BROADENING THE GENETIC BASE OF CROP PRODUCTION

BROADENING THE GENETIC BASE OF CROP PRODUCTION

Edited by

H.D. Cooper
Food and Agriculture Organization of the United Nations, Rome, Italy, and Secretariat of the Convention on Biological Diversity, Montreal, Canada

C. Spillane
Institute for Plant Biology, Zurich University, Zurich, Switzerland

T. Hodgkin
International Plant Genetic Resources Institute, Rome, Italy

CABI Publishing
in association with

Food and Agriculture Organization of the United Nations

and

International Plant Genetic Resources Institute

 2001 by IPGRI and FAO. A catalogue record for this book is available from the British Library, London, UK. Library of Congress Cataloging-in-Publication Data Broadening the genetic base of crop production / edited by H.D. Cooper, C. Spillane, and T. Hodgkin. p. cm. Includes bibliographical references. ISBN 0-85199-411-3 (alk. paper) 1. Food crops-- Germplasm resources. 2. Plant genetics. 3. Genebanks, Plant. I. Cooper, H. D. (H. David) II. Spillane, Charlie. III. Hodgkin, T. IV. Food and Agriculture Organization of the United Nations. V. International Plant Genetic Resources Institute. SB123.3.B76 2000 333.9534-- dc21 Published jointly by: CABI Publishing, a division of CAB International CABI Publishing, CAB International, Wallingford, Oxon OX10 8DE, UK Tel.: +44(0)1491 832111; Fax: +44 (0)1491 833508; Email: [email protected]; Web site: www.cabi.org CABI Publishing, 10 E 40th Street, Suite 3203, New York, NY 10016, USA Tel: +1 212 481 7018; Fax: +1 212 686 7993; Email: [email protected] Food and Agriculture Organization of the United Nations (FAO) Viale delle Terme di Caracalla, 00100 Rome, Italy Tel.: +39 06 57051; Fax: +39 06 57053152; Email: [email protected]; Web site: www.fao.org International Plant Genetic Resources Institute (IPGRI) Via delle Sette Chiese 142, 00145 Rome, Italy Tel.: +39 06 51892; Fax: +39 06 5750309 Email: [email protected]; Web site: www.ipgri.cgiar.org ISBN 0 85199 411 3 (CABI) ISBN 92 5 104438 4 (FAO) ISBN 92 9043 458 9 (IPGRI) The designations employed and the presentation of material in this publication do not imply the expression of any opinion whatsoever on the part of the Food and Agriculture Organization of the United Nations or the International Plant Genetics Resources Institute concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. The designations developed and developing economies are intended for statistical convenience and do not necessarily express a judgement about the stage reached by a particular country, territory or area in the development process. The views expressed herein are those of the authors and do not necessarily represent those of the Food and Agriculture Organization of the United Nations and the International Plant Genetic Resources Institute. All rights reserved. Reproduction and dissemination of material in this publication for educational or other non-commercial purposes are authorized without any prior written permission from the copyright holders provided the source is fully acknowledged. Reproduction of material in this publication for resale or other commercial purposes is prohibited without written permission of the copyright holders. IPGRI is an autonomous international scientic organization, supported by the Consultative Group on International Agricultural Research (CGIAR). IPGRIs mandate is to advance the conservation and use of genetic diversity for the well-being of present and future generations. IPGRIs headquarters is based in Rome, Italy, with ofces in another 19 countries worldwide. It operates through three programmes: (1) the Plant Genetic Resources Programme, (2) the CGIAR Genetic Resources Support Programme, and (3) the International Network for the Improvement of Banana and Plantain (INIBAP). FAO is the largest specialized agency in the United Nations system and the lead agency for agriculture, forestry, sheries and rural development. An intergovernmental organization, FAO has 180 Member Nations plus a member organization, the European Community. Since its inception in 1945, FAO has worked to alleviate poverty and hunger by promoting agricultural development, improved nutrition and the pursuit of food security the access of all people at all times to the food they need for an active and healthy life. Typeset in 10/12pt Garamond by Columns Design Ltd, Reading Printed and bound in the UK by Biddles Ltd, Guildford and Kings Lynn

00044455

Contents

Contributors Foreword M. Duwayri and G. Hawtin Preface Acronyms and Abbreviations PART I GENERAL PRINCIPLES 1 2 Broadening the Genetic Base of Crops: an Overview H.D. Cooper, C. Spillane and T. Hodgkin Evolutionary and Genetic Perspectives on the Dynamics of Crop Genepools C. Spillane and P. Gepts Base-broadening: Introgression and Incorporation W. Spoor and N.W. Simmonds The Role of Local-level Geneow in Enhancing and Maintaining Genetic Diversity J. Berthaud, J.C. Clment, L. Emperaire, D. Louette, F. Pinton, J. Sanou and S. Second Regulatory Aspects of Breeding for Diversity N.P. Louwaars

ix xv xvii xxiii 1 1

25 71

3 4

81

105

vi

Contents

Decentralized and Participatory Plant Breeding for Marginal Environments S. Ceccarelli, S. Grando, A. Amri, F.A. Asaad, A. Benbelkacem, M. Harrabi, M. Maatougui, M.S. Mekni, H. Mimoun, R.A. El-Einen, M. El-Felah, A.F. El-Sayed, A.S. Shreidi and A. Yahyaoui Empowering Farmers and Broadening the Genetic Base: Agricultural Research and Resource Management R. Salazar

115

137

PART II: CROP CASE STUDIES 8 9 The State of Millet Diversity and its Use in West Africa O. Niangado The State of the Use of Maize Genetic Diversity in the USA and Sub-Saharan Africa S.P. Tallury and M.M. Goodman The State of the Use of Potato Genetic Diversity R. Ortiz The State of the Use of Cassava Genetic Diversity and a Proposal to Enhance it G. Second and C. Iglesias The State of the Use of Musa Diversity S.L. Sharrock, J.-P. Horry and E.A. Frison 147

159 181

10 11

201 223

12

PART III: POPULATION MANAGEMENT 13 Dynamic Management of Genetic Resources: a 13-year Experiment on Wheat I. Goldringer, J. Enjalbert, J. David, S. Paillard, J.L. Pham and P. Brabant Genetic Base-broadening of Barley (Hordeum vulgare L.) in the Nordic Countries M. Vetelinen and E.A.J. Nissil Evolutionary Changes in Cambridge Composite Cross Five of Barley K.M. Ibrahim and J.A. Barrett Genetic Base-broadening in the West Indies Sugar Cane Breeding Programme by the Incorporation of Wild Species A.J. Kennedy Potential of Genetic Resources and Breeding Strategies for Basebroadening in Beta L. Frese, B. Desprez and D. Ziegler 245

14

261 271

15 16

283

17

295

Contents

vii

18

HOPE, a Hierarchical, Open-ended System for Broadening the Breeding Base of Maize L.W. Kannenberg The Germplasm Enhancement of Maize (GEM) Project: Private and Public Sector Collaboration L.M. Pollak and W. Salhuana A French Cooperative Programme for Management and Utilization of Maize Genetic Resources A. Gallais, J.P. Monod and others

311

19

319

20

331

PART IV: OTHER APPROACHES TO BROADENING THE GENETIC BASE OF CROPS 21 Broadening the Genetic Base of Lentil in South Asia W. Erskine, S. Chandra, M. Chaudhry, A. Sarker, B. Sharma, M. Tufail and M.C. Tyagi Genetic Diversity of Barley: Use of Locally Adapted Germplasm to Enhance Yield and Yield Stability of Barley in Dry Areas S. Grando, R. von Bothmer and S. Ceccarelli Breeding Phaseolus for Intercrop Combinations in Andean Highlands J.P. Baudoin, F. Camarena, M. Lobo and G. Mergeai Improving Potato Resistance to Disease under the Global Initiative on Late Blight B.R. Trognitz, M. Bonierbale, J.A. Landeo, G. Forbes, J.E. Bradshaw, G.R. Mackay, R. Waugh, M.A. Huarte and L. Colon A Mexican Bean Breeding Programme for Comprehensive Horizontal Resistance to all Locally Important Pests and Diseases R. Garca Espinosa, R.A. Robinson, P. Ramrez Vallejo, F. Castillo Gonzlez and F. Romero Rosales The Impact of Decentralized and Participatory Plant Breeding on the Genetic Base of Crops J.R. Witcombe Base-broadening for Client-oriented Impact: Insights Drawn from Participatory Plant Breeding Field Experience L. Sperling, J. Ashby, E. Weltzien, M. Smith and S. McGuire 341

22

351 373

23 24

385

25

399

26

407

27

419

Final Statement by Participants in the Rome Workshop on Broadening the Genetic Base of Crops, September 1997 Index

437 441

Contributors

A. Amri, International Center for Agricultural Research in the Dry Areas (ICARDA), PO Box 5466, Aleppo, Syria F.A. Asaad, Field Crops Research Institute, Agricultural Research Center (ARC), PO Box 12619, 9 Gamaa Street, Giza, Egypt J. Ashby, International Centre for Tropical Agriculture (CIAT), Cali, Colombia A. Baratin, CAUSSADE Semences, Z.I. Avenue de Meaux, 82300 Caussade, France M. Baron, LIMAGRAIN Gntique, BP 1, 63720 Chappes, France J.A. Barrett, Department of Genetics, University of Cambridge, Downing Street, Cambridge CB2 3EH, UK A. Barrire, INRA, Station dAmlioration des Plantes Fourragres, 86600 Lusignan, France J.P. Baudoin, Facult Universitaire des Sciences Agronomiques de Gembloux, 2 Passage des Dports, 5030 Gembloux, Belgium A. Benbelkacem, Institut Technique des Grandes Cultures (ITGC), BP 16, El-Harrach, Algiers, Algeria J. Berthaud, Institut de Recherche pour le Dveloppement (IRD), 911 Avenue Agropolis, BP 5045, 34032 Montpellier cedex, France G. Berthe, LIMAGRAIN Gntique, BP 1, 63720 Chappes, France M. Bonierbale, International Potato Center (CIP), PO Box 1558, Lima, Peru P. Brabant, INA-PG, 16 rue Claude Bernard, 75005 Paris, France J.E. Bradshaw, Scottish Crop Research Institute (SCRI), Invergowrie, Dundee DD2 5DA, UK A. Boyat, INRA, Station dAmlioration des Plantes, Domaine de Melgueil, 34130 Mauguio, France F. Camarena, Universidad Nacional Agraria La Molina, Apdo 456, Lima, Peru M. Cambolive, PIONEER Gntique, BP 5, 31840 Aussonne, France P. Carolo, RUSTICA Prograin Gntique, Domaine de Sandreau, Mondonville, 31700 Blagnac, France

ix

Contributors

F. Castillo Gonzlez, Colegio de Postgraduados, IREGEP, Texcoco, Mxico 56230, Mxico S. Ceccarelli, International Center for Agricultural Research in Dry Areas (ICARDA), PO Box 5466, Aleppo, Syria S. Chandra, Indian Institute for Pulses Research, Kanpur 208024, India A. Charcosset, INRA-UPS-INA.PG, Station de Gntique Vgtale, Ferme du Moulon, 91190 Gif-sur-Yvette, France M. Chaudhry, Ayub Agricultural Research Institute, Faisalabad, Pakistan J.C. Clment, IRD, 213 Rue Lafayette, 75480 Paris cedex 10, France L. Colon, Centre for Plant Breeding and Reproduction (CPRO-DLO), PO Box 16, Wageningen 7600 AA, The Netherlands H.D. Cooper, Food and Agriculture Organization of the United Nations. Present address: Secretariat of the Convention on Biological Diversity, World Trade Centre, Suite 300, 393 St Jacques, Montreal, Quebec, Canada H2Y 1N9 J. David, ENSAM INRA, UFR de Gntique et Amelioration des Plantes, Domaine de Melgueil, 34130 Mauguio, France J. Dallard, INRA, Station dAmlioration des Plantes, Domaine de Melgueil, 34130 Mauguio, France M. Derieux, INRA, Domaine de Brunehaut, 80200 Estres-Mons, France B. Desprez, Florimond Desprez Co., BP 41, 59242 Cappelle en Pvle, France J.L. Desselle, INRA, Station dAmlioration des Plantes, Domaine de Melgueil, 34130 Mauguio, France F. Devaux, MAISADOUR, Silo de Haut Mauco, Route de Saint Sever, 40000 Mont de Marsan, France P. Dubreuil, INRA-UPS-INA.PG, Station de Gntique Vgtale, Ferme du Moulon, 91190 Gif-sur-Yvette, France J. Durovray, NOVARTIS, 12 route de Crossay, Senneville, 28700 Francourville, France H. Duval, INRA, Station dAmlioration des Plantes, Domaine de Melgueil, 34130 Mauguio, France R.A. El-Einen, Field Crops Research Institute, Agricultural Research Center (ARC), PO Box 12619, 9 Gamaa Street, Giza, Egypt M. El-Felah, Institut National de la Recherche Agronomique de Tunisie (INRAT), Rue Hedi Karray, 2880 Ariana, Tunis, Tunisia A.F. El-Sayed, Field Crops Research Institute, Agricultural Research Center (ARC), PO Box 12619, 9 Gamaa Street, Giza, Egypt L. Emperaire, IRD/ISA, SCLN 210, Bloco C, sala 112, 70862-530, Brasilia (DF), Brazil J. Enjalbert, INRA, Laboratoire de Pathologie Vgtale, Grignon, 78850 ThivervalGrignon, France W. Erskine, International Center for Agricultural Research in Dry Areas (ICARDA), PO Box 5466, Aleppo, Syria M. Ferranet, MAISADOUR, Silo de Haut Mauco, Route de Saint Sever, 40000 Mont de Marsan, France G. Forbes, International Potato Center (CIP), Experimental Station at Santa Catalina, PO Box 17-21-1977, Quito, Ecuador L. Frese, Federal Centre for Breeding Research on Cultivated Plants (BAZ), Gene Bank, Braunschweig, Germany E.A. Frison, International Network for the Improvement of Banana and Plantain, Parc Scientique Agropolis II, 34397 Montpellier cedex 5, France

Contributors

xi

A. Gallais, Station de Gntique Vgtale, INRA-PROMAIS programme, Ferme du Moulon, 91190 Gif-sur-Yvette, France R. Garca Espinosa, Colegio de Postgraduados, IFIT, Texcoco, Mxico 56230, Mxico P. Garnier, INRA-UPS-INA.PG, Station de Gntique Vgtale, Ferme du Moulon, 91190 Gif-Sur-Yvette, France P. Gepts, Department of Agronomy and Range Science, University of California, 1 Shields Avenue, Davis, CA 95616-8515, USA. Email: [email protected] I. Goldringer, INRA-UPG-INA.PG, Station de Gntique Vgtale, Ferme du Moulon, 91190 Gif-sur-Yvette, France M.M. Goodman, Department of Crop Science, North Carolina State University, Raleigh, NC 27695, USA B. Gouesnard, INRA, Station dAmlioration des Plantes, Domaine de Melgueil, 34130 Mauguio, France S. Grando, International Center for Agricultural Research in Dry Areas (ICARDA), PO Box 5466, Aleppo, Syria M. Harrabi, Institut National Agronomique de Tunisie (INAT), 43 Avenue Charles Nicole, 1082 Tunis-Belvdre, Tunisia T. Hodgkin, International Plant Genetic Resources Institute, Via delle Sette Chiese 142, 00145 Rome, Italy J.-P. Horry, CIRAD-FLHOR, Dpartement des Productions Fruitires et Horticoles, BP 153, 97202, Fort de France cedex, Martinique M.A. Huarte, Instituto Nacional de Tecnologa Agropecuaria (INTA), Ruta 226, km 73.5, Balcarce, Buenos Aires 7620, Argentina K.M. Ibrahim, School of Biological Sciences, University of East Anglia, Norwich NR4 7TJ, UK C. Iglesias, 1400 Woodland, West Lafayette, IN 47906, USA L.W. Kannenberg, Plant Agriuclture Department, University of Guelph, Guelph, Ontario, Canada N1G 2W1 A.J. Kennedy, West Indies Central Sugar Cane Breeding Station, Groves, St George, Barbados A. Lacombe, VAN der HAVE France, Domaine de la Revanche, 32700 Lectoure, France B.A. Lambert, VERNEUIL Recherche, Avenue de la Croix Blanche, 49630 Corn, France J.A. Landeo, International Potato Center (CIP), PO Box 1558, Lima, Peru V. Lavergne, INRA-UPS-INA.PG, Station de Gntique Vgtale, Ferme du Moulon, 91190 Gif-sur-Yvette, France M. Lefort, Bureau des Ressources Gntiques (BRG), 16 rue Claude Bernard, 75231 Paris cedex 05, France M. Leipert, SDME, Centre de Slection, rue du Bois Musquet, 28300 Champhol, France M. Lobo, Corporacin Colombiano de Investigacin Agropecuaria, C.I. La Selva, Apartado Aereo 100 470, Rionegro, Colombia D. Louette, Chemin de la Croix, 30260 Brouzet les Quissac, France N.P. Louwaars, Plant Research International, PO Box 16, 6700 AA, Wageningen, The Netherlands M. Maatougui, Institut Technique des Grandes Cultures (ITGC), BP 16, El-Harrach, Algiers, Algeria

xii

Contributors

G.R. Mackay, Scottish Crop Research Institute (SCRI), Invergowrie, Dundee DD2 5DA, UK A. Majest, COOP DE PAU Semences, Avenue Gaston Phoebus, 64230 Lescar, France S. McGuire, Technology and Agrarian Development, Wageningen University, Nieuwe Kanaal 11, 6709 PA Wageningen, The Netherlands M.S. Mekni, Institut National de la Recherche Agronomique de Tunisie (INRAT), Rue Hedi Karray, 2080 Ariana, Tunis, Tunisia G. Mergeai, Facult Universitaire des Sciences Agronomiques de Gembloux, 2 Passage des Dports, 5030 Gembloux, Belgium P. Miclo, INRA, Station dAmlioration des Plantes, Domaine de Melgueil, 34130 Mauguio, France H. Mimoun, Institut Technique des Grandes Cultures (ITGC), BP 16, E-Harrach, Algiers, Algeria D. Misevic, PIONEER Gntique, BP 5, 31840 Aussonne, France J.P. Monod, PROMAIS, 20 rue Bachaumont, 75002 Paris, France Y. Montalant, INRA, Domaine de Brunehaut, 80200 Estres-Mons, France O. Niangado, Foundation Novartis, BPE 1449, Bamako, Mali E.A.J. Nissil, Boreal Plant Breeding Ltd, FIN-31600 Jokioinen, Finland P. Nourrissat, NOVARTIS, 12 route de Crossay, Senneville, 28700 Francourville, France R. Ortiz, International Crops Research Institute for the Semi-Arid Tropics (ICRISAT), Patancheru, 502 324 Andhra Pradesh, India S. Paillard, INRA-UPS-INA.PG, Station de Gntique Vgtale, Ferme du Moulon, 91190 Gif-sur-Yvette, France A. Panouill, INRA, Unit Exprimentale du Mas, 40390 Saint-Martin-de-Hinx, France J.L. Pham, Genetic Resources Center, International Rice Research Institute, PO Box 933, 1099 Manila, Philippines F. Pinton, Universit Paris-10 Nanterre, Ladyss, Bat. K, 92001 Nanterre cedex, France M. Pollacsek, INRA, Unit dAmlioration des Plantes, Domaine de Crouelle, 63039 Clermont-Ferrand cedex, France L.M. Pollak, USDA-ARS Corn Insects and Crop Genetics Research Unit, Department of Agronomy, Iowa State University, Ames, IA 50011, USA M. Raillard, CARGILL Gntique Europe, Croix de Pardies, 40300 Peyrehorade, France P. Ramrez Vallejo, Colegio de Postgraduados, IREGEP, Texcoco, Mxico 56230, Mxico C. Rebourg, INRA-UPS-INA.PG, Station de Gntique Vgtale, Ferme du Moulon, 91190 Gif-sur-Yvette, France P. Rimieri, INRA, Station dAmlioration des Plantes Fourragres, 86600 Lusignan, France R.A. Robinson, Retired Crop Scientist, 445 Prevost Lane, Fergus, Ontario, Canada N1M 2N3 F. Romero Rosales, Colegio de Postgraduados, IFIT, Texcoco, Mxico 56230, Mxico T. Ronsin, NICKERSON Semences, Route de Longu, 49160 Longu-Jumelles, France R. Salazar, SEARICE, 83 Madasalin St., Sikatuna Village, Quezon City, Philippines W. Salhuana, Pioneer Hi-Bred International, Inc., 6204 SW 16th Ct, Miami, FL 33183, USA J.P. Sampoux, RUSTICA Prograin Gntique, Domaine de Sandreau, Mondonville, 31700 Blagnac, France J. Sanou, INERA/CRRA de Farako B, 01BP910, Bobo-Dioulasso 01, Burkina Faso A. Sarker, Bangladesh Agricultural Research Institute, Joydebpur, Dhaka, Bangladesh

Contributors

xiii

G. Second, Institut de Recherche pour le Dveloppement (IRD, ex-ORSTOM), 911 Avenue Agropoplis, BP 5045, 34032 Montpellier cedex 1, France S. Second, Institut de Recherche pour le Dveloppement (IRD), 911 avenue Agropolis, BP 5045-34032 Montpellier cedex 1, France D. Segonds, RAGT, Avenue St Pierre, BP 31, 12033 Rodez cedex 5, France B. Sharma, Genetics Division, Indian Agricultural Research Institute, New Delhi 110012, India S.L. Sharrock, International Network for the Improvement of Banana and Plantain, Parc Scientique Agropolis II, 34397 Montpellier cedex 5, France A.S. Shreidi, Agircultural Research Center (ARC), PO Box 2480, Tripoli, Libya N.W. Simmonds, 9 McLaren Road, Edinburgh EH9 2BN, UK M. Smith, Department of Plant Breeding, Cornell University, Ithaca, New York, USA L. Sperling, Systemwide Program for Participatory Research and Gender Analysis for Technology Development and Institutional Innovation (PRGA), International Center for Tropical Agriculture (CIAT), c/o Frankenstraat 26, 2582SL The Hague, The Netherlands C. Spillane, Institute for Plant Biology, Zurich University, Zollikerstrasse 107, CH-8008 Zrich, Switzerland W. Spoor, Crop Science Department, Scottish Agricultural College (SAC), West Mains Road, Edinburgh EH9 3JG, UK C. Tabel, RAGT, Avenue St Pierre, BP 31, 12033 Rodez cedex 5, France S.P. Tallury, Department of Crop Science, North Carolina State University, Raleigh, NC 27695, USA B.R. Trognitz, International Potato Center (CIP), PO Box 1558, Lima, Peru M. Tufail, Ayub Agricultural Research Institute, Faisalabad, Pakistan M.C. Tyagi, Genetics Division, Indian Agricultural Research Institute, New Delhi 110012, India M. Vetelinen, Nordic Gene Bank, PO Box 41, S-230 53 Alnarp, Sweden P. Vincourt, RUSTICA Prograin Gntique, Domaine de Sandreau, Mondonville, 31700 Blagnac, France R. von Bothmer, Department of Plant Breeding Research, The Swedish University of Agricultural Sciences, SE-268 31 Svalv, Sweden R. Waugh, Scottish Crop Research Institute (SCRI), Invergowrie, Dundee DD2 5DA, UK E. Weltzien, Genetic Resources and Enhancement Program, International Centre for Research in the Semi Arid Tropics (ICRISAT), Bamako, Mali J.R. Witcombe, Centre for Arid Zone Studies, University of Wales, Bangor, Gwynedd LL57 2UW, UK A. Yahyaoui, Ministre de lAgriculture, Institution de la Recherche et de lEnseignement Suprieure Agricole (IRESA), 30 rue Alain Savary, 1002 Tunis, Tunisia D. Ziegler, Federal Centre for Breeding Research on Cultivated Plants (BAZ), Gene Bank, Braunschweig, Germany

Foreword

The worlds political leaders, meeting in Rome at the 1996 World Food Summit, made a public commitment to halve the number of under-nourished people by 2015. The World Summit Plan of Action sets out a component of this task: to pursue, through participatory means, sustainable, intensied and diversied food production. The better use of plant genetic diversity and resources will be a prerequisite to meeting this challenge. Greater use of plant genetic diversity will be required in order to produce varieties adapted to the extreme and highly variable environments of low-productivity or marginal areas. The need to combine sustainable productivity increases with mounting pressure to reduce the use of agrochemicals, and make more efcient use of water and nutrients, is likely to place even greater reliance on the utilization of a wider range of plant genetic resources in high-productivity areas. The importance of improving conservation and use of the genetic diversity of useful plants has been recognized in a variety of ways over the last decade. The Convention on Biological Diversity reected a strengthened global recognition of the importance of maintaining biodiversity for development. The Global Plan of Action for the Conservation and Sustainable Use of Plant Genetic Resources for Food and Agriculture, which was adopted by 150 countries at the International Technical Conference on Plant Genetic Resources (held in Leipzig in June 1996), identied a series of specic ways in which conservation and use of plant genetic diversity could be improved. The Conference of Parties to the Convention on Biological Diversity in November, 1996, highlighted key elements of the Global Plan of Action. These are: broadening the genetic base of major crops; increasing the range of genetic diversity available to farmers; strengthening the capacity to develop new crops and varieties that are specically adapted to local environments; exploring and promoting the use of under-utilized crops; and strategies that deploy genetic diversity to reduce crop vulnerability. This emphasis on the wise use of genetic resources, and the contribution that this can make to socioeconomic development and food security, is reected in the strategy and programmes of the Food and Agriculture Organization and the International Plant Genetic Resources Institute.
xv

xvi

Foreword

The very limited amounts of plant genetic variation that are present in modern varieties of some crops has been identied as an area of major concern by a number of international and national bodies and expert commentators. Substantial losses in production have resulted from the narrow genetic base present in some crops, the loss of many millions of tons of production of maize in the USA in the early 1970s being one of the most famous examples. Increasingly, production tends to depend on fewer varieties drawn from a narrower genetic base. Additionally, farmers choice of varieties to plant is often limited by poor access to genetic resources, particularly in developing countries. Concerned with the limited use of the available genetic diversity, the FAO Global Plan of Action identied broadening the genetic base of crops as a key element of the activities that should support improved use of plant genetic resources. The plan is concerned to see genetic vulnerability reduced and increased amounts of diversity made available to farmers. It recognizes that ways need to be found whereby increased amounts of useful genetic diversity can be brought into breeding programmes and into crop production systems. The ways in which this might be done are many and varied, and this book examines the rationale and need for activities to broaden the genetic base of crop production, and examines past experience and discusses new methodologies. We hope that it will stimulate more research and practical activities in this area. Mahmud Duwayri Director, Plant Production and Protection Division Food and Agriculture Organization of the United Nations Rome Geoffrey Hawtin Director General, International Plant Genetic Resources Institute Rome

Preface

Plant breeding and the production of new cultivars is widely regarded as underpinning agriculture and the development of society. Yet crop failures and risks associated with genetic uniformity on large cultivated areas, yield stagnation (below potentially attainable levels), and persistent failures to achieve sustainable production increases in important local ecologies are widespread problems. What is frequently not recognized is that continuing success requires a long-term and sustainable commitment to the effective utilization of plant genetic resources by enhancing and expanding the genetic base from which future cultivars will be generated. This is not a question of more collecting, storage and evaluation we have more than 6 million accessions in the worlds genebanks. Nor is it enough to introduce selectively a few genes. Indeed, the future needs for an enhanced base cannot be predicted and the particular needs of farming systems should not be prejudged. Because of this, and of the short term interests of many involved in current national and international breeding programmes, there is a clear need for an international commitment by FAO and others to take a lead in championing base broadening of the worlds food and bre crops. This commitment should take many forms: initiating specic crop programmes; promoting awareness; and supporting and linking local, national and regional activities. It should be aimed at ensuring that farmers and breeders have increased access to useable diversity. The commitment should be an essential component of an ecologically sound approach to agricultural production based on the more effective use of genetic diversity, employing, as appropriate, decentralized plant breeding driven by the specic needs of farmers and consumers. Strengthening the participation of farmers communities is vital to achieving these goals.

The above statement was prepared by a group of experts meeting in Rome, September 1997, to examine ways to further the objectives of Activity 10 of the FAO Global Plan of Action for the Conservation and Sustainable Use of Plant Genetic Resources for Food and Agriculture: Increasing genetic enhancement and base-broadening efforts. The full statement, with recommendations, and a list of participants, is provided on p. 437.
xvii

xviii

Preface

Following that meeting, it was decided to produce the present book, to bring together, in one volume, a number of papers on various approaches which contribute to broadening the genetic base of crops. By broadening the genetic base of crops we mean, in particular, increasing the amount of genetic diversity used to produce new crop varieties or used in agricultural production. This reects a continuing concern with the fact that much crop production depends on very limited amounts of genetic variation and that, in many cases, the genetic diversity present in a crop is not easily available to the plant breeder or the producer interested in making use of it. The book has a similar title to one published 20 years ago (A.M. van Harten and A.C. Zeven (1979), Broadening the Genetic Base of Crops. Proceedings of the conference, PUDOC, Wageningen, The Netherlands). The earlier book, and the conference on which it was based, emphasized collecting and evaluating germplasm, and methods for overcoming barriers to the use of wild relatives. The present book focuses more on the ways in which the amounts of genetic diversity in production can be increased and the ways in which breeding programmes can make use of greater amounts of genetic variation. A number of papers describe incorporation work (sensu Simmonds, Biological Reviews 66, 189241, 1991), that is, the large-scale development of locally adapted populations from unimproved or exotic sources, through long-term population-based approaches. It also reects the many developments that have taken place over this period in participatory plant breeding and in situ conservation of genetic resources. This book could not have been prepared without the help of the many individuals who generously contributed their time, energy and expertise to its writing and production. Above all, the authors of the various chapters are thanked for their contributions as well as for their patience and forbearance in adapting to changing deadlines. Special acknowledgement is due to Norman Simmonds, whose seminal paper on Introgression and incorporation: strategies for the use of genetic resources (Biological Reviews (1993) 68, 539562) stimulated the development of a specic activity on the issue in the Global Plan of Action. In the Food and Agriculture Organization (FAO), particular thanks go to Mahmud Duwayri, Niek van der Graaf, Helena Gomez, Erik Kueneman, Tony Putter, Marcio Porto and Peter Kenmore, for encouraging and guiding the initiative. Nina Dudnik worked tirelessly in keeping the editors more organized than they otherwise would have been. Finally, thanks go to Paul Stapleton of the International Plant Genetic Resources Institue, Shalini Dewan of FAO and Tim Hardwick of CABI Publishing for their cooperation in publishing this work.

Acronyms and Abbreviations

ACMV ADG AFLP APIC ARI ASTA BAZ BCMV BRG BTI BUCAP CBD CENARGEN CGIAR CIAL CIAT CILSS CIMMYT CIP CIRAD cms CORAF CORPOICA

African cassava mosaic virus Andigena group amplied fragment length polymorphism Association of Potato Intergenebank Collaborators advanced research institute American Seed Trade Association Bundesanstalt fr Zchtungsforschung an Kulturpanzen (Federal Centre for Breeding Research on Cultivated Plants), Germany bean common mosaic virus Bureau des Ressources Gntiques, France Boyce Thompson Institute Biodiversity Use and Conservation in Asia Programme Convention on Biological Diversity Centro Nacionalde de Recursos Genticos e Biotechnologia (of EMBRAPA), Brazil Consultative Group on International Agricultural Research local committee for agricultural research Centro Internacional de Agricultura Tropical Comit Permanent Inter-Etats de Lutte contre la Scheresse dans le Sahel Centro Internacional de Mejoramiento de Maz y Trigo, Mexico (International Wheat and Maize Improvement Centre) Centro Internacional de la Papa, Peru (International Potato Center) Centre de Coopration Internationale en Recherches Agronomiques pour le Dveloppement, France cytoplasmic male sterility Confrence des responsables de recherche agronomique africains Corporacin Colombiana para la Investigactin en Agricultura, Colombia
xix

xx

Acronyms and Abbreviations

CRBP CRSP DGER DM DUS EBN EMBRAPA ESD FAO FFS FHIA GCA g-cms GE GEM GILB HOPE HR HYE HYV IAEA IARC IBPGR ICA ICARDA ICRISAT ICTA IDRC IIRB IITA INIBAP INIFAP INRA INSAH INTA IPGRI IPR IRAT IRRI ISNAR ISO ISWYN ITC IWIS KUL

Centre des Recherches Rgionales sur Bananiers et Plantains, Cameroon collaborative research support programmes Direction Gnrale de lEnseignement et de la Recherche, France dynamic management distinctiveness, uniformity and stability endosperm balance number Empresa Brasiliera de Pesquisa Agropecuaria equal seed descent Food and Agriculture Organization of the United Nations farmer eld school Fundacin Hondurea de Investigacin Agrcola general combining ability geniccytoplasmic male sterility genotype environment Genetic Enhancement of Maize Project Global Initiative on Late Blight hierarchical, open-ended population enrichment horizontal resistance high-yielding environments high-yielding variety International Atomic Energy Agency International Agricultural Research Centre International Board for Plant Genetic Resources Instituto Colombiano Agropecuario International Center for Agricultural Research in Dry Areas International Crops Research Institute for the Semi-Arid Tropics Imperial College of Tropical Agriculture International Development Research Centre International Institute of Sugarbeet Research, Belgium International Institute for Tropical Agriculture International Network for the Improvement of Banana and Plantain Instituto Nacional de Investigaciones Forestales y Agropecuarias Institut National de la Recherche Agronomique, France Institute for the Sahel Instituto Nacional de Tecnologa Agropecuaria, Argentina International Plant Genetic Resources Institute intellectual property rights Institut de Recherches Agronomiques Tropicales et de Cultures Vivrires, France International Rice Research Institute International Service for National Agricultural Research International Standards Organization International Spring Yield Wheat Nursery INIBAP Transit Centre International Wheat Information System Katholieke Universiteit Leuven, Belgium

Acronyms and Abbreviations

xxi

LAMP LSC LYE MBP MGIS NARS NCRPIS NGO NSSL NTB OFM OPV PBR PCIM PGRFA PLRV PPB PPRI PRACIPA PRAPACE PRECODEPA PROCIPA PVP PVS QDPI QTL QUT RAPD R&D RFLP RRS RYD SCA SCRI SEARICE SI SIFT SMBR SSR TBG TBRI TRIPs TSG UNALM UPOV UPWARD

Latin American Maize Program Lancaster Sure Crop low-yielding environments modied bulk population Musa Germplasm Information System National Agricultural Research System North Central Regional Plant Introduction Station, USA non-governmental organization National Seed Storage Laboratory, USA Neotuberosum on-farm management open-pollinated variety plant breeders rights Programa Cooperativo de Investigaciones en Maz plant genetic resources for food and agriculture potato leaf roll virus participatory plant breeding Plant Protection Research Institute Programa Andino Cooperativo de lnvestigacin en Papa Programme Rgional de lAmlioration de la Culture de la Pomme de Terre et de la Patate Douce Programa Regional Cooperativo de Papa Programa Cooperativo de Investigaciones en Papa plant variety protection participatory varietal selection Queensland Department of Primary Industry, Australia quantitative trait loci Queensland University of Technology, Australia random amplied polymorphic DNA research and development restriction fragment length polymorphism reciprocal recurrent selection Reid Yellow Dent specic combining ability Scottish Crop Research Institute, UK South East Asia Regional Institute for Community Education self-incompatible standard international eld trial Sierra de Manantlan Biosphere Reserve simple sequence repeats tuberosum group Taiwan Banana Research Institute Trade Related Intellectual Property Rights technical steering group Universidad Nacional AgrariaLa Molina, Peru International Union for the Protection of New Varieties of Plants Users Perspective with Agricultural Research and Development

xxii

Acronyms and Abbreviations

USDA VCU VR WBN WISBEN

United States Department of Agriculture value for cultivation and use vertical resistance World Beta Network West Indies Sugarcane Breeding and Evaluation Network

1
1Food

Broadening the Genetic Base of Crops: an Overview

H.D. COOPER1, C. SPILLANE2 AND T. HODGKIN3
and Agriculture Organization of the United Nations. Present address: Secretariat of the Convention on Biological Diversity, World Trade Centre, Suite 300, 393 St Jacques, Montreal, Quebec, Canada H2Y 1N9; 2Institute for Plant Biology, Zurich University, Zollikerstrasse 107, CH-8008 Zrich, Switzerland; 3International Plant Genetic Resources Institute, Via delle Sette Chiese 142, 00145 Rome, Italy

Introduction
Crop improvement through plant breeding, like crop evolution in general, occurs through selection operating on genetic variability. Selection by plant breeders or by farmers can be intense and has resulted in major improvements in our crops. However, continued success in plant breeding can only be realized in so far as new variability is available for selection. Such variability provides adaptability, that is the capacity for genetic change in response to selection. Genetic diversity is therefore essential for crop improvement. In many crops the amount of variability that is available for selection is rather limited. A number of crops have an intrinsically narrow genetic base caused by bottlenecks at domestication, by migration or by epistatic and disease effects. Others have a narrow genetic base arising from a lack of diversity in previous breeding practices. Plant breeders tend to use favoured existing cultivars as the basis for generating new ones, because they need to generate short-term gains, and because, often, they are forced to meet stringent market requirements for uniformity. In fact, there is a tendency to give more attention to adaptation through selection than to the generation of new variability or the maintenance of adaptability (Simmonds, 1962). This represents opportunities lost in terms of the utilization of genetic resources, and in its worst manifestations can lead to a progressively narrower genetic base, slower progress (genetic gain) in plant breeding and increased risk of crop vulnerability. Clearly visible consequences of this have been limited to relatively few calamitous events such as the Irish potato famine, caused by late blight, and the southern leaf blight epidemic in the US maize crop in 1970, which was the result of cytoplasmic uniformity. The latter event led to a major report by the US Academy of Sciences on the Genetic Vulnerability of Major Crops (National Research Council, 1972) which recommended a broadening of the genetic base of major staple crops. However, to date there have been few concerted genetic base-broadening efforts. One reason for this may be the perception that, even in crops with a narrow genetic base, there remains
IPGRI/FAO 2001. Broadening the Genetic Base of Crop Production (eds H.D. Cooper, C. Spillane and T. Hodgkin)

H.D. Cooper et al.

sufcient genetic variability within breeding populations to continue making improvements (Duvick, 1984; Rasmusson and Phillips, 1997). Since the 1970s, work on the conservation of crop genetic resources has increasingly become a large-scale independent activity detached from crop improvement efforts. A substantial germplasm collecting effort was launched in the 1970s in response to concerns about genetic erosion and crop vulnerability. Over 1000 genebanks have been established, holding about 6 million accessions (FAO, 1998). Much has been invested in characterization, evaluation and documentation, yet, for various reasons, the use of many collections has actually been rather limited (FAO, 1998). This situation has led to criticism by some, including both plant breeders (e.g. Simmonds, 1993) and non-governmental organizations (e.g. Cooper et al., 1992), that the genetic resources maintained in genebanks could be more usefully deployed in both plant breeding and farmers elds. In more industrial agricultural systems, the separation between genetic resources conservation and crop improvement is also reected at eld level. Most modern cropping systems do not contain substantial crop diversity at eld level, but rather rely on diversity in reserve (Duvick, 1984). More traditional farming systems, on the other hand, often possess considerable diversity in production systems both within individual crops and through use of a wider variety of crops. Indeed the within-crop variability, acted upon by farmer selection (conscious and unconscious), was the basis for crop evolution from the beginnings of agriculture until the advent of modern plant breeding. On-farm crop improvement remains essential to many farmers today, particularly in more economically marginal areas. In recent years there has been growing awareness of the role played by farmers, especially women, in crop improvement and genetic resources management, and a renewed emphasis on strengthening the links between the conservation of genetic resources and their use. These trends are reected in the Global Plan of Action for the conservation and sustainable use of plant genetic resources for food and agriculture (PGRFA) adopted at the International Technical Conference on Plant Genetic Resources held in 1996 (FAO, 1996). The Plan includes, among its priority activities, increasing genetic enhancement and base-broadening efforts, promoting sustainable agriculture through diversication of crop production and broader diversity in crops and supporting on farm management and improvement of PGRFA. The importance of maintaining diversity in crop production systems was highlighted by the Conference of Parties of the Convention on Biological Diversity, when it considered agricultural biological diversity in 1996. Countries agreed to develop national strategies, programmes and plans, which should focus on, inter alia, the key elements of the Global Plan of Action, such as broadening the genetic base of major crops; increasing the range of genetic diversity available to farmers; strengthening the capacity to develop new crops and varieties that are specically adapted to local environments; exploring and promoting the use of underutilized crops; and deploying genetic diversity to reduce crop vulnerability (UNEP, 1997). Thus, there is a perception and concern that genetic diversity is limited, both within production systems and in breeding programmes, and that there is a need for continuing and concerted efforts to broaden the genetic base of crops. In the wide sense used here, broadening the genetic base of crops can be considered at three levels:

Overview

1. Increasing the extent of useful diversity available to breeders, that is broadening the genetic base of breeders material through genetic enhancement, or pre-breeding. 2. Increasing the range of useful diversity available to farmers as planting material. 3. Increasing the diversity of crops and varieties grown in the eld. In this book we have brought together papers on all of three of these areas. The book seeks to present and explore the main concepts related to broadening the genetic base of crops (see, in particular, Part I), and discuss the extent to which some recent trends (including the greater use of participatory plant breeding and on-farm genetic resources conservation) can contribute to these base-broadening goals. This chapter introduces the topic, provides an overview of the main concepts and surveys some examples, drawing upon the various chapters of the book, as well as the results of meetings on the topic held in 1997 and 1999 (FAO, 1999, 2000).

The Rationale for Broadening the Genetic Base of Crops

Variability, adaptability and the genetic base Adaptability is the capacity for genetic response to selection that results in adaptation (Simmonds, 1962). There is an inherent conict between adaptation and adaptability, since the more completely a population or variety is adapted to its environment the less adaptability it is likely to have for evolutionary change. Adaptability is essential for continued crop improvement through selection by farmers or in plant breeding programmes, but tends to be reduced by such selection. There is a continuing need to balance this tendency. Adaptability is dependent on the presence of genetic diversity within a population or crop. The extent of genetic diversity can vary substantially depending on the crop or population under consideration and not all of it contributes to adaptability. None the less, variation is an essential prerequisite of adaptability. Genetic diversity in crop plant populations does not occur at random but is structured in ways that depend on crop biology, distribution and the environment in which the crop occurs. Levels of diversity between and within plants can vary dramatically from virtual absence in a pure variety of modern rice to extremely high levels in populations of outbreeding forest trees. From the perspective of the user, whether a farmer or a breeder in the formal sector, the genetic base of a crop includes all the genetic diversity that is readily available to them for adaptation to any particular environment(s) of interest. It includes the genetic diversity of varieties, both traditional and modern, already in use in any one production environment, as well as the materials currently being used to develop new varieties for that environment. Less readily available material exists in different environments (in situ) and in genebanks (ex situ) and wild genepools. This exotic, unadapted material is generally not considered part of the genetic base, but can be brought into it through genetic enhancement or base-broadening programmes. The genetic diversity of locally adapted varieties includes variation within and between varieties as well as variation that is temporarily concealed as recessives in heterozygotes. It is supplemented by mutation (in the short term, to a low extent) and by

H.D. Cooper et al.

introduction of material from other populations. Recombination and selection allow new combinations of this variability to be exposed. In many traditional agricultural systems such variability allows crop evolution to continue, on-farm. Chapter 4 by Berthaud et al. provides examples of this. In traditional varieties of outcrossing annual crops such as maize (for example in Mesoamerica) and millet (West Africa), very high levels of variation are present within varieties (Simmonds, 1962). There is substantial variability between genotypes, which are themselves highly heterozygous. The recombination that results from outcrossing allows continued adaptation, through selection, at the level of the population, such that gene frequencies continually change, but gene combinations do not become xed. Within this diverse genetic background, farmers maintain a number of varieties that are phenotypically distinct for characteristics of importance to them (such as colour and end-product quality). While these characteristics are maintained by selection, there may be an enormous ux of other genetic characteristics between and within these varieties. As shown by Louette and colleagues (Louette, 1994; Louette et al., 1997) and by Berthaud et al. (Chapter 4, this volume), introgression of new genetic material can occur continually as a result of cross-pollination with material from neighbouring farms, and material brought in from other areas. Recombination of locally adapted material, together with introgression of material from other areas, continuously replenishes the available variability, and may help to balance ongoing selection by nature and farmer. Similar situations are observed for outbreeding perennials that are propagated by seed (such as coconuts and oil palm). Even crops which are usually clonally propagated, such as potatoes (in the Andes), bananas (in Uganda and Southeast Asia), sugar cane (in New Guinea) and cassava (in South America), can show high heterogeneity between clones. As shown by Second and Iglesias (Chapter 11, this volume), in some traditional cropping systems new cassava genotypes may be obtained when the latter are produced from seed. In inbreeding annuals such as rice (in Asia), wheat and barley (in the Mediterranean and the near East), heterozygosity is low, but in traditional agricultural systems there is substantial variability both within and between varieties. Though the extent of outcrossing is low compared with crops such as maize and millet, it is large enough to allow continuous recombination and crop improvement through selection over time. Such processes have allowed a large number of local varieties to evolve, as in the case of rice, where there are estimated to be over 100,000 known varieties (FAO, 1998). Outcrossing can also be signicant over short time periods as demonstrated in the work on dynamic management of wheat populations by Goldringer et al. (Chapter 13, this volume). Despite the presence of locally adapted variability on-farm, there may, nevertheless, be a need for inuxes of new material. This will be the case especially in areas outside the centres of diversity, and where the market imposes demands for new traits. Increasing the range of genetic material available to farmers can often be important even in areas where diversity levels are already high. In modern, industrial agricultural systems, locally adapted variability is very much reduced. The diverse varieties of outbred annual crops have been largely replaced by uniform F1 hybrids, as is the case with maize. Though substantial variability can still be present, as heterozygotes, in practice there is a major disincentive to its use by the

Overview

farmer, since second generation crops would give reduced yields and insufcient uniformity. Single clones of crops such as potatoes, banana and rubber are often grown on a large scale in agribusiness-oriented agriculture. Similarly, many local varieties of crops such as wheat, barley and rice have been replaced, to a large extent, by uniform stands of single lines which have low adaptability (Simmonds, 1962). The new crop varieties are highly adapted to their production environments (which, in parallel, have often also been improved through fertilizers and irrigation), and are highly productive. They often show broad adaptation to a very wide geographical range. However, the low levels of locally available variability have two major implications for plant breeding and genetic resources management. Firstly, crop evolution is separated from production in modern agricultural systems it no longer takes place on-farm, but in breeding programmes. To respond to changing needs and circumstances (such as changes in market requirements, or the emergence of a new or newly resistant pest or disease), farmers in such systems are dependent on a stream of new varieties produced in plant breeding programmes. This has been described as provision of diversity in time by Duvick (1984). This phenomenon is of course very advanced and widely accepted in many parts of the world, where it represents a rational and economically efcient division of labour. But it has not yet taken place in many marginal areas. In fact, in many such areas the incentives, capacity and infrastructure required to produce and distribute new varieties are not present. The loss of variability on-farm also reduces population adaptation as a factor in production (that is, the positive interactions between the components of varietal mixtures (Simmonds, 1962)), and might raise problems of crop variety vulnerability in the face of attacks by new pathogen races (Smale, 1997). Secondly, crop breeders tend to rely increasingly on a narrow set of improved elite materials for future plant breeding. Broadening the base of these materials through genetic enhancement becomes an increasingly important and increasingly difcult proposition. For a variety of reasons, including biological and economic constraints, there is a tendency to give this low priority. Additionally, as industrial agriculture expands, there are fewer reserves of crop genetic variability in situ, and an increasing need, therefore, to rely on ex situ collections.

Base-broadening and breeding programmes Most modern varieties, and the populations with which breeders work, consist of elite germplasm which has been carefully built up over periods of perhaps 1050 years of research, hybridization and selection. These outstanding genotypes are destroyed by crosses with unimproved exotic germplasm (Duvick, 1984). Moreover, breeders usually nd that, even within a narrow genetic base, improvements can continue to be made from further crossing and selection (Wych and Rasmusson, 1983). Thus, there is a major disincentive for breeders to access other germplasm sources that appear to be agronomically inferior. Unless outside sources of germplasm are considered absolutely necessary and possess specic desirable traits, breeders will usually try to nd new genetic material within their own breeding lines. Over the long term, this can lead to a dependence on a narrower and narrower elite germplasm base for crop improvement.

H.D. Cooper et al.

The variability of the available material that is likely to be well adapted to any one specic production environment is small in comparison with the variability of any crop as a whole. As pointed out by Simmonds (1962), it seems inconceivable that all of the potentially useful combinations have been assembled in any single group of locally adapted stocks (at least in areas outside the centres of diversity). Exotic materials, even crop wild relatives, often contain potentially useful genes and gene combinations, many of which are not expressed as obvious phenotypes in their original genetic background. Moreover, the immediate performance of exotic material itself is often poor. The problem is that mainstream breeding methods cannot accommodate, at reasonable cost, the numbers of combinations that would need to be examined (Simmonds, 1962). Instead, therefore, it has been proposed that dedicated programmes of prebreeding or genetic enhancement are required. While pre-breeding programmes of various types have previously been a well-established component of some publicly funded agricultural research programmes, currently there is concern that the support given to such programmes has declined in recent years. Pre-breeding programmes take different forms depending on the crop and the issues of concern, and include the introduction of new characters from wild relatives or the development of specic selections from which desirable improved inbreds can be obtained. Panmictic populations established on a very broad base, such as the barley composite cross or populations (Harlan and Martini, 1929, 1938), can also constitute pre-breeding programmes. Such mass reservoirs (Simmonds, 1962) of genetic adaptability can also carry out a valuable conservation function (Shands, 1990; Goldringer et al., Chapter 13, this volume). Simmonds (1993) described approaches that involve the large-scale development of locally adapted populations from unimproved or exotic sources, through long-term population-based approaches, as incorporation, distinguishing them from the introgression of single genes (typically for qualitative traits such as disease resistance) into breeders materials. In contrast to the periodic introgression of a few selected genes or gene complexes from exotic germplasm into elite germplasm, to meet specic pre-identied needs, incorporation approaches are based upon the construction of genetically diverse and dynamic plant populations, which through sequential cycles of genetic recombination and selection are used to develop locally adapted germplasm pools. Initially diversity is maximized by drawing upon material of wide genetic and geographic origins. Selection intensity is initially weak, though progressively increased as the populations are decentralized to the target environments and end-users. The process of selection is subject to so many constraints that it cannot possibly achieve perfection (Lewontin, 1984; Mayr, 1988; Eriksson et al., 1993). One objective of base-broadening is to facilitate the production of populations of increased adaptive tness. Wright (1922, 1977) combined migration, drift and selection into a unied process of evolution called the shifting balance theory, which views most populations as being substructured into small, periodically drifting groups where tness is due to many interacting loci. Wright envisaged a process where a population can be trapped at a certain tness level where only a limited number of genotypes are well adapted. However, evolutionary rearrangements of allele frequencies due to drift or other factors followed by selection can reassort the available genes into a new type more t than the original and which can reach a higher adaptive peak. Adaptive valleys, where the tness of genotypes is at a minimum, act as constraints to geneow and to reaching

Overview

higher adaptive peaks. While adapted germplasm may be at one type of adaptive peak, by denition corresponding exotic germplasm will be at a different type, with a major maladaptive valley preventing geneow between them (Whitlock et al., 1995). Basebroadening can be considered as a means of building ridges between different tness peaks (see also, Spillane and Gepts, Chapter 2, this volume; Goldringer et al., Chapter 13, this volume). Base-broadening programmes may also be important in maintaining or increasing the current levels of diversity and in avoiding further inadvertent narrowing through plant breeding programmes. Base-broadening programmes may be particularly important as a necessary complement to (or component of ) new crop improvement approaches (for example: micropropagation, F1 hybrid breeding, the possible introduction of apomixis in new crops and recombinant gene transfer) which, depending on the approach taken, may either contribute to broadening, or lead to further narrowing, of the genetic base. Pre-breeding is often considered to be an activity at the interface between germplasm conservation and utilization. As a result, which sector (private or public) and who (curators or breeders) should be responsible for this activity is not clear. Generally, it is considered a pre-competitive activity which commercial breeders cannot afford in the short term. Private incentives for such activities are few, and pre-breeding is usually regarded as an international or national public good. Indeed, most public research institutions, universities or funding agencies have done pre-breeding in the past, but with the increasing withdrawal of public funding, pre-breeding activities are now often left unfunded or underfunded (Duvick, 1984; Anderson et al., 1993). The scarcity of long-term research funds as well as the lack of academic recognition for such a long-term activity have contributed to the neglect of pre-breeding activities. This neglect has negative long-term implications for maintaining or increasing the use of germplasm stored in genebanks, and needs to be overcome. (Kannenburg and Falk, 1995; FAO, 1998).

Increasing the range of useful diversity available to farmers Farmers have three possible sources or channels of access for new crop varieties: (i) the private sector; (ii) government seed supply schemes; and (iii) informal farmer to farmer distribution of seed. Type (i) is involved in seed production for commercially viable markets and clients. Because resource-poor farmers express little nancial demand, types (ii) and (iii), on the other hand, provide the bulk of the staple crop varietal needs of resource-poor farmers in developing countries (FAO, 1998). When provided with the opportunity and means to do so, farmers often express a demand for access to a wider range of useful crops and varieties. Traditional farmers and those operating in more industrialized or market-oriented conditions alike are ready to test new varieties whether, for example, it is the latest ofcially released cultivar, or an interesting local variety observed during a visit to a relatives farm (Johnson, 1972; Grisley, 1994). Access to suitable new varieties at the level of the farm may in fact be considerably limited by a range of factors. These can include poor seed distribution systems (Cromwell and Wiggins, 1993), breakdown in informal seed supply networks (Bellon, personal communication), an inappropriate regulatory framework

H.D. Cooper et al.

(Louwaars, Chapter 5, this volume) and the lack of suitable varieties. This last reason may result from insufcient plant breeding capacity in the country concerned or the use of breeding strategies that do not result in varieties with the traits needed by farmers (Cromwell and Wiggins, 1993; Witcombe, Chapter 26, this volume). Because of these latter factors, improved varieties can often show low adoption rates. On a local level, it has been suggested that some existing local varieties, or groups of local varieties, may actually have quite narrow genetic bases as a result of a previous genetic bottleneck due to past events or widely uctuating environmental conditions (Wood and Lenn, 1996). This might be the case particularly where crops have been introduced into a region or locality where they did not previously exist, or where they have been subject to extreme biotic or abiotic stresses which have narrowed the genetic base in particular locations. Such limited genetic diversity in locally adapted materials can limit the response to selection that breeders (or farmers) may achieve. Often farmers do not have the opportunity to formulate and put forward their need for access to a wider range of planting materials (Spillane et al., 1999). Communities and farmers organizations have little choice but to accept the materials distributed by national organizations or by private companies. New plant breeding approaches, such as participatory variety selection and participatory plant breeding, seem likely to result in increased ranges of useful material for farmers (Witcombe, Chapter 26, this volume; Sperling et al., Chapter 27, this volume). Other imaginative approaches may be needed to ensure that farmers retain access to the full range of diversity that they have found useful in the past, or get access to a continuing supply of new useful materials.

Increasing the diversity of crops and varieties grown in the eld The diversity of crops and varieties grown in the eld can be important for many reasons. It can reduce crop vulnerability, that is the likelihood of crop failure due to pest and disease epidemics or unpredictable climatic effects. More generally, it can provide insurance to farmers against risk and uncertainty. This is especially important for farmers operating in economically marginal areas where there is little alternative provision of such livelihood insurance. Crop or varietal diversity in the eld can also contribute to the conservation of plant genetic resources. Most signicantly for the topic of this book, diversity in the eld allows crop evolution to continue, and this process can be accelerated through farmer participatory plant breeding, in those cases where farmers select from segregating populations or a mixture of a large number of selng lines. On-farm crop evolution is important in self- and cross-pollinated crops and in clonal crops (e.g. Berthaud et al., Chapter 4, this volume). The continuous and dynamic patterns of crop variety change and evolution seem to be a major feature of traditional crop production in many areas. In many industrial farming systems there is very little diversity on farm, at least at the level of a single farm or production unit. In these situations some of the functions of on-farm diversity are replaced by farm inputs such as pest control chemicals, and by alternative strategies for managing risk and uncertainty, such as crop insurance and alternative employment opportunities (Swanson et al., 1994). Nevertheless, in some cases, where large contiguous areas planted to single varieties or varieties with the same

Overview

resistance genes, crop vulnerability may be a concern (Smale, 1997). Additionally, as noted earlier in this chapter, in the more industrial farming systems there is a reliance on genetic diversity kept off-farm (in genebanks) and on temporal diversity the continuing production and introduction of new varieties. It is therefore vital that sufcient attention is provided to the other two aspects of broadening the genetic base, that is to increasing the diversity available to breeders through genetic enhancement, and ensuring that new varieties are actually available to farmers.

Identifying Priorities for Base-broadening

The needs and priorities for base-broadening will vary from crop to crop and also, for any one crop, from production area to production area. There are two main groups of criteria and indicators to determine or predict the likely need for base-broadening for particular crops in particular situations: ex ante and ex post. Some of these are explored in further detail, using an evolutionary perspective, by Spillane and Gepts, Chapter 2, this volume. The rst group of indicators and criteria refer to evidence of a bottleneck in the availability of genetic diversity historically during domestication or crop migration, or during modern plant breeding, or seed distribution:

During domestication, only a small proportion of the total diversity of the wild population is likely to be sampled (the founder effect: Ladizinsky, 1985). Subsequent geneow between the new cultivar and its wild progenitors might be restricted by breeding barriers, depending on the breeding system of the crop concerned, and nature of the domestication event. Crops which might be expected to have a narrow genetic base because of bottlenecks at domestication include rice, bread and durum wheat, Phaseolus beans, tomato, pigeon pea, chickpea, Citrus, and possibly Musa and yam (Spillane and Gepts, Chapter 2, this volume; Participants in the Rome Workshop, 1997, personal communication). Bottlenecks may have arisen through additional founder effects when crops were distributed between continents and the initial sample of material from which all material in the new location(s) were derived was very small. During transfer between latitudes, there may be a further narrowing of the genetic base because the population would not be well adapted to the new day-length conditions, and so only a small number of the genotypes would survive (epistatic effects, e.g. potato in temperate areas). Finally, there may be other chance occurrences that narrow the genetic base such as disease epidemics, which may decimate populations. Examples of crops with narrow genetic bases arising during migration may include soybean (in the USA), maize (in Africa and the US; see Tallury and Goodman, Chapter 9, this volume), sorghum and millet (in South Asia), lentil (in South Asia; see Erskine et al., Chapter 21, this volume), cocoa and coffee (Participants in the Rome Workshop, 1997, personal communication). Bottlenecks during modern plant breeding might be indicated, for example, by a small number of parents or high degree of relatedness between parents, as elucidated from pedigree analyses. The narrow genetic base of maize in the US is the classic example (Goodman, 1985, 1990; Tallury and Goodman, Chapter 9; Pollack and

10

H.D. Cooper et al.

Salhuana, Chapter 19, this volume). Another is rice in the US and Japan (Nakagahara et al., 1997; Xu et al., 1997). Further, narrowing of the genetic base may result from specialization within crops: for example, breeding of winter wheat has been mainly done by using only winter germplasm and breeding of spring wheat by using only spring germplasm (Spillane and Gepts, Chapter 2, this volume). The genetic base of hot peppers (Capsicum annum) is partitioned by the specialized requirements for distinct uses (e.g. thin pericarp required for drying, whilst fresh use requires small fruit). The genetic base of Brassica is similarly partitioned into different morphological types (Pickersgill in FAO, 2000). The diversity available in any one locality is inevitably a small fraction of the total available in the crop at large. Both formal and informal (farmer-to-farmer) seed distribution systems are likely to have a major impact in crop geneow within any country or region (Cromwell and Wiggins, 1993; Sperling et al., 1996; Sthapit et al., 1996; Witcombe et al., 1996). Factors such as harsh topography, lack of transport or communications can limit geneow between regions and thus farmers who live in peripheral regions are likely to have less access to novel germplasm (Brush and Meng, 1998). Genetic bottlenecks to such geneow can occur because of social factors such as income status, gender, family, ethnicity or other social groupings (Ferguson, 1992; Sperling et al., 1996). Both the diversity of material made available to farmers and its suitability to their production systems can be limited by inadequate plant breeding capacity, and inappropriate regulations for variety release and seed certication (Louwaars, Chapter 5, this volume).

The second group of indicators and criteria refer to actual evidence that performance of the crop is being limited by a narrow genetic base. These include, for example:

When a yield plateau or low rate of progress (genetic gain) is apparent in breeding programmes. This has been suggested to be the case for potatoes in northern temperate areas (Simmonds, 1993; Bradshaw in FAO, 2000), and for sugar cane in the Caribbean (Kennedy, Chapter 16, this volume). When breeders are increasingly forced to source qualitative genes from the secondary and tertiary genepools (often using costly high technology approaches). When there are repeated instances of crop failure, or crop vulnerability to particular biotic or abiotic stresses. Sugar cane in several islands of the West Indies provides an example of this situation (Rao and Gardiner, 1997). Concerns over the vulnerability of maize in the USA and Europe have led to base-broadening efforts (Pollack and Salhuana, Chapter 19, this volume; Gallais et al., Chapter 20, this volume). When the needs of farmers for a sufciently diverse range of planting material are not being met (this can include the needs of particular groups of farmers or sections of the community such as women, landless farmers, children or the old, and particular needs, such as that for labour-saving crops or crops required for particular products, including those of cultural importance).

Bottlenecks may be: (i) genetic that is, while the total genetic diversity of the species or genus may not be limiting, barriers to breeding occur between the primary genepool and other genepools. Additionally, barriers may arise due to differences in adaptation that have arisen through geographical bottlenecks; (ii) spatial or geographic that is, while the total genetic diversity worldwide may not be limiting, that available, or

Overview

11

adapted to, particular areas may be; and (iii) institutional or regulatory that is, that availability is limited by economic, legal or other regulatory factors, such as seed and variety release regulations, access laws or lack of purchasing power. Since the most appropriate way to overcome a suspected bottleneck may depend in part on its cause, the indicators and criteria outlined above may provide initial clues as to the most suitable base-broadening approach to use. Critical assessments of the state of diversity of crops, and of the state of use of diversity of those crops, would help to provide a more objective basis for determining future needs and priorities. Such assessments might include the following components:

Assessments of the genetic diversity present within the various genepools of the crop, comparing the in toto variation (that is the sum of the variation in the 1, 2 and 3 genepools) with that available in the primary genepool. Assessments of the genetic diversity present in the primary genepool of the crop, comparing that in situ (aggregating all areas), with that available (i) in collections, and (ii) in breeding pools. Examination of the extent that such available genetic diversity is actually utilized in breeding programmes of the crop, including pedigree analyses. Examination of the extent to which such genetic diversity used in breeding programmes is actually available to farmers in the full range of geographical areas and for the full range of purposes (including specic attention to past and ongoing efforts to produce improved varieties of use to farmers in both high-potential and marginal environments).

The chapters in Part II of this book provide information on some of these points: for millet in West Africa (Niangado, Chapter 8), for maize in the USA and sub-Saharan Africa (Tallury and Goodman, Chapter 9), for potato (Ortiz, Chapter 10), for cassava (Second and Iglesias, Chapter 11) and for Musa (Sharrock et al., Chapter 12).

Approaches to Broadening of the Genetic Base of Crops

Several approaches to base-broadening are discussed in this book. Many are strategies for genetic enhancement or pre-breeding with an emphasis on the incorporation approach, as noted earlier. Others also include elements of genetic resource conservation and plant breeding per se.

Incorporation programmes, involving the generation and maintenance of populations The main elements of a base-broadening programme involving incorporation (Simmonds, 1993; Spoor and Simmonds, Chapter 3, this volume), include:

Making use of the broadest possible starting materials, consistent with the specic objectives of the programme. Evaluation of such material is, in this context, often irrelevant, since most of the starting material is expected to be un-adapted to the target environments anyway. However, for reasons of feasibility, some genetic enhancement programmes start with a subset of material that is selected for likely

12

H.D. Cooper et al.

usefulness. Core collections may be useful tools to select genetically diverse starting material (Van Hintum et al., 1999). A need for extensive recombination. (If natural outcrossing rates are sufciently high, recombination is easy; otherwise controlled crossing or the use of male sterility genes may be necessary.) Weak and progressively decentralized selection. Whenever possible, selection should be based on multiple large populations and carried out over several generations in target environments, in order to exploit genotype by environment interaction; inbreeding should be minimized; simple and cheap methods should be used as far as is possible. (If adaptation is reected in production, as with seed grain crops, seminatural selection can be exploited as in the composite cross populations.) Long-term programmes of the order of a decade or more are required in order to generate usable material, though the duration depends on the specic objectives and crop concerned. The maintenance of the above process as a programme distinct from conventional breeding programmes, until usable material is produced. At this stage, material may be used directly in formal breeding programmes, crossed with existing adapted material or, potentially, made available for further selection in farmer participatory approaches.

Of course the actual methodologies have to be adapted to the biology of the crop, and to the specic objectives, scale of operation and time horizon of the programme. One way of grouping population-based approaches to base-broadening follows (see also Table 1.1, and Part III of this book). A: Synthetics and composite cross-populations Populations are established by crossing a number of genetically different accessions in order to maximize the importance of genetic and geographic origins. These populations are then split into the different locations where a need for new variability has been identied and each sub-population is allowed to adapt progressively to local conditions. Such base-broadening thus involves the evolutionary development over time of different locally adapted populations of genetic resources from an initially genetically diverse pool of unadapted exotic germplasm. Approaches of this type include the classical composite cross populations (Harlan and Martini, 1929, 1938; Ibrahim and Barrett, Chapter 15, this volume). Newer examples include the development of broad-based barley pools for Nordic conditions (Vetelinen and Nissil, Chapter 14, this volume). Simmonds (1962) considered such composite cross-populations as reconstituted local varieties or mass reservoirs of locally adapted variability, while Suneson (1969) considered that the changes in such populations constituted an evolutionary plant breeding method. Similar concepts have been used for sugarbeet (Frese et al., Chapter 17, this volume). Experiments are now being carried out to explore the use of such populations for the dynamic in situ management of genetic resources (Goldringer et al., Chapter 13, this volume). Dynamic management, as practised on composite wheat populations, is also proposed for cassava (Second and Iglesias, Chapter 11, this volume), mimicking the management of cassava cultivars in Amazonian villages. Such on-farm dynamic management is also documented for maize and millet (Berthaud et al., Chapter 4, this volume). In crops where the seed is the harvested product, the normal cycle of planting and harvest may provide sufcient selection to produce agronomically useful material.

Overview

Table 1.1. Base-broadening: examples of incorporation programmes involving population management. Crop and objective A: Composite crosses Barley (composite crosses, e.g. CCII) To broaden the local genetic base; to produce new varieties; to maintain longterm breeding reserve. Starting material 28 lines (15 adapted to N. America, 13 Old World) Selected for promising phenotypic characters from USDA collection. All 378 crosses made; polyallele crossing. 30 diverse varieties, crossed in pairs, followed by pairwise crosses of F1s, for 3 generations. Harvest bulked. Selection process Grown as unselected bulk for 30 generations at Davis, and distributed to barley breeders in various countries. Status and outputs 15 commercial varieties by 1956; more than 50 lines eventually developed into varieties, others used as parents for other varieties. References Harlan and Martini, 1929, 1938.

Barley (Cambridge CCV)

Grown as unselected bulk for 30 generations at Davis; 10,00015,000 seeds/season. Samples of F10, F20, F30 maintained at Cambridge, UK. Dynamic genepools Evaluation data

Ibrahim and Barrett, Chapter 15, this volume.

Barley (Nordic) To produce broad-based barley genepools, resistant to various diseases, adapted to Nordic conditions.

Unadapted material: 5 wild Not yet in progress. barley, 10 landraces from Asia. Adapted material from Sweden or Finland. Selected for disease resistance and wide diversity. Controlled crossing for 6 generations (pairwise and diallel), with selection avoided.

Vetelinen and Nissil, Chapter 14, this volume.

continued 13

Table 1.1. continued Crop and objective Wheat (dynamic management, France) To maintain genetic diversity for use in breeding programmes by mimicking natural processes responsible for its diversication and conservation. Cassava (EMBRAPA, Proposed) 1 To conserve diversity in centres of diversity. 2 To produce new, broadly based varieties in new areas mimicking traditional practices. Starting material Three populations: PA (16 parents, domestic sources); PB (16 parents, exotic sources); and PS (as PA, with male sterility). 3 years of bulk multiplication. Selection process Status and outputs References

14

The 3 populations distributed Diverse set of locally adapted Goldringer et al., over 7 to 12 locations in France. populations with mostly Chapter 13, this volume. No conscious selection; 2 levels agronomically favourable traits. of management: intensive farming; extensive farming. In latter stages geneow between locations to be introduced to create metapopulations. Second and Iglesias, Chapter 11, this volume. 1 Dynamic conservation of varieties. 2 Development of locally adapted varieties.

Planting at distances to allow balance between adaptation 1 In centre of origin: pool and geneow. Selection for local varieties in single eld. tuber quality and productivity. 2 In new areas: pool Also, in the case of (2), maximum diversity metapopulation established in a single eld. through periodic and limited exchange between material in a network of sites. Crosses with wild B. vulgaris germplasm, made in 1986, 1990 and 1994. In 1990: male sterile sugarbeet plants were chosen as the female parent to obtain F1 plants. F1s intercrossed. Bulked to produce F3 families. At least two recombination cycles were allowed without selection. No prior selection on useful characters was done. Mild selection on root shape and bolting resistance. After ve cycles of mass selection some of the progenies started to resemble the sugarbeet.

Sugarbeet (USDA/ARS) To overcome narrow genetic base.

Started in 1986. USDA germplasm releases.

Frese et al., Chapter 17, this volume.

H.D. Cooper et al.

Overview

Sugarbeet (European Breeding companies) To overcome narrow genetic base.

Beet x wild crosses, using 22 different French wild beet populations. 50:50 combination maintained in progeny using Dogget populations.

Subesequently selection for agronomic characters.

Started in 1996; selection for agronomic characters now under way.

Frese et al., Chapter 17, this volume.

B: Incorporation to overcome historical bottlenecks Potato (John Innes/Cornell Large number of samples etc.) from Andigena cultivars of To overcome bottlenecks S. America. caused by: (i) migration No evaluation. from Andes to northern temperate zones; and (ii) late blight epidemic. Potato (John Innes/North Carolina ) As above. Large number of samples from cultivated diploids of S. America. No evaluation. Wild diploids of S. America, crossed with dihaploids derived from TBR. Clones from upper Amazon.

Weak mass selection, over 10 to 20 years, under long days and disease attack, alternating between seedlings and tubergrown selections.

Neo-tuberosum: promising selections, especially in crosses with temperate tuberosum cultivars.

Simmonds, 1993; Spoor and Simmonds, Chapter 3, this volume.

As above.

Material to be crossed with temperate cultivars of tuberosum and neotuberosum groups.

Carroll and de Maine, 1989; Simmonds, 1993.

Potato (Sturgeon Bay) As above.

Material of 25% wild and Ortiz, Chapter 10, this 75% tuberosum germplasm. volume.

Cocoa (BAL/CDC) To overcome narrow genetic base of crop.

Test crosses in well-structured New clones and parents. mating designs. Weak mass selection, longterm, alternating between seedlings and clonal selections. Proposed programme. New plantation material.

Simmonds, 1993.

Rubber (RRI, Malaysia) Wild Amazonian material. To overcome the very narrow genetic base derived from very small original population derived from the centre of origin.

Simmonds, 1993; Simmonds 1989.

15

continued

Table 1.1. continued Crop and objective Oil palm (PORIM) To re-domesticate crop in Asia from W. African material. Sugar cane (WISCBS) Repetition, on large genetic base, of substantial part of the crop evolution. Starting material Collections from W. Africa. Selection process Selection in Asia. Status and outputs Proposed programme. New plantation material. References Simmonds, 1993.

16

Large range of Noble polycrosses, and over 50 S. spontaneum parents. Crossing at plant breeding station. No evaluation. Tropical races.

Weak selection, on large populations, alternating seed/clonal cycles. Multilocational selection throughout Caribbean.

New parents to cross with existing commercial parents (CM).

Spoor and Simmonds, Chapter 3, this volume; Kennedy, Chapter 16, this volume.

Sorghum (conversion) To adapt exotic sorghum to long days.

Introgression of long day length into tropical material (i.e. in reverse to normal direction).

Goodman, 1990.

C: Genetic enhancement to overcome effects of intensive breeding and historical bottlenecks Maize (NCSU) Superior tropical maizes: Selection for adaptation to To overcome extremely hybrids or inbred lines. temperate conditions, over narrow genetic base in the Identied on basis of 56 generations. Resulting USA caused by: evaluation. material (100% tropical) (i) daylength bottleneck Large diallel cross. selfed for 2 generations, and between tropical and US tested by top-cross with elite maize; and (ii) intensive temperate. hybrid breeding.

14 inbred lines suitable for breeding or production and 2 lines for breeding only.

Goodman, 1990, 1985; Simmonds, 1993; Tallury and Goodman, Chapter 9, this volume. H.D. Cooper et al.

Overview

Maize (GEM) As above, specically to improve traits for agronomic productivity, disease and insect resistance; and value-added characteristics such as starch, oil and protein grain composition and qualities. Private/public sector cooperation. Maize (INRA/PROMAIS) To overcome narrow genetic base in France, especially that caused by intensive hybrid breeding Private/public sector cooperation.

Originally 50 elite tropical and temperate LAMP accessions (identied by evaluation data), and 7 commercial tropical hybrids from DeKalb Genetics.

Selected populations crossed to proprietary inbred lines, under short days, by each private cooperator. Resulting material (50% tropical) selfed. Some also crossed to elite by another cooperator and topcrossed to test, or rst, crossed to elite to give 25% tropical.

Breeding crosses (50%, 25%) evaluated by testcrosses. Best used to develop breeding lines by cooperators. Access to this material is limited to GEM cooperators. Data collected are freely available. GEM-enhanced lines and synthetics will be freely available after their public release.

Tallury and Goodman, Chapter 9, this volume; Pollak and Salhuana, Chapter 19, this volume.

1236 accessions from Europe, Accessions grouped into Asia, and N. and S. America. pools for different uses. Mostly landraces (open pollinated varieties); some bred lines from USA. All evaluated (per se and testcross value) at 2 French locations.

Differentiated breeding pools Gallais et al., Chapter 20, for different uses. this volume. Evaluation and molecular data.

continued

17

18

Table 1.1. continued Crop and objective Maize (HOPE, Guelph) To overcome narrow genetic base in Canada caused by: (i) daylength bottleneck between tropical and temperate maize; and (ii) intensive hybrid breeding. Starting material Selection process Status and outputs References

Open ended: initially 43 entries Hierarchical: (adapted OPVs, synthetics, L level (all introductions): composites, and 4 doublestratied mass selection. cross hybrids of foreign H level (from L level, and origin). To date c. 1000 few introductions): S2 entries included. Worldwide progeny selection. in origin. Includes germplasm E level (from H level, in 2 with poor agronomic heterotic blocks): full-sib performance. Most short reciprocal recurrent season, suitable for Guelph. selection. Around 5% longer season exotics from elite improved sources (e.g. CIMMYT).

HOPE populations that vary Kannenberg, Chapter 18, considerably among this volume. themselves and represent markedly different germplasm from that deployed commercially. 2 Guelph inbreds, CG91 and CG96, released in 1997, contain HOPE germplasm.

BAL/CDC: BAL plantations Sdn Bhd/Commonwealth Development Corporation; CIMMYT: International Maize and Wheat Improvement Centre; EMBRAPA; Empresa Brasiliera de Pesquisa Agropecuaria; GEM: genetic enhanacement of maize; HOPE: hierarchical, open-ended population enrichment; INRA: Institut National de la Recherche Agronomique; LAMP: Latin American Maize Programme; NCSU: North Carolina State University; OPV: open-pollinated variety; PORIM: Palm Oil Research Institute of Malaysia; RRI: Rubber Research Institute; USDA: United States Department of Agriculture; WICSCBS: West Indies Central Sugar Cane Breeding Station.

H.D. Cooper et al.

Overview

19

A similar process the Multiple Population Breeding Strategy has been proposed for tree species (Namkoong et al., 1980). B: Incorporation programmes to overcome historical bottlenecks caused during crop migration and/or disease attack In these cases, diverse material from the centre of diversity of the crop is used as the starting material, and subjected to repeated cycles of recombination and mild selection in the target environment. Thus a stage of the evolution of the selected crop is repeated, in order to overcome genetic bottlenecks that took place during crop evolution through founder effects, epistatic effects or, for example, disease epidemics. The examples of such crop reconstruction approaches to date include potato, sugar cane, cocoa, rubber and oil palm (Simmonds, 1993; Spoor and Simmonds, Chapter 3, this volume; Kennedy, Chapter 16, this volume). In the case of potato, the programmes were aimed at overcoming bottlenecks of all three types mentioned. In all of these cases, as in those of category A on page 12, the starting materials are chosen mainly in order to maximize diversity. Evaluation of the starting material is otherwise not practised. A related case, in which an alternative approach is used for the same objective, is the conversion of tropical sorghum germplasm to long days through the introgression of genes for short day into tropical material such that it could enter the genepool of the temperate crop (Goodman, 1990). C: Genetic enhancement to overcome effects of intensive breeding, as well as historical bottlenecks As highlighted earlier, temperate commercial maize has a very restricted genetic base for various reasons. In response to this, Goodman (1985) used a base-broadening approach similar to that outlined under Category B tropical germplasm was crossed, and selected under temperate conditions for several generations. However, in this case the starting material was selected on the basis of productivity. This reects the exacting requirements for commercial maize production. Similarly, in the Genetic Enhancement of Maize (GEM) programme, material was selected on the basis of the evaluation data of the previous Latin American Maize Programme (LAMP) (Pollak and Salhuana, Chapter 19, this volume). Material in the French INRA/PROMAIS programme was also evaluated (Gallais et al., Chapter 20, this volume). The HOPE programme uses a hierarchical system to separate diverse introductions from elite material (Kannenberg, Chapter 18, this volume). Private sector support is provided in the GEM and PROMAIS programmes. As a consequence, not all of the enhanced products are freely available.

Other approaches to base-broadening There are various other approaches to broadening of the genetic base of crops. Examples are provided in Part IV of this book. Erskine et al. (Chapter 21, this volume), used several approaches to overcome an ancient bottleneck in lentil. Lentils in South Asia are of a specic ecotype and have a narrow genetic base. This was overcome rst by the introduction of pre-adapted germplasm from West Asia, and secondly by the production of new hybrids. Baudoin et al. (Chapter 23, this volume) used inter-specic

20

H.D. Cooper et al.

crosses to develop composite populations adapted to maize/bean mixtures. Farmers will practise further selection within these mixtures. Grando et al. (Chapter 22, this volume) highlight the value and importance of barley landraces as a genepool for breeding programmes. Selections from diverse landrace populations can reveal high-yielding genotypes, which can be used directly, crossed in breeding programmes, or used in reconstituted mixtures. Espinosa et al. (Chapter 25, this volume), and Trognitz et al. (Chapter 24, this volume) emphasize the importance of a wide genetic base in breeding for horizontal resistance to diseases. Finally, there are many approaches to increasing the diversity available to farmers through participatory varietal selection, participatory plant breeding and decentralized plant breeding (Ceccarelli et al., Chapter 6; Salazar, Chapter 7; Witcombe, Chapter 26; Sperling et al., Chapter 27, this volume). As discussed by Witcombe, in cases where selection is decentralized, with or without farmer participation, there is likely to be broadening of the genetic base on-farm.

Concluding Remarks
The approach chosen for a base-broadening programme will reect crop biology, the nature of the perceived problems, the interests and needs of the farmers and, not least, the interests of those involved in developing and managing the programme. Some work will be solely concerned with introducing useful diversity into a plant breeding programme; other work with ensuring that farmers have access to more diverse seed stocks. Neither approach is necessarily more important than the other, and the two are not mutually exclusive. The long-term effect of these different emphases and approaches is equally uncertain both in respect of their impact on production and in respect of their impact on available genetic diversity for crop production systems. At the present time, a number of new approaches offer advances in crop improvement. These include gene-transfer and the use of marker-assisted selection as tools for managing useful trait diversity and improving the efciency and scope of both conventional plant breeding and genetic engineering. An increasing range of techniques is available to facilitate wide crosses in order to access germplasm in the secondary and tertiary genepools. There is substantial investment in many of these approaches, although such investments are heavily biased towards the agricultural systems of developed countries and more export-oriented crops (Spillane, 1999). The various chapters in this book show that a range of approaches to crop improvement are available that are expected to generate reasonable results with minimal resources, provided that there is a sufciently long-term commitment. These approaches may not need expensive inputs from biotechnology or molecular genetics. They can often be carried out in ways that bring together farmers and plant breeders and link together the formal and informal sectors. Such approaches are able to make use of a wider range of crop diversity, and are an important complement to the more high-technology approaches being pursued at the present time. In this book we have tried to bring together a range of chapters on various different aspects of base-broadening. Part I includes chapters on some of the basic principles and the factors that limit available genetic diversity in plant breeding and crop produc-

Overview

21

tion, both from crop genetic and from socioeconomic and human perspectives. Part II looks critically at the available diversity in selected crop genepools and in specic production situations. The chapters provide illustrative examples of specic situations and are not in any sense meant to be exhaustive. Part III presents recent work from different incorporation programmes where the emphasis has been on creating and maintaining populations. Part IV includes chapters which cover documented examples of basebroadening in different crops by means of approaches other than the creation and longterm management of composite populations. Both formal plant breeding programmes and farmer participatory approaches are included; wherever possible the emphasis has been on documenting past work and bringing together known cases that reect different experiences and situations.

References
Anderson, J.R., Pardey, P.G. and Roseboom, J. (1993) Sustaining growth in agriculture: a quantitative review of agricultural research investments. Agricultural Economics 10, 107123. Brush, S.B. and Meng, E. (1998) The value of wheat genetic resources to farmers in Turkey. In: Evenson, R.E., Gollin, D. and Santaniello, V. (eds) Agricultural Values of Plant Genetic Resources. CAB International, Wallingford, UK. Carrol, C.P. and De Main, M.J. (1989) The agronomic value of tetraploid F1 hybrids between potatoes of Group Tuberosum and Group Phureja/Stenotomum. Potato Research 32, 447456. Cooper, D., Hobbelink, H. and Vellve, R. (1992) Why farmer-based conservation and improvement of plant genetic resources? In: Cooper, D., Vellve, R. and Hobbelink, H. (eds), Growing Diversity. Intermediate Technology Publications, London. Cromwell, E. and Wiggins, S. (1993) Sowing Beyond the State: NGOs and Seed Supply in Developing Countries. ODI, London. Duvick, D.N. (1984) Genetic diversity in major farm crops on the farm and in reserve. Economic Botany 38(2), 161178. Eriksson, G., Namkoong, G. and Roberds, J.H. (1993) Dynamic gene conservation for uncertain futures. Forest Ecology and Management 62, 1537. FAO (1996) The Global Plan of Action for the Conservation and Sustainable Use of Plant Genetic Resources for Food and Agriculture. Food and Agriculture Organization of the United Nations, Rome. FAO (1998) The State of the Worlds Plant Genetic Resources for Food and Agriculture. Food and Agriculture Organization of the United Nations, Rome. FAO (1999) Broadening the genetic base of crops, report of a workshop, Rome, September 1997. Information paper for the Commission on Genetic Resources for Food and Agriculture (CGRFA-8/98/Inf.) Food and Agriculture Organization of the United Nations, Rome. FAO (2000) Broadening the genetic base of crops, report of a workshop, Edinburgh, November 1999. Scottish Agricultural College, Edinburgh (unpublished) Ferguson, A.E. (1992) Differences among women farmers: implications for African agricultural research programs. In: Proceedings of the Workshop on Social Science Research and CRSP, Cornahan Conference Centre, University of Kentucky, Lexington, Kentucky, June 911. Goodman, M.M. (1985) Exotic maize germplasm: status, prospects and remedies. Iowa State Journal of Research 59, 497527. Goodman, M.M. (1990) Genetic and germplasm stocks worth conserving. Journal of Heredity 81, 1116.

22

H.D. Cooper et al. Grisley, W. (1994) Farmer to farmer transfer of new crop varieties: an empirical analysis on small farms in Uganda. Agricultural Economics 11, 4349 Harlan, H.V. and Martini, M.L. (1929) A composite hybrid mixture. Journal of the American Society of Agronomy and Crop Science Society of America 21, 487490. Harlan, H.V. and Martini, M.L. (1938) The effect of natural selection in a mixture of barley varieties. Journal of the American Society of Agricultural Research 57, 189199. Johnson, A.W. (1972) Individuality and experimentation in traditional agriculture. Human Ecology 1 149159. Kannenberg, L.W. and Falk, D.E. (1995) Models for activation of plant genetic resources for crop breeding programs. Canadian Journal of Plant Science 75, 4553. Ladizinsky, G. (1985) Founder effect in crop-plant evolution. Economic Botany 39, 191199. Lewontin, R.C. (1984) Adaptation. In: Sober, E. (ed.) Conceptual Issues in Evolutionary Biology, MIT Press, Cambridge, Massachusetts, pp. 235251. Louette, D. (1994) Gestion traditionnelle de varits de mas dans la rserve de la biosphre Sierra de Manantlan (RBSM, tats de Jalisco et Colima, Mexique) et conservation in situ des ressources gntiques des plantes cultives. ENSA, Montpellier. Louette, D., Charrier, A. and Berthaud, J. (1997) In situ conservation of maize in Mexico: genetic diversity and maize seed management in a traditional community. Economic Botany 51(1), 2038. Mayr, E. (1988) Towards a New Philosophy of Biology: Observations of an Evolutionist. The Belknap Press of Harvard University, Cambridge, Massachusetts. Nakagahara, M., Okuno, K. and Vaughan, D. (1997) Rice genetic resources: history, conservation, investigative characterization and use in Japan. Plant Molecular Biology 35, 6977. Namkoong, G., Barnes, R.D. and Burley, J. (1980) A Philosophy of Breeding Strategy for Tropical Forest Trees. Tropical Forestry Paper 16, Oxford University, Oxford. National Research Council (1972) Genetic Vulnerability of Major Crops. National Academy of Sciences, Washington, DC. Rao, P.S. and Gardiner, J. (1997) Varietal change and diversity in the Sugar Association of the Caribbean member countries. Proceedings of the XXVI West Indies Sugar Technologies Conference, Guyana, pp. 4144. Rasmusson, D.C. and Phillips, R.L. (1997) Plant breeding progress and genetic diversity from de novo variation and elevated epistasis. Crop Science 37, 303310. Shands, H.L. (1990) Plant genetic resource conservation: the role of the genebank in delivering useful genetic materials to the research scientist. Journal of Heredity 81, 710. Simmonds, N.W. (1962) Variability in crop plants: its use and conservation. Biological Reviews 37, 422465. Simmonds, N.W. (1989) Rubber breeding. In: Webster, C.C. and Baulkwill, W.J. (eds) Rubber. Longman, London, pp. 85124. Simmonds, N.W. (1993) Introgression and incorporation. Strategies for the use of crop genetic resources. Biological Reviews 68, 539562. Smale, M. (1997) The Green Revolution and wheat genetic diversity: some unfounded assumptions. World Development 25, 12571269. Spillane, C. (1999) Recent Developments in Biotechnology as they Relate to Plant Genetic Resources for Food and Agriculture. Background Study Paper No. 9, FAO Commission on Genetic Resources, Rome. Spillane, C., Engels, J., Fassil, F., Withers, L., Gass, T. and Cooper, D. (1999) Strengthening National Programmes for Plant Genetic Resources for Food and Agriculture. Issues in Genetic Resources No. 8, IPGRI, Rome. Sperling, L., Scheidegger, U. and Buruchara, R. (1996) Designing Seed Systems with Small Farmers: Principles Derived from Bean Research in the Great Lakes region of Africa. Agricultural Research and Extension Network Paper No. 60, London, ODI.

Overview

23

Sthapit, B.R., Joshi, K.D. and Witcombe, J.R. (1996) Farmer participatory crop improvement. III. Participatory plant breeding, a case for rice in Nepal. Experimental Agriculture 32, 479496. Suneson, C.A. (1969) Evolutionary plant breeding. Crop Science 9, 119121. Swanson, T.M., Pearce, D.W. and Cervigni, R. (1994) The Appropriation of the Benets of Plant Genetic Resources for Agriculture: an Economic Analysis of the Alternative Mechanisms for Biodiversity Conservation. Background Study Paper No. 1, FAO Commission on Plant Genetic Resources, FAO, Rome. UNEP (1997) The Biodiversity Agenda: Decisions from the Third Meeting of the Conference of Parties to the Convention on Biological Diversity, Buenos Aires, 415 November, 1996. United Nations, New York. Van Hintum (1999) The general methodology for creating a core collection. In: Johnson, R.C. and Hodgkin, T. (eds) Core Collections for Today and Tomorrow. IPGRI, Rome, pp. 1017. Whitlock, M.C., Phillips, P.C., Moore, F.B.G. and Tonsor, S.J. (1995) Multiple tness peaks and epistasis. Annual Review of Ecological Systematics 26, 601629. Witcombe, J.R., Joshi, A., Joshi, K.D. and Sthapit, B.R. (1996) Farmer participatory crop improvement. I. Varietal selection and breeding methods and their impact on biodiversity. Experimental Agriculture 32, 445460. Wood, D. and Lenn, J.M. (1996) The conservation of agrobiodiversity on-farm: questioning the emerging paradigm. Biodiversity and Conservation 5, 121 Wright, S. (1922) Evolution in Mendelian populations. Genetics 16, 97159. Wright, S. (1977) Evolution and the Genetics of Populations, Vol. 3. University of Chicago Press, Chicago, Illinois. Wych, R.D. and Rasmusson, D.C. (1983) Genetic improvement of malting barley cultivars since 1920. Crop Science 23, 10371040. Xu, Y., Beachell, H., Dliday, R. and McCouch, S.R. (1997) A genome-wide view of rice germplasm. Plant Genome Conference, San Diego, CA, January 1216 1997. http://probe.nalusda.gov:8000/otherdocs/pg/pg5/abstracts/p-3b-62.html

Evolutionary and Genetic Perspectives on the Dynamics of Crop Genepools

C. SPILLANE1 AND P. GEPTS2
1Institute

for Plant Biology, Zurich University, Zollikerstrasse 107, CH-8008 Zrich, Switzerland; 2Department of Agronomy and Range Science, University of California, 1 Shields Avenue, Davis, CA 95616-8515, USA

Nothing makes sense except in the light of evolution (T. Dobzhansky)

Introduction
Genetic diversity is not randomly distributed across or within plant populations. Various factors such as reproductive system, life history and gene ow can inuence the level and distribution of genetic diversity within and among populations (Hamrick, 1997). Evolutionary and historical information on a particular crops domestication, geographic dissemination and the extent of use of its various genepools is extremely valuable for the rational planning of the use of germplasm in any germplasm-enhancement or base-broadening initiatives (Smartt, 1986; Gepts and Debouck, 1991; Gepts, 1993; Simmonds, 1993). However, crop genetic diversity in itself is not necessarily an economically productive goal from a farmers perspective. It may even be counterproductive if resources are expended to increase diversity randomly when they instead might have been used to target or adapt available germplasm diversity to situations and locations where it is needed (Smith and Duvick, 1989). There are instances where signicant breeding progress has been made based upon relatively narrow genetic bases (Leng, 1974; Wych and Rasmussen, 1983; Hallauer, 1986; MacKey, 1986; Dudley and Lambert, 1992; Manninen and Nissila, 1997; Rasmussen and Phillips, 1997). Different mechanisms for the endogenous generation of genetic variability have been posited as possible explanations for such instances (Rasmussen and Phillips, 1997). On the other hand, theory predicts that the response to selection diminishes as the genetic variability decreases (Falconer, 1989). Indeed, there are situations where lack of useful genetic variability is considered the potential rate-limiting factor for genetic progress (Shands and Weisner, 1991, 1992; Simmonds, 1993; Kannenberg and Falk, 1995). For all crops, both farmers and breeders have to date only used a fraction of the
IPGRI/FAO 2001. Broadening the Genetic Base of Crop Production (eds H.D. Cooper, C. Spillane and T. Hodgkin)

25

26

C. Spillane and P. Gepts

plant genetic resources that are available. Although gains have been achieved from the tapping of this fraction of genetic resources by plant breeding, these gains have been limited (Fehr, 1981; Gepts, 1998). Current breeding efforts, for valid reasons of practical expediency, often tend still to exploit the genetic variability which was inherent in early founder populations and which resulted from early cycles of selection. In such instances breeding progress becomes increasingly difcult with each cycle until a genetic ceiling on improvement is approached, unless signicant new variability is introduced into breeding populations (Kannenberg and Falk, 1995; Gepts, 1998). While yield increases for some crops have been signicant in some countries, in other countries (or agroenvironments) they have been much less spectacular. For instance, yields of maize in the Central American region have remained well below the global average since 1961. The same applies to sorghum and millet in East Africa, cassava in Central Africa, potato in South America and sweet potato in East Africa (FAO, 1998). For some crops, there has been little progress in almost all environments. For example, dry bean yields have remained on a plateau globally since the 1950s (McClean et al., 1993; Gepts, 1998). On a local level, some landraces may actually have a quite narrow genetic base because of genetic bottlenecking due to past events (Wood and Lenn, 1996). Such bottlenecks can limit the response to selection that breeders (or farmers) can achieve in improving their crops. In some instances there may be a need to facilitate the broadening of breeding or cultivated populations in key staple crops to remove (or reduce) barriers to yield increases due to limited genetic diversity (Simmonds, 1993; Kannenberg and Falk, 1995). For other crops, it may be pointless to broaden the genetic base per se. In distinguishing between such instances it is the authors assertion that it will be useful to initially identify in what geographic regions the genetic base has been narrowed due to its evolutionary history since (and during) domestication, and then look at other factors (yield plateau, yield gain rate, etc.) to determine whether a genetic base-broadening exercise may be warranted or not. This chapter therefore looks at a range of factors over both time and geography that can affect the levels of genetic diversity found in crop plants. In particular the chapter focuses on factors that can cause genetic bottlenecks or loss of useful genetic diversity to occur. The authors recognize that genetic diversity is lost as a result of anthropogenic or natural selection and do not equate such purging/purifying selection (elimination of deleterious or undesirable alleles) with the largely undened concept of genetic erosion (Allard, 1996; Smale, 1997; Wood and Lenn, 1997).

Genepools for Crop Improvement

An essential starting point for any meaningful genetic base-broadening effort will be to assess the extent of utilization of the crops genetic resources to date and whether this can be (or needs to be) improved to meet future socioeconomic needs (Smartt, 1986). The genepool concept of Harlan and de Wet (1971) provides a good basis for categorizing plant genetic resources according to their ease of use for plant breeding or cultivation (see Box 2.1). Most crop genepools are typically depauparate for some agronomically desirable traits (e.g. protein quality, biotic or abiotic stress tolerance, etc.) that may only be avail-

The Dynamics of Crop Genepools

27

Box 2.1. Genepools of crop genetic resources. In the classication of Harlan and de Wet (1971), for each crop species there are three genepools of genetic resources:

The primary genepool (GP-1), which refers to germplasm within which sexual recombination is readily accomplished as a result of cross-hybridization (i.e. the true biological species). This category includes landraces, cultivars in current use, obsolete cultivars and many breeders lines. This category also includes the conspecic wild progenitor (Dobzhansky, 1937; Mayr, 1963). The secondary genepool (GP-2), which refers to germplasm within which crosses with the cultivated species are difcult but can be achieved using conventional plant breeding methods. Hybrids may be weak or partially sterile, chromosomes may pair poorly and there may be differences in ploidy levels. This category includes non-conspecic wild relatives of the cultivated species, as well as other cultivated species belonging to the same genus. The tertiary genepool (GP-3), which refers to germplasm where sexual recombination is possible only by advanced breeding or biotechnological techniques. This category would include many distant wild relatives of the cultivated species (or even different genera) from which gene transfer through sexual recombination is possible but is very difcult. Embryo culture of either the hybrid or its offspring may be needed or bridging crosses may be necessary to transfer germplasm from this genepool to GP-1.

able in the sexually inaccessible genepools of other crops or species. Many transgenic approaches to crop improvement hence arise from a lack of suitable conventional approaches to dealing with a particular agronomic problem or need (Spillane, 2000). Where lack of an agronomic trait seriously limits agricultural productivity, transgenic approaches may provide new options for phenotypic base-broadening where current options for trait selection or generation are lacking in their efcacy or existence (Thro and Spillane, 2000). It is now necessary to consider a fourth (or quaternary, GP-4) genepool to reect recent technological advances in plant molecular biology and genomics. Such a genepool would refer to any useful germplasm, which can be harnessed from species that are sexually incompatible with the crop species. Plant transformation techniques such as Agrobacterium-mediated transformation and biolistics allow the precise transfer of genes from any organism into either plant nuclear (Walden and Wingender, 1995) or chloroplast genomes (Svab and Maliga, 1993; Daniell et al., 1998). The many examples of the application of gene transfer from microorganisms through genetic engineering techniques range from the introduction of vaccine antigen genes to aluminium tolerance genes to food plants (e.g. Mason et al., 1996; de la Fuente, 1997; Arakawa et al., 1998; Cheng et al., 1998). Many isolated plant genes are now also being transferred between sexually incompatible crop plant species (e.g. Whitham et al., 1996; Molvig et al., 1997; Wilkinson et al., 1997). The evolutionary relationships between sexually compatible (and incompatible) crop genepools are being elucidated. Recent advances in comparative molecular mapping between different crops (e.g. the Gramineae) open hitherto unknown possibilities for uniting the genetics of different crop species in the grass family (Bennetzen, 1998; Gale and Devos, 1998; Kellogg, 1998; McCouch, 1998). It may even be possible to develop a unied genetic map of higher plants, which spans both monocots and dicots

28

C. Spillane and P. Gepts

(Paterson et al., 1996; Bevan and Murphy, 1999). The application of molecular markers to genetic linkage maps of a wide range of crops is allowing the identication of the chromosomal (physical) locations of genes for improving yield and other complex traits important to agriculture (McCouch et al., 1997). The underlying genetics of a wide range of quantitative agronomic traits is being dissected through the identication of quantitative trait loci (QTLs) using a combination of molecular markers and advanced statistical breeding procedures. These developments will provide increased opportunities for understanding the genetic basis of adaptation across different crop species and applying the knowledge gained from one crop genepool to promoting the development of adapted germplasm in the primary genepool of another crop (Hoisington et al., 1996; Devos and Gale, 1997; McCouch, 1998; Sasaki, 1998). Genomics is a term that is used to describe the development and application of large-scale, high-throughput and parallel-processing approaches to the functional analysis of entire genomes or genetic systems (Bouchez and Hofte, 1998). Genomics technologies include large-scale genome sequencing efforts such as those now completed for Arabidopsis and rice (Rounsley et al., 1998). Emerging genomics technologies such as DNA chip and microarray technologies can allow the function of tens of thousands of different genes to be analysed simultaneously in parallel (Baldwin et al., 1999; Kurian et al., 1999; Zweiger, 1999). Genomics initiatives are both generating genetic markers and identifying genes that can be used either for marker/gene-assisted selection, or the development of transgenics with improved novel agronomic properties (Phillips and Freeling, 1998; Dellapenna, 1999; Sommerville and Sommerville, 1999; Suarez et al., 2000). Genomics technologies will provide powerful tools for both the analysis and broadening of the genetic base of crop genepools (Lee, 1998; Sniegowski, 1999; Spillane, 2000).

Promoting Useful Geneow for Crop Improvement

Geneow refers to the movement of genes among populations of crops and related taxa, or the introduction of genes from related taxa into the recipient genepool of a crop. Such geneow can be induced by humans (as in plant breeding) or can occur spontaneously between crops and wild relatives. The latter phenomenon is less well understood. Using the Harlan and de Wet (1971) classication, it is evident that geneow can occur at different taxonomic levels intraspecic, interspecic and intergeneric. The extent of spontaneous geneow is governed by many factors such as the breeding system of the crop, the distribution pattern of the crop and its wild relatives, the mode of pollination and dispersal, planting design of the crop, inter-plant distances, longevity of individuals and population sizes (Levin and Kerster, 1974; Klinger et al., 1992; Hamrick et al., 1995). The more geneow capacity is limited, the more between-population genetic differentiation is predicted to increase (Wright, 1965). High levels of gene ow can homogenize population structure and counteract the effects of drift and diversifying selection. High levels of gene ow into small populations can also reduce local variation, prevent local adaptive differentiation and reduce tness (Slatkin, 1987). While both breeders and farmers generally focus their crop improvement efforts on what materials are available to them from the primary genepool, some plant

The Dynamics of Crop Genepools

29

breeders and biotechnologists also attempt to move useful genetic resources (i.e. promote geneow) from the secondary, tertiary and quaternary genepools into the cultivated primary genepools. For most crops, much of the useful genetic diversity available in the primary, secondary, tertiary or quaternary genepools has not, to date, been used over a wide range of agroenvironments either by breeders or by farmers (Smartt, 1984). Table 2.1 provides some examples of the genepools of a number of crop species. While the direction of geneow can be both to and from the crop, in this Chapter we are concerned with geneow into and within the primary genepool. We do not focus on outward geneow from the primary genepool to the secondary and tertiary genepools, which is a topic that has recently become subject to much investigation
Table 2.1. The primary, secondary and tertiary genepools of selected crops. Crop Barley Bread wheat Primary genepool Hordeum vulgare H. spontaneum Triticum aestivum Secondary genepool Tertiary genepool H. bulbosum Triticum spp. Aegilops Secale Thinopyrum Haynaldia Leymus Hordeum spp. Triticeae References Bothmer et al., 1983; Valkoun et al., 1997

Other Triticeae spp. Dubin et al., 1997

Cassava Common bean

Manihot esculenta

Over 80 wild Manihot spp. P. acutifolius, P. lunatus and other Phaseolus spp.

Bonierbale et al., 1997 Debouck, 1991; Schmit and Debouck, 1991; Schmit et al., 1993; Llaca et al., 1994; Freytag and Debouck, 1996; Hidalgo and Beebe, 1997

Phaseolus vulgaris P. coccineus, P. costaricensis, P. polyanthus

Chickpea

Cicer arietinum; None C. reticulatum; C. echinospermum

C. bijugum; Muehlbauer et al., 1994; C. pinnatidum; Singh et al., 1997 C. judaicum; C. chorassanicum; C. montbretii Arachis spp. L. nigricans Smartt, 1984; Singh and Nigam, 1997 Muehlbauer et al., 1994; Robertson and Erskine, 1997

Groundnut Lentil

Arachis hypogaea A. monticola Lens culinaris

Section Arachis None

Sugar cane

Saccharum ofcinarum; S. barberi S. edule; S. sinense; S. robustum; S. spontaneum

Erianthus spp.; Imperata spp.; Anderson and Miller, 1995 Miscanthus spp.; Narenja spp.; Nephia spp.; Neyrudia spp.; Sclerostachya spp.; Veteveria spp.

Sweet potato Ipomoea batatas

Approx. 400 wild Ipomoea spp.

Huaman and Zhang, 1997

30

C. Spillane and P. Gepts

because of perceived risks of geneow from transgenic crops to their wild relatives. Such biosafety-related geneow continues to be well researched and reviewed elsewhere (Schmitt and Linder, 1994; Lefol et al., 1996; Mikkelsen et al., 1996; Linder et al., 1998). This chapter looks at some of the factors that are barriers to deliberately promoting useful geneow from the secondary and tertiary genepools into the primary cultivated genepool. We also look at barriers to useful geneow within the primary cultivated genepool and how this is important for the development of germplasm adapted to specic agroenvironments. All of these barriers can represent constraints to broadening the genetic base of the primary cultivated genepool of any crop.

Genetic Bottlenecks in the Continuing Evolution of the Primary Cultivated Genepool

During their evolutionary history, all crops have been subjected to genetic bottlenecks and subsequent founder effects, over both time and geography. A genetic bottleneck occurs when the genetic diversity of a population is reduced due to a decrease in its population size because of some event (Nei et al., 1975). Founder effects refer to situations where a new effective population is derived from a limited number of individuals representing a small sample of the genetic pool from which they were derived (Mayr, 1942). Smaller population sizes are more prone to loss of rare alleles or allelic combinations because of genetic drift (Wright, 1965). The overall result is decreased genetic variability in the new population. Founder effects can affect subsequent response to both articial and natural selection pressures (James, 1971). Founder effects can sometimes lead to a morphologically distinct species or race, when the founder populations genepool undergoes what Mayr (1954) called a genetic revolution to develop new coadapted gene complexes, in response to new selection pressures (Templeton, 1981). Quantitative variation may harbour potentially strong epistatic components, and such cryptic epistatic variation may become important when allele frequencies are substantially disturbed by selection or a population bottleneck (Goodnight, 1987; Wade, 1992). Alterations in a small number of key genes could potentially have major effects on the genotype and morphology as a whole (Wilson et al., 1974; Hedrick and McDonald, 1980; Carson and Templeton, 1984; Lande, 1985). This may have been the case in the domestication of some of the cultivated cereals and legumes (Doebley et al., 1995, 1997; Paterson et al., 1995; Koinange et al., 1996). Indeed, recent work further suggests that changes in the regulatory (rather than protein encoding) region of key genes (e.g. teosinte branched 1) involved in domestication may be responsible for some of the major phenotypic effects that occurred during domestication (Doebley and Lukens, 1998; Wang et al., 1999). In any crops evolutionary history, genetic bottlenecks and subsequent founder effects are likely to have occurred at all levels of geographical scale. For instance, domestication events will have contributed to global bottlenecks for most crops while crop migration events are likely to have contributed to more regional bottlenecks. However, genetic bottlenecks will also have occurred at national and local levels. For many crops, genetic diversity is likely to have been lost over time from what was domesticated due to random genetic drift caused by small population sizes resulting

The Dynamics of Crop Genepools

31

from either new crop introductions or crop failures. Such founder events will have been an extremely important factor in narrowing the genetic base of many crops, especially where their cultivation has moved beyond their centre of diversity/origin. The following sections attempt to look at factors that contribute to genetic bottlenecks of crop genetic diversity/resources over time, geography or both.

Founder effect of domestication As humans migrated over the past 10,000 years, they moved domesticated crops beyond their initial centres of origin and diversity (Smith, 1995; Diamond, 1997). The earliest protocrops are likely to have spread along with the cultural diffusion of the idea of cultivating plants. It is probable that the rst farmers chose to grow the rst domesticates where they could be grown, rather than domesticate a new set of crop plants for each new location. In many crop plants, the number of genes controlling the essential agronomic traits of non-shattering of seed and reduction of seed dormancy is small. During the domestication process involving the sowing of wild seeds, it is likely that once these agronomic traits had become xed in a cultivated population there would have been little rationale to continue sowing seeds of the wild plant, once the domesticated type had been established (Ladizinsky, 1985). Some plant species with good potential in domestication could therefore quite feasibly due to phytogeographic accidents have been totally neglected by the earliest farmers (Smartt, 1986). The earliest domestication events for instance were biased in favour of annual rather than perennial cereal crops (Francis, 1990; Pimm, 1997). They were also probably biased in favour of sexually reproducing crops which easily generated visible variation in their progeny for subsequent selection (Jefferson and Bicknell, 1997). The proto- or initial domesticates did not inherently have an outstanding agronomic potential nor did they have high levels of genetic diversity. In many locations, this limited initial diversity is likely to have been added to only marginally since (Simmonds, 1993; Rasmusson and Phillips, 1997). The selection pressures for many traits (e.g. ripening, seed size, germination, nondehiscence, palatability, etc.) associated with domestication are likely to have xed alleles of major effect (Knight, 1948; Hilu, 1983; Gottlieb, 1984). Major effect genes are proposed to be the basis for the selection of the tomato fruit (Paterson et al., 1991; Paterson, 1995) and the transformation of the teosinte inorescence into the maize ear (Doebley et al., 1997). One gene, Sh1, accounts for the difference between shattering and non-shattering sorghum types (Paterson et al., 1995). Paterson et al. (1995) demonstrated that the domestication of sorghum, rice and maize is due predominantly to a small number of major effect homologous genes conferring phenotypes such as large seed size, seed dispersal and daylength insensitive owering that were convergently selected for in the three species. Such genetic changes largely dene the basic biological differences between wild and cultivated germplasm. The rst genetic bottleneck event, for any crop selected for domestication, is likely to be the actual initial domestication of the crop, the earliest of which took place some 700010,000 years ago. Most of our major food crops were initially domesticated by 5000 years ago, at which stage only a small part of the wild genepool is likely to have been brought under cultivation (Ladizinsky, 1985; Doebley, 1989; Gepts, 1995). The

32

C. Spillane and P. Gepts

initial domestication of each crop species by the earliest farmers represented a selection process, which de facto would have reduced genetic diversity in the incipient crops (Ladizinsky, 1985; Allard, 1988; Debouck and Tohme, 1989; Doebley, 1992; Gepts, 1995). Numerous studies of the relative genetic variability found in cultivated versus wild progenitors demonstrate that the wild species are generally more variable than the corresponding crop (Chapman, 1989; Gepts, 1995). Among many crops, this has now been shown for barley (Brown and Clegg, 1983; Petersen et al., 1994), soybean (Doyle, 1988), pearl millet (Gepts and Clegg, 1989), lentil (Havey and Muehlbauer, 1989), sunower (Arias and Rieseberg, 1995; Cronn et al., 1997) and rice (Cordesse et al., 1990). However, this is not an absolute rule. In comparisons between the cultivated and wild forms of sorghum, few novel alleles were identied in the wild genepool (Deu et al., 1995; de Oliveira et al., 1996). In addition, some parts of the genome may show increased variation caused by selection under domestication (Gepts, 1995). For most crops, a wild conspecic progenitor species forms part of the primary genepool (GP-1) and crosses of the conspecic progenitor are usually possible with the cultivated forms. Nevertheless, it is likely that much of the available genetic diversity in the wild progenitors of many crops has rarely been used or comprehensively harnessed either by farmers or by breeders since the earliest domestication. Evolutionary studies (Table 2.2), which conclusively identify the wild progenitors of crop plants, therefore provide valuable information for initiatives to broaden the genetic base of the cultivated genepool of crops (Gepts, 1995). Gene exchange between many crops and their non-conspecic wild relatives is often highly restricted (Harlan, 1965). Various isolating mechanisms prevent natural hybridization between sympatric cultivated and wild plants (e.g. Robert et al., 1991). If geneow occurs, it may be more likely to occur in the direction from the cultivated to

Table 2.2. Examples of closest wild relative of diploid species identied or conrmed using molecular markers (see Gepts, 1995 for references). Crop Avocado Barley Common bean Groundnut Lentil Lettuce Maize Pea Pearl millet Rice Rice Sorghum Soybean Tomato Cultigen Persea americana var. guatemalensis Hordeum vulgare ssp. vulgare Phaseolus vulgaris var. vulgaris Arachis hypogea Lens culinaris ssp. culinaris Lactuca sativa Zea mays Pisum sativum Pennisetum glaucum Oryza sativa Oryza glaberrima Sorghum bicolor Glycine max Lycopersicon esculentum Closest wild relative P. steyermarkii (female) P. nubigena (male) Hordeum spontaneum P. vulgaris var. aborigineus and mexicanus A. monticola L. culinaris ssp. orientalis L. serriola and other unidentied wild taxa Z. mays ssp. parviglumis (Balsas and Jalisco) P. humile P. glaucum ssp. monodii O. rupogon O. breviligulata S. bicolor ssp. arundinaceum G. soja L. exculentum var. cerasiforme

The Dynamics of Crop Genepools

33

the wild populations, because of farmer selection against wild domesticated hybrids. However, spontaneous geneow between a crop and its wild relatives has been documented for many crops, including barley (Svitashev et al., 1994), beans (Singh et al., 1991a,b; Beebe et al., 1997), carrot (Wijnheijmer et al., 1989), cotton (Brubaker et al., 1993), foxtail millet (Boudry et al., 1993), maize (Doebley, 1990), quinoa (Wilson and Manhart, 1993), radish (Klinger et al., 1992), rice (Langevin et al., 1990), sugarbeet (Boudry et al., 1993), squash (Wilson, 1990), sunower (Arias and Rieseberg, 1994) and tomato (Rick, 1974). The agronomic signicance of such geneow is unresolved (Wood and Lenn, 1996). Even today, the induced geneow of genetic resources from both wild relatives and conspecic progenitors via plant breeding into the cultivated genepools is still limited and often concerns the introgression of a few major genes for disease or pest resistance via recurrent backcrossing (Simmonds, 1993; Rasmusson and Phillips, 1997).

Polyploidy-induced domestication bottlenecks The adaptive benet of polyploidy is largely unknown (de Wet, 1979; Leitch and Bennett, 1997; Wendel, 1999). However, almost 80% of crop species are polyploid (Grant, 1963). Some crops, such as wheat, have been domesticated or have arisen as a result of polyploidization events. The tness of novel polyploids relative to their progenitors can be partially increased if the new polyploid is self-fertile and perennial. Hence, polyploidy is much more common in perennial species than annual ones, and most allopolyploid species are self-fertile (MacKey, 1970; Stebbins, 1971). In general, polyploidization may offer novel avenues for phenotypic response to selection (e.g. Wright et al., 1998). Repeated cycles of hybridization and polyploidy have played an important role in the evolution of wheats, sugar cane, potato, sweet potato, yams and plantain. Wild polyploid complexes can generate variants that are of agricultural value, as in the case of taro where numerous cultivated races have been selected from the wild (Kuruvilla and Singh, 1981). Where diploid progenitor species produce measureable amounts of unreduced gametes (e.g. in potato, cassava, cotton, strawberry and cherry), it is more likely that derived polyploids will have originated repeatedly in time and geography (Harlan and de Wet, 1975). However, polyploidization events, which have occurred infrequently or over a local scale, can represent a genetic bottleneck to geneow from the wild progenitors into the cultivated genepool. Of the two major types of polyploids, this bottleneck effect applies in particular to allopolyploid crops because allopolyploids have originated from chromosome doubling of sterile interspecic diploid hybrids or the union of unreduced gametes (Ladizinsky, 1985). Many important crops such as hexaploid wheats (Triticum aestivum), tobacco (Nicotiana tabacum), rape (Brassica napus) and cotton (Gossypium hirsutum) are allopolyploids. Allopolyploids, unlike autoployploids, can lead to entirely new gene combinations and hence novel phenotypes. Cotton provides a good example of a likely genetic bottleneck due to allopolyploidy. Most commercial cotton varieties belong to G. hirsutum L. (upland cotton). The development of upland cotton from hybridization and polyploidization in the New World between G. raimondi and the introduced species G. herbaceum is likely to have constituted a signicant genetic bottleneck event in the domestication of upland cotton (Ladizinsky,

34

C. Spillane and P. Gepts

1985). Polyploidization can also lead to new and useful gene interactions. This feature has been suggested by Leitch and Bennett (1997) in general for plants and by Jiang et al. (1998) specically for cotton. The latter authors observed a bias of QTLs for seed bre traits towards the D genome, although the D genome ancestor does not produce bres. While the progenitors of many allopolyploid species are known to possess useful genetic variation, such genetic resources are not easily accessible because of chromosomal and endosperm imbalances in crosses. Hence, they have rarely been used in plant breeding to date. F1 hybrids between allopolyploids and their progenitors are usually sterile. In most instances, promoting geneow from the progenitors into the primary genepool of allopolyploids cannot be done by regular crosses and requires the use of embryo rescue and other wide crossing technologies. However, for some other crops such polyploid barriers are not insurmountable. For instance, many wild Solanum species can readily be crossed with the cultivated potato once their ploidy level is experimentally adjusted by chromosome doubling or haploidization. However, accessing genetic variability in some of the wild species such as S. etuberosum and S. brevidens requires more complex bridging crosses and chromosome adjustment (Hermsen, 1989). Polyploidy may be a barrier to interspecic gene ow in some legumes (e.g. Phaseolus) but not in others (e.g. Arachis) (Smartt, 1986). Polyploidy can sometimes act as a genetic bridge between different diploid genepools (Dewey, 1980), as in the case of the wild Aegilops species of wheat (Zohary, 1965).

Crossability barriers limit geneow While the genepools of many crop wild relatives may be known to contain agronomically useful germplasm, there are often signicant technical barriers to obtaining viable hybrids between the primary genepool and wild species in the secondary and tertiary genepools (Muehlbauer et al., 1994). Many types of reproductive isolating barriers can reduce the level of geneow possible between species or even between populations. Combinations of such barriers can result in reproductive isolation, i.e. no geneow. A signicant element of plant breeding to broaden the genetic bases of crops is concerned with nding ways to overcome such isolating barriers, so as to gain access to novel and useful genetic variation. There may be extensive genetic variation in crossability of some cultivated genotypes with wild relatives, as is the case for wheat (Jiang et al., 1994) and Phaseolus sp. (Parker and Michaels, 1986; Koinange and Gepts, 1992). Reproductive isolation, incompatibility barriers, embryo or endosperm abortion, hybrid sterility and limited levels of genetic recombination are signicant obstacles to the greater use of wild germplasm. These obstacles are in addition to those of undesirable linkages to non-agronomic traits once geneow has been achieved. Reproductive isolating barriers can be of two general types: (i) pre-mating barriers or (ii) post-mating barriers. There can be different types of pre-mating barriers to geneow between populations of the same or different species. These include ecogeographic, temporal, oral and gametic incompatibility. Ecogeographic barriers exist when habitats are distinct and hence the populations rarely come into contact. For instance, an ecogeographic barrier exists between the common bean, P. vulgaris, and the runner bean, P. coccineus. The former is found in warm to temperate areas and the

The Dynamics of Crop Genepools

35

latter is found in the cool, humid uplands of Mexico and Central America (Delgado Salinas et al., 1988; Debouck, 1991). In addition, populations of P. coccineus can be reproductively isolated because of differences in pollinator distribution: at higher altitudes, P. coccineus owers are visited by hummingbirds, whereas at lower altitudes they are visited by bumblebees (Brquez and Sarukhan, 1980). Seasonal isolation can occur when plants mature at different times with little to no overlap between the owering periods to allow hybrid formation. Gametic incompatibility occurs when foreign pollen grains cannot germinate on another species stigma or cannot grow down the style to reach the ovaries. Alternatively, pollen grains sometimes do function on an allochtonous stigma or style, but they are less competitive than the autochtonous pollen (e.g. Robert et al., 1991; Carney and Arnold, 1997). A wide range of gametic incompatibility barriers is observed in crosses between white and purple owered species of Capsicum (Zijlstra et al., 1991). There is also a range of post-mating barriers to geneow between populations of the same or different species. These include hybrid inviability, hybrid sterility and hybrid breakdown. Hybrid inviability occurs when hybrids are weak and have poor survival, such as occurs with most crosses between species in the secondary and primary crop genepools. An example of such hybrid inviability is provided by endosperm balance number (EBN) differences between species (Ehlenfeldt and Ortiz, 1995; Spillane et al., 2000). To achieve normal endosperm development, EBNs must typically be in a 2:1 maternal:paternal ratio (Johnston and Hanneman, 1982). EBNs can, however, be manipulated by ploidy changes to overcome endosperm development problems and obtain new hybrids (e.g. Carputo et al., 1997). Hybrid sterility occurs when the hybrids do not produce functional gametes and are hence sterile, as occurs in crosses between the cultivated rices Oryza sativa and O. glaberrima. Hybrid breakdown is observed when F2, backcross and later generation hybrids have reduced viability or fertility as is seen in many of the F2 populations of interspecic bean crosses (Hucl and Scoles, 1985). Geneow can be limited by both chromosomal (structural) and genic factors. Rieseberg and colleagues (1995a,b) have provided evidence for both types of barriers in Helianthus sp. They showed that rapid genomic reorganization following hybridization appeared to have taken place in Helianthus anomalus, relative to its two parental species, H. annuus (sunower) and H. petiolaris (Rieseberg et al., 1995a). In effect, this reorganization isolated H. anomalus from its parents and allowed it to diverge to become a separate species. Between H. annuus and H. petiolaris both chromosomal and genic factors play a role, as shown by the level of introgression of mapped markers into advanced backcross populations (Rieseberg et al., 1995b). Because of such mating barriers, the wild relatives of many crops (e.g. cowpea, chickpea, groundnut, millets) are often organized into sections based on their chromosome numbers, morphology and their crossability with each other and the cultigen(s). The dening feature of a section is often a genetic barrier to crossability with another section. For instance, there are three crossibility sections in the Cicer genus, which contains the cultivated chickpea C. arietinum. Of the 40 species in the Cicer genus, only C. reticulatum and C. echinospermum can be crossed with cultivated chickpea (Singh and Ocampo, 1997). In such instances, major constraints to geneow exist between different sections (Muehlbauer et al., 1994). Similarly, over 1500 Asian rice cultivars which were classied into six different phyletic groups by molecular analysis exhibit various

36

C. Spillane and P. Gepts

levels of sterility in inter-group crosses but not in intra-group crosses (Glaszmann, 1987; Khush, 1997). The research area of wide hybridization, especially of the less commercial crops, is considered to be a neglected area for research funding and for international coordinated efforts which are strongly linked to breeding programmes (Duvick, 1989). Yet, continuing advances in wide crossing techniques such as hybrid embryo culture (Sharma et al., 1996) and the development of novel crossing strategies (e.g. bridge crosses) are making the wild genepools of many crops ever more accessible (Stalker, 1980; Muehlbauer et al., 1994). For instance, wide crosses in wheat have become very successful since the advent of successful embryo rescue techniques (Jiang et al., 1994; Sharma, 1995). The success rate of gene transfer in such wide crosses can be increased by knowledge of the chromosome pairing mechanisms and their genetic control. Such knowledge is essential to promote recombination between heterologous or homologous chromosomes if one is to minimize (or maximize) the size of the introgressed chromosome segment(s) (e.g. Luo et al., 1996a,b). It is likely that continuing advances in structural genomics (e.g. comparative mapping) and genetic engineering (e.g. crossability transgenes) will result in new strategies for wide hybridization.

Inbreeding and outbreeding depression affect geneow Outbreeding depression (or hybrid dysgenesis) is the population level counterpart to genetic mechanisms separating species or subspecies. Hybrid plants within individual populations of species are usually normal, but hybrids between different populations of the same species are sometimes weak or inviable (Wallace, 1968). Within crop species, this can be a barrier to geneow between different populations. For example, the largeand small-seeded genepools of common bean (P. vulgaris) from the Andes and Mesoamerica, respectively, sometimes give rise, when crossed, to lethal or weak progeny. In the F1, the lethality is due to two complementary loci (Gepts and Bliss, 1985; Gepts, 1988). In subsequent generations, numerous genes appear to be involved (Welsh et al., 1995; Singh and Molina, 1996; W.C. Johnson and P. Gepts, unpublished results). In cultivated populations that evolved sympatrically with their wild progenitors, mechanisms that prevent or restrict geneow among populations are selectively advantageous. Marchais (1994) has suggested the existence of a genetic isolating mechanism between pearl millet and its wild relative in West Africa. Nevertheless, efcient reproductive isolating mechanisms are rare within the primary genepool, even among races/populations that have remained sympatric for thousands of generations. Among differently adapted cultivated and wild populations, disruptive selection is the primary mechanism that restricts geneow. The most common mechanisms isolating sympatric races of the same crop species from each other are differences in owering time and gametophytic or sporophytic systems that prevent cross-fertilization. Cultivated elds of African rice (O. glaberrima) are often invaded by populations of its wild relative, O. barthii. A dominant allele in each species makes hybrid progeny weaker than either parent and hence reduces geneow from one species to the other (Chu and Oka, 1972). Inbreeding depression has been documented for most investigated plant species,

The Dynamics of Crop Genepools

37

even highly selng species (Charlesworth and Charlesworth, 1987). In plant breeding, inbreeding depression can be an impediment to the increased use of exotic germplasm (Spillane et al., 2000). For example, use of open-pollinated tropical maize populations in a US maize breeding programme indicated that inbreeding depression for yield may be severe in some exotic populations (Goodman, 1992). Plants can also exhibit a phenomenon of optimal outcrossing distance which may have a major bearing on what types of crosses yield the most t or immediately favourable progeny (Waser, 1993). Such a phenomenon can be associated with either inbreeding or outbreeding depression (Lynch, 1991). These latter two phenomena are still poorly understood genetically yet can affect what germplasm is of more immediate utility in plant breeding (Spillane et al., 2000).

Mating system effects on geneow One of the most important evolutionary factors affecting geneow within a crop genepool is the mating behaviour of the crop. The mating system of a crop affects the geographic distribution of genetic variation within the crop (Schoen and Brown, 1991). Plants may be either mainly self-fertilized (inbreeding species) or there may be inherent biological barriers to self-compatibility such that an egg cell is normally only capable of being fertilized by pollen of a different genotype of that species (outbreeding species). The mode of reproduction of the crop in farmers elds may sometimes differ from that of the species from which the crop is derived, as is the case for vegetatively propagated crops such as sugar cane, potato, orchard crops and soft fruits. Most vegetatively propagated crops are outbreeders in which methods of clonal reproduction are used to bypass sexual reproduction. Also, among sexually propagated crops such as tomato there has been a long-term trend towards increase in autogamy (Rick et al., 1977). This trend is attributable to several causes, such as the preference for true-totypeness and the possible absence of suitable pollinators in areas outside the domestication area. Some genetic systems limit genetic recombination and therefore can act to maintain coadapted gene complexes. For instance, chromosomal rearrangements and tight genetic linkage can hold coadapted genes together by limiting recombination (Stebbins, 1971). Inbreeding and asexual (apomictic) reproduction can also have the same effect, although these systems tend to x whole genomes rather than blocks of genes. Asexual reproduction can occur because of agamospermy in which viable seeds arise without fertilization or by vegetative propagation via structures such as stolons, tubers, rhizomes or suckers. Many important crops are vegetatively propagated through tubers (potato, yams, sweet potatoes) or articially via cuttings (banana and cassava). Very few crops are obligate apomicts. In some crops such as Bracharia, sugar cane and some fruit crops, apomixis, polyembryony and predominant asexual reproduction are barriers to geneow and account for limited utilization of germplasm due to the difculty of breeding such crops (Savidan, 2000). Self-pollination is also a barrier to geneow. While outbreeding crop species are usually characterized by wide genetic polymorphism, high levels of heterozygosity and recombination in each sexual cycle, self-fertilization brings about drastic reductions in recombination and the splitting of the progeny population into homozygous, true

38

C. Spillane and P. Gepts

breeding lines (genotypes), which maintain genetic variation in tight multilocus gene combinations (Allard, 1996; Prez de la Vega, 1996). While selection operates mainly at the level of the individual genes in outcrossing plants, it operates at the level of the individual genotypes in the case of self-fertilizing plants. The multilocus genotypes in selng plants can be very tightly associated with denable micro-niches, soil types and climatic conditions. In populations of selng plants, genetic variability is maintained by the coexistence of different homozygous lines and genetic exibility is maintained by low levels of outcrossing, which result in the recombination and production of numerous new homozygous combinations (Allard, 1996; Prez de la Vega, 1996). On the other hand, an outcrossed, heterozygous species has a considerable amount of genic exibility with which to meet environmental change, but selng homozygous species may be better adapted to specic unchanging conditions. An intermediate strategy with variable outcrossing rates may yield the greatest long-term success in an unpredictable world; this may be why few extreme selfers or outcrossers are found (Allard, 1988).

Founder effects of geographic movements of crops Since the advent of agriculture, long-distance crop movements or dissemination have always occurred and continue to this day. By their nature, such crop movements do not lead to the even distribution of the genetic variability of crops, and many crops are likely to have been subjected to signicant genetic bottlenecks during movement from one location to another (Simmonds, 1995). Crop migration followed by natural selection over time in the new environement can lead to ecogeographic differentiation of the genetic variation at the new environment (e.g. Allard, 1990; Saghai-Maroof et al., 1990; Zhang et al., 1990). Major crop movements occurred in pre-Columbian times: for example, the movement of African crops such as sorghum, millet (Panicum spp.), cowpea (Vigna unguiculata), sesame (Sesamum indicum) and niger seed (Guizota abyssinica) to semi-arid areas of India (Simmonds, 1995). For instance, bananas (Musa spp.) had probably reached parts of upland East Africa from their centre of diversity in Asia by the rst millennium AD. The dissemination of mung bean (Vigna radiata) from its centre of origin in western Asia to Southeast Asia led to a reduction in seed protein diversity (Tomooka et al., 1992). The Columbian exchange after 1492 was a process which led to an increase in crop movements, in particular of crops previously unknown either within or outside of the Americas or Europe (maize, potato, cassava, sugar cane, banana, tomato, squash, etc.) (Bermejo and Len, 1994; Butzer, 1995). Maize emerged as an essential staple in Africa and as a minor staple in India and parts of China. Cassava, sweet potato and groundnut dispersed across Africa. Sweet potato became an extremely important food crop both within and outside of its primary centre of diversity in the Andean region (FAO, 1998). Since its spread beyond Latin America from the 16th century onwards, sweet potato is likely to have undergone many major geographical genetic bottlenecks, especially in its dissemination between smaller islands (Huaman and Zhang, 1997). Much of the genetic diversity available in Latin America has never been used outside Latin America (Z. Huaman, personal communication). Amaranth (Amaranthus hypochondriacus) represents a crop that only reached Asia

The Dynamics of Crop Genepools

39

(via Europe) in the early 19th century. It has been subjected to signicant bottlenecks since its migration from its centre of diversity in the Andean region in the 16th century. Spread of amaranths to higher latitudes in Asia has left behind any races that have owering delayed by long days. Although far more amaranth is now produced in Asia than in the Americas, most production has relied on a few broadly adapted Mexican A. hypochondriacus and A. cruentus selections. The extensive diversity present in Latin America appears to have played a limited role in developing regionally adapted germplasm (Kauffman and Weber, 1990). Genetic bottlenecks due to geographic movements may also have occurred very recently. For instance, present-day US sunower cultivars may be derived from a number of movement-related bottlenecks. These occurred when sunower was bred for a variety of uses in Europe and subsequently re-imported back to America, its centre of domestication, during the latter half of the 19th century (Cronn et al., 1997). Similarly, while cultivated soybean (Glycine max) originated in China at least 3000 years ago, it was not widely grown outside of Asia until the 1900s and its diffusion was subject to major genetic bottlenecks (Hymowitz and Kaizuma, 1981). The genetic base of cultivated soybean outside of Asia is considered to still be extremely narrow (Shoemaker et al., 1986; FAO, 1998; Dashiell and Fatokun, 1997). In some regions it is thought that initial crop introductions were based upon extremely narrow genetic bases, such as for potato (Europe), banana (Caribbean) and sugar cane (Caribbean) (Simmonds, 1995). However, it is likely that crops in most locations have been subjected to recurrent introductions of new germplasm following the initial introductions (Simmonds, 1995). Examples include sugar cane to the Americas and potato to the northern temperate environments of Europe and the USA. In some instances, crop movements may have reached the same destination via different routes or from different locations, resulting in different samplings of the primary genepool. The coconut is thought to have reached Mesoamerica independently via two routes, one trans-Atlantic from Africa and the other trans-Pacic, with the result that the coconuts of the two coasts of Mesoamerica are different (Simmonds, 1995). It is unlikely that such initial introductions represented anywhere near a full or optimal sampling of the genetic diversity available in the primary genepool of the crops concerned. In addition, the initial introductions are most likely to have led to genetic and phenotypic inertia effects which may have had knock on effects on dictating which new germplasm was subsequently useful for addition to the crop genepool at that location. In some instances, domesticated plants came into contact and hybridized through cultivation in new locations with previously isolated wild or cultivated relatives. The apple, Malus pumila, was originally cultivated in the region between the Caspian and Black seas and as its cultivation spread to new locations it hybridized with wild relatives, such as M. sylvestris (Europe) and M. baccata (East Asia). Interspecic hybridization with wild species following human migration also played an important role in developing better adapted germplasm of grape (Vitis vinifera) and peach (Prunus persica). The modern strawberry cultivars are the result of the accidental cross in France between two octoploid species introduced from the Americas Fragaria chiloense and F. virginiana. In spite of subsequent introgression of additional genetic diversity by breeders, the modern cultivated strawberry genepool is quite narrow, given the effect of polyploidization and long-distance human dispersal (Graham et al., 1996). For

40

C. Spillane and P. Gepts

Table 2.3. Some molecular genetic studies which demonstrate a narrower cultivated genetic base in one region/race relative to another. Crop Cocoa Scientic name Narrower genetic base Broader genetic base region/race/type region/race/type Forastero, Criollo and Trinitario types Papua New Guinea cultivars Other USA hexaploid wheat cultivar classes A. cruentus Reference Lerceteau et al., 1997; Whitkus et al., 1998 He et al., 1995 Chen et al., 1994 Transue et al., 1994

Theobroma cacao Nacional type USA cultivars

Sweet potato Ipomoea batatas Wheat Amaranth

Triticum aestivum USA hard red spring wheat cultivars Amaranthus spp. A. hypochondriacus A. caudatus

example, only two plants of F. chiloense were introduced initially in France (Sauer, 1993). Many modern plantation crops such as oil palm, coffee, cocoa and rubber have relatively narrow genetic bases due to inadequate initial genetic diversity sampling (Sauer, 1993; Smartt and Simmonds, 1995). Phenology is an important factor in the success of new crop germplasm introductions. Phenology refers to the study of how periodic biological events such as owering, seed dormancy, etc. relate to climate and other environmental factors. Germplasm from different latitudes and altitudes often has specic temperature and photoperiod requirements and hence will not ower when grown in new locations. Soybean provides an interesting example of a daylength-sensitive crop for which eastwest movement is much easier than northsouth. Therefore, movement of soybean across latitudes has been very limited (Hymowitz and Kaizuma, 1981). Novel germplasm can only be of utility in a new location if its phenology is well matched to the resources and constraints of its new production environment (Shorter et al., 1991). The timing of phenological events is generally modulated by responsiveness to photoperiod and temperature, with much variation in responsiveness between genotypes (Summereld et al., 1996). Yet, durations from sowing to owering are of critical importance if crops are to have the potential to yield well in a new environment (Summereld et al., 1997). Efforts to broaden the genetic base of lentil in South Asia have had to address phenology as a barrier to geneow (Erskine and Muehlbauer, 1991; Erskine, 1997). In many instances, unless such germplasm is subjected to sexual recombination (or gene transfer) and then re-selected, the genetic variability is not usually accessible in the new location. It is now known that important phenological traits may not necessarily be under complex genetic control. For instance, owering time genes have a major effect on adaptation of genotypes to specic environments. In wheat, these include the genes that control vernalization (Vrn genes), response to photoperiod (Ppd genes) and developmental rate genes (Snape, 1996). There are multiple alleles at all of these loci: for example, the winter versus spring habit in wheat varieties is controlled by one to four major loci; in Pisum, four major loci control owering time (Snape et al., 1976; Murfet, 1977). As a result of increased understanding of the underlying genetics of phenological traits, there may now be some potential to ne-tune the owering time of

The Dynamics of Crop Genepools

41

crops to specic environments (Laurie, 1997). For instance, Snape et al. (1995) have converted a UK winter wheat into a spring wheat by transferring a vernalization-insensitive allele at the Vrn1 locus. Such work may open new routes for geneow between crop genepools previously isolated as a result of phenological barriers.

Disease-induced genetic bottlenecks Even where several resistance genes are available for combatting specic pests or pathogens, there is often a reliance at the eld level on a few such genes when farmers exhibit preferences for the most successful varieties (Heisey et al., 1997). This results in a risk of crop genetic vulnerability if large areas are planted to a few of the most popular crop varieties which have identical genes for resistance against particular pests or diseases (Smale, 1997). In such instances, neo-virulent pathogens that arise through natural selection or immigration can lead to serious losses of all susceptible genotypes (Mundt, 1994). Crop epidemics due to genetic vulnerability can lead to genetic bottlenecks at the regional or local level and the loss of locally adaptive traits in susceptible varieties. Such an event occurred in the USA when a chestnut blight (Cryphonectria parasitica) epidemic destroyed large numbers of chestnut trees which exhibited signicant overall genetic heterogeneity but were uniformly homogenous at susceptible genetic loci (Anagnostakis, 1982). European and North American potatoes are also likely to have passed through a major (Phytopthera infestans) disease-induced bottleneck in the 1840s (Simmonds, 1993). For some important crops there is little to no effective resistance in the cultivated genepool against important pests and diseases. Examples include leaf roll virus in potato (Corsini et al., 1994), black sigatoka in plantains (Swennen and Vuylsteke, 1991), bruchids and weevils in coffee, rice grassy stunt virus (Swaminathan, 1982), rice hoja blanca virus (Madriz et al., 1998) and bean golden mosaic virus (Hidalgo and Beebe, 1997). For instance, an epidemic of African cassava mosaic virus (ACMV) suddenly occurred in Madagascar in 19341936 in which all of the susceptible local varieties were largely destroyed (Cours et al., 1997). Sufcient levels of ACMV resistance were not found in the available Manihot esculenta genepool (Jennings, 1994; Cours et al., 1997). Subsequently, this disease-imposed bottleneck necessitated an ultimately successful exercise to develop ACMV-resistant varieties by interspecic crossing of cultivated cassava (M. esculenta) with ceara rubber (M. glaziovii). However, ACMV still causes serious crop losses in much of sub-Saharan Africa and has recently reached epidemic levels in Uganda (Otim-Nape et al., 1997). To counter such problems, Lenn and Wood (1991) have emphasized the importance of broadening the genetic base of disease resistance in crops, including through the use of wild germplasm and gene deployment strategies (Smale, 1997).

Geneow in and between farmers elds There have been few systematic studies of geneow at the farmers eld level as either a result of germplasm exchange between farmers or other modes of transmission (Brush, 1995; Louette et al., 1997). However, farmer management of crop varieties can be

42

C. Spillane and P. Gepts

highly dynamic and open systems with a high turnover of local and introduced germplasm over even a few crop generations (Louette et al., 1997; Rice et al., 1997). The lack of studies of this process may reect the complexity of undertaking such studies as such geneow can be conditioned by biological, physical and social factors. Yet there is much potential for such local-level geneow occurring whether it is between landraces, modern varieties and/or sympatric wild relatives. Little of predictive scientic value is known of how farmers may consciously or unconsciously affect (or effect) local-level geneow. Yet recent genetic studies of farmers landraces of pearl millet in West Africa have demonstrated that the crop management practices of neighbouring farmers led to the selection of different genotypes of the same named landrace and similar genotypes of different named landraces (Busso et al., 2000). In this instance, while the phenotypic characteristics which identied a landrace were maintained across farmers, it was found that the genotypes of two different landraces grown by the same farmer were more similar than those of the same landrace grown by two different farmers (Busso et al., 2000). Both formal and informal (farmer-to-farmer) seed distribution systems are likely to have a major impact on crop geneow within any country or region (Cromwell et al., 1993; Sperling et al., 1996; Sthapit et al., 1996; Witcombe et al., 1996). Genetic bottlenecks to such geneow can occur because of social factors such as income status, gender, family, ethnicity or other social groupings. Local-level crop geneow can be affected by the relative wealth of farmers (Ferguson, 1992; Sperling et al., 1996). Factors such as harsh topography, lack of transport or communications can limit geneow between regions, and farmers who live in peripheral regions are likely to have less access to novel germplasm (Brush and Meng, 1998). A study of the partitioning of genetic diversity in Andean potato landraces demonstrated high levels of geneow between commercial landrace populations as a result of seed tuber exchange among farmers. This geneow was of such magnitude that it counteracted ecogeographic partitioning and local adaptation (Zimmerer and Douches, 1991). However, lower levels of geneow were inferred for those potato types used solely for subsistence (Zimmerer and Douches, 1991). In some instances, geneow can occur between introduced modern varieties and local landraces, leading to the rustication of the introduced varieties (Wood and Lenn, 1996; Louette et al., 1997). While the detrimental effects of weeds on agriculture are well known, little is known of the evolutionary or agronomic role of weed forms or companion weeds. These could act as natural bridges to useful geneow between conspecic wild progenitors and crops (Pickersgill, 1981; Harlan, 1992; Wood and Lenn, 1996; Linder et al., 1998). In most modern agricultural practices, companion weeds and crops are now only likely to coincide at the boundaries of the cultivated crops (Cubero, 1997). There are some examples of traditional cultural practices that can promote geneow between fruit crops and their wild relatives (Zohary and Hopf, 1993; Hanelt, 1997). Conversely, farmers can actively select against any crop wild relative hybrids because of their undesirable seed, phenological and other characteristics (R. Papa, A. Delgado Salinas, J. Acosta and P. Gepts, unpublished results). These traits are likely to show up already in the F1 generation because they are generally controlled by dominant genes. If undesirable hybrids can be recognized by farmers before owering, negative selection can strongly limit geneow towards the crop. If such selection can only operate during and after owering, selection will only be partially effective. The existence of farmer

The Dynamics of Crop Genepools

43

selection can actually lead to asymmetric geneow whereby geneow from the crop has a stronger effect than geneow to the crop (R. Papa and P. Gepts, unpublished results). The breeding system of a farmers crop can have a strong inuence on geneow at the local level (Allard, 1975). While plant species may range from obligate outcrossers to completely selfed, most species will lie somewhere in between. However, even strongly self-pollinating crops such as rice, barley, wheat and oats will have some degree (e.g. 15%) of outcrossing (Allard and Kahler, 1971). Local geographic separation of populations can therefore be a signicant barrier to geneow between different populations of the same crop species. In plants, geneow is usually effected by two carriers: seed and pollen. The bulk of pollen from both wind- and insect-pollinated plant species can often only travel 510 m (Golenberg, 1987). On the other hand, pollen of some species (e.g. beet) can be dispersed at distances up to 1 km (Ellstrand and Hoffmann, 1990; Klinger et al., 1991). Unaided seed dispersal is often limited to several metres (Levin and Kerster, 1969), yet much seed dispersal is via articial means (animals, trade, soil, transport) as has been well demonstrated for companion weeds of crops (Mack, 1991; Boudry et al., 1993). Even within a population, limited geneow may result in a highly substructured population in which only adjacent plants have a high likelihood of mating. Local-level geneow within plant populations can and does occur (Ellstrand, 1992; Hamrick et al., 1995; Ellstrand et al., 1996). As expected, the extent of local-level geneow between populations is likely to be affected by the relative sizes of each population (Klinger et al., 1991). It is possible that farmers manipulate distance between crops to regulate undesirable geneow.

New strategies for promoting useful geneow between exotic and adapted genepools Hallauer and Miranda (1981) cogently dened exotic germplasm as all germplasm that does not have immediate usefulness without selection for adaptation to a given environment. The denition of what is adapted or exotic material within the cultivated primary genepool is perspective dependent i.e. what is adapted germplasm for one grower may be unadapted for another depending on their relative environmental locations. The denition of exotic germplasm therefore also implies a parallel denition of an environment where an extreme or unsuitable environment is one to which the germplasm in question is poorly adapted. Both elite and landrace material within the primary genepool can be adapted or unadapted (exotic) depending on the environment in question. However, regarding the utility of wild germplasm to farmers or breeders, all such germplasm is exotic according to the denition of Hallauer and Miranda (1981). Most plant breeders are reluctant to utilize exotic or unadapted material due to its initial detrimental effects on elite breeding material (Kannenberg and Falk, 1995). Breeders seldom explore much of the genetic variability contained within wild relatives for useful traits since the use of such unadapted materials presents major practical problems in breeding (Duvick, 1996). Crosses with exotic material can result in the concurrent introduction of inferior alleles and disruption of useful co-adapted gene complexes in the elite material. Even after 20 or more years of conventional breeding, a single gene transferred by backcrossing from a wild species could be associated with enough linked chromosomal DNA to contain more than 100 other potentially undesirable genes

44

C. Spillane and P. Gepts

(Young and Tanksley, 1990). Exotic germplasm, whether cultivated or wild, can negatively affect adaptedness when introduced into a locally adapted genetic base. The linkage drag inherent in elite or locally adapted breeding material is a major constraint to increased utilization of unadapted or exotic germplasm over the longer term. Although it may be desirable over the longer term, there are major barriers (both genetic and practical) to promoting useful geneow from exotic genepools into the adapted genepools of different breeders and farmers. Before exotic germplasm can be of utility, in many instances it must undergo conversion or pre-breeding such as was performed for sorghum and maize in the USA (Simmonds, 1993). Despite such isolated efforts, it is thought for most crops that the genetic gap between elite adapted genepools and exotic pools is growing larger with each breeding cycle (Holley and Goodman, 1988; Martin et al., 1991). Yet it is well known that exotic germplasm contains useful alleles that could further improve elite germplasm or be used to develop de novo adapted germplasm (Harlan, 1976; Hawkes, 1977; Stalker, 1980; Rick, 1982). For instance, Uhr and Goodman (1995a,b) were able to identify maize inbreds derived from tropical germplasm that had inbred or testcross yields comparable with that of temperate inbreds. Better knowledge of the underlying genetic differences between adapted and exotic germplasm could help to overcome such barriers to geneow. The genetic basis of adaptation is poorly understood (Barton and Turelli, 1989; Orr and Coyne, 1992; Milligan et al., 1994; Hawtin et al., 1996). The phenotypic traits conferring adaptation to novel environments were traditionally considered to have a polygenic basis composed of many loci of small effect (e.g. Falconer, 1989). However, it is now accepted that often a few of these loci can account for a major portion of the observed genetic variance (e.g. Edwards et al., 1987). Such loci have been termed quantitative trait loci (QTL) because they control a quantitatively inherited trait (e.g. Tanksley, 1993). The recent use of DNA marker-assisted mapping in conjunction with new statistical methods (e.g. Lander and Botstein, 1989) has provided an efcient means of identifying major-effect QTLs underlying quantitative agronomic traits (e.g. Hayes et al., 1993; Nodari et al., 1993; Lu et al., 1997; McMullen et al., 1998). While little is yet known of the genetic origins or basis of adaptedness, it is now considered that marker-assisted dissection of the genetic basis of adaptedness is feasible (Allard, 1996). Some important adaptive traits (owering time, vernalization requirement, etc.) in both domesticated and wild plants can be due to a few genes or QTLs of major effect. This has been found for both domesticated (Ferreira et al., 1995; Teutonico and Osborn, 1995) and wild plants (Coupland, 1995; Mitchell-Olds, 1996; Kuittinen et al., 1997). For instance, comparative QTL analysis has demonstrated that the conversion of exotic sorghum races into more agronomically desirable genotypes could largely be attributed to fewer than ten QTLs imparting reduced height or earlier owering (Lin et al., 1995). However, there are also some instances where many QTLs have been found to be responsible for such adaptive traits (Jansen et al., 1995). While QTL mapping holds great promise for crop improvement, such techniques require adequate levels of genetic variation to work optimally. Most of the breedingrelated QTL studies undertaken to date have been based on the manipulation of quantatitive trait variation already existing within elite or adapted germplasm (Tanksley and Nelson, 1996). Frequently, elite germplasm (e.g. in self-pollinated crops) has reduced

The Dynamics of Crop Genepools

45

levels of polymorphism making QTL analysis difcult (Helentjaris et al., 1985; Miller and Tanksley, 1990; Wang et al., 1992; Anderson et al., 1993). A more long-term and broader focus on the germplasm starting materials (including exotic germplasm) is necessary to avoid a situation where molecular markers are being used to manipulate the same alleles that breeders have been manipulating for many years through classical breeding procedures (Tanksley and Nelson, 1996). One approach may be more proactively to select genetically diverse genotypes from existing or QTL-mapping tailored crop core collections for inclusion in such QTL analysis studies (Van Hintum et al., 1998). Where previously the use of exotic germplasm was limited mainly to prospecting (through evaluation) for monogenic qualitative traits such as disease resistance genes (e.g. Lenn and Wood, 1991), it is now clearly the case that exotic wild germplasm can contain useful loci for many quantitative agronomic traits (Tanksley and McCouch, 1997). It has been known for some time that such hidden or cryptic wild loci contributing to quantitative traits cannot be screened for unless evaluated in the genetic background of the cultivated genepool i.e. cultivars or breeders lines (Frey et al., 1984; Bramel-Cox and Cox, 1988; Vetelainen, 1994; Tanksley and McCouch, 1997). It is most likely that wild relatives, regardless of their phenotype, contain alleles that can improve most quantitative traits of interest (Frey, 1975; Cox et al., 1984; Singh et al., 1995; Tanksley and McCouch, 1997). Unfortunately, until recently there has been no effective strategy for identifying and recombining the few agronomically favourable alleles, which are masked by the preponderance of undesirable alleles, in wild relatives (Tanksley and McCouch, 1997). However, exciting new strategies, combining an appropriate mating system with marker-assisted selection, are now emerging for the identication of agronomically important QTLs in phenotypically inferior wild relatives and for their precise transfer into the primary genepools of crops (Haghighi and Ascher, 1988; deVicente and Tanksley, 1993; Anderson et al., 1996; Tanksley and Nelson, 1996; Tanksley et al., 1996; Tanksley and McCouch, 1997). The selection of superior lines may also be aided by the availability of molecular markers that are linked to genes for the quantitative trait of interest or to deleterious traits (lack of viability or fertility, lack of adaptation, etc.). Positive selection for the former (foreground selection) and selection against the latter (background selection) can accelerate the selection process (Tanksley et al., 1989; Paterson et al., 1991; Edwards, 1992; Lee, 1995). This innovative use of molecular maps and markers is likely radically to alter and improve the way that exotic germplasm is utilized in plant breeding and genetic enhancement in the decades ahead (McCouch, 1998).

Geneow constraints in plant breeding Plant breeding as a science is in a constant state of ux in its efforts to maintain or increase food production levels in the face of continually changing biotic and abiotic stresses. All plant breeding is based upon the application of articial selection pressures to select improved germplasm from unimproved (Duvick, 1996). By its very denition, such selection, whether natural or human selection, generally leads to a reduction in the genetic diversity of the population undergoing selection at a particular location.

46

C. Spillane and P. Gepts

This is a necessary aspect of plant breeding and crop improvement (e.g. Burton et al., 1990; Marocco et al., 1992). Indeed, high selection intensity breeding of crop varieties may have narrowed the adapted or cultivated genepool of many of the major crops (Tanksley and Nelson, 1995). Such a phenomenon may be especially acute in selfpollinated crops where the level of genetic variation in cultivated varieties is often only a small fraction of that available in the overall genepool (Miller and Tanksley, 1990; Wang et al., 1992). Recent genetic diversity studies of the varieties in production of crops as diverse as strawberry (Graham et al., 1996), hard spring wheat (Mercado et al., 1996), rice (Yu and Nguyen, 1994) and peppers (Izioka et al., 1997) have indicated a general need to broaden the genetic base of the crops concerned. A number of comparative studies have shown greater levels of genetic diversity in landraces compared with modern cultivars (Sonnante et al., 1994; Jo et al., 1997). For example, a comparison of the levels of gene diversity in US commercial common bean cultivars gives an indication of the high level of selection intensity that have been applied to the primary genepool of bean in the USA (Sonnante et al., 1994) (see Table 2.4). The existence of a relatively broader genetic base for less productive landraces may not on its own be sufcient justication to warrant the broadening of the genetic base of more productive elite germplasm. A wider range of factors (e.g. yield plateaus, lack of useful traits, past bottlenecking events, recurrent disease problems) should be considered in deciding whether the elite germplasm of a crop requires aggregate infusions of novel genetic variability, or not. Sonnante et al. (1994) show that modern plant breeding per se is not the only culprit in the reduced genetic diversity levels in the primary genepools of crops. The domestication of crops by the earliest farmers is likely to have involved only a small sample of the available genetic diversity in the wild relative genepool(s) from which the crop was selected. Informal mass selection efforts by farmers followed by widespread adoption of popular landraces can also have a narrowing effect on genetic diversity. Areas such as secondary centres of diversity which exhibit high levels of phenotypic diversity may in some instances actually have low levels of genetic diversity (Wood and Lenn, 1996). The data presented in Table 2.4 are based on molecular diversity analyses. When considering phenotypic data (growth habit, seed colour, etc.), a trend in the opposite direction is often observed, namely cultivated materials can show increased phenotypic diversity compared with their wild counterparts. This discrepancy can be attributed to genome sampling effects. Phenotypic data for genetic diversity are often (but not exclusively) based on traits with high heritability (e.g. seed weight) or controlled by major genes. This type of inheritance probably reects the type of selection practised during domestication. In contrast, unlinked molecular marker loci are generally neutral (i.e. they have a small or neutral phenotypic effect) and are therefore much less likely to sur-

Table 2.4. Levels of gene diversity in US commercial common bean germplasm classes and Latin American germplasm from which they were derived (Sonnante et al., 1994). Genepool Middle American Andean Wild ancestor 0.24 0.20 Landrace group 0.09 (Durango) 0.17 (Nueva Granada) US commercial class 0.00 (Pink) 0.06 (Kidney)

The Dynamics of Crop Genepools

47

vive in populations than genes with a strong selective advantage (Crow and Kimura, 1970). Molecular markers, because of their small or neutral phenotypic effect, are therefore much more representative of the majority of genes in the genome that have small phenotypic effects. Therefore, the downward trend revealed by molecular markers is thought to represent more accurately the reduction in genetic diversity in crop genomes. Initial successes with conventional plant breeding, coupled to the fact that genetic progress can be made on a relatively narrow genetic base, are likely to lead to inertia effects in the pedigree history of modern varieties (Duvick, 1984; Rasmussen and Phillips, 1997). The inertia effect of existing adapted germplasm can have a major effect on what exotic germplasm can now be chosen for recombination with existing adapted germplasm. Of the 250300 described races of maize, only one forms the basis of US agriculture, 2% provide the foundation for the temperate zones and less than 5% have been used in maize breeding globally (Goodman, 1985). The estimation of general (GCA) and specic combining ability (SCA) can dictate what germplasm can be used for incorporation of multigenic traits (Grifng, 1956; Holland and Goodman, 1994). Some heterotic groups of inbreds are used preferentially over others in hybrid seed production (Mumm and Dudley, 1994). It is also likely that parents used for initial crosses were chosen for their adaptation traits of most utility in the dened target environment. In many cases plant breeders have conned their efforts to using genetic variation within dened intraspecic groups. Breeding of winter wheat has been mainly done by using only winter germplasm and breeding of spring wheat by using only spring germplasm. Similar examples occur in other crops, such as rice, groundnut, common bean, chickpea, rape seed and mustard. In European barley, a few studies have demonstrated that genetic variation in the cultivated genepool has not increased signicantly since scientic plant breeding began (Linde-Laursen et al., 1987; Fischbeck, 1992; Melchinger et al., 1994). This was considered to be mainly due to the use of a few superior landraces as the founder materials (in the 1880s1920s) and the further recurrent use of genetically derived elite cultivars since (Fischbeck, 1992). Since 1900, no more than 20 sources of exotic germplasm have contributed to varieties that make up the largest proportion of US barley acreage (Hayes et al., 1997). Similarly, the genetic base of the US rice industry is considered to be narrow and can be largely traced to 22 plant introductions in the southern rice belt and to 23 plant introductions in the western rice belt (Xu et al., 1997). Japanese rice cultivation is also considered to have a relatively narrow genetic base (Nakagahara et al., 1997). A recent study of soft winter wheat (T. aestivum) breeding lines showed that the eastern US soft winter wheat genepool had been developed from a narrow genetic background, mainly due to the recurrent use of a small number of the same parents in breeding (Kim and Ward, 1997). Diversity studies of cotton (G. hirsutum) cultivars from Pakistan (Iqbal et al., 1997) and Australia (Multani and Lyon, 1995) have indicated high levels of genetic similarities between most of the varieties tested. The level of genetic diversity found in a range of G. hirsutum cultivars was also found to be low (Brubaker and Wendel, 1994). Most such examples relate to comparatively narrow genetic bases in specic environments or regions as identied by molecular diversity analyses at a particular point in time. However, for some crops and regions there are examples based on pedigree analysis

48

C. Spillane and P. Gepts

over time which suggest that the numbers of exotic landraces introgressed into elite germplasm can be increasing over time. CIMMYTs International Wheat Information System (IWIS) has yielded data that show that the number of parental landraces in CIMMYTs bread wheats has increased over the past four decades, from six ancestors in Yaqui 50 to 68 ancestors in Weaver. The IWIS has also been used to show that the cytoplasmic diversity in CIMMYT wheats was restricted and that landraces from certain regions of wheats centre of origin rarely appear in the pedigrees of modern wheats (Dubin et al., 1997). It has also been shown that in the major wheat-growing areas of the world, the concentration among leading cultivars has tended to decline as agricultural research and seed systems have matured (Smale and McBride, 1996; Smale, 1997). Pedigree analysis, based on systems such as IWIS, is a highly useful and costeffective tool for designing rational strategies to broaden the genetic base of cultivated genepools, where such broadening may be needed. In plant breeding it is likely that selection intensity is correlated to some extent with the time horizon the plant breeder faces in providing germplasm that is marketable or acceptable to his or her client farmers. In the past, public-sector plant breeders may have had a longer time horizon than private-sector breeders and subsequently could play a larger role in pre-breeding and genetic enhancement. Ironically, just at the time when improved research tools for genetic enhancement and pre-breeding are becoming increasingly available, declining public sector support for plant breeding may result in a serious and widening pre-breeding gap between the exotic germplasm in genebanks and the elite germplasm actually used in breeding programmes (FAO, 1998).

Recombination rates the effect of mating and selection strategies on geneow Mating and selection strategies employed in breeding affect the evolutionary trajectory of primary crop genepools over time. Breeding (or mating) strategies depend on the reproductive biology of the crops and generally on what has been successful in the past (Jensen, 1988). The practical recombination of genetic variability is an issue that can be neglected relative to selection and isolation in plant breeding. Simmonds (1962) stated that the recombination phase of a breeding programme is often throught of as something to be passed as quickly as possible by rigorous inbreeding or backcrossing. Stebbins (1959) argued that, because of the low endogenous mutation rate, genetic recombination was the most likely source of novel genetic variability for breeding and that recombination-generated diversity could be maximized by hybridization between populations with different adaptive norms. Intragenic recombination coupled to epistastic effects has been suggested to be an important source of novel and potentially useful genetic variability in developing elite germplasm (Schnable et al., 1996; Rasmusson and Phillips, 1997). Crop plant genomes will differ in their permeability to introgression of different genes or chromosomal regions by wide cross recombination with different wild relatives (Rieseberg et al., 1996). This is a problem for accessing exotic rice germplasm (Brar and Khush, 1997). Recombination rates are typically reduced in crosses between divergent taxa of maize/teosinte (Doebley and Stec, 1993), rice species (Causse et al., 1994) and tomato species (Paterson et al., 1988). In addition, inbreeding following a cross may

The Dynamics of Crop Genepools

49

also reduce recombination rates (Srivastava, 1980). The movement of alleles across reproductive barriers can be enhanced by increasing recombination rates between parental linkage blocks (Hanson, 1959a,b; Wall, 1970; Rieseberg et al., 1996). The fact that some mating strategies promote higher levels of recombination than others has important implications for genetic enhancement and genetic base-broadening schemes if inadvertent bottlenecks to geneow are to be avoided. Recurrent backcrossing is often considered the most inefcient mechanism for breaking up parental linkage blocks (Hanson, 1959a,b). Two mating designs that enhance recombination between parental genomes are selng and sib-mating. Unfortunately, populations resulting from either selng or sib-mating will generally exhibit lower levels of fertility than progeny from recurrent backcrossing (Wall, 1970). Mating designs employing one or more generations of sib-mating interspersed with backcrossing may be an effective strategy for efcient geneow from wild relatives into the primary genepool (Wall, 1970; Rieseberg et al., 1996). Because the two parents are usually quite different, a backcrossing programme to one or the other parent is generally necessary to increase the overall frequency of favourable alleles (Dudley, 1982). Such mating strategies may be important for genetic enhancement or prebreeding, especially in situations where the resources to conduct marker-assisted introgression are not currently available. Recombination rates are also a vital consideration in the development of inbred lines in breeding for heterosis. Use of slow rates of inbreeding via sib-mating strategies which promoted higher recombination rates has been considered to be a more effective method than selng for extracting inbred lines from tropical maize germplasm (Holley and Goodman, 1988). However, mating strategies based on sib-crossing will disrupt favourable coadapted multilocus combinations more than selng and hence require longer periods of breeding to reselect for more favourable multilocus combinations (Allard, 1996). This has led to suggestions to combine backcross and inbreeding to retain useful parental gene combinations while concurrently generating new, potentially useful gene combinations. Additional backcrossing schemes that have actually been used include the recurrent, inbred and congruity backcross methods: 1. In the recurrent backcross, as illustrated by the introgression of small seed size from Glycine soja into G. max (LeRoy et al., 1991), successive backcrosses are performed to the same parent with or without selection between backcross generations. This mating and selection scheme is more effective for traits with higher heritability, which are often mono- or oligogenic and segregate in qualitative fashion. 2. In the inbred backcross method, a limited number of backcrosses (usually one to three) are performed, followed by several generations of selng (Wehrhahn and Allard, 1965). The inbred backcross method results in a population of more than 50 lines that are homozygous, have a common genetic background similar to that of the recurrent parent, but differ for specic genomic regions inherited from the unadapted donor parent. Because the lines can be used in replicated trials, this method has been used successfully to transfer quantitative traits from exotic germplasm into elite, adapted germplasm as illustrated by improvement of nitrogen xation and seed protein content in common bean (Bliss, 1985). 3. The congruity backcross, originally proposed by Haghighi and Ascher (1988), involves backcrossing to both parents in alternate generations. This method was used to recover progeny with increased fertility from P. vulgaris P. acutifolius crosses that are

50

C. Spillane and P. Gepts

normally highly sterile (Haghighi and Ascher, 1988). In the progeny of these crosses, it was possible to obtain progenies with intermediate seed morphologies rather than morphologies resembling one or the other parent (Anderson et al., 1996). The recovery of these morphologically intermediate yet viable and fertile individuals was attributed to the increased levels of effective recombination through the increased levels of heterozygosity. Urrea and Singh (1995) compared the inbred backcross and the congruity backcross in an intraspecic P. vulgaris cross between two diverged common bean cultivars, belonging to races Mesoamerica and Durango, respectively, of the Mesoamerican gene pool. Their results showed that the congruity backcross generated progenies which were intermediate between the two parents, whereas the inbred backcross led to progenies resembling very much the recurrent parent even after a single backcross. Their results suggested that further recombination among advanced, superior backcross lines may be needed to ensure sufcient recombination, as illustrated by St Clair and Bliss (1991).

Factors contributing to bottlenecks in crop cytoplasmic genetic diversity The genomes of cytoplasmic organelles (chloroplast and mitochondria) of most crop plant species also exhibit signicant genetic diversity (e.g. Kemble et al., 1983; Komarnitsky et al., 1990; Belhassen et al., 1991) and hence can also be subjected to genetic bottlenecks, drift and founder events (Rieseberg and Soltis, 1991). In most angiosperms, cytoplasmic organelles are maternally inherited (Sears, 1980). In general, chloroplast DNA has proved to be less genetically variable than mitochondrial DNA (Palmer, 1987). Repetitive use of the same female parents in breeding programmes can lead to a bottleneck in cytoplasmic genetic diversity among cultivars, where all cultivars share similar maternally inherited cytoplasmic DNA. For instance, in the case of all commercial sugar cane (Saccharum ofcinarum) hybrids, cytoplasmic genetic diversity is derived only from S. ofcinarum. Nuclear genetic diversity is derived from S. ofcinarum and one or more of S. spontaneum, S. sinense, S. barberi and S. robustum (Irvine, personal communication). Although S. spontaneum contains the most signicant levels of cytoplasmic genetic diversity, it has rarely been used as a female parent in crosses. Cytoplasmic bottleneck effects may be common in crops where the cytoplasmic genomes can, among many other traits, confer cytoplasmic male sterility (cms), which in conjunction with nuclear restorer genes is of immense practical value in plant breeding (Horner and Palmer, 1995). For instance, in maize three different genetic types of geniccytoplasmic male sterility (g-cms) have been described (g-cms-T, g-cms-S and gcms-C) on the basis of specic nuclear restorer genes capable of restoring fertility (Horner and Palmer, 1995). While such g-cms traits are extremely useful, for example in recurrent selection or hybrid seed production, the use of the same male sterile parents in pedigree breeding schemes can lead to a situation where many cultivars contain genetically identical cytoplasmic genomes. The geographic distribution of genes for fertility restoration in hybrid rice based on cytoplasmic male sterile germplasm has been studied from this perspective (Li and Zhu, 1989). The US maize blight epidemic of 1969/70 (where 15% of the crop was

The Dynamics of Crop Genepools

51

lost) was caused by the use of the same Texas (or T) cytoplasmic male sterility in developing many maize varieties which were inadvertently uniformly susceptible to a race of the southern leaf blight fungus, Helminthosporium maydis, as a result of a mitochondrial disease susceptibility locus (NRC, 1972). More recently, breeding schemes that limit the contribution of the cytoplasm of the male sterile to the progeny have been devised and are important in avoiding breeding-induced cytoplasmic bottlenecks (Horner and Palmer, 1995). Nevertheless, F1 rice hybrids today grown over 18 million hectares in China contain a single source of cytoplasmic male sterility (WA) (R.S. Paroda, personal communication). Recent analyses of genetic variation in the cytoplasmic genomes (chloroplast, mitochondria) of Triticum (wheat) and Aegilops have suggested that Aegilops speltoides is the cytoplasmic donor/progenitor (i.e. female parent) of all polyploid wheats (Wang et al., 1997). Interestingly, the same study showed that the evolution of the two organellar genomes has been parallel only to a limited extent, even though they have been maternally inherited as a set. Similarly, a study of chloroplast DNA diversity in populations of wild and cultivated barley indicated that the great majority of the cultivated samples tested had chloroplast DNA derived from only one wild progenitor (Neale et al., 1988).

Changing agricultural practices and consumer preferences The changing nature of food supply, consumer preferences and agricultural modernization can also affect the levels of genetic diversity in crop production. In many instances, this is inevitable and necessary, especially from a farmers perspective. Consumer and trader preferences for uniformity of produce that has been graded into particular categories can exert selection pressures for particular sizes, shapes, colours, tastes, etc., of crop produce. Consumer preferences which are developed over generations in any given sociocultural context can also have a signicant effect on which alleles (or allelic combinations) contributing to quality traits such as palatability, cooking time and storability before and after cooking are acceptable to different groups of consumers. It may often be difcult to introduce novel germplasm which does not meet such consumer criteria. For instance, tequila is made from a single clone of agave away from which tequila producers are reticent to diversify genetically in case tequila quality suffers. Selecting for specic genes which confer quality traits can lead to overall reductions in allelic diversity at particular loci in a crop genepool. This may account for the regional predominance of only a few glutenin subunits which are controlled by specic alleles on the long arm of wheat group 1 chromosomes (Morgunov et al., 1993; Pena, 1995). Similarly, the quality requirements for good malting barley are wholly different from those for good feed barley. The special quality requirements for malting barley have been shown to have signicant effects on the genetic diversity of malting barley cultivars (Wych and Rasmusson, 1983; Saghai-Maroof et al., 1994). On the other hand, consumer preferences for a steady supply of locally produced fresh products can result in the selection of a wide range of maturity times in certain crops (e.g. Brassica). For example, the cauliower crop in Italy occurs as distinct types in several regions, differing in curd colour and conformation, and in leaf and plant type (Crisp and Astley,

52

C. Spillane and P. Gepts

1985). These examples indicate that overall breeding-induced reduction in genetic diversity is due in part to producer- and consumer-driven requirements. Due to globalization and comparative advantages in crop production, crop products are increasingly traded and hence transported over large distances. This necessarily imposes selection pressures for non-spoilage of transported varieties/produce, with a concomitant bias against genetic variation conferring early or sequential ripening, which may not be such a problem for produce that is quickly consumed or processed at a more local level. Agricultural mechanization involving more precise seeding and harvesting requires varieties with more uniform phenotypes to work efciently, and largescale cultivation of such varieties is necessary to realize economies of scale relating to the use of such machinery. The standardization of agricultural machinery for harvesting by institutions such as the International Standards Organization (e.g. ISO 6689 on combine harvesters or ISO 11520 on grain dryers) may have selection effects on the phenotypic requirements of crop varieties (Duranton and Abram, 1996). While some types of food processing (e.g. malting barley for brewing) may require strict phenotypic standards, others such as starch production may require less stringent phenotypic criteria. Hence, some types of food processing may promote genetic diversity in crop production.

Conclusions
Many factors conspire to reduce the genetic diversity of crop genepools. While diversity is not a breeding objective per se, it is useful to remember that genetic diversity needs to be maintained or enhanced in order to obtain signicant future progress from selection. Given the small fraction of total genetic diversity that has been included since domestication in the domesticated genepool of most crops, there is probably a need to re-domesticate or further domesticate most crops, i.e. to reconstitute or regenerate the domesticated genepools through introgression or incorporation of genetic diversity from exotic, wild and/or cultivated genepools. However, what is not clear is whether the necessary public-sector funding will be available from governments for such longterm germplasm enhancement work. To ensure that enhanced germplasm is actually used, any such pre-breeding or base-broadening work would have to be closely integrated from the outset with existing, more downstream, breeding efforts and seed supply channels. Yet the payoffs to society of long-term germplasm enhancement could be substantial (Hoisington et al., 1999). Of relevance to planning future evolutionary change in crop genepools will be shifts towards substituting genetic resources (genes and gene complexes) for chemical inputs wherever it is possible agronomically (Harwood, 1989; Coffmann and Smith, 1991). Breeding for sustainability is likely to become increasingly a process of tting cultivars to an environment rather than altering the environment by adding inputs such as fertilizer, water and pesticides (Coffmann and Smith, 1991). The benecial social, economic and energy externalities and spillovers of moving agriculture towards a more biological resource intensive activity will be signicant. Since developing a long-term base-broadening programme will incur costs and planning by comparison is cheap, it is essential that any such germplasm-enhancement

The Dynamics of Crop Genepools

53

or base-broadening programmes take into account all available evolutionary and historical information on the crop which is of relevance to the geographical scale of the programme. This applies whether the genetic base broadening is to be undertaken at the global, regional, national or local level.

References
Allard, R.W. (1975) The mating system and microevolution. Genetics 79, 115126. Allard, R.W. (1988) The Wilhelmine E. Key 1987 invitational lecture. Genetic changes associated with the evolution of adaptedness in cultivated plants and their wild progenitors. Journal of Heredity 79, 225238. Allard, R.W. (1990) The genetics of hostpathogen coevolution: implications for genetic resource conservation. Journal of Heredity 18, 16. Allard, R.W. (1996) Genetic basis of the evolution of adaptedness in plants. Euphytica 92, 111. Allard, R.W. and Kahler, A.L. (1971) Allozyme polymorphisms in plant populations. Stadler Symposia 3, 924. Anagnostakis, S.C. (1982) Biological control of chestnut blight. Science 215, 466471. Anderson, D.L. and Miller, J. (1995) The botanical classication of sugar cane. Sugar Journal, December 1995. Anderson, J.A., Churchill, G.A., Autrique, J.E., Tanksley, S.D. and Sorrells, M.E. (1993) Optimising parental selection for genetic linkage maps. Genome 31, 181186. Anderson, N.O., Ascher, P.D. and Haghighi, K. (1996) Congruity backcrossing as a means of creating genetic variability in self pollinated crops: seed morphology of Phaseolus vulgaris L. and P. acutifolius A. Gray. Euphytica 87, 211224. Arakawa, T., Chong, D.K.X. and Langridge, W.H.R. (1998) Efcacy of a food plant-based oral cholera toxin B subunit vaccine. Nature Biotechnology 16, 292297. Arias, D.M. and Rieseberg, L.H. (1994) Gene ow between cultivated and wild sunowers. Theoretical and Applied Genetics 89, 655660. Arias, D.M. and Rieseberg, L.H. (1995) Genetic relationships among domesticated and wild sunowers (Helianthus annuus, Asteraceae). Economic Botany 49, 239248. Baldwin, D., Crane, V. and Rice, D. (1999) A comparison of gel-based, nylon lter and microarray techniques to detect differential RNA expression in plants. Current Opinion in Plant Biology 2, 96103. Barrett, S.C.H. (1983) Crop mimicry in weeds. Economic Botany 37, 255282. Barton, N.H. and Turelli (1989) Evolutionary quantitative genetics: how little do we know? Annual Review of Genetics 23, 337370. Beebe, S., Toro, O., Gonzlez, A., Chacn, M. and Deouck, D. (1997) Wild-weed-crop complexes of common bean (Phaseolus vulgaris L., Fabaceae) in the Andes of Peru and Colombia, and their implications for conservation and breeding. Genetic Research and Crop Evolution 44, 7391. Belhassen, E., Atlan, A., Couvet, D., Gouyon, P.H. and Quetier, F. (1993) Mitochondrial genome of Thymus vulgaris L. (Labiate) is highly polymorphic between and among natural populations. Heredity 77, 462472. Bennetzen, J.L., SanMiguel, P., Chen, M., Tikhonov, A., Francki, M. and Avramova, Z. (1998) Grass genomes. Proceedings of the National Academy of Sciences USA 95, 19751978. Bermejo, J.E.H. and Leon, J. (1994) Neglected Crops: 1492 From a Different Perspective. FAO Plant Production and Protection Series No. 26, FAO, Rome. Bevan, M. and Murphy, G. (1999) The small, the large and the wild: the value of comparison in plant genomics. Trends in Genetics 15, 211214. Bliss, F.A., (1985) Breeding for enhanced dinitrogen xation potential of common bean

54

C. Spillane and P. Gepts (Phaseolus vulgaris L.). In: Ludden, P.W. and Burris, J.E. (eds) Nitrogen Fixation and CO2 Metabolism. Elsevier, Amsterdam, pp. 303310. Bliss, F.A., (1989) Utilization of genetic resources for crop improvement: the common bean. In: Brown, A.H.D., Clegg, M.T., Kahler, A.L. and Weir, B.S. (eds) Plant Population Genetics, Breeding, and Genetic Resources. Sinauer, Sunderland, Massachusetts, pp. 317333. Bonierbale, M., Guevara, C., Dixon, A.G.O., Ng, N.Q., Asiedu, R. and Ng, S.Y.C. (1997) Cassava. In: Fuccillo, D., Sears, L. and Stapleton, P. (eds) Biodiversity in Trust. Cambridge University Press, Cambridge, pp. 120. Bouchez, D. and Hofte, H. (1998) Functional genomics in plants. Plant Physiology 118, 725732. Boudry, P., Morchen, M., Saumitou-Laprade, P., Vernet, Ph. and van Dijk, H. (1993) The origin and evolution of weed beets: consequences for the breeding and release of herbicide resistant transgenic sugar beets. Theoretical and Applied Genetics 87, 471478. Bramel-Cox, P.J. and Cox, T.S. (1988) Use of wild germplasm in sorghum improvement. Proceedings of the 43rd Annual Corn and Sorghum Research Conference 1325. Brar, D.S. (1997) Alien introgression in rice. Plant Molecular Biology 35, 3547. Brown, A.H.D. (1978) Isozymes, plant population genetic structure and genetic conservation. Theoretical and Applied Genetics 52, 145157. Brown, A.H.D. and Clegg, M.T. (1983) Isozyme assessment of plant genetic resources. Isozymes. Current Topics in Biological and Medical Research 11, 285295. Brubaker, C.L. and Wendel, J.F. (1994) Re-evaluating the origin of domesticated cotton (Gossypium hirsutum; Malvaceae) using nuclear restriction fragment length polymorphisms (RFLPs). American Journal of Botany 81, 13091326. Brubaker, C.L., Koontz, J.A. and Wendel, J.F. (1993) Bidirectional cytoplasmic and nuclear introgression in the New World cottons, Gossypium barbadense and G. hirsutum (Malvaceae). American Journal of Botany 80, 12031208. Brush, S. (1995) In situ conservation of landraces in centers of crop diversity. Crop Science 35, 346354. Brush, S.B. and Meng, E. (1996) Farmers valuation and conservation of crop genetic resources. Paper presented at Symposium on Genetic Resources for Agriculture: the Economics of Valuation and Conservation, Tor Vergata University, Rome, 13 May, 1996, pp. 134. Brush, S.B. and Meng, E. (1998) The value of wheat genetic resources to farmers in Turkey. In: Evenson, R.E., Gollin, D. and Santaniello, V. (eds) Agricultural Values of Plant Genetic Resources. CAB International, Wallingford, UK, pp. 97113. Brquez, A. and Sarukhn, J. (1980) Biologa oral de poblaciones silvestres de Phaseolus coccineus L. I. Relaciones plantas-polinizador. Boletin de la Sociedad Botanica Mexicana 39, 525. Burton, J.W., Koinange, E.M.K. and Brim, C.A. (1990) Recurrent selfed progeny selection for yield in soybean using genetic male sterility. Crop Science 30, 12221226. Busso, C.S., Devos, K.M., Ross, G., Mortimore, M., Adams, W.M., Ambrose, M.J., Alldrick, S. and Gale, M.D. (2000) Genetic diversity within and among landraces of pearl millet (Pennisetum glaucum) under farmer management in West Africa. Genetic Resources and Crop Evolution 47, 561568. Butzer, K.W. (1995) Biological transfer, agricultural change and environmental implications of 1492. In: CSSA, International Germplasm Transfer: Past and Present. CSSA Special Publication 23, pp. 129. Carney, S. and Arnold, M. (1997) Differences in pollen-tube growth rate and reproductive isolation between Louisiana irises. Journal of Heredity 88, 545549. Carputo, D., Barone, A., Cardi, T., Sebastiano, A., Frusciante, L. and Peloquin, S.J. (1997) Endosperm balance number manipulation for direct in vivo germplasm introgression to potato from a sexually isolated relative (Solanum commersonii Dun.). Proceedings of the National Academy of Sciences USA 94, 1201312017. Carson, H.L. and Templeton, A.R. (1984) Genetic revolutions in relation to speciation phenom-

The Dynamics of Crop Genepools

55

ena: the founding of new populations. Annual Review of Ecology and Systematics 15, 97131. Causse, M.A., Fulton, M.T., Cho, Y.G., Ahn, S.N., Chunwongse, J., Wu, F.S., Xiao, J.H., Ronald, P.C., Harrington, S.E., Second, G., McCouch, S.R. and Tanksley, S.D. (1994) Saturated molecular map of the rice genome based on an interspecic backcross population. Genetics 138, 12511274. Ceccarelli, S., Erskine, W., Grando, S. and Hamblin, J. (1994) Genotype environment interaction and international breeding programmes. Experimental Agriculture 30, 177187. Ceccarelli, S., Grando, S. and van Leur, J.A.G. (1995) Barley landraces of the Fertile Crescent offer new breeding options for stress environments. Diversity 11, 112113. Chapman, C.G.D. (1989) Collection strategies for the wild relatives of eld crops. In: Brown, A.H.D., Frankel, O.H., Marshall, D.R. and Williams, J.T. (eds) The Use of Plant Genetic Resources. Cambridge University Press, Cambridge. Charlesworth, D. and Charlesworth, B. (1987) Inbreeding depression and its evolutionary consequences. Annual Review of Ecology and Systematics, 18, 237268. Chen, H.B., Martin, J.M., Lavin, M. and Talbert, L.E. (1994) Genetic diversity in hard red spring wheat based on sequence-tagged-site PCR markers. Crop Science 34, 16281632. Cheng, X., Sardana, R., Kaplan, H. and Altosaar, I. (1998) Agrobacterium-transformed rice plants expressing synthetic cryIA(b) and cryIA(c) genes are highly toxic to striped stem borer and yellow stem borer. Proceedings of the National Academy of Sciences USA 95, 27672772. Chu, Y.E. and Oka, H.I. (1972) The distribution and effects of genes causing F1 weakness in Oryza brevigulata and O. glaberrima. Genetics 70, 163173. Coffman, W.R. and Smith, M.E. (1991) Roles of public, industry, and international research center breeding programs in developing germplasm for sustainable agriculture, In: Plant Breeding and Sustainable Agriculture: Considerations for Objectives and Methods. CSSA Special Publication no. 18. Cordesse, F., Second, G. and Delseny, M. (1990) Ribosomal gene spacer length variability in cultivated and wild rice species. Theoretical and Applied Genetics 79, 8188. Corsini, D.L., Pavek, J.J., Martin, W.W. and Brown, C.R. (1994) Potato germplasm with combined resistance to leafroll virus and viruses X and Y. American Potato Journal 71, 377385. Coupland, G. (1995) Genetic and environmental control of owering time in Arabidopsis. Trends in Genetics 11, 393397. Cours, G., Fargette, D., Otim-Nape, G.W. and Thresh, J.M. (1997) The epidemic of cassava mosaic virus disease in the 1930s1940s: lessons for the current situation in Uganda. Tropical Science 37, 238248. Cox, T.S., House, L.R. and Frey, K.J. (1984) Potential of wild germplasm for increasing yield of grain sorghum. Euphytica 33, 673684. Crisp, P. and Astley, D. (1985) Plant Genetic Resources Newsletter. IPGRI, Rome. Cromwell, E., Wiggins, S. and Wentzel, S. (1993) Sowing Beyond the State: NGOs and Seed Supply in Developing Countries. ODI, London. Cronn, R., Brothers, M., Klier, K., Bretting, P.K. and Wendel, J.F. (1997) Allozyme variation in domesticated annual sunower and its wild relatives. Theoretical and Applied Genetics 95, 532545. Crow, J.F. and Kimura, M. (1970) An Introduction to Population Genetics Theory. Harper & Row, New York. Cubero, J.I. (1997) Conservation through plant breeding and in situ conservation. In: Valdes, B., Heywood, V.H., Raimondo, F.M. and Zohary, D. (eds) Proceedings of the Workshop on Conservation of the Wild Relatives of European Cultivated Plants, Herbarium Mediterraneum Panormitanum: Palermo, pp. 207209. Daniell, H., Datta, R., Varma, S., Gray, S. and Lee, S.B. (1998) Genetic engineering of herbicide

56

C. Spillane and P. Gepts resistance through the chloroplast genome and prevention of outcross to weedy relatives. Nature Biotechnology 16, 345348. Dashiell, K. and Fatokun, C. (1997) Soyabean. In: Fuccillo, D., Sears, L. and Stapleton, P. (eds) Biodiversity in Trust. Cambridge University Press, Cambridge, pp. 181190. Debouck, D.G., (1991) Systematics and morphology. In: Van Schoonhoven, A. and Voysest, O. (eds) Common Beans; Research for Crop Improvement. CAB International, Wallingford, UK, pp. 55118. Debouck, D.G. and Tohme, J. (1989) Implications for bean breeders of studies on the origins of common beans, Phaseolus vulgaris L. In: Beebe, S. (ed.) Current Topics in Breeding of Common Bean. CIAT, Cali, Colombia. de la Fuente, J.M., Ramirez-Rodriguez, V., Cabrera-Ponce, J.L. and Herrera-Estrella, L. (1997) Aluminum tolerance in transgenic plants by alteration of citrate synthesis. Science 276, 15661568. Delgado Salinas, A. (1988) Variation, taxonomy, domestication, and germplasm potentialities in Phaseolus coccineus. In: Gepts, P. (ed.) Genetic Resources of Phaseolus Beans. Kluwer, Dordrecht, pp. 441463 DellaPenna, D. (1999) Nutritional genomics: manipulating plant micronutrients to improve human health. Science 285, 375379. de Oliveira, A.C., Richter, T. and Bennetzen, J.L. (1996) Regional and racial specicities in sorghum germplasm assessed with DNA markers. Genome 39, 579587. Deu, M., Hamon, P., Chantereau, J., Dufour, P., Dhont, A. and Lanaud, C. (1995) Mitochondrial DNA diversity in wild and cultivated sorghum. Genome 38, 635645. de Vicente, M.C. and Tanksley, S.D. (1993) QTL analysis of transgressive segregation in an interspecic tomato cross. Genetics 134, 585596. Devos, K.M. and Gale, M.D. (1997) Comparative genetics in the grasses. Plant Molecular Biology 35, 315. Dewey, D. (1980) Some applications and misapplications of induced polyploidy in plant breeding. In: Lewis, W.D. (ed.) Polyploidy: Biological Relevance. Plenum Press, New York. DeWet, J.M. (1979) Origins of polyploids. Basic Life Science 13, 315. Diamond, J. (1997) Guns, Germs, and Steel. Norton, New York. Dobzhansky, T. (1937) Genetics and the Origin of Species. Colombia University Press, New York. Doebley, J. (1989) Isozymic evidence and the evolution of crop plants. In: Soltis, D.E. and Soltis, P.S. (eds) Isozymes in Plant Biology. Dioscorides Press, Portland, pp. 165191. Doebley, J. (1990) Molecular evidence for gene ow among Zea species. Bioscience, 40, 443448. Doebley, J. (1992) Mapping the genes that made maize. Trends in Genetics 8, 302307. Doebley, J. and Lukens, L. (1998) Transcriptional regulators and the evolution of plant form. Plant Cell 10, 10751082. Doebley, J.F. and Stec, A. (1993) Inheritance of the morphological differences between maize and teosinte. Genetics 129, 285295. Doebley, J., Stec, A. and Gustus, C. (1995) Teosinte branched1 and the origin of maize: evidence for epistasis and the evolution of dominance. Genetics 141, 333346. Doebley, J., Stec, A. and Hubbard, L. (1997) The evolution of apical dominance in maize. Nature 386, 485488. Doggett, H. and Majisu, B.N. (1968) Disruptive selection in crop development. Heredity 23, 123. Doyle, J.J. (1988) 5S ribosomal gene variation in the soybean and its progenitor. Theoretical and Applied Genetics 75, 621624. Dubin, H.J., Fischer, R.A., Mujeeb-Kazi, A., Pena, R.J., Sayre, K.D., Skovmand, B. and Valkoun, J. (1997) Wheat. In: Fuccillo, D., Sears, L. and Stapleton, P. (eds) Biodiversity in Trust. Cambridge University Press, Cambridge, pp. 309320. Dudley, J.W. (1982) Theory for transfer of alleles. Crop Science 22, 631636.

The Dynamics of Crop Genepools

57

Dudley, J.W. and Lambert, R.J. (1992) Ninety generations of selection for oil and protein in maize. Maydica 37, 8187. Duranton, V. and Abram, F. (1996) Down on the Farm: Fast Growing Developments in ISO Technical Committee TC 23. ISO Bulletin May 1996, pp. 912. Duvick, D.N. (1984) Genetic diversity in major farm crops on the farm and in reserve. Economic Botany 38, 161178. Duvick, D.N. (1989) The romance of plant breeding. Stadler Genetics Symposium 19, 3954. Duvick, D.N. (1996) Plant breeding, an evolutionary concept. Crop Science 36, 539548. Edwards, M. (1992) Use of molecular markers in the evaluation and introgression of genetic diversity for quantitative traits. Field Crops Research 29, 241260. Edwards, M.D., Stuber, C.W. and Wendel, J.F. (1987) Molecular marker-facilitated investigation of quantitative trait loci in maize. I. Number, distribution and types of gene action. Genetics 116, 113125. Ehlenfeldt, M.K. and Ortiz, R. (1995) On the origins of endosperm dosage requirements in Solanum and other angiosperm genera. Sexual Plant Reproduction 8, 189196. Ellstrand, N.C. (1992) Gene ow by pollen: implications for plant conservation genetics. Oikos 63, 7786. Ellstrand, N.C. and Hoffman, C.A. (1990) Hybridization as an avenue for escape for engineered genes. Strategies for risk reduction. Bioscience 40, 438442. Ellstrand, N.C., Whitkus, R. and Rieseberg, L.H. (1996) Distribution of spontaneous plant hybrids. Proceedings of the National Academy of Sciences USA 93, 50905093. Erskine, W. (1997) Lessons for breeders from land races of lentils. Euphytica 93, 107112. Erskine, W. and Muehlbauer, F.J. (1991) Allozyme and morphological variability, outcrossing rate and core collection formation in lentil germplasm. Theoretical and Applied Genetics 83, 119125. Falconer, D.S. (1989) Introduction to Quantitative Genetics. Longman, Harlow, UK. FAO (1998) The State of the Worlds Plant Genetic Resources for Food and Agriculture. FAO, Rome. Fehr, W.R. (1981) Genetic Contributions to Yield Gains of Five Major Crop Plants. CSSA Special Publication Number 7. CSSA/ASA, Madison, Wisconsin. Ferguson, A.E. (1992) Differences among women farmers: implications for African agricultural research programs. In: Proceedings of the Workshop on Social Science Research and CRSP, Cornahan Conference Centre, University of Kentucky, Lexington, Kentucky, June 911, 1992. Ferreira, M.E., Satagopan, J., Yandell, B.S., Williams, P.H. and Osborn, T.C. (1995) Mapping loci controlling vernalisation requirement and owering time in Brassica napus. Theoretical and Applied Genetics 90, 727732. Fischbeck, G. (1992) Barley cultivar development in Europe success in the past and possible changes in the future. Barley Genetics IV, 885901. Francis, C.A. (1986) Multiple Cropping Systems. Macmillan, New York. Frey, K.J. (1975) Plant breeding in the seventies: useful genes from wild plant species. Egyptian Journal of Genetics and Cytology 5, 460482. Frey, K.J., Cox, T.S., Rodgers, D.M. and Bramel-Cox, P. (1984) Increasing cereal yields with genes from wild and weedy species. In: Chopra, V.L., Joshi, B.C., Sharma, R.P. and Bansal, H.C. (eds) Genetics: New Frontiers, Vol. IV, pp. 5168. Freytag, G.F. and Debouck, D.G. (1996) Phaseolus costaricensis, a new wild bean species (Phaseolinae, Leguminosae) from Costa Rica and Panama, Central America. Novon 6, 157163. Gale, M.D. and Devos, K.M. (1998) Comparative genetics in the grasses. Proceedings of the National Academy of Sciences USA 95, 19711974. Gepts, P. (1988) A Middle American and and Andean gene pool. In: Gepts, P. (ed.) Genetic Resources of Phaseolus Beans. Kluwer, Dordrecht, pp. 375390.

58

C. Spillane and P. Gepts Gepts, P. (1993) The use of molecular and biochemical markers in crop evolution studies. Evolutionary Biology 27, 5194. Gepts, P. (1995) Genetic markers and core collections. In: Hodgkin, T., Brown, A.H.D., van Hintum, T.J.L. and Morales, E.A.V. (eds) Core Collections of Plant Genetic Resources. Cambridge University Press, Cambridge, pp. 127145. Gepts, P. (1998) Origin and evolution of common bean: past events and recent trends. Horticultural Science 33, 11241130 Gepts, P. and Bliss, F.A. (1985) F1 hybrid weakness in the common bean: differential geographic origin suggests two gene pools in cultivated bean germplasm. Journal of Heredity 76, 447450. Gepts, P. and Clegg, M.T. (1989) Genetic diversity in pearl millet (Pennisetum glaucum [L.] R.Br.) at the DNA sequence level. Journal of Heredity 80, 203208. Gepts, P. and Debouck, D.G. (1991) Origin, domestication, and evolution of the common bean, Phaseolus vulgaris. In: Voysest, O. and Van Schoonhoven, A. (eds), Common Beans: Research for Crop Improvement. CAB International, Wallingford, UK, pp. 753. Gill, K.S. (1989) Use of collections by public breeders. In: Brown, A.H.D., Frankel, O.H., Marshall, D.R. and Williams, J.T. (eds) The Use of Plant Genetic Resources. Cambridge University Press, Cambridge. Glaszmann, J.C. (1987) Isozymes and classication of Asian rice varieties. Theoretical and Applied Genetics 74, 2130. Golenberg, E.M. (1987) Estimation of gene ow and genetic neighbourhood size by indirect methods in a selng annual Triticum dicoccoides. Evolution 41, 13261334. Goodman, M.M. (1985) Exotic maize germplasm: status, prospects and remedies. Iowa State Journal of Research 59, 497527. Goodman, M.M. (1992) Choosing and using tropical maize germplasm. Proceedings of the Annual Corn and Sorghum Research Conference 47, 4764. Goodnight, C.J. (1987) On the effect of founder events on epistatic genetic variance. Evolution 41, 8091. Gottlieb, L.D. (1984) Genetics and morphological evolution in plants. American Naturalist 123, 681709. Graham, J., McNicol, R.J. and McNicol, J.W. (1996) A comparison of methods for the estimation of genetic diversity in strawberry cultivars. Theoretical and Applied Genetics 93, 402406. Grant, V. (1963) The Origins of Adaptations. Colombia University Press, New York. Grifng, B. (1956) A generalized treatment of the use of diallel crosses in quantitative inheritance. Heredity 10, 3150. Haghighi, K.R. and Ascher, P.D. (1988) Fertile, intermediate hybrids between Phaseolus vulgaris and P. acutifolius from congruity backcrossing. Sexual Plant Reproduction 1, 5158. Hallauer, A.R. (1986) Progress to date in the use of exotic materials in conventional breeding programs. Southern Africa Maize Symposium Proceedings 4, 3540. Hallauer, A.R. and Miranda, J.B. (1981) Quantitative Genetics in Maize Breeding. Iowa State University Press, Ames, Iowa. Hamilton, C., Frary, A., Lewis, C. and Tanksley, S. (1996) Stable transfer of intact high molecular weight DNA into plant chromosomes. Proceedings of the National Academy of Sciences USA 93, 99759979. Hammer, K. and Schubert, I. (1994) Are Vavilovs law of homologous series and synteny related? Genetic Resources and Crop Evolution 41, 123124. Hamrick, J. (1997) Allozyme diversity in cultivated crops. Crop Science 37, 2630. Hamrick, J.L., Godt, M.J.W. and Sherman-Broyles, S.L. (1995) Gene ow among plant populations: Evidence from molecular markers. In: Hoch, P.C. and Stephenson, A.G. (eds)

The Dynamics of Crop Genepools

59

Experimental and Molecular Approaches to Plant Systematics. Missouri Botanical Garden, Missouri. Hanelt, P. (1997) Gene ow between crops and related taxa some case studies. In: Valdes, B., Heywood, V.H., Raimondo, F.M. and Zohary, D. (eds) Proceedings of the Workshop on Conservation of the Wild Relatives of European Cultivated Plants, Herbarium Mediterraneum Panormitanum: Palermo. pp.5162. Hanson, W.D. (1959a) Early generation analysis of lengths of chromosome segments around a locus held heterozygous with backcrossing or selng. Genetics 44, 833837. Hanson, W.D. (1959b) The breakup of initial linkage blocks under selected mating systems. Genetics 44, 857868. Harlan, J.R. (1965) The possible role of weed races in the evolution of cultivated plants. Euphytica 14, 173176. Harlan, J.R. (1976) Genetic resources in wild relatives of crops. Crop Science 16, 329333. Harlan, J.R. (1992) Crops and Man, 2nd edn. American Society of Agronomy and Crop Science Society of America, Madison, Wisconsin. Harlan, J.R. and deWet, J.M.J. (1971) Toward a rational classication of cultivated plants. Taxon 20, 509517. Harlan, J.R. and deWet, J.M.J. (1975) On: Winge and a prayer: the origins of polyploids. Botanical Review 41, 361390. Harwood, R.R. (1989) History of sustainable agriculture: U.S. and international perspective. In: Edwards, C.A. et al. (eds) Sustainable Agricultural Systems. Soil and Water Conservation Society, Ankeny, Alaska. Havey, M.J. and Muehlbauer, F.J. (1989) Variability for restriction fragment lengths and phylogenies in lentil. Theoretical and Applied Genetics 77, 839843. Hawkes, J. (1977) The importance of wild germplasm in plant breeding. Euphytica 26, 615621. Hawtin, G., Iwanaga, M. and Hodgkin, T. (1996) Genetic resources in breeding for adaptation. Euphytica 92, 255266. Hayes, P.M., Liu, B.H., Knapp, S.J., Chen, F., Jones, B., Blake, T., Franckowiak, J., Rasmusson, D., Sorrells, M., Ullrich, S.E., Wesenberg, D. and Kleinhofs, A. (1993) Quantitative trait locus effects and environmental interaction in a sample of North American barley germ plasm. Theoretical and Applied Genetics 87, 392401. Hayes, P.M., Cerono, J., Witsenhoer, H., Kuiper, M., Zabeau, M., Sato, K., Kleinhofs, A., Kudrna, D., Kilian, A., Saghai-Maroof, M., Hoffman, D. and the North American Barley Genome Mapping project (1997) Characterizing and exploiting genetic diversity and quantitative traits in barley (Hordeum vulgare) using AFLP markers. Journal of Quantitative Trait Loci 3,(2). http://probe.nalusda.gov:8000/otherdocs/jqtl/jqtl1997-02/ He, G., Prakash, C.S. and Jarret, R.L. (1995) Analysis of genetic diversity in a sweet potato (Ipomoea batatas) germplasm collection using DNA amplication ngerprinting. Genome 38, 938945. Hedrick, P. and McDonald, J.F. (1980) Regulatory gene adaptation: an evolutionary model. Heredity 45, 8397. Heisey, P.W., Smale, M., Byerlee, D. and Souza, E. (1997) Wheat rusts and the costs of genetic diversity in the Punjab of Pakistan. American Journal of Agronomy and Economics 79, 726737. Helentjaris, T., King, G., Slocum, M., Siedenstrang, C. and Wegman, S. (1985) Restriction fragment length polymorphisms as probes for plant diversity and their development as tools for applied plant breeding. Plant Molecular Biology 5, 109118. Hermsen, J.G. Th. (1989) Current use of potato collections. In: Brown, A.H.D., Frankel, O.H., Marshall, D.R. and Williams, J.T. (eds) The Use of Plant Genetic Resources. Cambridge University Press, Cambridge.

60

C. Spillane and P. Gepts Hidalgo, R. and Beebe, S. (1997) Phaseolus beans. In: Fuccillo, D., Sears, L. and Stapleton, P. (eds) Biodiversity in Trust. Cambridge University Press, Cambridge, pp. 139155. Hilu, K. (1983) The role of single-gene mutation in the evolution of owering plants. In: Hecht, M.K., Wallace, B. and Prance, G.T. (eds) Evolutionary Biology. pp. 97128. Hoisington, D., Jiang, C., Khairallah, M., Ribaut, J.M., Bohn, M., Melchinger, A., Willcox, M. and Gonzalez-de-Leon, D. (1996) QTL for insect resistance and drought tolerance in tropical maize: prospects for marker assisted selection. Symp. Soc. Exp. 50, 3944. Hoisington, D., Khairallah, M., Reeves, T., Ribaut, J.M., Skovmand, B., Taba, S. and Warburton, M. (1999) Plant genetic resources: what can they contribute toward increased crop productivity? Proceedings of the National of Academy of Sciences USA 96, 59375943. Holland, J.B. and Goodman, M.M. (1994) Combining ability of tropical maize accessions with U.S. germplasm. Crop Science 35, 767773. Holley, R.N. and Goodman, M.M. (1988) Yield potential of tropical maize hybrid derivatives. Crop Science 28, 213218. Horner, H.T. and Palmer, R.G. (1995) Mechanisms of genic male sterility. Crop Science 35, 15271535. Huamn, Z. and Zhang, D.P. (1997) Sweetpotato. In: Fuccillo, D., Sears, L. and Stapleton, P. (eds) Biodiversity in Trust. Cambridge University Press, Cambridge, pp. 2938. Hucl, P. and Scoles, G.J. (1985) Interspecic hybridization in the common bean: a review. HortScience 20, 352357. Hymowitz, T. and Kaizuma, N. (1981) Soybean, seed protein electrophoresis proles from 15 Asian countries or regions: hypotheses on paths of dissemination of soybeans from China. Economic Botany 24, 408421. Iqbal, M.J., Aziz, A., Saeed, N.A., Zafar, Y. and Malik, K.A. (1997) Genetic diversity evaluation of some elite cotton varieties by RAPD analysis. Theoretical and Applied Genetics 94, 139144. Irvine, J.E., TAES, Texas A&M University, USA. Izioka, E.E.K., Gilabert, C.E., Marino, C.L., Castillo, M.E.C. and Lopes, C.R. (1997) RAPD analysis of Capsicum anuum cultivars aiming breeding programs. Abstract P64 Plant Genome V, January 1216, 1997, San Diego, California. James, J.W. (1971) The founder effect and response to articial selection. Genetic Research 16, 241250. Jansen, R.C., Van Ooijen, J.W., Stam, P., Lister, C. and Dean, C. (1995) Genotype-by-environment interaction in genetic mapping of multiple quantitative trait loci. Theoretical and Applied Genetics 91, 3337. Jennings, P.R. and Aquino, R.C. (1968) Studies on competition in rice. III. The mechanism of competition among phenotypes. Evolution 22, 529542. Jennings, D.L. (1994) Breeding for resistance to African cassava mosaic geminivirus in East Africa. Tropical Science 34, 110122. Jensen, N.F. (1988) Plant Breeding Methodology. John Wiley & Sons, New York. Jefferson, R.A. and Bicknell, R. (1996) The potential impacts of apomixis: a molecular genetics approach. In: Sobral, B.W.S. (ed.) The Impact of Plant Molecular Genetics. Birkhuser, Boston, Massachusetts, pp. 87101. Jiang, C.-X., Wright, R., El-Zik, K. and Paterson, A. (1998) Polyploid formation created unique avenues for response to selection in Gossypium (cotton). Proceedings of the National Academy of Sciences USA 95, 44194424. Jiang, J., Friebe, B. and Gill, B.S. (1994) Recent advances in alien gene transfer in wheat. Euphytica 73, 199212. Johnston, S. and Hanneman, R. (1982) Manipulations of endosperm balance number overcome crossing barriers between diploid Solanum species. Science 217, 446448. Jo, O., Xu, Y., Chen, X., Park, W.D., Beachell, H.M., Dilday, R.H., Goto, M. and McCouch,

The Dynamics of Crop Genepools

61

S.R. (1997) Comparative evaluation of within-cultivar variation of rice (Oryza sativa L.) using microsatellite and RFLP markers. Genome 40, 370378. Kannenberg, L.W. and Falk, D.E. (1995) Models for activation of plant genetic resources for crop breeding programs. Canadian Journal of Plant Science 75, 4553. Kauffman, C.S. and Weber, L.E. (1990) Grain amaranth. In: Janick, J. and Simon, J.E. (eds), Advances in New Crops. Timber Press, Portland, Oregon, pp. 127139. Kawano, K. and Thung, M.D. (1982) Intergenotypic competition and competition with associated crops in cassava. Crop Science 22, 5963. Kellogg, E.A. (1998) Relationships of cereal crops and other grasses. Proceedings of the National Academy of Science USA 95, 20052010. Kemble, R.J., Gunn, R.E. and Flavell, R.B. (1983) Mitochondrial DNA variation in races of maize indigenous to Mexico. Theoretical and Applied Genetics 65, 129144. Khush, G.S. (1997) Origin, dispersal, cultivation and variation of rice. Plant Molecular Biology 35, 2534. Kim, H.S. and Ward, R.W. (1997) Genetic diversity in Eastern U.S. soft winter wheat (Triticum aestivum L. em. Thell.) based on RFLPs and coefcients of parentage. Theoretical and Applied Genetics 94, 472479. Klinger, T.P., Elam, D.R. and Ellstrand, N.C. (1991) Radish as a model system for the study of engineered gene escape rates via cropweed mating. Conservation Biology 79, 14311435. Klinger, T., Arriola, P.E. and Ellstrand, N.C. (1992) Cropweed hybridization in radish (Raphanus sativus): effects of distance and population size. American Journal of Botany 79, 14311435. Knight, R.L. (1948) The role of major genes in the evolution of economic characters. Journal of Genetics 48, 370387. Koinange, E.M.K. and Gepts, P. (1992) Hybrid weakness in wild Phaseolus vulgaris L. Journal of Heredity 83, 135139. Koinange, E.M.K., Singh, S.P. and Gepts, P. (1996) Genetic control of the domestication syndrome in common-bean. Crop Science 36, 10371045. Komarnitsky, I.K., Samoylov, A.M., Redko, VV., Peretyayko, V.G. and Gleba, Yu.Yu. (1990) Intraspecic diversity of sugar beet (Beta vulgaris) mitochondrial DNA. Theoretical and Applied Genetics 80, 253257. Kuittinen, H., Sillanpaa, M.J. and Savolainen, O. (1997) Genetic basis of adaptation: owering time in Arabidopsis thaliana. Theoretical and Applied Genetics 95, 573583. Kurian, K.M., Watson, C.J. and Wyllie, A.H. (1999) DNA chip technology. Journal of Pathology 187(3), 26771. Kuruvilla, K.M. and Singh, A. (1981) Karyotypic and electrophoretic studies on taro and its origin. Euphytica 30, 405413. Ladizinsky, G. (1985) Founder effect in crop-plant evolution. Economic Botany 39, 191199. Ladizinsky, G. (1995) Domestication via hybridization of the wild tetraploid oats Avena magna and A. murphyi. Theoretical and Applied Genetics 91, 639646. Langevin, S.A., Clay, K. and Grace, J.B. (1990) The incidence and effect of hybridization between cultivated rice and its related weed red rice (Oryza sativa L.). Evolution 44, 10001008. Lande, R. (1985) Expected time for random genetic drift of a population between stable phenotype states. Proceedings of the National Academy of Sciences USA 82, 76417645. Lander, E. and Botstein (1989) Mapping Mendelian factors underlying quantitative traits using RFLP linkage maps. Genetics 121, 185199. Lark, K.G., Chase, K., Adler, F., Mansur, L.M. and Orf, J.H. (1995) Interactions between quantitative trait loci in soybean in which trait variation at one locus is conditional upon a specic allele at another. Proceedings of the National Academy of Sciences USA 92, 46564660.

62

C. Spillane and P. Gepts Laurie, D.A. (1997) Comparative genetics of owering time. Plant Molecular Biology 35, 167177. Lee, M. (1996) Comparative genetic and QTL mapping in sorghum and maize. Symposiam of the Society for Experimental Biology 50, 3138. Lee, M. (1998) Genome projects and gene pools: new germplasm for breeding? Proceedings of the National Academy of Sciences USA 95, 20012004. Lefol, E., Danielou, V. and Darmency, H. (1996) Predicting hybridization between transgenic oilseed rape and wild mustard. Field Crops Research 45, 153161. Leitch, I. and Bennett, M. (1997) Polyploidy in angiosperms. Trends in Genetics 2, 470476. Leng, E.R. (1974) Results of long-term selection for chemical composition in maize and their signicance in evaluating breeding systems. In: Dudley, J.W. (ed) Seventy Generations of Selection for Oil and Protein in Maize. Crop Science Society of America, Madison, Wisconsin. Lenn, J. and Wood, D. (1991) Plant diseases and the use of wild germplasm. Annual Review of Phytopathology 29, 3563. Lerceteau, E., Robert, T., Petiard, V. and Crouzillat, D. (1997) Evaluation of the extent of genetic variability among Theobroma cacao accessions using RAPD and RFLP markers. Theoretical and Applied Genetics 95, 1019. LeRoy, A.R., Fehr, W.R. and Cianzio, S.R. (1991) Introgression of genes for small seed size from Glycine soja into G. max. Crop Science 31, 693697. Levin, D. and Kerster, H. (1969) Density-dependent gene dispersal in Liatris. American Naturalist 103, 6174. Levin, D. and Kerster, H. (1974) Gene ow in seed plants. Evolutionary Biology 7, 139220. Li, Z. and Zhu, Y. (1989) Rice male sterile cytoplasm and fertility restoration. In: IRRI, Hybrid Rice. International Rice Research Institute, Manila, pp. 85102. Lin, Y.R., Schertz, K.F. and Paterson, A.H. (1995) Comparative analysis of QTLs affecting plant height and maturity across the Poaceae, in reference to an interspecic sorghum population. Genetics 141, 391411. Linde-Laursen, I., Nielsen, G. and Johansen, H.B. (1987) Distribution of isoenzyme markers at 37 loci in a pedigree of European spring barley. Hereditas 106, 241251. Linder, C., Taha, I., Seiler, G., Snow, A. and Rieseberg, L. (1998) Long-term introgression of crop genes into wild sunower populations. Theoretical and Applied Genetics 96, 339347. Llaca, V., Delgado Salinas, A. and Gepts, P. (1994) Chloroplast DNA as an evolutionary marker in the Phaseolus vulgaris complex. Theoretical and Applied Genetics 88, 646652. Louette, D., Charrier, A. and Berthaud, J. (1997) In situ conservation of maize in Mexico: genetic diversity and maize seed management in a traditional community. Economic Botany 51, 2038. Lu, C., Shen, L., Tan, Z., Xu, Y., Xu, Y., He, P., Chen, Y. and Zhu, L. (1997) Comparative mapping of QTLs for agronomic traits of rice across environments by using a doubled-haploid population. Theoretical and Applied Genetics 94, 145150. Luo, M., Dubcovsky, J. and Dvorak, J. (1996) Recognition of homeology by the wheat Ph1 locus. Genetics 144, 11951203. Luo, M., Dubcovsky, J., Goyal, S. and Dvorak, J. (1996) Engineering of interstitial foreign chromosome segments containing the K+/Na+ selectivity gene Kna1 by sequential homeologous recombination in durum wheat. Theoretical and Applied Genetics 93, 11801184. Lynch, M. (1991) The genetic interpretation of inbreeding depression and outbreeding depression. Evolution 45, 622629. Mack, R.N. (1991) The commercial seed trade: an early dispenser of weeds in the United States. Economic Botany 45, 257273. MacKey, J. (1970) Signicance of mating systems for chromosomes and gametes in polyploids. Hereditas 66, 165176.

The Dynamics of Crop Genepools

63

MacKey, J. (1986) Recombination in future plant breeding. Svalof 18861986. Research and Results in Plant Breeding, pp.275280. Madriz, J., de Miranda, J.R., Cabezas, E., Oliva, M., Hernandez, M. and Espinoza, A.M. (1998) Echinochloa hoja blanca virus and rice hoja blanca virus occupy distinct ecological niches. Journal of Phytopathology 146, 305308. Manninen, O. and Nissila, E. (1997) Genetic diversity among Finnish six-rowed barley cultivars based on pedigree information and DNA markers. Hereditas 126, 8793. Marchais, L. (1994) Wild pearl millet population (Pennisetum glaucum, Poaceae) integrity in agricultural Sahelian areas. An example from Kieta (Niger). Plant Systematics and Evolution 189, 233245. Marocco, A., Cattivelli, L., Delogu, G., Lorenzoni, C. and Stanca, A.M. (1992) Performance of S2 winter barley progenies from original and improved populations developed via recurrent selection. Plant Breeding 108, 250255. Marshall, D.R. and Brown, A.H.D. (1975) Optimum sampling strategies in genetic conservation. In: Frankel, O.H. and Hawkes, J.G. (eds) Genetic Resources for Today and Tomorrow. Cambridge University Press, Cambridge, pp. 5380. Martin, J.M., Blake, T.K. and Hockett, E.A. (1991) Diversity among North American spring barley cultivars based on coefcients of parentage. Crop Science 31, 11311137. Mason, H., Ball, J., Shi, J.-J., Jiang, X., Estes, M. and Arntzen, C. (1996) Expression of Norwalk virus capsid protein in transgenic tobacco and potato and its oral immunogenicity in mice. Proceedings of the National Academy of Sciences USA 93, 53355340. Mayr, E. (1942) Systematics and the Origin of Species. Colombia University Press, New York. Mayr, E. (1954) Change of genetic environment and evolution. In: Huxley, J., Hardy, A.C. and Ford, E.B. (eds) Evolution as a Process. Allen and Unwin, London. Mayr, E. (1963) Animal Species and Evolution. Harvard University Press, Cambridge, Massachusetts. McClean, P.E., Myers, J.R. and Hammond, J.J. (1993) Coefcient of parentage and cluster analysis of North American dry bean cultivars. Crop Science 33, 190197. McCouch, S. (1998) Towards a plant genomics initiative: thoughts on the value of cross-species and cross-genera comparisons in the grasses. Proceedings of the National Academy of Sciences USA 95, 19831985. McCouch, S.R., Chen, X., Panaud, O., Temnykh, S., Xu, Y., Cho, Y.G., Huang, N., Ishii, T. and Blair, M. (1997) Microsatellite marker development, mapping and applications in rice genetics and breeding. Plant Molecular Biology 35, 8999. McMullen, M.D., Byrne, P.F., Snook, M.E., Wiseman, B.R., Lee, E.A., Widstrom, N.W. and Coe, E.H. (1998) Quantitative trait loci and metabolic pathways. Proceedings of the National Academy of Sciences USA 19962000. Melchinger, A.E., Graner, A. and Singh, A. (1994) Relationships among European barley germplasm: I. Genetic diversity among winter and spring cultivars revealed by RFLPs. Crop Science 34, 11911199. Mercado, L.A., Souza, E. and Kephart, K.D. (1996) Origin and diversity of North American hard spring wheats. Theoretical and Applied Genetics 93, 593599. Mikkelsen, T.R., Andersen, B. and Jrgensen, R.B. (1996) The risk of crop transgene spread. Nature 380, 3131. Miller, J.C. and Tanksley, S.D. (1990) RFLP analysis of phylogenetic relationships and genetic variation in the genus Lycopersicon. Theoretical and Applied Genetics 80, 437448. Milligan, B.G., Leebens-Mack, J. and Strand, A.E. (1994) Conservation genetics: beyond the maintenance of marker diversity. Molecular Ecology 3, 423435. Mitchell-Olds, T. (1996) Genetic constraints on life history evolution: quantitative-trait loci inuencing growth and owering in Arabidopsis thaliana. Evolution 50, 140145.

64

C. Spillane and P. Gepts Molvig, L., Tabe, L.M., Eggum, B.O., Moore, A.E., Craig, S., Spencer, D. and Higgins, T.J. (1997) Enhanced methionine levels and increased nutritive value of transgenic lupins (Lupinus augustifolius L) expressing a sunower albumin gene. Proceedings of the National Academy of Sciences USA 94, 83938398. Morgunov, A.I., Pena, R.J., Crossa, J. and Rajaram, S. (1993) Worldwide distribution of Glu-1 alleles in bread wheat. Journal of Genetics and Breeding 47, 5360. Muehlbauer, F.J., Kaiser, W.J. and Simon, C.J. (1994) Potential for wild species in cool season food legume breeding. Euphytica 73, 109114. Multani, D.S. and Lyon, B.R. (1995) Genetic ngerprinting of Australian cotton cultivars with RAPD markers. Genome 38, 10051008. Mumm, R.H. and Dudley, J.W. (1994) A classication of 148 US maize inbreds: I. Cluster analysis based on RFLPs. Crop Science 34, 842851. Mundt, C.C. (1984) Techniques to manage pathogen coevolution with host plants to prolong resistance. In: Teng, P.S., Heong, K.L. and Moody, K. (eds) Rice Pest Science and Management. International Rice Research Institute, Manila, pp. 193205. Murfet, I.C. (1977) Environmental interaction and the genetics of owering. Annual Review of Plant Physiology 28, 253278. Nakagahara, M., Okuno, K. and Vaughan, D. (1997) Rice genetic resources: history, conservation, investigative characterization and use in Japan. Plant Molecular Biology 35, 6977. National Research Council (1972) Committee on Genetic Vulnerability of Major Crops. National Academy of Sciences, Washington, DC. Neale, D.B., Saghai-Maroof, M.A., Allard, R.W., Zhang, Q. and Jorgensen, R.A. (1988) Chloroplast DNA diversity in populations of wild and cultivated barley. Genetics 120, 11051110. Nei, M., Maruyama, T. and Chakraborty, R. (1975) The bottleneck effect and genetic variability in populations. Evolution 29, 110. Nicholls, N. (1997) Increased Australian wheat yield due to recent climate trends. Nature 387, 484485. Nodari, R.O., Tsai, S.M., Guzmn, P., Gilbertson, R.L. and Gepts, P. (1993) Towards an integrated linkage map of common bean. 3. Mapping genetic factors controlling hostbacteria interactions. Genetics 134, 341350. Orr, H.A. and Coyne, J.A. (1992) The genetics of adaptation: a reassessment. American Naturalist 140, 725742. Otim-Nape, G.W., Bua, A., Thresh, J.M., Baguma, Y., Ogwal, S., Ssemakula, G.N., Acola, D., Byabakama, B. and Martin, A. (1997) Cassava Mosaic Disease in Uganda: the Current Pandemic and Approaches to Control. NARO/NRI Publication, Natural Resources Institute, Chatham. Palmer, J.D. (1987) Chloroplast DNA evolution and biosytematic uses of chloroplast DNA variation. American Naturalist 130, S6-S29. Parker, J.P. and Michaels, T.E. (1986) Simple genetic control of hybrid plant development in interspecic crosses between Phaseolus vulgaris L. and P. acutifolius A. Gray. Plant Breeding 97, 315323. Paterson, A.H. (1995) Molecular dissection of quantitative traits: progress and prospects. Genome Res. 5, 32133. Paterson, A.H., Lander, E.S., Hewitt, J.D., Peterson, S., Lincoln, S.E. and Tanksley, S.D. (1988) Resolution of quantitative traits into Mendelian factors by using a complete linkage map of restriction fragment length polymorphisms. Nature 335, 721726. Paterson, A.H., Damon, S., Hewitt, J.D., Zamir, D., Rabinowitch, H.D., Lincoln, S.E., Lander, E.S. and Tanksley, S.D. (1991) Mendelian factors underlying quantitative traits in tomato: comparison across species, generations, and environments. Genetics 127, 181197.

The Dynamics of Crop Genepools

65

Paterson, A.H., Lin, Y.-R., Li, Z., Schertz, K.F., Doebley, J.F., Pinson, S.R.M., Liu, S.-C., Stansel, J.W. and Irvine, J.E. (1995) Convergent domestication of cereal crops by independent mutations at corresponding genetic loci. Science 269, 17141718. Paterson, A.H., Lan, T.-H., Reischmann, K.P., Chang, C., Lin, Y.-R., Liu, S.-C., Burow, M.D., Kowalski, S.P., Katsar, C.S., DelMonte, T.A., Feldmann, K.A., Schertz, K.F. and Wendel, J.F. (1996) Toward a unied genetic map of higher plants, transcending the monocotdicot divergence. Nature Genetics 14, 380382. Pena, R.J. (1995) Quality improvement of wheat and triticale. In: Rajaram, S. and Hettel, G.P. (eds). Wheat Breeding at CIMMYT: Commemorating 50 years of Research in Mexico for Global Wheat Improvement. Wheat Special Report No. 29. CIMMYT, Mexico, pp. 100111. Petersen, L., Ostergard, H. and Giese, H. (1994) Genetic diversity among wild and cultivated barley as revealed by RFLP. Theoretical and Applied Genetics 89, 676681. Phillips, R.L. and Freeling (1998) Plant genomics and our food supply: an introduction. Proceedings of the National Academy of Sciences USA 95, 19691970 Pickersgill, B. (1981) Biosystematics of cropweed complexes. Kulturpanze 29, 377388. Pimm, S.L. (1997) In search of perennial solutions. Nature 389, 126127. Rasmussen, D.C. and Phillips, R.L. (1997) Plant breeding progress and genetic diversity from de novo variation and elevated epistasis. Crop Science 37, 303310. Reiseberg, L.H. and Soltis, D.E. (1991) Phylogenetic consequences of cytoplasmic gene ow in plants. Evolutionary Trends in Plants 5, 6584. Rieseberg, L., Van Fossen, C. and Desrochers, A. (1995a) Hybrid speciation accompanied by genomic reorganization in wild sunowers. Nature 375, 313316. Rieseberg, L., Linder, C. and Seiler, G. (1995b) Chromosomal and genic barriers to introgression in Helianthus. Genetics 141, 11631171. Rieseberg, L.H., Arias, D.M., Ungerer, M., Linder, C.R. and Sinervo, B. (1996) The effects of mating design on introgression between chromosomally divergent sunower species. Theoretical and Applied Genetics 93, 633644. Rice, E., Blanco, J.L. and Smale, M. (1997) Farmers Use of Improved Seed Selection Practices in Mexican Maize: Evidence and Issues from the Sierra de Santa Marta. Economics Working Paper 9704. CIMMYT, Mexico. Rick, C. (1982) The potential of exotic germplasm for tomato improvement. In: Vasil, I., Scowcroft, W. and Frey, K. (eds) Plant Improvement and Somatic Cell Genetics. Academic Press, New York, pp. 142. Rick, C.M., Zobel, R.W. and Forbes, J.F. (1974) Four peroxidase loci in red fruited tomato species: genetics and geographic distribution. Proceedings of the National Academy of Sciences USA 71, 835839. Rick, C., Fobes, J. and Holle, M. (1977) Genetic variation in Lycopersicon pimpinellifolium: evidence of evolutionary change in mating systems. PSE 127, 139170. Robert, T., Lespinasse, R., Perns, J. and Sarr, A. (1991) Gametophytic competition as inuencing gene ow between wild and cultivated forms of pearl millet (Pennisetum typhoides). Genome 34, 195200. Robertson, L.D. and Erskine, W. (1997) Lentil. In: Fuccillo, D., Sears, L. and Stapleton, P. (eds) Biodiversity in Trust. Cambridge University Press, Cambridge, pp. 128138. Rounsley, S., Lin, X. and Ketchum, K.A. (1998) Large-scale sequencing of plant genomes. Current Opinion in Plant Biology 1, 136141. Saghai-Maroof, M.A., Allard, R.W. and Zhang, Q. (1990) Genetic diversity and ecogeographical differentiation among ribosomal DNA alleles in wild and cultivated barley. Proceedings of the National Academy of Sciences USA 87, 84868490. Saghai-Maroof, M.A., Zhang, Q. and Chojecki, J. (1994) RFLPs in cultivated barley and their application in the evaluation of malting quality cultivars. Hereditas 121, 2129.

66

C. Spillane and P. Gepts Sasaki, T. (1998) The rice genome project in Japan. Proceedings of the National Academy of Sciences USA 95, 20212026. Sauer, J. (1993) Historical Geography of Plants. CRC Press, Boca Raton, Florida. Savidan, Y. (2000) Apomixis: genetics and breeding. Plant Breeding Reviews 18, 1386. Schmit, V. and Debouck, D.G. (1991) Observations on the origin of Phaseolus polyanthus Greenman. Economic Botany 45, 345364. Schmit, V., Dujardin, P., Baudoin, J.P. and Debouck, D.G. (1993) Use of chloroplast DNA polymorphisms for the phylogenetic study of 7 Phaseolus taxa including P. vulgaris and P. coccineus. Theoretical and Applied Genetics 87, 506516. Schmitt, J. and Linder, C.R. (1994) Will escaped transgenes lead to ecological release? Molecular Ecology 3, 7174. Schnable, P.S., Xu, X., Civardi, L., Xia, Y., Hsia, A.-P., Zhang, L. and Nikolau, B.J. (1996) The role of meiotic recombination in generating novel genetic variability. In: Sobral, B.W.S. (ed.) The Impact of Plant Molecular Genetics, Birkhauser, Boston, pp. 103110. Schoen, D.A. and Brown, A.H.D. (1991) Intraspecic variation in population gene diversity and effective population size correlates with the mating system in plants. Proceedings of the National Academy of Sciences USA 88, 44944497. Sears, B. (1980) Elimination of plastids during spermatogenesis and fertilization in the plant kingdom. Plasmid 4, 233255. Shands, H.L. and Weisner, L.E. (1991) Use of Plant Introductions in Cultivar Development, Part I. CSSA Special Publication no. 20, CSSA and ASA, Madison, Wisconsin. Shands, H.L. and Weisner, L.E. (1992) Use of Plant Introductions in Cultivar Development, Part 2. CSSA Special Publication no. 20, CSSA and ASA, Madison, Wisconsin. Sharma, H.C. (1995) How wide can a wide cross be? Euphytica 82, 4364. Sharma, D., Kaur, R. and Kumar, K. (1996) Embryo rescue in plants. Euphytica 89, 325337. Shoemaker, R.C. et al. (1986) Choroplast DNA variation in the genus Glycine subgenus soya. Journal of Heredity 77, 2630. Shorter, R., Lawn, R.J. and Hammer, G.L. (1991) Improving genetic adaptation in crops a role for breeders, physiologists and modellers. Experimental Agriculture 27, 155176. Simmonds, N.W. (1962) Variability in crop plants, its use and conservation. Biology Review 37, 442465. Simmonds, N.W. (1979) Principles of Crop Improvement. Longman, London. Simmonds, N.W. (1993) Introgression and incorporation. Strategies for the use of crop genetic resources. Biology Review 68, 539562. Simmonds, N.W. (1995) Food crops: 500 years of travels. In: CSSA, International Germplasm Transfer: Past and Present. CSSA Special Publication 23, pp.3145. Simmonds, N. and Smartt, J. (1995) Evolution of Crop Plants, Longman, London. Singh, A.K. and Nigam, S.N. (1997) Groundnut. In: Fuccillo, D., Sears, L. and Stapleton, P. (eds) Biodiversity in Trust. Cambridge University Press, Cambridge, pp. 114128. Singh, K.B. and Ocampo, B. (1997) Exloitation of wild Cicer species for yield improvement in chickpea. Theoretical and Applied Genetics 95, 418423. Singh, S. and Molina, A. (1996) Inheritance of crippled trifoliolate leaves occurring in interracial crosses of common bean and its relationship with hybrid dwarsm. Journal of Heredity 87, 464469 Singh, K.B., Pundir, R.P.S., Robertson, L.D., van Rheenen, H.A., Singh, U., Kelley, T.J., Parthasarathy Rao, P., Johansen, C. and Saxena, N.P. (1997) Chickpea. In: Fuccillo, D., Sears, L. and Stapleton, P. (eds) Biodiversity in Trust. Cambridge University Press, Cambridge, pp. 101113. Singh, S.P., Gutierrez, J.A., Molina, A., Urrea, C. and Gepts, P. (1991) Genetic diversity in cultivated common bean: II. Marker-based analysis of morphological and agronomic traits. Crop Science 31, 2329.

The Dynamics of Crop Genepools

67

Singh, S.P., Nodari, R. and Gepts, P. (1991) Genetic diversity in cultivated common bean. I. Allozymes. Crop Science 31, 1923. Singh, S.P., Molina, A. and Gepts, P. (1995) Potential of wild common bean for seed yield improvement of cultivars in the tropics. Canadian Journal of Plant Science 75, 807813. Slatkin, M. (1987) Gene ow and the geographic structure of natural populations. Science 236, 787792. Smale, M. (1997) The Green Revolution and wheat genetic diversity: some unfounded assumptions. World Development 25, 12571269. Smale, M. and McBride, T. (1996) Understanding global trends in the use of wheat diversity and international ows of wheat genetic resources. In: CIMMYT 1995/6 World Wheat Facts and Trends: Understanding Global Trends in the Use of Wheat Diversity and International Flows of Wheat Genetic Resources. CIMMYT, Mexico. Smartt, J. (1984) Gene pools in grain legumes. Economic Botany 38, 2435. Smartt, J. (1986) Exploitation of grain legumes. VI. The future the exploitation of evolutionary knowledge. Experimental Agriculture 22, 3958. Smartt, J. and Simmonds, N.W. (eds) (1995) Evolution of Crop Plants. Longman, London. Smith, B., (1995) The Emergence of Agriculture. Scientic American Library, New York. Smith, J.S.C. and Duvick, D.N. (1989) Germplasm collections and the private breeder. In: Brown, A.H.D., Marshall, D.R., Frankel, O.H. and Williams, J.T. (eds) The Use of Plant Genetic Resources. Cambridge University Press, Cambridge, pp. 1731. Snape, J.W. (1996) The contribution of new biotechnologies to wheat breeding. In: Reynolds, M.P.S., Rajaram, S. and McNab, A. (eds) Increasing Yield Potential in Wheat: Breaking the Barriers. CIMMYT, Mexico, pp. 167181. Snape, J.W., Law, C.N. and Worland, A.J. (1976) Chromosome variation for loci controlling ear emergence time on chromosome 5A of wheat. Heredity 37, 335340. Snape, J.W., Quarrie, S.A. and Laurie, D.A. (1995) Comparative QTL mapping and its application in cereal breeding. In: Proceedings FAO/IAEA International Symposium on the Use of Induced Mutations and Molecular Techniques for Crop Improvement. IAEA, Vienna, pp. 3949. Sniegowski, P. (1999) The genomics of adaptation in yeast. Current Biology 9, R897898. Somerville, C. and Somerville, S. (1999) Plant functional genomics. Science 285, 3803. Sonnante, G., Stockton, T., Nodari, R.O., Becerra Velsquez, V.L. and Gepts, P. (1994) Evolution of genetic diversity during the domestication of common-bean (Phaseolus vulgaris L.). Theoretical and Applied Genetics 89, 629635. Sperling, L., Scheidegger, U. and Buruchara, R. (1996) Designing Seed Systems with Small Farmers: Principles Derived from Bean Research in the Great Lakes Region of Africa. Agricultural Research and Extension Network Paper No. 60. ODI, London. Spillane, C. (2000) Can agricultural biotechnology contribute to poverty alleviation? AgBiotechNet, March 2000 (in press). http://www.agbiotechnet.com Spillane, C., Vielle-Calzada, J.-P. and Grossniklaus, U. (2000) Parent-of-origin effects and seed development: genetics and epigenetics. In: Khachatourians, G. et al. (eds) Handbook of Transgenic Food, Vol. 1. Srivastava, H.K. (1980) Heterosis for chiasma frequency and quatitative traits in common bean (Phaseolus vulgaris L.). Theoretical and Applied Genetics 56, 2529. Stalker, H.T. (1980) Utilization of wild species for crop improvement. Advances in Agronomy 33, 111147. St Clair, D.A. and Bliss, F.A. (1991) Intrapopulation recombination for 15N-determined dinitrogen xation ability in common bean. Plant Breeding 106, 215225. Stebbins, G.L. (1959) The role of hybridization in evolution. Proceedings of the American Philosophical Society 103, 231251.

68

C. Spillane and P. Gepts Stebbins, G.L. (1971) Chromosomal Evolution in Higher Plants. Addison-Wesley, Reading, Massachusetts. Sthapit, B.R., Joshi, K. D. and Witcombe, J. R. (1996) Farmer participatory crop improvement. III. Participatory plant breeding, a case for rice in Nepal. Experimental Agriculture 32, 47996. Suarez, M.C., Bernal, A., Gutierrez, J., Tohme, J. and Fregene, M. (2000) Developing expressed sequence tags (ESTs) from polymorphic transcript-derived fragments (TDFs) in cassava. Genome 43, 627. Summereld, R.J., Ellis, R.H. and Craufurd, P.Q. (1996) Phenological adaptation to cropping environment. From evaluation descriptors of times to owering to the genetic characterisation of owering responses to photoperiod and temperature. Euphytica. 281286. Summereld, R.J., Ellis, R.H., Craufurd, P.Q., Aiming, Q., Roberts, E.H. and Wheeler, T.R. (1997) Environmental and genetic regulation of owering of tropical annual crops. Euphytica 96, 8391. Svab, Z. and Maliga, P. (1993) High frequency plastid transformation in tobacco by selection for a chimeric aadA gene. Proceedings of the National Academy of Sciences USA 90, 913917. Svitashev, S., Bryngelsson, T., Vershinin, A., Pedersen, C., Sll, T. and von Bothmer, R. (1994) Phylogenetic analysis of the genus Hordeum using repetitive DNA sequences. Theoretical and Applied Genetics 89, 801810. Swaminathan, M.S. (1982) Beyond IR36 rice research strategies for the 80s. Paper presented at the CGIAR International Centres Week, World Bank, November 10, 1982, Washington, DC. Swaminathan, M.S. (1996) Farmers Rights and Plant Genetic Resources Recognition and Reward. MSSRF, Madras. Swennen, R. and Vuylsteke, D. (1991) Bananas in Africa: diversity, uses, and prospects for improvement. In: Ng, N.Q., Perrino, P., Attere, F. and Zedan, H. (eds) Crop Genetic Resources of Africa. IITA, Ibadan, Nigeria. Tanksley, S.D. (1993) Mapping polygenes. Annual Review of Genetics 27, 205233. Tanksley, S.D. and McCouch, S.R. (1997) Seed banks and molecular maps: unlocking genetic potential from the wild. Science 277, 10631066. Tanksley, S.D. and Nelson, J.C. (1996) Advanced backcross QTL analysis: a method for the simultaneous discovery and transfer of valuable QTLs from unadapted germplasm into elite breeding lines. Theoretical and Applied Genetics 92, 191203. Tanksley, S.D., Grandillo, S., Fultin, T.M., ZAmir, D., Eshed, Y., Petiard, V., Lopez, J. and BeckBunn, T. (1996) Advanced backcross QTL analysis in a cross between an elite processing line of tomato and its wild relative L. pimpinellifolium. Theoretical and Applied Genetics 92, 213224. Tanksley, S.D., Young, N.D., Paterson, A.H. and Bonierbale, M.W. (1989) RFLP mapping in plant breeding: new tools for an old science. Bio/Technology 7, 257264. Templeton, A.R. (1981) Mechanisms of speciation a population genetics approach. Annual Review of Ecology 12, 2341. Teutonico, R.A. and Osborn, T.C. (1995) Mapping loci controlling vernalisation requirement in Brassica rapa. Theoretical and Applied Genetics 91, 12791283. Thoday, J.M. (1961) Location of polygenes. Nature 191, 368370. Thro, A.M. and Spillane, C. (2000) Biotechnology-assisted participatory plant breeding: complement or contradiction? CGIAR System Wide Initiative on Farmer Participatory Research: Cali, SWP PRGA Working Document no. 4 , February 2000. www.prgaprogram.org/prga/ Tomooka, N, Lairungreang, C., Nakeeraks, P., Egawa, Y. and Thavarasook, C. (1992) Center of genetic diversity and disemination pathways in mung bean deduced from seed protein electrophoresis. Theoretical and Applied Genetics 83, 289293.

The Dynamics of Crop Genepools

69

Transue, D.K., Fairbanks, D.J., Robison, L.R. and Andersen, W.R. (1994) Species identication by RAPD analysis of grain amaranth genetic resources. Crop Science 34, 13851389. Uhr, D.V. and Goodman, M.M. (1995a) Temperate maize inbreds derived from tropical germplasm: I. Testcross yield trials. Crop Science 35, 779784. Uhr, D.V. and Goodman, M.M. (1995b) Temperate maize inbreds derived from tropical germplasm: II. Inbred yield trials. Crop Science 35, 785790. Urrea, C.A. and Singh, S.P. (1995) Comparison of recurrent and congruity backcross for interracial hybridization in common bean. Euphytica 81, 2126. Valkoun, J., Ceccarelli, S. and Konopa, J. (1997) Barley. In: Fuccillo, D., Sears, L. and Stapleton, P. (eds) Biodiversity in Trust. Cambridge University Press, Cambridge, pp. 191212. van Hintum, Th.J.L., Brown, A.H.D., Spillane, C. and Hodgkin, T. (1999) Core Collections of Plant Genetic Resources. IPGRI, Rome. Vavilov, N.I. (1951) The origin, variation, immunity and breeding of cultivated plants. Chronica Botanica 13, 1366. Vetelainen, M. (1994) Exotic barley germplasm: variation and effects on agronomic traits in complex crosses. Euphytica 79, 127136. von Bothmer, R., Flink, J., Jacobsen, N., Kotimaki, M. and Landstrom, T. (1983) Interspecic hybridization with cultivated barley (Hordeum vulgare L.). Hereditas 99, 219244. Wade, M.J. (1992) Sewall Wright: gene interaction and the shifting balance theory. Oxford Surveys of Evolutionary Biology 8, 3362. Walden, R. and Wingender, R. (1995) Gene-transfer and plant regeneration techniques. Trends in Biotechnology 13, 324331. Wang, G.-Z., Miyashita, N.T. and Tsunewaki, J. (1997) Plasmon analyses of Triticum (wheat) and Aegilops: PCR-single-strand conformational polymorphism (PCR-SSCP) analyses of organellar DNAs. Proceedings of the National Academy of Sciences USA 94, 1457014577. Wang, R.L., Stec, A., Hey, J., Lukens, L. and Doebley, J. (1999) The limits of selection during maize domestication. Nature 18, 236239. Wang, Z.Y., Second, G. and Tanksley, S.D. (1992) Polymorphism and phylogenetic relationships among species in the genus Oryze as determined by analysis of nuclear RFLPs. Theoretical and Applied Genetics 83, 565581. Waser, N.M. (1993) Population structure, optimal outbreeding, and assortative mating in angiosperms. In: Thornhill, N.W. (ed.) The Natural History of Inbreeding and Outbreeding. University of Chicago Press, Chicago, pp. 173199. Wehrhahn, C. and Allard, R.W. (1965) The detection and measurement of the effects of individual genes involved in the inheritance of a quantitative trait in wheat. Genetics 51, 109119. Welsh, W., Bushuk, W., Roca, W. and Singh, S.P. (1995) Characterization of agronomic traits and markers of recombinant inbred lines from intra- and interracial populations of Phaseolus vulgaris L. Theoretical and Applied Genetics 91, 169177. Wendel, J.F. (1999) Genome evolution in polyploids. Plant Molecular Biology 42, 225249. Whitham, S., McCormick, S. and Baker, B. (1996) The N gene of tobacco confers resistance to tobacco mosaic virus in transgenic tomato. Proceedings of the National Academy of Sciences USA 93, 87768781. Whitlock, M.C., Phillips, P.C., Moore, F.B.G. and Tonsor, S.J. (1995) Multiple tness peaks and epistasis. Annual Review of Ecological Systematics 26, 601629. Wijnheijmer, E.H.M., Brandenburg, W.A. and TerBorg, S.J. (1989) Interactions between wild and cultivated carrots (Daucus carota L.) in the Netherlands. Euphytica 40, 147154. Wilkinson, J.Q., Lanahan, M.B., Clark, D.G., Bleecker, A.B., Chang, C., Meyerowitz, E.M. and Klee, H.J. (1997) A dominant mutant receptor from Arabidopsis confers ethylene insensitivity in heterologous plants. Nature Biotechnology 15, 444447. Wilson, H.D. (1990) Gene ow in squash species. Bioscience 40, 449455.

70

C. Spillane and P. Gepts Wilson, H.D. and Manhart, J. (1993) Crop/weed gene ow: Chenopodium quinoa Willd. and C. berlandieri Moq. Theoretical and Applied Genetics 86, 642648. Wilson, A.C., Sarich, V.M. and Maxson, L.R. (1974) The importance of gene rearrangements in evolution: evidence from studies on rates of chromosomal, protein and anatomical evolution. Proceedings of the National Academy of Sciences USA 71, 30283030. Witcombe, J.R., Joshi, A., Joshi, K.D. and Sthapit, B.R. (1996) Farmer participatory crop improvement. I. Varietal selection and breeding methods and their impact on biodiversity. Experimental Agriculture 32, 44560. Wood, D. and Lenn, J.M. (1997) Conservation of agrobiodiversity on-farm: questioning the emerging paradigm. Biodiversity and Conservation 6, 109129. Wright, R.J., Thaxton, P.M., El-Zik, K.M. and Paterson, A.H. (1998) D-subgenome bias of Xcm resistance genes in tetraploid Gossypium (cotton) suggests that polyploid formation has created novel avenues for evolution. Genetics 149, 19871996. Wright, S. (1922) Evolution in Mendelian populations. Genetics 16, 97159. Wright, S. (1965) The interpretation of population structure by F-statistics with special regard to systems of mating. Evolution 19, 395420. Wych, R.D. and Rasmusson, D.C. (1983) Genetic improvements in malting barley cultivars since 1920. Crop Science 23, 10371040. Xu, Y., Beachell, H., Dilday, R. and McCouch, S.R. (1997). A Genome-wide View of Rice Germplasm. Plant Genome Conference, San Diego, CA, January 1216 1997. http://probe.nalusda.gov:8000/otherdocs/pg/pg5/abstracts/p-3b-62.html Young, N.D. and Tanksley, S.D. (1989) RFLP analysis of the size of chromosomal segments retained around the Tm-2 locus of tomato during backcross breeding. Theoretical and Applied Genetics 77, 353359. Yu, L. and Nguyen, H. (1994) Genetic variation detected with RAPD markers among upland and lowland rice cultivars (Oryza sativa L.). Theoretical and Applied Genetics 87, 668672 Zhang, Q., Saghai-Maroof, M.A. and Allard, R.W. (1990) Worldwide pattern of multilocus structure in barley determined by discrete log-linear multivariate analysis. Theoretical and Applied Genetics 80, 121128. Zijlstra, S., Purimahua, C. and Lindhout, P. (1991) Pollen tube growth in interspecic crosses between Capsicum species. HortScience 26, 585586. Zimmerer, K.S. and Douches, D.S. (1991) Geographical approaches to crop conservation: the partitioning of genetic diversity in Andean potatoes. Economic Botany 45, 176189. Zohary, D. (1965) Colonizer species in the wheat group. In: Baker, H.G. and Stebbins, G.L. (eds) The Genetics of Colonizing Species. Academic Press, New York. Zohary, D. and Hopf, M. (1993) Domestication of Plants in the Old World, 2nd edn. Clarendon Press, Oxford. Zweiger, G. (1999) Knowledge discovery in gene-expression-microarray data: mining the information output of the genome. Trends in Biotechnology 17, 429436.

3
1Crop

Base-broadening: Introgression and Incorporation

W. SPOOR1 AND N.W. SIMMONDS2
Science Department, Scottish Agricultural College (SAC), West Mains Road, Edinburgh EH9 3JG, UK; 29 McLaren Road, Edinburgh EH9 2BN, UK.

Introduction
Serious attention was paid to genetic resources of crops and to genetic diversity in general only after the beginnings made in the 1920s by the Russian school of plant collection led by N.I. Vavilov. Considerable collections of a few crops were built up locally and scattered round the world in the following decades, but really serious efforts were only begun in the 1960s; they were followed by the foundation of the International Board for Plant Genetic Resources (IBPGR, now IPGRI) in 1974, which has since played a leading role in work on genetic resources of crops. Though collections were widely regarded, in both Consultative Group on International Agricultural Research (CGIAR) and Food and Agriculture Organization (FAO) circles, as being highly desirable, even necessary, there was little understanding of how they might be used and to what purpose. In the 1970s and 1980s, indeed, there was widespread confusion about the function of curators of collections, who might themselves both characterize (i.e. identify) and evaluate (i.e. determine the utility of ) items in their care. It took a long time for the genetic resources community to realize that this idea was unworkable. In the longer run the point was taken that evaluation can only be for plant breeders working locally, because evaluation is beset by genotype environment problems, which cannot possibly be resolved, even understood, at a single (usually distant) place. It is now universally agreed that collections are best accumulated, cared for, maintained and characterized at chosen central sites but that utility/usefulness can only be determined by plant breeders working in distant places, in the sites to which adaptation is desired. Holden et al. (1984, 1993) and Bretting and Widrelechner (1995) comprehensively review the essentials. The foregoing does not mean, however, that collections went completely unused in earlier times. From the 1920s there were many examples of what we call introgression (borrowing a genecological word); that is, items from collections were chosen as having some desirable character (nearly always a disease resistance) and crossed and backcrossed to breeding stocks in the hope of transferring the desired character. This
IPGRI/FAO 2001. Broadening the Genetic Base of Crop Production (eds H.D. Cooper, C. Spillane and T. Hodgkin)

71

72

W. Spoor and N.W. Simmonds

sometimes worked but sometimes did not. All too often, the desired character was either polygenic or was a vertical resistance (VR), due to a major gene, readily nullied by evolution of a new pathogenicity in the pathogen. The wheat rusts, potato late blight, and leaf diseases of rice barley and grain legumes provide scores of examples. Generally, introgression has been widely practised but has only very rarely been useful. Thus, at least 11 R-genes have been backcrossed from Solanum demissum into cultivated potatoes, to no useful effect. To the plant breeder they are merely nuisances.

Recognition of Wider Functions

Recognition of the need to go beyond introgression to widen the genetic bases of diverse crops, either narrow at the start or seriously narrowed by subsequent selection, came much later and even now the concept has not been fully assimilated. The rst plain statement was that of Simmonds (1962), which coincided with the initiation of the AndigenaNeotuberosum potato project described later. The principle was simply to recognize that the large-scale loss of genetic variability due to response to selection had to be repaired somehow and that the best and most economical approach was to adopt a very widely founded base-broadening attack. The approach aims at wide incorporation of genetic variability which, of its very nature, cannot be pre-selected. It generally turns out that the earlier evolutionary history of the crop must be reconstructed if good new parents are to be synthesized. In effect, then, we recognize two basic ways of using a crop collection, namely: (i) introgression, adopting a pairwise crossing and backcrossing technique to introduce specic genes and (presumptively useful) characters; and (ii) incorporation, which attempts a wide-scale incorporation of genetic variability, the value of which cannot be known until the incorporation has been nearly achieved. In contrast to the preceding, incorporation must go forward on a wide genetic base, must in a sense repeat crop evolution and must usually be regarded as a continuing process. Figures 3.1 and 3.2 provide general illustrations: the former treats both introgression and incorporation; the latter elaborates on the theme of incorporation.

Functions of the Two Approaches

Introgression will have its uses but it will commonly continue to fail in future, as it has failed in the past. Either the desired character is not simply monogenic (and therefore not readily backcrossable), or it is a VR susceptible to the emergence of a new pathotype. It may be that major-gene virus resistances will prove to be more useful and durable than others because the agents may be less liable to generate new pathotypes, but this is quite speculative and far from certain as a generalization. By contrast, incorporation has indenitely wide applications and has yet been but little explored. The need sometimes arises from a narrow base ab initio, sometimes from a base narrowed by prolonged selection. A good deal is known about its application to potatoes, sugar cane and maize (Simmonds, 1993; and Figs 3.3 and 3.4; see also Kennedy, Chapter 16, this volume). Its use for barleys is plainly foreshadowed by the great Californian Composite Cross experiments, and its application to several tropical tree crops has been outlined

Base-broadening: Introgression and Incorporation

73

Basebroadening

Collection Poorly adapted

Local genetic base Well adapted Plant breeding

Incorporation

New populations

Introgression

Fig. 3.1. The general pattern of introgression and incorporation in crop improvement.
From Biological Reviews of the Cambridge Philosophical Society (68, 542, 1993) by kind permission of the Editor and Cambridge University Press.
X1 Enhanced crop

Advancement

Incorporation

Initial improvement

Whole base

Original narrow base X2

Fig. 3.2. The general pattern of incorporation in crop improvement.

Two arbitrary characters, X1 and X2, are plotted; they might well be major components of an index of general worth. Widths of circles roughly represent genetic variances.

74

W. Spoor and N.W. Simmonds

in, for example, rubber (Fig. 3.5), cocoa and oil palm. However, the great array of other crops, both temperate and tropical, that would benet from incorporation remain untouched, both experimentally and even conceptually. There has been very little awareness and thinking; the small grain cereals such as wheats, rices and millets and the inbred grain legumes are conspicuous examples of neglect. There have, however, been some signs of interest in soybeans.

Genetic Principles of Incorporation

The genetic principles underlying introgression are simple and obvious; those underlying incorporation are not. The latter may be listed as follows: 1. Large scale of operation is essential if a wide range of variability is to be caught in the net; a narrow start simply ensures an even narrower nish. 2. Non-adaptation of the introduced materials will usually have to be assumed, even if the assumption is sometimes mistaken. The primitive potatoes and sugar canes looked
Wild allopolyploids 2n = 48, 72 SC inbreeders e.g. demissum, acaule, stoloniferum

Wild diploids 2n = 24 SI outbreeders

Human selection for alkaloid-free tubers

Occasional allopolyploids with acaule triploid, AAC and pentaploid, AACCC

Cult. diploids 2n = 24 SI outbreeders

Cult. autotetraploids 2n = 48 SC outbreeders Andigena Group Cult. triploids 2n = 36

Evolution of variability

Migration to Europe

Experimental selection

*
Tuberosum Group 2n = 48 SC outbreeders

*
Breeding introgression

Dihaploid and diploid breeding 2n = 48

Tetraploid populations on expanded genetic base 2n = 48

Neotuberosum 2n = 48

Fig. 3.3. The long-term place of base-broadening (= incorporation) as a feature of potato evolution and current improvement.
Entries of a special consequence for incorporation are marked with an asterisk. From Smartt and Simmonds (1995, p. 467), by kind permission of Blackwell Scientic.

Base-broadening: Introgression and Incorporation

75

useless but many primitive-looking cereals and tree crops have attractions. The cereal pot trials described by Spoor and Simmonds (1993) provided some surprises: examined barleys generally appeared to have no value but this concealed a substantial minority that were intrinsically well adapted to Scotland 3. The process must be complementary to conventional breeding: testcrosses may sometimes be undertaken, but any general crossing with adapted materials is not only inappropriate but also inappropriate because it must corrupt the process of adaptation itself and will tend to corrupt, even destroy, the base-broadening effort. 4. The methods and breeding patterns to be adopted will depend on the biology of the crop, its breeding system and reproductive behaviour. The general rule must be quick and dirty but always within the bounds of security and reliability. The crops already at least partly understood, all demanded different methods (Simmonds, 1993). An attractive possibility, if it could be proved to be useful, would be to exploit pot tests (e.g. in barley: Spoor and Simmonds, 1993).
spontaneum wild, Southeast Asia (also Africa and Pacific) 2n = 40128 Miscanthus connection? robustum wild, New Guinea 2n = 60, 80

Natural crossing

officinarum primitive chewing cvs 2n = 80 Migration and somatic mutation

sinense thin cvs Southeast Asia, esp. India 2n = 82124

Nobilization, 1920s onwards Base-broadening, 1960s onwards

*
Nobilized populations 2n = 100125

officinarum noble cvs for sugar manufacture 2n = 80

Base-broadening, 1960s onwards

*
Contemporary breeding

* *
New nobilized populations 1970s onwards 2n = 100125

Fig. 3.4. The long-term place of base-broadening (= incorporation) as a feature of evolution and current improvement of sugar cane. (See also Kennedy, Chapter 16, this volume.)
Entries of a special consequence for incorporation are marked with an asterisk. From Simmonds (1976, p. 106), by kind permission of Blackwell Scientic.

76

W. Spoor and N.W. Simmonds

Locally adapted genetic base

Collections assembled

Collections conserved and analysed

Conventional cyclical GAM improvement

Test crosses, potential parents identified

2 Phenotypic selection 3

Improved clones in agriculture

INCORPORATION

1 Polycrosses, selection, genetic enhancement D2

Genetic worth (G) and diversity (D)

D3 G3 D2

D1

G2 G1

1920s

1980s Date

>2000

Fig. 3.5. The genetic base and use of base-broadening (= incorporation) in rubber breeding.
This programme is projected rather than actually realized but it observes the principles set out on pages 7477 and the analogy with potatoes and sugar canes (Figs 3.3 and 3.4) will be clear. From Webster and Baulkwill (1989, pp. 116117) by kind permission of Blackwell Scientic. GAM:

Base-broadening: Introgression and Incorporation

77

5. It is prudent to assume little inherent adaptation in the material, and therefore to use large populations, to maximize recombination (by cyclical crossing if necessary) and to maintain only weak selection. Rigorous selection can only narrow the base unnecessarily, but usually without concomitant gain. 6. Utilization of such broadly based populations will require considerable development of local adaptation before true potential can be recognized. To minimize undue narrowing of the base and further genetic loss, replication over several diverse sites is highly desirable. 7. An important feature of local genetic adaptation is horizontal resistance (HR) to disease. Most important crops have 510 (or more) more or less serious diseases, and new ones regularly appear (as in sugar cane). Usually, it will be sufcient to grow populations in disease hot-spots and let natural selection do the work; sometimes, however (as in potatoes and sugar cane), some intervention by the plant breeder is needed. The profound importance of HR (generally, but especially for the small farmer) has been strongly emphasized by FAO (1986, 1988) and genetic features have been reviewed by Simmonds (1991). 8. Progress may well be remarkably rapid on evolutionary time-scales (it was in potatoes and sugar canes) but seemingly rather slow on human/plant breeding time-scales. This is unavoidable but it implies a long-term commitment and assured continuity in relevant programmes. But, even in years, time scales are not always long and both potatoes and canes showed major advances in a decade: a mere moment in crop evolutionary time (Figs 3.3 and 3.4). Strictly annual crops could well be quicker. 9. The outcome of an effective base-broadening programme will be enhanced genetic variance in economic characters (Figs 3.1, 3.2 and 3.5) and either good materials per se or good parents for crossing into established programmes. Both outcomes are known and both are to be expected. A closing comment is justied because it is often a source of difculty. The term prebreeding is used in various senses and should, in our opinion, usually be avoided. It may refer to the development of potential parents either from adapted stocks or from what might be regarded as a base-broadening programme. Only confusion follows from loose usage.

An Example from Potatoes

Frequent reference has been made above to base-broadening of potatoes, perhaps the leading example of incorporation because it was the earliest and arguably the most successful. The work was summarized, with references, in Simmonds (1993). Figure 3.3 herein illustrates the evolution of the group, including base-broadening. The potatoes are ancient Andean cultivars which started as diploids (2x = 24) from diploid ancestors; they are all outbred and self-incompatible (SI). In the Andes, at great altitudes, autotetraploid derivatives (4x = 48), the Andigena group (ADG), came to dominate and still do. They are adapted to short tropical days but to cool weather. A few ADG cvs were taken to Europe in the 16th century and slowly became adapted to temperate climates though they tended to be late (because of short-day

78

W. Spoor and N.W. Simmonds

adaptation) and were very susceptible to diseases. In Europe, and also in North America, local adaptation emerged in the Tuberosum group (TBR), which is effectively Andigena with minor morphological differences and adaptation to long summer days. The importance of day-length adaptation cannot be over-emphasized. Very few ADG were brought to the north and the crop made good genetic progress on a very narrow genetic base up to the middle 19th century when it was catastrophically attacked by the newly introduced late-blight fungus, Phytophora infestans, introduced from Mexico to Europe and North America in the 1840s. By then, TBR was a major food crop. Blight still further narrowed the base, though the crop responded well to selection. Pedigree information showed that millions of hectares of a few clones all went back to half a dozen or so introductions. Attempts to use ADG for direct crossing to TBR were useless; products were sometimes vigorous but had many bad features, including disease susceptibility. The only solution, and the method in fact adopted, was to recreate the TBR group from ADG stocks, of which a large sample was available in the Commonwealth Potato Collection at the John Innes Institute, Hertford, England. An initial sample (1959) of about 3300 seedlings was grown, selected roughly and the selections carried on vegetatively in the next year. Thereafter seedling and tuber propagation alternated; seed was collected from isolated tuber plots in the hope that it was mostly crossed, even though the ADG potatoes are fairly self-fertile. This was a mistake, later remedied by making crosses, when it was found that selng predominated. Little conscious selection was practised, though plots were deliberately grown every year in a blight hot-spot. Responses were very largely to natural selection for survival, growth and yield (i.e. tolerance of long days), with minimal interference by the breeder. Daylength adaptation and disease resistance improved very rapidly and in three 2year generations (i.e. before 1970), the better materials rivalled TBR in performance, had a great range of tuber types, cooking qualities and resistances in them and showed some truly spectacular heterosis in crosses with TBR. So good were they that the name Neotuberosum (NTB) was invented for them. In the 1970s about 60,000 plants were grown, so the demands, though non-trivial, were not great. The work was carried on in Scotland in later years and many crosses made into a conventional breeding programme. Stocks were widely distributed round the world. A few current commercial varieties in the UK, Europe and North America are half NTB in constitution and many more are in prospect. The material is, of course, freely available to bona de potato breeders and, if wisely used, can hardly fail to revolutionize potato breeding, temperate and tropical. It may be that the greatest potential of all will be in the tropics, for Third World agriculture (Simmonds, 1997). Diploid selections can readily be introduced into the programme if required (Fig. 3.3).

Summary and Conclusions

Many crop plant collections were built up and maintained in various places over the last 6070 years of the 20th century. They have provided some interesting scientic data but have been but little exploited for plant breeding. The evidence of the need for base-broadening of a crop plant lies in historical information, pedigrees and poor progress under selection.

Base-broadening: Introgression and Incorporation

79

Two means of using a collection may, broadly, be distinguished: introgression implies the introduction of a character by backcrossing a single gene or a small group of polygenes; incorporation implies the large-scale introduction of genetic variability by mass crossing followed by mass selection under some approximation to random mating. The former has been widely used for monogenic disease resistances (usually VR) which have generally failed, for obvious reasons. The latter has been successfully exploited in potatoes and sugar cane and is being ever more widely adopted as a longerterm strategic measure for the improvement of other crops. In principle, it is applicable to all crops, and a great many (e.g. all the inbred cereals) would benet if so treated. Incorporation, if correctly exploited, generates substantial reserves of adapted parental materials for incorporation in plant breeding programmes. It may itself generate new crop varieties but is not primarily intended to do so: its function is to widen the genetic base.

References
Bretting, P.K. and Widrelechner, M.P. (1995). Genetic markers and plant genetic resource management. Plant Breeding Reviews, 13, 1186. FAO (1986). Breeding for Durable Resistance in Perennial Crops. Plant Production and Protection Paper 70. FAO, Rome. FAO (1988). Breeding for Durable Disease and Pest Resistance, 2nd edn. Plant Production and Protection Paper, 55. FAO, Rome. Holden, J.H.W. and Williams, J.T. (eds) (1984) Crop Genetic Resources: Conservation and Evaluation. Allen and Unwin, London. Holden, J.H.W., Peacock, J. and Williams, J.T. (1993) Genes, Crops and Environment. Cambridge University Press, Cambridge. Simmonds, N.W. (1962) Variability in crop plants, its use and conservation. Biological Reviews 37, 422465. Simmonds, N.W. (1976) Evolution of Crop Plants. Longman, London, and Blackwell, Oxford. Simmonds, N.W. (1991) Genetics of horizontal resistance to diseases of crops. Biological Reviews 66, 189241. Simmonds, N.W. (1993) Introgression and incorporation. Strategies for the use of crop genetic resources. Biological Reviews 68, 539562. Simmonds, N.W. (1997) A review of potato propagation by means of seed as distinct from clonal propagation by tubers. Potato Research 40, 191214. Smartt, J. and Simmonds, N.W. (1995) Evolution of Crop Plants, 2nd edn. Longman, London, and Blackwell, Oxford. Webster, C.C. and Baulkwill, W.J. (1989) Rubber. Longman, London, and Blackwell, Oxford.

4
1Institut

The Role of Local-level Geneow in Enhancing and Maintaining Genetic Diversity

J. BERTHAUD1, J.C. CLMENT2, L. EMPERAIRE3, D. LOUETTE4, F. PINTON5, J. SANOU6 AND G. SECOND1
de Recherche pour le Dveloppement (IRD), 911 Avenue Agropolis, BP 5045, 34032 Montpellier cedex, France; 2IRD, 213 Rue Lafayette, 75480 Paris cedex 10, France; 3IRD/ISA, SCLN 210, Bloco C, sala 112, 70862-530, Brasilia (DF), Brazil; 4Chemin de la Croix, 30260 Brouzet les Quissac, France; 5Universit Paris-10 Nanterre, Ladyss, Bat. K, 92001 Nanterre cedex, France; 6INERA/CRRA de Farako-B, 01BP910, Bobo-Dioulasso 01, Burkina Faso

Introduction
In traditional agricultural systems, it is often thought that plant varieties are conserved without evolution over long periods of time by farmers because local-level geneow and its role in maintaining and enhancing genetic diversity of crops is often underestimated. Local-level geneow can be dened as gene exchange between varieties at the local level, i.e. within farmers elds and rural communities. In this chapter we present a few case studies of seed management by farmers in different locations and production systems which demonstrate the impact of farmer activity in the management of plant diversity. Knowledge related to farmer management is still limited but is required when we want to propose strategies for broadening the genetic base of crops which take advantage of this existing former management or which produce new genetic material to be included in such management. The selected cases are not typical of all the situations encountered but offer the opportunity to stress specic and important points. Some of these cases were presented in Bamako during the workshop: Genetic resources management in savannah of Africa in March 1997 (Anonymous, 1997).

IPGRI/FAO 2001. Broadening the Genetic Base of Crop Production (eds H.D. Cooper, C. Spillane and T. Hodgkin)

81

82

J. Berthaud et al.

Case Studies
Maize in a traditional community in Mexico The valley of Cuzalapa and the Sierra de Manantln Biosphere Reserve (SMBR) This biosphere reserve was established in a area where teosinte, the wild parent of maize, is found, and especially where the species called Zea diploperennis occurs. It is an area of traditional agriculture in a centre of maize diversity. The indigenous community of Cuzalapa is located in a valley in the southern section of the buffer zone of the Sierra de Manantln Biosphere Reserve (SMBR), in the municipality of Cuautitln, in the state of Jalisco, on the Pacic Coast of Mexico. The Cuzalapa watershed covers nearly 24,000 ha (most of which lie within the boundary of the biosphere reserve) of mountainous land of extremely irregular topography, ranging from an elevation of 550 m to 2660 m. Each year, about 1000 ha of maize may be sown in Cuzalapa; of this area, 600 ha are irrigated (Martinez and Sandoval, 1993) using traditional techniques based on gravity. Cultural practices continue to be relatively traditional when compared with those found outside the Sierra de Manantln. Maize (Zea mays ssp. mays) is the dominant crop in the valley. It is grown during the rainy season from June to November, associated with squash (Cucurbita spp.), on an average of 2 ha per farmer. It is also planted under irrigation in the dry season, extending from December to May, intercropped with beans (Phaseolus vulgaris), on an average of more than 2 ha per farmer. Irrigation and intercropping are reported to have been a feature of agriculture in Cuzalapa in pre-colonial times (Laitner and Benz, 1994). Maize and beans are produced essentially for home consumption; some surplus is sold at local markets, yet the Cuzalapa community is poorly linked to commercial markets. Cuzalapa is located in one of the most marginalized municipalities of the region, in terms of housing quality and level of education (Rosales and Graf, 1995). At the time of the study (19891991), these localities were all remote from main roads and urban areas. Based on its farming and socioeconomic characteristics, Cuzalapa is broadly representative of many indigenous, poor and isolated rural areas in Mexico. It is one of the many traditional communities which are being slowly incorporated into commercial marketing systems while maintaining features of indigenous society. Documenting the exchange of seed lots and varieties To document which maize varieties are cultivated and to record the exchange of seeds and varieties in the community and between the valley of Cuzalapa and other regions, 39 farmers (one-fth of Cuzalapa farmers) were surveyed during six cropping seasons covering three calendar years (the 1989, 1990 and 1991 rainy and dry seasons). For each farmer and cropping season, data were collected on varieties cultivated and seed sources. Varieties included those grown on farmers own elds, on rented elds and shared cropping elds. Each variety was recorded with the name given by the farmer. The seed source was classied into three classes: (1) own seed (seed selected by farmers from their own harvest); (2) seed acquired in Cuzalapa (seed obtained in the valley of Cuzalapa from another farmer); and (3) introduced or foreign seed (acquired outside the Cuzalapa watershed). The data therefore give a clear representation of the

Local-level Geneow and Genetic Diversity

83

extent of seed exchange and of local level of geneow, but they understate the importance of seed of foreign origin introduced in Cuzalapa because seeds acquired in Cuzalapa can be of foreign origin in the previous generation. Regular introduction of new varieties This community does not function as an isolated area. On the contrary, foreign varieties are regularly introduced to be tested. From the 26 varieties identied, only the varieties Blanco, Amarillo ancho, Negro, Tabloncillo, Perla and Chianquiahuitl are local (Table 4.1); the other varieties are foreign. Local means that varieties had been grown continuously for at least one farmer generation in the valley of Cuzalapa. Only the introduction date of the Chianquiahuitl can be traced back to the 1950s and all the varieties can be classied as Tabloncillo race. Although few in number, the local varieties cover more than 80% of the area planted with maize. Blanco, Tabloncillo and Perla are white-grained and used essentially to make tortillas, the Mexican staple food. Amarillo ancho is yellow-grained and used for feeding poultry. Negro is purple-grained and is essentially consumed roasted at the milky stage. The remaining 20 of the 26 varieties that Cuzalapa farmers were growing during the survey period are classied as foreign. Each foreign variety covered less than 5% of the maize area planted in each season, and most varieties were cultivated by only a few farmers at a time. The composition of this group of varieties changed from season to season. Only three of these varieties (Argentino, Enano and Amarillo) had been regularly cultivated over the preceding 4 or 5 years by a signicant percentage of farmers (1012%). Most varieties had been used for the rst time recently or during the survey period and had only been planted once or twice. The origin of the foreign varieties is often difcult to ascertain. Farmers are able to indicate in which community they acquired a variety, but not its original source. Even the original name of the variety can disappear or take on a different meaning when farmers exchange seed. Based on the information collected, foreign varieties can be

Table 4.1. Importance of 26 varieties cultivated in Cuzalapa (Mexico). Grain colour White White White White Yellow Purple

Varieties Six local Blanco Chianquiahuitl Tabloncillo Perla Amarillo ancho Negro 20 foreign 3 most cultivated Argentino Enano Amarillo 17 minor varieties

% maize area 51 12 5 0.4 8 3

% farmers 59 23 6 0.2 23 34

5 3 3 < 3 per variety

10 12 11 < 4 per variety

White White Yellow Mainly white

84

J. Berthaud et al.

classied into three groups: 15 farmers varieties (landraces); four farmers advanced generations of improved varieties; and one variety which was a recent generation of an improved variety. The group of foreign varieties is morphologically diverse, including white-, yellow- and purple-grained material. Representatives of different races are found in this group. Most varieties were introduced from communities of southwestern Jalisco, less than 100 km from Cuzalapa, although a variety cultivated by one farmer originated in the United States. Variable composition of a variety Signicant differences in origin were associated with the dominance of the variety in terms of planted area (Fig. 4.1). Seed of the most widely grown varieties including the local varieties and the three most important foreign varieties is less likely to have been obtained from farmers outside Cuzalapa than seed of the more minor foreign varieties (7.9% of local varieties and 5.3% of important foreign varieties seed lots were introduced, compared with 36.4% of minor foreign varieties seed lots). Nevertheless, it is difcult to establish a pattern for the minor foreign varieties, because each variety appears to be a special case, dened by the time of its introduction and the number of farmers planting it. Seed of local varieties is essentially reproduced by each farmer. Among local varieties, farmers manage the seed for Chianquiahuitl and Negro more conservatively. More than 70% of the seed for these varieties is selected from farmers own maize harvests. In fact, farmers plant such a small area with the variety Negro that, on average, seed equivalent to only 27 ears is required per farmer to plant their area (Louette, 1994). This amount of seed, if in good condition, is carried over easily from one cycle to the next, and farmers do not need to get seed from another farmer. Chianquiahuitl is a variety of unknown origin that is believed to be no longer widely cultivated outside the study zone, and farmers in the Cuzalapa Valley must rely on their own stocks. The case of Blanco contrasts with that of Chianquiahuitl. Of all the local varieties, Blanco has the highest proportion of seed obtained from farmers outside the study zone (15%). This result reects the importance of Blanco in terms of area cultivated in Cuzalapa and regions nearby. Because Blanco is important for household subsistence, an insufcient number of ears suitable for seed may remain at planting time. Farmers then search for seed from other farmers in and outside the community. We have considered as a seed lot all kernels of a specic type of maize selected by a farmer and sown during a cropping season to reproduce this particular type of maize. Farmers use an important percentage of their own seed lots for both major and minor foreign varieties (42.1 and 39.4%, respectively), in the rst case to reproduce them and in the second case to test foreign varieties over several seasons. Major and minor foreign varieties can be distinguished by their pattern of diffusion. The percentage of seed brought from other regions is small for the most widely grown foreign varieties (5.3%), while for some of the minor varieties introduced late in the survey period all seed lots were introduced (average 36.4%). On the other hand, farmers in the valley much more frequently exchange seed of the major (52.6%) than of the minor foreign varieties (24.2%). Major varieties were introduced some years ago by these local farmers and, because they have demonstrated characteristics of value, their seed is redistributed to other farmers in Cuzalapa. In contrast, survey farmers who did not plant the minor

Local-level Geneow and Genetic Diversity

85

70 60 50 % of seed lots 42.1 40 30 20 10 0 Local Foreign major Origin of varieties Seed acquired in Cuzalapa Foreign minor 7.9 5.3 34.2 24.2 39.4 36.4 57.9 52.6

Own seed

Introduced seed

Fig. 4.1. Origin of seed planted in Cuzalapa by origin of variety.

varieties during the study period may not yet be convinced of their advantages and do not look for seed. In summary, for the local varieties, there is a moderate level of diffusion inside the watershed and little infusion from other regions. Recently introduced foreign varieties are infused from outside the valley. Older foreign varieties that have attained a moderate level of acceptance are also diffused inside the watershed. The pattern of diffusion of the varieties is therefore linked essentially to the local acceptance of the variety, the time it has been introduced in the region, and the availability of seed inside and outside the region. The general patterns of maize seed exchange that we have just described conceal major differences among survey farmers. Three major groups can be identied. At one extreme, we nd 17 farmers who always use seed selected from their own maize harvests. They sow the same varieties regularly and only modify the proportion of maize area planted with each variety in every cropping season. These farmers are considered suppliers of seed of local varieties (they always have seed). Fourteen farmers use their own seed lots in addition to seed acquired in the community or introduced from other regions, and the proportion of each type of seed varies from season to season depending on each farmers objectives and constraints. These farmers are generally regarded as suppliers of introduced seed, and some are known in the community for their curiosity concerning new varieties. At the other end of the spectrum are six farmers who have never used their own seed lots (0% of own seed lots) and had recourse throughout the study period to seed acquired within and outside the Cuzalapa community. This group of farmers includes those who do not have rights to land and cannot plant maize each season and those who farm small areas on which they cannot harvest enough maize for both family con-

86

J. Berthaud et al.

sumption and seed. Farmers in this group are obliged to look for seed from other farmers when they want to plant maize. Geneow The survey and the observation of the sowing pattern in an area of 10 ha during three cultivation seasons indicate that there is no traditional strategy to actively prevent the sowing of seed lots of different varieties in contiguous areas. Farmers sow on average 2.5 varieties (one to seven) per cycle in the same eld, independently of those sown on the contiguous elds. There is no physical isolation between local and foreign varieties and between locally reproduced seed lots and seed lots originating from other areas. In addition, the date of sowing does not lead to a sufcient difference of owering dates to permit reproductive isolation Phenotypic observations on six farmers elds conrmed the presence of geneow and gave an idea of the level at which it occurs. It was observed, as indicated by literature (Paterniani and Stort, 1974), that the level of introgression of one variety by another diminished rapidly with distance. We observed the presence of 1020% of purple grains in the rst row of the white or yellow variety, levelling down to 1% after the rst 2 or 3 m. The level then stabilized over a great distance. In Cuzalapa, the traditional management of sowing, leading to the development of different varieties on contiguous areas, favours genetic exchange between all variety types, independently of the origin and growing cycle of the different varieties.

Maize in Burkina Faso Crop management Maize is cultivated in Burkina Faso in the southern region where the annual rainfall is over 900 mm and in a secondary zone with a lower annual rainfall of 600800 mm. According to FAO, during the last decade, maize has been cultivated on 200,000 ha with an average of 1.31.6 t ha 1. The crop was introduced in Africa in the 16th century from germplasm of various origins (Chastenet, 1998). Since then, this crop evolved in this region and became locally adapted. New introductions have been very limited before this century. Surveys and collections in Burkina Faso have shown that maize varieties are strongly structured, based on genetic and cultural components (Sanou et al., 1997). Based on several criteria (Table 4.2), three types of elds have been characterized: 1. Backyard elds, very small, high organic fertilization (manure from cattle). These elds are cultivated by women farmers. Varieties in these eld are specic, called household type, very early maturing and used to bridge the gap between main harvests. 2. Bush elds. They can be located far away from the village, are of a large size, 310 ha, and are relied upon for the main maize production. They are managed by men. The corresponding varieties are late maturing and produce grain for local consumption and for urban markets. They are of the open-eld type. 3. Village elds. They are intermediate, in size and location, between the two former types. In these elds, open-eld, late-maturing varieties are also used.

Local-level Geneow and Genetic Diversity Table 4.2. Main characteristics of the elds cultivated with maize in Burkina Faso. Field typology Backyard eld (0.10.5 ha) Area Fertilization Type of cultivation Main and secondary zone +++ Intercropping Village eld (25 ha) Main and secondary zone ++ Intercropping or single crop Open eld CV ++ Bush eld (310 ha) Main zone

87

Maize variety types Household CV Competition between local and elite varieties or cash crops Sex of farmers Female , Absent; +, some; ++, high; +++, very high.

+ Single crop or intercropping Open eld CV +++

Male

Male

Maize is mainly intercropped with cowpea and groundnut, but also with cotton, pearl millet or yam. Origin of varieties Both open-eld and household types of varieties are considered family heritage and are transmitted through the male lineage, but in some cases they can be exchanged or borrowed. Weddings offer such opportunities: women introduce the household varieties of their family into the collection of varieties of their family-in-law. In other cases, it is a specic trait of a variety that leads to the exchange (early maturing, culinary properties, etc.). Many of these exchanges involve household varieties for cultivation in backyard elds and transfer is mostly from the marginal to the main (southwest of Burkina Faso) area of cultivation. Elite varieties and local varieties Experiments were conducted in Burkina Faso to compare elite varieties (SR22 from the International Maize and Wheat Improvement Centre (CIMMYT), Maka, Irat 171, from the Centre de Coopration Internationale en Recherches Agronomiques pour le Dveloppement (CIRAD)) with local household (early maturing) and local open-eld varieties in conditions comparable to traditional agriculture (with no fertilizer) and intercropping on one hand, and to intensive agriculture (using fertilizers, single crop and high density planting) on the other hand (Sanou, 1996). In the conditions of traditional agriculture, most of the local varieties had a better productivity than the best elite variety (SR22) selected in a single-crop environment. In the conditions of intensive agriculture, many local varieties (60%) had a production very similar to the elite variety. This was especially true for the early maturing varieties. However, a higher planting density (from 50,000 to 100,000 plants/ha) affected the local varieties more than the elite variety. Intensifying the maize cropping system with local varieties could bring lodging problems. Complex hybrids produced between local varieties and the elite variety SR22 showed an improvement of productivity of 12 t ha 1, a small

88

J. Berthaud et al.

increase in height (less than 20% for ear height) and a better reaction to Maize Streak virus when compared with their respective local variety parents. This improvement is produced by the favourable traits existing in the elite variety. Geneow between elite and local varieties Differences were already mentioned between local household and open-eld varieties. Almost no substitution from elite varieties is observed in this group. Elite varieties are selected for traits in relation to their commercial interest: productivity, grain colour, late maturing. They can substitute for eld varieties but not the household varieties. Geneow was evaluated by: 1. Observation of cultivation practices (Fig. 4.2). Backyard, village and bush elds constitute separated entities but it was observed that farmers establish temporary shelters in their bush elds and plant household varieties around them for their own consumption before the harvest of bush elds. It is an opportunity for the dissemination of genes of the household early material. 2. Use of isozyme markers. The elite variety SR22 has some specic isozyme alleles IDH1-2 and IDH2-4,2 which are not found in the local varieties. On 100 local varieties tested, ten showed introgression from the elite variety. These ten varieties were found in villages where the elite variety was widely distributed and grown in rotation with cotton or where the extension service had established demonstration elds. Contrary to what might have been predicted by observation of cultivation practices, introgression from the elite variety into the household varieties occurred (Sanou, 1996). Evolution in this system In Burkina Faso, the local varieties are structured in two groups. Geneow facilitates gene transfer from one group to the other to a limited extent. The household variety group is relatively autonomous when compared with the eld varieties, but within each group gene exchange can be very extensive. In planning to produce new genetic material for this crop management system, it is important to take into account the geneow pattern. The nal product (i.e. new varieties) has to meet the criteria used by farmers to choose their existing varieties (maturing time, culinary quality, etc.). Dynamic management of this diversity and participatory breeding should therefore be conducted on both pools separately. To broaden the genetic base for utilization outside the area, it is possible to put these two pools together. However, the new products will be very different from what currently exists in this production area. Conclusion The genetic structure we observed for maize varieties in this area is derived from the social structure of the communities that use them. This social structure surely buffers possible changes at the level of the genetic material to be used by farmers. At the same time, it does not freeze evolution of these varieties but promotes a gradual and continuous crop evolution.

Local-level Geneow and Genetic Diversity

89

Fig. 4.2. Typical maize eld distribution and geneow in the main cultivation area in Burkina Faso. 1, Backyard elds with household cultivars; 2, village elds with open eld cultivars; 3, bush elds with open eld cultivars; 4, household cultivars around temporary settlements.

Pearl millet landraces in West Africa The results in this section are based on surveys conducted by one of the authors (J.-C. Clment) in the Sahelian region of West Africa, from Mauritania to Chad, over a 20year period to collect pearl millet landraces (Clment, 1985, 1997) and complement the work by Niangado (Chapter 9, this volume). During these surveys, both genetic material and information from the farmers of this area were collected; in several regions, collections were made several times between 1968 and 1988. During this period, this part of Africa experienced very severe droughts. From the available data we can describe the factors that had most impact on stability and variation in this plant material. The large intra- and intervarietal polymorphism noted in pearl millet varieties of this region is expressed in traits which are used as selection criteria by farmers. They are mainly inorescence agromorphological traits and include: inorescence shape, grain colour, size and vitrosity and non-shattering of spikelets. These traits can be seen by the naked eye and used to distinguish the main phenotypes at the local and regional scale. Another trait, the duration of the sowing to harvest period, is also useful for discriminating between varieties and relates to the daylength sensitivity of the varieties. Early maturing varieties do not respond to photoperiod, while late-maturing varieties are quite sensitive to daylength. Millet landraces of Mali display a unique range of maturity in West Africa. In the area of Kolokani (Mali), farmers cultivate several varieties independently in adjacent plots: early-maturing varieties called sounas (souna di, souna boab) maturing in 7595 days, and three late-maturing varieties called sanios maturing in 100160 days (sanio tiotioni, sanio makangoulou, sanio boab). This range of maturity is associated with some overlapping of owering times which could facilitate gene exchange between cultivated varieties (early and late maturing) as well as with wild and weedy forms of this species.

90

J. Berthaud et al.

Stability of cultivated types and areas When data from the surveys from the 1980s and from older sources are compared, two facts are worth noting: a stability in the area of cultivation of the main types and a surprising phenotypic stability of the various types cultivated in these areas. This stability is especially noted in the central area Mali, Burkina, Niger where the largest West African varietal diversity is found. In this area, landraces and families of varieties are still cultivated with their former characteristics. Generally, the main varietal types in this central area are still morphologically distinct despite adjacent or overlapping areas of cultivation. This can be explained by farmers selecting seed only from inorescences corresponding to the types they are interested in. In fact, the farmer strategy for seed production is related to a mass selection within each type of variety, renewed at each generation. The on-going search for better food security (i.e. to lower the consequences of climate variation on production) leads to the utilization and the maintenance of the broadest range of varieties and maturity times, making these crop management systems in situ genetic reservoirs. Genetic isolation can also exist. In southeast Burkina Faso, North Togo and Benin a variety named Iniadi is cultivated: its inorescence is short and conical and it matures quickly, in 90 days. This variety is the only early one within a set of late varieties that are well adapted to a climate with more than 1100 mm rainfall. The genetic isolation conferred by this different owering time could explain the stability of this landrace and its specicity, which makes it one of the main genetic groups used for F1 hybrid production in pearl millet. Andrews and Kumar (1996) have described the use of this variety in the F1 hybrid production in the USA, India and Africa. Using isozyme markers, Sdogo and Tostain (1996) have shown that this variety has strongly diverged from the others found in this area. Variation and evolution of landraces in relation to climatic changes Monitoring changes in the crop management systems and varieties is possible in this region where many surveys were conducted over several decades. In areas where early, semi-early and late varieties are planted, the latter two types usually evolve toward a shorter maturing time. This is a farmer strategy to adapt to climatic changes. This strategy favours progressive genetic changes within varieties more than a substitution of varieties. In pearl millet, the genetic control of maturity is relatively simple and selection can be quite effective for this trait (Bilquez, 1963; Bilquez and Clment, 1963). In Senegal the sanio landrace is no longer cultivated with its former maturity time of 140160 days. However, it can still be found in some areas such as the Serere country, where farmers modied it for a shorter maturity time of 120130 days maintaining all the other traits typical of this landrace. In southeast Niger, the semi-early variety maewa (120130 days) has become a 100120 days variety and grain colour has changed from blue-grey to yellow or grey-yellow.
EVOLUTION TOWARDS EARLINESS DROUGHT AND INTRAPOPULATION MIXING In Mauritania, due to the severe drought, farmers stopped cultivating pearl millet and replaced it with sorghum. Their millet seeds were lost. When climate became more favourable, they planted their elds with

Local-level Geneow and Genetic Diversity

91

varieties from adjacent areas where the cultivation had been maintained or from a souna landrace cultivated in West Senegal and distributed by the relief teams. Landraces cultivated in the south and west part of Mauritania were related to sounas from the Senegal river; those cultivated in the east were related to sounas from Mali. Loss of seed due to drought promoted a new mixing of genes in the area when the pearl millet cultivation was re-established from these two souna types and their intermixing. In Niger, a similar phenomenon was noted, where cultivation favoured a mixing within groups of varieties along a northsouth axis.
GENETIC CHANGE AND CLIMATE Severe drought was experienced by farmers from 1968 to 1973, 1976 to 1979 and 1983 to 1985, and caused substantial losses of millet types. However, its consequences should be placed in perspective with other factors. In the area of Toubouro in Cameroon and Bossangoa in the north of the Central African Republic, the disappearance of pearl millet is due to a change in agricultural activity redirected towards cotton production. In Guinea, in the Kankan area and in the south of Sigiri, pearl millet is being abandoned and replaced by rice and maize. A similar case can be observed in the east of Senegal. In the area of the Senegal river, from Bakel to Saint Louis, the tiotande, an after ood (decrue) specic ecotype, is now almost a relict. It is replaced by maize and sorghum with a higher production potential and a better resistance to pests and diseases. Sorghum is also replacing pearl millet in many typical Sahelian areas (500 to 600 mm rainfall) in Mali.

Conclusion For pearl millet, detailed surveys have been conducted on a very large area and over a long period of time. This area, the Sahel, with its low rainfall is relatively marginal for agriculture and often suffers climatic crises (drought). In this harsh environment, farmers have been able to maintain most of their pearl millet varieties. This apparent stability does not preclude genetic evolution. Exploiting intravarietal diversity, farmers have been able to shorten the maturing time of some of their varieties leading to a better adaptation to a drier climate. They have also recreated their varieties exploiting intervarietal diversity with introductions from adjacent areas, and by favouring geneow and selection. The main characteristic of this management is the existence of a large amount of genetic diversity and extensive geneow in this crop. If one of these elements, i.e. diversity or geneow, is reduced, the system will be affected. Modications of this system can arise from farmers changing their main crop from pearl millet to another crop due to the many cultural and economic changes that occur in this region. Up to now, farmers have demonstrated their ability to overcome crises, managing their varieties as dynamic entities, a concept that could be used when we think of strategies to broaden the genetic basis of some of our crops.

Cassava in Amazon (Rio Negro, Brazil) This case study has been described in detail by Emperaire et al. (1998). A general discussion of the use of diversity in this crop is given by Second and Iglesias in this volume, Chapter 11.

92

J. Berthaud et al.

Study site location In Brazil, Supiranga is a village located on the Rio Negro, between Barcelos and Santa Isabel, where nine families live. These families are of very different ethnic origin (Tukano, Desana, Pira-Tapuia, Tariano, Bar). Slash-and-burn cultivation of cassava is the basis of the production system in this village. Cassava our is very strategic in home economics, local exchanges and regional markets, and cassava plots are located in a radius of 23 km around the farmers homes. Every year, men clear a plot of 0.51 ha in the more than 10-year-old regrown forest. Mostly women then take care of the plot and the harvest. A plot is kept under cultivation for 23 years, i.e. two cassava cultivation cycles, and then enriched with fruit trees before being abandoned for the forest to re-grow. Farmer selection criteria of varieties (Table 4.3) In this study, a variety is dened as a group of plants which are given the same name by farmers. As cassava is a vegetatively propagated crop, it can be assumed that most of the plants with the same name are of unique genetic origin.
USAGE SELECTION CRITERIA

Today, in Supiringa, cassava is mostly eaten or sold as our (farinha), as pancake made from the mixture of grated meal and starch (beijus) and sometimes of pure starch (curads). Colour is a very important attribute of the food, and especially of the our. It has a specic value but is also considered as an integrated meta-trait to describe other characteristics, related to nutritional and culinary values, such as starch, water and bre contents.

AGRICULTURAL SELECTION CRITERIA

Selecting varieties is also a part of the production strategy that promotes food security. Combining varieties with various maturing and

Table 4.3. Selection criteria used by farmers from Supiranga. Use criteria Colour = integrated criterion for : Starch content Fibre content Water content Yield Production Colour of processed products Ease of peeling Ease of uprooting Yield criteria Yield = integrated criterion for : Yield potential Broad adaptation Resistance to pests and diseases Resistance to splitting Resistance to lodging Yield stability Propagation rate Branching Height Maturity time Storage time Drought tolerance of cuttings Non-use criteria Symbolic value Emotional value Filiation value Novelty value Purple-red colour Unusual shape

Local-level Geneow and Genetic Diversity

93

storage times helps to ensure a stable production over the year in the cultivated plots (roas). Early maturing varieties (67 months) are systematically planted in newly established plots. Tuber storage underground varies from 10 to 36 months and offers farmers the opportunity of abandoning the plot for a few months and carrying out other activities (e.g. forest product harvesting and staying in town). Factors related to cultivation, either to the genetic material (yield capacity, pest and disease resistance, ecological adaptation, lodging resistance) or to knowledge of the agricultural techniques (evaluation of the state of a plot, cutting, maintenance) are rarely identied by farmers but are globally assessed through yield. Boster (1984) had already perceived the same behaviour in relation to the management of the cassava variety diversity by the Aguaruna in Brazil.
AESTHETIC SELECTION CRITERIA

Farmers combine non-use value or aesthetic selection criteria with the previous criteria. A variety can be appreciated for its origin: e.g. it was cultivated by a mother or a grandmother, and becomes an element of family continuity. A collection of varieties is also the backdrop to a womans working life: a cassava eld with some purple-red plants or with varieties of an unusual shape can give proud, aesthetic pleasure and entertainment as can a well cleaned plot or healthy plants. These aesthetic aspects of establishing a variety collection are real factors involved in the selection process.

Local management of variety collections The number of varieties cultivated per family is relatively homogeneous, and varies between 28 and 40. From a total of 61 varieties found in the area, only ve (8%) are cultivated by all the ve families studied, 13 are cultivated by four of them, nine by three families, 20 by two families, but 14 by only one family. These gures show the specicity of each family variety portfolio and the individual management level of these portfolios. Different families prefer different selection criteria. The very rare or very frequent varieties are found in the three main categories of bitter cassava (mandiocas): white, semi-yellow and yellow. However, a comparison based on plant frequency within varieties leads to a general scheme for the distribution of diversity in these portfolios:

Half the plants belong to two to three yellow or semi-yellow varieties. White varieties represent only 215% of the plants.

Within this framework, each farming woman manages her own variety portfolio according to her own preferences and cutting availability. Selection criteria have been modied in search of a better adaptation to a specic market based on non-timber forest products in the Brazilian Amazon. This change in variety diversity occurred during one human generation and had already been noted by Dufour (1993) for the Tukano of the Colombian Amazon. Vegetative propagation Cassava varieties are distributed in the eld in small spots, which allow, during the two or three production cycles, a serial harvesting from the centre to the border. It is a way of preserving the eld (roa) centre where some medicinal and magic species are cultivated away from prying eyes and spiteful actions. Every 15 days, some tens of square

94

J. Berthaud et al.

metres containing a limited number of varieties are harvested (Dufour, 1993 noted an average of 4.3 varieties per harvest). This part of the eld is almost immediately replanted with new cuttings taken from cassava stem bundles of the harvested varieties (feixes). By this technique, the spatial distribution is maintained unchanged throughout the cultivation period. Ecological conditions required for sexual seed production, dispersal and germination are not very well dened. It is possible that not all varieties are equally tted for seed production, in which case the genetic broadening of varieties would rely only on a few of them. Our own data and information gathered from farmers indicate that the frequency of plants grown from sexually recombined seed is one or two per new roa, i.e. every 5 years: a low but signicant frequency. This sexually induced broadening of diversity has already been described by other authors (Chernela, 1986; McKey and Beckerman, 1993) in the context of low pressure for land use, as in Supiranga. A possible link between modes of land management and production of new varieties through sexual seeds should be evaluated. The social origin of the cuttings is very different from one family to another. Some individuals received varieties from their mother or mother-in-law during their wedding ceremony: these varieties give the couple the capacity to be self-sufcient and to become integrated into the community when they start their rst eld. The cuttings of the initial variety portfolio are an important element of the process of social reproduction, being a legacy from the women and recognized as a patrimony by men, representative of the past and valuable for the future (Chastel, 1986). These cuttings are also a form of identity assertion. Those who make the choice of favouring agriculture take great care of this genetic patrimony and try to broaden it by practices that relate to the social as well as the agricultural environment. Genetic diversity The genetic diversity of this material was studied using DNA markers to test the signicance of the high diversity index existing among the varieties when estimated by the many attributed names. Several questions had to be answered:

What is the degree of diversity of cassava from Supiranga compared with a representative cassava world collection? Are two plants with the same name genetically distinct? Cassava is vegetatively propagated and one expects to nd identical plants under one name. How frequently can exceptions be found? When two plants with the same name are found to be genetically different, how different are they? And how does this difference compare with the genetic differences between different varieties?

Universal markers (restriction fragment length polymorphism (RAPD) and amplied fragment length polymorphism (AFLP)TM) were used. We analysed the diversity among 42 plants representing 18 names, including three plants sem nome (with no name), presumably directly grown from germination, all planted in one eld of this village. The plants were also compared with 40 accessions from a world collection at CIAT. Two series of analyses were conducted using the same DNA extracts and AFLP (Second et al., 1997) or RAPD (Colombo, 1997) markers. The main results of these studies reviewed in Emperaire et al. (1998) show that the genetic diversity found in this

Local-level Geneow and Genetic Diversity

95

cassava eld compared well with the tested world-collection diversity but that there was no overlap. This shows that the diversity maintained in the Amazon is very large and that some specic alleles and gene combinations are maintained in this area. Through these analyses, 25 genotypes were distinguished between the 42 plants of the single Amazonian eld. This shows that many of the plants with the same name are genetically identical, that all names represent different genotypes, and that the plants with no name (sem nome) were all distinct. These three sem nome plants are genetically related and their sexual origin is conrmed. Genotypes with the same name appeared as genetically related as compared with total diversity. These preliminary results show that farmers recognized the interest and utilized the process of sexual recombination, which is favoured by the grouping of several varieties in a eld. This management of variety diversity was something not expected for a vegetatively propagated crop. A traditional variety can be considered a family of clones sharing: (i) some traits of direct interest for the farmers: quality, productivity, resistance to pests, diseases and other stress; and (ii) some morphological traits useful for their identication. However, the full description of a variety should incorporate the culinary and post-harvest quality. These ndings are being conrmed by work currently in progress. Main points from this cassava case study A vegetatively propagated crop such as cassava can go through steps of sexual reproduction under farmer management. Both natural selection and farmer selection are applied after genetic recombination, i.e. after new associations of genes are produced. The impact of farmers on the evolution of varieties is quite strong. Modication of food habits and economic exchanges bring changes to the content of the individual cassava variety portfolios. The evolution of this traditional agriculture is still active and translates into an evolution of the plant material at all levels including the individual farmer level. During our current study, we observed that the unit for the management of diversity is the variety portfolio and not the variety. This shows that farmers are aware of the diversity existing in their plant material and that, to t their needs, they require various varieties, each sharing some interesting traits. Social and exchange networks are basic factors for the conservation and management of cassava in Brazil. Through these networks, varieties are maintained even when catastrophic events occur. These networks are at the root of the establishment of the variety portfolios. Farmers select their activities according to complex motivations, based on food, social and economic interests. All these motivations lead to the maintenance of a large choice of varieties and to a broad cassava variety portfolio. They have a structuring effect on biological diversity of this crop at the village level we studied, but this effect should also be found at a higher level. These farmers surviving in a traditional system of agriculture and using a vegetatively reproduced crop could in theory have been involved in a perfectly genetically frozen system. The study conducted showed that, on the contrary, there is a strong dynamism promoted by the social system and, as a consequence, a strong dynamic equilibrium for the genetic diversity involved in the production system.

96

J. Berthaud et al.

Geneow between varieties and wild relatives Situations of maizeteosinte introgression BALSA AREA Most of the F1 hybrids between maize and teosinte are fully fertile. In the Balsa region, teosinte is a wild plant that can be found along roads and outside maize elds. Wilkes (1967) recognized both spatial and temporal isolation of teosinte and maize, and noted that teosinte owers 23 weeks later than maize. However, when teosinte is found beside maize elds, F1 hybrids are detected. In their collections of Balsa teosinte, Sanchez and Ordaz (1987) reported that ten out of 55 populations showed F1 hybrids. So, in this area, there are situations where teosinte and maize are not as isolated as is currently reported.
VALLEY OF MEXICO In the valley of Mexico, teosinte belongs to the race Chalco, a representative of Zea mays ssp. mexicana. In this area, teosinte has all the traits of a weed. It is almost only found within and along the cultivated maize elds. In some elds, teosinte accounts for 10% of the plants (Wilkes, 1967) or more (J. Berthaud, personal observation). There is no spatial isolation, and temporal isolation is only partially operative (Wilkes, 1967, p. 73). Flowering of teosinte is about 2 weeks later than maize, and Wilkes (1967) noted a proportion of F1 maize teosinte hybrids of 25% within teosinte and teosinte-like plants.

Teosinte found in this area (north of state of Michoacan and south of state of Guanajuato) belongs to the same subspecies as in the Chalco area: Zea mays ssp. mexicana. This subspecies also includes the populations discovered around the town of Durango. In this region, populations of teosinte are found wild or as weeds, and plants grow both in maize elds and in places where there is no maize (Wilkes, 1967). There is no spatial isolation, but there is temporal isolation, with teosinte owering 2 weeks after maize. Wilkes (1967, pp. 8081) mentioned that in some small populations teosinte plants exhibited non-brittle rachis and paired spikelets, which are maize traits. This suggests that introgression from maize occurred in these small populations and that teosinte populations can survive with some domestication traits. Another case of introgression is described by Doebley (1990) when he found that the chloroplast DNA in a CMS-S cytoplasmic male-sterile maize was only shared by three populations of Central Plateau teosinte, from Copandaro. This CMS-S maize also shared the same mitochondrial type as the three teosinte populations: further evidence that introgression has occurred. In this case, introgression would have started on a teosinte plant, and successive backcrosses from maize pollen would have resulted in a plant with maize nuclear genes in a teosinte cytoplasm.
CENTRAL PLATEAU ZEA DIPLOPERENNIS, A WILD ZEA SPECIES CONSERVED IN SITU

The perennial and diploid teosinte, Zea diploperennis, was discovered by R. Guzman in 1977 and described by Iltis et al. (1979). It is only found in the Sierra de Manantln (state of Jalisco, Mexico). Four populations are reported, at an altitude extending from 1350 to 2440 m above sea level. Populations from Manantln and Rincon de Manantln are small, with a recent origin, while populations from La Joya and Valley of San Miguel are distributed over 40 and 320 ha respectively.

Local-level Geneow and Genetic Diversity

97

The ecological requirements of Zea diploperennis make it quite dependent on human activity. Its populations grow on former maize elds or on intermittently grazed areas. Manantln and Rincon de Manantln populations were most probably established by farmers for pasture. In San Miguel, the teosinte population is actively managed: farmers sow seeds where density of teosinte is low. A genetic use of this population is also reported by Benz et al. (1990). A controlled introgression of maize by Zea diploperennis is promoted by farmers planting some teosinte along maize elds. They plant the F1 hybrids the following year and the backcrosses the next years. The 4th year, these seeds are mixed with those of the local varieties. Farmers say that it is a way to improve maize, especially for disease resistance. However, actual introgression is yet to be conrmed. A biosphere reserve was created which protects this Zea species and conserves it in situ. At the beginning, this implied preserving the wild populations in their original localities. Later observations have shown that survival of these populations is linked to the activity of farmers who keep their land open through cultivation. When land is left fallow, forest regenerates after less than 30 years, trees and shrubs compete with teosinte for light and the teosinte is eliminated. This example of in situ conservation underlines the strong link existing between a wild species and human activities. In this case, when geneow exists between wild and cultivated forms of the species, it is largely due to the activity of farmers, who are the key element of the system. Genetic differentiation in maize and teosinte populations Studies already carried out by Smith et al. (1984, 1985) and Doebley et al. (1984, 1987) based on isozyme diversity have shown:

Plants from section Luxuriantes Z. diploperennis, Z. perennis, Z. luxurians form a group different from the Mexican annual teosintes. The race Balsa, Z. mays ssp. parviglumis, has a wider range of variation. Accessions from Central Guerrero are close to accessions from Central Plateau, taxonomically belonging to Z. mays ssp. mexicana. A closer similarity between Balsa and maize could be due in part to the existence of more alleles (in low frequencies) in this race. As a consequence, Balsa would have more alleles in common with maize.

From these studies, several genetic parameters have been estimated (Doebley et al., 1984). Diversity in populations of teosinte is quite comparable with diversity of Mexican maize accessions (Eyre-Walker et al., 1998), and a lot of variation is retained within populations. Does introgression occur between maize and teosinte ? Doebley (1990) has already discussed this issue. The most convincing example given is from Z. diploperennis, where he found a plant with alleles from maize at two linked loci. Frequency of these alleles was 0.01, which can be considered as high. Kato (1984) did not nd knob distribution in maize and teosinte that would have suggested introgression. For example, abnormal chromosome 10 type 2 is only found in teosinte and not in sympatric maize. In conclusion, presence in the elds of F1 hybrids between maize and teosinte has been well documented (Wilkes, 1967; Sanchez and Ordaz, 1987) but there is still a

98

J. Berthaud et al.

need to set up experiments to follow the evolution of F1 hybrids when they are backcrossed by maize or by teosinte, and estimate the size of gene exchange between these forms of Zea.

Geneow, Farmers and the Base-broadening Process

Comparison of dynamic (DM) and on-farm management (OFM) of genetic resources The dynamic management of genetic resources (DM: Goldringer et al., Chapter 13, this volume) consists of distributing the genepool to be conserved in a number of small populations, representing a metapopulation. In so doing, processes of genetic drift and selection are favoured. This system promotes the conservation of alleles at the population level but not at that of genotypes or individuals and it takes advantage of the effects of genetic drift and natural selection to enrich the small populations with locally advantageous genes and gene combinations. Dynamic management can also be seen as a base-broadening or pre-breeding technique. When we compare this DM with what we described as on-farm management (OFM), we see that in OFM the selection process depends on natural selection and on selection exerted by farmers (mainly directed to traits useful to the farmers). OFM is conducted on many farms and the total population size is much greater than that of the more controlled condition of DM, and can explore many more niches. However, the two types of management are similar because both are more directed at the conservation of alleles than the conservation of genotypes. Evaluating the bread-wheat dynamic management in France over a 13-year period, Goldringer et al. (Chapter 13, this volume) draw the conclusion that morphological characters evolved differently from site to site during this period, and that evolution also occurred for the relationship between plants and pathogens. A comparable evolution is expected within OFM of genetic resources: direction of selection is always changing and genetic material evolving due to geneow and recombination. Equilibrium is seldom reached but this dynamic maintains genetic diversity. In OFM, some varieties are used in different conditions in different sites: for instance, the Blanco variety of maize in Cuzalapa or the backyard varieties in Burkina Faso and cassava in Amazonia. These varieties experience varied selection pressures and even selection cycles, for example from rain-fed cultivation to irrigated cultivation and back. A broader adaptation can be expected from these varieties compared with genetic material produced through dynamic management alone. OFM goes beyond dynamic management as it takes into account management at the level of a collection (variety set or portfolio) which can enable more diversity to be conserved than DM and can also enable a complementary type of diversity management to be carried out. OFM is an open system that allows complementary introductions of varieties and a rapid evolution and genetic adaptation when needed. However, the system is susceptible to replacement of varieties by new introductions, i.e. new varieties or new crops that totally replace the landraces. This is one way farmers deal with changing conditions in production systems and it can lead to the disappearance of some local varieties when they are not cultivated in many different places. It is the strength and weakness of this system that it responds to economic and social forces.

Local-level Geneow and Genetic Diversity

99

The broadening of the genetic base of some crops could be obtained by introgression from wild ancestors or related weeds. The management of this type of introgression requires almost necessarily a monitoring of geneow. It will be more easily established and run in a dynamic management system than in the OFM system where it could be difcult for farmers to assess the introgression process. However, this may not always be the case, and situations have been described where geneow between wild relatives and crops are under the control of farmers (see page 97).

Promoting geneow through farmer participation In the cases studied, several types of genetic exchange have been observed which are generated by the activity of farmers on their farms. The three levels of exchange described are as follows: 1. Exchange, introduction and loss of varieties in farmers collections leading to a change in the number of varieties available to the farmers in a specic place. 2. Genetic exchange between landraces and modern varieties, and between wild and cultivated plants. 3. Genetic exchange within varieties as the varieties are usually seed reproducing and are not genetically xed. The base-broadening activities could be modelled according to the situation described on-farm. Exchange between collections of varieties The rst level of farmer management of genetic resources is that of the collection of varieties maintained. The diversity within these collections (the portfolio) is a farmers answer to agricultural uncertainty and to varying needs, according to productivity and usage criteria. When dening strategies for base-broadening, we should take into consideration that the new products (varieties) will be used in association with others and managed as portfolios of varieties. These new products will have to complement the diversity already existing in these collections, which implies a very delicate combination of new traits and adaptation to different requirements. Base-broadening could also play a role in restoring the farmers collections where they have disappeared for such reasons as war or drought. Exchange between varieties Exchanges occur between different types of material, i.e. between wild and cultivated plants, between modern and traditional varieties, or between traditional landraces. This intervarietal diversity and close proximity of varieties in the farmers elds favour gene exchange between varieties. The varieties generally have very distinct morphological and genetic traits and farmers maintain through selection the most important characteristics that identify these varieties and are the basis for their agricultural interest. Gene exchange can also allow farmers to select for different characteristics, such as a shortening of the time to maturity of a crop. It is quite important to recognize the existence of such a management system, which is certainly a process promoting the genetic diversity found in landraces.

100

J. Berthaud et al.

Base-broadening can use this gradual introgression as a model. It can also take advantage of this existing OFM to propose new progenitors to farmers for incorporating new genes into their own varieties. This should be a way for farmers to obtain new (modied) landraces which respond better to their changing needs. Exchange within varieties Farmers recognize their varieties by certain agronomic and morphological traits and apply selection pressure for these traits. Varieties are more or less stable for these traits but not for others. Each variety is far from being genetically homogeneous and this makes them a potential reservoir for genetic progress. Base-broadening actions should consider diversity not only at the intervariety level but also at the intravariety level when it comes to decide on base-broadening strategy. This is also true for crops with vegetative propagation such as yam and cassava: their varieties are often polyclonal, and genetic recombination is a mechanism exploited by the farmers. This reservoir of diversity within varieties also offers a possibility for breeding progress or for developing broad adaptation. We have seen that the variety Blanco is cultivated in Cuzalapa under irrigation but also in other villages under rainfed conditions. Seeds of this variety are exchanged between these villages. This management confers a broad adaptation to the variety and, at the same time, favours the maintenance of a within-variety genetic diversity.

General Conclusion
Peasant farming systems have been considered as crop evolutionary laboratories (Brush and Meng, 1998). Our case studies have shown three levels of complexity in these laboratories between collections of varieties, between varieties and within varieties with geneow occurring at these three levels. Maintenance of these varieties or landraces is not based only on interest for maximum yield but also on their private value and their comparative advantage due to relatively small advantages (Brush and Meng, 1998). In their concept of variety, farmers integrate a dynamic component; we are very often far away from the concept of a commercial variety with its constraints of distinctness, homogeneity and stability. The concept of a dynamic variety is well in agreement with biological models of metapopulations, which consider sets of populations with history of extinction and migrations (Hanski, 1998). An evolution of farmers varieties towards more stability and homogeneity, i.e. closer to the concept of commercial variety, would introduce an erosion of the geneow and lead to a loss in our capability to maintain genetic diversity and a broad genetic base for our crops. The dynamic of the biological structure of genetic material, farmers varieties and landraces is strongly dependent on the social structure organizing the life of the farmers: we can consider that the former is a reection of the latter. Selection of seeds as a continuous and iterative process is emphasized by Rice et al. (1998) for maize. We could dene plant varieties in these farming systems as continuous cultural constructions which integrate constraints from yield, use and aesthetic criteria, but these activities come at a biological and social cost. More detailed studies should be conducted and theoretical approaches completed to evaluate and compare different strategies, taking into account genetic load and other biological constraints.

Local-level Geneow and Genetic Diversity

101

How can we take advantage of this system to conduct base-broadening of our crops? We saw that varieties managed by farmers are open genetic systems. Farmers are very active in experimenting with new genetic material and they have a strong interest in innovation. This is a very favourable situation for experimenting with base-broadening involving farmers in this process. The dynamic of these systems can be used as models when we dene base-broadening strategies. As a result, it should produce varieties and populations of progenitors dynamically maintained. When we want farmers to be part of the process, we have to look beyond biological criteria and take into account the social and farming systems in and/or for which the base-broadening will be carried out. Incentives to promote this base-broadening will have to be compatible with the main component of these integrated systems. As a very general conclusion for our study, it can be said that OFM of seeds and varieties is a complex system with many different options chosen by farmers. When we propose systems of improvement and conservation of plant varieties we should be careful that they are not too simplistic and not very far away from what is being done by these traditional farmers. Continuing to document this complex system will certainly reveal even more sophisticated relationships between people and their crops, challenging our strategies for broadening the genetic base of our crop and the conservation of their genetic resources.

References
Andrews, D.J. and Kumar, K.A. (1996) Use of the West African pearl millet landrace Iniadi in cultivar development. Plant Genetic Resources Newsletter 105, 1522. Anonymous (1997) Gestion des ressources gntiques des plantes en Afrique des Savanes. BamakoMali: 2428 fvrier 1997. IER; BRG; Solagral, Paris. Benz, B.F., Sanchez-Velasquez, L.R. and Santana-Michel, F.J. (1990) Ecology and ethnobotany of Zea diploperennis: preliminary investigations. Maydica 35, 8598. Bilquez, A.F. (1963) Etude du mode dhrdit de la prcocit chez le mil pnicillaire (Pennisetum typhodes Stapf et Hubbard). I. Dterminisme gntique des diffrences de sensibilit la longueur du jour existant entre les mils du groupe Sanio et ceux du groupe Souna. AgronomieTropicale 24, 12491253. Bilquez, A.F. and Clment, J.C. (1963) Etude du mode dhrdit de la prcocit chez le mil pnicillaire (Pennisetum typhodes Stapf et Hubbard). II. Dterminisme gntique des variations de prcocit des mils du groupe Souna. Agronomie Tropicale 24, 258261. Boster, J.S. (1984) Classication, cultivation and selection of Aguaruna cultivars of Manihot esculenta (Euphorbiaceae). Advances in Economic Botany 1, 3447. Brush, S.B. and Meng, E. (1998) Farmers valuation and conservation of crop genetic resources. Genetic Resources and Crop Evolution 45, 139150. Chastel, A. (1986) La notion de patrimoine. In: Nora, P. (ed.) Les lieux de la mmoire (vol. 2). Gallimard, Paris, pp. 405450. Chastenet, M. (1998) Introduction et place du mas au Sahel occidental (Sngal-Mauritanie). In: Chastenet, M. (ed.) Plantes et paysages dAfrique, une histoire explorer. Karthala-CRA, Paris, pp. 251282. Chernela, J.M. (1986) Os cultivares de mandioca naara do Uapes (Tukano). In: Ribeiro, B.G. (ed.) Suma tenologica brasileira. Vol. 1: Etnobiologia. FINEP, Petropolis, pp. 151158.

102

J. Berthaud et al. Clment, J.C. (1985) Les mils pnicillaires de lAfrique de lOuest, International Board for Plant Genetic Resources, Rome. Clment, J.C. (1997) Les varits traditionnelles de mil en Afrique sahlienne. Facteurs de stabilit et de variation. Anonymous Actes du colloque gestion des ressources gntiques des plantes en Afrique des savanes, Bamako, Mali, 2428 fvrier 1997. IER-BRG-Solagral, pp. 133142. Colombo, C. (1997) Etude de la diversit gntique de maniocs amricains (Manihot esculenta Crantz) par les marqueurs molculaires (RAPD et AFLP). PhD thesis, Ensa-Montpellier, France. Doebley, J. (1990) Molecular evidence and the evolution of maize. Economic Botany 44, suppl. 627. Doebley, J., Goodman, M.M. and Stuber, C.W. (1984) Isoenzymatic variation in Zea (Gramineae). Systematic Botany 9, 203218. Doebley, J., Goodman, M.M. and Stuber, C.W. (1987) Patterns of isozyme variation between maize and Mexican annual teosinte. Economic Botany 41, 234246. Dufour, D.L. (1993) The bitter is sweet: a case study of bitter cassava (Manihot esculenta) use in Amazonia. In: Hladik, C.M., Haladik, A., Linares, O.F., Pagezy, H., Semple, A. and Hadley, M. (eds) Tropical Forests, People and Food. Unesco, Paris, pp. 575588. Emperaire, L., Pinton, F. and Second, G. (1998) Gestion dynamique de la diversit varitale du manioc en Amazonie du Nord-Ouest. Nature, Science, Socit 6, 2742. Eyre-Walker, A., Gaut, R.L., Hilton, H., Feldman, D.L. and Gaut, B.S. (1998) Investigation of the bottleneck leading to the domestication of maize. Proceedings of the National Academy of Sciences USA 95, 44414446. Hanski, I. (1998) Metapopulation dynamics. Nature 396, 4149. Iltis, H.H., Doebley, J.F., Guzman, R. and Pazy, B. (1979) Zea diploperennis (Gramineae): a new teosinte from Mexico. Science 203, 186188. Kato, Y.T.A. (1984) Chromosome morphology and the origin of maize and its races. Evolutionary Biology 17. Laitner, K. and Benz, B.F. (1994) Las condiciones culturales y ambientales en la reserva de la biosfera Sierra de Manantln en tiempo de la conquista: una perspectiva de los documentos etnohistoricos secundarios. Estudios del Hombre 1, 1545. Louette, D. (1994) Gestion traditionnelle de varits de mas dans la rserve de la biosphre Sierra de Manantln (RBSM, tats de Jalisco et Colima, Mexique) et conservation in situ des ressources gntiques des plantes cultives. PhD thesis, Ensa-Montpellier, France. Martinez, R. and Sandoval, J.J. (1993) Levantamiento taxonomico de suelos de la subcuenca de Cuzalapa, Sierra de Manantln, Jalisco. Terra 11, 311. McKey, D. and Beckerman, S. (1993) Chemical ecology, plant evolution and traditional manioc cultivation systems. In: Hladik, C.M., Haladik, A., Linares, O.F., Pagezy, H., Semple, A. and Hadley, M. (eds) Tropical Forests, People and Food. Unesco, Paris, pp. 83112. Paterniani, E. and Stort, A.C. (1974) Effective maize pollen dispersal in the eld. Euphytica 23, 129134. Rice, E., Smale, M. and Blanco, J.L. (1998) Farmers use of improved seed selection practices in Mexican maize: evidence and issues from the Sierra de Santa Marta. World Development 26, 16251640. Rosales, A.J.J. and Graf, M.S.H. (1995) Diagnostico sociodemograco de la reserva de la biosfera Sierra de Manantln y su region de inuencia. IMECIO, Universidad de Guadalajara, El Grullo. Sanchez, G.J.J. and Ordaz, S.L. (1987) El teocintle en Mexico. Distribucion y situacion actual de las poblaciones. (2): Systematics and ecogeographical studies of genepools. International Board for Plant Genetic Resources, Rome. Sanou, J. (1996) Analyse de la variabilit gntique des cultivars locaux de mas de la zone de

Local-level Geneow and Genetic Diversity

103

savane ouest-africaine en vue de sa gestion et de son utilisation. PhD thesis, EnsaMontpellier, France. Sanou, J., Gouesnard, B. and Charrier, A. (1997) Isozyme variability in West African maize cultivars (Zea mays L.). Maydica 42, 111. Second, G., Costa Allem, A., Emperaire, L., Ingram, C., Colombo, C., Mendes, R.A. and Carvalho, L.J.C.B. (1997) AFLP based Manihot and cassava numerical taxonomy and genetic structure analysis in progress. Implications for dynamic conservation and genetic mapping. African Journal of Root and Tuber Crops 2, 1017. Sdogo, M.C. and Tostain, S. (1996) Diversit enzymatique des mils cultivs (Pennisetum glaucum (L.) R.Br.) du Burkina Faso. Plant Genetic Resources Newsletter 105, 814. Smith, J.S.C., Goodman, M.M. and Stuber, C.W. (1984) Variation within teosinte. III. Numerical analysis of allozyme data. Economic Botany 38, 97113. Smith, J.S.C., Goodman, M.M. and Stuber, C.W. (1985) Relationships between maize and teosinte of Mexico and Guatemala: numerical analysis of allozyme data. Economic Botany 39, 1224. Wilkes, G.H. (1967) Teosinte: the closest relative of maize. PhD thesis, Bussey Institute, Harvard University, Cambridge, Massachusetts.

Regulatory Aspects of Breeding for Diversity

N.P. LOUWAARS
Plant Research International, PO Box 16, 6700 AA Wageningen, The Netherlands

Introduction
Seed regulatory frameworks aim to promote varietal and seed quality, and thereby to protect farmers from planting sub-standard seed. Seed laws commonly regulate variety testing and release, seed certication and seed quality control, and they establish the institutional framework of national seed councils and certication agencies. Variety release systems aim at making only varieties of proven value available to farmers. Seed certication aims at controlling the varietal identity and purity throughout the seed chain. Seed quality control checks on seed quality such as viability, purity and health. Seed quality control also protects bona de seed producers from competition by less scrupulous colleagues. Additionally, intellectual property rights on varieties are included in seed laws of some countries in order to protect the breeder from competitors that may provide seed without any R&D investment through the granting of an exclusive right over the exploitation of the variety. The regulatory frameworks that have been developed in various countries reect different levels of state involvement. For example, in North America certication is, in many cases, a voluntary service, and variety release is fully the responsibility of the company. This reects the general condence in the regulatory effects of the free market itself: suppliers of poor quality seed would be punished by the customers through declining demand for their products. In various European countries, on the other hand, public institutions have developed a signicant mandate for policing seed quality. In some countries such as The Netherlands, certication agencies have developed as independent foundations managed by farmers, seed producers and breeders, but they operate within a legal framework. In most developing countries, formal seed production has developed as part of a strategy based on the Green Revolution paradigm of agricultural development in which plant breeding would increase yield potential under optimum growing conditions, and seed production would therefore be a necessary vehicle for technology transfer. Seed and other inputs have often been heavily subsidized in order to increase adoption of the
IPGRI/FAO 2001. Broadening the Genetic Base of Crop Production (eds H.D. Cooper, C. Spillane and T. Hodgkin)

105

106

N.P. Louwaars

new technologies. Within this paradigm, centralized seed production units have been built to resemble the successful European and North American seed industries. These formal seed systems subsequently developed specialized seed quality control institutions to create a quality awareness and to safeguard the interests of farmers, similar to the ofcial seed certication agencies in the north. Especially in the era of privatization of public institutions at the end of the 1980s, these seed quality control institutions became the driving force behind the development of seed legislation (Louwaars and van Marrewijk, 1996). Such legislation would give these institutions a legal backing necessary to perform its police tasks vis--vis (private) seed producers. The result of this development is that many seed laws in the south strongly resemble those in the north. However, whereas in the north the farmers interest was often represented by a strong voice in the seed quality control systems, in several countries in the south seed regulations t into the existing bureaucratic structures and are imposed upon both seed producers and users. Seed laws are not usually intended to inuence the direction of plant breeding. However, there are signicant indirect effects of the variety release systems and of seed certication requirements on plant breeding methodologies and the resulting varieties. Breeders tend to target favourable farming conditions, wide adaptation and varietal uniformity as a result. This chapter analyses the effects of seed regulatory frameworks in relation to the objectives of broadening the genetic base of plant breeding, with particular focus on the role of alternative breeding strategies.

Effects of the Various Components of the Regulatory Framework on Breeding

Variety registration and testing Variety release may become a goal in itself when regulatory systems are too rigid. Release is the yardstick by which the effectiveness of public plant breeding is measured. The reward system for breeders is commonly based on the number of varieties released, not on their widespread use by farmers. Hence, the objectives of plant breeders are likely to be adapted to the variety release procedure instead of the farmers needs. A variety release system commonly incorporates the following steps: 1. Application with a formal variety release committee and variety registration, including a variety description. 2. Testing for the value for cultivation and use (VCU) of the variety involving a prescribed number of sites and seasons. 3. Testing for distinctiveness, uniformity and stability (DUS). 4. Analysis of the test results by the committee leading to approval or rejection for formal release. In each of these stages there can be a bias favouring particular types of varieties. Application for variety release commonly includes payment of a fee. The global trend of reducing public expenditures has raised the nancial contribution by the breeder to the testing system in most countries. The result is that both public and

Regulatory Aspects of Breeding for Diversity

107

private breeders reduce the number of varieties submitted for ofcial release to those that are likely to perform well in all test locations and to pass other registration standards such as the DSU requirement. Varieties with specic adaptation to particular agroecological niches or uses are less likely to be presented. The management of many variety testing systems further reduces the number of approved varieties and delays release. High input levels are often used to improve the trial from a statistical point of view. Sometimes, this is also a deliberate policy to represent the conditions of the better farmers and to motivate other farmers to follow; also, high input levels give beautiful crops that make a trial presentable to visitors. But high inputs such as fertilizers conceal environmental variations in the trial, thus reducing residual variance. Poor trial design and management increases residual variances and thus reduces the chance that actual differences in quantitative traits are identied. This may require additional trials to be performed, which delays release, or it may obstruct release altogether. Moreover, poor site selection and high input levels make it doubtful whether the trial results have any value for most farmers. For example, it is unlikely that the ofcial sorghum trial results in India are valuable for the majority of farmers where average yields in the 1989/90 trials were three times the farmers average yields (Virk et al., 1996). Breeding for favourable agroecological conditions and for monoculture is encouraged. The evaluation of trials using simple statistical analysis methods leads to a bias in favour of breeding approaches for wide adaptation. Since trials are pooled in one calculation, the variety having the highest average yield is considered the best. However, this may not be the best variety in any of the testing sites. Standard variety release procedures rarely accept a variety that is specically adapted to particular conditions, even though national variety lists may contain regional recommendations. The trial system is biased against breeding for partial (horizontal) resistance, which is in most cases polygenic and more durable. Such varieties are resistant, but not immune to disease and thus they commonly carry disease symptoms, and for this reason are liable to be rejected in a release system, even if uniform. Additionally, the small size of the research plots makes it difcult to identify horizontal resistance (Ceccarelli, 1989). Evaluation of variety trials by release committees is usually totally dependent on gures, with the result that only yield becomes the decisive characteristic, and important characteristics for smaller-scale farmers may not be taken into account. These include, for example, aptitude to intercropping, shattering (e.g. soybean), lodging when harvesting is delayed (e.g. maize), cooking time of the produce (e.g. beans), and the yield and quality of secondary products (e.g. straw for construction or fodder). Breeding thus tends to concentrate on yield only, without giving credit to the diverse needs of farmers. Variety release committees commonly consider the appropriateness for the production of certied seed as an important criterion. A variety needs to be morphologically identiable, and thus distinct from existing varieties, and stable. Both factors contribute to the need for a certain level of genetic uniformity. The uniformity standards of seed certication systems are commonly very high (allowing only one dozen or a few dozens off-type plants per hectare); much higher even than International Union for the Protection of New Varieties of Plants (UPOV) standards for plant variety protection (see page 109). Releasing varieties to a seed certication system thus implies breeding for uniformity, even where this has no agronomic advantage.

108

N.P. Louwaars

Finally, lack of participation and transparency in the closed system of formal variety release leads to conservative trial designs and management. Parallel (demonstration) trials by the extension service, non-governmental organizations (NGOs) or private seed companies are rarely taken into account in the release decision. Ofcial on-farm variety trials are becoming increasingly popular with variety release systems. This positive development, however, hardly ever contributes to releasing more adapted varieties because such on-farm trials are either completely researcher managed, and thus similar to station trials, or the results cannot be easily analysed statistically, leading to denial of such results: the non-numerical comments by farmers are difcult to include in reports. In developing countries, farmers are rarely well represented in the variety release committee, or in the evaluation of varieties. Improved representation would be likely to lead to a higher transparency of the system. In conclusion, standard variety release procedures commonly result in the approval of few, uniform and widely adapted varieties that often do not respond to the diverse needs of farmers.

Seed certication and quality control Seed certication, being the check on varietal identity and purity, has a marked effect on breeding strategies. Varieties have to be stable in order to guarantee varietal identity; only uniform varieties can supply the required level of stability. Further effects on breeding largely depend on the denitions and interpretation of the seed law. In some countries all seed planted falls within the scope of the law. Examples can be found in the former Soviet Union and various countries in Africa and Asia. In other countries all seed in the market fall within the denition of the law, thus basically making any exchange of seed of non-tested seed and non-released variety seed strictly illegal. More recently some more open seed regulatory systems are being designed in developing countries. Examples are the regulation of seed of so-called notied crops only in Bangladesh, and the specic rules to (de-)regulate the local and formal production of landrace seed in the draft Eritrean seed law. Strict seed laws can mean that the distribution of non-regulated seed is illegal, thus participatory plant breeding strategies that rely on informal dissemination of new selections may not be legal in a strict interpretation of many seed laws. Even on-farm testing of non-released varieties has been obstructed in some countries (e.g. Uganda at the end of the 1980s) on these grounds. It also discourages the emergence of small-scale seed enterprises that nd a niche market by selling seed of locally preferred varieties. In conclusion, seed quality control regulations tend to limit the number of varieties of seed that can be multiplied, and reduces the possibilities for alternative breeding and seed production strategies.

Intellectual property rights Intellectual property rights (IPRs) can be considered as a contract between the inventor (author, breeder, etc.) and society, by which an exclusive right to use or commercialize the protected subject matter is granted in return for public disclosure. Plant varieties

Regulatory Aspects of Breeding for Diversity

109

can, in some countries, be protected by a special type of IPR, called plant variety protection (PVP) or plant breeders rights (PBR); in some others by patents. The many countries that do not protect plant varieties, but are members of the World Trade Organization, are required to do so by patents or by an effective sui generis system through Article 27.3(b) of the Agreement on Trade Related Intellectual Property Rights (TRIPs). Furthermore, in a growing number of countries, gene constructs and biotechnological methods can now be patented. Implications for plant breeding and particularly for broadening the genetic base can be envisaged and are described in the following paragraphs. The term plant variety protection is commonly used in connection with the UPOV, but it may be used for any sui generis system to protect plant varieties. Under the UPOV convention, an exclusive right is granted to the breeder or discoverer of a variety that is distinct, homogeneous (uniform), stable and new (novelty relates to the marketing of seed of the variety). This means that the breeder is likely to be granted benets only for those varieties that conform with these DUS-N standards. Applying for PVP is only useful if the variety will be commercialized, i.e. multiplied and sold to a fairly wide range of customers. Homogeneity standards are not extremely strict compared with seed certication standards. However, as the number of varieties of any one crop species registered increases, a larger number of characteristics have to be uniform in order to distinguish a new application. A landrace of a crop for which few varieties have been registered so far could thus comply with the homogeneity requirement, but it would not pass the novelty requirement. It may, in fact, not be worthwhile to have PVP for most landraces since it is unlikely that there would be a large-scale commercial market. Specic to PVP amongst IPRs, are the breeders exemption and the farmers privilege. The breeders exemption rules that any protected variety may at any time be used to produce a new variety on which PVP can be granted. This means that PVP does not monopolize the genetic base of plant breeding. Revisions introduced in the most recent UPOV Convention (1991) may even have a positive effect on broadening the genetic base of plant breeding, since, in its clause on essentially derived varieties, varieties that have been developed by small manipulations in a protected variety are considered dependent, i.e. that the rights on the new variety are shared with the holder of the rights of the original one. This means that in order to reap the full benet of the PVP system, a breeder has to develop a new variety through more substantial changes, such as through a crossing programme, preferably with more distant parents. The farmers privilege rules that farmers are allowed to produce seed of a protected variety without consent from the owner of the variety either for his own planting, or for non-commercial purposes (the details differ according to national laws, some of which include an allowance for over-the-fence trading of seed). The farmers privilege is particularly important in developing countries where by far the largest quantities of seeds are produced within the local informal seed system (that is, saving of seed onfarm, and local diffusion of seed through farmer-to-farmer exchange: Almekinders and Louwaars, 1999). This privilege has, however, been tightened in the latest UPOV Convention (1991), and now requires a positive act by national authorities to permit it, on a crop-by-crop basis. Depending on its application then, PVP may thus restrict the use of protected varieties in local seed systems. Though designed to stimulate conventional plant breeding, PVP does not restrict

110

N.P. Louwaars

alternative breeding strategies. Indeed the breeders privilege allows for any protected variety to be used in any breeding programme, including alternative ones. However, most products of alternative breeding are either likely to be ineligible for protection, or are unlikely to benet from such protection since they are mostly intended to be fed into local seed systems. The patent system was designed for industrial inventions. Living matter had been explicitly excluded from the system for ethical and practical reasons. The practical reason was mainly that plants could not be described as precisely as methods or articles in the inanimate world. However, it is now widely considered that genes can be described, and some countries allow genes and methods for gene manipulation to be patented. Since existing patent systems do not make provision for a farmers privilege and the research exemption of the patent system is narrowly dened (compared with the breeders exemption in PVP), germplasm containing patented genes is not freely available for breeding even if the patented gene has entered a plant through natural crossing. Patents thus can cause a serious limitation in the availability of parent material in a breeding programme. Arguably, in alternative breeding programmes where a wide range of genetic resources are being used, this problem may even become more serious than in a conventional programme. In conclusion, IPRs designed to promote commercial plant breeding may sometimes be at the cost of public plant breeding for less endowed farmers, but may stimulate the use of a wider genetic base in conventional plant breeding. Strong IPRs are likely to create obstacles in the use of germplasm in alternative breeding programmes.

Regulations governing access to genetic resources, the sharing of benets and farmers rights Finally, implementation of the Convention on Biological Diversity (CBD) and the revised International Undertaking on Plant Genetic Resources for Agriculture will have implications for the availability of genetic resources. Regulations on access to genetic resources, and IPR-type arrangements for the protection of the genetic resources of local and indigenous communities, are being introduced or considered in a number of countries, and are being discussed both under the CBD, and in the Commission on Genetic Resources for Food and Agriculture. This includes, in the latter forum, elaboration of the concept of farmers rights a term introduced by the Food and Agriculture Organization (FAO) to recognize the contributions of farmers in the development of the materials that are the basis of modern plant breeding. Under the CBD, national governments have the authority to determine access to genetic resources, which is subject to national legislation and should be on mutually agreed terms. Unless determined otherwise, access is subject to prior informed consent of the providing country. Major restrictions on access to genetic resources could lead to serious obstacles for the broadening of the genetic base of plant breeding, and hamper alternative breeding strategies. As indicated in other chapters (Berthaud et al., Chapter 4; Salazar, Chapter 7, both this volume), there is often substantial geneow within and between farmers crop varieties. In certain circumstances, restrictions on access could, in fact, harm efforts by farmers themselves to improve their own varieties. The wider the parental materials used in breeding, the more barriers that may be encountered.

Regulatory Aspects of Breeding for Diversity

111

Farmers rights are meant to stimulate the conservation and use of plant genetic resources through recognition and remuneration. Farmers rights are as yet insufciently dened to fully analyse their impact on plant breeding. If farmers rights are interpreted as the rights of farmers to continue their traditional saving and exchanging of seed, and the right to receive a remuneration for the continued use of diversity, the concept can strongly support strategies for broad-based plant breeding. On the other hand, IPR-like interpretations of the concept would be likely to impede geneow and thus, in fact, be contrary to the principle of farmers rights. The same considerations apply to the elaboration of mechanisms to respect, preserve and maintain the knowledge, innovations and practices of indigenous and local communities, as required under Article 8j of the CBD. In conclusion, the full control over germplasm by communities or nations, though inspired by thoroughly ethical concerns, may in fact reduce genetic diversity and hinder farmers own efforts at plant breeding. Such restrictions could also result in it being easier to use modern varieties as sources of genetic material (because of the breeders exemption) than to take local landraces as a starting point of breeding.

Macro-economic policies that promote the narrowing of the genetic base While seed laws and IPRs greatly inuence the economics of plant breeding and seed supply, other economic factors are also important, and thus are briey mentioned here (McGuire, 1997). Fundamental changes have occurred in the organization of plant breeding in developed countries since the 1980s. The privatization of plant breeding research (as in the case of the Plant Breeding Institute, Cambridge, UK), or its commercialization (as in the case of many universities and research departments), has increased the inuence of economic forces on plant breeding. Competitive plant breeding is likely to have a positive effect on the rate of development of new nished varieties, but it may reduce the opportunities for commercial breeders to invest in new approaches to breeding. Base-broadening or genetic enhancement programmes are typical examples: the investment in the use of more diverse germplasm means a high initial cost and risk for the company, whereas successful results would be available to all competitors as soon as the rst variety is released. Competitive plant breeding, where crossing the two best varieties in the market at any particular moment is likely to yield a slightly better one in a short period, thus results in a narrowing of the genepool that is used for actual breeding. This applies particularly to self-fertilizing species such as wheat and barley, where commercial prospects for the sale of seed are poor. In developing countries also there is a trend towards the commercialization of public agricultural research as research institutions increasingly have to compete for public funds and concentrate on earning revenues in the market. Since they produce concrete products, breeding departments of national agricultural research systems (NARS) are particularly susceptible to these pressures. The result is that such breeding programmes concentrate even more on those sectors of the farming economy likely to afford to pay a seed price that includes a research bonus. This again strengthens the trend of producing varieties that t the variety release and seed certication systems and relatively uniform farming conditions.

112

N.P. Louwaars

Impact of Regulatory Frameworks on Alternative Breeding Methods

Where conventional breeding methods may be very successful to develop varieties for uniform and favourable farming conditions, they rarely manage to develop varieties that are adapted to ecologically diverse and marginal conditions. Farmers breeding methods for such conditions generally result in genetically diverse landraces. Such landraces are maintained through a combination of conscious and unintentional selection by both the farmers themselves and the biotic and abiotic environment. Farmers may also consciously look out for off types in their crop and experiment with them. Improvement of landraces through mixing introduced materials into a landrace, or promotion of cross-fertilization by planting different materials close to each other, is also part of the traditional process of landrace development. To benet from the value of genetic diversity in particular farming systems, scientic breeders are emulating some traditional methods. Strategies broadly include:

Breeding for plasticity within a variety, e.g. genes or gene combinations that respond to multiple stresses. Breeding for diversity within varieties, e.g. many alleles of one gene in one genotype. Breeding for specic adaptation, i.e. for diversity among varieties.

In order to be effective, participatory methods have to be used, especially in the latter two strategies. The importance of breeding for diversity can be illustrated in the case of breeding for disease resistance (Louwaars, 1997). A monogenic (vertical) resistance can be broken by the pathogen, causing massive losses. Such losses are more disastrous for riskprone, small-scale farmers in low external input farming conditions, compared with their more commercial colleagues. The seed replacement rate in commercial farming is high, so when a resistance is broken farmers can purchase seeds of a new variety the next season, assuming that the breeders have done their job in being ahead of nature. Their loss has to be absorbed for 1 or 2 years only, which can be done in the presence of effective rural credit. This was the case for the massive destruction of maize due to blight in the early 1970s in the USA. Breeders for low external input farmers, who are mostly in the public sector, have different responsibilities. Their clients obtain seed to a large extent through the local seed exchange mechanisms, and variety replacement is slow. In the absence of effective rural credit schemes and commercial seed supply systems, a serious outbreak of a disease cannot be tackled by the market. Durability of resistances to diseases is therefore even more important in such systems. Breeding for polygenic resistance or multilines should be promoted, even though the methodologies are more complex. This is currently discouraged due to very ineffective variety release systems. Breeding for partial (horizontal) resistance, for example, is biased against, since, as explained on page 107, such varieties commonly do carry some disease symptoms and are rejected by the release system. Breeding multilines can also encounter problems. An illustrative case is a multiline wheat variety bred by The Netherlands cooperative seed company Zelder in the late 1970s. This novelty put a lot of pressure on The Netherlands variety release system. The variety, aptly named Tumult (Dutch for commotion), was insufciently uniform, and thereby could not be released. The nal option was to register the con-

Regulatory Aspects of Breeding for Diversity

113

stituent lines as separate varieties. These are hardly distinctive because they are different only for the resistance gene. Moreover, testing of the lines for VCU would cause problems: the resistance to yellow rust of individual lines was below standard, but when tested as a multiline the variety was superior. Eventually the registration authority was convinced by the breeder to test the multiline variety as one entity. Commercially, however, Tumult turned out to be a failure. The breeding and release procedure caused the multiline variety to be outdated before it even appeared on the market. Additionally, maintaining the lines and producing the seed separately were too expensive. The Zelder seed company has not produced multiline varieties since. In breeding for diversity among varieties, conventional variety release systems provide quite effective sieves that allow only a very limited number of varieties to reach the farmers. Where the number of different (uniform) varieties has to constitute genetic diversity in the eld, any regulated system is going to frustrate such strategies. Breeding for diversity among varieties basically means the selection of many different lines for many different conditions. Participatory plant breeding commonly leads to such results, when farmers are stimulated to select from wide germplasm in on-station or regional nurseries. Again, the physical release of the material is not connected with subsequent formal seed production, so in this case strategies are obstructed not supported. The main risk of the use of wide germplasm in such breeding programmes is that in the future more and more proprietary material will enter the scene, which may create restrictions. The same accounts for strong farmers rights or strict implementation of national sovereignty. The importance of such regulatory frameworks in breeding strategies has been acknowledged only very recently. Regulations are an impediment to novel approaches, such as breeding for specic adaptation and participatory plant breeding, which are very valuable tools for targeting less endowed farmers.

Conclusions and Recommendations

There are a variety of reasons why current public seed regulation is unsatisfactory: it is not efciently organized, often uses inappropriate standards, does not offer opportunities for farmer and seed producer participation, and is not sufciently transparent. Thus, adjustment to seed regulatory frameworks is necessary because of signicant changes in national seed systems, including reductions in budget for public agricultural research; the failure of many seed parastatals; pressure for the establishment of plant variety protection; the increasing contributions of commercial seed enterprises; and the emergence of innovative local-level variety development and seed production initiatives, as well as increasing concern about plant genetic diversity (Tripp, 1997; Tripp et al., 1997). There are a number of options for regulatory reform. In plant breeding, more emphasis should be placed on decentralizing variety testing, breeding for particular niches, and making site selection, trial management and analysis more representative of farmers conditions. In variety regulation, simpler registration procedures are required, and the demands of plant variety protection should not be allowed to bias or limit the development and use of public and farmer varieties. Variety performance testing for release should be made more exible. In seed quality control, standards should be

114

N.P. Louwaars

re-examined for their relevance to particular farming conditions, and much of the responsibility for monitoring seed quality should be passed to seed producers and merchants, accompanied by well-dened public oversight and enforcement mechanisms. Introduction of intellectual property rights, rules on access to genetic resources, and mechanisms to reward local communities should be done in such a way as to avoid or minimize negative impacts on the use of genetic diversity.

References
Almekinders, C.J.M. and Louwaars, N.P. (1999) Farmers Seed Production. New Approaches and Practices. Intermediate Technology Publications, London. Ceccarelli, S. (1989) Wide adaptation: how wide? Euphytica 40, 197205. Louwaars, N.P. (1997) Regulatory aspects of breeding for eld resistance in crops. Biotechnology and Development Monitor 33, 68. Louwaars, N.P. and van Marrewijk, G.A.M. (1996) Seed Supply Systems in Developing Countries. CTA, Wageningen. McGuire, S. (1997) The effects of privatisation on winter-wheat breeding in the UK. Biotechnology and Development Monitor 33, 11. Tripp, R. (1997) New Seeds and Old Laws. Intermediate Technology Publications, London. Tripp, R, Louwaars, N, van der Burg, W.J. and Witcombe, J.R. (1997) Alternatives for Seed Regulatory Reform. An Analysis of Variety Testing, Variety Regulation, and Seed Quality Control. Agricultural Research and Extension Network paper No. 69, Overseas Development Institute, London. Virk, D.S, Packwood, A.J. and Witcombe, J.R. (1996) Varietal Testing and Popularisation and Research Linkages. Discussion papers series, Centre for Arid Zone Studies, Bangor.

Decentralized and Participatory Plant Breeding for Marginal Environments

S. CECCARELLI1, S. GRANDO1, A. AMRI2, F.A. ASAAD3, A. BENBELKACEM4, M. HARRABI5, M. MAATOUGUI4, M.S. MEKNI6, H. MIMOUN4, R.A. EL-EINEN3, M. EL-FELAH6, A.F. EL-SAYED3, A.S. SHREIDI7 AND A. YAHYAOUI8.
Center for Agricultural Research in Dry Areas (ICARDA), PO Box 5466, Aleppo, Syria; 2National Institute for Agricultural Research (INRA), BP 415, Rabat, Morocco; 3Field Crops Research Institute, Agricultural Research Center (ARC), PO Box 12619, 9 Gamaa Street, Giza, Egypt; 4Institut Technique des Grandes Cultures (ITGC), BP 16, El-Harrach, Algiers, Algeria; 5Institut National Agronomique de Tunisie (INAT), 43 Avenue Charles Nicole, 1082, Tunis-Belvdre, Tunisia; 6Institut National de la Recherche Agronomique de Tunisie (INRAT), Rue Hedi Karray, 2080 Ariana, Tunis, Tunisia; 7Agricultural Research Center (ARC), PO Box 2480, Tripoli, Libya; 8Ministre de lAgriculture, Institution de la Recherche et de lEnseignement Suprieure Agricoles (IRESA), 30 Rue Alain Savary, 1002 Tunis, Tunisia
1International

Introduction
Formal plant breeding has been benecial to farmers in favourable environments, and to those who could protably modify their environment to suit new cultivars. It has been less benecial to those farmers, the poorest, who cannot afford to modify their environment through the application of additional inputs (Byerlee and Husain, 1993; Eyzaguirre and Iwanaga, 1996; Trutmann, 1996). Poor farmers in marginal environments continue to suffer from chronically low yields, crop failures and, in the worse situations, malnutrition and famine, while farmers in favourable environments who use large quantities of inputs are now concerned with the adverse environmental effects and the reduction of genetic diversity. Because of the successes in favourable environments, plant breeders have tried to solve the problems of poor farmers living in unfavourable environments by simply extending the same methodologies and philosophies applied to favourable, high-potential environments, without considering the possible limitations associated with the presence of large genotype environment (GE) interactions.
IPGRI/FAO 2001. Broadening the Genetic Base of Crop Production (eds H.D. Cooper, C. Spillane and T. Hodgkin)

115

116

S. Ceccarelli et al.

In unfavourable environments, progress with empirical breeding has been negligible, with Australia where most of the selection takes place in the target environments as one well-known exception. As a result, the yield of some important staple crops has shown only modest increase or remained virtually unchanged. This has been attributed to the difcult nature of the target environment (Passioura, 1986; Blum, 1988), and has led to the widespread habit of conducting most selection work in favourable conditions (Simmonds, 1991). Rather than questioning the dangers of selecting in an environment different from the target environment, much research has been done, and resources spent, to seek alternatives, such as physiological traits, to empirical breeding for unfavourable conditions. GE interactions are almost unanimously considered to be among the main factors limiting response to selection and the efciency of breeding programmes in general. GE interactions become important when the rank of breeding lines changes in different environments (crossover GE interaction Baker: 1988). GE interactions in general, and GE interactions of crossover type in particular, have a negative impact on the progress of breeding programmes, particularly when breeders try to avoid them by searching for widely adapted cultivars. However, breeders may exploit GE interactions by selecting for specic adaptation (Ceccarelli, 1989; Hildebrand, 1990; Stroup et al., 1993). This is particularly relevant when breeding for adaptation to unfavourable environments and subsistence agriculture. Throughout this chapter, unfavourable environments are dened as those where crop yields are commonly low due to the concomitant effects of several abiotic and biotic stresses. The semi-arid areas of Syria, where barley is the predominant rain-fed crop, are a good example of such environments where not only low annual rainfall, but also rainfall distribution, low winter temperatures, and high temperatures and hot winds from anthesis to grain-lling are important abiotic stresses. Although the frequency, timing, intensity and duration of each of these stresses, as well as their specic combinations, vary from year to year, pre-anthesis water stress is common and postanthesis water stress is the rule (Fig. 6.1). Because of the high probability of low yields and crop failures in unfavourable environments, the use of inputs such as fertilizers, pesticides and weed control is seen by farmers as risky. Therefore, the adoption of improved agronomic practices has been limited, and the only economic solution to increase crop yields in unfavourable environments is through breeding. However, empirical plant breeding for these environments has been historically much less successful than it has for more favourable or for high-potential environments. This chapter describes a strategy to exploit GE interactions in national and international breeding programmes for such areas. Issues such as the choice of the selection environment, specic versus wide adaptation, and genetic uniformity versus genetic diversity are discussed, and some implications for the genetic base are noted.

Challenging Conventional Breeding Concepts

Most plant breeders assume that breeding for environments where drought and other stresses are unpredictable and variable is too slow and too difcult. The target is hard to dene; heritabilities, and hence responses to selection, are too low to achieve

Plant Breeding for Marginal Environments

117

Fig. 6.1. Cumulative frequency distribution of barley yields in two climatic zones of Syria (zone C receiving less than 250300 mm annual rainfall, and zone B receiving 300350 mm annual rainfall) between 1982/83 and 1996/97. The distributions are based on 97 and 134 locationyear combinations for zones C and B, respectively. In zone C, low yields of barley are common, crop failures occur one year out of 10, and yields above 2.5 t ha 1 are expected about 15% of the time. In zone B, yields of 1 t ha 1 or less have a frequency of about 7%, and yields above 2.5 t ha1 occur about 45% of the time.

meaningful results. Therefore, most of the breeding for stress environments has actually been conducted using the same basic approach that has been very successful in areas where lack of water (and other abiotic stresses) is seldom important. With few exceptions, many breeding programmes share the following concepts:

Selection has to be conducted under the high-input conditions of research stations, where environmental noises can be kept under control, error variances are smaller and response to selection are higher. Cultivars must be genetically homogenous (pure lines, hybrids, clones) and must be widely adapted over large geographical areas; locally adapted landraces must be replaced because they are low-yielding and disease-susceptible. Seed of improved cultivars must be disseminated through mechanisms and institutions such as variety release committees, seed certication schemes and governmental seed-production organizations. The end-users of new cultivars are not involved in selection; they are only involved at the end of the consolidated routine (breeding, researcher-managed trials, verication trials), to verify whether the choices made for them by others are appropriate or not.

Until recently, breeders seldom questioned these assumptions. When they have, it has been found that:

Selection in high-input research stations tends to produce cultivars that are superior to local landraces only under improved management, not under the low-input conditions typical of the farming systems of stress environments. The result is that

118

S. Ceccarelli et al.

although many new cultivars out-yield local landraces on research stations and some are released few (if any) are actually grown by farmers in difcult environments. Poor farmers in stress environments tend to maintain genetic diversity in the form of different crops, different cultivars within the same crop, and/or heterogeneous cultivars to retain adaptability (Simmonds, 1962), i.e. to maximize adaptation over time (stability or dependability), rather than adaptation over space (Martin and Adams, 1987). Diversity and heterogeneity serve to disperse or buffer the risk of total crop failure due to environmental variation. This is in sharp contrast with the trend of modern plant breeding towards uniformity. Resource-poor farmers seldom use the formal seed-supply systems. They frequently rely on their own or on neighbours seed (Almekinders et al., 1994). Therefore, when the appropriate cultivar is selected, adoption is much faster through non-market methods of seed distribution (Grisley, 1993). When farmers are involved in the selection process, their selection criteria may be different from those of the breeder (Hardon and de Boef, 1993; Sperling et al., 1993; Sperling et al., Chapter 27, this volume). Typical examples are crops used as animal feed such as barley, where breeders often use grain yield as the sole selection criterion while farmers are usually equally concerned with forage yield and the palatability of both grain and straw.

Because the concepts of conventional plant breeding are not questioned, the blame for the non-adoption of new cultivars is variously attributed to the ignorance of farmers, the inefciency of extension services and the unavailability of seed of improved cultivars. Thus, enormous resources continue to be invested in a model of breeding which is unlikely to succeed in unfavourable agroclimatic conditions. The contrast between the reality of the farming systems and the plant breeding philosophies is particularly striking in developing countries. This is not surprising: most of the breeders from developing countries have received their training in those rarely questioned breeding principles enshrined in developed countries.

Genotype by environment interactions of crossover type Since GE interactions of crossover type pose major problems in conventional breeding programmes, the question of how frequently these interactions occur is important. In general, when different lines or cultivars of a given crop are evaluated in a sufciently wide range of environments, GE interactions of crossover type seem to be very common. Examples of GE interactions of crossover type can be found in the literature in a range of crops and environments, and for drought, temperature and salinity stresses: Blum and Pnuel (1990) and Calhoun et al. (1994) in wheat; Breese (1969) in cocksfoot; Arboleda-Rivera and Compton (1974), Muruli and Paulsen (1981), Hildebrand (1984), Lofer et al. (1986), Latte and Edmeades (1994), Byrne et al. (1995) and Bnziger et al. (1997) in maize; Simmonds (1984a) in sugar cane; Lawn (1988) in chickpea; Ceccarelli (1989), Ceccarelli and Grando (1991), Jackson et al. (1993) and Muoz et al. (1998) in barley; Virk and Mangat (1991) in pearl millet; Chisi et al. (1966) in sorghum; and Shannon and Francois (1978) in muskmelon.

Plant Breeding for Marginal Environments

119

In the barley breeding programme of the International Center for Agricultural Research in Dry Areas (ICARDA), the breeding material is routinely tested in sites differing in total rainfall, soil fertility and winter temperatures, with an average fourfold difference in mean yields between the two most contrasting sites. When we compared different lines at each of the two most contrasting sites, a consistent pattern was observed over a period of 14 years (Table 6.1). When compared in low-yielding sites (LYE), those lines which had the highest yield in high-yielding sites (GH) yielded signicantly less than the lines which had the highest yield in low-yielding sites (GL). When the comparison was made in high-yielding sites (HYE), those lines which had the highest yield in the lowest-yielding sites (GL) yielded signicantly less than the lines which had the highest yield in high-yielding sites (GH). A similar picture emerges from the few published data on genetic correlation between yield measured in high- and low-yielding conditions (Atlin and Frey, 1989, 1990; Ceccarelli et al., 1992; Ud-Din et al., 1992; Cooper and DeLacy, 1994). This indicates that, as a general phenomenon, high-yielding lines under optimum growing conditions do not perform well under poor growing conditions, and vice versa. This is hardly surprising. Physiologists have long recognized, with specic reference to drought, that different physiological mechanisms and different phenologies are associated with high yield in favourable conditions and high yield in unfavourable conditions (Hsiao, 1982; Blum, 1993). Figure 6.2 shows an example of GE interactions of crossover type in barley. Lines selected for high grain yield in high-yielding sites (YP) yielded more than lines selected in low-yielding sites (YD) in the medium- to high-yielding locationyear combinations, and the YD lines yielded more than the YP lines in the low- and very low-yielding locationyear combinations (Ceccarelli et al., 1998). Between the medium- and
Table 6.1. Average grain yield (t ha 1) in the lowest (LYE) and highest (HYE) yielding sites in each of 13 cropping seasons, of the 5% highest-yielding lines in the lowest-yielding site (GL), and of the 5% highest-yielding lines in the highest-yielding site (GH). LYE Year 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 Mean GL 1.2830.073 1.9350.026 1.0760.013 4.1990.048 1.2870.018 0.7940.017 1.6930.024 1.3050.026 0.8510.018 1.7020.024 0.6800.10 1.8600.032 1.4560.005 1.2730.066 1.528 GH 0.7670.076 1.3400.079 0.6330.050 3.3830.131 0.6580.052 0.429 0.046 0.9530.110 1.0300.046 0.5310.041 1.3770.081 0.4510.085 1.3700.093 0.8730.091 0.3200.051 1.008 GL 3.4620.121 4.1390.078 2.6870.090 4.6720.173 4.8740.315 3.0660.118 4.7050.254 4.7990.171 3.7690.107 3.2100.107 3.8720.317 4.0890.425 3.6730.058 4.0000.538 3.930 HYE GH 4.1390.029 4.9700.070 3.5470.037 6.1000.067 7.8140.107 4.1220.027 6.0730.065 5.7930.099 4.0740.085 3.6630.047 5.1090.068 5.2140.078 6.2180.061 7.0330.155 5.276

120

S. Ceccarelli et al.

low-yielding locationyear combinations the YP and YD lines cross over (note that the range of environments in Fig. 6.2, and their associated yield levels, may represent either variation over time within one given geographical area or variation over space, i.e. between different geographical areas within or across countries). Four points emerge from the analysis of multi-environment trials such as those illustrated in Table 6.1 and Fig. 6.2. First, the denition of unfavourable environments plays a key role in determining breeding strategies. Dening as unfavourable environment an environment with an average yield at or above the crossover point is often justied for some crops. Taking Syria as an example again, bread wheat is cultivated only in areas with more than 300 mm annual rainfall, and the probability of yields lower than 2 t ha1 is considerably less than for barley: this may result in the absence of crossover and hence in different breeding strategies for the two crops (Calhoun et al., 1994). Secondly, crossover may be affected by the type of comparisons made in multi-environment trials. The superiority of lines such as the YP lines in Fig. 6.2 over the mean of all the lines in the trial, or against a local check, may be taken as an indication of broad adaptation. Very rarely is the comparison made with lines specically selected in stress environments below the crossover point (such as the YD lines) because few breeding programmes select specically for performance under stress conditions. When this is done, the YP lines are still broadly adapted, but to a narrower range of environments. Thirdly, the presence of a crossover GE interaction is often neglected by conducting selection and testing only above the crossover point, and particularly in high-input research stations. When breeding materials are evaluated in many locations and selec-

Fig. 6.2. Crossover type of GE interaction: YP and YD are lines selected on the basis of the 19881990 data in high- and low-yielding environments respectively; and then tested in 21 locationyear combinations between 1991 and 1994.

Plant Breeding for Marginal Environments

121

tion is made for high average yield across locations, this is equivalent to selection for responsiveness to favourable conditions (Simmonds, 1991). Even shuttle breeding between two contrasting environments can miss the presence of crossover GE interactions if both environments are above the crossover point. Fourthly, while the existence of a crossover GE interaction has been allegedly disproved by the release of cultivars for low-yielding environments (Byerlee, 1994), this argument neglects the fact that many of these cultivars have never been adopted by farmers in the intended environment (Maurya et al., 1988; Jansen et al., 1990; Byerlee and Husain, 1993). An argument commonly used in favour of selecting in good environments is that heritabilities are higher there than in poor environments (Blum, 1988). But the experimental evidence does not consistently suggest that heritability in low-yielding conditions is lower than that in high-yielding conditions (Table 6.2), and even when it is lower, it is often not sufciently lower to offset the effect of a low correlation coefcient. Our own experience with barley (Singh and Ceccarelli, 1995) also suggests that there is no relationship between yield level and magnitude of heritability.

Table 6.2. Heritability estimates of grain yield at low and high yield levels and ratio between the root squares of the two heritabilities in different crops (modied from Ceccarelli, 1994). Crop Wheat Wheat Wheat Wheat Wheat Cocksfoot Common bean Common bean Flax Maize Maize Maize Barley Barley Barley Oat Oat Oat Oat Sunower Sunower Sunower Sorghum Sorghum Sorghum Soybean High (hy) 0.25 0.78 0.33 0.89 0.93 0.89 0.59 0.19 0.44 0.52 0.46 0.62 0.47 0.65 0.47 0.56 0.45 0.38 0.67 0.86 0.74 0.79 0.63 0.17 0.69 0.56 Low (hx) 0.03 0.32 0.68 0.74 0.43 0.50 0.55 0.20 0.56 0.71 0.58 0.44 0.54 0.66 0.68 0.63 0.32 0.52 0.32 0.57 0.44 0.24 0.43 0.58 0.32 0.31 Ratio hy/hx 2.89 1.56 0.70 1.10 1.47 1.33 1.04 0.97 0.89 0.86 0.89 1.19 0.93 0.99 0.83 0.94 1.18 0.85 1.45 1.23 1.3 1.81 1.21 0.54 1.47 1.34 Reference Roy and Murty (1970) Allen et al. (1978) Pederson and Rathjen (1981) Pfeiffer (1988) Cooper et al. (1997) Breese (1969) Schneider et al. (1997) Schneider et al. (1997) Allen et al. (1978) Selmani and Wassom (1993) Latte and Edmeades (1994) Bnziger et al. (1997) Allen et al. (1978) Weltzien and Fischbeck (1990) Singh and Ceccarelli (1995) Allen et al. (1978) Frey (1964) Johnson and Frey (1967) Atlin and Frey (1990) Alza and Fernandez-Martinez (1997) Alza and Fernandez-Martinez (1997) Alza and Fernandez-Martinez (1997) Zavala-Garcia et al. (1992) Chisi et al. (1996) Zavala-Garcia et al. (1992) Allen et al. (1978)

122

S. Ceccarelli et al.

A positive interpretation of GE interactions (Ceccarelli, 1989, 1994; Hildebrand, 1990; Stroup et al., 1993; Gauch and Zobel, 1997) recognizes the importance of specic adaptation and leads to the selection of lines specically adapted to favourable environments (such as the YP lines in Fig. 6.2), and of lines specically adapted to unfavourable environments (such as the YD lines in Fig. 6.2). Cooper and De Lacy (1994) and Cooper et al. (1996) use a similar strategy in the presence of large and repeatable GE interactions. The ICARDA barley breeding programme has shown (see Grando et al., Chapter 22, this volume) that it is indeed possible to improve the production of a typically lowinput crop such as barley, grown in environments with low and poorly distributed rainfall, low temperatures in winter, high temperatures and drought during grain-lling, low soil-fertility and poor agronomic management by using strategies and methodologies based on: (i) decentralized selection for specic adaptation in the target environment; (ii) use of locally adapted germplasm; (iii) use of suitable plot techniques and experimental designs to control environmental variation; and (iv) participation of farmers in selection. These strategies and methodologies have been described in detail by Ceccarelli and Grando (1996), and the rest of this chapter will concentrate on how to use the concept of specic adaptation in international plant breeding programmes.

Decentralization: using specic adaptation in international breeding programmes International breeding programmes aim to assist national programmes to increase agricultural production by developing superior cultivars. This is traditionally done through very large breeding programmes that develop xed or semi-xed lines with an average good performance across many environments (often high-input research stations). This type of interaction between international and national plant breeding programmes has been largely a one-way, top-down process (Simmonds and Talbot, 1992) where international programmes develop germplasm, distribute it as international nurseries, and national programmes test and eventually release selections as cultivars. This top-down approach has excluded the use of locally adapted germplasm, which is specically adapted to particular conditions and often performs poorly in the favourable conditions of research stations, and has, in fact, encouraged its displacement. The distribution of germplasm from international centres to national breeding programmes has indeed historically included segregating populations. However, such segregating populations were derived from crosses designed by the international breeders, were the same for all the countries and were not targeted to a specic environment. The adoption of a positive interpretation of GE interaction by international breeding programmes has been advocated as a way to address the need of smallholder, resource-poor farmers, who have been bypassed by the Green Revolution (Stroup et al., 1993). To exploit specic adaptation fully and make positive use of GE interactions, Ceccarelli et al. (1994) propose that international breeding programmes should decentralize (that paper uses the term devolution; decentralization is a better term) most of the selection work to national programmes by gradually replacing the traditional international nurseries either with targeted segregating populations, or with nurseries that

Plant Breeding for Marginal Environments

123

contain material selected only for highly heritable traits. Early distribution of breeding material reduces the danger of useful lines being discarded because of their relatively poor performance at some test sites. This problem is illustrated by 288 barley lines evaluated both in the Maghreb countries (Libya, Tunisia, Algeria, Morocco) and in ICARDAs preliminary yield trials grown at three sites in Syria (ranging from moderately favourable to unfavourable) in 1991/92. In Syria, 103 entries were selected and in the Maghreb 154; but only 49 were selected in both. More than half of the lines selected in Syria were discarded in the Maghreb, and almost 70% of those selected in the Maghreb were discarded in Syria. This gives a measure of the danger, in a centralized breeding programme, of discarding lines potentially useful in other areas. In 1991, ICARDAs barley breeding programme started a gradual process of decentralization of selection work to the four Maghreb countries (Ceccarelli et al., 1994). This was extended to Iraq in 1992, to Egypt in 1995, and it is gradually being implemented in the Mediterranean highlands of Turkey and Iran. When the term decentralization is used in a plant breeding context and it is referred to as decentralized breeding, it may actually refer to two fundamentally different processes, namely decentralized selection and/or decentralized testing. Decentralized selection is a term rst used by Simmonds (1984a) and dened as selection in the target environment(s). In the case of self-pollinated crops, it consists of selection of early segregating populations (such as F2) in a number of locations representing the target environment(s) (climate, soil, farming system and management) the breeding programme aims to serve. Decentralized selection becomes selection for specic adaptation when the selection criterion is the performance in specic environments rather than the mean performance across environments. Decentralized selection is different from decentralized testing, which is a common feature of breeding programmes and takes place, usually in the form of multilocation trials and on-farm trials, after a number of cycles of selection in one or a few environments (usually with high levels of inputs). In the type of decentralization implemented in North Africa (Fig. 6.3), nationalprogramme scientists identify suitable parents, crosses are made at ICARDA, the material is advanced as bulks (without any selection) to the F3 generation, and then distributed to the national programmes. The reason for advancing material to F3 is to ensure the availability of sufcient seed for distribution. Selection between populations is made in the target environments in each country. Since barley is mostly used as an animal feed in most of the countries in North Africa and the Near East, uniform cultivars are not required, and successful bulks could, in theory, be released directly as cultivars after adequate testing. Unfortunately, however, a high degree of genetic uniformity is still required for variety release, even in those countries where farmers grow heterogeneous landraces. To meet these requirements, selection within the best populations is done within each country and at ICARDA through single seed descent, and the best lines are used for further cycles of crosses. The decentralization to Iraq is similar, in principle, to the one described for North Africa, but much simpler in its implementation. Two types of targeted crosses are made at ICARDA with elite germplasm identied in Iraq, one for the Baghdad-type of environment (irrigated areas, warm winters, high temperature and salinity stress, barley used for dual purpose) and one for the Mosul-type of environment (low rainfall,

124

S. Ceccarelli et al.

Crosses are made at ICARDA

The best F5 are yield tested in specific yield trials for each country

F3 bulks grown at 22 locations in Egypt, Libya, Tunisia, Algeria and Morocco Selection within the best bulks is made in each country NARS

Second cycle of selection of F4 bulks at the same locations in the five countries

Fig. 6.3. Scheme of decentralized barley breeding between ICARDA and ve national programmes in North Africa.
ICARDA: International Center for Agricultural Research in Dry Areas.

continental climate with cold winters). The F2s are sent to Iraq and all the successive stages of selection are conducted in Iraq without backup from ICARDA. Released cultivars and/or promising lines are sent back to ICARDA for crosses. Other breeding activities at ICARDA, such as lentil breeding for the Anatolian plateau in Turkey and for South Asia, are based on the same philosophy (see Erskine et al., Chapter 21, this volume). These examples show that decentralization does not have to follow a recipe, but can take several forms depending on the nature, the strength and the expertise of the national programmes. Since it started, the process of decentralizing the barley breeding programme has been gradually extended to other countries and regions (Table 6.3). For each country and/or region where the two most important conditions for decentralized breeding exist namely large GE interactions with ICARDA research station(s) and availability of local expertise in plant breeding decentralization generally follows three steps: rst, we send a special nursery to identify suitable parents; secondly, we start a specic crossing programme aimed at developing a specic germplasm pool for that country/region; and, thirdly, we distribute the segregating populations. When fully implemented, the rst step is replaced by the routine in-country screening of various germplasm sources. This, together with the decentralized screening for resistance to pests and diseases, ensures that more and more parental material is supplied by the individual countries. Another type of decentralization has been practised in Ethiopia, where, in 1984, the barley breeding programme was almost entirely based on exotic germplasm, and where, now, the major thrust is on Ethiopian landraces, tested in farmers elds at low input levels. In this case we have decentralized a methodology (selection of landraces in the target environment) rather than a germplasm (Berhane et al., 1997; Yitbarek et al., 1998). Decentralized selection is very powerful in maintaining genetic diversity. As an

Plant Breeding for Marginal Environments Table 6.3. Countries and regions where decentralized barley breeding has been initiated. Country/region North Africa Near East Near East East Africa/Red Sea Ethiopia Central Asia Turkey Cyprus Far East Pakistan Gulf countries Ecuador Countries/area (Egypt, Lybia, Tunisia, Algeria, Morocco) Central Iraq (Baghdad area) Nortern Iraq (Gezira) (Yemen, Eritrea, Tigray) The whole country except Tigray Uzbekistan, Kirgystan, Kazakhstan, Armenia (India, Thailand, China) (Saudi Arabia, Qatar, Oman) Status

125

Fully implemented Fully implemented Fully implemented First crosses made in 1998 Use of local landraces fully implemented; rst crosses in 1998 First special nursery in 1997 First nursery planned for 1999 First special nursery in 1995; rst crosses in 1998 First special nursery in 1996, rst crosses in 1997 First special nursery in 1997 First crosses made in 1992 First nursery planned for 1999

example, we show the frequency of F4 bulks and of F5 bulks which were selected from the F3 bulks distributed in 1994, 1995 and 1996 (Table 6.4) following the scheme shown in Fig. 6.3. The selection pressures used in decentralized selection appear much milder than those commonly applied in breeding programmes, so that much of the original diversity present in the original sets of F3 bulks is still present in the subsequent cycles, with as much as 80% of the bulks still being retained after two cycles of selection in the sets distributed in 1994 and 1995. In the third set, the frequency of bulks after two cycles of selection drops to 56% (mostly because no data were returned from Algeria, and consequently those bulks that would have been selected only in Algeria are missing). While the national programmes were accepting decentralization very positively, we started recognizing that decentralization per se will not necessarily respond to the needs of resource-poor farmers in less-favoured areas, if it is only a decentralization from the research station(s) of ICARDA to the research stations of the national programmes, and if the research stations of the national programmes do not represent, as is often the case, the difcult environments where the crop is predominantly grown. To exploit the potential gains from specic adaptation to low-input conditions, breeding must be

Table 6.4. Number and frequency of bulks selected in two successive cycles of selection in four North African countries starting from F3 bulks distributed in 1994, 1995 and 1996. F3 bulks Year 1994 1995 1996 No. 224 208 269 Year 1995 1996 1997 F4 bulks No. 197 185 257 % 87.9 88.9 95.5 Year 1996 1997 1998 F5 bulks No. 180 169 151 % 80.4 81.3 56.1

126

S. Ceccarelli et al.

decentralized from research stations to farmers elds. Although decentralization and farmer participation are unrelated concepts, decentralization to farmers elds almost inevitably leads to the participation of farmers in the selection process. Therefore, the ICARDA barley programme initially considered farmer participation as a type of decentralized selection to exploit GE interactions and to make use, within a formal breeding programme, of the farmers knowledge about the crop, its specic uses and its specic adaptation.

Maximizing Specic Adaptation Farmers Participation

The idea of farmers participating in the development of new technologies is neither new nor revolutionary (Rhoades and Booth, 1982; Sperling, et al., 1993; Farrington, 1996). It was introduced in 1982 (Rhoades and Booth, 1982) as the farmer-back-tofarmer model, later modied into the farmer-rst-and-last model (Chambers and Ghildyal, 1985) and more recently discussed by Sperling et al. (1993), Stroup et al. (1993) and Ceccarelli et al. (1996, 1997). Using Sperlings terminology, formal breeding programmes can be described as a sequential and cyclical process in which: (i) an extremely large amount of genetic variability is continuously created; (ii) this variability is drastically reduced through cycles of selection (we have seen that this is often done in conditions which have little in common with those of resource-poor farmers); and (iii) the few lines surviving step (ii) are presented to farmers who are asked to verify if the choices made for them are appropriate. Besides GE interactions discussed earlier, there is evidence that when farmers are involved in the selection process their selection criteria may be different from those of the breeder (Hardon and de Boef, 1993; Sperling et al., 1993). This is shown by the data of Table 6.5. One of the special nurseries distributed to Tunisia as part of the decentralized barley breeding, described above, was planted near Tejerouine, a village in southern Tunisia close to the border with Algeria. The nursery included 207 barley types, mostly early segregating populations (F4), together with check cultivars. The latter were both the landraces grown in North Africa and improved cultivars. The farmer was more selective than the breeder. This is expected since breeders usually select not only lines that could become cultivars, but also those with some other potentially useful specic traits, while farmers are presumably only interested in lines that could become cultivars. The type of material selected was of interest. Martin, the local landrace, was present once among the 40 lines selected by the breeder, twice among the 13 lines selected

Table 6.5. Number and percentage of lines selected by a breeder, a farmer and the farmers wife, from a nursery of 207 barley breeding lines (Tejerouine, Tunisia, 1996). Selected by Breeder Farmer Farmers wife No. (%) 40 (19.3) 13 (6.3) 14 (6.8) In common with: Farmer Farmers wife 2 3 0

Plant Breeding for Marginal Environments

127

by the farmer, and was never selected by the farmers wife. She was the only one to select twice one of the two Algerian landraces, Tichedrette, and once the other Algerian landrace, Saida. The farmer also selected Saida twice, but from two plots not selected by his wife (Saida was present three times in the nursery). Therefore, although chosen from different plots, the farmer and his wife eventually agreed on at least one cultivar. Neither the farmer nor his wife selected improved cultivars. Examples of different selection criteria between farmers and breeders can be expected in crops used as animal feed, such as barley. Breeders often use grain yield as the sole selection criterion which usually brings with it high harvest index and lodging resistance. However, in unfavourable environments lodging is seldom a problem because of moisture stress, and farmers are interested not only in grain yield, but also in straw yield and in the palatability of both grain and straw. A drastic difference between the breeders and farmers selection criteria was found during the initial stages of farmers participation in the ICARDA barley breeding programme. This informal participation led to the formulation of a farmers barley ideotype for dry areas characterized by tall plants and soft straw. A crop which remains tall even in very dry years is important to farmers, because it reduces their dependence on costly hand-harvesting (Ceccarelli et al., 1995), while soft straw is considered important for palatability. It is obvious that these two characteristics represent a drastic departure from the typical selection criteria used in breeding modern cultivars of cereal crops short plants with stiff straw and high harvest index. It is also obvious that cultivars possessing the two characteristics considered important by farmers in dry areas will not be suited for cultivation in high-yielding environments because of their lodging susceptibility a further indication of the importance of specic adaptation. There is also evidence that, when breeders and farmers select in the same environment, farmers selection can be effective, implying that farmers possess considerable knowledge which is almost totally neglected in formal plant breeding programmes. Data from Rwanda illustrate the results of the participation of farmers in bush bean (Phaseolus vulgaris). Even though the selection was conducted on-station, the farmers were able to select a number of cultivars which kept their superiority when evaluated on-farm, while the professional breeders failed to do so. A participatory barley-breeding research project started in Syria in 1997 with 208 barley lines and populations planted in nine farmers elds and on two research stations. Selection was conducted by the farmers and a breeder both in farmers elds and on-station. This has shown that farmers can be effective in identifying high-yielding entries both in their elds (Table 6.6) and on the research stations. The lines selected by the farmers in six out of the nine locations had a signicantly higher average grain yield than the population mean (Table 6.6); the breeders selections were also signicantly higher yielding than the population mean in six out of the nine farmers elds and on the two research stations. The lines selected by the farmers and by the breeder in the nine locations never differed signicantly in grain yield with the exception of one village (Al Bab), where the farmers selections yielded signicantly more grain and more biomass than the breeders selections. On the research stations (Tel Hadya and Breda), we used the frequency of the 20 top-yielding lines included among the selected lines as a measure of selection efciency. This frequency was 14.3% and 16.4% among the breeders selections in Tel Hadya and Breda, respectively. Among the farmers selections in Tel Hadya, in only one case was

128

S. Ceccarelli et al.

Table 6.6. Grain yield (kg ha1) and total biological yield (kg ha1) of the lines selected by the farmers and by the breeder in nine farmers elds in Syria. Grain yield Location (code) Ibbin (01) Ebla (02) Tel Brak (03) Jurn El-Aswad (04) Baylonan (05) Al Bab (06) Melabya (07) Bari Sharki (08) Sauran (09) Farmer 4,615*** 3,498* 4,235 2,049* 454* 649*** 915 1,366* 2,561 Breeder 3,971*** 3,199** 4,020* 1,724** 324 488*** 920*** 1,129 2,654 a n.s. n.s. n.s. n.s. n.s. *** n.s. n.s. n.s. Farmer 10,687** 8,743 8,729* 10,535** 3,198 2,272*** 4,127** 5,276 6,796 Biological yield Breeder 9,686*** 8,233 8,036 8,429* 2,816 1,787*** 3,246* 4,708 7,257 a n.s. n.s. n.s. n.s. n.s. *** n.s. n.s. n.s.

Source: Ceccarelli et al. (2000). n.s. = non-signicant; * P <0.05; **P <0.01; ***P <0.001 relative to the comparisons with the population mean. aComparison between breeders and farmers selections based on t-test for samples of unequal size; ***P <0.001 relative to the comparisons between farmer and breeder selections.

the frequency of the highest-yielding lines higher than that of the breeder, while in Breda the frequency of the highest-yielding lines among the farmers selections was higher than that of the breeder in seven out of nine cases; in Breda the differences between entries were much more obvious at the time the selection was made. One of the hypotheses tested with these experiments on participatory plant breeding was that farmers selection will contribute more to maintain and/or enhance diversity than breeder selection does. In the case of the experiment in Syria, the similarity of selection done in different environments by the breeder and by the individual farmers was evaluated with the similarity analysis based on the dice coefcient (Czekanowski, 1917). Figure 6.4 shows the dendogram of the nine farmers and of the breeder based on cluster analysis of their selections in the nine farmers elds. There was a high degree of similarity among the selections done by the breeder in the various farmers elds, with dice coefcients always higher than 0.57, and a much lower degree of similarity among the selections done by the farmers in their elds, with dice coefcient always lower than 0.33. Also there was little similarity between the selections done by the breeder and the farmer in the same eld. The large dissimilarity between the lines selected by the farmers in different locations should result in a higher diversity in the breeding material after one cycle of selection by the farmers than after one cycle of selection by the breeder. These data show that decentralized breeding can give different results depending on whether it is participatory or not. Decentralized selection is a powerful way of maintaining a high level of diversity. Table 6.7 shows the number and the percentage of different entries selected in all four possible combinations of centralized versus decentralized and of participatory versus non-participatory selection. Participation nearly doubles the percentage of lines retained after one cycle of centralized selection (from 34% to 61%), probably as a consequence of more people being involved in participatory than non-participatory selection. Decentralization further increases the percentages of lines retained after one cycle of selection to about three-quarters, and with minor differences due to participation.

Plant Breeding for Marginal Environments

129

0.0

0.2

0.4

0.6

0.8 BREED01 BREED02 BREED03 BREED09 BREED04 BREED06 BREED08 BREED05 BREED07 FA 03 FA 06 FA 08 FA 09 FA 05 FA 04 FA 01 FA 02 FA 07

Fig. 6.4. Dendogram of nine farmers and of a breeder based on cluster analysis of their selections in farmers elds (FA = farmer, BREED = breeder). Individual locations are indicated with the location code used in Table 6.6. Table 6.7. Number and percentage of entries selected after one cycle of different types of selection from a population of 208 barley entries. Type of selectiona Centralized, non-participatory Centralized, participatory Decentralized, non-participatory Decentralized, participatory No. of selected entries 70 127 164 154 % of selected entries 33.7 61.1 78.8 74.0

Source: Ceccarelli et al. (2000). aParticipatory selection was done by nine farmers, while non-participatory selection was done by one breeder.

These data indicate not only that farmers are expert, but that they predominantly select for specic adaptation. In contrast with breeders, who usually search for a small number of widely adapted cultivars, farmers in different areas need different cultivars to meet specic requirements, either in relation to crop utilization or to the particular environment of their farms. Decentralized, participatory selection intended as selection done by farmers in their elds and in the early stages of selection inevitably leads to the selection of a large number of different cultivars, with obvious advantages for the maintenance of genetic diversity within a crop. However, what has been described so far, not only in this chapter but also in most of the literature on farmer participation, is not participatory plant breeding, but an experiment on participatory plant breeding that lacks the cyclical nature of plant breeding. This means that a truly participatory breeding programme is very complex, with farmers growing different cycles of selection, until the identication and adoption of

130

S. Ceccarelli et al.

new cultivars. These are then utilized by the formal breeding programme in further cycles of recombination, and by the farmers in further cycles of selection. The seed production of the new cultivars can be done either by the formal seed system or by the farmers themselves (informal seed system) depending on the crop and the infrastructures. The entire process may require too much land to be hosted by individual farmers at each location. The important role of the selection environment, the interest shown by neighbouring farmers in visiting the plots, and the theoretical advantage of testing in more locations with less replications in each location suggest that community participation (Ashby et al., 1996), intended as participation of several farmers in a given geographical area, rather than individual participation, could be a more realistic and feasible way to implement a decentralizedparticipatory plant breeding programme. This not only has the advantage of a much wider awareness among the farmers of the newest genetic material available, but it allows the use of different experimental designs and plot sizes that is usual in formal breeding programmes. A schematic example of a decentralizedparticipatory breeding programme is shown in Fig. 6.5 for one target environment. The same scheme can be applied to any number of target environments. The rst level of selection (new breeding material generated by the formal breeding programme) could be grown each year, and used for selection, by one farmer (level 1 farmers) in each of n areas representative of the target population of environments. The breeding lines selected in the rst cycle of selection could then be grown, within the same area, by a number of farmers (between ve and ten) neighbouring the level 1 farmer. These will be the level 2 farmers and will probably grow between ten and 20

Formal breeding programme generates variability

Formal/informal seed systems

Fig. 6.5. Decentralizedparticipatory plant breeding. The gure is a simplication of the text.
indicates the level 1 farmers, indicates level 2 farmers, and indicates level 3 farmers.

Plant Breeding for Marginal Environments

131

new breeding lines each year. At this level of selection, all lines could be replicated at each location, or different farmers could host either different replications or different incomplete blocks (if an incomplete block design is used). At the third level, the lines selected by each of the level 2 farmers could be grown in larger plots by ve to ten farmers neighbouring the level 2 farmers. As for the level 2 selection, there is a wide choice of experimental designs and eld layout at this level. Adoption and seed multiplication of superior bulks can begin at this stage (only 5 years after the crosses are made), but at the same time the formal breeding programme can purify the superior bulks (by single seed descent or double-haploids) in those cases (crops and/or countries) where uniform cultivars are needed by farmers and for future use as parental material in the crossing programme. The formal breeding programme will continue to generate the new breeding material distributed to the level 1 farmers, produce seed for the trials at the different levels, and recycle the best populations and/or lines from level 3 selection both for crossing and for re-selection. Therefore, the new breeding lines distributed each year to level 1 farmers will be both early segregating populations and pure lines selected from the more successful bulk identied in the previous 3-year selection period.

Conclusions
Decentralized breeding, intended as decentralized selection and testing in the target environment by the users of the future cultivars using the genetic variability created by a formal breeding programme, is a powerful means of maintaining genetic diversity. This is due to concomitant effects of differences between selection environments and between different users which result in the selection of different breeding material in different selection sites. In addition to the most obvious type of genetic diversity, i.e. the presence of different cultivars of the same crop grown in the same country or geographical area at the same time, decentralized selection between early segregating populations done directly by farmers may lead to the adoption and cultivation of bulks, i.e. of genetically heterogeneous cultivars similar to the landraces which are currently grown in many marginal environments. Because marginal environments tend to be more heterogeneous than favourable environments (Ceccarelli et al., 2000), the effect of decentralized selection on genetic variability is likely to be larger in marginal environments. However, decentralized selection might also be useful in favourable environments to obtain a better adaptation of crops to the physical environment and therefore to rely less on the use of inputs. In addition to the maintenance of different types of genetic variability, decentralized selection in international agricultural research centres is a powerful means of creating true partnership with the scientists in the national programmes. In a decentralized breeding programme the role of national programme scientists changes from merely testing material selected elsewhere, to selecting from within segregating populations derived from crosses that they decided. In decentralized and participatory breeding, farmers can play a similar role. Eventually, and because decentralized breeding based on decentralized selection is a highly dynamic process, it is also an efcient way to continuously adapt the breeding material to a continuously changing agricultural environment.

132

S. Ceccarelli et al.

Acknowledgements
The authors thank the OPEC Fund for International Development for supporting the decentralized barley breeding in North Africa, the Government of The Netherlands for supporting the barley breeding programme in Ethiopia, the Government of Italy for supporting the barley breeding programme at ICARDA, Der Bundesminister fr Wirtschaftliche Zusammenarbeit (BMZ) for supporting farmer participation in Syria, and the International Development Research Centre (IDRC) for supporting farmer participation in Tunisia and Morocco. The authors thank Dr R.H. Booth, former Assistant Director General for International Cooperation at ICARDA, for his strong support during the rst stages of decentralized breeding.

References
Allen, F.L., Comstock R.E. and Rasmusson D.C. (1978) Optimal environments for yield testing. Crop Science 18, 747751. Almekinders, C.J.M., Louwaars, N.P. and de Bruijn, G.H. (1994) Local seed systems and their importance for an improved seed supply in developing countries. Euphytica 78, 207216. Alza, J.O. and Fernandez-Martinez, J.M. (1997) Genetic analysis of yield and related traits in sunower (Helianthus annuus L.) in dryland and irrigated environments. Euphytica 95, 243251. Arboleda-Rivera, F. and Compton, W.A. (1974) Differential response of maize (Zea mays L.) to mass selection in diverse selection environments. Theoretical and Applied Genetics 44, 7781. Atlin, G.N. and Frey, K.J. (1989) Predicting the relative effectiveness of direct versus indirect selection for oat yield in three types of stress environments. Euphytica 44, 137142. Atlin, G.N. and Frey, K.J. (1990) Selecting oat lines for yield in low-productivity environments. Crop Science 30, 556561. Ashby, J.A., Gracia, T., Guerrero, M., Quirs, C.A., Roa, J.I. and Beltrn, J.A. (1996) Innovation in the organization of participatory plant breeding. In: Eyzaguirre, P. and Iwanaga, M. (eds) Participatory Plant Breeding. Proceeding of a Workshop on Participatory Plant Breeding, 2629 July 1995, Wageningen, The Netherlands. IPGRI, Italy, pp. 7797. Baker, R.J. (1988) Tests for crossover genotypeenvironmental interactions. Canadian Journal of Plant Science 68, 405410. Bnziger, M., Betrn, F.J. and Latte, H.R. (1997) Efciency of high-nitrogen selection environments for improving maize for low-nitrogen target environments. Crop Science 37, 11031109. Lakew, B., Semeane, Y., Alemayehu, F., Gebre, H., Grando, S., van Leur, J.A.G. and Ceccarelli, S. (1997) Exploiting the diversity of barley landraces in Ethiopia. Genetic Resources and Crop Evolution 44(2), 109116. Blum, A. (1988) Plant Breeding for Stress Environments. CRC Press, Boca Raton, Florida. Blum, A, (1993) Selection for sustained production in water-decit environments. In: International Crop Science I. Crop Science Society of America, Madison, Wisconsin, pp. 343347. Blum, A. and Pnuel, Y. (1990) Physiological attributes associated with drought resistance of wheat cultivars in a Mediterranean environment. Australian Journal of Agricultural Research 41, 799810. Breese, E.L. (1969) The measurement and signicance of genotypeenvironment interactions in grasses. Heredity 24, 2744.

Plant Breeding for Marginal Environments

133

Byerlee, D. (1994) Modern Varieties, Productivity, and Sustainability: Recent Experience and Emerging Challenges. CIMMYT, Mexico, D.F. Byerlee, D. and Husain, T. (1993) Agricultural research strategies for favoured and marginal areas, the experience of farming system research in Pakistan. Experimental Agriculture 29, 155171. Byrne, P.F., Bolaos, J., Edmeades, G.O. and Eaton, D.L. (1995) Gains from selection under drought versus multilocation testing in related tropical maize populations. Crop Science 35, 6369. Calhoun, D.S., Gebeyehu, G., Miranda, A., Rajaram, S. and van Ginkel, M. (1994) Choosing evaluation environments to increase wheat grain yield under drought conditions. Crop Science 34, 673678. Ceccarelli, S. (1989) Wide adaptation. How wide? Euphytica 40, 197205. Ceccarelli, S. (1994) Specic adaptation and breeding for marginal conditions. Euphytica 77, 205219. Ceccarelli, S. and Grando, S. (1991) Environment of selection and type of germplasm in barley breeding for stress conditions. Euphytica 57, 207219. Ceccarelli, S. and Grando, S. (1996) Drought as a challenge for the plant breeder. Plant Growth Regulation 20, 149155. Ceccarelli, S., Grando, S. and Hamblin, J. (1992) Relationships between barley grain yield measured in low and high yielding environments. Euphytica 64, 4958. Ceccarelli, S, Erskine, W, Grando, S. and Hamblin, J. (1994) Genotype environment interaction and international breeding programmes. Experimental Agriculture 30, 177187. Ceccarelli, S., Grando, S. and van Leur, J.A.G. (1995) Barley landraces of the Fertile Crescent offer new breeding options for stress environments. Diversity 11, 112113. Ceccarelli, S., Grando, S. and Booth, R.H. (1996) International breeding programmes and resource-poor farmers: crop improvement in difcult environments. In: Eyzaguirre, P. and Iwanaga, M. (eds) Participatory Plant Breeding. Proceeding of a Workshop on Participatory Plant Breeding, 2629 July 1995, Wageningen, The Netherlands. IPGRI, Italy, pp. 99116. Ceccarelli, S., Bailey, E., Grando, S. and Tutwiler, R., (1997) Decentralized, participatory plant breeding: a link between formal plant breeding and small farmers. In: New Frontiers in Participatory Research and Gender Analysis. Proceedings of an International Seminar on Participatory Research and Gender Analysis for Technology Development. CIAT, Cali, Colombia, pp. 6574. Ceccarelli, S., Grando, S. and Impiglia, A. (1998) Choice of selection strategy in breeding barley for stress environments. Euphytica 103, 307318. Ceccarelli, S., Grando, S., Tutwiler, R., Baha, J., Martini, A.M., Salahieh, H., Goodchild, A. and Michael, M. (2000) A methodological study on participatory barley breeding. I. Selection phase. Euphytica 111, 81104. Ceccarelli, S., Grando, S., Bailey, E., Amri, A., El-Felah, M., Nassif, F., Rezgui, S. and Yahyaoui, A. (2000) Participatory barley breeding in Syria, Morocco and Tunisia. Euphytica (in press). Chambers, R. and Ghildyal, B.P. (1985) Agricultural research for resource poor farmers: the farmer-rst-and-last model. Agricultural Administration 20, 130. Chisi, M., Bramel-Cox, P.J., Witt, M.D., Claassen, M.M. and Andrews, D.J. (1996) Breeding for grain yield stability using full-sib family recurrent selection in sorghum. Crop Science 36, 10831087. Cooper, M. and DeLacy, I.H. (1994) Relationships among analytical methods used to study genotypic variation and genotype-by-environment interaction in plant breeding multi-environment experiments. Theoretical and Applied Genetics 88, 561572. Cooper, M., Brennan, P.S. and Sheppard, J.A. (1996) A strategy for yield improvement of wheat which accommodates large genotype by environment interactions. In: Cooper, M. and Hammers, G.L. (eds) Plant Adaptation and Crop Improvement CAB International, Wallingford; ICRISAT, Andra Pradesh, India; IRRI, Manila, Philippines, pp. 487511.

134

S. Ceccarelli et al. Cooper, M., Stucker, R.E., DeLacy, I.H. and Harch, B.D. (1997) Wheat breeding nurseries, target environments, and indirect selection for grain yield. Crop Science 37, 11681176. Czekanowski, J. (1917) Zarys method statystycznyck. Warsaw. Eyzaguirre, P. and Iwanaga, M. (1996) Farmers contribution to maintaining genetic diversity in crops and its role within the total genetic resources system. In: Eyzaguirre, P. and Iwanaga, M. (eds) Participatory Plant Breeding. Proceeding of a Workshop on Participatory Plant Breeding, 2629 July 1995, Wageningen, The Netherlands. IPGRI, Italy, pp. 918. Farrington, J. (1996) Farmers participation in agricultural research and extension: lessons from the last decade. TAA Newsletter June 1996, 910,15. Frey, K.J. (1964) Adaptation reaction of oat strains selected under stress and non-stress environmental conditions. Crop Science 4, 5558. Gauch, H.G. Jr and Zobel, R.W. (1997) Identifying mega-environments and targeting genotypes. Crop Science 37, 311326. Grisley, W. (1993) Seed for bean production in sub-Saharan Africa, issues, problems, and possible solutions. Agricultural Systems 43, 1933. Hardon, J.J. and de Boef, W.S. (1993) Linking farmers and breeders in local crop development. In: de Boef, W., Amanor, K., Wellard, K. and Bebbington, A. (eds) Cultivating Knowledge. Genetic Diversity, Farmer Experimentation and Crop Research. Intermediate Technology Publications, London, pp. 6471. Hildebrand, P.E. (1984) Modied stability analysis of farmer managed, on-farm trials. Agronomy Journal 76, 271274. Hildebrand, P.E. (1990) Modied stability analysis and on-farm research to breed specic adaptability for ecological diversity. In: Kang, M.S. (ed.) Genotype-by-Environment Interaction and Plant Breeding. Dept of Agronomy, Louisiana Agricultural Experimental Station, Baton Rouge, Florida, pp. 169180 Hsiao, T.C. (1982) The soilplantatmosphere continuum in relation to drought and crop production. In: Drought Resistance in Crops with Emphasis on Rice. International Rice Research Institute, Los Baos, Philippines, pp. 3952. Jackson, P.A., Byth, D.E., Johnston, R.P. and Fischer, K.S. (1993) Genotype environment interactions in progeny from a barley cross. 1. Patterns of response among progeny lines for grain yield and time to anthesis. Australian Journal of Experimental Agriculture 33, 619627. Jansen, H.G.P., Walker, T.S. and Barker, R. (1990) Adoption ceilings and modern coarse cereal cultivars in India. American Journal of Agricultural Economics 72, 653663. Johnson, G.R. and Frey, K.J. (1967) Heritabilities of quantitative attributes of oats (Avena sp.) at varying levels of environmental stress. Crop Science 7, 4346. Latte, H.R. and Edmeades, G.O. (1994) Improvement for tolerance to low soil nitrogen in tropical maize. I. Selection criteria. Field Crops Research 39, 114. Lawn, R.J. (1988) Breeding for improved plant performance in drought-prone environments. In: Bidinger, F.R. and Johansen, C. (eds) Drought Research Priorities for the Dryland Tropics. ICRISAT, Patancheru, India, pp. 213219 Lofer, C.M., Salaberry, M.T. and Maggio J.C. (1986) Stability and genetic improvement of maize yield in Argentina. Euphytica 35, 449458. Martin, G.B. and Adams, M.W. (1987) Landraces of Phaseolus vulgaris (Fabaceae) in Northern Malawi. II. Generation and maintenance of variability. Economic Botany 41, 204215. Maurya, D.M., Bottral, A. and Farrington, J. (1988) Improved livelihoods, genetic diversity and farmer participation: a strategy for breeding in rainfed areas of India. Experimental Agriculture 24, 311320. Muoz, P., Voltas, J., Araus, J.L. and Romagosa, I. (1998) Changes in adaptation of two- and six-rowed barley releases over time in Northern Spain. Plant Breeding 117, 531535. Muruli, B.I. and Paulsen, G.M. (1981) Improvement of nitrogen use efciency and its relationship to other traits in maize. Maydica 26, 6373.

Plant Breeding for Marginal Environments

135

Passioura, J.B. (1986) Resistance to drought and salinity: avenues for improvement. Australian Journal of Plant Physiology 13, 191201. Pederson, D.G and Rathjen, A.J. (1981) Choosing trial sites to maximize selection response for grain yield in spring wheat. Australian Journal of Agricultural Research 32, 411424. Pfeiffer, W.H. (1988) Drought tolerance in bread wheat analysis of yield improvement over the years in CIMMYT germplasm. In: Klatt, A.R. (ed.) Wheat Production Constraints in Tropical Environments. CIMMYT, Mexico D.F., pp. 274284. Rhoades, R. and Booth, R. (1982) Farmer-back-to-farmer: a model for generating acceptable agricultural technology. Agricultural Administration 11, 127137. Roy, N.M. and Murty, B.R. (1970) A selection procedure in wheat for stress environments. Euphytica 19, 509521. Schneider, K.A., Rosales-Serna, R., Ibarra-Perez, F., Casares-Enriques, B., Acosta-Gallegos, J.A., Ramirez-Vallejo, P., Wassimi, N. and Kelly, J.D. (1997) Improving common bean performance under drought stress. Crop Science 37, 4350. Selmani, A. and Wassom, C.E. (1993) Daytime chlorophyll uorescence measurement in eld grown maize and its genetic variability under well-watered and water-stressed conditions. Field Crops Research 31, 173184. Shannon, M.C. and Francois, L.E. (1978) Salt tolerance of three muskmelon cultivars. Journal of the American Society for Horticultural Science 103, 127130. Simmonds, N.W. (1962) Variability in crop plants, its use and conservation. Biological Reviews 37, 422465. Simmonds, N.W. (1984a) Principles of Crop Improvement. Longman, London. Simmonds, N.W. (1984b) Decentralized selection. Sugar Cane 6, 810. Simmonds, N.W. (1991) Selection for local adaptation in a plant breeding programme. Theoretical and Applied Genetics 82, 363367. Simmonds, N.W. and Talbot, M. (1992) Analysis of on-farm rice yield data from India. Experimental Agriculture 28, 325329. Singh, M. and Ceccarelli, S. (1995) Estimation of heritability using varietal trials data from incomplete blocks. Theoretical and Applied Genetics 90, 142145. Sperling, L., Loevinsohn, M.E. and Ntabomvura, B. (1993) Rethinking the farmers role in plant breeding, local bean experts and on-station selection in Rwanda. Experimental Agriculture 29, 509519. Stroup, W.W., Hildebrand, P.E. and Francis, C.A. (1993) Farmer participation for more effective research in sustainable agriculture. In: Ragland, J. and Lal, R. (eds) Technologies for Sustainable Agriculture in the Tropics. American Society of Agronomy Special Publication 56 ASA, CSSA, and Soil Science Society of America, Madison, Wisconsin, pp. 153186. Trutmann, P. (1996) Participatory diagnosis as an essential part of participatory breeding: a plant protection perspective. In: Eyzaguirre, P. and Iwanaga, M. (eds) Participatory Plant Breeding. Proceeding of a Workshop on Participatory Plant Breeding, 2629 July 1995, Wageningen, The Netherlands. IPGRI, Italy, pp. 3743. Ud-Din, N., Carrer, B.F. and Clutter, A.C. (1992) Genetic analysis and selection for wheat yield in drought-stressed and irrigated environments. Euphytica 62, 8996. Virk, D.S. and Mangat, B.K. (1991) Detection of crossover genotype by environment interactions in pearl millet. Euphytica 52, 193199. Weltzien, E. and Fischbeck, G. (1990) Performance and variability of local barley landraces in Near-Eastern environments. Plant Breeding 104, 5867. Yitbarek, S., Berhane, L., Fikadu, A., van Leur, J.A.G., Grando, S. and Ceccarelli, S. (1998) Variation in Ethiopian barley landrace populations for resistance to barley leaf scald and netblotch. Plant Breeding 117, 419423. Zavala-Garcia, F., Bramel-Cox, P.J., Eastin, J.D., Witt, M.D. and Andrews, D.J. (1992) Increasing the efciency of crop selection for unpredictable environments. Crop Science 32, 5157.

Empowering Farmers and Broadening the Genetic Base: Agricultural Research and Resource Management
R. SALAZAR
SEARICE, 83 Madasalin St., Sikatuna Village, Quezon City, Philippines

Introduction
Farmers struggle for access and control of land is inviolable. Genuine agrarian reform remains the basis for rural and national development. However, the increasing dominance of technology in agricultural production requires that rural communities also need to act on the issue of access and control in the development of agricultural technologies. This concern is becoming decisive as technological development makes information and its higher form, knowledge a major economic resource, or factor of production. Power and wealth now, and more so in the future, will be determined by those who own and control information and knowledge. This covers all economic sectors, including agriculture. Intellectual innovation in agriculture has been in the hands of farmers for millennia. It was only in the last 30 years of the 20th century in developing countries, and from 1900 onwards in industrial countries, that agricultural research was transferred from farmers elds into the laboratories of the formal and centralized research institutions in the public and private sectors (Berg et al., 1991; Fowler, 1994). For farmers, these developments can constitute a disempowering process: they can become mere objects of research, passive receivers of the nished products of research institutions, and, being tied thus, have changed into labourers on their own lands (Lewontin, 1998). They have become end-users or consumers, rather than active creators. This contributes to the mystication of agricultural science, which becomes something esoteric, beyond the reach of farming communities, beyond understanding and therefore powerful. As usual, language reinforces this disempowerment. Farmers knowledge is considered backward while products of institutional research are called modern, no matter if they are inappropriate. Farmers varieties are called primitive cultivars; those of formal research institutions are advanced materials. Since formal agricultural research is not controlled by the farmers, they therefore cannot ensure that the research questions are relevant to their needs. Public research institutions and well-meaning scientists can only approximate farmers needs and
IPGRI/FAO 2001. Broadening the Genetic Base of Crop Production (eds H.D. Cooper, C. Spillane and T. Hodgkin)

137

138

R. Salazar

objectives. Commercial research institutions are straightforward: their research is for prot. These corporations will try to t their research to the needs and objectives of their target farmer market that is simple business sense but in the end both these research approaches are top-down. Most top-down approaches end up mainly beneting those at the top. Most importantly, top-down research centralizes power at the top. Understanding the negative implications of centralized and formal research institutions does not mean that these institutions are not important. In fact, this chapter will argue that these institutions will continue to have important and distinct roles: in particular, in ensuring a supply of genetic materials and other public goods. These institutions become counterproductive only when they attempt to disregard the role of farmers research and try to replace rather than strengthen farmers intellectual innovation. The most decisive areas in agricultural innovation are related to the use of plant genetic resources through plant breeding, seed distribution and, now, biotechnology. Seed is the most important key to agricultural production: it is therefore a major arena of battle for control of the different sectors and interests involved in agriculture, from the small farmers in the South to transnational corporations mostly based in the North (Kloppenburg, 1988). Farmers traditional innovation created the crops the world depends on today. The great diversity of these crops and the diversity within crop species are products of the different agroecological conditions where these crops are grown, and are results of the intellectual innovation of farmers all over the world (Fowler and Mooney, 1990). These traditional systems of plant genetic resource (PGR) conservation and improvement largely remain today: for example, almost all of the upland rice varieties cultivated in Asia today are farmers varieties. The general rule is that the varieties used by farmers under less favourable conditions are their own varieties. For so-called minor crops or crops of local importance, such as traditional vegetables and small grain cereals, almost all of the varieties are farmer varieties. Practically all taro, yams and sweet potato varieties in the Philippines and throughout Southeast Asia are farmer varieties. New and diverse sweet potato varieties are periodically created by Filipino farmers, not by formal research institutions. More generally, small-scale farming is a conspicuous feature of agriculture today and is likely to remain so (Whitten and Settle, 1998). There are about 4.5 million small-scale farmers worldwide, including about 2 million who are using mainly traditional technologies (Pretty, 1995). They have the potential to play a major role in maintaining a broad genetic base in production systems.

The Asian Irrigated Rice Farming System

The irrigated rice farming system is strongly linked to the market economy; rice is grown not only for subsistence but mostly for cash income. The year long availability of water through irrigation canals radically altered the ecosystem that now allows for at least two agricultural seasons. Farmers varieties that were bred for subsistence systems and for one agricultural season per year were easily replaced by modern cultivars. The initial results of the use of modern rice high-yielding varieties (HYVs), produced by the International Rice Research Institute (IRRI) in the Green Revolution,

Empowering Farmers and Broadening the Genetic Base

139

and the package of technologies that came with it was quite positive, or even dramatic. Yield increases of 3070% were reached. Countries predicted to suffer serious food shortages and even famine, such as India, reversed the trend and produced enough food. While this chapter will provide criticism on some of the policies that promoted the use of rice HYVs, in the shorter term it is also true that farmers voted for it with their farms. In fact, it was only a matter of time before limits to the sustainability of the approach became apparent. The limited number of varieties that can be produced by research institutions narrowed the genetic base of irrigated rice and made it extremely susceptible to pests and diseases. The tungro virus wiped out rice production in Asia in 1972 and 1973, and the virus continues to be the most devastating disease today. The brown plant hoppers reduced the harvest of Vietnam by 20% in 1992. Rice blast fungus destroys wide areas on a scale never known before genetically narrow rice HYVs were widely used. Pests and diseases are able to evolve around the vertical resistance developed by centralized breeding centres (Gallagher, 1988); the massive use of chemicals only accelerated this process (Gallagher et al., 1994; Heinrichs et al., 1994; Settle et al., 1996). Even more alarming, the breeding programmes of formal research institutions are often losing the race against biotic stresses (Robertson, 1996). Increasingly, disease resistance and other desirable traits bred into the rice plant break down sooner than the successful breeding of new varieties, so farmers in irrigated areas all over the region are clamouring for more and newer varieties. For example, farmers in central and northern Vietnam, during a random survey conducted in preparation for the Biodiversity Use and Conservation in Asia Programme (BUCAP), unanimously complained that the modern cultivars they cultivate are fast deteriorating and that they are not receiving enough varieties from their research institutions.

Working with farmers in the Philippines The Philippines had around 3500 rice cultivars before the use of Green Revolution seeds and technologies. Traditionally, each farmer family would cultivate two to ve rice varieties per season and there could be 50100 rice varieties in a 5000 ha area. Today, ve to seven related varieties can cover several thousands of hectares. At one point, almost 50% of the rice farms were planted to IR36. In the beginning, productivity also increased by at least 30%; however, as noted above, problems related to the narrowing of the genetic base of the rice varieties, most of which share a common pedigree, surfaced. Farmers started to select and evolve new varieties to respond to their needs and to adapt the varieties to the conditions of their farms. In the island province of Bohol in the Philippines, all rice varieties released by the government soon developed red grain types as the people of the island prefer red rices. Thus, there are red 36, red 64, red 66, etc., that traced their origin from IRRI releases as well as local germplasm. All over the country, rice farmers and the environmental conditions of their farms selected and evolved new varieties from what they received. In 1991 in the province of North Cotabato in southern Philippines, 30% of the irrigated rice lands were planted to a

140

R. Salazar

farmer variety called bordagol. Bordagol was preferred because it was resistant to the tungro virus, and required less fertilizer. It also had superior eating quality and fetched a higher price in the market. In the same province, around ten farmer varieties less popular than bordagol were cultivated by the farmers for other traits that bordagol and the rice HYVs did not have (Salazar, 1992a). The same developments can be observed in all irrigated rice areas in Southeast Asia. Rice farmers have been breeding their own varieties using both modern introduced cultivars and traditional cultivars as raw materials. The South East Asia Regional Institute for Community Education (SEARICE), a Philippines-based non-governmental organization (NGO), has been promoting farmer-management of plant genetic resources. Its work is mainly focused on the irrigated prime rice lands where diversity has been lost and where the knowledge system of local farmers has been displaced. The objectives were to recover diversity and strengthen the role of local farmers as PGR managers and researchers. SEARICE selected a town in North Cotabato province mentioned above with a contiguous irrigated rice area of around 3500 ha in 1993. SEARICE also set up an NGO called CONSERVE to assist the local farmers. More than ten varieties (collected outside the town and from nearby provinces) were distributed to the farmers for observation and adaptation. In 1995, the project CONSERVE assisted the distribution of 18 bulk-bred populations from an alternative breeding programme called MASIPAG. Eight new lines (F7) bred at CONSERVE were also distributed. MASIPAG or FarmerScientist Partnership for Development Association, established in 1986, comprises 21 peoples organizations, ten NGOs, three national farmer movements and scientists from three educational institutions. The group collects local rice varieties and teaches farmers how to perform hybridization and subsequent seed selection themselves (Salazar, 1992b). New seeds coming out of this programme are rapidly spreading throughout the Philippines and over 10,000 farmers are estimated to be using such seeds (Zamora, 1993). At the same time, in the CONSERVE project, local farmers were trained in crosshybridization including the enhancement of their traditional selection techniques. Out of more than 200 farmers trained in hybridization and selection, only four became effective and committed in continuing hybridization work and working on highly segregating materials, and 31 families continue to be involved in intensive selection work. As of the rst season of 1998, 60% of the 3500 ha irrigated area was planted to varieties that were adapted and selected by local farmers. The bordagol variety, which had deteriorated, was planted to less than 3% of the rice lands. The rest of the area was planted to government-released varieties. The number of varieties in the whole area increased to 32 (although eight varieties, four from the government and four from the farmers, are the most dominant and popular varieties). At the same time, new lines with high potential are emerging from the breeding and selection efforts of those trained by the project and from farmer neighbours who learned from those who were trained. It is estimated that these new farmer varieties will spread in the community over the coming years. Farmers in the area are again playing a major role in plant genetic resource management, creating and conserving diversity. The project has started the process of broadening the genetic base of the rice crop in the irrigated ecosystem, increasing the number of varieties from a few modern cultivars to 32. More importantly, and intimately linked with broadening the genetic base,

Empowering Farmers and Broadening the Genetic Base

141

is the process of broadening the actors involved in plant genetic resource research and development. Broadening the actors also means empowering these actors at the sociopolitical and technical levels. It is also important to note shortcomings of our work at CONSERVE. The project initially had a strong conservation bias, as a reaction to the ongoing problem of genetic erosion. However, the conservation, for its own sake, of specic traditional cultivars in a production system that was increasingly linked to the market where the agroecosystem had been altered by year-round availability of water, and where farmers economic conditions, pressures and needs were rapidly changing proved not to be viable. Farmers conserve only by utilizing resources that respond to their needs. Thus, except for a few glutinous local rice varieties that were continually planted because they had distinct desirable traits, most of the varieties that the farms cultivated were either introduced from research institutions, or were developed by themselves or other local farmers from such introductions (and most possibly from their local varieties too). It is clear that the conservation of specic materials for the far future must be a strategic concern of institutions such as governments and international research centres and other international organizations. Small farmers can only conserve through active utilization and constant improvement of these materials. While long-term in situ conservation is necessary (for its distinct function) to complement the ex situ genebanks, this approach may require constant support and subsidy in so far as some of the materials may not be useful to small farmers as conditions change (see also Wood and Lenn, 1997). The second shortcoming was the lack of systematic pre-breeding or access to diverse materials to effectively provide the raw materials for farmers use when the project shifted to participatory plant breeding. While the farmers were enthusiastic, the materials they had in hand were very few. Furthermore, the sources of germplasm were limited to materials in the province, and to materials from other farmer organizations and NGOs in the country. The rich source of PGR diversity, the national and international research institutions, was not tapped.

Building from the farmer eld school approach of integrated pest management CONSERVE built from the successes and lessons of related efforts to assist local farmers. On sociological principles and approaches, the project beneted from a number of community organizing and consciousness raising experiences in the Philippines. The most important contribution at the technical level came from the experiences of the Food and Agriculture Organization (FAO) integrated pest management (IPM) programme. The FAO IPM programme, based on farmer eld schools (FFS), is one of the earliest efforts to merge scientic soundness with a research and education process that empowers farmers. FFS are held in the rice elds thoughout the season, from land preparation to harvest time. Every week, insect populations are monitored and the rice plant is assessed in all its major stages. The data are gathered and analysed by a community of farmers with simple research instruments developed by the IPM programme. This approach builds on local knowledge systems and relies on the farmers as researchers. The approach therefore induces farmers to learn. First, it avoids the banking approach to education where farmers are mere receivers of knowledge. Secondly,

142

R. Salazar

the approach is premised on praxis. Praxis is an actionreection method where knowledge is gained through experience (Habermas, 1987). Thirdly, FFS relies on collective approaches where experience and knowledge are gained by a community, not by a select few. Farmers can become experts in applying ecological principles to crop management; some become trainers or genuine scientic researchers (Dilts, 1999). The FFS approach does not idealize but is realistic about farmers capacities. Practical research tools and instruments are provided to: (i) help farmers plan their research objectives; (ii) gather data more systematically; and (iii) analyse the data. Additional data and information not normally reached by local communities but by research institutions are added. One of the results of the IPM programme is that pesticide use in the same area has been reduced by more than 50%, and the use of insecticides has almost been eliminated. This reduction is directly due to the enhanced level of biodiversity of natural enemies of plant pests and of other components of biodiversity, including the food sources of these natural enemies. These farmers may be considered, therefore, to be managing biodiversity in ecosystems in support of sustainable agriculture (Settle et al., 1996; Whitten and Settle, 1999). To date, over 1,000,000 Indonesian farmers have graduated from FFS: over 400,000 in Vietnam, and over 170,000 in the Philippines. The programme has been extended to several other Asian countries, and more recently to many countries in Africa and elsewhere. It has also been extended to other crops such as vegetables, maize and cotton. Work is consolidated through the development of community IPM programmes (Dilts, 1999). The IPM approach has been adapted to include plant genetic resource and sustainable soil management concerns. Using the farmer-eld-school participatory and experimental approach, the science of plant breeding is slowly being returned to farmers control, in the same manner that the IPM approach was returning the science of pest management to farming communities.

Lessons Learnt
The need for institutional support and access to germplasm The presence of an institution, in this case SEARICE and CONSERVE, is needed for technical assistance and links to formal-sector institutions. One of its most important functions was to provide the farmers with access to diverse genetic resources. Farmers traditional systems of access to genetic resources are too limited to be effective compared with that offered by formal research institutions. The other important role was to link and provide the local farmers with related technologies such as IPM, and selecting and breeding. The third role that must be mentioned is the assistance provided in the conduct of on-farm research with local farmers. In short, local farmers in their on-farm PGR research and development efforts also require institutional support. Pre-breeding, including the selection of parental lines, is a major supportive function requiring institutional support. While the analysis of the state of the plant genetic resources in the community can be undertaken by the local farmers as a basis for determining breeding objectives, the resources required are mostly beyond the means of the

Empowering Farmers and Broadening the Genetic Base

143

small farmers. While the local farmers can efciently evaluate their local materials, they will have little knowledge about plant genetic resources outside their area and those in the hands of national and international research institutions. As reported above, only four farmers continued with hybridization and in the handling of early generation materials, yet the number of farmers interested in later generation lines was signicant. As noted earlier, conservation, on-farm, needs to be linked to utilization if it is to be successful: it is not possible for farmers to maintain varieties that are no longer useful. In the end, the majority of the varieties introduced should also be deposited in the national rice genebank, in a black-box arrangement. More efforts to adapt formal research systems to farmers realities and capacities are needed. Replication of varietal trials in small plots is too laborious and maybe too complicated for farmers management. The number of breeding lines a farmer family can handle is low compared with the capacity of research institutions. Estimating varietal performance by using mathematical extrapolation (needed for small test plots) is less effective in demonstrating results compared with the conclusions farmers can reach observing a new variety cultivated at a production scale. These few examples highlight the need for more creative development of research approaches and instruments for onfarm participatory PGR management. Building from the local knowledge system, the need to strengthen this, using both traditional and institutional science, is an important task. There is a need to develop methodologies, research instruments and techniques that are appropriate to on-farm research by local farmers. Again, the model of the farmer eld schools of the IPM approach can be used as a guide. Under the FFS, simplied research instruments have been developed so that farmers are able to gather the data, and to analyse these data themselves. The research instruments for an ecological pest management approach under the IPM programme need to be developed further. Farmers practising on-farm PGR research and development require wider and free access to genetic resources. The pressures that bear on the lives of farmers families and the stresses faced by their farms are not only local but global as well. They need many sources of raw material to be able to respond and succeed. Farmers breeding efforts need to be formally recognized. For example, SEARICE and CONSERVE convinced the provincial government to honour and award the farmer who developed the bordagol variety. Community gatherings are used to recognize families that bred or are breeding new varieties. This recognition should include the qualication of farmers varieties to avail their developers with agricultural credit and insurance. Related to the issue of recognition is the issue of benet sharing when and if farmers varieties are used in successful commercial breeding. The absence of international and national agreements and legislation on this matter is creating a tension in farming communities. First, farmers traditionally exchanged seeds freely. Their traditional system depended on free exchange and it is imbedded in farming cultures. Yet developments today of proprietary ownership of genetic resources, and of national restrictions on the exchange of germplasm, threaten to prevent them from doing so. Both the farmers and agricultural development in general will lose out from this situation. We know that farmers traditional systems depended on free exchange of PGR, and that these resources move from village to village and great distances, and were continually improved. Thus seeds and the knowledge about them had always been

144

R. Salazar

exchanged freely, and this is part of the culture (or maybe ethos) of small farmers. While a single individual may be recognized, by naming a good variety after him or her, the seeds and the knowledge had always been collectively owned. The experience at CONSERVE was that as the information about the commercial privatization of seeds was discussed with farmers that seeds and the knowledge that goes with them have become a potential source of wealth there was some confusion. Many complained that it is unthinkable for them, as farmers, not to share seeds when they are asked to do so. The idea that these can be a source of wealth was not easy to handle either. A few of the farmers trained in participatory plant breeding had begun naming the varieties after themselves (which is not too problematic), and some did not want to share early generation materials and/or breeding lines, preferring instead to release only nished varieties (which could be a problem). We are thus faced with two opposing reactions. The rst of these shows that bringing the farmers system into the commercial sphere could undermine the system that created and maintained this diversity. The second reaction, although very insignicant at the moment, shows that some farmers will change and adapt to the commercial framework. We do not know at this stage whether or not farmers working within this framework will be able to continue to create and maintain diversity. There is a need to revise the seed certication systems of the country. Farmers varieties are frequently non-uniform populations, for obvious reasons. Farmers also like to select and evolve their varieties. This contradicts the traditional seed certication requirement of stability and uniformity. However, the multilocation trials required for nationwide adaptation do not always apply to farmer varieties that have more specic adaptation to microenvironments (see Louwaars, Chapter 5, this volume). There is a need to support the post-production requirements of farmers PGR management: the drying, milling, bagging and storage system that exists today in the country and in the region is designed for genetically uniform harvests. There is also a need for marketing development that will support the genetically diverse output of farmer breeding. This is a longer-term investment, and appropriate research and modelling are required. Many of these issues will be explored further in the Biodiversity Use and Conservation in Asia Programme (BUCAP)mentioned above. BUCAP will use the participatory research and development approach of the FFS and community IPM programmes to develop a model of farmer-led plant genetic resource research and breeding, which, inter alia, will help maintain genetic diversity on-farm, focusing on areas shifting from subsistence to market-based production; or, where the genetic base has been narrowed (for example in irrigated areas), attempt to restore genetic diversity.

Related sociopolitical concerns It is important to note that the area in North Cotabato province where this project was conducted is an area with strong farmers organizations. These organizations were born and grew in the farmers struggle for agrarian reform and against the Marcos government; village-level formal and informal organizations exist with leaders tested in the past. The highest form of this organization in the project area is the farmers cooperative, which has more than 2000 families as formal members.

Empowering Farmers and Broadening the Genetic Base

145

Reliance on their own strengths and capabilities and the capacity to be critical of top-down approaches and technologies is part of the farmers history and tradition. The decision to seek alternative agricultural technologies was already taken by their organizations even before SEARICE and CONSERVE arrived: the farmers in this town had sent petitions for the banning from the market of endosulfan pesticide compounds; they have signied their opposition to the genetic engineering of rice with Bt. The same is true of the farmers working on a similar project in the island province of Bohol: the initial effort was within the umbrella of a province-wide alliance of peasant organizations working for agrarian reform, for democratic rights, and for peasant concerns. One of the strongest PGR conservation and development efforts by local farmers of yams and sweet potato goes hand-in-hand with their opposition to the construction of an airport in their town. This chapter highlights these social and political bases and processes to emphasize the point that they are needed in on-farm PGR conservation and development catering to the interest of local and poor people. The presence of a strong farmers multipurpose cooperative, especially its credit programme, allowed farmers to use their own varieties. The farmers varieties qualify for the cooperatives production credit programme. Even before the project of SEARICE and CONSERVE, the farmers were widely planting the bordagol variety (developed by one of the members of the farmers cooperative) and a few other local varieties because it was encouraged by the cooperative. Moreover, the cooperative promoted and marketed bordagol, which was fetching a higher price because of its eating quality, creating more income for the farmers. In general, there is a need for an alternative social infrastructure that will serve as the power base for on-farm research. Without this, the research problems and directions will be determined by the richer sectors of the rural population, and scientists who are supposed only to assist will end up dominating the interaction.

References
Berg, T., Bjrnstad, ., Fowler, C. and Skrppa, T. (1991) Technology Options and the Gene Struggle. Development and Environment No. 8. NORAGRIC Occasional Papers Series C. Dilts, R. (1999) Facilitating the emergence of local institutions: reections from the experience of the Community IPM Programme in Indonesia. In: Asian Productivity Organisation Role of Institutions in Rural Community Development: Report of the APO Study Meeting in Colombo, 2129 September 1998. pp. 5065. Fowler, C. (1994) Unnatural Selection. Technology, Politics, and Plant Evolution. Gordon and Breach. Fowler, C. and Mooney, P. (1990) Shattering: Food, Politics, and the Loss of Genetic Diversity. The University of Arizona Press. Gallagher, K.W. (1988) Effects of host plant resistance on the microevolution of the rice brown planthopper, Nilaparvata lugers (Star) (Homoptera: Delphacidae). Dissertation, University of California, Berkeley, California. Gallagher, K.D., Kenmore, R.E. and Sogawa, K. (1994) Judicial use of insecticides deter planthopper outbreaks and extend the life of resistant varieties in southeast Asian rice. In: Denno, R.E. and Perfects, T.J. (eds) Planthoppers, Their Ecology and Management. Chapman & Hall, New York, pp. 599614.

146

R. Salazar Habermas, J. (1987) The Philosophical Discourse of Modernity. Oxford University Press, Oxford. Heinrichs, E.A. (1994) Impact of insecticides on the resistance and resurgence of rice planthoppers. In: Denno, R.F. and Perfect, T.J. (eds) Planthoppers, Their Ecology and Management. Chapman & Hall, New York, p. 799. Kloppenburg, J.R. (1988) First the Seed: the Political Economy of Plant Biotechnology. Cambridge University Press, Cambridge. Lewontin, R.C. (1998) The maturing of capitalist agriculture: farmer as proletarian. Monthly Review, 50(3), 7284. Pretty, J.N. (1995) Regenerating Agriculture: Policy and Practice for Sustainability and Self Reliance. Earthscan Publications, London. Robinson, R.A. (1996) Return to Resistance: Breeding Crops to Reduce Pesticide Dependence. AgAccess /IDRC. Salazar, R. (1992a) Community plant genetic resources management: experiences in Southeast Asia. In: Cooper, D., Vellv, R. and Hobbelink, H. (eds) Growing Diversity. Genetic Resources and Food Security. Intermediate Technology Publications. Salazar, R. (1992b) MASIPAG: alternative community rice-breeding in the Philippines. Appropriate Technology, 18(4) 2021. Settle, W.H., Ariawan, H., Tri Astuti, E., Cahyana, W., Hakim, A.L., Hindayana, D., Lestarei, A.S., Pajarningsih and Sartanto (1996) Managing tropical rice pests through conservation of generalist natural enemies and alternative prey. Ecology 77(7), 19751988. Whitten, M.J. and Settle, W.H. (1998) The role of the small-scale farmer in preserving the link between biodiversity and sustainable agriculture. In: Chou, C.H. (ed.) Frontiers in Biology: Biotechnology, Biodiversity and Sustainable Agriculture. Proceedings of the 25th AGM, International Union of Biological Sciences, Taiwan Province, 1723 November 1997. Wood, D. and Lenn, J.M. (1997) The conservation of agrobiodiversity on-farm: questioning the emerging paradigm. Biodiversity and Conservation 6, 109129. Zamora, O.B. (1993) Agricultural genetic resources management by the informal sector in Southeast Asia: challenges and needs. Paper presented at the Nordic Seminar on Growing Diversity Farmer Management of Agricultural Genetic Resources, Liding, Sweden.

The State of Millet Diversity and its Use in West Africa

O. NIANGADO
Foundation Novartis, BPE 1449, Bamako, Mali

Introduction
Africa is an important centre of diversity for several crops including millet, sorghum and rice as well as local crops such as fonio. The great variation in rainfall regimes and soil types has led to a wide range of production systems (Spencer and Sivakumar, 1987). Pearl millet, Pennisetum glaucum (L.), was domesticated in the Sahelian belt about 30004000 years ago. Portres (1976) suggested that it arose through multiple domestications from the wild progenitor, P. glaucum spp. monodii, and isoenzyme studies provide evidence to support this theory (Tostain and Marchais, 1989). African farmers have selected crop varieties adapted to a wide range of conditions; the large diversity of millet is the result of this work (Bono, 1973). Major movements between ethnic groups have also contributed to the growth of diversity. Pearl millet is the most widely cultivated cereal in Africa, south of the Sahara. In Western Africa, 17 countries grow pearl millet on 10.8 million ha (Mha), but ve countries account for 90% of the area: Niger (3.5 Mha), Nigeria (3.2 Mha), Burkina Faso (1.1 Mha), Mali (1.0 Mha) and Senegal (0.9 Mha). The other main area of production is South Asia, particularly India which accounts for 10.5 million of the 11.2 Mha total production in this region. In West Africa, production is in the hands of small farmers and yields are low and variable, rarely passing 1000 kg ha1. A number of constraints, primarily climatic and soil-related, are responsible for this poor performance. Various crop improvement programmes have been put in place in West Africa but, while signicant gains have been made with cash crops, very little progress has been achieved for subsistence crops. For millet, it is estimated that only 5% of the cultivated area is sown to improved varieties (Matlon, 1985). This contrasts with the situation in India where there have been major gains, with high adoption rates for hybrid varieties. However, it is important to note that in India varieties are adopted in relatively high production areas: soil and climatic factors are
IPGRI/FAO 2001. Broadening the Genetic Base of Crop Production (eds H.D. Cooper, C. Spillane and T. Hodgkin)

147

148

O. Niangado

much less favourable in West Africa than in India (Stoop et al., 1982; Matlon, 1985). This chapter describes the diversity available in the millet genepool, and the characteristics of millet varieties; surveys plant breeding programmes in West Africa and reviews their degree of success; describes the institutional arrangements for millet breeding in the sub-region at national and sub-regional levels; and, nally, draws some conclusions concerning the use of millet diversity in the region.

Millet Diversity
Millet genepools The genus Pennisetum consists of more than 140 species (Clayton, 1972). Stapf and Hubbard (1934) recognized 13 cultivated species, six annual wild relatives and 15 intermediate forms. The number of species was reduced by Clayton (1972) and Brunken et al. (1997). Though the name P. americanum was used by these authors, the name P. glaucum is preferred (de Wet, 1987). Thus, the primary genepool includes:

The cultivated form: P. glaucum, including the races Typhoides, Nigritarium, Globosum and Lonis, characterized on the basis of seed shape (Brunken et al., 1977). The wild form: P. glaucum ssp. monodii (= violaceum) with three races: Mollissium, Violaceum and Fallax. An intermediary, hybrid, form: P. glaucum ssp. stenostachyum, which possesses some features of the cultivated form, but, like the wild form, is characterized by the nonpersistence of the husks.

The secondary genepool includes those species that can be articially crossed with millet, such as elephant grass (P. purpureum), which is resistant to numerous diseases. The tertiary genepool includes other wild species such as P. orientale and P. villosum. Some of these species possess potentially useful characteristics such as apomixy, cold and drought resistance, and cytoplasmic diversity. Gene transfer is not possible using classical techniques (Hanna, 1987).

Collections The International Crops Research Institute for the Semi-Arid Tropics (ICRISAT) in collaboration with the International Plant Genetic Resources Institute (IPGRI) and national programmes has conducted a number of collecting missions in 15 countries of Africa (22 missions, 1654 accessions) and in India (14 missions, 2690 accessions). According to Kumar and Rao (1987), ICRISAT held over 17,000 accessions including 11,000 accessions of local varieties, 600 of wild relatives and 6000 from breeding programmes. A more recent estimate puts the ICRISAT collection at 21,000, representing 58% of the world collection. Other important collections are held by the USA, France, Canada and Uganda (FAO, 1998). Overall, African material constitutes more than half of the world collection, and is fairly well represented in it. Nevertheless gaps remain: priority areas for collecting as

Millet Diversity and Use in West Africa

149

identied by ICRISAT include Benin, Chad, Guinea-Bissau, Cte DIvoire and Mauritania (FAO, 1998).

Variability within the cultivated form P. glaucum shows great phenotypic variability for a number of characters (Bono, 1977; Grouzis, 1980; Marchais et al., 1993).

Length of the panicle: the greatest variation is found in Nigeria where it varies from 8 to 20 cm. The conical form is the most widespread. The vitrosity of the grain: this is very important in indicating organoleptic and conservation properties. Early varieties have higher vitrosity. Colour: dominant grain colours are yellow (especially in Niger) and grey. Except for Ghanaian and Togolaise varieties, the early types are yellow, and the later types from Mali and Senegal are more grey. Taste: a very important characteristic. Presence or absence of awns (bearded grain): in Senegal this allows early types to be distinguished from late ones. Elsewhere, presence of the awn is independent of length of the growing cycle. Root systems: generally, local varieties have efcient root systems which facilitate nutrient uptake even from poor soils (Raymond, 1962; Jacquinot, 1972). Relatively strong tillering allows adaptation to rainfall variations and compensation from disease attacks.

Isoenzyme studies by IRD (formerly ORSTOM) of 188 populations have distinguished ve groups in West Africa. The same studies have identied the following major groups (Tostain, 1987, 1992; Tostain and Marchais, 1993):

late millets of West Africa; millets of East and southern Africa; millets of south Asia.

Length of the growing season In many villages it is not unusual to nd four to seven millet varieties. Matlons (1985) analysis of data collected by ICRISAT in Burkina Faso shows that farmers use this diversity according to the season, soil types and food preferences. Classication of cultivated millet varieties is usually made according to the length of the growing season and sensitivity to photoperiod. Most farmers maintain both early (souna) and late (sanio) varieties to ensure some stability in production (and to cover the hungry period). Clement (1985) groups cultivated millet varieties into three groups: early varieties with a growing period of 7090 days, intermediate types with 90120 days, and late types of 120180 days. However, earliness and lateness are relative concepts, depending on the region. There is in fact a northsouth gradient: the varieties with the shortest season, in an absolute sense, occur in the drier north (the Sahelian zone), while the varieties with the longest growing period occur in the south (the Sudanian zone).

150

O. Niangado

Millet varieties of Mali show particularly high levels of diversity between regions (Marchais, 1982): this is especially marked in one particular range with respect to the length of the growing season. In the region of Kolokani, farmers cultivate two sounas or early varieties (souna die with a growing period of 7585 days, and souna boaba of 8595 days), and three sanios or late varieties (sanio totioni, 100110 days; sanio makangoulou, 115140 days; and sania boaba, 130160 days). These are grown in separate elds that are nevertheless in close proximity to each other and thus there is a large amount of geneow between these varieties, as well as between cultivated and wild varieties. Despite this geneow, farmers maintain distinct phenotypes by selecting for visible characteristics such as length and thickness of the panicle, colour, size and vitrosity of the grain, as well as the length of the growing period (Clement, 1997; see also Berthaud et al., Chapter 4, this volume).

Photoperiodism In some varieties, owering occurs when the day-length falls below a certain number of hours critical for each variety (millets are short-day plants). In fact, owering typically occurs 20 days before the end of the rainy season, though this varies, such that there are early or late varieties, which are none the less photoperiodic. This phenomenon is important in the millet-growing areas of West Africa, where rainfall, besides being low, is also highly variable, both in terms of annual precipitation and timing of the rains. In fact, for areas of similar total precipitation, the growing season is 2030 days shorter in West Africa than in India (Oran, 1977; Vaksmann and Traor, 1994). The major constraint is the variability of the date of the start the rainy season, which leads to great uncertainty over the date of sowing. Photoperiodism allows the length of the growing season to be adjusted, by photosensitivity of owing time, such that the varieties mature at the end of the rainy season (Vaksmann et al., 1996). Thus, photoperiodism provides great phenotypic plasticity, allowing adjustment of the cycle with the duration of the rainy season and providing farmers with much greater exibility in choosing the date of sowing. The earlier sowing this allows makes the crop more competitive in relation to weeds and makes better use of the peaks of nutrients that are released by mineralization from soil organic matter at the start of the rains. However, in order to have more control over the length of the cycle and, indirectly, over plant height (early-sown seeds would give rise to taller plants), as well as to overcome constraints on the use of the same genetic material in different ecologies or in different seasons, the elimination of daylength sensitivity has generally been a major objective of plant breeding programmes. The great droughts of the 1970s reinforced these objectives because it seemed that direct reduction of the cycle could facilitate drought tolerance. However, evaluation of the impact of agricultural research on rural development has shown that the improved varieties which resulted from such approaches are not well adapted to the existing production systems, due, perhaps, to their lack of exibility in the face of variable environments (Lambert, 1983). In fact the countries of this region are characterized by a large climatic variation combined with very low soil fertility. Farmers have domesticated and selected millet varieties that, while low yielding, are adapted to the environments. Stability of produc-

Millet Diversity and Use in West Africa

151

tion and resistance to environmental stresses are the main characteristics of the ecotypes of this region (Stoop et al., 1981; Matlon, 1985). The major challenge now faced by plant breeding programmes is to integrate these important characteristics in future varieties of higher yield potential.

Use of Millet Diversity

Worldwide, there are extensive millet breeding programmes, but nevertheless only a small fraction of the available diversity has been utilized. Photoperiod-sensitive material has been particularly under-utilized (Rai et al., 1997). Millet breeding has been more or less successful in India, and also in breeding for Namibia and other smaller areas throughout southern Africa. The successful varieties such as Okashana 1 were developed at ICRISAT in India using material derived from Togo in West Africa (Andrews and Kumar, 1996). There has been much less success, however, in developing improved varieties for West Africa itself. The following paragraphs describe crop improvement efforts in West Africa. Three main types of approach were taken: (i) improvement of local populations; (ii) creation of composite populations and synthetic varieties using local and/or introduced materials, and the descendants of other crosses; and (iii) improvement of the harvest index through the introduction of the dwarng gene.

Improvement of local populations Local populations were improved for characteristics such as earliness, length of the panicle, grain size and tillering, This is possible because of the substantial genetic variation that exists, and because many of these traits appear to be controlled by genes with additive effects (Bono and Leclerc, 1963; Lambert, 1983). A range of selection methods was used. For example, mass selection was used in Niger in 1967 to improve the local variety Haini Kirei, producing both normal and early varieties. In Mali the populations M9D3 and M12D1 were produced in this way. In Senegal, from 1961, recurrent selection was used to improve ve populations (three early and two late). The best progeny of the three early populations were combined to give a synthetic population souna 2, which was used as the basis of a new selection programme. The best in topcross were used to develop the souna 3 synthetic, which is actually in use in some countries of the Sahelian region. Similar methods were used to improve the local variety Zongo in Niger, and to produce the synthetic varieties PS71 in Mali and saria in Burkina Faso. This approach has provided the majority of the local improved varieties in West Africa and some are still being promoted in the sub-region. But, in general, they have not shown better production, on farmers elds, than the original local varieties (Matlon, 1983). Only rarely, have they been widely diffused. This lack of success may have many causes, but probably the main one is environmental variation: it is necessary to select for each ecological niche.

152

O. Niangado

Creation of composite populations and synthetic varieties using local and/or introduced materials, and the descendants of other crosses Some workers have considered that the variability in local varieties is limited (Kumar and Rao, 1987; Clement, 1985). For this reason, composite populations have been developed in a number of countries, including over 50 such populations developed by ICRISAT (Rai et al., 1997). For example:

Nigerian Composite, developed at the Institute for Agricultural Research in Nigeria, made up from 200 S1 derived from 275 accessions from Nigeria and 54 from other areas of West Africa. A World Composite, composed of 144 S4 derived from 1000 accessions of the world collection, including material from East Africa, and Nigeria (the Gero collection). This has been successfully used in the development of successful varieties for India (Andrews, 1986). Compel (Composite pis longs), created for Mali from six ecotypes, with long panicles, from Mali and Niger. Genetic Pools also developed for Mali, composed of crosses between varieties from Mali and other countries of West Africa. These include the SoxSa composites of early (souna) and late (sanio) varieties (Niangado and Kumar, 1989). Further improvement through recurrent selection produced the variety Soxsat, which is being promoted in Nigeria and Mali.

These methods have been shown to be efcient in improving millet for various traits (such as grain weight, tillering, length of the ear, resistance to mildew). Nevertheless, the varieties produced have not generally been accepted on farms in West Africa. Again this is probably due to genotype environment interactions between sites, as well as differences in farmer preferences.

Improvement of the harvest index through the introduction of the dwarng gene Early work by Blondel (1967) showed that, among millet varieties, some are more responsive to nitrogen fertilizers. According to Jacquinot (1972), traditional millet varieties have characteristics that are useful to traditional agriculture (hardiness and adaptation to traditional agricultural practices) but are poorly adapted to intensication (Etasse, 1972). In order to produce short-straw varieties, more adapted to the use of inputs, the dwarng gene D2 from Indian material was introduced at the Institute of tropical agricultural research (IRAT) into several local varieties in Mali, Niger, Nigeria and Senegal. A more elaborate programme, nanced by the European Development Fund, was later carried out. None of the varieties produced was acceptable on-farm. In the light of the results it has to be questioned whether it is appropriate to go to such lengths to develop dwarf varieties. The normal intensive conditions for such varieties high fertilization, high planting density, mechanical harvesting are far from those present in West Africa. In any case, much diversity for grain:straw ratio exists among local varieties. Selection among such varieties may be more fruitful since the material would, in general, be better adapted to local conditions. On the other hand, it may be that the reserves accumu-

Millet Diversity and Use in West Africa

153

lated in the long stems of traditional varieties are important in sustaining grain development in the case of water stress, as is the case for maize (Duncan, 1975) and rice (Reyniers et al., 1982).

Conclusions According to Matlon (1985), of some 3000 improved varieties examined by ICRISAT, none superior to the local types has been identied. In fact, in West Africa, millet yields have barely increased since the 1960s (FAO, 1988). Given the diversity in production environments as well as large year-to-year variation in climatic conditions, it is perhaps not surprising that the formal plant breeding efforts based on wide adaptation have failed (see also Ceccarelli et al., Chapter 6, this volume). The removal of the photoperiodic response and concentration of the dwarf varieties of high yield potential (but low yield under farmers actual conditions) have proven to be equally inappropriate.

Institutional Arrangements for Millet Breeding in West Africa

National programmes In francophone West Africa, formal millet plant breeding programmes were initiated in 1931 at the agronomy research centre of Bambey, Senegal, under the auspices of IRAT. The Bambey station served the entire francophone sub-region. Another progamme was initiated at the Institute for Agricultural Research in Nigeria. In the 1960s, national research structures began to be developed. At the same time, ICRISAT and CILSS (Permanent Inter-state Committee for Drought Control in the Sahel) were established. In the 1980s, a major regional project was initiated for the improvement of millets, sorghum, cowpea and maize. At this time the regional organization CORAF (Confrence des Responsables de Recherche Agronomique Africains, later Conseil Ouest et Centre Africain pour la Recherche et le Dveloppement Agricole) was also established. In most of Africa, the organization of research programmes and institutions continues to follow the colonial model: research is still organized by product, with one station per crop, and programmes are therefore centralized. Recently, however, some countries of West Africa have decided to decentralize certain aspects of their work. In Mali, regional centres have been developed with participation of users facilitated by a non-governmental organization: the Regional Commission of Research Users (Niangado, 1997). In Benin, INRAB also developed its strategic plan on a regional basis. However, the plant breeding programme remains national.

Collaborative research Following independence in the West African states, a number of efforts have been initiated for networking at the sub-regional level:

154

O. Niangado

The Institute for the Sahel (INSAH), operating under the auspices of CILSS, and with funding from the European Community, which has developed a network for the evaluation of new millet varieties in the region. Since INSAH has ceased activities, ICRISAT has held a number of annual workshops in various countries of the region. The Intsormil network, following the model of the Collaborative Research Support Programmes (CRSP) funded by USAID, focused on capacity building in the national agricultural research systems. A number of national scientists have been trained at US universities, but there has been little attention paid to building links within the sub-region. Rocafremi, the Millet Research network for West and Central Africa, modelled on the other networks promoted by the International Agricultural Research Centres. This network, created in 1990, involves the NARS of all countries of the subregion. It has the objectives, with regard to millet, to: strengthen the NARS; promote training; develop multilateral collaboration; ensure the circulation of scientic information among the scientists of the sub-region; dene common constraints and research priorities; facilitate the exchange of technologies between countries, and internationally; encourage the transfer of results to farmers by means of trials and demonstrations in farmers elds; and encourage collaboration between existing networks in the sub-region. It is governed by a general assembly of all participating countries.

ICRISAT has played an important role in strengthening the capacity of millet scientists in the region and the implementation of the millet network. The ICRISAT Sahelian Centre has been a centre of excellence in networking with its NARS partners. The centre has provided germplasm to the breeders in the region. While materials developed by the centre have generally been too early, some of these materials, mostly from the various bilateral projects, have been released in the region. Collaborative research networks allow participating countries to evaluate the varieties developed in other countries and to select and keep those better adapted to their own situation. Unfortunately, no network is currently engaged in the elaboration of a large genetic base material with a view of making it available to the NARS of the region; rather, countries tend to contribute by providing nished products (local improved varieties, synthetics or composites) which may not be adapted to other localities.

Conclusions
West Africa is part of the centre of origin of pearl millet. The great variability of millets in West Africa should constitute a great asset for the development of adapted varieties of higher potential, but after decades of effort there have been few improvements in production in the region. The progress made to date shows no indication of any ceiling having been reached in the genetic improvement of this crop. The great variability available in the cultivated handcars and their wild relatives is a good basis for genetic improvement in grain yield and should provide optimism about opportunities for future adaptation. The main thrust of research in the region has been to produce

Millet Diversity and Use in West Africa

155

varieties that would respond well to improved agronomic conditions, and these varieties have not been well received by farmers since they are not adapted to their specic environments. To avoid this kind of problem, it will be necessary to integrate the farmers in the breeding process, by implementing a participatory plant breeding process. A major mechanism for adaptation to variable agroclimatic conditions photoperiodism has been selected against in most formal breeding programmes. There has been little use of local diversity, and where diversity from the region has been used, for example in creating composite populations, there has been little or no attention to selection for specic adaptation to local environments. It is clear that a change of emphasis is necessary if farmers are to benet from improved varieties in the region. First, there is a need to make maximum use of locally available diversity, including factors such as photoperiodism that facilitate adaptation to the variable climatic conditions. Secondly, efforts to broaden the genetic base, through the use of composites, must be complemented by decentralized selection in the target environments to allow for specic adaptation. This is especially important in marginal environments such as those of the Sahelian zone and must take place on farmers elds (see also Ceccarelli et al., Chapter 6, this volume). This has institutional implications. Global, regional and sub-regional networks, and international centres, are important in order to access a wide diversity and to develop composite populations. Selection, however, must be decentralized to, and within, national systems. There is, therefore, a need for both centralized and decentralized structures. Existing networks, such as Rocafremi, could be refocused to promote and facilitate such approaches to these activities.

References
Andrews, D.J. (1986) Current results and prospects in sorghum and pearl millet breeding. In: Davis, T.J. (ed.) Sixth Agriculture Sector Symposium: Development of Rainfed Agriculture under Arid and Semi-arid Conditions. World Bank, Washington, DC. Andrews, D.J. and Kumar, K.A., (1996) Use of the West African pearl millet landrace Iniadi in cultivar development. Plant Genetic Resources Newsletter, 105, 1522. Blondel, D. (1971) Contribution ltude de la croissance matire sche et lalimentation azote des crales de culture sche au Sngal. Agronomie Tropicale 26, 707720. Bono, M. (1973) Contribution la morpho-systmatique des Pennisetum annuels cultivs pour leurs grains en Afrique Occidentale Francophone. Agronomie Tropicale 28(3), 229355. Bono, M. and Leclerecq, P. (1963) Mthodes damlioration varitale des mils et sorghos utilises au CRA Bambey. Agronomie Tropicale 18(1), 3352. Brunken, J., de Wet, J.M.J. and Harlan, J.R. (1977) The morphology and domestication of pearl millet. Economic Botany 31(3), 163174. Clayton, W.D. (1972) Graminae. In: Hutchinson, J. and Dalziel, J.M. (eds) Flora of West Tropical Africa, 2nd edn. Crown Agents for Overseas Governments and Administrations, London. Clement, J.C. (1985) Les mils pnicillaires de lAfrique de lOuest. Prospections et collectes. ORSTOM-IBPGR, Abidjan-Rome. Clement, J.C. (1997) Les varits traditionnelles de mil en afrique sahlienne: facteurs de stabilit et de variation. In: Gestion des ressources gntiques des plantes en afrique des savanes. Actes du Collogue, Bamako, Mali, 2428 February, 1997. de Wet, J.M.J. (1987) Pearl millet (Pennisetum glaucum) in Africa and India. In: Witcombe, J.R.

156

O. Niangado and Beckerman, S.R. (eds) Proceedings of the International Pearl Millet Workshop, 711 April, 1986, ICRISAT Center, India. Duncan, W.G. (1975) Maize. In: Evans, L.T. (ed.) Crop-Physiology. Some Case Histories. Cambridge University Press, Cambridge, pp. 2350. Etasse, C. (1972) Amlioration du mil (Pennisetum typhoides Stapf ) pour lagriculture intensive. Agronomie Tropicale 27, 783790. FAO (1998) The State of the Worlds Plant Genetic Resources for Food and Agriculture. Food and Agriculture Organization of the United Nations, Rome. Grouzis, M. (1980) Sur le Pennisetum violaceum sensus lato en Afrique de lOuest. Formes, cologie et distribution gographique. Bulletin de 1IFAN no.2. Hanna, W.W. (1987) Utilization of wild relatives of Pearl Millet. In: Witcombe, J.R. and Beckerman, S.R., (eds) Proceedings of the International Pearl Millet Workshop, 711 April, 1986, ICRISAT Center, India. Jacquinot, L. (1972) Rsultats et perspectives des recherches effectues au Sngal sur la potentialit du mil cralier (Pennisetum typhoides). Agronomie Tropicale 27(8), 815821. Kumar, A. and Rao, A. (1987) Diversity and utilization of pearl millet germplasm. In: Witcombe, J.R. and Beckerman, S.R. (eds) Proceedings of the International Pearl Millet Workshop, 711 April, 1986, ICRISAT Center, India. Lambert, C. (1983) Inuence de la prcocit sur le dveloppement du mil (Pennisetum typhoides Stapf et Hubbard) en conditions naturelles. Agronomie Tropicale 38(1), 715. Marchais, L. (1982) La diversit phnotypique des mils pnicillaires cultives au Sngal et au Mali. Agronomie Tropicale 37, 6880. Matlon, P.J. (1983) The technical potential for increased food production in the West African semi-arid tropics. Paper presented to the Conference on Accelerated Agricultural Growth in SubSaharan Africa, 29 August1 September 1983, Victoria Falls, Zimbabwe. Matlon, P.J. (1985) Sorghum and millet improvement in the West African semi-arid tropics. A critical review of objectives, methods, and progress to date. Paper presented at the runion sur la technologie approprie destine aux paysans des zones semi-arides ouest africaines; 25 avril 1985, Farming System Unit, SAFGRAD, Ouagadougou, Burkina Faso. Niangado, O. (1997) La rgionalisation de la recherche agricole au Mali. In: Shaping Effective Collaboration among Stakeholders in Regional Agricultural Research and Development in SubSaharan Africa. Workshop Proceeding. Niangado, O. and Kumar, K.A. (1989) Souna Sanio. Paper presented at the Regional Millet Workshop, ICRISAT Sahelian Centre, Sandore, Niger, 47 September, 1989. Oran, P.A. (1977) Agriculture in the semi-arid regions: problems and opportunities. In: Proceedings of the International Symposium on Rainfed Agriculture in Semi-arid Regions. Consortium for Arid Lands Institute. 1722 April 1977, Riverside, California. Portres, R. (1976) African cereals: Eleusine, fonio, black fonio, teff, Brachiaria, Paspalum, Pennisetum, and African rice. In: Harkan, J.R., de Wet, J.M.J. and Stemler A.B.L. (eds) Origins of African Plant Domestication. Moulton, The Hague. Rai, K.N., Rao, A.S. and Reddy K.N. (1997) Pearl millet. In: Fucillo, D., Sears, L. and Stapelton, P. (eds) Biodiversity in Trust: Conservation and Use of Plant Genetic Resources in CGIAR Centres. Cambridge University Press, Cambridge, pp. 243258. Ramond, C. (1962) Pour une meilleure connaissance de la croissance et du dveloppement du mil Pennisetum. Agronomie Tropicale 13, 844863. Reynier, F.N., Truong-Binh, Jacquinot, L. and Nicov, R. (1982) Amlioration de la tolrance du riz pluvial la scheresse. Agronomie Tropicale 37, 270287. Stapf and Hubbard, C.E. (1934) Flora of Tropical Africa. IX (Pennisetum). Stoop, W.A., Pattanayak, C.M., Matlon, P.J. and Root, W.R. (1982) A strategy to raise productivity of subsistence farming systems in the West African semi-arid tropics. Sorghum in the Eighties: Proceedings of the International Symposium 27 November 1981, ICRISAT Center,

Millet Diversity and Use in West Africa

157

Patancheru, India International Crops Research Institute for the Semi-Arid Tropics, pp. 519526. Tostain, S., Riandey, M.F. and Marchais, L. (1987) Enzyme diversity in pearl millet (Pennisetum glaucum). 1. West Africa. Theoretical and Applied Genetics 74, 188193. Tostain, S. and Marchais, L. (1989) Enzyme diversity in pearl millet (Pennisetum glaucum). 2. Africa and India. Theoretical and Applied Genetics 77, 634640. Tostain, S. (1992) Enzyme diversity in pearl millet (Pennisetum glaucum L.). 3. Wild millet. Theoretical and Applied Genetics 83, 733742. Tostain, S. and Marchais, L. (1993) Evaluation de la diversit gntique les mils (Pennisetum glaucum (L) R. BR) au moyen de marqueurs enzymatiques et relations entre formes sauvages et cultives. In: Le Mil en Afrique: Diversit gntique et agrophysiologique: potentialits et contraintes pour lamlioration gntique et lagriculture. ORSTOM. Edition. Vaksmann, M. and Traore, S. (1994) Adquation entre risque climatique et choix varital du mil. In: Reyniers, F.N. and Netoyo, L. (eds) Bilan hydrique agricole et scheresse en Afrique Tropicale. John Libbey Eurotext, Paris, pp. 113123. Vaksmann, M., Traor, S. and Niangado, O. (1996) Le photopriodisme des sorghos africains. Agriculture et Dvloppement 9, 1318

The State of the Use of Maize Genetic Diversity in the USA and Sub-Saharan Africa
S.P. TALLURY AND M.M. GOODMAN
Department of Crop Science, North Carolina State University, Raleigh, NC 27695, USA

Introduction
Globally, maize (Zea mays L.) is the most widely grown of all the cereal grain crops. It is unique in its adaptation to diverse ecological regions of the world, ranging from cool temperate to hot tropical and sub-tropical zones, from tropical highlands to lowland regions. It is cultivated in about 70 countries, including 53 developing countries (Timothy et al., 1988; Dowswell et al., 1996). The developing countries constitute about 64% of world maize area and 43% of world production The United States produces about 40% of the crop on only 20% of world maize area and is the worlds leading national producer, with average yields of 8.0 t ha-1 (Dowswell et al., 1996; USDA, 1999). The average per hectare maize yield of the African continent as a whole (about 40 maize-growing countries) is only about 1.6 t (USDA, 1999). Further, the average yields of the developed nations is about 2.53 times that of the developing countries. This disparity is mainly due to the lack of improved production technology and unfavourable environment that have inuenced maize production in developing countries. Maize is mostly consumed as staple human food in Latin America, Africa and Asia, whereas, in the developed countries it is mostly fed to animals. It is also used in a myriad of manufactured products, including, for example, breakfast cereals, sweeteners, ethanol fuel and beer. Increasing and stabilizing grain yield has been the principal concern for maize breeders around the world. Pandey and Gardner (1992) reported that, among 48 maize researchers spread over four continents, all but two emphasized yield improvement as a high-priority objective and invested about 46% of their resources toward production of high-yielding populations or hybrids. This is particularly true in developing countries where urbanization is taking away land from agricultural production. As a result, increased yield per unit area is the most viable option to sustain maize production. The best materials available to achieve this objective are popular elite lines developed in the target environments where hybrids are employed. Recurrent selection of segregating progenies from elite elite line crosses has been used to develop superior
IPGRI/FAO 2001. Broadening the Genetic Base of Crop Production (eds H.D. Cooper, C. Spillane and T. Hodgkin)

159

160

S.P. Tallury and M.M. Goodman

inbred lines to produce high-yielding hybrids. Maize breeders realized that this approach would lead to the narrowing of the germplasm base. However, most are disinclined to introduce unadapted germplasm into existing breeding programmes, principal reasons being that sufcient genetic progress is made using elite line crosses, the longer time period involved with exotic materials to achieve progress because of undesirable linkage drag, and the fear that the introduction of exotic germplasm might dilute the genetic make-up of the existing cultivars (Kannenberg and Falk, 1995). Beginning in the early 1900s, in developed countries, high-yielding varieties (HYVs) and hybrids gradually replaced the landraces and farmer cultivars of maize. The sprouting of commercial seed companies and improved technology coupled with the discovery of hybrid vigour played a major role in the shift from open-pollinated varieties to hybrids. This led to genetic uniformity and narrowing of the gene base which, in turn, resulted in genetic vulnerability to insect pests and fungal pathogens. A striking example of the effect of genetic uniformity leading to genetic vulnerability involved the susceptibility of maize with T-cytoplasm to southern leaf blight in 1970 in the USA. Since that time, US maize breeding efforts, both private and public, have recognized and emphasized the need to widen the genetic base of commercial maize germplasm. In spite of this emphasis, Goodman (1998a) indicated that US commercial maize contains only 2.9% exotic germplasm, which includes both temperate and tropical, and that exotic tropical germplasm is about a tenth of that. The exotic tropical germplasm includes introduced tropical germplasm that has undergone extensive selection and testing for adaptation to temperate US conditions and usually constitutes at most 25% of the exotic tropical germplasm of any one inbred line (and most often constitutes 12%). However, it is not genetic diversity per se that is needed to make progress in maize. Breeders are interested in useful genetic diversity, which depends upon the target environments and growers preferences. Although genetic diversity in time (newer cultivars replacing older ones) has occurred, these newer cultivars are still closely related to the ones they replaced, and thus the genetic base of commercial maize has remained narrow (Duvick, 1984). The scenario is different in developing countries where landraces and openpollinated varieties (OPVs) are still being grown by farmers. Recently, Taba (1997a) indicated that landraces still account for about 42% of the developing country maize area. However, this trend is gradually changing as more and more HYVs and hybrids adapted to different ecologies are becoming available, primarily through the efforts of International Agricultural Research Centres (IARCs), such as the International Maize and Wheat Improvement Centre (CIMMYT), the International Institute for Tropical Agriculture (IITA) and multinational commercial seed companies.

Maize and its Genepools

Origin Maize (Zea mays L.) is a monocotyledonous species and belongs to the family Gramineae in the tribe Maydeae. It is native to the Americas, with the centre of origin in southcentral Mexico. The earliest archaeological samples consisting of small eight-

Maize Diversity in the USA and Sub-Saharan Africa

161

rowed soft cobs, often a little larger than an ordinary pencil eraser were discovered in the caves of Tehuacn in highland southcentral Mexico (Goodman, 1988; Taba, 1997a). These were later identied as the remains of the earliest cultivated maize. Nevertheless, maize has left an evolutionary trail obscured by complexity as there are no extant intermediate forms between wild maize and the approximately 50 Mexican maize landraces, many of which are still being cultivated in Mexico (Wilkes and Goodman, 1995). The tribe Maydeae contains seven genera and some have been proposed to ll the obscurity surrounding the progenitor(s) of maize. Three (Zea, Euchlaena and Tripsacum) are native to the New World whereas the remaining four (Coix, Sclerachne, Polytoca and Chionachne) are of Asiatic origin (Hallauer and Miranda, 1988; Goodman, 1995). All seven genera, however, share the common trait of monoecy, i.e. separate male and female inorescences on the same plant. Although the Asiatic genera have not been studied as extensively as the New World genera, there has been an occasional speculation that Coix (Jobs tears) is more closely related to maize than the other Asiatic genera. Several hypotheses have been proposed for the origin and evolution of maize, but no convincing, denitive evidence exists, although teosinte may have played an important role in its lineage (Galinat, 1995). According to Goodman (1995), two hypotheses have gained approval from several researchers for the origin of the New World Maydeae members. One hypothesis suggests that maize, teosinte and Tripsacum are descendants of a common ancestor (Weatherwax, 1955). The other hypothesis suggests that maize is derived from teosinte (Beadle, 1939). Recent advances in molecular systematics using isozymes, chloroplast DNA and quantitative trait loci mapping of kernel weight in maize, teosinte and their hybrids, prompted Doebley (1990) and Doebley et al. (1994) to suggest that maize was domesticated from teosinte; teosinte is ancestral to maize and that Z. mays L. ssp. parviglumis (race Balsas) was the direct ancestor of maize.

Historical and global germplasm movement of maize There is general agreement that maize was under domestication about 50008000 years ago in southcentral or southwestern Mexico (Goodman, 1988). There is also a consensus that following its domestication in Mexico it spread widely throughout the Americas, extending as far north as southeastern Canada, before and after the European colonization. Christopher Columbus rst clearly noted the presence of maize on the north coast of Cuba in 1492 (Mangelsdorf, 1974; Goodman, 1988). He later took it with him to Spain where he introduced it into the Old World in the spring of 1493. There is a possibility that earlier European voyages to the New World might have occurred and maize could have been present in the Old World at an earlier date (see Jeffreys, 1967). Maize was widely grown in Spain, Italy and southern France by the end of the 1500s. In the early 1500s, maize entered Africa from Spain and Italy, mainly through Portuguese traders (Obilana and Asnani, 1978; Taba, 1995a, 1997a; Dowswell et al., 1996). Caribbean yellow int was the most predominant type in East Africa during the 16th century and by the 1700s became a major food crop in West and Central Africa. Portuguese traders also introduced several lowland tropical races from Brazil, the

162

S.P. Tallury and M.M. Goodman

Guyanas and the Paran basin to the West African coast. Also in the mid-17th century, Dutch settlers brought oury and int types to southern Africa, and in the late 19th and early 20th centuries, southern dents from the US and northern Mexico were imported into South Africa (Brandolini, 1970; Taba, 1995a, 1997a). Maize made its entry into south Asia in the early 1500s probably via the Mediterranean traders who introduced it along the western coast of India (Dowswell et al., 1996). By the time of Magellans voyage in the Pacic, maize had been introduced into the Philippines. By the mid-1600s, it became well established in the Philippines, Indonesia and Thailand (Taba, 1995a, 1997a; Dowswell et al., 1996). From there it spread to China, Japan and Korea by the mid-1700s.

Maize racial diversity Maize exhibits a tremendous amount of morphological variability among the different races and varieties. From the original collection of 20,000 accessions, assembled under sponsorship of the Rockefeller Foundation and the US National Research Council, a set of typical collections amounting to about 1200 were characterized and grouped into about 250300 named races (Goodman, 1983). Most of these racial collections trace their origins to the tropics and are estimated to represent about 95% of maize diversity. This vast amount of diversity is housed mainly in germplasm banks at CIMMYT and the Instituto Nacional de Investigaciones Forestales y Agropecuarias (INIFAP) in Mexico, Instituto Colombiano Agropecuario (ICA) in Colombia, Programa Cooperativo de Investigaciones en Maiz (PCIM) in Peru, the United States Department of Agriculture (USDA) North Central Regional Plant Introduction Station (NCRPIS) in Ames, Iowa, the USDA National Seed Storage Laboratory (NSSL) in Fort Collins, Colorado, and the Maize Breeding Program at North Carolina State University in Raleigh. Brown (1979) estimated that there might be 150180 distinct racial groups worldwide, of which about 130 come from Latin America. These racial complexes provided a systematic classication of the variability in maize and are a source to broaden the genetic base of commercial maize. They formed the bases for several thousand OPVs and improved populations for commercial use (Brown and Goodman, 1977; Dowswell et al., 1996). It was indicated by Brown (1979) and Goodman (1985a) that, on a racial basis, only 2% of the available germplasm is being used in temperate maize improvement; this includes the races Corn Belt Dent, Northern Flint and Cateto (Argentine) Flints. On a worldwide basis, excluding subsistence farming, at most, 5% of the races are represented in any maize of commerce (Goodman, 1983). Whether genetic diversity is measured by the number of races or racial groups of current commercial importance or by the number of alleles represented among the breeding materials, Goodman (1983) indicated that only a small proportion of the total variability available is currently being used in the temperate breeding programmes, and this is even today. Brown and Goodman (1977) classied US maize germplasm, excluding the popcorns and sweet corns, into nine broad racial complexes based on plant, ear, kernel characteristics and geographic distribution. Most of these racial complexes have had little impact on temperate maize of commerce, but one of these racial complexes, Corn

Maize Diversity in the USA and Sub-Saharan Africa

163

Belt Dent, provides essentially all US maize germplasm. The Corn Belt Dents evolved in the eastern US Corn Belt in the late 19th century. They arose from a germplasm pool consisting of about 7580% Southern Dents and 2025% Northern Flint materials (Doebley et al., 1988). Presumably, this event occurred by the accidental hybridization in the eld when poor stands of the Southern Dents were replanted with early Northern Flints (Wallace and Brown, 1956). Virtually all of the maize now produced in the US and most other temperate regions of the world traces back to the Corn Belt Dents. Some of the most prominent open-pollinated varieties of this racial complex are Reids Yellow Dent, Lancaster Sure Crop, Leaming, Boone County White, Krug Yellow Dent and Midland Yellow Dent.

Genepools of maize The racial descriptions by Brown and Goodman (1977) provided a broad grouping of the maize germplasm which forms the most important part of the primary genepool of Harlan and deWet (1971). Primary genepool The primary genepool consists of all of the maize and related germplasm that hybridizes freely with cultivated maize. The racial complexes, OPVs, inbreds, tropical, sub-tropical populations and diploid teosintes form the germplasm of this genepool. Teosinte, a weedy annual, has been considered as the closest relative of maize (Doebley, 1990). It shares the same chromosome number of 2n = 2x = 20, and similar chromosome morphology, although more terminal knobs are seen on teosinte chromosomes. Morphologically, teosinte plants often resemble maize. It crosses relatively freely with maize, and hybrids are fertile and show regular meiotic pairing (Iltis, 1983; Goodman, 1995). The annual Mexican teosintes have been considered as subspecies of maize by Doebley (1990). Two perennial forms of teosinte, a tetraploid, Z. perennis (2n = 4x = 40), and a diploid, Z. diploperennis (2n = 2x = 20), have also been found in Mexico. Although most annual teosinte races are interfertile with maize, Z. diploperennis exhibits minor fertilization barriers with maize (Taba, 1997b). Teosinte received much attention from maize researchers because of its speculated role in the origin and evolution of maize. Otherwise, it has been of very limited use for maize improvement or hybrid development (Goodman, 1988; Taba, 1995b, 1997b). According to Goodman (1988), no inbred line, hybrid or variety contains intentionally used teosinte germplasm, although two inbred lines N159 and N160, derived in part from teosinte germplasm were released in Nebraska; neither has been used in hybrid development. Additionally, Northrup King once used a privately developed inbred line derived in part from teosinte germplasm (Goodman, 1985a). Secondary genepool The species belonging to the genus Tripsacum constitute the secondary genepool of maize. Tripsacum species are perennials with chromosome numbers in multiples of x = 18. Most species exhibit a polyploid series from diploid to tetraploid and in some cases pentaploid and hexaploid (Berthaud et al., 1997), and many are obligate

164

S.P. Tallury and M.M. Goodman

apomicts. Tripsacum can be crossed with maize, generally with difculty, and the offspring show varying degrees of sterility. Chromosome morphology and knob positions of Tripsacum species differ greatly from those of maize (Goodman, 1995). Also, vegetatively, the various Tripsacum species are very variable and species identication is laborious. In general, the main use of Tripsacum species has been as fodder, and there is no documented use of Tripsacum germplasm in commercial maize production. Galinat (1977) showed that introgression between maize and Tripsacum is possible. A research programme, sponsored by the French research organization IRD, to transfer apomixis from polyploid Tripsacum populations to maize lines is underway at CIMMYT (Berthaud et al., 1995, 1997); a second such programme exists in the US at a USDA research station in Oklahoma (Kindiger, 1997). The potential benets of these projects, if and when accomplished, will greatly help farmers in the developing world, who could replant hybrid seed generation after generation. Tertiary genepool The distantly related oriental or Asiatic genera such as Coix, Sclerachne, Polytoca and Chionachne constitute this genepool. Chionachne and Sclerachne have 20 chromosomes; Polytoca has 40; and different species of Coix have 10, 20 and 40 chromosomes. Genera in the tribe Andropogoneae can also be considered to form the tertiary genepool. While wide crosses between maize and other genera have been reported since the 1940s, none of these has ever contributed to maize improvement and none has ever been backcrossed to maize. Overall, in our opinion, the primary genepool of maize has substantial genetic diversity for commercial exploitation. We do not foresee any immediate necessity to utilize taxa from secondary and tertiary genepools for commercial use, although transfer of apomixis from Tripsacum to maize would greatly benet maize growers of the developing countries.

Maize in the United States

Potential sources for broadening the germplasm base in the USA Between the late 19th century and early 20th century, the US maize production was predominated by thousands of open-pollinated varieties. However, only a few of these are represented in the current US maize breeding programmes, both public and private (Baker, 1984; Goodman, 1998a). Although open-pollinated varieties formed the basis for initial maize research programmes, the concept of heterosis coupled with the development of hybrids beginning in the 1930s, and the epidemic of southern maize blight in 1970, led private and public maize researchers to look for additional sources to diversify cultivated maize germplasm. The following section provides a review of the potential sources for broadening maize germplasm in the US. Temperate-adapted sources Historically, OPVs of maize were the source materials used in the temperate maize improvement programmes. Beginning in the early 1900s, the discovery of hybrid

Maize Diversity in the USA and Sub-Saharan Africa

165

vigour which led to the development of the inbred-hybrid concept (East, 1908; Shull, 1909, 1910; Jones, 1918) revolutionized the maize breeding programmes in the US. The rapid diffusion of the inbred-hybrid development led to the replacement of thousands of OPVs. By the 1960s, production of high-yielding hybrids in temperate breeding programmes was based largely on inbreds from the OPVs Reid Yellow Dent (RYD) and Lancaster Sure Crop (LSC) (Baker, 1984; Smith, 1988; Goodman, 1998a). The seed parents of present-day high-yielding hybrids are derived from the Iowa Stiff Stalk Synthetic source, which is predominantly of RYD origin, and mostly trace back to four inbred lines (B14, B37, B73, B84). The pollen parents, derived chiey from C103, Mo17 and Oh43, which often contribute to plant health, are derivatives of LSC. Iodent lines and Minnesota 13 lines have added early dry-down, allowing earlier harvesting with less articial drying, and permitting hybrid maize to be grown further north than ever before. Midland, Krug, Leaming, Northwestern Dent, Pride of Saline and Jarvis are other important OPVs that have also contributed to current breeding programmes, but at a much lower level than Reid and Lancaster (Gracen, 1986; Goodman, 1998a). Although about 250 or so described maize races constitute the maize genepool, only one of these races, Corn Belt Dent, represents the commercial maize germplasm grown in the entire US and almost all of the temperate regions of the world. Pedigree information on most US germplasm is available in Gerdes et al. (1993). Temperate exotic sources Most of the exotic germplasm currently employed in the US is temperate in origin and is represented by two specic sources. Maz Amargo from Argentina was used as a source of insect and disease resistance, mostly via the inbred B68 from Iowa State. The second source of temperate exotic germplasm is represented by two French lines: F2 and F7. These lines have outstanding germinability under adverse (cold, wet, cloudy) environmental conditions and are widely used for breeding and production of hybrid seed across the northern maize-growing parts of both the Old and New World. Tropical exotic sources The tropical sources most widely used for inbred line development in the US are the Cuban Flints, the Tusns and other Caribbean materials (Goodman, 1985b). The most widely used race in the tropical maize breeding, Tuxpeo, has not proved to be a very suitable germplasm source for temperate breeders. Many of the Tuxpeo collections are too late, even under short-day temperate conditions, for effective evaluation. Also, it is presumed to be one of the putative grandparents of the Corn Belt Dents and may not express high heterosis when combined with some US materials. However, direct evaluation of tropical germplasm in the US Corn Belt is virtually impossible, due mainly to the photoperiod sensitivity of the tropical materials in long-day temperate environments. A wide array of tropical materials are available from national and international maize germplasm banks. These range from OPVs to improved synthetics and are available from CIMMYT or national programmes such as ICA of Colombia. An extension of the Latin American Maize Program (LAMP) called Germplasm Enhancement of Maize (GEM), nanced by the USDA and several private seed companies (Pollak and Salhuana, Chapter 19, this volume), has been identifying tropical sources for use in US

166

S.P. Tallury and M.M. Goodman

maize breeding programs. In the maize breeding programme at North Carolina State University, private and public tropical hybrids from Central and South America have proven useful sources to extract inbred lines of all-tropical origin but adapted to temperate maize breeding conditions (Goodman, 1992). Theoretically, initiatives similar to LAMP and GEM could be undertaken to broaden the genetic base of maize in the developing countries too. However, we do not foresee an immediate necessity for such an endeavour except in those areas where hybrids have been in extensive use for several decades. Taba (1997a) indicated that openpollinated landraces and varieties are still cultivated on 42% of the developing country maize area because these OPVs have genetic attributes preferred by the farmers. Although some are inherently low-yielding, these OPVs are genetically diverse and are themselves a source of genetic variability. However, as improved populations and newer hybrids gain popularity among the farmers of the developing countries, the landrace populations may eventually disappear from farmers elds. As (and when) such a situation is approached, initiatives similar to LAMP and GEM can be undertaken to broaden the genetic base of maize in the developing world. A wealth of materials are available from CIMMYT, IITA and many national programmes, such as in Mexico and Thailand, to help maize farmers from the developing world to meet any such future challenges. Synthetic sources Several tropical and sub-tropical composites suitable for US use have been described by Gracen (1986), and several others have been developed by Arnel Hallauer at Iowa State University. These have been used to a limited extent for derivation of inbred lines, but are not used for production.

Biological and Environmental Barriers to the Movement of Maize Germplasm

Maize is an outbreeding species and grows from 58 north to 40 south latitude. It is cultivated in more diverse areas of the world than any other major crop, from sea level to 3800 m elevation, and from desert oases to areas with 11,000 mm of rainfall (Timothy et al., 1988). It is undoubtedly the most widely adapted crop and is cultivated in Canada, northern Europe, Russia, South Africa, the Himalayas, China, India, Southeast Asia and Australia (Hallauer and Miranda, 1988; Timothy et al., 1988). Adaptation to altitude inuences the type of maize grown in these areas. As a consequence of its diverse cultivation, several different local races and types have evolved and adapted to each ecological region. Essentially, each of these local types could form a source of exotic germplasm for another maize-growing region. However, attempts to introduce and use these exotic germplasm sources may be hampered by several biological and environmental constraints. Susceptibility to local insect pests and fungal pathogens is a commonly observed biological constraint. In North Carolina, we often experience this problem in our maize programme with CIMMYT tropical lines and other tropical germplasm sources. During the summer months, the CIMMYT lines become highly susceptible to smut (Ustilago maydis (DC.) cda.), other ear-rot fungi and aphids (Rhopalosiphum maidis). Similar phenomena occur when temperate materials are introduced into tropical environments.

Maize Diversity in the USA and Sub-Saharan Africa

167

Of the environmental constraints, sensitivity to day-length (photoperiod) has been observed as the major factor inuencing movement of maize from one ecology to another. When tropical materials adapted to short-day lengths (1012 h daylight) are introduced into long-day (1518 h daylight) summer growing seasons of temperate areas, they grow extremely tall (1218 ft/3.75.5 m), have more foliage, exhibit poor tassel-silk synchrony, weak root strength and much lodging. Thus, photoperiod can be an impediment for effective evaluation and movement of maize germplasm from one locale to another. Drought and heat susceptibility is the other major environmental barrier that restricts the movement of germplasm. CIMMYT has successfully developed several drought-resistant elite populations. One of them, Tuxpeo Sequia, exhibited outstanding yield increases even under severe moisture stress (Dowswell et al., 1996), although maize is not competitive with such drought-tolerant crops as cotton and sorghum. Other constraints for the movement of maize across regions include grain characteristics preferred by the consumers such as grain colour (yellow vs. white) and grain texture (int vs. dent vs. our).

Heterosis and Development of Heterotic Groups

In the US Corn Belt, heterosis for yield is almost always observed when inbred lines derived from genetically different backgrounds are used to produce hybrids. Following Shulls demonstration of heterosis in the early 1900s, OPVs served as source populations to derive inbred lines for use as parents of double-cross hybrids. Later, especially with the switch of commercial maize production to single-cross hybrids in the 1960s, progenies from elite inbred line crosses became the source populations to derive superior inbred lines. The tremendous yield gains obtained from superior double-cross and single-cross hybrids demonstrated to maize breeders that the identication and maintenance of heterotic groups are valuable in the classication of maize germplasm and a necessity for continued yield improvements. Although heterosis is a manifestation of genetic diversity of the parental lines, Moll et al. (1965) clearly showed that there is a limit to this manifestation (there is no notable yield heterosis in teosinte maize or Tripsacum maize hybrids) and, as a result, a knowledge of these relationships will help to increase the efciency of a breeding programme (Hallauer and Miranda, 1988). There are several ways to classify and designate the source populations and lines into different heterotic groups. Commonly, establishing inbred lines from already known heterotic patterns is the rst step: for example, in temperate regions, inbred lines from Stiff Stalk and Lancaster genetic backgrounds; and in the tropics, lines from Tuxpeo, ETO (Estacion Tulio Ospia), Tusn, etc. The source populations of mixed or unknown assignment, such as the broad germplasm pools of CIMMYT or tropical populations from GEM, can be crossed to a set of standard, well-established inbred lines. The population line crosses are then tested in replicated yield trials. Based on the yield data, the source populations can be grouped into corresponding heterotic patterns. As a corollary, the racial collections or CIMMYT germplasm pools can be intercrossed among themselves and the hybrids tested in replicated trials to identify the most heterotic pools (Crossa et al., 1990; Sinobas and Monteagudo, 1996). Either

168

S.P. Tallury and M.M. Goodman

intra- or interpopulation recurrent selection procedures can be conducted in these heterotic pools to derive superior inbred lines from each pool. Additionally, if the germplasm pools are broad-based populations, then selection within the pool may also produce lines which can be combined with lines selected from another pool to evaluate heterosis and produce desirable hybrids. Preliminary data from crosses of broad-based populations are often of little guidance in actual use of inbred lines, since the lines vary greatly. Thus each line developed from a broad-based population must be tested against lines of known heterotic pattern. At North Carolina State University, we screen new lines from broad-based sources (commercial tropical hybrids, synthetics) against B73 Mo17 (or an equivalent), then screen survivors against Stiff Stalk and Lancaster lines separately.

Ways of Genetic Base-broadening in the USA

In the US, exotic germplasm includes domestic unadapted germplasm, temperate foreign germplasm, and tropical or semi-tropical germplasm (Goodman, 1985a). However, elite inbred lines are used to produce superior commercial single-cross hybrids. Consequently, any attempt to use exotic germplasm should begin with an approach to tailor it, both phenotypically and genotypically, to acceptable levels for use in commercial production. Therefore, in our view, genetic base-broadening of US maize should proceed as outlined below. More detailed and comprehensive approaches may be found in the chapter on GEM (Pollak and Salhuana, Chapter 19) and that by Kannenberg (Chapter 18) in this volume.

Identify the best materials in the tropics In our experience of working with exotic germplasm at North Carolina State University, we believe that either private tropical hybrids or public tropical inbred lines are the most promising germplasm sources. Yield trial information or information from tropical breeders can be used to screen tropical hybrids. Also, the LAMP project (Salhuana et al., 1997) provides maize breeders with much of the useful information required to make judicious selection of tropical germplasm accessions. Winter nursery (short-day) conditions can be used to compare maturities with those of current breeding materials with and lodging (a weak point for most tropical materials) can be scored there at the same time as initial crosses are made.

Adaptation of selected tropical materials to temperate conditions This, in our opinion, is a critical step and could be very frustrating. The day-length sensitivity (photoperiod) of tropical materials when grown in long summer days in the US has been the main bottleneck for their large-scale use in temperate maize-growing areas. We feel that photoperiod sensitivity can be eliminated or at least reduced from segregants of even 100% tropical materials with skill and luck. To do this, we suggest either of two approaches.

Maize Diversity in the USA and Sub-Saharan Africa

169

One of the approaches is to conduct a long-term programme (ve to six generations) of plant-to-plant mating with selection for earlier, shorter, erect plants with good tassel and silk nick and some evidence of prolicacy. The selected plants are then selfed for at least two generations with visual selection to obtain 100% tropical semi-inbred lines adapted to temperate growing conditions. These lines are then topcrossed with elite temperate lines. Replicated testing of the topcrosses with commercial hybrid checks will provide a measure of the performance of these tropically derived temperateadapted inbred lines. The second approach is that being used in the GEM project. Selected tropical populations are crossed to elite lines (e.g. private lines such as LH132, FR1064, LH51 or FR697 or other private-company equivalents) under short-day conditions. These 50% exotic populations are selfed and the partially inbred lines are topcrossed with commercial testers from a complementary heterotic group to test in replicated yield trials. Also, selected 50%-exotic lines are crossed to elite lines to derive 25% tropical lines. The lines derived from these 25% exotic crosses can be tested in topcrosses to identify productive inbred lines.

Maize in Sub-Saharan Africa

Maize is the most important staple food crop of sub-Saharan Africa. Although its production has rapidly expanded in the past three decades, its share of total cereal production has remained about 40% since 1970 (Byerlee and Heisey, 1996). An increased demand for maize production of about 3.2% per year will exist in sub-Saharan Africa for the next two decades due to steady population growth, urbanization and economic growth (Byerlee and Saad, 1993). Maize is an introduced crop in Africa. Several different types were introduced into diverse ecologies of the continent and represent a vast amount of diversity (Brandolini, 1970). Although Jeffreys (1967) suggested that maize was present in southern Africa before Columbus discovered the New World, it clearly arrived in Africa by the 16th century with Portuguese traders who introduced Cateto or Caribbean Flints on both the eastern and western coasts (Obilana and Asnani, 1978; Efron, 1985; Taba, 1995a, 1997a; Byerlee and Heisey, 1996). The Dutch introduced several types into South Africa in the mid-1600s. Brandolini (1970) indicated that the Dutch settlers brought Caribbean Orange Flint, Brazilian Floury and Hickory King types to South Africa in the 17th century. Southern Dents from the US were introduced into South and East Africa in the early 1900s (Brandolini, 1970; Obilana and Asnani, 1978). In East Africa, the local yellow maize was derived from early introductions of the Caribbean Flint and Yellow Dents from South Africa. It appears that highaltitude Andean maize was the last distinct type to enter East Africa by way of missionaries before the First World War (Obilana and Asnani, 1978). In West Africa, two distinct types, Caribbean Flints and Southern Flours or Coastal Flours (related to oury types of Central and South America), were observed. As a result of the longterm cultivation with different introduced maize types in Africa, the crop became adapted to several different ecologies, which ultimately led to the isolation of landraces and farmer cultivars.

170

S.P. Tallury and M.M. Goodman

Useful genetic diversity available to farmers in sub-Saharan Africa Although maize originated and evolved in the tropics, temperate germplasm has been the most productive for farmers. Temperate germplasm has evolved through cycles of genetic improvement and diverged into heterotic groups, while tropical germplasm has mainly been exploited as open-pollinated populations. With the establishment of CIMMYT and IITA, genetically improved populations (synthetics, composites, etc.) of tropical germplasm became accessible to sub-Saharan African countries. The reader is referred to CIMMYT (1985), Eberhart et al. (1988), Timothy et al. (1988) and Dowswell et al. (1996) for details on improved populations and inbreds developed by IARCs and national programmes in Africa. Dowswell et al. (1996) have outlined the most up-to-date and comprehensive description of useful maize genetic diversity available in sub-Saharan Africa and formed the basis for the following description of useful materials for maize researchers in that area. Inbreds from CIMMYT and IITA A breeding effort initiated at IITA with later collaboration from CIMMYT resulted in the release of several tropical inbred lines with resistance to maize streak virus and/or to Striga and having good stalk and root strength (Kim et al., 1987). Several of these lines contained US germplasm in their lineage (Gracen, 1986; Kim et al., 1987). Improved populations from IITA TZSR-W and TZSR-Y were the rst high-yielding populations with maize streak resistance released by IITA for lowland tropics. The former is a white semi-int and the latter is a yellow semi-int (Dowswell et al., 1996). These two populations were used as donors for streak resistance and led to the release of resistant inbreds for use in subSaharan Africa (Kim et al., 1987). The tropical mid-altitude populations, TZMSR-W and TZEMSR-W, are also good yielding germplasm sources for sub-Saharan Africa. The former is a late-maturing, white, semi-int population, whereas the latter is an early-maturing white population that performed well in Nigeria, Cameroon and Zaire (Dowswell et al., 1996). Zimbabwean maize programme Most of the maize production area in Africa lies within the mid-altitude subtropical environment and requires resistance to streak virus. The national maize programme of Zimbabwe developed some of the best maize germplasm materials from the open-pollinated varieties Southern Cross, Salisbury White and Hickory King for this production environment. The rst commercial single-cross hybrid, SR52, was released in 1960. The three-way hybrids, R200, R201 and R205 are also very popular in Zimbabwe (Dowswell et al., 1996). Additionally, CIMMYTs regional programme in Zimbabwe developed several improved populations and inbred lines for mid-altitude areas of subSaharan Africa (Dowswell et al., 1996). Highland transition-zone germplasm CIMMYT is the major player in the development of improved germplasm for this ecology in eastern and southern Africa. Pool 9A from CIMMYT is the best germplasm source for this ecology. It is a late-maturing, white semi-dent population and is derived

Maize Diversity in the USA and Sub-Saharan Africa

171

from Kitale Synthetic II, Ecuador 573 and SR52. Another diverse germplasm complex adapted to this ecology was developed cooperatively by IITA and Cameroons maize improvement programme from materials based on CIMMYTs Pool 9A, TZMSR (IITA), V301, V304 (Guatemalan), DeKalb 690 (US), SR52 and ZS206 (Zimbabwe) (Dowswell et al., 1996). Kitale highland germplasm At the Kitale Research Station in Kenya, several breeding populations were assembled from the best local maize varieties, which mostly trace back to Hickory King. These populations, when crossed with Ecuador 573 and Costa Rica 76 (both Latin American accessions), resulted in signicant yield increases. Two outstanding germplasm complexes, Kitale I and Kitale II, resulted from intercrosses of local materials with the above Latin American varieties. A variety-cross hybrid, H611 (K II Ecuador 573), was released for commercial use. Ecuador 573 and Kenya Flat White are very useful sources to develop inbreds in Kenya and eastern Africa (Dowswell et al., 1996).

Constraints to Maize Production in Sub-Saharan Africa

As in most developing countries, many of Africas agricultural production constraints are a result of complex interactions of man-made and natural calamities. These include drought, infertile soils, insect pests and diseases, combined with lack of improved technology, and indifference of political support for agriculture (Gelaw, 1985; Byerlee and Jewell, 1997). Moreover, the colonial inuence in most countries of the continent led to the production of cash crops such as tobacco, coffee or tea for export rather than food crops such as maize. Forty-seven countries with different ecological environments, colonial histories and an array of food staples comprise the African continent (Byerlee and Eicher, 1997). It is impossible to design a standard maize production package for such diverse ecologies. Different categories of farmers demand varieties or hybrids that best t their needs, environmental conditions and subsistence management practices which, in turn, necessitate the development of ecologically specic varieties (Kim and Ajala, 1996). Overall, the major constraints to maize production in sub-Saharan Africa can be classied as environmental, technological, biological, socioeconomic and governmental (Efron, 1985; Gelaw, 1985; Eicher and Byerlee, 1997). However, the following discussion is mainly limited to the genetic (technological) constraints to maize production in sub-Saharan Africa. The adoption of improved OPVs and hybrids in sub-Saharan Africa has been promising over the past decade, and an estimated 3350% of maize area is planted with improved materials. However, the availability of these materials to small-scale farmers is patchy and ranges from 20% of maize area in Tanzania and Ethiopia to nearly 100% of the area in Zimbabwe (Byerlee and Heisey, 1996; Byerlee and Jewell, 1997). Moreover, the relative emphasis on improved OPVs versus hybrids varies from Central and West Africa to eastern and southern Africa. Although often inherently low-yielding, OPVs appeal to subsistence farmers because: (i) seed can be stored from year to year; (ii) non-availability, until recently, of hybrids for the lowland and midaltitude tropics that predominate in this area; and (iii) the principal sources of improved tropical materials for this area have been mostly OPVs bred to provide yield

172

S.P. Tallury and M.M. Goodman

stability and food security. CIMMYT and IITA have developed several improved populations and hybrids suited to the major ecologies of sub-Saharan Africa, and the reader is referred to Gelaw (1985), Dowswell et al. (1996) and Byerlee and Heisey (1996) for additional details. Overall, the main genetic constraints to yield improvement include: (i) inherently low-yielding OPVs grown on marginal soils; (ii) paucity of good source populations in the national programmes to derive high-combining inbreds; (iii) practice of inadequate and inappropriate breeding methodologies due to the lack of technical knowledge and/or facilities; (iv) improper and/or lack of seed production and maintenance facilities; and (v) pre- and post-harvest susceptibility to insects and diseases.

Maize Improvement Networks Operating in Africa

Although the National Agricultural Research Systems (NARSs) in sub-Saharan Africa have been instrumental in providing resource-poor, subsistence farmers with improved germplasm, they lack the resources necessary for continued and sustained research and training activities. Two of the IARCs, CIMMYT in Mexico and IITA in Nigeria, have been involved in the development of improved maize technology for resource-limited, smallholder farmers of developing nations and provide support to NARSs (Dowswell et al., 1996). CIMMYT has an international mandate for maize research and conducts the largest public maize improvement programme in the world. Its major focus has been on the development of improved populations for use in the developing countries of the world. IITAs maize research has been mainly on developing appropriate improved maize germplasm and production technology in Western and Central Africa. Following on this lead by IITA, CIMMYT established a regional maize research station in Zimbabwe in 1985 (Dowswell et al., 1996). According to Byerlee and Jewell (1997), an estimated 43% of the maize research budget of the IARCs was spent in Africa in the mid-1980s. Many US agricultural universities and the USDA have also been major players in the maize improvement networks for sub-Saharan Africa. Given the subsistence nature of maize in Africa and the constraints of resources for smallholder farmers, IITA and the regional branches of CIMMYT have become important networks to develop a research pipeline to complement maize research at NARSs. The release within 3 years of semi-int hybrids, MH17 and MH18, in Malawi by using dent materials from Malawi and int materials from CIMMYT is an outstanding example of collaboration between Malawian and CIMMYT scientists, which reduced both the time to release and the cost to the national research programme (Smale, 1995). Byerlee and Heisey (1996) indicated that germplasm from the IARCs contributed to about three-quarters of the maize varieties released in Africa, although data on actual use are limited. CIMMYTs extensive international testing trials in tropical environments indicated environmental similarities across several areas, leading to the concept of megaenvironments (Paterniani, 1990). These mega-environments share climatic adaptation, length of growing season, occurrence of specic diseases and pests, and so on. Such environments can be extremely useful for testing germplasm sources. CIMMYT and IITA are instrumental not only in developing improved maize germplasm, but also in the training and education of researchers and farmers through workshops. IITA led the

Maize Diversity in the USA and Sub-Saharan Africa

173

development of maize germplasm with resistance to streak virus and tolerance to infestation by the parasitic witch weed, Striga (Kim et al., 1985). Recent biotechnological innovations have the potential to help research efforts of the NARSs in most countries of sub-Saharan Africa, particularly in combating insect pests and diseases. Genetic marker techniques can help identify and select resistant genotypes even in the absence of the disease or pests. NARSs can certainly benet by collaborating with IITA, which has the research infrastructure, technology and personnel to provide training and conduct biotechnological research. In the past 5 years, an informally developed US maize genome project, an ougrowth of the Maize Genetics Co-op, has provided detailed and extensive genetic linkage maps of maize. A wealth of maize genome information is available for the public on the Internet, and the Maize Database contains current knowledge about the maize genome and its expression. There are several web sites (e.g. www.nal.usda.gov) that can access this information from anywhere in the world. One of the exciting areas of biotechnological applications in maize is to exploit genomic sequence data to efciently and effectively utilize naturally occurring variation for agronomically important traits. Recently, John Doebley from the University of Minnesota proposed that once the nucleotide sequence diversity in a set of candidate genes for important traits is determined, this information can then be used to test whether specic DNA sequence polymorphisms are associated with variation in the phenotype for these traits. Consequently, we can pinpoint specic lines in the maize germplasm pool with the desirable variation, and also the regions of maize chromosomes where useful variation is most apt to be found. It is anticipated that such studies will eventually aid in understanding the distribution of genetic diversity within the maize genome. Sub-Saharan African countries need to develop and promote strong and stable linkage with IITA and collaborate with neighbouring countries with similar environments to develop an improved maize production package for the subsistence farmers of the region. However, as Eicher and Byerlee (1997) indicated, African governments and donors will have to make an extended commitment to support research at IARCs and NARSs with focus on developing locally adapted, improved maize germplasm that can serve African farmers into the next century.

A Proposal for Broadening the Genetic Base of Maize in Sub-Saharan Africa

Background Population improvement and inbred line development have been the major components of maize breeding programmes. The choice between the two is largely inuenced by breeding objectives, available resources and time constraints; however, inbred line development is the major emphasis in private breeding programmes. Maize breeders have used several breeding methods to either improve existing populations or to derive new inbred lines for hybrid production. In sub-Saharan Africa, where OPVs still predominate on farms, the trend is gradually changing toward hybrids. A critical

174

S.P. Tallury and M.M. Goodman

evaluation of breeding strategies and selection methods is essential to broadening the genetic base and also to providing yield stability for the subsistence farmer. In our view, the experience and knowledge gained from the research investment to widen the genetic base of temperate maize germplasm could provide the framework to broaden the germplasm base of tropical maize in sub-Saharan Africa. Analogously to the use of tropical germplasm for temperate maize improvement, elite temperate germplasm could be used to widen the sub-Saharan maize germplasm base. The major limitation of this approach is, of course, the susceptibility of temperate sources to tropical insect pests and diseases (especially viruses), in addition to the potential lack of adaptation to low-input, limited rainfall agriculture characteristic of sub-Saharan Africa. CIMMYT and IITA have been conducting trials to identify and select temperate sources useful to improve tropical germplasm. Gracen (1986) summarized studies of tropical and sub-tropical CIMMYT topcrosses to the two then most widely used US dent inbreds, B73 and Mo17. The highest yielding combinations were Pool 24 (tropical late white dent) B73, and Pool 25 (tropical late yellow int) B73. The best combining crosses included Population 21 (Tuxpeo-1) with both B73 and Mo17, and Population 43 (La Posta) Mo17. CIMMYT developed these germplasm pools and populations from extensive international testing. The most commonly used CIMMYT populations in Africa are Population 21, which combines well with both Mo17 and B73, Population 30 (Blanco Cristalino-2), Population 32 (ETO Blanco) and Population 43 (La Posta). While proposing a comprehensive breeding system for African hybrid maize breeding programmes, Eberhart et al. (1988) stressed the importance of heterotic patterns in selecting germplasm to maximize population-cross performance. Because this performance determines the rate of improvement of the derived inbreds and hybrids, they expressed the view that the use of a superior common heterotic pattern among maize breeders across countries would facilitate the exchange of elite inbreds and lead to improved commercial maize hybrids. It was observed that excellent population-cross performance resulted from a Tuxpeo Dent with Caribbean Flint heterotic pattern. They also advocated multi-trait improvement, particularly during early generations because a large number of lines can be screened. Over the years, CIMMYT has developed about 20 elite maize populations adapted to lowland tropics (Dowswell et al., 1996). These materials, including lines and hybrids developed from them, can be good starting germplasm to derive inbred lines adapted to tropical African environments and to widen the sub-Saharan maize germplasm base.

Procedure to derive new inbred lines from elite exotic materials Basically, any attempt to develop new inbred lines involves choosing suitable donor parental germplasm, continued inbreeding (sib-mating or selng) with selection, topcross testing of the selected lines and identication of good inbred lines with high combining abilities. However, if elite exotic materials are chosen as source populations, we propose the following scheme to produce new lines. A scheme very similar to this is being followed at North Carolina State University. The time period can be reduced by half if two suitable growing environments (seasons) are available.

Maize Diversity in the USA and Sub-Saharan Africa

175

1. Grow F1 hybrids between local and elite exotic materials. Make plant-to-plant crosses (sib-mate). Select the best plants at harvest with desirable traits (usually earlier maturity, insect, disease and lodging resistance, and yield potential) from the donor parent. 2. Grow F2 ear-to-row progenies. Repeat as before to develop families. 3. Grow F3 families. Repeat as before. 4. Continue the same procedures (about 56 sib-mating generations) with selection for resistances against insects and diseases and until phenotypically acceptable progenies are obtained. Then, self one or two generations with selection to derive S1 or S2 lines. 5. Produce topcrosses of these lines in isolation for combining ability. Use an adapted commercial hybrid, related line cross or another inbred (if available from a different heterotic group and adequately vigorous) as a tester. 6. Conduct replicated topcross yield trials in target environments. Select the best yielding topcrosses. Based on yield and other data, select the best lines for release as inbreds. Goodman (1998b) outlined a scheme on how to maximize the favourable effects of a variable exotic source through incorporation of exotic germplasm into elite maize lines. He emphasized early topcross evaluation (individual F1s or F1S1s topcrossed to elite lines) to identify the most promising fraction of a variable source population as quickly as possible and to concentrate line development efforts on that fraction. A similar procedure can be used to derive inbred lines from crosses of elite temperate or tropical materials (from Thailand and Brazil, for example) crossed with improved open-pollinated populations or lines developed at CIMMYT.

Recommendations for Future Maize Research

In our view, the major emphasis in sub-Saharan Africa should be toward the development of varieties and hybrids tolerant to drought, temperature extremes and improved resistance to economically important diseases, pests and parasites, especially to maize streak virus and Striga. Additionally, the development of early maturing varieties would expand the area planted to maize, particularly in the drier savannah. As we have indicated in this chapter, these objectives can be achieved with less cost and in a shorter time span by collaborative research efforts between the NARSs and the maize programmes of IITA and CIMMYT. Another often ignored but important aspect is to pay attention to the needs of small-scale farmers. Low adoption of improved varieties in Africa and elsewhere is often due to lack of attention devoted to their needs. In our opinion, the farmers role as the sole consumers of the technology should change to their active participation in generating the appropriate technology for their use. Researchers must acknowledge farmers skills and innovations that have contributed to global agricultural developments and should understand their needs for future improvements of agricultural production. They should develop greater scientic respect for and collaboration with those who have gained the wisdom of generations of non-scientic farming (Haugerud and Collinson, 1990). Often one of their most pressing needs is protection from insect and rodent damage to stored grain.

176

S.P. Tallury and M.M. Goodman

Conclusions
Conventional maize breeding should continue to provide benets to small-scale farmers of sub-Saharan Africa. In our view, development of early maturing, locally adapted, disease- and pest-resistant germplasm should provide yield stability and food security for subsistence farmers. Considerable progress has been achieved with IITA and CIMMYT research efforts in the area of developing disease-resistant maize germplasm. Promising preliminary results have been obtained from drought-tolerant germplasm from the IARCs to ght the effects of periodic drought stresses (Byerlee and Heisey, 1996). Drought-tolerant materials from CIMMYT were also more efcient in utilizing available soil nitrogen and performed well even in marginal soils. Although fertilizer consumption has risen in Africa, leading to production increases, the most suitable option for resource-limited subsistence farmers is the use of genetically improved highyield-potential materials. In conclusion, as increasing population growth exerts pressure on land, genetic improvement of the breeding materials would be the single most valuable input for future maize yield increases in sub-Saharan Africa.

References
Baker, R.J. (1984) Varietal origins of inbreds. Illinois Corn Breeders School Proceedings 20, 119. Beadle, G.W. (1939) Teosinte and the origin of maize. Journal of Heredity 30, 245247. Berthaud, J., Savidan, Y., Barr, M. and Leblanc, O. (1995) Tripsacum: its diversity and conservation. In: Taba, S. (ed.) Maize Genetic Resources. Maize Program Special Report. CIMMYT, Mexico, pp. 7485. Berthaud, J., Savidan, Y., Barr, M. and Leblanc, O. (1997) Tripsacum. In: Fuccillo, D., Sears, L. and Stapleton, P. (eds) Biodiversity in Trust. Cambridge University Press, Cambridge, pp. 227233. Brandolini, A. (1970) Maize. In: Frankel, O.H. and Bennett, E. (eds) Genetic Resources in Plants Their Exploration and Conservation. IBP handbook no.11. F.A. Davis Company, Philadelphia, Pennsylvania, pp. 273309. Brown, W.L. (1979) Development and improvement of the germplasm base of modern maize. Eucarpia Corn and Sorghum Section Proceedings 9, 93111. Brown, W.L. and Goodman, M.M. (1977) Races of corn. In: Sprague, G.F. (ed.) Corn and Corn Improvement. American Society of Agronomy, Madison, Wisconsin, pp. 4988. Byerlee, D. and Eicher, C.K. (1997) Africas food crisis. In: Byerlee, D. and Eicher, C.K. (eds) Africas Emerging Maize Revolution. Lynne Rienner Publishers, Boulder, Colorado, pp. 38. Byerlee, D. and Heisey, P.W. (1996) Past and potential impacts of maize research in subSaharan Africa: a critical assessment. Food Policy 21, 255277. Byerlee, D. and Jewell, D. (1997) The technological foundation of the revolution. In: Byerlee, D. and Eicher, C.K. (eds) Africas Emerging Maize Revolution. Lynne Rienner Publishers, Boulder, Colorado, pp. 127143. Byerlee, D. and Saad, L. (1993) CIMMYTs Economic Environment to 2000 and Beyond: a Revised Forecast. Draft paper, CIMMYT, Mexico. CIMMYT (1985) To Feed Ourselves: a Proceeding of the First Eastern, Central and Southern Africa Regional Maize Workshop. March 1017. CIMMYT, Mexico. Crossa, J., Taba, S. and Wellhausen, E.J. (1990) Heterotic patterns among Mexican races of maize. Crop Science 30, 11821190.

Maize Diversity in the USA and Sub-Saharan Africa

177

Doebley, J. (1990) Molecular evidence and the evolution of maize. Economic Botany 44(3, suppl), 627. Doebley, J., Wendel, J.D., Smith, J.S.C., Stuber, C.W. and Goodman, M.M. (1988) The origin of Corn Belt maize: the isozyme evidence. Economic Botany 42, 120131. Doebley, J., Bacigalupo, A. and Stec, A. (1994) Inheritance of kernel weight in two maizeteosinte populations. Implications for crop evolution. Journal of Heredity 85, 191195. Dowswell, C.R., Paliwal, R.L. and Cantrell, R.P. (1996) Maize in the Third World. Westview Press, Boulder, Colorado. Duvick, D.N. (1984) Genetic diversity in major farm crops on the farm and in reserve. Economic Botany 38, 161178. East, E.M. (1908) Inbreeding in corn. Annual Report of the Connecticut Agricultural Experimental Station for 1907. 419428. Eberhart, S.A., Kim, S.K., Mareck., J., Darrah, L.L. and Goodman, M.M. (1988) A comprehensive breeding system for maize improvement in Africa. In: Ng, N.G., Perrino, P., Attere, F. and Zedan, H. (eds) Proceedings of the International Conference on Crop Genetic Resources of Africa, Vol. 2, Ibadan, Nigeria. International Board for Plant Genetic Resources (IBPGR), Rome, pp. 175193. Efron, Y. (1985) Use of temperate and tropical germplasm for maize breeding in the tropical area of Africa. In: Brandolini, A. and Salamini, F. (eds) Breeding Strategies for Maize Production Improvement in the Tropics. FAO, Florence, Italy, pp. 105131. Eicher, C.K. and Byerlee, D. (1997) Accelerating maize production: synthesis. In: Byerlee, D. and Eicher, C.K. (eds) Africas Emerging Maize Revolution. Lynne Rienner Publishers, Boulder, Colorado, pp. 247262. Galinat, W.C. (1977) The origin of corn. In: Sprague, G.F. (ed.) Corn and Corn Improvement. ASA, Madison, Wisconsin, pp. 147. Galinat, W.C. (1995) The origin of maize: grain of humanity. Economic Botany 49, 312. Gelaw, B. (1985) CIMMYTs maize improvement role in east, central and southern Africa. In: To Feed Ourselves: a Proceeding of the First Eastern, Central and Southern Africa Regional Maize Workshop. March 1017. CIMMYT, Mexico, pp. 208228. Gerdes, J.T., Behr, C.F., Coors, J.G. and Tracy, W.F. (1993) Compilation of North American Maize Breeding Germplasm. CSSA publication, Madison, Wisconsin. Goodman, M.M. (1983) Racial diversity in maize. In: Gordon, D.T., Knoke, J.K., Nault, L.R. and Ritter, R.M. (eds) International Maize Virus Disease Colloquium and Workshop. Ohio University Press, Wooster, Ohio, pp. 2940. Goodman, M.M. (1985a) Exotic maize germplasm: status, prospects and remedies. Iowa State Journal of Research 59, 497527. Goodman, M.M. (1985b) Use of tropical and subtropical maize and teosinte germplasm in temperate conditions. In: Brandolini, A. and Salamini, F. (eds) Breeding Strategies for Maize Production Improvement in the Tropics. FAO, Florence, Italy, pp. 93104. Goodman, M.M. (1988) The history and evolution of maize. CRC Critical Reviews in Plant Sciences 7, 197220. Goodman, M.M. (1992) Choosing and using tropical maize germplasm. Annual Corn and Sorghum Industry Research Conference Proceedings 47, 4764. Goodman, M.M. (1995) Maize. In: Smartt, J. and Simmonds, N.W. (eds) Evolution of Crop Plants. Longman, London, pp. 192202. Goodman, M.M. (1998a) Research policies thwart potential payoff of exotic germplasm. Diversity 14, 3035. Goodman, M.M. (1998b) Incorporation of exotic germplasm into elite maize lines: maximizing favorable effects of the exotic source. Theoretical Population Biology (Accepted) Gracen, V.E. (1986). Sources of temperate maize germplasm and potential usefulness in tropical and subtropical environments. Advances in Agronomy 39, 127172.

178

S.P. Tallury and M.M. Goodman Hallauer, A.R. and Miranda, J.B. (1988) Quantitative Genetics in Maize Breeding, 2nd edn. Iowa State University Press, Ames, Iowa. Harlan, J.R. and deWet, J.M.J. (1971) Toward a rational classication of cultivated plants. Taxon 20, 509517. Haugerud, A. and Collinson, M.P. (1990) Plants, genes and people: improving the relevance of plant breeding in Africa. Experimental Agriculture 26, 341362. Iltis, H.H. (1983) From teosinte to maize: the catastrophic sexual transmutation. Science 222, 886894. Jeffreys, M.D.W. (1967) Pre-Columbian maize in southern Africa. Nature 215, 695697. Jones, D.F. (1918) The effects of inbreeding and crossbreeding upon development. Connecticut Agriculture Experimental Station Bulletin 207, 5100. Kannenberg, L.W. and Falk, D.E. (1995) Models for activation of plant genetic resources for crop breeding programs. Canadian Journal of Plant Science 75, 4553. Kim, S.K. and Ajala, S.O. (1996) Combining ability of tropical maize germplasm in West Africa. 1. Open-pollinated varieties. Maydica 41, 127134. Kim, S.K., Efron,Y., Fajemisin, J. and Khadr, F.H. (1985) Evolution and progress of hybrid maize project at IITA. In: Brandolini, A. and Salamini, F. (eds) Breeding Strategies for Maize Production Improvement in the Tropics. FAO, Florence, Italy, pp. 367384. Kim, S.K., Efron, Y., Khadr, F., Fajemisin, J. and Lee, M.H. (1987) Registration of 16 maize streak virus resistant tropical maize parental inbred lines. Crop Science 27, 824825. Kindiger, B. (1997) Apomictic corn: progress, potential, and opportunity. Proceedings of the 52nd Annual Corn and Sorghum Research Conference. pp. 260270. Mangelsdorf, P.C. (1974) Corn: Its Origin, Evolution and Improvement. Belknap Press, Harvard University, Cambridge, Massachusetts. Moll, R.H., Lonnquist, J.H., Fortuno, J.V. and Johnson, E.J. (1965) The relationship of heterosis and genetic divergence in maize. Genetics 42, 139144. Obilana, A.T. and Asnani, V.L. (1978) Genetic resources of maize (Zea mays L.) in Africa. In: Crop Genetic Resources in Africa. Proceedings of a Workshop Organized Jointly by the Association for the Advancement of Agricultural Sciences in Africa (AAASA) and the IITA. January 46. Ibadan. pp. 188192. Pandey, S. and Gardner, C.O. (1992) Recurrent selection for population, variety, and hybrid improvement in tropical maize. Advances in Agronomy 48, 187. Paterniani, E. (1990) Maize breeding in the tropics. Critical Reviews in Plant Science 9, 125154. Salhuana, W., Sevilla, R. and Eberhart, S.A. (1997) Final Report: Latin American Maize Project. Pioneer. Hi-Bred International, Johnston, Iowa. Shull, G.H. (1909) A pureline method of maize breeding. American Breeders Association Report 5, 5159. Shull, G.H. (1910) Hybridization methods in maize breeding. American Breeders Magazine 1, 98107. Simmonds, N.W. (1962) Variability in crop plants, its use and conservation. Biological Reviews 37, 422465. Sinobas, J. and Monteagudo, I. (1996) Heterotic patterns among US corn belt and Spanish maize populations. Maydica 41, 143148. Smale, M. (1995) Maize is life: Malawis delayed green revolution. World Development 23, 819831. Smith, J.S.C. (1988) Diversity of United States hybrid maize germplasm; isozymic and chromatographic evidence. Crop Science 28, 6369. Taba, S. (1995a) Maize germplasm: Its spread, use, and strategies for conservation. In: Taba, S. (ed.) Maize Genetic Resources. Maize Program Special Report, CIMMYT, Mexico, pp. 758. Taba, S. (1995b) Teosinte: geographic variations and conservation. In: Taba, S. (ed.) Maize Genetic Resources. Maize Program Special Report, CIMMYT, Mexico, pp. 5973.

Maize Diversity in the USA and Sub-Saharan Africa

179

Taba, S. (1997a) Maize. In: Fuccillo, D., Sears, L. and Stapleton, P. (eds) Biodiversity in Trust. Cambridge University Press, Cambridge, pp. 213226. Taba, S. (1997b) Teosinte. In: Fuccillo, D., Sears, L. and Stapleton, P. (eds) Biodiversity in Trust. Cambridge University Press, Cambridge, pp. 234242. Timothy, D.H., Harvey, P.H. and Dowswell, C.R. (1988) Development and Spread of Improved Maize Varieties and Hybrids in Developing Countries. USAID, Washington, DC. USDA (1999) Food Grains Agricultural Statistics. USDA, Washington, DC, pp. 2830. Wallace, H.A. and Brown, W.L. (1956) Corn and its Early Fathers. Michigan State University Press, East Lansing. Weatherwax, P. (1955) History and origin of corn. In: Corn and Corn Improvement. Academic Press, New York, pp. 116. Wilkes, H.G. and Goodman, M.M. (1995) Mystery and missing links: the origin of maize. In: Taba, S. (ed.) Maize Genetic Resources. Maize Program Special Report, CIMMYT, Mexico, pp. 116.

10

The State of the Use of Potato Genetic Diversity

R. ORTIZ
International Crops Research Institute for the Semi-Arid Tropics (ICRISAT), Patancheru, 502 324 Andhra Pradesh, India

Introduction
The potato (Solanum tuberosum) belongs to the family Solanaceae. Potatoes are grown in more countries than any other crop except maize, and, with an annual production approaching 300 million t, this crop is the fourth largest crop worldwide (after rice, wheat and maize) (Anonymous, 1998). China is the worlds largest potato producer. Nowadays, in excess of one-third of the global potato output comes from the developing world, underlying the importance of potato as a source of food, employment and income in the developing world, whose countries are becoming integrated into the international potato trade of frozen French fries (chips), chips (crisps) and starch. The yield potential of potato expected from an ideotype adapted to the northern hemisphere has been calculated as 100 t ha 1. However, the actual world yield average for 1996 was 16.1 t ha 1, while one of the highest national average yields (43.7 t ha 1) was recorded in The Netherlands. Potato yields are affected by some major diseases and pests; lack of improved, locally adapted potato cultivars; limited access to chemical fertilizers; and a shortage of a high-quality planting materials (Ghislain et al., 1997). Late blight (caused by Phytophthora infestans) ranks rst among constraints for potato production worldwide. Breeders in the northern hemisphere also consider in their programmes resistance to viruses (potato leaf roll luteovirus and potato virus Y potyvirus), cyst nematode (Globodera pallida), golden nematode (G. rostochiensis) and specic insect pests. Resistance to late blight, viruses, bacterial wilt (Pseudomonas solanacearum), cyst and root-knot nematodes (Meloidogyne spp.), and potato tuber moth (Phthorimaea operculella) are desired characteristics of new cultivars for the tropics. Tuber-bearing Solanum spp. have provided resistance genes for most diseases or pests affecting potato. Besides breeding for resistance to these biotic stresses, potato breeders aim to improve yield, adaptation and tuber quality (for a review, see Tarn et
Previous address: The Royal Veterinary and Agricultural University, Department of Agricultural Sciences, 40 Thorvaldsensvej, DK-1871, Frederiksberg C, Copenhagen, Denmark IPGRI/FAO 2001. Broadening the Genetic Base of Crop Production (eds H.D. Cooper, C. Spillane and T. Hodgkin)

181

182

R. Ortiz

al., 1992). Genes from some wild and landrace genetic resources of potato have been used by potato breeders to improve these characteristics (Innes, 1992). The aims of this chapter are as follows:

To discuss the recent evolution of economically important potato species. To examine the nature and availability of genetic diversity within the group. To assess the utilization of the available genepool within breeding programmes. To look for the future needs of the crop.

Genetic Diversity and Resources Available within the Genepools of Potato

Hawkes (1990) indicated that there are 235 wild and cultivated potato species. Most of them are diploids, followed by tetraploids. The basic chromosome number (x) of potato is 12. A polyploid series occurs in the tuber-bearing Solanum species. Besides the diploid species (2n = 2x), there are seven species of triploid (3x), 27 tetraploids (4x) and three pentaploids (5x), each of which include cultivated species. Hexaploidy (6x) also occurs in addition to these ploidy levels in wild species. The most popular among the cultivated species are S. tuberosum ssp. tuberosum (adapted to long days) and S. tuberosum ssp. andigena (adapted to short days). Both subspecies are tetraploids with polysomic inheritance. An endosperm dosage system known as endosperm balance number (EBN) (Johnston et al., 1980) has inuenced the speciation of polyploid forms from diploid Solanum species. The EBN was dened as a unifying concept to predict endosperm function in intraspecic, interploidy and interspecic crosses (Ortiz and Ehlenfeldt, 1992). Every Solanum species has an effective ploidy (the EBN), which must be in a 2:1 maternal to paternal ratio in the endosperm for crosses to succeed. The EBN was determined for most Solanum species by crossing each with species of known EBN (Hanneman, 1994). Most North and Central American diploid Solanum species are 1EBN, whereas tetraploids are 2EBN and hexaploids are 4EBN. Conversely, most South American diploid species are 2EBN, whereas the polysomic tetraploids are 4EBN. Species of the same ploidy are isolated from each other by EBN, whereas geneow may occur between species of different ploidy but similar EBN. Crosses between species with different EBN could be made when one of the species produces 2n gametes (or gametes with the sporophytic chromosome number). Most tuber-bearing Solanum species are relatively easy to cross and show little chromosome differentiation between their chromosomes. Accordingly, Ortiz (1998) indicated that the primary genepool of potato consists of old and modern tetraploid cultivars, tetraploid Andean landraces and tetraploid breeding populations. Diploid cultivars or breeding populations, diploid (2EBN) tuber-bearing wild species producing 2n gametes and hexaploid (4EBN) species also belong to this primary pool. The secondary genepool consists of disomic tetraploid species (2EBN) and diploid (1EBN) tuber-bearing Solanum species. These species may cross with the crop primary genepool after isolation barriers (mainly due to EBN) are overcome. The tertiary potato genepool consists of diploid (1EBN) non-tuber-bearing wild species of the series Etuberosa and other Solanum species. This tertiary genepool could cross with the primary potato

Potato Genetic Diversity

183

genepool via bridge species and assisted by special techniques such as embryo rescue. Most diploid potatoes are self-incompatible due to a gametophytic system. Selng occurs in tetraploid potatoes, but outcrossing rates vary from 0.10 to 0.74 (Brown, 1993b), which suggests that it would be possible to select for higher outcrossing rates to develop broad-base populations by natural pollination. The International Potato Centre (CIP, Lima, Peru) holds one of the largest potato collections in the world (Table 10.1). In excess of two-thirds of this collection are cultivated genetic resources. Other important ex situ genebanks are the Commonwealth Potato Collection in the Scottish Crop Research Institute (Innes, 1992), the GermanDutch potato genebank of the Braunschweig Genetic Resources Center in Germany and the Center for Genetic Resources in The Netherlands (van Soest et al., 1984; Mix-Wagner, 1996), the US Potato Introduction Project (Hanneman, 1989; Spooner and Bamberg, 1991) and the N.I. Vavilov Institute of Plant Industry in Russia (Budin, 1992). The importance of these ex situ genebank collections to preserve diversity of potato landraces has been illustrated by Huaman and Schmiediche (1991).
Table 10.1. Composition of landrace genetic resources held at CIP, and desired characteristics reported in this and other potato germplasm collections. Species Number Characteristics Origin

Cultivated landrace genetic resources Diploids S. ajanhuiri 10 Frost tolerance, bitterness S. goniocalyx 48 Flavour, yellow esh S. phureja 170 Lack tuber dormancy, high dry matter content, resistance to late blight, PVY S. stenotomun 268 Resistance to PVY Triploids S. chaucha 97 S. juzepczukii 31 Tetraploid (S. tuberosum) ssp. andigena 2644

Bolivia, Peru Bolivia, Costa Rica, Peru Bolivia, Colombia, Ecuador, Mexico, Peru Argentina, Bolivia, Colombia, Ecuador, Peru Argentina, Bolivia, Ecuador, Peru Argentina, Bolivia, Peru Argentina, Bolivia, Chile, Ecuador Guatemala, Mexico, Peru, Venezuela Argentina, Bolivia, Ecuador, Guatemala, Peru

Flavour Frost tolerance, bitterness Resistance to many pests and diseases (e.g. late blight, PVX, PVY, nematodes) Long-day adaptation, tuber appearance, disease resistance (e.g. PVX) Frost resistance, bitterness

ssp. tuberosum

48

Pentaploids S. curtilobum

11

Argentina, Bolivia, Peru, Venezuela Bolivia, Brazil, Chile, Colombia, Ecuador, Mexico, Paraguay, Peru, Venezuela

Wild genetic resources 93 wild species 1500

Resistance to many pests and diseases, tolerance to abiotic stresses

Sources: Hoopes and Plaisted (1987); Jellis (1992); Huaman et al. (1997). PVX: potato virus X. PVY: potato virus Y.

184

R. Ortiz

However, DNA-marker analysis reveals signicant genetic differences between accessions preserved in the genebank and those re-collected in the original sites (del Rio et al., 1997b). It seems that a relatively fast evolution occurs in the wild environment owing to more genetic recombination in each season than in the ex situ genebanks, in which this recombination takes place only during seed increase. Re-collection of potato genetic resources in areas of host plantpathogen coevolution may be an alternative approach to obtain desired genetic resources for resistance breeding. Resistance to specic biotic stresses in wild tuber-bearing Solanum species does not appear to be evenly distributed throughout America. For example, North American species are more resistant to pests such as green peach aphid (Myzus persica), potato aphid (Macrosiphum euphorbiae), potato ea beetle (Epitrix spp.) and potato leafhopper (Empoasca fabae) (Flanders et al., 1997). Latitude and altitude of the collection site affect insect resistance in wild tuber-bearing Solanum species. Species collected in hot and arid areas are associated with resistance to Colorado potato beetle (Leptinotarsa decemlineata), potato ea beetle and potato leaf hopper (Flanders et al., 1992).

Accessing Solanum Germplasm from the Distinct Potato Genepools

A compatible interspecic cross permits the development of a broad base breeding population. Ploidy manipulations with haploids and 2n gametes (Fig. 10.1) offer the best option for incorporation of diploid genetic resources into the cultivated tetraploid genepool (Peloquin and Ortiz, 1992). Mendiburu and Peloquin (1976, 1977a,b) dened the terminology and the method for the systematic utilization of ploidy manipulations for potato breeding. The potato has been regarded as a crop species whose chromosome sets can be easily managed (Peloquin et al., 1989a). The most common germplasm enhancement strategy in potato involves species, haploids, 2n gametes and the EBN (Peloquin et al., 1989a). The species are the source of genetic diversity, haploids provide a method for capturing the diversity, and 2n gametes and EBN are involved in an effective and efcient method of transmitting diversity to cultivars (Jansky et al., 1990). This approach in potato breeding has been reviewed recently by Ortiz (1998). Pollen with 24 chromosomes (or 2n pollen) from diploid species results mostly due to a recessive meiotic mutant known as parallel spindles (ps), which appears to be ubiquitous among tuber-bearing Solanum species (Ortiz, 1994). Similarly, 2n eggs (with 24 chromosomes) are mostly controlled by another recessive meiotic mutant known as omission of second division (os) in diploid species (Ortiz and Peloquin, 1991). Haploids (or sporophite with the gametic chromosome number, i.e. 2n =24) of the tetraploid potato are easily obtained through interspecic interploidy 4x2x crosses. Tetraploid seed parents (either from S. tuberosum ssp. tuberosum or ssp. andigena) are crossed with a S. phureja pollinator. After such a cross, maternally derived haploids (with 24 chromosomes) are obtained through parthenogenesis (Peloquin et al., 1996). These haploids have 2EBN and are easily crossed with diploid 2EBN species with the desired characteristics. The 2EBN haploid-species hybrids producing 2n gametes are crossed directly with 4EBN tetraploid cultivars or advanced selections to obtain 4EBN tetraploid hybrids, or with other 2EBN diploid stocks for further breeding at the diploid level (Ortiz et al., 1994).

Potato Genetic Diversity

185

4x1, 4EBN cultivar

Haploid inducer
parthenogenesis

4x1, 4EBN cultivar

n egg

4x, 2EBN species

2n pollen

Maternal haploid (2x, 2EBN)

n gamete

2x, 2EBN species

n gamete

6x, 4EBN hybrid

n gamete

4x2, 4EBN cultivar

n gamete

4x2, 4EBN cultivar

n gamete

Haploid-species 2x1, 2EBN hybrid

2n gamete

5x, 4EBN hybrid

n gamete

4x3, 4EBN cultivar

n gamete

Haploid-species 2x2, 2EBN hybrid

2n gamete

4EBN addition lines 2x, 2EBN species

n pollen

4x, 4EBN hybrid Sexual depolyploidization and polyploidization

2x, 1 EBN species

2n eggs

4x1, 4EBN cultivar

n eggs

3x, 2EBN hybrid 4x2, 4EBN cultivar

n gamete

2n pollen

Haploid inducer
parthenogenesis

5x, 4EBN hybrid

n gamete

2x, 2EBN species

n gamete

4EBN addition lines 2EBN haploids

n gamete

2EBN addition lines Chromosome engineering

Fig. 10.1. Ploidy manipulations and chromosome engineering for introgression or incorporation of wild and landrace genetic resources of tuber-bearing Solanum species into the cultivated potato genepool.
Sources: Bo Fu et al. (1996); Schmiediche (1997).

Evolutionary History of Potato: Genetic Bottlenecks in Specic Locally Adapted Germplasm

Cultivated potatoes evolved from wild tuber-bearing Solanum species in the high plateau of the Andes between Peru and Bolivia (Simmonds, 1995). The crop was domesticated by ancient farmers in this area owing to its tubers, which contain carbohydrates, proteins, minerals and vitamin C. The Andean tetraploid cultivated species arose by bilateral sexual polyploidization (2x 2x cross) between S. stenotomum and an unidentied wild species (Grun, 1990). Later, 2n pollen of several diploid species fertilized n eggs of S. tuberosum spp. andigena, which led to extensive gene incorporation from wild or cultivated diploid species into the cultivated tetraploid genepool of the Andes. The potato was brought to Europe by the Spaniards in the 16th century. The crop was rst harvested at New Hampshire in North America in the early 1700s. In the mid-18th century, potato was extensively grown in England and Ireland, although the crop had been rst planted in the late 16th century. Mass selection of early-maturing clones with big tubers among seedlings derived from open pollination was the method to obtain potato germplasm adapted to long days. Brown (1993a) provides a recent review of the adoption of potato as a food in the northern hemisphere. Late blight hit Ireland in the mid-1840s, destroying most of the potato crop. As a result, 1 million people died due to famine and another 1 million left Ireland. Many researchers have suggested that S. tuberosum ssp. tuberosum from Chile could have been

186

R. Ortiz

the source of the new germplasm adapted to the long days of the northern hemisphere, especially after the late blight outbreak of the 19th century. However, Simmonds (1995) indicates that total crop destruction was unlikely and that perhaps there was a substantial introgression of Chilean germplasm into the cultivated genepool adapted to the northern latitude. One of these exotic accessions was Rough Purple Chili, which is in the pedigree of many potato cultivars (Hawkes, 1979). In the early years of potato breeding, seedlings derived from open pollination were grown for further selection of the most promising clones. With this method, todays number one cultivar of North America Russet Burbank was developed more than 100 years ago (Fig. 10.2). It was assumed initially that such open-pollinated-derived offspring was the result of selng (Glendinning, 1979; Plaisted and Hoopes, 1989). However, a recent investigation with genetic markers (Douches et al., 1991) suggests a hybrid origin of some of these selected offspring that became new cultivars in the USA. The genetic structures of S. tuberosum ssp. andigena from the Andes and ssp. tuberosum from Chile were investigated with the aid of isozyme markers (Ortiz and Huaman, unpublished results). These isozymes have been studied in detail in North American cultivars and breeding lines (Douches and Ludlam, 1991; Douches et al., 1991). Further analysis indicated that allozyme frequency changes were not always due
1851: Rev. Goodrich obtained Rough Purple Chili along with another seven South American landraces via Panama open pollination Garnet Chili1 selected by Rev. Goodrich open pollination

1861: long white tuber cultivar Early Rose1 selected by Albert Breese open pollination 1876: Luther Burbank selected a long white potato1 derived from a seedball found in his mother patch sport 1914: Russet Burbank released as new potato cultivar

1990s: Russet Burbank still grown on c. 40% of North American potato acreage

Fig. 10.2. The development of North Americas most popular potato cultivar Russet Burbank traces back to exotic germplasm and three generations of selection of hybrid offspring derived from open pollination.
Source: Plaisted and Hoopes (1989). 1Hybrid origin as revealed by genetic markers (Douches et al., 1991).

Potato Genetic Diversity

187

to genetic drift, but also to directional selection of a quantitative trait locus affecting an agronomic or quality characteristic, which is linked to a specic isozyme locus, thereby reecting the genome manipulation by potato breeders (Ortiz, 1999). These allozyme surveys among the potato genepools suggest that a genetic bottleneck, as measured by the number of allozymes per locus, occurred only in certain chromosomes of North American cultivars (Ortiz and Huaman, unpublished results). In contrast, some allozymes not observed in the farmers selections from the Andes or Chile were identied in the North American cultivars. This result conrmed that potato breeders have incorporated in their improvement programmes genes from wild and other primitive cultivated tuber-bearing Solanum species.

Cytoplasm Diversity
Potato breeders prefer S. tuberosum ssp. tuberosum cytoplasm in crosses with S. tuberosum ssp. andigena because a tuber yield advantage appears to be associated with the former, especially in northern latitudes (Plaisted, 1977). However, geneticcytoplasmic male sterility interactions control male sterility in potato (Iwanaga et al., 1991). This male sterility results from the interaction between sensitive factors in the cytoplasm of one species and nuclear genes from another species. There are some species whose cytoplasms are not sensitive, or other species that do not possess genes that interact with sensitive cytoplasm. Ortiz et al. (1993b) indicated that tetraploid cultivars differ in their cytoplasm as determined by investigating male sterility in their derived offspring from interspecic crosses. Early analysis of chloroplast DNA with restriction endonucleases demonstrated that cytoplasm of most cultivated species, except S. tuberosum ssp. tuberosum, derived from S. stenotomum (Hosaka et al., 1984). Extensive research has revealed wide chloroplast diversity in the Andean potatoes (Hosaka and Hanneman, 1988b). Intraspecic chloroplast variation appears to be common in both cultivated and wild species. The variation range of cultivated diploid species corresponds to that of the tetraploid species, which indicated that chloroplast diversity at the tetraploid level originated many times from the diploid populations. A recent investigation determined that the A chloroplast genome arose in central Peru and the T chloroplast genome originated on the BoliviaArgentina boundary (Hosaka, 1995). Based on this analysis of chloroplast genome, Hosaka (1995) proposed that potatoes were wild species differentiated from time to time and place to place to form an ancestral species complex. Consequently, a wide chloroplast diversity among Andean tetraploid potatoes was formed through sexual polyploidization, and the limited diversity found in Chilean tetraploid potatoes was derived from further selection from this tetraploid Andean genepool. The cultivars Garnet Chili and Early Rose (Fig. 10.2) have the same T type as Chilean S. tuberosum ssp. tuberosum, while the old European cultivar Myatts Ashleaf has a type A chloroplast genome similar to S. tuberosum ssp. andigena (Hosaka and Hanneman, 1988a). This nding and other surveys of chloroplast diversity in European tetraploid cultivars (Waugh et al., 1990; Powell et al., 1993) conrm the early introduction to the northern hemisphere of S. tuberosum ssp. andigena followed by the introduction of S. tuberosum ssp. tuberosum germplasm after the late blight

188

R. Ortiz

epidemics. Most modern Japanese cultivars have a T chloroplast genome, although some cultivars possess a W chloroplast genome derived from S. demissum, and very old cultivars an A chloroplast genome (Hosaka, 1993). These results conrm the similar pattern of introduction of potato cytoplasm in the northern hemisphere.

Genetic Diversity in Breeding Programmes

Cross-breeding of potato started after the late blight epidemics, although selection from open pollinated offspring was still performed until the 20th century. Interspecic crosses were made about 1850 (Hawkes, 1979) however, systematic utilization of wild species in potato breeding did not begin until the 20th century (Simmonds, 1995). Wild species were crossed to different breeding stocks or cultivars, and their selected hybrid offspring backcrossed to the recurrent cultivar parent(s). Table 10.2 lists the most important wild and landrace tuber-bearing Solanum genetic resources used during this century to develop modern cultivars in the northern hemisphere. Despite the wealth of genetic resources available for potato breeding, only a few species have been included in the genetic improvement programmes. These species were chosen to introgress resistance genes into the tetraploid cultivated genepool of North America and Europe (Ross, 1986; Plaisted and Hoopes, 1989). Extra genetic variation was brought into this genepool together with the chromosomes segments bearing the introgressed gene(s) of interest. Co-ancestry analyses have been proposed in tetraploid potato cultivars to determine their genetic relationships and inbreeding based on potato pedigrees (Mendoza and Haynes 1974a; Glendinning, 1997). This analysis suggested a narrow genetic base in US cultivars (Mendoza and Haynes, 1974b). Furthermore, Douches et al. (1996) reported that genetic yield potential has not improved among the North American cultivars released in the 20th century. Likewise, phenotypic diversity cultivars based on tuber characteristics among selected cultivars do not appear to reect the year of cultivar release (Fig. 10.3), which might suggest erratic changes owing to breeders selection during the 20th century. None the less, breeding gains have been reported for chip-

Table 10.2. Most important wild and landrace tuber-bearing Solanum genetic resources used in crop breeding of modern potato cultivars in the northern hemisphere. Ploidy Diploid Europe S. chacoense, S. commersonii S. maglia, S. microdontum, S. phureja, S. sparsipilum, S. vernei, S. verrucosum North America S. berthaultii, S. chacoense, S. kurtzianum, S. maglia, S. microdontum, S. phureja, S. raphanifolium, S. spegazzinii, S. stenotomun, S. toralapanum, S. vernei S. acaule, S. fendleri, S. tuberosum spp. andigena S. demissum

Tetraploid Hexaploid

S. acaule, S. stoloniferum, S. tuberosum spp. andigena S. demissum

Sources: Ross (1986); Plaisted and Hoopes (1989).

Potato Genetic Diversity

189

processing characteristics in this cultivated North American tetraploid germplasm (Douches et al., 1996), especially after 1960, when attention was given by breeders to potato-chip quality (Love et al., 1998). The release of the cultivar Lenape in 1967, which was withdrawn due to its amount of tuber glycoalkaloids, marked the beginning of a sustained increased in tuber solids in most breeding programmes of North America (Love et al., 1998). Lenape has S. chacoense in its pedigree, and has been the ancestor of Atlantic and Snowden, two of the most important processing cultivars in the North American market. These results also demonstrate that problems associated with tuber glycoalkaloid content could be overcome through conventional cross-breeding in potato.

(a)

Irish Cobler 1976 Sebago 1938 Norchip 1968 Atlantic 1978 Katahdin 1932 Kennebec 1940 Snowden 1990 Superior 1961

0.5

0.4

0.3 0.2 Average distance between clusters

0.1

0.0

(b)

Irish Cobler 1976 Superior 1961 Norchip 1968 Katahdin 1932 Sebago 1938 Kennebec 1940 Atlantic 1979 Snowden 1990

0.5

0.4

0.3 0.2 Average distance between clusters

0.1

0.0

Fig. 10.3. Dendrogram of tetraploid North American potato cultivars based on tuber characteristics recorded in Michigan (a) or Idaho (b). Data for average linkage cluster analysis taken from Douches et al. (1996) and Love et al. (1998), respectively.

190

R. Ortiz

Greater genetic advance for tuber yield may be expected from enhanced genetic diversity in the breeding population. However, maximum heterozygosity appears to be more important in 4x 4x crosses between adapted breeding lines than in adaptednon-adapted crosses (Bonierbale et al., 1993). As suggested by Amoros and Mendoza (1979), heterosis for yield depends on heterozygosity, but the parental material should be rst selected for adaptation. Different genetic marker systems based on protein or DNA electrophoresis are available to distinguish commercial cultivars and to assess polymorphism in potato (Douches and Ludlam, 1991; Hosaka and Hanneman, 1991; Grg et al., 1992; Hosaka et al., 1994; Demeke et al., 1996; Provan et al., 1996a,b; Sosinski and Douches, 1996; Schneider and Douches, 1997; Kim et al., 1998). Some of these marker systems have demonstrated the important role of wild potato species and exotic germplasm in the development of modern S. tuberosum cultivars (Powell et al., 1991).

Utilization of Exotic Landrace and Wild Germplasm in the Northern Hemisphere

Researchers from North America and Scotland developed independently two populations derived from S. tuberosum ssp. andigena adapted to long-day environments since the 1960s (Glendinning, 1979; Plaisted, 1982). This material has been named as Neotuberosum and provides new sources of adapted genetic resources with desired characteristics to potato breeding programmes of the northern hemisphere. For example, 50% of the genetic composition of the US cultivar Rosa derives from this Neotuberosum germplasm (Plaisted et al., 1981). Similarly, long-day-adapted populations of the cultivated diploid species S. phureja and S. stenotomum were bred in Scotland and the USA. This germplasm adapted to long-day temperate environments has also performed well under the short photoperiods of the tropics (Haynes and Mendoza, 1977; Plaisted et al., 1987). Selection for tuberization under long days improved the adaptation to the northern hemisphere of cultivated diploid and tetraploid Andean potatoes. In this regard, Fukumoto et al. (1991) indicated that selection for adaptation of Andean tetraploid germplasm followed by hybridization with adapted northern germplasm gave a greater rate of improvement in yield than early hybridization between unadapted and adapted germplasm followed by selection for adaptation. Conversely, Jacobsen and Jansky (1989) observed no apparent differences in tuberization response between haploidspecies hybrids derived from selected and unselected wild diploid species parents. These researchers reported that some wild diploid species were better than others in producing hybrids that tuberize, whereas there were no differences in the breeding value for this characteristic between haploid parents.

Landraces and In Situ Evolution in the Andes

Isozyme surveys revealed that most of the potato genotypic diversity that exists in Cusco (Peru) appears to be between rather than within populations (Brush et al., 1995). This genotype endemism indicates the role of the genotype as the unit for pre-

Potato Genetic Diversity

191

serving landrace populations. In contrast, the majority of allelic diversity was contained within geographical and landrace populations (Zimmerer and Douches, 1991). Common parentage, shared introgression and high rates of seedtuber exchange explained these results. These ndings suggest that Andean farmers are able to keep most of the potato genetic diversity in their small landholdings. In situ conservation of cultivated potato genetic resources may be required because econometric analysis suggests that adoption of high-yielding cultivars by Andean farmers in Peru leads to a reduction in but not complete loss of diversity (Brush et al., 1992). Furthermore, insect pollination between cultivated and wild species facilitates the ongoing evolution of the potato in the Andes (Johns and Keen, 1986). Electrophoretic surveys have conrmed this considerable geneow between diploid, triploid and tetraploid Andean cultivated species (Quiros et al., 1992) or between weedy and cultivated diploid species (Rabinowitz et al., 1990). This was not surprising because botanical seed propagation has been used for disease elimination, stock rejuvenation and generation of new cultivars, despite the fact that Andean farmers also preserve potato genetic diversity by clonal propagation of tubers. Hence, as suggested by Quiros et al. (1992), Andean potatoes form a large and plastic genepool amplied and renovated by outcrossing followed in some cases by human selection of desirable genotypes.

Breeding Potato for Marginal Environments

Adaptation of potato to stressful environments that limit crop performance has been the focus of the breeding programme at CIP. High and low temperature, humidity or drought, and host susceptibility to pests or diseases are among the major constraints for potato production in some agro-ecozones of the developing world. CIP strategy to breed for tolerance or resistance to these stresses and develop locally adapted improved germplasm was based on the concept that the commercial value of a clone depends on genes for adaptation, yield per se and resistance to pests and diseases (Mendoza and Rowe, 1977). Population improvement in a highly heterozygous and heterogeneous wide-based germplasm, which was derived from wild or landrace genetic resources and commercial cultivars, has been the breeding tool for genetic enhancement at CIP (Mendoza, 1989). In this way, desired genes or genetic combinations that would be at low frequency or in small segments of the potato genepool were incorporated into tetraploid breeding materials. As a result of this work, advanced tetraploid breeding populations with resistances to pests and diseases were developed for the cool and warm tropics (Golmirzaie et al., 1991). Rapid progress in resistance breeding at CIP depended on efcient screening procedures and high heritability of the characteristic. From these advanced materials many new tetraploid cultivars have been released in the developing world (Mendoza, 1992). The most recent releases combine resistance to two or three stresses. Diploid potato germplasm was also developed by CIP breeders (Watanabe et al., 1994). This diploid germplasm, which was derived from haploids of tetraploid cultivars or advanced selections, and diploid wild or cultivated species possess specic resistances, acceptable agronomic and tuber characteristics, and 2n pollen to transfer the desired gene(s) to the tetraploid level (Ortiz et al., 1994). These selected diploid stocks are another important source for widening the genetic base of potato breeding.

192

R. Ortiz

Potato Improvement Networks in the Developing World and Crop Genetic Enhancement
Cooperative research networks in the developing world have been an important partner for CIP success in the genetic betterment of the potato crop. Such networks are, among others, Programa Andino Cooperativo de Investigacin en Papa (PRACIPA), Programme Rgional de lAmelioration de la Culture de la Pomme de Terre et de la Patate Douce en Afrique Centrale et de lEst (PRAPACE), Programa Regional Cooperativo de Papa (PRECODEPA) in Central America and the Caribbean, and Programa Cooperativo de Investigaciones en Papa (PROCIPA) in South America southern cone. Another important network that brings users into the research process is Users Perspective with Agricultural Research and Development (UPWARD) in Asia. All these networks have received technical support from CIP. The history of CIP-24 provides an example of research partnership without borders for germplasm exchange and improvement (Bo Fu et al., 1996). Chinese farmers today grow CIP-24 in more than 250,000 ha (Schmiediche, 1997). This cultivar was bred in Argentina with wild and cultivated genetic resources assembled in the northern hemisphere (Fig. 10.4). This success story shows the importance of free availability of potato genetic resources as well as breeders cooperation for the genetic betterment of the crop. Cooperation also exists among potato genebank curators. They have formed the Association of Potato Intergenebank Collaborators or APIC (Bamberg et al., 1995), which holds in excess of 13,000 accessions (Huaman, 1998). International, regional and national genebanks are participating in APIC to address common problems and facilitate efcient management of potato genetic resources from a global perspective. APIC has made progress by developing joint passport and evaluation databases, coordinating collecting efforts, preserving germplasm with backup storage, and exchanging technical information on genebank management and in the free but safe movement of potato germplasm. Since 1996, 11 European genebanks are cooperating on the EU project Potato genetic resources, including preservation, characterization and use of secondary potato varieties for ecological production systems in Europe (Schuler and Hoekstra, 1997). Late blight still remains a top priority in potato breeding, and recently has been the focus for international collaboration among researchers. A Global Initiative on Late Blight (GILB) was launched in March 1996 with participants from 19 nations of Africa, America, Asia and Europe, and CIP. This GILB stimulates, integrates and coordinates research and development on late blight. As a global network, it also assists in developing common efforts that will require funding. For example, researchers of the GILB are studying the genotype environment interaction for host response to late blight to determine whether resistance to this disease shows stability across sites.

Outlook: Broadening the Genetic Base of Potato Production via Evolutionary Breeding and Biotechnology
Despite advances in introgression or incorporation of wild species and landrace germplasm in potato breeding populations, new approaches are needed to enhance the

Potato Genetic Diversity

193

Erwin Baur collection assembled in Germany with wild material collected in Mexico (1930s) US cultivars Earlaine and Saranac released by USDA in 1930s1940s Huinkil MAG (Earlaine Crosses between S. acaule, S. demissum and S. stoloniferum made at Max Planck Institute (Germany) in the 1950s MPI 61.375/23 sent by MPI to Argentina in 1967 MPI 61.375/23 and Huinkil are crossed in Argentina in 1971 Selection B71240.2 from such cross sent to CIP (Peru) in 1976 Saranac) released in 1948 in Argentina

B71240.2 sent by CIP to Institute of Plant Industry (Burnley, Australia) for cleaning B71240 sent from Australia to China as CIP-24 in 1978 Cultivar testing of CIP-24 in Inner Mongolia in 1980s CIP-24 released as new cultivar by China in 1984 CIP-24 grown in 250,000 ha in the 1990s in China B71240.2 released as Achirana-INTA in 1986

Fig. 10.4. Germplasm sources and development of B-71240.2 (also known as Achirana-INTA or CIP-24), a photoperiod-insensitive, early-maturing cultivar with resistance to PLRV, some resistance to late blight, and tolerance to Al. B-71240.2 has been released as a cultivar in 12 further tropical or subtropical countries.
Sources: Peloquin et al. (1989a,b); Ortiz (1998). MPI: Max Planck Institute. CIP: International Potato Centre.

utilization of these genetic resources in the betterment of the cultivated potato genome. Evolutionary crop breeding should be pursued in potato to widen the genetic base of the cultivated germplasm. In this breeding scheme, heterozygous wild species and landraces are the source of allelic diversity, which should be incorporated into locally adapted populations. Owing to the importance of non-additive gene interaction in tetraploid potatoes, selection for specic combining ability (SCA) on segregating populations should result in high-yielding germplasm. This concept of SCA considers both the cross combination and the specic individual within the cross, i.e. outstanding hybrids should be selected from the best crosses. A true potato seed synthetic cultivar from open pollination may broaden the genetic base of crop production in potato (Ortiz, 1997). Selected tetraploid and diploid parental genotypes, based on their SCA, may be chosen for isolated polycrosses to obtain tetraploid offspring via unilateral or bilateral sexual polyploidization (Ortiz, 1998). Likewise, synthetic breeding populations derived from these polycrosses could be shared with other breeders, who will select in cooperation with their local farmers promising genotypes for further cultivar release. The participation of farmers in the selection process requires that an interesting amount of variation still exists in the

194

R. Ortiz

breeding population. Farmers selection within adapted germplasm may be required for cultivar development in some regions because tuber yield per se in optimum environments seems to be a poor indicator of acceptability (Prain et al., 1992), or sometimes breeders and farmers choices diverge (Thiele et al., 1997). For example, culinary quality inuences choice of new germplasm by farmers and consumers. A core collection of Solanum species will improve the utilization of Solanum genetic resources in potato breeding. This core collection must contain chosen wild and cultivated accessions representing with minimum redundancy the genetic variability of the whole tuber-bearing Solanum germplasm and closely related non-tuber-bearing Solanum species (Ortiz, 1998). CIP has been developing a core collection of Latin American landraces using hierarchical classication based on their taxonomic species, ecogeographical origin, morphological characteristics, molecular markers and host response to diseases and pests (Huaman, 1998). Recently, CIP scientists have dened a core sub-set of S. tuberosum ssp. andigena, which contains 306 farmers landraces (Huaman et al., 1999). The number of accessions included in the core was determined primarily by the square root of the number of accessions where andigena was collected (Huaman et al., 2000a). A phenogram was constructed from the descriptor data using a matching coefcient and the unweighted pair group method using arithmetic averages. Accessions were retained in each cluster considering data on resistance to diseases and pests, dry matter content and number of duplicate accessions identied in the original collection. Allozyme frequencies validated the sampling procedure for this core collection (Huaman et al., 2000b). The most frequent allozymes in the core collection were those observed in the entire collection, and their frequency distributions were homogeneous for most loci. Core sub-sets for other cultivated species will be based on sampling with DNA markers. There are several cross-breeding methods to incorporate most of the genetic resources into potato breeding populations (Fig. 10.1). Nevertheless, protoplast fusion (or somatic hybrids) offers another means for asexual interspecic gene transfer in potato breeding (Hegelson, 1989). Likewise, genetic engineering provides a source of new alleles, especially for resistance breeding, in potato. Molecular maps based on DNA and biochemical markers have been developed in potato (Bonierbale et al., 1988; Gebhardt et al., 1989, 1991; Freyre et al., 1994; Eck et al., 1995; Jacobs et al., 1995). The number of genetic markers, available for mapping and assisted breeding, makes potato one of the most fully covered plant genomes (Tanksley et al., 1992). A Solgenes database may be accessed through the World Wide Web interface (http://probe.nalusda.gov:8300/cgi-bin/browse/solgenes) maintained by the National Agricultural Library in Beltsville (Maryland, USA). Genetic-aided analysis with molecular markers proved that quantitative variation for many complex characteristics was under polygenic control. Likewise, resistance gene clusters have been located in chromosomes 5 and 11 after mapping major resistance genes to similar or close chromosome segments. DNA markers may assist gene introgression into breeding pools. Marker-aid selection could facilitate breeding for some quantitative characteristics. However, some markers linked to these characteristics are inconsistent across populations and environments (Bonierbale et al., 1993; Ortiz et al., 1993a; Freyre and Douches, 1994). Hence, candidate markers for assisted selection must be validated in independent populations and across environments. The lack of common markers among different populations also indicates that

Potato Genetic Diversity

195

potato breeders have been able to develop improved germplasm by manipulating distinct chromosome regions in their respective populations. This nding suggests that genetically enhanced populations may be developed in potato with the aid of DNA markers by combining specic chromosomal regions controlling the desired characteristic(s). One of the most important targets will be pyramiding available resistance genes to pests and diseases with the aid of these DNA markers in broad-base breeding populations. Broadening the genetic base of potato with locally adapted, pest- and disease-resistant germplasm will ensure the sustainable and environment-friendly production of this crop. However, as demonstrated by the allozyme surveys reported earlier (Ortiz, 1999), the need for broadening the genetic base in potato may be for specic chromosomes or regions within chromosome arms.

References
Amoros, W.R. and Mendoza, H.A. (1979) Relationship between heterozygosity and yield in autotetraploid potato. American Potato Journal 56, 455. Anonymous (1998) CIP Potato Facts: Production, Use, Consumption, Trade. International Potato Centre, Lima, Peru. Bamberg, J.B., Huaman, Z. and Hoekstra, R. (1995) International cooperation in potato germplasm. In: Duncan, R.R. (ed.) International Germplasm Transfer: Past and Present. Crop Science Society of America Special Publication 23, American Society of AgronomySoil Science Society of America, Madison, Wisconsin, pp. 177182. Bo Fu, S., Weining, T., Jimin, W., Chunlin, W., Zhengui, Y., Shegnwu, W. and Huarte, M. (1996) Economic impact of CIP-24 in China. In: Walker, T.S. and Crissman, C.C. (eds) Case Studies of the Economic Impact of CIP-related Technology. International Potato Centre, Lima, Peru, pp. 3149. Bonierbale, M.W., Plaisted, R.L. and Tanksley, S.D. (1988) RFLP maps based on a common set of clones reveal modes of chromosomal evolution in potato and tomato. Genetics 120, 10951103. Bonierbale, M.W., Plaisted, R.L. and Tanksley, S.D. (1993) A test of maximum heterozygosity hypothesis using molecular markers in tetraploid potatoes. Theoretical and Applied Genetics 86, 481491. Brown, C.R. (1993a) Origin and history of the potato. American Potato Journal 70, 363373. Brown, C.R. (1993b) Outcrossing rates in cultivated autotetraploid potato. American Potato Journal 70, 363373. Brush, S.B., Taylor, J.E. and Bellon, M.R. (1992) Technology adoption and biological diversity in Andean potato agriculture. Journal of Developmental Economics 39, 365387. Brush, S.B., Kesseli, R., Ortega, R., Cisneros, P., Zimmerer, K. and Quiros, C. (1995) Potato diversity in the Andean center of crop domestication. Conservation Biology 9, 11891198. Budin, K.Z. (1992) The USSR potato collection: its genetic potential and value for plant breeding. Diversity 8, 1213. del Rio, A.H., Bamberg, J.B., Huaman, Z., Salas, A. and Vega, S.E. (1997b) Assessing changes in the genetic diversity of potato gene banks. 2. In situ vs. ex situ. Theoretical and Applied Genetics 95, 199204. Demeke, T., Lynch, D.R., Kawchuk, L.M., Kozub, G.C. and Amstrong, J.D. (1996) Genetic diversity of potato determined by random amplied polymorphic DNA analysis. Plant Cells Reports 15, 662667. Douches, D.S. and Ludlam, K. (1991) Electrophoretic characterization of North American potato cultivars. American Potato Journal 68, 767780.

196

R. Ortiz Douches, D.S., Ludlam, K. and Freyre, R. (1991) Isozyme and plastid DNA assessment of pedigrees of nineteenth century potato cultivars. Theoretical and Applied Genetics 82, 192200. Douches, D.S., Maas, D., Jartrzebski, K.R. and Chase, R.W. (1996) Assessment of potato breeding progress in the USA over the last century. Crop Science 36, 15441552. Eck, H.J. van, Voort, J.R. van der, Draaistra, J., Zand voort, P. van, Enckevort, E. van, Segers, B., Peleman, J., Jacobsen, E., Helder, J. and Bakker, J. (1995) The inheritance and chomosomal localization of AFLP markers in a non-inbred potato offspring. Molecular Breeding 1, 397410. Flanders, K.L., Hawkes, J.G., Radcliffe, E.B. and Lauer, F.I. (1992) Insect resistance in potato sources, evolutionary relationships, morphological and chemical defenses, and ecogeographical associations. Euphytica 61, 83111. Flanders, K.L., Radcliffe, E.B. and Hawkes, J.G. (1997) Geographic distribution of insect resistance in potatoes. Euphytica 93, 201221. Freyre, R. and Douches, D.S. (1994) Development of a model for marker-assisted selection of specic gravity in diploid potato across environments. Crop Science 34, 13611368. Freyre, R., Warnke, S., Sosinski, B. and Douches, D.S. (1994) Quantitative trait locus analysis of tuber dormancy in diploid potato (Solanum spp.). Theoretical and Applied Genetics 89, 474480. Fukumoto, O., Plaisted, R.L. and Ewing, E.E. (1991) Comparison of two techniques for introgression of unadapted andigena germplasm into temperate germ plasm. American Potato Journal 68, 391404. Gebhardt, C., Ritter, E., Debener, T., Schachtschabel, U., Walkemeier, B., Uhrigh, H. and Salamini, F. (1989) Restriction fragment length polymorphism analysis and linkage map in Solanum tuberosum. Theoretical and Applied Genetics 78, 6575. Gebhardt, C., Ritter, E., Barone, A., Debener, T., Walkemeier, B., Schachtschabel, U., Kaufmann, H., Thompson, R.D., Bonierbale, M.W., Ganal, M.W., Tanksley, S.D. and Salamini, F. (1991) RFLP maps of potato and their alignment with the homoeologous tomato genome. Theoretical and Applied Genetics 83, 4957. Ghislain, M., Querci, M., Bonierbale, M., Golmirzaie, A. and Nelson, R. (1997) Biotechnology and the Potato: Applications for the Developing World. International Potato Centre, Lima, Peru. Glendinning, D.R. (1979) Enriching the potato gene pool using primitive cultivars. In: van Harten, A.M. and Zeven, A.C. (eds) Broadening the Genetic Base of Crops. Pudoc, Wageningen, pp. 3945. Glendinning, D.R. (1997) Estimation of inbreeding in potato pedigrees. Potato Research 40, 277284. Golmirzaie, A.M., Mendoza, H.A., Vallejo, R.L., Espinoza, J. and Serquen, F. (1991) Breeding potaoes for warm tropics. In: Production, Post-harvest Technology and Utilization of Potato in the Warm Tropics. Proceedings of a Workshop held 2327 July 1990, Reduit, Mauritius. Mauritius Sugar Industry Research Institute, Reduit, Mauritius, pp. 1221. Grg, R., Schachtschabel, U., Ritter, E., Salamini, F. and Gebhardt, C. (1992) Discrimination among 136 tetraploid potato varieties by ngerprints using highly polymorphic DNA markers. Crop Science 32, 815819. Grun, P. (1990) The evolution of cultivated potatoes. Economic Botany 44(Suppl.), 3955. Hanneman, R.E. Jr (1989) Potato germplasm resources. American Potato Journal 66, 655667. Hanneman, R.E. Jr (1994) Assignment of endosperm balance number to the tuber bearing solanums and their close non-tuber bearing Solanums. Euphytica 74, 1925. Hawkes, J.G. (1979) Genetic poverty of the potato in Europe. In: van Harten, A.M. and Zeven, A.C. (eds) Broadening the Genetic Base of Crops. Pudoc, Wageningen, pp. 1927. Hawkes, J.G. (1990) The Potato, Evolution, Biodiversity, and Genetic Resources. Bellhaven Press, London.

Potato Genetic Diversity

197

Haynes, F.L. and Mendoza, H.A. (1977) The value of cultivated diploid species in breeding. Potato Research 20, 250. Hegelson, J.P. (1989) Somatic hybridization of wild Solanum species with potato: a potential source of diversity for breeders. In: Louwes, K.M., Toussaint, H.A.J.M. and Dellaert, L.W.L. (eds) Parental Line Breeding and Section in Potato Breeding. Pudoc, Wageningen, pp. 8794. Hoopes, R.W. and Plaisted, R.L. (1987) Potato. In: Fehr, W.R. (ed.) Principles of Cultivar Development, Vol. 2. Macmillan, New York, pp. 385436. Hosaka, K. (1993) Similar introduction and incorporation of potato chloroplast DNA in Japan and Europe. Japanese Journal of Genetics 68, 5561. Hosaka, K. (1995) Successive domestication and evolution of the Andean potatoes as revealed by chloroplast DNA restriction endonuclease. Theoretical and Applied Genetics 90, 356363. Hosaka, K. and Hanneman, R.E. Jr (1988a) The origin of cultivated tetraploid potato based on chloroplast DNA. Theoretical and Applied Genetics 76, 172176. Hosaka, K. and Hanneman, R.E. Jr (1988b) Origin of chloroplast DNA diversity in the Andean potatoes. Theoretical and Applied Genetics 76, 333340. Hosaka, K. and Hanneman, R.E. Jr (1991) Seed protein variation within accessions of wild and cultivated potato species and inbred Solanum chacoense. Potato Research 34, 419428. Hosaka, K., Ogihara, Y., Matsubayashi, M. and Tsunewaki, K. (1984) Phylogenetic relationships between the tuberous Solanum species as revealed by restriction endonuclease analysis of chloroplast DNA. Japanese Journal of Genetics 59, 349369. Huaman, Z. (1998) Collection, maintenance and evaluation of potato genetic resources. Plant Varieties and Seeds 11, 2938. Huaman, Z. and Schmiediche, P. (1991) The importance of ex situ conservation of germplasm: a case study. Diversity 7, 12. Huaman, Z., Golmirzaie, A. and Amoros, W. (1997) The potato. In: Fuccillo, D., Sears, L. and Stapleton, P. (eds) Biodiversity in Trust. Cambridge University Press, Cambridge, pp. 2128. Huaman, Z., Ortiz, R. and Gomez, R. (1999) A proposed Solanum tuberosum ssp. andigena core collection. In: Impact on a Changing World: Program Report 19971998. International Potato Centre, Lima, Peru, pp. 185194. Huaman, Z., Ortiz, R. and Gomez, R. (2000a) Selecting a Solanum tuberosum ssp. andigena core collection using morphological, geographical, disease and pest descriptors. American Potato Journal 77, 183190. Huaman, Z., Ortiz, R., Zhang, D. and Rodriguez, F. (2000b) Isozyme analysis of entire and core collection of Solanum tuberosum ssp. andigena potato cultivars. Crop Science 40, 273276. Innes, N.L. (1992) Gene banks and their contribution to breeding of disease resistant cultivars. Euphytica 63, 2331. Iwanaga, M., Ortiz, R., Cipar, M.S. and Peloquin, S.J. (1991) A restorer gene for genetic cytoplasmic male sterility in cultivated potatoes. American Potato Journal 68, 1928. Jacobs, J.M.E., van Eck, H.J., Arens, P., Verkerk-Bakker, B., te Lintel Hekkert, B., Bastiaanssen, H.J., El-Kharbotly, A., Pereira, A. and Jacobsen, E. (1995) A genetic map of potato (Solanum tuberosum) integrating molecular markers, including transposons, and classical markers. Theoretical and Applied Genetics 91, 289300. Jacobsen, T.L. and Jansky, S.H. (1989) Effects of pre-breeding wild species on tuberization of Solanum tuberosum haploidwild species hybrids. American Potato Journal 66, 803811. Jansky, S.H., Yerk, G.L. and Peloquin, S.J. (1990) The use of potato haploids to put 2x wild species germplasm into a usable form. Plant Breeding 104, 290294. Jellis, G.J. (1992) Multiple resistance to diseases and pests in potato. Euphytica 63, 5158. Johns, T. and Keen, S.L. (1986) Ongoing evolution of the potato in the Altiplano of western Bolivia. Economic Botany 40, 409424. Johnston, S.A., den Nijs, A.P.M., Peloquin, S.J. and Hanneman, R.E. Jr (1980) The signicance

198

R. Ortiz of genetic balance to endosperm development in interspecic crosses. Theoretical and Applied Genetics 57, 59. Kim, J.H., Joung, H., Kim, H.Y. and Lim, Y.P. (1998) Estimation of genetic variation and relationship in potato (Solanum tuberosum L.) cultivars using AFLP markers. American Journal of Potato Research 75, 107112. Love, S.L., Pavek, J.L., Thompson-Johns, A. and Bohl, W. (1998) Breeding progress for potato chip quality in North American cultivars. American Journal of Potato Research 75, 2736. Mendiburu, A.O. and Peloquin, S.J. (1976) Sexual polyploidization and depolyploidization: some terminology and denitions. Theoretical and Applied Genetics 48, 137143. Mendiburu, A.O. and Peloquin, S.J. (1977a) Bilateral sexual polyploidization in potatoes. Euphytica 26, 573583. Mendiburu, A.O. and Peloquin, S.J. (1977b) The signicance of 2n gametes in potato breeding. Theoretical and Applied Genetics 49, 5361. Mendoza, H.A. (1989) Population breeding as a tool for germplasm enhancement. American Potato Journal 66, 639653. Mendoza, H.A. (1992) Breeding potatoes for combined resistance or tolerance to the main biotic and abiotic stresses. In: Rousselle-Bourgeois, F. and Rousselle, P. (eds) Proceedings of the Joint Conference on EAPR Breeding and Varietal Assessment Section and the EUCARPIA Potato Section, Landerneau, France, 1217 January 1992. INRA, Ploudaniel, France, pp. 123130. Mendoza, H.A. and Haynes, F.L. (1974a) Some aspects of breeding and inbreeding in potatoes. American Potato Journal 50, 216222. Mendoza, H.A. and Haynes, F.L. (1974b) Genetic relationship among potato cultivars grown in the United States. HortScience 9, 328330. Mendoza, H.A. and Rowe, P.R. (1977) Strategy for potato population breeding for adaptation to the lowland tropics. American Potato Journal 54, 488. Mix-Wagner, G. (1996) Management of the in vitro genebank for old potato cultivars at the Institute of Crop Science in Braunschweig. Landbauforschung Volkenrode 46, 59. Ortiz, R. (1994) El mutante meitico huso paralelos (ps) en la evolucin de las especies tuberferas del gnero Solanum. Boletn Lima 16, 363379. Ortiz, R. (1997) Breeding for potato production from true seed. Plant Breeding Abstracts 67, 13551360. Ortiz, R. (1998) Potato breeding via ploidy manipulations. Plant Breeding Reviews 16, 1586. Ortiz, R. (1999) Nordic Professorship in Plant Genetic Resources: Final Report (19971999). The Royal Veterinary and Agricultural University, Frederiksberg, Denmark. Ortiz, R. and Ehlenfeldt, M.K. (1992) The importance of endosperm balance number in potato breeding and the evolution of tuber bearing Solanums. Euphytica 60, 105113. Ortiz, R. and Peloquin, S.J. (1991) Breeding for 2n egg production in haploid species 2x potato hybrids. American Potato Journal 68, 691703. Ortiz, R., Douches, D.S., Kotch, G.P. and Peloquin, S.J. (1993a) Use of haploids and isozyme markers for genetic analysis in the polysomic polyploid potato. Journal of Genetics and Breeding 47, 283288. Ortiz, R., Iwanaga, M. and Peloquin, S.J. (1993b) Male sterility and 2n pollen in 4x progenies derived from 4x 2x and 4x 4x crosses in potatoes. Potato Research 36, 227236. Ortiz, R., Iwanaga, M. and Peloquin, S.J. (1994) Breeding potatoes for developing countries using wild tuber bearing Solanum spp. and ploidy manipulations. Journal of Genetics and Breeding 48, 8998. Peloquin, S.J. and Ortiz, R. (1992) Techniques for introgressing unadapted germplasm to breeding populations. In: Stalker, T.P and Murphy, J.P. (eds) Plant Breeding in the 1990s. CAB International, Wallingford, UK, pp. 485507. Peloquin, S.J., Jansky, S.H. and Yerk, G.L. (1989a) Potato cytogenetics and germplasm utilization. American Potato Journal 66, 629638.

Potato Genetic Diversity

199

Peloquin, S.J., Yerk, G.L., Werner, J.E. and Darmo, E. (1989b) Potato breeding with haploids and 2n gametes. Genome 31, 10001004. Peloquin, S.J., Gabert, A.C. and Ortiz, R. (1996) Nature of pollinator effect in potato haploid production. Annals of Botany 77, 539542. Plaisted, R.L. (1977) Andigena as a source of genetic diversity. Potato Research 20, 255. Plaisted, R.L. (1982) Progress and future plans for the use of neotuberosum populations. In: Utilization of the Genetic Resources of the Potato III. International Potato Centre, Lima, Peru, pp. 6579. Plaisted, R.L. and Hoopes, R.W. (1989) The past record and future prospects for the use of exotic potato germplasm. American Potato Journal 66, 603627. Plaisted, R.L., Thurston, H.D., Sieczka, J.B., Brodie, B.B., Jones, E.D. and Cetas, R.C. (1981) Rosa: a new golden nematode resistant variety for chipping and tablestock. American Potato Journal 58, 451455. Plaisted, R.L., Mendoza, H.A., Thurston, H.D., Ewing, E.E., Brodie, B.B. and Tingey, W.M. (1987) Broadening the range of adaptation of andigena (neo-tuberosum) germplasm. CIP Circular 15(2), 15. Powell, W., Phillips, M.S., McNicol, J.W. and Waugh, R. (1991) The use of DNA markers to estimate the extent and nature of genetic variability in Solanum tuberosum cultivars. Annals of Applied Biology 118, 423432. Powell, W., Baird, E., Duncan, N. and Waugh, R. (1993) Chloroplast DNA variability in old and recently introduced potato cultivars. Annals of Applied Biology 123, 403410. Prain, G., Uribe, F., Scheidegger, U. and Rhoades, R.E. (1992) The friendly potato: farmer selection of potato varieties for multiple uses. In: Moock, H.L. (ed.) Diversity, Farmer and Knowledge. Cornell University Press, Ithaca, New York, pp. 5268. Provan, J., Powell, W. and Waugh, R. (1996a) Microsatellite analysis of relationships within cultivated potato (Solanum tuberosum). Theoretical and Applied Genetics 92, 10781084. Provan, J., Powell, W. and Waugh, R. (1996b) Analysis of cultivated potato using intermicrosatellite amplication. Genome 39, 767769. Quiros, C.F., Ortega, R., van Raamsdock, L., Herrera-Montoya, M., Cisneros, P., Schmidt, E. and Brush, S.B. (1992) Increase of potato genetic resources in their center of diversity: the role of natural outcrossing and selection by the Andean farmer. Genetic Resources and Crop Evolution 39, 107112. Rabinowitz, D., Linder, C.R., Ortega, R., Begazo, D., Murguia, H., Douches, D.S. and Quiros, C.F. (1990) High levels of interspecic hybridization between Solanum sparsipilum and S. stenotomum in experimental plots in the Andes. American Potato Journal 67, 7381. Ross, H. (1986) Potato Breeding Problems and Perspectives. Advanced Plant Breeding Supplement 13. Paul Parey Verlag, Berlin. Schmiediche, P. (1997) Overview of CIP work in East and Southeast Asia and the Pacic. In: International Potato Centre Program Report 19951996. International Potato Centre, Lima, Peru, pp. 1517. Schneider, K. and Douches, D.S. (1997) Assessment of PCR-based simple sequencing repeats to ngerprint North American potato cultivars. American Potato Journal 74, 149160. Schuler, K., and Hoekstra, R. (1997) Kartoffelgenbanken in Deutschland und der EU. Kartoffelbau 48, 122123. Simmonds, N.W. (1995) Potatoes. In: Smartt, J. and Simmonds, N.W. (eds) Evolution of Crop Plants, 2nd edn. Longman Scientic & Technical, Harlow, pp. 466471. Sosinski, B. and Douches, D.S. (1996) Using polymerase chain reaction-based DNA amplication to ngerprint North American potato cultivars. HortScience 31, 130133. Spooner, D.M. and Bamberg, J.B. (1991) Prole: the Inter-Regional Potato Introduction Project (IR-1), US center for potato germplasm. Diversity 7, 3235.

200

R. Ortiz Tanksley, S.D., Ganal, M.W., Prince, J.P., de Vicente, M.C., Bonierbale, M.W., Broun, P., Fulton, T.M., Giovanonni, J.J., Grandillo, S., Martin, G.B., Messeguer, R., Miller, J.C., Paterson, A.H., Pineda, O., Roder, M.S., Wing, R.A., Wu, W. and Young, N.D. (1992) High density molecular linkage maps of the tomato and potato genomes. Genetics 132, 11411160. Tarn, T.R., Tai, G.C.C., De Jong, H., Murphy, A.M. and Seabrook, J.E.A. (1992) Breeding potatoes for long-day temperate climates. Plant Breeding Reviews 9, 217332. Thiele, G., Gardner, G., Torrez, R. and Gabriel, J. (1997) Farmer involvement in selecting new varieties. Experimental Agriculture 33, 275290. van Soest, L.J.M., Dambroth, M. and Lamberts, H. (1984) The GermanNetherlands potato collection. Plant Genetic Resources Newsletter 58, 1115. Watanabe, K.N., Orrillo, M., Iwanaga, M., Ortiz, R., Freyre, R. and Perez, S. (1994) Diploid potato germplasm derived from wild and land race genetic resources. American Potato Journal 71, 599604. Waugh, R., Glendinning, D.R., Duncan, N. and Powell, W. (1990) Chloroplast DNA variation in European potato cultivars. Potato Research 33, 505513. Zimmerer, K.S. and Douches, D.S. (1991) Geographical approaches to crop conservation: the partitioning of genetic diversity in Andean potatoes. Economic Botany 45, 176189.

11
1Institut

The State of the Use of Cassava Genetic Diversity and a Proposal to Enhance it

G. SECOND1 AND C. IGLESIAS2

de Recherche pour le Dveloppement (IRD, ex-ORSTOM), 911 avenue Agropolis, BP 5045, 34032 Montpellier cedex 1, France; 21400 Woodland, West Lafayette, IN 47906, USA

Introduction
Cassava is a perennial arbustive plant cultivated for its tuberous roots and, together with its wild relatives, comprises the widely diverse genus Manihot, found across the tropical zones of the American continent. Cassava ranks fth in the production of world food crops, after maize, rice, wheat and potato. In 1998, the production of fresh roots was estimated at 166 million t (Mt) produced in 92 countries. The largest production was in Africa (51%), followed by Asia (29%) and South America (19%). Five countries produce more than 15 Mt each: Nigeria, Brazil, Thailand, Democratic Republic of Congo and Indonesia (FAO, 1998a). The areas of production are in the tropical zones where annual rainfall exceeds 600 mm and the mean temperatures 13C (Raffaillac, 1996). Cassava thrives in light and drained soil, rich in potassium. Nitrogen tends to favour the development of aerial parts, and the presence of endomycorrhiza favours phosphorus nutrition. Cassava tolerates a low soil pH (Howeler, 1991) and long dry seasons (El-Sharkawy, 1993). These characteristics confer adaptation to marginal areas where cereal crops are in decline, as in southern Africa. Cropping and production systems based on cassava are very diverse. The major part of world production is obtained under traditional cropping systems, with little or no inputs and frequent intercropping practices. Agroindustrial production complexes are rare but cash cropping is also practised. Cassava can produce large quantities of dry matter under optimal conditions: the total biomass at 12 months can reach 45 t in the south of Cte dIvoire, for example. The potential annual yield has been estimated at 30 t of dry matter per hectare (Cock, 1985) and some newly bred cultivars come close to this yield (Raffaillac, 1996). Farmers often grow cassava after other staple food crops because it can produce in soils where other plants cannot. Cassava production is mainly for human consumption. Only in Asian countries such as Thailand, Indonesia, China and Vietnam is most of the crop produced for
IPGRI/FAO 2001. Broadening the Genetic Base of Crop Production (eds H.D. Cooper, C. Spillane and T. Hodgkin)

201

202

G. Second and C. Iglesias

animal feed and industry. International trade accounts for only 11% of the world production. Thailand alone produces 81% of these exports, Indonesia 10% and China 6%. Some cultivars have sweet esh in the roots and can be consumed directly after boiling, while others have a bitter esh and need a specic process to become edible (McKey and Beckerman, 1996). The bitterness is due to cyanhydric acid liberated, after a wound, from cyanoglucosides contained in variable amounts according to the cultivar and environment. Various processes, known from antiquity by the Amerindians and transmitted to Africans, allow the detoxication of the roots of high cyanogenic cultivars. Used separately or in combination, processes such as peeling, crushing, slicing or grating, water expressing, decanting, sun or smoke drying or frying, fermenting by soaking in water, heaping, stacking and boiling or steaming can reduce cyanhydric acid, which is water soluble and volatile. These processes lead to a variety of products for direct consumption or for short- or long-term storage. Cassava produces a starch appreciated by nutritionists for its excellent digestibility. Cassava roots are low in proteins. Young leaves, rich in proteins, vitamins and minerals are consumed in northeast Brazil as well as in some West and Central African countries. Following recent studies in Brazil, they are now also added as a dry powder to infant foods (Motta et al., 1994). Research activity in cassava has been relatively scarce, although early work on the selection of clones of cassava dates to 1899 in the Brazilian state of Bahia (Zehntner, 1919). Various breeding programmes were started in the 1930s in Indonesia, Africa, Madagascar, India and Brazil, including the use of wild species. However, with a few exceptions, new cultivars have not had success comparable with improved cultivars of other major food crops. Yet diseases, epidemics and changes in cropping systems imposed upon farmers by the scarcity of arable lands in some areas add to the challenges that could be addressed by new varieties. Recent work shows that enormous diversity exists for cassava even in a single eld in the Amazon basin, and that new cultivars arise from genetic recombination and germination in the same elds. This chapter reviews the state of cassava genetic diversity available for use, and, based on these observations, proposes new ideas in the management of cassava diversity. Also, some new hypotheses about the basis of cassava domestication and the importance of genetic diversity among the wild Manihot relatives of cultivated cassava are addressed.

An Overview of the Genetic Diversity and Resources Available Within the Genepools of Cassava
The whole genepool of cassava is included in the genus Manihot. This genus, distributed naturally only in the New World tropics, is part of the family Euphorbiaceae, which is one of the largest in the Dicotyledons, falling within subfamily Crotonoideae, which also includes Hevea. The native range of Manihot extends from southern Arizona (USA) to northern Argentina. The species of genus Manihot are rather sporadic in their distribution and never become dominant members of the local vegetation. Most of them are found in areas with a long dry season but some of them are adapted to rain

Cassava Genetic Diversity and its Enhancement

203

forests or their margins, usually proliferating in openings of the forest, whether manmade or naturally caused. They are all perennial but sensitive to frost, and thus their area of distribution does not exceed 2000 m in altitude. They are also found, however, in sub-tropical areas where frost is not rare but predictable. Manihot species from these sub-tropical ecosystems become dormant during winter months, as other species do during the dry season in tropical areas. Most of the species are found on acidic soil but some can be found on lime-derived soils. The limits of the genus Manihot are well dened. On the other hand, the limits of the various species within the genus are very controversial. The last available published review (Rogers and Appan, 1973) reduced the number of species from 171 to 98, on the basis of mainly vegetative characteristics. They are grouped into 17 sections, one of which includes the cultivated species only. Cassava is found most closely related to two sections: the central American section Parvibracteatae, which includes M. aesculifolia (considered as the most closely related wild species); and the South American section Heterophyllae which includes most of the wild forms now assembled by Allem (1994) under the species M. esculenta s.l. The main centre of diversity of the Manihot genus is found in Brazil, with 77 species out of 80 South American species. Southwest Mexico appears as the other centre of diversity for the 17 Central American species. Another classication is proposed in a monograph in preparation (A.C. Allem, personal communication). Of the 77 species identied by Rogers and Appan (1973) for Brazil, this new study recognizes only 38, as well as describing a few new species. The lack of correspondence between the two classications suggests an enormous difculty in dening a clear line of discontinuity within the genus. This is a case where molecular markers may help to clarify the situation. Apart from the cassava that was rst introduced by the Portuguese from the east coast of Brazil to West Africa, several wild species were introduced to many Old World tropical areas as potential rubber crops (Serie, 1989). In particular, M. glaziovii adapted well and subsisted in many places including Africa, India and Indonesia as an ornamental or shade tree (Rogers and Appan, 1973). Evidence of its spontaneous introgression into cultivars and the emergence of spontaneous wild hybrid forms was suggested not only in Brazil but also in Africa (see page 204). The rst attempt at molecular characterization of the genus Manihot was based on Southern blot restriction fragment length polymorphism (RFLP) analysis of chloroplast and nuclear ribosomal DNA (Bertram, 1993). Representative sets of the Central and North American species were used, but only one wild species from South America: M. carthaginensis from Colombia. Cassava was also included. The results showed: (i) a relatively low level of nucleotide divergence for the more divergent species (0.1% for chloroplast DNA); and (ii) that the major divergence found was between M. carthaginensis and cassava on one hand, and the Central/North American species on the other hand. They rejected Rogers and Appans suggestion of M. aesculifolia being the species most directly related to cassava, and indirectly pointed to South America as the probable area of domestication of cassava. A survey of a few South and Central American species by Colombo (1997) using the random amplied polymorphic DNA (RAPD) technique also agreed with those ndings. Studies including amplied fragment length polymorphism (AFLP) (Roa et al., 1997) and microsatellites (Roa et al., 1998) characterization of cassava across four

204

G. Second and C. Iglesias

South American species (wild M. esculenta, M. tristis, M. carthaginensis and M. brachyloba) and M. aesculifolia conrmed the view of Allem (1994) that cassava is closely related to what he considers as wild M. esculenta. A small proportion of molecular markers, however, differ between the wild and cultivated forms of M. esculenta. The detailed results are consistent with the hypothesis that ancestors of cassava can be found within the Brazilian group of Manihot species (Allem, 1994). Another AFLP survey (Second et al., 1997; Second, 1998) examined all Brazilian species along with cassava and a few other South and Central American species. It demonstrated or conrmed several points in relation to the genetic structure and species relationship within Manihot (Fig. 11.1): the wild forms of M. esculenta are the most closely related to cassava but do not explain all of its diversity by descent. Most Brazilian species cluster together and are difcult to individualize. Several presumed spontaneous hybrids were conrmed among them, even between species considered

Fig. 11.1. Distribution in the plan dened by the rst two eigenvectors of a principal coordinate analysis of the simple matching coefcient matrix of similarities in pairwise comparison of 364 accessions representative of the genus Manihot, including cassava. From the bulk of the molecular diversity of genus Manihot, the rst axis extracts cassava and its direct wild ancestor, while the second axis extracts the glaziovii group. Presumed hybrids are found between cassava and glaziovii, with not only tree cassava but also some cultivars.
e = cassava (the same genotypes as in Fig. 11.2); other symbols = wild species or hybrids; G = M. carthaginensis s.l. or wild glaziovii group; T = tree cassava as found in northeast Brazil; F = M. esculenta ssp. abellifolia and peruviana; * = 217 accessions representing 53 other species, mostly originated in Brazil. Source: Second et al. (1997).

Cassava Genetic Diversity and its Enhancement

205

morphologically as the most distant. M. carthaginensis sensu lato (including M. glaziovii) is most divergent from cassava, among the Brazilian species. It forms, however, spontaneous hybrids with cassava, including some forms recognized as M. glaziovii, or tree cassava or even cultivars. This last point is in line with earlier observations made at morphological and isozyme levels on cassava, tree cassava and M. glaziovii (Lefvre, 1989; Wanyera et al., 1994), and validates the hypothesis of a contribution of M. glaziovii not only in the domestication of cassava but also in its adaptation to the Old World, in particular Africa where it met the cassava mosaic disease (CMD), which has particular virulence. In this regard, it is interesting to note that M. aesculifolia is also morphologically related to M. carthaginensis (A.C. Allem, personal communication), although widely divergent at molecular level. A pioneer phylogenetic study based on the non-coding sequence of a single-locus gene further supports the hypothesis that domestication occurred from the wild forms of M. esculenta originating from the border of the Amazonian basin forest rather than from the cerrado area (Olsen and Schall, 1999). Earlier work (that cannot be reviewed in detail here for lack of space available) indicates that in Manihot, interspecic hybrids occur frequently and can be obtained relatively easily. In general, no strong reproductive barrier was evidenced between cassava and various wild species tested. Nevertheless, some failures in attempts to hybridize M. glaziovii with cassava have been noted in Brazil in the context of other successful hybridizations (Nassar et al., 1986). A partial pollen sterility was also observed in Africa among F1 hybrids of M. glaziovii with cassava (Lefvre, 1989). However, even in this case, where it appears likely that the wild stocks of M. glaziovii presented a strong reproductive barrier with cassava, this was gradually overcome in spontaneous as well as articial hybridization (the fact that many of the cassava trees around the world are relatively easy to reproduce through cuttings is by itself an indication that they are often hybrid derivatives with cassava, not pure M. glaziovii, even if sometimes considered as such). Clearly, the work on this line is incomplete, but available evidence favours the view that most wild Manihot species are part of the primary or secondary genepool of cassava.

Diversity in cassava: evidence from molecular markers Several reports have addressed the question of the overall genetic diversity of cassava, based mainly on arbitrary molecular markers such as RAPD and AFLP. Collections of both American and African cultivars have been studied, although generally in separate studies, making the comparison between the diversity found on the two continents difcult. Also, Asian cultivars have been covered only marginally in these studies. Colombo (1997; see also Emperaire et al., 1997, and Second et al., 1999) has conducted the most extensive study using RAPD and AFLPTM markers on a set of cultivars combining a representative set of Brazilian collections, a reduced core collection of accessions held in the Centro Internacional de Agricultura Tropical (CIAT) and a small collection extracted from a single traditional eld in the Amazon basin. In line with other studies, Colombos results showed a large amount of variation but little genetic structure. In other words, although such studies allow a comparison of the amount of diversity found in various areas of origin, and suggest certain grouping of cultivars, the

206

G. Second and C. Iglesias

Cassava Genetic Diversity and its Enhancement

207

diversity appears to be in complete continuum, with only little genetic linkage disequilibrium, even between extremes in the distribution. An amazing nding was, however, that a molecular diversity nearly equivalent to that described by Colombo (1997) was evidenced among genotypes extracted from a single eld in the Amazon basin. Moreover, the collection from this single eld showed, in comparison with the world collection, the largest (although relatively minor) divergence found within cassava. In other words, the diversity of cassava in the Amazon basin is large and is not well represented in the available world collection (Fig. 11.2). Further, a traditional cultivar characterized by a local name appears to be often composed of several genetically interrelated but different clones: a concept of a dynamic management of genetic diversity according to an Amerindian tradition emerges from this study (Emperaire et al., 1997). Work on this line is still underway, particularly in Amerindian elds of French Guyana, Guyana, Brazil and Ecuador, conrming these ndings while showing that an even more divergent diversity exists on the Guyanian plateau. A conrmation that this little characterized genetic diversity involves not only recombinational diversity but also alleles that are under-represented, or not represented in available collections comes from a study at the level of microsatellite loci that is underway in collaboration with CIAT. As expected, the few cultivars originated in Africa or Asia included in Colombos studies did not show any particular divergence from American cultivars. Interestingly, however, this remark also extended to one cultivar (NGA2) known to result from a cross with M. glaziovii and used in the construction of the available RFLP genetic map (Fregene et al., 1997). This cultivar clustered with a Colombian cultivar (Col22) and a few other cultivars mostly originated from northeast Brazil. This raised the possibility that M. glaziovii or another species closely related to it such as M. carthaginensis has also spontaneously contributed to the diversity of cassava in America (Second, 1998). Various studies have addressed the diversity of African cultivars on the basis of isozyme variation (Lefvre, 1989), RFLP (Beeching et al., 1993) and RAPD (Marmey et al., 1994). All show a large amount of variation comparable with that found among American cultivars. An isozyme marker presumably introgressed from M. glaziovii was found generally associated with resistance to CMD. However, this resistance was also found associated with various genetic backgrounds in a different study based on RAPD. It was thus presumed to depend on various combinations of genes (Mignouna and Dixon, 1997). The isozyme study showed that introgression from M. glaziovii in Africa has probably contributed to the diversity of cassava on this continent. At the level of chloroplast DNA only, an RFLP study suggests a relatively large cytoplasmic diversity within cassava, with three different cytotypes, as compared with the diversity found among wild species (Fregene et al., 1994). The conclusions from this overview are:

Fig. 11.2. A UPGMA dendrogram based on the simple matching coefcient of similarity for the presence or absence of 143 polymorphic fragments (60 AFLP and 83 RAPD) between 82 cassava plants. Forty cultivars are representative of a world collection (W) and 25 genotypes were distinguished among 42 plants collected in a single clearing plot in the Amazon basin (A). The two groups of cultivars form two different clusters, with a similar overall diversity, except for one landrace of the world collection group found among the single plot group. The origin of this landrace is also from the Amazon basin (Para State, Brazil). Source: Emperaire et al., 1997.

208

G. Second and C. Iglesias

A direct wild ancestor for cassava has been identied. However, domestication included some interspecic events that have not yet been properly characterized. Genetic diversity in cassava is high but with little linkage disequilibrium. A high genetic diversity can be found in a single Amerindian eld and a traditional cultivar appears to be a family of clones. In spite of the vegetative multiplication generally used to propagate cassava, the domestication process appears to have included: (i) assembling a large diversity in the same eld; and (ii) numerous genetic recombination events. No clear bottleneck in the transfer of genetic diversity from America to Africa has been evidenced. De novo selection of original cultivars in Africa appears to have also included introgressions from the wild species M. glaziovii. The complete analysis of cassava genetic diversity in America and of its transfer to other continents has not yet been thoroughly accomplished, and there is indication that not all the available diversity is yet assembled in the world collections. The genus Manihot represents a wide genepool, with most if not all species part of the primary or secondary genepool of cassava.

Extent of Genetic Diversity Utilized to Date

A considerable proportion of the total effort to collect, preserve and characterize Manihot genetic resources must have a user-oriented focus. The benet to nal users of Manihots genetic resources is through crop genetic improvement and the diffusion of cultivars that combine desirable agronomic and quality traits. It is extremely important to have farmers participation from the onset of germplasm collections, in order to gather most of the information related to traits they consider desirable in the germplasm they possess and characteristics that they judge to be missing or in need of some improvement (Gulick et al., 1983; Fukuda and Guevara, 1996). Hershey (1994) estimated that close to 10% of the cassava germplasm accessions held at CIAT have been used in the recombination programme to generate genetic diversity that can be used as the basis for breeding programmes. At the same time, he stated that the number that actually has contributed to elite genotypes for release to farmers was considerably less, perhaps around 1%. Organized breeding for cassava is relatively recent, and it has not had the success that breeding has had in other crops. A great percentage of cassava production is still based on landraces (Hershey, 1987). But landraces themselves are dynamic: each genotype has a relatively short life in production, and is then substituted by other introduced material, or selection from native seedlings. In most national programmes the initial step in the formulation of a breeding programme is an extensive germplasm collection, to catalogue and evaluate what there is in the country. In large countries (such as Brazil) it is common that landraces introduced from other areas are well adapted and become new cultivars in other areas (e.g. Rosa from Sergipe, recently released in Cear). In countries where there is considerable variability in the eld, we should encourage this type of approach, before resorting to the introduction of foreign germplasm (such as improved populations from international research centres). Africa, which produces more than 50% of world production (mainly for human consumption), was struck early by CMD, the effects of which were so devastating that

Cassava Genetic Diversity and its Enhancement

209

all the national breeding programmes concentrated on obtaining resistance to it. The programmes of the colonial powers (Belgium, France, UK) since the 1930s, and then of the national programmes promoted the collection and exchange of landraces between several agronomic research stations. However, a stage of global exchange was not reached until the international centres (International Institute for Tropical Agriculture (IITA) and CIAT) started to develop and interchange their cassava germplasm collections integrated in their breeding efforts in partnership with the national centres (Dixon et al., 1994; Porto et al., 1994). The use of genetic diversity from some Manihot wild species was promoted in the 1930s in Africa to circumvent the lack of good resistance to CMD in cultivars. Several species were used, including M. glaziovii. Germplasm derived from the former East African breeding programme was for a long time the main source of resistance being used (Jennings, 1976; Jennings and Hershey, 1985). Additional useful genetic variability has been detected in the wild relatives of cassava, although it has not been used extensively (Nassar et al., 1986; Bonierbale et al., 1997). IITA has used several Manihot species in interspecic hybridization to transfer their desirable genes into cassava (Asiedu et al., 1994). Spontaneous sexual and asexual polyploids from both intra- and interspecic crosses have also been tested for their value as cultivars. These crosses look very promising and represent a new avenue to enlarge the genetic base of cassava cultivars (Hahn et al., 1990, 1994; Sreekumari et al., 1995). We suggest that this potential for polyploidy in cassava should be put in perspective with various observations that facultative apomixis can be found in interspecic cassava hybrid derivatives (Nassar et al., 1998). In nature, apomixis is generally associated with polyploids (Asker and Jerling, 1992). When facultative, that is when there subsists a small percentage of sexual reproduction, apomixis still allows genetic recombination and breeding to take place in farmers elds. There is an inherent risk in the reduction of genetic diversity through the adoption of a single dominant cultivar in an area where cassava represents the basic staple. For example, in the early 1970s, bacterial blight nearly decimated many elds in the Congo that were cultivated with a single traditional cultivar, and created a temporary acute food shortage. On the other hand, not only bacterial blight but also mealybugs and green mites might have been introduced in Africa through importation of cuttings. There are about 28,000 accessions of cassava held in ex situ collections worldwide (Rao et al., 1995; FAO, 1998a). Most are in eld genebanks, though IITA and Brazil also hold substantial in vitro collections. Most accessions are of landraces. However, the number of discrete landraces may be substantially less than this gure suggests because of recent losses of some collections and duplication between collections: for example, Bonierbale et al. (1995) reported 11% duplication within the CIAT collection. Regional estimates of cassava accessions include:

Costa and Morales (1994) reported a total of 7500 accessions maintained in Latin America; 55% corresponded to Brazilian germplasm and 40% to Colombian accessions. In the case of Africa, Msabaha (1994) reported 847 landraces maintained by national programmes in East and southern Africa; Benett-Lartey (1994) reported 2860 for West and Central Africa; while Ng et al., (1994) referred to 350 African accessions maintained at IITA.

210

G. Second and C. Iglesias

Tan (1994) reported a total of 1110 local landraces being maintained by different national programmes in Asia.

The demand for genetic diversity by breeders still centres on traits such as root yield potential, root dry matter content, earliness, yield stability and reaction to pests and diseases (Fukuda, 1994). Recently, there has been an increasing interest in root quality traits, post-harvest deterioration, sprouting ability under adverse conditions, photosynthetic capacity and nutrient use efciency. The demand for genetic diversity has been determined by the needs of farmers, processors and consumers. Genetic diversity studies conducted by different scientists show considerable variability among cassava accessions for most of the traits of interest. However, the case of resistance to CMD, derived from an introgression of M. glaziovii, and the potential for nding novel genetic variants for starch quality within wild germplasm calls for a greater future need to concentrate on the sections of wild Manihot species that are the most promising. Not all the accessions maintained in cassava germplasm collections have been properly characterized for traits of present importance. This should be a high priority, in order to clarify the needs for further collection, use of wild relatives and resorting to genetic transformation. Even when morphological characterization of cassava germplasm accessions has been completed, basic standardization in the scale of measurement and growing conditions is often lacking. A group of 13 basic descriptors has now been dened (Fukuda and Guevara, 1996). Core collections (Iglesias et al., 1995) provide a more manageable sample of the world cassava collection that can be evaluated across different ecosystems and years in order to determine the genotype environment effects for important traits. However, core collections should continually be improved to make them more representative of global cassava genetic diversity (Emperaire et al., 1997). The agronomic evaluation of germplasm accessions in different ecosystems has resulted in the selection, multiplication and release of cultivars from other regions. Introduction has been one of the most ancient breeding methods, and it has the advantage of dealing with materials that have already passed through the lter of farmers selection in other places. Agronomic evaluation usually leads to the selection of accessions with excellent performance for certain traits, but also with weaknesses for other traits. Therefore, after the initial evaluation phase, there is a recombination process to generate segregating progenies from which individuals combining the best of both parents can be selected. The development of these segregating populations constitutes one of the most important steps to link cassava germplasm collection held at international centres or large national programmes with other programmes working in particular regions. The diffusion of genetic diversity from such large programmes to others represented a much larger movement of germplasm in the last decades of the 20th century than had occurred in the previous 450 years. Table 11.1 presents a summary of cassava cultivars released in Latin America and Asia, and their relationship to accessions in the germplasm collection at CIAT and/or landraces. We can observe that all cultivars can be traced back to a given accession. There have not been more than three or four cycles of selection and recombination for most of the active cassava breeding programmes that started in the early 1970s. The importance of the germplasm collections held at international centres and national pro-

Cassava Genetic Diversity and its Enhancement

211

350 300 Planted area (1000 ha) 250 200 150 100 50 0 R1 R3 R60 R90 K50 R5

Fig. 11.3. Area planted with the traditional variety (Rayong 1) and alternative new varieties released in the late 1980s and early 1990s in Thailand.

grammes can be pictured through the example of Thailand where, until the mid-1980s, there were more than 1.5 million ha planted to a single cultivar (Rayong 1). Nowadays, farmers have at least three options in terms of improved cassava cultivars, and the area under cassava cultivation has diversied its genetic base (Fig. 11.3), reducing its vulnerability (Kawano et al., 1998). However, it may be stressed that Thailand has relatively quickly adopted new improved cultivars, because cassava is mostly a commercial crop to be exported for animal feed and not a basic staple food in this country. In the case of Nigeria, the emerging possibilities to commercialize cassava associated with advanced processing have been recognized as the main driving force in the adoption of high-yielding cultivars (Nweke et al., 1994). This conrms that farmers are more prone to adopt new cultivars when they have accessible and uent market channels for their product other than its consumption at the family/community level. In a breeding programme, clearly not only the quantity but also the quality of the end-product is important. It is paramount to take into consideration the cropping systems as practised by the farmers. The relative importance of single crop or mixed cropping should be taken into consideration. Also, the importance of the aerial part of the plant should not be overlooked, since the success of the following cropping season depends also on the quantity and quality of stem cuttings. Assistance to the farmer in relation to propagation material is rarely provided in a sustainable way. Also, it should be kept in mind that, in various cases, consumption of the leaves can substantially complement the quality of the diet based on cassava. The extent of available genetic diversity in the future will be broadened by the use of wild genetic resources and the incorporation of genetic transformation protocols. Those two alternatives will help address those traits for which there is restricted genetic diversity (e.g. starch quality) or none available (e.g. resistance to stem borers). In spite of the fact that a very restricted percentage of the total available genetic diversity has been used up to the moment, present work being developed in relation to genetic distances, genome complementation and detection of genomic regions with large effects

Table 11.1. CIATs cassava germplasm utilization and impact. Original name or pedigree MCOL 1684 MCOL 1468 MMEX 59 MPAN 51 MCOL 1468 MCOL 1684 MCOL 1468 MCOL 1684 MCOL 1468 MCOL 1505 MMEX 59 MCOL 1468 MCOL 1684 MCOL 2215 Year of release 1993 1981 1981 1981 1982 1982 1983 1983 1984 1984 1986 1988 1989 1992 1986 1987 1990 1990 1990 1991 1993 1993 1994

212

Country

Name

Main features High yield Early maturity, high yield Drought tolerance, high yield Disease resistance, acid soil adaptation High yield, resistant to CBB High yield, resistant to CBB Early maturity, casabe quality Casabe quality Early maturity, high yield High yield, Diplodia resistance High yield Table use, high yield High yield High starch content High yield High yield High yield High starch content, CBB resistant High yield, CBB resistant High yield, starch, Diplodia resistant High yield and starch content High yield High starch

a) Landrace provided by CIAT Philippines VC 5 Cuba CMC-40 Mexico Costea Mexico Sabanera Dom. Rep. M Col 1468 Dom. Rep. M Col 1684 Haiti Mdme Jacques Haiti M Col 1684 Colombia Manihoica P-11 Colombia Manihoica P-12 Brazil M Mex 59 Philippines VC 2 Philippines M Col 1684 Ecuador Portoviejo 650

b) CIAT clones introduced to national programmes Philippines VC 1 MCOL 22 MMEX 59 China Nanzhi 188 MCOL 22 MVEN 270 Panama Dayana Colombia ICA-Catumare MCOL 655A MCOL 1515 Colombia ICA-Cebucan (MCOL 1438 MCOL 647) (MCOL 638 Colombia ICA-Costea MMEX 11 MCOL 65 Colombia ICA-Negrita MCOL 22 (MCOL 655A MCOL 1515) Vietnam KM-60 Rayong 60 (MCOL 1684 R1) Vietnam KM-94 Kasetsart 50 (R1 R90)

G. Second and C. Iglesias

MVEN 218)

Cassava Genetic Diversity and its Enhancement

c) Clones selected from CIAT seed introductions Thailand Rayong 3 MMEX 55 MEVEN 307 Thailand Rayong 2 MCOL 113 MCOL 22 Malaysia Perintis (MCOL 22 MVEN 270) MCOL 1684 Brazil CNPMF 8339/11 (MVEN 185 MCOL 22) (SM 7666 MVEN 218) Brazil CNPMF 8347/19 (SM 76-66 MCOL 638) (MCOL 655A MCOL 1515) Philippines VC 3 (MCOL 638 MCOL 655A) MCOL 22 Malaysia MM-92 (CM 477-3 MCOL 1684) (MMEX 55 MPAN 114) Indonesia Malang 1 CM 4049-2 (CM 105-19 CM 849-1) Indonesia Malang 2 CM 4031-10 (CM 922-2 CM 507-37) Philippines VC 4 CM 4014-3 (CM 728-3 CM 681-2) d) Clones selected from local CIAT crosses Thailand Rayong 60 Thailand Rayong 90 Thailand Sriracha 1 Thailand Kasersart 50 Thailand Rayong 5 Vietnam Rayong 60 Rayong 1 MCOL 1684 MCOL 1505 V43 MKU2-151 (R1 CM 305) Rayong 1 (MCOL 1505 V43) CMR25-105-112 (27-77-10 R3) 21-1 (CMC 76 V43)

1984 1985 1988 1989 1989 1990 1992 1992 1992 1993 1987 1990 1991 1992 1994 1994 1986 1989 1990 1990 1992 1995 1995

High starch content Good for snack food High yield on peat soils High yield High yield Dual purpose Early maturity, high yield Drought tolerance High starch High yield High yield, early maturity High starch content High starch High yield and starch content High yield, early harvest High starch High starch yield Cold tolerance Root rot tolerance Earliness, intermediate root rot Tolerance Root rot resistance High starch

e) Clones selected from local crosses or landraces with CIAT assistance Indonesia Adira 4 M31 China SC-124 E-24 Brazil Mae Joana Local landrace Berreiruinha (AM) Brazil Zolhundinha Local landrace Irancuba (AM) Brazil EMBRAPA 8 Local landrace Urucar (AM) China SC-8002 ZM8002 China SC-8013 ZM8013 High yield Total 46 cultivars

Source: CIAT, Cassava Program, Breeding and Economics Section les (1993, 1995).

213

214

G. Second and C. Iglesias

upon traits of large importance will boost the use of germplasm that was overlooked from its phenotypic evaluation, as has happened in other crops (Tanskley and McCouch, 1997). There will always be a trade-off between the need to gather the most representative sample of cassava germplasm before it disappears, and the need to make the best use of available genetic resources in collections that are expensive to maintain. Cassava genetic resources conservation and utilization is a multi-disciplinary task, which should involve scientists (germplasm specialists, breeders, pathologists, entomologists, food quality specialists, bio-technologists, etc.) as well as nal users. From that concept we derive the importance of networking in the collection, preservation and use of cassava genetic resources.

Networks for the Conservation and Use of Cassava Genetic Resources

There are a series of formal and informal networks related to the conservation and use of cassava genetic resources operating in the three continents and other networks with global coverage (Table 11.2). Most of the networks operating in Africa have a formal structure supported by nancial resources from special projects (Benett-Lartey, 1994; Msabaha, 1994), facilitated by IITA and furthering their objective of promoting the evaluation, adaptive selection and diffusion of improved cassava germplasm. In the case of Latin America, most of the networks represent potential resources more than effective institutional set-ups. Most of them have depended on CIAT in terms of operational resources and coordination, and lack formal support from donors (Iglesias, 1994). In the case of Asia, the network is informal. While there is considerable support from CIAT in terms of coordination, there are several strong national programmes that can take the lead in coordinating activities in the near future (Kawano, 1995). The Cassava Genetic Resources Network, which was initially sponsored by the International Plant Genetic Resources Institute (IPGRI), included three key elements in its plan of action: (i) the implementation of an international database on cassava genetic resources; (ii) a worldwide strategy for rational and safe conservation of the germplasm, using available techniques; and (iii) collaborative activities for better use of cassava genetic resources (Perret, 1994). For a crop that lags behind in terms of resources invested for research and development, it is very important to have a network such as this actively working.

A Proposal to Enhance Genetic Base-broadening of Cassava

This proposal is for enhancing the dynamic conservation of cassava, both of wild species and traditional cultivars. Such a dynamic conservation would be associated with cassava breeding, in particular through the evaluation of the useful diversity in agronomic research institutions but also largely through farmer-participatory breeding as initiated by IITA in Nigeria (Robinson, 1996, p. 379). It would complement and, if successful, perhaps partly substitute for the present efforts to exchange, conserve and use germplasm in cassava genetic improvement. The nal expected product is to preserve, in a simple yet sustainable way for future generations, a large genetic basis for the

Cassava Genetic Diversity and its Enhancement

Table 11.2. Cassava research networks operating in Latin America, Africa and Asia. Region Latin America Network Cassava Research and Development Network Panamerican Cassava Breeders Network Southern Cone Network East and Southern Africa Root Crops Research Network (ESARRN) Collaborative Group on Root and Tuber Improvement and Systems Research (CORTIS) International Society for Tropical Root Crops African Branch (ISTRCAB) Cassava Research and Development Users Perspective with Agricultural Research and Development (UPWARD) Cassava Genetic Resources Network (CGRN) Cassava Biotechnology Network (CBN) Type of network Informal Informal Informal Formal Formal Formal Informal Formal Formal Formal Secured nancing No No No Yes Yes No No Yes No Yes

Africa

Asia Global

215

216

G. Second and C. Iglesias

improvement of cassava from both traditional cultivars and wild species. The scheme includes considerable activities related to intercontinental exchange or introduction of germplasm, making such material available for evaluation in the eld and ready for immediate use in breeding initiatives. From a practical as well as a theoretical point of view, there are several aspects that make the genus Manihot and cassava good candidates for dynamic conservation:

Optimum conditions for the ex situ conservation of seeds are not yet known for any of the species, even though most species are apparently orthodox in relation to seed conservation. While in vitro conservation of plantlets is successful to conserve the cultivars, it is expensive and has limited possibilities for present application with most wild species. As evidenced by the difculty in classifying the genus and difculties in recognizing limits between species, extensive interspecic hybridization takes place. Up to the end of the 20th century, the domestication process of cassava included: (i) gathering a large genetic diversity in the same eld; (ii) permanent selection of new genotypes arising through genetic recombination; and (iii) introgression from wild species widely different from the direct ancestor.

Model for dynamic conservation of Manihot species The basic model consists of grouping in an arbitrary order accessions per species, and growing them in the eld. Each species is relatively isolated from the others by a minimum of 30 m, but pollination is not controlled. It is justied to apply the model in different ways according to the biome of the collections maintenance site, as follows. An optimal dynamic conservation in the area of origin of the species concerned should not induce excessive genetic recombination but rather conserve the species as they are. Only a relocation of the populations to save them from destruction is advisable. Pooling various populations in a single articial one can be justied for practical reasons and to save them from disappearing in the wild, rather than to produce new genotypes. Entries pooled should be collected in a relatively small and ecologically homogeneous area. In this case, it will be relatively easy to decide the limits of the species. When some accessions die, it will be preferable to replace them from the same or similar accessions rather than from reintroducing their offspring obtained in exotic locations. At the other extreme, when introducing the species to an exotic location (different ecology, different continent, etc.) accessions may be pooled from their total area of origin. In this case, the limit between the species is much less well dened. The admixture of such groups of accessions from a large geographic area will lead to large genetic recombinations. Such recombinations should actually boost the chances of successful adaptations to a new environment as exemplied in the most successful cases of domestication of plants. It is possible to imagine a sustainable, yet relatively simple scheme of dynamic conservation as follows: (i) a network for initialization of the dynamic conservation in the area of origin of the species (dened sensu stricto), with grouping of entries for each species found in a certain biome but not introducing accessions from other biomes;

Cassava Genetic Diversity and its Enhancement

217

and (ii) a network of dynamic conservation plots in various areas of the tropical world, with species (sensu lato) represented by entries originated from their whole area of distribution. On a long-term basis, periodic exchanges of seeds between these various populations from exotic locations in the world should mimic natural geneow between populations in the wild and maintain a large diversity in each population. This scheme of metapopulation (see Olivieri et al., 1990; Goldringer et al., Chapter 13, this volume) should realize, in a sustainable way, the conservation of the genus Manihot in the long term, coupled with its evaluation and utilization. In September 1994, a test of a dynamic conservation of wild Brazilian species of Manihot (Mendes et al., 1998; Second et al., 1998) was initiated in the elds of CENARGEN (Centro Nacional de Pesquisa de Recursos Genticos e Biotechnologia), with focus on the wild species most closely related to cassava (Allem, 1999) and/or to the most threatened species of the Cerrado area. The success of the chosen scheme will be evaluated in the long term. The monitoring of genetic changes occurring in the articial populations during dynamic conservation, as well as in farmers participatory breeding, should be largely simplied with the rapid development of new technologies that allow the quantication of the amount of hybridization of a template DNA to a very large number of probes (or oligonucleotides set up in micro-arrays or on a DNA chip). It will become possible to bulk many plants in a single sample for analysis and to obtain the information on allelic frequencies at many loci in a single experiment.

Managing an intraspecic Manihot esculenta dynamic conservation We propose that two aspects of a dynamic conservation of cultivars could be applied: conservation ex situ as seeds in partial substitution of in vitro conservation and dynamic on-farm conservation of cultivars. Because we conrmed that the use of sexual reproduction is frequent in the maintenance of traditional cultivars, we propose that most of them could be conserved through seeds. Conservation of seeds represents a dynamic conservation since the original gene combinations of the cultivar are modied. It should be efcient in conserving the useful genetic diversity, if not the genotypes. Also, since a dynamic management of the diversity of cassava cultivars is at the origin of the present stage of the domestication of cassava, it is natural to think of ways in which this tradition could be maintained in order to contribute signicantly to the conservation of cassava genetic diversity. Dynamic conservation has been approached so far mostly for grain crops (see Enjalbert et al., 1998; Goldringer et al., Chapter 13, this volume) and it should be noted that there is a fundamental difference in a dynamic conservation approach between root and grain crops. In grain crops, the cycle of repeated sowing and reaping can act as a selective drive towards high yield because the propagule is the object of the crop. In the case of root crops, however, it is likely that a repeated cycle of planting and reaping will not preserve most of the attributes of the cultivars. To this aim, it should be necessary that a control of the quantity and the quality of the harvested roots be performed at each cycle and that a selection be imposed. Genotype environment interactions would in this way be evaluated in a variety of cases, which is paramount in a crop such as cassava with high levels of morphological plasticity

218

G. Second and C. Iglesias

(Raffaillac and Second, 1997). Then, encouraging farmer-participatory breeding appears as the most appropriate option for the conservation, evaluation and utilization of cassava genetic resources. This would apply to both the continent of origin of cassava particularly the Amazon basin and northeast Brazil and the areas of major present production humid tropical Africa and Asia.

Acknowledgements
The authors are grateful to two anonymous reviewers and to Dr Jean-Pierre Raffaillac for reading the manuscript and making many valuable suggestions.

References
Allem, A.C. (1994) The origin of Manihot esculenta Crantz (Euphorbiaceae). Genetic Resources and Crop Evolution 41, 133150. Allem, A.C. (1999) The closest wild relatives of cassava (Manihot esculenta Crantz). Euphytica 107, 123133. Asiedu, R., Hahn, S.K., Vijaya Bai, K. and Dixon, A.G.O. (1994) Interspecic hybridisation in the genus Manihot progress and prospects. Acta Horticulturae 380, 110113. Asker, S.E. and Jerling, L. (1992) Apomixis in Plants. CRC Press, Boca Raton, Florida. Beeching, J.R., Marmey, P., Gavalda, M.-C., Noirot, M., Haysom, H.R., Hugues, M.A. and Charrier, A. (1993) An assessment of genetic diversity within a collection of cassava (Manihot esculenta Crantz) germplasm using molecular markers. Annals of Botany 72, 515520. Benett-Lartey, S.O. (1994) The status of cassava genetic resources programmes in some West and Central African countries. In: Iglesias, C. and Iwanaga, M. (eds) International Network for Cassava Genetic Resources. IPGRI, Rome, pp. 3340. Bertram, R.B. (1993) Application of molecular techniques to genetic resources of cassava (Manihot esculenta Crantz. Euphorbiaceae): inter-specic evolutionary relationships and intra-specic characterization. PhD dissertation. University of Maryland, USA. Bonierbale, M., Iglesias, C. and Kawano, K. (1995) Genetic resources management of cassava at CIAT. In: MAFF International Workshop on Genetic Resources: Root and Tuber Crops. Tsukuba, Japan. pp. 3952. Bonierbale, M., Guerava, C., Dixon, A.G.O., Ng, N.Q., Asiedu, R. and Ng, S.Y.C. (1997) Cassava. In: Fuccillo, D., Sears, L. and. Stapleton, P. (eds) Biodiversity in Trust. Cambridge University Press, Cambridge, pp. 120. Cock, J.H. (1985) Cassava: New Potential for a Neglected Crop. IADS-CIAT-GTZ series, Westview Press, Boulder, Colorado. Colombo, C. (1997) Etude de la diversit gntique de maniocs amricains (Manihot esculenta Crantz) par les marqueurs molculaires (RAPD et AFLP). PhD thesis, ENSAM, Montpellier. Costa, I.R.S. and Morales, E.A.V. (1994) Cassava genetic resources in South America. In:. Iglesias, C. and Iwanaga, M. (eds) International Network for Cassava Genetic Resources. IPGRI, Rome, pp. 1620. Dixon, A.G.O., Asiedu, R. and Hahn, S.K. (1994) Genotypic stability and adaptability: analytical methods and implications for cassava breeding for low-imput agriculture. Acta Horticulturae 380, 130137.

Cassava Genetic Diversity and its Enhancement

219

El-Sharkawy, M.A. (1993) Drought-tolerant cassava for Africa, Asia, and Latin-America. Bioscience 43(7), 441451. Emperaire, E., Pinton, F. and Second, G. (1997) Gestion dynamique de la diversit varitale du manioc en Amazonie du Nord-Ouest. Natures, Sciences, Socits 6(2), 2742. Enjalbert, J., Goldringer, I., David, J. and Brabant, P. (1998) The relevance of outcrossing for the dynamic management of genetic resources in predominantly selng Triticum aestivum L. (bread wheat). Genet. Sel. Evol. 30(suppl. 1), S197S211. FAO (1998) FAOSTAT. http://apps.fao.org/debut.htm FAO (1998) The State of the Worlds Plant Genetic Resources for Food and Agriculture. Food and Agriculture Organization, Rome. Fregene, M.A., Vargas, J., Ikea, J., Angel, F., Thome, J., Asiedu, R.A., Akoroda, M.O. and Roca, W.M. (1994) Variability of chloroplast DNA and nuclear ribosomal DNA in cassava (Manihot esculenta Crantz) and its wild relatives. Theoretical and Applied Genetics 89, 719727. Fregene, M.A., Angel, F., Gomez, R., Rodriguez, F., Roca, W.M., Tohme, J. and Bonierbale, M. (1997) A genetic linkage map of cassava (Manihot esculenta Crantz). Theoretical and Applied Genetics 95, 431441. Fukuda, W.M.G. (1994) Cassava breeders needs for genetic diversity. In: Iglesias, C. and Iwanaga, M. (eds) International Network for Cassava Genetic Resources. IPGRI, Rome, pp. 141144. Fukuda, W.M.G. and Guevara, C.L. (1996) Taller sobre manejo de recursos geneticos de yuca Manihot esculenta Crantz. Memorias de la Reunin sobre Recursos Genticos de Yuca. Cruz das Almas, Bahia, Brazil 2328 October 1995. Working Document, CIAT-CNPMF. Gulick, P., Hershey, C.H. and Esquinas-Alcazar, J. (1983) Genetic Resources of Cassava and Wild Relatives. IBPGR, Rome. Hahn, S.K., Bai, K.V. and Asiedu, R. (1990) Tetraploids, triploids, and 2n pollen from diploid interspecic crosses with cassava. Theoretical and Applied Genetics 79, 433439. Hahn, S.K., Bai, K.V., Chukwuma, Asiedu, R., Dixon, A. and Ng, S.Y. (1994) Polyploid breeding of cassava. Acta Horticulturae 380, 102109. Hershey, C.H. (1987) Cassava genetic resources. In: Hershey, C.H. (ed.) Cassava Breeding: a Multidisciplinary Review. CIAT, Cali, Colombia, pp. 124. Hershey, C.H. (1994) Manihot genetic diversity. In: Iglesias, C. and Iwanaga, M. (eds) International Network for Cassava Genetic Resources. IPGRI, Rome, pp. 111134. Howeler, R.H. (1991) Long-term effect of cassava cultivation on soil productivity. Field Crops Research 26, 118. Iglesias, C.A. (1994) Interfase Entre los Programas de Mejoramientos, los Campos de los Agricultores y los Mercados de la Yuca en Latinoamrica. Documento de Trabajo 138. CIAT, Cali, Colombia. Iglesias, C., Bonierbale, M., El-Sharkawy, M., Lozano, C., Bellotti and Wheatley, C. (1995) Focusing basic research for cassava variety improvement. In: Howeler, R.H. (ed.) Cassava Breeding, Agronomy Research and Technology Transfer in Asia. CIAT, Bangkok, pp. 4060. Jennings, D.L. (1976) Cassava, Manihot esculenta (Euphorbiaceae). In: Simmonds, N. (ed.) Evolution of Crop Plants. Longman, London, pp. 8184. Jennings, D.L. and Hershey, C.H. (1985) Cassava breeding: a decade of progress from international programmes. In: Russell, G.E. (ed.) Progress in Plant Breeding, Vol 1. ButterworthHeinemann, London, pp. 89116. Kawano, K. (1995) Green revolution and cassava breeding. In: Howeler, R.H. (ed.) Cassava Breeding, Agronomy Research and Technology Transfer in Asia. CIAT, Bangkok, pp. 355367. Kawano, K., Narintaraporn, K., Narintaraporn, P., Sarakarn, S., Limsila, A., Limsila, J., Suparhan, D., Sarawat, V. and Watananonta, W. (1998) Yield improvement in a multistage breeding program for cassava. Crop Science 38, 325332.

220

G. Second and C. Iglesias Lefvre, F. (1989) Ressources gntiques et amlioration du Manioc, Manihot esculenta Crantz, en Afrique. TDM ORSTOM no.57, Paris. Marmey, P., Beeching, J.R., Hamon, S. and Charrier, A. (1994) Evaluation of cassava (Manihot esculenta Crantz) germplasm collections using RAPD markers. Euphytica 74, 203209. McKey, D. and Beckerman, S. (1996) Ecologie et volution des produits secondaires du manioc et relations avec les systmes traditionnels de culture. In: Hladik, C.M. et al. (eds) Lalimentation en fort tropicale Interactions bioculturelles et perspectives de dveloppement. Vol. 1: Les ressources alimentaires: production et consommation. UNESCO, Paris, pp. 165202. Mendes, R.A., Allem, A.C. and Second, G. (1998) Estabelecimento de uma coleo dinmica de Manihot (Euphorbiaceae). Pesquisa em Andamento 12, 14. Mignouna, H.D. and Dixon, A.G.O. (1997) Genetic relationships among cassava clones with varying levels of resistance to ACMV using RAPD markers. In: Thro, A.M. and Akoroda, M.O. (eds) Proceedings of the Third International Scientic Meeting, Cassava Biotechnology Network, Kampala, Uganda, 2630 August 1996. African Journal of Root and Tuber Crops, Vol. 2, Nos 1 and 2, pp. 2832. Motta, J. da S., Fukuda, W.M.G., Souza, L.C.B. and Costa, Z.M.F. (1994) A farinha da folha da mandioca uma alternativa como complemento alimentar. Mandioca em Foco 4, 2. Msabaha, M.A.M. (1994) Status of cassava genetic resources programmes: priorities in East and Southern Africa for cassava genetic resources network. In: Iglesias, C. and Iwanaga, M. (eds) International Network for Cassava Genetic Resources. IPGRI, Rome, pp. 2332. Nassar, N.M.A., Da silva, J.R. and Viera, C. (1986) Hybridaao interespecica entre mandioca e especies silvestres de Manihot, Cincia e Cultura 38(6), 10501055. Nassar, N.M.A., Viera, M.A., Viera, C. and Grattapaglia, D. (1998) A molecular and embryonic study of apomixis in cassava (Manihot esculenta Crantz). Euphytica 102(1), 913. Ng, N.Q., Asiedu, R. and Ng, S.Y.C. (1994) Cassava genetic resources programme at IITA. In: Iglesias, C. and Iwanaga, M. (eds) International Network for Cassava Genetic Resources. IPGRI, Rome, pp. 7176. Nweke, F.I., Hahn, S.K. and Ugwu, B.O. (1994) Circumstances of rapid spread of cultivation of improved cassava varieties in Nigeria. Journal of Farming Systems Research-extension 4(3), 93119. Olivieri, I., Couvet, D. and Gouyon, P.H. (1990) The genetic of transient populations: research at the metapopulation level. Tree 5, 207210. Olsen, K.M. and Schaal, B. (1999) Evidence on the origin of cassava: phylogeography of Manihot esculenta. Proceedings of the National Academy of Sciences USA 96, 55865591. Perret, P.M. (1994) Introductory remarks. In: Iglesias, C. and Iwanaga, M. (eds) International Network for Cassava Genetic Resources. IPGRI, Rome, pp. 12. Porto, M.C.M., Asiedu, R., Dixon, A. and Hahn, S.K. (1994) An agroecologically-oriented introduction of cassava germplasm from Latin America into Africa. Acta Horticulturae 38, 118129. Raffaillac, J.P. (1996) La fertilit en zone tropicale humide et le manioc. In: Pichot, J. et al. (eds) Sminaire sur la fertilit du milieu et les stratgies paysannes sous les tropiques humides. CIRAD, Montpellier, pp. 286298. Raffaillac, J.-P. and Second, G. (1997) Le manioc. In: Charrier, A., Jacquot, M., Hamon, S. and Nicolas, D. (eds) Lamlioration des plantes tropicales. Collection Repres. CIRAD et ORSTOM, pp. 429456. Rao, V.R., Debouck, D.G. and Iwanaga, M. (1995) The role of international organizations in root and tuber crops conservation. In: MAFF International Workshop on Genetic Resources: Root and Tuber Crops. Tsukuba, Japan. pp. 721. Roa, A.C.M., Maya, M., Duque, M.C., Tohme, J., Allem, A.C. and Bonierbale, M.W. (1997) AFLP analysis of relationships among cassava and other Manihot species. Theoretical and Applied Genetics 95, 741750.

Cassava Genetic Diversity and its Enhancement

221

Roa, A.C., Chavarriaga-Aguirre, P., Duque, M.C., Maya, M.M., Bonierbale, M.W., Iglesias, C. and Tohme, J. (2000) Cross-species amplication of cassava (Manihot esculenta) (Euphorbiaceae) microsatellites: allelic polymorphism and degree of relationship. American Journal of Botany 87, 16471655. Robinson, R.A. (1996) Return to Resistance: Breeding Crops to Reduce Pesticide Dependence. AgAcess, Davis, California, and IDRC, Ottawa. Rogers, D.J. and Appan, M. (1973) Flora Neotropica. Monograph No. 13. Manihot Manihotoides (Euphorbiaceae). Hafner Press, New York. Second, G. (1998) Manihot glaziovii contributed to the genetic make-up of cassava and represents an example of dynamic conservation and on-farm breeding of genetic resources. Proceedings of CBN IV. Salvador da Bahia, November 98. Second, G., Allem, A.C., Emperaire, L., Ingram, C., Colombo, C., Mendes, R.A. and Carvalho, J.C.B. (1997) Molecular markers (AFLP) based Manihot and cassava genetic structure analysis and numerical taxonomy in progress. Implications for their dynamic conservation and genetic mapping. African Journal of Root and Tuber Crops 2(1 and 2), 140147. Second, G., Colombo, C., Mends, R.A. and Berthaud, J. (1998) A scheme for a dynamic conservation of the genetic resources of wild Manihot and cultivated cassava in America and Africa and its extension to yams. Paper presented for publication in the Proceedings, regional workshop for the conservation and utilization of cassava, sweetpotato and yam germplasm in sub-saharan Africa. 1113th November 1997 at ILRI Nairobi, Kenya. Organised by: CIP, IPGRI, IITA. Second, G., Raffaillac, J.P. and Colombo, C. (1999) Manioc. In: Hamon, P., Seguin, M., Perrier, X. and Glaszmann, J.C. (eds) Diversit gntique des plantes tropicales cultives. Collection Repres, CIRAD, pp. 271308. Serie, J.-B. (1989) Historico da disseminacao da manicoba fora do Brazil. In: Anais do primeiro encontro nordestino de manioba, 1820 X 1988. Colecao Mossoroense: Serie C (BRA), Recife (BRA): SUDHEVA, IPA, vol. 469, 131 p. Sreekumari, M.T., Abhraham, K. and Jos, J.S. (1995) The potential of triploids in cassava improvement. Cassava Newsletter 16(2), pp. 67. Tan, S.L. (1994) Cassava genetic resources programmes in Asia. In: Iglesias, C. and Iwanaga, M. (eds) International Network for Cassava Genetic Resources. IPGRI, Rome, pp. 4155. Tanskley, S.D. and McCouch, S.R. (1997) Seed banks and molecular maps: unlocking genetic potential from the wild. Science 277, 10631066. Wanyera, N.M.W., Hahn, S.K. and AkenOva, M.E. (1994) Introgression of ceara rubber (Manihot glaziovii Muell-Arg) into cassava (M. esculenta Crantz): a morphological and electrophoretic evidence. Proceedings of the 5th Symposium of ISTRC-AB, pp. 125130. Zehntner, L. (1919) Estudo sobra algumos variedades de mandiocas Brasileiras. Sociedad Nacionale de Agricultura, Rio de Janeiro.

12

The State of the Use of Musa Diversity

S.L. SHARROCK1, J.-P. HORRY2 AND E.A. FRISON1

Network for the Improvement of Banana and Plantain, Parc Scientique Agropolis II, 34397 Montpellier cedex 5, France; 2CIRADFLHOR, Dpartement des Productions Fruitires et Horticoles, BP 153, 97202 Fort de France cedex, Martinique
1International

Importance of Musa
Bananas1 are cultivated throughout the tropical regions of the world, where they constitute a major staple food crop for many hundreds of millions of people as well as providing a valued source of income through local and international trade. They are grown over a harvested area of approximately 10 million ha, with an annual production of around 88 million tonnes (Mt), of which around one-third is produced in each of the African, Asia-Pacic, and Latin America and Caribbean regions. In terms of gross value of production, they are the fourth most important food crop in developing countries after rice, wheat and maize. The vast majority of producers are small-scale farmers growing the crop either for home consumption or for local markets. As well as being a cheap and easily produced source of energy, they are also rich in vitamins A, C and B6. Furthermore, with growing urbanization, bananas and plantains are becoming increasingly important as cash crops, often providing the main source of income for rural populations, thus playing an important role in poverty alleviation. In Africa, bananas and plantains provide more than 25% of food energy requirements for around 70 million people. In the East African Highlands region alone, around 15 Mt are produced annually and it is here that bananas reach their greatest importance as a staple food crop. In countries such as Uganda, Burundi and Rwanda, annual per capita consumption of bananas has been estimated at more than 250 kg; the highest in the world. The largest producer of bananas in the world is Uganda, which produces over 10 Mt of the crop annually. Bananas for export account for little more than 10% of global production. They are grown predominantly in Latin America and the Caribbean where the cultivation of banana and plantain is of importance to the food and income security of nearly 200 million people. Even in this region, however, bananas produced for export constitute only around 30% of total production.
1 The term banana is used throughout the text to refer to all types of bananas including cooking bananas and plantains.

IPGRI/FAO 2001. Broadening the Genetic Base of Crop Production (eds H.D. Cooper, C. Spillane and T. Hodgkin)

223

224

S.L. Sharrock et al.

Musa Genetic Resources

Wild species The family Musaceae consists of two genera: Musa and Ensete. The genus Ensete is distributed from Africa to New Guinea, and no more than eight species have been reported (Stover and Simmonds, 1987). The genus is not of economic importance on a global scale, although Ensete ventricosum is locally very important as a staple food crop in parts of southern Ethiopia. Edible starch is extracted from the corm and pseudostem and fermented to make kocho, an important food source for several million people in the region (Demeke, 1986). The genus Musa is divided into four sections, the members of which include both sexual (seeded) and asexual types. Two of the sections contain species with a basic chromosome number of 10 (2n = 20) (Callimusa and Australimusa), while the species in the other two sections (Eumusa and Rhodochlamys) have a basic chromosome number of 11 (2n = 22). The species in the sections Callimusa and Rhodochlamys are of ornamental interest only and do not produce edible fruit. The section Australimusa contains Musa textilis (Abaca) from which Manila hemp is produced, and it is within this section that the seedless, and therefore edible, Fei bananas, found mainly in the Pacic islands, have been selected and propagated by humans. Virtually all banana cultivars arose from the Eumusa group of species. This section is the biggest in the genus and the most geographically widespread, with species being found throughout Southeast Asia from India to the Pacic Islands. The section contains some 11 species (Horry et al., 1997). Most cultivars are derived from two species, Musa acuminata (A genome) and Musa balbisiana (B genome). However, Shepherd and Ferreira (1982) identied cultivars derived from hybridizations with M. schizocarpa (S genome) in the Musa germplasm collection in Papua New Guinea, and this was subsequently conrmed by Carreel et al. (1993). In addition, a Philippine clone (Butuhan) is considered to be the result of an ancient hybrid between M. balbisiana and M. textilis (T genome), and landraces containing the three genomes acuminata, balbisiana and textilis have been found in Papua New Guinea (Carreel, 1995). Musa acuminata is the most widespread of the Eumusa species, being found throughout the range of the section as a whole. Chromosome structural changes, which have occurred spontaneously, or as a result of recombination events, have resulted in the development of natural reproductive barriers within the species, causing sub-species divergence and genetic diversity in the species as a whole. Nine subspecies have been recognized (Horry et al., 1997). The centre of diversity of the species is thought to be either Malaysia (Simmonds, 1962) or Indonesia (Nasution, 1991; Horry et al., 1997). The majority of species in the section Eumusa produce separate male and female owers on the same inorescence. These owers are produced asynchronously and pollination occurs exogenously. In a few cases (M. schizocarpa, M. acuminata ssp. banksii, M. acuminata ssp. errans), hermaphrodite owers are produced and in these species the level of heterozygosity is low less than 10%. Even in the non-hermaphrodite subspecies, the heterozygosity rate averages only 25% (Horry et al., 1997). M. balbisiana is distributed through the seasonally dry, northern and eastern fringes of the Eumusa area. Although originally thought to be a genetically homoge-

Musa Diversity

225

nous species, recent work using PCR-based techniques indicates that the species may in fact be more heterozygous than previously suggested (Jarret and Gawel, 1995).

Cultivars The export banana trade is almost completely dependent on only one variety: Cavendish. On the other hand, a great deal of morphological variation exists within the cultivated, non-export bananas, and at least 500 varieties are thought to exist worldwide (Purseglove, 1988). The great amount of genetic variation in cultivated bananas has arisen from three sources: sexual recombination within and between the original parents (M. acuminata and M. balbisiana), ploidy-level differences and somatic mutations. Although details of the early cultivation and spread of banana cultivation remain a matter for conjecture, bananas are considered to have been among the rst plants to be domesticated by humans. In Oceania, the earliest agriculture has been dated at around 8000 BC, and it is in this region that the Fei bananas are of some importance. Their early cultivation has been described as proto-agriculture: that is, they were gathered from the wild rather than planted (Price, 1995). The domestication of Eumusa bananas is thought to have occurred at around the same time in Southeast Asia (Simmonds, 1962) and it has been suggested that the earliest uses of these plants may have been non-food, or at least not involving the fruit. The plants would have produced a bre, which could be used in shing nets, and leaves for constructing shelters. In addition, various parts of the plant apart from the fruit are edible, and the male bud is still widely used in parts of Asia as a vegetable. Edibility of mature fruits of diploid Musa acuminata (AA) came about as a result of two mutation events, resulting in female sterility and parthenocarpy, and such edible types would no doubt have been selected and maintained by humans. Triploid AAA cultivars arose from these diploids, perhaps as a result of crosses between edible diploids and wild M. acuminata sub-species, resulting in a wide range of AAA genotypes. In most parts of Southeast Asia these triploids, which are more vigorous and have larger fruit, have replaced the original AA diploids. However, in New Guinea, AA diploids remain agriculturally signicant and a wide range of diversity is still found in cultivation. This last remaining area of AA diploid diversity is, however, now under threat due to the increasing cultivation of high-yielding triploid varieties. Many of Papua New Guineas primitive diploids were collected in the late 1980s during a series of collecting missions funded by the then International Board for Plant Genetic Resources (IBPGR). The material collected is now established in germplasm collections and is available to the Musa research community. The diploid and triploid acuminata cultivars were taken by humans to areas where M. balbisiana is native, and natural hybridizations resulted in the formation of hybrid progeny with the genomes AB, AAB and ABB. The Indian subcontinent is thought to have been the major centre for hybridization of acuminata types with the indigenous M. balbisiana, and the region is noted for the wide variety of AAB and ABB cultivars. M. balbisiana is considered to be more drought and disease resistant than M. acuminata, and such characteristics are often found in cultivars containing a B genome. Hybridization would have given rise to a wide range of edible types of banana, some of which would have survived and been multiplied under domestication. Consequently, a

226

S.L. Sharrock et al.

diverse selection of cultivars of Musa is thought to have arisen in Southeast Asia along with the earliest development of agriculture many thousands of years ago (Price, 1995). It is thought that the subsequent dispersal of edible bananas outside Asia was brought about solely by humans (Simmonds, 1962) and the history of banana cultivation is therefore closely linked to the early movement of human populations. This early dispersal of banana cultivars resulted in the development of distinct sub-groups of varieties in different geographic locations. Secondary diversication within the major groups of cultivated bananas is thought to have been the result of somatic mutations rather than sexual reproduction. Mutations affecting traits of economic or horticultural interest have been selected by farmers over the years and multiplied by vegetative propagation to produce morphotypes. Movement eastwards resulted in the development of a distinct group of AAB bananas cultivated throughout the Pacic Islands. These are known as the Maia Maoli/Popoulu group, and the progenitors of these bananas are thought to have been carried eastwards by proto-Polynesians from an area in or near the Philippines more than 4000 years ago (De Langhe, 1996). It is argued by some that bananas existed in South America in pre-Columbian times, and this is taken as evidence for early Polynesian contact with America (Langdon, 1993). A very distinct type of cooking banana (plantain) is found growing in the wet tropical zones of West Africa. Plantains are rare in most of Asia as well in other parts of Africa, and their origin in West Africa is shrouded in mystery. It is thought that they have been cultivated in this region for more than 3000 years, but the identity of the people responsible for such cultivation is unknown (De Langhe et al., 1996). It is possible that the same proto-Polynesians that carried the banana east to the Pacic islands also carried it west to Africa. As dry suckers can survive storage for several months, long sea voyages would not have been a problem. Such a hypothesis ts with the nding that plantains must have reached Africa more than 3000 years ago, but archaeological evidence for such voyages is unlikely to be found. Recent research suggests that tetraploid Musa hybrids, which evolved through sexual polyploidization, may have been the progenitor of the original plantains, rather than diploids as previously believed (Ortiz and Crouch, 1997). Molecular analysis of a large number of plantain landraces using random amplied polymorphic DNA (RAPD) indicates that most plantains have a very high level of genetic similarity to one another (Crouch et al., 1998). This supports the suggestion that somatic mutations are responsible for the diversity of morphotypes of plantains now found in West Africa. Another distinct group of bananas are found in the East African Highlands. These are thought to have been introduced between the 5th and 10th centuries, and a wide range of unique varieties now exists here. This area of secondary diversity is clearly the work of East Bantu-speaking people (De Langhe, 1996), but the origins of these bananas remain unknown. The occurrence of distinct sub-groups of cultivars in particular geographic locations is signicant in relation to accessing and using diversity. For example, plantain diversity is mainly found naturally in West Africa, although diversity in the Indian plantains (Nendran varieties) has been reported recently. Therefore, although it might be possible to introduce new diversity of this type of banana from other regions, the main way to produce new diversity in the plantain subgroup in West and Central Africa is through breeding. The same is true for the East African Highland bananas.

Musa Diversity

227

A review of the history of banana cultivation is provided by Price (1995) and further information on the number of varieties found within each of the major subgroups of bananas is given in Table 12.1. Studies on diversity in Musa germplasm have been carried out by several researchers. Isozyme analysis of banana and plantain cultivars has identied a large amount of variation in the AAB group. This is apparently due to the genetic heterogeneity of the progenitor species, particularly M. acuminata, over the wide geographic range in which the hybridizations took place (Lebot et al., 1993). These results support the evidence of Horry and Jay (1990) that suggested two independent centres of domestication for M. acuminata: one in Southeast Asia and the other in Papua New Guinea. Further details of studies on genetic diversity of wild and cultivated Musa are given by Careel (1995), Jarret and Gawel (1995), Jarret et al. (1992) and Lebot et al. (1993).

Opportunities and Constraints to Accessing Diversity for Breeding/Use

Availability of genetic resources The Asia Pacic region is the centre of diversity of the genus Musa, and activities related to the collecting and conservation of banana germplasm have historically been given priority by countries in the region. Collecting missions have already covered much of the region, and representative diversity of the material collected is being conserved in genebanks both within and outside the region. Widespread general collecting is therefore no longer considered necessary. However, targeted collecting in specic locations is still ongoing by national programmes in the region. At present, these include China, northeastern India, Indonesia, Pemba and Zanzibar. Other areas that have been identied as priority areas for germplasm collecting include the Andeman Islands and Sabah and Sarawak states in eastern Malaysia. Prospection in Malaysia would complement the efforts of Indonesia to collect, characterize and conserve the Musa germplasm from Kalimantan, the Indonesian part of Borneo where forest res annually destroy large areas of biodiversity, including Musa germplasm. In recent years, germplasm collecting in Papua New Guinea, Vietnam and China has been carried out in collaboration with the International Network for the Improvement of Banana and Plantain (INIBAP). As a result, duplicates of all distinct accessions collected have been, or will be, provided to INIBAP for inclusion in the INIBAP Musa germplasm collection. This collection is now the largest in vitro collection of Musa germplasm in the world and, following an agreement with the Food and Agriculture Organization (FAO), is held in trust and is freely available to bona de users worldwide. Guidelines for the Safe Movement of Musa Germplasm have been developed and published by FAO/IPGRI (Diekmann and Putter, 1996), and INIBAP has put in place the necessary virus-indexing capability to ensure that all germplasm distributed from the INIBAP Transit Centre (ITC) is in accordance with these guidelines. Breeding programmes thus have unlimited access to a wide range of Musa genetic resources for use in improvement programmes. This is particularly important in view of the fact that none of the major Musa breeding programmes is located in Southeast Asia, the centre of diversity of the genus.

228

Table 12.1. Review of available information on the extent and centre of diversity of the main sub-groups of bananas and plantains.a Genome Fei AA Subgroup Centre of diversity Pacic Islands PNG Malaysia Philippines East Africa Gros Michel Cavendish Red/Green Red Lujugira-Mutika China No. of clones/morphotypesb 2 main clones 1020 morphotypes 80+ 10 18 10 3 8 main clones 30+ morphotypes 2 9 main clones 140+ cooking morphotypes 80+ beer morphotypesc 23 3 1 45 12 23 ? 2 main clones 60+ morphotypes 1 main clone 10+ morphotypes 4 main clones 110+ morphotypes Valmayor et al., 1981 Additional references MacDaniels, 1947 Sharrock, 1995 Pascua, 1988

AAA

Houbin and Zehuai, 1997; Daniells, 1990 Karamura and Karamura, 1994

East Africa

AB AAB

Ney Poovan Pisang Raja Pisang Kelat Mysore Silk Pome Laknao Maia Maoli/ Popoulu Iholena Plantaind

India

India India India Philippines Pacic Hawaii West and Central Africa

S.L. Sharrock et al.

Lebot, 1994 Lebot, 1994 Swennen, 1990

Musa Diversity

ABB

Bluggoe Pisang Awak Monthan Kalapua

India Thailand India Papua New Guinea India Philippines Philippines N.E. India S.E. Asia/PNG

57 45 45 56? 10 ? 1012 2 1015

Valmayor et al., 1981

Sharrock, 1990 Singh and Uma, 1996 Valmayor et al., 1981 Singh and Uma, 1996

BBB AAAB, AABB, ABBBe

Saba

a The information provided in this table is taken exclusively from the available literature and it should not be regarded as denitive. It is merely an attempt to present, in one place, existing information. The table is derived mainly from Stover and Simmonds (1987); additional references are given in the table. b The terminology used by different authors in describing types of bananas is confusing. The terms cultivar, variety, clone, morphotype, type are used interchangeably. In this table we use the term clone to describe a main cultivar type and the term morphotype for the mutations derived from the main clone. The number of clones/morphotypes given in each subgroup is based on the best available information. The widespread occurrence of synonyms and duplications from country to country, as well as the high level of mutations in Musa, makes it extremely difcult to give accurate gures. In some cases, the different varieties described have not been divided into clones and morphotypes. c It is a common phenomenon for cooking cultivars to change to beer cultivars, especially in highland regions above 1400 m above sea level. d Although plantains are generally rare in Asia, several varieties, Nendran, are widely grown in India. e Naturally occurring tetraploid varieties are generally uncommon.

229

230

S.L. Sharrock et al.

Movement of germplasm Musa germplasm is also maintained in eld genebanks in many countries where it is readily available for characterization and evaluation studies (Horry, 1999). Such germplasm is also available for direct use by farmers within the country, but in many cases the lack of evaluation data on accessions in collections means that potentially useful cultivars have not been identied. Germplasm maintained in eld collections is not readily accessible to potential users outside the country due to the strict quarantine measures imposed by most producing countries. It is generally required that all germplasm introductions should be in accordance with the FAO/IPGRI Guidelines for the Safe Movement of Musa Germplasm to prevent the inadvertent introduction of pests and diseases, especially virus diseases. The guidelines require all material to be distributed in vitro, and only after completing virus indexing according to the recommended tests. Three virus-indexing centres are presently in operation. These are located in France at the Centre de Coopration Internationale en Recherche Agronomique pour le Dveloppement (CIRAD); Australia, at the Queensland Department of Primary Industry (QDPI); and South Africa, at the Plant Protection Research Institute (PPRI). The time required for virus indexing is around 1012 months per accession. Virus indexing is a major bottleneck in the distribution of material from the ITC, with some 38% of accessions at present unavailable for distribution as they still have not completed indexation or are infected with viruses. Effective virus therapy techniques remain under development and, until such techniques have been perfected, accessions that are currently virus infected cannot be made available.

Use of genetic resources in breeding Banana breeding rst started in the 1920s at the Imperial College of Tropical Agriculture (ICTA) in Trinidad, and soon after this a parallel programme was developed in Jamaica. Initial efforts were directed towards breeding for resistance to Panama disease (Fusarium wilt), already present in the Caribbean and to which the principle export clone, Gros Michel, was susceptible. The continuing spread of Panama disease and the ensuing destruction of Gros Michel plantations led to the initiation of a breeding programme in Honduras in 1959 by the United Fruit Company. Efforts to produce a Fusarium-resistant Gros Michel-type cultivar proved unsuccessful, and the naturally selected Cavendish variety replaced Gros Michel as the main export variety. The United Fruit Company decided to withdraw from genetic improvement research in 1984 and donated its programme to the Honduran Government. Nevertheless, Musa breeding continued and the programme is now maintained by the Fundacin Hondurea de Investigacin Agrcola (FHIA). A major constraint for Musa breeding programmes is the breeding behaviour of the crop. Most important cultivars, both for export and local consumption, are triploid, and triploidy leads to problems during meiosis due to uneven numbers of chromosomes. Consequently, Musa cultivars exhibit low seed set and poor seed germination. Indeed, some cultivars, such as the important export variety Cavendish, are almost completely sterile and can only be used with difculty in breeding. The complicated nature of breeding bananas, together with a lack of appreciation of its importance

Musa Diversity

231

as a staple food crop, has meant that relatively few funds have been directed by either the private or public sector to banana breeding. Indeed, until the 1980s there were effectively only two banana breeding programmes in existence (Honduras and Jamaica), both of which focused on improving bananas for the export market. The basic breeding procedure developed in the late 1920s in Trinidad and Jamaica (ICTA) relied on the use of natural diploids and later on the production of improved elite diploids, which were then used as male parents for crossing onto the desired triploid as the female parent in the production of improved tetraploid progeny. One limitation to this strategy is the low level of female fertility in desirable triploids. Cavendish, for example, being highly sterile, is a very poor female parent. While the quest to develop a dessert-type variety able to replace Cavendish continues, the advanced diploids are now also being used to great effect in the improvement of other types of bananas and plantains. Breeding work in Honduras now has a global focus and efforts are being made to produce plantains and other types of cooking bananas suitable for smallholder production in Africa and Asia as well as in Latin America. The basic breeding scheme developed by ICTA has been adopted by several new breeding programmes that have emerged, including those of the International Institute for Tropical Agriculture (IITA) in Nigeria and Uganda; CIRAD, France; Centre Rgional de Recherches sur Bananiers et Plantains (CRBP), Cameroon; and Empresa Brasiliera de Pesquisa Agropecuaria (EMBRAPA), Brazil. All these programmes use breeding schemes involving either natural diploids (wild seeded species or edible landraces) or improved diploids for the production of improved tetraploid hybrids. In addition, CIRAD has developed a new breeding scheme that has resulted in the synthesis of triploid varieties solely from natural or improved diploid varieties (Horry et al., 1997). Breeding using genetic transformation technology is also being carried out at a number of institutes, including the Katholieke Universiteit Leuven (KUL), Belgium; Queensland University of Technology (QUT), Australia; CIRAD-FLHOR, France; the Boyce Thompson Institute (BTI), USA; and the University of Hawaii, USA. Most of the main sub-groups of bananas are now being addressed by one or more of the breeding programmes (see Table 12.2). Despite the efforts of the various Musa improvement programmes, production for both the export market and local consumption still depends essentially on farmers varieties selected from naturally occurring germplasm. The absence, until very recently, of improved cultivars has been attributed to the difculties of breeding Musa at the genetic and practical levels (Ortiz et al., 1995). Genetic improvement has been severely hampered by the lack of useful genetic variability (i.e. in cultivars that are either male or female fertile) and low levels of female fertility (for crossing with improved diploids developed as male parents).

Applications of biotechnology in breeding Musa improvement programmes are making increasing use of a wide range of biotechnological tools in order to improve breeding efciency in bananas. A comprehensive review of this is provided by Crouch et al. (1998). Techniques such as embryo rescue are now used routinely to overcome some of the barriers to hybridization. Indeed, embryo rescue has increased seed germination rates by a factor of 310 at IITA, and

232

S.L. Sharrock et al. Table 12.2. Focus and aims of the main Musa breeding programmes (information obtained from Frison et al., 1997a). Breeding programme FHIA Honduras

Focus Export bananas Plantains Cooking bananas

Aims Sigatoka resistance Dwarf, disease resistance Dwarf, disease resistance Disease resistance (Sigatoka), high stable yield, wide adaptability, fruit quality, disease resistance (Fusarium), high stableyield fruit quality Disease and pest resistance Disease and pest resistance Disease and pest resistance Disease resistance (Sigatoka, Fusarium, Moko) Disease resistance (Sigatoka, Fusarium) Dwarf, disease resistance Fusarium resistance, dwarf, short cycle

Production area LAC, Asia, Australia West and Central Africa, LAC, India Asia, East Africa West and Central Africa

IITA Nigeria

Plantains

IITA Uganda CIRAD, France

AAAEA Highland bananas AAA export bananas

East Africa ACP countries, French West Indies West Africa, LAC, East Africa, Pacic West and Central Africa South America, Asia

AAA/AAB cooking/ dessert bananas CRBP, Cameroon EMBRAPA, Brazil Plantains AAB dessert bananas

Banana Board, AAA export bananas Jamaica IAEA, Vienna TBRI, Taiwan AAACavendish AAACavendish

LAC, Australia LAC/Asia Asia

ACP: Africa, Caribbean, Pacic; CIRAD: Centre de Coopration Internationale en Recherche Agronomique pour le Dveloppement; CRBP: Centre Rgionale de Recherches sur Bananiers et plaintains; EA: East African; EMBRAPA: Empresa Brasiliera de Pesquisa Agropecuaria; FHIA: Fundacin Hondurea de Investigacin Agrcola; IAEA: International Atomic Energy Agency; IITA: International Institute for Tropical Agriculture; LAC: Latin America and Caribbean; TBRI: Taiwan Banana Research Institute.

micropropagation has played a key role in the rapid propagation of large numbers of male and female parents for crossing blocks (Ortiz et al., 1995). Colchicine treatment has also been used extensively to double the chromosome number of selected diploid clones, and this forms part of the breeding strategy adopted by the CIRAD breeding programme (Horry et al., 1997).

Musa Diversity

233

A number of research groups are also working on Musa molecular genetics with the aim of developing marker-assisted breeding strategies. Genetic transformation Most banana breeding, both for export and local cultivars, is focused on the introduction of pest and disease resistance. Although sources of resistance to the major pests and diseases (Sigatoka diseases, Fusarium wilt and nematodes) have been identied, the introduction of such useful genes into sterile cultivars may only be possible through the use of genetic transformation techniques. Research in this area has progressed rapidly in recent years, and stable, genetically transformed banana plants have been produced (May et al., 1995; Sgi et al., 1995, 1999).

Extent of Diversity Utilized by Musa Breeding Programmes

Conventional breeding Diploids, both wild and cultivated, are widely used in breeding programmes as sources of resistance to diseases and pests. The wild species M. acuminata and M. balbisiana are generally used as male parents in crosses, particularly in the development of elite diploid hybrids for further crossing on to triploid cultivars as female parents. While resistance traits may be transferred from wild species, often these are transferred together with unfavourable agronomic features, such as poor bunch size and long fruiting cycles. Other wild Musa species have not been used by breeding programmes, although interspecic fertility is not necessarily a problem. It is known, for example, that M. schizocarpa hybridizes readily with M. acuminata in the wild, and this species has proved to be involved in the ancestry of some cultivars (Careel et al., 1993). Other species, such as M. textilis, have also been implicated in the domestication of bananas (Careel, 1995) but are not presently being used by breeding programmes. Cultivars are used in breeding mainly for their favourable agronomic characteristics.

Breeding for resistance Pests and diseases are the main constraints to banana production at the global level. Resistant varieties are the base upon which sustainable systems of production at the smallholder level can be developed. Genetic improvement is therefore aimed at the introduction of resistance. Black Sigatoka leaf spot disease caused by Mycosphaerella jiensis is considered to be the most serious constraint to banana production globally. This pathogen, to which many important and widely grown cultivars are susceptible, can cause severe leaf necrosis reducing yields by 3050% and losses of up to 100% of marketable quality fruit. The cost of controlling the disease in Costa Rica alone is estimated at $50 million annually and involves some 3850 fungicide-spraying operations. Such control measures are obviously beyond the means of the majority of smallscale producers.

234

S.L. Sharrock et al.

Fusarium wilt is a soil-borne disease which affects many important cultivars of banana and plantain. This disease has a long and destructive history. Caused by the fungus, Fusarium oxysporum f. sp. cubense, it almost destroyed the export trade, which was exclusively based on the susceptible cultivar Gros Michel, during the rst half of the 20th century. Gros Michel plantations were then replaced by Cavendish, a resistant dessert-type cultivar. Cavendish varieties are, however, susceptible to race 4 of the pathogen, which causes losses in subtropical areas such as Taiwan, Australia and South Africa, and a tropical race 4 has recently been reported in Southeast Asia. In addition, a complex of plant parasitic nematodes (Radopholus similis, Pratylenchus spp. and Helicotylenchus multicinctus) cause serious crop losses in all regions. Nematode infestations interfere with nutrient uptake and transport, resulting in slow growth, reduced fruit lling and susceptibility to wind lodging. Sources of resistance to these major pests and diseases have been identied, together with genes controlling a number of other traits. Genetic studies at IITA, for example, have resulted in the identication of genes controlling resistance to black Sigatoka and weevil borer, virus susceptibility, dwarsm, albinism, suckering behaviour, parthenocarpy and male/female fertility (Horry et al., 1997). However, identied genes for resistance are still generally considered to be from too narrow a base, and greater efforts are required in identifying further sources of resistance (INIBAP, 1998). The main source of black Sigatoka resistance originally utilized in breeding was Calcutta 4, a wild, non-edible diploid banana. A model to explain the inheritance of this resistance to black Sigatoka has been developed by Ortiz and Vuylsteke (1994a). The resistance in plantainCalcutta 4 hybrids appears to involve one major recessive gene and two additive minor genes with dosage effects (Craenen and Ortiz, 1997). However, the high level of resistance shown by Calcutta 4, although thought to have a polygenic basis, has already been overcome by the pathogen in the Pacic (Fullerton and Olsen, 1993). Alternative sources of resistance are obviously required and IITA have now identied more than 30 sources of resistance in their germplasm collection, most of them diploid M. acuminata, (AA) both wild and cultivated types (Vuylsteke et al., 1997). Different sources of resistance are now being incorporated in a single genotype by crossing tetraploid hybrids of Calcutta 4 parentage with other resistant diploids with the hope of generating durable host plant resistance. In relation to Fusarium wilt, although natural sources of resistance do occur in wild and cultivated Musa, different resistances are required for the different races of the pathogen. Segregation in progenies derived from crosses between three susceptible Musa spp. and the resistant cultivated diploid banana Pisang Lilin suggested in the latter the presence of a single dominant gene for resistance to Fusarium race 1. However, resistance to race 4 seems to be under polygenic control (Rowe and Rosales, 1993). Further sources of resistance to tropical race 4 are particularly required in view of the fact that the widely grown Cavendish cultivars are susceptible to this race of the pathogen, and no chemical control measures are available. Resistance to the nematode species R. similis has been identied in the cultivar Pisang Jari Buaya, and this has been used extensively by banana breeders. It appears that one or more dominant alleles control this resistance (Rowe and Rosales, 1993). Further sources of resistance to this pathogen as well as to the other nematode species affecting Musa are still required.

Musa Diversity

235

In all cases, the identication of sources of resistance requires time and money. The development of early screening tests, whereby seedlings or in vitro plantlets can be tested for resistance under laboratory or nursery conditions, will go a long way towards facilitating this work. This, along with increased characterization of existing collections, will assist breeding programmes in identifying germplasm with useful traits with which they can enlarge the number of cultivars or wild species utilized in the breeding of improved varieties. Recent reviews of Musa breeding activities are given by Ortiz et al. (1995), Bakry et al. (1997) and Ortiz and Vuylsteke (1994b)

Mutation breeding and somaclonal variation Spontaneous mutations have contributed to the genetic diversication of the Musa genepool, and the induction of mutations through the irradiation of in vitro cultures of banana is currently performed by several teams, including the International Atomic Energy Agency (IAEA; Austria), the Queensland Department of Primary Industry (QDPI; Australia) and Instituto de Biotecnologa de las Plantas (IBP; Cuba). Although few clones of interest have so far been produced, an early owering Cavendish clone has been selected in Malaysia from a mutant Cavendish clone produced through irradiation by the IAEA (Ortiz et al., 1995), while a Fusarium-resistant Gros Michel has been identied in Cuba (INIBAP, 1997a). Somaclonal variation, or genetic variation resulting from the in vitro process, can also be exploited in Musa breeding. Somaclonal variation, however, closely parallels natural variability (Israeli et al., 1995) and thus should not be considered as a source of new diversity. Nevertheless, a Fusarium race 4-resistant Cavendish clone has been selected by the Taiwan Banana Research Institute (TBRI) in Taiwan (Horry et al., 1997). Other examples of useful somaclonal variants are a Mons Mari (AAA) variant with long fruit (Smith and Drew, 1990) and a female fertile variant in plantain (Vuylsteke, 1991).

Possibilities for decentralized breeding The possibilities for developing decentralized breeding in plantains are presently being explored by IITA (Ortiz, 1997). Plantain-derived tetraploid hybrids are routinely crossed with diploid accessions for the production of triploid hybrids for evaluation. However, natural open pollination of the same tetraploid hybrids also produces viable seed, and the elite performance of certain hybrids generated in this way offers possibilities for breeding. Under the proposed programme, primary tetraploid germplasm would be crossed with selected diploids using controlled articial hand-pollination, or through polycrosses among selected parents. Synthetic populations derived from these crosses could then be released to other improvement programmes for evaluation and selection of promising clones. Such a programme would allow breeding programmes and national agricultural research systems (NARSs) early access to potentially elite material and, importantly, allow selection of promising lines under local agroenvironmental conditions.

236

S.L. Sharrock et al.

Diversity and the Smallholder Farmer

At the beginning of the 21st century, smallholder farmers throughout the tropics continue to cultivate a wide diversity of Musa germplasm. Up to very recently, there has been virtually no replacement of local cultivars by improved material from breeding programmes. Thus, throughout Africa and Asia a wide range of landraces is still being cultivated by farmers (Sharrock and Engels, 1997). Despite more than 70 years of banana breeding, it is only in the last few years of the 20th century that breeding programmes produced material that is considered to be superior to naturally occurring clones. Widescale, multilocational testing of improved varieties from breeding programmes rst started in 1991 with the establishment of the International Musa Testing Programme, coordinated by INIBAP and funded by the UN Development Programme. As a result of this programme, three improved hybrids from FHIA were identied as performing well over a range of environments and showing signicant levels of disease resistance. These were therefore recommended for further evaluation and testing at the national level, and since 1993 they have been distributed to more than 50 countries. In most countries, evaluation is carried out by the public sector, although in some instances non-governmental organizations (NGOs), universities and private companies may also be involved. One of the main constraints to the rapid distribution of new varieties to farmers is the slow rate of propagation of bananas under eld conditions. In addition, the cycling period of the crop means that evaluation over two to three cropping cycles takes a period of more than 2 years. In most countries, therefore, improved varieties are still undergoing evaluation and are not yet widely available to farmers. One exception to this is Cuba, where the necessary infrastructure has been put in place to allow the rapid mass-propagation of banana planting material using bio-factories. The Cuban government has put much effort into identifying new, disease-resistant varieties, introducing these into the country, mass propagating them and distributing them to farmers. By early 1998, around 8000 ha had been planted with a range of improved varieties: this represents some 7.6% of the total banana production area. These resistant cultivars are increasingly replacing black-Sigatoka-susceptible banana and plantain clones on the island. Despite the lack of introduced improved varieties, losses of diversity in farmers elds are undoubtedly occurring. Increasing trends towards commercialization, particularly for sale in expanding local and regional markets, in India and other Asian countries favour the establishment of single-cultivar plantations over mixed farming. In addition, the spread of diseases such as black Sigatoka into areas of smallholder production where chemical control measures are unavailable may result in the loss of the most susceptible cultivars. However, even in areas where farmers still cultivate a range of different varieties, there is a continuing need for diversity to be made available. The spread of pests and diseases into new areas results in a need for resistant germplasm. Likewise, increasing urbanization means that there is a growing demand for varieties suitable for sale in local markets. In many countries the lack of systems for the mass propagation and distribution of germplasm is a major bottleneck to the availability of new cultivars. Many countries possess germplasm collections, but incomplete characterization and evaluation of accessions in the collections means that potentially useful accessions have not been identied. Finally, the acquisition of germplasm from other countries is con-

Musa Diversity

237

strained by quarantine requirements, which usually makes the introduction of new varieties a costly and time-consuming exercise. An important consideration in the introduction of new diversity to farmers is the very specic needs of both producers and consumers in both pre- and post-harvest qualities of the crop. Needs vary considerably from place to place and these may not easily be addressed by acquiring improved germplasm from breeding programmes that operate at the global, or even regional level. A similar constraint exists with traditional varieties acquired from other areas. It is clear that the level of resources required for breeding Musa precludes the development of breeding programmes in every country. Furthermore, the reproductive nature of the crop means that farmer-participatory breeding schemes are much more difcult than with seed-propagated crops. One possible solution to these problems, however, could lie in the adoption of the type of decentralized breeding scheme described on page 235. Above all, mechanisms to allow the feedback of information from farmers to breeding programmes are essential, and farmer-participatory evaluation of germplasm can play an important role in this regard.

The Role of Networks

Networks at the global level The number of Musa improvement programmes in existence today is very small considering the scale of the problems to be addressed. In an effort to maximize the output and accelerate the impact of these limited Musa improvement efforts, PROMUSA, the Global Programme for Musa Improvement, was established in 1997 as a joint initiative of INIBAP and the World Bank. This programme was developed as an innovative mechanism to bring together research carried out both within and outside the CGIAR, creating new partnerships between NARSs and research institutes in both developing and developed countries. The programme specically aims to bring together, at the global level, all the major players in Musa improvement research. Within the framework of PROMUSA, a Musa genetic improvement working group has been established, bringing together all the major Musa breeding programmes, including those focusing on breeding using conventional hybridization techniques and those using mutation breeding and genetic engineering approaches. At the rst meeting of the Genetic Improvement Working Group, priority research needs were identied, which included the need for enhanced utilization of plant genetic resources by breeding programmes. It was noted that breeders need better access to existing collections for increased availability of Musa germplasm. Where gaps are identied in these collections, targeted exploration and collection of germplasm should be considered, with a particular focus on wild species. It was recognized that attention should be given to the whole range of Musa species, not just M. acuminata and M. balbisiana. To improve access to existing collections, it was recommended that more effort should be put into characterization and evaluation of the germplasm and that the ensuing databases/information should be exchanged between collections and breeders (Frison et al., 1997a). INIBAP is presently addressing this issue with the development of the Musa Germplasm Information System (MGIS). The software for this system has recently

238

S.L. Sharrock et al.

been completed, and curators of collections are now able to input data on the accessions in their collections. The aim is to collect together in a single format characterization and evaluation data for Musa germplasm in existing collections worldwide and make these available to interested parties. During the rst meeting of the Genetic Improvement Working Group, it was also agreed that molecular knowledge of the complex genome structure of bananas and plantains is essential in order to better understand Musa genetics. It was recognized that the lack of segregating populations is a major obstacle to rapid progress in the identication of molecular markers and their use in Musa breeding, and that molecular markers are necessary to increase the efciency and effectiveness of banana improvement efforts. Within the framework of PROMUSA, INIBAP has therefore recently initiated a project for the production of additional segregating populations and the identication of molecular markers using these. The various working groups that have been established within the framework of PROMUSA (Sigatoka, Fusarium, Nematodes, Viruses and Genetic Improvement) operate as networks, within which the exchange of information, germplasm (such as parental material from breeding programmes), etc., is facilitated. The networking approach encourages the development of collaborative projects and the creation of synergies. All network members participate in the identication of priorities for the group as a whole and are fully involved in the decision-making process. INIBAPs International Musa Testing Programme, which now operates within the framework of PROMUSA, also adopts a networking approach for the evaluation of germplasm. In this programme, NARS partners use a common evaluation format developed through a participative process to evaluate germplasm provided by a number of Musa breeding programmes. Improved material as well as potential breeding parents may be evaluated within this programme. The multilocational nature of the programme allows genotype environment effects to be studied in improved varieties. The inclusion of an evaluation mechanism in PROMUSA also provides the necessary opportunity for the feedback of information to the breeding programmes. The advantages of bringing interested parties together in a global programme such as PROMUSA to address specic research issues are many (Frison et al., 1997b). They include:

Development of new and innovative partnerships between programme participants, who may be advanced research institutes (ARIs), international agricultural research centres (IARCs), Universities, NARSs, NGOs, private sector. Global prioritization of research needs. Improved possibilities for funding for programme participants. Improved access to information and resources. Opportunities for interdependent research projects.

Networks at the regional level Regional banana research networks have been established in Latin America and the Caribbean (INIBAP-LACNET), Asia and the Pacic (INIBAP-ASPNET), eastern and southern Africa (BARNESA), and West and Central Africa (MUSACO). INIBAP provides the coordination and secretariat function for these networks, which have

Musa Diversity

239

proved to be powerful tools in priority setting and in the development of regional collaborative activities. Results and information are exchanged between members of the networks, thus facilitating the creation of synergies and minimizing the duplication of efforts within the region. Regional networks are also responsible for organizing meetings, training courses, workshops and seminars as appropriate for network members. In relation to the use of Musa diversity, the regional networks have an important role to play. They provide a forum for the exchange of information on Musa germplasm available within the region and outside. Collaborative projects and scientic exchange visits between network members allow national scientists to become familiar with varieties being cultivated in other countries in the region. INIBAP provides a global dimension, keeping each region in touch with developments in other regions. INIBAP also plays a service role in germplasm conservation and distribution. Through the regional networks, INIBAP is able to maintain close contact with national research activities and provide them with appropriate germplasm. Networking is an important means of ensuring efciency and effectiveness in germplasm evaluation and distribution activities. At the regional level, both IITA and CRBP collaborate with NARSs for the evaluation of germplasm. A key feature in the success of this type of collaboration is the development of horizontal relationships between the network partners (Ortiz and Vuylesteke, 1994b).

Networks at the national level Within the support of the regional networks, national partners are now beginning to develop national Musa research networks. In addition to research institutes and universities, countries are encouraged to include NGOs, farmers organizations and extension agencies in such networks, thus providing an important link with the ultimate end-user of germplasm, the farmer. The bringing together of users and extension agencies is a particularly important step in facilitating and encouraging the use of diversity. National Musa networks of varying levels of formality have now been established in a number of countries, in several cases as a direct result of encouragement and support from the regional network. Thailand provides a useful example of the development and benets of networking at the national level. A functioning national Musa network has been established and a coordinated National Banana Programme developed. Twelve institutes, with approximately 80 scientists, participate in banana research activities coordinated by the network. A Banana Working Group set up under the supervision of the Horticultural Research Institute complements the network. An example of the resulting increased activity is national funding for selection and improvement of two of Thailands most important banana cultivars for domestic consumption and trade (Kluai Kai and Kluai Namwa) (INIBAP, 1997b).

Conclusions and Recommendations

Although locally important in several countries, the use of diversity is still at an early stage in global banana and plantain production. This is true as much for the direct use

240

S.L. Sharrock et al.

of existing diversity by farmers as it is for attempts at genetic base-broadening in Musa improvement programmes. There has been relatively little movement of germplasm from country to country or region to region, and few varieties have had any impact outside their traditional areas of production. It is only in recent years that a system has been put in place for the safe movement of Musa germplasm and the necessary virusindexing capabilities established. In the past, the international movement of Musa germplasm was severely restricted by the high risks involved. Increased efforts in the following areas are required in order to facilitate and enhance the use of Musa germplasm:

Further evaluation of existing cultivars, particularly in relation to disease resistance, in order to identify material that may have potential for direct use by farmers. Increased characterization of material in collections and use of the MGIS in order to allow the rapid dissemination of information to potential users (breeders, extensionists, etc.). Greater knowledge of the resources available in the wild species is needed before breeders will be prepared to make use of them in breeding programmes. Tools, such as molecular markers, are required in order to reduce the time taken to breed new varieties. Virus indexing remains a bottleneck. More efcient indexing techniques, together with increased capacity, are required to speed up virus-indexing procedures. Virus therapy techniques are required in order to make accessions presently infected with virus available for international distribution. Increased, sustainable support to breeding programmes is essential to ensure the continued production of a diverse range of new hybrids for evaluation.

Finally, and perhaps more importantly, is public awareness. There is a continuing need to raise the awareness amongst scientists, policy makers and indeed the general public of the importance of bananas and plantains as a staple food crop. In the last few years of the 20th century, some of the barriers to hybridization in Musa were overcome and the rst improved varieties ever produced by breeding programmes have been released for widespread testing. However, many research needs remain and, particularly in this time of scarcity of resources, it is important that a crop with the global signicance of bananas and plantains does not continue to be neglected by researchers and donors alike.

References
Bakry, F., Careel, F., Caruana, M.-L., Cte, F., Jenny, C. and Tzenas du Montcel, H. (1997) Les bananiers. In: Charrier, A., Jacquot, M., Hamon, S. and Nicolas, D. (eds) Lamlioration des plantes tropicales. CIRAD and ORSTOM, France, pp. 109139. Carreel, F. (1995) Etude de la diversit gntique des bananiers (genre Musa) laide des marqueurs RFLP. PhD thesis, INA Paris-Grignon, Paris, France. Carreel, F., Faur, S., Gonzlez de Lon, D., Lagoda, P.J.L., Perrier, X., Bakry, F., Tezenas du Montcel, H., Lanaud, C. and Horry, J.-P. (1993) Evaluation de la diversit gntique chez les bananiers diplodes lIRFA-CIRAD. Fruits (numro spcial), 2540. Craenen, K. and Ortiz, R. (1997) Effect of the bs1 gene in plantainbanana hybrids on response to black Sigatoka. Theoretical and Applied Genetics 95, 497505.

Musa Diversity

241

Crouch, J.H., Vuylsteke, D. and Ortiz, R. (1998) Perspectives on the application of biotechnology to assist the genetic enhancement of plantain and banana (Musa spp.). Electronic Journal of Biotechnology 1 (1). Daniells, J. (1990) The Cavendish subgroup. Distinct and less distinct cultivars. In: Jarret, R.L. (ed.) Identication of Genetic Diversity in the Genus Musa. Proceedings of an International Workshop held at Los Baos, Philippines, 510 September. INIBAP, Montpellier, France, pp. 2935. De Langhe, E.A.L. (1996) Banana and Plantain: the Earliest Fruit Crop? Focus Paper No.1. INIBAP Annual Report 1995. INIBAP, Montpellier, France. De Langhe, E.A.L., Swennan, R. and Vuylsteke, D. (1996) Plantain in early Bantu world. Azania (2930) special double volume for 19941995. Demeke, T. (1986) Is Ethiopias Ensete ventricosum crop her greatest potential food? Agriculture International 38, 362365. Diekmann, M. and Putter, C.A.J. (1996) FAO/IPGRI Technical Guidelines for the Safe Movement of Germplasm. No.15. Musa, 2nd edn. FAO, Rome; IPGRI, Rome. Escalant, J.V., Cte, F., Grappin, A., Auboiron, E., Legavre, T., Lagoda, P., Carreel, F., Bakry, F., Horry, J.-P., Kerbelec, F., Teisson, C. and Tezenas du Montcel, H. (1994) Biotechnologies for Musa improvement strategies and results. In: Banana and Plantain Breeding: Priorities and Strategies. INIBAP, Montpellier, France, pp. 2226. Frison, E.A., Orjeda, G. and Sharrock, S.L. (eds) (1997a) PROMUSA: a Global Programme for Musa Improvement. Proceedings of a Meeting held in Gosier, Guadeloupe, March 5 and 9, 1997. INIBAP, Montpellier, France, World Bank, Washington, DC. Frison, E.A., Collins, W.W. and Sharrock, S.L. (1997b) Global Programs: a New Vision in Agricultural Research. Issues in Agriculture 12. CGIAR, Washington, DC. Fullerton, R.A. and Olsen, T.L. (1993) Pathogenic diversity of Mycosphaerella jiensis Morelet. In: Ganry, J. (ed.) Breeding Banana and Plantain for Resistance to Diseases and Pests. CIRAD and INIBAP, Montpellier, France. Horry, J.-P. (1999) Banana (Musa spp.) eld genebanks. In: Englemann, F. (ed.) Management of Field and In Vitro Germplasm Collections. Proceedings of a Consultation Meeting 1520 January 1996, CIAT Cali, Columbia. International Plant Genetic Resources Institute, Rome, Italy, pp. 4347. Horry, J.-P and Jay, M. (1990) An evolutionary background of bananas as deduced from avenoids diversication. In: Jarret, R.L. (ed.) Identication of Genetic Diversity in the Genus Musa. Proceedings of an International Workshop held at Los Baos, Philippines, 510 September 1988. INIBAP, Montpellier, France, pp. 4155. Horry, J.-P., Ortiz, R., Arnaud, E.,Crouch, J.H., Ferris, R.S.B., Jones, D.R., Mateo, N., Picq, C. and Vuylsteke, D. (1997) Banana and plantain. In: Fuccillo, D., Sears, L. and Stapleton, P. (eds) Biodiversity in Trust. Cambridge University Press, Cambridge, pp. 6781. Houbin, C. and Zehuai, W. (1997) Banana collections in Yunnan and Guangxi Provinces of China. INFOMUSA 6(1), 3132. INIBAP (1997a) MusaNews. INFOMUSA 6(2), 52. INIBAP (1997b) Networking Banana and Plantain. INIBAP Annual Report 1996. INIBAP, Montpellier, France. INIBAP (1998) PROMUSA Genetic Improvement Working Group meeting developing collaborative strategies. INFOMUSA 7(2), PROMUSA No. 2, iiiiv. Israeli, Y., Lahav, E. and Reuveni, O. (1995) In vitro culture of bananas. In: Gowen, S. (ed.) Bananas and Plantains. Chapman and Hall, London, pp. 147178. Jarret, R.L. and Gawel, N. (1995) Molecular markers, genetic diversity and systematics in Musa. In: Gowen, S. (ed.) Bananas and Plantains. Chapman and Hall, London, pp. 6683. Jarret, R.L., Gawel, N., Whittemore, A. and Sharrock, S. (1992) RFLP-based phylogeny of Musa species in Papua New Guinea. Theoretical and Applied Genetics 84, 579584.

242

S.L. Sharrock et al. Karamura, D.A. and Karamura, E.B. (1994) A Provisional Checklist of Banana Cultivars in Uganda. INIBAP, Montpellier, France. Langdon, R. (1993) The banana as a key to early American and Polynesian history. Journal of Pacic History 28, 1335. Lebot, V., Aradhya, K.M., Manshardt, R. and Meilleur, B. (1993) Genetic relationships among cultivated bananas and plantains from Asia and the Pacic. Euphytica 67, 163175. Lebot, V., Meilleur, B.A. and Manshardt, R.M. (1994) Genetic diversity in Eastern Polynesian Eumusa bananas. Pacic Science 48(1), 1631. MacDaniels, L.H. (1947) A Study of the Fei Banana and its Distribution with Reference to Polynesian Migrations. Bernice P. Bishop Museum Bulletin 190. Honolulu, Hawaii. May, G., Afza, R., Mason, H., Wiecko, A., Novak, F. and Arntzen, C. (1995) Generation of transgenic banana (Musa acuminata) plants via Agrobacterium-mediated transformation. Bio/Technology 13, 486492. Nasution, R.E. (1991) A taxonomic study of the species Musa acuminata Colla with its intraspecic taxa in Indonesia. Memoirs of Tokyo University of Agriculture 23, 1122 Ortiz, R. (1997) Secondary polyploids, heterosis and evolutionary crop breeding for further improvement of the plantain and banana (Musa spp. L) genome. Theoretical and Applied Genetics 94, 11131120. Ortiz, R. and Crouch, J.H. (1997)The efciency of natural and articial pollinators in plantain (Musa spp. AAB group) hybridization and seed production. Annals of Botany 80, 693695. Ortiz, R. and Vuylesteke, D.R. (1994a) Inheritance of black Sigatoka disease resistance in plantainbanana (Musa spp.) hybrids. Theoretical and Applied Genetics 89, 146152. Ortiz, R. and Vuylesteke, D.R. (1994b) Future strategy for Musa improvement. In: Banana and Plantain Breeding: Priorities and Strategies. INIBAP, Montpellier, France, pp. 4042. Ortiz, R., Ferris, R.S.B. and Vuylesteke, D.R. (1995) Banana and plantain breeding. In: Gowen, S. (ed.) Bananas and Plantains. Chapman and Hall, London, pp. 110146. Pascua, O.C. (1988) A tentative key to Philippine bananas. In: Jarret, R.L. (ed.) Identication of Genetic Diversity in the Genus Musa. Proceedings of an International Workshop held at Los Baos, Philippines, 510 September 1988. INIBAP, Montpellier, France, pp. 6775. Price, N.S. (1995) The origin and development of banana and plantain cultivars. In: Gowen, S. (ed.) Bananas and Plantains. Chapman and Hall, London, pp. 112. Purseglove, J.W. (1988) Tropical Crops, Monocotyledons, Vol. 2. Longman, London. Rowe, P. and Rosales, F. (1993) Genetic improvement of bananas, plantains and cooking bananas in FHIA, Honduras. In: Ganry, J. (ed.) Breeding Banana and Plantain for Resistance to Diseases and Pests. CIRAD/INIBAP, Montpellier, France, pp. 243266. Sgi, L., Panis, B., Remy, S., Schoofs, H., De Smet, K., Swennen, D. and Cammue, B.P.A. (1995) Genetic transformation of banana and plantain (Musa spp.) via particle bombardment. Bio/Technology (now Nature Biotechnology) 13, 481485. Sgi, L., Remy, S., Prez Hernndez, J.B., Cammue, B.P.A. and Swennen, R. (1999) Transgenic banana (Musa species). In: Bajaj, Y.P.S. (ed.) Biotechnology in Agriculture and Forestry, Vol. 47, Transgenic Crops II. Springer, Berlin. Sharrock, S. (1990) Report on 4th IBPGR/QDPI Banana Germplasm Collecting Mission to Papua New Guinea. IBPGR, Rome. Sharrock, S. (1995) Collecting the Musa genepool in Papua New Guinea. In: Guarino, L., Ramanatha Rao, V. and Reid, R. (eds) Collecting Plant Genetic Diversity, Technical Guidelines. CAB International, Wallingford, UK, pp. 647658. Sharrock, S. and Engels, J. (1997) Focus paper 1: complementary conservation. In: Networking Banana and Plantain: INIBAP Annual Report 1996. INIBAP. Montpellier, France, pp. 69. Shepherd, K. and Ferreira, F.R. (1982) The PNG Biological Foundations Banana Collection at Laloki, Port Moresby, Papua New Guinea. IBPGR/SEAN Newsletter 8(4), 2834. Simmonds, N.W. (1962) Evolution of the Bananas. Longman, London.

Musa Diversity

243

Singh, H.P. and Uma, S. (1996) Banana Cultivation in India. Offset Press, IARI Campus, New Delhi. Smith, M.K. and Drew, R.A. (1990) Current applications of tissue culture in plant propagation and improvement. Australian Journal of Plant Physiology 17, 267269. Stover, R.H. and Simmonds, N.W. (1987) Bananas. Longmans, London. Swennen, R. (1990) Limits of morphotaxonomy: names and synonyms of plantains in Africa and elsewhere. In: Jarret, R.L. (ed.) Identication of Genetic Diversity in the Genus Musa. Proceedings of an International Workshop held at Los Baos, Philippines, 510 September 1988. INIBAP, Montpellier, France, pp. 172210. Valmayor, R.V., Rivera, F.N. and Lomuljo, F.M. (1981) Philippine Banana Cultivars, Names and Synonyms. University of the Philippines at Los Baos, Philippines. Vuylsteke, D., Swennan, R. and De Langhe, E. (1991) Somaclonal variation in plantains (Musa spp. AAB group) derived from shoot tip culture. Fruits 46, 429439. Vuylsteke, D., Ortiz, R., Ferris, R.S.B. and Crouch, J.H. (1997) Plantain improvement. Plant Breeding Reviews 14, 267320.

13

Dynamic Management of Genetic Resources: a 13-year Experiment on Wheat

I. GOLDRINGER1, J. ENJALBERT1, J. DAVID2, S. PAILLARD1, J.L. PHAM3 AND P. BRABANT1,4

1INRA-UPS-INA.PG,

Station de Gntique Vgtale, Ferme du Moulon, 91190 Gif-sur-Yvette, France; 2ENSAM INRA, UFR de Gntique et Amlioration des Plantes, Domaine de Melgueil, 34130 Mauguio, France; 3Genetic Resources Center, International Rice Research Institute, PO Box 933, 1099 Manila, Philippines; 4INA-PG, 16 rue Claude Bernard, 75005 Paris, France

Introduction
Dynamic management (DM) of genetic resources was described by Henry et al. (1991) as a complementary strategy to genebanks for conserving genetic resources. The principle of DM is to maintain the evolutionary mechanisms in subdivided populations grown in different environments. Whereas there is an obvious risk of losing some individuals or some alleles, new alleles and genotypes are expected to appear in this system by means of recombination, mutation and migration. Such a DM approach has been experimentally tested in France since 1984 using wheat composite populations (Henry et al., 1991). More recently, it was proposed that DM could also be used as a kind of pre-breeding strategy (Goldringer et al., 1994). The idea was to obtain a gradual increase in the specic adaptation of the different populations to the local conditions of their multiplication sites in order to select interesting parents. In fact, developing these aspects brings DM very close to approaches earlier described as mass reservoirs or base-broadening (Simmonds, 1962). Still earlier, Suneson (1956) had proposed the use of composite cross populations as a base for a new breeding approach, which he called an evolutionary plant breeding method. To support this idea, the author demonstrated the continuous increase in yield of four barley populations growing under competitive natural selection for 1229 generations. Hence, we think that any kind of DM programme can be conducted with the objective of either variability conservation or germplasm enhancement, or maybe both at the same time. Great advantages of these methods are that they do not need high-technology equipment, they are easy to decentralize, and hence they may be carried out by farmers, breeders and local technical centres. Moreover, they allow farmers to be involved in developing the parents of new varieties. The positive aspects of farmer-participatory
IPGRI/FAO 2001. Broadening the Genetic Base of Crop Production (eds H.D. Cooper, C. Spillane and T. Hodgkin)

245

246

I. Goldringer et al.

breeding, particularly in the developing countries, have now been acknowledged (Ceccarelli et al., 2000) where the specic needs of many farmers are often not considered by the breeders. The purpose of this chapter is, rst, review the available results concerning the wheat DM experiment and, secondly, to discuss the possible use of the DM approach as a conservation strategy as well as a pre-breeding or base-broadening method.

The Wheat Experiment of Dynamic Management: Principle and Experimental Protocol

Principle of the dynamic management of genetic resources The aim of a DM approach is to maintain or mimic natural processes responsible for the diversication and conservation of genetic diversity. Rather than choosing and protecting one or several sites signicant for the diversity of a species (for example, in the gene centres) as in the case of classical in situ conservation, polymorphic composite cross populations are introduced into contrasting environments so that the different sub-populations evolve under the effects of natural selection, genetic drift and mutation. While each sub-population is expected to lose some of its initial variability due to drift and selection in a particular environment, diversity is expected to be maintained at the level of the metapopulation due to genetic differentiation. As random losses of alleles are expected to be different in each sub-population, the conservation of neutral allelic diversity will depend on the total number of sites. On the other hand, selected variability will be maintained only if the sites differ sufciently in respect of the selective pressures. In such conditions, the network of populations will also allow new favourable mutations to be captured in the individual sub-populations. If there is no migration between populations, evolution of the sub-populations and their adaptation to the environmental conditions will depend on the initial genetic variability in the population, on the size of the sub-populations combined with the strength of the selection pressures, and on the mating system. For the latter, theoretical studies showed that, in a nite population under directional selection, the optimal adaptive response (accumulation of positive alleles) was obtained with a low outcrossing rate (David et al., 1993). Such a mixed mating system combines rapid xation of the good genotypes due to a high level of selng, and mid- to long-term adaptation due to recombination between these genotypes. Moreover, it allows efcient purging of the mutational load, which might accumulate in populations of pure selfers that had a relatively low effective size. In a sub-population, the relative effect of selection compared with genetic drift will be all the lower as the population size is small. Even for medium-sized populations, useful variability might be lost within a population through genetic drift and hitch-hiking effects of selection. Hence, it is necessary to maintain some migration between populations so as to renew the variability within each population. The amount of geneow between populations has to be determined taking account of the effective size of the populations and the degree of differentiation between populations that is desirable. Arguments in favour of migration between populations also come from the shifting balance theory of Wright (1931, 1982). This predicts that populations will evolve

Dynamic Management of Genetic Resources: Wheat

247

until they reach the nearest adaptive peak of the tness surface (known as adaptive landscape) and then will not be able to evolve to a better combination of favourable genes since this would involve crossing a maladaptive valley. Such crosses would be possible owing to changes in allele frequencies in the population after a bottleneck event or with migration from other populations. Dynamic management has been proposed for woody species (Eriksson et al., 1993). For crop species, the pioneering study is that conducted on barley populations in the United States since 1928 (Harlan and Martini, 1929; Allard, 1988, 1990), despite the fact that only two environments were used initially. This experiment on barley inspired the experiment on dynamic conservation of genetic resources of wheat described below.

The dynamic management experiment Institutional framework A pilot programme for the DM of genetic resources of winter wheat (Triticum aestivum L.) has been conducted by the Direction Gnrale de lEnseignement et de la Recherche of the Ministry of Agriculture (DGER, France) and by the Institut National de la Recherche Agronomique (INRA, France) since 1984 (Henry et al., 1991). The experiment relies on the willingness of agricultural teachers who manage many of the populations. One important aim of this programme is to increase the awareness, in future farmers and agricultural technicians, of genetic diversity and the necessity of conserving genetic resources. In the French agricultural context, specic equipment adapted to experimental cultivation and labour with the required technical skill are mostly to be found at agricultural schools or research stations. It seems difcult to involve farmers who are trained and equipped for large-scale eld work. However, this may not be the case in areas where a more traditional agriculture still exists. Plant material Three initial composite populations (PA, PB and PS) were created by crossing 16 parental lines pyramidally for PA and PB populations (Thomas et al., 1991; David et al., 1997) and by two successive random crosses of 62 lines for PS population. Whereas PA and PB were predominantly selng, outcrossing was forced in PS in each generation using the recessive male-sterility nuclear gene ms1b (McIntosh, 1988). The genetic base of PB was wider than that of PA due to the presence in the former of more exotic lines among the parents. Dwarsm genes were present in disjunction in all three initial populations. Network and protocol In 1984, after 3 years of bulk multiplication, seed samples of the three initial populations were distributed to the sites of the multilocational experimental network (Fig. 13.1). The populations were not represented in each locality but there were seven locations for PA, nine for PB and 12 for PS (for more details, see Henry et al., 1991). Two methods of cultivation were used in each site: an intensive farming method, which corresponded to the method usually used for wheat in the area, and an extensive farming method, with one-third of the nitrogen fertilizer used in the intensive method

248

I. Goldringer et al.

Fig. 13.1. Network for the dynamic management of the wheat genetic resources programme (DGER of the French Agriculture Ministry, INRA, INA-PG and BRG).

and no fungicide treatments. Since 1984, each of the local populations has been cultivated every year without any conscious human selection, using seeds of the same population harvested the previous year in the same location under the same culture condition. The allogamous mating system in PS is maintained because only seeds from the male-sterile plants are harvested each year. A 100 m2 cultivated area was generally used so as to obtain around 10,00015,000 plants per population. For PS populations, male-sterile plants were identied at owering time and around 2000 plants were harvested. For the self-pollinating populations, PA and PB, 10,000 individuals corresponded to about 5000 individuals of a panmictic population. A demographic size of a few thousand individuals was chosen since it was expected that the genetic drift would be very limited. The outcrossing PS populations were sometimes preferred by the teachers participating in the work because they gave the opportunity of doing practical work with the students and because no specic equipment was required for the harvest. All the populations were isolated from each other and from neighbouring crops to avoid unwanted cross-pollination.

Dynamic Management of Genetic Resources: Wheat

249

Results
Each year, a sample of the seeds harvested for each population is stored in a centralized cold room. These have been used for comparative studies in different experiments: seeds from some of the samples have been grown in order to evaluate plants from different populations in the same location. Six years of multiplication were sufcient for the rst signicant changes to appear.

Evolution of agromorphological traits The mean plant height of all the populations of the autogamous PA and PB pools increased compared with the initial situation (David et al., 1992; Le Boulch et al., 1994). This increase was interpreted as the result of the competition for light between plants of a local population. Because the founding populations contained both dwarf and non-dwarf alleles at two major loci (Rht1 and Rht2), there was large phenotypic variability. Competition for light therefore led to rapid evolution in the populations (Pontis, 1992). Plants taller than the tallest plants of the initial populations were also found, indicating that recombination and selection of quantitative polymorphic loci had occurred. Populations PA and PB cultivated in intensive conditions became taller than the populations grown extensively. This may well have been due to a more severe competition between plants in intensive conditions, since the higher nitrogen input amplied height differences (Le Boulch et al., 1994). A northsouth gradient of earliness was found among the PA and PB populations (David et al., 1992). The populations cultivated in the south of France became earlier than those cultivated in the north (Fig. 13.2). This was interpreted as an adaptation to climatic constraints. In the south of France, heat and water stress appear early during plant development, and seed lling of late maturing plants may be hindered, whereas in the north these stresses seldom happen and late-maturing plants can accumulate more dry matter (Le Boulch et al., 1994). Besides selection pressures due to climate and soil (physical environment), the response of the populations to biotic selection pressures was studied. The resistance to powdery mildew (Erysiphe graminis f. sp. tritici) was selected for investigation, since the disease is widespread and can lead to yield reductions in wheat monoculture. Le Boulch et al. (1994) found signicant changes in the frequencies of some specic resistance genes to powdery mildew in the PA and PS populations after 8 years of multiplication. Although these changes differed with multiplication site and initial population, they were not related to the spectrum of virulence genes of the pathogen populations in the different sites. In each genetic pool, individuals were found that accumulated more resistance genes than any individual analysed in the initial population (three genes versus two genes in PA; four genes versus one in PB; more than four genes versus three in PS). Adult eld resistance results from the combined expression of specic resistance genes and other quantitative factors that control partial resistance. In the DM populations, adult resistance also differed with multiplication site and initial population (PA, PB or PS). Comparing mean adult resistance of all the populations with the resistance of the corresponding initial population showed no signicant difference for pool PA, a slight decrease in pool PB, and an increase in pool PS. These results differ markedly

250

I. Goldringer et al.

38 r = 0.802* Mean heading date (days after 5 May) 37 Vervins 36

35 Chlons 34 Nrac Montreuil 33

Rennes Le Moulon

32 Toulouse 31 43 44 45 46 47 Latitude of the site (N) 48 49 50

Fig. 13.2. Northsouth earliness gradient in pool A populations (sixth generation) evaluated at Le Moulon in 1991.
Source: Le Boulch et al., 1994.

from those obtained for barley populations in which Allard (1990) and Ibrahim and Barrett (1991), respectively, found an increase of resistance to Rhynchosporium secale (after about 40 generations) and to Erysiphe graminis f. sp. hordei (after nine generations). In the wheatpowdery mildew interaction, the rapid evolution of the pathogen due to its sexual mating system and its high dissemination level leads to a rapid rotation of its avirulence genes at any given site. This may explain a response that consisted of the accumulation of different resistance genes rather than a general increase in the frequency of one gene. It also underlines the recombination which allowed the pyramiding of genes involved in adaptation to changing environments. Resistance to eyespot was also studied on a sub-set of the short lines extracted from six PA and six PB populations after ten generations. A major resistance gene (Pch1) derived from the interspecic crosses with Aegilops ventricosa was polymorphic in the initial populations. This gene can be easily identied with seedling tests. Fifty inbred lines per population were tested at INRA Le Rheu in 1996 and 1997. The results showed signicant differences in the frequency of the gene according to the multiplication site (Table 13.1). Differences were much greater in PA than in PB. In PA, Pch1 frequency increased in the sites where the pathogen was more likely to be found, i.e. Vervins, Rennes Chlons, in the north of France. In both pools, the mean frequency over the populations was not different from the initial one, giving evidence of the maintenance of the gene at the metapopulation level. Some important traits such as grain yield proved to have evolved in each population according to local conditions, although it was not possible to identify the specic physical or biotic characteristic of the environment to which these evolutions were

Dynamic Management of Genetic Resources: Wheat

251

Table 13.1. Frequency of the phenotype resistant for Pch1 locus (homozygous + heterozygous lines) in 11 PA and 15 PB populations (after ten generations) and in the initial populations PA0 and PB0. Frequency of resistant phenotype at Pch1 PA Populations Initial population Final populations: Chlons Le Chesnoy Le Moulon Montreuil Rennes Toulouse Venours Vervins Mean of the nal populations (Int. + Ext.)
a Means

PB Extensive Intensive 0.27 0.25 0.17 0.00 0.37 0.30 0.23 0.10 0.47 0.09 0.50 0.07 0.00 0.37 0.19a 0.21 0.24 0.12 0.15 0.08 0.14 0.07 Extensive

Intensive 0.24 0.25 0.09 0.08 0.35 0.11 0.81 0.25a

were not signicantly different from the frequencies of the initial populations.

related. This is illustrated by the correlation found (r = 0.78, Fig. 13.3) between yields evaluated at Le Moulon and Clermont-Ferrand in 1991 of PA and PB populations originating from the same site and the same cultural condition (David, 1992). Such positive and signicant correlations were found for other traits, although correlations between PA or PB populations and the corresponding PS populations were never signicant, again indicating that the outcrossing populations showed specic behaviour. The outcrossing populations (PS) behaved quite differently from the predominantly self-pollinating ones. Though the PS populations have been less studied than the others, it seems that, as expected, for the rst ten generations within-population variability was maintained at a higher level, whereas evolution and adaptation to local conditions were slower.

Evolution of biochemical and molecular markers Two-dimensional electrophoresis of proteins was used to investigate the differentiation between populations at the biochemical level (David et al., 1997). The study was restricted to PB populations cultivated for eight generations in six different sites with two agronomic practices. For each population, extracts of several plants were pooled together and six replications (different extracts and gels) were tested. Spot intensity was measured by automatic image analysis. Analysis of variance revealed that 39 spots out of the 162 reproducible ones were quantitatively polymorphic between populations. The site effect was the predominant contribution to the observed differentiation (60% of the between-population variance), whereas the agronomic practices accounted for

252

I. Goldringer et al.

68 r = 0.78 Grain yield (qx ha1) of PB populations 66 Chlons-Int

64 Chlons-Ex 62 Le Moulon-Ext 60 Le Moulon-Int 58 Nerval-Int Nerval-Ext Toulouse-Ext

Toulouse-Ext Montreuil-Ext Montreuil-Int

56 56 57 58 59 60 61 62 63 Grain yield (qx ha1) of PA populations 64 65 66

Fig. 13.3. Relationship between the mean over two locations (Clermont-Ferrand and Le Moulon) of grain yield evaluated in 1991 of PA and PB populations (sixth generation).

only 10% and the conditionsite interaction accounted for 30% of the variance. This was also conrmed by the multivariate analysis. This convergent evolution of populations cultivated at the same site suggested that natural selection for local adaptation contributed to the differentiation of the protein patterns of the populations. Indeed, hitch-hiking effects amplied by the preferential selng in the PB pool may have led to convergent evolution of populations submitted to the same selection pressures even for neutral traits, and we can not conclude whether these 39 proteins have any selective value or not. The diversity of RFLP markers maintained after ten generations of multiplication was studied on three extensive PA populations (Le Moulon, Rennes, Toulouse), three extensive PB populations (Le Moulon, Rennes, Venours) and the two initial populations (PA0 and PB0). The 14 restriction fragment length polymorphism (RFLP) markers that were polymorphic in the 32 parental lines (see Enjalbert et al., 1999a, for more details) produced 38 polymorphic loci (reecting the hexaploid genome of wheat). The mean number of alleles lost in one population was 3.5 of the 72 initially present for PA, and 2.5 of 75 for PB, but taking the three populations as a whole, no allele was lost in PA whereas only one was lost in PB (Fig. 13.4). This suggests that if we had analysed all the populations of each pool, we would certainly have found all the initial alleles. Estimates of the effective size of the six populations (from 42 to 208) were much lower than expected under genetic drift only (>5000) considering the number plants cultivated for each population (~10,000). Moreover, signicant differentiation (estimated with the Wright Fst parameter) was found within both PA and PB pools (Enjalbert et al., 1999b). These large variations of the allelic frequencies of RFLP markers can be

Dynamic Management of Genetic Resources: Wheat

253

Fig. 13.4. Mean number of missing alleles in dynamic management population samples (number of RFLP alleles initially present = 75).

explained either by hitch-hiking due to linkage with loci subject to selection or by more indirect effects, since strong selection on some loci leads to a high variance of the reproductive contribution of each individual to the next generation and hence to an increase of genetic drift over the whole genome. Yet, we found no convergence between the PA and PB populations cultivated in the same sites (Enjalbert et al., 1999b). We suggest that the difference in evolution found for protein patterns was due to the fact that proteins were studied in populations derived from the same initial pool (PB) and hence the populations had the same initial linkage disequilibriums between markers and selected genes. However, for the molecular markers, completely different linkage disequilibriums may have been present in the initial PA and PB populations. Even though allelic frequencies of biochemical and molecular markers change under genetic drift combined with direct selection or selection by hitch-hiking, the metapopulation appeared to maintain the diversity of such loci overall.

The mating system Mid- to long-term evolution of the traits involved in individual tness in the different populations will depend upon the recombination level within each sub-population. Indeed, PS populations appeared to behave in a completely different way to the populations of the two self-pollinated pools. As previously noted, the optimal adaptive

254

I. Goldringer et al.

response would be obtained with a low (but not zero) outcrossing rate. Thus a knowledge of the outcrossing rate of the different self-pollinated populations is important for understanding the evolution of traits within each sub-population, and for determining gene ow between populations. For this, we used the data from an RFLP codominant locus on the six PA and PB populations previously described. Outcrossing rates were estimated in each population from the multilocus heterozygote deciency. Estimations ranged from 2.4% to 10.1%, depending on the population (Enjalbert et al., 1998). The corresponding proportions of heterozygous individuals were signicantly higher than expected under strict self-pollination. Outcrossing rate did not differ from one pool to the other but it appeared to be affected by the multiplication site. Populations from Rennes had a higher level of outcrossing than populations from Le Moulon. The interaction between the genetic pool and the cultivation site was also signicant since differences were found between the estimation of PA and PB populations grown in the same site. These results indicated that the self-pollinated populations did not evolve towards strict self-pollination but rather towards a mixed mating system. Such a mating system seems to have a genetic basis even though it is probably inuenced by climate. Recombination within sub-populations appears to be sufcient for future evolution. We think that the residual outcrossing rate can also be used for geneflows between populations.

Discussion
Assessment of the dynamic management programme Efciency of the network for maintaining genetic diversity Because of the diversity of the environments in the network, the DM system proved to be able to maintain the initial variability for all the agronomic traits studied except for genes poorly adapted to interindividual competition (such as dwarsm genes). The loss of variability within each local population is compensated by the differentiation between populations. Competition between plants seems to be the only selection pressure that leads to unidirectional evolution. However, a high competitive ability (tall plants) is often antagonistic to the agronomic value (low harvest index). In order to limit inter-plant competition effects, we propose to reduce seed sowing density and nitrogen fertilizer supply, i.e. switching to very extensive conditions. Such cultivation conditions seem convenient for developing countries since agricultural systems may be closer to these low-input conditions. In European countries, more limited operations such as cutting the tallest plants before harvest should be preferred since low-input conditions are not generally applied. In any event, we think that both solutions are able to circumvent the specic problem of increase in height. The DM system also proved to be efcient for the conservation of most of the neutral diversity assessed on the basis of biochemical and RFLP markers. Adaptation of the populations to local conditions Most of the evolution observed for agronomic traits could be related to local selection pressures. This was especially obvious when some convergence was found for evolution of different populations in the same site (e.g. between PA and PB for yield, or between

Dynamic Management of Genetic Resources: Wheat

255

extensive and intensive farming conditions for the proteins revealed by two-dimensional electrophoresis). Disease resistance, earliness and grain production have evolved under the physical and biotic selection pressures of the different multiplication sites. Note that the biotic constraints are expected to uctuate more than physical ones. Increases in the frequencies of some resistance genes to powdery mildew were found to be independent of the virulence spectrum of the parasite of that site. This means that evolution of this type can not be easily predicted from the characterization of the pathogen populations of each site. This might be different for other pathogens such as yellow rust or leaf rust, the races of which are more stable from one year to the next. In any case, the characterization of the intensity of the parasitic pressure of a given site may provide some elements for predicting the evolution of the adult stage resistance of a population cultivated in this site. Though the increase in earliness in the populations cultivated in the south of France was believed to be related to selection for heat and drought tolerance, we did not determine whether these populations had developed other mechanisms for drought resistance. We think that such adaptations are likely to happen provided that the initial genetic variability for the trait was sufcient. Assessment of the agronomic value of the dynamic management material It should be noted that whatever the material we evaluated and the experiment undertaken, the genotypes from extensive farming methods were never poorer in agronomic performance than genotypes from the intensive farming method. For example, we selected lines from the different populations after 10 years of multiplication mainly on the basis of their height. In 1996/97, 201 were evaluated for grain yield at Le Moulon without replication and 46 at Montpellier with two replications (data not published). In neither of the experiments were intensive lines found to be signicantly superior to the extensive ones for yield and kernel weight. In another experiment (Goldringer et al., 1998), when comparing the extensive and intensive populations from Rennes and Le Moulon, the extensive populations were found to be more resistant to leaf rust (Puccinnia recondita) and powdery mildew (Erysiphe gramini) than the intensive ones. Moreover, extensive populations proved to be shorter than intensive ones. Generally speaking, the populations that have been cultivated under extensive farming methods show an equivalent or greater agronomic potential than the intensive populations. A particular issue with these experiments remains the management of plant height. Unlike the barley composite cross populations (Harlan and Martini, 1929; Suneson, 1956; Allard, 1988), in our wheat populations pure natural selection and betweenplant competition did not lead to lines close to varieties adapted to classical cultural conditions. While increase of plant height would not be such a negative property in some developing countries where low-input agriculture is applied (Ceccarelli et al., 1996), in the European agricultural context there was a substantial difference between mean height of the populations and that of cultivated varieties or of improved lines derived from recurrent selection programmes. To use DM populations as a source of variability in a breeding scheme or recurrent selection programme, we have suggested that articial selection should be introduced in some DM populations in order to preserve their cultivated pattern. We would like to emphasize this point because the selection of new combinations of genes might depend on the genetic context of the Rht loci. Dwarng genes are known to display strong pleiotropic effects (Pinthus and Gale,

256

I. Goldringer et al.

1990) on other traits, and the selection of gene combinations in the dwarf genetic background would be much more advisable. The preliminary agronomic evaluation of lines selected from the different tengeneration populations showed that the very best lines had a good grain production level compared with lines extracted from a recurrent selection programme and the current varieties (Fig. 13.5). Hence, this indicates that interesting parents may be found after one or two steps of intensive selection within the populations. This stage of selection would be reduced if plant height of the populations was under control.

Some propositions for dynamic management programmes We can identify two different levels of dynamic management more or less connected to each other and open to use in supporting variety breeding: 1. One DM system, oriented towards genetic resources conservation with frequent introduction of interspecic crosses, that is more likely to be used by the national and international research institutes. 2. One DM system (could also be termed low-intensity recurrent selection) oriented towards pre-breeding with some degree of articial selection to maintain the cultivated pattern and to improve some economically important traits. This system should be used by experimental stations of research institutes as well as by agricultural or agronomic schools, and, in developing countries, it could also be used by farmers interested in selection.

Fig. 13.5. Grain yield of lines derived from the dynamic management populations of PA and PB pools compared with lines derived from recurrent selection and with commercial varieties.
0,

lines derived from recurrent selection; +, lines derived from dynamic management.

Dynamic Management of Genetic Resources: Wheat

257

The idea of this second approach is to recreate types of traditional landraces for species and areas where these have been replaced by a small number of commercial varieties or to structure such a system so as to renew the genetic variability and improve its use for breeding. The two systems might differ mainly in the amount of genetic variability that is accepted by the people managing the populations, by the emphasis that is put on the conservation of the initial variability, and by the amount of migration that could be set between populations. It is clear that there is no absolute dichotomy between the two systems but it is rather a continuum in which the different programmes may nd their place depending on the species, breeding context, country and economic level. Choosing the sites The benet of using sites with high and contrasting parasitic pressures has already been emphasized. The quantitative knowledge of the intensity of parasitic pressure for a given disease seems useful to predict response at the adult stage and hence could be a rst indicator. Characterization of the climate, among which latitude can be a good predictor of the response for earliness, provides a second set of criteria. Though the genetic correlation between grain yield of PA and PB populations cultivated in the same site suggested that some sites were more favourable for high yield, the a priori characterization of the sites for their ability to select high-yielding genotypes needs further investigations. Choosing the sites should not be very different depending on whether DM is oriented towards genetic resources conservation or pre-breeding. However, in the rst case, environments may be numerous, contrasting or even extreme, while, in the second case, sites should be chosen to maximize the possibility of adaptation to the local conditions and should represent the cultural conditions of the area. Maintaining the desirable crop characteristics This should be easy in pre-breeding-oriented DM since articial selection would be applied. For the populations managed by farmers in developing countries, selection towards an ideotype adapted to the local conditions and also corresponding to the traditional phenotype cultivated by the farmers can be applied, except that selection intensity should not be too strong. The phenotype of the populations has also to be under control in the DM programme oriented towards genetic resources conservation. If part of the aim is to be able to provide new variability to the local farmers or breeders, the material should not be too distant from the cultivated type. Hence, exotic or interspecic variability should be introduced at given generations, and a few generations of adaptation of the populations should interspace the introduction. Material should be distributed to farmers or breeders only after the adaptation period. What initial variability and how to make the composite population The amount of initial variability depends much on the aims of a given programme (short-, mid- or long-term) and on the problem that is to be addressed: is there a lack of variability for a given trait such as a resistance, or is there a general need to renew the variability? Sometimes, interspecic crossing will be the only way to nd resistance to a pathogen. However, the more interspecic variability is introduced, the longer will be the time needed to return to a cultivated pattern adapted to local conditions.

258

I. Goldringer et al.

Populations directly derived from the crossing of interspecic parents should only be distributed in the genetic resources conservation-oriented DM programme. Manual crosses (pyramidal or circular) between a limited number of parental lines (~20) chosen according to agronomic traits, geographical origins or pedigrees to maximize genetic variability appear to be a good solution for pre-breeding programmes. For DM oriented towards genetic resources conservation, the number of parental lines of a population can be considerably increased using male sterility systems in self-pollinating species or natural open-pollination in out-crossing species. Mating system, population size and number of populations Our study was performed on a preferentially self-pollinating species and it seems difcult to generalize the conclusions to true out-crossing species, even though we had turned the mating system of one of the populations into allogamy. Yet, for self-pollinated species, increasing recombinations with a male sterility gene appears interesting especially for DM oriented towards genetic resources conservation. Elimination of the gene (recessive or dominant) is easy when the material is to be used in the context of the pure line breeding of self-pollinating species. For pre-breeding DM, a more limited level of recombination should be recommended to maximize the medium-term selection response. Indeed, this is what is often found in so-called self-pollinated species. Our results showed that the size of the populations should be relatively large since the selection due to the adaptation of the populations to their new environments leads to a drastic reduction in the effective population size. At least 10,000 individuals should be cultivated in each population. Yet, even with the low effective sizes we found for the wheat populations, genetic variability was maintained at the level of the metapopulation. Hence, we would emphasize the importance of maintaining sufcient numbers of different populations at a time. Based on our experience, we would propose at least ten populations for a given genetic pool in order to ensure a safe maintenance of the variability in the medium term. Geneows between populations Some theoretical developments are available concerning the modalities of the geneow that it is desirable to establish between populations, but there is a need for further developments to provide more precise indications. The general idea is that a small amount (small enough so as not to break the adaptation processes) of pollen exchange between the sub-populations would be favourable (see pages 246247). In the case of the pre-breeding DM populations, geneows could be based on occasional seed exchange between farmers.

References
Allard, R.W. (1988) Genetic changes associated with the evolution of adaptedness in cultivated plants and their wild progenitors. Journal of Heredity 79, 235238. Allard, R.W. (1990) The genetics of hostpathogen coevolution. Implication for genetic resources conservation. Journal of Heredity 81, 16. Ceccarelli, S., Grando, S. and Booth, R.H. (1996) International breeding programmes and resource-poor farmers: crop improvement in difcult environments. In: Eyzaguirre, P. and

Dynamic Management of Genetic Resources: Wheat

259

Iwanaga, M. (eds) Participatory Plant Breeding. Proceeding of a Workshop on Participatory Plant Breeding, 2629 July 1995, Wageningen, The Netherlands. IPGRI, Italy, pp. 99116. Ceccarelli, S., Grando, S., Tutwiler, R., Baha, J., Martini, A.M.L., Salalieh, H., Goodchild, A. and Michael, M. (2000) A methodological study on participatory barley breeding. I. Selection phase. Euphytica 111, 91104. David, J.L. (1992) Approche mthodologique dune gestion dynamique des ressources gntiques chez le bl tendre (Triticum aestivum L). PhD thesis, lInstitut National Agronomique Paris Grignon. David, J.L., Savy, Y., Trottet, M. and Pichon, M. (1992) Mthode de gestion dynamique de la variabilit gntique. Exemple dun rseau exprimental de populations composites de bl tendre. In: BRG (ed.) Complexes despces, ux de gnes et ressources gntiques. Lavoisier, Cachan, France, pp. 337350. David, J.L., Savy, Y. and Brabant, P. (1993) Outcrossing and selng evolution in populations under directional selection. Heredity 71, 642651. David, J.L., Zivy, M., Cardin, M.L. and Brabant, P. (1997) Protein evolution in dynamically managed populations of wheat: adaptive responses to macro-environmental conditions. Theoretical and Applied Genetics 95, 932941. Enjalbert, J., Goldringer, I., David, J.L. and Brabant, P. (1998) The relevance of outcrossing for the dynamic management of genetic resources in predominantly selng Triticum aestivum (bread wheat). Genetics, Selection, Evolution 30 (suppl.1), 197211. Enjalbert, J., Boeuf, C., Belcram, H. and Leroy, P. (1999a) Use of multiparental inbred populations to determine allelic relationships of molecular markers. Plant Breeding 118, 8890. Enjalbert, J., Goldringer, I., Paillard, S. and Brabant, P. (1999b) Molecular markers to study genetic drift and selection in wheat populations. Journal of Experimental Botany 50(332), 283290. Eriksson, G., Namkoong, G. and Roberds, J.H. (1993) Dynamic gene conservation for uncertain futures. Forestry Ecology Management 62, 1537. Goldringer, I., Pham, J.L., David, J.L., Brabant, P. and Gallais, A. (1994) Is dynamic management of genetic resources a way of pre-breeding? In: Balfourier, F. and Perretant, M.R. (eds) Eucarpia Genetic Resources Section Meeting Evaluation and exploitation of genetic resources pre-breeding. Clermont-Ferrand, France, March 1518 1994. pp. 163170. Goldringer, I., Paillard, S., Enjalbert, J., David, J.L. and Brabant, P. (1998) Divergent evolution of wheat populations conducted under recurrent selection and dynamic management. Agronomie 18, 413425. Harlan, H.V. and Martini, M.L. (1929) A composite hybrid mixture. Journal of the American Society of Agronomy 21, 487490. Henry, J.P., Pontis, C., David, J.L. and Gouyon, P.H. (1991) An experiment on dynamic conservation of genetic resources with metapopulations. In: Seitz, A. and Loeschcke, V. (eds) Species Conservation: a Population Biological Approach. Birkhaser Verlag, Basel, pp. 185198. Ibrahim, K.M. and Barrett, J.A. (1991) Evolution of mildew resistance in a hybrid bulk population of barley. Heredity 67, 247256. Le Boulch, V., David, J.L., Brabant, P and De Vallavieille-Pope, C. (1994) Dynamic conservation of variability: responses of wheat populations to different selective forces including powdery mildew. Genetics, Selection, Evolution 26(Suppl. 1), 221240. McIntosh, R.A. (1988) Catalogue of gene symbols. In: Miller, T.E. and Koebner, R.M.D. (eds). Proceedings of the 7th International Wheat Genetics Symposium, Cambridge, UK, 1319 July 1988, Vol. 2, pp. 12251323. McIntosh, R.A. (1998) Breeding wheat for resistance to biotic stresses. Euphytica 100, 1934. Pinthus, M.J. and Gale, M.D. (1990) The effects of gibberellin-insensitive dwarng alleles in wheat on grain weight and protein content. Theoretical and Applied Genetics 79, 108112.

260

I. Goldringer et al. Pontis, C. (1992) Utilisation de marqueurs gntiques pour le suivi de la variabilit de 3 composites de bl tendre dhiver (Triticum aestivum L.) menes en gestion dynamique. PhD thesis, LInstitut National Agronomique Paris Grignon. Simmonds, N.W. (1962) Variability in crop plants, its use and conservation. Biological Reviews 37, 442465. Simmonds, N.W. (1991) Selection for local adaptation in a plant breeding programme. Theoretical and Applied Genetics 82, 363367. Suneson, C.A. (1956) An evolutionary plant breeding method. Agronomy Journal 48, 188191. Thomas, G., Rousset, M., Pichon, M., Trottet, M., Doussinault, G. and Picard, E. (1991) Mthodologie de lamlioration de bl tendre (Triticum aestivum L.). I. Cration par croisements et analyse dune population articielle a 16 parents, base de cette tude mthodologique. Agronomie 11, 359368. Wright, S. (1931) Evolution in Mendelian populations. Genetics 16, 97159. Wright, S. (1982) The shifting balance theory and macrovolution. Annual Review of Genetics 16, 119.

14
1Nordic

Genetic Base-broadening of Barley (Hordeum vulgare L.) in the Nordic Countries

M. VETELINEN1 AND E.A.J. NISSIL2

Gene Bank, PO Box 41, S-230 53 Alnarp, Sweden; 2International Plant Genetic Resources Institute, Regional Ofce for Asia, the Pacic and Oceania, PO Box 236, UPM Post Ofce, 43400 Serdang, Selangor Darul, Ehsan, Malaysia

Introduction
The genetic base of cultivated barley has been narrowed since the beginning of modern plant breeding (Linde-Laursen et al., 1987; Martin et al., 1991; Fischbeck, 1992; Melchinger et al., 1994). A lack of useful genetic variation, decreasing or even imperilling long-term genetic gain, can be especially encountered in breeding programmes targeted to specic environments such as marginal areas or specialized products (malting barley for example) (Wych and Rasmusson, 1983; Melchinger et al., 1994; SaghaiMaroof et al., 1994; Manninen and Nissil, 1997). The primary genepool of barley includes cultivated barleys, Hordeum vulgare L., and wild barley, Hordeum vulgare L. ssp. spontaneum. It still contains a very wide range of genetic variation. Since wild barley is cross-compatible with modern cultivated forms, it is available for incorporation into germplasm used in breeding programmes (Brown et al., 1978; Nevo, 1979, 1992; Nevo et al., 1986; Jana and Pietrzak, 1988; Chalmers et al., 1992; Zhang et al., 1993). Wild barley and other exotic, i.e. unadapted (Hallauer and Miranda, 1981), barleys such as landraces are attractive for breeding since they may be a source of novel genetic variation. One difculty in utilization and selection of exotic sources for breeding is that the favourable genes may be hidden in this non-adapted material. This applies especially to quantitative genes. Thus, how the novel variation can be included in breeding populations in the most effective and sustainable way is an important issue not merely from the breeding point of view, but also for continued evolution in the primary genepool of barley. Genetic enhancement of unadapted germplasm whether by introgression (backcrossing programmes) or incorporation (large-scale development of locally adapted populations) is one of the most costly and time-consuming components of any breeding programme. Consequently, plant breeders may not always have enough resources to work with exotic germplasm requiring extensive pre-breeding. In the Nordic countries an incorporation project was initiated by scientists at the Swedish University of Agricultural Sciences, Department of Plant Breeding in Svalv to link the efforts of
IPGRI/FAO 2001. Broadening the Genetic Base of Crop Production (eds H.D. Cooper, C. Spillane and T. Hodgkin)

261

262

M. Vetelinen and E.A.J. Nissil

researchers and breeders. The rst aim of the programme was to form two dynamic barley genepools which possessed a high level of resistance to various barley diseases since there was a general lack of resistance sources in Nordic barley breeding material. The second aim was to recombine genes from unadapted exotic germplasm with locally adapted varieties and breeding lines in order to broaden the overall genetic variation in Nordic barley breeding materials and to avoid future problems associated with a narrow genetic base. The establishment of Dynamic Genepools of Nordic Barley was carried out separately for germplasm adapted to southern (Sweden) and northern parts of Fennoscandia (Finland). The crossing programmes to develop the dynamic genepools for barley began in both Finland and Sweden during 1991/92.

Materials and Methods

The same unadapted material was included in both Finnish and Swedish genepools, i.e. ve Hordeum vulgare L. ssp. spontaneum (hereafter called H. spontaneum, or wild barley) accessions from Syria and Jordan and ten landraces from different parts of China, Tibet and Pakistan. The adapted material (25 genotypes) for the Swedish genepool originates from the southern part of Sweden, while the Finnish material (25 genotypes) is adapted to more northern conditions (Fig. 14.1). Both exotic and adapted founder lines possessed resistance to at least one major pathogen and were selected to ensure that there was wide phenotypic diversity among them. The development of the dynamic genepools was based on a controlled crossing programme in order to maintain the pedigree information on each line. All of the founder lines contributed to the genepool with equal frequency. The material was intercrossed for six generations using both pairwise crosses and half-diallel designs so that new multiples were obtained (Fig. 14.1) and were used (Table 14.1). Crosses were made in a glasshouse in order to avoid natural selection during the establishment phase, and no conscious selection was carried out during the establishment. Selected F2 progeny were used for further studies.

Results and Discussion

Exotic germplasm as sources of novel genetic variation for Nordic barley In order to verify that the selection of the material for genetic diversity was successful, the parental lines were tested for variation at 11 allozyme loci, of which nine showed polymorphism. Neis (1975) gene diversities within parental lines used in both the Swedish and Finnish genepool were 0.13 (adapted parents), 0.16 (landraces) and 0.25 (H. spontaneum). Cluster analysis revealed that adapted parents, landraces and H. spontaneum were genetically divergent (Fig. 14.2). Thus the genetic studies showed that exotic germplasm possessed greatest diversity and that genetic variation was structured within the germplasm used. It also conrmed the potential utility of exotic germplasm as a source of new genetic variation for Nordic barley. It was also essential to check that the exotic alleles from unadapted material were incorporated in the genepools and maintained through all the stages of the crossing

Genetic Base-broadening of Barley

= adapted materials Swedish varieties and lines 25 genotypes H. vulgare landraces 10 genotypes H. spontaneum 5 genotypes Finnish varieties and lines 25 genotypes = unadapted materials H. vulgare landraces 10 genotypes H. spontaneum 5 genotypes

263

Crossing population 6 generations of recombination

Crossing population 6 generations of recombination

276 progeny

276 progeny

Selfing Multiplication

Selfing Multiplication

Utilization by Nordic plant breeders Southern parts of the Nordic countries Northern parts of the Nordic countries

Fig. 14.1. Development of dynamic genepools for Nordic barley breeding.

programme. In the Swedish genepool, tests on the third crossing generation showed that exotic marker genes were preserved. Novel recombinants were also observed (Vetelinen, unpublished data).

Agronomic performance and adaptive traits Modern varieties have been developed during decades under strong directional selection for adaptation, high-quality standards and yield. This leads to difculties in introducing new sources of genetic diversity into the breeding programmes. The introduction of unadapted germplasm usually lowers agronomic performance due to the disruption of

264

M. Vetelinen and E.A.J. Nissil

Table 14.1. Crossing scheme for the establishment of the dynamic genepools. Generation 0 1 2 3 4 5 6 No. of progeny produced 40 20 190 95 48 24 276 Crossing design Pairwise Half diallel Pairwise Pairwise Pairwise Half diallel Effects of exotic germplasm on agronomic traits Test on multiple resistance, allele frequencies and variability in important agronomic traits Effects of exotic germplasm on agronomic traits Studies Genetic and quantitative analysis of parent material

inter-loci interactions or by expression of inferior alleles. However, it may also be useful to break up existing gene complexes and increase the likelihood of new valuable recombinants. Agronomic performance of selected progeny drawn from the different crossing generations was tested and compared with their adapted parents/grandparents (Vetelinen, 1994; Vetelinen et al., 1996, 1997). Ten progeny taken from both Finnish and Swedish genepools were evaluated together with their adapted parents for agronomic performance. The results showed that transgressive segregants, tested against the best adapted parent mean, were already produced in the early phase of the base-broadening programme (Table 14.2). Progeny from the different types of exotic sources were also compared, and progeny from unadapted landraces performed better than those from H. spontaneum parents. However, even progeny with wild parents produced transgressive segregants for overall agronomic performance. However, with wild gene sources, the need for longer-term pre-breeding is likely to be an unavoidable consequence of the simultaneous introduction of undesirable wild-traits, such as seed shattering and non-synchronous tillering habit (Vetelinen, 1994; Vetelinen et al., 1996). In the northern parts of the Nordic countries, the utilization of exotic materials is especially difcult due to marginal environmental conditions demanding a very specic combination of adaptive traits (Tigerstedt, 1994). Cultivation in this area is possible even north of the Arctic Circle due to effects of the Gulf Stream. However, low temperatures, specic light conditions and early frosts create unique circumstances during the short but intensive growing season. Since in northern marginal conditions several traits can limit adaptation, it was necessary to consider in which type of genetic base and in which kind of specic genotypes exotic material should be incorpo-

Genetic Base-broadening of Barley

Dices similarity coefficient 0.4 0.5 0.6 0.7 0.8 0.9 1.0

265

Kara Sv 898219 Sv 893133 Sv 89425 Aisling Magda Golf Alfa Ida Sv 898287 Sv 892508 Milka Hulda Nancy Sv 87368 Sv 892596 L H 7614 Ariel Korinna Ellinor Sv 87443 Shirley Sv 892052 Jet L H 4166 L H 4246 L H 4248 L H 4241 L H 7405 L H 4175 L H 4086 L H 4035 Sv 892368 L H 7009 Sv 89412a Sv 89412b S H 3038 S H 3042 S H 3057 S H 3064 S H 3059

Fig. 14.2. Dendogram of 41 parent lines in the Swedish dynamic genepool revealed by UPGMA cluster analysis based on Dices coefcient.
L: landrace; S: Hordeum spontaneum.

rated. With both the dynamic genepools, the majority of parents were adapted to the different northern environments, and two different incorporation schemes were tested: (i) progeny in which the same exotic material was introduced into a genetic background adapted to the agroecological conditions of the extreme Fennoscandian north; or (ii) introduction in the background adapted to the conditions of the southern part

266

M. Vetelinen and E.A.J. Nissil Table 14.2. Percentages of plants from ve Finnish (Fi) and ve Swedish (Sw) progeny with an indexa above the best adapted parent mean. Unadapted parental lines. Adapted parent 1 Jo 1621 Kalle Kinnan Hja 83054 Arttu Alfa Korinna Golf Nancy Ida Unadapted parent 1 H 4241 (L) H 4248 (L) H 4035 (L) Ob 264 H 4175 (L) H 4241 (L) H 4248 (L) H 4035 (L) Sv 89412 H 4175 (L) Adapted parent 2 Hja 80138 Meltan Pohto Mette Jo 1545 Aisling Sv 892052 Jet Sv 898219 Milka Unadapted parent 2 H 7614 (L) H 7405 (L) H 3064 (S) H 3042 (S) H 4246 (L) H 7614 (L) H 7405 (L) H 3064 (S) H 3042 (S) H 4246 (L) Index 2-row 58 38 10 27 50 24 36 11 45 6-row 27 11 8 13

Progeny Fi 1 Fi 2 Fi 3 Fi 4 Fi 5 Sw 1 Sw 2 Sw 3 Sw 4 Sw 5
a Index=

i ear emergence per days + i straw length cm + i no. of ears per plant + i heading synchrony + i yield per plant + i seed shattering + i persistence of awns in threshing. L: unadapted landrace; S: Hordeum vulgare ssp. spontaneum line.

of this geographical region. The eld experiments to study these two incorporation schemes were conducted in the northern conditions of Finland using early maturity as the evaluation criterion of the progeny. It was realized that crosses having a parent specically adapted to the northern conditions had a higher frequency of transgressive segregants for early heading than those in which adapted parents originated from the more southern part of the region (Fig. 14.3). Exotic sources affected transgression for earliness (Vetelinen et al., 1996), but the main conclusion was that incorporation of exotic material was most successful in the local genetic background. The selection and inclusion of adapted genotypes in base-broadening is also important to the success of such programmes, especially those targeted to harsh marginal areas.

Evaluation of exotic

adapted progeny

The evaluation strategy used for the exotic adapted progeny is of special importance when the aim is to renew the genetic base of the crop. Strict evaluation of exotic adapted progeny can hinder the success of pre-breeding and prevent new plant types from entering breeding populations. In this study, evaluation based on an index composed of several agronomic traits (including traits indicating wild status) was compared with breeders phenotypic evaluation of a sample of the third-generation progenies. Correlation between the two measures was only moderate (Table 14.3). This indicated deciency of the index composed and/or the conservatism of the breeders phenotypic evaluation (for example, plants with unusual outlooks were discarded). In the latter case, breeders may have to reconsider their evaluation methods in the early phase of parent development when working with exotics in order to allow new plant types to enter breeding populations.

Genetic Base-broadening of Barley

267

EAR EMERGENCE Finnish crosses (a) 100

80

Frequency (%)

60

40

Fi 1 Fi 2 Fi 3 Fi 4 Fi 5

20

0 60 6170 Days 7180 81

Swedish crosses (b) 100

80

Frequency (%)

60

40

Sw 1 Sw 2 Sw 3 Sw 4 Sw 5

20

0 60 6170 Days 7180 81

Fig. 14.3. Frequency distribution for ear emergence (days from planting to heading) in (a) Finnish (Fi 1Fi 5) and (b) Swedish (Sw 1Sw 5) progeny.

268

M. Vetelinen and E.A.J. Nissil

Table 14.3. Spearman correlation coefcients between index (7a = composed of seven traits, 4b = four traits and 3c = three traits) and breeders evaluation. Trait Breeders evaluation Index 7 4 3 2-row 0.44** 0.26** 0.21** 6-row 0.55** 0.23** 0.12*

a Traits included: ear emergence, days; straw length, cm; number of ears per plant; heading synchrony; yield, g per plant; seed shattering; persistence of awns in threshing. b Ear emergence, days; straw length, cm; number of ears per plant; heading synchrony. c Ear emergence, days; straw length, cm; number of ears per plant. *,**: signicant at 5% and 1% levels respectively.

It was noted that parentprogeny regressions for several traits were low, which indicated that the parents were poor predictors of their progenys performance and supported the hypothesis that inter-loci interactions were importnant in agronomic performance of barley. Consequently, the overall genetic potential of exotic barleys should only be tested after incorporation in an appropriate genetic background (Vetelinen et al., 1997).

Utility of the programme to breeders The dynamic genepools were basically established to develop three different types of sources of genetic diversity currently important in Nordic barley breeding programmes. The development of genotypes with multiple resistance to several barley diseases through extensive recombination was the aim in the development of both dynamic genepools. For this purpose exotic genotypes were used which possessed resistance to at least one major pathogen, and 15 phenotypically diverse genotypes were selected for the dynamic genepools. The adapted parents were also selected with a specic focus on the differences in disease resistance. It was planned to screen the progeny from the sixth generation for multiple resistance and to incorporate them into resistance breeding programmes. With the Finnish genepool a special aim was to develop novel variation for adaptive traits critical in breeding for the northern marginal areas. Current six-row breeding material originates from the Nordic landraces and the genetic base of this material is narrow (Manninen and Nissil, 1997). The most critical adaptive traits for which useful new variation is needed are lodging resistance + drought resistance + early maturity and netblotch resistance + early maturity. The emphasis with the Finnish genepool was also to develop wider variation for the critical yield components and new plant types to enhance yield potential of early maturing six-row barley. In the short-season conditions of the north, the number of tillers per plant is more or less xed and likely to be only one to three tillers per plant to ensure early maturity. In addition, all the current high-yielding and early-maturing six-row barleys tend to have relatively small seed size. Therefore, the aim was also to produce new desirable recombinants expressing good yielding potential, yield stability and larger seed size.

Genetic Base-broadening of Barley

269

The evaluation of the genepools for multiple resistance and adaptive quantitative traits has not yet been extensively started, although in the establishment phase some very promising new recombinants were taken into the six-row barley breeding programme. A genome map for mapping of quantitative trait loci (QTLs) based on the cross between two highly adapted six-row barley lines was developed concurrently with the dynamic genepools. It is planned to use this for identifying desirable wild QTLs from the dynamic genepool and for testing genepools in divergent and directional selection sites.

Resource use The development phase of the dynamic genepools was carried out as a collaborative project between research and breeding sectors. Plant breeders beneted with new parental materials, while the research sector could provide education in the elds of applied plant breeding research and pre-breeding. The human resources used in the establishment were one technical assistant per site for crossing work and management of the plant material from planting to harvesting; one PhD student who carried out the research activities including laboratory and eld work; and a consultative group including academic supervisors and a plant breeder. The time taken for the establishment of the genepools was 3 years per site.

Conclusions
The studies carried out in the establishment phase of the dynamic genepools of Nordic barley showed the following: 1. Exotic germplasm used in the establishment of dynamic genepools provided new genetic and agronomic variation for Nordic barley. In order to determine the genetic potential of the exotic germplasm, these lines were tested in progeny derived from complex adapted exotic crosses. 2. Phenotypes of exotic lines were poor predictors of progeny performance. Thus, exotic germplasm should be tested when incorporated in an appropriate genetic background. A crossing scheme that enhanced extensive recombination was useful for recovering new desirable genotypes. 3. Exotic sources can be used in improving overall agronomic performance. If shortterm progress is desired, unadapted cultivated sources are preferred to wild ones. 4. Reconsideration of evaluation methods for progeny after exotic adapted crosses may be needed in order to allow new plant types to enter breeding populations. This is of special importance if a breeder wishes to renew the genetic base of the breeding material. 5. Incorporation was most successful when done in the local genetic base. 6. The adaptive value of a genotype can depend on several traits and their interaction with the environment. Therefore, further testing of transgressive segregants in different environments is needed to verify their value as parental material.

270

M. Vetelinen and E.A.J. Nissil

References
Brown, A.H.D., Nevo, E., Zohary, D. and Dagan, O. (1978) Genetic variation in natural populations of wild barley (Hordeum spontaneum). Genetica 49(2/3), 97108. Chalmers, K.J., Waugh, R., Watters, J., Forster, B.P., Nevo, E., Abbott, R.J. and Powell, W. (1992) Grain isozyme and ribosomal DNA variability in Hordeum spontaneum populations from Israel. Theoretical and Applied Genetics 84, 313322. Fischbeck, G. (1992) Barley cultivar development in Europe success in the past and possible changes in the future. Barley Genetics IV, 885901. Hallauer, A.R. and Miranda, J.B. (1981) Quantitative Genetics in Maize Breeding. Iowa State University Press, Ames, Iowa. Jana, S. and Pietrzak, L.N. (1988) Comparative assessment of genetic. Genetics 119, 981990. Linde-Laursen, I., Nielsen, G. and Johansen, H.B. (1987) Distribution of isoenzyme markers at 37 loci in a pedigree of European spring barley. Hereditas 106, 241251. Manninen, O. and Nissil, E. (1997) Genetic diversity among Finnish six-rowed barley cultivars based on pedigree information and DNA markers. Hereditas 126, 8793. Martin, J.M., Blake, T.K. and Hockett, E.A. (1991) Diversity among North American spring barley cultivars based on coefcient of parentage. Crop Science 31, 11311137. Melchinger, A.E., Graner, A. and Singh, A. (1994) Relationships among European barley germplasm: I. Genetic diversity among winter and spring cultivars revealed by RFLPs. Crop Science 34, 11911199. Nei, M. (1975) Molecular Populations Genetics and Evolution. Elsevier/North Holland Publishing Company, Amsterdam. Nevo, E. (1979) Genetic diversity and environmental associations of wild barley, Hordeum spontaneum, in Israel. Evolution 33(3), 815833. Nevo, E. (1992) Origin, evolution, populations genetics and resources for breeding of wild barley, Hordeum spontaneum, in the Fertile Crescent. In: Shewery, P.R. (ed.) Barley: Genetics, Biochemistry, Molecular Biology and Biotechnology. CAB International, Wallingford, UK, pp. 1943. Nevo, E., Beiles, A. and Zohary, D. (1986) Genetic resources of wild barley in the Near East: structure, evolution and application in breeding. Biological Journal of the Linnean Society 27, 355380. Saghai-Maroof, M.A., Zhang, Q. and Chojecki, J. (1994) RFLPs in cultivated barley and their application in the evaluation of malting quality cultivars. Hereditas 121, 2129. Tigerstedt, P.M.A. (1994) Adaptation, variation and selection in marginal areas. Euphytica 77, 171174. Vetelinen, M. (1994) Exotic barley germplasm: variation and effects on agronomic traits in complex crosses. Euphytica 79, 127136. Vetelinen, M., Nissil, E., Tigerstedt, P.M.A. and von Bothmer, R. (1996) Utilization of exotic germplasm in Nordic barley breeding and its consequences for adaptation. Euphytica 92, 267273. Vetelinen, M., Suominen, M. and Nissil, E. (1997) Agronomic performance of crosses between Nordic and exotic barleys. Euphytica 93, 239248. Wych, R.D. and Rasmusson, D.C. (1983) Genetic improvement in malting barley cultivars since 1920. Crop Science 23, 10371040. Zhang, Q., Saghai-Maroof, M.A. and Kleinhofs, A. (1993) Comparative diversity analysis of RFLPs and isozymes within and among populations of Hordeum vulgare ssp. spontaneum. Genetics 134, 909916.

15
1School

Evolutionary Changes in Cambridge Composite Cross Five of Barley

K.M. IBRAHIM1 AND J.A. BARRETT2

of Biological Sciences, University of East Anglia, Norwich NR4 7TJ, UK; 2Department of Genetics, University of Cambridge, Downing Street, Cambridge CB2 3EH, UK

Introduction
The breeding methods commonly used in barley, Hordeum vulgare, rely on selecting desirable genotypes from genetically heterogeneous landraces (pure-line breeding) or from the segregating progeny of crosses between lines with superior traits (pedigree breeding). Another method that can be used is the bulk-population method, where the segregating progeny of multiple crosses between superior lines are harvested and propagated in bulk. Nilsson-Ehle is believed to have been the rst breeder to use the bulkpopulation method (Allard, 1960). He adopted it in handling the segregating generations of hybrids made to combine winter hardness and high yield in wheat. The action of natural selection was assisted by discarding plants that had suffered winter damage during successive generations. Growing a large population increased the chances of high-yielding types appearing among the favoured winter-hardy selections and also disposed of the need for keeping detailed pedigree records. After a few generations of bulking, homozygosity among the selfed plants would be at a level such that selected plants could be expected to breed true. In 1929, Harlan and Martini proposed the composite cross method (CC method) of breeding barley. This method is essentially similar to the bulk-population method except that the hybridization stage involves crossing a large number of varieties of diverse origin, and genetic make-up, in order to create a heterogeneous population of recombinant genotypes, as well as the parental types. This population would then be grown under normal agricultural conditions in the areas where locally adapted varieties were needed. Natural selection acts on the available variation during the successive generation, leading to an increase in the frequency of locally adapted genotypes without creating a genetically uniform population. Mak and Harvey (1982) state that this method not only produces new genetic variation during the hybridization and segregation stages but also preserves the variation in an exploitable form. Suneson (1956) referred to the technique as an evolutionary plant breeding method.
IPGRI/FAO 2001. Broadening the Genetic Base of Crop Production (eds H.D. Cooper, C. Spillane and T. Hodgkin)

271

272

K.M. Ibrahim and J.A. Barrett

Composite cross populations can be created in a number of ways. Hybridization may be carried out either by hand-crossing or using genetically controlled male-sterility to facilitate out-crossing. The majority of the early composite crosses of barley were synthesized by hand-crossing diverse accessions and mixing the F2 seeds in known proportions or by using bridge crosses in a hierarchical way to produce a nal stock of hybrid seed. An example of the former is composite cross two (CCII) which was created by mixing equal numbers of F4 seeds from 378 hybrids obtained from pairwise crosses among 28 varieties of barley (Harlan and Martini, 1929). Composite cross ve (CCV), on the other hand, was created by crossing 30 diverse varieties in pairs followed by pairwise crosses with the Fl hybrids for three generations. Composite crosses can be used as: 1. Dynamic reservoirs of genetic variation in projects of germplasm conservation especially when maintenance of locally adapted diversity in an exploitable form is required. 2. Experimental populations for use in studies of the effects of natural selection on heterogeneous populations of cultivated crops that are maintained under natural conditions. 3. Sources of locally adapted new varieties by single plant, or progeny selection once the composite cross populations attain high levels of homozygosity. 4. As animal feed. 5. As food crops where legislation does not preclude the marketing of heterogeneous varieties and where performance under adverse conditions, and not maximum yield under optimal input, is the main objective. These and other uses of the composite crosses of barley have been reviewed by Allard (1988) and Simmonds (1993). Suneson (1956) showed that the yields of four composite crosses was lower than that of a commercial variety of barley, Atlas, for the rst eight to 15 generations, but all exceeded Atlas after between ten and 15 generations. Similar trends of rapid increase in yield to the levels of commercial pure-line varieties, followed by slower but steady increase have been observed in CCII, CCV and CCXXI (Allard, 1988). The major drawback of the CC method is the length of time required to produce new varieties. While 56 years is normal for the development of cereal varieties using standard breeding methods, successful barley varieties derived from CCs have all been selected from well-advanced generations, all beyond F15. This led Simmonds (1962) to argue that the CC method should not be seen as a breeding technique per se, but rather as an adjunct to breeding where reservoirs of locally adapted variability can be maintained permanently. This is, however, only valid to the extent that the adaptive phenotypes are also agronomically desirable. Various studies of metrical characters in CCs have conrmed that the populations become tter and more adapted to local environments (Jain and Marshall, 1967; Luckett and Sharif, 1987; Allard, 1988; Ibrahim and Barrett, 1991). Among the selected phenotypes were longer straw, late owering, prostrate growth habit and smaller seed, which are, in the main, agronomically undesirable (see also Goldringer et al., Chapter 13, this volume). However, Ibrahim (1990) has argued that the ease of implementation of the CC method, when compared with pedigree breeding, has many advantages for crop improvement in developing countries, for example in Eritrea.

Evolutionary Changes in CCV of Barley

273

The Origin of Cambridge CCV

CCV was developed by the late H.V. Harlan of the United States Department of Agriculture from intercrosses among 30 varieties of barley chosen to represent the major barley-growing regions of the world (Suneson, 1956). The hierarchical crossing scheme was started in 1938 when 15 lots of hybrid seeds were produced from paired crosses between the selected varieties. In the following 2 years, two cycles of pairwise crossings generated four lots of hybrid seed. Lastly, in 1940, these were crossed in all six possible combinations (no reciprocals) and the harvest was bulked to form the F1 generation of CCV. Subsequent generations were propagated in Davis, California, under normal agricultural practices, growing 10,00015,000 plants every season (Jain and Allard, 1960). In 1974, seed samples from the F10, F20 and F30 generations were brought to Cambridge, England, and have been maintained ever since as parallel populations designated populations 1, 2 and 3 of CCV. Successive generations are produced by sowing about 10,000 seeds from the previous harvest as spring barley with no conscious selection and no application of fungicides or pesticides. Thus populations 1, 2 and 3 of Cambridge CCV, although from the same initial segregating population, had undergone 10, 20 and 30 generations of selng and selection, respectively, under Californian conditions, when they were established in Cambridge. Maintaining the populations in parallel enables the detection of selection-induced evolutionary changes such as directional changes in allele frequencies or changes in the phenotypic traits of the three populations. It may also be possible to attribute signicant changes that are not consistent in the parallel populations, either in magnitude or direction, to the differences in the genetic composition of the initial samples of seed transferred to Cambridge.

The Usefulness of Cambridge CCV

By the beginning of the 21st century, the three populations of Cambridge CCV had been maintained under the same environmental conditions and agricultural practices in a major barley-growing region of the UK for almost a quarter of a century. They offer a unique opportunity to evolutionary and population geneticists as well as to applied breeders who want to assess the potential agronomic value of composite cross populations. In the following sections, we describe briey some studies conducted so far using Cambridge CCV material.

Hostpathogen interaction in Cambridge CCV Powdery mildew caused by Erysiphe graminis f. sp hordei is an economically important fungal disease of barley. Ibrahim and Barrett (1991) derived inbred lines from selected generations of the three populations of Cambridge CCV and screened them for variation in mildew resistance. Host reaction to open-air infection as well as to selected isolates of known virulence genotype was scored. Increased resistance to both was observed in the advanced generations of all three populations. There were signicant differences between the populations in the rate of change towards higher resistance.

274

K.M. Ibrahim and J.A. Barrett

Population 1 had the highest proportion of resistant plants (Fig. 15.1). Mildew attack, which is known to cause yield losses of up to 20% in untreated commercial varieties, appears to have conferred selective advantage to the individual plants that carry resistance genes. This work also identied specic mildew resistance loci where strong directional changes in allele frequencies over successive generations were apparent. A comparison of inbred lines extracted from population 1 and the 30 parental lines (Fig. 15.2) using a number of mildew isolates of known virulence genotypes identied the resistance genes in the host plants which increased in frequency in response to mildew attack (Ibrahim and Barrett, 1991). The predominantly selng Cambridge CCV populations consist of a mixture of lines that are highly homozygous. With regard to the evolution of disease resistance, these lines are expected to be affected by epidemic development in the same manner as variety mixtures (Barrett, 1978). In general, when their composition with respect to a known trait is observed to change in a particular direction over time, that trait is usually identied as a target of natural selection. Such directional change can also be brought about by the hitch-hiking of the genes which control that trait with other genes that are the target of selection. However, comparative studies have identied mildew resistance as a strong candidate target of natural selection (Ibrahim et al., 1996). Of course, overall, directional selection targets the combination of genes that produce the best phenotype.

Variation in discrete and metrical characters The three Cambridge CCV populations have also been studied extensively for variation in many discrete morphological characters for which the mode of inheritance is known. These include: growth habit (prostrate, erect), chaff colour, aleurone colour, awn type (smooth, rough), ear type (brittle, tough) and rachilla hair (long, short). Metrical characters such as owering time, plant height, seed number, seed weight and other measurements on the inorescence have also been investigated. Luckett and Sharif (1987) reported that within 6 years of their introduction to Cambridge, the CCV populations
Population 1 76 62 47 40 22 18 10 F12 F15 F18 F21 F22 F25 F28 F31 F32 F35 F38 F41 21 25 20 40 44 Population 2 Population 3

Fig. 15.1. Frequency (%) of plants resistant to open-air spores of powdery mildew in the three parallel-grown populations of Cambridge CCV.

Evolutionary Changes in CCV of Barley

275

74 CC/1 CC/99 CC/128 36.6 22.45 16.6 6.67 F12 lines F21 lines 16.33 66 66

74

Parental

Fig. 15.2. Frequency (%) of resistance to three mildew isolates in the CCV parental lines (n = 30) and inbred lines from F12 (n = 49) and F21 (n = 50) of population 1. Isolate CC/1 has the Va8 and Va12 mildew virulence genes. CC/99 and CC/128 are more complex. See Ibrahim and Barrett (1991) for their virulence genotypes.

evolved towards later owering, increased height at maturity and higher yield compared with the original introductions. Ibrahim (1989) observed similar trends in comparisons between generations from all three populations spanning 10 years propagation under Cambridge conditions. In the case of owering time, it was apparent that an increase in the frequency of few genes with major effect was the cause of the shift towards later owering. A bi-modal distribution of owering time was observed and, in the advanced generations, more lines fell in the late-owering class compared with the early generations, as shown in Fig. 15.3 for the samples from population 1. A similar shift towards later owering was also observed when CCXLII was grown over several generations in Cambridge (Knight, 1991). The concurrent generations from populations 2 and 3 also showed similar trends. A nested experimental design using 50 inbred lines from each of ve population 1 generations spanning 10 years showed that mean owering time had increased by 1 day per generation, on average. The mean number of days from sowing to the rst appearance of awns from the ag leaf were 42, 45, 52 and 52 for generations F12, F15, F18 and F21, respectively. Comparisons of means were statistically signicant except for that between F18 and F21. Furthermore, there was signicant variation in the mean owering time of each of the lines within each generation. Lines extracted from the Californian CCV generation concurrent to F21 of Cambridge CCV owered on average 42 days after sowing. The most striking change in the discrete morphological characters was the marked increase in the number of plants with prostrate growth habit (Table 15.1). The single recessive gene responsible for prostrate growth habit, sh, is known to cause the pleiotropic effect of late owering compared with the dominant allele (Whitehouse et al., 1972). The shifts observed in the frequencies of the sh allele could therefore be the result of selection acting on either phenotype. A 2 test of the association between early or late owering and growth habit in the four generations is shown in Table 15.2, which also gives the frequencies of the four phenotypic combinations. The partitioning

276

K.M. Ibrahim and J.A. Barrett

F12

F15

F18

F21

Fig. 15.3. Distribution of the average time to owering (days from sowing date) in lines derived from four generations of population 1, Cambridge CCV. The bars represent 3-day intervals starting at 3033 days and ending with owering time of more than 69 days from sowing.

of the total variation in this three-dimensional contingency table reveals that the frequency of growth habit and owering time variants varied signicantly across generations. The signicant interaction between the two is the expected outcome of pleiotropy. The marginally signicant triple interaction, which is problematic to interpret when the two-way interactions are signicant (Everitt, 1977), implies that the association between any pair of the three variables may differ in degree or in direction in the different categories of the third. It is clear that early-owering erect lines have declined considerably in the advanced generations. Neither could, however, individually account for the shift in the other, contrary to the ndings of a previous study (Luckett, 1980) where prostrate growth habit was proposed as the selected phenotype. In the case of the morphological characters that showed frequency uctuations across generations, a number of causes can be hypothesized. The chance association of different alleles with alleles at loci that are targets of selection could bring about this effect. Changes in the environment that result in uctuations in the optimal phenotype selected may also lead to signicant non-directional phenotype shifts. For those characters in which directional trends were apparent, it is again difcult to ascertain that the characters are the actual targets of selection but it is possible to use frequency shifts in these loci to demonstrate the effect of selection on the CCV populations. If the characters studied are just markers for other selected alleles, the strongly inbreeding nature of barley makes them effectively targets of selection, because when recombination is restricted due to inbreeding the whole genome is effectively linked (Jain and Allard, 1960; Jain, 1961a,b).

Evolutionary Changes in CCV of Barley

277

Table 15.1. Morphological variation of 50 inbred lines in each of four generations of population 1, Cambridge CCV. Character Chaff colour Aleurone colour Anthocyanins Awn type Rachilla length Phenotype Black White Blue White Present Absent Smooth Rough Long Medium Short Long Short Brittle Tough Long Short Erect Prostrate F12 4 46 25 25 10 40 31 16 1 26 20 8 39 11 36 11 39 28 22 F15 12 38 24 26 11 39 25 24 4 36 9 2 47 9 40 13 37 29 21 F18 3 47 34 16 2 48 35 13 2 31 15 1 47 7 41 13 37 10 40 F21 1 49 36 14 4 46 42 8 4 29 17 1 49 10 40 9 41 7 43
2

15.56** 9.36* 10.06* 13.09**

8.37* 12.78* 9.41* 1.24 34.75**

Rachilla hair Ear type Neck type Growth habit

* = P < 5%; ** = P < 1%.

Table 15.2. Percentage of different phenotypic combinations in 50 inbred lines in each of four generations of population 1, Cambridge CCV and 2 tests of associations. Phenotype Erect and early Erect and late Prostrate and early Prostrate and late Source Total Two-way interactions Generation owering time Generation growth habit Flowering time growth habit Three-way interaction Generation owering time growth habit F12 28 0 8 14 F15 20 9 8 13 df 10 3 3 1 3 F18 5 5 5 35
2

F21 4 3 7 36

141.10 40.49 34.74 57.30 8.57

278

K.M. Ibrahim and J.A. Barrett

Variation in biochemical markers Edwards (1975) investigated the isozyme variation of the rst seeds of CCV harvested in Cambridge and, in a subsequent study, Luckett and Edwards (1986) reported that polymorphism at four esterase loci changed signicantly in successive generations grown in Cambridge but the trends were not similar in the three populations. They concluded that the esterase genes were hitch-hiking through chance initial associations with other selected genes, when CCV was introduced into Cambridge. An extensive survey of polymorphism at three storage protein loci (hordeins) was carried out by Ibrahim et al. (1996). The B hordeins are encoded by a single structural gene, Hor2, located on the short arm of chromosome 5, 78 cM distal to the Hor1 locus which codes for the C hordeins (Jensen et al., 1980; Shewry et al., 1981). The Hor3 locus of the D hordeins is located on the long arm of chromosome 5 (Shewry et al., 1983). The estimated map distance between Hor1 and Hor3 is 65 cM and the gene order is Hr2Hor1Hor3 (Shewry et al., 1984). Hordein polypeptides of the three classes separated using SDS-PAGE revealed extensive band pattern variation in the three populations of Cambridge CCV. The frequencies of single locus banding patterns as well as pattern combination for Hor1 and Hor2 showed marked directional changes in successive generations. This trend was highly correlated to the observed increase in resistance to powdery mildew (Ibrahim et al., 1996). Because the genes that encode the hordeins Hor1 and Hor2 are linked to major mildew and yellow rust loci, Mla, Mlat, MlGa, Mlk, Mlnn, Mlra and Yr4 (Shewry et al., 1981; Kreis et al., 1985; Jorgensen, 1994), it was concluded that the observed shifts in hordein banding patterns were the combined result of this linkage, the selective advantage of resistance to mildew and the inbreeding nature of barley. The association between mildew resistance and specic B hordein and C hordein polypeptide variants makes the latter good markers for selecting for mildew resistance. For example, the overall probability of randomly selecting a resistant line in the sample of 400 lines screened (P1) was 0.45. If lines are rst selected on the basis of the Hor B and Hor C patterns most commonly associated with mildew resistance, the probability (P2) is 0.81 and 0.77, respectively. When Hor B and Hor C band combinations are used, the probability of an otherwise randomly chosen line being resistant to mildew is 0.84. Thus, gains in selection efciency, ((P2 P1)/P1), of 7585% could have been attained by using these markers when screening for resistance to infection by the prevailing mildew population in the eld.

Maintenance of genetic diversity in Cambridge CCV The dynamics of the genetic make-up of the three populations of Cambridge CCV observed in the various studies described above is complex, but there is strong evidence to support the conclusion that resistance to powdery mildew has been selectively advantageous. Other phenotypic variants such as prostrate growth habit, tallness and early owering also appear to have been favoured. However, in spite of the detectable effect of natural selection favouring particular phenotypes, the surveys of variation in morphological, quantitative, disease resistance and biochemical characters in Cambridge CCV have clearly shown that a substantial amount of genetic variation still

Evolutionary Changes in CCV of Barley

279

exists in the populations. Most of the individual variants, be they morphological phenotypes or biochemical markers, were still present in the advanced generations, albeit in frequencies signicantly different from the early generations and the parental lines. However, the number of phenotypic combinations was signicantly reduced. For example, in the ve dimorphic morphological characters (chaff colour, rachilla hair, ear type, awn type and growth habit) shown in Table 15.1, the total number of possible phenotypic combinations is 25 = 32. The actual numbers detected in F12, F15, F18 and F21 were 16, 15, 9 and 9, respectively, clearly indicating some loss of diversity over time. The same trend was also apparent in quantitative characters. The variance between inbred lines within generation declined with time, showing reductions of 24.9%, 39.6% and 47.9% for owering time, height and number of seeds, respectively (Ibrahim, 1989). In a further study, using both SDS-PAGE of hordeins and restriction fragment length polymorphism (RFLP) analysis using a probe for the Hor2 locus, Danquah (1993) showed not only the same association of hordein variation and mildew resistance, but also that the most frequent B hordein variant was the same as that for the parental line cv. Algerian, which is known to carry the mildew resistance genes Ml-a1 and Ml-at. The hordein loci showed the greatest amount of genetic diversity of the markers studied in Cambridge CCV so far. Ibrahim et al. (1996) compared two samples of 100 seeds seven generations apart (F12 and F21) and were able to detect in the advanced generation (F21) all but the four rarest Hor2 alleles of the 19 present in the early generation (F12). The mean expected panmictic heterozygosity averaged over the three hordein loci was 0.55, 0.64, 0.44 and 0.40 in F12, F15, F18 and F21, respectively. This indicates that genetic diversity at the hordein loci has not been seriously eroded when measured on a per locus basis. However, there was a signicant decline in the diversity of multilocus pattern combinations from 26 in the F12 samples to 15 in F21.

Conclusion
The usefulness of composite cross populations depends on how their genetic make-up is affected by natural selection. Their use as mass reservoirs of genetic diversity was rst proposed by Simmonds (1962). Marshall and Brown (1974), however, have argued that mass reservoirs are of little value in preserving variation because they retain only a small proportion of the parental diversity. The Cambridge CCV studies outlined above have clearly shown that while a substantial amount of single locus diversity is still retained in CCV some 40 generations after the initial hybridization, multilocus genotypic diversity has declined demonstrably. Because this decline in the overall heterogeneity of CC populations is correlated with adaptedness to local environment (Ibrahim, 1989; Allard, 1990, 1996), it can be seen as an effective method of retaining locally useful diversity. On the other hand, the magnitude of the reduction in multi-locus genotypic combinations observed within the relatively short period of the propagation of Cambridge CCV clearly precludes the use of CCs for conservation purposes per se. The use of CC populations as food, or feed, crops can only be considered under two conditions: (i) where variety release and seed regulatory legislation do not preclude their use and plant breeders rights are not sought; and (ii) where emphasis is on

280

K.M. Ibrahim and J.A. Barrett

breeding for reliability of yield in less than optimum management conditions. Both conditions are characteristic of the agricultural systems of many developing countries. Apart from the initial crossing phase, no highly skilled labour or detailed record keeping is required, and the combination of natural selection to local conditions, combined with selection by local growers to their own needs in the later generations, can make this approach very cost-effective. Ibrahim (1990) has argued that for the example for Eritrea, adoption of the CC method could be benecial particularly in barley. In the traditional barley cultivation in Eritrea, genetically diverse landraces are grown for a variety of uses and in proximity to the wild relatives of barley. The potential free movement of genes, even with limited outcrossing, the possibility of coevolution of the crop with its pests and diseases, and the diversity arising from selection by growers for different uses of the grain as well as straw produce has created these landraces. However, the long liberation war in Eritrea and the famines of the 1980s have forced farmers to consume their traditional seed stocks and thereby break the chain that maintains the locally adapted diversity. The unavoidable encroachment of modern agriculture which uses uniform varieties is yet another potential cause of the irretrievable loss of diversity that appeared after the end of the war. We argue that the creation of composite crosses from the most desirable surviving landraces, perhaps combined with some introgression of elite material from other areas, could form the basis of a national barley improvement programme for Eritrea, without the risk of loss of genetic diversity and vulnerability to diseases and pests. Several questions need to be addressed in order to assess the potential use of CCs for this purpose.

How many parental lines to use and how diverse do they need to be both genetically and in terms of their geographic origin? Should the choice aim at providing maximum diversity in specic characters such as drought tolerance or disease resistance? Should the method of maintenance simulate the traditional farming systems or the recommended practices? Should limited articial selection be allowed in order to direct the evolution of the population? How many generations of propagation are required before the population can be used by the local growers?

Furthermore, conditions of seed storage and schemes for rejuvenation of stored seeds would also need to be developed. Yield trials at different locations, different seasons and possibly different management practices would need to be conducted. Finally, on utilization, should the farmers maintain their own seed stocks of the CC populations?

References
Allard, R.W. (1960) Principles of Plant Breeding. John Wiley & Sons, New York. Allard, R.W. (1988) Genetic changes associated with the evolution of adaptedness in cultivated plants and their wild progenitors. Journal of Heredity 79, 225238. Allard, R.W. (1990) The genetics of hostpathogen co-evolution: implications for genetic resources conservation. Journal of Heredity 81, 16.

Evolutionary Changes in CCV of Barley

281

Allard, R.W. (1996) Genetic basis for the evolution of adaptedness in plants. Euphytica 92, 111. Barrett, J.A. (1978) A model of epidemic development in variety mixtures. In: Scott, P.R. and Bainbridge, A. (eds) Plant Disease Epidemiology. Blackwell, Oxford, pp. 129137. Danquah, E.Y. (1993) Natural selection and components of tness in Composite Cross Five (CCV) of barley. PhD Thesis, University of Cambridge, Cambridge, UK. Edwards, K.J.R. (1975) Natural selection and biochemical properties of polymorphic esterases in barley. Barley Genetics 3, 2329. Everitt, B.S. (1977) The Analysis of Contingency Tables. Chapman and Hall, London. Harlan, H.V. and Martini, M.L. (1929). A composite hybrid mixture. American Society of Agronomy Journal 21, 487490. Ibrahim, K.M. (1989) Genetic variation in Composite Cross Five of barley. PhD Thesis, University of Cambridge, Cambridge, UK. Ibrahim, K.M. (1990) The need to conserve the crop genetic resources of Eritrea. RIEC Bulletin 1, 711. Ibrahim, K.M. and Barrett, J.A. (1991) Evolution of mildew resistance in a hybrid bulk population of barley. Heredity 67, 247256. Ibrahim, K.M., Hayter, J.B.R. and Barrett, J.A. (1996) Frequency changes in storage protein genes in a hybrid bulk population of barley. Heredity 77, 231239. Jain, S.K. (1961a) Stabilising selection for heading time in a bulk hybrid population of barley. Nature 191, 11231124. Jain, S.K. (1961b) Studies on the breeding of self-pollinated cereals, the composite cross bulk population method. Euphytica 10, 315324 Jain, S.K. and Allard, R.W. (1960) Population studies in predominantly self-pollinated species. I: Evidence for heterozygote advantage in a closed population of barley. Proceedings of the National Academy of Sciences USA 46, 13711377. Jain, S.K. and Marshall, D.R. (1967) Genetic changes in barley populations analysed in terms of some life cycle components of selection. Genetica 38, 355374. Jensen, J., Jorgensen, J., Jensen, H., Giese H. and Doll, H. (1980) Linkage of the hordein loci Hor1 and Hor2 with the powdery mildew resistance loci Ml-k and Ml-a on barley chromosome V. Theoretical and Applied Genetics 58, 2731. Jorgensen, J.H. (1994) Genetics of powdery mildew resistance in barley. Critical Reviews in Plant Science 13, 97119. Knight, S.J. (1991) Genetic variation in Composite Cross XLII of barley. PhD Thesis. University of Cambridge, Cambridge, UK. Kreis, M., Shewry, P.R., Forde, B.G., Forde, J. and Miin, B.J. (1985) Structure and evolution of seed storage proteins and their genes with particular reference to those of wheat, barley and rye. In: Oxford Surveys of Plant Molecular and Cell Biology, Vol. 2. Oxford University Press, Oxford, pp. 253317. Luckett, D.J. (1980) An investigation of natural selection in a genetically variable barley population. MPhil Thesis, University of Cambridge, Cambridge, UK. Luckett, D.J. and Edwards, K.J.R. (1986) Esterase genes in parallel composite cross populations. Genetics 114, 289302. Luckett, D.J. and Sharif, A.L. (1987) Metrical characters in parallel-grown sub-populations of barley Composite Cross Five. Australian Journal of Biological Science 40, 6577. Mak, C. and Harvey, B.L. (1982) Exploitable genetic variation in a composite bulk population of barley. Euphytica 31, 8592. Marshall, D. and Brown, A.H.D. (1974) Optimum sampling strategies in genetic conservation. In: Frankel, O.H. and Hawkes, J.G. (eds) Crop Genetic Resources for Today and Tomorrow. Cambridge University Press, Cambridge, pp. 5380. Shewry, P.R., Wolfe, M.S., Slater, S.E., Parmar, S.R., Faulks, A.I. and Miin, B.J. (1981) Barley storage proteins in relation to varietal identication, malting quality and mildew resistance.

282

K.M. Ibrahim and J.A. Barrett In: Barley Genetics IV, Proceedings of the Fourth International Barley Genetics Symposium, Edinburgh, pp. 596603. Shewry, P.R., Finch, R.A., Parmar, S., Franklin, J. and Miin, B.J. (1983) Chromosomal location of Hor3, a new locus governing storage proteins in barley. Heredity 50, 179189. Shewry, P.R., Miin, B.J. and Kasarda, D.D. (1984) The structural and evolutionary relationships of the prolamin storage proteins of barley, rye and wheat. Philosophical Transactions of the Royal Society, London B 304, 297308. Simmonds, N.W. (1962) Variability in crop plants, its use and conservation. Biological Reviews 37, 442465. Simmonds, N.W. (1993) Introgression and incorporation. Strategies for the use of crop genetic resources. Biological Reviews 68, 539562. Suneson, C.A. (1956) An evolutionary plant breeding method. Journal of Agronomy 48, 188191. Whitehouse, R.N.H., Jenkins, G. and Johnson, J.D. (1972) Annual Report, Plant Breeding Institute. Cambridge, UK, pp. 8082.

16

Genetic Base-broadening in the West Indies Sugar Cane Breeding Programme by the Incorporation of Wild Species

A.J. KENNEDY
West Indies Central Sugar Cane Breeding Station, Groves, St George, Barbados

Introduction
Sugar cane is a large perennial grass. In agriculture, it is propagated vegetatively as stem cuttings, each node having a bud and a ring of root initials. Varieties of cane are therefore clones and once selected are xed genetically. Although sugar canes are tropical in origin, through breeding and selection the geographical range of their production has been extended to sub-tropical areas such as South Africa, Louisiana and Texas in the USA and S~ Paulo in Brazil. ao Modern sugar cane varieties are complex interspecic hybrids and are highly polyploid with a degree of aneuploidy. Their possible evolution is shown in Fig. 16.1, which is simplied from Roach (1995). Sugar cane is a crop of ancient cultivation, the history of which is summarized by Roach (1995). As a crop, it has been subject to improvement through breeding since the late 19th century. Although progress was initially good, by the 1960s it was realized that the exceedingly narrow genetic base of the existing clones was beginning to impede further progress. Walker (1962) noted direct evidence of this in the West Indies. The need for the introduction of new genetic material into breeding programmes was generally accepted. Several new collections of wild relatives of sugar cane were made, particularly of Saccharum spontaneum. A wider collection of the cultivated forms of Saccharum ofcinarum was also accumulated. This chapter describes a genetic base-broadening programme in the West Indies that utilizes this wide diversity of genetic material. Essentially, the programme repeats the nobilization stage of the crops evolution, on a wider genetic base.

Origin of sugar cane The basic genome of x = 10 is generally accepted for the Saccharum genus. There are ve main species recognized as being involved in the sugar cane complex, and the tax IPGRI/FAO 2001. Broadening the Genetic Base of Crop Production (eds H.D. Cooper, C. Spillane and T. Hodgkin)

283

284

A.J. Kennedy

Unknown ancestors

Narenga spp. Erianthus spp. Saccharum spontaneum

Miscanthus spp. Sclerostachya spp.

The Saccharum complex

Erianthus arundinaceus S. spontaneum E. arundinaceus Two New Guinea S. spontaneum clones S. robustum S. barberi

Modern hybrid sugar cane clones

Six S. officinarum clones

S. sinense

The part of the evolution in bold outlines has been repeated on a large scale by the WI Base-broadening Programme

Fig. 16.1. A simplied evolution of sugar cane. From Roach (1995).

onomy and evolution of these are reviewed in detail by Daniels and Roach (1987) and summarized by Roach (1995). A brief description of these species is given here. 1. Saccharum spontaneum L. (2n = 40128): this is a highly polymorphic species with a wide ecological range in the tropics and sub-tropics from 8N to 40S. Its centre of origin and of diversity is in India, but it is widely distributed from Africa to the Solomon Islands and even Japan. 2. Saccharum robustum Brandes and Jeswiet (ex Grassl) (2n = 60 or 80): this species is a name given to a wide variety of populations that probably arose through hybridiza-

Genetic Base-broadening of Sugar Cane

285

tion and introgression between S. spontaneum and other genera, such as Erianthus and Miscanthus in New Guinea. 3. Saccharum barberi Jesw. (2n = 81124): Daniels et al. (1991) descibe the taxonomy of this species and separate it into six different horticultural groups. It is native to northern India where it was cultivated for sugar production. It is thought to be a natural hybrid between S. ofcianarum and S. spontaneum. 4. Saccharum sinense Roxb. (2n = 115120): this species is very similar to S. barberi and is described by Daniels et al. (1991). It too is a hybrid between S. ofcianarum and S. spontaneum. It was cultivated in India and China. 5. Saccharum ofcinarum L. (2n = 80): this species is found only in cultivation, with centres of both origin and diversity in New Guinea. Selected clones of S. ofcinarum formed the original basis of sugar cane plantation cultivation and are known as the noble canes. 6. Modern commercial cultivars. As can be seen from the above descriptions, sugar canes are very complex in their genetic make-up, often being the product of both interspecic and intergeneric hybridization. It is now accepted that modern sugar canes as we know them have arisen from what is termed the Saccharum complex, which comprises the interbreeding genera Saccharum, Erianthus, Sclerostachya, Narenga and Miscanthus. However, the two major contributors are S. ofcinarum and S. spontaneum.

History of breeding and variety development During the 16th century sugar cane cultivation shifted from S. sinense and S. barberi to a more extensive cultivation of S. ofcinarum, the noble canes. These were cultivated successfully for over 300 years. However, they were unthrifty, requiring frequent replanting, and were susceptible to many diseases. The noble canes spread throughout the tropics wherever sugar was produced. In 1858 it was recognized for the rst time that sugar cane could produce fertile true seed. This was reported in the Barbados Liberal newspaper in February 1859 (Parris, 1954). A worker on Highlands Plantation named Harper spotted the seedlings and reported them to the overseer named Parris. Although 4.5 acres of seedlings were grown on the estate, they appear not to have had favourable characteristics (possibly because they would have been selfs from the original clone, Cheribon, and displayed inbreeding depression) and nothing more was done. In 1888 Harrison and Bovell in Barbados (Harrison and Bovell, 1888) and Soltewedel in Java (Kobus, 1893) simultaneously rediscovered the fertility of sugar cane and began breeding programmes. The rst bred variety in Barbados was named Burke and was selected from seedlings produced in 1888. From 1902 to 1919 Bovell began a series of controlled pollination crosses and seedling selection that gave rise to BH10(12) and Ba11569. All of the breeding in Barbados and Java was initially done using the existing noble varieties. Between 1910 and 1925 in Java, breeding was to take a new path. Forms of noble canes were crossed with the wild relative S. spontaneum, mimicking what had occurred in the speciation process in nature. This combined the sugar-producing characters of nobles with the vigour and disease resistance of the spontaneum clones. Similar work was done in India using a local noble called Vellai and Indian forms of spontaneum.

286

A.J. Kennedy

Two very important varieties came out of this work. POJ2878 from Java and Co205 from India. These two varieties were soon distributed very widely. In the Caribbean, noble (S. ofcinarum) breeding continued until interspecic hybrids were imported from Java and India in 1928. From then, hybrids between these and the locally improved nobles formed the basis of the breeding programme. Some very well-known varieties such as B 34104 and B 37161 were produced. Because of the high degree of heterozygosity and polyploidy, these programmes were initially able to generate a large variability and so were very successful for many years, gradually producing better and better varieties adapted to local conditions. By the 1960s, however, the initial variability had been exploited and progress slowed considerably (Walker, 1972).

Collections and the beginning of genetic base-broadening programmes During the 1950s and early 1960s, it became apparent that there were limitations on the rate at which improvements could be made because of lack of genetic variability and that gains made from the initial phase of hybridization could not be sustained. All of the breeding then in existence was based on parental material originating from only two forms of S. spontaneum and six forms of S. ofcinarum (Arceneaux, 1965). In the West Indies, the results of this narrow base were indicated by diminishing returns from conventional breeding (Walker, 1962). Many cane breeders began then to accumulate collections of wild forms and species to supplement their breeding by the introduction of new genes. The designated world collections of sugar cane germplasm are maintained by the Governments of India and the USA at centres in Kerala and Florida, respectively. Collections of S. ofcinarum, S. spontaneum, S. robustum, Erianthus arundinaceus, Miscanthus oridulus and M. sinensis were made from Papua New Guinea in 1957 and 1977, and from the Indonesian Islands in 1976 and 1984 (Warner and Grassl, 1958; Berding and Koike, 1980; Krishnamurthi and Koike, 1982; Tew et al., 1986). Over 400 clones of S. spontaneum were collected from the Indian subcontinent (Panje, 1956). Further collections have been made from the Philippines (Medina et al., 1986), northern Thailand (Srinivasan and Sadakorn, 1983) and Taiwan (Lo and Sun, 1969). The world collections, therefore, contain a very wide range of wild materials that may be at the disposal of sugar cane breeders. The current germplasm collection, held by the West Indies Central Sugar Cane Station, is some 3030 clones from different origins. These are given in Table 16.1. The clones in this germplasm are kept as a live eld collection occupying about 15 ha. It must be managed such that elds are replanted frequently to maintain adequate growth for good owering. The actual numbers in each category are not stable since new hybrids and new selections are added each year while older clones may be discarded as being less desirable. The proportions of each group do stay more or less the same. The commercial clones in the collection are predominantly of regional origin but there is an ongoing policy to import and exchange breeding varieties with other breeding institutions around the world. There are clones from Australia, Brazil, Cuba, South Africa, Fiji, Taiwan, India, the USA (Louisiana, Hawaii and Florida), Mauritius,

Genetic Base-broadening of Sugar Cane Table 16.1. West Indies Central Sugar Cane Station Collection. Type of accession Commercial clones (regional and worldwide origin) Near commercial products of base-broadening (see below) Base-broadening backcrosses BB1 and BB2 clones Base-broadening noble S. spontaneum F1 S. ofcinarum (nobles) S. robustum S. spontaneum Erianthus arundinaceus

287

Number of clones 1780 394 302 287 110 29 114 14

Reunion and Mexico in the collection. The regional varieties are supplemented every year with the latest selections from member countries that comprise the West Indies Sugarcane Breeding and Evaluation Network (WISBEN): that is, Guyana, Trinidad, Barbados, St Kitts, Dominican Republic, Jamaica and Belize. In the early interspecic crossing programmes it was generally considered that the noble canes would contribute the ability to store sucrose, while the vegetative vigour, good ratooning and disease resistance would come from S. spontaneum. More recently it became clear that there was scope for selection for higher sucrose storage capacity among spontaneum clones (Brown et al., 1969). Some degree of genetic base-broadening has been utilized in several sugar cane breeding programmes other than in Barbados, most notably in the USA (Louisiana and Hawaii), Australia, Taiwan, India and South Africa. Berding and Roach (1987) comprehensively review all these efforts. In Louisiana the two aims were to obtain cold resistance and resistance to mosaic. S. spontaneum accessions were screened for these characters thus limiting the genotypes that were used in crossing. In Hawaii, basebroadening was used to obtain selections that would tolerate the extremes of environment found in those islands, such as high elevation and low-nutrient, acid soils. In Taiwan, effort was concentrated on using Miscanthus spp. as a source for tolerance to drought and soil salinity, and resistance to downy mildew. Thus, in general, these programmes are characterized by having well-dened, but limited, goals using the introgression of specic traits. On the other hand, in the West Indies a more general incorporation approach (Simmonds, 1993) was taken with no prior selection among the large number of S. spontaneum clones used and the philosophy was that of incorporation of new germplasm on a population basis. The details of this work will be described later.

Nobilization The process of crossing S. spontaneum on to S. ofcinarum is known as nobilization. The noble parent is usually the seed bearer or female. Although S. ofcinarum (2n = 80) in intraspecic crosses shows normal n + n inheritance, when crossed with S. spontaneum it displays the unusual pattern of transmission of the somatic chromosome number. The progeny are, therefore, 2n + n (Bremer, 1925; Price, 1961; Roach, 1969).

288

A.J. Kennedy

Thus, if a 64 chromosome spontaneum is used, the F1 progeny have 112 (80 + 32) chromosomes, whereas the F1 from a 112 chromosome spontaneum will have 136 (80 + 56) chromosomes. This pattern is also followed in the subsequent generations when the F1 is backcrossed1 to noble canes although here the degree of aneuploidy is highly variable and there is usually a loss of spontaneum chromosomes. By the second backcross to a noble, only one-quarter, or less, of the spontaneum complement remains in the hybrid. Crosses between F1s backcrossed populations and commercial varieties all show n + n transmission. Since, with this mode of chromosome transmission, the proportion of spontaneum chromosomes is reduced with each generation of backcrossing, it is generally considered that it is unproductive to beyond the rst backcross generation before resorting to crossing on to commercial hybrid varieties where n + n inheritance is found. Roach (1978) suggests that n + n chromosome transmission predominates where spontaneum chromosomes are present in both parents. Heterosis for vigour and sugar content has been shown in F1 F1 crosses that combine chromosomes from two different S. spontaneum clones (WICSCBS, 1975). Using in situ DNA hybridization on chromosomes it has been possible to distinguish the chromosomes coming from S. ofcinarum from those coming from S. spontaneum in the commercial sugar cane variety R 570 (DHont et al., 1996). This work also showed the new fact that there has been crossing over between the chromosomes of the two species. This has implications for the degree of new variability generated from such crosses. This was previously thought not to occur. If it has happened in hybrid varieties based on the rst two S. spontaneum clones used in hybrid production, then presumably it may also occur in our base-broadening hybrids using new sources of spontaneum chromosomes. Only about 10% of the chromosomes in the hybrid variety came from S. spontaneum.

Genetic Base-broadening in the West Indies

The objectives of the base-broadening work in Barbados were very general. The characters of interest were adaptations to various environments within the Caribbean region and better overall vigour and ratooning ability. The range of environments that the breeding programme in Barbados serves spans from Guyana (8N), on the South American mainland, through the Caribbean island chain as far north as Jamaica (18N). This represents an enormous diversity of sugar cane environments. Although the crossing is done at the Central Station, selection is distributed across the region. The West Indies Sugar cane Breeding and Evaluation Network (WISBEN) comprises the sugar industries of Guyana, Trinidad, Barbados, St Kitts, Dominican Republic, Jamaica, Belize and the Central Breeding Station. Different forms of S. spontaneum had been used in the programme in Barbados as early as 1935 and had given rise to a very popular variety, B 41211, which was known locally as juicy cane and was popular for chewing. From the 1960s, a concerted effort was made to introduce a wide genetic base from wild species. A deliberate programme
1In

this work, and in the base-broadening programme described later, the generation-wise process is usually termed backcrossing, even though the recurrent noble may be a different genotype in each generation.

Genetic Base-broadening of Sugar Cane

289

to broaden the genetic base of our breeding clones was established by 1965 and has been described by Walker (1972), and Berding and Roach (1987). Base-broadening clones are the products of deliberate crosses using new wild germplasm. The approach was to repeat the evolution of the crop on a very wide genetic base. Alternating clonal (parental) and seedling cycles (selection) with a short generation time were used. Large populations and weak selection pressure carried a large number of potential parents from a wide range of crosses forward to each successive generation. Starting from an original cross between a clone of S. spontaneum and a noble cane (S. ofcinarum), the resulting seedlings are selected for the best types. These are termed an F1 generation. These, in turn, are then crossed to either a different noble variety or to a commercial sugar cane clone to produce the rst BB1 generation selections, which are crossed to a different commercial sugar cane clone to give the second BB2 generation. This process is repeated until truly commercial-level selections are produced. These we term near-commercial or semi-commercial clones. Seed from the second and later BB generations is sent to the WISBEN countries for selection in their very diverse environments. By starting from many sources of S. spontaneum and S. ofcinarum, and by avoiding inbreeding at the various backcross generations, a substantially new genepool has been created from which to select improved, more productive clones. Further, by distributed selection in WISBEN territories, genepools with specic adaptation to each local environment have been developed. The early generation base-broadening clones are selected in Barbados but the near commercial clones are, as indicated above, selected throughout WISBEN. The overall breeding programme makes between 800 and 900 crosses every year. On average about 25% of these are of the base-broadening crosses of various generations, although they have at various times taken up between 30% and 40% of the crossing effort. The fertility of interspecic crosses is low and about half of the crosses attempted fail to set any seed (WICSCBS, 1978). There are two unique aspects of the base-broadening effort in Barbados when compared with other programmes worldwide. First, there was a deliberate attempt to improve the noble population through a generation-wise polycross and selection of resulting progeny. Three generations of selection were performed with the aim of improving sugar content and reducing bre content (Walker, 1972). The starting point was 80 noble clones selected on their visual appearance from the USDA World Collection combined with 35 noble varieties bred in Barbados (Walker, 1972). These 115 clones were planted in a high rainfall area 300 m above sea level in St Vincent to encourage owering. Ninety-two clones owered and were put into a polycross; 57 of them gave viable seed. The seedlings were planted in both Barbados and Trinidad. Selection was for sucrose content, softness (correlated with bre content), ratooning ability and vigour. The second and third cycles of crossing were made in Barbados. At each generation about 100 selected clones formed the breeding population. Improvement for sugar content was limited, probably because of the lack of variation in the original population. There was, however, signicant reduction in bre content and general improvement of vigour. Such a programme was possible in Barbados because of the large collection of nobles available and the relative ease with which nobles ower in the Caribbean. Nobles hardly ower at all in most other locations that have breeding programmes and elaborate photoperiod facilities are required to induce owering. The second distinctive feature of our programme has been its wide scope. In all,

290

A.J. Kennedy

over 70 new S. spontaneum clones have been incorporated into the programme and many breeding lines developed with two or more different spontaneums in their pedigree. The whole programme may be viewed as population improvement aimed at incorporation of completely new wild germplasm into our existing commercial genetic background. The breeding programme in the West Indies is largely funded by a levy on the revenues from sugar sales of all the owner member states. It is to the credit of these individually small industries that they were prepared to keep faith with such a long-term project. It was, after all, over 20 years before commercial breeding began to benet directly from the base-broadening efforts. They not only supported the genetics and breeding work, but also supported and funded essential research on the physiology of owering that allowed photoperiod treatments to synchronize the owering of S. ofcinarum, S. spontaneum and commercial varieties of sugar cane. The effective networking of distributed selection across WISBEN guaranteed that a wide genetic diversity was maintained, not only in the base parental material, but also in the selected products of base-broadening crosses. The base-broadening programme in sugar cane is described in the various annual reports of the station (WICSCBS, 1975, 1978, 1984, 1985) and has been summarized by Simmonds (1993).

Synchronization of owering times The owering times of the very diverse sugar cane germplasm in Barbados range from September to the end of December. The noble canes tend to ower from late October to the end of November. The collection of S. spontaneum clones, however, has a very wide origin from 9.5S of the equator to 32N. They display an equally wide range of owering times in Barbados (13N). The natural owering range is from July or August through to the following January (Walker et al., 1977). These authors point out that generally the owering time is related to the latitude of their origin. To make many desirable crosses it was necessary to develop photoperiod control systems in order to synchronize the owering of the genotypes to be crossed. Simple procedures to achieve this were developed by MacColl (1977) and Midmore (1980). We now routinely use night light breaks to delay owering in early-owering clones to bring them in line with the mainstream and late-owering clones.

The use of Saccharum robustum and Erianthus arundinaceus Crosses with S. robustum have in general proved to be disappointing by comparison with those from S. spontaneum. Although the hybrids show potential for sugar content they do not show the same vigour and lack ratooning ability (characters of S. robustum itself ). The exploitation of S. robustum was discontinued for this reason. Interest in the use of hybrids between commercial sugar cane and E. arundinaceus began in 1980, largely as a response to the possibility of producing varieties, not for sugar production but a source of biomass production for other purposes. However, the initial hybrids looked very promising as potential sugar canes (WICSCBS, 1984, 1985). The vigour, appearance and sugar content of the F1 hybrids were comparable

Genetic Base-broadening of Sugar Cane

291

with those of the BC2 hybrids derived from S. spontaneum. The morphology (D.I.T. Walker, personal communication) and chromosome numbers (P.S. Rao, unpublished) suggested that true hybrids had been achieved. However, the use of PCR markers has so far failed to nd Erianthus DNA in the putative hybrids (A. DHont, personal communication). It is possible that what were thought to be exciting hybrids were in fact the result of self-pollination of the commercial female parent. Both restriction fragment length polymorphism (RFLP) and PCR techniques have been used to conrm pedigrees of crosses involving wild relatives of sugar cane (DHont et al., 1997).

Products from the base-broadening programme There is evidence now that products from the base-broadening programme are regularly reaching late selection stages in all of our member countries. This is illustrated by Table 16.2 showing the accumulated results from crosses made between 1979 and 1994. As more advanced generations of crosses between commercial varieties and basebroadening clones are made, the distinction between the two programmes becomes more and more blurred. We are at the point now where an increasing number of crosses each year involve parents that can be described as semi-commercial which have been derived from base-broadening programmes. Presently about 20% of WISBENs crosses each year are of this nature. In Barbados, two varieties have been released that were derived from new S. spontaneum germplasm: B 79474 and B 80251. B 79474 has in its pedigree three old wild sources SC12/4, a noble; Toledo, a spontaneum; and 28NG251, a robustum to which have been added a new noble, 57NG68, and a new spontaneum, SES49. B 80251 has one old spontaneum, Moentai, to which a new noble, BNS55, and a new spontaneum, SES84/58, have been added. To reduce vulnerability to disease, a relatively large number of varieties are grown in each country throughout the West Indies (Rao and Gardiner, 1997). The products of the base-broadening programme are now

Table 16.2. Results of crosses made between 1979 and 1994 in the West Indian Sugarcane Breeding and Evaluation Network (WISBEN). Number of clones with new germplasm in their pedigree Country Guyana Trinidad Barbados Jamaica Dominican Republic Belize Total Reaching third selection stage 40 * ** 7 13 19 79 Reaching nal selection stages 16 13 24 6 3 3 65

* Trinidad does not have this selection stage in its programme. ** Large numbers have reached both Stage 2 and Stage 3 selections from the base-broadening crosses for many years and have been incorporated back into the commercial breeding programme.

292

A.J. Kennedy

beginning to contribute to this broad genetic diversity in the eld. Although WISBENs programme has been long term (35 years) and required considerable effort and resources, it is felt that its benets are now becoming apparent and it was well worth the devotion to it. The generation of large populations of clones from the West Indies sugar cane genetic base-broadening was directed to the selection of new varieties for the production of sucrose. It is now in the phase where its products are being absorbed into the commercial breeding or combined with clones developed especially for very high sucrose content. The range of genetic diversity in these new clones remains very large. There is potential among the base-broadening selection for the development of canes with alternative uses. The vigour in early generations means that they have potential as producers of biomass for such uses as bre production to make paper or board. Fibre content of sugar canes is in the range of 1015% with most being about 13%. In the base-broadening clones, bre content as high as 20% is encountered, with potential to select even higher types. Similarly, types with high total sugar content, but a relatively low sucrose component, are encountered. Such clones could be developed into varieties suitable for molasses or syrup production or for the production of rum. The impressive genetic variation generated by this kind of base-broadening is such that appropriate selection pressures could generate varieties with a wide range of uses. As such, it is maintained as a resource that may be more widely exploited in the future.

Acknowledgements
I cannot conclude without paying a tribute to Professor N.W. Simmonds, whose help, ideas and enormous encouragement kept the project alive for so many years, and to Mr D.I.T. Walker and Dr P.S. Rao, who carried out much of the early work and whose ideas guided the programme from its inception.

References
Arceneaux, G. (1965) Cultivated sugarcanes of the world and their botanical derivation. Proceedings of the International Society for Sugar Cane Technology XII, 844854. Berding, N. and Koike, H. (1980) Germplasm conservation of the Saccharum complex: a collection from the Indonesian archipelago. Hawaii Planter Record 59, 1176. Berding, N. and Roach, B.T. (1987) Germplasm collection, maintenance and use. In: Heinz, D.J. (ed.) Sugar Cane Improvement Through Breeding. Elsevier Scientic Publishers, Amsterdam, pp. 143210. Bremer, G. (1925) The cytology of the sugarcane. The chromosomes of primitive forms of the genus Saccharum. Genetica 7, 293322. Brown, A.D.H., Daniels, J. and Latter, B.D.H. (1969) Quantitative genetics of sugarcane. II. Correlation analysis of continuous characters in relation to hybrid sugar cane breeding. Theoretical and Applied Genetics 39, 110. Daniels, J. and Roach, B.T. (1987) Taxonomy and evolution. In: Hienz, D. (ed.) Sugar Cane Improvement Through Breeding. Elsevier Scientic Publishers, Amsterdam, pp. 784. Daniels, J., Roach, B.T., Daniels, C. and Paton, N.H. (1991) The taxonomic status of Saccharum barberi Jeswiet and S. sinense Roxb. Sugar Cane 1991(1), 1116.

Genetic Base-broadening of Sugar Cane

293

DHont A.P., Grivet, L., Feldmann, P., Rao, S., Berding, N. and Glaszmann, J.-C. (1996) Characterisation of the double genome structure of modern sugarcane cultivars (Saccharum spp.) by molecular cytogenetics. Molecular and General Genetics 250, 405413. DHont, A., Rao, P.S., Alleyne, S., Glaszmann, J.-C. and Feldmann, P. (1997) Application of molecular markers in sugar cane breeding. Proceedings of the WIST XXVI Conference, Guyana, pp. 7779. Harrison, J.B. and Bovell, J.R. (1888) Report on the Results Obtained upon the Experimental Fields at Dodds Reformatory, Barbados. (Quoted in Stevenson, 1965.) Kobus, J.D. (1893) Historisch Overzicht over het Zaaien van Suikerriet. Archief Java Suikerind 1. Krishnamurthi, M. and Koike, H. (1982) Sugarcane collecting expedition, Papua New Guinea, 1977. Hawaii Planter Record 59, 273313. Lo, C.C. and Sun, S. (1969) Collecting wild cane in Taiwan. In: Liu, K.-C. (ed.) Proceedings of the International Society of Sugar Cane Technologists XIII, 10471055. MacColl, D. (1977) Some aspects of owering of sugarcane in Barbados and its control in a breeding programme. Annals of Botany 41, 191207. Medina, R.E., Krishnamurthi, M., Lapastora, E., Dosayla, R., Divinagracia, N., Silverio, G. and Cano, I.B. (1986) Sugarcane germplasm collection in the Republic of the Philippines. In: Clayton, J.L. and Handojo, H. (eds) Proceedings of the International Society of Sugar Cane Technologists XIX, 522527. Midmore, D.J. (1980) Effects of photoperiod on the owering and fertility of sugarcane. Field Crops Research 3, 6581. Panje, R.R. (1956) Studies in Saccharum spontaneum: a note on the objectives and scope of work. Proceedings of the International Society of Sugar Cane Technologists IX, 750756. Parris, G.K. (1954) James W. Parris: discoverer of sugarcane seedlings. The Garden Journal, SeptOct. Price, S. (1961) Cytological studies in Saccharum and allied genera.VII. Maternal chromosome transmission by S. ofcinarum in intra- and interspecic crosses. Botanical Gazette 122, 298305. Rao, P.S. and Gardiner, J. (1997) Varietal change and diversity in the Sugar Association of the Caribbean member countries. Proceedings of the WIST XXVI Conference, Guyana, pp. 4144. Roach, B.T. (1969) Cytological studies in Saccharum. Chromosome transmission in interspecic and intergeneric crosses. In: Liu, K.-C. (ed.) Proceedings of the International Society of Sugar Cane Technologists XIII, 901920. Roach, B.T. (1978) Utilization of Saccharum spontaneum in sugarcane breeding. In: Reis, F.S. and Dick, J. (eds) Proceedings of the International Society of Sugar Cane Technologists XVI, 4358. Roach, B.T. (1995) Sugar canes. In: Smart, J. and Simmonds, N.W. (eds) Evolution of Crop Plants. Longman, London, pp. 160166 Simmonds, N. (1993) Introgression and incorporation: strategies for the use of crop genetic resources. Biological Reviews 68, 539562. Srinivasan, T.V. and Sadakorn, J. (1983) Exploration and collection of Saccharum germplasm in Thailand. IBPGR South East Asia Newsletter 7, 79. Stevenson, G.C. (1965) Genetics and Breeding of Sugar Cane. Longmans, Green and Co., London, pp. 4445. Tew, T.L., Purdy, L.H., Lamadji, S. and Irawan (1986) Indonesian sugar cane collecting expedition 1984. Sugar Cane 1986, 1517. Walker, D.I.T. (1962) Family performance at early selection stages as a guide to the breeding programme. In: Williams, J.R. (ed.) Proceedings of the International Society of Sugar Cane Technologists XI, 469482. Walker, D.I.T. (1972) Utilisation of noble and Saccharum spontaneum germplasm in the West

294

A.J. Kennedy Indies. In: Henderson, M.T. (ed.) Proceedings of the International Society of Sugar Cane Technologists XIV, 224232. Walker, D.I.T., MacColl, D. and Rao, P.S. (1977) Aspects of the use of Saccharum spontaneum in the West Indies programme. In: Reis, F.S. and Dick, J. (eds) Proceedings of the International Society of Sugar Cane Technologists XVI, 291303. Warner, J.N. and Grassl, C. (1958) The 1957 cane expedition to Melanesia. Hawaii Planters Record 55, 209236. WICSCBS (1975) Annual Report of the West Indies Central Sugar Cane Breeding Station 197475. WICSCBS (1978) Annual Report of the West Indies Central Sugar Cane Breeding Station 197778. WICSCBS (1984) Annual Report of the West Indies Central Sugar Cane Breeding Station 198384. WICSCBS (1985) Annual Report of the West Indies Central Sugar Cane Breeding Station 198485.

17
1Federal

Potential of Genetic Resources and Breeding Strategies for Basebroadening in Beta

L. FRESE1, B. DESPREZ2 AND D. ZIEGLER1

Centre for Breeding Research on Cultivated Plants (BAZ), Gene Bank, Braunschweig, Germany; 2Florimond Desprez Co., BP 41, 59242 Cappelle en Pvle, France

Introduction
In Europe, cultivated forms of Beta (leaf beet, garden beet, fodder and sugarbeet) are grown on more than 5.6 million ha, to which can be added approximately 550,000 ha in North America; 600,000670,000 ha in China (with a promising production and market potential for sugarbeet); and a number of smaller areas in Chile, Egypt, Iran, India, Japan, Morocco, Tunisia, Turkey and Syria. Altogether, sugarbeets are grown on about 7.5 million ha worldwide. Breeding efforts are focused on the sugarbeet crop, while leaf, garden and fodder beet breeding is of regional importance only. The genus Beta is native to Europe and adjacent areas. Sections Nanae (Greece) and Procumbentes (Canary Islands) have a limited distribution area, while wild species of section Beta occur along the coastline from the south of Sweden to Morocco and from the Canary Islands to Iran. Section Corollinae occurs at altitudes higher than 800 m. It has a large distribution area in Turkey and neighbouring countries. The centre of diversity is probably located where the species distribution of sections Beta and Corollinae overlap (eastern Turkey and the western part of Transcaucasia). The domestication of beets probably started in the Euphrates and Tigris region and continued in Turkey and Greece, from where cultivated beets were introduced to northern Europe (Boughey, 1981). Cultivated beets have occurred in China since the 5th century (Sun Yi Chu, 1994) and also in Arabic countries. One of the youngest cultivated forms, the sugarbeet, has become a cash crop of worldwide importance which has been cultivated on a large scale only since 1811, when Napoleon decreed that beet should be grown for sugar. As the sugarbeet was probably selected from one single cultivated population only, the White Silesian, the genetic base of the crop is supposed to be very narrow. The White Silesian beet had a rather low sugar content. However, the German scientist Achard considered the good root shape of this fodder-beet-like type as a favourable trait and started to select within this population for higher sugar content and yield. The sugar content increased from 4% to 16.5% between 1784 and 1981 (Winner, 1981).
IPGRI/FAO 2001. Broadening the Genetic Base of Crop Production (eds H.D. Cooper, C. Spillane and T. Hodgkin)

295

296

L. Frese et al.

Before 1960, open-pollinated multigerm varieties were developed using family selection methods and the breeding material had a comparatively broad genetic variation (Desprez and Desprez, 1993). Compared to potato, barley and other economically important crops, sugarbeet did not seriously suffer from pest and disease attacks or adverse environmental conditions in the main production areas (Lewellen, 1992). Though almost all beet pests and diseases were already known, the sugarbeet crop was considered as a relatively healthy crop until the 1960s. However, because of the growing acreage, the sugarbeet was increasingly cultivated in short crop rotation, amplifying disease problems. During the 1960s the rising cost of hand labour became an even more pressing problem. The ordinary sugarbeet seed ball contains three to four seeds; seedlings emerge in clumps and had to be thinned by hand. Savitsky (1950) found in a seed eld of about 1.5 ha one monogerm, homozygous plant. This character was essentially inherited by a single recessive gene, which became of great economic importance in sugarbeet breeding and production. Lines derived from that plant, such as SLC101 (Savitsky, 1952), were extensively used in breeding programmes. In 1942 Owen (1954) discovered cytoplasmic male-sterile germplasm. After the Second World War, sugarbeet breeders focused their work on the development of monogerm, cytoplasmic malesterile hybrid varieties with a high sugar quality, a high yield of recoverable sugar (Oltmann et al., 1984) and a good level of eld resistance to diseases. Only one monogerm line (C562) was resistant to bolting and all of the rst monogerm varieties from the Hilleshg company, which reached more than 60% of the European market (cv. Monohill and others), were derived from this line. In addition, all hybrid varieties were based on a single source of cytoplasmic male sterility, the Owen cms (Owen, 1948; Bosemark, 1979). After the introduction of cytoplasmic male sterility and monogermness, the female breeding pool went through a genetic bottleneck. Accordingly, breeders have systematically enlarged the female genepool by using their own breeding stock. Even though exotic germplasm was not used, after 30 years of broadening the female genepool, the genepool has sufcient variability. Today, breeders wish to broaden the whole breeding pool as such (female and male pools). Strikingly, for sugarbeet breeders the crop itself still is the most important genetic resource used for the development of improved varieties today. One could argue that the genetic base of the crop as such is not as narrow as generally assumed. It seems rather that it is a lack of specic traits which hampers breeding progress. Indeed, because commercial plant breeders use a large number of different, heterozygous pollinator populations, hybrid varieties still have much genetic variation. Additionally, sugarbeet is a wind-pollinated, strongly outcrossing crop. Therefore, plant breeders today may prot from exotic germplasm that was introgressed in the sugarbeet breeding pool, either by chance or deliberately by breeders. The rst introductions of wild germplasm into sugarbeet probably occurred at the beginning of the 20th century, in Russia, the USA and Italy. Cultivated wild beet crosses were, for example, described by Tjebbes (1933), who used Beta vulgaris ssp. maritima from the North Sea coast with a sugar content ranging from 15.7% to 17.6%; and Munerati (1932), who crossed a population from the Po estuary with sugarbeets to introduce genetic variation for resistance to Cercospora beticola. The Munerati material, at least, has been widely used in sugarbeet breeding. Probably because of these early experiences with wild beet crosses, in the 1970s and early 1980s there was a great fear that introgression of undesirable genes of wild or

Base-broadening in Beta

297

exotic germplasm along with the desired disease resistance trait would destroy the results of costly selection on high sugar quality and bolting resistance. However, the view on potential benets arising from the utilization of exotic germplasm began to change when soil-borne diseases such as the beet cyst nematode (Heterodera schachtii) (Hellinga, 1943) or the beet necrotic yellow vein virus (BNYVV) (Grnewald et al., 1983), already identied as very harmful disease agents in sugarbeet elds, began to spread and threaten the sugarbeet production in the whole northern hemisphere. Though many lines were tested, no major disease resistance genes against these important pests and diseases were detected in the crop. In addition, the level of tolerance to the cyst nematode proved to be insufcient in the breeding genepool (Curtis, 1970; Heijbroek, 1977). Interestingly, in some Italian varieties (cv. Roxane, Java, Alba), there was some resistance or tolerance in an agronomically satisfactory genetic background. It is assumed that Italian varieties still contained some wild genes originating from the Munerati material. These sources have been used to develop the rst varieties with Rizomania tolerance/resistance (Desprez and Desprez, 1999). In 1956 Savitsky (1960) detected strong Heterodera schachtii resistance genes in Beta section Procumbentes. However, due to strong crossing barriers between section Beta and section Procumbentes, the utilization of this source proved to be very difcult and time-consuming. It is easy to understand that this specic experience did little to promote a broader use of exotic material in breeding programmes (Desprez and Desprez, 1996). In the 1980s, the continued collecting and evaluation efforts of the USDA/ARS programme yielded more and more exciting results on new sources of resistances, for example to the Rizomania disease (Doney and Whitney, 1990) in B. vulgaris ssp. maritima, which crosses easily with sugarbeet. Since then, the interest in utilization of Beta genetic resource collections has been increasing worldwide. Breeders are mainly searching for disease resistance genes in exotic germplasm to supplement their breeding pool. The introduction of additional genetic variation for sugar content and yield genes from exotic germplasm is thereby welcomed as a positive side effect that can benet breeding progress in the long run.

The Sugarbeet Breeding Research Community

Because of the small number of remaining large sugarbeet seed companies, collaboration between experts is no longer restricted to national projects. In the non-competitive sector particularly, there is a strong willingness to cooperate at international level. Knowledge as well as germplasm is exchanged across the northern hemisphere where sugarbeets are mainly produced. Pre-selected wild material from Europe can be found in Chinese breeding gardens, and Chinese leaf beets in European evaluation programmes. Due to the exchange of scientists between universities and the fusion of breeding companies at international level, national projects are becoming more and more an integral part of international activities. The different partners can be grouped as follows:

Institutes and companies in Europe and the USA, with a strong interest in novel genetic variation for pest and disease resistance, drought and salt tolerance, which are

298

L. Frese et al.

required to develop varieties meeting the demand for an ecologically sound sugarbeet production. Governmental institutes developing varieties (China, India, Iran, Poland, Ukraine, Bulgaria), with interest in access to high-yielding, high-quality breeding material, resistant germplasm and high-temperature tolerance.

Sugarbeet breeders and researchers from the commercial and public sector convene in different associations. The Study Group Genetics and Breeding of the International Institute of Sugarbeet Research (IIRB) meets once a year and organizes joint research projects at international level that interest all breeding companies. Another international forum with a wide coverage of nationalities and scientic disciplines is provided by the World Beta Network (WBN). WBN meetings take place every 3 years and deal with joint activities for genetic resources conservation, documentation and utilization. Smaller, but nevertheless very important groups are the USDA/ARS Sugarbeet Crop Advisory Committee (annual meetings), other similar national associations and the French Beta Network which can also cooperate with partners outside the country. Discussion on the utilization of Beta genetic resources collections has become a permanent topic on the agenda of all these groups.

The Genus Beta and its Useful Characters

The genus, which is the raw material for breeders, consists of four sections, which can be grouped into three genepools (Table 17.1). Breeders have successfully tapped the primary and tertiary genepools, while only a few attempts were made to use the secondary genepool. There are a number of reasons for this. Allthough stronger crossing barriers exist between section Corollinae and section Beta than between species within section Beta, interspecic hybrids can be produced. First attempts to introgress yellowing virus resistance from the Corollinae section into the sugarbeet (Dalke cited in Jassem, 1985) failed probably because of lack of sufcient chromosome homology between the Corollinae and the sugarbeet chromosomes, which is a prerequisite for crossing-over and recombination. In addition, while breeders were still engaged in the introgression of the nematode resistance genes from section Procumbentes into the sugarbeet, large-scale screening of the B. vulgaris ssp. vulgaris and ssp. maritima germplasm yielded donors of interest to breeders. For the time being there is therefore no real pressure for an urgent utilization of the secondary genepool. A third reason is lack of evaluation of the Corollinae section due to a practical problem arising from the fact that all Corollinae species are hard seeded. Thus, the pericarp cap has to be removed manually to facilitate germination, which is very time-consuming and has deterred investigators from large-scale screening. Section Nanae (B. nana) has never been taken into consideration for base-broadening projects. As a species adapted to high altitudes it is very difcult to handle at locations where sugarbeet breeding work is generally conducted. It is not known whether this species has ever been successfully multiplied ex situ, not to mention successfully grown for evaluation purposes. All three genepools contain useful as well as undesirable wild characters (Dale et al., 1985; Van Geyt et al., 1990; Lewellen, 1992; Paul et al., 1992; Stanescu, 1994; Bttner et al., 1997; Mesbah et al., 1997; Yu, 1997; Bosemark, 1998; Michalik et al., 1998; Panella, 1998). Examples are provided in Table 17.2.

Base-broadening in Beta Table 17.1. Taxonomy of the genus Beta. Primary genepool Section Beta syn. Vulgares Ulbrich B. vulgaris L. ssp. vulgaris (cultivated beets) Leaf beet group Garden beet group Fodder beet group Sugarbeet group ssp. maritima (L.) Arcang. ssp. adanensis (Pamuk.) Ford-Lloyd & Will. B. macrocarpa Guss. B. patula Ait. Section Corollinae Ulbrich Base species B. corolliora Zosimovich B. macrorrhiza Steven B. lomatogona Fisch & Meyer Hybrid species B. intermedia Bunge B. trigyna Wald. & Kid. Section Nanae Ulbrich B. nana Boiss. & Heldr. Section Procumbentes Ulbrich syn. Patellares B. procumbens Smith B. webbiana Moq. B. patellaris Moq.

299

Secondary genepool

Tertiary genepool

Bottlenecks to the Utilization of Beta Genetic Resources

The utilization of Beta genetic resources is confronted by several different kinds of problems. Accordingly, different strategies and methods have been developed for the introduction of new genetic material into the sugarbeet breeding pool. The rst problem is the fear of breeders that introgression of wild beet germplasm would require excessively high investments to recover the root shape, root yield, sugar yield and sugar quality of the cultivated parents. Hence, breeders tend to search in the sugarbeet breeding genepool rst sometimes in vain as in the case of resistance to Heterodera schachtii. O. Bosemark, head breeder of a Swedish company, was the rst to say that this fear had no rational basis (Bosemark, 1989). In fact, the discussion initiated by Bosemark helped to promote the breeding approaches of researchers like Munerati (1932). Bosemark demonstrated in selection experiments that after crossing exotic material with sugarbeet, only a few selection cycles are required to regain a reasonable yield and root shape. In view of the narrow genetic base of the crop, he considered the potential prot for a breeder higher than the loss of funds arising from selection against wild characters in cross progenies of cultivated exotic crosses. Annual, quick-bolting wild types of section Beta form the second problem. The

300 Table 17.2. Characters relevant to beet breeding and their distribution over species.
Taxon code Trait Annual life cycle Monogerm seed balls Hard seededness Seed shattering CMS Genic male sterility Salt tolerance Frost tolerance Curly top Yellowing viruses Beet mosaic virus BNYV virus Yellow wilt Peronospora farinosa Erysiphe betae Rhizoctonia solani Cercospora beticola Polymyxa betae Black leg disease Erwinia ssp. Heterodera schachtii Heterodera trifolii Meloidogyne hapla Myzus persicae Pegomya ssp. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

L. Frese et al.

16

17

18

Taxon codes used: (1) B. vulgaris ssp. vulgaris; (2) B. vulgaris ssp. vulgaris leaf beet group; (3) garden beet group; (4) fodder beet group; (5) sugarbeet group; (6) B. vulgaris ssp. maritima; (7) ssp. adanensis; (8) B. macrocarpa; (9) B. patula; (10) B. corolliora; (11) B. macrorrhiza; (12) B. lomatogona; (13) B. intermedia; (14) B. trigyna; (15) B. nana; (16) B. procumbens; (17) B. webbiana; (18) B. patellaris. = variation detected.

evaluation for disease resistance is sometimes difcult to realize directly on such annual types and even if a useful trait is detected it is not always transferable to the crop as it may not be expressed in the genetic background of cultivated material. Because of these difculties it has been suggested that annual wild types are rst crossed with cultivated germplasm and the resulting biennial material is then screened. In the case of the leaf spot disease, for example, early-bolting plants cannot be evaluated precisely in the eld test, while greenhouse and laboratory tests have a low correlation with eld results (W. Mechelke, KWS, personal communication). Screening in the eld is equally difcult because plants are already senescent when the disease develops. In the case of the Cercospora beticola leaf spot disease, it is then very difcult or even impossible to score. Similar difculties are known for various other diseases. In addition, annual types when

Base-broadening in Beta

301

tested in the eld may contaminate the soil with seeds and contribute thereby to the weed beet problem. The third problem arises from crossing barriers between section Beta and sections Corollinae, Nanae and Procumbentes, respectively (Jassem, 1992). They are particularly strong in the case of Procumbentes species. However, section Corollinae and Procumbentes contain characters that have not yet been detected in section Beta, such as insect resistance. Stanescu (1994) noted that B. corolliora may be resistant to attack by Mamestra brassicae L. and Noctuide loxostege stricticalis L. larvae. Only few accessions of these hard-seeded species have been evaluated today. Breeders will perhaps one day nd strong resistance genes acting against C. beticola in the secondary genepool.

Breeding Approaches to Broaden the Genetic Base of Beta

The different breeding programmes which have been developed to overcome these difculties can be grouped in two categories: 1. In previous years, breeders only concentrated on the introgression (Simmonds, 1993; Spoor and Simmonds, Chapter 3, this volume) of desirable traits into the sugarbeet crop as fast as possible. Fast can mean a few years as in the case of the primary genepool or more than a decade, as in the case of the tertiary genepool (see examples below). 2. Today, breeders have also started programmes that do not concentrate on a specic character but aim at broadening the genetic base of the crop in general. This type of approach has been described by Simmonds (1993; see also Spoor and Simmonds, Chapter 3, this volume) as incorporation. Two examples are given below for incorporation from the primary genepool.

Introgression from the primary genepool Examples of conventional breeding approaches are described by Desprez and Desprez (1999) and in more detail by Bttner et al. (1997). The BGRC 54817 (RNR 870909) held by the BAZ Genebank and collected in Normandy, France by Dutch scientists in 1970 showed variation for Rizomania resistance. Through selection of resistant single plants, followed by selng and selection of resistant plants in the S1-line, the character Rizomania resistance could be xed. A parallel backcross programme was started for inheritance studies and for broadening the genetic base of Rizomania resistance in sugarbeets. Bttner et al. (1997) applied a well-known breeding method, which is very successful if the donor parent crosses easily with the sugarbeet and if the character is simply inherited.

Introgression from the tertiary genepool Strong crossing barriers exist between the Procumbentes species (B. procumbens, B. webbiana and B. patellaris) and the section Beta. Phylogenetic research using DNA

302

L. Frese et al.

ngerprinting (Jung et al., 1993) even suggests that the section Procumbentes could be considered as a separate genus. Recent results presented by Shen et al. (1997) also suggest that the section Procumbentes diverged from the other Beta species at a rather early evolutionary stage. This explains why introgression of resistance genes against the cyst nematode Heterodera schachtii into sugarbeet has been so difcult and time-consuming. Savitsky (1960) was the rst who crossed B. vulgaris with nematode-resistant B. procumbens. In Germany, crosses between cultivated forms and the wild species B. procumbens, B. webbiana and B. patellaris were done by Lptien (1984) at the University of Hanover. Today, the resulting varieties are mainly used to decrease the nematode population density in infested elds. The rst steps were production of alien monosomic resistant addition lines; the second step was production of diploid sugarbeet translocation lines carrying a small fragment of the wild chromosome with the resistance gene; the third step was genetic localization of the gene(s), cloning and sequencing of the nematode resistance gene Hs1pro-1, and testing of the function of Hs1pro-1 in transformation experiments (Cai et al., 1997; Jung, 1997). By means of genetically transformed sugarbeet an important disadvantage of the conventionally developed nematode-resistant varieties insufcient agronomic performance may be overcome. The nematode resistance gene originating from B. procumbens encodes a 282amino acid protein, which has features quite similar to disease resistance genes previously cloned from other higher plants. This gene is not only useful for sugarbeet resistance breeding, but may also be of interest to breeding programmes outside the genus Beta, such as rapeseed (Jung, 1997).

Incorporation from the primary genepool Colleagues from the USDA/ARS started a systematic sugarbeet enhancement programme in 1986. Crosses with wild B. vulgaris germplasm were made in 1986, 1990 and 1994. In 1990, male-sterile sugarbeet plants were chosen as the female parent to obtain F1 plants. F1 plants of each cross were intercrossed to produce F2-families, which were then bulked to produce F3 families. At least two recombination cycles were allowed before mild selection on root shape and bolting resistance was started. After ve cycles of mass selection some of the progenies started to resemble sugarbeet. In this case no prior selection on useful characters was done. Testing on disease resistance followed later. Our USDA colleagues are very satised with this programme, which has steadily started to produce USDA germplasm releases (Doney, 1998; Panella, 1998). When ofcially registered by the Journal of Crop Science, the releases can be used by any breeder and fed into elite breeding programmes. French colleagues have recently started a similar but more sophisticated breeding programme. It was launched by the company Florimond Desprez at the request of the Bureau des Ressources Gntiques (BRG). Almost all European breeding companies actively participate in this programme, which uses wild material of French origin. One of the interesting features is that the value of the wild sources is only partly known. Wild material from Corsica, for example, contains variation for Rizomania resistance. Yet, no selection on this specic character is done before crossing. The difference compared with the introgression work done by Bttner et al. (1997) is the long-term

Base-broadening in Beta

303

strategy behind it. It is a declared aim of the project: (i) to allow maximum recombination between the cultivated and wild sources; and (ii) to keep a half-and-half equilibrium of the wild and cultivated genome (Doggett and Eberhard, 1968). For that purpose, so-called Doggett populations containing the alleles MM for multigermicity, aa for genetic male sterility, and the Sf allele for self-fertility are used as female parents (Owen, 1942, 1954). Crosses between MMaaSfSf and wild Beta vulgaris ssp. maritima populations were made in 1996. The presence of the aa-allele allows the identication of male sterile plants in the population. Seeds are harvested on male-sterile plants only, which ensures maximum outcrossing and recombination. The programme is designed in such a way that each participant receives two French wild beet populations and produces two buffer populations. It was agreed to exchange seed samples of the two progenies amongst the 11 participants in 1999 when the pre-competitive character of the programme ended. At that stage each partner owned pools originating from 22 different French wild beet populations. Subsequently, selection started for agronomic characters (especially root shape) and with the introduction of selng cycles to produce inbred lines and crossing cycles to recombine the different sub-populations. Though sugarbeets are self-incompatible, the production of inbred lines is possible because of the presence of the Sf allele in each population. How to exploit this genetically broad material, which will get adapted to the environmental conditions prevailing at each selection site, is now at the discretion of each partner. This differentiation process will also contribute to a diversication of the elite breeding pool amongst companies and breeding institutions. A simplied description of the base-broadening programme is given in Scheme 17.1. The activities of the French Beta Network follow similar principles as described by Mitteau (1997) for bread wheat and barley. The Beta programme allows collaborative projects on genetic resources management since the common material has no property rights on it. This is also an open programme so that any new participant can enter by contributing two new populations which can then be shared with the already participating partners without much disturbance. New participants, however, need to contribute actively to the network programme before they can prot from the work done by the others. In the long run it will produce material that could become useful for many international cooperative studies.

Incorporation from the secondary genepool A fraction of the Corollinae material that is used in a research programme at the University of Kiel (Germany) was collected by a German scientist in Turkey in the 1970s. The Genebank of the Federal Centre for Breeding Research on Cultivated Plants (BAZ) maintains this material and has provided the Chinese University at Harbin in northeast China with Corollinae accessions. The aim of the project is the establishment of a set of alien monosomic addition lines, which were developed by the Chinese counterpart using Corollinae accessions.1 The actual value of the Corollinae sources is not known, and it will be determined only after all theoretically possible alien
1 See http://www.plantbreeding.uni-kiel.de July 1998; personal communication of Professor C. Jung.

Scheme 17.1. Production of a buffer population using wild Beta from France. Step 1: Production of F1 seeds. Doggett population (DP) genotypes 1/2 Aa; 1/2 aa 1 MM 1 S1S1 1 bb 1 aa Female parent (DP) Sugarbeet Wild population (WP) genotypes 1 AA 1 MM 1 S/ ? 1 B/ ? 1 AA

304

Year 1

Generation F0

Male parent (WP) Seabeet

Explanation of the breeding step Production of sugarbeet and seabeet stecklings (= small beets cultivated for seed production, only)

F0

Sugarbeet

Seabeet

Before owering Aa sugarbeet genotypes are discarded. Seabeet plants pollinate male-sterile sugarbeet plants. Seeds are harvested on aa sugarbeet genotypes separately (half-sibs). Seeds of the seabeet population are harvested as a bulk to maintain the original accession.

Explanations of genotypes: AA = genic male-fertile; aa = genic male-sterile type; MM = multigerm; mm = monogerm type; SfSf = self-fertile type; S1n/ ? segregating self-incompatible, sterile type, Sf is dominant to all alleles of the S-series; BB = annual, bb = biennial type, B/? = segregating annual type .

Step 2: Production of the F3 generation starting from F1 seeds. Year 2 3 3 4 4 Generation F1 F1 F2 F2 F3 Male-fertile (AA, Aa) /sterile (aa) genotypes 1 Aa 1 Aa 1/4 AA, 1/2 Aa, 1/4 aa 1/4 AA, 1/2 Aa, 1/4 aa 2/3 Aa, 1/3 aa Explanation of the breeding step Sowing of separately harvested half-sib families. The genome consists of 50% sugarbeet and 50% seabeet genes. Singling of F1 plants to an equal eld stand per half-sib family to ensure about equal genetic contributions of each of half-sib family to the F2. The F2 seed sample is harvested on each half-sib family separately. Sowing of half-sib families to produce F2 stecklings. All aa genotypes are earmarked at owering and seeds are harvested only on aa genotypes. Mailing of F3 seed samples to each participant. Mutual exchange of F3 seed material.

L. Frese et al.

Step 3: Production of the buffer population. Male-fertile (AA, Aa) /sterile (aa) genotypes 2/3 Aa, 1/3 aa

Base-broadening in Beta

Year 4

Generation F3

Explanation of the breeding step Sowing of the separately harvested half-sib families derived from an individual sugarbeet seabeet accession cross DP WP1n to produce aa genotypes of this specic F3. Other F3 families produced by programme partners are sown accordingly. Depending on the number (k) of F3 s within a particular DP WP cross exchanged between participants, a maximum of n k half-sib families will jointly ower and intercross. Seeds are again harvested on aa genotypes, only. Within each DP WP1n family, aa genotypes can be harvested as a bulk. The Doggett population is at gene equilibrium. The individual DP WP1n should be kept separately during the next generations of intercrossing. Plants still contain about 50% sugarbeet and 50% seabeet genome. Sowing of n (DP WP) families for mild mass selection on agronomic characters such as root shape and bolting resistance. Production of inbred lines, if considered to be already useful, is possible through the segregating Sf allele.

F3

2/3 Aa, 1/3 aa

F4

1/2 Aa, 1/2 aa

F4

1/2 Aa, 1/2 aa

WP1 = wild population no. 1; WP2 = wild population no. 2; etc. = generative phase.

305

306

L. Frese et al.

monosomic addition lines have been established. By testing these addition lines for disease resistance, the researchers will try to localize resistance genes on the wild beet chromosomes. After backcrossing with sugarbeet, diploid, resistant recombinants can perhaps be selected as basic material for breeding. This approach has a basic similarity to the French Doggett population concept: the specic value of the wild Corollinae parent is not known precisely.

Future Role of Genebanks in Germplasm Enhancement Programme

As a consequence of the long-term base-broadening programmes initiated in the USA and France, considerable fractions of genetic diversity will be maintained in buffer populations. Genebank accessions which were used to create these buffer populations are further maintained as individual accessions in genebanks. Hence, genetic diversity stored in genebanks will become duplicated in base-broadening programmes. Today, our Beta collections are static. Genebank managers keep nicely classied and documented seed samples in their stores. In future, at least in the case of outcrossing species, genebanks could assume new functions. Besides the maintenance of rationalized static collections, genebanks could become responsible for the maintenance of more dynamically managed genepools resulting from different kinds of programmes. In an outcrossing crop, the role of a genebank manager may become more that of a genepool manager, coordinating and linking: (i) in situ maintenance of natural populations; and (ii) sampling and ex situ conservation of populations; with (iii) base-broadening work. Base-broadening work can last for a long period aiming at the creation of new diversity through continued recombination and evolution. Since genebanks already have long-term responsibilities, these institutions may be in a good position to follow long-term germplasm maintenance and breeding strategies. In France, native B. vulgaris ssp. maritima populations are already managed in situ in their natural environment, which enables seed harvest on request for evaluation purposes, and helps to avoid multiplication and storage of accessions in an ex situ collection. In addition, by grouping a number of wild populations of similar origin, the French base-broadening programme decreases the number of accessions that need to be manipulated. Furthermore, better adapted, dynamically managed material is of higher interest to breeders. In the eld of base-broadening, genebanks could:

Run simple selection projects that are required to adapt wild germplasm to routine screening methods, as in the case of annual wild Beta vulgaris. Maintain buffer populations in the deep-freeze store if there is a temporary lack of funds required for continuation of the work, or if populations have fullled their current purpose. Maintain information on the purpose and breeding history of buffer populations for future users. Maintain donor lines with high frequencies of useful genes, as has already been briey described by Bttner et al. (1997).

Natural evolution, as well as breeding, is a dynamic process. An integrated germplasm conservation and utilization programme linking in situ maintenance, ex situ

Base-broadening in Beta

307

conservation and base-broadening can therefore better mediate between evolution and breeding than static genebank collections. It is at least our impression that the integration of these three elements of germplasm conservation and utilization has made considerable progress in sugarbeet in recent years.

References
Bosemark, N.O. (1979) Genetic poverty of the sugarbeet in Europe. In: Proceedings of Conference Broadening the Genetic Base of Crops. Wageningen, 1978. Pudoc, Wageningen, pp. 2933. Bosemark, N.O. (1989) Prospect for beet breeding and use of genetic resources. In: Report of an International Workshop on Beta Genetic Resources. International Crop Network Series no. 3. International Board for Plant Genetic Resources, Rome, pp. 8997. Bosemark, N.O. (1998) Genetic diversity for male sterility in wild and cultivated beets. In: Frese, L. Panella, L., Srivastava, H.M. and Lange, W. (eds) International Beta Genetic Resources Network. A Report on the 4th International Beta Genetic Resources Workshop and World Beta Network Conference held at the Aegean Agricultural Research Institute, Izmir, Turkey, 28 February3 March 1996. International Crop Network Series 12. International Plant Genetic Resources Institute, Rome, pp. 3445. Boughey, C.L. (1981) Evolutionary and taxonomic studies in wild and cultivated beets. PhD thesis, University of Birmingham. Bttner, G., Frese, L. and Steinrcken, G. (1997) Selektion von Rizomania-Resistenzgenen aus Wildrben (Beta vulgaris L.). Beitrge und Poster. In: IIRB-Kongressen 5/1997. Cuvillier, Gttingen, pp. 111. Cai, D., Kleine, N., Kie, S., Harloff, H., Sandal, NN., Marcker, K.A., Klein-Lankhorst, R.M., Salentijn, E.M.J., Lange, W., Stiekema, J., Wyss, U., Grundler, F.M.W. and Jung, C. (1997) Positional cloning of a gene for nematode resistance in sugar beet. Science 275, 832834. Curtis, G.J. (1970) Resistance of sugar beet to the cyst-nematode Heterodera schachtii Schm. 1970. Annals of Applied Biology 66, 167177. Dale, M.F.B., Ford-Lloyd, B.V. and Arnold, M.H. (1985) Variation in some agronomically important characters in a germplasm collection of beet (B. vulgaris L.). Euphytica 34, 449455. Desprez, M. and Desprez, B. (1993) volution de methods de slection de la betterave sucrire des origines nos jours. Compte Rendu de lAcademie dAgriculture Francais 79(6), 7184. Desprez, M. and Desprez, B. (1996) Importance and difculties of using genetic resources illustrated by the example of beet cyst nematode. Poster presented at the 4th International Technical Conference on Plant Genetic Resources, Leipzig, Germany, 1723 June 1996. Desprez, M. and Desprez, B. (1999) La lutte contre la rhizomania: le rle de la slection et ses perspective. Symposium Rhizomania in Europe Budapest, 1617 Septembre 1998. IIRBInfo no. 3. Doggett, H. and Eberhart, S.A. (1968) Recurrent selection in sorghum. Crop Science 8, 199121. Doney, D.L. (1998) Beta evaluation and sugar beet enhancement from wild sources. In: L. Frese, Panella, L., Srivastava, H.M. and Lange, W. (eds) International Beta Genetic Resources Network. A Report on the 4th International Beta Genetic Resources Workshop and World Beta Network Conference held at the Aegean Agricultural Research Institute, Izmir, Turkey, 28 February3 March 1996. International Crop Network Series. 12. International Plant Genetic Resources Institute, Rome, pp. 7376. Doney, D.L. and Whitney, E.D. (1990) Genetic enhancement in Beta for disease resistance using wild relatives: a strong case for the value of genetic conservation. Economic Botany 44(4), 445451.

308

L. Frese et al. Grnewald, I., Horak, I. and Schlsser, E. (1983) Rizomania. III. Verbreitung im Hessischen Ried und im Raum Worms sowie Beziehungen zum Boden-pH und zur Fruchtfolge. Zuckerind 108(7), 650652. Heijbroek, W. (1977) Partial resistance of sugarbeet to beet cyst eelworm (Heterodera schachtii Schm.). Euphytica 26, 257261. Hellinga, J.J.A. (1943) Verslag over het onderzoek van grondmonsters op bietenaaltjes verricht in samenwerking met de suikerbietenfabriken in 1941 and 1942. Verslagen van het Institute voor Rationele Suikerproduktie 13, 4766. Jassem, B. (1985) Einbeziehung von Wildrben zur Verbesserung der Resistenz und anderer Eigenschaften bei Zuckerrben. Tag.-Ber. Akad. Landwirtsch.-Wissen 229, 213218. Jassem, B. (1992) Species relationship in the genus Beta as revealed by crossing experiments. In: Frese, L. (ed.) International Beta Genetic Resources Network. A Report on the 2nd International Beta Genetic Resources Workshop held at the Institute of Crop Science and Plant Breeding, Braunschweig, Gemany, 2428 June 1991. International Crop Network Series no. 7. International Plant Genetic Resources Institute, Rome, pp. 5561. Jung, C. (1997) Die Klonierung des Hs1pro-1-Gens fr Nematodenresistenz aus Beta procumbens. Vortrge Panzenzchtung 37, 1830. Jung, C., Pillen, K., Fhr, S. and Melchinger, A.E. (1993) Phylogenetic relationships between cultivated and wild species of the genus Beta revealed by DNA ngerprinting. Theoretical and Applied Genetics 96, 449457. Lewellen, R.T. (1992) Use of plant introductions to improve populations and hybrids of sugarbeet. In: Use of Plant Introductions in Cultivar Development. Part 2. CSSA Special Publication No. 20, pp. 117136. Lptien, H. (1984) Breeding nematode resistant beets. II. Investigations into the inheritance of resistance to Heterodera schachtii Schm. in wild species of the section Patellares. Zeitschrift fr Panzenzchtung 93, 237245. Mesbah, M., Scholten, O.E., de Bock, Th.S.M. and Lange, W. (1997) Chromosome localisation of genes for resistance in Beta species of the section Procumbentes. [Abstract]. 60th IIRB Congress, Cambridge, UK, 30 June4 July 1997. Michalik, B., Grezebelus, D. and Baranski, R. (1998) Promotion of the Use of East European Beta vulgaris Germplasm Collection. Report on the IPGRIECP/GR funded project 97/011. Mitteau, M. (1997) Collection Nationale France. Catalogue des ressources gntiques bl tendre et Orge. Bureau des ressources gntiques (BRG), Paris. Munerati, O. (1932) Sull incrocio della barbabietola coltivata con la beta selvaggia della costa adriatica. LIndustria Saccarifera Italiana 25, 303304. Oltmann, W., Burba, M. and Bolz, G. (1984) Die Qualitt der Zuckerrbe. Bedeutung, Beurteilungskriterien und zchterische Manahmen zu ihrer Verbesserung. Paul Parey, Berlin. Owen, F.V. (1942) Some Mendelian characters in Beta vulgaris and linkages observed in the Y.R.B. group. Journal of Agronomic Research 65, 155171. Owen, F.V. (1948) Utilization of male-sterility in breeding superior-yielding sugar beets. Proceedings of the American Society for Sugar Beet Technology 5, 156161. Owen, F.V. (1954) Hybrid sugarbeets made by utilizing both cytoplasmic and Mendelian male sterility. Proceedings of the American Society for Sugar Beet Technology 8, 64. Panella, L. (1998) Screening and utilizing Beta genetic resources with resistance to Rhizoctonia root rot and Cercospora leaf spot in a sugar beet breeding programme. In: Frese, L., Panella, L., Srivastava, H.M. and Lange, W. (eds) International Beta Genetic Resources Network. A Report on the 4th International Beta Genetic Resources Workshop and World Beta Network Conference held at the Aegean Agricultural Research Institute, Izmir, Turkey, 28 February 3 March 1996. International Crop Network Series 12. International Plant Genetic Resource Institute, Rome, pp. 6272. Paul, H., Henken, B., de Bock, Th.S.M. and Lange, W. (1992) Resistance to Polymyxa betae in

Base-broadening in Beta

309

Beta species of the section Procumbentes, in hybrids with B. vulgaris and in monosomic chromosome additions of B. procumbens in B. vulgaris. Plant Breeding 109, 265273. Savitsky, H. (1950) Monogerm sugar beets in the United States. Proceedings of the American Society for Sugar Beet Technology 6, 156159. Savitsky, H. (1960) Meiosis in an F1 hybrid between a Turkish wild beet (Beta vulgaris ssp. maritima) and Beta procumbens. Journal of the American Society for Sugar Beet Technology 11, 4967. Savitsky, V.F. (1952) Methods and results of breeding work with monogerm beets. Proceedings of the American Society for Sugar Beet Technology 7, 344350. Shen, Y., Ford-Lloyd, B.V. and Newbury, H.J. (1997) Genetic relationships within the genus Beta determined using both PCR-based marker and DNA sequencing techniques. Heredity 80, 624632. Stanescu, Z. (1994) Contribution of Romanian scientic research to the activities of the World Beta Network. In: Frese, L. and Doney, D.L. (eds) International Beta Genetic Resources Network. A Report on the 3rd International Beta Genetic Resources Workshop and World Beta Network Conference held at the North Dakota State University, Fargo, USA, 46 August 1994. International Crop Network Series 11. International Plant Genetic Resources Institute, Rome, p. 31. Sun Yi Chu (1994) Beta germplasm collection and its application status in China. In: Frese, L. and Doney, D.L. (eds) International Beta Genetic Resources Network. A Report on the 3rd International Beta Genetic Resources Workshop and World Beta Network Conference held at the North Dakota State University, Fargo, USA, 46 August 1994. International Crop Network Series 11. International Plant Genetic Resources Institute, Rome, pp. 2729. Tjebbes, K. (1933) The wild beets of the North Sea region. Botaniska Notiser 14, 305315. van Geyt, J.P.C., Lange, W., Oleo, M. and de Bock, Th.S.M. (1990) Natural variation within the genus Beta and its possible use for breeding sugar beet. Euphytica 49, 5776. Winner, C. (1981) Zuckerrbenbau. DLG-Verlag, Frankfurt/Main. Yu, M.H. (1997) Sugarbeet root-knot nematode resistance and breeding aspects [Abstract]. 60th IIRB Congress, Cambridge, UK, 30 June4 July 1997.

18

HOPE, a Hierarchical, Openended System for Broadening the Breeding Base of Maize

L.W. KANNENBERG
Plant Agriculture Department, University of Guelph, Guelph, Ontario N1G 2W1, Canada

Rationale and Objectives

The hierarchical, open-ended population enrichment (HOPE) breeding system was designed to broaden signicantly the genetic base presently deployed in commercial maize breeding programmes in Canada and similar short-season maturity areas. This base is extremely narrow and relies heavily on the earlier maturing derivatives of the six inbred pedigrees (Goodman, 1990) that dominate the US hybrid maize industry. Some potential consequences of a constricted breeding base are an eventual reduction in the rate of breeding progress, greater vulnerability to biotic and abiotic stresses, and the lack of alternative or additional gene complexes which may exist in untapped germplasm. As Frankel (1977) points out, the use of such germplasm may result in breeding material with new developmental pathways or ecological adaptations. The latter may, for example, be important as agricultural practices change to reduced or no tillage, lower inputs, increased use of marginal lands, etc. The objective of the HOPE breeding system is to develop a breeding source for inbred lines that have markedly different genetic backgrounds from lines presently used in commercial breeding and that have the agronomic qualities and combining ability to serve as a parent or parents of commercially successful hybrids, thereby enhancing the genetic diversity available to the maize grower.

Methods and Strategy

The HOPE breeding system The HOPE breeding system is presently undergoing its second revision. The original design, described in Cramer and Kannenberg (1992), was employed from 1977 to 1985. From 1986, the system was modied to the one described in detail by Kannenberg and Falk (1995). Following a recent evaluation (Popi, 1997), the
IPGRI/FAO 2001. Broadening the Genetic Base of Crop Production (eds H.D. Cooper, C. Spillane and T. Hodgkin)

311

312

L.W. Kannenberg

HOPE system is now undergoing a radical alteration, resulting in considerable simplication. The version of the HOPE system described by Kannenberg and Falk (1995) involves A and B heterotic sets, each consisting of four genepools, arranged in a hierarchy of performance levels: low (L), intermediate (I), high (H) and elite (E) (Fig. 18.1). The performance hierarchy is to separate materials with poorer agronomic characteristics from immediately more useful germplasm. The A and B heterotic sets are meant to provide sources at the E level for, respectively, A and B inbred lines that have good combining ability, i.e. the hybrid formed by crossing the two inbreds will exhibit good heterosis (hybrid vigour). The genepools are open-ended: thus, new germplasm (introductions) can be continually introduced into the system based upon level of performance and heterotic pattern, and improved germplasm can be moved stepwise up the hierarchy of performance levels. Increasingly stringent selection procedures are employed at each successively higher hierarchical level: this allows shaping of the extensive variability at the lower levels into diverse, but potentially more useful, germplasm at the E level, the level from which inbred lines are developed. The newest version of HOPE, as proposed by Popi (1997), maintains the same overall concept but in a considerably simplied form. Genetic analysis of the system has shown that only three hierarchical levels are necessary and that the A and B sets are useful only at the E level (Fig. 18.2). Essentially, Popi found that the various recurrent selection procedures employed in the HOPE breeding system increased agronomic performance at all levels of the hierarchy. In the most recent cycles of selection, the L genepools had similar grain yields to the I and H level genepools for the respective A and B sets; however, grain moisture at harvest (an indicator of maturity) and stalk breakage were higher for the L level genepools, thus distinguishing performance of the L level germplasm from the I and H level genepools for these important traits in short-

EA RRS S2s of best A Introd.

Elite

EB RRS S2s of best B Introd.

HA S2-progeny selection

High

HB S2-progeny selection IB

Half-sibs of better A Introd.

IA Half-sib selection Intermediate

Half-sibs of better B Introd.

Half-sib selection LB All B Introductions

All A Introductions

LA Stratified mass selection Low

Stratified mass selection

Fig. 18.1. The version of the HOPE breeding system that was used from 1986 to 1997. The A and B sets represent complementary heterotic groups.

Base-broadening in Maize

313

season corn. Agronomic performance of the I- and H-level genepools in the most recent cycles was similar within each set. Further, the degree of heterosis of the cross between genepools at the L, I and H levels, respectively, remained low (less than 10%) for grain yield even after several cycles of selection. Consequently, it was decided to combine LA and LB into a single L genepool and, as well, the four genepools (IA, IB, HA and HB) into a single H-level genepool. At the E level, heterosis of the EA EB cross increased from 9% in the original cycle to about 21% in the current cycle, indicating that the reciprocal selection practised at the E level has been effective; consequently, both the EA and EB genepools are being kept in the newest version of HOPE. Relative to the breeding procedures, stratied mass selection will continue to be employed at the L level, and selfed progeny recurrent selection (Hallauer and Miranda, 1988) at the H level, but the number of families tested at the H level will be approximately doubled to 196. The number of families selected per cycle will also be increased to 3040 to reduce random drift effects. Selfed progeny selection has been chosen because selng unmasks deleterious recessive genes, which is an especially important consideration when working with highly variable germplasm. However, at the E level, instead of reciprocal recurrent selection (RRS), reciprocal full-sib selection (RFS), as described in Hallauer and Miranda (1988), will be employed. Compared with RRS, RFS doubles the number of families tested without changing the number of test plots. This is because each RFS plot is a cross of an A family with a B family, whereas with RRS each A family and each B family is crossed to a different tester. In the newest revision of HOPE, the number of families tested at the E level will also be approximately doubled to 196 and the number of families selected increased to 2025 in order to reduce random drift effects. The system will remain open-ended with new introductions included into the L level and, when performance is sufcient, into the H level as well. However, no testing of heterotic pattern will be required. Germplasm will also be moved upward in the hierarchy as before, except that in each cycle at least one entry will be advanced to each genepool of the E level: this will ensure some geneow from the H to the E level in each cycle. The procedures to initiate this revision of the HOPE system were begun in 1997.

EA S2s of best Introd.

Full-sib RRS

EB

Elite

H S2-progeny selection

High

All introductions L Stratified mass selection Low

Fig. 18.2. The newest version of the HOPE breeding system.

314

L.W. Kannenberg

This newest version of HOPE will substantially reduce the complexity, human resources, eld requirements and costs relative to the previous HOPE systems.

The synthesis of HOPE and subsequent germplasm introductions The initial synthesis of each of the H, I and L genepools involved a total of 43 components, consisting of adapted open-pollinated varieties, synthetics, composites and four double-cross hybrids of foreign origin. Six to eight of the 43 components were assigned to each of the six genepools based on agronomic performance and heterotic pattern as deduced from historical information from the Guelph breeding programme. These six to eight components were then extensively intercrossed to form the original cycle of that genepool. Subsequently, the E-level genepools were developed by intercrossing within each set S2 lines of the best entries from the rst cycle of selection in the H-level genepools. Introductions have continually been added to the HOPE system. During the rst 5 years of HOPE, 408 introductions were made. In the following 11 years (to 1992), an additional 448 introductions were added. To date (2000), approximately 1000 introductions have been included in the HOPE system. All introductions are incorporated into the L-level genepools, but some (less than 3%) have also been introgressed into the I or H levels, depending on their relative performance. No introductions are put directly into the E-level genepools. Most introductions have been of short-season maturity suitable for Guelph and are worldwide in origin. The majority of introductions during the rst years of HOPE were accessions from the North Central Regional Plant Introduction Station in Ames, Iowa. Subsequently, germplasm has been received from short-season breeding programmes in many parts of the world. The only criterion for incorporating such germplasm into the HOPE system is that it can produce mature grain in the Guelph environment. Even germplasm with poor agronomic performance is included in HOPE on the assumption that desirable attributes may be hidden by obvious faults. About 5% of introductions have been so-called exotics, i.e. germplasm requiring a longer growing season than available at Guelph. In the case of exotic materials, only elite improved sources, such as improved populations from the International Maize and Wheat Improvement Centre (CIMMYT), are chosen for inclusion in the HOPE system because of the time and cost of adapting such germplasm to the necessary maturity. For these reasons, no efforts have been made to introgress germplasm of wild progenitors of maize, such as teosinte or Tripsacum ssp., into HOPE.

Results
Utility of HOPE germplasm Popi (1997) used random amplied polymorphic DNA (RAPD) analyses to determine the genetic distances among the HOPE genepools and also relative to 11 short-season commercial hybrids. Principal component analysis indicated that the HOPE populations vary considerably among themselves and represent markedly different germplasm

Base-broadening in Maize

315

10 5 0 PC2 5 10 15 20 20 15 HB EB 10 5 0 PC1 5 10 15 20 LA HA IA L IB
B

Most advanced cycles of HOPE breeding system gene pools EA

Eleven commercial hybrids

Fig. 18.3. Principal component analysis* of HOPE populations compared with 11 commercial hybrids.
EA, HA, IA, LA are, respectively, elite, high, intermediate and low levels of the A set of HOPE populations. EB, HB, IB, LB are, respectively, elite, high, intermediate and low levels of the B set of HOPE populations. The number of selection cycles represented are four at the E level, 11 at the H level, and 15 at the I and L levels. *Note that the vertical axis of the graph has been reduced relative to the horizontal axis.

from that deployed commercially (Fig. 18.3). Thus, HOPE germplasm has the potential to contribute to the genetic diversity of commercial germplasm. We have some evidence that this potential may be realized: two University of Guelph inbreds (CG91 and CG96, released in 1997), contain HOPE germplasm. They were derived from a cross of a selected HOPE family to a commercial hybrid. We have found that HOPE germplasm, as with many improved populations, although competitive in yield performance, is typically lacking in such agronomic traits as stalk quality, resulting in more stalk breakage than acceptable. Crossing population germplasm to commercial hybrids with their long history of intensive selection for agronomic qualities provides a breeding source that combines agronomic eliteness with the unique and selected germplasm from the populations. Our testcross data indicate that CG91 and CG96 can produce hybrids competitive in performance with current commercial materials. Whether CG91, CG96 and subsequent HOPE inbred releases will be used commercially will be dependent on their capacity in hybrid combinations to perform under the intensive and extensive testing to which commercial materials are subjected. Stalk quality is being intensively selected for in the HOPE genepools and, if experience with other Guelph populations is any indication, future inbred releases from the HOPE programme may be derived solely from a HOPE background rather than the cross of HOPE material to a commercial hybrid. In any event, we are optimistic that HOPE-based inbreds will ultimately be used in commercial breeding programmes, directly as a parent in a hybrid and/or indirectly as a component in a company breeding programme. In either case, the HOPE breeding system will meet its goal of diversifying the commercially deployed genetic base.

Lessons from the HOPE experience Popis (1997) evaluation of the HOPE breeding system showed little heterosis in A B crosses at the L, I and H levels, respectively, while at the E level, which was subjected to

316

L.W. Kannenberg

reciprocal recurrent selection, heterosis was satisfactory (over 20% for grain yield). This indicated that the A and B heterotic grouping was desirable only at the E level. Further, little difference in performance was observed among the L, I and H levels of each set. Therefore, these levels could be combined. However, we believe it is still important to separate the L-level materials from the E germplasm; in particular, selng at the H level helps to decrease inbreeding depression because of unmasking of deleterious recessives before these materials are elevated to the E level. Thus, the H level functions as a buffer between the L- and E-level genepools. These changes (depicted in Fig. 18.2) have greatly simplied the logistics of the previous HOPE scheme. The relatively good performance of the L-level germplasm has been surprising, especially considering that about 1000 accessions have been introduced into the L level over the past 20 years. The fact that introductions were crossed to the E level before inclusion probably helped, but stratied mass selection would also appear to be an effective selection tool in broad-based germplasm.

Requirements for HOPE In general, breeding systems such as HOPE require personnel trained in plant breeding, especially population improvement methods. One exception is the stratied mass selection procedure used at the L level: farmers, village teachers or other interested peoples could easily be taught this procedure as a means for systematic improvement of local varieties. The technologies and costs of the HOPE breeding system are equivalent to those of other maize breeding programmes. The L-level requirements are for an isolation area of about 1 ha and one cycle of stratied mass selection can be completed each season. The selfed progeny selection at the H level and reciprocal full-sib selection at the E level both require 3 years per cycle with a winter nursery. Of the 3 years, one is for replicated eld trials at two or more locations and the other 2 years include summer and winter breeding nursery activities. Details of the above procedures can be found in Hallauer and Miranda (1988) and, specic to HOPE (except for RFS selection), in Kannenberg and Falk (1995).

Conclusion and recommendations The HOPE breeding system has two goals. The rst is to provide a source of inbred lines that are genetically very different from those currently used in commercial breeding programmes. The second goal is that HOPE inbreds must be at least comparable in performance with current commercial inbreds both in per se performance and in hybrid combinations. Popis (1997) molecular analyses showed that the rst goal has been achieved. Whether or not HOPE germplasm will be used commercially depends on the success of meeting the second goal. The hybrid maize industry is highly competitive and our experience has been that commercial breeders will not expend time and resources working unproven breeding germplasm. Commercial breeders, however, will evaluate inbred lines from other than their own programme if these inbreds are considered to have potential. In the approximately 20-year history of the HOPE programme, only four inbreds have been released to the industry. The rst two HOPE inbreds did

Base-broadening in Maize

317

not have adequate performance to serve directly as parent lines in a hybrid programme, but were released because we thought they might have potential as genetically different germplasm in a breeding programme. Although several commercial companies obtained this germplasm, to our knowledge, no commercial inbreds have been developed from it. The two 1997 inbred releases, which are a combination of HOPE and commercial germplasm, have, according to our data (which are not extensive because of our limited resources), the potential for direct commercial exploitation as a parent line of a hybrid and/or a source of new and different germplasm in a breeding programme. Future HOPE inbred releases should be more frequent, based more or less on the 3 years per cycle of selection required at the E level, and should be incrementally improved, perhaps to the point where the HOPE selections can be inbred without rst crossing to commercial germplasm. Regardless, the ultimate measure of the success or failure of the HOPE breeding system will be whether inbred lines with HOPE germplasm are used in commercial hybrids. Only then will the HOPE breeding system have contributed to diversication of the commercially deployed genetic base of maize. The HOPE breeding system is in effect a continuing experiment. The two evaluations (Cramer, 1986; Popi, 1997) to date have resulted in substantial modications to the HOPE system, especially from Popis study. A number of questions remain. For example, although we are inclined to favour reciprocal full-sib selection at the E level, primarily because of the increased testing efciency (2:1) relative to other procedures, a half-sib programme using two commercially recognized inbred testers of opposite heterotic pattern might be a desirable alternative. Also, the kind and extent of genetic diversity within the system need to be characterized, especially at the E level and especially relative to the genetic diversity available from other more traditional and simpler population breeding projects, such as the eight other broad-based populations undergoing recurrent selection in the Guelph breeding programme.1 Essentially, we need to know how unique and how useful HOPE germplasm is. Another question is whether we should continue to add new introductions regardless of their agronomic quality or should only accessions with superior performance to existing HOPE germplasm be used? For that matter, are introductions adding to desirable genetic diversity, or should the HOPE breeding system be closed to introductions so that selection within the existing HOPE germplasm might be more effective in developing desirable gene combinations? Modern molecular technologies may be helpful in the analyses required to resolve these and other questions relevant to the HOPE breeding system. Although the HOPE breeding system has been used to date only for developing breeding germplasm suitable for the short-season maturity of Guelph, there is of course no reason why HOPE or breeding systems of similar intent cannot be developed on a wider scale. Indeed, we have devised a scheme in which interfacing HOPE-like systems were developed for the northern, central and southern maize-growing areas of Ontario. Commercial companies agreed to contribute evaluation trials and breeding nursery space. However, external funding was necessary to support the programme and we were not successful in obtaining this. Nevertheless, breeders recognize the critical need to diversify the breeding base of maize and they are prepared to apportion a share of their physical resources in collaborative ventures to accomplish this endeavour.
1A

description of these and other Guelph population can be found at: http://131.104.232.5/research/corn_breeding/populations2.htm

318

L.W. Kannenberg

References
Cramer, M.M. (1986) An evaluation of HOPE, the hierarchical, open-ended maize (Zea mays L.) breeding system. PhD Dissertation, University of Guelph, Guelph, Ontario. Cramer, M.M. and Kannenberg, L.W. (1992) Five years of HOPE: the hierarchical open-ended maize breeding system. Crop Science 32, 1131171. Frankel, O.H. (1977) Genetic resources. Annals of the New York Academy of Science 287, 332344. Goodman, M.M. (1990) Genetic and germplasm stocks worth conserving. Journal of Heredity 81, 1116. Hallauer, A.R. and Miranda, J.B. (1988) Quantitative Genetics in Maize Breeding, 2nd edn. Iowa State University Press, Ames, Iowa. Kannenberg, L.W. and Falk, D.E. (1995) Models for activation of plant genetic resources for crop breeding programs. Canadian Journal of Plant Sciences 75, 4553. Popi, J. (1997) A critical evaluation of the HOPE breeding system as a means for broadening the deployed germplasm base in maize. PhD Dissertation, University of Guelph, Ontario.

19

The Germplasm Enhancement of Maize (GEM) Project: Private and Public Sector Collaboration

L.M. POLLAK1 AND W. SALHUANA2

1USDA-ARS

Corn Insects and Crop Genetics Research Unit, Department of Agronomy, Iowa State University, Ames, IA 50011, USA; 26204 SW 146th Ct., Miami, FL 33183, USA

Rationale and Objectives of LAMP and GEM

GEM could not exist if the Latin American Maize Project (LAMP) (Salhuana et al., 1991; Pollak, 1993; Sevilla and Salhuana, 1997) had not come rst. LAMP provided the information necessary to efciently select germplasm bank accessions for enhancement. In this regard, LAMP served as the rst step to sharing promising maize materials from the germplasm banks with breeders. GEM completes the process by returning to the germplasm bank enhanced materials developed from the accessions, which can be directly used in breeding programmes. The primary objective of LAMP was to evaluate accessions found in Latin American and US germplasm banks so they could be used in breeding programmes. This project was the rst coordinated international effort to deal with the evaluation of the genetic resources of a major world crop. LAMP was based on the cooperative effort of 12 countries: Argentina, Bolivia, Brazil, Colombia, Chile, Guatemala, Mexico, Paraguay, Peru, United States, Uruguay and Venezuela. The results of LAMP indicate that there are accessions (native and foreign) that show good yield potential on which enhancement can be initiated in order to improve yield, agronomic characteristics and adaptability before being incorporated into breeding programmes. In 1991, a catalogue and CD-ROM of data of 12,113 accessions evaluated in LAMPs rst stage, and 2794 selected (primarily on yield) accessions evaluated in the second stage, in 59 different locations of 32 regions of the 12 countries, was published (LAMP, 1991). Based on those data, the principal investigators in each country selected a total of 268 elite accessions that were crossed with the best testers of each region. Thirty-one testers were used for crossing with the elite accessions. Within similar homologous areas, principal investigators exchanged testcrosses among other principal investigators in the same homologous area, so that testcross evaluations were done in more than one country. Data from the testcross evaluations were published in a catalogue and updated CD-ROM in 1995 (Salhuana and Sevilla, 1995), and a nal report published in 1997 (Salhuana et al., 1997). In the fth stage of LAMP, each principal
IPGRI/FAO 2001. Broadening the Genetic Base of Crop Production (eds H.D. Cooper, C. Spillane and T. Hodgkin)

319

320

L.M. Pollak and W. Salhuana

investigator was to enhance selected germplasm to meet his or her countrys breeding objectives, yet funding for only a year of small-scale enhancement was available. In the USA, the seed source for virtually all maize grown commercially is the seed industry. For any of the LAMP accessions to be useful in maize improvement, some mechanism had to be established to enhance them so they could enter commercial maize breeding channels. The competitive nature of the seed industry made it unlikely that any one company would support an enhancement effort. Public breeders are poorly funded, and there are few if any grant sources for germplasm enhancement, so it was unlikely that public breeders could nd the nancial resources to support an enhancement effort. It was clear that a coordinated and cooperative effort among public and private sectors was needed before the LAMP materials would be used in US breeding programmes.

US Maize Diversity Prior to GEM

Maize is the USAs major crop, where over 30 million ha are planted each year, and is extremely important to the US economy due to the amount produced, its value to industry and its export value. Through feeding livestock that is processed into meat and dairy products, maize affects nearly everyone in American society. The USA is also the world leader in maize production. Genetic variability is essential in plant breeding programmes (Michelini and Hallauer, 1993), but improvements in crops by plant breeding are usually followed by decreased genetic diversity, particularly in materials that reach commercial production. Thus farmers hybrids face increased genetic vulnerability while farmers face increased economic risk. Since the early 1960s, there have been frequent and urgent warnings about maizes genetic vulnerability and the potential of exotic germplasm to decrease this vulnerability (Anon., 1972; Brown, 1975; Walsh, 1981; Wilkes, 1989; Goodman, 1990). Concerns increased after a southern maize leaf blight (incited by Bipolaris maydis (Nisikado) Shoemaker, race T) epidemic in 1970, which was due to the widespread genetic uniformity of the crop speeding development of the disease to epidemic proportions. The concerns regarding maizes genetic vulnerability have been reinforced by studies documenting a reduction in genetic variability among domestic lines and hybrids. The most recent survey of germplasm sources for the maize crop (Darrah and Zuber, 1985) found that 88% of 1984 US maize seed produced for 1985 planting included germplasm derived from one variety of the Corn Belt Dent race: Reid Yellow Dent. From a biochemical data perspective, US maize cultivation and breeding remain dependent upon the inbred lines B73, A632, Oh43 and Mo17, or closely related derivatives (Smith, 1988). In the US Corn Belt, exotic germplasm includes unimproved domestic as well as foreign populations (Stuber, 1986). A comprehensive survey conducted in 1983 on the use of exotic germplasm in commercial maize revealed that less than 1% of the US germplasm base consisted of exotic germplasm (Goodman, 1985). In spite of the above concerns, maize breeders have continued to focus on shortterm breeding goals largely because of the predominance of the private sector in maize breeding and its need for short-term results. This results in a very narrow genetic base of maize produced on the farm, with many companies selling closely related hybrids (Smith, 1988). This may lead to a yield plateau, greatly increases vulnerability to pests,

Germplasm Enhancement of Maize

321

and makes it difcult to meet new market demands. Jenkins (1978) found the genetic base of maize breeding programmes reduced because greater emphasis was given to developing recycled lines instead of lines developed from improved populations or synthetics. Private companies are, however, growing increasingly concerned about their narrow germplasm pools, although tough competition in the industry results in the tendency to increase diversity by focusing on elite proprietary exotics from branch stations within a company. Although breeders are still making 1.52.0% per year of genetic gain for yield in adapted materials, it is not known how long these gains will continue. Geadelmann (1984) suggested that incorporation of exotic strains into adapted germplasm would increase the available genetic variability and give rise to additional heterotic vigour, lessening chances for a yield plateau.

Objectives and Goals of GEM

GEMs objective is to provide to the maize industry materials developed using germplasm enhancement of useful exotic germplasm, with the ultimate aim of improving and broadening the germplasm base of maize hybrids grown by American farmers. Because proprietary germplasm is used to make breeding crosses, access to breeding materials is limited to GEM cooperators, but the opportunity to become a cooperator is available to all. Data collected on GEM materials are freely available, and GEMenhanced lines and synthetics will be freely available through the US North Central Regional Plant Introduction Station (NCRPIS) after their public release. Traits targeted for improvement are agronomic productivity, disease and insect resistance, and valueadded characteristics.

Methods and Strategy Used in GEM

Development and organization of the project GEM began in 1993 when several public and private maize breeders, with support from the American Seed Trade Association (ASTA), developed a proposal for a public/private collaborative effort to enhance the accessions identied as useful by LAMP. In addition to enhancement, the proposal also addressed other concerns of ASTA member companies including the diminishing role of the public sector in maize breeding, federal and state budgetary constraints causing decreased nancial support of public programmes, and relative lack of public support for maize research compared with other crops despite the greater economic importance of maize to the nation. In 1994 the proposal and a breeding protocol (modied pedigree breeding procedure) was sent to all ASTA membership with maize breeding programmes asking for their willingness to become a member of GEM. Participation as a private-sector collaborator in GEM involved signing an agreement and following the protocol. The protocol required them to:

Cross exotic accessions assigned by the coordinator to their proprietary inbred lines and return the crosses to the coordinator.

322

L.M. Pollak and W. Salhuana

Contribute in-kind support for the breeding effort (winter and summer nursery rows, yield trial plots and disease observation rows). Not distribute proprietary exotic germplasm outside the company.

In-kind support by industry was considered important for lobbying the US Congress, for providing the necessary number of testing environments, for ensuring that the enhanced materials have commercial relevance, and for providing public programmes with routine cooperation and guidance. In return, a participating company received early access to GEM materials. The following points addressed proprietary concerns.

It is extremely difcult to extract an inbred line once it has been hybridized, especially with an accession. Pedigrees are coded so that only the coordinator knows which company made a proprietary cross. A companys S2 lines are considered proprietary. Publicly released material from company breeding efforts are S2 synthetics, as opposed to S3 lines from public breeders.

The proposal also was sent to public-sector maize breeders giving them the opportunity to become collaborators, although they were not required to make breeding crosses. Signing an agreement allowed them access to GEM materials, potential nancial or inkind support to participate, and signalled to the US Congress and the US Department of Agriculture (USDA) their belief in the value of the project. A sub-committee from ASTA lobbied key legislators of the US Congress for permanent base funding to the Agricultural Research Service (ARS) to support the public effort at ARS and university locations. Besides recognizing the value of the work, Congress was impressed by the $1.5 million contribution Pioneer Hi-Bred International made to LAMP, and the in-kind support from companies for GEM that was estimated to be worth approximately $450,000 per year. This amount underestimated the actual value of the contributions because it ignored the value of the proprietary germplasm, overhead costs, and advice and service of the private breeders. In 1995, $500,000 of permanent yearly funding was appropriated by Congress to fund GEM. In 2000, there were 38 companies, 42 public scientists and one international agency (Empresa Brasiliera de Pesquesa Agropecuaria EMBRAPA) serving as GEM cooperators. A cooperators meeting is held once a year at the Corn & Sorghum ASTA meetings to discuss progress. A Technical Steering Group (TSG) meets three times a year to discuss policies, protocol and results. The TSG is presently composed of eight members from industry, two from USDA-ARS (members are limited to the GEM coordinator and North Carolina project leader and are ex ofcio due to conict of interest) and one from a university. Members from companies and universities serve a staggered 3-year term. The GEM coordinator in Ames manages seed curation, line evaluations and release, cooperative nurseries and yield tests using in-kind support, data analysis and management, public cooperator research plans and nances, and public relations. In addition, the Ames location conducts value-added trait research and develops enhanced lines for improved yield and value-added traits. The Raleigh project leader manages enhancement of the 50% tropical breeding crosses for the southern USA, enhancement

Germplasm Enhancement of Maize

323

of the 50% tropical breeding crosses for moving to the Corn Belt, and public cooperator projects in the southern USA. From the US Congresss appropriated funds, some public cooperators receive partial nancial support for conducting GEM-related research. In-kind support from private cooperators each year provides over 6500 summer and 2000 winter nursery rows. Over 7000 rows of yield test plots throughout the eastern, southern and Corn Belt sections of the USA provide a wide diversity of environments. Additional GEM-related nursery rows and yield test plots are grown in the Ames and Raleigh locations.

Starting materials used for breeding GEM is an ongoing project, but to initiate enhancement 51 elite tropical and temperate LAMP accessions were chosen on the basis of yield performance in LAMP, and seven commercial tropical hybrids were provided by DeKalb Genetics. The enhancement protocol is for one of the private cooperating companies to cross an exotic material by a proprietary inbred line to make a 50% exotic breeding cross, then for another private cooperator to cross the 50% cross with their proprietary line of the same heterotic pattern to make a 25% exotic breeding cross. All 50% and 25% breeding crosses are evaluated for yield as testcrosses, and the best used to develop breeding lines by cooperators.

Breeding activities and results to date In 1994 the coordinator assigned four accessions or tropical hybrids (from the 51 accessions/seven tropical hybrids initial starting materials) to each private cooperator to cross to a proprietary inbred line in the 1994/95 winter nursery. The coordinator specied the heterotic pattern Stiff Stalk (SS) or non-Stiff Stalk (nSS) to use, based on information from LAMP testcrosses. If the heterotic pattern of the exotic material was unknown, both crosses were made. Private cooperators returned exotic private inbred breeding crosses to the coordinator, then balanced samples of the two-way breeding crosses were sent to two different companies for making three-way SS or nSS breeding crosses in 1995 summer nurseries or day-neutral locations. Each private cooperator was assigned eight crosses. In the 1995/96 winter nurseries, two companies (Golden Harvest and Pioneer) made nSS testcrosses with all available SS breeding crosses, and two companies (Cargill and Holdens) made SS testcrosses with nSS breeding crosses. Cooperative yield testing during 1996 involved 16 experiments grown in more than six locations, evaluating 564 testcrossed 50% and 25% exotic breeding crosses. From these results, breeding crosses were selected for advancement to line development. Results for the best topcrosses of Stiff Stalk breeding crosses and non-Stiff Stalk breeding crosses are shown in Table 19.1, expressed as a percentage over the ve check hybrids (Holdens Foundation Seeds Single Cross LH195/LH212, Holdens Foundation Seeds Single Cross LH195/LH59, Pioneer Brand Hybrid 3489, Pioneer Brand Hybrid 3525 and Pioneer Brand Hybrid 3163). Some results are similar to the mean of the checks or better; thus we expect the testcrosses of the best lines developed from these breeding crosses to be superior to the

324

L.M. Pollak and W. Salhuana Table 19.1. Relative yields and moisture values of some testcrossed GEM breeding crosses evaluated in 1996. Percentage of ve check hybrids Experiment Pioneer nSS tester 50% tropical breeding crosses 25% tropical breeding crosses 50% temperate breeding crosses Locations 5 4 5 GEM breeding cross DKB844:S16 CUBA117:S15 CUBA164:S52008a FS8A(S):S09 UR10001:S18 Yield (%) 98 98 103 93 92 Grain moisture (%) 116 108 93 103 105

Golden Harvest nSS tester 50% tropical 9 breeding crosses 25% tropical breeding crosses 50% temperate breeding crosses 25% temperate breeding crosses Cargill SS tester 50% tropical breeding crosses 25% tropical breeding crosses Holdens SS tester 50% tropical breeding crosses 25% tropical breeding crosses 25% temperate breeding crosses 8 8 7

CUBA164:S15 CHIS775:S19 CHIS740:S1411a CHIS740:S1415 AR16035:S19 FS8B(S):S03 AR16021:S0908b FS8A(S):S0915

97 96 101 96 100 96 106 99

100 102 91 98 104 105 92 91

6 8

ANTIG03:N12 CHIS775:N19 DKXL370:N11a20 CHIS775:N1920

96 92 101 101

110 105 109 105

6 9 5

DKB844:N11b DREP150:N20 BR51501:N11a08d BR51501:N11a12 FS8B(T):N11a08a UR13085:N0215

96 94 101 99 103 102

107 117 103 108 98 105

check hybrids. Because any line selected through this procedure will probably need agronomic improvement, they are best used as breeding lines. Line development was started in a few breeding crosses based on LAMP and grown as S2 testcrosses in yield tests in 1997. Results for testcrosses are shown in Table 19.2. After the initial seasons of developing and evaluating breeding crosses, GEMs seasonal breeding programme has developed into a balance of the above elements

Germplasm Enhancement of Maize

325

Table 19.2. Results from experiments evaluating topcrosses (Holdens SS tester) of lines developed from two breeding crosses using the Chilean temperate accession CH05015, grown in 1997 in Corn Belt locations. Moisture Yield (kg ha 1) 9,041 8,493 Lodged (%) (%) 20.7 18.5 Stalk 3.7 5.4 Root 4.7 4.7

Pedigree CH05015:N15-98-1 CH05015:N15-76-1 Check hybrids LH195/LH212 LH195/LH59 Pioneer 3163 Pioneer 3489 Pioneer 3525 Checks mean LSD0.05 CH05015:N12-183 CH05015:N12-186 Check hybrids LH195/LH212 LH195/LH59 Pioneer 3163 Pioneer 3489 Pioneer 3525 Checks mean LSD0.05

10,262 8,644 10,175 8,981 9,210 9,454 20.3 10,670 10,638

20.0 18.7 22.4 17.8 16.4 19.0

4.3 2.7 9.6 5.5 14.2 7.2

0.9 2.5 6.4 0.5 0.0 2.1

20.4 19.8

4.8 6.8

0.0 0.0

10,518 8,607 10,774 10,309 10,684 10,178 32

19.3 18.5 20.4 17.6 16.4 18.5

6.5 6.2 17.9 15.2 8.6 10.9

0.0 0.0 0.0 0.0 1.2 0.2

(developing and evaluating new breeding crosses) plus ongoing line development at various stages. To date, we have developed over 350 breeding crosses, and instituted pedigree breeding for line development in over 46 breeding crosses. Over 3200 topcrossed lines were in Corn Belt and southern US yield tests in 1998. Results indicate that these lines have true yield potential. In experiments grown by Pioneer HiBred International, Inc. to test S2 lines from CUBA164 breeding crosses topcrossed with a Pioneer non-Stiff Stalk tester, a CUBA164 topcross was the second-highestyielding entry behind Pioneer 3525, in six replications. Seven other lines yielded more than Pioneer 3163, ranking ve to eleven after Pioneer checks 3489 and 34G81. In another experiment with seven replications, eight CUBA164 topcrosses yielded more than Pioneer 3163, ranking two to nine after Pioneer 33A14. In an experiment with three replications, seven CUBA164 topcrosses yielded more than the rst check, LH195/LH212, two with yields over 200 bushels per acre. Twenty-ve topcrosses yielded more than the rst Pioneer check, 34E79. Another experiment of four replications grown by DeKalb Genetics evaluated lines from a 50% exotic breeding cross with an accession from Argentina, AR01150, test-

326

L.M. Pollak and W. Salhuana

crossed with a DeKalb Stiff Stalk line. Six AR0150 topcrosses were the highest-yielding entries, beating the highest-yielding check, Pioneer 3163. Four more AR01150 topcrosses beat the next highest-yielding check, DK621. Based on previous experience with LAMP germplasm, GEM cooperators are condent of nding many economically important traits in this germplasm, which will lead to inbred lines improved for a wide variety of characteristics as well as productivity. For example, 691 Peruvian maize accessions were evaluated for resistance to European maize borer leaf feeding (Abel et al., 1995). Eleven resistant varieties were identied, with all 11 commonly grown in coastal valleys of Perus northern coast. Further analysis indicated that the resistance factor was unrelated to that found in Corn Belt-adapted maize, DIMBOA concentration. Lines developed from breeding crosses of the resistant accessions with Corn Belt lines also show resistance. None of the 11 accessions were included in the top 5% of LAMP accessions, representing the highest-yielding accessions, indicating the importance of screening a wide variety of germplasm for nonyield-related traits.

Social and Economic Impact of LAMP and GEM

Overall utility to breeders This is the rst coordinated public/private effort that addresses the need to reduce genetic vulnerability by diversifying the germplasm base and providing enhanced, improved germplasm to help private industry meet changing markets in a major crop. In spite of industry consolidation, the number of private cooperators has increased since the projects inception, cooperating companies have increased their donation of in-kind support, and end-users are starting to participate. There is increasing interest in participating by foreign companies. As the unique germplasm is publicized, new public cooperators come from a diversity of backgrounds, such as animal nutrition and food science. The impact on US farmers and agriculture in general will be enormous as the germplasm diversies and extends the use of the maize crop.

Opportunities and constraints demonstrating the importance of genetic resources The importance of genetic resources will be easier to demonstrate when economic payoffs ensue from safeguarding those resources, which are the product of several thousands of years of evolution and human experimentation. This is very difcult to demonstrate since precise information about the use of genetic resources does not exist. However, the impact of the utilization of new germplasm can be examined through the examples of LAMP and GEM. The success of the LAMP project was founded upon the cooperative efforts of the principal investigators, who believed that it was necessary to collaborate with strategies focused on success without being daunted by the limitations that would inevitably occur. The results of the LAMP project (Salhuana and Sevilla, 1995) show that for each country there exists germplasm from other countries that performs better than native

Germplasm Enhancement of Maize

327

germplasm. It is imperative to exchange germplasm among countries without restrictions since experiences from several countries have shown that the best germplasm is often derived from foreign accessions. The utilization of selected germplasm for the tropics, temperate and highland areas will bring to other regions the increase in yield and other agronomic characteristics for which the germplasm was selected in the native environment, resulting in a higher quality grain supply for human and animal consumption. Another cooperative project is GEM, in which private industry is participating by contributing the proprietary inbred lines in crosses with the best LAMP accessions as well as with in-kind support in the form of nursery rows, yield test plots and isolation rows. Alliances among countries and collaboration of private industry will help ensure the success of present and future projects in genetic resources, especially in this period of scarce nancial resources, because genetic resource activities must compete for funds from other investments that can yield more immediate, but fewer long-term benets. Collaboration helps additional important traits to be identied in exotic germplasm since some cooperators have the facilities and expertise that others do not have. The results obtained in GEM shows that the contribution of exotic germplasm for disease and insect resistance and value-added traits accompanied with yield will have a large impact on maize productivity.

Opportunities and constraints use of biotechnology New capabilities to introduce exotic genes are becoming increasingly available from biotechnology. Transformation technologies make it easier to transfer genes into maize from other genera than it is to transfer useful genetics from related species through crossing and repeated backcrossing. Conventional procedures impose 37 years of additional time and effort to segregate and recombine genes and break up linkage blocks that would otherwise render the genotype poorly adapted to agriculture. It is critical to realize that the most important traits in agriculture are controlled by groups of genes, which by themselves have a small effect. For the foreseeable future, the most useful genetic diversity will be found in the crop species itself. Advances in biotechnology can help to utilize more effectively a wider variety of diversity. Biotechnology can contribute in at least two broad areas in plant breeding. Genetic markers can identify important genes or chromosome regions, and genetic transformation can move potentially useful exotic genes into inbred lines. Unfortunately, considering the large numbers of accessions in germplasm banks, limited resources do not permit using biotechnology methods to establish variability and genetic distances among exotic and elite that will permit utilization of exotic germplasm in the most efcient way. The results from tomato and rice indicate that exotic germplasm does contain useful genes that can signicantly enhance agricultural production. The preliminary results of GEM also demonstrate the same results for yield, disease and insect resistance, and value added-traits. After identifying the regions of DNA of the trait that we would like to transfer from the exotic material it will possible to transfer this to the elite material by using transformation or by a backcross technique, identifying and selecting the offspring that contain the favourable genes from both parents.

328

L.M. Pollak and W. Salhuana

Challenges and Future Tasks

The immediate challenges must therefore be: 1. To conserve genetic diversity securely. 2. To characterize genetic diversity. 3. To evaluate genetic diversity more completely in pre-breeding and enhancement programmes. 4. To provide resources to accomplish these tasks. 5. To develop and apply new biotechnologies to improve the effectiveness of germplasm conservation, evaluation and utilization. 6. To continue to improve the efciency of agriculture and its degree of harmony with the environment by successful plant breeding.

References
Abel, C.A., Wilson, R.L. and Robbins, J.C. (1995) Evaluation of Peruvian maize for resistance to European corn borer (Lepidoptera: Pyralidae) leaf feeding and ovipositional preference. Journal of Economic Entomology 88, 10441048. Anonymous (1972) Corn. In: Genetic Vulnerability of Major Crops. National Academy of Sciences, Agricultural Board, National Research Council, Washington, DC, pp. 97118. Brown, W.L. (1975) A broader germplasm base in maize and sorghum. Annual Corn and Sorghum Research Conference Proceedings 30, 8189. Darrah, L.L. and Zuber, M.S. (1985) United States farm maize germplasm base and commercial breeding strategies. Crop Science 26, 11091113. Geadelmann, J.L. (1984) Using exotic germplasm to improve Northern corn. Annual Corn Sorghum Research Conference Proceedings 39, 98110. Goodman, M.M. (1985) Exotic maize germplasm: status, prospects, and remedies. Iowa State Journal of Research 59, 497527. Goodman, M.M. (1990) Genetic and germplasm stocks worth conserving. Journal of Heredity 81, 1116. Jenkins, M.T. (1978) Maize breeding during the development and early years of hybrid maize. In: Walden, D.B. (ed.) Maize Breeding and Genetics. John Wiley & Sons, New York, pp. 1328. LAMP (1991) Catalogo del germoplasma de maz, Vols 1 & 2. USDA-ARS, Washington, DC. Michelini, L.A. and Hallauer, A.R. (1993) Evaluation of exotic and adapted maize (Zea mays L.) germplasm crosses. Maydica 38, 275282. Pollak, L.M. (1993) Evaluation of Caribbean maize accessions in Puerto Rico. Tropical Agriculture (Trinidad) 70, 812. Salhuana, W. and Sevilla, R. (eds) (1995) Latin American Maize Project (LAMP), Stage 4 Results from Homologous Area 1 and 5. USDA-ARS Special Publication, Beltsville, Maryland. Salhuana, W., Jones, Q. and Sevilla, R. (1991) The Latin American Maize Project: model for rescue and use of irreplaceable germplasm. Diversity 7, 4042. Salhuana, W., Sevilla, R. and Eberhart, S. (1997) Latin American Maize Project (LAMP) Final Report. Pioneer Hi-Bred International, Inc. Special Publication G12083. Johnston, Iowa. Sevilla, R. and Salhuana, W. (1997) General description of the plan and execution of Latin American maize project (LAMP). In: Salhuana, W., Sevilla, R. and Eberhart, S.A. LAMP Final Report. Pioneer Hi-Bred International Inc. Special Publication G12083. Johnston, Iowa, pp. 136.

Germplasm Enhancement of Maize

329

Smith, J.S.C. (1988) Diversity of United States hybrid maize germplasm: isozymic and chromatographic evidence. Crop Science 28, 6369. Stuber, C.W. (1986) Use of exotic sources of germplasm for maize improvement. In: Dolstra, O. and Miedema. P. (eds) Breeding of Silage Maize. Pudoc, Wageningen, pp. 1931. Walsh, J. (1981) Genetic vulnerability down on the farm. Science 214, 161164, Wilkes, G. (1989) Germplasm preservation: objectives and needs. In: Knutson, L. and Stoner A.K. (eds). Biotic Diversity and Germplasm Preservation, Global Imperatives. Kluwer Academic Publishers, Dordrecht, pp. 1341.

20

A French Cooperative Programme for Management and Utilization of Maize Genetic Resources

Groupe Mas INRA: A. GALLAIS1, A. BARRIRE2, A. BOYAT3, A. CHARCOSSET1, J. DALLARD3, M. DERIEUX4, J.L. DESSELLE3, P. DUBREUIL1, H. DUVAL3, P. GARNIER1, B. GOUESNARD3, V. LAVERGNE1, M. LEFORT5, P. MICLO3, Y. MONTALANT4, A. PANOUILL6, M. POLLACSEK7, C. REBOURG1, P. RIMIERI2 PROMAIS: J.P. MONOD8, A. BARATIN9, M. BARON10, G. BERTHE10, M. CAMBOLIVE11, P. CAROLO12, F. DEVAUX13, J. DUROVRAY14, M. FERRANET13, A. LACOMBE15, A. LAMBERT16, M. LEIPERT17, A. MAJEST18, D. MISEVIC11, P. NOURRISSAT14, M. RAILLARD19, T. RONSIN20, J.P. SAMPOUX12, D. SEGONDS21, C. TABEL21, P. VINCOURT12
Station de Gntique Vgtale, Ferme du Moulon, 91190 Gif-Sur-Yvette, France; 2INRA, Station dAmlioration des Plantes Fourragres, 86600 Lusignan, France; 3INRA, Station dAmlioration des Plantes, Domaine de Melgueil, 34130 Mauguio, France; 4INRA, Domaine de Brunehaut, 80200 Estres-Mons, France; 5Bureau des Ressources Gntiques (BRG), 16 rue Claude Bernard, 75231 Paris cedex 05, France; 6INRA, Unit Exprimentale du Mas, 40390 Saint-Martin-de-Hinx, France; 7INRA,Unit dAmlioration des Plantes, Domaine de Crouelle, 63039 Clermont-Ferrand cedex, France; 8PROMAIS, 20 rue Bachaumont, 75002 Paris, France; 9CAUSSADE Semences, Z.I. Avenue de Meaux, 82300 Caussade, France; 10LIMAGRAIN Gntique, BP 1, 63720 Chappes, France; 11PIONEER Gntique, BP 5, 31840 Aussonne, France; 12RUSTICA Prograin Gntique, Domaine de Sandreau, Mondonville, 31700 Blagnac, France; 13MAISADOUR, Silo de Haut Mauco, Route de Saint Sever, 40000 Mont de Marsan, France; 14NOVARTIS, 12 route de Crossay, Senneville, 28700 Francourville, France; 15VAN der HAVE France, Domaine de la Revanche, 32700 Lectoure, France; 16VERNEUIL Recherche, Avenue de la Croix Blanche, 49630 Corn, France; 17SDME, Centre de Slection, Rue du Bois Musquet, 28300 Champhol, France; 18COOP DE PAU Semences, Avenue Gaston Phoebus, 64230 Lescar, France; 19CARGILL Gntique Europe, Croix de Pardies, 40300 Peyrehorade, France; 20NICKERSON Semences, Route de Longu, 49160 Longu-Jumelles, France; 21RAGT, Avenue St Pierre, BP 31, 12033 Rodez cedex 5, France
IPGRI/FAO 2001. Broadening the Genetic Base of Crop Production (eds H.D. Cooper, C. Spillane and T. Hodgkin)
1INRA-UPS-INA.PG,

331

332

A. Gallais et al.

Introduction
It has been observed in several countries, particularly in the USA and France, that there are very few effective genitors of commercial maize hybrids, in spite of a great number of hybrid varieties. This is due to the fact that few elite inbred lines can yield very good hybrids and that, when such a good line is developed, it is used in several combinations, as well as in pedigree selection to develop new lines. The consequence is relatedness among hybrids because they share a same parent or related parents. In France the situation for early maize was particularly alarming around 1985. At this time, the famous int line F2, derived from the Lacaune population, was involved in about 85% of early maturing hybrids. Such a situation may present a risk for agriculture due to the lack of diversity among varieties used by farmers. The risk can be a susceptibility to environmental conditions and in particular to disease, as illustrated by the Helminthosporium epidemic in 1970 in the USA, following a large-scale use of a cytoplasm allowing male sterility for the production of hybrids. Another risk for the breeder of early maize was a limit in the genetic advance due to a narrow base in breeding material, particularly for int material. To stimulate and prepare long-term genetic advance through the broadening of the genetic base of the breeding material, mainly in early maturing maize, a cooperative programme was developed, in a rst phase from 1983 to 1993, between INRA and PROMAIS (an association grouping 13 private breeding companies). The technical objectives were:

To collect, multiply and conserve a large number of populations adapted to French conditions, with particular emphasis on early-maturing material. To study the performance of the populations and their combining abilities, for grain and also for silage yield because silage production is important in northern Europe and the combining ability groups of lines are not necessarily the same for grain and silage productions. To develop a restricted number of pools with good usefulness for selection, i.e. with a good average performance and a large genetic variance for traits of agronomic interest.

Two other important objectives were: (i) to develop methodology for the management of populations as genetic resources in maize; and (ii) to develop a permanent organization to maintain maize genetic resources in France. In a second phase, developed since 1993, studies with restriction fragment length polymorphism (RFLP) molecular markers were initiated: (i) for classifying the populations in groups; (ii) for predicting the genetic variance within each population; and (iii) for studying the origin of int material to have better management and use of this type of material in breeding programmes. As this phase was a continuation of studies begun in the rst phase, it was integrated into the whole programme.

Management and Utilization of Maize Genetic Resources

333

Material and Methods

Collecting and multiplication A total of 1236 adapted accessions were involved in this programme. Most of them were landraces or open-pollinated varieties. They originated from Europe, Asia, North and South America (see Table 20.1). Special emphasis was put on European landraces. In addition to 267 French populations, many populations were obtained from Eastern European countries. One hundred and twelve accessions were obtained by collecting missions to Chile and Argentina during 1984. Synthetic accessions (323) from bred material were also gathered, mainly from the USA. Although not considered in the present report, about 200 exotic accessions were also obtained from CIMMYT (Centro Internacional de Mejoramiento de Maiz y Trigo) and CIRAD (Centre de Coopration Internationale en Recherches Agronomiques pour le Dveloppement). These accessions, which originated from tropical countries (South America or Africa), were used by INRA to develop tropical temperate pools, and to study the introgression of exotic material within the temperate material (Gouesnard et al., 1996). All adapted accessions were multiplied, with controlled pollination, in 1983, 1984 and 1985, on the basis of 200 full-sib progeny for each population, i.e. 400 plants. They were maintained at Mauguio INRA Station under long-term storage conditions with safety duplication.

Table 20.1. Geographic origin of the accessions. Country Asia North America USA Canada Mexico South America Europe France Iberian peninsula Italy Bulgaria Hungary Romania Former Yugoslavia Poland Czechoslovakia Other European Others Total Landraces 55 45 38 4 3 131 676 267 68 22 85 87 57 35 18 21 16 6 913 Synthetics 27 157 137 13 7 13 125 64 0 15 13 10 17 3 3 0 0 1 323 Total 82 202 175 17 10 144 801 331 68 37 98 97 74 38 21 21 16 7 1236

334

A. Gallais et al.

Evaluation of the populations Per se and testcross values All accessions have been evaluated for their per se and testcross value. For maize, as hybrids are developed, it is important to classify the accessions according to their behaviour in combination with testers representative of different heterotic groups. Accession performance was tested at two locations with two replicates per location. Data were recorded for silking, plant and ear heights, length and width of ear leaf, ear characteristics (length, diameter, etc., for a subset of populations), kernel texture and for qualitative traits such as lodging resistance, prolicacy, protandry and health of ear. Based on maturity, the populations were distributed into six groups. In a second step, 262 populations originating from France were evaluated simultaneously for the previous morphological and ear traits including kernel size and weight (Gouesnard et al., 1997). The evaluation of testcross performances was carried out according to maturitybased groups for grain and silage production. Each population was evaluated in combination with three testers for grain, and with two testers for silage at two locations and three replicates. Grain yield, grain moisture, lodging resistance, and plant and ear height were observed in the grain trials whereas dry-matter yield, dry-matter content, lodging resistance and some morphological traits were recorded in the silage trials. The overall analysis of all data was made difcult by the distribution of earliness groups across different locations. Some locations were common to different groups. Data of two locations were grouped only when the residual variancecovariance matrices were homogeneous. Analysis of interactions was only studied in these situations of homogeneity (Miclo and Desselle, 1991). Polymorphism of molecular markers
ENZYMATIC LOCI (STUDIES IN THE FIRST PHASE)

Enzymatic polymorphism were studied on 204 populations (Lavergne, 1988; Lefort-Buson et al., 1991; Garnier, 1992). The results were used for the classication of accessions into groups and for the prediction of within-population variability for quantitative traits. Each population was represented by 1060 individuals and characterized up to 27 polymorphic loci.

RFLP (STUDIES IN THE SECOND PHASE)

RFLP studies were undertaken to take into account the limitations of enzymatic markers, and to examine further genetic polymorphism and classication in groups. In a rst step, 117 inbred lines were characterized for 63 probe enzyme combinations (Dubreuil, 1996). This study showed that RFLP markers are very efcient for classifying inbred lines in heterotic groups. Thus, they were used in a second step to characterize ten populations, each represented by 30 individuals, with 35 probe enzyme combinations (Dubreuil and Charcosset, 1998). To apply this technique for the classication of a great number of populations at a reasonable cost, RFLP analyses were subsequently done using bulks of ten individuals and three bulks per population (Rebourg, 1996). These methods are now being applied to about 200 European populations to study their origin and their polymorphism. A study was also undertaken to see whether the within-population RFLP polymorphism could give some information on the genetic variability for quantitative traits.

Management and Utilization of Maize Genetic Resources

335

Development of pools To facilitate the management of the variability and to prepare the material for breeders, it was necessary to create a small number of operational pools of germplasm accessions. Several approaches, such as principal component analysis (Desselle, 1991), illustrated that the variability among populations had a continuous distribution. No clear relationship between the geographical origin of the populations and their agromorphological characteristics could be established in test-cross trials (Lavergne, 1988). Thus, geographical origin of the populations was not considered as a classication criterion for the development of pools. Instead, they were created using utilization criteria. Using an index that combined per se and testcross traits (grain yield, grain moisture, lodging resistance for grain pools, and dry-matter yield, dry-matter content and lodging resistance for silage pools), the poorer populations were not selected for inclusion in germplasm pools (but always maintained in a coldroom). For grain and silage and for a given earliness, two types of pools were developed: (i) pools with populations exhibiting a good average value in crosses with the different testers (good combining ability GCA); and (ii) pools developed from the best populations in combination with a given tester (GCA + SCA specic combining ability). In all cases, emphasis was put on early-maturing germplasm. For each year of population evaluation, a series of pools was developed. Using previous criteria for denition of groups, populations can be present in several pools. Populations with a good general combining value can also be present within a pool having good combining value with a given tester. Also, many populations are present both in grain and silage pools. About half of the populations (630) were included in pools. A total of 31 broad-based pools were developed (Table 20.2). Nine additional restricted-based pools were developed by intermating about ten populations for each pool, chosen on the basis of their combining ability with a given tester. They were expected to have a better performance and a lower level of genetic variability than broad-based pools. After a generation of intercrossing in isolation, all developed pools were evaluated for their per se and testcross value with three testers in a multisite experiment.

Genepool enhancement In spite of selection for lodging resistance among populations to form the pools, they were still very susceptible to lodging and therefore difcult to use in classical breeding programmes. To improve adaptation traits, and particularly lodging resistance, 24 of the 31 pools were crossed to six elite inbred lines or three single-crosses of the same heterotic group. Two generations of random mating were developed, and then the derived improved broad-based pools were evaluated for their per se and testcross value.

Breeding potential of developed pools The breeding potential of four grain and four forage improved broad-based pools was assessed by one cycle of recurrent selection for combining ability with a tester

336 Table 20.2. Number and characteristics of developed pools. Good SCAa with int tester F2 F257 4 4 1 2 4 4 2 2 4 4 3

A. Gallais et al.

Heterotic group Grain early pools broad-based improved broad-based restricted-based selected improved broad-based Silage early pools broad-based improved broad-based restricted-based selected improved broad-based Grain half-late pools broad-based improved broad-based Grain late pools broad-based
aGCA:

Good GCAa 4 4

Good SCA with early dent tester F250 F252 4 4 2 2 4 4 2 2

Good SCA with dent tester A632 F742 or MBS847b 4 2

general combining ability (average of crosses with three testers); SCA: specic combining ability. bIn 1985 and 1986.

(Table 20.2). The tester was chosen for its combining ability with the pool. For each pool, 400 plants were selfed, and 200 were selected on their S1-value basis and then crossed to the tester. The progeny were evaluated in several locations in comparison with the best commercial hybrid check. According to an index combining agronomic traits for grain or silage production, 40 S1 families were nally selected and intercrossed.

Results
Evaluation of populations At the level of per se value, for all measured traits, a large variation was observed among populations. A part of the observed variation for morphological traits was explained by variability in owering earliness. However, variability was still present when the owering earliness differences among populations were taken into account in the analysis (Miclo and Desselle, 1991). Variability for ear morphology traits among populations appeared to be particularly large (Duval et al., 1994). Recently, the study of all French populations by Gouesnard et al. (1997) has shown a relationship between their morphological traits and their geographical origin; studies are in progress for the development of a core collection applied to this set of populations. Population location interactions of the testcrosses were signicant in most situa-

Management and Utilization of Maize Genetic Resources Table 20.3. Yield (q ha 1) and lodging (%) for per se value of grain pools. Good GCA Heterotic group Broad-based pools Improved broad-based pools Restricted-based pools Yield Lodging 53 62 34 26 Good SCA on int tester F2 F257 Yield Lodging 46 64 56 43 34 29

337

Good SCA on early dent T. F250 F252 Yield Lodging 52 57 54 31 28 20

GCA: good combining ability; SCA: specic combining ability.

tions studied; however, they had a smaller magnitude than the population effects. The population tester interactions were also lower in magnitude than those of the tester effects and the population effects. This interaction was greater for grain yield than for grain moisture (Miclo and Desselle, 1991). Lavergne et al. (1991) and Miclo and Desselle (1991) found positive correlations between per se and combining ability values in crosses with different testers. This suggests the predominance of additive gene effects for traits of interest. This also emphasizes the value of preliminary selection on per se value for these materials.

Value of the pools For per se value, yield increase of improved broad-based pools relative to unimproved broad-based pools was signicant, mainly for grain pools (Table 20.3). Grain moisture and lodging resistance were also improved (1% and 1 to 6%). Nevertheless, improvement depended on combining ability group: the increase in yield was about 30% for dent pools, and about 10% for the int pools. The improvement for the restricted pools was lower than for improved pools. The improvement according to an agronomic index for silage pools was lower than the improvement for grain pools. The progress in testcross value was lower, because the tester buffers the genetic variance (Table 20.4, for dent pools). Yield of dent pools increased by about 20%, and that of int pools by 10%. In reference to hybrid checks (100%), int material progressed from 84% to 92%, and dent material progressed from 80% to 93%. Progress on grain moisture was about 1.3% obtained both on improved and

Table 20.4. Grain yield (q ha 1) and grain moisture (%) of grain dent pools in crosses with int tester. Pools Broad-based pools Improved broad-based pools Restricted-based pools F250 F252 Grain yield 54.2 64.0 57.3 69.4 Grain moisture 29.8 29.1 28.7 27.5

338

A. Gallais et al.

restricted pools. This study shows that the improvement of agronomic traits of broad-based pools is possible using these two methods, either by crossing with elite hybrids, or by using restricted pools. Although the rst one appears more efcient, the choice of material is very important. In this study, it appears that the int elite inbreds used for improvement had a lower agronomic value than the dent inbreds. Flint inbreds are generally rst-generation lines, and so their value is expected to be close to that of int populations, relative to the difference in value between dent populations and dent inbreds, which have been selected for several cycles. In addition, crossing with elite inbreds makes the pools closer to a specic combining ability group, which orients its further utilization.

Breeding potential for recurrent selection The results show a great variability among S1 families for yield, lodging resistance and earliness, on both per se and topcross evaluations. The expected gain from selection in one cycle was calculated, using the actual selection rate (from 10% to 20%). On grain pools, the expected genetic gain was about 0.4 t ha 1 for yield and 1% for grain moisture. On silage pools, it was about 0.7 t ha 1, which allows a yield of about 90% of that of the commercial check.

Polymorphism at marker loci and relationship with genetic variation Enzymatic polymorphism Most loci displayed two or three alleles. It appears that within-population variability was important relative to the differentiation among populations. On average, the within-population diversity represented 75% of the total allozyme diversity (LefortBuson et al., 1991; Gallais et al., 1992). It appeared that two groups of populations which were clearly differentiated for their specic combining ability with two testers displayed similar allelic frequencies. From this result, isozymes appeared to be of limited value to dene relevant heterotic groups. This may be due to a low number of loci and alleles per locus and indicates the value of alternative marker techniques such as RFLP. RFLP The study showed, on average, six alleles per locus, which is clearly superior to isozymic polymorphism for similar studies (Dubreuil et al., 1996). Furthermore, it showed that classications of inbred lines, established based on RFLP information, are highly consistent with the origin of these lines and their known combining ability (heterotic groups). The study of populations for their RFLP and enzymatic polymorphism showed that within-population diversity represents about 75% of the total diversity for both techniques. However, unlike isozymes, groups of populations dened by using 2535 RFLP probe enzyme combinations share common origins (Dubreuil and Charcosset, 1998) or specic combining abilities (Rebourg, 1996). This result can be explained both by the greater polymorphism that was observed at RFLPs and by the larger number of loci that were considered.

Management and Utilization of Maize Genetic Resources

339

Relationship between genetic diversity and genetic variation Besides the establishment of genetic groups, another potential interest of the markers could be the prediction of the variance of a population for agronomical traits through the amount of genetic diversity. Isozyme data (Garnier, 1992; Gallais et al., 1992) showed that populations that had the lowest isozyme diversity also displayed the lowest variance. Thus, populations that display a low diversity should be considered for direct selection only if their average value is high. Complementary investigations on another set of populations and using RFLP markers (Dubreuil, 1996) conrmed such a relationship within populations from the same origin.

Conclusion
This comprehensive genetic resource conservation and utilization programme depended upon the long-term cooperation between French public and private breeders. Important results included the following:

The collection of a large number of populations of maize and their evaluation for agronomic molecular traits. The development of a restricted number of broad-based genetic pools, which facilitates the management and use of broad diversity. The improvement of some pools to stimulate their use and to broaden the genetic base of material used by breeders.

All participants in the programme were supplied with all pools and have, to some extent, exploited them for their breeding proposes. Today, breeders are aware of the value of these genetic resources and willingly share the tasks associated with the longterm conservation of the populations. The programme was also very informative concerning the methodology for the management and use of genetic resources. First, with the use of different types of data, it was possible to classify the populations into groups, and to develop pools useful for the breeders. RFLPs appear to have a greater potential value than isozymes for the classication into groups and the prediction of within-population variability for quantitative traits; they can allow populations without sufcient potential to be discarded. Finally, the strategy that was used to prepare new material useful in the relatively short term for the plant breeder, i.e. development of pools after low selection intensity between populations and introgressing these pools by elite material, seems very efcient. However, it is clear that to make the maximum use of the genetic variability of such promising pools, several cycles of recurrent selection are necessary. This kind of strategy should be all the more important as the difference in performance between elite germplasm and genetic resources of the collection increases.

References
Desselle, J.L. (1991) Rapport sur le Programme Populations Sources mas. 2me partie: tude de la base de donnes par analyses en composantes principales. INRA, Montpellier.

340

A. Gallais et al. Dubreuil, P., Dufour, P., Krejci, E., Causse, M., De Vienne, D., Gallais, A. and Charcosset, A. (1996) Organization of RFLP diversity among inbred lines of maize representing the most signicant heterotic groups. Crop Science 36, 790799. Dubreuil, P. (1996) Etude de lapport des marqueurs RFLP pour lanalyse de la diversit gntique et sa structuration chez le mas. Relations avec les caractristiques agronomiques de populations traditionnelles. PhD Thesis, South Paris University. Dubreuil, P. and Charcosset, A. (1998) Genetic diversity within and among maize populations: a comparison between isozyme and nuclear RFLP loci. Theoretical and Applied Genetics 96, 577587. Duval, H., Gouesnard, B., Boyat, A. and Dallard, J. (1994) Populations sources mas. Programme coopratif INRA-Etablissements de slection-Ministres, 19831993. Document interne. INRA, Montpellier. Gallais, A., Duval, H., Garnier, P. and Charcosset, A. (1992) Un exemple de gestion des ressources gntiques en vue de la slection. In: Colloque en hommage Jean Perns, Bureau des Ressources Gntique, Lavoisier, Cachan, France, pp. 477490. Garnier, P. (1992) Contribution ltude de la variabilit gntique inter et intra population chez le mas (Zea mays L): valorisation dinformations agromorphologiques et enzymatiques. PhD Thesis, INAPG, Paris. Gouesnard, B., Dallard, J., Panouill, A. and Boyat, A. (1997) Classication of French maize populations based on morphological traits. Agronomie 17, 491498. Gouesnard, B., Sanou, J., Panouill, A., Bourion, V. and Boyat, A. (1996) Evaluation of agronomic traits and analysis of exotic germplasm polymorphism in adapted exotic maize crosses. Theoretical and Applied Genetics 92, 368374. Lavergne, V. (1988) Variabilit entre populations de mas (Zea mays L.). Approche biomtrique et approche enzymatique. PhD Thesis, INAPG. Lavergne, V., Lefort-Buson, M., Daudin, J.J., Charcosset, A., Sampoux, J.P. and Gallais, A. (1991) Genetic variability among populations of maize germplasm. 1. Comparative analysis of top cross values and per se values of populations. Maydica 36, 227236. Lefort-Buson, M., Lavergne, V., Daudin, J.J., Charcosset, A., Sampoux, J.P. and Gallais, A. (1991) Genetic variability among populations of maize germplam. 2. Enzymatic polymorphism and its relationship to quantitative trait diversity. Maydica 36, 237246. Miclo, P. and Desselle, J.L. (1991) Rapport sur le Programme Populations Sources mas. 1re partie: tude de la base de donnes par analyses de variance. INRA, Montpellier, France. Rebourg, C. (1996) Analyse de la diversit de populations de mas laide de marqueurs molculaires. Application la gestion des ressources gntiques. Diplme dEtudes Approfondies, Pierre et Marie Curie University, Paris.

21

Broadening the Genetic Base of Lentil in South Asia

W. ERSKINE1, S. CHANDRA2, M. CHAUDHRY3, A. SARKER4, B. SHARMA5, M. TUFAIL3 AND M.C. TYAGI5

1International

Center for Agricultural Research in Dry Areas (ICARDA), PO Box 5466, Aleppo, Syria; 2Indian Institute for Pulses Research, Kanpur 208024, India; 3Ayub Agricultural Research Institute, Faisalabad, Pakistan; 4Bangladesh Agricultural Research Institute, Joydebpur, Dhaka, Bangladesh; 5Genetics Division, Indian Agricultural Research Institute, New Delhi 110012, India

Introduction
Lentil (Lens culinaris Medikus) was developed from Lens culinaris ssp. orientalis (Boiss.) Ponert in the eastern Mediterranean region as one of the rst domesticated plants (Zohary and Hopf, 1988). Evidence of the spread of lentil eastward into the IndoGangetic Plain dates to around 2000 BC, but previous contacts between Mohenjo-Daro and the Sumerians and Akkadians of Mesopotamia are well documented, and it may have been introduced into the Indus valley earlier (Cubero, 1981). De Candolle (1882) wrote that on linguistic evidence It may be supposed that the lentil was unknown in this country [India] before the invasion of the Sanskrit-speaking race. Today, approximately half of the worlds area (46.3%) of lentil is in South Asia (FAO, 1996) (Table 21.1), where indigenous lentils are of a specic ecotype (pilosae) and exhibit a marked lack of variability. This lack of variability has limited breeding progress. This chapter describes the genetic bottleneck in lentil in South Asia and the introduction of new genetic diversity into the region.

Genetic and Phenotypic Variation in South Asian Lentils

Qualitative morphological characters In a monograph on lentil in 1930, Barulina described a special endemic group of cultivated lentil in South Asia, characterized by strong pubescence on the vegetative canopy. The foliage of this group, named grex pilosae, is grey-green because of the presence of soft hairs, and it contrasts with the green colour common for lentil.
IPGRI/FAO 2001. Broadening the Genetic Base of Crop Production (eds H.D. Cooper, C. Spillane and T. Hodgkin)

341

342

W. Erskine et al. Table 21.1. Average area (ha) sown with lentil in selected lentil-producing countries in 1996 (FAO, 1996), together with the mean coefcient of variation (CV%) over nine agro-morphological characters measured on accessions (acc.) in the evaluation of a world germplasm collection. The number of accessions in each country sample is also given. Country Afghanistan Bangladesh India Iran Lebanon Nepal Pakistan Syria Turkey
SEDb

Area (1000 ha) n.a. 206 1124 175 7 175 63 128 630 3389

No. of accessions 32 n.a. 100 100 25 n.a. 14 31 77

CVa (%) 19.6 n.a. 15.7 19.0 21.2 n.a. 16.6 19.0 19.1 2.6

World

n.a.: data not available a Coefcients of variation (%) calculated for each trait in each country. The mean CV over traits is presented. Data from Erskine et al. (1989). b SED from a one-way analysis of variance of CV over countries for nine traits.

Additionally, pilosae lentils either lack tendrils or have only short or rudimentary tendrils. These two qualitative characters differentiate the lentil of South Asia from other cultivated lentils.

Quantitative morphological characters Regional groups were revealed in a study of the quantitative agro-morphological variation in germplasm from 13 major lentil-producing countries (Erskine et al., 1989). Landrace accessions from India and Ethiopia were strikingly similar for the nine quantitative traits surveyed, and clustered together in one group. The group was characterized by early owering and maturity, low biological yield, short stature and lowest pod height, and small seeds. This similarity for quantitative traits was previously noted by Barulina (1930), but she indicated that germplasm from these countries differed markedly for qualitative characters, with the Ethiopian material characterized by an elongated pod apex and the Indian material characterized by the pilosae traits. Using data of Erskine and Witcombe (1984) and Erskine et al. (1989), it is possible to assess if germplasm from various countries differs in phenotypic variability. To this end, the coefcient of variation among accessions from a country for each trait was subjected to a one-way analysis of variance on the basis of country of origin undertaken over characters (Table 21.1). India and Pakistan have the lowest CV despite the fact that India is the worlds largest lentil producer (Table 21.1) and has considerable heterogeneity among its growing environments.

Base-broadening in Lentil

343

Adaptation
When lentil was rst introduced into the Indo-Gangetic Plain, it was exposed to environmental conditions very different from those at its origin in the Fertile Crescent of West Asia. A possible reconstruction of the phenological problems associated with the initial spread of the crop into the Indo-Gangetic Plain was inadvertently made when lentil selections from West Asia were introduced there through early International Center for Agricultural Research in Dry Areas (ICARDA) international nurseries. When lentils selected in West Asia were sown in India and Pakistan, most came to ower when the indigenous lentils were maturing (Ceccarelli et al., 1994). This nding prompted research on environmental factors that control owering in lentil and it is now known that temperature and photoperiod modulate owering (Summereld et al., 1985). In a study of a broad spectrum of germplasm under a range of temperatures and photoperiods, Indian germplasm was more sensitive to temperature and less responsive to photoperiod than germplasm from West Asia (Erskine et al., 1990, 1994). This suggests that selection for local adaptation during the spread of lentil into the IndoGangetic Plain at least 4000 years ago has greatly affected phenology and the genetic response to environmental factors. Thus, pilosae germplasm is also strikingly more susceptible to cold damage than germplasm from West Asia (Erskine et al., 1981). There is also evidence for the specic adaptation of pilosae germplasm to the edaphic factors of iron (Fe) and boron (B) deciencies, providing a further contrast to genetic material from the Middle East. Fe-deciency symptoms are observable on some accessions of lentil grown in calcareous soil with pH 8.0, as is common in the Middle East (Erskine et al., 1993). In a germplasm collection, originating from 18 countries, very few landraces from Syria and Turkey exhibited Fe-deciency symptoms. Those landraces exhibiting symptoms of Fe-deciency mostly originated from relatively warm climates, such as India and Ethiopia. This resulted from either the chance introduction to these regions of susceptible to low levels of iron in the soil, or unconscious selection pressure in favour of such types. Recently, a sample of the world lentil germplasm collection was screened for boron (B)-deciency tolerance in a soil with low boron content at Rampur, Nepal. All accessions showed severe B-deciency symptoms except pilosae accessions from Bangladesh, India, Nepal and Pakistan (W. Erskine, personal observation). Recent evidence from seven isozyme loci and 22 random amplied polymorphic DNA (RAPD) loci show that germplasm from South Asia (Bangladesh, India, Nepal and Pakistan) and Afghanistan differs markedly from that from the rest of the world. Additionally, South Asian and Afghanistan germplasm was found to have low levels of genetic diversity (Ferguson et al., 1998). These data conrm the morphological and physiological distinctiveness of pilosae germplasm. However, Afghan germplasm is not of the pilosae type. One explanation for the similarity of pilosae lentils and Afghan material at the isozyme and DNA level is that Afghanistan was the route of entry into the Indian sub-continent. On average, Afghan germplasm is among the latest to ower in the world, owering on average 3 weeks later than Indian material (Erskine et al., 1989). Afghan (highland) lentils are now reproductively isolated from the lowland pilosae type because of the mismatch in owering time. A detailed study of germplasm from Pakistan including material from highland areas adjacent to Afghanistan and from the lowland Indo-Gangetic Plain may provide the key to further understanding of

344

W. Erskine et al.

the introduction of lentil into South Asia. However, the evidence from isozyme and DNA studies clearly indicates that lentil was not introduced into India from Ethiopia along the same route as crops such as sorghum (Sorghum bicolor (L.) Moench.) and cowpea (Vigna unguiculata (L.) Walp.), despite the similarity in quantitative morphological characters between Ethiopian and the pilosae lentil of South Asia (Simmonds, 1976). In summary, by comparison with other Asian lentils, the pilosae group of South Asia is characterized by two endemic qualitative morphological traits, precocity in owering and maturity, low biomass, increased sensitivity of the temperature control of owering, and high frequency of boron efciency and of iron inefciency. Furthermore, lentil germplasm from India is the least variable among lentil-producing countries, despite India being the largest producer in the world. It further seems likely that the asynchrony in owering between exotic material from West Asia and the indigenous germplasm in South Asia enforced reproductive isolation of pilosae lentil. A genetic bottleneck occurred which was maintained by differing owering times and resulted in germplasm with reduced genetic diversity for agro-morphological, isozyme and molecular traits.

Broadening the Genetic Base in South Asia

Plant introduction A narrow genetic diversity limits the scope for new recombinations, and the simplest approach to widening the variation available to breeders in South Asia is plant introduction. As noted above, the introduction of material from West Asia into the IndoGangetic Plain is difcult because of non-overlapping owering. The introduction of some exotic, earlier-owering material, however, has been more fruitful. For example, Precoz (ILL 4605) was the rst early-owering, large-seeded (macrosperma) (4.5 g per 100 seeds) lentil introduced into South Asia, where endemic germplasm has a seed weight 3.0 g per 100 seeds. Precoz thrives in the wetter areas of Pakistan, such as the National Capital Territory and parts of Northwest Frontier Province, where it has been released for cultivation as Manserha 89. Its early owering synchronizes with the owering of indigenous germplasm, allowing articial crossing without the need for extended photoperiod (Tyagi and Sharma, 1981). Precoz and/or its derivatives have been included in every crossing block in India. For the upland areas fringing the Indo-Gangetic Plain, the introduction of germplasm and breeding material from West Asia is important because of the specic needs for cold tolerance and a medium-duration crop. For example, in Baluchistan, Pakistan, the medium-maturing accession ILL 5865 was released as ShirAZ-96 in 1996 (ICARDA, 1997).

Hybridization The genetic variation in a region may also be increased through hybridization within the primary genepool. The use of the early-owering Precoz in South Asia is one

Base-broadening in Lentil

345

example of this. However, a major systematic effort to widen the genetic diversity available to breeders in South Asia has been made at ICARDA by extensive crossing of pilosae lentils with primary genepool germplasm from other origins since 1981. An extended photoperiod (18 h) has been used in the crossing block to improve synchrony in owering between diverse parents. The F1 generation was grown in a high-elevation (890 m above sea level) summer nursery at Terbol, Lebanon, for rapid generation advancement, allowing two generations per year. Segregating populations were then either distributed to national programmes as part of the International Legume Testing Network in an international F3 or F4 nursery or sent to specic national programmes. Selections were made locally for adaptation and disease resistance. This system exploited the comparative advantage at ICARDA in crossing and rapid generation advancement with selection in the target production area. In this manner, targeted crosses for Bangladesh were made at ICARDA, Syria, to incorporate resistance to rust, a major production problem. Selections were then made in Bangladesh of adapted rust-resistant plants from the segregating populations. As a result, Bangladesh released its rst rust-resistant lentil Barimasur-2 (selected from cross ILL 4353 ILL 353) in 1993. Barimasur-2 has yield advantage of c. 20% over the standard check (Table 21.2) with wide adaptation. A further selection Barimasur4 (selected from L5 FLIP 84-112L) was released in 1995 for high yield (53% increased yield over the check) with combined resistance to rust and stemphylium blight (Table 21.2). In a pilot production project, the results of 12 farmer-managed onfarm trials (1996/97) showed that the new cultivars were superior to the local cultivar by 23110% in Bangladesh. With an erect plant stature the variety Barimasur-4 is suitable for inter-cropping with sugar cane and mixed-cropping with mustard and linseed. In Pakistan the variety Masoor 93 was released for cultivation in the Punjab province of Pakistan with a 31% increased yield over the check and improved disease resistance and seed size (Tufail et al., 1995) (Table 21.2). It comes from the cross 18-12 (local) ILL 4400 (from Syria) made at Faisalabad, Pakistan, using hybridization techniques learned at ICARDA. The cultivar Masoor 93 is now grown on about 2025% of the lentil area in Pakistan. In India, the intensive use of Precoz and its derivatives in hybridization by local lentil improvement programmes has resulted in the development of extra-bold and extra-early lines. The All-India coordinated lentil trial programme has recently started

Table 21.2. Varieties released in South Asia from crosses between pilosae lentils from South Asia and representatives of other genepools.
Country Bangladesh Variety Barimasur-2 Barimasur-4 Masoor-93 Year of release 1993 1995 1995 Parentage ILL4353 ILL353 L5 FLIP84112L 1812 ILL4400 Yield (t ha 1) 1.8 2.3 1.7 % increase over check Other important traits 20 53 31 Resistant to rust Erect, resistant to rust and Stemphylium blight Resistant to rust, Botrytis and Ascochyta blights and sclerotinia stem rot

Pakistan

346

W. Erskine et al.

two new categories of trial, as a result of the broadening of the genetic diversity of lentil in South Asia: (i) extra-bold seeds ( 35 g per 1000 seeds); and (ii) extra-early (maturity 115 days). In the 1996/97 cropping season there were a total of 93 lines in the All-India Coordinated Trials, of which 44% (predominantly in the extra-bold seeded and extra-early nurseries) had exotic ICARDA parentage. In India, the extra-early nursery is particularly important because it opens up a large new potential cropping niche for lentil in India, namely late-sown lentil (using an early-maturing cultivar) following the harvest of a long-season rice crop. At present, there are 4 million ha left fallow in winter after the harvest of long-season rice in India annually. With intensive cropping, the newly developed early lines can be tted in late planting conditions. These lentil lines mature in 115 days compared with 135145 days for landraces from North India.

Evolution of the ICARDA Breeding Programme

The above is illustrative of a changing approach by ICARDA and national programmes as national capacities for lentil breeding have risen and the need for specic adaptation has been recognized (Ceccarelli et al., 1994). The lentil breeding programme at ICARDA is built upon the foundation of the germplasm collection, as the source of variation for breeding. Additionally, the germplasm collection has contributed philosophically to breeding because the strategy of the improvement programme ows from an understanding of the evolutionary patterns of variability of landraces and their adaptation. ICARDAs lentil breeding programme aims to produce genetic material for a series of separate, but nely geographically targeted streams, linked closely to national breeding programmes (Robertson and Erskine, 1997). The breeding strategies used for this annual, diploid, self-pollinated food legume have changed with time. In Stage 1, the variation in the world lentil germplasm collection was directly exploited. Selection was made among and within locally adapted landraces. These selections were distributed to national programmes through the International Nursery Network to test for local adaptation. As a result, many cultivars released by national programmes are actually selections from landraces in the ICARDA germplasm collection (Fig. 21.1) (Robertson et al., 1996). These Stage-1 registrations emphasize rstly the value of direct exploitation of landraces and secondly the underexploited nature of lentil germplasm. The particular combinations of characters required for specic regions were often not found on the shelf in the germplasm collection. Consequently, ICARDA started hybridization, and selections from segregating populations made at ICARDA in Syria to produce Stage 2 material. Such selections were then distributed after multiplication to the national programmes to select in their respective agroclimatic conditions. This has resulted in the release of a number of cultivars in different regions (Fig. 21.1). However, lentil lines developed from selection at ICARDA in West Asia are mostly limited to adaptation to the home region. As a result, the breeding programme has decentralized to work closely with national programmes. For the other regions, at Stage 3, crosses are agreed upon with national-level cooperators and then made at ICARDA, Syria. Country-specic segregating populations are then shipped to national cooperators for local selection. Approximately 200 such crosses are made annually at ICARDA.

Base-broadening in Lentil

347

Fig. 21.1. Lentil varieties released by NARS (19821996). Germplasm: Stage 1 selections among and within accessions in ICARDA germplasm collection. Hybrids ICARDA selected: Stage 2 selections made at ICARDA, Syria in segregating populations from ICARDA crosses. Hybrids NARS selected: Stage 3 selections made by NARS from segregating populations from ICARDA crosses. NARS crosses: Stage 4 selections made by NARS from segregating populations from NARS crosses with parents from ICARDA.

Selections made by national programmes are then fed back into the International Nursery Network for wider distribution. In Stage 4, the national programmes use ICARDAderived material in their own crossing programmes and selections are made locally.

Future Prospects
In addition to broadening the genetic base of lentil production in South Asia, the hybridization of pilosae lentils with germplasm from other origins is producing spinoffs for regions outside South Asia. In particular, the early owering of the pilosae type is useful in drought escape through the avoidance of terminal drought in dry areas of the Mediterranean region (Silim et al., 1993). Furthermore, shifting from spring to winter planting of lentil in the highlands requires winter-hardy varieties. Winter-hardy germplasm is available to exploit this niche to boost lentil production in these regions. These lines are also being used in hybridization by ICARDA to generate high-yielding lines with improved cold tolerance. These examples illustrate the progress made in increasing the genetic variation of lentil in South Asia and the removal of an ancient genetic bottleneck. Plant introduction is of immediate value for lowland areas only for early owering and maturing germplasm; it is of greater value in the highland edges of South Asia, where the

348

W. Erskine et al.

cropping season is longer. Hybridization has introduced resistance to rust, and Stemphylium and Ascochyta blights, increased seed size together with extra-early owering into South Asia; it may also be useful in introducing a rapid early growth rate for the late- planting crop. Mutation breeding or gene transformation technology may be useful to incorporate resistance currently unavailable in the primary genepool to grey mould (caused by Botrytis cinerea), broomrape (Orobanche sp.) and pea leaf weevil (Sitona sp.).

References
Barulina, H. (1930) Lentils of the USSR and other countries. Bulletin of Applied Botany, Genetics and Plant Breeding, Leningrad Supplement 40, 265304 (Russian, English summary). Ceccarelli, S., Erskine, W., Hamblin, J. and Grando, S. (1994) Genotype by environment interaction and international breeding programs. Experimental Agriculture 30, 177188. Cubero, J.I. (1981) Origin, taxonomy and domestication. In: Webb, C. and Hawtin, G. (eds) Lentils. CAB International, Wallingford, UK, pp. 1538. De Candolle, A.P. (1882) Origins of Cultivated Species. Reprint 1967, Hafner, London. Erskine, W., Myveci, K. and Izgin, N. (1981) Screening a world lentil collection for cold tolerance. LENS Newsletter 8, 58. Erskine, W. and Witcombe, J.R. (1984) Lentil Germplasm Catalog. International Center for Agricultural Research in the Dry Areas, Aleppo, Syria. Erskine, W., Adham, Y. and Holly, L. (1989) Geographic distribution of variation in quantitative traits in a world lentil collection. Euphytica 43, 97103. Erskine, W., Ellis, R.H., Summereld, R.J., Roberts, E.H. and Hussain, A. (1990) Characterization of responses to temperature and photoperiod for time to owering in a world lentil collection. Theoretical and Applied Genetics 80, 193199. Erskine, W., Saxena, N.P., and Saxena, M.C. (1993) Iron deciency in lentil: yield loss and geographic distribution in a germplasm collection. Plant and Soil 151, 249254. Erskine, W., Hussain, A., Tahir, M., Baksh, A., Ellis, R.H., Summereld, R.J. and Roberts, E.H. (1994) Field evaluation of a model of photothermal owering responses in a world lentil collection. Theoretical and Applied Genetics 88, 423428. FAO (1996) Production Yearbook 1996. FAO, Rome, Italy. Ferguson, M.E., Robertson, L.D., Ford-Lloyd, B.V., Newbury, J. and Maxted, N. (1998) Genetic relationships of lentil landraces from sixteen countries. Euphytica (in press). ICARDA (1997) Legume Program Annual Report for 1996. International Center for Agricultural Research in the Dry Areas, Aleppo, Syria. Robertson, L.D. and Erskine, W. (1997) Lentil. In: Fucillo, D., Sears, L. and Stapleton, P. (eds) Biodiversity in Trust. Cambridge University Press, Cambridge, pp. 128138. Robertson, L.D., Singh, K.B., Erskine, W. and Abd El Moneim, A.M. (1996) Use of genetic diversity in germplasm collections to improve food and forage legumes for West Asia and North Africa. Genetic Resources and Crop Evolution 43, 447460. Silim, S.N., Saxena, M.C. and Erskine, W. (1993) Adaptation of lentil to the Mediterranean environment. I. Factors affecting yield under drought conditions. Experimental Agriculture 29, 919. Simmonds, N.W. (1976) Evolution of Crop Plants. Longman, London. Summereld, R.J., Roberts, E.H., Erskine, W. and Ellis, R.H. (1985) Effects of temperature and photoperiod on owering in lentils (Lens culinaris Medic.). Annals Botany 56, 659671. Tufail, M., Malik, I.A., Choudhary, M., Ashraf, M. and Saleem, M. (1993) Genetic resources

Base-broadening in Lentil

349

and breeding of lentil in Pakistan. In: Erskine, W. and Saxena, M.C. (eds) Lentils in South Asia. ICARDA, Aleppo, Syria, pp. 5875. Tufail, M., Ahmad, M. and Ali, A. (1995) Masoor-93: an ideal combination of characters for Punjab Province, Pakistan. LENS Newsletter 22, 5254. Tyagi M.C. and Sharma, B. (1981) Effect of photoperiod and vernalization on owering and maturity in macrosperma lentils. Pulse Crops Newsletter 1(2), 40. Zohary, D. and Hopf, M. (1988) Domestication of Plants in the Old World. Clarendon Press, Oxford, UK.

22
1The

Genetic Diversity of Barley: Use of Locally Adapted Germplasm to Enhance Yield and Yield Stability of Barley in Dry Areas

S. GRANDO1, R. VON BOTHMER2 AND S. CECCARELLI1

International Center for Agricultural Research in Dry Areas (ICARDA), PO Box 5466, Aleppo, Syria; 2Department of Plant Breeding Research, The Swedish University of Agricultural Sciences, SE-268 31 Svalv, Sweden

Introduction
The area of the Near East that includes parts of Jordan, Lebanon, Palestine, Syria, southeast Turkey, Iraq and western Iran is known as the Fertile Crescent (Fig. 22.1). The domestication of several of our main crops, including wheat and barley, took place in this area some 10,000 years ago (Zohary and Hopf, 1993) and this made it possible for man to move from hunting and food collecting to farming. Probably the most important of the early cereals was barley, and archaeobotanical material from the region clearly shows that the rst barleys were two-rowed (Harlan and Zohary, 1966). The wild progenitor of cultivated barley, Hordeum spontaneum, is still widely distributed along the Fertile Crescent where, particularly in the driest areas, it can be easily identied at a distance because of its height. Over the course of time since early domestication, barley has spread, together with agriculture, to new environments rst in close proximity to the primary area of domestication and then east and west to other areas in Europe, Asia and North Africa. Due to the exposure to new environments and repeated conscious selection by man for different purposes, new gene complexes for adaptation, resistance, quality and other traits have been developed. Natural selection gave rise to locally adapted landraces. Barley thus developed into a crop with high genetic diversity over the temperate areas of the old world. Like H. spontaneum, landraces constitute a major genepool of importance for breeding programmes. Barley is still one of the most important cereal crops in the Fertile Crescent, covering an area of approximately 5 million hectares. It is a typical crop of marginal, lowinput, drought-stressed environments (Ceccarelli, 1984) except for few pockets where it is also grown in favourable conditions. Barley grain and straw are the most important source of feed for small ruminants, mostly sheep. Conventional breeding and highyielding (HY) varieties had virtually no success in this area. But this lack of success has
IPGRI/FAO 2001. Broadening the Genetic Base of Crop Production (eds H.D. Cooper, C. Spillane and T. Hodgkin)

351

352

S. Grando et al.

Fig. 22.1. The Fertile Crescent, where crops such as barley, wheat, lentil, stone fruits and olives were domesticated (modied from Harlan and Zohary, 1966). Experimental sites used by ICARDAs barley project are shown.

had a positive effect in preserving biodiversity because in these environments all the barleys grown are landraces (Weltzien, 1988) that have evolved directly from the wild progenitor in hostile environments; they are very popular among farmers for their good feeding quality as both grain and straw. In Syria, farmers distinguish two major groups of landraces, largely on the basis of seed colour, namely Arabi Abiad (white seed) and Arabi Aswad (black seed). The rst is common in environments receiving between 250 and 400 mm annual rainfall; the second is cultivated in harsher environments with less than 250 mm annual rainfall. Although collected as early as 1950 by Vavilov, little is known about these two groups. A few accessions were included in the worlds genebank collections but, as in many other crops, little use has been made of these valuable genetic resources. Generally, in both developing and developed countries, the use in barley breeding of exotic or primitive germplasm, such as H. spontaneum and landraces, is not common. Elite breeding based on modern lines and newly released varieties prevails. Though these modern lines were originally based on material extracted from previous landraces of a particular area, modern breeding, private or public, concentrates its breeding efforts on a narrow genepool. This is a vulnerable strategy and basebroadening needs to receive higher priority from genebank curators, practical plant breeders and society in general. Pre-breeding efforts, mainly in the shape of testing of exotic material for disease resistance and backcrossing to highly bred varieties, have been carried out (Lehmann and Bothmer, 1988), but programmes where base-broadening is the major concern

Genetic Diversity of Barley

353

have only recently been initiated in some countries and should be further encouraged. So, for example, particular composite cross populations dynamic genepools composed of both landraces and H. spontaneum have been made separately for Swedish and for Finnish conditions (Vetelinen, 1997; Vetelinen and Nissil, Chapter 14, this volume). In the early 1980s it was postulated that because barley landraces have been grown continuously since domestication in unfavourable and stress environments and without inputs, their evaluation could teach a barley breeder a few lessons on adaptation to lowinput, stressful environments (Ceccarelli, 1984). It was also postulated that the lesson(s) could be useful to other breeders in countries where barley landraces are still predominant, as well as to breeders of crops grown prevalently in stress environments. This chapter illustrates the contribution that landraces and the wild progenitor of cultivated barley, H. spontaneum, can make to increase agricultural production, particularly in areas with severe climatic and nutritional stresses such as the Syrian steppe.

Experimental Conditions
From 1984, all breeding material was tested under typical northern-Syrian growing conditions, i.e. strictly rainfed, predominantly in areas with low and erratic rainfall, and without fertilizers, pesticides or herbicides. Three experiment sites (Fig. 22.1), representing three distinct agricultural systems, were used: Tel Hadya, where ICARDAs headquarters are located, is a typical highinput favourable environment with a wide choice of different crops; Bouider, rented from a farmer, is the opposite extreme, with a typical low-input highly risky environment and barley being the only possible rainfed crop; Breda, an ICARDA eld station, is intermediate between the two. The three sites are geographically close, being located at 35 (Tel Hadya), 60 (Breda) and 80 km (Bouider) south-southeast of Aleppo, respectively, and this is an enormous advantage in terms of eld operations. Table 22.1 shows the total rainfall at the three sites since the work on landraces began. Although rainfall does not convey all the information about climate, rainfall distribution and winter temperatures also play a determinant role; it is evident that there is a consistent rainfall gradient between the three sites, which makes the area unique in providing large climatic contrasts within short distances.

Source of Material
Wild barley The wild progenitor of cultivated barley was earlier treated as a separate species (Hordeum spontaneum C. Koch) but due to the full compatibility in crosses, no postzygotic sterility barriers and morphological close resemblance to cultivated barley it is now treated as a subspecies, Hordeum vulgare ssp. spontaneum (C. Koch) Thell. (Asfaw and Bothmer, 1990; Bothmer et al., 1995). For simplicity, however, the taxon is called H. spontaneum in this chapter. H. spontaneum is an annual, brittle, two-rowed, diploid (2n = 14), predominantly

354

S. Grando et al. Table 22.1. Total rainfall (mm) in the three experimental sites used by the barley breeding programme in northern Syria. Year 1984/85 1985/86 1986/87 1987/88 1988/89 1989/90 1990/91 1991/92 1992/93 1993/94 1994/95 1995/96 Long-term average Tel Hadya 372.6 316.4 357.9 504.2 234.4 233.4 293.5 352.6 290.1 373.3 312.9 404.5 337.2 Breda 276.6 218.3 244.6 414.0 194.8 183.2 241.3 263.2 283.0 291.2 244.2 359.8 267.8 Bouider 203.0 176.2 385.7 189.0 148.7 213.4 249.6 224.2 245.6 203.1 316.0 232.2

self-pollinated and colonizing species. There is considerable morphological similarity between H. spontaneum and cultivated two-rowed varieties, the major difference being in their modes of seed dispersal. In H. spontaneum, ears are brittle and fragment at maturity into individual arrow-like triplets, each containing one seed and two sterile lateral spikelets. Following dispersal, the seeds insert themselves into the ground driven by the movement of twisting awns due to changes in temperature and/or humidity. Under cultivation, this specialization, which ensures the survival of the plant in nature, is undesirable, and was replaced by non-brittle mutants, which were selected by man for reaping, threshing and sowing. The difference is controlled by one or two very closely associated genes (Bt and Bt1) on chromosome 3. Brittle rachis is dominant over non-brittle. H. spontaneum is also characterized by an extremely long period of seed dormancy, very long and rough awns, and densely haired empty glumes. The wild form occupies primary habitats such as dry mountain slopes, steppes and semideserts or more ephemeral biotopes. It is more open pollinated than modern types of cultivated barley; up to 10% outcrossing has been reported (Brown et al., 1978). It has thus a higher degree of heterozygosity and heterogeneity than cultivated barley in its populations. The high degree of genetic variation in H. spontaneum is visualized by the fact that there is a strong effect of natural selection for microhabitats (shady, exposed, dry, moist, etc.) leading to development of sub-population systems within very small areas (Nevo, 1992). In areas where H. spontaneum grows in contact with the barley crop, crosses and gene introgression are fairly common. This has certainly contributed to geneow between the two taxa over the entire period of cultivation in the Middle East, giving rise to a very versatile genetic system favouring the adaptive radiation of barley over the years. Most probably H. spontaneum still contributes to the evolutionary processes of barley landraces through a continuous introgression of genes. There is ample evidence in the literature that H. spontaneum can contribute useful genes for several characters. Resistance to powdery mildew (Moseman et al., 1983; Gustafsson and Claesson, 1988; Nevo, 1992; Lehmann et al., 1998), leaf rust

Genetic Diversity of Barley

355

(Moseman et al., 1990; Nevo, 1992) and other diseases (Nevo, 1992) has been identied in H. spontaneum, and its use in breeding for disease resistance has been reported by several authors. The species showed variation for important agronomic traits such as earliness, biomass, grain yield, grain protein, and tolerance to salinity and drought. This vast potential store of genetic resources is as yet largely unexploited. H. spontaneum is thus expected to have a potential in contributing useful genes in barley breeding as a donor of adaptive traits to extreme stress conditions, as suggested by its distribution in the driest areas of the region. In general, H. spontaneum is less vigorous in early growth than cultivated barley, although lines with good vigour are present. The level of cold tolerance is similar to that of H. vulgare but within the wild genepool accessions much earlier than H. vulgare are available. It was interesting to nd a large number of lines with a long peduncle as well as with the ability to extrude the spike above the ag leaf. High straw protein content may offer additional justication for the use of H. spontaneum (Table 22.2). The decisive year for including H. spontaneum in the breeding activities for stress environments was 1987 when a group of lines were evaluated at Bouider. That year, rainfall was only 176.2 mm and there was a crop failure (Ceccarelli, 1994) with virtually no lines able to produce grain. The only lines able to head and to produce some grains were two accessions of H. spontaneum. In these, some photosynthetic activity was found between 0700 and 0800 hours, the average net photosynthesis was 2.7 0.4 mol CO2 m 2 s 1 and stomatal conductance was 0.060 0.003 mol m 2 s 1. At the same time the stomata of the locally adapted landrace, Arabi Aswad, were closed. By midday the stomatal conductance of the H. spontaneum accessions decreased to 0.025 mol CO2 m 2 s 1 at a leaf water potential of 4.6 MPa while the leaf water potential of Arabi Aswad was 3.5 MPa and the stomatal conductance was zero. These lines combined earliness with an acceptable level of cold resistance, they were able to maintain good plant height under drought conditions and, despite the very dry conditions, they were also able to extrude part of the last internode outside the ag leaf. From the preliminary evaluation it was concluded that the most useful traits that H. spontaneum could contribute in relation to adaptation to stress environments are earliness combined with adequate level of cold tolerance and plant height under drought. The importance of plant height under drought is associated with the fact that one of the most evident effects of drought is a reduction in plant height. In environments where both grain and straw have a value to the farmers as animal feed, plant height is an important trait as it is related to biological yield. Also in dry years the crop is too short to be harvested by combine and farmers either leave it for grazing or harvest it by hand at much higher cost.

Barley landraces In 1981 a large collection of barley landraces was made by E. Weltzien in Syria and Jordan: from the elds of 70 farmers (60 in Syria and 10 in Jordan), who had been using their own seed for generations, 100 spikes were collected at random (Weltzien, 1988). The spikes were kept separate, unlike what most conventional collectors, such as germplasm specialists, often do. This was a key in the subsequent utilization of the collection.

356

S. Grando et al.

Table 22.2. Mean and range of seven characters measured in 55 accessions of H. spontaneum and in nine cultivars of H. vulgare. Charactera Cold damage Early growth vigour Days to heading Peduncle length Peduncle extrusion Straw yield (t ha 1) Straw protein (%) H. spontaneum Mean Range 3.0 2.9 141.8 44.2 18.4 4.4 5.1 1.05.0 1.05.0 131.5152.5 32.850.3 8.226.3 2.86.5 2.77.0 H. vulgare Mean 3.7 1.6 141.1 23.5 4.4 5.1 3.5 Range 1.55.0 1.02.5 138.5145.5 20.332.1 1.010.5 4.16.1 2.84.7

aCold damage (1 = resistant, 5 = susceptible); early growth vigour (1 = good, 5 = poor); days to heading (days from emergence to awn appearance); peduncle length (distance from last node to base of spike, in cm); peduncle extrusion (distance from collar of ag leaf to base of spike, in cm).

When the seed collected was multiplied off-season (planting in summer) as headrows, two main characteristics were noted: rst, a high degree of seed dormancy was observed, with the material collected from the south showing a higher percentage of germination; secondly, few of the rows were able to head and produce seed, with differences in the material collected at the same sites (Weltzien, 1982). These were the rst lessons the landraces were teaching, both by indicating traits of adaptive signicance (such as vernalization requirement and seed dormancy), and by expressing the variability that these populations harbour. Additional information on the structure of the variation between and within collection sites was obtained when the material was evaluated under eld conditions as individual rows (Weltzien, 1988, 1989) or as plots (Weltzien and Fischbeck, 1990). Signicant genetic variation was found for seed colour, growth habit, awn barbing, days to heading, culm length, leaf width, awn length, early growth vigour, lodging and powdery mildew resistance. Three important indications, which were conrmed later, emerged from these studies. Firstly, that the genetic variability within the landraces was also expressed in stress sites, where the heritability was even higher than in a non-stress site; secondly, in a stress site the majority of landraces outyielded the check (improved) cultivars; thirdly, in the non-stress site the checks outyielded the landraces but not always signicantly. Arabi Abiad and Arabi Aswad the two barley landraces widely grown in Syria show a large and signicant variability for virtually all characters (growth habit, early growth vigour, cold damage, plant height, days to heading, days to maturity, grain-lling duration, grain yield, spike length, peduncle extrusion, 1000 kernel weight, protein content and lysine content). This is based on the evaluation in 1984/85 of 420 single-head progenies (lines) in three trials at Breda (Ceccarelli et al., 1987). In each of the rst two trials, 140 lines were evaluated (10 for each of 14 collection sites) together with four checks (Arabi Abiad and Arabi Aswad, and two improved cultivars Harmal and Rihane-03). In the third trial, 70 lines for each of two collection sites were evaluated. The data quantied some of the differences between the white-seeded and the black-seeded landraces. These are of particular interest to plant breeders to verify the

Genetic Diversity of Barley

357

rm belief of Syrian farmers that the black-seeded landrace is better adapted to dry areas and it is a better feed for sheep than the white-seeded landrace. The use of lines with the required seed colour could be important to ensure quick adoption. The differences between the two landraces are summarized in Table 22.3 using the data of the third experiment where number and length of seminal roots these are the only roots produced in very dry conditions and coleoptile length were measured. The black-seeded landrace usually has less-vigorous early growth, is more cold tolerant, matures slightly earlier, has a shorter grain-lling period, is taller under drought, is more productive under stress, has smaller kernels, a shorter coleoptile length, and shorter and fewer seminal roots. Some of these differences such as those related to phenology, and cold tolerance, growth vigour, kernel size and plant height are easily interpretable as related to adaptation to dry and cold areas where the black-seeded landrace is predominantly cultivated. The advantage of plant height under drought is often quoted by farmers in Syria as one of the main reasons for preferring Arabi Aswad to Arabi Abiad in the drier areas. The most interesting aspect of this early work was the extraordinary amount of variability found within landraces as shown by the intervals of variation in Table 22.3. The fact that landraces are composed of several genotypes is neither new nor original. In fact this has been reported for several crops, such as lentil (Erskine and Choudhary, 1986), sorghum (Blum et al., 1991), bread and durum wheat (Damania and Porceddu, 1983; Porceddu and Scarascia Mugnozza, 1984; Spagnoletti-Zeuli et al., 1984; Damania et al., 1985; Lagudah et al., 1987; Blum et al., 1989; Elings and Nachit, 1991), beans (Martin and Adams, 1987a, b), both cultivated and wild barley (Brown, 1978, 1979; Asfaw, 1989) and others. In addition to grain yield and other agronomic, morphological and physiological characters, an unexpected amount of variability was found for resistance to yellow rust,
Table 22.3. Differences between the black-seeded (Arabi Aswad) and the white-seeded (Arabi Abiad) barley landraces commonly grown in Syria. Charactera Cold damage Early growth vigour Days to heading Days to maturity Grain-lling duration Plant height (cm) Grain yield (kg ha 1) Grain protein content (%) Grain lysine content (%) 1000 kernel weight (g) Seminal root number Seminal root length (mm) Coleoptile length (mm)
aCold

Arabi Aswad (black) Mean SE Range 2.10 0.06 2.99 0.09 147.4 0.22 171.9 0.30 24.5 0.26 52.1 0.47 1769 36 10.5 0.05 0.43 0.00 35.7 0.29 5.7 0.06 55.8 1.23 47.5 0.44 3.081.02 4.651.52 153.0141.5 177.6168.8 30.519.50 61.840.9 2480944 11.69.7 0.450.41 43.531.1 7.14.4 86.337.16 55.439.4

Arabi Abiad (white) Mean SE Range 3.26 0.08 2.31 0.10 147.7 0.14 173.8 0.28 26.1 0.25 43.1 0.46 1542 40 10.6 0.06 0.43 0.00 41.9 0.35 6.2 0.05 69.1 1.11 52.4 0.55 4.681.58 5.001.06 150.4145.3 178.5168.8 30.520.5 53.433.4 2324920 11.99.9 0.460.40 47.934.6 7.45.0 99.343.3 61.441.4

damage (1 = resistant; 5 = susceptible); early growth vigour (1 = good; 5 = poor); days to heading (days from emergence to awn appearance); days to maturity (days from emergence to yellowing of entire plant); grain-lling duration (days between heading and maturity); plant height (measured at maturity, distance from base of plant to top of spike, excluding awns).

358

S. Grando et al.

powdery mildew, scald and covered smut (van Leur et al., 1989). With the exception of covered smut, there was a signicant variation both between and within collection sites. The reaction to diseases varied from absolutely or partially resistant types to highly susceptible lines. This was of particular interest because the common belief is that landraces are disease susceptible, and therefore not worth the attention of modern plant breeders. The data indicated, however, that although landraces look disease susceptible because the majority of plants are susceptible, they contain a small amount of resistant individuals which are an important source of genes for disease resistance within an adapted genetic background. In the case of Syrian landraces, the presence of such a high level of heterogeneity is not as obvious at rst sight as it is, for example, in Ethiopian or Nepalese barley landraces. This might explain why Syrian farmers do not select within landraces either before or after harvesting. One might have hypothesized that thousand of years of natural and human selection in a stress environment could have reduced the amount of heterogeneity by continuously selecting for the most adapted genotypes. However, this does not seem to be the case and the variation available within populations appears to be large and of great value to a breeding programme for stress environments and low-input conditions. This is evident by looking at the yield advantage of some of the pure lines extracted from landraces over the original landraces and some improved (modern) cultivars (Table 22.4) in low rainfall conditions and with little or no use of inputs. These data have been conrmed in many comparisons between different types of germplasm in that type of environment and explain the failure of introducing modern cultivars
Table 22.4. The highest-yielding pure lines extracted from landraces in 1984/85 in Breda (277 mm rainfall), compared with the two commonly grown landraces and two improved cultivars. Entry SLB 45-48 SLB 39-31 SLB 39-58 (Arta) SLB 45-83 SLB 45-95 SLB 45-40 SLB 39-05 SLB 45-04 SLB 39-10 SLB 45-90 SLB 45-34 SLB 45-76 Checks Arabi Abiad (landrace) Arabi Aswad (landrace) Rihane-03 (modern) Harmal (modern) LSD0.05
LSD, least signicant differerence.

Seed colour Black White White Black Black Black White Black White Black Black Black

Plant height (cm) 50.8 42.4 45.1 55.9 53.7 61.5 45.3 55.2 45.0 61.8 53.1 53.6 45.4 47.7 49.4 45.9 5.4

Grain yield (kg ha 1) 2480 2324 2287 2232 2227 2216 2189 2180 2162 2153 2146 2122 1666 1547 1013 1017 453

% increase over Arabi Abiad 149 139 137 134 134 133 131 131 130 129 129 127

Genetic Diversity of Barley

359

into the area. The most important information from a breeding point of view is about the amount of improvement it is possible to achieve by simply utilizing the variability available within landraces. The presence of large genetic diversity within populations adapted to an environment where conventional breeding has failed suggested that, in addition to the need for continuous collection and both ex situ and in situ conservation, there was the almost unexplored possibility of using this large reservoir of useful genetic variation in these landraces for plant improvement. There were four ways to do that: 1. Develop highest-yielding pure lines extracted from landraces into pure line varieties. 2. Utilize pure lines extracted from landraces that are superior for yield, as well as for other characters including resistance to pests and diseases and quality characteristics, as parents in the crossing programme to introduce additional desirable characters in an adapted genetic background. 3. Develop mixtures or multi-line varieties, constructed with a variable number of pure lines properly characterized for a set of agronomic characters. This permits exploitation of the buffering capacity of genetically heterogeneous populations in relation to stability and will conserve a certain amount of the evolutionary process within populations. 4. Evaluate lines with contrasting expressions of specic attributes to quantify the adaptive role of specic morphological traits in stress environments, and to identify, localize and tag gene complexes or quantitive trait loci (QTLs) associated with adaptation with molecular techniques. The four avenues were initiated within a few years of each other, with the exception of the molecular approach, which has started only recently. The remainder of this chapter describes the results for the rst three approaches. These activities had two main objectives. On one hand, they had the objective of generating new cultivars for the dry areas of Syria; on the other hand, they were intended to generate a methodology of utilization of landraces to be used in those countries where landraces, even of other crops, are still available. With this second objective in mind, the methods of exploiting the genetic variation between and within landraces were kept as simple as possible in view of their utilization by breeders in developing countries with limited resources. Another key aspect of the methodology was to employ the level of inputs used by farmers to make sure that the products (pure lines and mixtures) would be benecial to poor farmers and that the yield increases could be sustained.

Landraces and H. spontaneum as Breeding Material

Pure line selection within landraces Since 1984/85 the collection of 7000 spikes described earlier has been systematically evaluated. This has been done by testing between 300 and 400 lines each year, with a classical pure-line selection method (the oldest and simplest of breeding methods), under typical farmers conditions. Farmers were invited to visit the plots and to make their own selection: their selection criteria (tall plants under drought and soft straw) were subsequently used as routine.

360

S. Grando et al.

After 10 years, and with about half of the collection evaluated, three pure lines (two with black seed, named Tadmor and Zanbaka, and one with white seed, named Arta the only one ofcially released) are already grown in farmers elds each on an area estimated as between 500 and 2000 ha. Before 1981, Tadmor, Zanbaka and Arta were three spikes, among millions, in three farmers elds. In 2000, the progeny of these three spikes grow in farmers elds and outyield the local landraces by between 10 and 25% without additional inputs. Figure 22.2 shows an example of the yield advantage that can be obtained in farmers elds with this strategy. Arta was compared with the local landrace (either Arabi Abiad or Arabi Aswad, depending on the location) in 69 farmers elds in ve provinces of Syria. The locations have been ranked in ascending order using the yield of the local landrace. The superiority of Arta is larger at low-yield levels than at higheryield levels: in all the 23 lowest-yielding locations Arta always outyielded the local landrace (only in one case were the yields similar), and therefore Arta is mostly benecial to farmers in difcult environments. Interestingly enough, as one of the rst few lines to be yield tested, Arta was already showing its superiority in 1984/85 (see Table 22.4; the collection number of Arta was SLB 39-58). The evaluation of the collection of landraces has not only produced three varieties rapidly spreading from farmer to farmer. During the past 10 years we also identied, within the landraces, sources of resistance to most of the major barley diseases such as powdery mildew, scald, yellow rust, covered smut, barley stripe and root rot, which have been selected to be used as parental stocks. The collection of pure lines is evaluated within the context of the breeding programme. Therefore, it has been possible during the years to make several comparisons between the landraces of Syria and Jordan and modern cultivars. A typical example of this type of comparison is given in Table 22.5, where 77 lines
5000

4000 Grain yield (kg ha1)

3000

2000

1000

0
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69

Locations Local

ARTA

Fig. 22.2. Grain yield of Arta compared with local barley in 69 farmers elds in ve provinces of Syria in 1996. Each cultivar was grown on plots of 1 ha.

Genetic Diversity of Barley

361

Table 22.5. Grain yield (kg ha 1) under stress and under non-stress of barley landraces and modern cultivars in Syria. Stressa Type of germplasm Modern (n = 155) Landracesc (n = 77) Best check
aAverage bAverage

Non-stressb Range Yield 3901 3413 4147 Range 23104981 23984610

Yield 488 788 717

0893 4861076

of two stress sites. of three non-stress sites. cPure lines obtained by pure-line selection within landraces.

from landraces were compared with modern cultivars using the average grain yield of two stress sites and the average grain yield of three non-stress sites. The landraces have an average yield advantage of 60% under stress while the modern cultivars have an average yield advantage of 14% under non-stress. Besides the mean performance of the two types of germplasm, the interval of variation is also very informative. All 77 lines yielded something under stress, while some of the modern failed even though the best modern yielded almost much as the best landraces. Under non-stress it was interesting to nd that the yield of some landraces was not much less than that of the best modern cultivars. The lower yield potential of the lines from landraces, specically selected for stress conditions, is not a serious problem because in the dry areas of Syria, the probability of yields exceeding 3 t ha 1 is about six times lower than the probability of yields less than 1.5 t ha 1 (Ceccarelli, 1996). One of the most important messages of Table 22.5 is about the choice of the selection environment. It is clear from the two examples that, had the selection been done only under the high-yielding conditions of a typically high-input experiment station, the landraces would have had a short life as breeding material (see Ceccarelli et al., Chapter 6, this volume). Today, a similar approach has begun in Ethiopia and Eritrea with barley, work on barley landraces is very active in Tunisia, and some work is being done in Iraq. New collections of barley landraces have recently been made in Nepal to begin a similar type of programme.

Crosses: building on adaptation The identication of agronomically superior pure lines within landraces, and of sources of disease resistance, opened a second avenue in exploiting adapted germplasm, i.e. their use as parental material in a breeding programme. This started about 3 years after the systematic evaluation of pure lines began (the rst crosses with H. spontaneum were made in 1987). One example of the value of using landraces in a crossing programme for dry areas is given in Table 22.6, where 514 breeding lines unrelated to landraces (improved) and 525 pure lines extracted from landraces are compared with lines derived from three

362

S. Grando et al.

Table 22.6. Grain yield (kg ha 1), biological yield (kg ha 1), plant height (cm) and harvest index in Breda and grain yield (kg ha 1) in Tel Hadya (both in 1995) of different types of breeding material. Breeding material Improved (n = 514) Mean Max Min Improved (n = 214) Mean Max Min landraces 3883 33 5206 2630 3657 23 5455 2250 775 10 1252 259 752 7 1232 320 2678 24 3658 1930 2549 16 4027 1529 25.1 0.3 38 16.9 21.4 0.1 30.5 13.1 29.1 0.3 37.9 11.6 29.8 0.2 39.9 16.5 Grain yield (TH95) 4125 27 5812 1375 Grain yield (BR95) 591 8 1201 69 Biological yield (BR95) 2559 17 4504 1559 Plant height (BR95) 23.2 0.2 40.3 14.8 Harvest index (BR95) 22.8 0.3 41.3 3.24

Landraces (n = 525) Mean Max Min Landraces Hordeum spontaneum (n = 133) Mean Max Min Improved Hordeum spontaneum (n = 17) Mean Max Min

2797 49 4489 1515

724 11 1077 369

2829 32 4007 2060

29.1 0.4 43.6 20.5

25.9 0.3 35.6 11.5

2814 118 3995 1780

537 37 907 306

2362 111 3681 1842

27.1 1.7 44.1 19.2

20.6 1.2 30.6 11.4

BR95 = Breda 1995; TH95 = Tel Hadya 1995.

types of crosses. The comparison was done in a very dry site and year (Breda in received 244 mm rainfall 1995) where grain yield, total biological yield, plant height and harvest index were measured, and in a relatively wet site (Tel Hadya with 313 mm rainfall) yield potential was measured. The landraces yielded, on average, more than the improved lines under stress but had a lower average yield potential. Under stress, landraces and improved lines had a similar biological yield, but the landraces were much shorter and had a higher harvest index two characteristics usually associated with high-yielding varieties. Crosses between landraces and improved germplasm generated breeding material that was as good as landraces for grain yield and total biological yield under stress, and superior for plant height while maintaining a relatively high harvest index. However, crosses between landraces and H. spontaneum generated breeding material which is almost as good as that derived from crosses between landraces and improved germplasm. The total biological yield and plant height are higher than in any other material, and both grain yield under stress and harvest index are probably underestimated because of the presence of some brittle-rachis genotypes. The last type of cross

Genetic Diversity of Barley

363

improved varieties H. spontaneum generated the least promising type of breeding material, except perhaps for plant height (Table 22.6). The superiority of the crosses with landraces suggests that the strategy of using adapted germplasm in a breeding programme is to capitalize on their specic adaptation to drought and low-input conditions rather than to consider them as sources of a few useful genes as done in most plant breeding programmes. Therefore, in breeding for stress environments, landraces should be regarded as the recipient of a few useful genes to be added to their adapted genetic background, rather than as donors of traits not available in elite germplasm. This is conceptually similar to what breeders in favourable environments do: once they have found genotypes with high yield potential and good adaptation to high-yielding conditions, they continue to build on them. The strategy is strengthened by the availability of genes for disease resistance within landraces. If a line extracted from landraces is agronomically superior but susceptible to a disease, the source of resistance is rst sought among lines from the same collection site to preserve as much adaptation as possible, and secondly sought among lines from neighbouring collection sites. Sources of disease resistance from germplasm adapted to different environments are the last resource. It is for these reasons that the best germplasm pool for the Fertile Crescent is now derived from crosses involving lines extracted from landraces. One type of cross landraces H. spontaneum has been the most promising avenue to improve plant height under drought: this characteristic, together with straw softness, is very often indicated by farmers as one of the most desirable traits, particularly in dry areas. As mentioned earlier, a crop that remains tall even in very dry years is important to farmers, because it reduces their dependence on costly hand-harvesting, while soft straw is considered important in relation to palatability. Table 22.7 shows the tallest of 1532 most recently developed breeding lines tested in 1994/95 at Breda.

Table 22.7. Plant height at Breda (244 mm rainfall) in 1995 of barley lines derived from crosses with Hordeum spontaneum, compared with the barley landrace most common in dry areas (Arabi Aswad) and with a cultivar selected specically for plant height under drought (Zanbaka). Cross/name Hordeum spontaneum 20-4/Arar 28//WI2291/Bgs SLB 45-40/Hordeum spontaneum 41-1 Zanbaka/Hordeum spontaneum 41-2 Zanbaka/Hordeum spontaneum 41-2 Moroc 9-75/Arabi Aswad//Hordeum spontaneum 41-3 Arabi Aswad Zanbaka Mean of all breeding lines Maximum Minimum LSD0.05 Plant height (cm) 43.5 43.0 42.5 41.5 41.0 24.8 26.0 23.5 43.5 12.5 5.6

364

S. Grando et al.

While the mean plant height of all the lines was 23.5 cm, the shortest lines were only 12.5 cm tall, and the most widely cultivated landrace (Arabi Aswad) was about 25 cm, some of the lines derived from crosses with H. spontaneum were taller than 40 cm. They were also signicantly taller than Zanbaka, a pure line selected from Arabi Aswad and already grown by some farmers for its plant height. It is obvious that these two characteristics tall plants and soft straw represent a drastic departure from the typical selection criteria used in breeding high-yielding cereal crops short plants with stiff straw and high harvest index. Cultivars possessing the two characteristics considered important by farmers in dry areas would be unsuitable for high-yielding environments because of their lodging susceptibility, and in a traditional breeding programme will not be made available to farmers a further indication of the importance of specic adaptation.

Mixtures Although successful, pure-line selection within landraces is potentially dangerous because it tends to replace genetically heterogeneous populations with genetically pure lines. The adoption by Syrian farmers of three different pure lines almost at the same time and in a relatively small geographical area some farmers even adopted two different cultivars at the same time suggests that the danger may be less dramatic than what happened with the spreading of single genotypes over very large areas as in the case of HY varieties. There is also evidence that in marginal environments replacement of landraces is often only partial (Brush, 1995). But, in principle, genetic uniformity is in contrast with the genetic diversity that characterizes the agricultural systems of difcult environments and poor farmers. In these systems diversity is preserved at one or more levels by using different crops in the same farm, by using different cultivars of the same crop, and by using heterogeneous cultivars. Diversity reduces the risk of crop failures due to abiotic and biotic stresses, while monoculture of a single genotype maximizes it. One wonders why millennia of natural selection operating in harsh environments on a crop such as barley in the Fertile Crescent have left us with heterogeneous populations rather than with a single or few genotypes with superior adaptation. Is it perhaps yet another lesson landraces are imparting: that it is the structure of the population, in addition to the genetic constitution of the components it harbours, which is the secret of adaptation to difcult and unpredictable environments. Therefore, constructing mixtures with a number of superior yet genetically different pure lines selected from landraces is the long-term objective of using landraces in the breeding programme. This would add the advantage of a population-buffering mechanism to the adaptation of the individual components (Simmonds, 1962; Grando and McGee, 1990; Lenn and Smithson, 1994). This is an additional way, perhaps more time-consuming and experimentally more complex than the rst two, to respond to the need of poor farmers for stable yields. Therefore, in the last 10 years of the 20th century we conducted trials with mixtures of variable numbers of superior, yet genetically different, pure lines selected from landraces to compare yield and yield stability of pure lines and of mixtures. The results have shown the superiority of some specic mixtures but also that some pure lines have similar yield and stability to mixtures.

Genetic Diversity of Barley

365

In the most recent of these trials, mixtures and pure lines within the two barley landraces, Arabi Abiad and Arabi Aswad, were compared. Mixtures with 72, 34, 17 and 5 components in the black-seeded group and 75, 34, 15 and 5 components in the white-seeded group were used. The constituent lines were either unselected (the more complex mixture) or derived from one, two or three cycles of selection. The material was evaluated from 1990/91 to 1994/95 in 22 environments with mean yields ranging from 614 to 4385 kg ha 1. Black-seeded material tends to have lower average grain yield, lower response and higher frequency of positive intercepts than white-seeded material. In both groups the mixtures with ve components had an advantage over the more complex mixtures. In the black-seeded group (Table 22.8), the mixture of ve lines did not have a clear advantage over the components. In particular, SLB 5-96 had a high average yield (2164 kg ha 1), combined with a relatively good response (b = 0.97) and a positive intercept (a = 99.9). In the white-seeded group (Table 22.9) the mixture of ve components had an advantage over the single lines, combining a high average grain yield (2263 kg ha 1), with a good response (b = 1.05) and positive intercept (a = 32.9). The only other line with a positive intercept (SLB 9-98) had a very low average grain yield (1833 kg ha 1) and low response (b = 0.79). The results indicate the possibility that the two Syrian barley landraces possess different buffering mechanisms. The advantage of both ve-component mixtures over the more heterogeneous mixtures would suggest that yield stability may be achieved with a modest degree of heterogeneity.

Table 22.8 Average grain yield (kg ha 1), regression coefcient (b) and intercept (a) of four black-seeded mixtures, ve lines and three checks. Material Mixtures MIX 72 MIX 34 MIX 17 MIX 5 Pure lines SLB 5-96 SLB 5-07 SLB 5-86 SLB 5-31 SLB 5-30 Checks Arabi Aswad Tadmor Zanbaka LSD0.05 Grain yield 2017 2060 2076 2131 2164 2179 1950 2266 1982 1896 1971 1946 164 b 0.88 0.96 0.97 0.93 0.97 0.98 0.84 1.05 0.86 0.83 0.86 0.84 a 147.4 10.2 11.8 150.1 99.9 97.6 167.0 35.7 144.2 116.7 140.0 154.9

366

S. Grando et al. Table 22.9. Average grain yield (kg ha 1), regression coefcient (b) and intercept (a) of four white-seeded mixtures, ve lines, and three checks. Material Mixtures MIX 75 MIX 34 MIX 15 MIX 5 Pure lines SLB 9-63 SLB 9-71 SLB 9-76 SLB 9-09 SLB 9-98 Checks Arabi Abiad Arta Harmal LSD0.05 Grain yield 2237 2209 2174 2263 2288 2302 2388 2328 1833 2202 2414 2204 164 b 1.15 1.17 1.08 1.05 1.14 1.15 1.24 1.13 0.79 1.14 1.19 1.15 a 226.7 277.1 139.1 32.9 144.4 152.3 248.5 86.8 146.1 222.9 117.9 248.8

Understanding adaptation to stress Landraces have also been extremely useful for understanding the adaptive role of given traits. For example, barley lines extracted from landraces collected from ve sites in the Syrian steppe (Table 22.10), when compared with barley lines extracted from landraces collected in Jordan and with a wide range of modern barley genotypes, showed a higher frequency of genotypes with prostrate or semi-prostrate growth habit, cold tolerance and short grain-lling period, and a lower frequency of genotypes with good growth vigour and early heading. Their average grain yield in unfavourable conditions (Bouider, 1989) was 984 kg ha 1 (ranging from 581 to 1394 kg ha 1), more than twice the average grain yield of modern genotypes (483 kg ha 1 ranging from crop failure to 1193 kg ha 1). The average yield in favourable conditions of the Syrian landraces (3293 kg ha 1) was 75% of the average yield in favourable conditions of the modern germplasm (4398 kg ha 1). Although this particular set of data is based on one environment only, it conrms the existence of a trade-off between yield in unfavourable conditions and yield in favourable conditions (Ceccarelli, 1989) found in other sets of data based on more environments. Landraces collected in Jordan, from sites with milder winters than the Syrian steppe, have a higher frequency of genotypes with better growth vigour, more erect habit, less cold tolerance, slightly longer grain-lling period and earlier heading than Syrian landraces. Their average grain yield in unfavourable conditions was only slightly lower (835 kg ha 1) than Syrian landraces, while their average yield in favourable conditions (3947 kg ha 1) was in between that of the Syrian landraces and

Genetic Diversity of Barley Table 22.10. Mean of morphological and developmental traits in 1041 modern (unrelated to Syrian or Jordanian landraces) barley genotypes compared with 322 pure lines extracted from Syrian landraces and 232 pure lines from Jordanian landraces. Landraces Traits1 Early growth vigour Growth habit Cold damage Days to heading Grain-lling duration Grain yield under non-stress Grain yield under stress
1Early

367

Modern (n = 1041) 2.5 b2 2.8 c 3.0 a 117.9 b 39.3 a 4398.0 a 483.1 c

Syria (n = 322) 3.2 a 4.0 a 1.3 c 121.2 a 35.5 c 3293.0 c 984.0 a

Jordan (n = 232) 2.4 b 3.1 b 2.3 b 116.9 c 37.4 b 3947.0 b 834.7 b

growth vigour (1 = good, 5=poor); growth habit (1 = erect, 5 = prostrate); days to heading (days from emergence to awn appearance); grain-lling duration (days between heading and maturity) and grain yield under non-stress (kg ha 1) were scored or measured at Tel Hadya in 1987/88 (504.2 mm rainfall); cold damage (1 = resistant, 5 = susceptible) was scored at Bouider in 1987/88 (385.7 mm rainfall); grain yield under stress (kg ha 1) was measured at Bouider in 1988/89 (189 mm rainfall) on 521 modern lines, 92 Syrian landraces and 86 Jordanian landraces. 2 Means followed by the same letter are not signicantly (P 0.05) different based on ttest for samples of unequal size.

the modern germplasm. The highest yield under stress of Syrian landraces is not due to an escape mechanism, as they are the latest group in heading, and therefore could be a combination of resistance (or tolerance) and avoidance (prostrate habit and cold tolerance result in good ground cover) mechanisms. Landraces are not only variable for above-ground characteristics. A recent study (Table 22.11) shows that considerable variation exists for both the number and the length of seminal roots (Grando and Ceccarelli, 1995) between different germplasm types. Seminal roots are very important because in dry years they represent the only roots the plant produces. It appears that during the domestication of barley the number of seminal roots has evolved from about three in H. spontaneum to ve to seven in

Table 22.11. Mean and range of variation for number of seminal roots and their maximum length at Zadoks stage 10 in three groups of barley germplasm. Number Germplasm group Modern Landraces Hordeum spontaneum LSD0.05
aLSD 0.05 bLSD 0.05

Length Range 4.6 6.1 4.4 5.9 3.0 3.8 0.4b Mean 96.5 118.8 107.3 14.9 Range 70.8 115.3 107.4 131.6 97.2 118.1 11.6

Mean 5.5 5.1 3.3 0.7a

for group means comparison. for entry means comparison.

368

S. Grando et al.

cultivated forms, while there has been a reduction in early root growth (root length) in modern varieties. In addition, the data show that also for below-ground characteristics probably important in relation to the use of water: one of the most limiting resources there is considerable variability within landraces. Therefore, the advantages of heterogeneity are probably also valid underground. The genetic structure of landraces then, may be considered as an evolutionary approach to survival and performance under arid and semi-arid conditions (Schulze, 1988). It follows that, during millennia of cultivation under adverse conditions, natural and articial selection have not been able to identify either an individual genotype possessing a key trait associated with its superior performance, or an individual genotype with a specic architecture of different traits. On the contrary, the combined effects of natural and articial selection have led to an architecture of genotypes representing different combinations of traits. This is shown more clearly in Table 22.12 where 321 lines derived from Syrian landraces collected in the steppe were classied according to the score for early growth vigour in three classes: good vigour (score 2.5), intermediate (score = 2.53.5) and poor vigour (score 3.5). Each class was then classied according to the score for growth habit (erect 2.5; semiprostrate = 2.53.5; prostrate 3.5). No genotypes were found in the good vigourerect, intermediate vigourerect, poor vigourerect, and poor vigoursemiprostrate classes. The groups were compared not only for the two traits used in their classication, but also for days to heading, cold tolerance and length of the grain-lling period. Lines with good early growth vigour tend to be less cold tolerant, earlier and with a longer grain-lling period. This small percentage of genotypes presumably will have a yield advantage in years with slightly milder winter temperatures, absence of late frosts and less severe terminal stress. The highest frequency of genotypes (71.3%) combines an intermediate early growth vigour with semiprostrate or prostrate growth habit. These genotypes are slightly more cold tolerant than the rst group, but are slightly later in heading. However, they are better
Table 22.12. Frequency of different combinations of early growth vigour (GV) and growth habit (GH), and mean values of cold tolerance (CT), days to heading (DH) and length of the grain-lling period (GF) in a sample of 322 lines of barley collected in the dry areas of Syriaa (traits as indicated in footnote of Table 22.10). Groups Good vigour erect Good vigour semiprostrate Good vigour prostrate Intermediate vigour erect Intermediate vigour semiprostrate Intermediate vigour prostrate Poor vigour erect Poor vigour semiprostrate Poor vigour prostrate LSD0.05
aCollection

% 0 1.2 5.3 0 6.2 65.1 0 0 22.1

GV 2.2 2.4 2.9 3.1 3.9 0.2

GH 3.3 3.9 3.4 4 4.2 0.1

CT 1.6 1.4 1.5 1.4 1.3 0.1

DH 118.8 119.8 119.7 121.2 121.9 0.6

GF 37.4 36.6 35.8 35.4 35.4 0.7

sites are all included in the Palmyra region as dened by Weltzien (1988). LSD, least signicant difference.

Genetic Diversity of Barley

369

equipped to escape terminal drought because of the shorter grain-lling period. About one-quarter of the genotypes (22.1%) have a poor early growth vigour but a very prostrate growth habit (GH = 4.2) and a high level of cold tolerance (1.3). Their slightly (although signicant) later heading is not necessarily a negative attribute, mostly because it is compensated by a very short grain-lling period. In the reference environment, a population with such an architecture of genotypes is probably the best solution to long-term stability.

Conclusions
The use of barley landraces from a section of the Fertile Crescent in the barley breeding programme at ICARDA contains a number of messages, potentially of general relevance. Landraces should be considered not only as genetic resources to conserve for the needs of tomorrow, but as breeding material to be used today, particularly in breeding programmes for stress environments and for poor farmers. Landraces are adapted to levels of inputs farmers can afford, yet are variable. So, they can be improved without requiring additional inputs. Landraces are adapted to their environment; they t in the farming systems of their area of adaptation; they are often essential in the diet; in many cases they are the only food or feed available: the welfare of people depending on landraces should and can be improved not by replacing landraces but by improving them. We have shown that this is possible provided that a number of conventional concepts in plant breeding are challenged, not least the concept of wide adaptation in a broad sense. The advantage of selecting for specic adaptation in space is the production of different varieties in different countries as well as different varieties, not necessarily homogenous, in different environments within a country, thus contributing to conserving biodiversity. Eventually, testing and evaluating landraces under the conditions of poor farmers will ensure that no additional inputs are required to benet from the new varieties. The participation of farmers during selection is essential to identify meaningful selection criteria and, later, to maximize the probability of adoption. The story of utilizing the genetic variability within a collection of landraces made in 1981 in Syria and Jordan is a success story. Yet the variability used is only what was sampled at a given moment of the evolutionary processes taking place within the landraces; the benets that variability has generated give a measure of the importance of keeping those processes alive through in situ conservation. Keeping the genes of landraces alive in breeding programmes is a moral obligation towards those many farmers who maintained landraces available for future generations.

Acknowledgements
The Government of Italy has supported the work on barley landraces and on H. spontaneum in Syria conducted after 1987.

370

S. Grando et al.

References
Asfaw, Z. (1989) Variation in hordein polypeptide pattern within Ethiopian barley, Hordeum vulgare L. (Poaceae). Hereditas 110, 185191. Asfaw, Z. and Bothmer, R. von (1990) Hybridization between landrace varieties in Ethiopian barley (H. vulgare ssp. vulgare) and the progenitor of barley (H. vulgare ssp. spontaneum). Hereditas 112, 5764. Blum, A., Golan, G., Mayer, J., Sinmena, B., Shpiler, L. and Burra, J. (1989) The drought response of landraces of wheat from the northern Negev Desert in Israel. Euphytica 43, 8796. Blum, A., Golan, G. and Mayer, J. (1991) Progress achieved by breeding open-pollinated cultivars as compared with landraces of sorghum. Journal of Agricultural Science, Cambridge 117, 307312. Brown, A.H.D. (1978) Isozymes, plant population genetic structure, and genetic conservation. Theoretical and Applied Genetics 52, 145157. Brown, A.H.D. (1979) Enzyme polymorphism in plant populations. Theoretical and Population Biology 15, 142. Brown, A.H.D., Zohary, D. and Nevo, E. (1978) Outcrossing rates and heterozygosity in natural populations of Hordeum spontaneum Koch in Israel. Heredity 41, 4962. Brush, S.B. (1995) In situ conservation of landraces in centers of crop diversity. Crop Science 35, 346354. Ceccarelli, S. (1984) Utilization of landraces and H. spontaneum in barley breeding for dry areas. Rachis 3(2), 811. Ceccarelli, S. (1989) Wide adaptation: how wide? Euphytica 40, 197205. Ceccarelli, S. (1994) Specic adaptation and breeding for marginal conditions. Euphytica 77, 205219. Ceccarelli, S. (1996) Adaptation to low/high input cultivation. Euphytica 92, 203214. Ceccarelli, S. and Grando, S. (1987) Diversity for morphological and agronomic characters in Hordeum vulgare spp. spontaneum C. Koch. Genetica Agraria 41, 131142. Ceccarelli, S., Grando, S. and Leur, J.A.G. van (1987) Genetic diversity in barley landraces from Syria and Jordan. Euphytica 36, 389405. Damania, A. and Porceddu, E. (1983) Variation in landraces of turgidum and bread wheats and sampling strategies for collecting wheat genetic resources. Proceedings of the 6th International Wheat Genetic Symposium, pp. 123136. Damania, A.B., Jackson, M.T. and Porceddu, E. (1985) Variation in wheat and barley landraces from Nepal and the Yemen Arab Republic. Zeitschrift Panzenzuchtung 94, 1324. Elings, A. and Nachit, M.M. (1991) Durum wheat landraces from Syria. I. Agro-ecological and morphological characterization. Euphytica 53, 211224. Erskine, W. and Choudhary, N.A. (1986) Variation between and within lentil landraces from Yemen Arab Republic. Euphytica 35, 695700. Grando, S. and Ceccarelli, S. (1995) Seminal root morphology and coleoptile length in wild (Hordeum vulgare ssp. spontaneum) and cultivated (Hordeum vulgare ssp. vulgare) barley. Euphytica 86, 7380. Grando, S. and McGee, R.J. (1990) Utilization of barley landraces in a breeding program. In: Biotic Stresses of Barley in Arid and Semi-arid Environments. Montana State University Press. Gustafsson, M. and Claesson, L. (1988) Resistance to powdery mildew in wild species of barley. Hereditas 108, 231237. Harlan, J.R. and Zohary, D. (1966) Distribution of wild wheats and barley. Science 153, 10741080. Lagudah, E.S., Flood, R.G. and Halloran, G.M. (1987) Variation in high molecular weight

Genetic Diversity of Barley

371

glutenin subunits in landraces of hexaploid wheat from Afghanistan. Euphytica 36, 39. Lehmann, L.C. and Bothmer, R. von (1988) Hordeum spontaneum and landraces as a gene resource for barley breeding. In: Jorna, M.L. and Slootmaker, L.A.J. (eds) Cereal Breeding Related to Integrated Cereal Production. Pudoc, Wageningen, pp. 190194. Lehmann, L.C., Jnsson, R. and Gustafsson, M. (1998) Identication of resistance genes to powdery mildew isolated from Hordeum vulgare ssp. spontaneum and landraces of barley. Journal of the Swedish Seed Association 108, 94101. Lenn, J.M. and Smithson, J.B. (1994) Varietal mixtures: a viable strategy for sustainable productivity in subsistence agriculture? Aspects of Applied Biology 39, 163172. Martin, G.B. and Adams, W.M. (1987a) Landraces of Phaseolus vulgaris (Fabaceae) in Northern Malawi. I. Regional variation. Economic Botany 41, 190203. Martin, G.B. and Adams, W. M. (1987b) Landraces of Phaseolus vulgaris (Fabaceae) in Northern Malawi. II. Generation and maintenance of variability. Economic Botany 41, 204215. Moseman, J.G., Nevo, E. and Zohary, D. (1983) Resistance of Hordeum spontaneum collected in Israel to infection with Erysiphe graminis hordei. Crop Science 23, 11151119. Moseman, J.G., Nevo, E. and El-Morsidy, M.A. (1990) Reactions of Hordeum spontaneum to infection with two cultures of Puccinia hordei from Israel and United States. Euphytica 49, 169175. Nevo, E. (1992) Origin, evolution, population genetics and resources for breeding of wild barley, Hordeum spontaneum, in the Fertile Crescent. In: Shewry, P.R. (ed.) Barley: Genetics, Biochemistry, Molecular Biology and Biotechnology. CAB International, Wallingford, pp. 1943. Porceddu, E. and Scarascia Mugnozza, G.T. (1984) Genetic variation in durum wheat. In: Proceedings of the 6th International Wheat Genetic Symposium, Kyoto, Japan, pp. 253264. Schulze, E.D. (1988) Adaptation mechanisms of non cultivated arid-zone plants: useful lesson for agriculture? In: Bidinger, F.R. and Johansen, C. (eds) Drought Research Priorities for the Dryland Tropics, ICRISAT, Patancheru, India, pp. 159177. Simmonds, N.W. (1962) Variability in crop plants, its use and conservation. Biological Reviews 37, 442465. Spagnoletti-Zeuli, P.L., De Pace, C. and Porceddu, E. (1984) Variation in durum wheat populations from three geographical origins. I. Material and spike characteristics. Euphytica 33, 563575. van Leur, J.A.G., Ceccarelli, S. and Grando, S. (1989) Diversity for disease resistance in barley landraces from Syria and Jordan. Plant Breeding 103, 324335. Vetelinen, M. (1997) Dynamic genepools of barley. Swedish University for Agricultural Science Agraria 36. von Bothmer, R., Jacobsen, N., Baden, C., Jrgensen, R.B. and Linde-Laursen, I. (1995) An Ecogeographical Study of the Genus Hordeum. Systematic and ecogeographic studies on crop genepools 7, IPGRI, Rome. Weltzien, E. (1982) Observation on the growth habit of Syrian and Jordanian landraces of barley. Rachis 1, 67. Weltzien, E. (1988) Evaluation of barley (Hordeum vulgare L.) landrace populations originating from different growing regions in the Near East. Plant Breeding 101, 95106. Weltzien, E. (1989) Differentiation among barley landrace populations from the Near East. Euphytica 43, 2939. Weltzien, E. and Fischbeck, G. (1990) Performance and variability of local barley landraces in near-eastern environments. Plant Breeding 104, 5867. Zohary, D. and Hopf, M. (1993) Domestication of Plants in the Old World. Oxford Science Publications, Oxford.

23
1Facult

Breeding Phaseolus for Intercrop Combinations in Andean Highlands

J.P. BAUDOIN1, F. CAMARENA2, M. LOBO3 AND G. MERGEAI1

Universitaire des Sciences Agronomiques de Gembloux, 2 Passage des Dports, 5030 Gembloux, Belgium; 2Universidad Nacional Agraria La Molina, Apdo 456, Lima, Peru; 3Corporacin Colombiana de Investigacin Agropecuaria, C.I. La Selva, Apartado Aereo 100 470, Rionegro, Colombia

Rationale or Impetus for Undertaking the Breeding Strategy

The common bean, Phaseolus vulgaris L., is by far the most important pulse crop in Latin America, its centre of origin and diversity. The species is the most valuable source of dietary protein, particularly among the low-income populations. Small farmers are the principal producers of beans, often as a secondary crop in association with cereals, root and tuber plants, and vegetables (Francis, 1986; Woolley et al., 1991). A high proportion of the seeds is directly consumed on the farm or traded only in local markets. Despite their nutritional importance, production growth rates have been declining in Latin America, with dry seed yields averaging less than 600 kg ha 1 (Woolley et al., 1991; Lobo, 1994). This is particularly true for the low-input intercropping systems of the Andean highlands, where beans are suffering from several constraints: high incidence of diseases and insect pests, soil mineral deciencies, cold or drought stresses, poor crop management, competition with associated crops and lack of improved seeds. Most of the landraces and improved varieties of the common bean are susceptible to one or more of these production constraints, preventing the realization of their full yield potential and causing production instability from one year to the next. One way to increase seed production of common bean in the Andean highlands is to develop a breeding process well adapted to the traditional multiple-cropping systems. In the small-scale subsistence farms of the regions, P. vulgaris and other Phaseolus cultigens are very often associated with maize, in small dispersed plots, mainly on sloping land of limited fertility and prone to soil erosion. In such marginal conditions, the association ensures reduced variability in total biomass and yield, arising mainly from compensatory effects among the crops, lower weed and erosion problems due to better soil cover, and reduced incidence of pests and diseases. Intercropping also gives a greater diversity of diet and a more stable source of income, particularly for farmers with limited land resources (Francis, 1986; Baudoin and Marechal, 1987; Baudoin and Camarena, 1994). However, when considering the benets of the association, we
IPGRI/FAO 2001. Broadening the Genetic Base of Crop Production (eds H.D. Cooper, C. Spillane and T. Hodgkin)

373

374

J.P. Baudoin et al.

should not underestimate one major obstacle: the substantial interspecic competition of the associated components in environments sometimes limited in natural resources (water, nutrients and light). This competition affects particularly understorey crops, such as the Phaseolus beans. In this study, we consider breeding Phaseolus beans for intercrop combinations in Andean regions, ranging from 2000 to 3600 m altitude. The data and discussions come mainly from collaborative projects at Gembloux Agricultural University (Belgium) developed with the Centro Internacional de Agricultura Tropical (CIAT), Colombia, the Corporacin Colombiana de Investigacin Agropecuaria (CORPOICA), Colombia, and the Universidad Nacional Agraria La Molina (UNALM), Peru. In the target regions of Colombia and Peru, two main cropping patterns prevail. In the rst pattern, found in areas with a single rainy season (Ancash, Cajamarca and Cusco in Peru; Cauca, Nario and Putumayo in Colombia), maize and beans are mainly planted simultaneously under mixed, row or hill intercropping, with beans climbing on the maize stems. In the second pattern, found in eastern Antioquia of Colombia under bimodal rainfall distribution, maize is planted in the rst rainy season while beans are broadcast into the maize during the second season at a period corresponding to the physiological maturity of the cereal. This system is called relay cropping, with 12 months overlap of maize and bean cycles. The stems of the strong-stalked highland maize offer a good support for the long and twisted branches of climbing beans. In the two patterns, bean genotypes are always of indeterminate growth habit, with a strong climbing ability and a long growth duration, requiring usually 710 months to reach the fruit ripening stage. This slow-maturing trait is very frequent in highland areas affected by low temperatures during the cultural season. The length of the bean growth cycle is similar to that found in local traditional maize varieties. In the cultural associations of the target regions, the cultivated P. vulgaris genotypes belong to a rather limited number of large-seeded classes, i.e. in Colombia: mainly Cargamento (with cream-beige speckled seed colour) and Mortio (with purple seed colour); in Peru: mainly Panamito and Caballero (with white seed colour), Canario and Bayo (with yellow seed colour) and Nuas (popping beans with striking colours and patterns). Whatever the type of cultivated genotypes and cropping systems, beans are very often considered as a high-risk crop, prone to many diseases and not receiving a good crop husbandry management by small farmers (such as use of improved seeds, regular weeding, organic matter application, chemical treatment against pest, etc.). We decided therefore to rank our objectives according to this situation. First priority was given to multiple disease resistance, a prerequisite for better response of beans to improved crop husbandry and food marketing. Secondly, priority was given to tting plant architecture and yield components into the frame of traditional intercrop combinations. To achieve these objectives, one key factor of our breeding strategy consisted of exploiting fully the interspecic diversity available in Phaseolus. Indeed, within the common bean primary genepool, insufcient genetic variation has been found to overcome several major production constraints (Baudoin et al., 1995). Past experiments conducted mainly in Colombia demonstrated the excellent potential of two Phaseolus species, of Mesoamerican and north Andean origin: P. coccineus L. and P. polyanthus Greenm. Both species are well adapted to Andean highlands (above 2000 m) and belong to the secondary genepool of P. vulgaris (Baudoin et al., 1992). The species

Breeding Phaseolus for Intercrop Combinations

375

show several desirable agronomic attributes, such as pest, insect and disease resistance, cold tolerance, lodging resistance due to thick stem bases, presence of tuberous or brous root systems allowing a perennial cycle, long epicotyls and racemes, and a large number of pods per inorescence (Schmit and Baudoin, 1987). The base germplasm collection is maintained at CIAT and represents genetically broad unimproved germplasm, found as a garden or backyard crop in small farms or as a weedy vine climbing on trees at the border of montane rain forest (Schmit and Debouck, 1991). In the genetic improvement of beans for multiple cropping systems, these two food legumes have a value in their own right and could be bred as distinct crop, particularly for higher and colder Andean regions. They can also be utilized in interspecic crosses with P. vulgaris, to introgress genes controlling useful traits not commonly present in the primary common bean genepool.

Methods and Strategy Used

Breeding approach The best way of achieving signicant progress in breeding for multiple cropping systems is to adopt a farming systems approach, involving diagnosis, on-farm experimentations and active farmer participation at an early stage (Steiner, 1990; Woolley and Davis, 1991; Sperling et al., 1993). Evaluation and breeding are carried out under cultural conditions representative of the traditional cropping systems but in conditions sufciently uniform to allow reasonably reliable selection. In this case, the experimental design should mimic the two prevailing cultural associations (relay cropping and simultaneous intercropping) at some stages of the breeding process. Although any genetic modication of one crop will have an effect on the other associated crop, it is more efcient to breed each species separately, the other partner being represented by a locally well-adapted cultivar. Indeed, Davis and Garcia (1983) demonstrated that when indeterminate climbing beans are intercropped with different maize cultivars, the interaction between the two associated crops is usually not signicant. It is therefore not worthwhile selecting both cereal and food legume simultaneously. In these investigations, priority was given to the understorey crop, the Phaseolus beans. Segregating hybrids or unimproved populations of both P. coccineus and P. polyanthus were selected in association with one or two representative maize cultivars for each of the two cropping systems prevailing in the target regions. These maize cultivars are well adapted in the Andean highlands, have strong stalks to provide support to long and twisted branches of beans with indeterminate growth and are relatively resistant to lodging. Although the desirability of selecting under the target intercrop combinations is obvious, the cost and complexity of such investigations often compel the breeder to carry out part of the eld trials under single cropping. This is usually advised for early hybrid generations and qualitative traits. Quantitative traits or traits of specic adaptation are evaluated in more advanced generations, in relevant multiple cropping situations (Francis, 1985, 1986). As the objectives of the breeding programme are not to introduce selectively a few genes in some elite cultivars but rather to broaden the genetic base of locally adapted

376

J.P. Baudoin et al.

varieties, a population improvement scheme was adopted. The aims were to break up unfavourable linkage blocks and to incorporate a wider diversity of genetic resources into forms that can be further exploited by small farmers or integrated in more conventional varietal improvement (Poehlman, 1987).

Description of the breeding method The breeding methodology relied mainly upon the genetic wealth available in the two species: P. coccineus and P. polyanthus. At the onset of the research project (1986), the CIAT base collection contained 1570 populations, divided as follows: 1013 and 496 cultivated forms of, respectively, P. coccineus and P. polyanthus, and 61 wild forms closely related to P. coccineus. Most populations (73%) came from Mexico and Guatemala; the other important countries of origin were Puerto Rico, Costa Rica, Colombia, Venezuela and Bolivia (Schmit and Baudoin, 1987). The genetic improvement of Phaseolus beans using this germplasm collection followed a gradual process involving several activities. Seed increase of the collection The two species are characterized by an outcrossing breeding system ascribed chiey to the extrorse stigma and its position relative to the anthers. An appropriate seed-increase method was therefore used which maintained genetic integrity and total variability of the separate populations. Multiplication was carried out in two research stations of Colombia: Popayan (altitude 1750 m, Cauca) and Rionegro (altitude 2200 m, Antioquia). For each population grown in mesh cages, daily manual pollinations on several plants were made with a pollen mixture from all the owering individuals (Vanderborght, 1983). This was a tedious and costly method but was worthwhile to prevent a dilution of useful genes, which would have occurred in a more open system of mass reservoir maintenance. Seed samples from about 800 populations were produced from this seed increase operation and were available for evaluation trials. Evaluation of the collection Trials were rst conducted in the two Colombian stations of Popayan and Rionegro. Priority was given to plant architecture and yield component characters (such as stout main stems, moderate branching, long racemes, high pod setting rates and large seed size), earliness in maturity, adaptation to poor soil, cold tolerance and, more especially, resistance to the major diseases prevailing in Andean highlands: Ascochyta leaf blight (due to Phoma exigua var. diversispora (Bubak) Boerema), anthracnosis (due to Colletotrichum lindemuthianum (Sacc. and Magn.) Scribner) and angular leaf spot (due to Phaeisariopsis griseola Sacc.). Among the three, Ascochyta leaf blight appeared the most destructive enemy of beans in the target regions. On local P. vulgaris varieties, early attack of Phoma exigua var. diversispora can completely destroy leaves and pods and cause 74% yield losses in the Andean highlands. For this reason disease nurseries were established in Rionegro and Popayan for the years 1987 and 1988. The nurseries included 285 populations selected on the basis of their geographical distribution, using a randomized complete block design with three replications, and three locally adapted large-seeded varieties of P. vulgaris as control. The bean genotypes were single cropped

Breeding Phaseolus for Intercrop Combinations

377

or intercropped with maize (Schmit and Baudoin, 1992). Disease evaluation of Ascochyta leaf blight relied either on natural or articial inoculation, and scores, based on a 19 severity scale (van Schoonhoven and Pastor-Corrales, 1987), were taken at three different stages: owering, pod lling and seed physiological maturity. Creation of interspecic hybrids The objectives were to increase the genetic diversity of the locally adapted Andean cultivars of P. vulgaris, adding useful genes from P. coccineus or P. polyanthus populations identied mainly for their high disease resistance (particularly to Phoma). Crosses were made in the greenhouses of the University of Gembloux, using two systems: direct hybrids between P. vulgaris and one of the two donor species (with the help of embryo culture for some specic combinations) or indirect hybrids involving three or four parents with a wild P. coccineus form as a rst maternal parent. Several genotypes of the latter were identied for their high combining ability with P. vulgaris in previous experiments (Baudoin et al., 1992). The presence of wild P. coccineus cytoplasm favours interspecic gene exchange between the recurrent P. vulgaris parent and the donor parent (either P. coccineus or P. polyanthus). Such complex crosses, with P. coccineus cytoplasm, avoid a quick reversal towards the recurrent P. vulgaris parent and consequently the loss of useful genes from the donor parent (Baudoin and Marchal, 1991). Field breeding in Colombia and Peru Breeding work was carried out in several research stations representative of the target Andean regions in Colombia: Popayan, Rionegro and Pasto (altitude 2710 m, Nario); in Peru: Chiquian (altitude 3550 m, Ancash) and other sites of Ancash, Cajamarca, Cusco and Lima Departments, at altitudes ranging from 2200 to 2800 m. Trials were carried out separately for the two types of material: P. coccineus and P. polyanthus bred as a distinct crop and the interspecic combinations created at Gembloux. For the two species, the selection units consisted of tested populations and single plants within the best populations. Open-pollinated seeds from progeny of selected plants or populations formed the next generation. In advanced generations, the most interesting populations were eld isolated and selected at within-population level. For the interspecic crosses, hybrid breakdown, zygotic elimination and reduction in heterogenetic recombination impeded the following up of any kind of single plant, mass or bulk selection. The most appropriate breeding methodology adopted was cumulative or recurrent selection, designed to maintain a good state of heterozygosity during several generations and to increase the probability of crossing over in between linked genes (Baudoin and Marchal, 1991). The breeding scheme included reselection at each generation with intermating of selected plants to provide genetic combination and breakage of unfavourable linkage. Another similar methodology adopted for such interspecic hybrids was congruity backcrossing. This method consisted of recurrent backcrossing to each parent (both the recurrent P. vulgaris and the donor species) alternately at each generation. This procedure gave an increased fertility level of hybrid genotypes and maintained a balanced set of genes from the two species over several generations (Baudoin et al., 1995). In these two breeding schemes, intermating was carried on during several seasons, with a constant ow of new interspecic hybrids added to the recombination nursery. The purpose was to develop mixed-genotype populations, with a high frequency of favourable genes and enough variability for

378

J.P. Baudoin et al.

adaptation to a wide range of microenvironments (representing the small-scale Andean farms). Field breeding of P. coccineus, P. polyanthus and interspecic combinations was conducted during the rst generations at on-station level, in single cropping. The emphasis was on traits of interest whatever the cropping system involved, e.g. disease resistance, seed coat colour, seed size or earliness in owering. For advanced generations, breeding trials were conducted at both on-station and on-farm levels, in relay or intercropping systems. Selection, at this stage, concerned traits with specic interest for multiple cropping systems and were mostly of a quantitative nature: for example, competitive ability with the associated maize crop, tolerance to shade, rapid seedling growth, vegetative vigour, harvest index and dry seed yield (Baudoin et al., 1997). Figure 23.1 illustrates the outline of the breeding scheme

Results to Date
Evaluation of P. coccineus and P. polyanthus germplasm collections The disease nursery trials in the two highland stations of Rionegro and Popayan demonstrated the high resistance levels of the two donor species to Ascochyta leaf blight, compared with the susceptibility of the control P. vulgaris variety (Table 23.1).

Evaluation of Phaseolus coccineus and Phaseolus polyanthus germplasm Popayan and Rionegro (Colombia)

Follow up evaluation P. coccineus and P. polyanthus Colombia and Peru

Interspecific hybridization with P. vulgaris University of Gembloux

Varietal improvement P. coccineus and P. polyanthus Single plant and family selections Rionegro (Colombia), Chiquian (Peru)

Recombination nursery Intermating of hybrids: Chiquian (Peru) Congruity backcrossing: Rionegro (Colombia)

Yield trials P. coccineus and P. polyanthus Colombia and Peru

Varietal improvement Masspedigreebulk selections Rionegro (Colombia), Chiquian (Peru)

On-station and on-farm trials of improved lines Colombia and Peru

Fig. 23.1. Outline of the breeding scheme.

Breeding Phaseolus for Intercrop Combinations Table 23.1. Field reaction of P. coccineus and P. polyanthus populations to Ascochyta leaf blight at Popayan and Rionegro (Colombia). No. of populations in the three reaction classesa Species and site Phaseolus coccineus Rionegro Popayan Phaseolus polyanthus Rionegro Popayan
aScores

379

Resistant (score 13) 76 7 110 60

Intermediate (score 46) 19 12

Susceptible (score 79) 1

recorded at the stage of seed physiological maturity.

The 170 P. polyanthus populations showed outstanding resistance, with most accessions (64%) exhibiting no symptoms in the two stations. The best ones originated from Colombia, Venezuela and Mexico. The reaction of P. coccineus was more variable, with 83 populations classied as resistant, 31 classied as intermediate and one classied as very susceptible. The best genotypes originated from Costa Rica, Mexico and Guatemala. We also noticed the highest level of resistance from populations collected from sites above 2000 m, where climatic conditions (high humidity and cool to moderate temperatures) favour infection by the fungus. In the same nurseries, all the P. coccineus and P. polyanthus populations also showed an excellent reaction to the other two prevailing fungus diseases: anthracnosis and angular leaf spot, while a severe attack of these diseases occurred on P. vulgaris controls. According to genotypes and locations, dry seed yield of P. coccineus and P. polyanthus ranged from 250 to 3500 kg ha 1, with a growing period of 67 months. Dry seed yields of P. vulgaris ranged from 150 to 1200 kg ha 1, but with a shorter growing period of 45 months. Additional trials were conducted from 1989 in the same Colombian stations and in Pasto (Nario) and Peru (mainly Chiquian), covering a wider geographical distribution of the germplasm collection (with a total of 400 populations). These observations conrmed the eld resistance to Ascochyta leaf blight in the two species and revealed high variability in growth duration, plant architecture and yield components.

Creation of interspecic hybrids Table 23.2 shows the interspecic hybrid types, developed in Gembloux from 1985 to 1989. A total of 360 genotypic combinations of direct crosses and 220 genotypic combinations of complex crosses were obtained. The highest rates of success were obtained in crosses using P. vulgaris as female parent and a single donor species. Hybrids with P. polyanthus cytoplasm required in vitro embryo culture, due to starvation of embryos at a very early developmental stage (Mergeai et al., 1997). In the complex crosses, we identied three wild forms of P. coccineus, from Mexico, which showed a good crossing compatibility with P. vulgaris. Hybrids with such wild cytoplasms could act as a bridge

380 Table 23.2. Creation of interspecic hybrids with Phaseolus beans. Direct crosses Female P. vulg.a cvd P. vulg. cv P. poly. cv P. cocc. wlde P. cocc. wld P. poly. cv Male P. cocc.b cv P. poly.c cv P. vulg. cv P. vulg. cv P. poly. cv P. cocc. cv Complex crosses Female (P. cocc. wld (P. cocc. wld (P. cocc. wld (P. cocc. wld (P. poly. cv (P. poly. cv
aP. dcv

J.P. Baudoin et al.

No. of genotypic combinations 120 220 2 13 2 3

Male P. vulg. cv) P. vulg. cv) P. vulg. cv) P. poly. cv) P. cocc. cv) P. vulg. cv) P. vulg. cv P. cocc. cv P. poly. cv P. vulg. cv P. vulg. cv P. vulg. cv

No. of genotypic combinations 14 88 18 92 4 4

vulg. = P. vulgaris; bP. cocc. = P. coccineus; cP. poly. = P. polyanthus; = cultivated forms; ewld = wild forms.

to combine genes from both the recurrent (P. vulgaris) and donor P. coccineus or P. polyanthus.

Field breeding At the onset of the breeding programme, selection focused on dry seed yield (above 15002000 kg ha 1 according to the sites), earliness in owering (from 90 to 130 days), seed size (with a preference for large-seeded genotypes with 100-seed weight above 45 g) and disease resistance (mainly to Ascochyta leaf blight). By doing so, we retained in each station 80100 populations from each of the two species, P. coccineus and P. polyanthus, and around 200 interspecic lines from the crosses developed in Gembloux. Single plant and family selections were carried out with P. coccineus and P. polyanthus materials, using different cropping patterns: single cropped, relay and intercropped with maize. The best lines, from 15 to 30 in each of the two species, were tested in multilocational yield trials with locally adapted cultivars of P. vulgaris. Table 23.3 shows the performance of promising lines in single and multiple cropping systems in two Colombian stations. No fertilizer treatment and no protection against diseases were practised. In Rionegro, the two improved lines of P. coccineus and P. polyanthus outyielded the local common bean variety. This reects partly their very good disease

Breeding Phaseolus for Intercrop Combinations

381

Table 23.3. Seed production of improved Phaseolus lines in single cropping and association (Colombia). Dry seed yield (kg ha 1) Phaseolus genotypes Rionegro: relay cropping with G 35099 (P. coccineus) G 35224 (P. polyanthus) ICA VIBORAL (P. vulgaris)2 maize1 Single cropping (August 1993 to January 1994) 875b 932b 146c Association 1770a 1769a 315c 3645.7a 3057.7ab 2874.7ab 1840.bc 1269.3a 1534.4c

Pasto: simultaneous intercropping with maize (October 1994 to April 1995) G 35544 (P. polyanthus) 6514.3a G 35522 (P. polyanthus) 5978a G 35521 (P. polyanthus) 2981b G 35560 (P. polyanthus) 5724a G 35516 (P. polyanthus) 2772.7b BOLON ROJO (P. vulgaris)2 2894.9 b
1Beans

sown at owering period of maize; 2best local variety used as a control. Means followed by the same letter are not signicantly different (P 0.05).

resistance, compared with the heavily attacked control. In the same location, seed production of P. coccineus and P. polyanthus genotypes was higher in relay cropping than in single cropping. This can be explained by the good complementarity and lack of competition between maize and bean in such cultural conditions. In Pasto, most P. polyanthus improved lines outyielded the dry seed production of the P. vulgaris control in both single and intercropping. However, we observed a signicant yield reduction of intercropped Phaseolus beans compared with single cropping. This reects the strong competition between the two associated crops. However, farmers adopting this cropping pattern are usually more interested in the overall productivity and stability of the intercrop combinations, rather than in the seed production of each crop component. A good parameter to assess this overall productivity is the land equivalency ratio (LER), as described by Willey (1979). Preliminary results from Peru show an LER 1 for some maizebean associations. The interspecic lines were integrated in a recombination nursery, under single cropping and with the purpose of developing a large source population combining useful traits of P. vulgaris, P. coccineus and P. polyanthus. This recombination nursery was maintained for at least 4 years at the two stations of Rionegro and Chiquian. At each cultural season, we selected for the crossing block around 2030 hybrid plants with complementary characters (particularly in terms of disease resistance, earliness in owering, branching, seed production and harvest index) and a sample of the parental genotypes involved in the initial hybrids. In the case of cyclic recurrent selection (Chiquian), intermating was done each season between the most interesting interspecic individuals. In case of congruity backcrossing (Rionegro), hybrids were backcrossed alternately with the recurrent (P. vulgaris) and donor (P. coccineus and P. polyanthus) parents. In the recombination nursery, two major conditions were required in order to obtain progress in breeding interspecic hybrids. First, disease pressure had to remain high in the crossing block to keep genes of resistance from the donor species

382

J.P. Baudoin et al.

from one generation to the next. Secondly, intermating always had to involve parents having vulgaris-like and donor species-like characters. This was a prerequisite to broaden the genetic base of the improved populations, to avoid the rapid reversal toward the recurrent parent and to break undesirable linkage blocks present in either P. coccineus or P. polyanthus. One major difculty in breeding interspecic hybrids was rupturing the linkage between high levels of resistance to Ascochyta leaf blight and some unfavourable traits, such as lateness in owering, profuse branching and low harvest index. Once good interspecic genotypes were identied, they were grouped according to similarity traits (mainly degree of branching, growth duration, seed size and colour, etc.) and submitted to mass, pedigree and bulk selection for 23 years. Such trials were rst conducted on-station under relay cropping (Rionegro) or simultaneous intercropping (Chiquian). Later hybrid generations were tested in both on-station and on-farm trials, using only traditional intercrop combinations. In Peru, the breeding scheme adopted for the interspecic hybrids generated two improved composite populations from the combination (P. coccineus wild form P. vulgaris) P. coccineus: Bulk 1H and Bulk 2H. In rst experiments made in Ancash and Lima Provinces, these two improved lines outyielded the P. vulgaris control (ANC-034 from Canario class), owering 20 days earlier and giving a higher seed index (7080 g 100 seeds 1 compared with 45 g 100 seeds 1 in ANC-034). The two lines also showed high eld resistance to both anthracnosis and Ascochyta leaf blight. Bulk 1H and Bulk 2H are being tested in different small-scale farms of the Province of Ancash. For this purpose, we identied around 20 farmers interested in the improved varieties. As each variety represents a composite material, farmers are encouraged to practise their own selection and to modify their crop husbandry.

Conclusions and Recommendations

Our research on Phaseolus breeding is based upon a large genetic reservoir made of local germplasm stocks of two New World species (P. coccineus and P. polyanthus) and their crosses with P. vulgaris. The breeding outline follows the principles of a population improvement programme and leads to bulk varieties keeping sufcient genetic variation to cope with a traditional and heterogeneous agrosystem. The strategy is well tted to breeding for multiple characters of both quantitative and qualitative inheritance and to developing varieties that combine good seed production, stress tolerance and competitive ability with the companion crops. In this process, small-scale farms can pick up improved populations with a broad genetic base from the breeding nurseries. This material has sufcient plasticity to respond favourably to various environmental constraints. Success in breeding for intercrop combinations, however, requires several pre-conditions. These include availability of a large body of germplasm collection and hybrid populations, sufcient knowledge of the major agronomic constraints, good understanding of the yield components in relation to competitive ability, resources to breed at both on-station and on-farm levels and to adopt experimental layout mimicking traditional cropping systems. In order to upgrade productivity of intercrop combinations, any genetic changes among the component crops should also be accompanied by

Breeding Phaseolus for Intercrop Combinations

383

modications in agronomic practices. The nal objective will be not to disturb markedly a traditional cropping system but rather to open it up to other innovations. The ultimate impact of the research efforts will be the farm families themselves, which are faced with a need for food security and better income. As stated before, the farming system approach will help to develop end-products that outperform existing varieties and are readily acceptable by the local populations. In this approach, on-farm trials are essential to back station-based investigations and to reorient the selection process according to the new constraints observed in a less uniform and more stressful environment. Testing in farmers elds will also enable the selection of improved materials not only on the basis of yield but also on the basis of economic value, stability, total cropping system production or labour input. The link between breeders, extension services and farmers is of course essential to achieve the breeding objectives. Such a close collaboration between these three should be nancially supported by public agencies involved in agricultural research and integrated rural development.

References
Baudoin, J.P. and Camarena, F. (1994) Food legumes adapted for multiple cropping systems. Grain Legumes (AEP)5, 2325. Baudoin, J.P. and Marchal, R. (1987) La culture associe cralelgumineuse. Systmes traditionnels et possibilits dintensication en milieu tropical. In: AUPELF (Association des Universits Partiellement ou Entirement de Langue Franaise) (ed.) Les lgumineuses alimentaires en Afrique. Colloquium organized by the University of Niamey, 1922 November 1985, pp. 1922. Baudoin, J.P. and Marchal, R. (1991) Wide crosses and taxonomy of pulse crops, with special emphasis on Phaseolus and Vigna. In: Ng, N.Q., Perrino, P., Attere, F. and Zedan, H. (eds) Crops Genetic Resources of Africa, vol. II. Proceedings of an international conference organized by the International Institute of Tropical Agriculture (IITA) and the National Research Council of Italy (CNR), in association with the International Board for Plant Genetic Resources (IBPGR) and the United Nations Environment Program (UNEP), and held in Ibadan, Nigeria, 1720 October 1988, pp. 287302. Baudoin, J.P., Camarena, F. and Lobo, M. (1995) Amlioration de quatre espces de lgumineuses alimentaires tropicales Phaseolus vulgaris, P. coccineus, P. polyanthus et P. lunatus. In: Dubois, J. and Demarly, Y. (eds) Quel avenir pour lAmlioration des Plantes? Quatrimes Journes Scientiques du Rseau Biotechnologies Vgtales de lUREF (Universit des Rseaux dExpression Franaise). Namur, 1821 October 1993. John Libbey Eurotext, Paris, pp. 3149. Baudoin, J.P., Camarena, F. and Schmit, V. (1992) Contribution une meilleure connaissance de la position phyltique de la lgumineuse alimentaire Phaseolus polyanthus Greenm. Bulletin des Recherdes Agronomiques Gembloux 27(2), 167199. Baudoin, J.P., Camarena, F. and Lobo, M. (1997) Improving Phaseolus genotypes for multiple cropping systems. Euphytica 96, 115123. Davis, J.H.C. and Garcia, S. (1983) Competitive ability and growth habit of indeterminate beans and maize for intercropping. Field Crops Research 6, 5975. Francis, C.A. (1985) Variety development for multiple cropping systems. Critical Reviews in Plant Science 3, 133168. Francis, C.A. (1986) Multiple Cropping Systems. Macmillan, New York.

384

J.P. Baudoin et al. Lobo, M. (1994) Phaseolus production in Colombia. Grain Legumes (AEP)5, 2627. Mergeai, G., Schmit, V., Lecomte, B. and Baudoin, J.P. (1997) Mise au point dune technique de culture in vitro dembryons immatures de Phaseolus. Biotechnology, Agronomy, Society and Environment 1(1), 4958. Poehlman, J.M. (1987) Breeding Field Crops. Avi Book, Van Nostrand Reinhold, New York. Schmit, V. and Baudoin, J.P. (1987) Multiplication et valuation de Phaseolus coccineus L. et Phaseolus polyanthus Greenman, deux espces intressantes pour lamlioration de la productivit des lgumineuses vivrires. Bulletin Recherches Agronomiques Gembloux 22(4), 235253. Schmit, V. and Baudoin, J.P. (1992) Screening for resistance to Ascochyta blight in populations of Phaseolus coccineus L. and P. polyanthus Greenman. Field Crops Research 30, 155165. Schmit, V. and Debouck, D.G. (1991) Observations on the origin of Phaseolus polyanthus Greenman. Economic Botany 45(3), 345364. Sperling, L., Loevinsohn, M.E. and Ntabomvura, B. (1993) Rethinking the farmers role in plant breeding: local bean experts and on-station selection in Rwanda. Experimental Agriculture 29, 509519. Steiner, K.G. (1990) Manuel dexprimentation en milieu paysan pour les projets de dveloppement rural. GTZ (Deutsche Gesellschaft fr Technische Zusammenarbeit), GmbH, Eschborn, Germany. Vanderborght, T. (1983) Increasing seed of Phaseolus coccineus L. Plant Genetic Resources News (IBPGR) 53, 1718. van Schoonhoven, A. and Pastor-Corrales, M.A. (1987) Standard System for the Evaluation of Bean Germplasm. CIAT, Cali, Colombia. Willey, R.W. (1979) Intercropping its importance and research needs. Part 1. Competition and yield advantages. Field Crop Abstracts 32, 110. Woolley, J. and Davis, J.H.C. (1991) The agronomy of intercropping with beans. In: van Schoonhoven, A. and Voysest, O. (eds) Common Beans: Research for Crop Improvement. CAB International/CIAT, Wallingford, UK, pp. 707735. Woolley, J.N., Lepiz-I., R., Potes e Castro, T. de A. and Voss, J. (1991) Bean cropping systems in the tropics and subtropics and their determinants. In: van Schoonhoven, A. and Voysest, O. (eds) Common Beans: Research for Crop Improvement. CAB International/CIAT, Wallingford, UK, pp. 679706.

24

Improving Potato Resistance to Disease under the Global Initiative on Late Blight

B.R. TROGNITZ1, M. BONIERBALE1, J.A. LANDEO1, G. FORBES2, J.E. BRADSHAW3, G.R. MACKAY3, R. WAUGH3, M.A. HUARTE5 AND L. COLON6
1International

Potato Center (CIP), PO Box 1558, Lima, Peru; 2International Potato Center (CIP), Experimental Station at Santa Catalina, PO Box 17-211977, Quito, Ecuador; 3Scottish Crop Research Institute (SCRI), Invergowrie, Dundee DD2 5DA, UK; 4Instituto Nacional de Tecnologia Agropecuaria (INTA), Ruta 226, km 73.5, Balcarce, Buenos Aires 7620, Argentina; 5Centre for Plant Breeding and Reproduction (CPRO-DLO), PO Box 16, Wageningen 7600 AA, The Netherlands

Introduction

Late blight is the most devastating plant disease, occurring in almost every potatogrowing area of the world. The disease affects both foliage and tubers and it can destroy an entire eld within a few days. Late blight was responsible for the Irish potato famine in the 1840s, and it causes estimated crop losses of nearly 14 million tons annually, equivalent to nearly $3 billion (International Potato Centre, 1997). Despite major and expensive use of crop protection chemicals to control potato diseases, more than 20% of the potato crops potential annual yield is lost (James, 1981). Whereas in the past late blight was barely known in the potato-growing areas of the South American Andes (Cox and Large, 1960), it is now reported to be the number-one constraint in potato production in Bolivia, Argentina, Peru, Ecuador and Colombia, where potato is a staple food of small-scale farmers. In developing countries, production has soared by nearly 200% in the past 30 years. It now constitutes about one-third of global production and is expected to continue to rise by about 3% a year. Blight is especially serious in tropical regions due to continuous production cycles, year-round persistence of inoculum, and limited access to or frequent misapplication of fungicides. The small-scale, labour-intensive farming systems typical of developing countries provide specic conditions that call for the development and implementation of new integrated crop management procedures. This has prompted a worldwide increase of efforts towards the efcient control of the disease, with a focus on the use of resistant varieties. In 1996, the Global Initiative on Late Blight (GILB)
IPGRI/FAO 2001. Broadening the Genetic Base of Crop Production (eds H.D. Cooper, C. Spillane and T. Hodgkin)

385

386

B.R. Trognitz et al.

was inaugurated by breeders and late blight researchers from 30 institutions of 16 countries. This Initiative unites late-blight researchers from institutions of developing and developed countries in their efforts to collaborate in the development of integrated pathogen management strategies in which new potato varieties with long-lasting resistance to late blight would play the pivotal role. It was recognized that conventional breeding and selection of potato are central to any attempt to enhance durable resistance in new potato varieties (French and Mackay, 1996). The initiative facilitates the exchange of information and germplasm, to carry out independent and collaborative research on the pathogens diversity in and among different regions of the world, and to better characterize and understand genetic resistance and its interactions with the environment. The principal focus of the initiative is the development of cultivars with durable resistance to numerous variants of the fungus that are appropriate for use in integrated disease management programmes in developing countries. This chapter focuses on those activities of the GILB that aim to broaden the genetic base for durable resistance to late blight, and examine genotype environment (GE) interactions for such resistance.

Background
The pathogen The causal agent of late blight is Phytophthora infestans. Epidemics result from asexual propagation through the formation of sporangia, and the mycelium can overwinter in plant tissue such as potato tubers and debris. Sporangia either germinate directly or they release infectious zoospores. Many species of the Solanaceae and Nolanaceae, including potato and tomato, are natural hosts of P. infestans, and inoculation experiments have revealed that it causes disease on more than 130 species of higher plants (Abad et al., 1995). The sexual form of propagation via oospores has been observed in Mexico for many years (Gallegly and Galindo, 1958), while in other parts of the world it is seldom observed in the eld (Goodwin and Drenth, 1997). P. infestans is heterothallic and strains can be distinguished by two mating types: A1 and A2. It is generally assumed that because the A2 mating type was not found in Europe or North America until 1980, populations of P. infestans outside of Mexico consisted of only the A1 mating type and, therefore, could not propagate sexually (Fry and Goodwin, 1997). The common European strain of P. infestans has been shown to be of clonal origin (Goodwin et al., 1994a), and the phenotypes of its limited virulence spectrum correspond to 12 well-studied virulence factors: 11 from S. demissum (Malcolmson and Black, 1966) and one from S. stoloniferum (Schick and Schick, 1959). This strain, which may have originated in Mexico, was introduced to Europe along potato trading lines (Goodwin et al., 1994a) and was also prevalent in North America, Africa, Asia and South America. Around 1980, migration from Mexico of new, more aggressive strains of the pathogen, comprising both mating types, resulted in a signicant increase in the number and severity of epidemics, reduced effectiveness of established fungicides due to resistance of the new strains, and increased the area where epidemics occur (Fry et al., 1993).

Improving Potato Resistance to Disease

387

Although discussion about the possible occurrence of sexual recombination of the fungus outside of Mexico before 1980 continues (Fry and Goodwin, 1997; Ristaino, 1998), the well-documented migration of two sexually compatible mating types provides the opportunity for even more aggressive genotypes to develop through sexual recombination (Drenth et al., 1994). Many pathogen populations have acquired resistance against xymoxanyl, the active compound of metalaxyl and several other popular fungicides used to control late blight (Goodwin et al., 1994b). As a result of these pathogen dynamics, the incidence of the disease and severity of epidemics have increased signicantly since the early 1980s.

Breeding for late blight resistance Breeding for resistance continued with varying degrees of success with the introduction of resistance from wild and weedy germplasm, such as S. edinense and S. demissum which started in around 1900 (Mller, 1951). These sources were employed to transfer to potato major resistance (R) genes conferring hypersensitive resistance and immunity. However, the R genes, interacting on a gene-for-gene basis (Flor, 1956) with the pathogens virulence, did not provide lasting protection in the eld because of the frequent and ubiquitous occurrence of pathotypes of P. infestans matching all 12 R genes. For the same reason, combining or pyramiding of these R genes into one genotype also did not lead to resistant varieties. Pyramiding of major resistance genes could result in durably resistant genotypes only when genes were employed that were not matched by the virulences of any pathotype occurring in the area where the crop would be grown. The duration of resistance of such a complex genotype also depends on the assumption that a pathotype able to overcome the rst acting gene in the pyramid is not able to establish a population on the host sufciently large to allow selection of mutants that can overcome the second gene, and so on (Wolfe and Gessler, 1992). In contrast to R-gene resistance, the quantitative expression in the eld of the horizontal, rate-limiting type of resistance can be measured as a partial reduction of the amount of disease relative to a susceptible standard. A small number of European potato varieties namely Robijn, Populair, Libertas, Pimpernel and Surprise have maintained their comparatively high resistance levels in the eld in middle Europe for up to 50 years (Colon et al., 1995). This evidence that these varieties still express unaltered levels of resistance in spite of the changing composition of the pathogen population allows the interpretation of their resistance as the horizontal type and more durable than R-gene resistance. Proof of the genetic nature of this resistance whether it is polygenic, oligogenic or governed by single genes can only be obtained through genetic analyses on segregating progeny from a self or cross of these varieties, yet to be carried out. The genetic nature of race-non-specic resistance to late blight has remained unknown, and diverse hypotheses about its genetic architecture have been put forward. Black (1945) concluded from empirical evidence that horizontal resistance is polygenic. The importance of selecting for partial resistance, in contrast to immunity, has been stressed by Parlevliet (1979) based on the underlying assumption that most combinations of minor resistance factors will not result in perfect, complete resistance or immunity and that, in contrast, extremely high levels of polygenic resistance may be

388

B.R. Trognitz et al.

confounded with single, major gene resistance. Since then, this viewpoint of partial resistance in the eld resulting from the additive action of many genes, each one contributing a minor positive effect, has been the basis of successful resistance breeding of potato and other crops by many breeders and pathologists (Simmonds and Malcolmson, 1967; Umaerus, 1970; Ross, 1986; Parlevliet, 1993). The key feature of horizontal resistance is its race-non-specicity, i.e. it is effective against all strains of a pathogen. Race-non-specic, durable major (R) gene resistance is known from several other pathosystems involving wheat (Johnson, 1988; Roelfs, 1988) and barley (Jorgensen, 1994). Whether race-non-specic resistance to potato late blight always depends on the action of additive polygenes or on other mechanisms remains to be elucidated, as does the need for developing more efcient breeding strategies for this characteristic. Recent efforts to determine on genetic maps the position of quantitative trait loci (QTLs) conferring plant disease resistance have revealed that QTLs are frequently located in the vicinity of major resistance gene loci (Gebhardt, 1994; Veldboom et al., 1994; Mitchellolds, 1996; Steffenson et al., 1996). Resistance genes often are clustered at specic regions of the plant genome (Michelmore, 1995). In an analysis of a large number of potato varieties with and without R genes, Darsow et al. (1987) observed an overall increased resistance level of varieties with R genes, which, as they proposed, could be the result of undetected introgression of resistance factors other than the R genes from the wild sources.

Outline of Activities
Breeding and broadening the genetic base for durable resistance to late blight at the International Potato Centre Evaluation of existing varieties (SIFT) Plant breeders have developed many potato varieties that are resistant to late blight under local conditions. Comparison of resistance levels in varieties has been problematic because of lack of knowledge about the extent to which different environments inuence the expression of resistance, because different pathogenic races have been used, and because local evaluations may use different methods. Therefore, a multilocation standard international eld trial (SIFT) was launched in 1998 (French and Mackay, 1996) and its task is to distribute and test the best available late-blight-resistant potato varieties and breeding lines, submitted by breeders from all around the world, for adaptation and acceptance in developing countries. In this way, producers should gain ready access to the best available resistant materials that are adapted to their needs. The trial is being carried out at seven locations in America, Africa and Asia. Breeding for horizontal resistance to late blight using advanced sources This activity focuses on developing improved varieties for a wide range of environments that possess high and stable levels of resistance to late blight in combination with resistance to viruses and suitable culinary and processing quality. Several sources of resistance are used in three groups of materials differing in their genetic base and degree of enhancement. The groups are:

Improving Potato Resistance to Disease

389

Group B1, in which native cultivated forms of short-day-adapted S. tuberosum ssp. andigena, which are assumed not to depend on R-gene resistance, are intercrossed. Group B2, in which clones of both subspecies of the common potato, tuberosum and andigena, are intercrossed. Materials in this group are adapted to short-day and they are gradually improved for resistance and agronomic traits. Group B2 germplasm is being improved for adaptation to long-day at the Instituto Nacional de Tecnologia Agropecuaria (INTA), Argentina, as will be explained on page 391. Group B3, which represents the agronomically most advanced group of International Potato Centre (CIP) breeding clones that do not rely on race-specic R-gene resistance, developed as described below. This group combines quantitative resistance from S. tuberosum spp. tuberosum and andigena, from Neotuberosum clones (Plaisted, 1987), and from complex hybrids of S. acaule, S. bulbocastanum, S. phureja and S. tuberosum ssp. tuberosum (Hermsen, 1994). Again, the role of INTA, Argentina, is to select, recombine and adapt for long-day the germplasm of this group.

Recurrent selection is carried out within each of these groups, to increase the frequency of alleles that contribute to quantitative resistance. Extensive testing for resistance and other agronomically important characteristics is carried out in each generation of the recurrent scheme. The procedures employed focus on:

Selection for quantitative, race-non-specic resistance. Selection against race-specic resistance that simplies testing in the following generations. Clones possessing quantitative resistance can then be detected by using any race of the pathogen, independently of its particular virulence. Selection in those geographic areas where the new varieties will be grown. In that way, genotypes adapted to a specic environment can be identied. Selection of clones possessing a high general combining ability for resistance and yield. These clones can then be used as parents in local breeding programmes of National Agricultural Research Systems (NARSs).

Description of group B3 To develop the B3 group of late-blight-resistant clones, rst crosses were done in 1990. Best-performing individuals were selected from segregating progeny through eld testing for quantitative resistance and desired agronomic characteristics. These clones were then screened to eliminate individuals that expressed race-specic quantitative resistance (Landeo and Turkensteen, 1989), by crossing with a tester that does not possess R genes for resistance. Segregating seedling populations resulting from these test crosses are inoculated with an avirulent isolate (0-race) of P. infestans, as an additional screen against race-specic resistance. Parents of seedling populations segregating in a way that is expected for R-gene resistance are discarded. Parental clones that produced seedling progeny of non-segregating partial resistance in that test are used for recurrent crosses. A set of 112 clones possessing race-non-specic resistance was selected from the second cycle of recurrent selection. When used as parents, the potential to obtain, under a wide range of environments, varieties with resistance to late blight and improved agronomic and quality attributes is increased. Field tests to quantify the resistance levels of a sample of partially resistant clones

390

B.R. Trognitz et al.

were carried out at four locations in Peru (two), Colombia and Mexico, using segregating progeny that had been obtained in a line tester mating design during 1992. Narrow-sense heritabilities for resistance were 0.50.7 and those for total tuber yield were 0.250.5, indicating that further improvement through breeding of these characteristics is possible. The values obtained may, however, overestimate the actual heritability coefcient because the cross progenies tested at each site consisted of unique seedling individuals. Screening of related germplasm and pre-breeding for resistance at CIP The genepool of potato is comprised of six cultivated species and over 200 wild tuberbearing relatives (see Ortiz, Chapter 10, this volume). Ploidy levels range from diploid (2n = 2x = 24) to hexaploid (2n = 6x = 72), but 80% are diploid and self-incompatible. Several thousand accessions of cultivated and wild potato are held at six major and several minor genebanks, and attempts to evaluate the value for late-blight resistance breeding of these sources are as old as the collections themselves (Schick, 1932; Lehmann, 1938). Most procedures employed in the past to characterize late-blight resistance used only one or a few isolates of P. infestans to test a small sample of genotypes drawn from an accession, and could assess only the average resistance of an entire accession. Specic individual plants of an accession found to be resistant were not saved. The efciency of these methodologies for providing resistant germplasm and information immediately applicable by plant breeders is therefore limited. More recent approaches aim at overcoming these difculties by providing information about the type of resistance and the genetic structure of an accession. One such example is the ne screening method of germplasm for resistance to Colorado potato beetle proposed by Bamberg et al. (1996). An approach taken at CIP illustrates this development for characterizing late blight resistance of genebank accessions. The collection of wild potatoes held at CIP holds 1500 accessions representing 110 species. Current investigations on this collection are designed to: (i) better characterize the frequency and relative levels of resistance available in the accessions; (ii) characterize the nature of resistance (race-specic, non-specic); (iii) estimate the diversity of resistance genes present; and (iv) provide new genetic sources for resistance breeding. To meet these objectives, a complex procedure involves testing of a comparatively large number of samples of each accession for their response to several P. infestans isolates that represent the widest possible range of known contrasting virulences. Simultaneously, resistance sources and hybrid individuals are evaluated for important traits. Selected genotypes may then be studied further by geneticists, or used by breeders for introgression of their resistance into potato. Because potato varieties are clonal, and high levels of heterozygosity are advantageous, most efcient breeding relies on parental clones of known resistance genotype. Testing and selection of one potato cross generation takes 35 years (Ross, 1986), and more than ve crossing generations may be needed to breed a variety. Therefore, reduction of the number of generations could considerably reduce the time for variety development. Generations can be saved when the inheritance of resistance is known. Several populations of wild species selection S. tuberosum dihaploids are being investigated (Trognitz et al., 1995, 1996). One example is a hybrid population of a cross between a highly resistant genotype of wild S. verrucosum and a susceptible geno-

Improving Potato Resistance to Disease

391

type of cultivated S. phureja, designated population VP. Hybrid seeds of this cross were provided by Dr Nelson Estrada, PROINPA, Bolivia. The progeny were tested for resistance in the eld in Peru during 3 years and analysed for occurrence and segregation of race-specic R genes (Trognitz et al., 1997). The high potential value of this population for increasing the resistance to late blight of potato is illustrated by its extremely high resistance level in the eld. The entire VP population had a higher average eld resistance (measured as the area under the disease progress curve) than the resistant variety Perricholi that was used as a standard. Its frequency distribution deviated, however, from the expected normal distribution typical for a polygenic, quantitative trait. Therefore, detached leaet tests were performed to study the occurrence of R-gene resistance expressed as a hypersensitive response to infection. Of 99 plants tested with ve isolates of P. infestans, none was resistant to two isolates, and with the remaining three isolates segregation into resistant and susceptible classes was observed. The ratios of resistant:susceptible individuals obtained with either isolate signicantly diverge from ratios expected for a model of single dominant genes. The complementary action of two or more resistance genes, or a resistance gene and a suppressor gene, were more feasible models (Trognitz et al., 1997).

Breeding for durable resistance to late blight under the long-day conditions of Argentina Testing for both foliar and tuber blight resistance is being performed at two locations in Argentina; Balcarce (38 S) and Tucuman (27 S). The rst location is situated in the centre of the main potato growing area of Argentina and the second in a highland valley with excellent natural conditions for seed production, but severe late-blight epidemics. The objective of the INTA breeding programme is to develop varieties for Argentina that combine resistances to late blight and viruses with early maturity, good tuber characteristics and culinary quality (Mendiburu and Huarte, 1987; Huarte et al., 1994). Breeding is carried out at both the tetraploid and the diploid level. The tetraploid breeding programme To select for resistance to late blight, in each year a minimum of 20,000 seedling plants are inoculated in the eld with P. infestans isolates that are representative of pathogenic strains occurring in the potato production areas of Argentina. A set of potato R-gene differential genotypes is planted each year in the experimental eld to monitor the actual virulence spectrum of Phytophthora throughout the growing season. Because several virus diseases are an important constraint to potato production in Argentina, advanced clones from the late blight programme are tested for virus resistance, using natural and articial inoculation methods. In the tetraploid breeding programme, progenitors obtained from the CIP as well as from European and North and South American potato breeders are used as pistillate parents in crosses with local clones. Argentinian clones that produce superior pollen quality are used as pollen parents.

392

B.R. Trognitz et al.

Breeding clones and varieties produced by the INTA programme combine high levels of eld resistance to several diseases including late blight (Mendiburu and Huarte, 1987). Tests of these varieties at CIP revealed a wide range of resistance levels (Table 24.1) although many varieties do not express any of the known R genes conferring race-specic resistance (G. Forbes, CIPQuito, unpublished data). The varieties that are adapted to long-day and have good tuber characteristics and resistance to viruses represent valuable progenitors used in the breeding programmes. Examples of the success achieved in the Argentinian national potato breeding programme are the varieties Serrana INTA and Achirana INTA, both possessing good general combining ability (GCA) for virus resistance, in particular potato leafroll virus (PLRV) (Huarte et al., 1990); and Pampeana INTA with high GCA for late-blight resistance (M. Huarte, unpublished). Tetraploid breeding for late blight resistance has been intensied since 1994 through the collaboration with CIP when CIP breeding clones were received and selected for several seasons in Balcarce and Tucumn. These materials were late-maturing and they had tuber quality characteristics not accepted by the Argentinian market although they possessed high degrees of resistance to late blight. The performance of the group B clones indicated the possible presence of R genes, as the instability of the resistance was evident. In 1995, selection work began on 20,000 seedlings from cross progeny of CIPs group B3 that does not express any of the known R genes. In 2000, after 2 years of eld trials, 150 clones of superior late-blight resistance in combination with other characteristics of agronomic importance were intercrossed to obtain a second generation of seedling plants for recurrent selection. Seed produced at INTA, of the same group B3 progeny tested at Balcarce, were also distributed for selection for late-blight resistance on seedlings and clonal plants to Toluca, Mexico, and Rionegro, Colombia. Linear correlation of progeny means of resistance rating obtained at Balcarce and the two shortday environments was low (r 0.46), whereas progeny means obtained at Toluca and Rionegro were highly correlated (r 0.84), indicating a signicant genotype environment interaction of resistance. Pathogen variability, photoperiod and temperature may have signicantly contributed to this genotype environment interaction. The development of long-day adapted potato varieties with high levels of horizontal resistance to late blight under high disease pressure has been lengthy and not free from con-

Table 24.1. Comparison of late-blight resistance of Argentinian with European potato cultivars in a 2-year eld trial. Resistance (AUDPC) Cultivar Pampeana INTA Serrana INTA Huinkul MAG Pimpernel Bintje Origin Argentina Argentina Argentina Holland Holland 1996/97 242 1075 1157 387 1786 261 1997/98 878 1147 1259 929 2695 357 Other characters High dry-matter content Resistance to PLRV Resistance to PVY Horizontal resistance to late blight Susceptible to late blight

SE

SE

Improving Potato Resistance to Disease

393

straints, which were imposed by these environmental conditions and the short-day adaptation of the initial genetic material available. The variety Pampeana INTA may be used to demonstrate the potential of a classical pedigree breeding scheme conducted at INTA. Pampeana INTA is moderately resistant to late blight, it is resistant to several virus diseases and produces round tubers of high dry-matter content and high mashing and baking quality. This variety was released in 1985 and since then it has been used as a progenitor of both resistance and quality traits, and good market acceptance. The diploid programme at INTA, Argentina Wild species have been exploited in the breeding programme for several decades. Three species are used at present (Capezio et al., 1997; Huarte et al., 1997): S. microdontum, S. commersonii and S. chacoense have been extensively screened for the presence of R genes and assessed for their level of horizontal, non-R-gene, resistance. Within these sources, intraspecic cross populations segregating for horizontal resistance to late blight have been obtained, and molecular markers are being identied to assess horizontal resistance (Micheletto et al., 1997). Crosses of selected wild species clones to S. tuberosum dihaploids are made and resulting hybrids tested for resistance. Superior diploid hybrids are then used in 4x 2x interploidy crosses, to select tetraploid progeny for their direct use as cultivars or progenitors in the tetraploid programme. Diploid segregating hybrid progeny, using Argentinian wild species as sources of resistance, are also developed for experimentation with molecular-marker-assisted selection for late-blight resistance, using random amplied polymorphic DNA (RAPD), amplied fragment length polymorphism (AFLP) and microsatellite (SSR) markers. The efciency of the initial markers to detect resistance loci could be diminished, as the markers may not be perfectly associated with the loci determining horizontal resistance. Therefore, markers will be modied through genetic engineering. Wild species progenitors possessing true quantitative resistance and no R genes conferring race-specic resistance are being selected for further molecular tagging of resistance genes. Lessons learnt from the Argentinian programmes The approach chosen to select resistant genotypes to late blight initially depended on eld selection under natural infection on favourable environments for the development of the disease (Tucumn and Balcarce). Out of that simple scheme, an outstanding variety like Pampeana INTA was obtained. Later on, in order to improve the frequency of favourable genes and the efciency of selection, articial inoculation, together with the introduction of CIP germplasm, was carried out. The rst population introduced from CIP bearing R genes demonstrated the need to work in a background lacking those genes, considering the instability of the resistance shown on many of those clones. On the other hand, several of these clones showed a quite stable performance, indicating some degree of durable resistance. This material has potential for direct use as varieties and as progenitors. Still, there is a need for shortening the growing period and improving the tuber aspect for the local market requirements.

394

B.R. Trognitz et al.

The approach taken with population B, under long days, is showing its advantages: a high level of partial resistance is present, and good tuber aspect and shorter growth cycles are found at increased frequencies. Consideration of genotype environment interactions is important if this material is to be used in other regions or programmes: multisite testing will be necessary in order to select the more stable genotypes. Also crossing to S. tuberosum clones and cultivars with no R genes is carried on as a way to improve adaptation. Genotype environment interactions have proven to be quite important, particularly for the resistance that relies on the existing R genes. On the diploid side, the Argentine species have shown a high level of partial resistance together with the presence of R genes.

International cooperation to investigate the genotype quantitative resistance to late blight

environment interaction for

At the initial planning meeting of GILB, the GE interaction was given a high priority as the success of resistance breeding at one location could not be extrapolated to other locations without any further experiments. One previous study by Parker et al. (1992) indicated that the variety Alpha, horizontally resistant in Europe and North America, was susceptible to late blight in Toluca, Mexico. In response to the lack of more information on the stability of resistance across environments, CIP and a group of eight collaborators carried out a 3-year GE trial series to study a standard set of varieties at several locations worldwide and with varying pathogenic strains. The countries and institutions are: Canada (Agriculture Canada), the United States (Cornell and Washington State University), Denmark (LKF), Ecuador (CIP), Argentina (INTA), Scotland (Scottish Crop Research Institute SCRI), The Netherlands (CPRO), and two sites in France (Institut National de la Recherche Agronomique INRA). These locations represent nine sites with long-day and one short-day site (CIP, Ecuador). The data obtained indicate that interaction between quantitatively resistant genotypes and the environment is not signicant (Forbes, 1999). Tropically adapted, resistant varieties, such as Cruza-148, also were resistant in the temperate zone, and Torridon, a resistant variety from Scotland, was resistant at CIPEcuador. Therefore, although the absolute level of resistance expressed by a variety may depend on the specic local conditions, this variety can be predicted to be relatively more resistant than a susceptible variety regardless of where it is grown. This rst international experiment was performed with unrelated, clonal potato varieties, thus it investigates the interaction of different environmental settings with xed genotypes. Breeders would, however, like to know the relative contribution to this interaction of resistance alleles that may act in dominant and additive ways. Allelic interactions can be studied ideally in segregating biparental cross progeny, and diploid materials would facilitate the genetic analysis of those alleles. Therefore, a GE experiment using a diploid progeny, denominated PD, from a cross of a horizontally resistant S. phureja selection (P) and a susceptible potato dihaploid (D), will be carried out. The PD progeny of 246 individuals will be tested in the eld by two institutions in three contrasting environments: CIP (the central Peruvian Andes), BAZ (northeastern Germany) and CIP (Lembang, Indonesia). A genetic map of the PD population is

Improving Potato Resistance to Disease

395

currently being developed (Ghislain and Trognitz, 1996), and molecular markers of resistance alleles will help to determine relative contributions of alleles to the GE interaction.

Conclusion and Recommendations

The development of durable resistance to late blight in potato has been one of the most challenging tasks in the history of plant breeding. Recurring phases of success through the introduction of resistance from diverse genetic resources and failure due to the pathogens genetic exibility to adapt to its hosts have been experienced. International collaboration uniting breeders, pathologists, physiologists and molecular biologists has proved to be crucial for obtaining new insights, as the diversity of plantpathogen interactions have to be studied in a wide range of agroclimatic settings with their specic P. infestans strains.

References
Abad, Z.G., Abad, J.A., Fernandez-Northcote, E.N. and Ochoa, C. (1995) Host range of Phytophthora infestans in central Peru and in the world since 1840s. Resistance on wild tomatoes. Phytopathology 85, 1173. Bamberg, J.B., Longtine, C.A. and Radcliffe, E.B. (1996) Fine screening Solanum (potato) germplasm accessions for resistance to colorado potato beetle. American Potato Journal 73, 211223. Black, W. (1945) Inheritance of resistance to blight (Phytophthora infestans) in potatoes: unbalanced segregations. Proceedings of the Royal Society of Edinburgh, section B (Biology), vol. LXII, part 2, no. 20, pp. 171181. Capezio, S., van Damme, M. and Huarte, M. (1997) Evaluacin de resistencia a tizn tardo en follaje y en tubrculos de papa. In: IX Congreso Latinoamericano de Fitopatologa, 1217 October 1997, Montevideo, Uruguay. Colon, L.T., Turkensteen, L.J., Prummel, W., Budding, D.J. and Hoogendoorn, J. (1995) Durable resistance to late blight (Phytophthora infestans) in old potato cultivars. European Journal of Plant Pathology 101, 387397. Cox, A.E. and Large, E.C. (1960) Potato Blight Epidemics Throughout the World. Agriculture Handbook 174, USDA-ARS, Washington, DC. Darsow, U., Gbel, S., Gtz, E., Oertel, H. and Schler, K. (1987) R-Gene und relative Resistenz der Kartoffelknolle gegen Phytophthora infestans (Mont.) de Bary. Archiv fr Zchtungsforschung 17, 387397. Drenth, A., Tas, I.C.Q. and Govers, F. (1994) DNA ngerprinting uncovers a new sexually reproducing population of Phytophthora infestans in the Netherlands. European Journal of Plant Pathology 100, 97107. Flor, H.H. (1956) The complementary genetic systems in ax and ax rust. Advances in Genetics 8, 2954. Forbes, G. (1999) Update on the genotype by environment (G E) study. GILB Newsletter 8, 24. French, E.R. and Mackay, G.R. (1996) Enhancing the Global Late Blight Network. Report of the Project Design Meeting on the Global Initiative on Late Blight. CIP, Lima, Peru, March 1720, 1996.

396

B.R. Trognitz et al. Fry, W.E. and Goodwin, S.B. (1997) Resurgence of the Irish potato famine fungus. Bioscience 47, 363371. Fry, W.E., Goodwin, S.B., Dyer, A.T., Matuszak, J.M., Drenth, A., Tooley, P.W., Sujkowski, L.S., Koh, Y.J., Cohen, B.A., Spielman, L.J., Deahl, K.L., Inglis, D.A. and Sandlan, K.P. (1993) Historical and recent migrations of Phytophthora infestans: chronology, pathways, and implications. Plant Disease 77, 653661. Gallegly, M.E. and Galindo, J. (1958) Mating types and oospores of Phytophthora infestans in nature in Mexico. Phytopathology 48, 274277. Gebhardt, C. (1994) RFLP mapping in potato of qualitative and quantitative genetic loci conferring quantitative resistance to potato pathogens. American Potato Journal 71, 339345. Ghislain, M. and Trognitz, B.R. (1996) Molecular breeding for late blight resistance in potatoes. CIP Circular 22(2), 1015. Goodwin, S.B. and Drenth, A. (1997) Origin of the A2 mating type of Phytophthora infestans outside Mexico. Phytopathology 87, 992999. Goodwin, S.B., Cohen, B.A. and Fry, W.E. (1994a) Panglobal distribution of a single clonal lineage of the Irish potato famine fungus. Proceedings of the National Academy of Sciences USA 91, 11,59111,595. Goodwin, S.B., Sujkowski, L.S. and Fry, W.E. (1994b) Metalaxyl-resistant clonal genotypes of Phytophthora infestans in the United States and Canada were probably introduced from northwestern Mexico. Phytopathology 84, 1079. Hermsen, J.G. (1994) Introgression of genes from wild species, including molecular and cellular approaches. In: Bradshaw, J.E. and Mackay, G.R. (eds) Potato Genetics. Scottish Crop Research Institute, Invergowrie, pp. 515538. Huarte, M.A., Butzonitch, I.P. and Mendiburu, A.O. (1990) Mejoramiento para resistencia a virus en el programa Argentino de papa. In: Hidalgo, O.A. and Rincn, H. (eds) Avances en el Mejoramiento Gentico de la Papa en los pases del Cono Sur. CIP, Lima, pp. 8594. Huarte, M.A., Mendiburu, A.O. and Mendoza, H.A. (1994) INTACIP research in potato breeding: a model for general application. CIP Circular 20(2), 27. Huarte, M., Butzonitch, I.P., van Damme, M., Colavita, M., Capezio, S., Micheletto, S., Cacace, E. and Clausen, A. (1997) Experiencias, dicultades y logros obtenidos en la bsqueda de resistencia durable a enfermedades. In: Danial, D. (ed.) Primer Taller de PREDUZA en Resistencia Duradera en Cultivos Altos en la Zona Andina. Quito, Ecuador, 2224 September 1997. pp. 154163. International Potato Centre (1997) CIP in 1996. In: International Potato Centre Annual Report. CIP, Lima, pp. 1820. James, C.W. (1981) Estimated losses of crops from plant pathogens. In: Pimentel, D. (ed.) Handbook of Pest Management in Agriculture, pp. 8094. Johnson, R. (1988) Durable resistance to yellow (stripe) rust in wheat and its implications in plant breeding. In: Breeding Strategies for Resistance to the Rusts of Wheat. CIMMYT, Mexico, pp. 6375. Jorgensen, J.H. (1994) Genetics of powdery mildew resistance in barley. Critical Reviews in Plant Sciences 13, 97119. Landeo, J.A. and Turkensteen, L. (1989) Assessment of partial resistance to late blight (Phytophthora infestans) of major genes in potato. American Potato Journal 66, 530. Lehmann, H. (1938) Geschichte und Ergebnisse der Versuche zur Zchtung krautfulewiderstandsfhiger Kartoffeln. Der Zchter 10, 7280. Malcolmson, J.F. and Black, W. (1966) New R-genes in Solanum demissum Lindl. and their complementary races of Phytophthora infestans (Mont.) de Bary. Euphytica 15, 199203. Mendiburu, A.O. and Huarte, M.A. (1987) Potato breeding in Latin America with special reference to Argentina. In: Recent and Future Development of the Potato in the World. Amsterdam, April 59, 1987. Acta Horticulturae 213, 4350.

Improving Potato Resistance to Disease

397

Micheletto, S., van Damme, M., Clausen, A. and Huarte, M. (1997) Evaluacin de tres especies silvestres de papa para resistencia a Phytophthora infestans. In: Primer Taller Internacional sobre Recursos Fitogenticos del Noroeste Argentino. 1923 May, 1997. Salta, Argentina. Michelmore, R. (1995) Molecular approaches to manipulation of disease resistance genes. Annual Review of Phytopathology 15, 393427. Mitchellolds, T. (1996) Genetic constraints on life-history evolution: quantitative trait loci inuencing growth and owering in Arabidopsis thaliana. Evolution 50, 140145. Mller, K.O. (1951) ber die Herkunft der W-Sorten, ihre Entwicklungsgeschichte und ihre bisherige Nutzung in der praktischen Kartoffelzchtung. Zeitschrift fr Panzenzchtung 29, 366387. Parker, J.M., Thurston, H.D., Villarreal-Gonzalez, M.J. and Fry, W.E. (1992) Stability of disease expression in the potato late blight pathosystem: a preliminary eld study. American Potato Journal 69, 635644. Parlevliet, J.E. (1979) Components of resistance that reduce the rate of epidemic development. Annual Review of Phytopathology 17, 203222. Parlevliet, J.E. (1993) What is durable resistance, a general outline. In: Jacobs, Th. and Parlevliet, J.E. (eds) Durability of Disease Resistance. Kluwer Academic Publishers, Dordrecht, pp. 2339. Plaisted, R.L. (1987) Advances and limitations in the utilization of Neotuberosum in potato breeding. In: Jellis, G.J. and Richardson, D.E. (eds) The Production of New Potato Varieties. Cambridge University Press, Cambridge, pp. 186196. Ristaino, J.B. (1998) The importance of archival and herbarium materials in understanding the role of oospores in late blight epidemics of the past. Phytopathology 88, 11201130. Roelfs, A.P. (1988) Resistance to leaf and stem rusts in wheat. In: Breeding Strategies for Resistance to the Rusts of Wheat. CIMMYT, Mexico, pp. 1022. Ross, H. (1986) Potato Breeding Problems and Perspectives. Advances in Plant Breeding, Vol. 13. Paul Parey, Berlin, Hamburg. Schick, R. (1932) ber das Verhalten von Solanum demissum, Solanum tuberosum und ihren Bastarden gegenber verschiedenen Herknften von Phytophthora infestans. (Vorluge Mitteilung zur Frage der biologischen Spezialisierung von Phytophthora infestans). Der Zchter 4, 233237. Schick, R. and Schick, E. (1959) Die Differenzierung der verschiedenen Rassen der Phytophthora infestans auf Smlingen von S. demissum (Lindl.) und S. stoloniferum (Schlecht. et Bouche). Der Zchter 29, 220225. Schmalz, H. (1980) Panzenzchtung. Deutscher Landwirtschaftsverlag, Berlin. Simmonds, N.W. and Malcolmson, J.F. (1967) Resistance to late blight in Andigena potatoes. European Potato Journal 10, 161166. Steffenson, B.J., Hayes, P.M. and Kleinhofs, A. (1996) Genetics of seedling and adult plant resistance to net blotch (Pyrenophora teres f. teres) and spot blotch (Cochliobulus sativus) in barley. Theoretical and Applied Genetics 92, 552558. Trognitz, B., Forbes, G. and Hardy, B. (1995) Resistance to late blight of potato from wild species. CIP Circular 21(1), 69. Trognitz, B.R., Ghislain, M., Crissman, C. and Hardy, B. (1996) Breeding potatoes with durable resistance to late blight using novel sources of resistance and nonconventional methods of selection. CIP Circular 22(1), 69. Trognitz, B., Ghislain, M., Forbes, G., Oyarzun, P., Eslava, M., Herrera, R., Portal, L., Ramn, P. and Chacn, G. (1997) Evaluation of late blight resistance in populations of diploid hybrids for genetic mapping. In: Program Report 199596. CIP, Lima, pp. 132137. Umaerus, V. (1970) Studies on eld resistance to Phytophthora infestans. 5. Mechanisms of resistance and applications to potato breeding. Zeitschrift fr Panzenzchtung 63, 123. Veldboom, L.R., Lee, M. and Woodman, L. (1994) Molecular-marker-facilitated studies in an

398

B.R. Trognitz et al. elite maize population. I. Linkage analysis and determination of QTLs for morphological traits. Theoretical Applied Genetics 88, 716. Wolfe, M.S. and Gessler, C. (1992) The use of resistance genes in breeding: epidemiological considerations. In: Boller, T. and Meins, F. Jr (eds) Genes Involved in Plant Defense. Springer, Vienna, pp. 323.

25

A Mexican Bean Breeding Programme for Comprehensive Horizontal Resistance to all Locally Important Pests and Diseases

R. GARCIA ESPINOSA1, R.A. ROBINSON2, P. RAMIREZ VALLEJO1, F. CASTILLO GONZALEZ1 AND F. ROMERO ROSALES1
1Colegio

de Postgraduados, IFIT, Texcoco, Mxico; 2Retired Crop Scientist, 445 Prevost Lane, Fergus, Ontario, Canada

Introduction
Mexico is one of the centres of origin and domestication of the common bean (Phaseolus vulgaris). This country accordingly has a wide genetic variability in beans as well as in bean parasites. Until recently, the local landraces were able to withstand the impact of local pests and diseases. This stable situation changed because of an increase in the areas cultivated by bean coupled with the introduction of intensive cropping systems employing high-yielding pure lines protected either with vertical resistances (i.e. gene-for-gene relationships) or with pesticides (Garca, 1987b; Prez et al., 1995). Other possible factors include the global distribution of pathogens, or the appearance of new races (pathotypes) of old pathogens, perhaps due to the unrestricted movement of germplasm within the country and among different countries (Gram, 1960; Thurston, 1984; Garca, 1987a). The pest and disease populations have now increased to the point where not even the local landraces could be cultivated without crop protection chemicals. As a result of these pest and disease problems, among others, the crop is no longer protable and it is being abandoned as a commercial crop by farmers in the Mixteca region of Mexico (Garca, 1987b). The Mixteca consumers have a strong preference for the local landraces over the considerably cheaper beans now being imported from other countries under the North American Free Trade Agreement (NAFTA). The challenge was to improve both the resistance and the yield of the local landraces, with a view to offer the Mixteca farmers varieties derived from their own local landraces, but with high enough levels of horizontal resistance to avoid the use of pesticides, and thereby to recover the commercial value of the local landraces. This would
IPGRI/FAO 2001. Broadening the Genetic Base of Crop Production (eds H.D. Cooper, C. Spillane and T. Hodgkin)

399

400

R. Garcia Espinosa et al.

restore the overall stability of bean cultivation, provide the preferred qualities, eliminate the use of pesticides and maintain genetic diversity.

Objective of the Programme

Most plant breeding efforts for resistance to crop parasites have involved single-gene vertical resistance, obtained by gene-transfer breeding techniques (Robinson, 1976). Vertical resistance has two important advantages: it usually provides a complete protection against the parasite in question, and it functions over a wide geographical range. However, it has a number of disadvantages, the most important being its liability to fail on the appearance of a new, matching strain of the parasite (Robinson, 1987). Horizontal resistance is usually polygenically inherited and its accumulation requires the entirely different breeding technique of recurrent mass selection. Its chief advantage is that it is durable, and it does not fail like vertical resistance. Its main disadvantage is that it is quantitative and it may not always provide a complete protection. Furthermore, the effectiveness of a horizontal resistance, in any one cultivar, is usually limited to a single agroecosystem, because the epidemiological competence of parasites varies greatly from one agroecosystem to another. A cultivar that is in perfect balance within one agroecosystem will then have too much resistance to some parasites, and too little to others, when cultivated in a different agroecosystem (Robinson, 1996). Thus breeding programmes for horizontal resistance must aim for specic adaptation, with selection carried out in the target area. A comprehensive horizontal resistance bean breeding programme was initiated in 1991 in the Mixteca region in Mexico. The project was conducted by a multidisciplinary team from the Colegio de Postgraduados, Mexico, in cooperation with the University of Guelph, Canada. The objective of the programme was to test the possibility of obtaining both high levels of horizontal resistance to all the locally important bean parasites and the high yields attained by commercial cultivars, while retaining the culinary qualities of the Mixteca landraces

Breeding Strategy
This horizontal resistance bean breeding programme was directed against the two main pathological problems in the region, bean common mosaic virus (BCMV) and bean common blight (Xanthomonas campestris pv. phaseoli), and the main entomological problem of white y Bemisia sp. and Trialeurodes sp., as well as all other parasites that are unavoidably present during the screening process. These less-important parasites included several soil-borne plant pathogens (fungi such as Pythium spp., Rhizoctonia solani, Macrophomina phaseolina and Fusarium spp. among others) and several insect pests. For a successful horizontal resistance breeding programme, it is necessary that the many polygenes responsible for the various quantitative resistances are present in the original parent population, and this necessitates a broad genetic base of the starting material. The breeding programme was initiated with landraces that have a high

Bean Breeding for Resistance to Pests and Diseases

401

inherent variability, making it unnecessary to introduce exotic germplasm. Out of 70 local landraces collected from the highlands of the Mixteca region, seven were chosen as parent populations. These were selected specically for their susceptibility to a designated pathotype of each of the two main diseases: bacterial blight and BCMV. Since the same designated pathotypes were used in all subsequent screening populations, this ensured that no vertical resistance genes were functioning during the breeding process. These seven parents were crossed in a half-diallel cross after several (more than three) generations of selng, and each cycle of recurrent mass selection was subjected to late selection. Late selection is necessary to eliminate any effects of heterosis (hybrid vigour) and to ensure that the effects of any recessive polygenes for horizontal resistance are expressed. All screening was conducted under eld conditions in the Mixteca region, during the normal growing season, and according to the local farming system. Strong selection pressure for resistance was applied by individually inoculating each plant in a population of 6000 plants with bacterial blight and BCMV. The best 1% of plants were kept as parents for the next cycle of recurrent mass selection. The primary screening criterion was yield. This was considered the best proxy for horizontal resistance under the conditions employed (in which every plant had been inoculated with a designated pathotype to ensure that no vertical resistance was functioning). The screening populations were free of all crop protection chemicals. Secondary selection criteria included earliness, growth habit, pod length, and seed size, colour and brightness.

The breeding cycle The breeding cycle, dened as all the operations that occur between one controlled pollination and the next, included cross-pollination, followed by two or three generations of self-pollination, and family selection in the eld. To complete a breeding cycle in the shortest time, the differing climates of various regions within Mexico were exploited. However, screening was conducted only in the area of future cultivation, during the summer, in the Mixteca region. Crossing through hand-pollination was conducted during the winter, at high altitude. This heavy load of very delicate work was done in the greenhouse at the Colegio de Postgraduados by specially trained students. Selng was conducted in the eld, both in the summer at high altitude, and during the winter at low altitude at Veracruz.

Selection pressures The selection pressures were high in order to maximize the rate of genetic advance. The usual selection coefcient was one plant in 100. Although the selection of parents for the next breeding cycle was done in a single location, the selection of genotypes to be released as new cultivars was based on eld evaluations made during 2 years in at least three locations, including the site of future cultivation. Two experiments were established in each location to identify the best genotypes grown with and without articial fertilizers.

402

R. Garcia Espinosa et al.

Lines Released and Yields

At the time of writing (June 1999), the programme had produced four new lines, derived from only the third breeding cycle. These are currently in the process of ofcial registration. Out of the 33 parents of the fourth breeding cycle, ve new lines were selected as elite lines to develop into new varieties. The 33 parents of the fourth breeding cycle were used in the 1997 crossing process that produced hundreds of new genotypes, out of which the 33 parents of the fth breeding cycle were selected; out of those, another ve elite lines will be selected for ofcial registration. These 33 parents were also used for further crosses in this ongoing programme. The new lines have been tested under various ecological conditions by farmers in the Mixteca region. This testing had the additional purpose of bulking seed but, in some cases, the production has been so successful that the farmers refused to sell back the seed of some of the lines, which under certain eld conditions have yielded as much as 1800 kg ha 1 (the average yield in the region, in a good year is 600 kg ha 1, and in 1998 the local varieties yielded around 200 kg ha 1).

Yields Table 25.1 shows the gains obtained in the third and fourth breeding cycles in an experiment conducted in the Mixteca region, in which 64 genotypes were included (six original parents, 15 parents of the third and 33 of the fourth breeding cycles, along with some regional and commercial varieties). The experiment was established in 1997 and 1998 under a randomized block design, with three replicates and complete competition among plants. No fertilizers or chemicals for protection were applied.

The accumulation of horizontal resistance It should be borne in mind that all the screenings were conducted under eld conditions and, therefore, under the selection pressure of all locally important parasites. The data discussed below are mainly related to BCMV, although some critical evaluations are described concerning soil-borne diseases. Data concerning bacterial common blight are currently under analysis.

Table 25.1. Yield gains for selection in the third and fourth breeding cycles (Tepexi de Rodriguez, Puebla, 1997). % gain over original parents 152.9 136.6 100.0 % gain of fourth cycle over third cycle 11.9

Parents Fourth cycle Third cycle Original parents

Yield (kg ha 1) 650.7 581.30 425.58

Bean Breeding for Resistance to Pests and Diseases

403

Even though the programme started precisely with seven highly susceptible parents (in order to avoid the possibility of introducing vertical resistance to either of the two diseases targeted), the 1998 results clearly demonstrate the accumulation of resistance in some of the advanced genotypes to both BCMV and soil-borne diseases. Highest yields (signicantly different from the rest of genotypes) were obtained by two of the selected parents of the fourth breeding cycle: key numbers 507 and 553, with yields of 865.46 and 854.74 kg ha 1, respectively. Neither genotype showed severe symptoms of BCMV (Fig. 25.1) and only 38% and 50%, respectively, showed mild symptoms (Fig. 25.2). Lethal diseases, as sometimes BCMV can be for beans, exert very strong selection pressure among plant populations. It has been argued that horizontal resistance against such diseases is of limited value if the crop carries natural genetic exibility such as the case of allogamous crops like maize. However, this is not a problem for autogamous species, such as common beans. For this reason, emphasis has been placed on monitoring the incidence of the severe symptoms of BCMV. Nevertheless, even when mild symptoms are considered, it is still possible to observe the accumulation of horizontal resistance against this disease, as shown by the negative correlation between the higher yield of some of the advanced genotypes of this programme and their lower incidence of both types of symptoms (Fig. 25.2). Since the virus isolate that has been used throughout this programme (the designated pathotype) is capable of producing a continuum of symptoms among the progeny derived from the original parents, from mild, to severe, to the death of plants, it is clear that the most advanced materials of higher yields are showing horizontal resistance when inoculated with the designated pathotype, i.e. produce only mild symptoms (Fig. 25.2).

Fig. 25.1. Yieldincidence relationship of severe BCMV symptoms (S2) among the 64 genotypes tested in Tepexi de Rodrguez, Puebla, 1998. Note: % incidence is the proportion of plants affected; each point is the mean of three replicates.
*P4C = fourth breeding cycle progenitors; P3C = third breeding cycle progenitors; PO = original parents; VC = commercial varieties.

404

R. Garcia Espinosa et al.

Fig. 25.2. Yielddisease incidence relationship of BCMV, severe symptoms (S2) and severe + mild symptoms (GL) showing the ten best and worst yielding among the included 64 genotypes. Tepexi de Rodrguez, Puebla, 1998.

There is an evident accumulation of horizontal resistance to BCMV when the incidence of both mild and severe symptoms is compared among the original parents and their progeny. The incidence was higher for the original parents than for the parents of the third and fourth breeding cycles. However, the incidence of both symptoms was even lower for those of commercial and regional varieties (Fig. 25.3), but neither group showed the highest yields and some varieties of these groups showed susceptibility to other diseases and pest problems, especially soil-borne diseases that render them
80 60 40 20 0 Original parents Parents 3rd breeding cycle Parents 4th breeding cycle Regional varieties Commercial varieties S2 25.50 14.50 11.83 2.50 2.83 S1+S2 72.83 69.00 52.16 53.00 34.00

Fig. 25.3. Average percentage incidence of severe symptoms (S2) and mild + severe (S1+S2) symptoms of the six best genotypes among the advanced ones (those of higher yields) and the six and four commercial and regional varieties, respectively.

Bean Breeding for Resistance to Pests and Diseases

405

highly vulnerable in the Mixteca region. In fact, neither of these varieties would have been suitable as original parents in this programme because of the possibility of introducing monogenes of resistance as no information was available regarding their reaction to the inoculation of the pathotypes (BCMV and common blight bacterium). Additionally, the commercial varieties are not accepted by the Mixteca farmers. One of the clearest improvements observed recently is the accumulation of resistance to soil-borne diseases. In eld experiments conducted in 1997, it was easy to identify the plots of some of the commercial varieties because they were destroyed by these kinds of diseases, whereas the most advanced materials of the horizontal resistance breeding programme appeared perfectly healthy. In 1998 eld experiments, signicantly higher plant survival was observed for the six best-yielding lines of the parents of the fourth breeding cycle, followed by the parents of the third breeding cycle (Fig. 25.4). Although these diseases were not originally included as targets for the project, they have been present throughout the breeding process, acting as selective pressure on the genetic variation of the bean populations used in this programme.

Farmer participation Many farmers expressed interest in cooperating with the programme, particularly as they are allowed to keep any lines they please. This widespread farmer participation ensures the adoption of the lines best suited to their needs. Because of the relative simplicity and low cost of the breeding process for quantitative resistance compared with that for single-gene, vertical resistance, this method offers new potential for strengthening farmer participation in plant breeding. In the future, the farmers, organized in breeding clubs (Robinson, 1996), could conduct their own breeding programmes, with

100 80 % plant survival 60 40 20 0 Global average Average of best* P3C** P4C PO VC VR

Fig. 25.4. Percentage of plant survival, after the initial attack of soil-borne diseases, showing the levels of resistance for both the total of genotypes per group and the six best-yielding lines, in the Tepexi de Rodrguez, 64 genotypes experiment, summer 1998.
*Global average represents the average percentage of germination of all genotypes belonging to each group, whereas average of best is the average based on the six highest-yielding genotypes of each group. **P4C = fourth breeding cycle progenitors; P3C = third breeding cycle progenitors; PO = original parents; VC = commercial varieties; VR = regional varieties.

406

R. Garcia Espinosa et al.

assistance from scientists limited to making crosses and the early stages of selection. The total breeding activity could thus be greatly increased. Unlike breeding for qualitative (single-gene) resistance, breeding for quantitative resistance is a continuous process, with each breeding cycle providing new and better cultivars (Robinson, 1996). The later stages of horizontal resistance breeding in the hands of farmers themselves could produce new high-yielding, high-quality lines capable of withstanding all the locally important parasites. Advanced lines should be tested by a large number farmers under diverse agroecological conditions in the variable, mountainous topography of the Mixteca.

Conclusions
It must be emphasized that this is an interim report on a programme that is far from complete. Nevertheless, the genetic advances obtained in only a few breeding cycles (averaging 18% per cycle in the rst three cycles) are greater than anticipated. They more than justify the testing of this method in other crops and in other areas. The results obtained so far provide evidence of the accumulation of horizontal resistance. They suggest that it is feasible to obtain high-yielding, high-quality cultivars that have sufcient durable resistance to all locally important parasites to eliminate the need for crop protection chemicals. This programme demonstrates that breeding for horizontal resistance is quite simple, and could be undertaken by farmers and amateur breeders with only a minimum of assistance from scientists.

Acknowledgement
For the development of our programme we have had the generous support of IDRC of Canada.

References
Garca, E.R. (1987a) Importancia de la topatologa tropical. In: Garca, E.R. (ed.) Taller de topatologa tropical. Sociedad Mexicana de Fitopatologa, Texcoco, Mexico, pp. 119. Garca, E.R. (1987b) Reexiones sobre la importancia de las enfermedades en cultivos agrcolas y su control, en la Mixteca Oaxaquea. In: Muoz, O.A. and Dimas, Ch.B. (eds) Como Aumentar la Produccion Agropecuaria y Forestal en la Region Mixteca Oaxaquea, Sarh. Gobierno del Edo. de Oaxaca, pp. 421429. Gram, E. (1960) Quarantines. In: Horsfall, J.G. and Dimond, A.E. (eds) Plant Pathology, an Advanced Treatise. Vol. 3. Academic Press, New York, pp. 313356. Prez, M.J., Ferrera, C.R. and Garca, E.R. (eds) (1995) Diversidad gentica y patologa del frijol. Colegio de Postgraduados en Ciencias Agrcolas, Texcoco, Mexico. Robinson, R.A. (1976) Plant Pathosystems. Springer Verlag, New York. Robinson, R.A. (1987) Host Management in Crop Pathosystems. MacMillan Publishing Co., New York. Robinson, R.A. (1996) Return to Resistance. AgAcces and IDRC, Davis, California, and Ottawa. Thurston, H.D. (1984) Tropical Plant Diseases. The American Phytopathological Society, St Paul, Minnesota.

26

The Impact of Decentralized and Participatory Plant Breeding on the Genetic Base of Crops

J.R. WITCOMBE
Centre for Arid Zone Studies, University of Wales, Bangor, Gwynedd LL57 2UW, UK

Introduction
The use of decentralized methods in crop improvement impacts on the genetic base of crops as deployed in farmers elds, particularly when participatory methods are employed. Participatory plant breeding (PPB) is increasingly being used in both inbred and open-pollinated crop species (see Sperling et al., Chapter 27, this volume). Two methods, participatory varietal selection (PVS) and participatory plant breeding (PPB), were dened and compared for their impact on varietal biodiversity in farmers elds by Witcombe et al. (1996). The effects of PVS were considered unpredictable as it could result in either an increase or a reduction in varietal diversity at a local level. The outcome depends on the original level of crop diversity and the genetic diversity among the introduced varieties that farmers adopt. The manner of replacement of the existing cultivars either partial or complete is also important. In contrast, PPB methods were considered to have a generally favourable impact on varietal biodiversity and could be considered a dynamic form of in situ genetic conservation. PPB enhances the effects of traditional farmers practices, that maintain variability within and between landraces, by improving the efciency of selection and by increasing the variability on which this selection acts (Witcombe et al., 1996). Does PPB, when considered either on a local level or on the wider national or international level, tend to narrow or broaden the genetic base of crops? To answer this question, farmer participation has to be considered in relation to decentralization, and distinctions made between the effects of different types of participation, consultative or collaborative, dened by Biggs (1989). This is considered by using examples from rice, wheat and maize, the most widely grown cereals in the developing world.

IPGRI/FAO 2001. Broadening the Genetic Base of Crop Production (eds H.D. Cooper, C. Spillane and T. Hodgkin)

407

408

J.R Witcombe

Decentralization Versus Participation

Decentralization in plant breeding is scale sensitive. It can be from an international to a national level, from a national to a regional level, or from a regional level to ever smaller and ever more precisely dened target zones. The most complete decentralization is when farmers participate in the breeding process in their own elds (Ceccarelli et al., 1994). Hence participation and decentralization are overlapping concepts and the methodological difculties determining their relative importance are considerable or insurmountable (Witcombe, 1996). Most comparisons were considered too difcult to make because decentralization is nearly always confounded with increased farmer participation. Inevitably, in any decentralized breeding programme, the smaller the target area the more likely it is that the farmers role in the breeding process will increase, even if that role is only consultative. Hence, although this chapter is largely concerned with the effects of farmer-participatory approaches to plant breeding on base-broadening, most of the conclusions also apply to decentralized breeding.

Using Wide Adaptation in Centralized Plant Breeding

Plant breeding can exploit wide adaptation by selecting varieties that yield well over large geographical areas such as the six mega-environments dened by the International Maize and Wheat Improvement Centre (CIMMYT) for spring wheat (Rajaram et al., 1995). Mega-environments are broad (frequently discontinuous transcontinental) areas, characterized by similar biotic and abiotic stresses, cropping system requirements and consumer preferences. Although the spring wheat breeding programme is mainly conducted in only two research stations in Mexico, varieties bred by the programme are tested in multinational, multilocational trials in the mega-environments. Targeting the mega-environments in this way provides a degree of decentralization whilst having wide adaptation as a major breeding objective. A similar approach is followed by other Consultative Group on International Agricultural Research (CGIAR) centres such as the International Rice Research Institute (IRRI), which adopts an ecoregional approach where regions are targeted within the main rice ecosystems (Fischer, 1996). For example, there are nurseries targeted at seven irrigated, three rainfed, one deepwater and two upland ecosystems (Chaudhary and Ahn, 1996). This approach of addressing broadly dened environments has been successful and has led to the large-scale adoption of widely adapted varieties in agroecological regions (e.g. Smale et al., 1996; Witcombe, 1999). Such centralized systems do not have to lead to a narrowing of the genetic base because a genetically diverse set of germplasm, ideally derived from a diverse set of parents and crosses, can be distributed for international testing. However, as the same genotypes are distributed to many countries, and it is common to distribute nearinbred lines, several national programmes may select and release the same one. Moreover, in most developing countries there are very strong lters that prevent much of this varietal diversity from reaching farmers a centralized system of national multilocational trials and commonly a policy of recommending only a few of the very best performing entries in these trials for ofcial release in the country (Witcombe et al., 1998). As a consequence, there has been a reduction in the varietal diversity deployed

Decentralized and Participatory Plant Breeding

409

in farmers elds (at least when weighted by the proportions each variety occupies). For example, rice varieties IR8, IR36 and IR64 occupied, or still occupy, large areas in several geographically distant countries over long periods of time. A few varieties of wheat, such as Sonalika, also dominate over large areas for long periods in countries such as India (Smale et al., 1996). One way of overcoming this widespread adoption of a few varieties is to decentralize breeding programmes to a more local level. Decentralized programmes will each produce their own, distinct germplasm and farmers should preferentially adopt the varieties bred for their target area. Such preferential adoption should be the common case since decentralized programmes are building on locally adapted germplasm, selecting in the target environment, and taking local farmers preferences into account.

Decentralization by International Breeding Programmes

Decentralization by rening target areas International Agricultural Research Centres (IARCs) distribute standard sets of genotypes to many national breeding programmes. These are called international nurseries or trials. For example, the International Spring Yield Wheat Nursery (ISWYN) of CIMMYT is distributed internationally and evaluated by many national research programmes. DeLacey et al. (1996) carried out a pattern analysis of the geographic locations in the ISWYN using the multilocational yield data of the entries in the nursery. A lack of genetic variation towards a stress, e.g. for drought or heat tolerance, among the varieties would mean that the pattern analysis would not be able to discriminate between regions that vary for this stress. However, there was sufcient genetic diversity in the trials for the pattern analysis, carried out on data on varietal phenotypes across locations, to reveal sub-groups of locations within the mega-environments. If all of these sub-environments are targeted with germplasm specically chosen for them, then such decentralization can lead to improved base-broadening.

Decentralization by targeting national programmes The barley breeding programme at the International Center for Agricultural Research in Dry Areas (ICARDA) targets individual countries in North Africa by making a different set of crosses for each target country. Since there tends to be a clear geographical distribution of variation within germplasm, the crosses for each target country involve landraces originating from it (Ceccarelli et al., 1994; Chapter 6, this volume). Hence, ICARDAs barley breeding programme is an example of a centrally coordinated programme where the decentralization strategy results in a deliberate maintenance of genetic diversity between the individual programmes. The extent of base-broadening of the crop in farmers elds between countries will depend on the extent of the differences among the germplasm targeted at the individual national programmes, and on what varieties from the programmes germplasm are adopted by farmers.

410

J.R Witcombe

Decentralization by shifting responsibilities When IARCs supply international nurseries or international trials, national programme breeders can depend on them as sources of new varietal variation for promotion to national-programme, multilocational trials. This sometimes has the unintended, but unfortunate, effect of de-emphasizing in national programmes the role of crossing and selection among segregating generations. Resources are limited in many national programmes and basing a breeding programme on nished, or nearly nished, varieties is cheap and has many successful precedents. Traxler and Pingali (1999) have analysed releases by national programmes in wheat and rice according to whether they are the result of crosses made by the national agricultural research system (NARS), or direct introductions from IRRI for rice or CIMMYT for wheat. In rice, the proportion of direct introductions from IRRI has fallen to only 9%. However, in wheat, introductions account for the majority of releases, and this proportion is increasing over time and now exceeds 50%. The reasons for this large difference have yet to be fully investigated, although Traxler and Pingali (1999) offer some explanations such as the greater importance of rice to many national programmes, and the greater diversity of rice ecosystems as compared with wheat. A desirable shift towards decentralization would be for IARCs to reduce the emphasis on international nurseries and trials and increase support to targeted crossing programmes in which national programme breeders are actively involved. Unfortunately, such a shift increases the demands upon national programmes at a time when many of them are reported by the International Service for National Agricultural Research (ISNAR) to be facing a real decline in resources. However, this change could be facilitated by the adoption of more participatory approaches, which rely less on expensive experimental infrastructure and increase the effectiveness of delivery. IARCs could supply germplasm from crosses targeted at agroecological regions within national programmes in which the minimum of selection has been made for example, advanced generations produced by single seed descent that, it is suggested below, are ideal for incorporation into PPB programmes. Depending on resources, such activities could replace, or be in addition to, some of the more traditional approaches.

Participatory Approaches to Decentralization by National Breeding Programmes

Decentralization a case study for maize in Gujarat In India, breeding programmes are decentralized to the state level, and often agroecological zones are targeted within the state. Lessons on base-broadening can be learnt from a case study of maize breeding in the state of Gujarat, India (J.R. Witcombe, A. Joshi and S.N. Goyal, in preparation). The breeding programme was already decentralized to the state level and to the rainfed agroecological zone. Greater farmer involvement was introduced using consultative methods to identify target traits, such as the need for very short duration, white-endosperm varieties that grow well under low

Decentralized and Participatory Plant Breeding

411

fertility and rainfed conditions. Collaborative breeding was used in some of the later generations, after random mating was completed, in which mass selection was carried out by farmers. However, since this was done on researchers land, and not on farmers elds, the usual benets of collaborative plant breeding increasing diversity on participating farmers elds were not apparent. A maize composite was created by crossing both white- and yellow-endosperm maize varieties. Only one of the parents was a local variety; all others were introduced into the area either from breeding programmes in distant states but also from those located within Gujarat and from bordering Madhya Pradesh. All of these varieties had been tested in PVS programmes in Gujarat, so it was known that all of them had at least some attributes that farmers liked and some adaptation to local conditions. Several varieties, both white and yellow grained, have been produced from this composite. One of them, GDRM 187, has a yield advantage over the local varieties of more than 20%, is considerably earlier to mature, and has superior grain quality. Farmers greatly prefer it over the several cultivars they are currently growing, and it is likely to become the predominant variety in the area. Hence, although there is possibly a huge broadening of the genetic base, because of the diverse ancestry of GDRM 187, compared with the local varieties described by Joshi and Witcombe (1996, 1998), its widespread adoption may well lead to a loss of genetic diversity in farmers elds. This loss of diversity needs to be dened since there are many types of diversity in agricultural systems (reviewed in Witcombe, 1999). In this case, the reduction will occur in the weighted diversity that takes into account both the genetic diversity between varieties grown by farmers, and the area that each cultivar occupies (Souza et al., 1994). This possible decline in weighted diversity is also a function of the existence of a range of diverse landraces in the target area. If GDRM 187 replaced a single, narrowly based variety, then its impact would be entirely benecial for the breadth of the genetic base of the crop in the eld. Moreover, the genetic base available to breeders and farmers in the state has increased. Although there were very important collaborative elements, the breeding of GDRM 187 is probably best described as consultative PPB. From a national viewpoint, consultative participation may promote diversity when used to identify parents and target traits in decentralized breeding programmes. It is less likely that individual breeding programmes will depend on the same germplasm: by consulting farmers on their needs for new varieties locally adapted germplasm will often be used in crosses, and the choices of exotic germplasm will also differ according to local needs. Locally adapted germplasm will also usually be locally distributed, although an exception, the rice variety Kalinga III, is discussed below.

Decentralization of plant breeding with collaborative participatory research Decentralization that employs collaborative techniques is one of the most certain ways to have benecial effects upon the genetic base because in the process of collaboration farmers grow a diverse range of novel germplasm on their elds. These benets will tend to be greater in the areas where the PPB programme is being conducted (Witcombe et al., 1996). Only a part of this diversity the most successful of the farmers selections will diffuse from the participating farmers, resulting in geographic

412

J.R Witcombe

patterns of diffusion from these foci as found by Huke and James (1969) and Witcombe et al. (1999). The extent to which diversity will increase is dependent on the participatory breeding method employed and whether the crop is an inbreeder or outbreeder. In outbreeding crops, such as maize, it is unlikely in a collaborative participatory programme that many composites can be made or selected by farmers in their elds. Indeed, it is probable, given the more limited resources of decentralized programmes, that most will only work on one or two composites at any one time. If collaborative methods prove possible and they are more difcult in outbreeders because effective progress from selection without undesirable inbreeding depression requires both a minimum population size and genetic isolation from other populations then base-broadening in the eld is possible. A composite can carry a large reservoir of variability, particularly if selection is minimized during the initial random mating generations. It can be given to farmers in a range of specic target environments (both physical and socioeconomic) and selection for local adaptation and local needs of farmers will create and maintain diversity between populations. However, it is difcult to generalize the ways in which different farmers select and maintain seed will be important in determining eventual changes in varietal diversity (e.g. Louette et al., 1997; see also Berthaud et al., Chapter 4, this volume). In inbreeding crops such as rice, wheat and barley PPB is not constrained by isolation distances and population sizes. Many methods are possible, depending on how early farmers are given seed in the procedure of generation advance after hybridization. Sthapit et al. (1996) gave F5 pedigree bulks (i.e. bulks derived from individual F2 plants) of rice to farmers located in two high-altitude villages in Nepal. They reported an increase in varietal biodiversity in these villages as several bulks were accepted by farmers. However, in subsequent studies (Joshi et al., unpublished) it was found that only one or two of the F5 pedigree bulks had persisted, so although there were long-term increases in local varietal biodiversity (new varieties were often additions to farmer varietal portfolios rather than replacements) some of the increases were temporary. Other methods can be used that should be much more effective in increasing biodiversity than providing farmers with a small number of pedigree bulks. For example, farmers can be given bulk seed that produces highly heterogeneous populations of nearly homozygous plants. These bulks are produced by single seed descent, or variations of this method, that are designed to minimize the effects of selection during the selng generations in order to maximize between-plant variability in the advanced generations. Since the individual plants in the bulk on a farmers eld are approaching homozygosity the plants will tend to breed true, i.e. their offspring will tend to resemble them. This should be more effective than giving farmers earlier generation bulks that are more heterozygous and which require a greater understanding of the difference in the mode of inheritance of traits for selection to be effective. For example, plants selected in early generations for recessively inherited traits will breed true. However, for traits that are dominantly inherited both homozygous and heterozygous plants will be selected and the latter will produce segregants in the subsequent generations. (In brief, the more advanced the selng generation the higher is the expected betweenplant heritability, assuming equivalent environmental variation, and hence the higher the rate of genetic advance from selection.)

Decentralized and Participatory Plant Breeding

413

Such methods are being used in rice by M. Subedi and K.D. Joshi of LI-BIRD (Local Initiatives for Biodiversity Research and Development), a Nepalese non-governmental organization (NGO), in collaboration with CAZS. They are employing equal seed descent (ESD), that advances the progeny of a cross to homozygosity whilst reducing the effects of selection, and also what can be termed as modied bulk population (MBP) breeding. In this latter technique, a segregating population from a single cross is divided into sub-populations on the basis of highly heritable traits. The cross in this PPB programme is Kalinga III IR64, and at least two traits have been used to form the sub-populations: plant height (the cross is between a tall and dwarf variety) and maturity (Kalinga III is much earlier to mature than IR64). In both ESD and MBP breeding, sufcient seed can be produced to distribute it to many farmers for selection in their own elds. Hence, the entire selection process can be replicated across many farmers and environments, a procedure that is impossible to do costeffectively in a non-collaborative decentralized breeding programme. Assuming there are desirable variants in the bulk, selection within it by several, or many, farmers who grow the bulk in their own elds is the method that is most likely to produce a large, long-term increase in the biodiversity deployed in the eld. From the performance of the bulks so far, it is expected that many different lines will be selected and retained by farmers. The above examples are only some of the many possibilities because participatory breeding programmes need to be adapted to local circumstances. For example, when farmers are reluctant to select from within populations, as is the case in barley in the Middle East and North Africa, then selection between populations of crosses can have greater emphasis (S. Ceccarelli, personal communication). Hence, farmers select from many crosses each of which has been made between parents that do not differ in qualitative traits of importance to them: e.g. in barley the parents would have a common husk colour, and both would either have either two-rowed or six-rowed spikes. Another example is the case of longer-duration varieties of any crop where only one generation a year can be grown. Where doubled haploid techniques are available, their use is advantageous for PPB. Homozygous lines the doubled haploids can be produced much more rapidly than by conventional means and farmers can be given homozygous products that breed true. The size of the target areas of such a programme has important implications for base-broadening. The cross Kalinga III IR64 is wide enough to produce segregants that are suitable for both rainfed and upland conditions, and for a range of altitudes in Nepal from 300 to 1300 m. The same cross is also being used by Birsa Agricultural University and a bilateral development project, the Eastern Indian Rainfed Farming Project, in collaboration with CAZS. This very broad target environment is made possible by the adaptation of the parents. Kalinga III has wide adaptation to rainfed environments (it performs well in eastern and western India and Nepal over a range of latitudes) and IR64 has wide adaptation to high potential systems. If this single cross produces varieties that succeed across such a wide range of target areas, then this would lead to a narrowing of the genetic base of rice across the target area. However, estimating varietal biodiversity by considering only the ancestry of these successful varieties may be misleading. Although the coefcient of parentage between any pair of inbred lines derived from the cross will be 0.5, in reality the genetic dissimilarity between lines for two contrasting environments, e.g. upland and irrigated conditions, is expected to

414

J.R Witcombe

be higher. Divergent selection for adaptation to these two contrasting environments will result in the xation of different alleles at loci determining adaptation. Moreover, if the recommended method of choosing parents in PPB is followed using a locally adopted parent then the Kalinga III IR64 cross targets only a small area in Nepal where Kalinga III has been adopted for specialized niches in the February-sown (Chaite) rice crop in Nepal. In the PPB programme in Nepal, crosses have also been made to the most important Chaite rice, CH45, and the most important main-season rice, Masuli. Only a few examples of many PPB breeding programmes in rice are described above. The amount of diversity generated by decentralized breeding programmes will depend on the extent of decentralization, i.e. the number of programmes, and on the genetic diversity among the parental germplasm used by those decentralized programmes. However, PPB programmes are still too new and insufciently widespread for any conclusions to be made on the pattern of varietal diversity that is likely to emerge from them.

Second Cycle Products

It is simple to select the initial parents for targeted crosses in a rst cycle of PPB. The best and most obvious strategy appears to be to cross the most adapted, widespread local variety with a few exotic, high-yielding parents from a centralized breeding programme that are carefully selected for having complementary traits (Witcombe et al., 1996). The need for exotic, high-yielding cultivars emphazises how PPB and centralized breeding complement each other and how there is a continuing need for both of them. PPB can be regarded as adaptive breeding (adapting germplasm for specic target environments), whereas centralized breeding is more strategic (generating novel widely adapted germplasm). Once farmers accept varieties produced from the rst cycle crosses, the most obvious crosses in the second cycle of PPB will be between the best varieties produced in the rst cycle of PPB and the latest farmer-adopted releases, i.e. the most elite material from centralized breeding. From a viewpoint of maintaining genetic diversity, attention needs to be paid to minimizing the relatedness of parents from the centralized breeding programme that are used in the rst and second cycles. Nevertheless, since the use in the initial crosses of the rst-cycle locally adapted parent will also appear in the pedigrees of varieties produced in the second cycle, this will inevitably result in a degree of genetic relatedness between them. Similar, but less compelling, arguments could be found for a reduction in varietal diversity with conventional breeding. Very popular, widely adapted varieties will tend to be used often as parents and occur in the ancestry of many varieties. However, because local adaptation is not a major criterion for parental selection in centralized breeding programmes, and because many crosses tend to be made in such programmes, new varieties can be genetically distant from the varieties that they replace. For example, HD2329 was the most widely grown wheat cultivar in the Indian Punjab from 1987 to 1996 and, in the space of 3 years, 1997 to 1999, was largely replaced by PBW343 to which it is effectively unrelated (Joginder Singh, Punjab Agricultural University, personal communication).

Decentralized and Participatory Plant Breeding

415

PPB and Centralized Breeding are Complementary Strategies

Centralized breeding and decentralized participatory breeding can support each other because they utilize different types and ranges of germplasm as parents. Parental choice in centralized breeding can be limited by the need to target mega-environments or entire agroecological zones. Hence, many local landraces would be expected to contribute little or nothing to wide adaptation, but would be ideal parents for the target micro-environments of a PPB programme. Often these will be socioeconomic niches, such as the preference by some farmers, whose livelihoods are more dependent on livestock, for higher straw yields and straw quality rather than simply higher grain yield. On the other hand, in no decentralized PPB programme can anything but a tiny fraction of the available genetic resources be screened, and decentralized programmes cannot enter into strategic breeding that broadens the genetic base by using wide crosses between cultivated and wild species. Such wide crosses have been extremely important in broadening the genetic base of rice and wheat, and much of this work has taken place in centralized breeding programmes in IARCs. Recent examples include the use at the West Africa Rice Development Association (WARDA) of Oryza glaberrima in the improvement of Oryza sativa (Jones et al., 1997); the introduction of chromosome segments of distantly related species of rice into cultivated rice (Brar and Khush, 1997); and the use at CIMMYT of synthetic hexaploids in wheat breeding derived from crosses between tetraploids and a diploid ancestral species (e.g. Villareal et al., 1996). These wide crosses introduce into the cultivated species novel resistances, in most cases against pathogens, and often the crosses increase the genetic yield potential of the cultivated species. Partnerships between centralized breeding and participatory plant breeding are needed. However, institutional issues determine how these two approaches can be integrated. For example, in WARDA, progeny from the tissue-culture-facilitated interspecic cross have been entered into participatory trials (WARDA, 1999). Here both the strategic (base-broadening) breeding and participatory varietal selection that broadens the deployed genetic base in farmers elds are carried out by the same institution. In other cases, the partnership is simply by utilization of germplasm, e.g. where an NGO or national programme uses CGIAR-bred material in crosses to locally adapted and locally distributed germplasm. In the long term, partnerships between centralized breeding and collaborative, participatory breeding can produce the best germplasm and the broadest genetic base in farmers elds.

Conclusions
Participatory plant breeding (PPB) is increasingly being used in both inbred and openpollinated crop species. By denition, PPB has signicant farmer participation but it also involves decentralization of the breeding process from research station to farmers elds. From the viewpoint of broadening the genetic base of crops, both decentralization and participation are important. Decentralization, irrespective of the degree of farmer participation, can provide greater diversity than a centralized system. Participation by farmers, particularly when it is collaborative, can result in much greater diversity in the elds of collaborating farmers, and provide a broader range of promising genotypes.

416

J.R Witcombe

A common strategy in PPB will be to cross locally adapted landraces with genetically unrelated high-yielding varieties from centralized breeding. In subsequent cycles of plant breeding, because locally adapted material will tend to be used as one of the parents, there is an increasing likelihood of genetic relatedness between the products of different cycles. Decentralization will allow genetic diversity between programmes if they depend on different parental germplasm. However, some varieties, even those for marginal environments, can be widely adapted, and several decentralized breeding programmes could all choose the same widely adapted parent, for in each area in which it is cultivated it will have local adaptation. Only centralized breeding programmes can undertake strategic research that is designed deliberately to broaden the overall genetic base of crops by, for example, screening perhaps thousands of lines from ex situ germplasm collections or by using tissue-culture-facilitated wide crossing. Hence, centralized programmes are involved in strategic research while PPB programmes are at the adaptive end of the research spectrum. Since these strategies will tend to complement each other, neither approach should be adopted exclusively.

References
Biggs, S.D. (1989) Resource-poor Farmer Participation in Research: a Synthesis of Experiences from Nine National Agricultural Research Systems. OFCOR-comparative study paper no. 3. Special series on the organisation and management of on-farm client oriented research (OFCOR). International Service for National Agricultural Research, The Hague. Brar, D.S. and Khush, G.S. (1997) Alien introgression in rice. Plant Molecular Biology 35, 3547. Ceccarelli, S., Erskine, W., Hamblin, J. and Grando, S. (1994) Genotype environment interaction and international breeding programmes. Experimental Agriculture 30, 177187. Chaudhary, R.C. and Ahn, S.W. (1996) International network for genetic evaluation of rice (INGER) and its modus operandi for multi-environment testing. In: Cooper, M. and Hammer, G.L. (eds) Plant Adaptation and Crop Improvement. CAB International, Wallingford, UK, pp. 2538. DeLacey, I.H., Basford, K.E., Cooper, M., Bull, J.K. and McClaren, C.G. (1996) Analysis of multi-environment trials an historical perspective. In: Cooper, M. and Hammer, G.L. (eds) Plant Adaptation and Crop Improvement. CAB International, Wallingford, UK, pp. 39124. Fischer, K.S. (1996) Research approaches for variable rainfed systems thinking globally, acting locally. In: Cooper, M. and Hammer, G.L. (eds) Plant Adaptation and Crop Improvement. CAB International, Wallingford, UK, pp. 2538. Huke, R.E. and James, D. (1969) Spatial aspects of HYV diffusion. In: Proceedings of the Seminar-workshop on the Economics of Rice Production. Department of Agricultural Economics, International Rice Research Institute, Los Baos, Philippines, pp. 212240. Jones, M.P., Dingkuhn, M., Aluko, G.K. and Seman, M. (1997) Interspecic Oryza sativa. L. O. glaberrima Steud. progenies in upland rice improvement. Euphytica 94, 237246. Joshi, A. and Witcombe, J.R. (1996) Farmer participatory crop improvement. II. Farmer participatory varietal selection in India. Experimental Agriculture 32, 461477. Joshi, A. and Witcombe, J.R. (1998) Farmer participatory approaches for varietal improvement. In: Witcombe, J.R., Virk, D.S. and Farrington, J. (eds) Seeds of Choice. Making the Most of

Decentralized and Participatory Plant Breeding

417

New Varieties for Small Farmers. Oxford IBH, New Delhi and Intermediate Technology Publications, London, pp. 172190. Joshi, K.D., Sthapit, B.R. and Witcombe, J.R. (2001) How narrowly adapted are the products of decentralized breeding? The spread of rice varieties from a participatory plant breeding programme in Nepal. Euphytica (Special Issue). Louette, D., Charrier, A. and Berthaud, J. (1997) In situ conservation of maize in Mexico: genetic diversity and maize seed management in a traditional community. Economic Botany 51, 2038. Rajaram, S., van Ginkel, M. and Fischer, R.A. (1995) CIMMYTs wheat breeding mega-environments (ME). In: Lin, Z.S. and Xin, Z.Y. (eds) Proceedings of the Eighth International Wheat Genetics Symposium, Vol. 2. China Agriculture Scientech Press, Beijing, pp. 11011106. Smale, M., Aquino, P., Crossa, J., del Toro, E., Dubin, J., Fischer, T., Fox, P., Khairallah, M., Mujeeb-Kazi, A., Nightingale, K.J., Ortiz-Monasteria, I., Rajaram, S., Sing, R., Skovmand, B., van Ginkel, M., Varughese G. and Ward, R. (1996) Understanding Global Trends in the Use of Wheat Diversity and International Flows of Wheat Genetic Resources. International Maize and Wheat Improvement Centre (CIMMYT), Mexico, D.F. Souza, E., Fox, P.N., Byerlee, D. and Skovmand, B. (1994) Spring wheat diversity in irrigated areas of two developing countries. Crop Science 34, 774783. Sthapit, B.R., Joshi, K.D. and Witcombe, J.R. (1996) Farmer participatory crop improvement. III. Farmer participatory plant breeding in Nepal. Experimental Agriculture 32, 479496. Traxler, G. and Pingali, P.L. (1999) International Collaboration in Crop Improvement Research: Current Status and Future Prospects. Economics Working Paper 99-11. International Maize and Wheat Improvement Centre (CIMMYT), D.F. Mexico. Villareal, R.L., Mujeeb-Kazi, A., Fuentes-Davila, G. and Rajaram, S. (1996) Registration of four synthetic hexaploid wheat germplasm lines derived from Triticum turgidum T. tauschii crosses and resistance to Karnal bunt. Crop Science 36, 218. WARDA (West Africa Rice Development Association) (1999) Participatory Varietal Selection. The Spark that Lit the Flame. WARDA, Bouak, Cte dIvoire. Witcombe, J.R. (1996) Decentralization versus farmer participation in plant breeding: some methodology issues. In: New Frontiers in Participatory Research and Gender Analysis. Proceedings of the International Seminar on Participatory Research and Gender Analysis for Technology Development. September 914 1996. CGIAR Systemwide Program on Participatory Research and Gender Analysis for Technology Development and Institutional Innovation. CIAT, Cali, Colombia, pp. 135154. Witcombe, J.R. (1999) Does plant breeding lead to a loss of genetic diversity? In: Wood, D. and Lenn, J. (eds) Agrobiodiversity Characterization, Utilization and Management. CABI International, Wallingford, UK, pp. 245274. Witcombe, J.R., Joshi, A., Joshi, K.D. and Sthapit, B.R. (1996) Farmer participatory crop improvement. I. Varietal selection and breeding methods and their impact on biodiversity. Experimental Agriculture 32, 445460. Witcombe, J.R., Packwood, A.J., Raj, A.G.B. and Virk, D.S. (1998) The extent and rate of adoption of modern cultivars in India. In: Witcombe, J.R., Virk, D.S. and Farrington, J. (eds) Seeds of Choice. Making the Most of New Varieties for Small Farmers. Oxford IBH, New Delhi, and Intermediate Technology Publications, London, pp. 5368. Witcombe, J.R., Petre, R., Jones, S. and Joshi, A. (1999) Farmer participatory crop improvement. IV. The spread and impact of a rice variety identied by participatory varietal selection. Experimental Agriculture 35, 117.

27

Base-broadening for Clientoriented Impact: Insights Drawn from Participatory Plant Breeding Field Experience

L. SPERLING1, J. ASHBY2, E. WELTZIEN3, M. SMITH4 AND S. MCGUIRE5

Program for Participatory Research and Gender Analysis for Technology Development and Institutional Innovation (PRGA), International Center for Tropical Agriculture (CIAT) c/o Frankenstraat 26, 2582SL, The Hague, The Netherlands; 2International Centre for Tropical Agriculture (CIAT), Cali, Colombia; 3Genetic Resources and Enhancement Program, International Centre for Research in the Semi-Arid Tropics (ICRISAT), Bamako, Mali; 4Department of Plant Breeding, Cornell University, Ithaca, New York, USA; 5Technology and Agrarian Development, Wageningen University, Nieuwe Kanaal 11, 6709 PA Wageningen, The Netherlands
1Systemwide

Introduction
Broadening of basic germplasm pools can only achieve its goal of promoting agricultural stability if the products are actually sown and resown by farmers on a signicant scale. To achieve this end, varieties developed from based-broadened pools need to meet certain criteria: (i) they have to be adapted to farmers local growing conditions; (ii) they must meet farmers preferences1; and (iii) they have to remain accessible to farmers on a sustained basis, that is, in both the short and longer term. In practice, 3050 years of fundamental base-broadening research will have little value if the resulting primary products remain solely in researcher-controlled elds, laboratories and genebanks. To anticipate some of the more downstream challenges implied by upstream base-broadening work, research and development (R&D) strategies need now to be tested and targeted for widening the base of germplasm use at the bottom, among end-users in rural communities. Strategies need to be contrasted and compared to ensure that diverse germplasm pools can provide varieties that are locally adapted, farmer acceptable and farmer accessible. Current breeding pools offer greater variability than actually reaches the end-user. Further, out of hundreds of varieties in any single national agricultural research system
IPGRI/FAO 2001. Broadening the Genetic Base of Crop Production (eds H.D. Cooper, C. Spillane and T. Hodgkin)

419

420

L. Sperling et al.

(NARS), farmers typically receive a handful to be tested each season. For instance, in Rwanda, the initial experimental set of about 250 cultivars narrows down to three to ve entries in on-farm trials (Sperling et al., 1993); in Syria (S. Grando, personal communication), a pool of 200 is narrowed down to three or four; and in India (pearl millet), 3080 entries may enter preliminary yield trials, with one or two reaching farmers only after ofcial release (E. Weltzien, personal communication). The current narrow provision of materials to farmers offers fertile opportunity for testing the principles of local-level base-broadening in anticipation that even wider pools may be available in a base-broadened future. This chapter rst briey describes an emerging novel eld, participatory plant breeding (PPB), which has among its central aims broadening the diversity of germplasm available to and used by farmers. In practice, upstream base-broadening efforts will have to be linked integrally to compatible or like-minded downstream R&D eld programmes if diverse and novel products are to be sown in thousands of site-specic contexts. PPB offers one such like-minded paradigm, as programmes tend to work very interactively with farming communities and often build in capacity for continued evolution of both genetic materials themselves and farmers skills to manage such materials. Drawing from a large body of actual eld experience, this chapter summarizes some of the lessons learned from PPB in terms of meeting the three fundamental principles of adaptability, acceptability and accessibility. While there is accumulating evidence to enhance our ability to meet these three preconditions, and thus to inform the design of base-broadening programmes, critical knowledge gaps remain. As such, each section of this chapter integrates analysis of steps made to date with those that still need to be pursued. Through such reection, the chapter suggests a practical farmerlevel research agenda for enhancing our capacity to widen the base of farmers germplasm use.

Overview of Participatory Plant Breeding2

What is PPB? PPB involves scientists, farmers and others such as consumers, extensionists, vendors, industry and rural cooperatives in plant breeding research. It is termed participatory because users can have a research role in all major stages of the breeding and selection process. Such users become co-researchers as they can help set overall goals, determine specic breeding priorities, make crosses, screen germplasm entries in the pre-adaptive phases of research, take charge of adaptive testing, and lead the subsequent seed multiplication and diffusion process (Sperling and Ashby, 1999). The fundamental rationale for a PPB programme is that joint efforts can deliver more than when each actor works alone. While some have cogently argued that commercial, private-sector, plant breeding has long been client-driven, or participatory under another name, the application of PPB to reach poor client groups, to breed for high-stress, hetereogeneous environments and to incorporate diverse traits to meet specic client preferences results in fundamental changes in the way in which plant genetic resources are managed by formal breeding

Base-broadening for Client-oriented Impact

421

programmes and farmers. It makes sense, therefore, to analyse PPB as a new approach to germplasm development, especially in the public sector. The Consultative Group on International Agricultural Research (CGIAR) Program on Participatory Research and Gender Analysis for Technology Development and Institutional Innovation (SWP/PRGA) currently has detailed documentation on 65 PPB programmes and projects (Hecht, 1999; McGuire et al., 1999; Weltzien et al., 1999). Most of the cases, whether located in public sector or non-governmental organization (NGO) crop improvement programmes, were begun in the 1990s. Under the PPB rubric, two broad approaches have been dened: when farmers join in breeding experiments that have been initiated by formal breeding programmes (formal-led PPB); and when scientists seek to support farmers own systems of breeding, varietal selection and seed maintenance (farmer-led PPB). Both formal-led and farmer-led work may prove key for enlarging the base of germplasm accessed, modied and evolved through use in farming communities. Table 27.1 indicates the range of crops on which PPB programmes have been initiated, as well as proposals received by the SWP/PRGA. Given its relatively short history, the range of crops is impressive although few can be considered minor.

Table 27.1. Crop foci of PPB programmes to date. Crop Barley Bean Black gram Cassava Chickpea Chaite rice Cotton Cowpea Lentil Maize Pearl millet Potato Rice Sorghum Sunower Trees Wheat Peas Sweet potatoes Faba bean Native potatoes Durum wheat Yam Crotelaria Formal-led * * * * * * * * * * * * * * * * * Farmer-led * * PRGA small grant submissions * *

* *

* * * * * *

* * * * * * * * *

422

L. Sperling et al.

What are the goals of PPB? PPB programmes can have a diversity of overarching goals. The most common goal has been to contribute to increased production in farmers elds and increased farmer incomes through the development and enhanced adoption of suitable, usually improved varieties. These are the basic goals of any formal-led breeding programme, and participatory approaches are often experimented with to achieve these goals more effectively and more efciently. In this context, PPB programmes sometimes seek to rene their knowledge of farmers needs or preferences or re-orient general breeding directions, such as the type of germplasm used, the priority traits sought, and the management of both on-station and on-farm trials. Enhancement of crop diversity on-farm is another broad goal towards which some PPB programmes strive. Participatory breeding programmes having this goal tend to work more often with the farmers own germplasm or a combination of local and exotic materials. Many also involve farmers in the screening of a wide range of varieties in the pre-adaptive stages of research, either in on-station trials or in community plots. In several cases, PPB programmes have also released populations or have purposely promoted breeding strategies that result in heterogeneous materials. Chapter 26, this volume, by Witcombe discusses this subject further. Another important goal of PPB programmes is to provide benets for specic types of users (e.g. the rural poor, women, farmers with marginal soils) or to address deliberately the needs of a broader range of users. Such a goal necessitates an extensive diagnosis among well-dened types of potential user and stakeholder groups. While addressing issues related to improved adoption of breeding products and/or enhancement of crop diversity, PPB programmes often nd themselves confronted with the need to address modications in policy, whether these be seed regulations or variety release criteria and procedures. Most modications are sought to accommodate expansion and institutionalization of approaches that better serve farmers aims. These may include modications in the scale of testing and the scale of desired variety adaptation, the kind of data required for release, and the number of varieties released at any one time (see Louwaars, Chapter 5, this volume). Finally, some programmes specically work towards enhancing the farmers own breeding process, i.e. providing technical knowledge and insights so that farmers themselves are more successful in their own selection and seed production efforts. This skillbuilding goal is often addressed together with a more general effort towards strengthening the capacities of farming communities to demand and derive benets from the formal research institutions. Overviews of PPB programmes show that, to date, primary goals set largely parallel those of classic breeding research: to increase production or increase the value of products through enhanced quality traits. However, an impressive 20% of PPB projects also have as an explicit goal varietal diversity enhancement in farmers elds (McGuire et al., 1999; Weltzien et al., 1999). The strengths, shortcomings and overriding challenges of PPB programmes very directly herald those which might be expected to enrol in more general local-level base-broadening initiatives. Developing strategies for rendering adapted, appreciated and accessible materials to farming communities lies at the heart of both PPB and base-broadening work.

Base-broadening for Client-oriented Impact

423

What are the environments in which PPB unfolds? Based on an inventory of about 65 formal-led and farmer-led PPB cases, maps are being devised to indicate the range of environments in which PPB unfolds. One parameter of the conceptual map describes the type of agroecological context. This has been constructed on a scale for environments from high stress to low stress based on actual vs. expected yields coupled with an index for incidence of crop failure. Agroecological environments potentially range from those that are primarily subsistence-oriented and highly unstable, to systems in which crop production is predictable, highly controlled, and often shaped by signicant input use. The second parameter suggests the broad economic environment of PPB, that is the degree of homogeneous demand versus heterogeneous demand for varieties. Mapping was based on a scale of 1 to 10 according to the leniency/narrowness of varietal characteristics demanded by end-users and the similarity/discordance between varieties use for home consumption and for sale. Contexts at the higher end (for example, 8 or 9) tend to correspond to a high degree of homogeneity in product and often favour a narrow range of grain, taste and cooking types. Such a strong degree of uniformity is often associated with contexts where farmers are producing for highly specialized markets. Many plant breeders consider PPB as most appropriate for environments that are high-stress (marginal) and where agriculture is low-input (i.e. where adaptability is the key issue). Certainly, conventional breeding has been less effective in such difcult environments and in reaching farmers with few resources; so the rationale for testing participatory approaches, which are often site-specic, is a solid one. Analysis of actual PPB cases, however, shows a more complex picture. Not all PPB is concentrated in high-stress environments with low-input agriculture. An unexpectedly large number of PPB programmes are being initiated in the intermediate areas where agroclimatic stress is less severe (Fig. 27.1). On the whole, these are cases where quality concerns, i.e. meeting exigent end-user preferences, is dened as the paramount challenge: acceptability is key. Figure 27.1 also shows that a signicant amount of PPB work is now occurring in low-stress areas (Green Revolution-type zones) where homogeneous enduser preferences are well dened in the market (for example, the Nepalese Terai: J. Witcombe, personal communication). Two reasons explain most of the cases in these areas. First, some of these PPB programmes aim to expand intracrop varietal diversity in what have become relatively uniform farming areas. Secondly, some programmes are run by NGOs or organized farmer groups with the primary goal of helping communities gain greater control over their breeding process or seed supply (McGuire et al., 1999, Salazar, Chapter 7, this volume). In these sites, the operative driving PPB principles are of increasing control and accessibility over germplasm and germplasm processes. While the three prerequisites of PPB are integral to each programme, this broad look at the environments of PPB suggests that the weight of each may vary by site. In the harshest zones, adaptibility is the prime criterion to be met that is, getting something that grows. In the mid-potential zones, identifying adapted germplasm is often of equal challenge to nding something that addresses producer and consumer needs (i.e. is acceptable). In the higher potential areas, acceptability holds some weight, but control over breeding and diverse breeding materials (i.e. accessibility) may lay at the base of many PPB programmes.

424

L. Sperling et al.
Unfavourable ..............................................................................................................Favourable Subsistence ...............................................................................................................Commercial CORPOICA/CIAT Colombia SE Brazil/EPAGRI USA/popcorn

CIAT/CIAL/Colombia (beans) PROINPA/Bolivia EAP-Zamorano/Honduras Sokoine/CRSP Tanzania Ethiopia/Awassa

CONSERVE/ Philippines

IRRI/India Partners Tanzania NARS/CIAT

CAZS/NARC-Nepal PNAP/CIP/Rwanda USDA

Market integration

Guanxi PNL/CIAT Zaire IRRI India partners EMPMF NE Brazil BBA/India ICRAF/Burundi (trees) UPWARD/Philippines KRIBHCO/ODA India NARC/DFID Nepal REST/Ethiopia ICARDA/Syria CIMMYT Mexico Sure/ICRISAT India Guanxi Narendra Dev/India IER/KIT Mali ICRISAT/Namibia SAVE/Sierra Leone

ISAR/CIAT/COOPIBU Rwanda

Fig. 27.1. Environmental contexts of PPB programmes.

Germplasm Adaptability at the Local Level: Lessons from PPB Field Experience
It is now widely accepted that to identify adapted varieties in the majority of the worlds farming communities, much of the selection has to be done, onsite, under comparable agroecological and farmer management conditions. For example, Ceccarelli (1994) highlights the importance of genotype environment crossover between farms and experiment stations, advocating testing under more typical conditions to develop varieties that perform best locally with stable yield over time (Ceccarelli et al., 1991). This is a basis for early decentralization and selection under low-input conditions in the barley programme at the International Center for Agricultural Research in Dry Areas (ICARDA) (Ceccarelli et al., Chapter 6, this volume). In the overview paper

Base-broadening for Client-oriented Impact

425

on formal-led PPB (Weltzein et al., 1999), many researchers cited high-stress, marginal conditions (including low or erratic rainfall, unpredictable highland climates or low input due to remoteness) as a reason for employing participatory approaches in their programmes. After 15 years of breeding work with little farmer adoption, cassava breeders in a region of Colombia with poor soils and 8001000 mm rain used PPB to quickly release three farmer-tested varieties. While some varieties identied through highly decentralized testing meet adaptation needs only in specic niches (see Sthapit et al., 1996), others prove to be surprisingly widely adapted (Witcombe et al., 1999). Obviously much depends on how representative the selection is and the heterogeneity of user preferences in surrounding zones. While, in principle, the need for site-specic screening is widely accepted, guidelines for the degree of decentralization necessary (Witcombe, 1996) still have to be elaborated. Those who use PPB specically to enhance varietal diversity on-farm (both phenotypic and genetic diversity) face two sets of major challenges to putting into operation decentralization on a broad scale. These challenges will lie similarly at the core of most local-level base-broadening programmes.

Technical approaches to decentralization The rst set of challenges centres on general technical approaches to enhancing varietal diversity at the community level. Theoretically, diversity can be broadened through several strategies: 1. Many xed lines can be proferred for screening in community-based plots. Farmers can then individually select lines which they project will perform well on their own farms and carry on all further home-based testing. This approach worked quite well in Rwanda, where 21 new bean phenotypes were adopted among three communities in a 3-year period (Sperling et al., 1993). 2. Segregating materials or scientically enhanced populations can be made available to farming communities who themselves then guide the subsequent adaptation and selection. Berg (1996) describes the theory of this process, drawing from the famous experiment with composite cross populations of barley at the University of California at Berkeley (Allard, 1988, 1992; Soliman and Allard, 1991; see also Ibrahim and Barrett, Chapter 15, this volume). There, populations were exposed to natural selection over many generations and eventually showed good disease resistance and yield stability although not high yield potential per se. In practice, excellent results were achieved in Nepal in identifying adapted and farmer-accepted chill-tolerant rice varieties through the placing of segregating (F3 to F4) materials on-farm. This work was heavily shaped by both researcher and farmer input (Sthapit et al., 1996). In another case, in the Philippines, breeders gave F3 to F4 rice materials to farmers, with subsequent selection left entirely to communities to F7/F8. The advantages of such a schema is that it is easy to run and can be encouraged at many locations (Witcombe, 1996). 3. A third strategy might focus on upgrading local material itself (in terms of various traits) through recombinations/introgression. This is commonly employed across classic breeding programmes but is a lengthy and costly process. 4. A focus of diversity enhancement might also be on skill-building itself, rather than

426

L. Sperling et al.

on supplying germplasm. An experimental programme run by Zamarano and Cornell University for farmers from Honduras, Nicaragua and El Salvador considerably honed participants ability to select and manage maize seed (Gomez, 1995; Gomez and Smith, 1996). The point is that there are many different strategies that should be tested at the local level for working with farmers and giving them the means to choose among materials or develop and modify varietal materials themselves. The effectiveness of different strategies will probably vary according to such variables as farmer breeding/selecting skills (often related to the degree of previous exposure to diverse materials), the narrowness/leniency of the market environment, and the agroecological exibility (i.e. there may be few options in the harshest locales). Programmes need to be set up to compare and contrast different approaches simultaneously and at a scale that allows lessons to be learned. Ultimately, we need sharpened insights into which base-broadening approaches are best for which ecological and socioeconomic environments.

Organizational issues in decentralization The second challenge revolves around organizational aspects of the base-broadening enterprise. To date, there has been very little work exploring the organizational options for decentralization across many sites (namely the work of the local committees for agricultural research (CIALs) in Central and Latin America, Ashby et al., 1999). This gap is glaring as it is generally believed that client-oriented programmes have to be decentralized to have local-level impact simply because farmers have differing needs specic to their own agronomic and socioeconomic situations. Effective decentralization of varietal testing is a task beyond the resources of most public-sector research services: in practice, most NARS budgets are declining. Might NGOs possibly have an edge in decentralizing efforts in just one or two communities? And what are the prospects for NGO scaling up? Might the farmer eld school (FFS) model be a useful one to link with participatory and decentralized programmes? (The emphasis on skill-building in the FFS approach might be particularly relevant for these innovative breeding efforts.) Would organized groups of farmers have the comparative advantage in leading local testing and for scaling up? Unless local-level programmes efforts can enlarge their scale, they remain but interesting, perhaps productive, single-site experiments. To date, within the eld of PPB, there are relatively few cases where the process has been scaled up through the creation of multiple decentralized programmes: that is, where the process of PPB work actually takes place at different sites independently. The CIALs described above are one exception: as of 1999, about 250 CIALs, farmer research committees, in several countries have engaged in independent variety testing on a range of crops (A. Braun, personal communication). Another example of this is PPB work with cassava in Colombia developed around farmer cooperatives dedicated to drying cassava chips: in its 10 years, many cooperatives have been involved and in 1999 alone, the programme (assisted by the Corporacin Colombiana de Investigacin Agropecuaria (CORPOICA) and the CIAT) is working with 13 different communities (C. Iglesias, personal communication). PPB work in Rwanda also expressly experi-

Base-broadening for Client-oriented Impact

427

mented with different organizational options for decentralization, comparing the strengths and weaknesses of collaborating with the formal extension service, womens cooperatives and self-organized farmer research groups (Sperling and Scheidegger, 1997). Evaluation or assessment of organizational options is one of the large and almost total gaps in current PPB work. The choice of collaborators largely determines how much the process can be scaled up and the technical divisions of labour, i.e. devolving responsibility and deciding who does what. The choice of collaborators (and the scale on which one works) also directly inuences the cost of the PPB work, both overall, and the cost and benets for each partner. It is important to emphasize that almost all scaling-up work in PPB has been done in programmes working with stabilized materials. When one considers generation of genetic variability and selection in segregating populations, there are few examples of farmer participation and all are with very small numbers of farmers (just one to ten). Whether scaling up of farmer participation in these activities is necessary depends on the extent to which they can be centralized while still effectively addressing varietal needs over a broad area (Weltzien et al., 1999; see also Ceccarelli et al., Chapter 6, this volume). To anticipate base-broadening work, there is a fundamental need to identify decentralization options. Chapter 13, this volume, by Goldringer et al. illustrates the potential role of high schools and agricultural colleges in decentralizing formal basebroadening programmes. Most likely, if they are to be cost-effective, they will have to be farmer-initiated and managed. However, this has to be examined systematically. Which organizational forms can handle which type of technical, decentralized experiments? What are the costs and for whom? What are the short-term and longer-term effects on farmer-held diversity?

Germplasm Acceptability at the Local Level: Lessons from PPB Fieldwork

It is easier said than done to identify appreciated varieties. Simply said: adaptation, which is often mistakenly equated with yield, is not an adequate predictor of adoption. An increasing number of studies show that traits other than yield strongly affect farmers cultivar choice (e.g. Haugerud and Collinson, 1990; Nazarea-Sandoval, 1995).

Varietal traits farmers deem important A great deal of qualitative evaluation has been generated within the course of PPB evaluations. In an overview paper synthesizing some of the critical social issues and analyses associated with PPB work, 54 criteria were highlighted as being routinely cited by farmers in their construction of ideotypes (Hecht, 2001). Table 27.2 outlines the elements mentioned by farmers across crop types, with the broad categories synthesized as environmental, agronomic and cultural trait groups. In her thoughtprovoking social analysis, Hecht comments that slightly fewer than half of these 54

428

Table 27.2. Farmer selection criteria.

Environmental criteria Tolerant of soil limitations (fertility; variable water conditions; textural variability) Tolerant of climatic factors heat; cold; wind; rainfall variability) Biotic pressures (insect pests; animal pests; human predators; weed; competition; weed supression) Disease tolerance Agronomic criteria General (intercropping ability; shade; tolerance; owering characteristics; dormancy features) Yield (earliness of production; plant form) Tuber distribution Flexibility in planting and harvest Crop morphology (size of tuber, grain or cob (bigger is not always better); shape of plant; tillering habits; colour) Culture, production and use Labour demand (germination characteristics; planting ease; cultivation; weeding; harvesting) Processing (ease of) (dehulling; pounding; peeling; milling; humidication; breakage characteristics) Culinary (taste; texture; aroma; quality of broths ritual uses; nutritional factors; quality in fermented products) Storage Exchange utility Commercial utility Source: Hecht, 2000. PRGA Programme. Ancillary features Quality and palatability of leaves for humans Quality and palatability of leaves for animals Calibre of stover fodder Ratooning ability Craft uses of bres Mulch materials Craft uses of bres Staking materials Mulch materials Construction materials Staking materials Construction materials

L. Sperling et al.

Base-broadening for Client-oriented Impact

429

traits are related to post-harvest, labour, ancillary, culinary or storage features that is, the cultural dimensions of crops. This suggests that farmers, especially women farmers, value these characters a great deal, since these are often highly gender-specic tasks (such as processing). These latter characteristics have rarely been taken into account in the usual pathways of technology development. Salient among the various farmer criteria cited within PPB evaluations were the following: 1. 2. 3. 4. 5. 6. Cultivar performance in intercropped systems. The importance of ancillary products as food for both humans and animals. Earliness of production. Labour demand characteristics. Post-harvest processing concerns. Culinary dimensions.

While, yield analysis remains the measure of choice for many agronomists for assessing whether the variety is doing well (with some moving toward the measure of yield stability), the data from the participatory evaluation trials suggest other factors pertaining to household needs (i.e. early yield, ancillaries) labour and cultural demands are probably equally important as agronomic criteria.

Diagnostic methods to determine farmer needs Accurate diagnosis of what users want and need can be effected through a range of techniques. These can include: farmer screening of diverse germplasm nurseries (Ipinge et al., 1996); detailed analyses of weaknesses and strengths in the materials farmers are already sowing (Weltzien et al., 1996); and exploratory trials using local and exotic material, which explicitly expose farmers to a wide range of traits and/or ranges within traits (ICARDA, 1994). These methods are in addition to the more common formal and informal surveys, focus group sessions or community meeting assessments. The choice of which method to use seems to depend more on the scale of investigation, money and time resources available, and training of the researchers involved than on the actual results or degree of detail achievable through any one technique. Despite the importance of accurate diagnoses in guiding the development of a lengthy breeding programme, there have been few critcial analyses comparing and contrasting the techniques available (Weltzien et al., 1996). One deliberate focus of a PPB programme currently funded by the UKs Department for International Development in ve sites in eastern Africa and Latin America centres on comparing the results of different diagnostic strategies within the same communities. It aims to evaluate the type of information received (e.g. specicity on varietal traits and trade-off assessments, insights into different user preferences) with the time, skill and scale on which the techniques can be used (L. Sperling, personal communication). One concern is that most diagnostic techniques can only be conducted on a very local level with larger-scale diagnostic skills (such as conjoint analysis) demanding highly specialized training. This is a methodological area in which much fundamental work remains to be done to sharpen both PPB and future base-broadening work.

430

L. Sperling et al.

Differentiating among farmers Good diagnosis of farmers preferred varietal traits (ranges of criteria/trade-offs) logically needs to be coupled with sufcient means for distinguishing among farmers: it does matter who wants what if a programme has aims relating to targeting specic groups of potential users. In differentiating between farmers, the current vogue often dictates a focus on gender differences (women may have different preferences to men) or wealth differences (the poor having divergent needs/wants from the rich). Field evidence from PPB programmes shows that gender can be an important factor in determining differential preferences, but not always. In Mali, maize evaluations showed men putting production and early maturity as the main criteria, with women focusing on organoleptic and processing aspects (DeFoer et al., 1997). Rice work in West Africa had a similar gendered division, with WARDA (West Africa Rice Development Associations) scientists reporting that men focused on yield and yieldrelated traits such as plant vigour, while women concentrated on quality attributes, such as bold grains (Lilja and Dalton, 1998). However, in many cases, gender-differentiated evaluations do not yield clear-cut preferences. Field evidence also shows that wealth can be a key distinguishing factor. The review of formal-led PPB programmes showed a number of trends among the nancially disadvantaged. Poor farmers often cited earliness as an important factor in shortening the hungry season and in maximizing production on their small landholdings. Poorer farmers also had important criteria related to multiple crop uses (such as cassava in Colombia, pearl millet in Rajasthan (India) and cowpea in Cameroon), as people relied on non-food parts for animal feed, or seasonal excess processed for the market (cassava our/Colombia). Where farmers rely on selling some of their products for scarce household cash income, quality criteria proved to be determined more by market demands than by grower preferences. With potatoes in Peru and cowpeas in Cameroon, farmers grow modern varieties as cash crops, and benet by earning as much as double for products with preferred size, colour or shape (Weltzien Smith, 1999). However, farmers do divide themselves into groups (or construct images of selfidentify) beyond gender and wealth. For example, caste, ethnicity or age may be important in any given region. The methodological concern centres on how can we nd out which populations differences or user differences are really key in shaping preferences?. That is, farmers may differ by age, wealth, religion and politics but not all of these variables are equally important in determining, say, lentil variety preferences. There is a compelling need to develop and rene diagnostic techniques that can be used at different stages of the breeding process (for example, shaping initial priorities and, at the other end, evaluating nishing lines), which can be used at various geographic scales, and which ensure that different user groups are given a voice.

Germplasm Accessibility at the Local Level: Lessons from PPB Fieldwork

PPB work suggests that the earlier farmers are involved in developing germplasm, the higher the likelihood that the resulting varieties and seed will have to be moved outside of formal channels. This proves to be true for several reasons and especially if the

Base-broadening for Client-oriented Impact

431

work is truly participatory: (i) a large range of varieties may be identied for different kinds of users; (ii) the site-specicity of varieties may discourage formal-sector multipliers from devoting their efforts; (iii) the intravarietal heterogeneity might prohibit formal multiplication; and, if farmerresearcher collaboration is taken seriously and ethically; (iv) undened or shared property rights would deter more routine multiplication.

Seed system issues Participatory plant breeding programmes often have to go hand-in-hand with recommendations to develop, or build on, local, more decentralized seed systems, which can provide location-specic varieties that farmers themselves effectively multiply and distribute. For instance, prior to the civil strife in the early 1990s, both Rwanda and Burundi, Ministries of Agriculture and Rural Development proposed plans for decentralizing seed services partly to accommodate more decentralized breeding (L. Sperling, personal communication). PPB programmes are also increasingly incorporating integrated and innovative seed production components to deliver rapidly the positive impacts that PPB can achieve. Good examples of this type of integration come from the PPB and seed work with cassava in Colombia and with potato in Bolivia, Peru and Ecuador through a set of collaborations between the national seed projects and the International Potato Centre (CIP) (Iglesias et al., n.d.; Thiele, 1999). Overall, relatively little of the current PPB work ties site-specic breeding with seed multiplication work. Either it is assumed that products will go to formal channels (which has happened only in a single case each in India and Nepal) or it is assumed that local seed systems can handle PPB products well, that varieties diffuse by themselves. Formal systems have some obvious constraints. In addition to the four points listed above, formal systems generally deal with a relatively narrow range of crops and reach out to a relatively narrow range of users. Global estimates show farmers accessing about 8090% of their seed directly on-farm or through local channels (Cromwell, 1997). For non-hybrids, the gure has sometimes been cited as less than 2% of all seed used by small farmers coming from the formal system (CIAT, 1982). Proponents of local systems sometimes point to the effectiveness of local-level exchange mechanisms and assert that varieties move themselves. However, the authors know of few studies that actually assess the efciency of local systems: how fast they move varieties, if they are equitable, if they can handle many new entries, and if they have wide geographic spread. Rather, one detailed case shows local systems to be: (i) variable in efciency according to agroecological, socioeconomic environment, and degree of varietal appreciation; (ii) of restricted access unless varieties are diffused in local markets; and (iii) susceptible to frequent loss of specic varieties especially if genetic materials are in the early stages of diffusion (Sperling and Loevinsohn, 1993). Local seed systems full a range of functions. They provide information about seed sources, seed availability, and new types of seeds and varieties. They govern the actual ow of seed and affect farmers seed security and use of external sources of seed. In many regions with marginal conditions for crop production, regions with a very high seasonal or ecosystem diversity, and for relatively minor crops, local seed systems are the only system that provides seed to farmers, especially to poor farmers.

432

L. Sperling et al.

Each seed system (formal, farmer or something in between) has different strengths and weaknesses, and a base-broadening programme should actively understand the opportunities and limitations in each, rather than just slide into one strategy or another. Choice of seed system should be guided by such questions as:

What number of varieties will be diffused? On what scale is diffusion desired and to which target groups? Are the property rights associated with each seed system acceptable to partners? Is the nal material homogeneous or heterogeneous?

In sum, to accommodate both the products of PPB and other types of base-broadening work, there is a clear need to diagnose/test the ability of local systems, or new intermediaries (e.g. cooperatives), to handle multiple new infusions of material and to distribute such material widely and equitably. This issue is particularly pressing for the cross-pollinated materials.

Property rights Joint collaboration should mean joint benet sharing. At this point, there are no readymade arrangements or best practices that one can suggest to guide the diffusion of materials emerging from farmerresearcher collaborations. Neither the farmers rights debates nor the formal breeder-rights discussions fully match the bill. The SWP/PRGA has recently started work to address property-rights issues within the participatory plant breeding arena. The process is unfolding briey as follows. Eight to ten typical PPB collaborative type cases have been identied. For each case, analysis is taking place in three realms: (i) existing legal frameworks that might constrain or stimulate the PPB relationship; (ii) associated ethical concerns; and (iii) best practice actions emerging from widespread public debate with practitioners. At the moment, most of the PPB programmes have simply skirted the propertyrights issues with two very different strategies: materials jointly developed by formal breeders and farming communities have been fed into the formal variety release and seed multiplication system (not recognizing farmers input at all), or the PPBdeveloped materials have been released or let go into farming communities with no ofcial launch of any kind. This has had a positive impact among farmers mostly with self-pollinated crops, where seed increase and quality issues are relatively easy for farmers to manage at their own acceptable levels.

Conclusions and Discussion

The term base-broadening most often conjures up images of longer-term commitments to basic evolutions of genepools. This chapter suggests that the success of basebroadening will be highly contingent on whether farmers specic needs for adapted, acceptable and accessible germplasm are met. PPB programmes can provide an array of community-based (and tested) approaches for increasing the adaptation, acceptability/ attractiveness and accessibility of base-broadened germplasm. The research agenda for successful base-broadening needs to be considerably

Base-broadening for Client-oriented Impact

433

expanded to include such issues as identication of rigorous user diagnostic methods, effective decentralization options (technical and organizational), and creation/support of farmer-responsive seed channels. Intellectual property-rights issues will also require serious attention and consideration as they could pose signicant obstacles to use of base-broadened pools and use of PPB products in general. In several cases, for many crops, current breeding pools are already wide enough to start testing varied strategies for widening the base of farmers germplasm use. However, it is not clear that they have been focused strongly enough on adaptation and acceptability/attractiveness for the non-commercial farming sector. Exploratory PPB programmes may be critical for moving base-broadened materials out to thousands of communities. However, they may be equally essential for feeding back critical end-useroriented information to enhance the likely relevance of upstream base-broadening decisions for rural populations.

Notes
1We

use the terms preferences and farmer-acceptance in a dynamic sense. Obviously concerns of colour, taste and cooking time would be encompassed in these notions. Equally, however, preferences imply a sense of attractiveness. To be used by farmers, base-broadened pools have to offer advantages not currently available to farmers: novel traits or trait combinations, or better levels of existing traits. 2This section draws from Weltzien et al., 1999, and Sperling and Ashby, 1999. 3As one reviewer noted, one might expect that in isolated, economically marginal areas, accessibility would also be an issue albeit in a different sense since formal seed supply systems are rarely operating effectively in such areas.

References
Allard, R.W. (1988) Genetic change associated with the evolution of adaptedness in cultivated plants and their wild progenitors. Journal of Heredity 79, 225238. Allard, R.W. (1992) Reproductive systems and dynamic management of genetic resources. In: Dattee, Y.C., Dumas, C. and Gallais, A. (eds) Reproductive Biology and Plant Breeding. Springer Verlag, pp. 325334. Ashby, J.A., Braun and IPRA team (1999) Investing in People: the FarmerResearcher as the Protagonist in Rural Development. International Center for Tropical Agriculture, IPRA program, Cali. Berg, T. (1996) Devolution of plant breeding. In: Sperling, L. and Loevinsohn, M. (eds) Using Diversity: Enhancing and Maintaining Genetic Resources On-farm. Proceedings of a Workshop held on 1921 June, 1995, New Delhi India. International Development Research Centre, New Delhi, pp. 116123. Ceccarelli, S. (1994) Specic adaptation and breeding for marginal conditions. Euphytica 77, 205219. Ceccarelli, S., Valkoun, J., Erskine, W., Weigand, S., Miller, R. and Leur, J.A.G.V. (1991) Plant Genetic Resources and Plant Improvement as Tools to Develop Sustainable Agriculture. International Center for Agricultural Research in the Dry Areas (ICARDA), Syria. CIAT (International Center for Tropical Agriculture) (1982) Proceedings of the Conference on Improved Seed for the Small Farmer, 913 August, Cali, Colombia.

434

L. Sperling et al. Cromwell, E. (1997) Local-level seed activities: opportunities and challenges for regulatory frameworks. In: Tripp, R. (ed.) New Seed and Old Laws: Regulatory Reform and the Diversication of National Seed Systems. Intermediate Technology Publications on behalf of the Overseas Development Institute, London. DeFoer, T., Kamara, A. and DeGroote, H. (1997) Gender and variety selection: farmers assessment of local maize varieties in southern Mali. African Crop Science Journal 5(1), 6576. Gomez, F. (1995) Maize breeding with smallholder farmers in hillside cropping systems. In: Cornell International Institute for Food, Agriculture and Development (CIIFAD) Annual Report 1995. Gomez, F. and Smith, M. (1996) Conservation and enhancement of maize with small farmers in Honduras. A collaborative project between Escuela Agricola Panamericana (EAP)Zamarano and the Cornell International Institute for Food, Agriculture and Development (CIIFAD). Project report. Mimeo. Haugerud, A. and Collinson, M.P. (1990) Plants, genes and people: improving the relevance of plant breeding in Africa. Experimental Agriculture 26, 341362. Hecht, S. (2001) Social Issues in Participatory Plant Breeding. Systemwide Program on Participatory Research and Gender Analysis for Technology Development and Institutional Innovation (PRGA), Cali. ICARDA (International Center for Agricultural Research in the Dry Areas) (1994) Farmer Participation and Use of Local Knowledge in Breeding Barley for Specic Adaptation. A Proposal Submitted to BMZ. Iglesias, C.A., Hernandez, L.A. and Lopez, A.J. (n.d.) Participatory Selection with Cassava Farmers in Northern Colombia. International Center for Tropical Agriculture, Cali, Colombia. Ipinge, S.A., Lechner, W.R. and Monyo, E.S. (1996) Farmer participation in the evaluation of priority plant and grain traits on-station; the case of pearl millet in Namibia. In: Droughttolerant Crops of Southern Africa: Proceedings of the SADC/ICRISAT Regional Sorghum and Pearl Millet Workshop, 2529 July 1994, Gabarone, Botswana. International Crops Research Institute for the Semi-Arid Tropics, Andhra Pradesh, India. Lilja, N. and Dalton, T.J. (1998) Developing African public goods: rice varietal selection in Cte dIvoire. Paper presented at American Agricultural Economic Association Annual Meeting, Salt Lake City, Utah, 1998. McGuire, S., Manicad, G. and Sperling, L. (1999) Technical and Institutional Issues in Participatory Plant Breeding: from the Perspective of Farmer Plant Breeding. CGIAR Systemwide Program on Participatory Research and Gender Analysis for Technology Development and Institutional Innovation. Working Document No. 3. Cali, Colombia Nazarea-Sandoval, V. (1995) Indigenous decision-making in agriculture: a reection of gender and socioeconomic status in the Philippines. In: Warren, D.M., Slikkerveer, L.J. and Brokensha, D.W. (eds) The Cultural Dimension of Development: Indigenous Knowledge Systems. Intermediate Technology Publications, London. Soliman, K.M. and Allard, R.W. (1991) Grain yield of composite cross populations of barley: effects of natural selection. Crop Science 31, 705708. Sperling, L. and Ashby, J.A. (1997) Participatory plant breeding: emerging models and future development. In: Tripp, R. (ed.) New Seeds, Old Laws. Intermediate Technology Publications, London. Sperling, L. and Ashby, J.A. (1999) Moving participatory breeding forward: the next steps. In: Collinson, M. (ed.) Book of History of Farming Systems Research. Sperling, L. and Loevinsohn, M. (1993) The dynamics of improved bean varieties among small farmers in Rwanda. Agricultural Systems 41, 441453. Sperling, L. and Scheidegger, U. (1995) Participatory Selection of Beans in Rwanda: Results, Methods and Institutional Issues. Gatekeeper Series no. 51, International Institute for Environment and Development, London.

Base-broadening for Client-oriented Impact

435

Sperling, L., Loevinsohn, M. and Ntambovura, B. (1993) Rethinking the farmers role in plant breeding: local bean experts and on-station selection in Rwanda. Experimental Agriculture 29, 509519. Sthapit, B., Joshi, K.D. and Witcombe, J.R. (1996) Farmer participatory crop improvement III: participatory plant breeding, a case study from Nepal. Experimental Agriculture 32, 479496. Thiele, G. (1999) Informal potato seed systems in the Andes: why are they important and what should we do with them? World Development 27(1), 8399. Weltzien, E., Whitaker, M.L. and Dhamotharan, M. (1996) Diagnostic methods for breeding pearl millet with farmers in Rajasthan. In: Sperling, L. and Loevinsohn, M. (eds) Using Diversity: Enhancing and Maintaining Genetic Resources On-farm. Proceedings of a Workshop held on 1921 June, 1995, New Delhi, India. International Development Research Centre, New Delhi, pp. 127139. Weltzien, E., Smith, M., Meitzner, L.S. and Sperling, L. (1999) Technical and Institutional Issues in Participatory Plant Breeding from the Perspective of Formal Plant Breeding: a Global Analysis of Issues, Results and Current Experience. CGIAR Systemwide Program on Participatory Research and Gender Analysis for Technology Development and Institutional Innovation. Working Document No. 3, Cali, Colombia. Witcombe, J. (1996) Decentralization versus farmer participation in plant breeding: some methodology issues. In: New Frontiers in Participatory Research and Gender Analysis. Proceedings of the International Seminar on Participatory Research and Gender Analysis for Technology Development, September 914, 1996. CGIAR Systemwide Program on Participatory Research and Gender Analysis for Technology Development and Institutional Innovation. Cali, Colombia, pp. 135154. Witcombe, J., Petre, R., Jones, S. and Joshi, S. (1999) Farmer participatory crop improvement IV. The spread and impact of a rice variety identied by participatory varietal selection. Experimental Agriculture 35, 471487.

Final Statement by Participants in the Rome Workshop on Broadening the Genetic Base of Crops, September 1997

It is widely accepted that future sustainable increases in crop productivity will require, amongst other things, an increased use of plant genetic resources, through plant breeding. Yet, despite the great amount of diversity assembled in collections, and available in situ in the world at large, lack of diversity/variation within the particular pools of genetic resources available to breeders and to farmers can often limit progress. For some crops, this can result in yield plateaus, inability to overcome biotic and abiotic constraints, or even crop failure. Historical factors such as domestication and crop movements have meant that, at global and local levels, only a fraction of the potential genepool for some crops has been tapped for agricultural production. For many specic agroenvironments there is an extremely limited pool of adapted germplasm available to farmers and breeders, by comparison with what could be made available to them. There is a need for the increased use of exotic or unadapted plant genetic resources, and this should be based on a clear strategy that is well founded on ecological principles. But a major problem hindering the utilization of plant genetic resources is that of getting genetic diversity into a form that can be more easily used by breeders and farmers. In fact, much of the germplasm available in the primary, secondary and tertiary genepools of crops is not in a form which is of immediate and direct utility. Much pre-breeding or genetic enhancement work is concerned with the introgression of particular genes or gene complexes conferring identied agronomic characteristics into existing elite germplasm. While useful in achieving particular goals for crop improvement, such approaches have little impact on the overall ow of genetic material into the cultivated genepool of the crop in question. There is a need for wellcoordinated programmes on a large scale to broaden the genetic base of identied important crops. This would involve the development of several locally adapted pools of genetic resources, from diverse and previously unadapted material. In a sense, such programmes might re-trace crop domestication or evolution. One approach is to establish composite plant populations by crossing a number of genetically different accessions with a view to maximizing the genetic diversity, and then allowing such populations to progressively adapt to local conditions in a range of different locations,
437

438

Final Statement

for each of which a need for novel germplasm or new variability is identied. Some scientists have called this evolutionary breeding, where natural selection does much of the early-stage work. While the techniques required are cheap and technically rather easy, long-term programmes of the order of a decade (though this depends on the crop concerned) are required to generate tangible results, in the form of material which is readily usable in breeding programmes. This approach is fundamentally different from the introgression of selected characteristics from diverse genepools into existing usable populations, and is complementary to it. Crops vary, both by species and by region of cultivation, in the amount of diversity present in the material currently available to breeders and farmers, and hence in the contribution that such base-broadening work could make. Priority crops would have to be identied on the basis of assessments of needs at the global, regional, national and local levels, of the extent of genetic diversity potentially available, and on feasibility. It may be useful to develop sets of indicators and criteria to facilitate assessment of which crops, environments or end-users might benet from genetic base-broadening. Initially the work should focus on key crop and production situations where diversity is available for selection appears in some way to be an important limiting factor in crop improvement. The adapted pools of crop genetic resources, developed through base-broadening programmes, can feed into formal plant breeding systems, and into informal crop improvement systems of farmers. For some crops and regions, plant breeders in the formal sector have already found the material developed from long-term composite crosses useful. The products of such base-broadening schemes have allowed them to introduce into their breeding programmes substantial new diversity which is already adapted to particular production environments. Such populations have proved to be useful sources of new adapted combinations of useful characters such as disease resistance, phenological adaptation and stress tolerance. There is a complementary need to broaden the genetic base of crop production at the farmers eld level. Increased access by farmers to diverse genetic resources, within and between species, would allow increased productivity and resilience of farming systems. Appropriately designed base-broadening strategies could provide adapted material for use in developing cultivars for use by farmers, including poor farmers in difcult environments, and potentially they may also make a valuable contribution to direct farmer-based crop-improvement efforts. Additionally, there is much scope for improving the availability to farmers of potentially useful germplasm that is already available in genebanks and formal plant breeding programmes. Decentralized approaches to plant breeding and varietal selection are particularly important in marginal production areas where there are likely to be benets from direct selection in the target environment. In turn, decentralized selection efforts through farmerbreeder cooperation may allow the development and rational deployment of a range of varieties specically adapted to local ecologies. Work of this type could build upon the experience of some 50 existing cases of participatory plant breeding, involving national agricultural research systems (NARS), non-governmental organizations (NGOs), farmers organizations and the international agricultural research system, and upon the methods and experience in farmer research and education pioneered through the Farmer Field Schools and similar decentralized research approaches (e.g. the local committees for agricultural research (CIALs) in Latin America). In turn this will require institutional support and capacity building at the national and community levels.

Final Statement

439

Already the worlds governments have recognized the need for large-scale broadening of the genetic base of some important crops, both in the FAO Global Plan of Action and through the Convention on Biological Diversitys decision on agricultural biological diversity. They have also recognized, however, that this is a classic public good problem where individual incentives to individual nations, breeders, companies and farming communities to enhance and adapt germplasm for strategic long-term needs are insufcient to mobilize action. Thus long-term cooperative approaches among different countries and institutions are needed in this area, most likely led by the public sector. Indeed, for many staple crops of poor people where there is little nancial incentive for the private sector, the public sector has a clear responsibility. Such activities must be founded on scientic, ecological principles, and be developed with the full and meaningful involvement of actors at all levels: local communities, national programmes and institutions, and international organizations. The facilitation and coordination of such long-term cooperative programmes is an area where the Food and Agriculture Organization (FAO), the International Plant Genetic Resources Institute (IPGRI) and the other centres of the Consultative Group on International Agricultural Research (CGIAR) have an important role, indeed a responsibility, to play.

Specic recommendations
FAO should launch a Base-broadening Initiative in close collaboration with IPGRI, other CGIAR centres, the NARS, existing crops networks and other relevant institutions, in order to: 1. Facilitate the establishment of international networks of partners (institutions, individuals) to promote base-broadening projects/programmes, on a crop-by-crop basis (making use of existing crop network structures where possible). 2. Carry out reviews of the state of diversity within the cultivated genepools of crops of importance to global and national food security, and develop indicators and criteria to identify crops in possible need of genetic base-broadening both globally and nationally. 3. Facilitate, for identied crops, the development of long-term projects and programmes on genetic base-broadening through multipartner approaches. 4. Promote, in collaboration with educational institutions, training in approaches to base-broadening and population genetics, especially for developing-country scientists and students. 5. Recognize, strengthen and build upon on-going management of genetic resources by communities, and take steps to increase the diversity of genetic material available to farmers and rural communities (for example, by promoting stronger linkages between formal plant genetic resource conservation institutions, genetic enhancement and plant breeding programmes with farmers, and exploring the potential, as appropriate, of existing mechanisms such as the Farmer Field Schools developed through the FAO IPM programmes. 6. Promote awareness on the need for and modalities of genetic base-broadening, at policy and technical levels, through appropriate meetings, publications and the Internet. High priority should be given to these areas by FAO, and the international agricultural research system. They should form a central part of the programme of work of FAOs

440

Final Statement

Division on Plant Production and Protection. Also, support for these activities should be developed within the joint FAO/IPGRI programme to facilitate implementation of the Global Plan of Action, and in collaborative efforts to implement the CBDs decision on agricultural biological diversity. FAO should actively seek the necessary nancial resources for such work, including from extra budgetary sources.

Participants
Julien Berthaud, IRD David Cooper, FAO Isabelle Goldringer, INRA Toby Hodgkin, IPGRI Peter Kenmore, Global IPM Facility, c/o FAO Eric Kueneman, FAO Tonie Putter, FAO Louise Sperling, CIAT Charlie Spillane, consultant Bill Spoor, SAC Edinburgh Murthi Anishetty, FAO Niek van der Graaf, FAO Amor Yahyaoui, Ministry of Agriculture, Tunisia

Index

Note: page numbers in italics indicate tables and gures.

acceptability 423, 427430 access 110111, 142144 accessibility 423, 433 local-level germplasm 430432 adaptability 3, 5, 423, 424427 adaptation 3, 77, 427 barley 353357, 361364, 366 breeding for specic 112, 116, 122126, 414 genetic basis of 28, 44 lentil 343344 maximizing specic 126131 over space/time 118 stress 354, 355, 366 tropical to temperate 168169 wide 408409 adaptive landscape 247 adaptive peaks/valleys 67 ADG see Andigena group adoption 118, 427 Aegilops spp. 34, 51, 250 Afghanistan 342, 343 AFLP see amplied fragment length polymorphism Africa, sub-Saharan 169175 African cassava mosaic virus 41 agamospermy 37 agave 51 Agricultural Research Service 322 agricultural systems changes in 5152 industrial 4, 5, 8 Agrobacterium 27 Algeria 123, 127 allopolyploids 33, 34 aluminium tolerance genes 27 Amaranthus spp. 39, 40 Amazon 12, 91, 98, 202, 218 amplied fragment length polymorphism 94, 203, 205, 393 Andes 373383 as centre of diversity 38, 39 cropping patterns in 374

potato in 185, 191, 385 sweet potato in 38 Andigena group 7778 AndigenaNeotuberosum project 72, 78 angiosperms 50 angular leaf spot 376, 379 anthracnosis 376, 379, 382 aphids 166, 183 APIC 192 apple 39 apomixis 7, 37, 164, 209 Arabidopsis 28 Arachis spp. 29, 32, 34 Argentina 165, 202, 319, 385, 391394 Ascochyta blights 347, 376, 377, 378, 379, 380, 382 ASTA 321322 Australia 47, 116, 166 autogamy 37 avoidance 367 avocado 32 backcrossing congruity 50, 377, 381 inbred 4950 recurrent 49 see also introgression bacterial blight 209, 401 bacterial wilt 181 Balsa 97 Bamako workshop 81 bananas 223240 breeding 230239 centres of diversity 228229 cultivars 225227, 230231 diseases 230, 233234, 236 domestication and spread 225227 Fei 224, 225 bre from 225 genetic resources 224227 geographic movement 38, 39

441

442
bananas continued as outbreeding perennial 4 production 223 research networks 238239 as single clones 5 variety distribution 236 vegetative propagation 37 wild species 224225 see also Musa; plantain Bangladesh 108, 342, 343, 345 Barbados 285, 287, 288, 289, 290 barley adaptive traits 263266, 268, 269, 353355 in crosses 361364 stress 366369 adoption 123 agronomic performance 263264 as animal feed 118 base-broadening 13, 261269 breeding methods 271 brittleness 354, 362 broad-based pools 12 buffering mechanisms 364365 Cambridge composite cross V 271280 chloroplast diversity 51 as composite cross 6, 353, 425 disease resistance 250, 268, 269, 273274, 278, 354355, 357358, 360, 363, 388 diseases, leaf 72 see also leaf rust; powdery mildew diversity 364 domestication and spread 351 drought-resistant 268 dynamic management 247 ear emergence 267 early growth traits 368 early-maturing 268 in Ethiopia 124 exotic/adapted progenies, evaluation of 266268 farmers ideotype 127 owering time 274, 275 GE interactions, crossover type 118119 genepools 29, 261, 262, 351 dynamic 262, 268, 353 genetic variation in 47 genetic diversity 351369 grain-lling period 369 growth habit 274, 275 hordeins 278, 279 ICARDA breeding programme 119, 122, 123, 124, 126, 127, 409, 424 as inbreeding annual 4, 43, 412 incorporation 72 landraces 20, 351352, 355369 black-seeded 352, 356358, 360, 365 as breeding material 359369 collections 355 harvest index 362 pure-line selection within 359361 white-seeded 352, 356358, 360, 365 lodging 268 malting 51, 52 markers, biochemical 278 mixtures, breeding 364365 non-adaptation 75 Nordic 261269

Index
participatory breeding 421 plant height 274, 357, 362, 363364 pleiotropy 276 pre-breeding 352353 root length, seminal 357, 367368 seed colour 352, 356358 seed dormancy 356 straw softness 363364 in Syria 116, 127128 variety releases 123 wild species 32, 261, 351, 353354 spontaneous geneow with 33, 354 variability in 32 yields 117, 119, 121, 358359, 360361, 362 enhancement of 351369 stress effects on 361, 366367 barley stripe 360 BARNESA 238 barriers to geneow crossability 3436 ecogeographic 3435 in maize 166167 mating systems as 3738 phenology 40 post-mating 35 reproductive isolation 34 base-broadening approaches to 1120 breeding programmes 57 client-oriented 419433 introgression and incorporation 7179 local-level 420, 422, 424427 phenotypic 27 priorities for 911 research management 137145 Base-broadening Initiative xxi BAZ 301, 303, 395 BCMV see bean common mosaic virus bean, bush 127 bean, common Andean 373383 backcrossing 50 breeding method/strategy 375382 disease resistance, horizontal 399406 domestication bottlenecks 9 ecogeographic barriers 34 genepools 29, 46 germplasm collection 375 growth cycle 374 intercropping, breeding for 373383 intraspecic breeding 47 landraces, variability in 357 maize intercropping 20, 82, 373383 land equivalency ratio for 381 Mexican breeding programme 399406 farmer participation 405406 line releases 402 selection pressures 401, 403 yields 402 outbreeding depression 36 participatory breeding 421 pest resistance 399406 in Rwanda 425 seed production 373 wild species 32 spontaneous geneow between 33 yields 26, 402, 403 see also Phaseolus

Index
bean, faba 421 bean, mung 38 bean, runner 35 bean common blight 400, 402, 405 bean common mosaic virus 400, 401, 402405 bean golden mosaic virus 41 beet see Beta; leaf beet; sugar beet beet cyst nematode 297 beet necrotic yellow vein virus 297 Bemisia sp. 400 benet sharing 110111, 143 Benin 153 Beta 295307 base-broadening 301306 bolting 296, 299300, 302 bottlenecks 299301 characters, useful 298 disease resistance 297, 300, 301, 302 Doggett populations 303 domestication 295 genebanks, future role 306307 genepools 298 incorporation 302306 introgression 301302 insect resistance 301 nematode resistance 302 pollen dispersal 43 sections 295, 297, 298, 301302, 303 wild species 295, 296, 299, 300, 301303, 306 see also leaf beet; sugar beet Beta Networks 298, 303 biodiversity, varietal 413 biolistics 27 biosafety 30 biosphere reserves 97 biotechnology 327 Bipolaris maydis 320 Birsa Agricultural University 413 black gram 421 black sigatoka leaf spot 41, 233, 234, 236 blocks, parental linkage 49 Bolivia 376, 385, 431 bolting 296 bordagol rice 140, 143, 145 bottlenecks in Beta 299301 domestication 9 polyploidy-induced 3334 founder effect 9, 3841 genetic 8, 10, 19, 26 crop cytoplasmic 5051 disease-induced 41 potato 185187 in primary cultivated genepool 3052 social factors and 42 geographic 3841, 208 global 30 historical 1516, 19 institutional or regulatory 11 modern plant breeding 9 spatial or geographic 10 Boyce Thomson Institute 231 Bracharia 37 Brassica 10, 52 Brazil 424 bananas in 231

443
cassava in 9195, 201, 202, 203, 204, 207, 212213, 218 maize in 161, 175, 319 sugar cane in 286 breeders exemption 109110 breeding adaptive 414 alternative 110, 112113 bulk-population 217 capacity limitations 10 centralized 414, 415, 416 wide adaptation in 408409 client-orientated 419433 competitive 111 composite-cross 271280 decentralized 20, 122126, 128, 131, 408, 409414, 415416 collaborative participatory 411414 international 409410 participatory 130131, 410414 for marginal environments 115132 multiple 426 organizational issues in 426427 by rening target areas 409 by shifting responsibilities 410 by targeting national programmes 409 technical approaches to 425426 for diversity, regulatory aspects of 105114 elite 352 evolutionary xx, 12, 193, 245, 271 geneow constraints in 4548 intensive 19 line, pure 271 for marginal environments 115132, 191192 participatory plant see participatory plant breeding pedigree 271 pre- see pre-breeding privatization of research into 111 public sector 48 shuttle 121 for stress environments 117118 transgenic 27 breeding material, elite 43 breeding pools 419420 breeding programmes banana 230239 barley 271280, 351369 base broadening 57 bean, common, in Mexico 399406 Beta 295307 cassava 202, 208, 210211 formal 126 international 122126, 409410 lentil 341348 maize 311317, 319328, 331339 national 410 participatory 410414 off-farm 5 Phaseolus for intercropping 373383 potato 188195, 385395 sugar beet 296 sugar cane 283292 breeding true 38, 412 bridge crosses 34, 36, 272 brown plant hoppers 139 bruchids 41

444
BUCAP 139, 144 bulks 123, 125, 131, 412, 413 Bureau de Ressources Gntiques 302 Burkina Faso maize in 8688, 98 pearl millet in 90, 147, 149, 151 Cameroon 91, 171, 430 Canada 166, 311 Capsicum 10, 35 Caribbean 39, 286 carrot 33 cash crops 430 cassava 4, 201218 agronomic evaluation 210 Amazonian 9195, 98, 202, 218 bitterness 202 breeding programmes 202, 208, 210211 base-broadening 214218 participatory 421, 425, 426, 431 centre of diversity 203 collections 207, 209, 210 conservation networks 214 cultivars 210, 212 disease resistance 41, 207, 208, 210 divergence 207 diversity molecular markers 205208 domestication and spread 38 dynamic management 12, 216 gametes, unreduced 33 genepools 29, 202 genetic diversity within 9495, 202208, 209210, 211 incorporation programmes for 14 landraces of 208, 209 multiple-cropping 430 research networks 215 tree 205 vegetative propagation 37, 100 wild ancestor 207 wild species 211 yields 26, 201 see also Manihot cassava mosaic disease 205, 207, 208, 209, 210 cauliower 52 CAZS 413 CBD see Convention on Biological Diversity ceara rubber 41 CENARGEN 217 Central African Republic 91 centres of diversity 19, 38, 46, 147, 203, 228, 284, 373 Cercospora beticola 296, 300 cereal pot trials 75 CGIAR xxi, 71, 237, 408, 415, 421 cherry 33 chestnut blight 41 chickpea 9, 29, 35, 47, 118, 421 China 38, 39, 51 banana in 227 barley in 125 Beta in 295, 297, 298, 303 cassava in 201, 212213 maize in 162, 166 potato in 181 sugar cane in 285

Index
Chionachne 161, 164 chromosome doubling 34 chromosome pairing 36 chromosome rearrangement 37 CIALs xx, 426 CIAT 94, 205, 207, 209, 210, 214, 374, 375, 426 Cicer spp. 29, 35 CILSS 153 CIMMYT 48, 87, 160, 162, 165, 166, 167, 170, 172, 174, 175, 314, 333, 408, 409, 410, 415 CIP 183, 191, 192, 194, 389, 390392, 393, 394395, 431 CIRAD 87, 230, 231, 232, 333 Citrus 9 client-oriented breeding 419433 climate change 9091 clonal reproduction 4, 5, 37 coadaptation 37 cocksfoot 118 cocoa 9, 15, 19, 40, 74 coconut 4, 39 coffee 9, 40, 41 Coix 161, 164 Colletotrichum lindemuthianum 376 Collaborative Research Support Programmes 154 collaborators 427 collections 5, 71, 72, 93 banana 227 cassava 207, 209 core 12 maize 162 millet 148 potato 192, 194, 390 Colombia 424 cassava in 203, 212, 425, 426, 430, 431 maize in 319 Phaseolus in 374, 376, 377, 379, 381 potato in 385, 390 Colorado potato beetle 183, 390 Columbian exchange 38 Columbus, Christopher 161 combining ability 47, 193, 335, 392 commercialization 111, 160 Commonwealth Potato Collection 183 composites 411412 Congo 201, 209 conjoint analysis 429 conservation dynamic 216218, 256 on-farm 138, 141, 143, 191, 217 CONSERVE project 140141, 142, 143, 144, 145 consumer preferences 5152 Convention on Biological Diversity xv, xxi, xxii, 2, 110111 conversion 44 cooperation, international 394395 cooperatives, farmer 145, 426 CORAF 153 CORPOICA 374, 426 Costa Rica 233, 376, 379 Cte dIvoire 201 cotton 3334, 4748, 87, 142, 421 covered smut 357, 360 cowpea 35, 38, 87, 344, 421, 430 CPRO 394 CRBP 231, 239

Index
criteria agronomic 429 farmer selection 428, 429 crop diversity, in-eld 89 crop evolution, on-farm 8 crop geographic movements 3841 crop reconstruction approaches 19 crop vulnerability 8, 9, 10, 41 crossability barriers 3436 crossability transgenes 36 crosses composite 1215, 72, 271280, 425 test- 75 wide 20, 34, 36, 415 cross-fertilization 36 crossover 116, 118122 cross-pollination 4, 432 crotelaria 421 Cryphonectria parasitica 41 Cuba 161, 165, 212, 235, 236, 286 Cucurbita spp. 82 cyanhydric acid 202 cytoplasm 5051, 187188 daylength 9, 19, 40 pearl millet and 89 potatoes and 78, 186, 190 see also photoperiodism decentralization see breeding, decentralized depression inbreeding 37 outbreeding 36 developmental rate 40 DGER 247 dice coefcient 128 diffusion pattern 8485 disease epidemics 9, 19 genetic-bottleneck inducing 41 disease resistance 78 breeding for 112 horizontal 20, 77, 387389, 400, 402405 vertical 112 diversity availability to farmers 78 breeding for 112 regulatory aspects of 105114 chloroplast DNA 51 genetic 3, 11, 25, 52 reduced 46 inter/intravarietal 91, 97 isozyme 97 on-farm 810, 100 phenotypic 46 temporal 5, 9 diversity enhancement 426 DNA chip 28 domestication bottlenecks 9 founder effect 3133 Dominican Republic 212, 287, 288 doubled haploid techniques 413 drift, genetic 6, 28, 30, 98, 246 drought 9091, 119, 167, 255, 347 DUS 106, 109 dwarng 152, 255256 dynamic management 12, 14, 98100, 245258 assessment of 254256 experimental protocol for 247248 principles of 246247 propositions for 256258 earliness 90 East Indian Rainfed Farming Project 413 Ecuador 207, 212, 431 Egypt 123, 295 elite germplasm 56 EMBRAPA 231, 322 embryo rescue 34, 36, 231 Empoasca fabae 183 endosperm balance number 35, 182, 183184 Ensete spp. 224 environments marginal 115132 unfavourable 116, 120 enzymatic polymorphism 338 epistatic effects 9, 19, 30, 48 Epitrix spp. 183 Erianthus spp. 285, 286, 290291 Eritrea 108, 125, 280, 361 Erysiphe graminis 249250, 255, 273 see also powdery mildew ESD see seed descent, equal Ethiopia 124, 171, 224, 358, 361, 424 EU 192 Euchlaena 161 Eumusa spp. 224, 225 eyespot 250

445

FAO xv, 71, 110, 141, 227 Global Plan of Action xv, xvi, xvii, xxi, 2 Farmer Field Schools xx, 141142, 143, 426 farmers access by 78, 11 empowerment of 137145 exchanges by 99100 geneow and 4143, 98100 participation by 8, 99100, 108, 113, 126131, 193, 208, 245, 405406 see also participatory plant breeding role of 2 selection by 43, 95, 112, 118, 126131, 194, 413 small-scale 122, 138, 175, 236237 farmers cooperatives 144145, 426 farmers needs, diagnosis of 429430 farmers privilege 109110 farmers rights 110111 Fertile Crescent 351, 364, 369 FHIA 230, 236 eld types 86 Finland 262, 266, 268, 353 tness 6 owering time 36, 4041, 44, 275 fonio 147 founder effects 9, 19, 30 of domestication 3133 of geographic movement 3841 Fragaria spp. 3940 France 208 Beta in 298, 301, 303, 306 maize in 161, 165, 331339 wheat in 14, 98

446
fruit crops 42 Fusarium spp. 400 Fusarium wilt 230, 233, 234, 235, 238 gametes, unreduced 33 gametic incompatibility 35 GE interactions 115116, 118122, 124, 126, 217, 392395 crossover 116, 119, 121, 424 GEM 19, 165166, 169, 319328 Gembloux 374, 377, 379 gene transfer 20, 27, 36, 40, 88 genes, patents for 110 genebanks 2, 3 geneow 10, 2830 biosafety-related 30 constraints in breeding 4548 crossability-barrier limited 3436 depressions affecting 3637 domesticant/wild 323, 9698 exotic/adapted genepool 4345 farmers and 98100, 110 eld 4143 local-level 43, 81101 mating system effects on 3738 phenology as barrier to 4041 recombination and 4850 weeds and 42 genepools 2829 for crop improvement 2628 dynamics 2553 genetic bottlenecks 3052 quaternary 27 wild 3, 36 genetic base, narrow 10, 26, 40, 46 macro-economics and 111 genetic gain 10 genetic isolation 90 genetic linkage 37 genetic resources, access to 110111 genetic revolution 30 genome, D 34 genomics 28 genotype environment, see GE interactions genotypes 38 germplasm adapted 7, 4345 elite 56, 43, 45 see also varieties, elite exotic 6, 7, 4344 local-level acceptability 427430 accessibility 430432 adaptability 424427 pre-adapted 19 wild 43 GILB 192, 386 Globodera spp. 181 Glycine spp. 32, 39, 49 see also soybean golden nematode 181 Gossypium spp. 3334, 47 see also cotton grain legumes 72, 74 Gramineae 27 grape 39 Greece 295 green mites 209 green peach aphid 183 Green Revolution 105, 122, 138, 139, 423 groundnut 29, 32, 35, 38, 47, 87 Guatemala 376, 379 Guelph, University of 400 Guinea 91 Guizota abyssinica 38 Gujarat 410411 Guyana 162, 207, 287, 288

Index

hand-pollination 235, 401 haploidization 34 Hawaii 231, 287 Helianthus spp. 35 Helicotylenchus multicinctus 234 Helminthosporium maydis 51, 332 hemp, Manila 224 heritability, high 46, 121 Heterodera schachtii 297, 299, 302 heterogeneity 4 Hevea spp. 202 hitch-hiking 252253 homogeneity standards 109 Honduras 230231 HOPE breeding system 19, 311317 Hordeum spp. 29, 32 Hordeum spontaneum 261269, 351355, 359364, 367 Hordeum vulgare 261269, 271 see also barley hordeins 278, 279 hybrid breakdown 35 hybrid dysgenesis 3637 hybrid embryo culture 36 hybrid inviability 35 hybrid sterility 34, 35 hybridization 4, 32, 33, 36, 39, 272 IAEA 235 IARCs 160, 170, 172, 173, 409, 410, 415 IBPGR 71, 225 ICA 162 ICARDA 119, 122, 123126, 127, 343, 345, 346347, 353, 369, 409, 424 ICRISAT 148, 151, 153, 154 ICTA 230, 231 IIRB 298 IITA banana and 231, 234, 235, 239 cassava and 209, 214 maize and 160, 166, 170, 172, 174, 175 inbred backcross method 49 inbreeders 4, 37, 412 incorporation 6, 1119, 7179, 261 Beta 301, 302306 genetic principles 7477 India 38, 431 Beta in 295, 298 cassava in 202 lentil in 342, 343, 345, 346 maize in 162, 166, 410411 pearl millet in 147148, 151, 420, 430 plantain in 226, 227, 228229

Index
sorghum in 107 sugar cane in 285, 286, 287 wheat in 414 Indonesia 142, 395 cassava in 201, 202, 213 Musa in 224, 227 inertia effect 47 inorescence traits 89 INIBAP 227, 236, 237239 INIFAP 162 INRA 153, 247, 332, 333, 394 INRA/PROMAIS programme 19 INSAH 153 institutional support 142144, 153154, 247 INTA 389, 391, 394 integrated crop management 385 integrated pest management 141142 intellectual property rights 108110, 114 intercropping 87, 107, 345, 373383 International Legume Testing Network 345 International Nursery Network 346347 introgression 4, 6, 7179, 86, 99, 261 Beta 301302 maizeteosinte 96, 97 permeability to 48 Intsormil network 154 IPGRI xv, xxi, 71, 148, 214 Ipomoea spp. 29, 40 Iran 123, 295, 298, 342 Iraq 123124, 361 IRAT 153 Irish potato famine 1, 185186, 385 IRRI 138, 139, 408, 410, 424 ISNAR 410 ISO 52 isolation genetic 90 reproductive 34 seasonal 35 isozyme diversity 97 isozyme markers 88, 90, 207 ISWYN 409 IWIS 48 Jamaica 230231, 287, 288 Japan 10, 47, 162, 284, 295 Java 285286 Jordan 355, 360, 366 Lactuca spp. 32 LAMP 19, 165166, 168, 319324, 326327 land equivalency ratio 381 landraces 43, 46, 47, 84, 415, 416 geneow exchange between 99 regulations for 112 leaf beet 295, 297 leaf blight 378 leaf rust 255, 354 leaf spot disease 300 Lebanon 345 Lens spp. 29, 32, 341348 lentil 9, 1920, 40, 124, 341348, 357 adaptation 343344 breeding programmes 346347, 421 genepools 29 hybridization of 344346, 347 pilosae ecotype 341344, 345, 347 plant introduction, South Asia 344 rust-resistant 345, 347 seed size 345346 variation in 341342 wild relatives of 32 Leptinotarsa decemlineata 183 lethality 36 lettuce 32 LI-BIRD 413 linkage drag 44 linseed 345 LKF 394 lodging 87, 107, 127, 335, 356 Lycopersicon spp. 32

447

macro-economics 111 Macrophomina phaseolina 400 Macrosiphum euphorbiae 183 Maghreb 123 maize 159176, 311317, 319328, 331339 accession performance 334 /bean intercropping 20 blight epidemic 112 breeding programmes 1618, 412, 421 biotechnology 327 decentralized 410411, 426, 430 French 331339 GEM 319328 Guelph 314, 315 HOPE 311317 collections 162 cytoplasmic male sterility 50 dents 162163, 165, 320 diseases 164, 166, 170, 175, 403 diversity 162163, 170171, 320321 dynamic management 12 elite lines 159, 168, 174175, 332 ints 161, 162163, 332 oury 162 GE interactions, crossover 118 genepools 160164 genetic base, narrow 9, 19, 320321 genetic base broadening 164166, 168169, 173175 genome project 173 geographical movements 38, 161 germplasm banks 162 germplasm barriers 166167 heterosis 167168 high-yielding varieties (HYVs) 160 hybrids 4, 160, 167, 168, 173174, 332 improvement networks 172173 inbreds 44, 49, 174175 inbreeding depression 37 inertia effect in 47 intercropping 82, 87, 373374, 375, 381 in Kenya 171 local-level geneow Burkina Faso 8688 Mexico 8286, 9698 major-effect genes 31 molecular markers 334, 338 OPVs 160, 162, 164166, 167, 170, 171 outcrossing 4

448
maize continued photoperiod sensitivity 167, 168169 pre-breeding 44 seed selection 100 stalk quality 314, 323, 326 stress vulnerability 10 in sub-Saharan Africa 169175 synthetic sources 166 T-cytoplasm 160 temperate 164165 /teosinte hybrids 49, 9697 tropical 165166 in USA 165169, 320 wild relatives 32, 33 yields 26, 159, 337 in Zimbabwe 170 maize borer, European 326 maize streak virus 88, 170, 175 maladaptation 7 Malawi 172 Malaysia 224, 227, 228, 235 Mali 8991, 147, 149150, 151, 152, 424, 430 Malus spp. 39 Mamestra brassicae 301 Manihot spp. 29, 201, 202208 dynamic conservation 216218 genetic diversity 208214 hybridization between 209 molecular markers 204, 205208 RAPD techniques 203, 205, 207 Manihot esculenta ACMV resistance 41 see also cassava mapping genetic/molecular 2728, 36 marker-assisted 44 QTL 4445 marginal environments 115132 markers biochemical 278 isozyme 88, 90, 207 molecular 28, 45, 47 MASIPAG 140 mass reservoirs 12 mating systems, effects of on geneow 3738, 4850 Mauritania 9091 mealybugs 209 mechanization, agricultural 52 mega-environments 408 Meloidogyne spp. 181 Mexico 424 cassava in 212 maize in 8286, 9698, 160, 161, 166 Phaseolus in 375, 379, 399406 potato in 386387, 390, 394 sugar cane in 287 wheat in 408 MGIS 237, 240 microarray technologies 28 migration 9 millet breeding programmes, national 153 collections 148149 diversity 148150 use of 151153

Index
dwarng 152 foxtail 33 genepools 148 growing season 149150 improvement 151152 pearl see pearl millet photoperiodism 149, 150, 151, 154155 research, collaborative 153154 variability within 149 varieties early/late 149 synthetic 151 Miscanthus spp. 285, 286, 287 modied bulk population 413 molecular diversity analysis 46 multilines 112 Multiple Population Breeding Strategy 19 mung bean 38 Musa spp. 223240 breeding programmes 230235, 237 biotechnology 231233 decentralized 235 mutation and somaclonal 235 chromosome numbers 224 collections 225, 227 cultivars 225227 disease resistance 233235 domestication bottlenecks 9 farmers varieties 231 owers, hermaphrodite 224 genetic resources 224233 geographic movements 38 germplasm, movement of 230 hybridization between 225 molecular marking 238 networks 237239 polyploidy 225, 226, 230 smallholders and 236237 virus indexing 230, 240 wild species 224225, 233, 237 see also banana; plantain muskmelon 118 mustard 47, 345 mutation 3 Mycosphaerella jiensis 233 Myzus persica 183 NAFTA 399 NARs xx, 111, 154, 172, 173, 175, 235, 237239, 389, 410, 419420, 426 NCRPIS 162, 321 nematodes 181, 234, 302 Nepal 342, 343, 358, 361, 412, 414, 423, 424, 425, 431 netblotch 268 Netherlands, The 181 Niger 90, 91, 147, 149, 151, 152 Nigeria 147, 149, 152, 153, 201, 211, 214 niger seed 38 Noctuide loxostege 301 Nordic countries 261269 adaptive traits for 263266 NSSL 162 nuclear restorer genes 50 nurseries 124, 376, 378, 381, 429 international 409, 410

Index
oats 43 oil palm 4, 16, 19, 40 on-farm management 9899, 101 on-farm testing 108 Oryza spp. 32, 35, 36, 415 outbreeding 3, 4, 37, 412 outbreeding depression 3637 outcrossing 4, 12, 43 optimal distance 37 Pakistan 47, 342, 343, 344 Panicum spp. 38 panmictic populations 6 Papua New Guinea 224, 225, 227, 228229, 285 parentage, coefcients of 413 participatory plant breeding 8, 20, 113 base-broadening insights drawn from 419433 farmer-led 421 formal-led 421, 425, 430 impact on genetic base of crops 407416 mapping 423 for marginal environments 115132 overview of 420423 patents 110 PCIM 162 pea 32, 421 peach 39 pearl millet 9, 421, 430 diversity 147155 dynamic management 12 GE interactions, crossover 118 geographic movements 38 intercropped with maize 87 landraces 42 outbreeding depression 36 outcrossing 4 production 147 variability within 149 in West Africa 8991, 147155 wild relatives 32, 35, 148 yields 26 see also millet pedigree analysis 11 Pennisetum spp. 148 Pennisetum glaucum 32, 147, 149 see also pearl millet peppers, hot 10, 46 Persea spp. 32 Peru 187, 374, 377, 379, 385, 390, 395, 430, 431 pesticides 142, 145 PGRFA 2 Phaeisariopsis griseola 376 Phaseolus bean breeding programme, Andean eld 377378, 380382 methods 375378 results 378382 seed increase method 376 collection 376 congruity backcrosses 50 crossability and 34 disease resistance 376, 377, 378379, 380 domestication bottlenecks in 9, 15 genepools 29 growing period 379 hybrids

449
dysgenesis in 36 interspecic 377, 379380 intercropping breeding for 373383 with maize 373374, 381 in Mexico 399406 parasites 399 polyploidy and 34 viruses 403 wild species 32 yields 379, 380381 see also common bean Phaseolus coccineus 35, 374383 Phaseolus polyanthus 374383 phenology 40 Philippines 138, 139141, 142, 212213 Musa in 224, 226, 228229 rice in 425 sugar cane in 286 Phoma exigua 376 photoperiod 40, 149150, 154155, 343, 345 see also daylength Phthorimaea operculella 181 Phytopthera infestans 41, 181, 386387, 389, 390, 391 pigeon pea 9 Pisum spp. 32, 40 plant breeders rights 109 plant breeding, commercialization of 111 Plant Breeding Institute 111 plant genetic resource (PGR) conservation of 138, 145 diversity of 141 on-farm research into 142143 plant height 127 plantain 223, 226227, 228229, 231 decentralized breeding of 235 disease resistance 41 polyploidy and 33 see also banana; Musa spp. plasticity 112 pleiotropy 276 pollen as geneow carrier 43 pollinators 35 polyembryony 37 polyploidy 3334 Portugal 161 potato 5, 181195, 430, 431 Andean 4, 42 bottlenecks 19, 41, 185187 breeding programmes 188193, 385395, 421 evolutionary 193195 long-day Argentinian 391394 for marginal environments 191192 chromosome number 182, 390 co-ancestry analyses 188 collections 183, 194, 390 cross-breeding 194 cytoplasm diversity 187188 disease 181 disease resistance 181 late blight 385395 see also potato late blight domestication 185 EBN 182, 183184 epistatic effects 9

450
potato continued gene banks 183, 184, 192, 390 genepools 182185, 390 genetic base broadening 193195 genetic diversity 182183, 188190 genome mapping 194 geographic movements 38, 39, 185186 hybrids 391, 393 improvement networks 192 landraces 190, 191 molecular markers 190, 194195, 393, 395 mutants 184 Neotuberosum 190 open pollination 186 pests 183 photoperiodism 186, 190, 389 polycrosses 193 polyploidy 33, 34, 182, 183185, 390 propagation 37 selng 183 stress resistance 183 viruses 181, 391392 wild resources 188, 190 see also Solanum spp. yield plateau 10 yields 26, 181, 189190 potato aphid 183 potato ea beetle 183 potato late blight 1, 181, 186, 188, 192, 385395 potato leaf hopper 183 potato leaf roll virus 41, 181, 392 potato tuber moth 181 potato virus Y potyvirus 181 powdery mildew 249250, 255, 273274, 278, 354, 356, 357, 360 PPRI 230 PRACIPA 192 PRAPACE 192 Pratylenchus spp. 234 praxis 142 pre-breeding xix, 6, 7, 44, 141, 142143, 256258, 261 PRECODEPA 192 PROCIPA 192 PROMAIS 19, 332 PROMUSA 237288 propagation, vegetative 37, 93, 95 property rights 431, 433 protoplast fusion 194 Prunus persica 39 Pseudomonas solanacearum 181 Puccinnia recondita 255 Puerto Rico 376 pyramiding 387 Pythium spp. 400 quantitative trait loci 28, 34, 44, 45, 359, 388 quinoa 33 radish 33 Radopholus similis 234 RAPD 94, 203, 205, 207, 393 rape 33, 47 recombination 4, 6, 7, 12, 34, 37, 38, 40, 95 intragenic 48

Index
see also gene transfer recombination rates 4850 regulations 105114 effect on alternative breeding methods 112113 seed 105 relay cropping 374 reproduction asexual 37 sexual 95 research centralization 137138 commercialization 111 empowering farmers with 137145 on-farm 142143, 145 strategic 416 resistance horizontal 77, 107, 112, 387, 388391, 393, 394, 399406 vertical 139 resistance genes 388 resource management 137145 RFLP 207, 252, 279, 332, 334, 338339 Rhizoctonia solani 400 Rhopalosiphum maidis 166 Rhynchosporium secale 250 rice 3, 4 Asian irrigated 138142 biotechnology for 327 bordagol 140, 143, 145 breeding centralized 408409 decentralized 410, 425 participatory 412414, 421, 430 wide-cross 415 domestication bottlenecks 9 genetic base narrowing 10, 46, 47 genome sequencing 28 hybrid sterility 3536, 51 intraspecic breeding 47 introgression permeability 48 major-effect genes 31 in Philippines 139141 recombination rates in 49 red 139 self-pollination 43 varieties high-yielding 138 number of known 4 wild relatives 32, 33, 36 yields 139 rice blast fungus 139 rice grassy stunt virus 41 Rizomania 297, 301, 302 Rocafremi 154 rubber 5, 15, 19, 40, 203 ceara 41 runner bean 35 rural credit 112 rust 345, 347 rustication 42 Rwanda 127, 223, 420, 427, 431 Saccharum spp. 29, 50, 283287 chromosome numbers 285 Saccharum robustum 290291

Index
Sahel 91, 151 scald 357, 360 SCRI 394 SEARICE 140, 142, 143, 145 seed, non-shattering 31 seed certication 105, 107, 108, 144 seed descent equal 413 single 123, 412 seed dispersal 43 seed distribution systems 10, 42, 118, 431432 seed dormancy 31 seed quality control 105106, 108, 114 seed size 49 selection 6 articial 45 background 45 cumulative 377 decentralized 12, 123125, 127, 129, 131 disruptive 36 foreground 45 marker-assisted 20, 45 negative 42 recurrent 377, 381, 389 semi-natural 12 selection criteria aesthetic 92 agricultural 92 farmers 2, 92, 118, 126, 427429, 430 usage 92 selection efciency 127 selection intensity 6, 46, 48 selection strategies, effect on geneow 4850 selng 49, 174, 401 self-pollination 37, 43, 46 Senegal 90, 91, 147, 149, 151, 152, 153 sesame 38 shattering 107 shifting balance theory 246247 sib-mating 174175 SIFT 388 skill-building 422, 426 slash-and-burn 92 smut 166 covered 357, 360 social factors 10, 42, 144145 Solanum spp. 34, 181, 182184, 187188, 190, 194, 386387, 389, 391, 393, 394 sorghum 9, 16, 19, 26, 31, 344, 357, 421 GE interactions, crossover 118 geographic movements of 38 pre-breeding 44 Sahelian 91 variety testing 107 wild species 32 southern leaf blight 1, 51, 160, 164, 320 soybean 9, 32, 39, 40, 107 specic combining ability (SCA) see combining ability squash 33, 38, 82 Stemphylium 345, 348 straw 107, 127 strawberry 33, 3940, 46 stress biotic 8, 10, 139 breeding for environmental 117, 118, 120 pre-anthesis water 116 Striga 170, 175 sugar beet 1415, 295298 bolting 296 breeding programmes 296, 297298 disease resistance 296, 297 evolutionary breeding 12 nematode resistance 298 pollination 296 seed 296 wild relatives, geneow with 33, 296297 see also Beta sugar cane 16, 283292 bottlenecks 19, 50 breeding history 285286 breeding programmes base-broadening 286287 West Indies 288292 clones 286, 289 heterogeneity 4 collections 286287 owering time 290 GE interactions, crossover 118 geneow barriers 37 genepools 29 geographic movements 38, 39 hybridization 33, 283, 290 intercropping 345 nobilization 288289 noble canes 285, 286, 289, 290 origin of 283285 ratooning 287, 288 seed 285 stress vulnerability in 10 yield plateau in 10 sunower 32, 33, 35, 39, 421 sustainability 141 Sweden 262 sweet potato 26, 29, 33, 37, 38, 40, 421 farmers varieties 138 SWP/PRGA 421, 432 synthetics 12, 19 Syria 295, 420, 424 barley in 123, 127131, 353369 lentils in 342, 343, 345 wheat in 116, 120

451

Taiwan 232, 235, 286, 287 taro 33, 138 temperature 40 teosinte 30, 31, 49, 82, 96, 97, 161, 163, 167, 314 tequila 51 testing, decentralized 123 Thailand 166, 201, 210211, 213, 229, 231, 239, 286 Theobroma cacao 40 tobacco 33 tomato 9, 31, 32, 33, 37, 38, 49, 327, 386 top-down approach 122, 138, 145 traits adaptive 44 agronomic 31 farmer-selected 427429 quality 51 transgenics 28 trees, outbreeding 3 Trialeurodes sp. 400

452
Trinidad 230 TRIP 109 Tripsacum 161, 163164, 167, 314 Triticum spp. 29, 33, 40, 47 tungro virus 139, 140 Tunisia 126 Turkey 123, 124, 295, 342 Uganda 108, 223 UNALM 374 UPOV 107, 109 UPWARD 192 US barley in 47 maize in 9, 10, 112, 159, 162, 164169, 319328 rice in 10, 47 USDA 14, 162, 165, 298, 322 users 117, 420, 422, 430 Ustilago maydis 166 vaccine antigen genes 27 variability 35, 6 on-farm 4, 5 variation, sum of 11 varieties farmers 110, 138, 144 local 8, 12 variety, denition of 92 variety collections, exchange between 99 variety portfolios 95, 98 variety protection, plant 109110 variety release systems 105, 106108, 113 variety selection, participatory 8, 20 variety testing 107108, 113114 VCU 106, 113 Venezuela 376, 379 vernalization 4041, 44, 356 Vietnam 139, 142, 201, 212213 Vigna spp. 38, 344 Vitis vinifera 39 WARDA 415, 430 WBN 298 weeds as geneow bridges 42 seed dispersal from 43 weevils 41 West Africa millet in 4, 147155 West Indies 10, 283292 wheat 4 bread 9, 29, 48, 98, 120, 357 breeding programmes centralized 408, 415 decentralized 409, 410 participatory 414

Index
crossability 34 cytoplasmic donor 51 domestication 351 durum 9, 357, 421 dwarng 255256 dynamic management of 12, 14, 98, 245258 experiment in 247248 mating system 253254 principles 246247 results of 249254 earliness in 249 GE interactions, crossover 118 glutenin subunits 51 hexaploid 33 hybridization 33 landraces, increase in parental 48 mildew resistance 249250, 255 molecular markers 251253, 254 multiline 112113 outcrossing 251, 254 plant height 249, 254, 255256 polyploidization 33, 34 pre-breeding 256, 257, 258 resistance genes 388 selection pressures 249, 254 self-pollination 43, 253254, 258 spring 10, 46, 47, 408, 409 traits 249251, 254 vernalization genes in 4041 winter 10, 47, 247 yields 250251 white y 400 wild relatives 32, 35, 42, 43, 45 geneow with varieties 9698 wild species, progenitor 3233 WISBEN 287, 288290, 291292 women 2, 10, 87, 93, 94, 427, 429, 430 World Bank 237 World Food Summit, Rome xv World Trade Organization 109 Xanthomonas campestris 400 xymoxanyl 387 yam 9, 33, 37, 87, 100, 138, 421 yellow rust 113, 255, 357, 360 yield, as criterion 107 yield barriers 26 yield plateau 10 yields 119, 121, 127 Zea spp. 32, 82, 96, 97, 161, 163 Zea diploperennis 96, 97, 163 Zea mays 159, 160 see also maize Zelder seed company 112113 Zimbabwe 170, 171