expression of Bt Maize

103

been characterized. For other commercial events not listed here, all of the amino acid modifications have been made publicly available. For Cry1Ab, the native protein as expressed in Bacillus thuringiensis is a 130.6–135-kDa protoxin comprised of 1155 amino acids. This is called a protoxin because in this form it is not toxic to insects, and must be digested by a protease, such as trypsin, to a shorter form to gain toxicity. The trypsin-resistant core protein toxin (the ‘activated toxin’) is 65–66.7 kDa comprised of 594 amino acids. Event 176 expresses an activated Cry1Ab toxin, but it is 72.6 kDa comprised of 648 amino acids (Table 4.6). Mon-810 is neither the protoxin nor an activated toxin, but is something intermediate, and is 91 kDa comprised of 816 amino acids (Table 4.6). Both retain the amino acid structure of the activated toxin. Equivalent information is not available for the public-sector lines. Biochemical characterization Finding 3: The concentration of toxin in Bt maize has not yet been adequately characterized for any of the Bt events. For the private-sector events, more times during plant growth and more maize tissues should be sampled. For the publicsector events, toxin concentration had not been reported, but should wait until a few focal events are identified. Both public- and private-sector events will need to be characterized in hemizygous and homozygous condition.

Table 4.6. Transgene products expressed in maize. Event

Inserted gene Toxin size expressed in plant

176a

Cry1Ab

Mon-810 1835c 3 6 10 58 93 127 216 223 396

Cry1Ab Cry1Ba Cry1Ba Cry1Ba Cry1Ba Cry1Ba Cry1Ba Cry1Ba Cry1Ab Cry1Ab Cry1Ab

aEvent

72.6 kDa (plus c. 60, 40 and 36 kDa productsb) 91 kDa Unknown Unknown Unknown Unknown Unknown Unknown Unknown Unknown Unknown Unknown

176 is not being considered for introduction into Kenya by the IRMA project, but they used it experimentally. bThree additional immunoreactive proteins weighing approximately 60, 40 and 36 kDa were detected in the leaves of event 176, but not in the pollen. It was suggested that these may represent breakdown products resulting from intrinsic proteolysis within the leaf tissue (AGBIOS, 2003). cThis is a first-generation event with ubi:cry1Ba, that is not being considered for introduction into Kenya, but is included in this table for comparative purposes. The second-generation events 3–127 are also ubi:cry1Ba events.

Environ. Risk. Chap 04

10/9/04

14:27

Page 104

104

D.A. Andow et al.

Methods for accurately quantifying the products of transgenes in plants have been improving, but additional effort is needed to establish accurate, repeatable measures for the various plant tissues and crops. Extraction artefacts such as incomplete extraction from plant cells and product modification during processing (e.g. proteolytic degradation) have not been clarified, so the repeatability of various methods remains to be conclusively established. The plant phenological stages and the tissues that should be sampled during those stages have been variously reported. It is necessary to determine the protein concentration (both relative and absolute) in different genetic backgrounds, environments, stages of development (germination, flowering, maturity) and tissues. We interacted with the non-target (Birch et al., Chapter 5, this volume), resistance (Fitt et al., Chapter 7, this volume) and gene flow (Johnston et al., Chapter 6, this volume) groups to find out how often and in what tissues toxin concentration should be measured to be relevant to their risk assessments. The non-target group wanted to know concentrations at least once during the generation time for a typical arthropod species that would occur in maize. Because the typical generation time of an arthropod species is about 1 month (though aphids, scales and several other species are faster), toxin concentration should be measured for non-target assessment about three times during the maize growing season, during the vegetative, flowering and maturation stages, about 1 month apart. The resistance group was concerned about how toxin concentration changed during the development of the target pest species; specifically, they were concerned that the concentration could change enough that the plant would, for example, change from being a high-dose plant to being a low-dose plant within the life cycle of the target pest species. Hence, they suggested that toxin concentration is measured every 2 weeks during maize development. The gene flow group had no specific requirements on the timing of quantification. The number of maize tissues that should be sampled is determined by the feeding behaviour of the species that consume maize plants. Specifically, the non-target group recommended that every tissue on which a non-target species could complete development should be sampled. The resistance group recommended that every tissue consumed by the target pest species should be sampled. The gene flow group had no specific requirements on the tissues to be sampled. Therefore, a sufficient number of tissues for Bt maize would include leaves, pith, phloem, pollen, male flowers, roots below ground, adventitious roots above ground, tillers, ear leaves, silk, kernels and cob. This information is partially available for the two commercially available events (Table 4.7), but is not yet available for the public-sector events. The data should be collected soon for the private-sector events. For the public-sector events, this should be done soon after the number of possible events is narrowed to a few most promising events.

TIMING AND LOCATION OF TRANSGENE PRODUCTS IN PLANTS.

EXPRESSION IN HEMIZYGOUS AND HOMOZYGOUS PLANTS. Discussions with the gene flow group (Johnston et al., Chapter 6, this volume) highlighted an important

M M 5.21–15.06 M M M M

M M M M M M M

M M M

M M M <0.008

X M X M X M X

X M X

X M X M

X X X 0.09 X X X

X X X 1.14–2.35; 2.4 X X X

Phloem Pollen

X X X M X X X

X X X

X X X M

M M M M M M M

M M M

M M M 0.008

Male Root flowers BG

X X X M M M M

M M M

X X X M

Root AG

X M M M M M X

M M X

X M M M

Tiller

X X X M M M M

M M M

X X X M

Ear leaf

X X X M X X X

X X X

X X X M

Silk

X X X X M M 0.19–0.91

M M <0.005

X X X X

Kernel

X X X X M M M

M M M

X X X X

Cob

Time of sample is approximated from published information. X indicates measurements that are either not possible or are unnecessary, and M indicates measurements that are missing. V are vegetative stages, F is the flowering stage, and R are the reproductive stages. BG, below ground; AG, above ground (adventitious roots). aThis is a scale of maize development (Ritchie and Hanway, 1984). The V stages are vegetative stages, corresponding to the number of leaves on the plant. A healthy maize plant typically produces two or slightly more leaves in a week. The R stages are reproductive stages. R1 corresponds to flowering, when receptive silks are exposed and anthesis is nearly complete. R2 is the ‘blister’ stage, when small, white, liquidy kernels are present. R4 is the ‘dough’ stage when the endosperm of the maize kernel is pasty. R6 is physiological maturity. There is about a 2-week period between all of the listed stages, except R4 and R6, which is closer to 3 weeks. bEPA (2000a), AGBIOS (2003). Event 176 is not being considered for introduction into Kenya by the IRMA project, but they used it experimentally. cAGBIOS (2003).

V6 V10 V14 R1 R2 R4 R6

0.6–1.16 M M 0.53–3.03; 2.13–3.27 M M 0.44–0.47

Pith

9:32

Mon-810c

V6 V10 V14 R1

176b

Leaves

23/8/04

R2 R4 R6

Timea

Event

Table 4.7. Toxin concentration in maize plants (µg/g fresh weight).

Environ. Risk. Chap 04 Page 105

Transgene Locus Structure and Expression of Bt Maize 105

Environ. Risk. Chap 04

23/8/04

9:32

Page 106

106

D.A. Andow et al.

consideration for Bt maize in Kenya. Bt maize hybrids in the USA are constructed in such a way that the hybrid expresses one copy of the transgene, in what is referred to as a hemizygous state. The transgene occurs on only one of the two homologous chromosomes, and because there is not a corresponding allele on the other chromosome, the transgene is considered hemizygous (having only one copy of the gene in a diploid organism), instead of heterozygous (having two different alleles at the same locus). Bt maize is expected to be available in Kenya both as hybrids and OPVs. We do not know what breeding strategy will be used to produce either hybrids or OPVs, but it is likely that hybrids will be either hemizygous or homozygous, and OPVs will be homozygotes (David Hoisington, Mexico City, 2004, personal communication). As noted by the gene flow group (Johnston et al., Chapter 6, this volume), if farmers save seed to plant next year’s crop, both hybrids and OPVs will produce a mixture of homozygous and hemizygous transgene genotypes. All of the private-sector varieties have been characterized in hemizygous varieties. It might be expected that homozygotes would produce higher quantities of toxin, but it is possible that the transgene alleles would interfere with each other, leading to reduced toxin expression. Functional characterization Finding 4: Some of the first-generation public- and private-sector events (Cry1Ab and Cry1B) show good efficacy against Chilo partellus; however, sufficient efficacy against Busseola fusca, one of the key stemborer pests, has not yet been demonstrated.

For Bt crops, functional characterization is essential for assessing resistance risk (Fitt et al., Chapter 7, this volume). The relevant Bt function is toxicity to target pests and there are several ways to determine toxicity to target pests. If toxin is available, the LC50 assay, described below, is an excellent method for characterizing the potential functional significance of a particular Bt toxin. Other methods, such as the leaf tissue assay described below, can also provide useful results. As summarized in Chapter 2 (Muhammad and Underwood, this volume), the key maize stemborer pests in Kenya are C. partellus [Lepidoptera: Crambidae] and B. fusca [Lepidoptera: Noctuidae]. Initially these tests can be conducted in the laboratory, but at some point, the effectiveness of an event must be evaluated in the field. These methods and results are reported in Odhiambo et al. (2002), and are repeated here for the convenience of the reader. Fresh imported Bt maize leaves were brought to Kenya in February 2001. Bioassays were carried out using this leaf tissue against five species of stemborers, C. partellus, Chilo orichalcociliellus, B. fusca, Eldana saccharina and Sesamia calamistis (Table 4.8; Odhiambo et al., 2002). Survival of larvae exposed for 5 days to the leaf tissue was measured (presumably this was conducted with neonate larvae). The number of replicate larvae per leaf, the number of replicate leaves and the number of replicate experiments (presumably one replicate experiment) are not provided.

LEAF TISSUE BIOASSAYS.

Environ. Risk. Chap 04

23/8/04

9:32

Page 107

Transgene Locus Structure and Expression of Bt Maize

107

Table 4.8. Mortality of larvae in leaf tissue bioassays. Percentage mortalityb Eventa

Cry toxin

Control CML216 Control H614 176c 5207 (T5)d 5601 (T5)d 1835 (T3)d 7 (T3)d 602 (T3)d

None None Cry1Ab Cry1Ac Cry1B Cry1B Cry1B-1Ab Cry1E

C. part.

B. fusca

C. ori.

E. sac.

S. cal.

13 8 98 85 100 99 99 14

11 18 59 30 47 26 35 10

21 25 81 41 59 82 52 29

2 10 27 8 10 24 24 7

9 5 96 31 19 8 87 11

Numbers are copied from graphs in Odhiambo et al. (2002). More recent data on the secondgeneration public-sector events were unavailable, so these data on the first-generation public-sector events are used for illustrative purposes. aT5 is the fifth generation since transformation and T3 is the third such generation. bSpecies names are abbreviated. C. part., Chilo partellus; B. fusca, Buseola fusca; C. ori., C. orichalcociliellus; E. sac., Eldana sacharina; S. cal., Sesamia calamistis. cEvent 176 is not being considered for introduction into Kenya by the IRMA project, but they used it experimentally. dThese five first-generation events were not continued for use in Kenya after 2001.

The data suggest that Cry1Ab and Cry1B have good efficacy against C. partellus. Events 176, 5207, 5601, 1835 and 7 all caused high mortality to C. partellus. None of the events had high efficacy against B. fusca, although 176 and 5601 caused significant mortality. None of the toxins was consistently effective against B. fusca. Combining Cry1Ab and Cry1B together did not appear to improve efficacy against either of the two species. Cry1E appears to be uniformly ineffective against these maize stemborers. Rarely did mortality exceed that of both controls for any of the species. Efficacy against C. orichalcociliellus was high for events 176 and 1835, but the toxins were not consistently effective across the various events. E. saccharina was not strongly affected by any of the events. High mortality in S. calamistis was associated with the two events with Cry1Ab, 176 and 7. In conclusion, Cry1Ab and Cry1Ba may be effective against C. partellus, but additional development work is needed to identify a toxin that is effective against B. fusca. Although there has been some additional testing associated with the second-generation public-sector events, we were unable to obtain the results for inclusion in this chapter. (LC50 ASSAYS). These methods can be conducted to determine the appropriate Cry toxins to be incorporated into the target plant and to evaluate the efficacy of potential transgene products. These bioassays should use a standardized methodology, usually resulting in <10% control mortality that simulates (as close as possible) the expression of the transgene in the target plant and the insect life stage primarily affected. (Note: This methodology can also be used in part or in its entirety in characterizing the

LABORATORY BIOASSAYS

Environ. Risk. Chap 04

23/8/04

9:32

Page 108

108

D.A. Andow et al.

function of a known Bt transgene or in association with subsequent monitoring for resistance in the target pest.) Variables to consider should include: 1. Use of a protein of similar size, structure and solubility as occurs in the plant. In the case of Bt plants that primarily express soluble truncated semiactivated proteins (in descending order of priority): (i) soluble truncated semi-activated proteins (almost impossible to acquire); (ii) purified solubolized and trypsinized (activated) toxin; (iii) purified solubolized protein (protoxin); (iv) purified protoxin inclusion bodies (crystals) containing a single protein; (v) formulated materials containing single Cry proteins. 2. It is also possible to use Bt plant tissue incorporated into artificial diets. 3. Use of appropriate pest life stage. For caterpillar pests, neonates (first instars) and preferably either second and/or third instars. If conducting bioassays with neonates is problematic, preliminary bioassays could be conducted to determine the correlation between neonates and another instar, and thereafter the other instar could be used. 4. Duration of bioassay. Because insects will be continuously feeding on transgenic tissue in most cases, and because Bt proteins typically require more time to kill than traditional pesticides, a relative long exposure time is recommended. A bioassay duration of 5 days would most likely be considered the minimum duration, with 7 days or even longer (depending on the specific insect species) being recommended. 5. Media. If possible, plant material similar to the target crop for transgene expression should be used (there are several methods to keep plant material moist for 5–7 days). However, conventional artificial diet can be used for initial bioassays with the understanding that the results may be different when the candidate protein is expressed in a particular plant. 6. Insect species. Recognizing that most crops are attacked by numerous insect pests, the number of species to bioassay should be restricted to those species that are considered economic pests on the target crop or other crops where either Bt sprays or Bt transgenics are used or being considered. In addition, these methodologies can be adapted for non-target species testing.

Selectable marker and bacterial gene expression in Bt maize Finding 5: Both private-sector events probably contain both plant- and bacterialselectable marker genes. Event 176 expresses a marker gene, but Mon-810 does not express a marker gene. The second-generation public-sector events are reported not to have marker genes.

Available information on marker genes in Bt maize is summarized in Table 4.9. All of the transformations included marker genes in the original constructs, but only a few express marker gene products in the transformed plant. The secondgeneration public-sector events (3–396) used co-transformation so that the marker gene is likely at a separate locus from the target gene. Bacterialselectable marker gene expression is often weak in maize and other plants

Environ. Risk. Chap 04

23/8/04

9:32

Page 109

Transgene Locus Structure and Expression of Bt Maize

109

Table 4.9. Summary of information related to marker gene expression for the Bt maize transformation events potentially available for use in Kenya (see text below). Event 176b Mon-810

1835e 3 6 10 58 93 127 216 223 396

Original marker gene (from Table 4.1)

Marker DNAa

Marker RNAa

Marker product

35S:bar bacterial?:bla 35S?:nptII 35S?:EPSPS 35S?:gox 35S:bar 35S:bar 35S:bar 35S:bar 35S:bar 35S:bar 35S:bar 35S:bar 35S:bar 35S:bar

Present Present Not presentd Not present Not present? Present Not presentf Not presentf Not presentf Not presentf Not presentf Not presentf Not presentf Not presentf Not presentf

Present? Not presentb

Expressed?c Not expressed

Present

No EPSPS product No gox product Expressed No bar productf No bar productf No bar productf No bar productf No bar productf No bar productf No bar productf No bar productf No bar productf

aPellegrischi

et al. (2002) report that ‘9 second generation events showing negative bar and positive cry genes were isolated’ from 12,634 plants of event 1835, 4593 plants of a ubi:cry1Ab event and 2526 plants of a act:cry1Ab event. ? indicates our uncertainty that the bar gene was eliminated versus the bar gene was inactivated. bEvent 176 is not being considered for introduction into Kenya by the IRMA project, but they used it experimentally. cbar product is insufficient to provide herbicide tolerance under field conditions. Southern blot analysis confirmed the presence of the bar gene in all plant tissues, but expressed protein was undetectable in leaves, pollen, roots or kernels of transgenic maize at a detection threshold of 0.2 p.p.m. Promoter of bla is reported to be inactive because it is bacterial, and no products were found with Northern blot analysis (AGBIOS, 2003). dEPSPS did not co-segregate with the target transgene (CFIA, 2001; AGBIOS, 2003). nptII is linked to the target gene, but did not transfer to Bt maize. gox was presumed to be transferred and probably did not co-segregate with the target transgene (AGBIOS, 2003). eThis is a first-generation event with ubi:cry1Ba, that is not being considered for introduction into Kenya, but is included in this table for comparative purposes. The second-generation events 3–127 are also ubi:cry1Ba events. f David Hoisington, Mexico City (2004), personal communication.

because the marker genes have not been modified to express like plant genes. Plant-selectable marker genes, which can be of bacterial origin, have been optimized for plant expression. Lack of marker gene expression could be because the marker DNA has been eliminated, which happened by chance with the EPSPS gene in Mon-810. It is also possible that the gene is present but is not transcribed in to RNA, or the RNA is not translated into protein. It is important to determine where in this pathway marker gene expression is stopped, because this has implications for risk assessment.

Environ. Risk. Chap 04

110

23/8/04

9:32

Page 110

D.A. Andow et al.

Event 176 expresses a bar plant-selectable marker gene, although the concentrations of gene product are low. It will be necessary to consider the potential risks associated with this marker gene during subsequent risk assessment. Mon-810 does not express a marker gene. Southern blot analysis indicated that the genes for glyphosate tolerance (CP4 EPSPS) and antibiotic resistance (nptII) were not transferred into the MON 810 transgene. It was presumed that the CP4 EPSPS and gox genes have been integrated in the initial transformant at loci separate from the cry1Ab gene locus and then subsequently lost through segregation. The absence of the CP4 EPSPS and gox gene products was confirmed by Western blotting. The status of the bar marker in the second-generation public-sector events is uncertain by our analysis. Pellegrischi et al. (2002) recovered nine secondgeneration events without bar from 19,735 plants screened. The scientific evidence using Southern hybridization to show that the DNA is not present should be provided. However, Southern and PCR data are available to demonstrate the absence of the marker DNA (David Hoisington, Mexico City, personal communication), and this is reflected in Table 4.9. If the marker gene is present, but there is no marker product expressed, then it will be necessary to provide some assessment of the likelihood that the marker gene could be expressed sometime in the future. Gene silencing is highly variable and has been shown to alternate from generation to generation (Pawlowski and Somers, 1998). Depending on the mechanism, the marker may be permanently incapable of expression, or expression might occur conditionally. If conditional expression is possible, it will be necessary to consider the potential risks with this marker in subsequent risk assessment. If the marker gene is expressed, then its products would need to be assessed similar to the way target gene expression is assessed.

Pleiotropic and epistatic effects and genotype–environment interaction Finding 6: Epistasis, pleiotropy and gene–environment interaction will need to be evaluated during field testing using whole plant methodologies.

A transgene has pleiotropic effects when it affects a trait other than the trait for which it is targeted. A transgene has epistatic effects when it interacts with the regulation and expression of other genes to affect a trait in addition to the trait for which it is targeted. For example, transgene products could be processed post-transcription or post-translation, or possibly interact with other compounds in the plant, such as known toxins or nutritionally important compounds, or lignin expression in Bt maize (Saxena and Stotzky, 2001). Gene–environment interaction means that the expression of the transgene varies among environments. For Bt maize, it may be necessary to study morphological, agronomic and secondary metabolic changes (e.g. alteration in exudates and nutritional requirements, wound-induced volatiles, yield, aflatoxin proliferation in maize, accumulation of cuticular waxes, metabolic products that could affect Striga germination). However, the most feasible approach to evaluate these

Environ. Risk. Chap 04

23/8/04

9:32

Page 111

Transgene Locus Structure and Expression of Bt Maize

111

possible changes is through multiple field tests to assess phenotypes such as increased susceptibilities or performance differences in the Bt maize lines. These aspects of transgene expression must be studied in the field under multiple environments in the whole plant. Hence, it cannot be evaluated for any of the Bt maize events in Kenya at this time. Methodologies need to be developed to ensure that these issues are appropriately addressed. We expect that some of these methodologies will be risk-based, in that the phenotype will be characterized at the same time a specific risk is being assessed (see discussion on whole plant methodologies in Birch et al., Chapter 5, this volume).

Transgene Transmission Finding 7: The information provided to assess transgene transmission to Kenyan adapted maize lines is insufficient at this time.

All of the Bt maize transgenes were produced in such a way that a new transgene locus (or several loci) is created in the nuclear genome. Given that the transgenes are nuclear, we predict that they would be inherited in a Mendelian fashion. Preliminary experiments (Table 4.2) suggest that several of the transgenes follow Mendelian ratios. This work needs to be extended for perhaps more than five generations. During the creation of hybrid varieties, usually only one of the inbred parents will carry the transgene, and in this parent, it is usually homozygous (i.e. the plant has two copies of the transgene). Thus, hybrids are generally hemizygous for the transgene (Johnston et al., Chapter 6, this volume). However, it is also possible to produce homozygous hybrids in which both parents are homozygous for the transgene. OPVs, on the other hand, may be produced as homozygotes or a mixture of hemizygous and homozygous individuals. It will be necessary to determine the percentage of plants in a finished variety (hybrid or OPV) not expressing Bt genes. Lack of expression can occur if not all of the seeds were produced with the intended parents or if there are other sources of contamination of the seed source. This is important in assessing the possibility of more rapid development of pest resistance.

Discussion Structured assessment of transgene Our structured approach to transgene characterization can be used for capacity building workshops. We divided the characterization problem into four parts: transgene design, genotypic characterization of the transgene locus, phenotypic characterization of transgene expression and transgene transmission. In each part, we focused on issues necessary for conducting an environmental risk

Environ. Risk. Chap 04

23/8/04

9:32

Page 112

112

D.A. Andow et al.

assessment of Bt maize. However, many of the approaches we have raised here will need to be used in a human and animal health risk assessment as well. This book, however, does not address these health issues. Much of what we have suggested is already required by some regulatory authorities around the world, but by structuring the process, we provide a clearer scientific context and justify the necessity of the work in a scientific and transparent manner. Design of Bt maize We have made several recommendations for improved design of Bt maize. These recommendations represent suggestions that would reduce the risk of resistance evolution or reduce the scope of risk assessment. A thorough analysis of how transgene design could affect risk assessment is important to consider in the future. The recommendations represent an ideal for transgenes that may be difficult to achieve at this time. Hence, they should not be considered requirements, but by working toward them, it will be possible to eliminate some of the work that would have been necessary for environmental risk assessment. Transgene locus sequencing We advocate sequencing the transgene locus and flanking genomic sequences, while recognizing that this is not an international regulatory standard. Some regulatory authorities appear to require this information (e.g. Office of the Gene Technology Regulator in Australia), but others do not (e.g. USDA Animal and Plant Health Inspection Service in the USA). We recognize that sequencing transgene loci has in recent years become relatively routine, and that it is likely that transgene loci in plants destined for commercialization are now sequenced. Hence, we believe that such a requirement would not be burdensome and would provide substantial benefits, as we describe in the section on unpredicted and unexpected effects below. Whole plant methods Many of the methods for characterizing Bt maize transgenes are laboratory based and involve extractions from Bt maize plants (e.g. molecular characterization of transgene products; development of a molecular probe to monitor transgene presence). All of these tests can be done prior to field release. However, several aspects of transgene expression must be done on whole plants. For example, the quantification of transgene product in plant parts must be done on whole plants, and any evaluation of gene–environment interaction must be done in the field. Indeed, some aspects of transgene expression assessment will need to be done as a part of the risk assessment process (e.g. Birch et al., Chapter 5, this volume). These considerations imply that the characterization of the transgene does not need to be done all at the same time. Indeed, they can be staged through the development of the genetically modified organism (GMO). In this chapter, we have not addressed the staging of the risk assessment process directly, but believe that this is a critical issue that needs to be developed in the future.

Environ. Risk. Chap 04

23/8/04

9:32

Page 113

Transgene Locus Structure and Expression of Bt Maize

113

For Bt maize in Kenya, we noted that there is a major difference in the development stage of the public-sector Bt events and the private-sector events. The private-sector events already have been commercialized successfully in other parts of the world. As of 2001, the public-sector events still need to be improved by reducing transgene locus number and selected for simple transgene structures. These are steps that need to be completed before substantial efforts are invested in characterizing transgene locus structure and expression as described here. Efficacy and status of Bt maize Bt maize is aimed to control maize stemborers in Kenya. The major stemborer pest species in Kenya are C. partellus and B. fusca (Muhammad and Underwood, Chapter 2, this volume). Only preliminary testing using leaf tissue assays of the efficacy of Bt maize against these stemborers was completed by the end of 2001 (Table 4.10). These preliminary results suggest that two different Cry toxins may be effective against C. partellus, but none of them show sufficient control of B. fusca. If effective control of B. fusca cannot be discovered, the potential importance of Bt maize in Kenya will be limited to those locations and times when C. partellus is the only significant stemborer pest. As indicated

Table 4.10. Expectations and predictions for identifying potential unexpected or unpredicted effects based on the transgene genotype. Expectation or prediction

Location in chapter

1. 2.

Table 4.5, Target gene expression Table 4.9, Marker gene expression

Transgene will express the target gene product Transgene will or will not express a marker gene product 3. Nuclear transgenes should follow Mendelian inheritance 4. All transgene protein products can be predicted from the transgene locus sequence 5. The likelihood of homologous recombination can be predicted from the transgene locus sequence 6. The likelihood of ectopic expression can be predicted from the transgene locus sequence 7. Disturbance to native genes can be determined from the sequence of regions flanking the transgene 8. The hypothetical possibility that a transgene can jump to another genomic location (be activated by a transposon) can be evaluated from the sequence of regions flanking the transgene 9. The target gene promoter leads to predictions about the tissue and ontogenic patterns of expression of the target gene product 10. The mechanism of suppression of marker gene expression enables predictions of ways that suppression might fail

Table 4.2, Transmission Table 4.3 and 4.6, Sequence of transgene Table 4.3, Sequence of transgene Table 4.3, Sequence of transgene Table 4.3, Sequence of transgene Table 4.3, Sequence of transgene

Table 4.5 and 4.7, Target gene promoter Table 4.9, Marker gene expression

Environ. Risk. Chap 04

23/8/04

9:32

Page 114

114

D.A. Andow et al.

in Chapter 2, these are the lowland tropical, and mid-altitude areas, where there are mainly small-scale farmers using OPVs, and the lower elevations of the transitional areas, where the majority of farmers use hybrid seed.

Unexpected and unpredicted effects One of the most troublesome and controversial aspects of GMOs is the concern that they will produce unexpected or unpredicted effects. Unexpected and unpredicted effects can be addressed by specifying the expectations and predictions in a transparent scientific way. Any observation that does not fit expectations or contradicts predictions would be an unexpected or unpredicted effect. We have proposed the ten expectations and predictions that will address any possibility that originates from the transgene structure itself (Table 4.10). These expectations and predictions do not address all aspects related to environmental risk. For example, there may be unexpected or unpredicted affects associated with RNA products from the transgene or metabolic products of the transgene product, such as glycosylation of transgene protein product or other post-translational processing of the product. Moreover, as mentioned previously, epistasis, pleiotropy and gene–environment interaction, which are all potential sources of unexpected or unpredicted effects, are not taken into account in Table 4.10. There are over 240 different possible pathways that a single transgene product and its direct metabolites could affect a non-target predatory species, and if gene–gene interaction is allowed, there may be over 1106 different pathways by which a transgene product could affect a nontarget predator. It will probably be impossible to predict all of these interactions. Instead, we have suggested that these potential effects be integrated directly into the risk assessment process, by linking their characterization with experiments to evaluate specific components of environmental risk. This idea is elaborated in Birch et al., Chapter 5, this volume.

References AGBIOS (2003) Essential Biosafety, 2nd edn. Merrickville, Ontario. www.essentialbiosafety.com (accessed 25 November 2003). CBD (Secretariat of the Convention on Biological Diversity) (2000) Cartagena Protocol on Biosafety to the Convention on Biological Diversity: text and annexes. Montreal: Secretariat of the Convention on Biological Diversity. www.biodiv.org/ doc/ legal/cartagena-protocol-en-pdf (accessed 25 November 2003). CFIA (Canadian Food Inspection Agency) (2001) Regulatory Directive Dir94-08: Assessing Criteria for Determining Environmental Safety of Plants with Novel Traits. Plant Health and Production Division, Plant Biosafety Office, Montreal. EPA (US Environmental Protection Agency) (2000a) Biopesticide Fact Sheet Bacillus thuringiensis Cry1Ab delta-endotoxin and the genetic material necessary for its production (Plasmid vector pCIB4431) in corn [Event 176] (006458). US Environmental Protection Agency, Washington, DC, EPA 730-F-00-003.

Environ. Risk. Chap 04

23/8/04

9:32

Page 115

Transgene Locus Structure and Expression of Bt Maize

115

EPA (2000b) Biopesticide Fact Sheet Bacillus thuringiensis Cry1Ab delta-endotoxin and the genetic material necessary for its production in corn [MON 810] (006430). US Environmental Protection Agency, Washington, DC, EPA 730-F-00-006. Estruch, J.J., Kadwell, S., Merlin, E. and Crossland, L. (1994) Cloning and characterization of a maize pollen-specific calcium-dependent calmodulinindependent protein kinase. Proceedings of the National Academy of Sciences USA 91, 8837–8841. Hudspeth, R.L. and Grula, J.W. (1989) Structure and expression of the maize gene encoding the phosphoenolpyruvate carboxylase isozyme involved in C4 photosynthesis. Plant Molecular Biology 12, 579–589. IRMA (2001) Insect Resistant Maize for Africa Annual Report 2000 KARI/CIMMYT IRMA Project. IRMA Project Document No. 4. KARI and CIMMYT (Kenya Agricultural Research Institute and International Maize and Wheat Improvement Centre), Nairobi. IRMA (2002) Insect Resistant Maize for Africa Annual Report 2001 KARI/CIMMYT IRMA Project. IRMA Project Document No. 6. KARI and CIMMYT (Kenya Agricultural Research Institute and International Maize and Wheat Improvement Centre), Nairobi. Koziel, M.G., Beland, G.L., Bowman, C., Carozzi, N.B., Crenshaw, R., Crossland, L., Dawson, J., Desai, N., Hill, M., Kadwell, S., Launis, K., Lewis, K., Maddox, D., McPherson, K., Meghji, M.R., Merlin, E., Rhodes, R., Warren, G.W., Wright, M. and Evola, S.V. (1993) Field performance of elite transgenic maize plants expressing an insecticidal protein derived from Bacillus thuringiensis. Bio/Technology 11, 194–200. Mani, G.S. (1985) Evolution of resistance in the presence of two insecticides. Genetics 109, 761–783. Monsanto Company (2002) Safety Assessment of Bollgard® Cotton Event 531, www.monsanto.com/monsanto/content/our_pledge/ (accessed 25 November 2003). Mugo, S., Gethi, M., Songa, J., Odongo, O., Ombakho, G., Gethi, J., Njoroge, K. and Ininda, J. (2002) Develop locally-adapted insect resistant maize germplasm. In: IRMA (eds) Insect Resistant Maize for Africa Annual Report 2001. KARI/CIMMYT IRMA Project. IRMA Project Document No. 6. KARI and CIMMYT (Kenya Agricultural Research Institute and International Maize and Wheat Improvement Centre), Nairobi, p. 7. Odhiambo, B., Mugo, S., Taracha, C., Hoisington, D., Songa, J., Bergvinson, D. and McLean, S. (2002) Introduction of Bt maize leaves and insect bioassays to identify Cry proteins effective against Kenyan stem borers. In: IRMA (eds) Insect Resistant Maize for Africa Annual Report 2001. KARI/CIMMYT IRMA Project. IRMA Project Document No. 6. KARI and CIMMYT (Kenya Agricultural Research Institute and International Maize and Wheat Improvement Centre), Nairobi, pp. 2–3. Pawlowski, W.P. and Somers, D.A. (1998) Transgenic DNA integrated into the oat genome is frequently interspersed by host DNA. Proceedings of the National Academy of Sciences USA 95, 12106–12110. Pellegrischi, A., Odhiambo, B., Hoisington, D., Mugo, S. and McLean, S. (2002) Genetic engineering activities. In: IRMA (eds) Insect Resistant Maize for Africa Annual Report 2001. KARI/CIMMYT IRMA Project. IRMA Project Document No. 6. KARI and CIMMYT (Kenya Agricultural Research Institute and International Maize and Wheat Improvement Centre), Nairobi, pp. 1–2. Ritchie, S.W. and Hanway, J.J. (1984) How a Corn Plant Develops. Iowa State University Cooperative Extension Service Special Report 48, Rev. 1984. Ames, Iowa.

Environ. Risk. Chap 04

116

23/8/04

9:32

Page 116

D.A. Andow et al.

Saxena, D. and Stotzky, G. (2001) Bt corn has higher lignin content than non-Bt corn. American Journal of Botany 88, 1704–1706. Svitashev, S., Ananiev, E., Pawlowski, W.P. and Somers, D.A. (2000) Association of transgene integration sites with chromosome rearrangements in hexaploid oat. Theoretical and Applied Genetics 100, 872–880. Windels, P., Taverniers, I., Depicker, A., Van Bockstaele, E. and De Loose, M. (2001) Characterisation of the Roundup Ready soybean insert. European Food Research and Technology 213, 107–112.

Environ. Risk. Chap 05

5

10/9/04

14:28

Page 117

Biodiversity and Non-target Impacts: a Case Study of Bt Maize in Kenya A.N.E. BIRCH, R. WHEATLEY, B. ANYANGO, S. ARPAIA, D. CAPALBO, E. GETU DEGAGA, E. FONTES, P. KALAMA, E. LELMEN, G. LØVEI, I.S. MELO, F. MUYEKHO, A. NGISONG, D. OCHIENO, J. OGWANG, R. PITELLI, T. SCHULER, M. SÉTAMOU, S. SITHANANTHAM, J. SMITH, N. VAN SON, J. SONGA, E. SUJII, T.Q. TAN, F.-H. WAN AND A. HILBECK Corresponding authors: Dr A. Nick E. Birch, E-mail: [email protected], and Dr R. Wheatley (soil ecosystems), E-mail: [email protected], Scottish Crop Research Institute, Invergowrie, Dundee DD2 5DA, UK.

Non-target Environmental Risk Assessment Model Biodiversity is recognized in the Convention on Biological Diversity (CBD, 1992) as having multiple values, including a critical role for meeting the food, health and other needs of the growing world population. The ecological value of biodiversity can be related to ecosystem functions that are vital for crop production in sustainable agricultural systems. For example, species assemblages in an agroecosystem fulfil a variety of ecosystem functions, and a change in these assemblages can possibly harm the agroecosystem, including the farmer. The environmental risk assessment model described here assesses the possible risks of transgenic crops on biodiversity by selecting species from these assemblages, identifying the potential for risk, and proposing research protocols to assess these risks. Any given cropping system will typically contain about 1000 or a few thousand species. Although it is possible to assess impacts on this biodiversity in its entirety, the pre-release assessment of transgenic crops will be in closed, controlled environments, such as the laboratory, greenhouse or small-scale field, which require selection of a relatively small number of species or species groups. Therefore, an important component of a case-specific risk assessment is that the most relevant species are selected for pre-release testing in a scientifically defensible and transparent way. A species-based approach can be accomplished for above-ground ecosystems, but below-ground ecosystems require a different approach. Soils are complex biological, chemical and physical environments, containing a large © CAB International 2004. Environmental Risk Assessment of Genetically Modified Organisms: Vol. 1. A Case Study of Bt Maize in Kenya (eds A. Hilbeck and D.A. Andow)

117

Environ. Risk. Chap 05

23/8/04

9:32

Page 118

118

A.N.E. Birch et al.

number of species. Soil microbial and macro-faunal communities are extremely complex and largely undescribed. Hence, any individual species may be nonrepresentative or poorly connected to significant ecological functions, and individual species are unlikely to be predictive of ecological impact. Instead, the approach to below-ground biodiversity taken here is based on aggregate soil ecosystem functions. The non-target environmental risk assessment model developed here involves five steps: Step 1: categorizing and listing potential non-target species and ecological functions, and identifying important interactions; Step 2: prioritizing species or functions for pre-release testing according to maximum potential exposure and potential adverse effect; Step 3: conducting exposure pathway analyses; Step 4: describing hazard scenarios and formulating testing hypotheses; and Step 5: developing ecologically meaningful testing methods and protocols. Each of these steps will be described in more detail and then applied to the Kenya case study.

Step 1: Functional groups and categorizing non-target organisms and functions A combination of qualitative field expertise and data from biodiversity assessments is crucial for determining the list of possible non-target species, their trophic relationships and relevant functions. The list will be specific to the crop and its cropping context and agroecosystem. For above-ground ecosystems: (i) functional groups are established; and (ii) the identified species are classified into these functional groups. For below-ground ecosystems: (i) is carried out generally, and (i) and (ii) can be carried out for macroorganisms where there is enough species information. Establishing functional groups Using ecological function allows one to focus on ecological processes and limit the number of species and functions tested. Two types of functional criteria can be used (Table 5.1) – anthropocentric functions and ecological functions (Andow and Hilbeck, 2004). Anthropocentric functions are related directly to human goals, and include secondary pest species, natural enemies, rare or endangered species, species used to generate income, and species of social or cultural value. Natural enemies are the organisms in agroecosystems (predators, parasitoids and parasites) that help maintain pest species at stable and often low levels, a function known as biocontrol. Ecological functions relate to ecosystem processes independent of human goals, and include primary consumption, secondary consumption, pollination, decomposition, nutrient recycling and seed dispersal. These functional groups are not mutually exclusive. For example, many species are both secondary pests and non-target primary consumers. Others are both natural enemies and secondary consumers.

Species that consume plant residues

Decomposers

Nutrient cyclers Seed dispersers Species of unknown function

Species that eat primary consumers Species that visit flowers and carry pollen between them

Non-target primary consumers Secondary consumers Pollinators

Ecological functions Competitors

Species of conservation concern Species that generate income Species of social or cultural value

Honeybees, silk moths, fungi Monarch butterflies, Morpho butterflies or honeybees

Predators, parasitoids, parasites, pathogens, weed-eating herbivores

Sporadic pests (e.g. locusts), minor pests (e.g. leafhoppers)

Predators, parasitoids, parasites Social and solitary bees, flies, beetles, ants Ants, collembolans, bacteria, fungi, nematodes, earthworms, mites Fungi, bacteria Birds, small mammals, ants Nearly half the arthropod species in a habitat

9:32

Species that perform functions that benefit the functioning of the ecosystem Species that compete for environmental resources with the crop Weeds (light, water, nutrients) Plant-consuming species that are not the target of the transgene

Functions that are of value (negative or positive) to humans Species that cause relatively minor damage because their abundance is restrained by the primary pests or other external factors (e.g. pesticides), but that can cause significant damage if these factors change or the primary pests are reduced Species that consume/kill or damage pests and competitors of the crop Rare or endangered species (e.g. IUCN Red List) or species of pre-determined value for biodiversity conservation

Anthropocentric functions Alternate or secondary pests

Examples

23/8/04

Natural enemies

Definition

Functional group

Table 5.1. Functional classification of terrestrial non-target organisms in or near agricultural systems for pre-release testing of transgenic plants (Andow and Hilbeck, 2004; see also Box 5.1).

Environ. Risk. Chap 05 Page 119

Biodiversity and Non-target Impacts 119

Environ. Risk. Chap 05

23/8/04

9:32

Page 120

120

A.N.E. Birch et al.

Classification of non-target species The non-target species relevant to the agroecosystem of the crop are classified into the functional groups using the information and expertise available. However, a vast number of species found in agricultural fields probably cannot be classified into any one of these functional groups. Hence, it is critical to consider a category of species with unknown function, so that these species are not inadvertently overlooked because of lack of knowledge.

Step 2: Prioritizing non-target species or functions This part of the process involves: (i) ranking the non-target species or functions according to ecological principles; and (ii) prioritizing a number of these for possible assessment, with particular emphasis on those that might be adversely affected by the transgenic crop and are significant for ecological or anthropogenic reasons. Prioritization using ecological principles In order to provide a rational and transparent approach to support the choice of species or functions, we developed a series of selection matrices. Each species is ranked for its maximum potential exposure (occurrence, abundance, presence, linkage), and for potential adverse effect (significance, such as potential secondary pest, disease vector) (see Box 5.1 for details). This will often differ in each agroecological zone where the crop is grown. Species, groups or functions given the highest priority (rank 1) are therefore the ones that have high maximum potential exposure (they are very abundant, present in the agroecosystem every year throughout the growing season and closely linked to the crop as host plant) and the ones that have a vital role in ecosystem functioning. In the workshop, species were only assigned to three ranks to simplify the ranking procedure because only rank 1 (highest priority) species or functions are likely to be considered for assessment. This approach overcomes the simplistic assumption of species abundance as a direct measure of ecological significance, as the final rank results from the combination of factors. It is important not to exclude species on the basis of only one criterion. Although many species have an unknown or uncertain ecological function or significance, our collective lack of knowledge does not imply that they have an insignificant ecological role. For example, the ecological and functional significance of microbial symbionts is only beginning to be appreciated (e.g. Wolbachia; Werren, 1997). To ensure a precautionary approach, we suggest that species with a high standing biomass or that are found in frequent association with the transgenic crop habitat should be prioritized for testing even if their significance is unknown. Selecting the high-priority categories to be tested Species and functions that were assigned to the highest priority become candidates for testing. The final selection process is an expert-driven process, but is transparent. It is possible to advocate that several species from each functional

Environ. Risk. Chap 05

23/8/04

9:32

Page 121

Biodiversity and Non-target Impacts

121

Box 5.1. Criteria for prioritizing non-target species in each functional group to facilitate selection of species to evaluate in stage one tests; these criteria are consistent with Annex III of the Cartagena Protocol (CBD, 2002). 1. Maximum potential exposure – This is based on geographic range, habitat specificity, local abundance (Rabinowitz, 1981), prevalence (proportion of suitable habitat that is occupied by the species) and temporal association with the crop. In the prioritization support matrices (Tables 5.4, 5.5 and 5.6), these are referred to as occurrence (occurs in crop agroecosystem, geographic range and prevalence), abundance (local abundance and prevalence), presence (temporal association with the crop) and linkage (habitat specificity, association with maize). These criteria can be evaluated independently of the specific transgenic crop. Species with a broad geographic range, specificity to the crop habitat, high local abundance, high prevalence and high temporal overlap with the crop are likely to have greater exposure to the transgenic crop. 2. Potential adverse effect – This is based on the potential value of an adverse effect on a non-target species, should one occur, based on ecological and anthropogenic significance (see Table 5.1). In the prioritization support matrices (Tables 5.4, 5.5 and 5.6), this is referred to as significance, and in the primary consumers matrix, particular significance as disease vector or damaging pest is separately listed. Ecologically significant species fulfil significant ecological functions, such as biological control, pollination or decomposition. Economically significant species are likely to have an economic impact if their abundance changes. Examples include disease vectors or damaging pests. Threatened species include those listed on Red and Blue lists or who are otherwise threatened or endangered. Species of cultural significance could be symbolic species that appear repeatedly in public in symbolic ways (flags, logos, advertisements, news, etc.), or species with unique attributes (e.g. social organization, mass migration, stunning and rare beauty, strength). 3. Potential likely exposure – Species likely to be exposed to the transgene product or metabolites in the crop ecosystem. This assessment must take into account the specific transgenic crop and expression levels of the transgene product in each tissue. Species that are not exposed directly or indirectly are less likely to be affected by the transgenic crop, and if they are affected it will probably be through another species that is directly exposed to the transgene product or metabolite. See text for additional discussion.

group should be tested. It is also possible to advocate that single species, or one or several functional groups from a subset of the functional groups should be tested. Clearly, the greater the level of precaution required in the assessment, the larger the choice of species or functions should be. However, the size of the species list is also likely to be influenced by other factors, including economic and political ones. If the decision making is transparent and based on the consensus of diverse experts, the decisions will be defendable and likely to be acceptable to many.

Step 3: Exposure pathway analyses This step analyses possible causal pathways of exposure to the genetically modified (GM) plant and toxin, and potential impacts of the GM plant

Environ. Risk. Chap 05

122

23/8/04

9:32

Page 122

A.N.E. Birch et al.

(including direct and indirect, intended and unintended effects) for each species or function identified as highest priority in the previous step. The purpose of this evaluation is to differentiate candidate test species likely and unlikely to be exposed to the Bt toxin, and for the former, to guide the design of the exposure system in the test protocols. Potential likely exposure can occur through many pathways. Transgenic plant material and transgene products and metabolites may affect non-target species and ecosystem functional dynamics directly via plant residues (aboveor below-ground) (Zwahlen et al., 2003), senescent leaves and sloughed roots, root exudates (Saxena et al., 1999; Saxena and Stotzky, 2000), pollen (Losey et al., 1999), and other plant parts that express the transgene, such as seeds, floral and extra-floral nectaries, guttation fluids and phloem sap (Hilbeck, 2002). Any non-target organism feeding on the transgenic plant or parts of the plant may come in contact with the transgene and its product. In addition, the transgene product might interact with existing plant compounds to affect non-target organisms (Birch et al., 2002). Transgenic plant material and transgene products can affect non-target species indirectly through another organism, such as an herbivore (Birch et al., 1999; Hilbeck et al., 1999) or honeydew from Homopteran species such as aphids, scales or whiteflies (Raps et al., 2001; Bernal et al., 2002). Non-target species could therefore be affected by: (i) transgene products in the plant, plant secretions, herbivore, herbivore excretions or other species containing transgene products; (ii) metabolites of the transgene products; or (iii) interaction effects of the transgene products with other plant or herbivore compounds that alter plant or herbivore composition or physiology (e.g. Saxena and Stotzky, 2001a; Birch et al., 2002). The number of possible pathways is immense. It has been estimated that there are over 250 different exposure pathways by which a transgene product or its metabolites could affect a secondary consumer, of which only a few are direct effects of the transgene product (Andow and Hilbeck, 2004). These include unintended changes in ecologically important plant primary and secondary metabolites in the transgenic plant. This multitude of potential exposure pathways has important implications for test methodology (see below), and complicates analysis of potential exposure (Hilbeck, 2002). Soil ecosystems are driven by the types and amounts of carboncontaining compounds entering soils, providing both energy and nutrients (Wheatley et al., 1990, 1991, 2001). These inputs are from plant residues and organic compounds released by the roots of growing plants. The constituents of these inputs vary according to plant physiology and structure, and the stage of plant growth. The forms and availability of these compounds can define both rate and choice of microbial function (Wheatley et al., 2001). Thus, soil ecosystem dynamics can be influenced by the type of plant driving it, and the impact of any plant cultivar must be evaluated on the basis of changes in these plant inputs. To guide the analysis of possible exposure or indirect effects we developed a series of questions that will be detailed in Step 3 of the Kenya case study non-target assessment further in this chapter.

Environ. Risk. Chap 05

23/8/04

9:32

Page 123

Biodiversity and Non-target Impacts

123

Step 4: Hazard identification and hypothesis development The information from the previous steps guides hazard identification and the development of testable assessment hypotheses. For example, will a particular non-target herbivore species be affected if it ingests a GM product via flower feeding (i.e. direct effect)? If it is not lethally affected, will it pass the GM product on to higher trophic levels, possibly affecting any of its natural enemies (i.e. tri- or multi-trophic effect)? Will a natural enemy species providing an important ecological service elsewhere decline in numbers because a main non-target herbivore prey source is adversely affected? For the soil ecosystem, hazard identification needs to address the possible effects of the transgene products and any other changes in plant inputs. This step in the risk assessment is particularly important when exposure (direct or multi-trophic) is unlikely yet indirect effects are still possible. Such indirect effects can arise through changes in the population dynamics of a nontarget organism and/or through a shift in the species composition of a particular community assemblage. This step requires careful examination of known impact pathways on species and functional interactions, and relies a great deal on the ecological field expertise available to the process. Functional dynamics in plant–soil ecosystems have been shown to be highly variable on both a temporal and spatial basis, and affected by many factors, such as temperature, rainfall and cultivation practices. Consequently, comprehensive baseline data will be required to determine natural variation within comparable soil ecosystems to be able to detect any effects of transgenic crops.

Step 5: Protocols and measurement endpoints (parameters to be measured) The next step is to develop appropriate methodologies and protocols to assess risk. Such methods can be designed once it is evident which life stage(s) of a non-target organism is likely to be exposed and through what route it is likely to receive the GM product, directly and/or mediated indirectly, through other nontarget prey. Similarly, when and where soil-ecosystem functions may be vulnerable can be used to design better assessment methods. Any ecologically meaningful experimental design will mimic identified exposure routes, test identified hazard scenarios and introduce ecological realism as far as possible. Methodology/protocol Two kinds of methodology are necessary. Firstly, conventional ecotoxicology methodologies can be modified to assess effects of exposure to the transgene products, although the majority of exposure pathways will be ignored, which creates ambiguity in the interpretation of the results (Hilbeck, 2002; Andow and Hilbeck, 2004). Secondly, a ‘whole plant’ methodology is required. In such a method, the effects of the whole transgenic plant are evaluated, not just the transgene product. Such methods require the use of appropriate experimental controls that mimic natural exposure, as it would occur in the field.

Environ. Risk. Chap 05

23/8/04

9:32

Page 124

124

A.N.E. Birch et al.

Appropriate genetic and ecological controls An ideal genetic control is a plant verified as being fully isogenic to the transgenic plant. Isogenic means that the control is genetically identical to the transgenic plant except for the inserted transgene. In practice, isogenic controls cannot be used as such isogenic plants do not exist for commercial or commercializable varieties, or may not be truly isogenic. Near-isogenic controls are available for some commercial or commercializable crop varieties, but these may differ from the transgenic variety by as much as 4% of the genome. Moreover, all such transgenic varieties have had some selection for agronomic characteristics that their near-isogenic varieties have not experienced, which can result in genetic and phenotypic differences from their near-isogenic lines. Consequently, there can be systematic differences in agronomic characters between the transgenic plant and its near-isogenic control plants. One way to address the lack of rigorous genetic controls is to run multiple comparisons between several pairs of transgenic varieties or events and their isogenic or agronomic controls. Agronomic controls would consist of locally grown varieties. If these pairs of varieties are sufficiently different from each other, then they will be less likely to share many genetic differences except for the difference in the transgene. Hence, running the same experiment on several transgenic control pairs (so long as multiple transgenic lines containing the same transgene exist), under similar environmental conditions, allows more definitive conclusions about the effects of a transgene. An ideal ecological control is a plant variety that would be grown in a production system in the region of interest (Andow and Hilbeck, 2004). This does not have to be a commonly grown variety. However, a plant variety that has not been screened for performance in a local production system would not be a useful ecological control. In addition, it is crucial that the plant is presented to the test species in a way that mimics the way the species would experience the plant in the field. In many cases, these studies will be done in the laboratory, using greenhouse-grown plants, and laboratory-reared herbivores and natural enemies. If appropriate care is not taken, the plants could be etiolated (elongated) with low specific leaf weights, have atypical primary and secondary plant metabolism, and the test species could be inbred, physiologically stressed, physiologically variable or even diseased. Metabolism of excised plant tissues quickly changes radically, so laboratory bioassays using excised plant material should be short in duration (c. 24–48 h maximum) and do not mimic effects of the growing GM crop in the field. Conducting lab tests and demonstrating no negative effects on natural enemies (particularly using artificial diets and purified toxins) should not be used as evidence that no further testing is needed and that the transgene product and GM crop is safe for natural enemies under field conditions. Other types of controls (for testing Bt maize) are strongly recommended. These include any available conventional pest-resistant varieties, a conventional maize variety with Bt sprays and a conventional variety protected with a commonly used pesticide regime for the region.

Environ. Risk. Chap 05

23/8/04

9:32

Page 125

Biodiversity and Non-target Impacts

125

Measurement endpoint An appropriate experimental endpoint (parameter to be measured) for initial testing is generational relative fitness or some component of relative fitness. Generational relative fitness is the relative lifetime survival and reproduction of the non-target species. Hence, survival experiments should last at least through all of the developmental stages of the non-target species, and adult life stage parameters should be measured, including age-specific mortality and female fecundity. In principle, the duration of the test should correspond to the time the non-target species would be exposed to the transgenic plant, plant parts and residues, and the temporal pattern of expression and persistence of the transgene product and its metabolites. Generational relative fitness is a particularly useful endpoint, because it relates directly to risk. If the transgenic plant were to adversely affect a non-target species in the environment, its effects would come through some component of relative fitness. Hence, the results from such initial testing would guide the design of further tests, by identifying the fitness components that would possibly be affected by the transgenic plant in the environment. Although a general testing methodology cannot be specified in detail, an ecologically realistic experiment should meet several key criteria so that results are scientifically sound and ecologically interpretable (Andow and Hilbeck, 2004). Initial testing experiments minimally would include: 1. Food (e.g. ecologically relevant plant and or prey species) that is used by test species in their relevant habitat should be used in laboratory tests. If transgene product is used, it should be identical to what is produced in the transgenic plant. 2. Verification that the food offered to the species actually contained the administered material at the intended concentration or dose throughout the investigation. 3. Verification that all life stages of the species are exposed appropriately to the transgene product and actually contact the product in relevant ways. 4. Either use intact plants or plant parts in the experimental system with verification that the plant parts used contain the transgene product, or use the transgene product at concentrations or doses much higher than normally expressed in the plant as a worst-case scenario for short-term exposure. 5. Have a proper scientific control or controls. 6. Have sufficient replication and sufficient numbers of individuals screened, so that statistical power of the experimental design is not an issue for interpretation of results (Marvier, 2001). It is strongly recommended that professional statisticians be consulted before conducting the experiments, as well as for analysis and interpretation.

Scope of the Kenya Case Study Non-target Environmental Risk Assessment This assessment focused on non-target and biodiversity effects associated with maize and nearby habitats. We did not attempt to extend the assessment to

Environ. Risk. Chap 05

23/8/04

9:32

Page 126

126

A.N.E. Birch et al.

cover potential effects associated with aquatic environments, or rare or fragile natural habitats in the landscape, or on intercropped plants mixed with maize (e.g. legumes; Muhammad and Underwood, Chapter 2, this volume). A further focus of these assessments was on initial risk assessment in closed, controlled environments prior to field releases of the transgenic crop. It is expected that our future activities will extend the analyses to cover experimental procedures for field assessments, guided by the data obtained in these initial tests. While the methodologies developed were focused on the Kenyan maize production context, they should extend beyond Kenya and be applicable in similar maize production systems of neighbouring countries (Andow and Hilbeck, Chapter 1, this volume). This will be addressed more formally in our future activities. At the workshop, we focused solely on evaluation of Bt maize, but the methods could be adapted to assess ecological impacts of other transgenic plants. We assumed that taxonomic knowledge in Kenya is incomplete; therefore where individual species are not identified, species should be classified by ecological function, e.g. feeding guilds such as ‘larval parasitoid’ or ‘egg parasitoid’, rather than taxonomically.

Step 1: Functional groups and categorizing non-target species and functions in maize in Kenya Based on the experts present and the available published arthropod surveys conducted in Kenya, Kenyan maize fields contain a diversity of arthropod species from at least 18 orders and 75 families probably comprised of hundreds of species (Table 5.2). This level of information is sufficient to categorize maizeassociated species and functions. The following functional categories were used: non-target maize herbivores, natural enemies of maize herbivores, pollen feeders and pollinators, weeds, and soil functions. We did not have time to consider vertebrates (such as rats, squirrels and bush pigs; Muhammad and Underwood, Chapter 2, this volume), species of conservation concern (such as elephants or monkeys), species of cultural significance, soil macroorganisms (such as Melolonthinae larvae or cutworms; Muhammad and Underwood, Chapter 2, this volume), storage pests that are not found in the field (e.g. Plodia interpunctella, Ephestia cautella or Ephestia kuehniella; Fitt et al., Chapter 7, this volume) or plant pathogens (e.g. Aspergillus flavus, Fusarium; Munkvold et al., 1997; Muhammad and Underwood, Chapter 2, this volume). Non-target maize herbivores Non-target herbivores are potential secondary pests of the crop. While Bt maize may reduce target pest species to insignificant levels, other pests may increase and become damaging pests (secondary pests), decrease or be unchanged. Nontarget herbivores may also function as vectors of diseases that may cause damage to the crop (e.g. leafhoppers/maize streak virus). In addition, they can serve as: 1. Food for biocontrol organisms (natural enemies) that regulate the target pest. 2. Carriers of plant compounds to higher trophic levels possibly inducing tritrophic effects.

Environ. Risk. Chap 05

23/8/04

9:32

Page 127

Biodiversity and Non-target Impacts

127

Table 5.2. Number of individuals of listed arthropod orders and families recovered from pitfall, water and sticky traps, in farmers’ maize fields in Kilifi, Kakamega and Masii during the short rains in 2000 (Josephine Songa, Nairobi, 2002, personal communication). Arthropod order/family Common name Diptera Tachinidae Sarcophagidae Syrphidae Dolichopodidae Stratiomyidae Sciaridae Calliphoridae Muscidae Phoridae Diopsidae Drosophilidae Otitidae Tephritidae Asilidae Rhagionidae Bombyliidae Mycetophilidae Lauxaniidae Agromyzidae Anthomyzidae Sepsidae Orthoptera Gryllidae Blattidae Acrididae Tetrigidae Tettigonidae Gryllacrididae Mantidae Dermaptera Forficulidae Labiidae Hymenoptera Formicidae Apidae Ichneumonidae Vespidae Pompilidae Sphecidae Cephidae Eumenidae Braconidae Chalcididae Megachilidae Tiphiidae

Tachinid flies Flesh flies Hover flies Long-legged flies Soldier flies Dark-winged fungus gnats Blow flies Muscid flies Humpbacked flies Stalk-eyed flies Vinegar flies Picture winged flies Fruit flies Robber flies Snipe flies Bee flies Fungus gnats Lauxaniid flies Leaf miner flies Anthomyzid flies Black scavenger flies

Kilifi

Masii

Kakamega

22 705 9 5 25 – 2 260 – – 1 – 9 1 – 1 – – 5 23 –

1 2,204 11 – 107 3 876 1,194 – 1 10 – 4 – – – – – – 22 –

30 382 26 888 34 7 815 5,518 3 2 158 11 19 1 11 – 78 2 1 70 2

12,187 32 2,812 4 29 4 164

522 11 21 – 3 – 1

10,838 359 123 – 8 – 1

Common earwigs Little earwigs

7 –

5 –

173 –

Ants Honeybees Ichneumons Vespid wasps Spider wasps Sphecid wasps Stem sawflies Potter wasps Braconid wasps Chalcidids Leaf-cutting bees Tiphiid wasps

6,365 23 12 66 18 109 6 5 13 10 3 2

361 29 7 233 9 193 5 – 12 – 25 1

7,996 191 13 157 2 52 26 – 15 1 7 –

Crickets Cockroaches Short-horned grasshoppers Pygmy grasshoppers Long-horned grasshoppers Camel cricket Mantids

Continued

Environ. Risk. Chap 05

23/8/04

9:32

Page 128

128

A.N.E. Birch et al.

Table 5.2. Continued. Arthropod order/family Common name Hymenoptera (Continued) Mutillidae Velvet ants Evaniidae Ensign wasps Ibaliidae Ibaliids Chrysididae Cuckoo wasps Halictidae Halictid bees Coleoptera Coccinellidae Ladybird beetles Carabidae Ground beetles Staphylinidae Rove beetles Tenebrionidae Darkling beetles Melyridae Soft-winged flower beetles Scarabaeidae Scarab beetles Mordellidae Tumbling flower beetles Chrysomelidae Leaf beetles Cerambycidae Long-horned beetles Curculionidae Maize weevil Elateridae Click beetles Lagriidae Long-jointed bark beetles Dasytidae Soft-winged flower beetles Bupestridae Metallic wood boring beetles Meloidae Blister beetles Hemiptera Miridae Plant bugs Cydnidae Burrower bugs Reduviidae Assassin bugs Berytidae Stilt bugs Pyrrhocoridae Stainers Pentatomidae Stink bugs Homoptera Cicadellidae Leafhoppers Cercopidae Spittlebugs Cicadidae Cicadas Membracidae Treehoppers Aphididae Aphids Isoptera Termitidae Termites Rhinotermitidae Damp-wood termites Thysanoptera Thrips Opiliones Harvestmen Lepidoptera Moths and butterfliesa Araneae Spiders Diplopoda Chilopoda Isopopoda Annelida Acari Coleoptera Larvae aMoths

Kilifi

Masii

Kakamega

1 – – 1 14

– 2 8 – –

– – – 23 9

32 1,465 8 2,646 725 159 13 55 3 8 1 4 277 9 17

642 64 24 5 24 190 6 181 4 14 – – 22 2 15

68 90 145 35 779 496 15 134 – 12 6 2 13 1 14

22 10 16 1 333 3

3 5 7 – 2 6

39 3 3 – 1 16

329 2 – 16 –

180 – – 14 –

556 – – 31 60

1 1 – – 534 509 3 1 217 – 1 3

84 – 2 1 267 453 3 1 4 – – 3

25 12 – 3 142 360 – – – 2 – 4

and butterflies could not be identified further as the scales had been removed in the water traps.

Environ. Risk. Chap 05

23/8/04

9:32

Page 129

Biodiversity and Non-target Impacts

129

3. Biocontrol organisms of plants outside and/or inside the cropping system, e.g. when feeding on weeds, thereby helping to suppress their damaging effect. 4. Seed dispersers and seed predators. Non-target herbivores and their natural enemies on beans and other legumes were not considered during the workshop, but were identified as an important intercrop of most maize cropping systems in Kenya (Muhammad and Underwood, Chapter 2, this volume), and should be considered subsequently. The workshop also did not look at potential impacts on species living on weeds or outside the maize field, e.g. on lepidopteran larvae on other plants associated with maize. The non-target herbivores evaluated are listed in Table 5.3. Pollinating and pollen-feeding insects on maize Maize is wind pollinated and produces copious quantities of pollen that is a significant food source for many insects that may be important pollinators of other crops, natural enemies, or otherwise significant. The working subgroup agreed that no systematic observations on flower-visiting species associated with maize have been conducted in Kenya. However, some pollen-feeders known to be present in significant numbers in maize in Kenya were included in Table 5.3. Natural enemies of maize herbivores Natural enemies are beneficial organisms that help reduce pest populations. Many synthetic pesticides have reduced the numbers of natural enemies in crop fields, and it has become apparent that sustainable agricultural production methods need to conserve natural enemies. Hence, new pest control technologies, such as Bt maize, need to be tested for their negative, neutral or positive effects on natural enemies, prior to large-scale use. Some of the parasitoid natural enemies attacking the four target stemborer pest species of Kenyan maize are fairly host specific, such as Dentichasmias busseolae (Heinrich), which attacks Chilo partellus (Swinhoe), but rarely attacks Busseola fusca (Fuller) (Zhou et al., 2003). Other parasitoid species are less host specific. Therefore, the reduction in stemborer prey on Bt maize is likely to lead rapidly to altered natural enemy diversity and abundance, and could affect their biocontrol function on other non-target herbivores. The deliberate introduction of several parasitoid species into Kenya is planned, and it is important to assess how Bt maize will affect them. Sufficient taxonomic and ecological information exists in Kenya to use qualitative ecological expertise to identify and prioritize parasitoids of maize herbivorous arthropods as potential species for non-target risk assessment. The main parasitoid species of stemborers are listed in Table 5.3, based on the expert knowledge available in the group and several recent publications (Bonhof et al., 1997; Songa et al., 2002; Zhou et al., 2003). While the main generalist predators could not be listed, it is understood that they are also an important group that must be considered in a full risk assessment. Investigations by Songa in 2000 in coastal and western Kenya (IRMA, 2001) found potential predator groups to be ants, earwigs (Forficulidae),

Environ. Risk. Chap 05

23/8/04

9:32

Page 130

130

A.N.E. Birch et al.

Table 5.3. The categorized non-target organisms/functional groups included in the analyses of Bt maize in Kenya. Non-target herbivores Natural enemies

Weeds

Below-ground functions

Defoliators Grasshoppers/crickets Spider mites Locusts Spodoptera spp. Sap feeders Aphids Leaf and planthoppers Stinkbugs Grain feeders Sitophilus zeamais (Motsch) Prostephanus truncatus (Horn) Sitotroga cerealella (Oliver) Silk and cob feeders Cryptophlebia leucotreta (Meyrick) Helicoverpa armigera (Hubner) Pollen feeders Meloidae Apis mellifera (L.) Wild bee spp. Ants Butterflies Root feeders Gryllotalpidae Cut worms Termites Melolonthinae Saprovores Mycetophagidae Carpophilus spp. Catatus spp. Tenebrionids Earwigs

Egg parasitoid Grass and sedge Carbon Trichogramma spp. weeds decomposition Trichogrammatoidea spp. Andropogon spp. Cellulose breakdown Telenomus spp. Brachiaria spp. Ammonification Larval parasitoids Cynodon spp. Nitrification Cotesia flavipes (Cameron) Cyperus spp. Denitrification (introduced) Digitaria spp. Nitrogen fixation Cotesia sesamiae (Cameron) Eleusine spp. Phosphorus and Goniozus indicus Eragrostis spp. micronutrient uptake Egg and larval parasitoid Hyparrhenia spp. Chelonus curvimaculatus Melinis spp. (Cameron) Panicum spp. Pupal parasitoid Pennisetum spp. Pediobius furvus Rottboellia spp. (Gahan) Sorghum spp. Dentichasmias Broad-leaf weeds busseolae (Heinrich) Commelina benghalensis (L.) Bidens pilosa (L.) Tagetes minuta (L.) Senna obtusifolia (L.) Irwin & Barneby Parasitic weeds Striga hermonthica (Del.) Benth. Striga asiatica Striga aspera Sorghum spp.

Total: 26

Total: 9

Total: 21

Total: 7

Environ. Risk. Chap 05

23/8/04

9:32

Page 131

Biodiversity and Non-target Impacts

131

spiders, carabid beetles (Carabidae), ladybird beetles (Coccinellidae) and rove beetles (Staphylinidae). Many or most of these are also important predators of Helicoverpa armigera in maize and cotton, and of Spodoptera spp., since several of these predators are polyphagous (van den Berg and Cock, 1993, 1995; Watmough and Kfir, 1995). Other predator species identified to feed readily on H. armigera were various species in the families of Chrysopidae and Anthocoridae (van den Berg et al., 1993). Oloo (1989) found that predation may be a significant mortality factor for stemborer eggs and small larvae. Maize-associated weeds The weed community in Kenyan maize fields is very diverse. Grasses and sedges are important components of biodiversity – about 600 species of grasses alone have been reported in Kenya, suggesting that Kenya might be a centre of origin for many of the grasses found in East Africa (Ibrahim and Kabuye, 1987; Boonman, 1992). Weeds can contribute substantially to crop losses. In particular, the parasitic Striga weeds can be extremely damaging in parts of Kenya, and much scientific and financial effort has gone into seeking methods to control them in Africa (Muhammad and Underwood, Chapter 2, this volume). The weed species considered in the assessment are listed in Table 5.3. Soil ecosystem functions Species diversity in soils is as great as in any other ecosystem (Curtis et al., 2002), so that more than 90% of the biodiversity in agroecosystems is in the soil. A comprehensive assessment of the non-target impacts of GM crops on biodiversity must therefore include some assessment of impacts on soil biodiversity. Soil biodiversity is related to soil-ecosystem functions essential to plant production. Whether nutrients are made available for plant uptake, or lost to the environment, is entirely dependent on microbial functioning in the soil. Soil macroorganisms, such as insects and their larvae, nematodes and earthworms, also play a vital part in soil nutrient cycling by breaking down and redistributing organic material. Key microbial groups involved in nutrient cycling and soil fertility are good indicators of important ecosystem functions, such as nitrogen fixation, mineralization, nitrification or cellulose degradation. These can be rapidly assessed on large numbers of well-replicated soil samples, and are wellestablished parameters of soil health, particularly in low-input agroecosystems. Seven microbial functions were selected as being particularly responsive to the amount and type of plant materials being introduced to the system (Table 5.3). These were: (i) carbon decomposition; (ii) cellulose breakdown; (iii) ammonification; (iv) nitrification; (v) denitrification; (vi) nitrogen fixation; and (vii) phosphorus uptake. Functions (i) to (vi) are within the carbon and nitrogen cycles, driven by plant residues, and are part of the nutrient cycling process vital to continued plant growth; (vi) and (vii) are of particular importance in maize cultivation because of the prominence of Phaseolus bean in the maize cropping system, and because of the intimate relationships of the mycorrhizal association between the plant roots and fungi. Indicator organisms were identified for each functional group.

Environ. Risk. Chap 05

23/8/04

9:32

Page 132

132

A.N.E. Birch et al.

Findings of Step 1 A total of 62 species, species groups and functions in five functional categories (non-target herbivores, pollen feeders and pollinators, natural enemies, weeds and soil functions) relevant to Kenyan maize production systems were compiled for further risk assessment. This list is not comprehensive and a number of categories, guilds and functions were not included, e.g. vertebrates, general predators, pathogens, soil macroorganisms, species of conservation concern and species of cultural or economic significance. It is recognized that these missing components are important, and a full risk assessment of Bt maize in Kenya would have to include these.

Step 2: Prioritizing non-target species or functions To prioritize the listed species and functions, we used a selection procedure based on a prioritization and decision support matrix (selection matrix). This was done separately for non-target herbivores, natural enemies, weeds and soil functions. For soil functions, a modified decision support matrix was developed that was more suitable for the task. The species and functions prioritized as most important for testing were ranked 1. A full risk analysis of Bt maize in Kenya could also include species that received lower rankings. We are not advocating that all the species/functions are tested; the aim of the workshop was to draw up options and identify knowledge gaps for future research effort, scaled and staged to available resources. Non-target maize herbivores The lepidopteran non-target herbivores identified as being most important in maize in Kenya are H. armigera and Spodoptera species (Table 5.4). H. armigera is a minor or sporadic pest of maize, but an important pest of several other crops. It is therefore considered an important potential secondary pest in Bt maize; particularly if the natural enemy population is altered to its benefit (see natural enemies below for a more detailed discussion). Spodoptera species were also ranked as very important because of their abundance in some Kenyan ecological zones and their high damage potential. These species were not discussed in detail; however, the group recommends that they be studied as a key non-target herbivore. Most of the protocols developed for testing of H. armigera can be adjusted readily for testing of Spodoptera spp. Among the non-Lepidoptera, locusts, spider mites, leafhoppers, Carpophilus spp., and the grain feeders Sitophilus zeamais and Prostephanus truncatus were also given highest priority (Table 5.4). Locusts were considered sporadic pests, with outbreaks approximately every 10 years, but with significant damage caused during outbreaks. Spider mites were ranked highly as they may occur in high abundance at any time during the growth cycle of maize, and have a high damage potential. Leafhoppers (Cicadulina spp.) received the highest index of significance for their ability to transmit maize streak virus (MSV) disease to the maize plant, which causes significant crop

Likely Certain if near cotton Certain

Leaf and planthoppers

Stinkbugs

Root feeders

Medium Medium Medium Abundant, site specific

Certain Occasional Occasional Occasional

Medium

Abundant

Abundant

Sometimes Sometimes Sometimes Sometimes Sometimes

Linkage

Vegetative Anytime but more important in maturity

Always Sometimes

Reproductive Always Vegetative Sometimes

Reproductive Always

Reproductive Always

Reproductive Always

Reproductive Always

Vegetative Anytime Anytime Anytime Anytime

Presence

Potential pest

Significant potential pest, also postharvest Significant potential pest, also postharvest Potential pest, also postharvest Significant potential pest Significant pot pest

Potential pest

Potential pest

Significance

Maize dwarf mosaic virus Maize streak virus (MSV) MSV

Disease vector

2 2

1 3

2

2

1

1

3

1

3 1 1 1 2

Rank

Continued

Low Heavy Heavy Heavy Sometimes

Damage

Biodiversity and Non-target Impacts

Melolonthinae (e.g. Schizonycha, Phyllophaga)

Helicoverpa armigera Gryllotalpidae (mole crickets) Cut worms Termites

Certain

Likely

Sitotroga cerealella

Silk and cobs Cryptophlebia leucotreta

Certain

Prostephanus truncatus

Abundant

Low

Medium

Low Abundant Abundant Abundant Medium

Abundance

Possible adverse effect

9:32

Grain feeders Sitophilus zeamais

Likely Occasional Occasional Occasional Certain

Occurrence

Grasshoppers/crickets Locusts Spider mites Spodoptera spp. Aphids

Assemblage

Maximum potential exposure

23/8/04

Sap feeders

Defoliators

Feeding guild

Table 5.4. Selection matrix for prioritizing non-target herbivore and saprovore arthropod species associated with maize in Kenya (species with blanks could not be covered during the workshop).

Environ. Risk. Chap 05 Page 133

133

Saprovores

Catatus spp. Tenebrionids (darkling beetles) Earwigs

Mycetophagidae (dusty brown beetles) Carpophilus spp.

Assemblage

Medium low Medium

Certain

Abundant

Certain Likely Likely

Low

Abundance

Likely

Occurrence Sometimes

Linkage

Reproductive Always

Reproductive Sometimes

Reproductive Sometimes

Anytime

Presence

May be mainly predator

Storage contaminant

Potential pest

Significance

Aspergillus flavus

Disease vector

Possible adverse effect

Low

Damage

2

3 3

1

3

Rank

9:32

Feeding guild

Maximum potential exposure

23/8/04

Table 5.4. Continued.

Environ. Risk. Chap 05 Page 134

134 A.N.E. Birch et al.

Environ. Risk. Chap 05

23/8/04

9:32

Page 135

Biodiversity and Non-target Impacts

135

losses in Kenya (Muhammad and Underwood, Chapter 2, this volume). Carpophilus spp. were also given the highest priority because of their role in spreading the fungus A. flavus, which causes high aflatoxin levels in maize grains (Sétamou et al., 1998). S. zeamais and P. truncatus were prioritized because of their ubiquity and because of their abundance and the importance of postharvest losses for Kenyan farmers (Muhammad and Underwood, Chapter 2, this volume). In summary, of the total of 26 herbivore species and groups identified as important in Kenyan maize production systems in Step 1, eight species received the highest ranking (i.e. ‘1’). Some of these will be used in subsequent analyses, to illustrate procedures that should be applied to the other highly ranked species. Pollinating and pollen-feeding insects on maize Given the lack of information presently available (Step 1), the selection matrix could not be used to prioritize and select these species but was used to identify knowledge gaps (Table 5.5). In the face of this uncertainty, general ecological information about species associated with maize pollen in other parts of the world was used to suggest at least four species be included in subsequent analysis, including at least one species of wild, native bee, honeybee, and two species of the more important pollen-feeding predator species, e.g. from the orders Forficulidae (earwigs) and Coccinellidae (ladybird beetles). We focused on the honeybee (Apis mellifera L.) because honeybee biology is well known, specific expertise is available at the local level, they are important pollinators, they visit maize as a preferred source of pollen in other countries (Nowakowski and Morse, 1982; Vaissiere and Vinson, 1994), and previous evaluations of Bt maize have focused on the European subspecies, while in East Africa the African subspecies dominates. Natural enemies of maize herbivores The main parasitoid species of maize stemborers in Kenya are ranked for the highlands where B. fusca dominates and the lowlands where C. partellus dominates (Table 5.6). Trichogramma spp. were identified as an important eggparasitoid species also attacking H. armigera, one of the most important identified non-target herbivores in maize in Kenya. The most common identified Kenyan species are Trichogramma sp. nr. mwanzai (Guang and Oloo, 1990), Trichogramma sp. nr. exiguum (Ochiel, 1989) and Trichogramma buornieri (Abera et al., 2000). The final choice of which species to test should be based on a more detailed analysis of their relative importance and abundance in the agroecological zone of interest, which could not be completed during the workshop. The selected species is therefore referred to as Trichogramma spp. in the further assessment. The two Cotesia spp. (larval parasitoids) also received the highest ranking and will be used in subsequent analyses. Larval and pupal parasitoids of stemborers are important in reducing the carryover population in crop residues

Meloidae (pollen beetles) Apis mellifera Wild bee spp. Ants Earwigs Coccinellids Butterflies

Pollen feeders:

Not on maize

Other than flower visiting

Flower visiting

Assemblage

Likely Certain Certain Certain

Likely

Occurrence

Abundant Medium Medium Medium

Abundance

Anytime Reproductive Reproductive Reproductive

Reproductive

Presence

Sometimes Always Always Through dispersed pollen

Linkage

Natural enemies: parasitic on grasshopper eggs Pollinators of other crops Pollinators of other crops Natural enemy May be mainly predator Predaceous Not feeding on crop

Possible adverse effect

9:32

Feeding guild

Maximum potential exposure

23/8/04

Table 5.5. Selection matrix for pollinator and pollen-feeding species associated with maize in Kenya.

Environ. Risk. Chap 05 Page 136

136 A.N.E. Birch et al.

Species

Medium

Medium Low Low Low

Certain Occasional Occasional Certain

Low Low

Certain Occasional

Likely

Medium Low–medium

Certain Certain

All season All season All season All season

All season

All season All season

All season All season

All season

Presence

Strong Strong Strong Strong

Strong

Strong Strong

Strong Strong

Strong

Linkage

1 3 3 2

2

2 3

1 2

1

Rank

9:32

Highland Kenya ecosystem (Busseola fusca dominates) Egg parasitoid Trichogramma spp. Trichogrammatoidea spp. (uncompleted) Telenomus spp. (uncompleted) Larval parasitoid Cotesia sesamiae Cotesia flavipes Pupal parasitoid Dentichasmias busseolae Pediobus furvus Predators (Uncompleted)

Medium

Abundance

Certain

Occurrence

23/8/04

Lowland Kenya ecosystem (Chilo partellus dominates) Egg parasitoid Trichogramma spp. Trichogrammatoidea spp. (uncompleted) Larval parasitoid Cotesia flavipes Cotesia sesamiae Goniozus indicus (uncompleted) Egg and larval parasitoid Chelonus curvimaculatus (uncompleted) Pupal parasitoid Pediobus furvus Dentichasmias busseolae, native Predators (Uncompleted)

Guild

Maximum potential exposure

Table 5.6. Selection matrix for prioritizing parasitoid natural enemy species associated with maize in Kenya (differentiated for lowland and highland maize ecosystems).

Environ. Risk. Chap 05 Page 137

Biodiversity and Non-target Impacts 137

Environ. Risk. Chap 05

23/8/04

9:32

Page 138

138

A.N.E. Birch et al.

that may give rise to the initial infestation of the next growing season. In addition, new parasitoid species that have recently been released (Xanthopimpla stemmator Thunberg) or are being evaluated for release in Kenya (Telenomus isis Polaszek) as biocontrol agents of stemborer species will also need to be considered in a programme of impact testing. Maize-associated weeds Species ranks were based primarily on occurrence, abundance and significance (Table 5.7). ‘Presence’ and ‘Linkage’ were always high and general, because weeds are a pre-selected subset of plants that are present throughout most of the cropping season and linked generally to the crop plant. Most species with high occurrence were abundant in both lowland and highland areas, but for others, abundance seemed to differ regionally. For example, while Bracharia spp. and Pennisetum spp. are at low densities in the lowlands, they are at high densities in the highlands. Several species are significant as potential alternative host plant species for maize stemborers (Fitt et al., Chapter 7, this volume), and may function as sources or sinks of pests and natural enemies. The attractiveness of some grasses to stemborers is used in the ‘push–pull’ strategy to suppress stemborer populations in maize (Khan et al., 1997). However, recent evidence suggests that B. fusca is hardly present in wild hosts, and that C. partellus is specific to some wild plant species, while others are very poor hosts for maize stemborers (B. Le Ru, Nairobi, 2004, personal communication). The exact mechanism of host switching and host suitability of wild plants for stemborers is identified as an important information gap (consequences for resistance management/refugia selection; Fitt et al., Chapter 7, this volume). Some species are used as animal fodder. Based on these considerations, four grass and sedge species groups – Sorghum spp., Pennisetum spp., Panicum spp. and Rottboellia spp. – were ranked 1. Because of their additional importance within the push–pull strategy and as possible refuge species for Bt maize, Sorghum spp. and Pennisetum spp. were analysed further. FREE-LIVING GRASS AND SEDGE WEEDS.

WEEDS. Common broad-leaf weeds include Commelina benghalensis, Bidens pilosa, Tagetes minuta and Senna obtusifolia. A selection matrix was not constructed because the expertise was not present at the workshop. The species should be included in a full risk assessment of Bt maize in Kenya.

BROAD-LEAF

Striga spp. were immediately identified as the single most important weed species in those areas where it occurs in Kenya, and as a high priority for testing the non-target effects of Bt maize (Table 5.7). Striga spp. occur in the moist mid-altitude zone in western Kenya (Muhammad and Underwood, Chapter 2, this volume). In infested areas, they are highly abundant, widely present and strongly linked to the maize crop (obligatory parasite), and can cause significant reductions in yields.

PARASITIC WEEDS.

L2

L2

L1

H/L1 H/L1 H/L1 H/L1 H/L1 H/L1 H/L1 H/L1 H/L1 H/L1 H/L1 H/L1 H/L1

Presence

High

General General General General General General General General General General General General General

Linkage

Damaging, difficult to eradicate

Important weed AH/NE-Reserv. Insect repellent used in PP-strategy AH/NE-Reserv. AH/NE-Reserv. used in PP-strategy AH/NE-Reserv. AH/NE-Reserv. used in PP-strategy

Important weed

AH/NE-Reserv.

Significance

1

2 3 3 2 2 2 2 2 3 1 1a 1 1a

Rank

14:28

H, highlands; L, tropical lowlands; 1, high–likely; 2, intermediate; 3, low–unlikely; AH, alternative host plant to stemborers; NE-Reserv., can serve as a reservoir for natural enemies of stemborers. Some species of these genera are alternative host plants for all of the four most important stemborer species (Khan et al., 1997).

H3/L1 H1/L3 H2/L3 H/L2 H1/L2 H1/L2 H/L1 H1/L2 H2/L3 H/L2 H1/L3 H/L2 H/L2

Abundance

H3/L2 H2/L3 H1/L2 H/L2 H/L2 H/L1 H/L1 H1/L3 H2/L3 H/L1 H/L1 H/L1 H/L1

Occurrence

Potential adverse effect

10/9/04

Grass and sedge weeds Andropogon spp. Brachiaria spp. Cynodon spp. Cyperus spp. Digitaria spp. Eleusine spp. Eragrostis spp. Hyparrhenia spp. Melinis spp. Panicum spp. Pennisetum spp. Rottboellia spp. Sorghum spp. Broad-leaf weeds: not completed Parasitic weeds Striga spp.

Weed species

Maximum potential exposure

Table 5.7. Selection matrix for prioritizing weeds associated with maize in Kenya (differentiated for low- and highland maize production systems).

Environ. Risk. Chap 05 Page 139

Biodiversity and Non-target Impacts 139

Environ. Risk. Chap 05

23/8/04

9:32

Page 140

140

A.N.E. Birch et al.

Soil ecosystem functions Because of the high species and process diversity in plant–soil systems, the selection procedure was pursued simultaneously with a function-based approach for microorganisms and a species-based approach for macroorganisms. Horizontal gene transfer in the soil was not considered in this workshop, but will be in future. A function selection matrix was developed and applied to prioritize soil ecosystem functions (Table 5.8). Exposure to Bt maize material and linkage to plant roots were criteria used to define ‘maximum potential exposure’. Criteria related to ‘potential adverse effect’ were the importance in nutrient cycling for maize production, and as indicator values for soil health. Higher priority was given to functions involved in the degradation of maize residues (carbon compound and cellulose breakdown) with the coincident release of plant nutrients. The organisms involved in these functions (bacteria, fungi and soil micro- and macroorganisms) may be continually exposed to the Bt toxins. Soil-ecosystem dynamics are limited by energy availability, and functions are responsive to plant inputs. Bt maize may have more lignin in stems than the non-Bt counterparts (Saxena and Stotzky, 2001a), which may lead to slower degradation. This in turn may affect other soil micro- and macroorganisms, e.g. an increase in populations of grey maize leaf spot (Cercospora zea-maydis), which survives between cropping seasons on plant residues. As Kenyan maize fields receive little fertilizer, higher priority was also given to functions essential for nitrogen and phosphate supply. Phaseolus bean, with nitrogen-fixing bacteria in its roots, is a primary inter- and rotation crop with maize in Kenya. The quantity and the composition of root exudates probably play a role in the colonization of roots by specific Rhizobia spp. and other bacteria (Parker et al., 1977). Although maize is not the host plant, changes in its root exudates could affect neighbouring Rhizobia spp., and other free-living nitrogen fixers. Mycorrhizas are associated with the provision of phosphate to plants, and are often found in plants grown in low-phosphate or nutrient-poor soils, under a range of tillage systems (McGonigle et al., 1999). Maize roots have been frequently shown to be colonized by arbuscular mycorrhizas (AM) (e.g. McGonigle et al., 1999; Fries et al., 2000), over a large range of climates and soils. AM associations have also been reported to enhance crop yield on the acid sandy soils of West Africa (Bagayoko et al., 2000), and in maize cultivation in Kenya (Beatrice Anyango, Nairobi, 2002, personal communication). So it was considered probable that maize–AM associations are important factors relating to crop yield in Kenya. As they have a particularly intimate intracellular association with maize roots, on which they are dependent for carbon, they require evaluation. There was insufficient expertise at the workshop to complete a selection matrix for soil macroorganisms, but some options are offered. Cry toxins persist associated with plant residues, and so effects could continue after the modified crop has been grown and harvested. Interactions can be complex. It is recommended that soil macroorganisms be prioritized by:

Phosphorus and micronutrient uptake

Cellulose breakdown Ammonification Nitrification Denitrification Nitrogen fixation

Carbon decomposition

Function

Macrofauna, fungi, Residues (leaf, stalk bacteria and roots) and exudates Fungi, bacteria Residues and exudates Bacteria Residues and exudates Bacteria Residues and exudates Bacteria Residues and exudates Rhizobia spp. Residues, exudates and root associations Mycorrhizas Residues, exudates and root associations

Source of Bt toxin and metabolites

High

Low Medium Medium Medium High

Low

Link to plant roots

P

C N N N N

C

Cycle

Essential

Essential Important Important Important Essential

Essential

Importance

High

High High High Low High

High

Soil health indicator

High

High Medium Medium Medium High

High

Impact of adverse effect on maize production

1

1 2 2 2 1

1

Rank

9:32

Indicator organisms

Importance in nutrient cycling for plant production

Potential adverse effect

23/8/04

Potential exposure

Table 5.8. Selection matrix for prioritizing soil functions for Bt maize in Kenya.

Environ. Risk. Chap 05 Page 141

Biodiversity and Non-target Impacts 141

Environ. Risk. Chap 05

23/8/04

9:32

Page 142

142

A.N.E. Birch et al.

1. Abundance – common species may be more likely to play significant ecological roles. 2. Functional significance – e.g. direct consumers of plant residue (e.g. earthworms, beetles, termites, mites, Melolonthinae larvae, e.g. Schizonycha, Phyllophaga) that degrade large pieces of residue into smaller pieces thereby facilitating/enhancing microbial degradation; organisms that are important for soil physicochemical structure, such as soil macro- and micropores, soil crumble structure (e.g. earthworms, termites). 3. Trophic relationships – predatory or saprophagous soil organisms important for regulation of soil pest species (e.g. predatory mites, nematodes, collembolans). Findings of Step 2 From a total of 62 species, species groups and functions, 24 were assigned highest priority and analysed further: eight non-target maize herbivore species, five natural enemies of maize herbivores, four pollen feeders, two weed species and four soil functions. The selection matrices proved to be a valuable tool that allowed us to make efficient, transparent, science-based decisions on which species and function to proceed with in the further assessment. The information required to complete a selection matrix depends on the cropping system and local environment, but does not rely on any information associated with the transgenic crop. For soil functions and weeds, the selection matrix was modified to accommodate the particular system characteristics.

Step 3: Exposure pathway analysis The purpose of this evaluation is to differentiate candidate test species/functions likely and unlikely to be exposed to the Bt toxin, and for the former, to guide the design of the exposure system that should be used in the test protocols (Step 5). Information on Bt toxin expression in maize tissues is not fully available (Andow et al., Chapter 4, this volume). Most of the promoters are constitutive, and it is expected that Bt toxins are expressed in all growing maize tissues, including pollen, cobs, silk, roots and root exudates. Non-target maize herbivores Because expression information is not available, all herbivores feeding on any Bt maize tissue should be expected to ingest Bt toxin. One exception may be exclusively phloem-feeding insects, such as aphids, because no Bt proteins have been detected either in the insects or in their excreted honeydew when feeding on Bt maize (Head et al., 2001; Raps et al., 2001; Dutton et al., 2002). However, in rice, Bernal et al. (2002) found some evidence for presence of Bt toxin in the honeydew of the brown leafhopper. Thus, although expression in the maize plant phloem is probably unlikely, it should be verified in each transgenic Bt crop event in case new events express toxin in this tissue, intentionally or unintentionally. Leafhoppers feed on phloem but also on mesophyll cells that will contain the Bt

Environ. Risk. Chap 05

10/9/04

14:28

Page 143

Biodiversity and Non-target Impacts

143

toxin in many maize events, and are therefore likely to be exposed (BosquePérez, 2000). They are important transmitters of virus diseases. Both H. armigera and several Spodoptera spp. feed on maize leaves or other plant tissues that will contain Bt toxin. Both species are susceptible to Cry1Ab and Cry1Ac proteins (Hilbeck et al., 1998; Dutton et al., 2002), although considerably less so than the target pest species. Sublethal or possibly even lethal effects can be expected and should be quantified. Both species spend their entire life cycle on maize, Spodoptera feeding primarily on leaves and green tissues, whereas H. armigera feeds preferably on maize ears, which may contain lower Bt toxin concentrations.

LEPIDOPTERAN HERBIVORES.

Bt proteins are very complex, highly bioactive toxins, and it cannot be excluded that they exert subtle, sublethal effects on non-lepidopteran herbivores (non-target). Hence, direct toxicity from direct exposure to the Bt protein should not be the only criterion for making the final selection of non-target herbivore species to be tested, and it is important to also consider non-lepidopteran herbivores that are important in maize in Kenya for risk assessment. Locusts are defoliators, and spider mites ingest the cell contents of leaves and other tissues where the Bt toxin is expressed; therefore both groups will be directly exposed to Bt toxin (Dutton et al., 2002), possibly affecting life cycle and behaviour and therefore population dynamics, and indirect effects are also possible. These species were not evaluated further in the workshop, but should be considered in the complete assessment.

NON-LEPIDOPTERAN HERBIVORES.

GRAIN FEEDERS. S. zeamais and P. truncatus feed on the maize cob and kernels and will therefore probably be directly exposed to tissues expressing Bt toxin, both before and after harvest. The amount of Bt toxin in the cob and kernel of each event needs to be determined to check if exposure is likely. Exposure is likely to be long term in storage, even if the dose of Bt toxin is low. SAPROVORES. Carpophilus spp. beetles feed primarily on the frass of lepidopteran herbivores such as H. armigera. Raps et al. (2001) reported that the frass of a non-target lepidopteran pest, Spodoptera littoralis (Boisd.), contained fairly high concentrations of Bt toxins when fed on Bt maize. It is therefore assumed that Carpophilus spp. may be exposed to the Bt toxin via the frass of exposed lepidopteran herbivores, such as H. armigera and Spodoptera spp.

Pollinating and pollen-feeding insects on maize Domesticated honeybees and wild bees visit maize flowers to collect pollen. In diverse systems, bees can collect from many different plant species. Honeybee foragers and other pollen feeders (e.g. parasitoid wasps such as Trichogramma spp., see below; Long et al., 1998) regularly visit maize. It is not known but likely that the Bt protein will be expressed in maize pollen with the currently used promoters (Andow et al., Chapter 4, this volume). Foraging bees carry the pollen to the bee colony, potentially exposing both larvae and adults to the Bt toxin. Maize plants do not have true nectaries, but they secrete guttation

Environ. Risk. Chap 05

10/9/04

14:28

Page 144

144

A.N.E. Birch et al.

fluids from vascular tissue sources, which are used by many insects (D.A. Andow, Nairobi, 2002, personal communication), although it is unlikely that these guttation fluids contain substantial Bt toxins. A number of natural enemies, including species of the Coccinellidae and Forficulidae, are also known to feed on pollen as adults and sometimes as larvae. Pollen accumulates on maize leaves at the collar and, if it gets moist, will ferment, with yeasts being a dominant component of the fermentation process. Alternatively, if there is not much moisture, pollen will tend to be scattered on the upper leaf blade and persist there, often colonized by fungi, including some facultative maize pathogens, providing a rich food source (D.A. Andow, Nairobi, 2002, personal communication). Pollen can also be consumed by herbivore species (e.g. butterfly larvae) on other plants around the maize field receiving drifting pollen. Natural enemies of maize herbivores Natural enemies can be exposed to the effects of Bt toxin via multiple pathways. It is therefore important to consider not only the primary tritrophic exposure route through the prey or host, but also bitrophic exposure through direct feeding on plant tissues, or feeding on herbivore excretions such as honeydew or frass. A series of questions were developed to facilitate exposure analysis for the Trichogramma and Cotesia species groups (Box 5.1). EGG PARASITOIDS OF NON-TARGET LEPIDOPTERA. Trichogramma spp. parasitize eggs of lepidopteran species on maize, including H. armigera, and are known for their importance as biocontrol agents (van den Berg and Cock, 1993; Sithanantham et al., 2001). The hatched parasitoid larvae feed on the egg contents, pupate inside the egg and leave the egg after adults emerge. Trichogramma spp. spatio-temporal occurrence overlaps well with that of their host species. Further data are required to check if lepidopteran spp. feeding on maize also feed and survive on neighbouring uncultivated plants. Recent evidence suggests that B. fusca is rarely found on wild plants, and that other maize stemborers are present only in certain weeds, which could prevent parasitoids from finding alternate hosts if Bt maize eliminates stemborers on maize (B. Le Ru, Nairobi, 2004, personal communication). Many unidentified noctuid species exist on wild grasses and sedges (B. Le Ru, Nairobi, 2004, personal communication), and are likely to be alternate hosts to some maize stemborers in some regions, but more information is needed (knowledge gap). In addition, we need to know what happens to parasitoids when maize stemborers are sublethally affected by Bt toxin (e.g. low dose exposure or low susceptibility to the toxin). In general, little information exists on the importance of egg parasitoids in East and southern Africa (identified information gap). Bitrophic exposure for Trichogramma spp. (Box 5.2): Adult Trichogramma spp. are usually highly mobile and seek to feed preferably on plant nectar, but also on maize pollen and guttation fluids (Long et al., 1998). Other direct plant tissue feeding has not been reported for these species. If Bt toxin is expressed in the pollen or guttation fluids of Bt maize in Kenya, then adults will be exposed. This potential exposure pathway can be readily evaluated.

Environ. Risk. Chap 05

23/8/04

9:32

Page 145

Biodiversity and Non-target Impacts

145

Box 5.2. Likely potential exposure analysis for egg parasitoids Trichogramma spp. and larval parasitoids Cotesia spp., based on questions (Q) and responses given to the best of knowledge of the experts present and the information available. Exposure analysis for egg parasitoid Trichogramma spp. and larval parasitoid Cotesia spp. Bitrophic exposure Q1: What is the spatio-temporal overlap of Trichogramma or Cotesia spp. feeding period and the Bt maize growth cycle? ⇒ Trichogramma spp.: In most parts of Kenya, maize growth stages overlap largely with Trichogramma spp. occurrence. ⇒ Cotesia spp.: Whole season, i.e. overlap is complete. Q2: Does Trichogramma or Cotesia spp. feed on the parts of the maize plant containing the Bt toxin? ⇒ Trichogramma spp.: Adult Trichogramma spp. feed on pollen, which could contain Bt toxin, depending on the promoter used. The nectar-like guttation fluid may be fed upon but is unlikely to contain Bt toxins. ⇒ Cotesia spp.: Possibly, on pollen (depending on promoter used and on field confirmation that pollen is a food source for Cotesia flavipes) and on nectar-like guttation fluid (which is less likely to contain Bt toxin than pollen). Q3: Is the Bt toxin and/or metabolites detectable in Trichogramma or Cotesia spp. or its excretions after feeding on the plant? ⇒ Trichogramma spp.: Possible, but no data available. We assume exposure through this route negligible or unlikely (see above). ⇒ Cotesia spp.: No data available but considered possible if Bt-containing pollen is consumed. Q4: Does Trichogramma or Cotesia spp. feed on host products/excretions, e.g. honeydew, faeces? ⇒ Trichogramma spp.: Possible, but little data – Trichogramma spp. feed on aphid/planthopper honeydew but to date Bt toxins have not been detected in phloem or xylem tissues of maize. Trichogramma adults are not known to feed on Lepidopteran faeces. ⇒ Cotesia spp.: Possibly feeds on frass of hosts and honeydew from aphids or leaf and planthoppers. Q5: Is the Bt toxin and/or metabolites detectable in the host products/excretions e.g. honeydew, faeces? ⇒ Trichogramma spp.: Unclear whether maize guttation fluids or honeydew contains Bt toxin. ⇒ Cotesia spp.: Possibly feeds on frass of hosts and honeydew from aphids or leaf and planthoppers. Q6: Is the Bt toxin and/or metabolites detectable in Trichogramma or Cotesia spp. or its excretions after feeding on the host products? ⇒ Trichogramma spp.: Unclear. See previous response under Q5. ⇒ Cotesia spp.: No data available (but see responses above). Tritrophic exposure Q7: What is the spatio-temporal overlap of Trichogramma or Cotesia spp. parasitism on the host Lepidoptera eggs on Bt maize? ⇒ Trichogramma spp.: In most parts of Kenya, oviposition by the host Lepidoptera overlaps through most of the crop growth stages with Trichogramma spp. oviposition. Continued

Environ. Risk. Chap 05

23/8/04

146

9:32

Page 146

A.N.E. Birch et al.

Box 5.2. Continued. ⇒ Cotesia spp.: Only certain life cycle stages of the host (third to fifth instar) are suitable for attack, usually during medium crop growth stage. But if the crop is at different growth stages in adjacent fields then third to fifth instars could be available all season. Q8: Is the Lepidoptera host feeding on the plant tissues containing the Bt toxin and/or metabolites? ⇒ Trichogramma spp: Yes, H. armigera feed on leaves and ears that contain the Bt toxin. ⇒ Cotesia spp.: Yes. Q9: Is the Bt toxin and/or metabolites detectable in the host? ⇒ Trichogramma spp.: Possible only if Bt toxin is transferred from caterpillar to adult and from adult to egg (food forTrichogramma spp. larva). ⇒ Cotesia spp.: Yes, most likely, although no data exist to date specific to the Kenyan case. However, published data exist where Bt toxin concentrations was measured in Spodoptera spp. feeding on Bt maize (Raps et al., 2001; Dutton et al., 2002). Q10: Is the Trichogramma or Cotesia spp. feeding on hosts that contain the Bt toxin and/or metabolites? ⇒ Trichogramma spp.: Unclear. No data on Bt toxin content of host eggs. ⇒ Cotesia spp.: Yes, most likely, although no data exist to date specific to the Kenyan case (see response to Q9 above). Q11: Is the Bt toxin detectable in the Trichogramma or Cotesia spp. after feeding on the host? ⇒ Trichogramma spp.: Unclear. No data. ⇒ Cotesia spp.: No data yet (but see responses above). Q12: If the Bt toxin is not detectable in the Trichogramma or Cotesia spp., can it be indirectly affected by Bt maize? ⇒ Trichogramma spp: Not certain. It is possible that the plant composition or constituents, which are not of toxicological concern, but of insect nutritional value for the parasitoid, could be affected through altered host-egg suitability for development. ⇒ Cotesia spp.: Not certain. It is possible that the plant composition or constituents that are not of toxicological concern, but of nutritional value for the parasitoid, could affect suitability for development through altered host larvae. Additionally, if host larva is lethally affected by the Bt toxin, C. flavipes larva will die with the host, increasing C. flavipes population mortality rates. Higher level exposure Q13: Do Trichogramma or Cotesia spp. cannibalize their own species and therefore might be exposed to the Bt toxin, its toxins or its effects via its own spp. hosts? ⇒ Trichogramma spp.: Not known so far whether Trichogramma spp. parasitize their own eggs in a host (gap in knowledge). ⇒ Cotesia spp.: C. flavipes develops at the expense of C. sesamiae (Ngi-Song et al., 2001). Q14: What feeding preference or other behaviour could increase exposure of Trichogramma or Cotesia spp.? ⇒ Trichogramma spp.: changes in oviposition behaviour and oviposition rates to favour Lepidopteran eggs on Bt maize, reactions to changes in host population density. ⇒ Cotesia spp.: The hyperparasite Aphanogmus fijiensis was identified as a next higher trophic level organism possible being exposed to the Bt toxin via its C. flavipes host. Further analysis is necessary to assess what the role of this natural enemy is and whether there is reason for concern if Bt-containing hosts would adversely affect this species.

Environ. Risk. Chap 05

23/8/04

9:32

Page 147

Biodiversity and Non-target Impacts

147

Tritrophic exposure for Trichogramma spp. (Box 5.2): The lepidopteran host larvae are known to feed on Bt maize and ingest the Bt toxin. However, it has not yet been determined whether the ingested Bt toxin is passed on to the lepidopteran adult stage and subsequently to the eggs that the lepidopteran adults produce. This would constitute the most likely tritrophic route through which Trichogramma spp. larvae could come into contact with Bt toxins or their metabolites. It is also possible that adult Trichogramma spp. could be exposed if the honeydew produced by aphids and leafhoppers, a food source for Trichogramma spp. (McDougall and Mills, 1997) is found to contain Bt toxin. The specialized Cotesia flavipes (Cameron) and Cotesia sesamiae (Cameron) together account for 83% of the parasitized borers in southern coastal Kenya (Zhou et al., 2003). C. flavipes is an exotic parasitoid introduced to control C. partellus, and will accept equally all four main Kenyan stemborer species; however, it fails to develop in B. fusca due to encapsulation of the eggs (Ngi-Song et al., 1995), unless the B. fusca larva is parasitized by both Cotesia species at the same time, in which case C. flavipes can develop at the expense of C. sesamiae (Ngi-Song et al., 2001). Both Chilo stemborer species and Sesamia calamistis (Hampson) are suitable hosts for C. flavipes, though mortality is higher in S. calamistis. When third, fourth, fifth and sixth instars of C. partellus were exposed to C. flavipes females, parasitoid development was least successful in third-instar hosts, and most successful in fifth-instar hosts, and developmental time was longer (on average 24 days in third instar, 16.2 days in sixth instar). In contrast, C. partellus is most likely to die when parasitized as third instar and least likely as sixth instar. Bitrophic exposure for C. flavipes (Box 5.2): Where C. flavipes occurs, it completely overlaps with the maize growth period. Adult C. flavipes are highly mobile, with a lifespan of a few days. They possibly feed on maize guttation fluids or pollen, or plant sap from wounds, as they are known to have a longer lifespan in the laboratory when fed honey (Potting et al., 1997). They might also feed on the fresh frass of the stemborer host larvae; however, this must be confirmed by field studies on the feeding behaviour of adult C. flavipes in Kenyan agroecosystems. If the Bt toxin is expressed in the pollen, phloem or guttation fluids, adults can be exposed. Stemborer frass has been reported to contain Bt toxins in fairly high concentrations, so it is possible that C. flavipes would be exposed via this route (Raps et al., 2001). Tritrophic exposure for C. flavipes (Box 5.2): Stemborer host larvae are known to feed on Bt maize and ingest Bt toxin, therefore Cotesia spp. larvae are very likely to be tritrophically exposed to Bt toxin and/or metabolites as long as the stemborers can survive on Bt maize.

LARVAL PARASITOIDS OF NON-TARGET LEPIDOPTERA.

Maize-associated weeds An exposure analysis was not considered useful for weeds. The group decided to incorporate exposure analysis into the hazard identification section (Step 4 below).

Environ. Risk. Chap 05

23/8/04

9:32

Page 148

148

A.N.E. Birch et al.

Soil ecosystem functions Exposure analysis of soil ecosystems is complex as the organisms are exposed to a combination of degrading and living transgenic plant material and root exudates. Furthermore, the novel transgenic products/proteins will interact physically and chemically with soil constituents such as humic acids, clay minerals and or colloids. Far less is known about the fate of complex organic molecules like proteins in soils than of smaller organic or inorganic chemicals such as pesticides or industrial pollutants. Therefore, the following parameters need to be determined for an exposure analysis. Firstly, routes of movement and transport of transgenic plant material and products need to be identified (e.g. root exudates, movement of plant residues, including roots, release of proteins from the plant residues, etc.). This requires a good working knowledge of protein expression, the soil–plant interface, and residue movement and management. Bt toxins enter the soil in several ways: (i) via root exudates; (ii) release during senescence; and (iii) via leachates from degrading plant material and injuries. Secondly, the fate of these plant materials and associated Bt proteins needs to be understood and quantified (e.g. adsorption to clay minerals (Tapp et al., 1994), or humic acids, and potential accumulation of the protein (Tapp and Stotzky, 1995b), immobilization or leaching of transgenic proteins). It is essential to know how long the toxin persists in plant residues, and whether it remains active. Saxena and Stotzky (2001a,b), and Saxena et al. (2002) reported that active Bt toxin from transgenic maize root exudates and degrading Bt maize biomass persisted in soil for up to 350 days, the longest time studied. In studies that have examined the persistence of Bt toxin from transgenic cotton, the toxin was still detectable when the experiments were terminated after 28–140 days (Palm et al., 1994, 1996; Sims and Ream, 1997). Other studies have also reported the persistence of purified Bt toxins in soil for up to 234 days, when the trials were terminated (Tapp and Stotzky, 1995a; Palm et al., 1996; Tapp and Stotzky, 1998). Soil micro- and macroorganisms may therefore be exposed to Bt toxin over long periods. Findings of Step 3 Table 5.9 lists all 16 maize-associated arthropods prioritized for pre-release non-target impact testing in Kenya, with a summary of significance and exposure. For six species or species groups, testing protocols were developed in Step 5. For seven of the arthropod species or groups that were analysed, direct bitrophic exposure to Bt toxin must be expected. For four species analysed there did not exist sufficient information on transgene expression in pollen, nectar or phloem to confirm or refute bitrophic exposure. This re-affirmed the importance of a thorough characterization of the Bt maize for risk assessment (Andow et al., Chapter 4, this volume). For Cotesia spp., a knowledge gap was identified on adult feeding behaviour. Carpophilus spp. and Cotesia spp. may be exposed via lepidopteran frass. Tritrophic exposure is unlikely to occur for Trichogramma spp., unless lepidopteran eggs or aphid/planthopper honeydew contain Bt toxin or metabolites and are consumed. However, other important

Defoliator Defoliator Grain feeder

Grain feeder

Sucker

Saprovore

Spider mites Locusts Sitophilus zeamais

Prostephanus truncatus

Plant and leafhoppers

Carpophilus spp.

Cotesia flavipes

Trichogramma spp.

Larval parasitoid

Pollen feeder, larval predator Pollen feeder, egg predator Egg parasitoid

Coccinellidae spp.

Forficulidae spp.

Pollen feeder

Biodiversity and Non-target Impacts

Continued

Yes

Yes

No

No

No

Yes

Yes

No

No

No No No

Yes

Yes

Protocol developed?

9:32

Wild bee spp.

Very likely to be exposed by feeding behaviour, susceptible to toxin Abundant, cause heavy damage Very likely to be exposed by feeding behaviour, susceptible to toxin Abundant, cause heavy damage Very likely to be exposed by feeding behaviour Cause heavy damage when present Very likely to be exposed by feeding behaviour Significant potential pest, always Very likely to be exposed by feeding behaviour present Significant potential pest, always Very likely to be exposed by feeding behaviour present Cicadulina spp. vector of maize Likely to be exposed by feeding on mesophyll cells streak virus Vector of Aspergillus flavus May be exposed by feeding on frass of Helicoverpa armigera Important pollinator of other crops Will be exposed if Bt toxin is present in pollen or guttation fluids Pollination function Will be exposed if Bt toxin is present in pollen or guttation fluids Important natural enemy Will be exposed if Bt toxin is present in pollen or guttation fluids Important natural enemy Will be exposed if Bt toxin is present in pollen or guttation fluids Important natural enemy of Will be exposed if Bt toxin is present in pollen, lepidopteran species phloem or honeydew, guttation fluids or lepidopteran eggs Important natural enemy of Will be exposed to Bt toxin in host larvae; may Chilo partellus be exposed if Bt toxin is present in pollen or lepidopteran frass

Significant potential pest

Exposure

23/8/04

Honeybee (Apis mellifera) Pollen feeder

Spodoptera spp.

Silk and cob feeder Defoliator

Helicoverpa armigera

Selected species (above-ground arthropods) Feeding guild Rationale

Table 5.9. Summary of non-target above-ground maize-associated arthropods prioritized for testing for non-target impacts of Bt maize in Kenya.

Environ. Risk. Chap 05 Page 149

149

Larval parasitoid

Egg or larval predator

Cotesia sesamiae

Predators, e.g. ants, anthocorids, chrysopids

Important natural enemies of lepidopteran eggs and larvae

Important natural enemy of stemborer species

Will be exposed to Bt toxin in host larvae; may be exposed if Bt toxin is present in pollen or lepidopteran frass Will be exposed to Bt toxin in lepidopteran larvae, will be exposed if Bt toxin is present in lepidopteran eggs

Exposure

No

No

Protocol developed?

9:32

Selected species (above-ground arthropods) Feeding guild Rationale

23/8/04

Table 5.9. Continued.

Environ. Risk. Chap 05 Page 150

150 A.N.E. Birch et al.

Environ. Risk. Chap 05

23/8/04

9:32

Page 151

Biodiversity and Non-target Impacts

151

indirect impact routes or effects on important disease vectors, natural enemies or weeds make it necessary to include species that are unlikely to be directly exposed to the Bt toxin. The identified soil functions will all be exposed to Bt toxin in plant residues and root exudates.

Step 4: Hazard identification and hypothesis development Hazards relate to the effects of the Bt toxin, its metabolites or any combination effects with plant secondary metabolites on the life history and fitness parameters of the non-target organism, such as development time, survival and reproduction. Behavioural parameters such as preferential host, or host plant and oviposition choices, are also of great ecological importance. From the exposure analyses, hazard scenarios can be identified and used to frame hypotheses relevant for risk assessment. In the following sections, hazards will be identified first, and subsequently the research hypotheses that address the identified hazards will be listed.

Non-target maize herbivores Eight non-target maize herbivore species are recommended for inclusion in a pre-release non-target effects testing programme on Bt maize in Kenya. All are likely to be directly exposed to the plant-expressed Bt toxins. Identification of hazards and research hypotheses was carried out for the three species below, and was not completed for S. zeamais, P. truncatus, leafhoppers, locusts or spider mites. LEPIDOPTERAN HERBIVORES. The hazard associated with these species is that they might become significant secondary pests on Bt maize. Resistance risks associated with them are discussed by Fitt et al. (Chapter 7, this volume). H. armigera is closely associated with maize when cotton is not readily available. Spodoptera spp. are more polyphagous and can survive on a broader range of host plants. Therefore, understanding the possible changes in the fitness of these herbivores with the introduction of a new GM crop will determine their pest status. The following research hypotheses were developed:

1. Non-target lepidopteran herbivores (H. armigera and Spodoptera spp.) will have higher reproductive rates and/or immature survival on Bt maize. 2. These herbivores will prefer to oviposit on Bt maize. Aflatoxin infections of maize are a serious health threat for subsistence farmers in Kenya. Carpophilus spp. may alter transmission and prevalence of A. flavus fungi on Bt maize. Consequently, it will be important to quantify whether the spread of A. flavus is increased, decreased or unchanged in Bt maize. The following research hypothesis was developed: SAPROVORE AND DISEASE VECTOR CARPOPHILUS SPECIES.

Environ. Risk. Chap 05

23/8/04

9:32

Page 152

152

A.N.E. Birch et al.

3. Carpophilus spp. transmits A. flavus from maize ear to maize ear when it consumes the frass of lepidopteran target and non-target species feeding on Bt maize. Pollinating and pollen-feeding insects on maize In this section, the focus was on the honeybee, A. mellifera, which is an important economic species and essential for pollination in many crops such as fruit trees. Hazards for honeybees can arise at different organizational levels, affecting individual fitness parameters but also the colony as a whole. Feeding on Bt maize pollen expressing the Bt toxin may have subtle, long-term impacts on any of the relevant parameters. Similar hazard scenarios – except on the colony level – are also possible for other pollen-feeding organisms. The following research hypotheses were developed: 4. Individual fitness parameters of pollen-feeding/pollinating species will be reduced by Bt maize compared with the non-transgenic varieties. 5. Bt maize is more attractive to honeybees than non-transgenic maize varieties. 6. Pollinator effectiveness will differ on Bt maize from that on non-transgenic varieties. 7. Colony development will be adversely affected by Bt maize compared to the non-transgenic varieties. Natural enemies of maize non-target herbivores The hazard associated with the selected parasitoid species is that Bt maize causes a decrease in their reproductive fitness on both target and non-target pests, thus preventing them playing a role in resistance management (Fitt et al., Chapter 7, this volume) and reducing their biocontrol capacity on non-target pests and on neighbouring non-Bt maize. For both parasitoid species groups, bitrophic exposure is considered likely if the Bt toxin is present in pollen, phloem and/or guttation fluids. For the larval parasitoid Cotesia spp., tritrophic exposure is also possible. Tritrophic exposure is unlikely to occur for Trichogramma spp. but is possible if they consume lepidopteran host eggs or aphid/planthopper honeydew that contain Bt toxin or metabolites.

PARASITOIDS.

EGG-PARASITOID TRICHOGRAMMA SPECIES. Trichogramma spp. parasitize lepidopteran eggs, including H. armigera and Spodoptera spp. eggs. Plant secondary metabolites (volatile compounds) influence the host search and oviposition behaviour of gravid female parasitoids (Bouwmeester et al., 2003). Such behavioural changes can have a profound impact on their biocontrol capacity (Turlings et al., 1995; Hoballah and Turlings, 2001; Hoballah et al., 2002). Little is known to date about possible changes in the volatile patterns produced by transgenic Bt maize and how this could affect both herbivore and parasitoid behaviour. Further, certain parasitoid species are known to track the population dynamics of their host species in a density-dependent fashion. It is conceivable

Environ. Risk. Chap 05

23/8/04

9:32

Page 153

Biodiversity and Non-target Impacts

153

that reductions in target and non-target lepidopteran species in Bt maize will lead to a significant decline in egg parasitoids and therefore reduced biological control of other pests. The following research hypotheses were developed: 8. Suitability of host eggs for Trichogramma spp. parasitoids is reduced on Bt maize. 9. Oviposition preferences of Trichogramma spp. parasitoids are altered on Bt maize. 10. Bt maize is less attractive than non-transgenic maize for Trichogramma spp. parasitoids. Negative effects on Cotesia spp. could arise either from direct effects of the Bt toxin in host larvae, or indirectly, through altered nutritional composition of host larvae for Cotesia spp. larvae. Overall, parasitoid population density could be affected by the premature death of Cotesia spp. larvae if the host larvae die. Sublethal effects on the host larvae could signify reduced value of the host for survival, development and reproduction of the parasitoid. Lepidopteran adult oviposition behaviour is affected by cues from their host plants as Trichogramma are (Pivnick et al., 1994). Bt maize could therefore affect parasitoid host-finding ability, thus affecting its biocontrol capacity. In laboratory experiments, female C. flavipes are attracted to odours from infested and uninfested maize, sorghum and Napier grass, but the parasitoid is significantly more attracted to maize infested with stemborers than to artificially damaged maize, larvae alone, host frass or uninfested maize, indicating that these parasitoids use both plant volatiles released from damaged maize plants and volatiles from C. partellus frass as key host-finding signals (Potting et al., 1993, 1995; Ngi-Song et al., 1996). The production of volatiles attractive to the parasitoids is not restricted to the infested stem-part but occurs systemically throughout the plant (Potting, 1997). Learning does not seem to play a role in host microhabitat location for C. flavipes, and no intraspecific variation in host selection behaviour has been found (Potting, 1997). However, different strains did show variation in reproductive success and C. sesamiae are unable to locate aestivating hosts (Mbapila and Overholt, 1997). The relative proportions of stemborers parasitized on maize and on wild grasses or other crops could also change (e.g. if maize stemborers are eliminated on Bt maize and cannot survive on nearby wild plants of low host quality), which could have implications for resistance management strategies (Fitt et al., Chapter 7, this volume). In dual choice tests, C. flavipes and C. sesamiae cannot, however, discriminate between maize infested with C. partellus and maize infested with B. fusca (Potting et al., 1995). Unidirectional incompatibility, possibly caused by Wolbachia, was found between coastal and inland populations of C. sesamiae (Ngi-Song et al., 1998), and a C. sesamiae population from the coast was found to be infected with Wolbachia (Mochiah et al., 2002). Wolbachia are known to affect the phenotype of the carrier through several mechanisms, including male killing, cytoplasmic incompatibility, induction of parthenogenesis, feminization and altered fertility. The effect of Bt maize on these complex interactions are LARVAL PARASITOID COTESIA SPECIES.

Environ. Risk. Chap 05

23/8/04

9:32

Page 154

154

A.N.E. Birch et al.

unknown but should be investigated. Wolbachia has never been found in C. flavipes (A. Ngi-Song, Nairobi, 2002, personal communication), therefore C. sesamiae is proposed as the test parasitoid. The following research hypotheses were developed: 11. Bt maize reduces the host-finding ability of C. flavipes. 12. Host suitability for C. partellus for C. flavipes is reduced when hosts feed on Bt maize. 13. Wolbachia has different effects on C. sesamiae when hosts have fed on Bt maize compared to non-transgenic control varieties. 14. Adult C. flavipes feed on maize pollen and guttation fluids under field conditions. Maize-associated flora FREE-LIVING WEEDS. While weeds live in association with maize plants, they do not live on or from the maize plants. Therefore, an exposure analysis as for the arthropod food web section was not useful for weeds. Indirect competitive effects are the most likely routes of hazard. With widespread production of Bt maize, the density of some stemborer species may decrease releasing weeds from stemborer damage. Based on the criteria used in the prioritization and decision support matrix (Table 5.7), two weed species, Sorghum spp. and Pennisetum spp., were selected for pre-release risk assessment, based on the hypothesis that stemborers have a significant impact in their fitness. For example, C. partellus has been found to utilize and be widespread on wild Sorghum spp. as alternative host plant (B. Le Ru, Nairobi, 2004, personal communication). Sorghum halepense (Johnson grass), also native to Africa, is a minor weed in maize in Kenya, but in other parts of the world it is among the most noxious weeds. Since Sorghum spp. are an important alternative host for stemborers, it is important to check if a possible decline in the abundance and densities of the stemborer complex in Bt maize and surrounding areas would result in an increased fitness of Sorghum weeds and, consequently, in a worsening of their weed status. The following research hypothesis was formulated:

15. Stemborer feeding reduces the fitness of Sorghum weeds. In the previous analysis, Striga spp. were identified as a priority test species for risk assessment (Muhammad and Underwood, Chapter 2, this volume). Striga spreads only through its seeds, which are moved primarily by humans, particularly in infested soil, contaminated crop fodder and crop seeds. The seeds are tiny and produced in large numbers. They will only germinate under favourable environmental conditions (e.g. correct moisture, temperature) and in the presence of a germination stimulus, usually a root exudate from the crop plant. This stimulant ensures that Striga, which stores minimal food reserves in the seed, does not germinate until it has received a signal that a host root is nearby. Various compounds have been identified as stimulants. Strigol is a major Striga seed germination stimulant in

PARASITIC WEEDS.

Environ. Risk. Chap 05

23/8/04

9:32

Page 155

Biodiversity and Non-target Impacts

155

maize and a minor one in Sorghum root exudates (Siame et al., 1993). However, there is no definite proof that one single signal compound or class of compounds induces germination of parasitic weeds in the field. In fact, strigol was shown to act as a germination stimulant in maize but was also found in plant root exudates of medicinal plants where no parasitism by Striga was reported (Yasuda et al., 2003). Once stimulated, the Striga hermonthica germination strategy is based on the regulation of ethylene biosynthesis (Sugimoto et al., 2003). The Striga radicle grows directly towards the source of stimulant (Chang et al., 1986). The radicle must penetrate the maize seedling within 3–5 days or the Striga seedling will perish. Once penetration has occurred, an internal feeding structure (haustorium) is formed, and the parasite establishes host xylem connections. The host photosynthate is then diverted to the developing parasite, which also utilizes the host root system for water and mineral uptake. The relative success of each stage of the life cycle governs the volume of seed production. Any of the steps in this process could be altered by the expression of the Bt toxin or other unintended modifications, e.g. secondary metabolites. There is substantial variation among and between sorghum and maize cultivars with regard to tolerance to Striga (Showemimo, 2002); therefore, it will also be important to evaluate cultivar effects in Bt maize. From this, the following research questions regarding possible Striga and Bt maize interactions were identified that should be addressed in pre-release risk assessment: 16. Production of biologically active germination stimulant of Striga is higher in Bt maize than in non-transgenic cultivars. 17. Bt maize improves the fitness of the Striga species compared to the nontransgenic varieties. 18. Bt maize reduces the effectiveness of the Striga-control component of the push–pull strategy. In Table 5.10, the above-ground maize-associated flora prioritized for nontarget testing are summarized. Soil ecosystem functions A healthy sustainable soil ecosystem requires the maintenance and stability of biodiversity. One potential hazard of Bt maize may be a shift towards particular microbial groups, such as ligninase organisms. Changes in the dynamics of specific functional groups or marker species may relate to changes in functional microbial diversity. Organic matter is decomposed by microbial exo-enzymes and Bt toxins may affect the functioning of such enzymes. Nitrogen-fixing bacteria are not directly associated with maize plants but maize roots are in close association with both legume roots and free-living nitrogen fixers in the soil. So any Bt toxin released from transgenic plants may adversely affect the functioning of nitrogen-fixing bacteria. Similarly, as the nitrogen-fixing bacteria rely on carbon flow from the plants, their functioning rate may be changed when the transgenic plants provide an altered carbon input (pleiotropic effect).

Environ. Risk. Chap 05

23/8/04

9:32

Page 156

156

A.N.E. Birch et al.

Table 5.10. Summary of above-ground maize-associated flora prioritized for testing for nontarget impacts of Bt maize in Kenya. Selected species

Guild

Rationale

Hazard identification

Sorghum spp. Competitive weed

Significant weed competitor in maize

Striga spp.

Parasite causing significant crop losses

Alternative host for stemborer – implications for resistance management; currently under biocontrol through stemborer? Possible exacerbation of negative impact on maize yield through effects of Bt maize

Parasitic weed

Protocol developed? Yes

Yes

Mycorrhizal associations involve an intimate intracellular physical association between the fungus and plant cells, so intracellular fungal structural development may be affected by the Bt toxins. In mycorrhizal associations, the fungus obtains carbon, for energy, from the plant. Since the plant’s physiology has been changed to produce the Bt toxin, the pattern of carbon flow through the root may also be changed. This could have consequences for the growth and function of the fungus, and subsequently on its ability to provide nutrients to the plant. Functional dynamics in plant–soil ecosystems are highly variable, temporally and spatially, and affected by many factors, such as temperature, rainfall and cultivation. Consequently, in order to be able to detect any effects of transgenic crops, comprehensive baseline data will be required to determine natural variation within comparable soil ecosystems. So the following hazard scenarios and hypotheses for soil ecosystem functions were identified: 19. Inputs from Bt maize will alter microbial genetic diversity in soils, compared to non-transgenic maize. 20. Organic matter decomposition rates will be slower in soils from Bt maize than in soils from non-transgenic maize. 21. How long does the soil-incorporated Bt toxin from roots and degrading plant materials persist in soils? How much of it persists? How does this vary between different Kenyan soil types? 22. Nitrogen fixation rates, both of nodulated intercropped plants and freeliving bacteria in the soil, will be reduced in the presence of the Bt toxins. 23. Is the rate of conversion of plant residues to inorganic nitrogen for plant uptake affected by Bt maize residues in soils compared to non-Bt maize residues? 24. Mycorrhizal fungal development, colonization and subsequent function will be reduced in Bt maize plants compared to non-transgenic maize plants. Table 5.11 lists the soil ecosystem functions prioritized for the impact assessment of Bt maize in Kenya.

Environ. Risk. Chap 05

23/8/04

9:32

Page 157

Biodiversity and Non-target Impacts

157

Table 5.11. Summary of the soil ecosystem functions prioritized for testing non-target impacts of Bt maize in Kenya. Selected function

Rationale

Carbon decomposition

Essential for supplying energy and nitrogen for microbial function; plant organic input dependent

Hazard identification

Changed root exudate/residue composition and production in Bt maize may cause changes in carbon decomposition, affecting all other microbial functions Cellulose Essential for supplying Changed root exudate/residue breakdown energy and nitrogen for composition and production in microbial function; plant Bt maize may cause changes in organic input dependent cellulose decomposition, affecting all other microbial functions Nitrogen fixation Essential for supplying Changed root exudate/residue nitrogen for plant production composition and production in in low-input maize; plant Bt maize may cause changes in organic input dependent rhizobia stimulation, affecting nitrogen supply for maize Phosphorus and Essential for supplying Changes in Bt maize may cause micronutrient nutrients for plant changes in mycorrhizal uptake production in low-input maize associations and activity, on poor soils; plant organic affecting nutrient supply for input dependent maize, particularly phosphate

Protocol developed? Yes

Yes

Yes

Yes

Findings of Step 4 The exposure analysis from Step 3 was used to develop ecological hazard scenarios and relevant research hypotheses. Ecological hazard scenarios were identified for most of the highest priority species and functions. Some could not be completed in the workshop (e.g. spider mites, locusts) but should be included in a full risk assessment for Bt maize in Kenya. Some relevant risk-related hypotheses for risk assessment of Bt maize in Kenya were formulated.

Step 5: Methods and protocols Research methods and experiments were developed for most of the potential hazards identified in step 4. This section: 1. Poses the research hypothesis. 2. Provides a brief rationale on why this hypothesis is important for risk assessment. 3. Describes the proposed protocols. 4. Defines the measured endpoints (parameters to be measured). 5. Describes the appropriate statistical analysis.

Environ. Risk. Chap 05

10/9/04

14:28

Page 158

158

A.N.E. Birch et al.

Non-target maize herbivores NON-TARGET LEPIDOPTERAN HERBIVORES (E.G. H. ARMIGERA, SPODOPTERA SPP.). For sublethally affected lepidopteran herbivore species, it is important to find out whether or not their pest status will change on Bt maize compared to the nontransgenic maize varieties. It is also important to understand their role and capacity to induce food-chain effects at higher trophic levels.

Research question 1: What is the level of susceptibility of non-target lepidopteran herbivores to the expressed Bt toxin in transgenic maize? What biological parameters are affected? Rationale: Information on this question will help to predict whether either one of the two species will become a more damaging or less damaging pest on Bt maize. Proposed protocols: (i) Laboratory feeding trials using artificial diet containing different concentrations of microbially produced, purified Bt toxin of the same type(s) that is (are) expressed in the transgenic Bt maize, compared to artificial diet without toxin. Five replications per trial using at least 30 first-instar larvae should be carried out. Larvae are allowed to develop through adult eclosion, and at least two consecutive generations are to be tested. (ii) Laboratory feeding trials using greenhouse or field-grown Bt maize and nontransgenic control plants. Ideally, the 30 first-instar larvae are allowed to feed on the intact, growing plants by caging them on leaves or cobs. If no living plants can be used, cut leaf or cob material must be replaced every day, or every other day, to simulate the continuous exposure that the larvae would experience on the plant itself. For all tests with Spodoptera spp., use maize plants when they still have ‘soft’ leaf material, as this is the preferred tissue of this species. Measured endpoints: Mortality rate, growth rate, stage-specific developmental time, fecundity (indirect measures using pupal weight is acceptable), adult emergence, sex ratio. The first three parameters should be recorded for each larval stage and for the entire immature life stage (hatch of larvae until eclosion of adults). Statistical analysis: Life history parameters are derived for each treatment, as well as determining LD50. Biological parameters are subjected to analysis of variance (ANOVA) to test for differences between Bt and non-Bt treatments. Insect pest survival/mortality curves can be compared between treatments using the lifetest procedure. Research question 2: Does oviposition preference of H. armigera and Spodoptera spp. differ between Bt maize and non-transgenic maize? Rationale: H. armigera is more closely associated with maize as a host plant, at least when cotton is not readily available, while Spodoptera spp. are more polyphagous and can survive on a broader range of host plants. Differential oviposition behaviour will significantly influence the pest status of each species. Proposed protocols: Plants should be grown in 2–4-l pots, with one seed/plant per pot. At least 75 pots per maize line, i.e. Bt and non-transgenic maize cultivars, should be included. Both choice and no-choice experiments are recommended. For all tests with H. armigera, start the experiments when

Environ. Risk. Chap 05

23/8/04

9:32

Biodiversity and Non-target Impacts

Page 159

159

the plants are at the soft dough stage, which is when H. armigera is commonly found on maize plants (see Bernal and Sétamou, 2003, for details). Choice experiment: At least eight plants per replication and cage are arranged in a circle of 1.5 m in diameter, with transgenic and non-transgenic plants alternating. Gravid H. armigera or Spodoptera spp. females are released into the cage at the rate of one female per plant, i.e. eight females for eight plants. Plants must be dissected to record the number of eggs per plant within 48 h after release of the females. No-choice experiment: The same experimental set-up and procedure as above should be used, except only one maize-line will be provided at a time, i.e. either only Bt maize or only non-transgenic maize. Measured endpoints: Number of transgenic and non-transgenic plants on which eggs were deposited, number of eggs per plant and/or cob. Statistical analysis: Non-parametric test to discriminate between oviposition of non-target lepidopteran species on Bt maize vs. non-transgenic maize. Research question 3: How are Carpophilus spp. affected by consumption of the frass of H. armigera feeding on Bt maize and how does this influence the distribution of A. flavus and other fungi and their associated aflatoxin levels? Rationale: Carpophilus spp. beetles feed primarily on the frass of lepidopteran herbivores such as H. armigera and are therefore likely to be exposed to Bt. Aflatoxin infections of maize are a serious health threat for subsistence farmers in Kenya. Thus, it is crucial to understand whether the ability of Carpophilus spp. to transmit and distribute A. flavus fungi will be altered on Bt maize. Proposed protocols: H. armigera should be reared on both Bt maize and the control. Both maize lines are to be kept in separate cages. Cobs are to be collected and disinfected using a 0.5% hypochlorite solution for 5 min and thoroughly rinsed afterwards. Infest the cobs with second-instar H. armigera and wait until the larvae reach the fourth instar (purple frass produced). Fresh frass from the colonies should be collected every day and 5 mg of frass weighed into a clean capsule. Five Carpophilus beetles should be released into each of the experimental containers. Carpophilus have to be collected from the field, as it is difficult to raise them in the laboratory. Mortality should be recorded daily. At least 500 individuals should be tested per treatment. One week after infestation, a subsample of 20 individuals per treatment is randomly selected and tested for mobility. Each individual is placed in a large Petri dish and its behaviour observed for 5 min. The time spent in movement and direction of movement is recorded (e.g. video tracking). Measured endpoints: Mortality rate, growth rate, stage-specific developmental time, fecundity (indirect measures using pupal weight is acceptable), adult emergence, sex ratio. Mortality rate, growth rate and stagespecific developmental time should be recorded for each larval stage and for the entire immature life stage (hatch of larvae until eclosion of adults). Statistical analysis: Life-test analysis will be used to compare the survival curve of individuals fed Bt vs. non-Bt maize. A t-test will be used to compare the time in movement, and the time to 50% or 95% mortality in each treatment.

Environ. Risk. Chap 05

23/8/04

9:32

Page 160

160

A.N.E. Birch et al.

For studies on A. flavus infection and aflatoxin levels, the cobs must be first infested with H. armigera eggs at the milk stage (at least two per cob). When the grain reaches the soft dough stage, each line of maize (Bt or non-Bt) will be subjected to four different treatments. The four treatments comprise a combination of: (i) artificial infestation with five Carpophilus sp. adults per cob or no infestation; and (ii) inoculation with A. flavus spores or no inoculation, in a factorial experimental design. At maturity of maize, five cobs per treatment have to be collected and, from each cob, ten randomly selected kernels taken. The kernels are shelled, mixed and ground for fungal and aflatoxin analysis. At least five replications of 100 kernels are plated per treatment in each maize line and three subsamples analysed for aflatoxin content. Measured endpoints: The percentage of kernels infected with A. flavus and other fungi, as well as the amount of aflatoxin B levels in samples, recorded per treatment and maize type. Statistical analysis: The percentage of kernels infected with A. flavus is arcsine-transformed and aflatoxin levels are log-transformed before analysis. A factorial ANOVA is used to evaluate the effects of both Bt maize and presence of Carpophilus sp. on the increase of A. flavus infection and aflatoxin levels. The increase in aflatoxin levels due to the presence of Carpophilus sp. is calculated via linear contrasts, and a t-test used to compare this increase in Bt vs. non-Bt maize. Pollinating and pollen-feeding insects on maize Research question 4: Will individual fitness parameters of pollenfeeding/pollinating species be affected by Bt maize compared to the nontransgenic varieties? If yes, how? Rationale: Possible long-term exposure of pollen feeders/pollinator species has been identified. Protocols are proposed at the individual level for honeybees and for other pollen-feeding species (Forficulidae and Coccinellidae). (i) Proposed protocols at organismal level: Assessment of anti-metabolic effects of the toxin on immature honeybees can be conducted in laboratory experiments that allow complete control of food source and quality, toxin concentration and distribution, and an optimal environmental situation. Bioassays can be conducted on artificial diet that allows honeybee larvae to develop until maturity (Brødsgaard et al., 1998). This provides a reliable method for testing the effects of Bt toxins at the concentrations that one can expect larvae to be exposed to in the field, after pollen incorporation into diet in hives. Larvae are reared in sterile tissue culture multi-wells and grafted daily to new wells with food; handling is reduced to one time per day. Toxins can be mixed into the standard food at the appropriate concentrations. Bee larvae and pupae are monitored once daily until adult emergence. Measured endpoints: Larval development time is calculated at the ‘spinning’ stage of the fifth larval instar (when larvae stop feeding and defecate to begin pupation) and at adult emergence. The newly emerged adults are also weighed to investigate possible differences in body mass. It is advisable to use

Environ. Risk. Chap 05

23/8/04

9:32

Biodiversity and Non-target Impacts

Page 161

161

first instars to start the experiment, as young individuals are likely to be the most sensitive ones to a large array of proteins. (ii) Proposed protocols for other pollen feeding species (Forficulidae and Coccinellidae): Laboratory feeding bioassays using pollen from transgenic Bt maize and the appropriate control are recommended. Life history parameters should be measured for the whole lifespan, including immature and adult life stages, for two consecutive generations. Measured endpoints: Mortality rate, growth rate, stage-specific developmental time, fecundity (indirect measures using pupal weight is acceptable), adult emergence, sex ratio. The first three parameters should be recorded for each larval stage and for the entire immature life stage (hatch of larvae until eclosion of adults). Statistical analysis: Life history parameters are derived for each treatment. Biological parameters are subjected to ANOVA to test for differences between Bt and non-Bt treatments. Insect pest survival/mortality curves can be compared between treatments using the lifetest procedure, as well as determining LD50. Research question 5: Will attractiveness of Bt maize for honeybees differ from that of non-transgenic maize varieties? Rationale: Unintended effects of genetic modification might affect the volatile production or flower appearance in plants. The attractiveness of transgenic plants compared to their controls can be measured by observing the pre-foraging behaviour of honeybees. Proposed protocols: These choice experiments can be done in semi-field conditions (shaded house). In this case, foragers are given the choice between Bt and non-Bt plants. The experiments can also be conducted either at the individual or colony level. For the first case, individual foragers are released in the arena. For the second case, free-living colonies in shaded houses are given the choice of transgenic and control flowering maize plants. Colonies should be checked for infections (e.g. Nosema) before starting experiments. Measured endpoints: For the individual-level experiments, the individual forager bee is observed and her choices tracked and recorded. For the colonylevel experiments, visual observations of flower visits have to be made and the number of foragers arriving in a given time span recorded. In both cases, a baseline record of the actual number of flowering plants per unit area is needed. Statistical analyses: To analyse behavioural data, non-parametric statistical tests should be considered, as these data are generally not normally distributed. A minimum of 100 individuals should be observed. More-detailed behavioural observations might also be conducted, in order to clarify differences during foraging activity, depending on the outcome of the above-proposed experiments. Research question 6: Will pollinator effectiveness differ on Bt maize from that on non-transgenic varieties?

Environ. Risk. Chap 05

23/8/04

9:32

Page 162

162

A.N.E. Birch et al.

While insect pollination is not very relevant for maize, it is crucial for the reproductive success and yield of other crops. It is possible to monitor pollination effectiveness both directly (via visual observations) or indirectly (like fruit set). We therefore recommend these activities to be part of the impact assessment for other transgenic crop plants. Research question 7: Will colony development be affected by Bt maize compared to the non-transgenic varieties? Rationale: Colony development can be studied. Several indicators of colony response to environmental stressors such as Bt maize can be measured by hive examination (as opposed to adult lifespan). Proposed protocol at colony level: Smaller size colonies can be prepared to reduce the work needed for experimental observations, allowing more replications to increase the reliability of the results. Colony development can be monitored in closed systems or under semi-field conditions, using a shaded house with a field crop of appropriate size and covered with meshing nets. Each treatment should be in a different unit so that the bees are not to be given the choice between Bt maize and isogenic food. In this way, colonies are only exposed to the appropriate treatment. Maize plants should be the main food source within the shaded houses. We assume the colonies have the same origin and basically the same age and composition. Measured endpoints: Number of combs occupied by bees and the degree (%) of brooding and occupation of each single comb must be recorded. Larval mortality through regular examination of young larvae from all combs using a dissecting microscope (at least twice per week) should be carried out. Queen fertility and fecundity should also be monitored to assess potential sublethal effects. Pupal weight should be recorded at least on a weekly base. Colony activity should also be monitored by recording (automatically or even manually) the number of flights in and out of the hive in a time span. Statistical analyses: ANOVA will furnish good estimates of variation sources, provided adequate numbers of observations are made. Natural enemies of maize herbivores Whilst we acknowledge that the workshop assessment was not comprehensive across all natural enemies, the focus for the following research protocols is on the highest priority parasitoids in maize in Kenya, Trichogramma spp., C. flavipes and C. sesamiae. However, if modified to accommodate differences in biology, these methods can serve as examples for other non-target natural enemies. Protocols to test the selected egg-parasitoid Trichogramma species Research question 8: Are there differences in suitability of host eggs from nontarget lepidoptera (H. armigera and Spodoptera spp.) compared with host eggs from target Lepidoptera (C. partellus and B. fusca) to Trichogramma spp. parasitoids on Bt maize or non-transgenic maize? Rationale: Whilst eggs are not directly exposed, adults that produce them have developed from larvae that consumed the host plant tissues (leaves,

Environ. Risk. Chap 05

23/8/04

9:32

Biodiversity and Non-target Impacts

Page 163

163

stems) potentially throughout their entire larval stage. Although it is not known whether Bt toxin is transferred to the egg, other unintended physiological changes in the transgenic plants might affect host egg suitability for the eggparasitoid Trichogramma spp. that merits this question. Trichogramma spp. are very important natural enemies and any change in their biocontrol capacity would have significant implications for pest control, in particular in subsistence maize production, where farmers have few or no other options. Research should investigate if the eggs resulting from a generation of non-target pest larvae that was highly exposed to the host plant differ in their suitability for parasitism (oviposition and initial development) and progeny emergence (complete development to adults). PROTOCOLS FOR NO-CHOICE EXPERIMENTS. In laboratory no-choice experiments, parasitization success of the candidate trichogrammatid species can be measured by presenting them with eggs or egg masses from the different pest lepidopteran hosts either fed as larvae on a diet including Bt or on a Btfree diet. The larvae can be fed on either artificial diet plus Bt maize vs. artificial diet, or on conventional maize plus Bt maize vs. conventional maize. Cut batches of about 40 eggs from the C. partellus eggs (laid on pieces of card), H. armigera eggs (laid on pieces of cloth), and B. fusca and Spodoptera spp. eggs (normally laid on strips of translucent paper). The eggs are viable for parasitism for up to 72 h, so must be used for the experiment within 24 h after oviposition. Present the egg batches of each lepidopteran species separately to 1-day-old mated, individual females of each of the chosen trichogrammatid species in 7020 mm glass vials. Plug the vials immediately with cotton wool, and provide minute streaks of honey (diluted by adding distilled water and gelatine in the ratio of 66:33:1 respectively) on photocopier paper as diet. After 24 h, transfer the eggs to fresh vials and discard the females. The experiment should be repeated with 20 different female individuals (as replicates) for each trichogrammatid species and for each pest Lepidoptera being tested. Keep the vials until the exposed eggs produce either larvae or parasitoids. Remove hatched larvae daily to prevent them from feeding on the parasitized eggs. Keep emerged parasitoids for a maximum of 2 weeks before data is taken. Measured endpoints: (i) Percentage of eggs parasitized (from a sample of c. 20 egg masses per plot received after 3 days of exposure); (ii) number of progeny produced (number of female and male adults in the progeny produced; to be estimated from a sub-sample of 50/100 adults, if there are more progeny). Statistical analysis: Data are subjected to ANOVA (Proc GLM, SAS Institute, 1988) followed by the Student Newman–Keul mean separation test when the ANOVA was significant (P<0.05). The proportion of blackened (i.e. parasitized) eggs is arcsine transformed before being subjected to the ANOVA (Sokal and Rohlf, 1995).

PROPOSED

Research question 9: Are there differences in oviposition preferences of Trichogramma spp. parasitoids when given the choice between non-target vs. target host eggs on Bt or non-transgenic maize?

Environ. Risk. Chap 05

23/8/04

9:32

Page 164

164

A.N.E. Birch et al.

This experiment tests host preference of the different Trichogramma spp. for non-target lepidopteran species (H. armigera, Spodoptera spp.) fed on Bt as larvae or not fed on Bt, compared to C. partellus/B. fusca as the target pest hosts partly fed on Bt as larvae (i.e. sublethally affected) or not fed on Bt. Use only egg masses from first-generation adults whose larvae were raised on Bt or non-transgenic maize plants. This research question could be explored either in a small plot cage exposure experiment (22-m fine mesh cages), or, alternatively, without field release permission, it can be carried out in closed, controlled environment systems (greenhouse or climate chamber). Dual choice experiments in the laboratory can be conducted by presenting pairs of egg batches to single mated female Trichogramma individuals in vials, as in the no-choice experiment above. The rearing host should be a storage lepidopteran (e.g. Ephestia sp.) to avoid any influence of rearing host on subsequent parasitoid behaviour. Measured endpoints: (i) Percentage of eggs parasitized (from a sample of c. 20 egg masses per plot recovered after 3 days of exposure); (ii) number of progeny produced (number of female and male adults in the progeny produced; to be estimated from a sub-sample of 50/100 adults, if there are more progeny); (iii) sex ratio of progeny from each female adult. Statistical analysis: Data are subjected to ANOVA (Proc GLM, SAS Institute, 1988) followed by the Student Newman–Keul mean separation test when the ANOVA was significant (P<0.05). The proportion of blackened (i.e. parasitized) eggs is arcsine transformed before being subjected to the ANOVA (Sokal and Rohlf, 1995).

PROPOSED PROTOCOLS FOR DUAL CHOICE EXPERIMENTS.

Research question 10: Will effects of host plant constituents/volatiles on behaviour and performance of adult Trichogramma spp. parasitoids differ between Bt maize and non-transgenic cultivars under semi-field conditions? Rationale: Trichogramma spp. egg parasitoids are known to be more sensitive/responsive to the attributes of host plant than to that of the host insect (pest) (Smith, 1996). It is useful to check if any subtle changes in the physical and/or chemical attributes of the foliage of the transgenic host plant could potentially affect the behaviour and field performance of the adult parasitoids, mainly in terms of parasitism of host eggs. Proposed protocols: This research question could be explored either in small plot cage exposure experiment, or, alternatively, without field release permission, it can be carried out in closed, controlled environment systems (greenhouse or climate chamber). Test plots of Bt maize and non-Bt maize are planted in close vicinity and should be caged using preferably a fine mesh size and cage size of 22 m. The Trichogramma spp. egg parasitoid should be a representative species/strain of Trichogramma relevant to the ecology of the target stemborer species. One-day-old mated females should be used. To provide for host eggs, adult moths of the species to be compared are released into the caged plots for 24 h to oviposit on the maize plants. They must be removed after that. There should be at least 100 freshly laid eggs in each cage. Subsequently, the freshly laid eggs will be exposed as hosts for 24 h to Trichogramma spp. females released into the cages for recording the field

Environ. Risk. Chap 05

23/8/04

9:32

Page 165

Biodiversity and Non-target Impacts

165

performance. The release ratio is one female Trichogramma to 100 freshly laid eggs. They are released for 24 h and then removed. After 6 days, the parasitized eggs become black (nigrescence), and the ratio of parasitized to unparasitized eggs gives the parasitism rate. Note: some stemborer eggs can host two Trichogramma individuals. Counting the eclosion (exit) holes on the surface of the parasitized eggs after 12 days will give adult number. Measured endpoints: Percentage parasitized eggs (mean for each exposure) and progeny produced (mean per adult parasitoid released) (see above). Protocols to test the selected larval parasitoid species C. flavipes and C. sesamiae Research question 11: Does Bt maize alter host-finding behaviour of C. flavipes? Rationale: Host-finding behaviour of C. flavipes is complex and influenced by many factors. One important factor is volatiles emitted by the host plant; here this is maize. Both the quantity and the composition of the emitted volatiles influence Cotesia spp. host finding (Steinberg et al., 1993; Hoballah et al., 2002). These are emitted from infested or uninfested plants, respectively. In Kenya, Nwanze and Nwilene (1999) found plant volatile-mediated differential reaction of parasitoid activity to various sorghum genotypes. Proposed protocols: A colony of C. flavipes collected from C. partellus in the coastal zone of Kenya should be used. After parasitization, maintain stemborers on artificial diet (Ochieng et al., 1985; Onyango and OchiengOdero, 1994) at 25°C, 65–70% relative humidity (RH) and 12:12 (L:D) photoperiod. Collect parasitoid cocoons in glass vials and keep in a clean Perspex cage until emergence of adults. Provide adult parasitoids with a 20% honey/water solution as diet. Use 1-day-old, mated, naive female parasitoids in all experiments. Similarly, C. partellus collected from maize fields near the Kenyan coast should be used for the experiments. Rear larvae of C. partellus on an artificial diet, as described by Ochieng et al. (1985). Prior to any experiment, remove stemborer larvae from the artificial diet and feed them on fresh maize stems for 48 h. Grow maize Bt-expressing lines and control lines in 20-l plastic buckets in a nursery. Potted plants can be kept under large field cages (222 m) covered with fine mesh (400 µm) netting to protect them from insect attack. Additional maize plants are grown in the field. Behaviour assays can be conducted using a Y-tube olfactometer (described by Steinberg et al., 1992). The odour sources are placed in two Perspex chambers (3030120 cm) sufficiently large to accommodate whole plants (2–3 months old). One of the square ends of the chamber is left open. For the system to be airtight, the open end of the each box is placed over the test material, which stands in water held in a plastic basin. The open end is submerged 15 cm below the meniscus. The two chambers are connected to the arms of the Y-tube with Tygon tubing from the top of the chambers. An inlet, through which clean air enters the chamber, is drilled 30 cm from the bottom of the chamber on one side. A vacuum pump (Cole–Parmer Air-Cadet) draws and pushes air through the closed system.

Environ. Risk. Chap 05

166

23/8/04

9:32

Page 166

A.N.E. Birch et al.

Air is pushed through activated charcoal filter into the two chambers and drawn into the Y-shaped glass tubing of the olfactometer. The airflow is set at 2.5 l/min for each arm. Parasitoids are released individually in the stem of the Y-tube and allowed 5 min to choose one of the arms. When the parasitoid remains more than 15 s beyond the finishing line (4 cm past the intersection), it is recorded as a choice. The number of non-responding females is also recorded. The connections of the odour source chambers to the arms of the olfactometer has to be reversed after testing five insects, to rule out the effect of asymmetrical bias in the olfactometer or its surroundings. A cream-white curtain can be used to separate the experimental area from the surroundings. Tests are typically conducted at 23–26°C, 65–75% RH and light intensity of 350–450 lux. Time of day for tests should also be standardized. All tests should be replicated at least three times with 20 parasitoids per replicate. Different series of experiments can be conducted with this set-up using uninfested or stemborer-infested maize plants. 1. Response to volatiles from uninfested maize plants: Uninfested 8–10-weekold Bt maize and control maize plants should be used in the experiments. All plant genotypes should be tested in single and dual choice experiments, in the Y-tube olfactometer. One series of experiments can be conducted to determine the attractiveness of odours from uninfested plants. Bt maize and control maize are tested in single-choice experiments. Individual parasitoids are given a choice between odour from potted Bt or control maize and air drawn over a pot with soil only. Another series of tests can be conducted to determine the parasitoid’s preference for odours from different uninfested maize genotypes (Bt maize vs. control maize). Plants are placed in chambers connected to both arms of the olfactometer. An approximately similar biomass of plants must be used for each arm. Attractiveness of volatiles from potted Bt maize is compared with attractiveness of volatiles from potted control maize. Field materials must be checked for damage before each test and dissected after a test to confirm that it was uninfested. 2. Response to volatiles from plants infested by one stemborer species, C. partellus: In the Y-tube olfactometer, parasitoids are given a choice between odours from Bt maize or control maize plants infested with C. partellus. Each one is tested against the control (uninfested plants of the same type). Infest potted maize plants with two fourth instars by boring two holes in the maize stem (1 cm deep) with a 4-mm cork borer and placing one larva in each hole. Larvae are allowed to feed overnight and tests are conducted 18–20 h after infestation. Make holes in control plants, without introducing any larvae. Transfer the plants into the chambers connected to the Y-tube in the pot in which they were grown for 30 min before observations are made (to allow time for volatiles to be released in the chamber). Pot-grown plants are used in this test. A second series of tests can be conducted to determine the parasitoids preference between Bt or control maize plants infested by different stemborer species. The following treatments can be tested:

Environ. Risk. Chap 05

23/8/04

9:32

Biodiversity and Non-target Impacts

Page 167

167

1. Bt maize infested with C. partellus larvae vs. Bt maize infested with Chilo orichalcociliellus larvae (repeat for control maize). 2. Bt maize infested with C. partellus larvae vs. Bt maize infested with S. calamistis larvae (repeat for control maize). 3. Bt maize infested with C. partellus larvae vs. Bt maize infested with B. fusca larvae (repeat for control maize). Measured endpoints: The number of parasitoids selecting each odour zone is recorded. Statistical analysis: Data is analysed via Friedman’s one-way ANOVA by ranks, to determine orientation and choice preference of parasitoids. Research question 12: Is the host suitability of C. partellus for C. flavipes altered if it feeds on Bt maize? Rationale: Intended and unintended physiological changes in the transgenic plants might affect host egg suitability for the larval parasitoid C. flavipes that merits exploring this question. C. flavipes is a significant natural enemy of C. partellus in the Kenyan coast and Dry Mid-altitude zones, and any change in its biocontrol capacity would have significant implications for pest control in subsistence maize production where options are limited. Research should investigate whether or not non-target lepidopteran larvae fed with Bt maize differ in their suitability for parasitism (oviposition and initial development) and progeny emergence (complete development to adults) from non-transgenic control maize. Proposed protocol: Rearing procedures for C. flavipes are described by Ngi-Song et al. (1996). Insects are kept at 25°C, 65–70% RH and at a photoperiod of 12:12 h (L:D). Adult parasitoids are provided a 20% honey/water solution as diet. One-day-old naive female parasitoids should be used in the experiments. C. partellus were reared on a medium developed by Ochieng et al. (1985). Prior to experiments, stemborer larvae are fed fresh maize stems for 48 h. (i) Host suitability and parasitoid efficacy trials in the laboratory: The suitability of C. partellus previously fed on Bt maize or control maize for the development of C. flavipes is assessed by exposing single fourth-instar stemborer larvae to individual adult parasitoids as described by Ngi-Song et al. (1995). Parasitized larvae are then reared on artificial diet in an incubator at 28±1°C, 30–55% RH and a 12:12 (L:D) photoperiod, and inspected daily for mortality or parasitoid emergence (Onyango and Ochieng-Odero, 1994). Measured endpoints: Brood size, sex ratio, developmental time of progeny (egg to adult), mortality inside the host, female weight and the proportion of hosts from which parasitoids emerged and produced cocoons are recorded. This is compared to the life history factors of parasitoids from the controls as host plants for the herbivorous host of the parasitoid. Statistical analyses: Data are subjected to ANOVA (Proc GLM, SAS Institute, 1988) followed by the Student Newman–Keul mean separation test when the ANOVA was significant (P<0.05). Insect counts are square root transformed before being subjected to analysis (Sokal and Rohlf, 1995). The

Environ. Risk. Chap 05

168

23/8/04

9:32

Page 168

A.N.E. Birch et al.

data on sex ratio, mortality and proportion of hosts from which parasitoids emerged and produced cocoons are arcsine transformed before being subjected to the ANOVA (Sokal and Rohlf, 1995). (ii) Host suitability and parasitoid efficacy trials in the field: Experimental plots are planted in a randomized block design with four blocks of two plots each. Each plot consists of 15 rows; rows are 5 m long separated by 0.3-m alleys. The between-plot distance is 5 m within each block and blocks are located 5 m apart from each other. Walk-in cages (2.52.52.5 m) are randomly placed in each plot at the reproductive stage of maize. The field cages are constructed of Nitex nylon mesh (362 µm mesh size) with a metal frame and 1-m long zippered opening. Each cage contains 24 maize plants (either Bt maize or control maize). Ten plants are randomly selected in each cage and artificially infested with two fourth-instar C. partellus larvae, previously reared on artificial diet (with or without Bt toxin at a defined concentration). Two holes are drilled in each stalk, one at 0.5 m above the ground and the other at 1.5 m, and one larva placed in each hole. Holes are plugged with glass vials to prevent larvae escaping. Two days after infestation, the glass vials are removed and ten pairs of 1-day-old C. flavipes are released into each cage. Infested plants are cut 1 week after parasitoid release and dissected to recover stemborer larvae. These are then individually reared on artificial diet (no Bt toxin) until formation of parasitoid cocoons or moth pupae (Sétamou et al., 2002). Measured endpoints: The proportions of recovered and parasitized larvae, numbers of parasitoid cocoons, and numbers and sex ratio of parasitoid adults per host will be recorded. Statistical analysis: Log-likelihood ratio tests are used to compare the percentages of larvae recovered and parasitized, and parasitoid sex ratios between Bt and non-Bt maize. Mean brood sizes per larva recorded in both maize lines will be compared via t-tests. Research question 13: Does C. sesamiae have a different interaction with Wolbachia when parasitizing herbivores fed on Bt maize or on its nontransgenic control varieties? Rationale: Wolbachia are known to affect the phenotype of the carrier through several mechanisms, including male killing, cytoplasmic incompatibility, induction of parthenogenesis, feminization and altered fertility. The C. sesamiae populations exposed to Bt or control maize via a bitrophic or tritrophic route (see above) are tested for differences in the presence of Wolbachia infections. Proposed protocol: Extract DNA from five individuals of each population that have been stored in 99% ethanol since emergence. Rehydrate individuals by shaking in 100 µl of TE buffer for 1 h. Remove the TE buffer and homogenize the parasitoids using a pestle in an Eppendorf tube (500 µl). Next, add 200 µl of Tris buffer (10 mM Tris, 2.5 mM MgCl2 and 50 mM KCl, pH=7.6), 10 µl of 20% sodium dodecyl sulphate and 200 µl of phenol and zirconium beads of 0.1 and 0.5 mm diameter. Shake the Eppendorf tube in a Bead Beater homogenizer for 3 min and then centrifuge for 10 min. Transfer

Environ. Risk. Chap 05

23/8/04

9:32

Biodiversity and Non-target Impacts

Page 169

169

the supernatant into a 1.5-ml Eppendorf tube and add 200 µl of chloroform/isoamyl alcohol (24:1). Shake vigorously for 30 s and then centrifuge the vial again for 5 min (14,000 rpm). Transfer the supernatant to a new 1.5-ml Eppendorf tube. Precipitate the DNA with 400 µl of absolute ethanol and 20 µl of 3 M NaAc (pH=4.8) and keep it overnight in a 20°C freezer. Next, centrifuge the Eppendorf tubes for 20 min (14,000 rpm) at 8°C, and wash the precipitate using ice-cold 70% ethanol. Vacuum dry, then resuspend the precipitate in 50 µl of water. Use this solution as the template in the PCR. Concentration and quality of the DNA templates for PCR are checked using the conserved D-extension of the 28S rDNA (Campbell et al., 1994). In subsequent tests for the presence of Wolbachia in the wasp DNA, approximately equal amounts of template DNA are used. Several primer combinations are used, including the Holden FTSZ primers and the A and B primer combinations of Werren et al. (1995). PCR programmes as described by Werren et al. (1995) are used. After PCR, the samples are run out on 1% agarose gels to determine if a product of the appropriate size had been obtained. The above plant-based protocol can also be adapted to test the hostmediated effects of purified Bt toxin, incorporated into the diet of C. partellus at a range of sublethal concentrations. Measured endpoints: Development time, longevity, egg load, fecundity and larval survival of C. sesamiae are recorded to discover any change in the influence of Wolbachia on the life table statistics of C. sesamiae, mediated via the herbivorous host feeding on maize. Statistical analysis: Data on development time, longevity and fecundity are subjected to ANOVA (Proc GLM, SAS Institute, 1988), followed by the Student Newman–Keul mean separation test if the ANOVA was significant (P<0.05). Insect counts are square-root transformed before being subjected to analysis (Sokal and Rohlf, 1995). The proportion of hosts from which parasitoids emerged and produced cocoons are arcsine transformed before being subjected to ANOVA (Sokal and Rohlf, 1995). Research question 14: Do adult C. flavipes feed on maize pollen and guttation fluids under field conditions? Rationale: This has been identified as a gap of knowledge in the exposure analysis. If C. flavipes adults feed on pollen and guttation fluids of Bt maize they will be bitrophically exposed, as well as being tritrophically exposed. Research protocol: Field observations on the feeding behaviour of adult C. flavipes should be made in representative climatic zones (e.g. coastal, highlands) and maize agroecosystems (e.g. small-scale, intercropped, push–pull, intensive large-scale). Maize-associated flora The analysis of potential hazards suggests that the effects of Bt maize on germination, competitive ability, weed fitness and compatibility with the push–pull system should be assessed.

Environ. Risk. Chap 05

170

23/8/04

9:32

Page 170

A.N.E. Birch et al.

Research question 15: What is the fitness impact of stemborer feeding on Sorghum weeds? Rationale: Stemborers also feed on Sorghum weeds where they act as biocontrol agents. If overall density levels of stemborers are reduced because of large-scale production of Bt maize this could lead to a release of Sorghum weeds from a main natural enemy. Proposed protocol: Fitness of selected Sorghum weeds with and without various levels of stemborer feeding pressure should be tested under field conditions. Sorghum weed plants will be cultivated in field plots containing at least 100 plants in a non-competitive field arrangement (depending on soil fertility). The plots should be arranged in a completely randomized block design with a minimum of four replications. This experiment should be repeated at different locations in the target region for Bt maize release(s) where Sorghum weeds occur. Varying levels of feeding pressure are achieved by releasing different numbers of stemborer larvae per plant – varying from 0 to 20 larvae per plant. Wild hosts of stemborers do not support high populations compared with maize, so the numbers released have been increased to counter natural mortality on suboptimal host plant species (Ofomata et al., 2000). The trials should be conducted with the locally most abundant, economically more important stemborer species known to feed on Sorghum weeds. The control treatment will be Sorghum weed plots that are protected from stemborer feeding by spraying repeatedly the plants with Bt insecticides. Systemic insecticides could be an alternative, but they will also kill other herbivore species that are not affected by the Bt toxin. Measured endpoints: It is suggested that three main plant features are analysed: (i) plant growth; (ii) seed and rhizome production; and (iii) propagule viability. Fifteen plants should be collected every 21 days and analysed for: plant height, shoot number, dry matter weight of leaves, stems, rhizomes, and roots, leaf area, rhizome length and bud number, seed number and seed viability. Statistical analyses: Seed viability should be analysed every 6 months for 2 years after each experiment. Plant growth characteristics should be analysed by using selected growth parameters; e.g. leaf area index (LAI), absolute growth rate (AGR), relative growth rate (RGR), net assimilation rate (NAR), leaf area ratio (LAR), specific leaf area (SLA). Seed number, rhizome length and bud number should be analysed using ANOVA. Seed viability should be fitted (time vs. germination index) and the parallelism in the curves be tested by a suitable statistical model. Research question 16: Does Striga germination stimulant production and its biological activity differ in Bt cultivars compared to the isolines and the most commonly used cultivars? Rationale: Striga seeds will not germinate in the absence of a chemical stimulant. These stimulants can be classified as host root exudate, non-host root exudate, natural leachates/compounds and synthetic germination stimulants. Proposed protocols: Protocols were taken from the IITA Manual (1997), developed by Striga scientists.

Environ. Risk. Chap 05

23/8/04

9:32

Biodiversity and Non-target Impacts

Page 171

171

Collection of germination stimulants: To prepare stimulants of maize root exudates, the simplest way is to grow plants in a ‘double pot’ system. This requires two tapered pots of the same size. Perforate the bottom of the one pot and fit it into a second unperforated pot. Plant Bt and non-Bt varieties in sand in the upper pot. Water will percolate through the sand and the holes in the upper pot to collect in the bottom pot. After growing the seedlings for 7–14 days, discard the water in the bottom pot, refill with 25 ml of water and collect the subsequent leachate. This root exudate should be refrigerated and can be used to stimulate Striga seed germination. In addition to this root exudate, synthetic germination stimulants (any of the strigol analogs) can be used as checks, since properly conditioned Striga seed (see below) should be stimulated to germinate with these compounds. A 10 mg/l solution in water should be used but these compounds first have to be dissolved in a small volume of acetone since they are water insoluble. After dissolving them in acetone, the compounds can be mixed with the final desired volume of water. (i) Determination of Striga seed germination rate using collected germination stimulants: Striga seeds will not germinate in the absence of a chemical stimulant or unless they have been conditioned (see IITA Manual, 1997, for detailed protocols). After harvest of Striga seeds, there is a period of 4–6 months when the seed are truly dormant and generally cannot be conditioned to germinate. After this time period, it takes 7–21 days of exposure to moisture to precondition the seeds so that they will respond to germination stimulant. Striga seed surface must be disinfected to eliminate microbial contamination by washing the seeds in a 1% sodium hypochlorite solution. Place the surface-disinfected Striga seeds into 30 ml of sterile water in a sterile Petri dish. Stir the seeds to force them to sink. Put the Petri dish in a dark place for 14 days. During this period, the water must be exchanged every 2 days. After this period, spread the seeds on moist filter paper in another Petri dish. A small paintbrush works well to spread the seeds evenly over the surface of the filter paper. After spreading the seeds, add enough stimulant to barely cover all of the seed. After 48 h, screen the Petri dish for germinated Striga seeds. Measured endpoints: Number of germinated seeds is counted. From this, the germination percentage is determined and analysed after arcsine transformation. (ii) Testing the biological activity of the collected stimulants with maize plants: Use 2-l pots that are perforated in the bottom and cover the bottom hole with filter paper to avoid losses of Striga seeds. Fill with clean topsoil to a level of 8 cm below the desired soil surface. Sprinkle the Striga seeds on to the soil. Use between 1500–2000 seeds per g of soil (the seed germination must be checked before this trial, protocol see above). Then add the remaining 8 cm of soil. Irrigate the pots carefully after sowing, in order to avoid the Striga seeds moving down the soil profile. The next irrigation must be carried out after 4 days. Leave 3 more days without sowing maize. Use between 30 and 50 maize plants of each tested cultivar (at minimum Bt maize and the isogenic control). Sow two to three maize seeds per pot, and 7–10 days after planting thin to leave only one healthy maize plant per pot. Add 20 ml of NPK

Environ. Risk. Chap 05

172

23/8/04

9:32

Page 172

A.N.E. Birch et al.

(4–14–8) fertilizer diluted in water per pot on to the soil surface at a rate of 60–90 kg/ha. Allow the crop to grow for 5–6 weeks with minimal watering every 2 days. When the first Striga emerges, carefully remove the maize plant from the pot and wash the soil off the roots. The roots must not be squeezed because the Striga seedling can easily become dislodged from the maize roots. To account for potentially dislodged Striga seedlings, the remaining soil in the pot and the soil washed off should be sieved and Striga seedlings counted. This experiment can also be conducted as a small field trial if a permit for Bt maize field release is granted. Measured endpoints: Numbers of Striga seedlings per pot and maize plant. An alternative testing method for Striga is described in Khan et al. (2002). Research question 17: Could Bt maize improve or decrease the fitness of Striga spp. compared to the non-transgenic varieties? Rationale: Above-ground Striga mortality and seed production capacity is influenced by Striga plant vigour. Therefore, it is useful to measure how vigorous the plant is at various stages and how this is affected by any difference between Bt maize and the control maize. For example, Sorghum produces larger Striga plants with more flowers and seed capsules than maize. Seed production is a direct result of development of the seed-bearing stalks and organs, which are influenced by a range of environmental conditions as well as by biological factors, such as host plant resistance. Proposed protocols: The pot experiment or small field trial described in detail above will be repeated, except maize plants will not be removed for Striga seedling counting at first emergence of Striga. Instead, maize and Striga will be allowed to continue their growth until Striga flowers and sets seeds. Striga plants do not all mature at the same time. They should be collected just as they mature which requires frequent, if possible daily, checking. The most common vigour and fitness measurements include biomass of Striga, and height. Striga can be harvested, dried and weighed at any stage of interest. Typically, this is done 10 weeks after maize planting. Striga height is highly correlated with biomass and capsule number per plant. Harvested capsules should be left drying for 10–14 days. After that, the capsules should be gently threshed by tapping the floral heads on plastic sheeting to force seed shed. After threshing, the material should be screened by passing it through a 150–250-µm sieve. Sieving helps remove most of the plant trash in the seed lot. Measured endpoints: Numbers of flowers and produced capsules should be recorded for every Striga stalk. Total biomass can be weighed fresh and after drying for seed threshing. Collected seeds should be weighed to determine total seed production per Striga plant and per maize plant and pot. Research question 18: Will Bt maize interfere with the Striga-control component of the push–pull strategy? Rationale: One component of the push–pull strategy is to suppress Striga infestation. This is largely dependent on the combination of host and non-host plants used. In order to optimize the push–pull strategy and minimize the risk of adverse interference, Bt maize should be tested in the field for its suitability and

Environ. Risk. Chap 05

23/8/04

9:32

Page 173

Biodiversity and Non-target Impacts

173

susceptibility to Striga infestation within the context of the other plants employed in the push–pull strategy. Proposed protocols: Regular field trials for optimal use of various plants within the push–pull strategy should include one treatment using Bt maize instead of the conventional maize used. Everything else should be conducted according to regular protocols for testing and optimizing that strategy (Khan et al., 2002). Measured endpoints and statistical analyses: According to standard protocols for testing the push–pull strategy (Khan et al., 2002). Soil ecosystem functions Approaches to soil ecosystem research have tended to look at measures of soil activity such as respiration, nitrification and ammonification, and then to extrapolate these data to known microbial groups associated with that function. More recently, DNA-based molecular tools have opened up new opportunities to look at the dynamics of microbial species. The study of the complexity of soil macroorganisms and their trophic interactions also presents difficulties, but several techniques such as assemblage studies enable adequate surveys. Changes in microbial community structure caused by transgenic plant inputs are best defined by using molecular techniques such as PCR-denaturing or temperature gradient gel electrophoresis (DGGE/TGGE), terminal restriction fragment length polymorphism (T-RFLP) and mRNA/cDNA studies. These studies are complex and resource consuming, and should be highly targeted. The methods are rapidly improving and will provide powerful tools to precisely assess changes in species composition, both type and numbers, and in functional activity. However, many of these methods are still not routine and require considerable experience to run properly: no one molecular test will give a definitive answer. We therefore consider the best approach is to combine a specific test, either function or species-specific, with a more general microbial activity assessment such as basal or SIR respiration. In all cases, protocol design requires a detailed appraisal of all elements in the system, and constant revision to incorporate novel techniques. Soil microbial communities Research question 19: Inputs from Bt maize will alter the genetic microbial diversity in soils compared to non-transgenic maize. Rationale: Plant inputs drive the soil microbial community. As different plants differ in their system inputs the resultant microbial communities may differ. Bt proteins could either have a toxic effect or be a novel food source. Changes in soil microbial diversity may adversely affect functional dynamics in the plant–soil system. Proposed protocols: Molecular techniques can describe the microbial population as a whole, constituent parts (e.g. fungi) or functional groups (e.g. nitrifiers). Procedures can be modified to provide more options by using specific primers. These methods overcome all the previous problems inherent in using culture techniques for identification.

Environ. Risk. Chap 05

174

23/8/04

9:32

Page 174

A.N.E. Birch et al.

PCR-based community-profiling techniques produce information on the structure of the whole community or specifically targeted portions. The primer set used determines specificity (Pennanen et al., 2004). Microbial diversity is defined by PCR amplification of whole soil DNA then separation of the DNA bands down a gel by applying techniques such as DGGE/TGGE and/or TRFLP. With DGGE/TGGE, separation of the 16S fragments is achieved on GC content; whereas with T-RFLP, after endonuclease restriction of the fragment, the labelled terminal fragment is observed in a way analogous to amplified fragment length polymorphisms (AFLPs). These banding patterns can then be compared between and across samples to provide a description of the soil microbial community of each soil (Pennanen et al., 2004). Measured endpoint: The degree of difference is compared to the null hypothesis: there is no difference in biodiversity in soil in which Bt maize has been grown compared with soil in which conventional maize has been grown. Research question 20: Organic matter decomposition rates will be slower in soils from Bt maize than in soils from non-transgenic maize? Rationale: The primary step in the functioning of soil ecosystems involves the breakdown of plant residues to provide energy. Microbial functional dynamics are dependent on the quality of this input. As the physiology of the GM plant will be different to that of the non-transformed plant, these inputs will change and so it is necessary to compare function rates between soils receiving GM and non-GM plant inputs. At the same time, nitrogen-containing compounds required for many other microbial functions are released. Approaches to this can be divided into: 1. Assessments of the effects of freshly incorporated, and therefore relatively intact, plant material, stems, leaves and roots. 2. Studying the effects of somewhat degraded material that has been incorporated into the soil-organic matter component of the soil. 3. Effects of the Bt toxins, which are proteins, both as a potential substrate and as a possible toxin. Proposed protocols: 1. Decomposition rates can be estimated: (i) Via loss of organic content from leaf litter confined in nylon bags, with a 1-mm mesh. Bury 10 g of dried leaves from both transgenic and nontransgenic Bt maize in the soil. Collect samples every 20 days, dry at 105°C and ash at 600°C for 4 h. Calculate the loss of organic matter using the equation described by Santos and Whitford (1981). (ii) By hydrolysis of cellulose and its derivatives. Cellulolytic enzymes hydrolyse -D (1–4) glucoside bonds. Estimate cellulase activity by using carboxymethyl-cellulose as a substrate for soils from GM and non-GM plants with five replicates, following the protocol of Gilligan and Reese (1954), and by determining rates of reducing sugar formation using the procedures described in Miller (1959). 2. Microbial biomass (microbial population size) is responsive to inputs, particularly plant residues, exudates, etc.:

Environ. Risk. Chap 05

23/8/04

9:32

Biodiversity and Non-target Impacts

Page 175

175

(i) Biomass determinations, to show any effects of the Bt crop on the ‘standing crop’ of microbes, by fumigation extraction (FE) method (Ritz et al., 1992); soil samples, 10 g fresh weight, fumigated with ethanol-free chloroform for 18 h at 25°C, control samples stored at 5°C. After removal of the chloroform, soils are extracted with 40 ml 1 M KCl (soil/solution 1:4) on a roller bed (Wheatley et al., 1989) followed by centrifugation (2500 g for 10 min) and filtration through a Whatman GF/F filter (Ritz et al., 1997). Analyse filtrate for dissolved organic carbon in a persulphate reagent using a segmented flow auto-analyser. Measured endpoint: Biomass carbon is calculated as the concentration of carbon in the fumigated samples minus that in the appropriate control soil. This gives an indication of the microbial biomass, and so whether the various inputs to the soil from Bt maize affect microbial population development differently from the isoline. 3. Microbial activity in soils is limited by availability of energy, and influenced by other inputs such as proteins (Wheatley et al., 2001): (i) Substrate-induced respiration (SIR) indicates the effects of inputs on microbial activity. Use five replicates of each soil from comparable plants, repeated at least five times across the field. Add glucose to 10 g fresh weight of soil to a final glucose concentration of 2000 g C/g. Incubate in sealed containers for 4 h at 20°C (Anderson and Domsch, 1978). Measure concentrations of CO2 in the headspace of the incubation bottles using a gas chromatograph fitted with a thermal conductivity detector. Calculate rates of C release per g of soil and compare. Measured endpoint: Microbial activities in soil are driven by the fixed carbon inputs from plants. These measurements will demonstrate whether inputs from the Bt maize affect microbial activity levels in a different way to the isoline. Research question 21: How long does the soil-incorporated Bt toxin persist in soils? How much persists, and does this vary between different Kenyan soil types? Rationale: Exposure to the Bt toxins, exuded from roots and released during the decay of Bt maize plant material may adversely affect soil microand macroorganism dynamics. Bt toxins persist in soils, so continual effects must be determined. Proposed protocols: Soil samples should be collected from each soil type in each agroecological zone of interest, and brought to the research facility for experimentation. (i) Bioassays with applied Bt toxins: Spray a range of different concentrations of the microbially produced, activated Bt toxin on to the soils. Five replicates should be used, and untreated soil must be included as a control. Take soil samples at 0, 2, 4, 8, 16, 32 and 64 weeks after adding the Bt toxin. Extract and analyse soil samples by ELISA, as described in Zwahlen et al. (2003).

Environ. Risk. Chap 05

176

23/8/04

9:32

Page 176

A.N.E. Birch et al.

Measured endpoints: Decline in Bt toxin concentrations over the 64 weeks, therefore the persistence of the toxin, which has implications in cropping management systems. (ii) Bioassays on soil in which plants have grown: Grow transgenic and non-transformed maize plants to maturity in greenhouse and field plots, of the different soil types. Any possible effects of the production of the inactive Bt toxin by indigenous Bacillus thuringiensis in the soils will be revealed in the control soils. Mix different quantities of soil into a standard artificial, i.e. material from a non-transformed plant, diet for lepidopteran larvae. Allow the larvae of C. partellus, the target pest of Bt maize in Kenya, to feed for 4–5 days. Measured endpoints: Determine the LD50 (Saxena and Stotzky, 2001b). This can then be assessed against the long-term implications of cropping Bttransformed plants on the macrofauna. If the laboratory or greenhouse trials indicate that the plant material and Bt toxins persist in Kenyan soils, extended, field-realistic testing periods should be considered. If field release trials can be conducted, soil samples should be monitored for at least 3 years of continuous cultivation of Bt maize. Research question 22: Will nitrogen fixation rates, both of nodulated intercropping plants and free-living bacteria in the soil, be reduced by the presence of the Bt toxins? Rationale: Nitrogen fixation, the conversion of atmospheric nitrogen to ammonium, by a variety of organisms, is a major function of the nitrogen cycle and a major provider of nitrogen to plants in most agricultural systems. Nitrogen provision from nitrogen fixation is of particular importance in Kenyan intercropping maize systems. Both the nitrogen-fixing bacteria in root nodules and other free-living microorganisms in the soil are driven by energy supplies from the plant, and influenced by signal compounds produced by the roots. So it is important that the effects of GM plants on N-fixation dynamics are compared to non-transformed plants, to assess any negative impact. Protocol: Remove nodulated roots and attached soil of legumes grown in association with transgenic and non-transformed plants. Assess nitrogen fixation rates of nodules and field soil with the acetylene-reduction method; incubate replicated nodules and soil samples in an atmosphere amended with acetylene, then analyse headspace samples on a gas-chromatograph fitted with a flame-ionization detector. Measurement endpoint: Relative rates of nitrogen fixation can be deduced from the rates of ethylene formation, to indicate whether the amounts of nitrogen made available to the maize in intercropping systems is affected in any way by the Bt transgene. Research question 23: Is the rate of conversion of plant residues to inorganic nitrogen for plant uptake adversely affected by Bt maize residues in soils compared to non-Bt maize residues? Rationale: The recycling of inorganic N from plant residues for further crop production is a very important function of the soil ecosystem. This process

Environ. Risk. Chap 05

23/8/04

9:32

Biodiversity and Non-target Impacts

Page 177

177

involves the interaction of a vast array of microflora and higher trophic groups of organisms, such as the micro- and mesofauna, in the soil food web. These interactions over a range of trophic levels are particularly relevant in the first step in the nitrogen cycle, ammonification, but because of the complexity of this function, and time and expertise limitations, it was decided to progress this at future workshops and to include a comprehensive review and suggested protocols in the final workshop report. Nitrification is the next step in the N cycle, during which the immobile ammonium form is converted to the mobile nitrate form, with consequent implications for environmental management. As previous work (Wheatley et al., 2001) had shown nitrification rates to be particularly susceptible to changes in carbon inputs, particularly proteins, this function was chosen for study. Nitrification assays: Potential nitrification rates can be estimated by the method of Belser and Mays (1980). Amend 25 g of each soil sample with (NH4)2SO4 and NaClO3 solutions to give a final concentration of 4 and 15 mM respectively. Incubate at 20°C for 48 h. Measurement endpoint: Nitrification rates are calculated from the rate of accumulation of NO2–N over time. This will give a comparison of the relative impacts of Bt maize on the system, in particular the rate at which nitrogen becomes available to the plant. Research question 24: Will mycorrhizal fungal development, colonization and subsequent function be reduced in Bt maize plants compared to non-transgenic maize plants? Rationale: The fungal structures of mycorrhizas function within the root cells of the Bt maize plant, and so will be continually exposed to the Bt toxins. Therefore, it is conceivable that the Bt toxin or any other physico-chemical alteration in the transgenic plant may adversely affect the efficacy of the mycorrhizal association. Proposed protocols: Remove roots and associated soil from transgenic and non-transformed maize plants grown in the field, greenhouse and nursery. Carefully separate intact root systems and attached hyphae from the soil by immersing in a tub of water and gently agitating. After washing, keep the root samples moist in plastic bags. If necessary, refrigerate (approximately 5°C) for several days. Process and preserve roots in 50% ethanol in tightly sealed plastic vials for transport and storage (Brundrett et al., 1996). The washed roots are cut into small sections, mixed and subsamples removed, and weighed. Make the roots translucent by autoclaving for 15–20 min at 121°C in 10% KOH (w/v), then stained with Chlorazol Black E (CBE) in a lacto-glycerol solution (Brundrett et al., 1984) in an autoclave for 15 min at 121°C, or by standing in the solution for several days. Roots can also be stained with trypan blue (Bevege, 1969; Philips and Hayman, 1970). Roots can be destained again by immersing them in 50% glycerol for several days prior to observation, to removes excess stain. Root colonization and root length are measured simultaneously with mycorrhizal colonization by a gridline intersection procedure (Giovanetti and Mosse, 1980) in which roots are randomly dispersed in a 9-cm diameter Petri plate with grid lines. Intersections between gridlines

Environ. Risk. Chap 05

23/8/04

9:32

178

Page 178

A.N.E. Birch et al.

and roots designated as either mycorrhizal or non-mycorrhizal are quantified with a dissecting microscope. Roots can also be mounted on slides and viewed with a compound microscope (McGonicle et al., 1990). Measured endpoint: Examination of this intimate association between plant and fungus will show if fungal development has been affected by continuous exposure to relatively high levels of the Bt toxins, or to changes in nutrient supply from the roots. Any changes in mycorrhizal function have implications for plant uptake of phosphorus and micronutrients, particularly in low-input systems. Findings of Step 5 Scientifically rigorous laboratory-based testing methodologies and protocols were designed that addressed most of the hazards and research questions identified for the selected species and functions in the previous step 4. There are several specific functions involved in the release and transformation of nitrogen for plant growth in soil. All of these are responsive to the gross inputs of plant material, as a source of C and N. The choice of which function occurs is also influenced by substrate availability and type. The rates of these functions can also be affected by relatively small changes in the inputs of specific compounds (Wheatley et al., 2001). It is highly recommended that the maize plant residues be allowed to decompose for a long time in all of the incubation experiments – essentially until they are completely decomposed. Zwahlen et al. (2003) reported that Bt toxin was detectable in plant material as long as it was present, in any state of decomposition. Hence, all soil organisms degrading such material will be continuously exposed to the Bt toxin. A full risk assessment programme for Bt maize in Kenya should include the development of protocols and methodologies for the remaining species that could not be addressed at this workshop.

References Abera, T.H., Sithanantham, S., Ogol, C.K.P.O., Monje, J.C. and Zebitz, C.P.W. (2000) Survey for native trichogrammatid species in Kenya. In: Gazzoni, D.L. (ed.) Proceedings of XXI International Congress of Entomology, Iguassu, Brazil, 20–26 August, 2000. Anderson, J.P.E. and Domsch, K.H. (1978) A physiological method for the quantitative measurement of microbial biomass in soils. Soil Biology and Biochemistry 10, 215–221. Andow, D. and Hilbeck, A. (2004) Science-based risk assessment for non-target effects of transgenic crops. Bioscience 54, 637–649. Bagayoko, M., George, E., Romheld, V. and Buerkert, A.B. (2000) Effects of mycorrhizae and phosphorus on growth and nutrient uptake of millet, cowpea and sorghum in West African soil. Journal of Agricultural Science 135, 399–407. Belser, L.W. and Mays, E.L. (1980) Specific inhibition of nitrite oxidation by chlorate and its use in assessing nitrification in soils and sediments. Applied and Environmental Microbiology 39, 505–510.

Environ. Risk. Chap 05

10/9/04

14:28

Biodiversity and Non-target Impacts

Page 179

179

Bernal, J.S. and Sétamou, M. (2003) Fortuitous antixenosis in transgenic sugarcane. Antibiosis-expressing cultivar is refractory to ovipositing herbivore pests. Environmental Entomology 32, 886–894. Bernal, C.C., Aguda, R.M. and Cohen, M. (2002) Effect of rice lines transformed with Bacillus thuringensis toxin genes on the brown planthopper and its predator Cyrtorhinus lividipennis. Entomologia Experimentalis et Applicata 102, 21–28. Bevege, D.I. (1969) A rapid technique for clearing tannins and staining intact roots for detection of mycorrhizas caused by endogene species, and some records of infection in Australasian plants. Transactions of the British Mycological Society 51, 808–810. Birch, A.N.E., Geoghegan, I.E., Majerus, M.E.N., McNicol, J.W., Hackett, C.A., Gatehouse, A.M.R. and Gatehouse, J. (1999) Tritrophic interactions involving pest aphids, predatory 2-spot ladybirds and transgenic potatoes expressing snowdrop lectin for aphid resistance. Molecular Breeding 5, 75–83. Birch, A.N.E., Geoghegan, I.E., Griffiths, D.W. and McNicol, J.W. (2002) The effect of genetic transformations for pest resistance on foliar solanidine-based glycoalkaloids of potato. Annals of Applied Biology 140, 143–149. Bonhof, M.J., Overholt, W.A., Van Huis, A. and Polaszek, A. (1997) Natural enemies of cereal stemborers in East Africa: a review. Insect Science and its Application 17, 19–35. Boonman, J.G. (1992) Tropical grasses from the East African Centre of Biodiversity. Biotechnology and Development Monitor 11, 9–10. Bosque-Pérez, N.A. (2000) Eight decades of maize streak virus research. Virus Research 71, 107–121. Bouwmeester, H.J., Matusova, R., Zhongkui, S. and Beale, M.H. (2003) Secondary metabolite signalling in host-parasitic plant interactions. Current Opinion in Plant Biology 6, 358–364. Brødsgaard, C.J., Ritter, W. and Hansen, H. (1998) Response of in vitro reared honeybee larvae to various doses of Paenibacillus larvae spores. Apidologie 29, 569–578. Brundrett, M.C., Piche, Y. and Peterson, R.L. (1984) New method for observing the morphology of vesicular-arbuscular mycorrhizae. Canadian Journal of Botany 62, 2128–2134. Brundrett, M.C., Ashwath, N. and Jasper, D.A. (1996) Mycorrhizas in the Kakadu region of tropical Australia 1. Propagules of mycorrhizal fungi and soil properties in natural habitats. Plant and Soil 184, 159–171. Campbell, B.C., Steffen, J.D. and Werren, J.H. (1994) Phylogeny of the Nasonia species complex inferred from an international transcribed spacer and 28S rDNA. Insect Molecular Biology 2, 225–237. CBD (1992) Convention on Biological Diversity: Convention Text, www.biodiv.org/ convention/articles.asp (accessed 25 November, 2003). CBD (2002) Cartagena Protocol on Biosafety: Status of Ratification and Entry into Force, www.biodiv.org/biosafety/signinglist.asp (accessed October 2002). Chang, M., Netzly, D.H., Butler, L.G. and Lynn, D.G. (1986) Chemical regulation of distance: characterization of the first natural host germination stimulant for Striga asiatica. American Chemical Society 108, 7858–7860. Curtis, T.P., Sloan, W.T. and Scannell, J.W. (2002) Estimating prokaryotic diversity and its limits. Proceedings of the National Academy of Sciences USA 99, 10494–10499. Dutton, A., Klein, H., Romeis, J. and Bigler, F. (2002) Uptake of Bt-toxin by herbivores feeding on transgenic maize and consequences for the predator Chrysoperla carnea. Ecological Entomology 27, 441–447.

Environ. Risk. Chap 05

180

23/8/04

9:32

Page 180

A.N.E. Birch et al.

Fries, L.L.M., Pacovsky, R.S., Safir, G.R. and Kaminski, J. (2000) Phosphorus effect on phosphatase activity in endomycorrhizal maize. Physiologia Plantarum 103, 162–171. Gilligan, W. and Reese, E.T. (1954) Evidence for multiple components in microbial cellulases. Canadian Journal of Microbiology 1, 90–107. Giovanetti, M.E. and Mosse, B. (1980) An evaluation of techniques for investigating vesicular arbuscular mycorrhizal infection in roots. New Phytologist 84, 489–500. Guang, L.Q. and Oloo, G.W. (1990) Host preference studies on Trichogramma sp. nr. mwanzai Schulten and Feijen (Hymenoptera: Trichogrammatidae) in Kenya. Insect Science and its Application 11, 757–763. Head, G., Brown, C.R., Groth, M.E. and Duan, J.J. (2001) Cry1Ab protein levels in phytophagous insects feeding on transgenic corn: implications for secondary exposure risk assessment. Entomologia Experimentalis et Applicata 99, 37–45. Hilbeck, A. (2002) Transgenic host plant resistance and non-target effects. In: Letourneau, D.K. and Burrows, B.E. (eds) Genetically Engineered Organisms: Assessing Environmental and Human Health Effects. CRC Press, New York, pp. 167–185. Hilbeck, A., Baumgartner, M., Fried, P.M. and Bigler, F. (1998) Effects of transgenic Bacillus thuringiensis corn-fed prey on mortality and development time of immature Chrysoperla carnea (Neuroptera: Chrysopidae). Environmental Entomology 27, 480–487. Hilbeck, A., Moar, W.J., Pusztai-Carey, M., Filippini, A. and Bigler, F. (1999) Preymediated effects of Cry1Ab toxin and protoxin and Cry2A protoxin on the predator Chrysoperla carnea. Entomologia Experimentalis et Applicata 91, 305–316. Hoballah, M.E. and Turlings, T.C.J. (2001) Experimental evidence that plants under caterpillar attack may benefit from attracting parasitoids. Evolutionary Ecology Research 3, 553–565. Hoballah, M.E., Tamo, C. and Turlings, T.C. (2002) Differential attractiveness of induced odors emitted by eight maize varieties for the parasitoid Cotesia marginiventris: is quality or quantity important? Journal of Chemical Ecology 28, 951–968. Ibrahim, K.M. and Kabuye, C.H.S. (1987) An Illustrated Manual of Kenya Grasses. FAO, Rome. IITA (1997) Berner, D.K., Winslow, M.D., Awad, A.E., Cardwell, K.F., Mohan Raj, D.R. and Kim, S.K. (eds) Striga Research Methods – a Manual, 2nd edn. Prepared by the IITA Striga Research Group for The Pan African Striga Control Network (PASCON). International Institute of Tropical Agriculture PMB 5320, Ibadan, Nigeria. www.iita.org/info/striga.pdf (accessed November 2002). IRMA (2001) Appendix III. In: Insect Resistant Maize for Africa Annual Report 2000 KARI/CIMMYT IRMA Project. IRMA Project Document No. 4. KARI and CIMMYT (Kenya Agricultural Research Institute and International Maize and Wheat Improvement Centre), Nairobi. Khan, Z.R., Chiliswa, P., Ampong-Nyarko, K., Smart, L.E., Polaszek, A., Wandera, J. and Mulaa, M.A. (1997) Utilization of wild gramineous plants for management of cereal stemborers in Africa. Insect Science and its Application 17, 143–150. Khan, Z.R., Hassanali, A., Overholt, W.A., Khamis, T.M., Hooper, A.M., Pickett, J.A., Wadhams, L.J. and Woodcock, C.M. (2002) Control of Witchweed Striga hermonthica by intercropping with Desmodium spp. and the mechanism defined as allelopathic. Journal of Chemical Ecology 28, 1871–1885. Long, R.I., Corbett, A., Lamb, C., Roberg Horton, C., Chandler, J. and Stimman, M. (1998) Beneficial insects move from flowering plants to nearby crops. California Agriculture 52, 23–26. Losey, J.E., Rayor, L.S. and Carter, M.E. (1999) Transgenic pollen harms monarch larvae. Nature 399, 214.

Environ. Risk. Chap 05

23/8/04

9:32

Biodiversity and Non-target Impacts

Page 181

181

Marvier, M. (2001) Ecology of transgenic crops. American Scientist 89, 160–167. Mbapila, J.C. and Overholt, W.A. (1997) Oviposition, development and searching behaviour of Cotesia flavipes Cameron and Cotesia sesamiae (Cameron) (Hymenoptera: Braconidae) in relation to aestivating and non-aestivating Chilo spp. (Lepidoptera: Pyralidae) larvae in maize. African Entomology 5, 35–39. McDougall, S.J. and Mills, N.J. (1997) Influence of hosts, temperature and food sources on the longevity of Trichogramma platneri. Entomologia Experimentalis et Applicata 83, 195–203. McGonicle, T.P., Miller, M.H., Evans, D.G., Fairchild, G.L. and Swan, J.A. (1990) A new method which gives an objective measure of colonization of roots by vesiculararbuscular mycorrhizal fungi. New Phytologist 115, 495–501. McGonigle, T.P., Miller, M.H. and Young, D. (1999) Mycorrhizae, crop growth, and crop phosphorus nutrition in maize–soybean rotations given various tillage treatments. Plant and Soil 210, 33–42. Miller, G.L. (1959) Use of dinitrosalicylic acid reagent for determination of reducing sugar. Analytical Chemistry 31, 429. Mochiah, M.B., Ngi-Song, A.J., Overholt, W.A. and Stouthammer, R. (2002) Wolbachia infection in Cotesia sesamiae (Hymenoptera: Braconidae) causes cytoplasmic incompatibility: implications for biological control. Biological Control 25, 74–80. Munkvold, G.P., Hellmich, R.L. and Showers, W.B. (1997) Reduced Fusarium ear rot and symptomless infection in kernels of maize genetically engineered for European corn borer resistance. Phytopathology 87, 1071–1077. Ngi-Song, A., Overholt, W.A. and Ayertey, J.N. (1995) Suitability of African gramineous stemborers for development of Cotesia flavipes and C. sesamiae (Hymenoptera: Braconidae). Biological Control 24, 978–984. Ngi-Song, A.J., Overholt, W., Njagi, P.G.N., Dicke, M., Ayertey, J.N. and Lwande, W. (1996) Volatile infochemicals used in host and host habitat location by Cotesia flavipes Cameron and Cotesia sesamiae (Cameron) (Hymenoptera: Braconidae), larval parasitoids of stemborers on Graminae. Journal of Chemical Ecology 22, 307–323. Ngi-Song, A.J., Overholt, W.A. and Stouthammer, R. (1998) Suitability of Busseola fusca and Sesamia calamistis (Lepidoptera: Noctuidae) for the development of two populations of Cotesia sesamiae (Hymenoptera: Braconidae) in Kenya. Biological Control 12, 208–214. Ngi-Song, A.J., Kimani-Njogu, S. and Overholt, W.A. (2001) Multiple parasitism by Cotesia sesamiae and Cotesia flavipes on Busseola fusca. Biocontrol Science and Technology 11, 381–390. Nowakowski, J. and Morse, R. (1982) The behaviour of honeybees in sweet corn fields in New York State. American Bee Journal January, 13–16. Nwanze, K.F. and Nwilene, F.E. (1999) Interactions of host plant resistance and biological control of stemborers in sorghum. Insect Science and its Application 18, 261–266. Ochiel, G.R.S. (1989) Biology of Trichogramma species near exiguum Pinto & Platner (Hymenoptera – Trichogrammatidae) on some lepidopterous hosts in South Nyanza, Kenya. M. Phil. Thesis, International Centre of Insect Physiology and Ecology (ICIPE), Nairobi. Ochieng, R.S., Onyango, F.O. and Bungu, M.D.O. (1985) Improvement of techniques for mass culture of Chilo partellus (Swinhoe). Insect Science and its Application 6, 425–428. Ofomata, V.C., Overholt, W.A., Lux, S.A., Van Huis, A. and Egwuatu, R.I. (2000) Comparative studies on the fecundity, egg survival, larval feeding and development of Chilo partellus and Chilo orichalcociliellus (Lepidoptera: Crambidae) on five grasses. Annals of the Entomological Society of America 93, 492–499.

Environ. Risk. Chap 05

182

23/8/04

9:32

Page 182

A.N.E. Birch et al.

Oloo, G.W. (1989) The role of local natural enemies in population dynamics of Chilo partellus Swinh. under subsistence farming systems in Kenya. Insect Science and its Application 10, 243–251. Onyango, F.O. and Ochieng-Odero, J.P.R. (1994) Continuous rearing of the maize stemborer Busseola fusca on an artificial diet. Entomologia Experimentalis et Applicata 73, 139–144. Palm, C.J., Donegan, K.K., Harris, D. and Seidler, R. (1994) Quantification in soil of Bacillus thuringiensis var. kurstaki δ-endotoxin from transgenic plants. Molecular Ecology 3, 145–151. Palm, C.J., Schaller, D.L., Donegan, K.K. and Seidler, R.J. (1996) Persistence in soil of transgenic plant produced Bacillus thuringiensis var. kurstaki δ-endotoxin. Canadian Journal of Microbiology 42, 1258–1262. Parker, C.A., Trinick, M.J. and Chattel, D.J. (1977) Rhizobia as soil and rhizosphere inhabitants. In: Hardy, R.W.F. and Gibson, A.H. (eds) A Treatise on Dinitrogen Fixation. Section IV. John Wiley & Sons, New York, pp. 312–352. Pennanen, T., Caul, S., Daniell, T.J., Griffiths, B.S., Ritz, K. and Wheatley, R.E. (2004) Community-level responses of metabolically-active soil microorganisms to the quantity and quality of substrate inputs. Soil Biology & Biochemistry 36(5), 841–848. Philips, J.M. and Hayman, D.S. (1970) Improved procedures for cleaning roots and staining parasitic and vesicular-arbuscular mycorrhizal fungi for rapid assessment of infection. Transactions of the British Mycological Society 55, 158–160. Pivnick, A., Jarvis, B.J. and Slater, G.P. (1994) Identification of olfactory cues used in host-plant finding by diamondback moth, Plutella xylostella (Lepidoptera: Plutellidae). Journal of Chemical Ecology 20, 1407–1427. Potting, R.P.J. (1997) Evolutionary and applied aspects of the behavioural ecology of the stemborer parasitoid Cotesia flavipes. Insect Science and its Application 17, 109–118. Potting, R.P.J., Osae-Danso, F., Overholt, W.A. and Ngi-Song, A.J. (1993) Host selection in Cotesia flavipes, parasitoid of tropical stemborers. Proceedings of Experimental and Applied Entomology 4, 47–52. Potting, R.P.J., Vet, L.E.M. and Dicke, M. (1995) Host microhabitat location by stem borer parasitoid Cotesia flavipes: the role of herbivore volatiles and locally and systemically induced plant volatiles. Journal of Chemical Ecology 21, 525–539. Potting, R.P.J., Overholt, W.A., Danso, F.O. and Takasu, K. (1997) Foraging behaviour and life history of the stemborer parasitoid Cotesia flavipes (Hymenoptera: Braconidae). Journal of Insect Behaviour 10, 13–29. Rabinowitz, D. (1981) Seven forms of rarity. In: Synge, H. (ed.) Biological Aspects of Rare Plant Conservation. John Wiley & Sons, New York, pp. 205–217. Raps, A., Kehr, J., Gugerli, P., Moar, W.J., Bigler, F. and Hilbeck, A. (2001) Immunological analysis of phloem sap of Bacillus thuringiensis corn and of the non-target herbivore Rhopalosiphum padi (Homoptera: Aphidae) for presence of Cry1Ab. Molecular Ecology 10, 525–533. Ritz, K., Griffiths, B.S. and Wheatley, R.E. (1992) Soil microbial biomass and activity under a potato crop fertilised with N and without C. Biology and Fertility of Soils 12, 265–271. Ritz, K., Wheatley, R.E. and Griffiths, B.S. (1997) Effects of animal manure application and crop plants upon size and activity of soil microbial biomass under organically grown spring barley. Biology and Fertility of Soils 24, 372–377. Santos, P.F. and Whitford, W.G. (1981) The effects of microarthropods on litter decomposition in a Chihuahuan desert ecosystem. Ecology 62, 654–663.

Environ. Risk. Chap 05

23/8/04

9:32

Biodiversity and Non-target Impacts

Page 183

183

SAS Institute (1988) Users’ Guide: Statistics, Release 6.03 ed. SAS Institute, Cary, North Carolina. Saxena, D. and Stotzky, G. (2000) Insecticidal toxin from Bacillus thuringiensis is released from roots of transgenic Bt corn in vitro and in situ. FEMS Microbiology Ecology 33, 35–39. Saxena, D. and Stotzky, G. (2001a) Bt corn has a higher lignin content than non-Bt corn. American Journal of Botany 88, 1704–1706. Saxena, D. and Stotzky, G. (2001b) Bt toxin uptake from soil by plants. Nature Biotechnology 19, 199. Saxena, D., Flores, S. and Stotzky, G. (1999) Insecticidal toxin in root exudates from Bt corn. Nature 402, 480. Saxena, D., Flores, S. and Stotzky, G. (2002) Bt toxin is released in root exudates from 12 transgenic corn hybrids representing three transformation events. Soil Biology and Biochemistry 34, 133–137. Sétamou, M., Cardwell, K.F., Schultess, F. and Hell, K. (1998) Effect of insect damage to maize ears, with special reference to Mussidia nigrivenella (Lepidoptera: Pyralidae), on Aspergillus flavus (Deuteromycetes: Moniliales) infection and aflatoxin production on preharvest maize in the Republic of Benin. Journal of Economic Entomology 91, 433–438. Sétamou, M., Bernal, J.S., Legaspi, J.C. and Mirkov, T.E. (2002) Parasitism and location of sugarcane borer (Lepidoptera: Pyralidae) by Cotesia flavipes (Hymenoptera: Braconidae) on transgenic and conventional sugarcane. Environmental Entomology 31, 1219–1225. Showemimo, F.A. (2002) Evaluation of sorghum lines for Striga resistance. Tropical Agriculture 79, 237–240. Siame, B.A., Weerasuriya, Y., Wood, K., Ejeta, G. and Butler, L.G. (1993) Isolation of strigol, a germination stimulant for Striga asiatica, from host plants. Journal of Agricultural and Food Chemistry 41, 1486–1491. Sims, S.R. and Ream, J.E. (1997) Soil inactivation of the Bacillus thuringiensis subsp. kurstaki CryIIA insecticidal protein within transgenic cotton tissue: laboratory microcosm and field studies. Journal of Agricultural and Food Chemistry 45, 1502–1505. Sithanantham, S., Abera, T.H., Baumgaertner, J., Hassan, S.A., Lohr, B., Monje, J.C., Overholt, W.A., Paul, A.V.N., Wan, F.H. and Zebitz, C.P.W. (2001) Egg parasitoids for augmentative control of Lepidopteran vegetable pests in Africa: research status and needs. Insect Science and its Application 21, 189–205. Smith, S.M. (1996) Biological control with Trichogramma: advances, successes, and potential for their use. Annual Review of Entomology 41, 375–406. Sokal, R.R. and Rohlf, F.J. (1995) Biometry. The Principles and Practice of Statistics in Biological Research, 3rd edn. Freeman, New York. Songa, J., Overholt, W.A., Okello, R.O. and Mueke, J.M. (2002) Control of Lepidopteran stemborers in maize by indigenous parasitoids in semi-arid areas of Eastern Kenya. Biological Agriculture and Horticulture 20, 77–90. Steinberg, S., Dicke, M., Vet, L.E.M. and Wanningen, R. (1992) Response of the braconid Cotesia glomerata to volatile infochemicals: effect of bioassay set-up, parasitoid age, experience and barometric flux. Entomologia Experimentalis et Applicata 63, 163–175. Steinberg, S., Dicke, M. and Vet, L.E.M. (1993) Relative importance of infochemicals from first and second trophic level in long-range host location by larval parasitoid Cotesia glomerata. Journal of Chemical Ecology 19, 47–59.

Environ. Risk. Chap 05

184

23/8/04

9:32

Page 184

A.N.E. Birch et al.

Sugimoto, Y., Ali, A.M., Yabuta, S., Kinoshita, H., Inanaga, S. and Itai, A. (2003) Germination strategy of Striga hermonthica involves regulation of ethylene biosynthesis. Physiologia Plantarum 119, 137–145. Tapp, H. and Stotzky, G. (1995a) Dot blot enzyme-linked immunosorbent assay for monitoring the fate of insecticidal toxins from Bacillus thuringiensis in soil. Applied and Environmental Microbiology 61, 602–609. Tapp, H. and Stotzky, G. (1995b) Insecticidal activity of the toxins from Bacillus thuringiensis subspecies kurstaki and tenebrionis adsorbed and bound on pure and soil clays. Applied and Environmental Microbiology 61, 1786–1790. Tapp, H. and Stotzky, G. (1998) Persistence of the insecticidal toxin from Bacillus thuringiensis subsp. kurstaki in soil. Soil Biology and Biochemistry 30, 471–476. Tapp, H., Calamai, L. and Stotzky, G. (1994) Adsorption and binding of the insecticidal proteins from Bacillus thuringiensis subsp. kurstaki and subsp. tenebrionis on clay minerals. Soil Biology and Biochemistry 26, 663–679. Turlings, T.C.J., Loughrin, J.H., McCall, P.J., Rose, U.S.R., Lewis, W.J. and Tumlinson, J.H. (1995) How caterpillar-damaged plants protect themselves by attracting parasitic wasps. Proceedings of the National Academy of Sciences USA 92, 4169–4174. Vaissiere, B. and Vinson, S.B. (1994) Pollen morphology and its effect on pollen collection by honey bees, Apis mellifera L. (Hymenoptera: Apidae), with special reference to upland cotton, Gossypium hirsutum L. (Malvaceae). Grana. 33, 128–138. van den Berg, H. and Cock, M.J.W. (1993) African Bollworm and its Natural Enemies in Kenya. International Institute of Biological Control Kenya Station and Natural Resources Institute (Overseas Development Administration), CAB International, Wallingford, UK. van den Berg, H. and Cock, M.J.W. (1995) Natural control of Helicoverpa armigera in cotton: assessment of the role of predation. Biocontrol Science and Technology 5, 453–463. van den Berg, H., Cock, M.J.W., Oduor, G.I. and Onsongo, E.K. (1993) Incidence of Helicoverpa armigera (Lepidoptera: Noctuidae) and its natural enemies on smallholder crops in Kenya. Bulletin of Entomological Research 83, 321–328. Watmough, R.H. and Kfir, R. (1995) Predation on pupae of Helicoverpa armigera Hbn. (Lep., Noctuidae) and its relation to stemborer numbers in summer grain crops. Journal of Applied Entomology 119, 679–688. Werren, J.H. (1997) Biology of Wolbachia. Annual Review of Entomology 42, 587–609. Werren, J.H., Zhang, W. and Guo, L.R. (1995) Evolution and phylogeny of Wolbachia: reproductive parasites of arthropods. Proceedings of the Royal Society of London Series B 251, 55–71. Wheatley, R.E., MacDonald, R. and Smith, A.M. (1989) Extraction of nitrogen from soils. Biology and Fertility of Soils 8, 189–190. Wheatley, R.E., Ritz, K. and Griffiths, B.S. (1990) Microbial biomass and mineral N transformations in soil planted with barley, ryegrass, pea or turnip. Plant and Soil 127, 157–167. Wheatley, R.E., Griffiths, B.S. and Ritz, K. (1991) Variations in the rates of nitrification and denitrification during the growth of potatoes (Solanum tuberosum L.) in soil with different carbon inputs and the effect of these inputs on soil nitrogen and plant yield. Biology and Fertility of Soils 11, 157–162. Wheatley, R.E., Ritz, K., Crabb, D. and Caul, S. (2001) Temporal variations in potential nitrification dynamics related to differences in rates and types of carbon inputs. Soil Biology and Biochemistry 33, 2135–2144.

Environ. Risk. Chap 05

23/8/04

9:32

Biodiversity and Non-target Impacts

Page 185

185

Yasuda, N., Sugimoto, Y., Kato, M., Inanaga, S. and Yoneyama, K. (2003) (+)-Strigol, a witchweed seed germination stimulant from Menispermum dauricum root culture. Phytochemistry 62, 1115–1119. Zhou, G., Overholt, W.A. and Kimani-Njogu, S.W. (2003) Species richness and parasitism in an assemblage of parasitoids attacking maize stemborers in coastal Kenya. Ecological Entomology 28, 109–118. Zwahlen, C., Hilbeck, A., Gugerli, P. and Nentwig, W. (2003) Degradation of the Cry1Ab protein within transgenic Bacillus thuringiensis corn tissue in the field. Molecular Ecology 12, 765–775.

Environ. Risk. Chap 05

23/8/04

9:32

Page 186

Environ. Risk. Chap 06

23/8/04

6

9:32

Page 187

Gene Flow and its Consequences: a Case Study of Bt Maize in Kenya J. JOHNSTON, L. BLANCAS AND A. BOREM Corresponding author: J. Johnston, University of Minnesota, Plant Biology, 250 Biological Sciences Center, 1445 Gortner Avenue, St Paul, MN 55108, USA. E-mail: [email protected]

Introduction Gene flow between transgenic crops and related crops or free-living populations has the potential to lead to irreversible genetic and ecological change in the crop relatives. Maintaining barriers to gene flow between crops and their wild relatives has been a concern for farmers since domestication of plants began. Removing individuals from crop populations that have inherited undesirable traits from uncultivated relatives requires diligent artificial selection over many generations. Even more difficult is controlling the spread of one set of crop traits to free-living populations or different cultivars. Gene flow occurs when crop pollen or seed moves beyond the intended area of cultivation. In areas where a mixture of cropping systems is used, pollen or seed from transgenic crops could mingle with local cultivars or landraces, potentially changing their genetic composition. Gene flow from transgenic crops may accelerate the evolution of weediness in free-living populations, or alter the gene pools of non-transgenic crops in ways that are difficult to undo. We have chosen to take the approach that if gene flow from engineered crops to recipient populations can happen, it will. There are several documented examples of spontaneous hybridization between crops and their wild relatives (Ellstrand, 2003), including maize. Recently, a case of crop-to-crop transgene flow was documented in Canada (Hall et al., 2000), highlighting the opportunity for gene flow between adjacent fields of the same crop. Furthermore, once transgenes have introgressed into new populations, recovering or eradicating the gene construct would be difficult if not impossible, due to the unpredictable and sometimes extensive nature of gene flow (Ellstrand, 2003). Once the transgene enters a new crop or wild population, it is subject to different evolutionary processes than previously. Natural selection on transgenes may lead to rapid genetic changes in recipient populations (Haygood et al., 2003). Gene flow has been implicated in creating a risk of increased weediness (reviewed in Ellstrand © CAB International 2004. Environmental Risk Assessment of Genetically Modified Organisms: Vol. 1. A Case Study of Bt Maize in Kenya (eds A. Hilbeck and D.A. Andow)

187

Environ. Risk. Chap 06

188

23/8/04

9:32

Page 188

J. Johnston et al.

and Shierenbeck, 2000) or extinction in wild relatives (Wolf et al., 2001). Hybridization or gene flow between a crop (transgenic or not) and a related species could have one or more of these negative impacts, and the addition of the transgene simply introduces more unknowns. Thus, we assume that the cooccurrence of a crop and a potential recipient population with any amount of cross-compatibility among the two will result in gene flow. In this report, we have developed a framework of questions that relate to assessing the risks associated with gene flow, and have applied them to the proposed introduction of Bt maize into Kenya (Box 6.1). The gene flow working group at the Kenya workshop evaluated three aspects of gene flow risks of the proposed Bt maize introduction in Kenya: (i) what is the likelihood that intraand inter-specific gene flow between Bt maize crops and recipient populations will occur? (ii) What is the likelihood that a specific transgene from the crop will increase in frequency following gene flow in recipient non-transgenic populations? (iii) What potential ecological and agronomic effects could result from the spread and persistence of Bt transgenes in Kenyan agriculture? Several types of recipient populations were considered, including weedy relatives of the crop, wild relatives that are rare and non-transgenic crop varieties. As the introduction of Bt maize is still hypothetical in Kenya, we have stated the assumptions that we made about farming practices and seed distribution wherever possible. Our approach was to identify information gaps and suggest protocols for filling them. When appropriate we have indicated where the gene flow questions link to topics addressed by other chapters of this volume. Other topics are included in a special section at the end of the chapter. A glossary of terms can be found in the appendix to this chapter. The intended geographical scope of this report is all of Africa and tropical regions where maize is grown in both commercial and subsistence agriculture. We must qualify the gene flow section of the report as being relevant only to areas with no wild relatives of maize and in non-centres of origin. The gene flow concerns would be considerably different in Mexico and Central America due to the presence of teosintes, maize wild relatives that are usually capable of hybridization with cultivated maize.

Likelihood of Gene Flow from Bt Maize to Recipient Populations in Kenya Bt maize would potentially be introduced into all five maize-producing agroecological zones in Kenya, according to representatives of the Kenyan Agricultural Research Institute (KARI). Large-scale, commercial maize farms only occur in western Kenya (Muhammad and Underwood, Chapter 2, this volume). Throughout the country, maize is planted on small-scale farms as an important part of subsistence agriculture. Currently, KARI is considering inserting Bt genetic constructs into both commercial hybrid varieties and improved open-pollinated varieties (OPVs) that are favoured by small farmers. The pattern of Bt maize introduction into Kenya will vary depending on what types of Bt maize lines are available. The most likely scenario is that

Environ. Risk. Chap 06

23/8/04

9:32

Page 189

Gene Flow and its Consequences

189

Box 6.1. Gene flow and its consequences: a question framework to guide risk assessment. 1. What factors affect the likelihood of intra- and interspecific gene flow? 1.1. Are related taxa present in the region? 1.1.1. In what regions is the crop likely to be cultivated? 1.1.2. What is the frequency and distribution of closely related species within this area? 1.2. To what extent does the crop species cross-pollinate with other plantings of the same crop? 1.2.1. To what extent does the crop outcross? 1.2.1.1. Are male-sterile lines used to produce hybrid seed? (Male-sterile plants will always cross-pollinate.) 1.2.2. Over what distances does cross-pollination occur? 1.3. To what extent do fertile hybrids occur between the crop and nearby relatives? 1.3.1. Are hand-crosses successful? 1.3.1.1. Does the direction of pollination affect the success of crosspollination (e.g. crop-to-wild vs. wild-to-crop)? 1.3.2. Does hybridization occur spontaneously under natural or experimental conditions? 1.3.2.1. Have putative hybrid or backcrossed plants been observed? 1.3.2.2. Have hybrid or backcrossed plants been confirmed using genetic markers? 1.3.3. Over what distances can cross-pollination occur? 1.4. How fit are crop–weed hybrids? (The fitness of hybrids could be lower, similar or higher; here, the term hybrid includes crosses between cultivated and wild forms of the same species, as well as interspecific crosses.) 1.4.1. Are F1 hybrids fit enough to survive and reproduce while growing among wild relatives or crop plants? 1.4.2. Are F2 or BC1 hybrids fit enough to survive and reproduce while growing among wild relatives or crop plants? 1.4.3. What proportion of F2 or BC1 plants exhibit lifetime fecundity that is equal to or greater than the average fecundity of wild plants grown in the same experimental conditions? 1.4.4. How does hybrid fitness compare among different field locations, growing conditions and years? 1.4.5. Does hybrid fitness differ depending on whether the hybrid seed formed on a crop or wild plant? 1.4.6. Can interspecific hybrids persist as perennial plants without sexually reproducing? 1.4.6.1. Could these perennial plants spread clonally? 1.5. Does the crop produce volunteers or establish feral populations? 1.5.1. Can feral populations persist for more than one or two growing seasons? 1.5.2. Do feral populations establish a seed bank that could potentially be a source of weedy plants for many years? 1.5.3. Are the plants perennial and, if so, are they clonal (e.g. grasses)? 1.5.4. Are volunteer or feral populations considered to be weedy in agricultural or unmanaged areas? 1.5.5. Do feral populations act as a ‘genetic bridge’, increasing the chances that crop genes will escape into populations of wild relatives? Continued

Environ. Risk. Chap 06

10/9/04

190

14:30

Page 190

J. Johnston et al.

Box 6.1. Continued. 2. What is the likelihood that a specific transgene from the crop will increase in frequency following gene flow or establishment of feral populations? 2.1. If the transgene spreads to non-transgenic crops is it likely to persist in an agricultural setting? 2.1.1. Do crops with introgressed transgenes have higher yields or better quality than non-transgenic crops (e.g. due to resistance traits) that make them more likely to be selected and propagated by farmers? 2.1.2. Do transgenic crops produce more pollen than non-transgenic versions of the same crop? 2.1.3. Is the largest proportion of a cultivated area planted in transgenic crops, such that transgenic pollen will be more abundant and sire more seeds on non-transgenic plants growing nearby? 2.1.4. Is the transgene stably inherited over many generations and many crop backgrounds? 2.1.5. Are seeds or vegetative propagules dispersed from the crop, either naturally or by people? 2.1.5.1. To what extent will transgenic seeds or vegetative propagules be transported among regions? 2.2. Is the transgene likely to spread and persist in free-living intra- or interspecific hybrid populations? (Compare populations in farmer’s fields, field margins, and unmanaged areas, including habitats of any rare species that hybridize with the crop.) 2.2.1. If the transgene construct under consideration is available for research purposes, what are the fitness effects of the transgene when tested empirically in hybrid or backcrossed plants in comparison to appropriate control plants? 2.2.1.1. Is a fitness benefit associated with the transgene under natural or artificial selective pressures (e.g. natural levels of disease, herbivory or herbicide use)? 2.2.1.1.1. What ecological factors have the greatest effect on components of fitness in wild populations, and how is the transgene likely to alter these effects? 2.2.1.1.2. Will the factors limiting the population size of wild relatives (herbivores, pathogens, etc.) limit populations that have introgressed the transgene? 2.2.1.2. Is a fitness cost associated with the transgene in the absence of selective pressures? 2.2.1.3. How do fitness comparisons between transgenic and nontransgenic wild relatives vary among different field locations, growing conditions and years? 2.2.1.4. Is the transgene stably inherited over many generations? 2.2.1.4.1. Is transgene expression consistent among different genetic backgrounds, generations and environmental conditions? 2.2.2. If transgenic crop plants are not available for experimental studies, can the effect of the transgene be mimicked (e.g. using insecticidal sprays)?

Environ. Risk. Chap 06

10/9/04

14:30

Gene Flow and its Consequences

Page 191

191

Box 6.1. Continued. 2.3. Is the transgene likely to spread and persist in feral crop populations? 2.3.1. Can feral populations redevelop seed dormancy that is often bred out of cultivated varieties? 2.3.1.1. Will seed banks establish, creating a source of feral transgenic plants for many years to come? 2.3.2. Do feral populations possess mechanisms for dispersing seed further than their relatives in cultivation? 2.3.2.1. If seeds have enhanced dispersal, are feral populations likely to invade new habitat? 2.3.2.2. Could spread of feral populations allow transgenes to come into contact with wild relatives that do not occur immediately in agricultural areas? 2.3.3. Could feral populations create a ‘genetic bridge’ through which crop genes are more likely to flow into wild populations than from direct cropweed pollen transfer? 2.4. What possible variations on the direct, expected consequences of gene flow could affect the persistence and spread of transgenes in the environment? 2.4.1. Will the introduction of the transgenic promoter alone alter the phenotype of introgressed crops or weeds? 2.4.2. How will genetic background affect the likelihood of transgene silencing following introgression? 2.4.3. Could the transgene cause an increase in one fitness component, but overall reduction in fitness (e.g. the case with salmon growth hormone), creating extinction concerns for related taxa? 2.4.4. If multiple transgenes are ‘stacked’ can you assume that fitness effects will be additive? 3. What potential ecological and agronomic effects could result from the spread and persistence of transgenes? 3.1. How could the crop-to-crop spread of transgenes affect local agricultural production? 3.1.1. Are there possible effects of transgenic insect resistance on resistance management, integrated pest management, or beneficial non-target species? 3.1.2. Are there possible effects of transgenic herbicide resistance on the efficacy on locally used herbicides (e.g. glyphosate)? 3.1.2.1. Could these types of effect reduce or enhance options for crop rotation or rotations of herbicides with different modes of action? 3.1.3. Are there other ways in which transgenic crops will alter local agricultural practices? 3.1.3.1. For example, will more growers use ‘no-till’ methods in conjunction with herbicide-resistant crops? 3.1.4. Will use of transgenic crops reduce negative environmental impacts of traditional agricultural practices using non-transgenics? 3.1.5. Could the spread of any type of transgene compromise the market value of non-GM crops? 3.1.6. Could transgenes that are not approved for human consumption spread to other plantings of the same crop? Continued

Environ. Risk. Chap 06

10/9/04

192

14:30

Page 192

J. Johnston et al.

Box 6.1. Continued. 3.2. How could the crop-to-crop spread of transgenes affect the genetic diversity of local landraces? 3.2.1. Is the area known to be a centre of origin for the crop? 3.2.2. Is there already gene flow from modern cultivars into non-transgenic landraces? 3.2.3. Could a strongly selected transgene or a highly popular transgenic crop lead to more rapid displacement of landrace gene pools in comparison to displacement that may already occur? 3.3. How could the crop-to-wild spread of transgenes affect the biology of wild relatives of the crop? 3.3.1. Could a strongly selected transgene reduce the genetic diversity of rare and non-weedy species, beyond effects that may already occur due to gene flow from non-transgenic crops? 3.3.2. Could the transgenic trait allow free-living relatives to become more abundant within their typical habitats? 3.3.3. Could the transgenic trait allow free-living relatives to occupy new ecological niches (e.g. due to cold or drought tolerance)? 3.4. How could the crop-to-wild spread of transgenes affect non-target species and biodiversity, either directly or indirectly? 3.4.1. What is the transgene product, and what are the likely effects of this trait on other plants, herbivores and beneficial organisms in the region? 3.4.1.1. Is the transgene product likely to affect multiple trophic levels? 3.4.1.2. Will belowground food webs be affected? 3.4.1.2.1. Is there enough baseline data to make this determination? 3.4.2. Does the transgene confer any unintentional but ecologically significant changes in the chemical composition of hybrid and backcrossed progeny? 3.4.2.1. If so, are these effects likely to affect plant competition, plant–insect, or plant–soil interactions? 3.4.3. If free-living relatives of the crop become more widespread due to the transgene, are they likely to competitively displace any native plant species in their typical habitat? 3.4.3.1. Could the increase in dominance by the transgenic weed destabilize the whole food web? 3.4.4. If free-living relatives of the crop become more widespread due to the transgene, are they likely to competitively displace any native plant species by extending their native range? 3.4.4.1. Will the food web or exosystem processes in the new habitat be severely altered?

commercial hybrid Bt maize will be available first (Andow and Hilbeck, Chapter 1, this volume). If this is the case, Bt genes will probably be introduced into large-scale farming in the western regions of the country and spread from there. If OPVs are available first, it is more likely that small-scale farmers across the country will adopt Bt maize first. We worked on the assumption that even if only hybrid varieties are available, the Bt gene will eventually be introduced into both small- and large-scale farms throughout the country passively via inadvertent pollen movement or actively through intentional seed movement.

Environ. Risk. Chap 06

23/8/04

9:32

Gene Flow and its Consequences

Page 193

193

There are three types of recipient populations that should be considered as potentially at risk of gene flow from a transgenic crop introduction: (i) wild relatives; (ii) landraces or local cultivated varieties; and (iii) non-transgenic commercial crops. The populations that are generally of most concern are wild relatives, especially those that are weedy pests or rare species. Special concern for gene flow is often expressed in areas that are the centre of diversity for the crop in question. For maize, the centre of diversity is found in Mexico and Central America (Guatemala). The wild relatives of maize that are found in Mexico are absent from the African flora. There are no other species in the genus Zea that are known to occur in Africa. However, Sorghum shares a tertiary genome with maize (Harlan, 1992). Eastern Africa is the centre of diversity for Sorghum, but the risk of gene flow between maize and Sorghum species is negligible without the intervention of biotechnology. We conclude that the opportunity of gene flow from Bt maize to free-living plants in Kenya is negligible. Because there are currently no weeds that can hybridize with Bt maize in Kenya, we did not consider large sections of the gene flow question framework (Box 6.1). For instance, questions regarding crop–weed hybrid fitness, and volunteer and feral populations were not addressed during the workshop. If a similar evaluation of gene flow risks were to be carried out in other parts of the world, it would be imperative to consider the effects of gene flow to wild recipient populations. Furthermore, while there are currently no wild relatives of maize in Kenya, plant species are constantly moved around the globe. We recommend that the presence of wild relatives be re-evaluated every few years. Landraces are present in all areas where Bt maize is likely to be introduced in Kenya. It was fairly straightforward to conclude that if the Bt gene increases seed yield in Kenyan landraces, it can and will become introgressed into landrace gene pools. When we examined local landraces we included any type of maize: (i) cultivated from recycled seed (seed produced the previous year); (ii) genetically differentiated landraces; (iii) widely crossed and genetically dynamic local cultivars; and (iv) recycled OPVs or hybrid varieties. KARI statistics show that most small farmers are recycling seed, while large-scale farms tend to grow, buy and plant commercial hybrid varieties of maize each year (Muhammad and Underwood, Chapter 2, this volume). Since there is no genetic or phenotypic diversity information available about maize landraces in Kenya, we made some assumptions regarding the genetic make-up and compatibility with hybrid crop varieties. We assume that there is full genetic compatibility between landraces and conventional hybrid varieties. It is our understanding that the genetic composition of the landraces is dynamic, and is therefore likely to be partially derived from hybrid varieties. We are taking the approach that if gene flow can happen it will, so we are treating any level of genetic compatibility as similar to full compatibility. Thus because we have no reason to believe there will be any substantial barriers to gene flow between commercial Bt maize and landraces, we are assuming full compatibility, which is the worst-case scenario. An experimental test of compatibility between commercial varieties and landraces could be performed, but unless complete incompatibility was found, very little useful information would come out of such a test. Research effort would go to better use elsewhere.

Environ. Risk. Chap 06

194

23/8/04

9:32

Page 194

J. Johnston et al.

Based on information gathered informally from Kenyan participants in the workshop, it is likely that most Kenyan landraces of maize have been widely crossed with several gene pools and are as a result more genetically diverse than conventional hybrid cultivars. This pattern is also seen in Mexican maize landraces, which have higher genetic variation in terms of DNA sequence polymorphism than US inbred lines (Tenaillon et al., 2001). In the Mexican case, individual landraces vary significantly in rare neutral alleles and phenotypic trait diversity, which suggests that they have been genetically isolated for many generations (Blancas, 2001). Kenyan landraces are suspected to have much more mixing among gene pools, whereas phenotypic differences are maintained by artificial selection in each line. Many questions about landrace genetic diversity will be clarified when the maize collections in Kenya’s National Gene Bank are analysed for genetic diversity. If the Bt gene becomes introgressed into local Kenyan maize landraces and causes a significant increase in maize yield on small farms, we predict that the transgene will quickly become a stable component of the gene pool for many landraces. The rate of transgene spread and subsequent agronomic and ecological effects may vary substantially in regions with differing insect pressures. There are two ways that we felt the Bt transgene is likely to spread if Bt maize were introduced in Kenya. If large-scale commercial farms are the first to obtain the technology, and able to purchase fresh hybrid Bt seed every year, we expect there will be significant movement of transgenes via pollen drift. The small-scale farmers growing OPVs, local cultivars and landraces will be the most likely to be affected by the Bt pollen drift. The other route of gene flow is likely to be intentional movement of seeds among small-scale farmers. In Mexico, there is evidence that farmers have benefited from intentionally crossing commercial hybrid maize with landraces (Baltazar and Schoper, 2002) and the same is likely to occur in Kenya. Currently, seed trade among different regions is frequent, even encouraged by KARI and other government programmes (Muhammad and Underwood, Chapter 2, this volume). If there is a real advantage to the subsistence farmer, we predict that the seeds will move quickly into the small-scale farms and the transgene will become integrated into many of the landraces and local cultivars. We are assuming that large-scale farms will plant primarily Bt maize if it is available. Bt maize is being developed and will be distributed in Kenya by government and non-profit research institutions. Thus, it seems likely that the cost of commercially released Bt maize in Kenya will be readily affordable to farmers that are currently purchasing seed annually for planting. If farmers choose to plant conventional commercial hybrid cultivars that are non-Bt, there will be some opportunity for gene flow into these fields. However, if seeds are not recycled, then potential propagules will be removed from the field upon harvest. There is some chance that introgressed Bt volunteer maize plants will germinate and grow, but since the seeds would most likely be harvested with the next season’s crop, there is little chance that this will be a significant pathway of transgene escape beyond the local farm.

Environ. Risk. Chap 06

23/8/04

9:32

Page 195

Gene Flow and its Consequences

195

The routes and rates of transgene escape will vary depending on the type of source population, and whether or not farmers save seeds to plant their crops the following year. Seeds of transgenic hybrid maize varieties are usually hemizygous, possessing only one copy of the Bt gene (Andow and Hilbeck, Chapter 1, this volume). Thus only half of the pollen grains and seeds from a hemizygous transgenic plant will contain the transgene. If a homozygous OPV was produced, it would possess two copies of the gene and all pollen or seeds it produced would have a single copy of the transgene (Fig. 6.1). Whether farmers save seed or not will likely be important in determining the persistence of transgenes in a population, and how quickly they spread. In situations where farmers are buying new seed stocks to plant each year, most transgenes that have mingled with non-transgenic cultivars will be harvested along with the crop. The chances that transgenic seeds are left behind are small, and if there are transgenic hybrids that volunteer the following year, they will most likely be harvested along with next year’s crop. In this way, farmers who purchase new seed stocks each year, such as those growing non-Bt improved hybrid varieties, will have reduced risk of transgene incorporation into their non-transgenic cultivars, provided non-Bt varieties are produced in such a way that contamination from Bt varieties does not occur. In contrast, most small-scale farmers who grow landraces, local cultivars or OPVs will save seeds from their harvest to plant the following year. In this case, there is a much higher likelihood that transgenes will become incorporated into non-transgenic cultivars, either intentionally or unintentionally. If the Bt gene has a positive effect on maize fitness, and farmers preferentially select transgenic seeds to plant the following year, transgene frequency in the local cultivar or OPV can increase rapidly (Fig. 6.2). If the Bt gene does not increase fitness, or the farmer does not preferentially select these seeds to plant the next year, the frequency of the transgene in the population may remain fairly constant or increase slowly from one generation to the next. If the same Bt gene is introduced into a non-transgenic population repeatedly, its frequency will increase even if it does not have a positive effect on fitness (Haygood et al., 2003). For example, maize grown by a small farmer who saves seed in a field near a large maizeproducing field might experience passive gene flow via pollen every year. Over time, the frequency of Bt will increase in the small farmers’ maize, which may or may not affect yield of whatever cultivars are growing on the small farm. Hybridization event 1. Hemizygous source of transgene (hybrid or OPV). Among progeny, 50% have one copy of transgene, and 50% are nontransgenic. Transgene frequency = 0.25.

2. Homozygous source of transgene (OPV). Among progeny, 100% have one copy of the transgene. Transgene frequency = 0.5.

Hybrid offspring

×

×

Fig. 6.1. Transgene transmission during hybridization. The Bt transgene is indicated by the black bar on the chromosome.

Environ. Risk. Chap 06

23/8/04

9:32

Page 196

196

J. Johnston et al.

Hemizygous source of transgene

Hybridization event

Homozygous source of transgene

×

×

Offspring of cross Offspring Generation 2 (backcrossing into non-transgenic crop) Selection for transgene

No selection for transgene

Selection for transgene

No selection for transgene

Offspring Generation 3 (backcrossing) Offspring Generation 4 (backcrossing) Offspring Generation 5 (backcrossing)

Fig. 6.2. Predicted behaviour of transgene frequency following gene flow from crop to nontransgenic cultivar. Pie diagrams show predicted proportions of each generation that would possess no copies of the transgene (white), one copy of the transgene (grey) or two copies of the transgene (black). In the four scenarios depicted above, we are assuming seeds are saved and replanted from one generation to the next. With each successive generation, the population or pool of offspring considered gets larger. In the left-hand column under each cross, seeds from plants with the transgene are saved and replanted more often than nontransgenic seeds, resulting in an increase in gene frequency in each generation of offspring. In the right-hand column, the transgene is not preferentially selected, so in each generation there is a dilution effect of backcrossing to non-transgenic individuals.

Source populations that are hemizygous for the transgene may have different rates of transgene spread than a source population that is homozygous for the transgene. To date, most, if not all, transgenic varieties of maize have been created for large-scale agriculture and are hemizygous hybrid varieties. In Kenya, it seems that there is interest in developing an OPV with the Bt gene for smaller-scale farmers to use and recycle the seed. Presumably, any Bt OPV would be homozygous for the transgene in order to breed true for Bt from one generation to the next. If seed from a hemizygous OPV or hybrid variety were recycled, the next year many crop plants would lack a Bt gene, which may complicate both pest management and resistance management. If, instead, the seed were saved from a crop that was homozygous for Bt, all the plants in the first generation progeny would possess the transgene. Therefore, a Bt homozygous OPV might lead to a greater rate of spread if gene flow were to

Environ. Risk. Chap 06

23/8/04

9:32

Page 197

Gene Flow and its Consequences

197

occur, because each gene flow event from Bt homozygous maize would introduce twice as many copies of the transgene as from Bt hemizygous maize (Fig. 6.1). A factor to consider in the creation of Bt maize lines for Kenya is where to locate breeding fields for generating seeds. These fields may also act as sources for Bt transgene escape into surrounding non-transgenic maize fields. If Kenya were interested in further developing conventional hybrid maize for export or developing exports of organic or genetically modified (GM)-free products there would be additional cause for concern about crop-to-crop Bt contamination. Currently, little maize is exported from Kenya and most maize is grown on small farms and consumed locally (Muhammad and Underwood, Chapter 2, this volume). With this cropping system and negligible international sales of maize, there is little chance at present that Kenya will suffer losses related to international trade if gene flow from Bt maize to non-Bt conventional hybrid maize occurred. However, there could be some economic losses due to gene flow from Bt maize to other cultivars if there were a domestic market for non-transgenic maize in Kenya, or if Kenyan exports of non-transgenic maize were to increase.

Likelihood that a Transgene Increases in Frequency Following Gene Flow We assumed for the purposes of the workshop that the genetic background of the introduced transgenic maize is similar enough to that of local landraces that compatibility between the two would be high. Therefore, we predicted that outbreeding depression and other reproductive isolation mechanisms would not significantly restrict or reduce gene flow rates. If natural selection favours the transgene, continued or recurrent gene flow will accelerate the increasing frequency of the transgene in the recipient population (Ellstrand et al., 1999; Haygood et al., 2003). Transgenes are present in every individual in the genetically modified Bt maize line, or source, and absent in any recipient population (i.e. landraces and any other non-transgenic maize population). If free-living or landrace populations are small, fewer gene flow events need to occur to produce a significant increase in transgene frequency. The introduction of the transgene may increase the recipient populations’ mean fitness relative to landrace populations without the introduced transgene, which could further increase the frequency of the transgene. The fitness effect of the transgene on landrace populations is difficult to estimate, because there will be a combination of natural and artificial selection acting on a transgene in a new landrace genetic background. Natural selection will act on germination, vigour and yield of landrace plants that possess a transgene. However, there will also be potentially strong artificial selection acting in the population when farmers select seed for planting the following year (Berthaud et al., 2002). While it seems likely that plants that express the Bt gene construct will be more vigorous and would make good candidates for replanting the following year, it is possible that either due to genetic complications or a lack of herbivory, the Bt gene will have a neutral or

Environ. Risk. Chap 06

198

23/8/04

9:32

Page 198

J. Johnston et al.

detrimental effect on the phenotypic characters that farmers use to select seed for recycling. We have used population genetic theory to try to predict rates and extent of transgene spread, but with such a large artificial selection component and current information gaps, this exercise is overly speculative. Furthermore, we do not know if the fitness effect of the Bt gene is similar in hybrid varieties and landraces. Generally, only transgenic constructs that are inherited as stable Mendelian traits for many generations receive sufficient regulatory approval to be released commercially. Assuming this would be the case in Kenya with Bt maize, we expect that Bt genes would be expressed at similar levels in the released varieties and in all other landrace genetic backgrounds. We do not know if the expression of the Bt gene is likely to confer the same fitness benefit to landraces as it does to hybrid varieties. Because we were informed that local cultivars are similar to and often crossed with hybrid and OPVs, it is possible that the fitness effects will be similar. The best way to estimate the likelihood that Bt will spread in Kenyan maize landraces is to directly measure the fitness effect of the gene on a set of recipient populations. Fitness effects of a transgene on recipient populations should be separately assessed for each transformation event under consideration. The first step in estimating the fitness effect of a specific maize transgene can be done by comparing maize isolines, i.e. maize inbred lines that are homologous at all loci, but that differ by the presence/absence of the transgene. The test should be performed in a range of environments, varying from those under strong selective pressure to those under no selective pressure for the target trait. Several fitness components to consider are: germination rate, survivorship to reproduction, total number of seeds produced and male fitness (relative success of pollen to sire offspring). The most robust way to measure fitness effect of a transgene is to monitor transgene frequency from one generation to the next. If the transgene increases in frequency (in the absence of additional gene flow) there can be no question that selection is favouring the transgene.

Potential Ecological and Agronomic Effects Resulting from Spread and Persistence of Bt Transgenes Landraces are not static ancient varieties, but highly variable open evolving genetic systems (e.g. Jarvis and Hodgkin, 1999). Levels of genetic diversity in maize landraces and local cultivars in many regions worldwide are unknown. Stuber and Goodman (1983) conducted one of the first genetic surveys of popular and historical lines of maize from the USA, Canada, Europe, Latin America and India using allozyme electrophoresis. More recently, genetic diversity studies of maize germplasm have incorporated morphological and DNA analyses of more extensive samples of traditional maize from Europe (Rebourg et al., 2001) and allozyme variation in southwestern China (Lu and Bernardo 2001). In the more extensive survey by Stuber and Goodman (1983), results showed that the most widely used inbred maize lines had only

Environ. Risk. Chap 06

23/8/04

9:32

Gene Flow and its Consequences

Page 199

199

21% of the total number of alleles found in 1600 maize lines and teosinte, the wild relative of maize. Furthermore, most maize inbred lines were homozygous (had only one form of most genes). As a whole, these studies point out the significance of the geographic distribution of genetic diversity resulting in population differentiation due to local adaptation and agronomic practices associated with landraces worldwide. Therefore, genetic diversity of maize landraces and local cultivar varieties is expected to be unique and different from that of commercial inbred maize lines. We predict that one of the largest potential ecological impacts of gene flow for Bt maize to recipient populations in Kenya would be the loss of genetic uniqueness in landraces and local cultivars, if they are in fact adapted to local conditions and possessing unique genetic variation as we suspect. The rate of gene flow varies among populations even when individuals are fully compatible, have overlapping flowering times and occur close together (Ellstrand et al., 1999). Differences in genetic background of the source (i.e. transgenic maize populations) and recipient populations (i.e. maize landrace) may result in different rates and effects of gene flow. The effect of gene flow on the genetic diversity of a landrace will depend upon the size of the recipient population and how much genetic diversity is present in the donor population. Commercially inbred lines of maize, and consequently transgenic maize lines, contain substantially less genetic variation than landraces or local maize cultivars (Stuber and Goodman, 1983; Ellstrand et al., 1999; Tenaillon et al., 2001). Thus, it is predicted that genetic variation in locally adapted landraces and maize cultivars is likely to decrease after recurrent gene flow from several generations of the same commercial maize inbred line whether it is transgenic or not (Ellstrand et al., 1999). If gene flow occurs once or infrequently, or different transgenic lines are planted in different years, the effect of gene flow may be to increase neutral genetic diversity overall in landraces experiencing gene flow. However, strong selection may result in the loss of unique variation among genes with close linkage to the transgene construct (Gepts, 2002). If the Bt transgene has a positive effect on fitness, the transgene frequency will increase in the recipient population (Ellstrand et al., 1999). If a transgene becomes quickly fixed in a recipient population, genetic diversity can be lost due to a selective sweep (Gepts, 2002). If unique genetic variation was initially present in the landrace, the transgene and other genes from the commercial variety may eventually replace it. If landraces or cultivars experience an increase in fitness because of the Bt transgene, this may lead to an increase in yield that farmers might well perceive as a benefit. However, even a moderate level of gene flow (c. 1–5% per generation) that introduces alleles that interfere or replace locally adapted alleles or gene complexes may ultimately result in the reduction of local fitness of maize landraces over the long term. This loss in fitness may manifest itself immediately. Prior to release, we suggest that commercial Bt maize varieties be crossed with Kenyan maize landraces for several generations to test for potential reduction in landrace fitness over time. However, the logistics of such a test would be daunting.

Environ. Risk. Chap 06

23/8/04

9:32

Page 200

200

J. Johnston et al.

Gene flow allows a transgene such as Bt to move and spread through the landscape, encountering different ecological niches along the way. In Kenya, if the Bt gene spread to landrace populations, it may come in contact with a new suite of insect pests that differ from those found in the large-scale farm fields where the Bt maize varieties were initially planted. Small farmers are more likely to plant multiple crops together, while large-scale farmers will tend to plant large areas with one variety of maize. This new association with weeds and intercrops or adjacent crops, as well as subtle changes in microclimate associated with small-scale farmers’ fields, may pose new non-target or biodiversity risks. Furthermore, the additional presence of transgenes in the landscape potentially alters the process of insect resistance evolution (Fitt et al., Chapter 7, this volume). In summary, one can expect some changes in the genetic composition of Kenyan maize landraces when grown in close proximity to commercial Bt maize varieties. If the Bt gene has a large effect on fitness, there is likely to be a greater effect on these landraces of gene flow from Bt maize than from nontransgenic commercial hybrid maize. Gene flow from Bt maize to landraces may lead to the loss of rare and low frequency alleles that contribute to population differentiation and local adaptation of Kenyan maize landraces. The most likely area of the genome to be altered by gene flow from a transgenic crop is the region adjacent to the transgenic construct. In extreme cases, if the transgene is strongly beneficial and gene flow occurs at sufficient levels, the landrace may lose genetic distinctiveness at the site of the transgene and closely linked loci, or throughout large sections of the genome.

Experimental Protocols While there is much that is known about the biology of maize in Kenya and the risks of gene flow from maize cultivars in other environments, some critical knowledge gaps remain. The unknown factors relevant to the effects of gene flow from Bt maize in Kenya fall into two general categories. At a plant phenotypic level, the fitness effect of the Bt gene once it has introgressed into a population other than its original variety is unknown. At a population genetic level, this fitness effect will influence the fate of the Bt gene once it has become incorporated into a new population. Here we propose protocols to help predict the fate of the Bt gene following a gene flow event, and to monitor what actually happens once the gene is incorporated into a new population.

Estimating the fitness effect of a Bt gene on cultivars of maize It is important to recognize that fitness effect of a Bt gene is likely to vary among the maize-growing regions of Kenya. As pressure from the target insects of Bt increase, the fitness effect is predicted to increase. Cultivars may differ in their levels of natural resistance to the insects targeted by Bt, which may alter the fitness effect of the transgene. Furthermore, the genetic background presented by different cultivars may change the levels of transgene expression.

Environ. Risk. Chap 06

23/8/04

9:32

Page 201

Gene Flow and its Consequences

201

As locations and cultivars are added to the experimental design, it will be easier to extrapolate to a wider variety of growing situations in Kenya. Two types of treatment factors, maize variety and insect pressure, were considered in the experimental design. To estimate the fitness effect the Bt gene may have on landrace fitness, we suggest choosing pairs of hybrid varieties that are as close to isolines as possible, differing only in the presence or absence of the transgene. Groups of >30 plants of each line should be planted. 1. 2. 3. 4.

Bt transgenic Commercial Maize Hybrid 1. Non-transgenic Commercial Maize Hybrid 1. Bt transgenic Commercial Maize Hybrid 2. Non-transgenic Commercial Maize Hybrid 2.

Because the fitness effect of Bt is likely to be highly dependent on the number and species of insect pest, testing several infestation densities of different species of stemborer will generate more useful information. However, if resources are limiting, a test with two species at moderate stemborer densities compared to no insect pressure will suffice. One scenario for this experiment could involve three replications of two pairs of isolines (A, B and C, D) arranged in randomized blocks (Fig. 6.3). Each block receives a randomly assigned insect pressure treatment (high, medium or zero insect density – indicated by grey shading) of either of the two main stemborer pests in Kenya (or both), Chilo partellus or Busseola fusca (Muhammad and Underwood, Chapter 2, this volume). The same Western Kenya

Eastern Kenya

Fig. 6.3. The simplest scenario for evaluating the fitness effect of the Bt gene on hybrid cultivars. Each letter represents a planting of one variety. Heavily shaded blocks represent high insect pressure, lightly shaded blocks represent ambient insect pressure and unshaded blocks represent no insect pressure.

Environ. Risk. Chap 06

23/8/04

202

9:32

Page 202

J. Johnston et al.

experimental arrangement could be replicated in multiple locations in Kenya. At the end of the experiment, fitness components should be measured on each entry within each block and analysed statistically using ANOVA. If landraces can be used for pre-release experimentation, we suggest a second group of treatments to incorporate into the study we have just described, or to perform as a second-year follow-up study. Landraces are likely to be variable by region and traits for which they are selected. To start, we suggest choosing two or three landraces that represent different regions or different suites of characters selected for by farmers. It is important to compare landrace fitness to that of Btlandrace hybrids as well as the Bt hybrid variety. 5. Transgenic maize variety (Bt) (omit if adding varieties F–I to study above, see Fig. 6.3). 6. Kenyan Maize Landrace 1. 7. Kenyan Maize Landrace 2. 8. Kenyan Commercial Bt Maize HybridKenyan Maize Landrace 1. 9. Kenyan Commercial Bt Maize HybridKenyan Maize Landrace 2. In the F1 generation from a cross between hybrid Bt maize and a landrace, only half of the offspring will possess a Bt gene. Hence, F1 plants could be screened to identify individuals possessing a Bt gene. It is possible that F1 hybrids could exhibit hybrid vigour that is unrelated to the presence of the transgene. Thus, the most robust comparison would be among F1 hybrids with and without the transgene, so that any fitness advantage would be likely to be due to the presence of Bt. If landraces cannot be used for experimentation with Bt gene constructs prior to release of the commercial Bt hybrids, sprays of microbially produced Bt toxin could be used instead to simulate a transgenic Bt landrace. While this is not as conclusive an indicator of Bt effect on landrace fitness, it would be better than no landrace-specific information, and it would allow the inclusion of as many landraces as deemed desirable. Unfortunately, the impacts of gene flow from Bt maize on landrace populations is difficult to predict and impossible to measure without releasing the transgenic crop into the environment. If the crop is approved for release, and regulatory personnel are interested in monitoring the effects that Bt maize has on landrace populations, we suggest the following protocols for evaluating changes in neutral genetic diversity, frequency of the transgene in landrace populations, and fitness differences in transgenic and non-transgenic maize varieties.

Measuring fitness differences in transgenic and non-transgenic maize cultivars Significance The role of hybrids and hybrid introgressants is significant in the persistence and transfer of the transgene beyond the F1 generation. Hybrid breakdown can result in the limited success of the transgene in maize landrace populations. If hybridization occurs initially at a high level but results in hybrid breakdown due

Environ. Risk. Chap 06

23/8/04

9:32

Page 203

Gene Flow and its Consequences

203

to severe fitness loss, there will be a cost to the loss of viable seed and ultimately a risk of extinction for populations of small local landraces. Method Compare the mean relative fitness (see methods in protocol 1 above) of: (i) transgenic maize; (ii) non-transgenic maize landrace; (iii) hybrids of nontransgenic maize landracestransgenic maize with and without the transgene; and (iv) backcrosses in the extant range of the recipient population.

Monitoring the genetic diversity of neutral loci Significance Initial measures of neutral genetic diversity and population genetic parameters give a baseline level of diversity present in the landrace population. The changes in neutral genetic diversity after hybridization and introgression via gene flow will indicate the potential for loss and extinction of rare or low frequency alleles and loss of genetic polymorphism in terms of levels of heterozygosity. Furthermore, genetic differentiation as measured by Nei’s (1973) FST values can help determine if landrace populations are at risk of genetic homogeneity and thus rapidly evolve towards the inbred transgenic maize line. Method Measure population genetic parameters and gene frequencies of neutral loci in spatially proximal landrace populations at times: tn (i.e. initial exposure), tn+1 (i.e. hybridization) and tn+i (i.e. subsequent transgene introgression) in the population or populations of interest (see Whitton et al., 1997, for an example).

Monitoring presence and frequency of the transgene in a landrace population Significance A change in the level of the gene frequency of the transgene after hybridization and subsequent introgression can indicate whether it will persist or increase in the population. Moreover, predictions can be made with respect to the loss of landrace diversity due to selective pressure strongly favouring the transgene and thus the low allelic diversity associated with transgenic maize genomes. Method Determine the gene frequency of the transgene at tn (i.e. initial exposure), tn+1 (i.e. hybridization) and tn+i (i.e. subsequent to introgression) in the population or populations of interest in the landrace after a single gene flow event, and in an identical landrace after multiple gene flow events of the same magnitude (i.e. source and recipient populations that are similar in size).

Environ. Risk. Chap 06

23/8/04

9:32

Page 204

204

J. Johnston et al.

Conclusions and Recommendations If Bt maize is introduced into Kenyan agriculture, maize landraces are the most likely recipient populations for gene flow from the transgenic crop. The most important factor that is expected to influence the rate of transgene spread through landrace populations is the fitness effect of the Bt gene. We suggest estimating the fitness effect of the Bt gene by measuring it in hybrid lines and assuming it would be of similar magnitude in landraces grown in the intended region of distribution of the hybrid line. Alternatively, the fitness effect of the Bt gene could be measured directly on a sample of landraces in a closed system (greenhouse facility) and extrapolated. The largest fitness component for landrace populations is probably selection of ears for seed recycling by subsistence farmers. A better understanding of how the Bt gene may affect a farmers’ recycling choices would only improve the estimates of fitness. Finally, we identified the loss of genetic diversity as the largest ecological risk of the introduction of the Bt gene into maize landrace populations; but this depends on the relative but as yet unknown levels of diversity in Kenyan maize landraces. Several types of monitoring were suggested. Gene flow from any commercial varieties is likely to alter the genetic makeup of maize landraces, which may be of great concern in the long term. The spread of a Bt transgene via gene flow may complicate resistance management programmes. The spread of the transgene into subsistence agriculture is likely to expose a different suite of non-target organisms to the transgene, potentially creating additional biodiversity and non-target risks. Ultimately, small-scale farmers may suffer the most negative impacts of gene flow from Bt maize in Kenya, due to their virtually sole dependence on beneficial insect populations for biocontrol purposes. Perhaps most troublesome is the fact that little is known about the fate of transgenes in crop populations where seed saving is practised. As subsistence farmers select and recycle seed, we have no way of predicting what will happen to the transgene or the landrace populations. Will the transgene persist through many generations? Will Bt homozygotes and hemizygotes have the same levels of Bt toxin expression? Will it be possible to predict the proportion of recycled seed that possesses the Bt gene? What implications will this have for resistance management? There are many questions we cannot begin to answer with the current state of our knowledge. It is for policy makers and regulators to decide when enough information has been acquired for a decision regarding the release of Bt maize in Kenya.

Key findings 1. There are no known barriers to gene flow between commercial Bt maize varieties and Kenyan maize landraces or local cultivars. 2. In Kenya, maize pollen (with or without Bt) can easily move the distances separating large commercial farms and small farms, or small farms from each other.

Environ. Risk. Chap 06

10/9/04

14:31

Gene Flow and its Consequences

Page 205

205

3. In areas with moderate to high lepidopteran pest pressure, Bt maize plants will likely have higher fitness than plants without the Bt gene. 4. It can therefore be expected that if introduced, gene flow via pollen will quickly spread the Bt gene through Kenyan maize landraces. 5. Intentional movement of Bt maize seed by small-scale farmers is likely to accelerate the gene flow of the Bt gene into Kenyan maize landraces. 6. Consequences of gene flow will likely affect the diversity of landraces and pest resistance development. 7. The spread of the transgene into landraces will expose locally differing suites of non-target organisms with unknown outcomes. 8. Movement of Bt genes to populations outside of the area where it is initially planted will likely complicate resistance management. Key knowledge gaps 1. The fitness effect of the Bt gene on Kenyan maize landraces is not known. 2. The role that environmental conditions (climate, insect pressure) play in creating the fitness effect of the Bt gene in Kenyan maize landraces is not known. 3. The amount of unique genetic diversity contained within Kenyan maize landraces is not known. 4. The effect that Bt gene introgression would have on genetic diversity of Kenyan maize landraces is not known.

Glossary of Terms Cultivar: A general term to describe any crop line that is intentionally planted and harvested. Fitness: The relative reproductive success of an individual, in terms of cultivated maize it is functionally defined as the number of seeds produced or sired relative to other individuals in the same gene pool. Gene flow: The movement of genetic alleles from one population to another. Gene pool: All allelic variation at all genes within a population. Genetic diversity: The number of different forms of a gene found at neutral or adaptive genetic loci; a proxy for evolutionary potential. Improved hybrid variety: A cultivar that has been improved through an institutional breeding programme and whose seeds are generated by crossing two distinct, true-breeding lines (see Andow and Hilbeck, Chapter 1, this volume, for additional information). In general, these seeds are not intended for recycling. Insect pressure: The reduction in fitness due to herbivory by insects that are targeted by the Bt toxin. Introgression: The transfer, via hybridization and genetic recombination, of genes between gene pools. Isolines: Two or more lines of a cultivar that are genetically identical except for one or a few genes or loci of interest.

Environ. Risk. Chap 06

206

23/8/04

9:32

Page 206

J. Johnston et al.

Landrace: A cultivar line that has been improved by the selection of one or a few farmers. We are assuming in this case that seeds are saved and replanted each year, and that the gene pool of a landrace will contain some unique alleles compared to other cultivars. Line: A population that was initiated by few individuals followed by several generations of inbreeding within the group. Generally individuals within a line tend to be very similar genetically while individuals in different lines tend to be genetically divergent. Local cultivar: Used in this report essentially as a synonym for landrace, referring to any cultivar that has been selected for traits that increase its fitness in local conditions. Open-pollinated variety (OPV): Cultivars that have been developed and improved through institutional breeding programmes, but whose seeds are created by mating within a crop gene pool, and whose seeds can be recycled and breed true. Population: A group of interbreeding individuals. Selection: In a strictly natural setting, this refers to the differential survival of some genotypes over others. In an agricultural setting, it refers to the possession of traits that a farmer finds desirable, resulting in preferential saving of seed and planting the following year.

References Baltazar, B.M. and Schoper, J.B. (2002) Crop-to-crop gene flow: dispersal of transgenes in maize, during field tests and commercialization. Proceedings of the 7th International Symposium on the Biosafety of Genetically Modified Organisms. Beijing, China. Berthaud, J., Pressoir, G., Ramirez-Corona, F. and Bellon, M.R. (2002) Farmers management of maize landrace diversity. A case study in Oaxaca and beyond. Proceedings of the 7th International Symposium on the Biosafety of Genetically Modified Organisms. Beijing, China. Blancas, L. (2001) Hybridization between rare and common plant relatives: implications for plant conservation genetics. PhD dissertation, University of California, Riverside, California. Ellstrand, N.C. (2003) Current knowledge of gene flow in plants: implications for transgene flow. Philosophical Transactions of the Royal Society of London, Series B 358, 1163–1170. Ellstrand, N.C. and Schierenbeck, K. (2000) Hybridization as a stimulus for the evolution of invasiveness in plants? Proceedings of the National Academy of Sciences USA 97, 7043–7050. Ellstrand, N.C., Prentice, H.C. and Hancock, J.F. (1999) Gene flow and introgression from domesticated plants into their wild relatives. Annual Review of Ecology and Systematics 30, 539–563. Gepts, P. (2002) Possible effects of transgenes on genetic diversity. Proceedings of the 7th International Symposium on the Biosafety of Genetically Modified Organisms. Beijing, China. Hall, L., Topinka, K., Huffman, J., Davis, L. and Allen, A. (2000) Pollen flow between herbicide-resistant. Brassica napus is the cause of multiple-resistant B. napus volunteers. Weed Science 48, 688–694.

Environ. Risk. Chap 06

23/8/04

9:32

Gene Flow and its Consequences

Page 207

207

Harlan, J.R. (1992) Crops and Man. American Society of Agronomy, Madison, Wisconsin. Haygood, R., Ives, A.R. and Andow, D.A. (2003) Consequences of recurrent gene flow from crops to wild relatives. Proceedings of the Royal Society of London, Series B 270, 1879–1886. Jarvis, D.J. and Hodgkin, T. (1999) Wild relatives and crop cultivars: detecting natural introgression and farmer selection of new genetic combinations in agro-ecosystems. Molecular Ecology 8, 159–173. Lu, H. and Bernardo, R. (2001) Molecular marker diversity among current and historical maize inbreds. Theoretical and Applied Genetics 103, 613–617. Nei, M. (1973) Analysis of gene diversity in subdivided populations. Proceedings of the National Academy of Sciences USA 70, 3321–3323. Rebourg, C., Gouesnard, B. and Charcossett, A. (2001) Large scale molecular analysis of traditional European maize populations. Relationships with morphological variation. Heredity 86, 574–587. Stuber, C.W. and Goodman, M.M. (1983) Allozyme Genotypes for Popular and Historically Important Inbred Lines of Corn. US Agricultural Research Service, Southern Series Number 16. Tenaillon, M.I., Sawkins, M.C., Long, A.D., Gaut, R.L., Doebley, J.F. and Gaut, B.S. (2001) Patterns of DNA sequence polymorphism along chromosome 1 of maize (Zea mays ssp. mays L.). Proceedings of the National Academy of Sciences USA 98, 9161–9166. Whitton, J., Wolf, D.E., Arias, D.M., Snow, A.A. and Rieseberg, L.H. (1997) The persistence of cultivar alleles in wild populations of sunflowers five generations after hybridisation. Theoretical and Applied Genetics 95, 33–40. Wolf, D.E., Takebayashi, N. and Rieseberg, L.H. (2001) Predicting the risk of extinction through hybridisation. Conservation Biology 15, 1039–1053.

Environ. Risk. Chap 06

23/8/04

9:32

Page 208

Environ. Risk. Chap 07

10/9/04

7

14:32

Page 209

Resistance Risks and Management Associated with Bt Maize in Kenya G.P. FITT, D.A. ANDOW, Y. CARRIÈRE, W.J. MOAR, T.H. SCHULER, C. OMOTO, J. KANYA, M.A. OKECH, P. ARAMA AND N.K. MANIANIA Corresponding author: G.P. Fitt, CSIRO Entomology, Long Pocket Laboratories, 120 Meiers Road, QLD 4068 Indooroopilly, Australia. E-mail: [email protected]

Introduction This chapter addresses the risk that insect pests associated with Bt maize may evolve resistance to Bt proteins. Insecticide resistance is a common response among insects to the selection pressure imposed by insecticides. The framework and concepts developed here are designed to be applicable to diploid arthropods evolving resistance to insecticidal transgenes expressed in trangenic plants. They are also relevant to resistance in haplodiploid and parthenogenetic arthropods, nematodes, virus, fungi or bacteria to nematicidal, viral, fungicidal or bactericidal transgenes, as well as to herbicide resistance, though the details appropriate for diploid arthropods would need re-orienting to develop relevant resistance management guidelines. Broadly, we establish a series of questions that may be addressed before any field release, other questions that may be addressed during field trials but before any commercial release and some ongoing issues that could be addressed after a commercial release. Most of this chapter deals with information needs and interpretations prior to field release. These allow a comprehensive assessment of the pest/plant system and ecological attributes of the pests that help to define the risk of resistance and indicate possible resistance management approaches. Additional research during field testing should be used to address key assumptions and develop an effective, workable and acceptable resistance management plan, and to establish details of the monitoring and response system. The chapter addresses the following issues that can be identified prior to field release: © CAB International 2004. Environmental Risk Assessment of Genetically Modified Organisms: Vol. 1. A Case Study of Bt Maize in Kenya (eds A. Hilbeck and D.A. Andow)

209

Environ. Risk. Chap 07

23/8/04

9:32

Page 210

210

G.P. Fitt et al.

● ● ● ● ●

The range of target and non-target organisms associated with the transgenic plant that are susceptible to the transgene. Any history of resistance evolution among key pests. Potential resistance risks and possible requirement for resistance management. Components of a potentially workable resistance management plan. Monitoring requirements and contingency responses required to implement the plan.

We then discuss some issues to be considered after field release, but before commercial release, and conclude by presenting a possible resistance management plan for use of Bt maize if deployed for management of stemborers in Kenya.

Identification of Species at Risk Finding 1: The two stemborer species most likely at greatest resistance risk in Kenya are Chilo partellus and Busseola fusca. In addition, non-target Lepidoptera such as Helicoverpa armigera, Spodoptera exempta, Ephestia cautella, Ephestia kueniella and Plodia interpunctella may also be at risk for resistance evolution.

Four stemborers – Chilo partellus (Crambidae), Chilo orichalcociliellus (Crambidae), Sesamia calamistis (Noctuidae) and Busseola fusca (Noctuidae) – are the most damaging pests of maize in Kenya and neighbouring countries (Tables 7.1 and 7.2). Of these, C. partellus and B. fusca are the dominant pest species, being the most abundant and widespread and consistently associated with maize and other grain crops. Hence, they represent the most likely resistance risk for Bt maize. C. partellus is most common across lowland regions, and B. fusca in the highlands (Fig. 2.3 in Muhammad and Underwood, Chapter 2, this volume). Therefore, resistance risk varies geographically, and our analysis will reflect this geographic variability. Specifically, lowland maize production areas have resistance risk associated primarily with C. partellus (Lowland Tropics, Dry Midaltitude and Dry Transitional zones; B. fusca becomes a risk in the higher altitude areas of these zones). In the maize production area around Kitale, B. fusca is the sole resistance risk (primarily in the single-cropped Moist Transitional and Highland Tropics zones). It is in this region that most of the large-scale commercial maize production occurs. In the maize production area near the Lake Victoria shore, both C. partellus and B. fusca are significant resistance risks (primarily the Moist Mid-altitude zone). In the highlands, C partellus is the important resistance risk below 1500 m altitude, while B. fusca is the important resistance risk above 1500 m altitude (Moist Transitional and Highland Tropics zones). H. armigera (Noctuidae) is the main non-target pest of concern for resistance risk. H. armigera is consistently associated with maize, but is regarded as a minor or sporadic pest. It is a significant pest of many other crops and vegetables. H. armigera is susceptible to Cry1Ab (e.g. Bt maize)

Environ. Risk. Chap 07

23/8/04

9:32

Page 211

Resistance Risks and Management Associated with Bt Maize

211

Table 7.1. Maize growing season, and major stemborer pests in the six maize agroecological zones in Kenya based on Hassan et al. (1998) and Zhou et al. (2001). Agroecological zone (see Fig. 2.1)

Maize cropping scheme

% Farmers following cropping scheme

Growing season

Major borer pest Chilo partellus C. partellus C. partellus C. partellus C. partellus C. partellus C. partellus C. partellus C. partellus Busseola fusca B. fusca C. partellus and B. fusca C. partellus and B. fusca C. partellus and B. fusca

Lowland Tropics

Single maize crop Double maize crop

65 35

Dry Mid-altitude/ Dry Transitional

Single maize crop Double maize crop

44 56

Moist Transitional below 1500 m

Single maize crop Double maize crop

60 40

Highlands, Moist Transitional above 1500 m Moist Mid-altitude

Single maize crop Double maize crop

78 22

Long rains Long rains Short rains Long rains Long rains Short rains Long rains Long rains Short rains Long rains Long rains

Single maize crop

40

Long rains

Double maize crop

60

Long rains Short rains

Table 7.2. Maize agricultural regions defined by major pest status, associated resistance risk, and typical farm size in each region (Hassan et al., 1998; Zhou et al., 2001). Region and agroecological zone

Resistance risk

Farm size

Eastern Region (Lowland Tropics/Dry Mid-altitude/ Dry Transitional) North of Lake Victoria Region (Western Moist Transitional/ Highland Tropics) Lake Victoria Shore Region (Western Moist Transitional/ Moist Mid-altitude) Central Highland Region (Moist Transitional/Highland Tropics)

Chilo partellus

Small-scale

Busseola fusca

Small-scale and large-scale

C. partellus B. fusca

Small-scale

C. partellus <1500 m B. fusca >1500 m

Small-scale

and Cry1Ac (e.g. Bt cotton) proteins though considerably less than other Heliothine species such as Heliothis virescens (the primary target for Bt cotton in the USA) (Liao et al., 2002). H. armigera feeds largely on reproductive structures, particularly in developing maize ears. In the absence of information on the level of expression and efficacy of Cry proteins in reproductive structures of the Bt maize events evaluated to date, we cannot assess their

Environ. Risk. Chap 07

23/8/04

9:32

Page 212

212

G.P. Fitt et al.

likely impact on H. armigera larvae and the associated resistance risk. However, because survival of Helicoverpa zea on commercial Bt maize (field and sweet) does occur in the USA, there will likely be some pertinent resistance risk associated with H. armigera on Bt maize in Kenya. The interaction of H. armigera with Bt maize may also alter the seasonal dynamics of H. armigera and its pest status. S. exempta is an occasional pest feeding on maize foliage. Similar to other Spodoptera spp., S. exempta is not effectively controlled with Cry1A proteins and the threat of resistance development may be negligible. There is also a suite of significant storage pests of maize. These include the grain weevil and greater grain borer (Sitophilus zeamays and Prostephanus truncatus) (Coleoptera: Curculionidae), and the moths, E. cautella, E. kueniella and P. interpunctella (Lepidoptera; Pyralidae). Because the Cry proteins currently proposed for use in Bt maize have only Lepidopteran activity, there is no resistance concern with the weevils. However, because P. interpunctella has already developed resistance to Bt formulations in grain storage facilities in the USA (primarily because of the long persistence of Bt proteins in storage), E. cautella, E. kuehniella or P. interpunctella may be a resistance concern if mature maize grains express Cry toxins at high enough concentration to cause significant selection pressure on these moths (McGaughey, 1985).

History of Insecticide Resistance and Resistance Risk Knowledge of the past history of pesticide resistance can be instructive in assessing resistance risk. This can be clearly illustrated by two pest species of Helicoverpa in Australia. H. armigera and Helicoverpa punctigera are both significant pests of cotton and other crops where they are often controlled with a range of insecticides. H. armigera has historically evolved resistance to all major classes of insecticides deployed against it (e.g. endosulfan and synthetic pyrethroids: Forrester et al., 1993). Although H. punctigera has been subjected to as much selection pressure as H. armigera in cotton in Australia, it has not developed field resistance to any insecticide. This difference reflects the differing host range and mobility of the two species (Fitt, 1989), which result in a substantial proportion of the H. punctigera population occurring in unsprayed crop and non-crop situations. Populations of H. punctigera appear to be re-established in cotton areas each spring by immigrants from unsprayed inland areas (Gregg et al., 1995). By contrast, H. armigera populations are more consistently exposed to selection pressure within the cropping areas (Fitt and Daly, 1990). With the introduction of Bt cotton in Australia, this history of insecticide resistance clearly identified H. armigera as a resistance risk and hence the target for a pre-emptive resistance management strategy (Fitt, 1997). For the four main stemborer species in Kenya, there is no evidence of pesticide resistance, although pesticides are not extensively applied to maize, except in the large-holdings that account for only 8% of the maize holdings in Kenya. Small-scale farmers do not have the resources for pesticides or application. Muhammad and Underwood (Chapter 2, this volume) identify five

Environ. Risk. Chap 07

23/8/04

9:32

Page 213

Resistance Risks and Management Associated with Bt Maize

213

pesticides which are currently used in maize: Thiodan (endosulfan), Dipterex (trichlorfon), Bulldock (beta-cyfluthrin), Ambush (permethrin) and Pymac (the residue from Pyrethrum processing). Two of these are synthetic pyrethroids (Bulldock, Ambush) and all are likely to be disruptive to integrated pest management (IPM) approaches. Bt sprays (which contain a mix of four Cry proteins) are not currently used in maize or other grain production systems, but may be used extensively in vegetable production. ICIPE has recently established a Bt production facility, largely for production of product for mosquito control (Bacillus thuringiensis israeliensis – Bti) and for use in horticultural crops (Btk) (ICIPE, 2003). Whether this development may lead to a change in the use patterns of Bt sprays on field crops seems unlikely, but would need to be monitored and considered in assessing future exposure of stemborers to Bt proteins and hence resistance risk. We have no information on the use of pesticides against H. armigera or S. exempta on crops other than maize. There is evidence of pesticide resistance in H. armigera (S. Sithanantham, Nairobi, Kenya, 2002, personal communication), although no systematic monitoring of pesticide resistance has been conducted in Kenya. There appears to be no evidence of stemborer resistance to pesticides elsewhere in Africa, although this could be because of lack of exposure to pesticides or lack of monitoring of pesticide resistance. For Crambidae generally, there are some cases of pesticide resistance (e.g. rice stemborers), and for Noctuids there are many cases of resistance among the Heliothines. This may suggest a greater propensity for resistance among the Noctuids, or that Crambids have had lower historical exposure to insecticides than Noctuids.

Potential Exposure Defining potential exposure of insects to selection by Bt toxin is an important aspect of resistance risk assessment. Host range and mobility, the spatial arrangement of host plants within a landscape, differing expression levels of Bt in plant structures and different patterns of feeding among structures will all influence the potential exposure of insects feeding on Bt maize to Bt toxin. Insect populations may not coincide in time or space with Bt maize, may have only a small proportion of the local breeding population on Bt maize or may feed on tissues where Bt is not expressed. In this section, we characterize the potential exposure of the main stemborer species by summarizing what is known of their population ecology, behaviour and the structure of cropping systems where Bt maize may be deployed.

Exposure on target crop The occurrence on maize of the main stemborer pests varies with location and season (Fig. 2.3, Tables 2.9 and 2.10 in Muhammad and Underwood, Chapter 2, this volume). There are several agroclimatic zones, and in some of them,

Environ. Risk. Chap 07

23/8/04

9:32

214

Page 214

G.P. Fitt et al.

maize is double-cropped during both the long rains (March–August) and the short rains (October–December). In the cooler, high-elevation zones, maize is cropped only once a year corresponding to the long rains season. Both of the key pests overlap substantially and extensively with maize production, although there is considerable variation in the severity of attack by region and time in the season. Both C. partellus and B. fusca enter diapause during the winter in colddry regions. In B. fusca, this appears to be an obligate diapause, while C. partellus may have a facultative diapause and breeds year round in the warm low-lying coastal regions. Diapause during a maize production period may potentially reduce exposure of either species to Bt maize, but would also serve to preserve resistance genes between favourable periods for breeding. Finding 2: Insufficient information is available to determine how larval or adult movement and adult mating behaviour might influence the evolution of resistance in the at-risk stemborers. This information is critical to determining the feasibility of several resistance management options, and should be collected. Most of this information can be collected prior to field introduction of Bt maize.

Larval movement Larval movement is important for resistance management because it indicates how Bt maize could select for resistance. Specifically, for high-dose events, in which resistance is functionally recessive, the degree of larval movement may determine the minimal spatial scale at which mixing of Bt maize and non-Bt maize can be tolerated before it contributes substantially to increasing the rate of resistance evolution. Larval dispersal allowing movement between plants has two main consequences when a ‘seed mixture’ (mix in the same field of Bt maize with non-Bt plants suitable for stemborers) is used for delaying the evolution of resistance. First, the heterozygous larvae (i.e. RS or R1S1/R2S2 genotype for a single dose or pyramid approach, respectively) that emerge on, and move from, Bt plants could potentially survive when colonizing a non-Bt plant. This could reduce functional dominance of resistance and accelerate the evolution of resistance (Mallet and Porter, 1992), although there are many situations in which this would not occur (Tabashnik, 1994; Carrière et al., 2004a). Second, susceptible larvae hatching on, and moving from, a stemborer-suitable non-Bt plant could be killed when feeding on Bt maize. This would reduce effective size of the internal refuge and potentially accelerate the evolution of resistance (Tabashnik, 1994). Extensive interplant movement would compromise the potential for using seed mixtures as a means to provide effective in-crop refuges. Although the efficacy of a seed mixture may be compromised by larval movement, this is not always the case (Tabashnik, 1994; Carrière et al., 2004a). To evaluate the merit of a seed mixture strategy, it is therefore necessary to obtain information on the ability of the Bt plants to deliver a high dose (a high dose is better than a low dose for resistance management) and the frequency of larval movement (low movement is better than high movement). Moreover, data on feeding preference of larvae for Bt maize and the non-Bt plants in a mixture are also needed. For example, imagine that Bt maize is

Environ. Risk. Chap 07

23/8/04

9:32

Page 215

Resistance Risks and Management Associated with Bt Maize

215

intercropped either with host A or B. Suppose the rank order of larval feeding preference, from the lowest to the highest rank, is as follows: host A, maize and host B. Accordingly, movement and feeding rate of the heterozygous larvae from Bt maize to the non-Bt host could differ between the two mixture types. Similarly, the tendency of the Bt-susceptible larvae to leave the non-Bt host and feed on Bt maize could differ. While overly simplified this example emphasizes that differences in larval behaviour between different mixture types could translate in different rates of resistance evolution. Larval stemborers may disperse as early instars by ballooning on silks (techniques for assessing such dispersal are described in Berger (1989)). Following an initial phase of feeding on leaf surfaces, they enter the stem and feed concealed within it. During development, some larvae may exit the stem and move to other plants. The proportion of larvae moving is highly variable and dependent on larval density and food quality. If seed mixtures were to be considered as a refuge option (either alone or in combination with an external refuge; Tabashnik, 1994) then more information on the extent of post-feeding movement and the influence of levels of Bt expression on this behaviour would be required. For example, in a moderately expressing plant, early-instar larvae may detect Bt and move more often than recorded in conventional plants. Quantifying larval movement can be accomplished through field experiments where individual plants are artificially infested with marked larvae of known instar or history, and their movement to adjacent plants monitored daily or less regularly (Ross and Ostlie, 1990; Davis and Onstad, 2000). With cryptic feeding species, this may require a destructive sampling approach and considerable thought about experimental design. A capacity to maintain experimental plots essentially free of natural infestations is also essential. In Kenya, it is expected that small-scale producers will save seeds for producing the next year’s maize crop. Some of these seeds could result from fertilization of non-Bt maize with pollen from Bt maize (and inversely). Through time, this could create fields with a mixture of Bt and non-Bt maize plants (Johnston et al., Chapter 6, this volume). Fertilization of non-Bt maize by Bt pollen could often yield intermediate toxin levels in ears of non-Bt maize (C.F. Chilcut and B.E. Tabashnik, Tucson, Arizona, 2003, personal communication). This might result in intermediate toxin levels in the mixed fields and contribute in accelerating resistance to Bt maize in pest feeding on ears. Hence it will be important to quantify whether the pests that pose the greatest resistance risk feed on maize ears and investigate how larval movement from plant to plant could affect resistance risk. Adult emergence and mating An understanding of the timing of adult emergence and adult mating behaviour is crucial to design of a management strategy. Identifying when mating occurs in relation to the site and time of eclosion is critical. If all mating occurs soon after eclosion and in the natal patch, then inter-mating among refuge and Bt crop populations may be compromised. Very little information is available on mating dynamics of any of the stemborer species. Researching this in the field

Environ. Risk. Chap 07

23/8/04

9:32

Page 216

216

G.P. Fitt et al.

is challenging. Using a combination of mark-release and recovery of mating pairs could provide useful information, but would be labour intensive. Key questions are: ● ● ●

When do moths eclose? When and where do they mate after eclosion? How far do they move before first mating and subsequently?

Resistant larvae could develop more slowly than susceptible larvae, which could contribute to greater levels of non-random assortative mating (Gould, 1994; Liu et al., 1999). The effect of a developmental delay on resistance evolution, however, is likely to vary geographically for these stemborers. For most regions with more than one generation per growing season, developmental delays during the final generation may be unlikely to increase non-random mating. Emergence of adults during the final generation may be prolonged, with the last adults emerging too late to successfully parent offspring that can mature sufficiently to survive to the next growing season. Consequently, resistant adults emerging late during the final flight may either be likely to encounter many susceptible mates or be too late to parent offspring successfully. B. fusca and C. partellus are reported to mate soon after emergence in maize and perhaps in the site of emergence. It is not clear if this occurs in the first night after emergence or later. If it occurs later, there is greater potential for mixing before mating. It is uncertain if similar processes occur in association with the non-maize hosts. Adult movement Information on adult movement is needed for two purposes. First, adult movement will indicate the expected distances males and females move prior to mating and females move after mating. The most important movement parameter is the average distance moved, because it is the bulk of the population that largely determines how movement affects resistance evolution at these small spatial scales. These distances are crucial for understanding the local spatial scale in which resistance management must operate. For example, a high-dose event can be managed using a non-toxic refuge that is close enough to the Bt fields that males can readily disperse from the refuge to the Bt fields where they can mate with females in the Bt fields. For low-dose events, this use of movement information is less critical. The second use of movement data is to understand how resistance might spread geographically. Resistance may become more common at one location, but through adult dispersal, it could spread and contaminate other areas, where selection was not as intense, or where resistance management was more effectively established. Here the key movement parameter is the long-distance tail of the movement distribution. This can be measured accurately only when the scale of the observations of movement correspond to the scale of the tail of the distribution. For example, if the borer spreads at a rate of 100 km per generation, observations must be taken at approximately a 100-km scale as

Environ. Risk. Chap 07

23/8/04

9:32

Page 217

Resistance Risks and Management Associated with Bt Maize

217

well. One method to estimate movement uses estimates of gene flow based on FST coupled with the location of the sampled populations (see next section). Dispersal distances for the two key stemborers are lacking. The limited information available on adult movement suggests that most of the population moves relatively short distances within the crop or among adjacent vegetation, although this will be influenced by the quality of vegetation where adults emerge. Information on longer-range movement distances is sketchy. Movement of C. partellus may be quite restricted, perhaps to a distance of ~1 km, while the noctuid, B. fusca, is likely to be capable of extensive migratory flights under the right conditions. Seasonal variation in migration from refuges to Bt crop fields (e.g. from natural vegetation to maize fields) could have implications for resistance management (Andow, 2001; Guse et al., 2002). If such variation is suspected, mark–recapture experiments could be performed at different times during the maize-growing season. Rare long-range dispersal events are unlikely to influence the efficacy of refuges, but these events will influence the rate of spread of resistance to other geographical areas. Observations on the range expansion of C. partellus might be useful for quantifying the scale of long-range movement for this species. Methodology for studies of adult movement may involve mark–recapture techniques (Legg, 1983; Fitt et al., 1995; Carrière et al., 2001; Hunt et al., 2001; Showers et al., 2001; Cameron et al., 2002), observations on geographic range expansion of the species (Chiang, 1972) or a trait (Showers, 1993), spatially explicit studies linking configuration of refuges and Bt crops to insect density (Carrière et al., 2003, 2004b), use of stable isotope (Gould et al., 2002), or direct observation of dispersal events (Chiang et al., 1965). Inferences drawn from trapping networks using pheromone or light traps provide weak evidence of dispersal unless they are combined with additional information. Flight mill or wind-tunnel methods may indicate the propensity for flight but do not quantify movement distances. Geographic population structure The geographical population genetic structure of B. fusca and C. partellus is not known. This information gap is probably most serious for B. fusca. Information on population structure is needed to determine the geographic scale of a deme, which determines the spatial scale of resistance management efforts. If the scale is smaller than the maize agroecological regions described above, then resistance that evolves locally will not necessarily spread rapidly to the whole agroecological region. Alternatively, if the scale encompasses all of east Africa, then international cooperation would be necessary to ensure more effective resistance management. Population genetic data can be used to measure geographic population structure. For B. fusca, this could be done using co-dominant neutral genetic markers, such as allozyme variation (Wang et al., 1995; Bourguet et al., 2000; Han and Caprio, 2002). Data can be analysed by estimating heterozygote deficiency and FST. These methods are probably inappropriate to estimate gene flow in C. partellus, given that it was recently introduced to Kenya. One

Environ. Risk. Chap 07

23/8/04

9:32

Page 218

218

G.P. Fitt et al.

of the assumptions of these methods is that the species has been present for a long enough time that geographic variation in neutral gene frequencies reaches a stationary distribution caused by random genetic drift in the face of migration. For large populations and low levels of migrations, it may take hundreds of generations to reach this stationary distribution. For C. partellus, however, the fact that it spread rapidly over an extensive region of Africa might imply that FST measures on co-dominant neutral genetic markers would not provide clear information on spatial population genetic structure. For the purposes of resistance management, it might be necessary to manage this species on a large, international spatial scale.

Exposure on other plants Finding 3: All of the stemborer species on maize in Kenya, including C. partellus and B. fusca, use other crop and non-crop plant species in addition to maize. In many locations, these plants are intercropped in or planted nearby to maize fields. It is possible that for some borers these alternative host plants could contribute toward resistance management, but the information is insufficient to determine this. Preliminary studies suggest that non-crop hosts for B. fusca and C. partellus may be insignificant as refuges. These findings need to be confirmed and thoroughly assessed to ensure the most appropriate management strategy is identified prior to field release of Bt maize.

Other crop hosts C. partellus, B. fusca, S. calamistis and C. orichalcociliellus all attack other monocot crop plants in addition to maize (Table 7.3). Alternative crop hosts include sorghum (all four stemborers), sugarcane (all four stemborers), rice (C. partellus, S. calamistis), finger millet (B. fusca, S. calamistis and C. orichalcociliellus) and pearl millet (B. fusca). It will be important to determine: ● ● ●

What is the crop area for each of these crops in Kenya? To what extent do they overlap spatially and temporally with maize? Will borers emerging from these crops mingle and mate with borers emerging from maize?

Table 7.3. Crop hosts of major Lepidopteran cereal stemborers in Kenya (Overholt and Maes, 2001).

Maize Sorghum Rice Sugarcane Pearl millet Finger millet

Chilo partellus

Busseola fusca

Sesamia calamistis

Chilo orichalcociliellus

+ + + +  

+ +  + + +

+ + + +  +

+ +  +  +

Environ. Risk. Chap 07

23/8/04

9:32

Page 219

Resistance Risks and Management Associated with Bt Maize

219

Although we expect that these data are available, they were not available to us at the workshop. In the many publications on stemborers, there is little quantitative data relevant to these questions. Forage crop and non-crop hosts In addition to these crop hosts, all four stemborer species have been reported to attack a range of grass hosts in the Gramineae, Cyperaceae and Typhaceae (Table 7.4). Most of these are wild grasses but farmers grow some of them as fodder crops. Maize is an exotic crop for Kenya and the wild grasses represent the original host plants of the native stemborer species. Two grass hosts of stemborers are used in an IPM strategy for stemborer control. Napier grass (Pennisetum purpureum) and Sudan grass (Sorghum vulgare var. sudanese) are used as the ‘pull’ components of the ‘push–pull strategy’, which is now in the early stages of implementation in regions where mixed arable-livestock farming is common.1 The vast majority of Kenyan maize farmers are smallholders who intercrop maize (Muhammad and Underwood, Chapter 2, this volume). The intercrop varies with the agroecological zone. Beans are a common intercrop except for the lowland tropics, where maize is more commonly intercropped with cassava or cowpea. Intercropping with other monocots is less common, but intercropping with sorghum occurs in some areas and rice is sometimes intercropped with maize in the lowland tropics. Intercropping with suitable hosts for stemborers, from the point of view of resistance management, translates into a seed mixture or seed mixture+external refuge strategy, depending on the location and availability of suitable wild hosts. Larger farms that grow maize as a monoculture are only present in the western highlands of Kenya. Kenyan scientists are studying the presence of wild hosts in the vicinity of maize crops. Studies of grass hosts surrounding plots of maize in the eastern coastal province of Kenya (part of the Lowland Tropics zone where C. partellus is the dominant stemborer) indicate that the spatial extent of wild grass hosts may represent a natural refuge of approximately 20% compared to the maize crop. Farms and fields in this area are small and farmers usually intercrop maize. In western Kenya, where B. fusca is most common, the presence of wild host plants for the stemborers is considered to only represent a refuge of <5% compared to the area under maize (J. Songa, Nairobi, Kenya, 2002, personal communication). However, despite the existence of incidence records for wild hosts of stemborers (Table 7.4), there is a paucity of published data on actual utilization of these plants by either B. fusca or C. partellus. B. Le Ru (Nairobi, 2003, personal communication) found very few B. fusca outside of crops, and

1The

push–pull strategy not only provides control of stemborers but also of parasitic witchweeds (Striga spp.), a major weed problem in some areas of Kenya. Napier and Sudan grass are cut for livestock fodder. See Muhammad and Underwood (Chapter 2, this volume) and Nelson et al. (Chapter 3, this volume), for more information on the push–pull system.

CL – – ++ ++ ++ – ++ ++ – ++ ++ ++ ++ ++ – – ++ ++ ++ ++ ++ – ++ ++ –

Plant species

Gramineae (Poaceae) Andropogon sp. Brachiaria spp. Cenchrus ciliaris L. Coix lacryma-jobi L. Dactyloctenium bogdanii S. M. Philips Digitaria spp. Echinochloa pyramidalis (Lam.) Echinochloa haploclada (Stapf) Hyparrhenia cymbaria (L.) Stapf Hyparrhenia filipendula (Hochst.) Stapf Hyparrhenia pilgerana C.E.Hubbard Hyparrhenia rufa (Nees) Stapf Panicum deustum Thunb. Panicum maximum Jacq. Pennisetum macrourum Trin. Pennisetum procerum W.G.Clayton Pennisetum purpureum Schumach. Pennisetum trachyphyllum Pilg. Phragmites sp. Rottboellia cochinchinensis W.D.Clayton Setaria inrassata (Hochst.) Hack Setaria sphacelata (Schumach.) Moss Sorghum arundinaceum (Desv.) Stapf Sorghum versicolor Anderss Sorghum versicilliflorum (Desv.) ++ – ++ – – – ++ ++ – ++ – ++ – ++ – – ++ – ++ – – ++ ++ ++ –

SC – – – – – – – – – – – – – ++ – – ++ – ++ ++ – – ++ ++ ++

ES – – – – – – ++ – – – – – – – – – – – ++ – – – – – –

PB – – – – – – – – – – – – – – – – – – – – – – – – –

BS – – – – – – – – – – – – – – ++ ?? ++ ?? – – – – – – –

PS

220

– ++ – – – ++ ++ – ++ – ++ ++ ++ ++ – – ++ ++ – ++ ++ ++ ++ ++ –

BF

Insect species

Table 7.4. Wild hosts of Lepidopteran stemborers recorded from Kenya (based on Khan et al., 1997).

Environ. Risk. Chap 07 23/8/04 9:32 Page 220

G.P. Fitt et al.

++

++

–

++ – – ++ ++ ++ ++ – ++

++ ++ ++ ++ –

– – – –

++ – ++ + –

–

++ ++ ++ ++

++ – – – –

–

– – – –

– – – – –

–

++ ++ ?? ??

– – – – –

–

– – – –

– – – – –

23/8/04 9:32

Information collected and compiled from different agroecological zones of Kenya: ++, recorded as host; –, not recorded as a host; ??, host status not confirmed; CL, Chilo sp.; BF, Busseola fusca; CS, Sesamia calamistis; ES, Eldana sacharina; PB, Phragmataecia boisduvalli; BS, Bactra stagnicolana; PS, Poenoma sp.

Sorghum vulgare Pers.var. sudanense Sporobolus pyramidalis P. Beauv. Sporobolus marginatus auct. Hochst Tripsacum laxum Nash. Vossia spp. Cyperaceae Cyperus distans L.f. Cyperus immensis Delile Cyperus maculates Boeck Cyperus papyrus L. Typhaceae Typha domingensis Pers.

Environ. Risk. Chap 07 Page 221

Resistance Risks and Management Associated with Bt Maize 221

Environ. Risk. Chap 07

23/8/04

9:32

Page 222

222

G.P. Fitt et al.

suggests that wild hosts cannot provide adequate refuge, while J. Kanya (Nairobi, 2003, personal communication) suggests that wild hosts may represent only an 8% refuge for stemborers. These studies need to be finalized before firm conclusions can be made on the adequacy of wild hosts for either of the highrisk species. Estimates of the adequacy of natural refuge must take account of oviposition preferences for different hosts and include estimates of their relative suitability for stemborer development compared to maize as a host. Although stemborers attack a wide range of grasses, these appear generally to be less suitable for larval development than maize. Based on laboratory experiments, survival of larvae is reduced and their development can be extended on many wild hosts. This might imply that mating success of moths from wild hosts would be lower than for moths from maize, reducing the efficacy of the wild hosts as external refuges. This implication needs to be tested scientifically. Wild hosts represent alternative habitats for stemborer populations during the time when no maize crops are present. Wild hosts present at these times do not contribute directly to resistance management. Recommendations: ●

●

Field studies comparing the phenology and fitness of stemborers in wild grasses and maize are recommended to ensure that emergence of moths from maize and the refuge overlap and that mating of individual moths from the crop and the refuge can occur. Further field studies are recommended to establish the proportion and quality (e.g. body size) of moths generated in a particular area by grass hosts and other crops compared to maize.

Genetic integration of populations on other plants Bourguet and colleagues (Martel et al., 2003; Thomas et al., 2003) demonstrated that European corn borer populations are divided into two spatially overlapping (sympatric) populations, one associated with maize and the other associated with a co-occurring weed, Artemisia vulgaris (mugwort). This implies that the substantial populations associated with mugwort would not be an effective refuge for maize, and without knowing that they do not interbreed, the effectiveness of the ‘refuge’ would be overestimated. Hence, it is necessary to confirm that populations of stemborers on alternate hosts actually do interbreed with the populations associated with maize. Based on geographical variation in host plant use and location of larvae feeding on sugarcane, it was suggested that different biotypes of the African sugarcane borer Eldana saccharalis biotypes may occur across Africa (Conlong, 2001). It is not certain whether more than one biotype overlaps in the same area (sympatric), or whether the biotypes are mainly isolated geographically. In some stemborer species, a fraction of the population may enter diapause in maize fields during the dry season while individuals exploiting wild hosts remain active (Kfir et al., 2002; Zhou et al., 2003). This would have no consequences for resistance management, as long as there are moths emerging from wild hosts or non-Bt maize to mate with moths produced by the diapausing larvae.

Environ. Risk. Chap 07

23/8/04

9:32

Page 223

Resistance Risks and Management Associated with Bt Maize

223

Other crops with Bt transgenes Other crops with Bt transgenes could increase exposure of the stemborers to Bt toxin and result in more rapid resistance evolution (Fitt, 1997). No Bt crops have so far been commercialized in Kenya or its neighbouring or nearby countries. The major African cereal stemborer species also attack sorghum, sugarcane, rice, finger millet and pearl millet in addition to maize (Table 7.2). Sugarcane and rice have been transformed with insect resistance genes such as Cry 1Ab elsewhere. Stemborers are also important pests in sorghum and transformation of sorghum with Bt genes may be considered in the future. Abundance of external or internal refuges for Bt maize may be reduced if Bt events of these crops were planted in Kenya. Other issues may arise with the African bollworm, H. armigera, if Bt cotton is proposed as a future management strategy for this species. Although not a key pest of maize, H. armigera feeds in the ears and cobs of maize. Exposure to Cry1Ab in cobs of Bt maize would add to selection pressure in a Bt cotton system that would likely express Cry1Ac in cotton-producing regions of Kenya, Uganda and Tanzania. Cry1Ab and Cry1Ac are similar in mode of action and are widely reported to confer cross-resistance (Gould et al., 1992; Akhurst et al., 2003). Cross-resistance to Cry1B is less certain. Although there is one report of crossresistance between Cry1Ac and Cry1B, there also are two reports showing no cross-resistance between Dipel-resistant diamondback moth (Plutella xylostella) and Cry1B (W. Moar, Nairobi, Kenya, 2002, personal communication). Potential introduction of all future Bt crops expressing Cry1A proteins will need to consider the potential overlap of pests like H. armigera. Exposure associated with the farming system Finding 4: It may be necessary to have different resistance management strategies for the large-scale producers in the higher altitude western Kenyan regions and for the smaller-scale producers throughout the rest of the country.

Most maize in Kenya is grown on small intercropped plots (Muhammad and Underwood, Chapter 2, this volume). Wild host plants of stemborers grow in close vicinity to these plots. The push–pull strategy encourages planting of grass hosts close to the maize crop. Both intercropping and the push–pull strategy may provide refuges for susceptible moths. However, the push–pull strategy currently recommends regular cutting of the grass host, to prevent stemborers from completing development and to use the grass as fodder. A modification of this strategy would be required for the fodder grass to serve as a refuge that produces susceptible moths. This modification may be related to the timing of cutting of the grass, so that it serves effectively at producing some moths as well as attracting oviposition. In the western Moist Transitional and Highland Tropics zone around Kitale, maize is often grown in large monocultures without intercropping (Muhammad and Underwood, Chapter 2, this volume). In these systems, the refuge represented by grasses is much more limited than in other areas of Kenya. Slow-maturing, higher-yielding hybrid cultivars are grown in this area with the

Environ. Risk. Chap 07

23/8/04

9:32

224

Page 224

G.P. Fitt et al.

result that maize may be available to stemborer populations for up to 9 continuous months. Indeed, a very small area of irrigated maize and sweetcorn may be grown year round. This is likely to result in stemborers having multiple generations in maize or even continuous breeding in maize in small areas. Potential exposure to Bt in Bt maize would be very high. In the rest of the country, maize is present only during the rainy seasons (short and the long rains), each lasting 3–4 months. In double-cropped areas, approximately 90% of the annual maize crop is grown in the long rains (March–August), with the remaining 10% during the short rains (October–December) (Muhammad and Underwood, Chapter 2, this volume). In these regions, there may be considerable value in restricting use of Bt maize to the long rains season only. Having no exposure during the remainder of the year would aid resistance management, but this level of management may be very difficult to achieve. The majority of subsistence-oriented smallholders may occasionally use purchased certified seed, but in most cases they plant recycled seed from their own farm or from other informal sources (Muhammad and Underwood, Chapter 2, this volume). Large farmers use purchased certified seed as a standard practice. Despite the widespread use of recycled seed, approximately 900,000 ha of hybrid maize in grown annually in Kenya. This represents 60% of the total crop area when total area is about 1.5 million ha. Projected future exposure The purpose here is to consider future scenarios that could change our present evaluations of potential exposure. These scenarios include any possible changes in the farming system (including pest management practices) or regional production system that could influence exposure to the transgenic plant. For example, if the species has two hosts, a repellent could be applied to the non-transgenic host, forcing greater exposure of the species to the transgenic host, or the transgenic plant could precipitate a change in rotation practices, leading to greater exposure of the species. Shifts in government supports or prices of crop commodities could lead to changes in cropping practices region-wide. Expansion of large, confined animal-production facilities could shift demand for some commodities, changing exposure. Finding 5: The effect of seed saving associated with transgenic open-pollinated varieties (OPVs) on resistance risk needs to be evaluated.

Adoption pattern According to Poland et al. (2003), the IRMA project is likely to deploy Bt genes firstly in farmer varieties (predicted for 2007), to provide a new pest management tool for resource-poor farmers. These varieties will be followed in 2009 by high-yielding Bt hybrid varieties and in 2011 by Bt OPVs, providing options for small farmers and for large-scale farmers in the western highlands of Kenya. While large growers purchase seeds every year, smallholders usually retain their own seed (Muhammad and Underwood, Chapter 2, this volume).

Environ. Risk. Chap 07

23/8/04

9:32

Page 225

Resistance Risks and Management Associated with Bt Maize

225

Most plants in Bt maize fields of large farmers are therefore likely to retain high levels of Bt toxin expression. Smallholders may experiment with Bt OPVs, hybrid varieties, and their own varieties and landraces. This could generate considerable variation in purity and segregation of genes within landraces leading to a proportion of non-expressing plants. Gene flow, seed saving and maize cropping area How quickly such introgression from hybrids or OPVs may occur is difficult to assess, but it could be rapid (Haygood et al., 2003). Johnson et al. (Chapter 6, this volume) provides a thorough analysis of the genetics of such gene flow. The conclusion from this analysis that is most relevant to resistance risk assessment is that seed saving and gene flow will result in heterogeneous Bt and non-Bt maize populations in the subsequent generations, which will vary in composition depending on the strength of selection for the Bt toxin gene and the amount of Bt maize used in the region. This is true whether the Bt variety is a hybrid, a homozygous OPV or an OPV that is a mixture of homozygous and hemizygous Bt plants. The effect of these heterogeneous mixtures on resistance risk has not been assessed. It is difficult to predict whether the availability of Bt maize would allow an expansion of the total maize cropping area overall. Changing market opportunities and local needs would likely have a stronger influence.

High Dose vs. Low Dose Finding 6: None of the presently available Bt maize varieties appears to be a highdose event against the key maize stemborers in Kenya.

A ‘high-dose’ transgenic event produces a high enough ‘dose’ or concentration of toxin that nearly 100% of heterozygous larvae are killed (i.e. larvae that bear a single copy of a major resistance allele). A ‘low dose’ does not accomplish this. High dose can be determined when strains of the target pest that survive on a Bt crop are available. In this case, a Bt variety provides a high dose when it can kill nearly 100% of the insects from a heterozygous strain (obtained by crossing individuals from a Bt-resistant strain and from a Bt-susceptible strain). Because there are no strains of African stemborers identified that can survive on Bt crops, it has become necessary to develop an operational, but imperfect definition of high dose. One such definition is that an operational high dose is one that is 25 times the LC99, as estimated with susceptible individuals. A high dose has not always been achieved, nor is it absolutely necessary to maintain effectiveness of a Bt crop over at least a moderate period (Tabashnik et al., 2003a). For example, Bt cotton does not provide a high dose for H. armigera and H. zea, and yet after being used for 8 years, Bt cotton is still effective for controlling these pests (Tabashnik et al., 2003a). In these cases, large refuges, and perhaps expression of fitness costs on alternative hosts, may have contributed in delaying the evolution of resistance (Carriere et al., 2002; Tabashnik et al., 2003a). Nevertheless, simulation models clearly show that a

Environ. Risk. Chap 07

23/8/04

9:32

Page 226

226

G.P. Fitt et al.

high dose can delay the evolution of resistance more effectively than a low dose (Comins, 1977; Taylor and Georghiou, 1979; Tabashnik and Croft, 1982; Roush, 1994; Alstad and Andow, 1995; Gould, 1998; Tabashnik et al., 2003b). A high dose may also be advantageous because it allows greater options for resistance management and allows for greater use of the transgenic crop. Moreover, high-dose resistance management may not put many restrictions on how the non-transgenic refuges are managed (Ives and Andow, 2002; but see Carrière and Tabashnik, 2001; Onstad et al., 2002; Storer et al., 2003 for a different and important perspective), and will thus be more readily implemented compared to low-dose events. Low-dose events will require larger non-transgenic refuges and/or restrictions on the management of these refuges. Table 4.8 (Andow et al., Chapter 4, this volume) summarizes the results from a preliminary evaluation of the level of mortality caused by several Bt maize events. While these events are not necessarily those proposed for release, these leaf tissue assays suggest that none of them express at high-dose levels for B. fusca. The highest mortality reported for this species was 59%. For C. partellus, the private sector event 176, and the public sector events 5601, 1835 and 7 may express at high-dose levels, because mortality was ≥98%. However, even for this species, additional testing would be needed to determine if the plants express toxin at sufficiently high concentrations to meet the operational definition of high-dose given above.

Bioassays Bioassays generating LC50 and LC95 data are recommended (or other bioassays measuring growth suppression or other sublethal effects that have been previously correlated with LC50s). Plant specific testing should include: 1. Plant tissue – plant tissues that the target insect pests either feed on or cause damage to should be tested. For Bt maize and stemborers, plant tissues to be tested should include leaves, stalks, pollen and fruiting tissue such as cobs. 2. Tissue age – especially for leaves and stalks, tissues of various ages or plant phenology should be tested. 3. Transgene expression – the actual amount of transgene expression should be quantified in each tissue to be evaluated. Additional details are provided in Andow et al. (Chapter 4, this volume).

Laboratory Selection for Resistance It is desirable to generate strains of stemborers that can survive on a Bt crop to obtain relevant information on factors that influence the evolution of resistance in the field, such as dominance of resistance or presence of fitness costs (Tabashnik et al., 2003a). One way to obtain resistant laboratory strains involves splitting of a field-collected strain into two groups and imposing selection for resistance on one group using bioassay selection methods (Bolin et al., 1999; Tabashnik et al.,

Environ. Risk. Chap 07

23/8/04

9:32

Page 227

Resistance Risks and Management Associated with Bt Maize

227

2000), while keeping the other group as a control. The rate at which resistance may be selected is highly dependent on the initial frequency of resistance alleles. It is recommended that laboratory experiments be initiated to select resistant strains for the target proteins in the primary key insect pest(s). If resistant colonies of the same insect species have already been selected elsewhere, these should be obtained to provide a full suite of potential resistance alleles for future research. Clearly, the genetic composition of insect populations will vary geographically and the genetic basis of resistance could vary from one region to the next.

Frequency of Resistance For Bt maize to be an effective control tactic, the frequency of resistance must be rare in the target pests. This is also necessary to enable effective management of the risk of resistance evolution. The frequency of resistance in any of the Kenyan maize stemborer pests or other species potentially at risk is unknown. For B. fusca, C. partellus and possibly H. armigera, geographically different populations should be monitored to determine potential variability of response to the Cry1Ab, Cry1B and Cry2A proteins as well as to Bt maize and to provide baseline data to which future monitoring can be compared. This assessment could be in the form of bioassays, keeping in mind that it would be most useful to estimate resistance allele frequency using one of the methods suggested below in the Monitoring section.

Potential Risk Using the information on history of resistance, exposure, dose and the frequency of resistance, the insect species can be ranked according to their potential resistance risk, allowing identification of the weak link in the pest management system that may be exploited for pre-emptive resistance management. This includes identifying species with no or low resistance risk, so that further analysis is not necessary, and identifying species with high risk, which must be considered further.

Operational definition of resistance Resistance is caused by genetic mutations that lower susceptibility to a toxin. Resistance is thus a trait of an individual. However, resistance occurs in the field when the frequency of resistant individuals (and hence the frequency of resistance alleles) is high enough to cause unacceptable damage to the target crop. Hence, resistance can also be defined as a characteristic of a population and could be manifested as a field failure a few generations after deployment of a Bt crop. Operational definitions of resistance in the laboratory might involve a change in survivorship or growth rate after exposure to a discriminating dose of toxin.

Environ. Risk. Chap 07

228

23/8/04

9:32

Page 228

G.P. Fitt et al.

Resistance must be characterized phenotypically as lower susceptibility and must be demonstrated to be genetic. Resistance needs to be defined operationally as an individual trait: (i) for characterizing individual insects in the lab, such as during selection for resistance colonies; and (ii) for evaluating individuals acquired during monitoring. In addition, because resistance occurs in populations of the insect pest, population-level operational definitions are needed to (iii) determine when control failures are imminent and (iv) document the occurrence of a control failure caused by resistant pests. Here we provide an operational definition for (ii) and (iv). Definition (ii): An individual is resistant if it can survive from neonate larva to adult when feeding on Bt maize. If it can survive to a more advanced instar than a susceptible insect, then it may be considered partially resistant. Depending on the dominance of resistance, a putatively resistant individual could be either a RR homozygote or a RS heterozygote. In many cases, it will not be possible to determine if an individual is resistant because when it was collected, it would already be too old to be reared for its entire life on Bt maize. In this case, offspring of the individual must be tested. It is possible to determine the phenotype of such a putatively resistant individual by evaluating its offspring when the individual is crossed with a known susceptible individual. If the individual truly bears one or more resistance alleles, then resistance could be recessive and the individual could be RR, or resistance could be more dominant and the individual could be RS or RR. In either case, some of its offspring will also be resistant. Conversely, if it was phenotypically susceptible, then all of its offspring will also be susceptible. Thus, evaluation of the F1 offspring from a cross between a putatively resistant and a susceptible individual could allow determination of resistance of individuals. Resistance has evolved in the field when the frequency of resistance alleles or the frequency of resistant individuals (hence the frequency of the resistance alleles) has increased. This increase can be compared statistically to a baseline frequency that was established before the toxin was first used or to any previously estimated frequency. Resistance is most convincingly demonstrated when the resistance frequency is continuously increasing over time in a series of observations. Definition (iv): Often we focus on the practical outcome of resistance. For example, resistance may be viewed as the point when the frequency of resistant individuals becomes high enough to cause significantly more damage to the Bt maize variety than when it was first commercialized, or as much damage as occurs on non-Bt varieties. Hence, an operational population criterion defining control failures could be when the damage to the Bt crop reaches unacceptable levels. Thus defined, resistance is a characteristic of an insect population. Alternatively, a laboratory strain can be considered operationally resistant to a toxin when a statistically significant difference in expression of a life history trait (survival, fecundity, development time, etc.) occurs between a susceptible strain and the resistant strain when both strains are exposed to the Bt toxin. More generally, resistance of field-derived strains can be assessed by comparing their response to a toxin to the response of a susceptible reference strain.

Environ. Risk. Chap 07

23/8/04

9:32

Page 229

Resistance Risks and Management Associated with Bt Maize

229

Variation in resistance among strains can be studied using exposure to a single or many toxin doses (e.g. critical dose, discriminating dose, dose–response curves), or to a Bt crop in the greenhouse (Tabashnik et al., 2000, 2002).

Risk of resistance Finding 7: B. fusca poses the greatest resistance risk in the single-cropping regions of western Kenya and the high elevation areas of the Central Highlands of Kenya. These include the areas in Kenya with large-scale production and high use of hybrids. C. partellus is a resistance risk in the Lowland Tropics, and the doublecropped and lower elevation regions of Kenya. These are the areas of Kenya with small-scale production and high use of OPVs.

In the absence of any data on the background frequency of Bt-resistance genes or a past history of resistance, relative resistance risk can be assessed only on the basis of ecological factors and apparent efficacy of the current assessed range of Bt plants. Unfortunately, very little is known about mating behaviour or adult movement of borer species, as reviewed above. These are critical attributes for determining resistance risk (Table 7.5). All species may have limited mobility and localized breeding populations, which could result in locally intense selection for resistance.

Table 7.5. Ecological/behavioural attributes of stemborers and the efficacy of Bt maize against each species (from Songa et al., 2002); information used to make preliminary resistance risk assessment. Attribute

Chilo partellus

Chilo orichalcociliellus

Sesamia calamistis

High?

Not high

Not high

Facultative

No

Larvae balloon one to several metres ? 4

Facultative, in dry season No diapause in coastal region Larvae balloon one to several metres ? >5

Larvae balloon one to several metres ? ?

Larvae balloon one to several metres ? >6

?

?

?

?

40–65

25–50

??

40–70

? 400 30–50

? 450 10–80

? 450? ?

? 350 10–40

Busseola fusca

Toxicity of current Not high, low? Bt plants Diapause Yes, 3–6 months over winter Larval dispersal

Adult dispersal Generations/year Generations in maize Duration of life cycle (days) Abundance in wild hosts Fecundity Egg batch size

Environ. Risk. Chap 07

23/8/04

9:32

230

Page 230

G.P. Fitt et al.

Based on the limited toxicity data (High Dose vs. Low Dose section) and relative pest status, we suggest that B. fusca poses the greatest resistance risk to Bt maize in Kenya among all of the species potentially at risk. In other words, B. fusca is the weak link in the system and wherever B. fusca dominates the pest fauna, resistance management should be designed for this species. Tables 7.1 and 7.2 indicate that this would be some regions of the western part of Kenya and the higher elevation areas of the Central Highland region. Within these regions are the large-scale maize production areas that use hybrid varieties. B. fusca is less of a resistance risk in the Tropical Lowlands, lower elevations and almost anywhere that double-cropping prevails. In these regions, we suggest that C. partellus has the greatest resistance risk, because it is the main stemborer pest. Hence, we suggest that B. fusca and C. partellus pose a different resistance risk in the different agroecological zones for maize. Resistance risks posed by the main non-target pest, H. armigera, are difficult to assess. Helicoverpa armigera occurs throughout maize, grain, legume and vegetable crops. It is exposed to some pesticides in maize, but may be exposed to intensive pesticide use in vegetables. As a highly mobile and polyphagous species, it would be necessary to assess the level of geographic separation of grain, legume and vegetable production systems (likely to be highly intermingled). Polyphagy and mobility will operate to reduce selection pressure of Bt maize on H. armigera, although localized intense selection may occur. More detail is required about H. armigera ecology in the region. S. exempta is likewise polyphagous and mobile, and relatively insensitive to Cry1Ab. Its potential exposure to Bt maize in relationship to other plant hosts needs to be assessed. The resistance risk of E. kueniella and P. interpunctella was not assessed.

Monitoring Methods Monitoring methods (Andow and Alstad, 1998; the following text follows Andow and Ives, 2002) that could be used in resistance management for maize stemborers are: (i) screening field-collected egg masses; (ii) screening fieldcollected larvae (Tabashnik et al., 2000, 2003a); (iii) an in-field Bt maize screen (Andow and Hutchison, 1998; Tabashnik et al., 2000; Venette et al., 2000b); and (iv) an F2 screen (Andow and Alstad, 1998, 1999). The first two methods rely on laboratory discriminating dose assays on either neonate or older larvae (Roush and Miller, 1986; Tabashnik et al., 2000, 2003a). The third method relies on either a discriminating dose bioassay or Bt plants to supply the discriminating dose. The F2 screen is a more sensitive genetic screen, in contrast to the others, which are phenotypic screens. Four additional methods are Bt field maize sampling (Pierce et al., 1998), screening against test stocks (Gould et al., 1997), a molecular-based screen (Morin et al., 2003) and farmerbased monitoring. Field maize sampling may be logistically inefficient, and will not be discussed further. Use of test stocks and a molecular-based screen are not yet possible for maize stemborers because resistance has not yet been recovered. Farmer-based monitoring will be discussed in a later section.

Environ. Risk. Chap 07

23/8/04

9:32

Page 231

Resistance Risks and Management Associated with Bt Maize

231

Egg mass screen A screen based on field-collected egg masses might be done by collecting egg masses from the field, delivering them to a lab for testing, hatching the egg masses in the lab and screening the neonates using a laboratory discriminating dose assay. The sample unit is the egg mass, not the larva, because all larvae in an egg mass are related (i.e. they are not independent samples from the population). Although statistical analysis of the data should also be adjusted for egg mass size, for simplicity we assume that each egg mass is the same size and there are no ambiguities in testing the neonates as resistant or susceptible. Because the method screens for resistant phenotypes, when resistance is rare the statistical detection limit is 1/4N for dominant alleles and 1/(4N)1/2 for recessive alleles, where N is the number of egg masses sampled. The method cannot be any more precise than this theoretical detection limit. It should be clear that this method is statistically inefficient for recessive alleles. For example, a sample of 100 egg masses has a detection limit of 0.0025 for dominant alleles and 0.05 for recessive alleles. This means that for a sample of 100 egg masses, the theoretically best resolution of allele frequency is 0.0025 for dominant alleles and 0.05 for recessive alleles. In practice, the statistical precision of this method is somewhat less than this theoretical detection limit. High-dose resistance management requires that resistance is recessive and <0.001 (Roush and Miller, 1986; but see Carrière and Tabashnik, 2001; Carrière et al., 2002; Tabashnik et al., 2003a, for a different perspective). If resistance allele frequency is low, egg mass screening provides little useful evidence. Nevertheless, resistance allele frequency has been found to be high in at least one pest–crop system (Tabashnik et al., 2000, 2003a). The costs for conducting egg mass screens can be estimated by calculating the costs per sample, Cs, and the travel costs associated with getting to the sample location along with fixed costs associated with that location, Cl. The travel costs associated with acquiring the sample are proportional to the number of person-days needed to collect the required sample. For a desired detection limit, D, the necessary sample size, N, for a recessive allele is 1/4D2 and for a dominant allele is 1/4D. Letting Nd be the average number of samples a person could acquire in a day, and cl be the per capita travel costs, then N

 Nd

  



C1 = c1Round 

where the Round function rounds up the bracketed expression to the nearest integer. For sampling European corn borer in the USA (note: these costs probably do not apply in Kenya specifically or in Africa generally, but are provided to illustrate how costs can be calculated), we assume that travel time is a 2-h roundtrip and labour is US$10/h, so cl=US$20. Nd will depend on egg mass density and the number of plants a person can examine in a day. We assume that a person can examine 720 plants in a 6-h day, and that the

Environ. Risk. Chap 07

23/8/04

232

9:32

Page 232

G.P. Fitt et al.

average egg mass density is 0.05/plant, so Nd=36. Sample costs include collecting costs and laboratory costs, both supplies and labour. Per sample collecting costs are (US$10/h)(6h)/Nd and laboratory expenses are estimated at US$2.00/egg-mass. Cs is calculated to be US$3.67/sample. The total variable cost of this method and all subsequent methods is Ct=Cl + CsN. Larval screen A screen of field-collected larvae might be done by collecting larvae in the field (presumably later-instar larvae), transporting them to the lab, and conducting individual discriminating dose assays on the collected larvae. The sample unit is the larva, because it is screened directly. If the larvae must be reared and mated before screening, the costs are considerably higher (Bolin et al., 1998). The statistical detection limit for the larval screen is 1/2N for dominant alleles and 1/N1/2 for recessive alleles, which is not as good as the egg mass screen. Similar to the egg mass screen, it is very inefficient for rare recessive alleles. It is even less efficient if the larvae must be reared and mated before testing. Note that, compared to the egg mass screen, the per-sample (larva vs. egg mass) detection limit is higher for larvae than for egg masses, because larvae contain only single alleles from each parent, while egg masses contain all resistance genotypes possible for progeny of the parental pair. The costs for conducting this assay can be calculated similarly to the egg mass screen above. Nd will be a function of how many plants a person can dissect in a day, the density of larvae that are recovered from those plants, and the survival rate of larvae for testing. In the USA for European corn borer, we assume that a person can dissect 72 plants in 6 h, recovering larvae at a density of 0.5 per plant. Of these larvae, about 40% may die before they can be tested (Venette et al., 2000a); we assume here that 75% will survive for screening. This means that Nd=27. Per sample collecting costs are (US$10/h)(6 h)/Nd and laboratory expenses are estimated at US$1.00 per larva. Cs is calculated to be US$3.22 per sample. In-field screen An in-field maize screen involves planting and observing larval survival on a Bt maize field nearby an agronomically similar unsprayed non-Bt maize field (Tabashnik et al., 2000; Venette et al., 2000b). These fields can be deliberately planted for this purpose, or can be fields that happen to be adjacent. The sample unit is the larva, and although statistical analysis is complicated, the statistical detection limit converges to that of the larval method, because this method, like the previous two methods, is also a phenotype screen. The detection limit for large samples converges to 1/2N for dominant alleles and 1/N1/2 for recessive alleles, so the required number of samples, N, to attain a given level of sensitivity, D, is similar to the previous methods.

Environ. Risk. Chap 07

23/8/04

9:32

Page 233

Resistance Risks and Management Associated with Bt Maize

233

The main advantage of this method over the previous ones is that it may be considerably less expensive, because it is accomplished primarily by using the actual Bt plants in the field to screen insects. The costs for the method are related to Nd, a function of density of larvae in the sampled plant part and the number of these plant parts a person can examine in a day. If the plant part can be inexpensively sampled, the method can be quite efficient. Some maize stemborers aggregate in the ear at certain times of the season. For example, European corn borer larvae aggregate in the ear tips after silking and before the grain begins to dry (Venette et al., 2000a). In the USA, this suggests that Nd=900 and Cs=US$0.067, about two orders of magnitude less than the previous methods. Cl also includes the fixed costs of growing the fields. This method can be operationally implemented to measure resistance for the stemborer species. A field represents a sampling unit. Thus, average survival across many fields for a species would establish its baseline survival level (e.g. average number of large larvae of a species in 100 or more plants sampled in each of five fields in a given area; we assume here that the large larvae survive to adulthood). Survival levels obtained in subsequent years in the same way could be compared with the baseline level to quantify the progression of resistance (e.g. by looking at the overlap between the 95% confidence intervals for the mean survival levels obtained in different years).

F2 screen An F2 screen is done by collecting mated adult females from the field, transporting them to the lab, collecting the eggs of those females, rearing the F1 larvae, sib-mating the F1 families, collecting egg masses, and exposing neonates to Bt maize in the field (Andow and Alstad, 1998). This method, unlike the previous ones, is a genetic screen and will be able to detect any resistance allele that is present in the collected female, including those of her mates. Leaving aside the issue of statistical accuracy, which is dealt with in Andow and Alstad (1998, 1999) and Schneider (1999), the statistical detection limit of this method is 1/4N for both dominant and recessive alleles. This screen is particularly efficient for recessive alleles. For example, a sample of 100 female lines has a detection limit of 0.0025 for recessive resistance alleles, which is about two orders of magnitude more sensitive than the phenotypebased screens for similar sample size. The F2 screen is particularly efficient for species that are not cannibalistic as larvae. This is the most labour-intensive monitoring method, with a per sample variable cost of US$14.90 per female European corn borer line in the USA (Andow et al., 2000). The collecting costs will depend on how many females can be trapped in a single night. If 100 females can be caught in one night (Nd), with 2 h of labour and 2 h of travel time, cl=US$40. Even though it is the most labour-intensive method, it can be the most informative method (Ives and Andow, 2002), as discussed below.

Environ. Risk. Chap 07

23/8/04

9:32

Page 234

234

G.P. Fitt et al.

Possibility for Resistance Management The purpose of this section is to determine if resistance management is biologically possible by evaluating possible methods for reducing the selective advantage of resistance. Feasibility, acceptability and implementation should be considered at this stage of the assessment process and thoroughly addressed when field testing commences. ● ●

Recommendation 1. Pre-emptive resistance management should be implemented consistent with successful IPM approaches already being used. Recommendation 2. Pyramiding of multiple Bt transgenes to attain a ‘highdose’ transgenic event is recommended to aid in resistance management. The Cry toxins should act independently to minimize the risk of crossresistance (Andow et al., Chapter 4, this volume). Finding 8: Refuges are a critical component of resistance management but more information about the productivity of potential refuge hosts is needed before a refuge implementation strategy can be developed. There is a need to evaluate theoretically the resistance risk associated with using both in-field seed mixtures and external refuges, because this is likely to occur commonly in Kenya. Restricting the use of Bt maize to the first growing season of the year when stemborers are a problem could help reduce resistance risks. More information needs to be developed to determine if methods to alter movement or mating frequencies could be effective in resistance management.

Methods to reduce exposure Reducing exposure of the target pest to Bt maize is one way to reduce the rate of resistance evolution, because it reduces the selective advantage of the resistant pests over susceptible ones. Some methods that could be used to reduce the exposure of the species to the transgenic crop include limitations on the area of the GM crop planted, size and placement of refuges, timing of planting, etc. Refuges are critical components of the management strategy. A refuge is a location where the target pest population is not exposed to the Bt toxin. Refuges are variously measured, but the most appropriate measure is the proportion of the target pest population that is not exposed to Bt toxin at a specific point in time. Because this is sometimes difficult to measure, refuge requirements are often translated into area requirements based on scientific information about the population density of insects associated with different refuge vegetation. Using known biology and ecology of the target species it would be possible to estimate required refuge proportions using simple models which are readily available. For high-dose Bt maize events in the USA, a minimum 20% maize refuge is required. It is likely that for low-dose maize events (i.e. not high-dose) a larger refuge would be needed.

Environ. Risk. Chap 07

23/8/04

9:32

Page 235

Resistance Risks and Management Associated with Bt Maize

235

In addition or alternatively, various host plants of the key pests could provide a refuge. All the stemborers are polyphagous on numerous species of graminaceous crops and wild plants. Host lists for each species are discussed in Birch et al. (Chapter 5, this volume). Refuges may be provided in a structured way with specific areas of the non-transgenic crop or alternative crop hosts of the key pests. Such refuges need to be planted in association with the Bt crop in an appropriate spatial arrangement and to provide temporal overlap with the Bt maize crop. Non-Bt maize could be one option. Each Bt maize field should have a refuge within a specified distance, which will depend on dispersal propensity of the target pest. The refuge should provide susceptible individuals at essentially the same time the individuals are emerging from the Bt maize so that there is a high probability of intermingling. In the Kenyan case, there is a desire to utilize the existing farming system and wild hosts to provide sufficient refuge. This will minimize disruption to farmers’ activities and reduce requirement for additional investment of labour or capital. However, adopting unstructured refuges requires extensive quantitative information on the relative productivity of moths from different crops, other hosts and habitats. Such data would be required across all major ecoclimatic zones and farming systems. Data requirements would include the phenology of wild and alternative crop hosts relative to maize, phenology of borer populations in them, the spatial distribution of wild and alternative crop hosts relative to maize, the area coverage and diversity of these plants, and estimates of the suitability of the key species for stemborers. Estimates of adult productivity from wild and alternative crop hosts would need to discriminate different borer species if possible. Counts of emergence holes may underestimate productivity of moths, if successive generations of larvae use common emergence holes. Moreover, presence of emergence holes does not address potential mismatch in stemborer phenology on maize and the wild and alternative crop hosts. Information on suitability for development and levels of mortality could be gained with simple glasshouse experiments, but must be backed up by field studies, given that natural enemies will most certainly play an important role in determining moth productivity of the different hosts. This information is particularly critical as most wild and alternative crop hosts appear to have lower suitability than maize for producing viable adults, e.g. a 20% area of these plant species may be functionally only a 10% refuge of the target population if these hosts are only half as suitable as maize. Preliminary data for Kenya suggest that coastal lowland areas dominated by smallholders may have up to 20% of the landscape in natural refuge areas. It is not yet known what proportion of moths is generated in these natural areas, or if these populations can persist on their own, and this needs to be established prior to designating these areas as refuges. At this time we remain unconvinced that wild hosts can provide sufficient refuge for the two key stemborers, B. fusca and C. partellus, in agreement with B. Le Ru (Nairobi, 2003, personal communication). In highland areas, with larger holdings, estimates are <5% natural refuge. Again, the effectiveness of these potential

Environ. Risk. Chap 07

23/8/04

9:32

Page 236

236

G.P. Fitt et al.

refuge areas is not known. However, this would appear to be insufficient and a structured refuge approach would be required. Additional research needs to continue and to be extended to cover the main production regions at least. One way to introduce a refuge approach may be to integrate with current IPM/push–pull strategies for management. Using repellent plants to push moths from the Bt crop could help to concentrate moths outside the crop for mating in refuge areas. In the push–pull system, the pull component is Napier grass (P. purpureum) on which oviposition is high but larval development is poor. Such a plant would not provide a useful refuge since adults would not be produced. However, Sudan grass (and perhaps other plants) is similarly attractive for oviposition and could produce significant numbers of moths to provide an adequate refuge provided it is not harvested prematurely. In this way, Bt maize could be integrated within the push–pull IPM approach to gain additional advantages. Another strategy for delaying resistance to Bt maize would consist of using both a seed mixture and an external refuge. Such a strategy sometimes performs as well or better than when an external refuge is used alone (Tabashnik, 1994). In fact, a seed-mixture and external refuge will certainly occur in Kenya if Bt maize were used by small-scale producers or in OPVs (Johnston et al., Chapter 6, this volume) with wild and alternate crop hosts acting as an external refuge. Rates of larval movement, larval feeding preference, efficacy of maize plants expressing the Bt toxin and the size/productivity of the external refuge will affect the evolution of resistance. The influence of these factors needs to be studied.

Other methods to reduce selective differential Methods other than refuges that could reduce the fitness advantage of resistant over susceptible phenotypes involve suppressing pests in the transgenic crops by using additional pest control tactics such as pesticides, physical methods, mating disruption, biological control, etc. We recommend strongly that Bt maize be integrated into successful IPM approaches already in place and that all attempts be made to utilize additional control methods (pepper/ash, natural enemies) in Bt maize, especially if larval survival occurs. Techniques to draw parasitoids into Bt maize even when larval densities may be extremely low may help to selectively reduce survival of potentially resistant larvae. Cultural control is one obvious approach to reduce survival of borer populations in Bt maize fields after each crop. Significant numbers of borer larvae remain in diapause in stalks after harvest and in some regions there is evidence that stemborer infestations of subsequent maize crops derive directly from residues of the previous one (Muhammad and Underwood, Chapter 2, this volume). Crop residue management (e.g. destruction through ploughing, burning, early cutting of stalk to leave on the soil surface, etc.) can greatly reduce survival of these larvae (Kfir et al., 1989) and so suppress potentially resistant survivors. In most cases, maize crop residues are left in the field or

Environ. Risk. Chap 07

23/8/04

9:32

Page 237

Resistance Risks and Management Associated with Bt Maize

237

harvested and stored nearby (Muhammad and Underwood, Chapter 2, this volume). Both practices will support the survival of stemborers and may exacerbate resistance development. Timely crop residue destruction could be considered a mandatory component for any management strategy for largeholders and could be strongly encouraged for smallholders. Areas of refuge crop or wild hosts on the other hand should be left undisturbed. In areas with two maize growing seasons per year, Bt maize could be grown only in the first growing season, because this is the season with serious stemborer problems. This would create a temporal refuge and contribute in diminishing the selective advantage of the resistant genotype over the susceptible one, especially if fitness costs associated with the evolution of resistance are present.

Methods to alter heterozygote advantage Pests that are heterozygous for resistance have only one resistance allele, whereas homozygous resistant pests have two. When resistance is rare, most resistance genes will be in heterozygous individuals, so resistance evolution can be greatly accelerated if these heterozygous pests were phenotypically resistant (dominant resistance, large heterozygote advantage), and it would be greatly delayed if these heterozygotes were phenotypically susceptible (recessive resistance, no heterozygote advantage). Heterozygote advantage could be reduced by increasing transgene expression to achieve a ‘high dose’, or altering natural enemy attack patterns. The pyramiding of additional transgenes is recommended strongly for two reasons. The primary advantage of pyramiding is to challenge the pest with two independently acting modes of action, thereby greatly reducing the likelihood that individuals potentially resistant to one protein will survive the presence of both proteins. Secondarily, pyramiding may help achieve a ‘high-dose’ event against all of the stemborer species. Preliminary bioassays (Table 4.8 in Andow et al., Chapter 4, this volume) indicate that none of the presently available transgenic maize lines clearly function at a ‘high dose’ against any of the important stemborers. However, because the maize lines tested were not all of the commercial maize lines, testing of additional commercial Bt maize lines (e.g. from industrial companies interested in registering their Bt maize in Kenya) is suggested, and may result in a ‘high dose’ as is observed for European corn borer in the USA. Candidate proteins for gene pyramiding should be selected based on their inherent activity against the target pests and whether they act independently of the other Cry protein. As the primary candidates for gene pyramiding with Cry proteins against Lepidoptera are other Cry proteins, competitive binding and cross-resistance studies are recommended. For example, Cry1Ac and Cry1B have shown cross-resistance in H. virescens (Gould et al., 1992) but Dipelresistant P. xylostella resistant to Cry1Aa, Cry1Ab and Cry1Ac were not resistant to Cry1B (Tang et al., 1996). The level of cross-resistance between Cry1A and Cry1B needs to be determined before the two toxins can be relied

Environ. Risk. Chap 07

23/8/04

9:32

Page 238

238

G.P. Fitt et al.

on as a pyramided product. All future introductions of Bt cotton in the USA and Australia will be pyramided varieties expressing Cry 1Ac and Cry 2Ab with other combinations likely in the future. Pyramiding of additional transgenes can make heterozygotes phenotypically more like susceptible homozygotes. However, studies should be conducted as to the best approach to maximize multiple gene expression (e.g. compare expression using a gene fusion, which contained two Bt Cry toxin genes in the same construct, vs. independent events and expression of two toxin genes). However, as noted above, because there also is a concern about gene segregation with small farmers, a gene fusion would be preferred. Additionally, increasing transgene expression could have similar results. Increasing transgene expression could be accomplished by numerous plant molecular techniques such as changes in promoters, codon usage, mRNA stability, expression in chloroplasts, etc. (Koziel et al., 1993; McBride et al., 1995). Kota et al. (1999) showed that Cry1Ac- and Cry2Aa-resistant H. virescens could still be controlled when Cry2Aa was over-expressed in plant chloroplasts. Increasing overall toxin expression could also be accomplished by choosing transgenes that show higher levels of expression (e.g. synthetic Cry2Ab; Greenplate et al., 2000).

Methods to alter movement or mating frequencies Methods to alter movement could reduce exposure of the pests to selection as discussed above. Here we consider cases where altering movement increases the frequency of mating between RS and SS or RR and SS individuals. Increasing this mating frequency increases the production of RS offspring relative to RR offspring, and can delay resistance evolution for high-dose events. Push–pull or similar semi-chemical based approaches could be used to manipulate oviposition behaviour and population density of the pests into certain habitats. Although this could maximize the value of refuges, it could also reduce mating activity within the Bt crop if the push effect deters mating. Such localized matings would otherwise result in isolated subpopulations, which would severely compromise the value of refuge approaches.

Possibility for Monitoring and Response Plan Finding 9: Monitoring goals: the goals of monitoring are: (i) to detect resistance so that remedial actions can be taken; (ii) to document control failures; (iii) to determine if the needs of farmers for adequate pest control are being met; and (iv) to monitor compliance to resistance management.

Monitoring methods: realistic, cost-effective monitoring methods need to be identified, developed and field tested. Management responses: the responses available to larger-scale growers will differ from the responses available to smaller-scale growers. Additional research is needed to determine which responses can be implemented in different parts of Kenya.

Environ. Risk. Chap 07

23/8/04

9:32

Page 239

Resistance Risks and Management Associated with Bt Maize

239

Identification of monitoring goals Three factors largely determine when ecological monitoring can be done (Andow and Ives, 2002). A monitoring method must exist that provides useful information at the right time and place. In many cases, the lack of such a method means that significant ecological effects cannot be monitored. A response action or group of actions that may actually alleviate the detrimental effects must exist, and the monitoring method must reveal sufficient information to organize the response (Carrière et al., 2001). Often potential responses come too late, are ineffective or are not well planned. Finally, there must be a social need that the monitoring effort meets. Monitoring can be costly, and if it does not meet a clear social need, sustained funding is unlikely and the effort is not likely to be maintained. The central goals for monitoring are: (i) to detect an increase in resistance in a population early enough that remedial actions can be taken; (ii) to document the failure of resistance management (document control failures); (iii) to determine if the needs of farmers for adequate pest control are being met; and (iv) to evaluate the effectiveness of the implementation of resistance management (compliance monitoring).

Possible monitoring methods Based on the biology of the species and the potential methods for monitoring, are there any monitoring methods to meet the goals identified above? All monitoring methodologies should be standardized with quality control procedures to ensure methodologies are carried out effectively. Achieving this is best accomplished through a centralized monitoring laboratory. Increase in resistance Monitoring to detect an increase in resistance was discussed above. These methods include (Andow and Ives, 2002) egg mass or larval screening using a discriminating dose bioassay (Roush and Miller, 1986; Tabashnik et al., 2000, 2003a), an in-field screen using sentinel plots (Tabashnik et al., 2000; Venette et al., 2000b), an F2 screen (Andow et al., 2000), or backcross to resistant individuals if they are available (Gould et al., 1997). Sentinel plots (Venette et al., 2000b) might be useful primarily for large-scale farmers (Tabashnik et al., 2000). All of these methods estimate the frequency of resistance phenotypes or alleles. When conducted in successive years, they allow the detection of regional changes in resistance frequency (Tabashnik et al., 2003a). Using the cost estimates for monitoring European corn borer in the USA as discussed above, it is possible to estimate the cost of each monitoring method in relation to its value in providing information, which is related to the statistical detection limit of the method (Andow and Ives, 2002). The cost and statistical detection limit are both functions of sample size.

Environ. Risk. Chap 07

23/8/04

9:32

Page 240

240

G.P. Fitt et al.

For dominant alleles, there is an intriguing cost-detection limit crossover among the monitoring methods (Fig. 7.1a). For high detection limits (>0.01), which would be useful to document control failures, the least expensive methods are the larval and egg-mass screens, and the in-field screen is the most expensive. At lower detection limits (<0.01), which would be useful to estimate an increase in resistance, the in-field screen is the least expensive, and the F2 10,000.00 F2

1,000.00

(a)

Larvae Egg mass

100.00

In-field 10.00

Total variable costs (US$)

1.00

0.10

0.01 0.000001

0.00001

0.0001

0.001

0.01

0.1

1,000,000.00 (b) 100,000.00 10,000.00 1,000.00 100.00 10.0 1.0 0.1 0.0001

0.001

0.01

0.1

Detection limit Fig. 7.1. Total direct variable costs of four monitoring methods to achieve the given detection threshold for (a) a dominant allele and (b) a recessive allele (from Andow and Ives (2002)). Costs are estimated for typical situations in the USA for European corn borer on Bt maize. Sampling methods are F2 screen (F2), discriminating dose assay on field-collected larvae (Larvae), discriminating dose assay on neonates from field-collected egg masses (Egg mass), and in-field Bt sweetcorn screen (In-field).

Environ. Risk. Chap 07

23/8/04

9:32

Page 241

Resistance Risks and Management Associated with Bt Maize

241

screen is the most expensive. These results occur because the costs of planting and maintaining the Bt sweetcorn make the in-field screen more expensive than the other methods until >50 individuals need to be sampled to achieve the desired detection limit. The in-field screen is a particularly sensitive screen for dominant alleles. For approximately US$2000 in the USA, the in-field screen can estimate dominant allele frequencies with a theoretical detection limit of better than 0.0001. For recessive alleles, the least expensive method for all desired detection limits is the F2 screen (Fig. 7.1b). This occurs because the phenotype methods require much larger sample sizes than the F2 screen to attain a similar detection limit. For example, a detection limit of 0.05 requires only five isofemale lines for an F2 screen, and 400 larvae or 100 egg masses for the other methods. The F2 screen can estimate recessive allele frequencies to a limit of about 0.001. This could provide up to 7–12 years to respond with alternative resistance management tactics (Andow and Ives, 2002). A detection limit of 0.01 can be provided by the F2 screen for ~US$500 in the USA, which is between two and four times less expensive than the in-field screen, and about 100 times less expensive than the discriminating dose assays (Fig. 7.1b). The most cost-effective range of detection limits is for allele frequencies of 0.001 and up. Resistance failure Monitoring for resistance failures (control failures) should begin with observations by all involved parties (farmers, consultants, extensionists, academics, non-governmental organizations, etc.) of increased damage or presence of insects at unacceptable levels. Follow-up testing in the same season and in the subsequent season is needed to confirm reports of field failure (Carrière et al., 2001). Farmer needs Monitoring for farmer needs: for the resource-poor smallholders, we propose essentially no additional management requirements. Resistance monitoring needs to be carried out, but resource-poor smallholders would not be primarily responsible. Less resource-poor, large-holders will have to participate in resistance management. The protocols must be based on sound science first, and then adapted to be practical. Compliance For refuge compliance (primarily large farms), farmers would need to keep a detailed map as to where their refuge was. Determination of compliance could be accomplished by: ● ● ●

Determining if refuges are of adequate size and location. Assaying the putative refuges for damage. Using antibody tests to determine presence/absence of Bt maize in the fields.

Environ. Risk. Chap 07

23/8/04

242

9:32

Page 242

G.P. Fitt et al.

Possible responses Andow and Ives (2002) discuss potential responses to increases in the frequency of resistance in European corn borer populations. Some of these may be broadly applicable for stemborers in Kenya. The first, and most obvious, are responses which reduce the selection differential between resistant and susceptible individuals either by increasing survival of susceptible individuals or by reducing survival of the resistant ones. Examples include spraying Bt maize with insecticides, otherwise killing stemborers in Bt maize, increasing the refuge size, creating super-refuges for susceptible individuals, releasing susceptible individuals into the field, mating disruption in Bt maize, and releasing of natural enemies in Bt maize. The second class of responses covers those that focus on the movement of adults to modulate the relative frequencies of the resistance allele between Bt and non-Bt fields. This class of response will be effective only for high-dose events. Movement patterns could be manipulated by planting small pest-aggregation sites inside Bt maize fields. These sites could retain moths within fields, thereby increasing the number of moths remaining in their natal fields. Another way to manipulate movement is to use pheromones to lure susceptible males into Bt fields where resistant females may occur. How practical any of these responses might be will need to be determined in conjunction with farmers and advisors. If resistance occurs in areas dominated by large-holders, the following responses are possible: 1. Use of other control strategies that result in absolute or high mortality of putative resistant population (e.g. pesticide overspray, inundative releases of parasitoids, removal of crop from field). 2. Increase of refugia size. 3. Eliminate planting of Bt maize in the affected area until susceptibility has returned. A decision on the area affected by resistance should be based on knowledge of pest dispersal propensity (Carrière et al., 2001). Note: Mitigation plans should be fully discussed with growers through extension channels as soon as possible to illustrate the importance of following recommended resistance management strategies. If resistance occurs among crops of smallholders, it is unlikely that Bt maize could be withdrawn from the market once released, particularly when in the hands of smallholders and having been introgressed into numerous OPVs and landraces. Provided it was possible to detect changes in resistance frequency early enough, there could be two responses: 1. Increase structured refuge area where that is used. 2. Increase alternative management approaches in the Bt maize. This could involve methods to enhance natural enemy activity, resume use of traditional controls, reintroduce highly effective pesticides (where these are affordable). A possible scenario is that resistance would occur in smallholdings and spread to large-holdings. If the Bt cultivar none the less remained efficient against

Environ. Risk. Chap 07

23/8/04

9:32

Page 243

Resistance Risks and Management Associated with Bt Maize

243

stemborers for a reasonable time period (i.e. at least the time that it took to develop it), then a novel Bt cultivar could be engineered and released to control the stemborers. That is to say, if control failures occur because of resistance it will be necessary to rely on some other form of pest control or to tolerate the pest damage. The capacity to respond will be highly dependent on rapid and accurate communication with all landholders. This response, however, would not be considered a component of risk management.

Some Issues During Field Testing, but Before Commercial Release Characterization of transgene expression would be established before any field release (Andow et al., Chapter 4, this volume). However, it will be essential to confirm patterns of expression in field-grown plants over time and in different plant structures under natural environmental conditions and stresses. This would involve growing Bt maize and control plots in a range of production regions and under a range of agronomic treatments relevant to each region (nutrition, water stress, intercropped, etc.). Patterns of efficacy or expression could be examined through artificial infestations, monitoring natural infestations or using laboratory techniques based on ELISA and/or bioassays with susceptible colonies of key pests. The effectiveness of refuges also needs to be established in field conditions with Bt maize. If the refuges are wild hosts, annual or seasonal variation in refuge productivity could affect the evolution of resistance. Moreover, efficient Bt crops are sinks for specialized insect pests while refuges are sources. If a large proportion of a landscape is planted to Bt maize, then even generalist herbivores may become ecological specialists. In such a case, pest population density may decline in areas where Bt crops are common and refuges may become less efficient (Ives and Andow, 2002; Carrière et al., 2003, 2004a,b). Refuge size might have to be increased to compensate for such a regional demographic effect. These kinds of effects can be evaluated only after Bt maize is planted in the field.

Some Post-commercialization Issues Monitoring should be considered a required aspect of resistance management for registration, and the quality of the monitoring efforts must be supported to guard against its degradation. A primary reason for degradation of monitoring over time is decline in funding, and therefore appropriate funding for monitoring should be guaranteed. As monitoring will be primarily for large farmers, a user fee (such as a check-off fee based on yield) could be assessed. Another reason for degradation of monitoring is apathy. This could be partly overcome by periodic meetings and audits. Periodic reporting of monitoring results to all concerned parties (farmers, regulators, etc.) is critical.

Environ. Risk. Chap 07

23/8/04

9:32

Page 244

244

G.P. Fitt et al.

Discussion and Conclusions Our analysis of the resistance risk associated with Bt maize in Kenya suggests that resistance can be managed proactively. Managing resistance in areas dominated by poor farmers with small-scale holdings has been considered a challenge (Cohen et al., 2000), but the circumstances in Kenya seem appropriate to have an optimistic view towards proactive resistance management. Toxin dose is critical information (high or low dose) not yet available for Bt maize in Kenya. Nevertheless, making worst-case assumptions, we can identify the ‘weak link’ among the species at risk for resistance, and tailor the management strategy accordingly. The weak link is the species with a history of pesticide resistance, high potential exposure to Bt maize, low mobility and for which available Bt maize events represent a low dose. For Bt maize in Kenya, B. fusca is the weak link wherever it is the major pest. In all other areas of Kenya, C. partellus is the weak link. From these general conclusions, we assessed the resistance risk to Bt maize and considered how to develop a resistance management strategy for Kenya. ●

Recommendation 3. We propose a resistance management strategy for Bt maize in Kenya that comprises:

1. High-dose maize varieties expressing two independently acting Bt genes (we strongly support the KARI proposal that no introduction of single-gene Bt plants be released). 2. Restriction of use of Bt maize to the ‘long rains’ season (March–August) when pest pressure is highest and need for Bt maize is greatest. In the ‘short rains’ (October–December), alternative crops to maize would be recommended or if it is grown then only non-Bt maize. Ongoing seed supply of non-Bt maize may be required. 3. Refuges – this component would vary for holdings of differing size: (i) Smallholders (growing <1 ha each), with subsistence production representing about 80% of maize area, would have limited capacity to implement structured refuges. Provided rigorous research demonstrates the potential productivity of wild plants and other hosts to produce adult borers in all regions then these could provide adequate refuge. (ii) Larger holders (growing an average of 300 ha of maize) would be required to implement a structured refuge of a suitable and compatible non-Bt crop. Options could be maize, sorghum or any other suitable host plants for the stemborer species. Required areas should be estimated using empirically derived data incorporated into realistic simulation models, but would likely be of the order of a minimum of 20%. Refuges would require similar management inputs (fertilizer, water, etc.) to the Bt crop, be planted at the same time, be of the same or phenologically similar variety and be positioned as embedded strips within Bt fields. Each field of Bt maize should have a refuge located within dispersal range of the pests.

Environ. Risk. Chap 07

23/8/04

9:32

Page 245

Resistance Risks and Management Associated with Bt Maize

245

4. Crop residue destruction – significant numbers of stalk borers may diapause in stems of maize after harvest. Management of this residue in Bt maize fields would kill a high proportion of surviving, potentially resistant larvae and so maximize the value of the high-dose/refuge approach. All growers should be encouraged to manage stalks by cutting, burning or incorporation of stubble. 5. Maintenance of IPM systems associated with Bt maize to provide additional biological suppression of survivors in Bt fields. 6. Monitoring and mitigation programme for resistance, which still needs to be developed. This is an integral part of a proactive resistance management plan and must not be ignored. Underlying this recommendation are several assumptions and significant educational needs. We assume that wild hosts of maize stemborers are sufficiently productive and large enough in area to act as refuges in smallholder maize environments. In addition, we assume that the areas devoted to maize and the areas of wild hosts will not change once Bt maize is available in these smallholder farming systems. Prior to any commercial release, a comprehensive and targeted education programme should be implemented. This should cover all aspects of the technology, but specifically emphasize the threat of resistance, the loss of value it would produce (i.e. food security, commercial profits, loss of valuable Bt genes), the biological basis for a management strategy and the ways it would be implemented. Monitoring must be implemented in the resistance management strategy. However, it will be very important to recognize that monitoring costs will be scale dependent and the small-scale farmers may have difficulty bearing these costs. It will also be very important to get widespread agreement on what kinds of problems will trigger particular kinds of responses or mitigation efforts before any problems actually occur. Considerable effort would be required to package and present an education/communication programme in ways to ensure wide reach to smallholders and their understanding of key issues. All stakeholders need to be involved in this effort. Finally, we suggest that the resistance risk assessment and the proposed resistance management strategy may be relevant to other parts of eastern Africa with similar agroecosystems and social structures. Indeed, many aspects may be applicable to Bt maize in Africa in general.

References Akhurst, R.J., James, W., Bird, L.J. and Beard, C. (2003) Resistance to the Cry1Ac, endotoxin of Bacillus thuringiensis in the cotton bollworm, Helicoverpa armigera (Lepidoptera: Noctuidae). Journal of Economic Entomology 96, 1290–1299. Alstad, D.N. and Andow, D.A. (1995) Managing the evolution of insect resistance to transgenic plants. Science 268, 1894–1896. Andow, D.A. (2001) Resisting resistance to Bt corn. In: Letourneau, D.K. and Burrows, B.E. (eds) Genetically Engineered Organisms: Assessing Environmental and Human Health Effects. CRC Press, Boca Raton, Florida, pp. 99–124.

Environ. Risk. Chap 07

246

10/9/04

14:32

Page 246

G.P. Fitt et al.

Andow, D.A. and Alstad, D.N. (1998) The F2 screen for rare resistance alleles. Journal of Economic Entomology 91, 572–578. Andow, D.A. and Alstad, D.N. (1999) Credibility interval for rare resistance allele frequency. Journal of Economic Entomology 92, 755–758. Andow, D.A. and Hutchison, W.D. (1998) Bt-corn resistance management. In: Mellon, M. and Rissler, J. (eds) Now or Never: Serious New Plans to Save a Natural Pest Control. Union of Concerned Scientists, Cambridge, Massachusetts, pp. 19–66. Andow, D.A. and Ives, A.R. (2002) Monitoring and adaptive resistance management. Ecological Applications 12, 1378–1390. Andow, D.A., Olson, D.M., Hellmich, R.L., Alstad, D.N. and Hutchison, W.D. (2000) Frequency of resistance to Bacillus thuringiensis toxin Cry1Ab in an Iowa population of European corn borer (Lepidoptera: Crambidae). Journal of Economic Entomology 93, 26–30. Berger, A. (1989) Ballooning activity of Chilo partellus larvae in relation to size of mother, egg batches, egg and larvae and age of mother. Entomologia Experimentalis et Applicata 50, 125–132. Bolin, P.C., Hutchison, W.D., Andow, D.A. and Ostlie, K.R. (1998) Monitoring for European corn borer (Lepidoptera: Crambidae) resistance to Bacillus thuringiensis: logistical considerations when sampling larvae. Journal of Agricultural Entomology 15, 231–238. Bolin, P.A., Hutchison, W.D. and Andow, D.A. (1999) Long-term selection for resistance to Bacillus thuringiensis Cry1Ac endotoxin in a Minnesota population of European corn borer (Lepidoptera: Crambidae). Journal of Economic Entomology 92, 1021–1030. Bourguet, D., Bethenod, M.-T., Pasteur, N. and Viard, F. (2000) Gene flow in the European corn borer Ostrinia nubilalis: implications for the sustainability of transgenic insecticidal maize. Proceedings of the Royal Society of London, Biological Sciences 267, 117–122. Cameron, P.J., Walker, G.P., Penny, G.M. and Wisley, P.J. (2002) Movement of potato tuber moth (Lepidoptera: Gelechiidae) within and between crops, and some comparisons with diamondback moth (Lepidoptera: Plutellidae). Environmental Entomology 31, 462–468. Carrière, Y. and Tabashnik, B.E. (2001) Reversing insect adaptation to transgenic insecticidal plants. Proceedings of the Royal Society of London B 268, 1475–1480. Carrière, Y., Dennehy, T.J., Pedersen, B., Haller, S., Ellers-Kirk, C., Antilla, L., Liu, Y.-B., Willot, E. and Tabashnik, B.E. (2001) Large-scale management of insect resistance to transgenic cotton in Arizona: can transgenic insecticidal crops be sustained? Journal of Economic Entomology 94, 315–325. Carrière, Y., Dennehy, T., Ellers-Kirk, C., Holley, D., Liu, Y-B., Simms, M. and Tabashnik, B.E. (2002) Fitness costs, incomplete resistance, and management of resistance to Bt crops. In: Akhurst, R.J., Beard, C.E. and Hughes, P. (eds) Biotechnology of Bacillus thuringiensis and its Environmental Impact. Proceedings of the 4th Pacific Rim Conference, Canberra. CSIRO, Canberra, pp. 82–92. Carrière Y., Ellers-Kirk, C., Sisterson, M., Antilla, L., Withlow, M., Dennehy, T.J. and Tabashnik, B.E. (2003) Long-term regional suppression of pink bollworm by Bt cotton. Proceedings of the National Academy of Sciences USA 100, 1519–1523. Carrière, Y., Sisterson, M. and Tabashnik, B.E. (2004a) Resistance management for sustainable use of Bacillus thuringiensis crops. In: Horowitz, A.R. and Ishaaya, I. (eds) Insect Pest Management: Field and Protected Crops. Springer, New York, pp. 65–95. Carrière Y., Dutilleul, P., Ellers-Kirk, C., Pedersen, B., Haller, S., Antilla, L., Dennehy, T.J. and Tabashnik, B.E. (2004b) Sources, sinks, and zone of influence of refuges for managing insect resistance to Bt crops. Ecological Applications (in press).

Environ. Risk. Chap 07

23/8/04

9:32

Page 247

Resistance Risks and Management Associated with Bt Maize

247

Chiang, H.C. (1972) Dispersion of the European corn borer (Lepidoptera: Pyralidae) in Minnesota and South Dakota, 1945 to 1970. Environmental Entomology 1, 157. Chiang, H.C., Sisson, V. and Ewert, M.A. (1965) Northerly movement of corn borer moth, Ostrinia nubilalis, in southern Minnesota. Proceedings of the Minnesota Academy of Sciences 33, 17. Cohen, M.B., Gould, F. and Bentur, J.S. (2000) Bt rice: practical steps to sustainable use. IRRN 25.2, 4–10. Comins, H.N. (1977) The development of insecticide resistance in the presence of migration. Journal of Theoretical Biology 64, 177–197. Conlong, D.E. (2001) Biological control of indigenous african stemborers: what do we know? Insect Science and its Application 21, 267–274. Davis, P.M. and Onstad, D.W. (2000) Seed mixtures as a resistance management strategy for European corn borers (Lepidoptera: Crambidae) infesting transgenic corn expressing Cry1Ab protein. Journal of Economic Entomology 93, 937–948. Fitt, G.P. (1989) The ecology of Heliothis species in relation to agro-ecosystems. Annual Review of Entomology 34, 17–52. Fitt, G.P. (1997) Risks, deployment and integration of insect resistant crops expressing genes from Bacillus thuringiensis. In: McLean, G.D., Waterhouse, P.M., Evans, G. and Gibbs, M.J. (eds) Commercialisation of Transgenic Crops: Risk, Benefit and Trade Considerations. CRC for Plant Science and Bureau of Resource Sciences, Canberra, pp. 273–284. Fitt, G.P. and Daly, J.C. (1990) Abundance of overwintering pupae and the spring generation of Helicoverpa spp. (Lepidoptera: Noctuidae) in northern New South Wales, Australia: consequences for pest management. Journal of Economic Entomology 83, 1827–1836. Fitt, G.P., Dillon, M.L. and Hamilton, J.G. (1995) Spatial dynamics of Helicoverpa populations in Australia: simulation modelling and empirical studies of adult movement. Computers and Electronics in Agriculture 13, 177–192. Forrester, N.W., Cahill, M., Bird, L.J. and Layland, J.K. (1993) Management of pyrethroid and endosulfan resistance in Helicoverpa armigera (Hubner) in Australia. Bulletin of Entomological Research Supplement Series No 1. Gould, F. (1994) Potential and problems with high-dose strategies for pesticidal engineered crops. Biocontrol Science and Technology 4, 451–461. Gould, F. (1998) Sustainability of transgenic insecticidal cultivars: Integrating pest genetics and ecology. Annual Review of Entomology 43, 701–726. Gould, F., Martinez-Ramirez, A., Anderson, A., Ferre, J., Silva, F.J. and Moar, W.J. (1992) Broad spectrum resistance to Bacillus thuringiensis toxins in Heliothis virescens. Proceedings of the National Academy of Sciences USA 89, 7986–7990. Gould, F., Anderson, A., Jones, A., Sumerford, D., Heckel, D.G., Lopez, J., Micinski, S., Leonard, R. and Laster, M. (1997) Initial frequency of alleles for resistance to Bacillus thuringiensis toxins in field populations of Heliothis virescens. Proceedings of the National Academy of Sciences USA 94, 3519–3523. Gould, F., Blair, N., Reid, M., Rennie, T.L., Lopez, J. and Micinski, S. (2002) Bacillus thuringiensis-toxin resistance management: stable isotope assessment of alternate host use by Helicoverpa zea. Proceedings of the National Academy of Sciences USA 99, 16581–16586. Greenplate, J.T., Penn, S.R., Shappley, Z., Oppenhuizen, M., Mann, J., Reich, B. and Osborn, J. (2000) Bollgard II efficacy: quantification of total lepidopteran activity in a 2-gene product. Proceedings of the Beltwide Cotton Conference 2000. National Cotton Council, Memphis, Tennessee, pp. 1041–1042.

Environ. Risk. Chap 07

248

23/8/04

9:32

Page 248

G.P. Fitt et al.

Gregg, P.C., Fitt, G.P., Zalucki, M.P. and Murray, D.A.H. (1995) Insect migration in an arid continent II. Helicoverpa spp. in Australia. In: Drake, V.A. and Gatehouse, A.G. (eds) Insect Migration: Physical Factors and Physiological Mechanisms. Cambridge University Press, Cambridge, UK, pp. 151–172. Guse, C.A., Onstad, D.W., Buschman, L.L., Porter, P., Higgins, R.A., Sloderbreck, P.E., Cronholw, G.B. and Peairs, F.B. (2002) Modelling the development of resistance by stalk-boring Lepidoptera (Crambidae) in areas with irrigated, transgenic corn. Journal of Economic Entomology 31, 676–685. Han, Q.F. and Caprio, M.A. (2002) Temporal and spatial patterns of allelic frequency in cotton bollworm (Lepidoptera: Noctuidae). Environmental Entomology 31, 462–468. Hassan, R.M., Corbett, J.D. and Njoroge, K. (1998) Combining geo-referenced survey data with agro-climatic attributes to characterize maize production systems in Kenya. In: Hassan, R.M. (ed.) Maize Technology Development and Transfer: GIS Application for Research Planning in Kenya. CAB International, Wallingford, UK, pp. 43–68. Haygood, R., Ives, A.R. and Andow, D.A. (2003) Consequences of recurrent gene flow from crops to wild relatives. Proceedings of the Royal Society of London, Series B 270, 1879–1886. Hunt, T.E., Higley, L.G., Witkowski, J.F., Young, L.J. and Hellmich, R.L. (2001) Dispersal of adult European Corn Borer (Lepidoptera: Crambidae) within and proximal to irrigated and non-irrigated corn. Journal of Economic Entomology 94, 1369–1377. ICIPE (2003) Africa’s first Bt factory, www.icipe.org (accessed December 2003). Ives, A.R. and Andow, D.A. (2002) Evolution of resistance to Bt crops: directional selection in structured environments. Ecology Letters 5, 792–801. Kfir, R., Van Hamburg, H. and Van Vuuren, R. (1989) Effect of stubble treatment on the post-diapause emergence of the grain sorghum stalk borer, Chilo partellus. (Swinhoe) (Lepidoptera: Pyralidae). Crop Protection 8, 289–292. Kfir, R., Overholt, W.A., Khan, Z.R. and Polaszek, A. (2002) Biology and management of economically important lepidopteran cereal stem borers in Africa. Annual Review of Entomology 47, 701–731. Khan, Z.R., Chiliswa, P., Ampong-Nyarko, K., Smart, L.E., Polaszek, A., Wandera, J. and Mulaa, M.A. (1997) Utilization of wild Gramineous plants for management of cereal stemborers in Africa. Insect Science and its Application 17, 143–150. Kota, M., Daniell, H., Varma, S., Garczynski, S.F., Gould, F. and Moar, W.J. (1999) Overexpression of the Bacillus thuringiensis (Bt) Cry2Aa2 protein in chloroplasts confers resistance to plants against susceptible and Bt-resistant insects. Proceedings of the National Academy of Sciences USA 96, 1840–1845. Koziel, M.G., Beland, G.L., Bowman, C., Carozzi, N.B., Crenshaw, R., Crossland, L., Dawson, J., Desai, N., Hill, M., Kadwell, S., Launis, K., Lewis, K., Maddox, D., McPherson, K., Meghji, M.R., Merlin, E., Rhodes, R., Warren, G.W., Wright, M. and Evola, S.V. (1993) Field performance of elite transgenic maize plants expressing an insecticidal protein derived from Bacillus thuringiensis. Bio/Technology 11, 194–200. Legg, D.E. (1983) Dispersion, distribution, and subsequent infestations of the European corn borer, Ostrinia nubilalis (Hübner), among southern Minnesota corn fields, PhD thesis, University of Minnesota, St Paul, Minnesota. Liao, C., Heckel, D.G. and Akhurst, R. (2002) Toxicity of Bacillus thuringiensis insecticidal proteins for Helicoverpa armigera and Helicoverpa punctigera (Lepidoptera: Noctuidae), major pests of cotton. Journal of Invertebrate Pathology 80, 55–63. Liu, Y.B., Tabashnik, B.E., Dennehy, T.J., Patin, A.L. and Bartlett, A.C. (1999) Development time and resistance to Bt crops. Nature 400, 519.

Environ. Risk. Chap 07

23/8/04

9:32

Page 249

Resistance Risks and Management Associated with Bt Maize

249

McBride, K.E., Svab, Z., Schaaf, D.J., Hogan, P.S., Stalker, D.M. and Maliga, P. (1995) Amplification of a chimeric Bacillus gene in chloroplasts leads to an extraordinary level of insecticidal crystal protein in tobacco. Bio/Technology 13, 362–365. McGaughey, W.H. (1985) Insect resistance to the biological insecticide Bacillus thuringiensis. Science 229, 193–195. Mallet, J. and Porter, P. (1992) Preventing insect adaptation to insect-resistant crops: are seed mixtures or refugia the best strategy? Proceedings of the Royal Society of London, Series B 250, 165–169. Martel, C., Réjasse, A., Rousset, F., Bethenod, M.-T. and Bourguet, D. (2003) Hostplant-associated genetic differentiation in northern French populations of the European corn borer. Heredity 90, 141–149. Morin, S., Biggs, R., Sisterson, M., Shriver, L., Ellers-Kirk, C., Higginson, D., Holley, D., Gahan, L., Heckel, D.G., Carrière, Y., Dennehy, T.J., Brown, J.K. and Tabashnik, B.E. (2003) Three cadherin alleles associated with Bt resistance in pink bollworm. Proceedings of the National Academy of Sciences USA 100, 5004–5009. Onstad, D.W., Guse, C.A., Porter, P., Buschman, L.L., Higgins, R.A., Sloderbeck, P.E., Peairs, F.B. and Gronholm, G.B. (2002) Modelling the development of resistance by stalk-boring lepidopteran insects (Crambidae) in areas with transgenic corn and frequent insecticide use. Journal of Economic Entomology 95, 1033–1043. Overholt, W.A. and Maes, K.V.N. (2001) Field Guide to the Maize and Sorghum Stemborer Larvae of Eastern and Southern Africa. ICIPE Science Press, Nairobi. Pierce, C., Weinzierl, R. and Steffey, K. (1998) First year results of a survey for European corn borer resistance to Bacillus thuringiensis. In: Proceedings of the Illinois Agricultural Pesticide Conference, 6–8 January, 1998. Cooperative Extension Service, College of Agricultural, Consumer and Environmental Sciences, University of Illinois at Urbana-Champaign, pp. 67–68. Poland, D., Mugo, S. and Hoisington, D. (2003) IRMA II Business Plan Developed. IRMA Updates, June 2003, Vol. 4, Issues 1+2, www.cimmyt.org/ABC/InvestInInsectResist/htm/InvestIn-InsectResist.htm (accessed December 2003). Ross, S.E. and Ostlie, K.R. (1990) Dispersal and survival of early instars of European corn borer (Lepidoptera: Pyralidae) in field corn. Journal of Economic Entomology 83, 831–836. Roush, R.T. (1994) Managing pests and their resistance to Bacillus thuringiensis: can transgenic crops be better than sprays? Biocontrol Science and Technology 4, 501–516. Roush, R.T. and Miller, G.L. (1986) Considerations for design of insecticide resistance monitoring programs. Journal of Economic Entomology 79, 293–298. Schneider, J.C. (1999) Confidence interval for Bayesian estimates of resistance allele frequencies. Journal of Economic Entomology 92, 755. Showers, W.B. (1993) Diversity and variation of European corn borer populations. In: Kim, K.C. and McPheron, B.A. (eds) Evolution of Insect Pests. John Wiley & Sons, New York, pp. 287–309. Showers, W.B., Hellmich, R.L., Derrick-Robinson, M.E. and Hendrix, W.H. (2001) Aggregation and dispersal behaviour of marked and released European corn borer (Lepidoptera: Crambidae) adults. Environmental Entomology 30, 700–710. Songa, J.M., Mugo, S.N., Mulaa, M., Taracha, C., Bergvinson, D., Hoisington, D. and De Groote, H. (2002) Towards development of environmentally safe, insect resistant maize varieties for food security in Kenya. Paper presented to the Syngenta Symposium, Washington, DC, 25 June, 2002, the Syngenta staff in Greensboro, North Carolina, 26 June, 2002, and to the Syngenta staff in Basel, Switzerland, 27 June, 2002.

Environ. Risk. Chap 07

250

24/8/04

12:46

Page 250

G.P. Fitt et al.

Storer, N.P., Peck, S.L., Gould, F., Van Duyn, J.W. and Kennedy, G.G. (2003) Spatial processes in the evolution of resistance in Helicoverpa zea (Lepidoptera: Noctuidae) to Bt transgenic corn and cotton in a mixed agro-ecosystem: a biologyrich stochastic simulation model. Journal of Economic Entomology 96, 156–172. Tabashnik, B.E. (1994) Delaying insect adaptation to transgenic plants: seed mixtures and refugia reconsidered. Proceedings of the Royal Society of London Series B 255, 7–12. Tabashnik, B.E. and Croft, B.A. (1982) Managing pesticide resistance in crop-arthropod complexes: interactions of biological and operational factors. Environmental Entomology 11, 1137–1144. Tabashnik, B.E., Patin, A.L., Dennehy, T.J., Liu, Y.-B., Carrière, Y. and Antilla, L. (2000) Tracking the frequency of resistance to Bacillus thuringiensis in field populations of pink bollworm. Proceedings of the National Academy of Sciences USA 21, 12980–12984. Tabashnik, B.E., Liu, Y.-B., Dennehy, T.J., Sims, M.A., Sisterson, M., Biggs, R. and Carrière, Y. (2002) Inheritance of resistance to Bt toxin Cry1Ac in a field-derived strain of pink bollworm (Lepidoptera: Gelechiidae). Journal of Economic Entomology 95, 1018–1026. Tabashnik, B.E., Carrière, Y., Dennehy, T.J., Morin, S., Sisterson, M., Roush, R.T., Shelton, A.M. and Zhao, J.-Z. (2003a) Insect resistance to transgenic Bt crops: lessons from the laboratory and field. Journal of Economic Entomology 96, 1031–1038. Tabashnik, B.E., Gould, F. and Carrière, Y. (2003b) Delaying evolution of insect resistance to transgenic crops by decreasing dominance and heritability. Journal of Evolutionary Biology 17(4), 904–912. Tang, J.D., Shelton, A.M., Van Rie, J., de Roeck, S., Moar, W.J., Roush, R.T. and Peferoen, M. (1996) Toxicity of Bacillus thuringiensis spore and crystal protein to resistant diamondback moth (Plutella xylostella). Applied Environmental Microbiology 62, 564–569. Taylor, C.E. and Georghiou, G.P. (1979) Suppression of insecticide resistance by alteration of gene dominance and migration. Journal of Economic Entomology 72, 105–109. Thomas, Y., Bethenod, M.-T., Pelozuelo, L., Frerot, B. and Bourguet, D. (2003) Genetic isolation between two sympatric host-plant races of the European corn borer, Ostrinia nubilalis Hübner. I. Sex pheromone, moth emergence timing, and parasitism. Evolution 57, 261–273. Venette, R.C., Luhman, J.C. and Hutchison, W.D. (2000a) Survivorship of fieldcollected European corn borer (Lepidoptera: Crambidae) larvae and its impact on estimates of resistance to Bacillus thuringiensis Berliner. Journal of Entomological Science 35, 208–212. Venette, R.C., Hutchison, W.D. and Andow, D.A. (2000b) An in-field screen for early detection and monitoring of insect resistance to Bacillus thuringiensis in transgenic crops. Journal of Economic Entomology 93, 1055–1064. Wang, R.J., Yan, F.M., Li, S.G. and Li, S. (1995) Allozyme differentiation among nine populations of the corn borer (Ostrinia) in China. Biochemistry and Genetics 33, 413–420. Zhou, G., Overholt, W.A. and Mochiah, M.B. (2001) Changes in the distribution of Lepidopteran maize stemborers in Kenya from the 1950s to 1990s. Insect Science and its Application 21, 395–402. Zhou, G., Overholt, W.A. and Kimani-Njogu, S.W. (2003) Species richness and parasitism in an assemblage of parasitoids attacking maize stem borers in coastal Kenya. Ecological Entomology 28, 109–118.

Environ. Risk. Chap 08

8

23/8/04

9:33

Page 251

Risk Assessment of Bt Maize in Kenya: Synthesis and Recommendations A. HILBECK, D.A. ANDOW, A.N.E. BIRCH, G.P. FITT, J. JOHNSTON, K.C. NELSON, E. OSIR, J. SONGA, E. UNDERWOOD AND R. WHEATLEY Corresponding author: A. Hilbeck, Geobotanical Institute, ETH Zurich, Zurichbergstrasse 38, CH-8044 Zurich, Switzerland. E-mail: [email protected]

In this book, we provide the scientific basis for conducting an environmental risk assessment of Bt maize in East Africa, focusing on Kenya. This assessment comes out of work conducted at an international workshop on scientific environmental risk assessment of Bt maize in Kenya on 25–29 November 2002 at the International Centre for Insect Physiology and Ecology (ICIPE) near Nairobi. A total of 52 scientists participated in the workshop, of which 24 were African scientists. In this chapter, we summarize some of the main findings from the workshop (detailed in Chapters 3–7, this volume), identify several key lessons learned from conducting the case study, discuss some of the next steps that could be taken in Kenya to complete an environmental risk analysis of Bt maize and identify scientific issues that need to be developed further to facilitate scientific risk assessment.

Kenyan Maize Breeding and Agriculture Both hybrids and open-pollinated varieties (OPVs) of maize are produced and used in Kenya, and the preferred varieties for human consumption are the white maize varieties. Maize hybrids are produced by crossing at least two inbred parents to create offspring with desired traits; progeny from these offspring, however, often will not maintain those desired traits for more than one generation. A Bt transgene is incorporated into a maize hybrid by backcrossing the transgene into one of the inbred parents, usually the seed parent. Hence a Bt maize hybrid is usually hemizygous for the transgene (having one copy of the transgene). Homozygous hybrids can be produced, but they take longer because both parents must be converted with the transgene. Maize OPVs are produced in several ways. One common way is to © CAB International 2004. Environmental Risk Assessment of Genetically Modified Organisms: Vol. 1. A Case Study of Bt Maize in Kenya (eds A. Hilbeck and D.A. Andow)

251

Environ. Risk. Chap 08

252

23/8/04

9:33

Page 252

A. Hilbeck et al.

start with a diverse, desirable pool of genes and by mass selection produce an improved population that maintains traits of interest (such as yield and seed colour) for multiple generations. A Bt transgene is incorporated into an OPV either in the initial gene pool or by recurrent backcrossing. In either case, the Bt transgene increases in frequency in the improved population. Ideally, all Bt OPV plants will have two copies of the Bt transgene. This can be accomplished by recurrent backcrossing, screening and selection. However, an OPV with considerable genetic variation will most likely contain a mixture of hemizygous and homozygous (having two copies of the transgene) plants. At the time of writing, one private-sector Bt maize transformation event and perhaps nine public-sector events are being considered for use in Kenya. Mon-810, a private-sector event, is a part of a USAID-funded research project (E. Osir, Nairobi, 2002, personal communication). The nine public-sector events were developed by CIMMYT (International Centre for the Improvement of Maize and Wheat), and are referred to as ‘second-generation’ events. They are in early stages of development in the Insect Resistant Maize for Africa (IRMA) project, a joint venture between CIMMYT and the Kenya Agriculture Research Institute (KARI), funded by the Syngenta Foundation. IRMA plans to have Bt maize available in 2007. It is likely that adapted varieties from the private-sector source could be available soon. Kenya is one of the leading producers and consumers of maize in the eastern and southern Africa region, producing about 2.38 million t/year on 1.5 million ha (an average yield of 1.57 t/ha). Most maize is grown intercropped on smallholdings of <2 ha in size. When possible, many farmers attempt two crops per year. The ‘long-rains’ season (the first season) is the most productive in the western areas and the coast, and the ‘short rains’ season in the eastern mid-altitude areas. In the areas above 1500 m, there is only one season of rains and a single crop is grown in the year. Problems common to all Kenyan farmers are inadequate, erratic or unreliable rainfall, and low soil fertility except in some high-yielding areas. The diversity of the Kenyan environment means, however, that constraints to maize production vary greatly between areas, summarized below according to agroecological zone. Bt maize is expected to control pest damage from stemborer pests. In Kenya, the main species are Busseola fusca, primarily at higher altitudes and Chilo partellus, primarily at lower altitudes, though distributions and densities are dynamic. The Lowland Tropics zone is characterized by smallholder farmers growing maize for subsistence in diverse intercrops with very low yields. Most of the farmers use recycled or local seed. Field pests, including stemborers, other insects, birds and mammals are considered among the most important yield loss problems. C. partellus is the most important stemborer. In the Dry Mid-altitude and Dry Transitional zones, the majority of farmers double-crop and intercrop maize on smallholdings on sloping and erosionprone soil. Yields are low and most farmers recycle seed. Pests and diseases, including stemborers, are considered important constraints on maize production, but rank below rainfall reliability, low soil fertility, poor extension services and cost of inputs such as seed and fertilizer. Rainfall duration and timing is a crucial limiting factor for maize production, and crop failures are

Environ. Risk. Chap 08

23/8/04

9:33

Page 253

Risk Assessment of Bt Maize in Kenya

253

regular. Below 1000 m, C. partellus is the major stemborer pest in both seasons. Above 1000 m, both C. partellus and B. fusca are important pests. The Moist Transitional and Highland Tropics zones have the highest maize yields and produce 80% of the total national maize production. Most maize is grown in a single season on smallholdings and most farmers purchase hybrid seed annually. Most small-scale farmers intercrop, but unlike anywhere else in Kenya, the larger-scale farmers (>8 ha) mostly grow maize in monoculture. The highest-yielding maize regions are in these zones, with large-scale farmers producing a quarter of Kenya’s maize production. Pest problems rank below other constraints on production, but stemborers are among the most important pests. B. fusca is the major stemborer pest in all areas. If a Bt hybrid is made available first, it is most likely to be taken up by farmers in these zones. In the Moist Mid-altitude zone, the majority of farmers double-crop and intercrop on smallholdings. Yields are low and most farmers recycle seed, many stating a preference for local varieties. Pests are ranked lower as a constraint on agricultural production than in other regions, although stemborers are still considered among the most important pests. In some areas, C. partellus dominates, while in other areas, B. fusca is the major stemborer, and their relative densities vary from year to year. An assessment of the impact of any new agricultural technology such as Bt maize must consider whether the proposed technology will have a negative impact by exacerbating the effects of any of the constraints to crop production not targeted by the technology. The potential impacts of Bt maize in Kenya will be influenced by the variety released (whether it is a hybrid or OPV) and the area it is targeted for. The following section details a method for assessing technology options and their potential impacts that takes into account these factors.

Problem Formulation and Options Assessment (PFOA) The PFOA sets the scope for any environmental risk assessment of a transgenic organism and provides the options to which the organism-based technology can be compared. A PFOA relies on public deliberation to accomplish these ends. The structured public deliberation about transgenic organisms provides a rational, science-driven planning process by which countries can assess their needs, evaluate the risks related to multiple future options, design policies to reduce societal risks and enhance the benefits provided by various options. In general, the problem is defined as an unmet basic need of society that requires change. Basic human needs include food, shelter and safety. For example, individuals have the basic need for a certain amount of calories per day or the security that their children will continue to live healthy lives as a minimum foundation for well being. Other human interests are stakeholderspecific and include enhanced economic opportunity, positive social interactions, cultural richness, etc. By focusing on basic human needs, the key roles of the transgenic organism can be clarified, constructive discussion is promoted and the social context in which environmental risk will be assessed is specified.

Environ. Risk. Chap 08

23/8/04

254

9:33

Page 254

A. Hilbeck et al.

The PFOA is a process by which gaps are identified and possible solutions are compared. It assumes the actual people affected are at the centre of the assessment; therefore, adequate and fair representation should be ensured during such assessments. The initial data gathering for a PFOA assessment can take the form of a participatory rural appraisal, participatory learning and appraisal, focus group discussions, or questionnaire interviews. Once the data are available, representatives of the various stakeholders should participate in a PFOA review of the transgenic organism.

Proper controls The assessment of options is a standard part of risk assessment methodologies (NRC, 1983). For example, a proposal to use a transgenic crop is a proposal to change how crops are grown; it focuses on the possibilities for the future. Societal evaluation and decision making is based on the belief that the future will be an improvement over the present. So, from a risk assessment perspective, it is common practice to compare the transgenic crop to present agricultural practices. Consequently, an important scientific control for risk assessment is present practice. However, only using present conditions as a control for risk is clearly inadequate (NRC, 2002). Doing so is tantamount to comparing one possible future to the present, as if the present were the only possible alternative future. In risk assessment, it is important to consider the various future options, because the risks associated with each future must be compared with those associated with an alternative future (NRC, 1983). The PFOA is based on some of these alternative futures, to which the transgenic crop can be compared. As the assessment process proceeds, the risk assessment should also use one or more of these alternative futures as a scientific control.

Kenya results In Kenya, the use of the ‘needs’ concept in the PFOA proved to be particularly useful for encouraging constructive dialogue and consensus. The PFOA provides a technology evaluation process in which all stakeholders can contribute to the public discussion about the role of transgenic organisms in their nation. It focuses discussion on broad societal concerns rather than narrow individual interests, creates the potential for planned development at the national level and encourages the exploration of potential agreements among many stakeholders. Although the group was not a complete multistakeholder group, it was composed of persons with diverse national and professional backgrounds. Factors that could have led to dissent and controversy were noted, e.g. moral issues, laws and litigation on transgenic crops, farmers’ rights on genetic resources, seed ownership issues, etc., but the structure of the PFOA continually refocused participants to the broader task at hand. Furthermore, because time was limited, only key points could be

Environ. Risk. Chap 08

23/8/04

9:33

Page 255

Risk Assessment of Bt Maize in Kenya

255

considered. For example, in the selection of various alternative technologies/options for addressing the stemborer problem, the participants selected only two contrasting scenarios for further discussion: Bt maize and the push–pull technology (Box 3.1, Nelson et al., Chapter 3, this volume). The PFOA was not able to complete a full assessment of the problem or the possible options, because the broad range of stakeholders were not present in the discussion. The group did determine that that PFOA process should be continued in Kenya, and despite the incomplete representation, did illuminate several key points. Although a full PFOA was not completed, the group proposed that the stemborers were a problem for maize in only some parts of Kenya during some growing seasons. Farm surveys (Muhammad and Underwood, Chapter 2, this volume) showed that stemborers are not always the most serious pest problem facing maize farmers in East Africa. In evaluating the two future technologies, the group suggested that for Bt maize, high-quality, low-cost seed maize must be readily available (this is difficult in many parts of Kenya), and extension programmes that demonstrate how it integrates with other farming and pest control practices must be developed. From these observations, Bt maize may be most readily available and appropriate for larger-scale farmers who use maize hybrids in the Moist Transitional and Highland Tropics zones. In contrast to Bt maize, which addresses only stemborer control, the push–pull system suppresses both stemborers and Striga, a severe parasitic weed, using habitat management and vegetation diversity. This system, however, requires livestock so that the grasses that pull the stemborers from maize can be used productively. Hence, the push–pull system may be most appropriate for small-scale farmers with livestock. These two possible future technologies seem to be complementary and best suited for different kinds of farmers. On the other hand, it may be possible to incorporate Bt maize with the ‘pull’ element of the push–pull system in a novel approach combining several strategies (Fitt et al., Chapter 7, this volume).

Transgene Expression and Locus Structure Characterizing transgene expression and locus structure provides the basic information necessary to conduct a risk assessment. Several key risk issues are addressed: How can transgenic plants be designed to reduce risk or the need for risk assessment? What is the transgene product (i.e. the compound synthesized by the transgene)? Are there any other products from the transgene? Is the transgene inherited predictably? This assessment addresses these questions through transgene design, locus structure (genotype), expression (phenotype) and transmission, emphasizing reliable, conclusive methods.

Need for DNA sequencing For risk assessment, it is essential that the transgene locus is sequenced in the plant. One of the most serious and vexing concerns about transgenic plants is that they might produce unexpected or inadvertent (unintended) gene

Environ. Risk. Chap 08

23/8/04

9:33

Page 256

256

A. Hilbeck et al.

products, which in turn could have serious environmental or human health effects (Andow et al., Chapter 4, this volume). This concern stems from three sources: uncertainty about gene expression from the transgene locus, interactions between the transgene products and normal gene products in the plant (pleiotropy or epistasis) and environmental interactions with gene expression. DNA sequencing of the transgene locus in the plant provides definitive scientific evidence for reducing the uncertainties associated with gene expression from the transgene locus. The transgene could produce unknown or unexpected gene products via several known mechanisms, including ectopic expression, spurious open reading frames, homologous recombination, disturbance of native plant genes, etc., all of which can be assessed from the DNA sequence of the transgene in the plant. We recognize that this is not now a requirement in several national genetically modified organism (GMO) risk assessment processes, but it is now technically feasible and standard practice, and could be readily incorporated into the assessment process.

Molecular and whole-plant methods Transgene expression can be predicted from the DNA sequence of the transgene locus, but these predictions will sometimes prove inadequate. Transgene expression is determined by both the transgene (genotype) and the environment in which the transgene is expressed. For example, the level of pest protection provided by Bt maize depends on the environment. In environments dominated by the target pest species, the level of protection can be excellent, but in other environments in which the target pest occurs with other dominant pest species, protection may be minimal. This means that molecular methods are necessary, but not sufficient, to characterize the transgene phenotype. It is also necessary to characterize expression in whole plants in the relevant environments in which they will be grown, because in addition to variation in expression in different plant tissues and during different life stages, expression can also vary due to different environmental conditions (such as soil types).

Kenya results We have used only publicly available data in characterizing Bt maize. We found that none of the Bt transgenes has been characterized sufficiently to allow a full science-based risk assessment. This is both understandable and acceptable for the public-sector transgenic events, which are in the early stages of development. The two private-sector events, however, have been commercialized elsewhere, but additional scientific data are still needed to conduct a risk assessment in Kenya. Trangenes can be designed to reduce actual environmental risks or to reduce the need for environmental risk assessment. For Bt maize in Kenya, some recommendations were to eliminate marker genes, to select transgenes with simple structures, and to use two Bt genes with independent modes of action and

Environ. Risk. Chap 08

23/8/04

9:33

Page 257

Risk Assessment of Bt Maize in Kenya

257

high control efficacy against the main target pests. We recognize that many of the present Bt maize events cannot be redesigned to fit these recommendations, and these do not necessarily have high environmental risks as a consequence. All of the Bt maize events need to be characterized further. The DNA sequence needs to be provided and better estimates of control efficacy need to be made. None of the presently available Bt maize transformation events exhibit excellent control against B. fusca, the stemborer pest responsible for most of the projected yield losses to stemborers in Kenya. Some of them show good efficacy against C. partellus, the second key stemborer pest, but these data need to be expanded to provide more detailed and rigorous data on efficacy (Fitt et al., Chapter 7, this volume). In addition, the phenotype of all of the events needs additional characterization, with Bt toxin concentration reported for more maize tissues and developmental stages than presently reported. Finally, when Bt maize varieties are developed, the proportion of each that is homozygous for the transgene, hemizygous for the transgene, or lacking the transgene should be determined.

Non-target and Biodiversity Effects Consistent with the Convention on Biological Diversity (CBD, 1992), biodiversity is recognized as a significant social value and therefore subject to protection. Use of transgenic, insecticidal Bt crops could directly or indirectly affect non-target organisms that currently exist in agroecosystems and provide significant services to ecosystem functioning. The focus of the present assessment is on potential non-target effects in maize and the nearby adjacent habitats. We considered maize herbivores, parasitoid natural enemies, pollen feeders, grass and sedge weeds, parasitic weeds, and soil ecosystem functions. We did not attempt to broaden the case study to cover aquatic environments, or rare or fragile native habitats in the landscape. We did not address systematically vertebrates, species of conservation concern, species of cultural significance, or plant pathogens because the expertise was not present at the workshop. For instance, maize streak virus and Aspergillus flavus (which synthesizes aflatoxin, a potent human carcinogen) are two plant diseases that would need to be addressed in the future. Below-ground microorganisms and some associated ecosystem processes were addressed, but not soil macroorganisms. Such scientific areas were identified as important gaps in available knowledge. Because Kenya has not allowed the planting of Bt maize anywhere in the country as of the beginning of 2003, we focused on laboratory tests of non-target effects that can be conducted prior to field release of the transgenic crop. However, we also considered some tests that would need to be conducted in the field after the first field release.

Species selection process The range of non-target biodiversity and ecosystem functions associated with maize in Kenya was divided into various ecological function groups. Within

Environ. Risk. Chap 08

23/8/04

9:33

258

Page 258

A. Hilbeck et al.

each group, we developed selection matrices to identify species and ecosystem functions for assessing the non-target effects of Bt maize in Kenya. The selection matrices were based on qualitative and quantitative information about the cropping system and its associated species and ecosystem functions, and can be used prior to the development of Bt maize. These matrices were rapidly completed and produced a short prioritized list of candidate test species or ecological functions from a much longer list of the biodiversity associated with maize in Kenya, in a transparent and scientifically justifiable manner. For each of these potential candidate species or functions, we conducted an assessment of the potential exposure of each to Bt maize, which further prioritized the species or functional group. About 20 species and ecosystem functions were identified as potential candidates for testing the non-target effects of Bt maize in Kenya. Some of the species were higher priority in the lowland regions of Kenya, while others were higher priority in the highland regions. Non-target effects are difficult to predict, so they need to be measured under various conditions (regions, seasons, a range of non-target species). For selected species, we constructed hazard scenarios describing the hypothesized way in which the species could be harmed by Bt maize. These hazard scenarios allowed us to develop specific experimental protocols to assess the potential environmental risk of Bt maize.

Test methods and protocols We developed several scientifically rigorous laboratory-based methods and protocols to test the potential hazard scenarios on the candidate non-target arthropod species/functions. In addition, we developed several protocols for assessing the hypotheses in small-plot field experiments. Some of the hypotheses could be most efficiently assessed in small-scale field experiments, especially those developed to assess effects on weeds and soil ecosystem functions. There is a need to take into account multiple interactions between a nontarget species and other species in the same or different trophic level, such as between plant–host insect–parasitoid or plant–prey insect–predator. For example, we developed a hypothetical hazard scenario by which impacts on Carpophilus spp. might affect aflatoxin infestation. Aflatoxins are created by A. flavus, a fungal species that infects maize stems and ears. Carpophilus spp. transmit the fungus from the stem to the ear, within the ear and from ear to ear. Carpophilus spp. might be exposed to Bt toxin because they feed on the frass of Helicoverpa armigera, and Bt toxin might pass through H. armigera larvae into the frass. If such exposure or any indirect effect leads to an increase in Carpophilus spp. populations, then an increase in aflatoxin contamination could occur. Conversely, if Carpophilus spp. decrease or are in any other way less able to transmit A. flavus on Bt maize, this would lead to a decrease in aflatoxin contamination and add a significant benefit. Clearly, this scenario is hypothetical, but because aflatoxin contamination is a serious human health risk, it is essential to evaluate any testable hazard scenario that could lead to

Environ. Risk. Chap 08

23/8/04

9:33

Page 259

Risk Assessment of Bt Maize in Kenya

259

increased aflatoxin in Bt maize. In this case, several possible experiments could be used to evaluate rapidly the existence of the exposure pathway. The same experiments can be used to determine if there is a decrease in the capacity of the target stemborer species to aid Aspergillus transmission on Bt maize. Striga is a serious weed of maize in the Moist Mid-altitude zone. It is parasitic, germinating in the presence of maize roots, which it infects and then grows underground, withdrawing nutrients and water from the maize. After killing or debilitating maize, it sprouts a flowering shoot above ground, sets seed and showers the soil with millions of tiny seeds. We have proposed that the effect of Bt maize on this pest needs to be evaluated. If Bt maize exacerbates Striga problems, this would be a serious limitation. However, if it reduces Striga attack, this would be a beneficial side-effect of Bt maize.

Gene Flow and its Consequences Transgene movement from the transgenic crop plant to other organisms is one of the key considerations in risk assessment. The risk problem, however, is framed differently in different parts of the world. In the US regulatory system, gene flow per se is not considered an environmental risk. This perspective is that the genes themselves do not cause environmental harm; it is only the consequences of gene flow that are considered a risk. In the European Union regulatory system, gene flow per se is considered an environmental risk. This perspective is that the movement of transgenes into another organism is an environmental problem because this will always cause environmental effects and they are sometimes unknown or unanticipated. By focusing on the scientific dimensions to the problem, we consider both perspectives here, because they are not mutually exclusive. The assessment of gene flow associated with Bt maize in Kenya followed a systematic process of: (i) identifying the potential sources of Bt transgenes; (ii) identifying potential recipients of these transgenes; (iii) identifying the potential routes by which the transgenes could move from the sources to the recipients; (iv) assessing the likelihood of transgene movement along the various routes; (v) assessing the likelihood that a Bt transgene will increase in frequency following gene flow to the recipient non-transgenic population; and (vi) identifying the potential ecological and agronomic consequences from the spread and persistence of Bt transgenes in Kenyan agriculture.

Sources and recipients The sources of the Bt transgene for gene flow are maize breeding material in Kenya, imported whole maize grain from regions commercially producing Bt maize, and improved, adapted maize varieties, including both hybrids and OPVs. If gene flow occurs frequently, recipient species and populations can become significant secondary sources. We considered several wild species and their potential for experiencing gene flow from Bt maize, and concluded that

Environ. Risk. Chap 08

23/8/04

9:33

Page 260

260

A. Hilbeck et al.

none of the species were related to maize closely enough to be potential recipients. The most likely plants to experience gene flow from Bt maize in Kenya and most of Africa is likely to be maize itself. There are many types of maize landraces and local cultivars used in small-scale agriculture in Kenya and the rest of Africa that could be recipient populations of Bt genes via pollen- or seed-mediated gene flow.

Routes and likelihood of movement There are no known barriers to gene flow between Bt maize hybrid or OPVs and Kenyan maize landraces or local cultivars. In Kenya, maize pollen (with or without Bt) can easily move the distances separating large commercial farms and small farms, or small farms from each other. Intentional movement of Bt maize seed by small-scale farmers is likely to accelerate the rate of gene flow of the Bt gene into Kenyan maize landraces. The rate of gene flow will probably depend directly on the amount of Bt maize planted or imported into the region.

Increase and spread The increase and spread of a transgene depends on the source size and the relative fitness of the transgene in the recipient population. If Bt maize is planted commonly and recurrently, the Bt transgene is expected to spread rapidly even if the transgene is mildly deleterious to the recipient maize population. However, in areas with moderate to high stemborer pressure, Bt maize plants are likely to have higher fitness than cultivars without the Bt transgene. If the transgene increases average fitness of the recipient population, then it is likely to spread rapidly even when the source population is small. Data are needed to demonstrate whether there is a positive fitness effect of the Bt transgene when it moves into Kenyan maize landraces via gene flow. Perhaps more importantly, data are needed to understand the role that environmental conditions (climate, insect pressure) play in determining the fitness effect of the Bt gene in Kenyan landraces. However, data from other maize crop systems leads us to expect that if Bt maize is introduced into Kenya, gene flow via pollen and seed will quickly spread the Bt gene through most of the Kenyan maize landraces.

Potential consequences Gene flow can affect genetic diversity, non-target species and resistance risk. Introgression of the Bt gene into Kenyan maize landraces will alter patterns of genetic diversity in the recipient populations. The genes that are most closely linked to the Bt transgene are the most likely to suffer a reduction in diversity following gene flow. Although it is likely that these landraces contain unique genetic diversity, data are needed to demonstrate this. Once it is understood

Environ. Risk. Chap 08

23/8/04

9:33

Page 261

Risk Assessment of Bt Maize in Kenya

261

how much unique diversity exists near the region of transgene insertion it may be possible to determine how much, if any, of the significant genetic diversity will be reduced by Bt transgene flow. Transgene flow will also broaden the spatial extent to which non-target effects could occur. This could increase the number of non-target species that are potentially exposed to the transgene as well as increase the magnitude of exposure for others. This also needs to be assessed. Transgene flow will also alter the resistance risk in pest populations. Movement of the Bt gene into landraces will likely create genotypic mixtures within maize fields in which plants have one, two or zero copies of the transgene. Mixtures of genotypes could result in more rapid evolution of insect resistance, making resistance management more difficult. This potential effect also needs to be assessed.

Resistance Risk and Management By controlling stemborers, Bt maize also exerts positive selection for borers that are resistant to Bt toxin. Resistant borers are capable of eating Bt maize and are no longer controlled by it. Resistance could result in reduced maize yields, increased insecticide use and associated environmental and human health effects. The group assessed the resistance risk associated with Bt maize in Kenya and as far as possible proposed management practices to mitigate this risk. Although the results of this assessment are specific to the case of Bt maize in Kenya, they are likely to be relevant to other parts of eastern Africa with similar agroecosystems and social structures.

Resistance can be managed proactively Kenyan maize is dominated by small-scale production. In these systems, the structured refuges used in the large-scale production systems in developed countries are considered impractical. Despite these concerns, the group concluded that proactive resistance management should be possible. Moreover, resistance management could be improved by considering how to implement it consistent with successful pest management practices already being used. Different resistance management strategies may be needed for the large-scale producers in the higher altitude western Kenyan regions vs. the smaller-scale producers throughout the rest of the country, although more information is needed before this can be determined. In addition, the effect of seed saving associated with transgenic OPVs on resistance risk should be evaluated, because this is likely to occur commonly in Kenya.

High or low dose Bt toxin dose is critical information. A high-dose transgenic plant is one on which pests require two resistance genes to survive. If they have only one

Environ. Risk. Chap 08

23/8/04

262

9:33

Page 262

A. Hilbeck et al.

resistance gene, the transgenic plant is too toxic and they die. Resistance can be delayed for long periods if the transgenic plant expresses a ‘high dose’ of the Bt toxin. If it does not, additional efforts must be taken to delay resistance. None of the presently available Bt maize varieties appears to be a high-dose event against the key maize stemborers in Kenya. It may be important to develop or improve existing Bt maize transgenes by combining multiple Bt transgenes to attain a ‘high-dose’ transgenic event. If this is done, the Bt toxins should act independently to minimize the risk of crossresistance.

Weak link Resistance management should be designed for the weakest link. The weakest link is the pest experiencing a ‘low dose’ of Bt toxin, a history of resistance and/or low mobility. For Bt maize in Kenya, there was insufficient information available to determine which species was the weak link.

At risk species The two stemborers most likely at greatest resistance risk in Kenya are C. partellus and B. fusca. B. fusca poses the greatest resistance risk in the single-cropping regions of western Kenya and the high-elevation areas of the Central Highlands of Kenya. These include the areas in Kenya with large-scale production and frequent use of hybrids. C. partellus is a resistance risk in the Lowland Tropics and the double-cropped and Mid-altitude regions of Kenya. These are the areas of Kenya with small-scale production and frequent use of OPVs. In addition, the non-target herbivores H. armigera, Spodoptera exempta, Ephestia cautella and Ephestia kuehniella and Plodia interpunctella are differentially affected by the Bt toxin, ranging from sublethal to lethal, and may therefore also be at risk for resistance evolution.

Refuges Refuges are places where the target pest is not exposed to Bt toxin. Refuges are a critical component of resistance management but more information about the productivity of potential refuge hosts is needed before a refuge implementation strategy can be developed. All of the stemborer species on maize in Kenya, including C. partellus and B. fusca, use numerous crop and non-crop plant species in addition to maize. In many locations, these plants are intercropped in maize fields. It is possible that some of these alternative host plants could be effective refuges and contribute toward resistance management, but the information is insufficient to determine this. This information should be collected, and can be collected prior to field release of Bt maize.

Environ. Risk. Chap 08

23/8/04

9:33

Page 263

Risk Assessment of Bt Maize in Kenya

263

Monitoring and response The goals of monitoring are: (i) to detect resistance so that remedial actions can be taken; (ii) to document control failures; (iii) to determine if the needs of farmers for adequate pest control are being met; and (iv) to monitor compliance to resistance management. Realistic, cost-effective monitoring methods need to be identified, developed and field tested. The management responses to the monitoring information that will be available to larger-scale growers will be different from those responses available to smaller-scale growers. This should be taken into account when additional research is conducted to determine which responses can be implemented in different parts of Kenya. Based on these principles and findings, an initial resistance management strategy for Bt maize in Kenya was proposed. This should be viewed as an initial proposal that will require modification as new information becomes available.

Key Lessons Learned Conducting a case study is very instructive. It provides participants with the opportunity to learn while focusing on an important topic, and is the utilitarian acid test for any concepts and principles that might guide the assessment. Consequently, many broad lessons were generated from the case study. Here, we wish to highlight a few major lessons from the workshop. Lesson 1: PFOA is an extremely useful process and should be implemented widely.

The PFOA provides an opportunity for multiple stakeholders to review the extent of the problem, the merits of a range of options that can address the problem, and make a choice to support or not support a technology based on its merits in relation to other options. When the Kenyan trial run of the PFOA was starting, nearly all of the participants indicated that they probably could not participate in the whole process, but after the first day, everyone came back, and as the PFOA Model was finishing up on the last day, one of the participants said, ‘This was a really good idea, we should do this for all of our agricultural technologies.’ Everyone in the room supported his statement and the discussion went on to identify ways a societal PFOA could become part of Kenya’s transgenic organism assessment process. The deliberative, problem identification and technology review process offers a rational, participatory approach to making decisions that will impact the nation, its people and its environment. Participants believed this review is especially important when implementing a technology based on a transgenic organism that has the possibility for irreversible consequences that could impact society beyond the user. Lesson 2: The workshop process enabled us to identify potential environmental risks, which had not been previously considered.

Environ. Risk. Chap 08

264

23/8/04

9:33

Page 264

A. Hilbeck et al.

Because the IRMA Project had been active prior to the workshop, many of the Kenyan scientists had spent time considering the potential environmental risks associated with Bt maize. In many cases, these scientists had already considered the potential risks to some extent before the workshop. The processes followed in the workshop, however, helped to organize these ideas into a coherent framework, which allowed us to systematically consider the potential environmental risks. As a consequence, categories of potential risks were identified, which had not been considered before, such as the potential effects of Bt maize on weeds, such as Striga, and plant disease risks, such as aflatoxin concentrations. Lesson 3: We have identified and prioritized risks and developed protocols to address them. This information could be incorporated into regulatory systems in Kenya and it could be used to request funding to support the necessary research.

For many participants, the general problem of assessing the risks of a transgenic organism, such as Bt maize, seemed like such a complicated problem that it could hardly be started, and once started, would likely never end. That we identified and prioritized potential risks within the limited time of the workshop was a major result for many participants, because it indicated that risk assessment was possible to accomplish, and a scientific and systematic approach can be used to accomplish it. Moreover, by identifying experiments and protocols that could provide results relevant to the risk assessment, participants became convinced that their scientific skills are useful in the process. The workshop transformed a seemingly insurmountable risk assessment problem into a do-able task that could be addressed efficiently with the assembled expertise. Lesson 4: Continuing the Kenya Bt maize case study will allow risk assessment to be carried out in a scientifically rigorous and transparent manner, as specified in the Cartagena Protocol on Biosafety (CBD, 2000, Annex III, paragraph 3).

We did not conduct a complete environmental risk assessment of Bt maize in Kenya. We began this process by identifying risks and discovering pathways by which Kenyans could continue to complete a risk assessment. This case study revealed that by using transparent, scientifically rigorous procedures, a risk assessment can be conducted consistent with the Cartagena Protocol on Biosafety. Lesson 5: Gene flow from Bt maize to non-transgenic open-pollinated maize varieties may be inevitable if Bt maize is effective. This could result in additional environmental risks to maize genetic diversity, non-target species and increase resistance risk.

We found that an important potential risk of Bt maize in Kenya involves gene flow from Bt maize hybrid and OPVs to the non-transformed open-pollinated maize varieties. If Bt maize is effective at controlling stemborers, then its use may become common in some regions of Kenya, from where it could spread rapidly to other maize varieties. If this occurs, it could have adverse effects on genetic diversity of maize, non-target species and insect resistance risk. All of these possibilities, however, require additional scientific investigation of the nature and magnitude, if any, of the potential risk.

Environ. Risk. Chap 08

23/8/04

9:33

Page 265

Risk Assessment of Bt Maize in Kenya

265

Completing a Risk Assessment in Kenya From the previous chapters and the discussion above, a full risk assessment of Bt maize in Kenya needs to be completed prior to final regulatory decision making. Several open questions and essential gaps in knowledge were identified during the workshop that would need to be addressed in such an analysis. Research programmes that aim at closing the identified gaps can be developed, and at the same time, the needs for capacity building and external assistance can be identified. Kenya will need to consider how and how much of the risk assessment procedures developed in this book can be incorporated into the national biosafety regulations and policy. In the following, we highlight the most important issues for consideration in the risk assessment procedures in Kenya.

PFOA Farmers in Kenya do not all regard the stemborer problem similarly. Surveys found that farmers identified many unmet needs in relation to maize production, many of which they regarded as more significant than stemborers. The risk assessment of Bt maize needs to ensure that Bt maize will not exacerbate other production problems. In addition, farmer training and educational programmes on integrated pest management (IPM) should be implemented, to ensure appropriate and sustainable use of the new strategies (e.g. implementation of resistance management programmes for Bt maize, effective use of ‘push–pull’ system). Finally, the PFOA test-run of this workshop was an excellent capacitybuilding exercise and could serve as a model for conducting a full PFOA, which should be a multi-stakeholder process minimally including KARI, farmers’ nongovernmental organizations, local farmer representation, consumer groups and universities. For it to be effective, the group recommended embedding the process in the national Kenya biosafety framework. One idea for doing this is by stages (other approaches are also possible): ●

●

Stage 1: A Kenyan government organization such as the Ministry of Agriculture (MoA) or Kenya Agricultural Institute (KARI) would receive the proposal on the transgenic organism and evaluate the general viability of the proposal in consultation with Kenya Plant Health Inspection Service (KEPHIS). This entry review would determine if the proposal is sufficiently well documented to merit conducting a PFOA. If so, MoA/KARI would forward the file and the National Biosafety Committee would make the decision that the technology is permissible under current national law and merits a PFOA. Stage 2: If the decision is to proceed, MoA/KARI and KEPHIS will convene a PFOA Board (PFOAB). Each PFOAB will be composed of individuals with specific expertise and knowledge pertinent to the cropping system under review. Representatives of the multiple stakeholders will be asked to participate based on the social sectors affected by the decision.

Environ. Risk. Chap 08

23/8/04

9:33

Page 266

266

A. Hilbeck et al.

●

Stage 3: The National Biosafety Committee would review the PFOA, which would frame the risk assessment and provide an evaluation of how this technology fits into Kenya’s policy and national interest.

Transgene expression and locus structure It is important to obtain the requisite information on transgene locus structure for the private-sector events, and ensure that the funding and expertise is available to analyse the public-sector events as they are developed, so that information is available when it is needed for risk assessment. For all of these events, the breeding methods and targets need to be revealed soon so that the risk assessment can be structured appropriately. Will the Bt transgenes be incorporated into maize hybrids that will be used by some farmers in some areas of Kenya or into OPVs that can be used much more widely? What is the likely time frame for development? This is essential to schedule the risk assessment process. It is also important to design and conduct realistic field trials. If small-scale subsistence farmers are the most likely users of Bt maize, field trials should be carried out on the types of marginal lands small farmers often only have available to them using smallholder management techniques under the extremes of the local environmental stresses. While we have not addressed field trials in the Kenya workshop, we feel it important to stress that such trials ought to be done prior to regulatory approval for commercial use. As a longer-term goal, it will be useful to consider how to develop the scientific capacity to implement the recommendations on transgene design (using two genetically linked Bt toxin genes, with independent actions and high-dose expression rates, that have no marker genes or non-expressing transgenes).

Biodiversity and non-target effects Several important questions remain open and will need to be addressed in a full risk assessment. What are the important pollinators and pollen-feeders, predators and soil macroorganisms that should be candidates for non-target species selection? Is there a need to assess impacts on biodiversity in addition to the species-specific approaches using the selection matrices? How can laboratory, semi-field and field experiments be combined in biosafety assessment of transgenic organisms? Closing these gaps of knowledge would generate ecological and agronomic knowledge necessary for conducting a full risk assessment for Bt maize, and would be essential for understanding and improving Kenyan maize production systems in a sustainable way. In addition, pathogen–Bt maize interactions should be investigated fully, to understand whether additional benefits or risks arise related to seed-borne fungal diseases or aflatoxin production, which are significant health issues in Kenya. Moreover, while Striga problems are restricted presently to certain areas in Kenya, there is a danger of it spreading to other parts of the country. Therefore, how Bt maize can interact with and impact Striga should be carefully investigated.

Environ. Risk. Chap 08

23/8/04

9:33

Page 267

Risk Assessment of Bt Maize in Kenya

267

Gene flow and its consequences Regional maize landrace performance and genetic diversity in Kenya is not well understood, but will be a critical factor in the success and impact of Bt maize. The susceptibility of widely used maize landraces to stemborer will determine whether smallholder farmers are likely to perceive Bt maize as acceptable and useful. Susceptibility profiles of these maize landraces to stemborers represent key information that will also help to predict the consequences of gene flow to landraces, such as the effects associated with target and non-target arthropods feeding on the landraces. In addition, the value and uniqueness of Kenyan maize landraces and the associated conservation issues should be evaluated. Considering the lack of barriers to gene flow and seed exchange, Kenya may want to consider its policies relative to segregation of transgenic and nontransgenic products, both as market options for farmers and consumers and with respect to international trade.

Resistance risk and management Knowledge gaps need to be filled before a resistance management programme can be refined. It is recommended that the compatibility of Bt maize cultivation and resistance management with IPM strategies, such as the ‘push–pull’ system be determined, prior to field release of Bt maize. Additionally, farmer training programmes on IPM and resistance management for Bt maize need to be established to ensure sustainable use of Bt maize. A starting point is the list of detailed recommendations made in Fitt et al. (Chapter 7, this volume). This training would best be part of a wider training programme as recommended in Nelson et al. (Chapter 3, this volume).

Reflections and Future Outlook The Kenya workshop was constructive, efficient and consensual, incorporating a wide range of expertise and scientists with widely varying opinions about Bt maize. The participants became engaged in the process and stayed motivated throughout. Because published data and information on many practical agricultural issues are often scarce in developing countries, the participation of experienced East African field experts was absolutely crucial for the success of the workshop. The close interaction among participants over several days built lasting professional relationships and helped to realize the scientist-to-scientist exchange designed into the process. Since the workshop, new international collaborations have been initiated, possible project proposals developed, scientific publications have been written (e.g. Capalbo et al., 2003; Hilbeck et al., 2004), and international scientific exchanges have been established, including international visits, hiring of PhD students and participation in international meetings of African, Brazilian and Vietnamese working group member scientists. The group continues to add more participant scientists, and

Environ. Risk. Chap 08

268

23/8/04

9:33

Page 268

A. Hilbeck et al.

membership is currently at over 200 members worldwide. Further, the project has been invited by the Swiss government to present an overview of its accomplishments at a side event of the first Meeting of the Parties (MOP1) to the Biosafety Protocol in Kuala Lumpur in February 2004. For all of the accomplishments documented in this book, the Kenya workshop is just the beginning of an evolving effort. The next developments in this effort will be documented in a workshop on risk assessment of Bt cotton in Brazil, followed by another workshop on risk assessment of Bt cotton in Vietnam. These next workshops will develop several significant additions to the risk assessment process. One of the unwarranted criticisms of risk assessment of transgenic crops is that it slows down the process of developing useful transgenic crops for countries that may need them urgently. However, we would argue that it is equally risky to accept a new product without evaluating how it can be useful, particularly with regard to sustainable crop management systems for subsistence farmers where livelihoods are at stake. The PFOA will help a country make appropriate evaluations in the context of risk assessment. Moreover, it must be recognized that risk assessment is itself a process, and when understood and implemented appropriately it can facilitate the development of useful transgenic crops. Key to accomplishing this facilitative role is a recognition that assessments in each scientific section need to be ordered in relation to the development of the transgenic plant. It takes several years to develop a potentially useful transgenic plant. During this time, distinct stages in development can be identified. The challenge for risk assessment is to ensure that critical information feeds into the development process at the appropriate time to further the aim of minimizing environmental risk, and avoid extra analyses and development costs at a later stage. Although it is possible to identify many ways to divide the development of a transgenic plant into stages, for risk assessment, there are several key transitions. How and when does it make sense to move from the small-scale contained laboratory or greenhouse experiments to small-scale outdoor field experiments? Similarly, how and when does it make sense to scale up from small field experiments to large-scale field experiment? By understanding the scientific rationale for how the risk assessment process dovetails into these questions, it should be possible to ensure that risk assessment experiments will be neither premature nor unnecessarily delayed, and that the development of the transgenic plant is not delayed unnecessarily. A second issue involves the generality of risk assessment data. It may be possible to make generalizations over space, time, cropping system, transformation event, and so on. For example, if an experiment is conducted in Kenya, are the results relevant for Tanzania? Ethiopia? Nigeria? India? Switzerland? Mexico? China? Japan? What are the limits to spatial generalization, and what are the scientific principles that justify such generalization? Similarly, when is it possible to generalize from one maize Cry1Ab transformation event to another maize Cry1Ab event? What about generalization to a maize Cry1Ac event? A maize Cry1B, Cry1E or Cry2A event? What about a cotton Cry1Ac event? The scientific basis for generalizing or not generalizing needs to be developed, so that risk assessment can be facilitated.

Environ. Risk. Chap 08

23/8/04

9:33

Risk Assessment of Bt Maize in Kenya

Page 269

269

These and other weighty issues will be addressed in the future workshops. The results from these workshops will be published in future volumes of this series. We believe that these efforts are a unique and innovative contribution from public-sector scientists to the environmental risk assessment of transgenic plants, supporting the process of implementation of the Biosafety Protocol.

References Capalbo, D.M.F., Hilbeck, A., Andow, D.A., Birch, A.N.E., Bong, B.B., Fitt, G.P., Fontes, E.M.G., Heong, K.L., Johnston, J., Osir, E.O., Snow, A., Songa, J. and Wan, F.-H. (2003) Brazil and the development of international scientific biosafety testing guidelines for transgenic plants. Journal of Invertebrate Pathology 83, 104–106. CBD (1992) Convention on Biological Diversity: Convention Text, www.biodiv.org/convention/articles.asp (accessed 25 November, 2003). CBD (2000) Cartagena Protocol on Biosafety to the Convention on Biological Diversity: Text and Annexes. Secretariat of the Convention on Biological Diversity, Montreal, www.biodiv.org/doc/legal/cartagena-protocol-en-pdf (accessed 1 December, 2003). Hilbeck, A., Andow, D.A., Birch, A.N.E., Bong, B.B., Capalbo, D., Fitt, G., Fontes, E., Heong, K.L., Johnston, J., Nelson, K., Osir, E., Snow, A., Songa, J. and Wan, F.-H. (2004) The GMO Guidelines Project: development of international scientific environmental biosafety testing guidelines for transgenic plants. In: Ehler, L.E., Sforza, R. and Mateille, T. (eds) Genetics, Evolution and Biological Control. CAB International, Wallingford, UK. NRC (National Research Council) (1983) Risk Assessment in the Federal Government: Managing the Process. National Academy Press, Washington, DC. NRC (National Research Council) (2002) Environmental Effects of Transgenic Plants: the Scope and Adequacy of Regulation. National Academy Press, Washington, DC.

Environ. Risk. Chap 08

23/8/04

9:33

Page 270

Environ. Risk. Index

23/8/04

9:33

Page 271

Index

Note: page numbers in italics refer to figures, tables and boxes aflatoxin 159–160, 257 contamination transmission 258–259 agroecological zones 22–23, 24 cropping systems 27, 28, 29 farming systems 252–253 fertilizer use 29 non-target species prioritization 120 stemborers 210, 211, 253 agroecosystem, non-target species 120 agronomic performance evaluation 6 anthropocentric functions 118, 119 aphids 142 Apis mellifera (honeybee) 135, 143–144, 149 attractiveness of Bt maize 161 colony development 162 hazard identification 152 arbuscular mycorrhizas 140 see also mycorrhizal associations arthropods diversity 126, 127–128 maize-associated 148, 149 Aspergillus flavus 37, 159–160, 257 transmission by Carpophilus 258–259 Bacillus thuringiensis (Bt) 2 backcrossing 5, 7 bacteria, free-living 176 bacterial gene expression 108–110 bar marker gene 110

bees, wild 143–144, 149 bioassay leaf tissue 106–107, 113 bioassays 226 laboratory 107–108 biocontrol organisms 129 biodiversity 117, 257–259 below-ground 117–118 intrinsic value 13 risk assessment 266 risks with gene flow 200 soil ecosystems 131 transgenic crop risks 117 as value 12–13 biolistic transformation 4, 90 biological control of stemborers 44 bitrophic exposure 144, 145, 148 Bt cotton 212, 213 Bollgard® 96, 97 pests 223 control 225–226 Bt cry genes 8 modification 8–9 Bt genes 2 fitness effects on maize cultivars 200–202 incorporation 7 other crops with 223 Bt maize adoption pattern 224–225 control efficacy estimates 257 271

Environ. Risk. Index

10/9/04

14:32

Page 272

272

Bt maize continued design 112 development 7–9 efficacy 113–114, 243 expression patterns 243 field testing 243 high-dose 244, 261–262 introduction to Kenya 188, 192 post-commercialization 243 potential impacts 253 seasonal restriction 244 smallholder experimentation 225 status 113–114 toxin dose 244, 261–262 worldwide use 3 Bt sprays 213 Bt toxin genes closely linked 94 strength of selection 225 Bt toxins dose 244, 261–262 expression level 225 insect potential exposure 213–219, 220–221, 222–225 intermediate level 215 persistence in soil 175–176 stemborer exposure 223 Busseola fusca (stemborer) 38, 39, 40 control level 257 crop hosts 218 importance 252, 253 resistance risk 210–212, 229–230, 262 Carpophilus (saprovore) 151–152 adult feeding 169 Aspergillus transmission 258–259 frass consumption 159 host-finding behaviour 165–167 host suitability alteration 167–168 Wolbachia interaction 168–169 Cartegena Protocol on Biosafety 1, 15, 16 risk assessment 264 cattle 29 Chilo (stemborer) 38, 39, 40 control level 257 crop hosts 218 importance 252, 253 resistance risk 210–212, 229–230, 262 spatial population genetic structure 218 CIMMYT (International Centre for the Improvement of Maize and Wheat) 252

Index

climate 22–23 production constraints 33 controls 13–14 isogenic 13 Convention on Biological Diversity (CBD) 12–13, 117 cornborer, European in USA 239–241, 242 Cotesia (larval parasitoid) 145–146, 147, 148, 149–150 hazard identification 153–154 counterfactuals, economic 14 crop residues destruction 45, 245 management 236–237 cropping systems 27, 28, 29 area 225 crops/crop plants 235 forage 219 resistance to Bt proteins 218–219 target 213–218 transgenes 85–86 see also transgenic crops cross-resistance Cry proteins 237–238 reduction 94 Cry toxins 8 cross-resistance 237–238 gene pyramiding 237–238 stemborer activity 113, 210–212 Cry1Ab toxin 223 efficacy 107 gene pyramiding 237–238 target transgene product 103 Cry1Ac toxin 223 Cry1B toxin 107 crystalline delta-endotoxins 8 see also Cry toxins cultivars 48 definition 205 local 198, 206 gene flow risk 260 genetic diversity 199 genetic uniqueness loss 199 see also landraces

dairying 29 diapause, stemborers 41, 214, 245 in maize fields 222 disease vectors 126, 132, 135 hazard identification 151–152

Environ. Risk. Index

23/8/04

9:33

Page 273

Index

dispersal of insect pests 217 DNA sequencing 255–256, 257 double-cropping 214, 252, 253 stemborers 42–43, 224

eclosion 215–216 ecological functions 118, 119 ecosystems, below-ground 117–118 ecotoxicology 123 ectopic expression 83–84 egg mass screen 230, 231–232, 239 cost 240 egg parasitoids 144, 145–146, 147 hazard identification 152–153 host egg suitability 162–163 host plant constituent effects on behaviour/performance 164–165 oviposition preferences 163–164 protocols 162–165 Eldana saccharina (stemborer) 38, 39, 40 elite inbred 5 Enviromental Protection Agency (US) 12 environment, transgene effects 85–86, 256 environmental risk assessment 9, 253 PFOA relation 58–59, 60–61, 62 environmental risk identification 263–264 enzyme-linked immunosorbent assay (ELISA) 99 epidemiological analysis 11 epistasis 84, 110–111, 114, 256 event 7 226 event 176 90, 92, 97 copy number per locus 97 mortality 226 plant-selectable marker gene 110 target transgene product 102–103 event 531 96 event 1835 226 event 5601 96, 226 event 5602 96 experimental studies 11 expert judgements 12 expert regulatory judgement 12 exposure definition 10 pathway analyses 121–122

F2 screen 230, 233, 239 cost 240–241

273

detection limit 241 farming systems 27, 29 agroecological zones 252–253 large 224–225 resistance management 242–243, 244 monoculture 223–224 small-scale 195 gene flow risks 200 seed recycling 224 stemborer resistance 223–224 see also intercropping; smallholders farms/farmers need determination for pest control 239, 241 productivity 26–27 size 23, 26 fertilizer use 29 field trials 266 fluorescent in situ hybridization (FISH) 95 forage crops 219 functional group establishment 118, 119 fungi mycorrhizal associations 131 pathogens 36–38 Fusarium spp. 37–38 futures, alternative 15, 254

gene flow 187–188, 259–261 Bt maize adoption 225 consequences 267 definition 205 experimental protocols 200–203 feral population establishment 190–191 to free-living plants 193 increase 260 inter-/intra-specific 189 landrace genetic diversity 199 to landraces 193–194 likelihood from Bt maize to recipient populations 188, 192–197 potential consequences 260–261 rate 260 recipients 259–260 risk assessment 189–192, 264, 267 risks to small-scale farmers 200 routes 260 seed saving 195 sources 259–260 spread 260

Environ. Risk. Index

23/8/04

9:33

Page 274

274

gene flow continued transgenes frequency increase 190–191, 197–198 predicted behaviour following 196 spread 194, 198–200 via pollen 260 gene gun 4 gene silencing 110 gene–environment interaction 101, 114 evaluation 112 gene–gene interaction 101 generational relative fitness 125 genetic diversity gene flow impact 260–261 neutral loci 203 genetically modified organisms (GMOs) 1 unexpected/unpredicted events 114 genomic sequence, bioinformatic evaluation 98 genotype–environment interaction 110–111 goats 29 grain feeders 143 grasses 131, 138, 139 wild hosts 219, 220–221, 222 see also Pennisetum (Napier grass); Sorghum (Sudan grass) guttation fluids 143–144 Carpophilus adult feeding 169

habitat management 255 harm identification 10 hazard identification 10 Helicoverpa (African bollworm) 223 resistance 212, 213 risk 230, 262 hemizygous transgenes 104, 106, 257 Bt gene copy 195 transgene spread rate 196 herbicide tolerance 2, 3 herbivores natural enemies 129, 131, 135, 137, 138 methods/protocols 162 non-target species 152–154 non-target species 126, 129, 132, 133–134, 135 exposure pathway analysis 142–143 hazard identification 151–157 lepidopteran 132, 143 methods 158–160 natural enemies 152–154

Index

protocols 158–160 resistance risk 262 see also lepidopteran non-target herbivores heterozygote advantage alteration 237–238 homozygous transgenes 104, 106, 195, 257 transgene spread rate 196 honeybee see Apis mellifera (honeybee) host-finding behaviour, Carpophilus 165–167 host plant resistance 46 host suitability alteration, Carpophilus 167–168 human needs, basic 253 hybrid plants 5, 6, 7 adoption 224–225 Bt 7 fitness effect of Bt gene 198 gene flow rate 196–197 hemizygous/homozygous 106, 111, 251 improved variety 205 introgression from 225 large-scale farming 192 percentage not expressing Bt genes 111 seed 30 use 251 hybridization 195

in-field screen 230, 232–233, 239 cost 240 sensitivity 241 insect pests dispersal 217 geographic population structure 217–218 grain feeders 143 maize resistance 8–9 mating frequency alteration 238 migration 217 movement alteration 238 population density manipulation 238 population genetic data 217–218 potential exposure to Bt toxin 213–219, 220–221, 222–225 potential overlap 223 production constraints 32, 34 refuges 262 susceptible individuals 242 see also leafhoppers; resistance to Bt proteins; saprovores; stemborers; named species

Environ. Risk. Index

23/8/04

9:33

Page 275

Index

Insect Resistant Maize for Africa (IRMA) 31–33, 46, 47 deployment of Bt genes 224 second-generation events 252 transformation events 87, 88 insecticides see pesticides integrated crop management integration of Bt maize 236 maintenance 245 push–pull system 45–46 intercropping 29, 45, 219 legumes 29, 45, 176, 219 smallholders 219, 252, 253 stemborer refuges 223

Kenya case study scope 15–17 climate 22–23 maize consumption 252 production 21–23, 24–25, 26–27, 28, 29–34, 252 PFOA 254–255, 265–266 risk assessment 265–267 transgene expression results 256–257 Kenya Agricultural Research Institute (KARI) 252 PFOA 77 Kenyan Plant Health Inspection Service (KEPHIS) 30 PFOA 77–78

laboratory bioassays 107–108 landraces conservation issues 267 definition 206 fitness differences from transgenic cultivars 202–203 fitness effects of transgenes 197–198, 200–202, 260 losses 199 gene flow risk 193–194, 260, 261 genetic diversity 199, 267 genetic uniqueness loss 199 genotype mixtures 261 regional performance 267 risk of genetic homogeneity 203 transgene presence/frequency monitoring 203

275

larvae development rate 216 dispersal 215 feeding preference 214–215 movement 214–215 larval parasitoids 145–146, 147 hazard identification 153–154 protocols 165–169 larval screen 230, 232, 239 cost 240 leaf tissue bioassay 106–107, 113 leafhoppers 132, 135, 142–143, 149 legumes intercropping 29, 45, 176, 219 nitrogen fixation 155 lepidopteran non-target herbivores 132, 143 egg parasitoid 144, 145–146, 147 hazard identification 151 larval parasitoid 145–146, 147 methods 158–160 oviposition preferences 158–159 protocols 158–160 susceptibility to Bt toxin 158 linkage maps 95 livestock 29 living modified organisms see genetically modified organisms (GMOs) locusts 132, 143, 149 macroorganisms 131 maize breeding 5–7 climatic areas 22–23, 24, 25 commercial production 26 consumption 21–22 cultivars 30–31 genotype mixtures 261 isolines 198 net buyers 26–27 net production 26 non-Bt fertilization by Bt pollen 215 non-transgenic 197 production in Kenya 21–23, 24–25, 26–27, 28, 29–34 areas 22–23 costs 26 seed use 30–31 white endosperm 31, 251 maize-associated flora protocols 169–173 see also weeds

Environ. Risk. Index

23/8/04

9:33

Page 276

276

Maize Data Base Project (MDBP) 31–33 maize residue degradation 140 maize streak virus (MSV) 36, 257 resistance 36 transmission 132, 135 maize–arbuscular mycorrhizal associations 140 marker genes elimination 93 lack of expression 109 plant-selectable 109 selectable 90, 108–110 mating 215–216 activity reduction 238 disruption 242 frequency alteration 238 non-random assortative 216 maximum potential exposure 121 microbial symbionts 120 migration of insect pests 217 Mon-810 event 91 marker gene expression absence 110 target transgene product 102–103 monitoring tools, DNA-based 99 monoculture 223–224 mycorrhizal associations 131, 140, 156 development/colonization/function 177–178

natural enemies 44, 135, 137, 138 maize herbivores 129, 131 exposure path analysis 144, 145–146, 147 methods/protocols 162 non-target species as food 126 herbivores hazard identification 152–154 pollen feeding 144 release 242 Nei’s FST values 203 nitrification assays 177 nitrogen fixation 140, 155, 176 plant residue conversion to inorganic 176–177 supply 140 non-crop plants, resistance to Bt proteins 218, 219, 220–221, 222 non-target effects 257–259

Index

risk assessment 266 non-target environmental risk assessment model 117–118, 119, 120–125 ecological controls 124 ecotoxicology 123 exposure pathway analyses 121–122, 142–151 functional groups 118, 119, 120, 126–132 generational relative fitness 125 genetic controls 124 hazard identification 123, 151–157 hypothesis development 123, 151–157 indirect effects 123 Kenya case study 125–178 laboratory test limitations 124 measurement endpoints 125 methods 157–178 protocols 123–125, 157–178 whole-plant methodology 123 non-target species biocontrol 129 categorizing 126, 127–128, 129, 130, 131–132 classification 119, 120 criteria for prioritizing 121 gene flow impact 260–261 high-priority category selection 120–121 high standing biomass 120 potential hazard assessment 258–259 prioritizing 120–121, 132–142 selection matrices 120, 121 see also herbivores, non-target species nptII gene 14 nuclear genome, transgene location 95–96 nuclear transformations 95 nutrient cycling 131

open-pollinated variety (OPV) 5, 7 adoption 224–225 Bt gene copies 195 definition 206 gene flow rate 196–197 hemizygous/homozygous mixture 111, 252 homozygous 106, 111 introgression from 225 percentage not expressing Bt genes 111 production 251–252 seed renewal 30

Environ. Risk. Index

23/8/04

9:33

Page 277

Index

small-scale farmers 192 use 31, 251 open reading frames (ORFs) new 98 unpredicted 84 organelle genome 95–96 organic matter decomposition 174–175 oviposition preferences behaviour manipulation 238 egg parasitoids 163–164 lepidopteran non-target herbivores 158–159 lepidopteran stemborers 222 wild grasses 236 ox-ploughing 29

parasitoids 44 bitrophic exposure 144, 145 pollen feeders 143 resistant larvae survival reduction 236 of stemborers 135, 137, 138 tritrophic exposure 145–146, 147 see also egg parasitoids; larval parasitoids Pennisetum (Napier grass) 154, 219, 236 pesticides additional control in transgenic crops 236, 242 botanical 44 extent of use 212–213 local 44 resistance 209, 212–213 synthetic 43, 49 pests 32, 34–38, 39, 40–46, 48–49 control additional tactics in transgenic crops 236 traditional methods 242 heterozygote advantage alteration 237–238 production constraints 34 secondary 126 storage 34, 37 see also insect pests Phaseolus bean maize cropping systems 131 nitrogen fixation 140 phosphate supply 140 plant compounds, trophic levels 126 plant nutrient release 140 plant residue conversion to inorganic nitrogen 176–177

277

plant tissue testing 226 plants, free-living, gene flow risk 193 plant–soil ecosystem functional dynamics 156 plasmid maps 90, 91 pleiotropy 84, 110–111, 114, 256 pollen, maize Carpophilus adult feeding 169 fertilization of non-Bt maize 215 gene flow 260 pollination/pollen-feeding insects 129, 130, 135, 136 colony development 162 effectiveness 161–162 exposure pathway analysis 143–144 hazard identification 152 individual fitness 160–161 methods/protocols 160–162 polymerase chain reaction (PCR) primers 99 potential adverse effect 121 potential likely exposure 121 private-sector events 87, 88, 90–91, 92 copy number per locus 97 development stage 113 event 176 90, 92 transgene locus structure 266 Problem Formulation and Options Assessment (PFOA) 58, 253–255 attributes 65–67, 72, 73 changes 68–69, 73 controls 254 data management/analysis 79 environmental risk assessment 58–59, 60–61, 62, 253 evaluation 74–78 GMO risk evaluation 78 group composition 75–76 human needs 253 impact 70–71 information management 80 Kenya 265–266 Bt maize 62, 64, 65–72, 73 results 254–255 multi-stakeholder 74 new technologies 78 process 60–61 risk assessment 254 spatial scale 79 stakeholders 77, 78–79 timing 79 topics 60–61, 76 uncertainties 75 value 263

Environ. Risk. Index

23/8/04

9:33

Page 278

278

problem identification 10 production constraints 31–34 public-sector events 87, 88, 252 copy numbers per locus 97 development stage 113 first generation 89, 95, 96 second generation 89, 91, 95, 96 events 252 transgene loci 97, 98 transgene locus structure 266 push–pull strategy 73 grasses 138 integrated crop management 45–46 stemborers grass hosts 219, 223, 236 suppression 255 Striga control component 172–173 suppression 255

rainfall 23, 25 refuges 234–236, 244–245 compliance 241 crop plants 235 effectiveness 243 intercropping 223 resistance management strategy 94, 244–245, 262 size increase 242, 243 stemborers 223, 235 undisturbed 237 wild hosts 235 regenerated plants 5 resistance 2, 3 cost-detection limit crossover 240–241 definition 227–229 failure 241 increase detection 239 management during field testing 243 failure 239 implementation effectiveness 239 post-commercialization 243 refuge-based strategy 94 potential responses 242–243 risk assessment 106 resistance alleles 94 frequency 228 initial 227 resistance to Bt proteins 209–210

Index

crop plants 218–219 cultural control 236–237 delaying 236 exposure reduction 234–236 frequency 227 changes 239 gene flow impact 260–261 genetic basis variation 227 geographical spread 216–217 heterozygote advantage alteration 237–238 high-dose events 226 increased detection 239–240 individual trait 227, 228 intermediate level toxins 215 laboratory selection 226–227 larval development 216 larval movement 214–215 management 234–238, 261–263, 267 strategy 244–245, 263 weak link 262 mitigation programme 245 monitoring 263 degradation 243 goals 238–243 methods 230–233, 239–241 programme 245 non-crop plants 218, 219, 220–221, 222 response plan 238–243, 263 risk 212–213, 229–230, 244–245, 261–263 knowledge gaps 267 potential 227–230 species at risk 210–212, 262 Rhizobia 140 rice intercropping 219 risk definition 10 management 11 valuation 10 risk analysis terms 10 risk assessment 9–11, 264 alternative futures 15, 254 case-by-case 85 Kenya 265–267 standards of evidence 11–12 transgenic plants 83–84

saprovores 133–134, 143 hazard identification 151–152

Environ. Risk. Index

23/8/04

9:33

Page 279

Index

Scientific Advisory Panel (SAP, US) 12 sedges 131, 138, 139 seed(s) annual purchase of new 195 breeding fields 197 certified 224 criteria 30–31 dispersers 129 exchange 30 improved 30, 31 local varieties 30–31 predators 129 recycling 30, 31, 224, 253 renewal 30 saving 195, 215, 225 selection matrices 258 sentinel plots 239 Sesamia calamistis (stemborer) 38, 39, 40 crop hosts 218 resistance risk 210–212, 229 smallholders experimentation with Bt maize 225 gene flow rate 260 intercropping 219, 252, 253 natural refuge areas 235 resistance management 242–243, 244 resource-poor 241 seed recycling 224 soil biodiversity 117–118 Bt toxin persistence 175–176 erosion 28, 29 genetic microbial diversity 173–174 management 29 measures of activity 173 microbial communities 173–178, 257 microbial systems 131, 173 nitrogen fixation rate 176 nutrients 29, 131 organic matter decomposition 174–175 suppression of weeds 35–36 soil ecosystems baseline data 123 dynamics 122 exposure pathway analysis 148 functions 131, 140, 141, 142 hazard identification 155–157 inputs 122 methods 173 non-target impacts 157 sorghum (crop)

279

BT transformation 223 intercropping 219 Sorghum (Sudan grass) 154, 156, 219, 236 shared tertiary genome with maize 193 stemborer fitness impact 170 Southern blot analysis monitoring tools 99 transgene locus 96–97 species selection 257–258 spider mites 132, 143, 149 stemborers 38, 39, 40–46, 48–49 adult emergence 215–216 adult movement 216–217 agroecological zones 253 alternative habitats 222 biological control 44 botanical control agents 44 Bt toxin exposure 223 chemical control 43 continuous breeding in monoculture 224 control level 257 control strategies 43–46, 47, 49, 113–114, 252 crop hosts 235 crop loss 42–43 Cry toxin efficacy 107 cultural control 45, 236–237 diapause 41, 214 in maize fields 222, 245 dispersal 217 ecology 40–42 emergence 216 emergence holes 235 fitness impact on Sorghum weeds 170 fungal spore transport 37 genetic integration on other plants 222 geographic population structure 217–218 grass hosts 219 high-dose transgenic event 225–226 host plants 41–42 resistance 46 suitability 138 switching 138 larvae development rate 216 dispersal 215 feeding preference 214–215 movement 214–215 local control agents 44, 49 mating 215–216 migration seasonal variation 217

Environ. Risk. Index

23/8/04

9:33

Page 280

280

stemborers continued movement patterns 242 multiple generations in monoculture 224 parasitoids 135, 137, 138 hazard identification 152–154 potential exposure on target crop 213–218 push–pull control system 45–46 refuges 223, 235, 262 resistance 210–212, 261 farming systems 223–224 frequency 227 non-crop plants 219, 220–221, 222 other crops 218–219 other crops with Bt transgenes 223 risk 212–213, 262 target crop 213–218 sorghum crops 223 species complex 38, 40 suppression in push–pull strategy 255 susceptible individuals 242 transgene fitness effects 201 wild hosts 235 grasses 219, 220–221, 222 Striga (witchweed) 35–36, 138, 139 Bt maize impact 259, 266 control component of push–pull system 172–173 crop losses 131 fitness 172 germination stimulant production/activity 170–172 hazard identification 154–155, 156 suppression in push–pull strategy 255 super-refuges 242 sympatric populations 222

T-DNA 4, 93 target gene expression 100–104, 105, 106–108 target gene promoter 102 target transgene product 102–103 theoretical models 11 tissue age 226 toxicological evaluations 13–14 toxins concentration in plants 104, 105 exposure pathway analyses 121–122 tractors 29 transformation events 87–88, 89

Index

laboratory bioassays 107–108 leaf tissue bioassays 106–107 transformed plants 90 transgene(s) agronomic effects 198–200 characterization 112–113 crop 85–86 design 85–86, 88, 90–91, 92, 93–94 DNA sequence 98–99 ecological effects 198–200 environment 85–86 epistatic effects 110–111 fitness effects 198, 199 on landraces 197–198 frequency increase from gene flow 190–191, 197–198 high-dose event 225–226 increase 260 insertions 4 intergenerational transmission 86 landraces fitness effects 197–198 presence/frequency monitoring 203 location in maize genome 95–96 movement 259–261 number of copies at each locus 97, 98 other crops with 223 persistence 191–192, 195, 198–200 pleiotropic effects 110–111 predicted behaviour following gene flow 196 pyramiding of additional 237–238 recipient populations 199 routes of escape 195 seed saving 195 silencing 4 spread 191–192, 194, 198–200, 260 structured assessment 111–114 target gene expression 100–104, 105, 106–108 transmission 87, 111, 195 unknown/unexpected gene product production 256 see also gene flow transgene expression 84–87, 100–104, 105, 106–111, 255–257 biochemical characterization 103–104, 105, 106 characterization 85–87, 243 environment 256 functional characterization 101, 106–108

Environ. Risk. Index

23/8/04

9:33

Page 281

Index

hemizygous/homozygous plants 104, 106 increased 238 Kenya results 256–257 molecular characterization 101–103, 256 phenotype characterization 86–87 risk assessment 266 target gene expression 100–104, 105, 106–108 testing 226 whole plant methods 112–113, 256 transgene locus bioinformatic evaluation 98 complexity 84 number in Bt maize 96–97 sequencing 112 Southern blot analysis 96–97 structure 83–87, 94–99, 255–257 characterization 85–87 risk assessment 266 transgene product 85 location in plants 104, 105 non-target species interactions 122 timing in plants 104, 105 transgenesis 2–9 techniques 3–5 transgenic crops 85–86 additional pest control 236 fitness differences from landraces 202–203 gene flow 187–188 worldwide use 2–3, 4 transgenic events high-dose 262 see also numbered events transgenic plants exposure pathway analyses 121–122 movement of material in soil 148 non-target species interactions 122

281

risk assessment 83–84 transport of material in soil 148 Trichogramma (parasitoid wasp) 143, 144, 145–146, 147, 148, 149 hazard identification 152–153 host egg suitability 162–163 host plant constituent effects on behaviour/performance 164–165 oviposition preferences 163–164 protocols 162–165 tritrophic exposure 145–146, 147, 148

valuation of risk 10 varieties of maize 99 vegetation diversity 255 viral diseases 36

water stress 29 weeds broad-leaf 138, 139 co-occurring with maize 222 exposure pathway analysis 147 free-living 154 hazard identification 154–155 maize-associated 131, 138, 139 parasitic 138, 139, 154–155 production constraints 33–36 protocols 169–173 soil suppression 35–36 see also Striga (witchweed) weighted risk 10 wild hosts 219, 220–221, 222 refuges 235 witchweed see Striga (witchweed) Wolbachia, Carpophilus interaction 168–169

Environ. Risk. Index

23/8/04

9:33

Page 282