Plant Breeding Reviews, Volume 17

PLANT BREEDING REVIEWS Volume 17 Plant Breeding Reviews is sponsored by: American Society for Horticultural Science C...

Author: Jules Janick

93 downloads 1075 Views 23MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

PLANT BREEDING REVIEWS Volume 17

Plant Breeding Reviews is sponsored by: American Society for Horticultural Science Crop Science Society of America Society of American Foresters National Council of Commercial Plant Breeders

Editorial Board, VoluIne 17 G. R. Askew F. A. Bliss M. Gilbert

PLANT BREEDING REVIEWS Volume 17

edited by

Jules Janick Purdue University

John Wiley 8' Sons, Inc. NEW YORK / CHICHESTER / WEINHEIM / BRISBANE / SINGAPORE / TORONTO

A NOTE TO THE READER This book has been electronically reproduced. from digital information stored at John Wiley & Sons, Inc. We are pleased that the use of this new technology will enable us to keep works of enduring scholarly value in print as long as there is a reasonable demand for them. The content of this book is identical to previous printings.

This book is printed on acid-free paper.

=

Copyright © 2000 by John Wiley & Sons, Inc. All rights reserved. Published simultaneously in Canada. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, scanning or otherwise, except as permitted under Sections 107 or 108 of the 1976 United States Copyright Act, without eitheJ: the prior written permission of the Publisher, or authorization through payment of the appropriate per-copy fee to the Copyright Clearance Center, 222 Rosewood Drive, Danvers, MA 01923, (978) 750-8400, fax (978) 750-4744. Requests to the Publisher for permission should be addressed to the Permissions Department, John Wiley & Sons, Inc., 605 Third Avenue, New York, NY 10158-0012, (212) 850-6011, fax (212) 850-6008, E-Mail: PERMREQ @WILEY.COM. This publication is designed to provide accurate and authoritative information in regard to the subject matter covered. It is sold with the understanding that the publisher is not engaged in rendering professional services. If professional advice or other expert assistance is required, the services of a competent professional person should be sought. Library of Congress Catalog Card Number: 83-641963 ISBN 0-471-33373-5 ISSN 0730-2207 Printed in the United States of America 1098765432

Contents List of Contributors

vii

1. Dedication: Longping Yuan: Rice Breeder and World

Hunger Fighter

1

liming Li and Yeyun Xin 2. Hybrid Rice: Genetics, Breeding, and Seed Production liming Li and Longping Yuan

I. II. III. IV. V. VI. VII. VIII.

Introduction Heterosis in Rice Male Sterility in Rice Breeding for Three-line System Hybrid Rice Breeding for Two-line System Hybrid Rice Wide Compatibility and Utilization of Intersubspecific Heterosis Hybrid Rice Seed Production Future Prospects Literature Cited

3. Rose: Genetics and Breeding Serge Gudin

I. II.

III. IV. V. VI. VII.

Introduction Systematics Cytogenetics and Genetics of Rosa Breeding Objectives Breeding Criteria and Selection Procedures Breeding Technology Conclusion Literature Cited

15 17 24 33 46 72 88 98 111 120

159 160 162 169 172 173 175 180 181

v

CONTENTS

vi

4. White Clover Molecular Genetics Derek W. R. White, Derek R. Woodfield, Brigitta Dudas, Richard 1. S. Forster, and David 1. Beck I. II. III. IV.

Introduction Transgenic Approaches to White Clover Improvement Marker-Assisted Selection Conclusions Literature Cited

5. The Rise and Fall of Overdominance James F. Crow I. II. III. IV. V. VI. VII.

Introduction Ear1y History Alternative Explanations of Heterosis Why Are Hybrids So Good? Physiological and Molecular Studies Is the Hybrid Breeding Method Best? Conclusions Literature Cited

6. An Informal History of Statistics

191

192 193 212 215 216

225 226 227 230 245 247 249 251 252

259

N. W. Simmonds I.

II. III. IV. V. VI. VII.

Introduction Statistics-l and Political Arithematic Probability Error Theory and the Central Limit Twentieth-Century Developments Conclusions Biographical Sketches Literature Cited

Subject Index Cumulative Subject Index Cumulative Contributor Index

260 262 269 279 288 300 302 314

317 319 335

Contributors David L. Beck, The Horticulture and Food Research Institute of New Zealand, Private Bag 92169, Auckland, New Zealand James F. Crow, University of Wisconsin, Genetics Department, Madison, WI 53706 Brigitta Dudas, New Zealand Pastoral Agriculture Research Institute, Private Bag 11008, Palmerston North, New Zealand Richard L. S. Forster, The Horticulture and Food Research Institute of New Zealand, Private Bag 92169, Auckland, New Zealand Serge Gudin, Universite d'Aix-Marseille III, Service 442, Avenue Escadrille Normandie-Niemen, France Jiming Li, Cornell University, Department of Plant Breeding, Ithaca, NY 148531902 N. W. Simmonds, 9 McLaren Road, Edinburgh, EH9 2BN, Scotland Derek W. R. White, New Zealand Pastoral Agriculture Research Institute, Private Bag 11008, Palmerston North, New Zealand Derek R. Woodfield, New Zealand Pastoral Agriculture Research Institute, Private Bag 11008, Palmerston North, New Zealand Yeyun Xin, China National Hybrid Research & Development Center, Changsha, Hunan Province, PR 410125, China Longping Yuan, China National Hybrid Research & Development Center, Changsha, Hunan Province, PR 410125, China

vii

Longping Yuan

1

Dedication: Longping Yuan Rice Breeder and World Hunger Fighter liming Li Department of Plant Breeding, Cornell University, Ithaca, New York 14853-1902. Yeyun Xin China National Hybrid Rice Research & Development Center, Changsha, Hunan Province, P.R. China 410125

It is an honor and a privilege to summarize the great contribution of Pro-

fessor Longping Yuan, renowned rice breeder and world hunger fighter. As graduate students and assistants, we had the good fortune to work with Professor Longping Yuan, and our admiration for his scientific career and personal character continues to grow. Because China for a long time had a closed-door policy, which was especially severe during the Cultural Revolution of the 19608 and 1970s, many people still lack an accurate understanding of China and her scientific achievements in agriculture. We hope that this dedicatory chapter honoring Professor Longping Yuan will not only contribute to an appreciation of his special contribution to developing hybrid rice, but will also help explain China's role in advancing agricultural technology. CHILDHOOD AND EDUCATION

Longping Yuan was born in Peking on September 7, 1930. His father was a staff member in a Railroad Bureau and his mother was a foreign language teacher. As a result of the Sino-Japanese war and the subsequent Plant Breeding Reviews, Volume 17, Edited by Jules Janick © 2000 John Wiley & Sons, Inc.

ISBN 0-471-33373-5

1

2

J.

LI AND Y. XIN

world conflagration, his schooling was constantly interrupted. He intermittently attended three primary schools in three provinces from 1936 to 1942, and three middle schools at three cities from 1942 to 1949. From 1949 to 1953, he studied at the Southwestern Agricultural College in Chongqing, majoring in agronomy (Xie 1996). As a young man, he became a strong swimmer, which contributed to his lithe figure and current good health. In 1964, he married Deng Ze at An-Jiang, Hunan Province, China, and is now the father of three sons: Ding-An, DingJiang, and Ding-Yang. INITIATION OF RESEARCH ON HYBRID RICE

After his graduation from college in 1953, Longping accepted a teaching job at An-Jiang Agricultural School in the western mountainous area of Hunan Province. He taught plant genetics and breeding and other courses there for 18 years until he was transferred to direct the Cooperative Group of Hybrid Rice Research at Hunan Academy of Agricultural Sciences. During the many years he was a teacher, he attempted experiments to graft hybrids between plant species following the now discredited theories of Trofim Lysenko, whose influence prevailed in the genetic research carried out in the former Soviet Union and China. The failure of these experiments and his own knowledge of English and Russian led him to search the literature on genetics published in the 1960s in the West, which led him to Mendelian genetics. At the risk of great political danger, he boldly taught his students the chromosomal and gene theories and their application to plant breeding. He also started to explore the mechanism of heterosis utilization in rice after the discovery of an extraordinary natural rice hybrid that he noticed in 1960. China's severe famine and the impoverished life of rural villagers during the early 1960s greatly affected Yuan and made him determined to develop a high-yielding rice. Male sterility in rice aroused his unflagging interest because it is so critical for the utilization of heterosis in a crop like rice, where large-scale hand emasculation is extremely difficult. He continued to pursue this interest even though he was under great pressure from the successive political turmoil of the Maoist movements. Finally, in 1964 and 1965, he found six naturally male sterile plants in early-season indica rice cultivars from farmers' rice fields. In 1966, he published the first research paper on rice male sterility in Kexue Tongbao (Scientific Correspondence), in Chinese with an English abstract, which initiated China's hybrid rice researches. In this landmark paper, he grouped the male sterile plants into three types (pollen-free

1. DEDICATION

3

type, pollen-abortive type, and partially male sterile type), and rightfully predicted that rice heterosis could be utilized to increase yields through the development of male sterile (or A) lines, maintainer (or B) lines, and male restorer (or R) lines. As early as 1926, Jones first reported the existence of male sterility in rice in the United States. Between the 1930s and 1960s, rice scientists in the United States, Japan, India, and the International Rice Research Institute in the Philippines reported some findings on male sterility. But the great difficulty for heterosis utilization by means of male sterility still depended upon the practicability of commercial hybrid seed production. Furthermore, even in the West, the prevailing dogma asserted that there should be no significant heterosis in rice because it was self-pollinated. The turmoil of the Great Proletarian Cultural Revolution brought many social troubles into Yuan's research and daily life, but these challenges did not make him and his assistants retreat. On the contrary, they strengthened their research efforts with the support of what was then the Chinese State Science and Technology Committee. During six years of indefatigable breeding effort for generation advance in Hunan Province, Yunnan Province, and Hainan Island beginning in 1964, Longping Yuan realized that these male steriles might derive from the crossing of parents with very little genetic diversity. Thus, he decided to search for male sterile plants in wild rice. In 1970 Longping Yuan's assistants coincidentally discovered a male sterile plant from a wild rice species (Oryza rufipogon Griff. or O. sativa f. Spontanea) during the course of looking for wild rice sources on Hainan Island. After confirming male sterility, Yuan initiated the development of male sterile lines from this material and confirmed that both cytoplasmic and nuclear genes controlled its sterility. This natural wild abortive plant, dubbed "WA," probably deriving from outcrossing between the local wild rice and a local cultivar, provided a crucial genetic tool in the birth of China's commercial hybrid rice and confirmed Pasteur's famous aphorism, "Chance favors the prepared mind only." From 1971 on, Yuan shared this male sterile material with many research institutes. In 1972, Yuan developed China's first commercial A and corresponding B line, 'Er-Jiu-Nan No.1 A(B)' derived from WA. His assistance also brought about the successful development of several commercially usable A and B lines, including 'Zhen-Shan 97 A(B)', 'V20A(B)' and 'V41A(B)'. The former two A(B) lines still occupy more than 70% of the area under hybrid rice production in China. In 1972, a national cooperative research group was established; among its members were several hundred researchers from various disciplines, including cytology, genetics, physiology, and ecology, and from research institutes

4

J. LI AND Y. XIN

and universities in 19 rice-growing provinces in China. Under the aegis of this cooperative research group led by Longping Yuan, several commercially feasible restorer lines were bred or screened from hundreds of thousands of rice cultivars in 1973. The success of the development of hybrid rice was declared at a rice conference in October, 1973. Shortly after the development of three lines (A, B, and R lines), some Chinese rice experts still denied the existence of rice heterosis, especially grain yield heterosis, and objected to the hybrid rice program, following the classical genetic assumption that "self-pollinated crops do not deteriorate by selfing and thus do not have heterosis by hybridization." Clearly, the results with rice indicate that heterosis can playa role in self-pollinated crops and this has been confirmed for other selfpollinated crops such as tomato. Yuan developed the first high-yielding commercial rice hybrid 'NanYou No.2' in 1974. He and his assistants also demonstrated to farmers how rice hybrids showed much higher yields than the best inbred check cultivars, refuting the arguments against rice hybrids with hard facts, and convinced the Chinese policymakers of the great yield potential of hybrid rice. In time, many large and effective seed companies were established, forming a beneficial agricultural business for the first time in the history of China. The most difficult obstacle for the commercial exploitation of hybrid rice was seed production because of the relatively short flowering period, small stigma, and other traits influencing self-pollination in rice. The low yield of hybrid rice seed also hindered the commercialization of hybrid rice in Japan and other countries. At first, the yield of hybrid seed only reached 83 kg/ha in the experimental seed production plots. Yuan summarized the solutions to low yield hybrid seed production after patient and minute observation of rice floral behavior; devised a series of yield-boosting techniques such as the determination of seeding dates, adjustment of flowering stages, supplementary pollination, and application of gibberellic acid; and published his findings in Yichuan Yuzhong (Genetics and Breeding) in 1977. As a result of the improvement of seed production techniques, an army of about 30,000 people from China's rice-growing provinces converged on Hainan Island to produce hybrid rice seeds, demonstrating the resolve of the Chinese government to make full use of this new technology to eliminate poverty and malnutrition, which had plagued China for centuries. Yuan was appointed as the general adviser and gave lectures and on-the-spot instructions during this campaign. This unprecedented massive agricultural campaign was highly successful and resulted in a sizable

1. DEDICAnON

5

increase in planting area containing hybrid rice: 0.14 million ha in 1976, 2.1 million ha in 1977, and then to 10 million ha by 1986. In 1980, a delegation sent by the Ford Foundation found to their astonishment that China had learned to produce hybrid rice, a feat which had eluded IRRI, the Central Rice Research Institute in India, and many researchers who had been involved in this effort. In the 1980s, hybrid rice technology was patented in the United States as China's first agricultural invention, and the patent was transferred to Ring Around, a U.S. company. CONTRIBUTIONS OF HYBRID RICE

Based on statistical data from the early 1980s to 1996, hybrid rice outyields the conventional rice by 20 to 30% in China. Currently, hybrid rice occupies 15 million ha in China, which accounts for 500/0 ofthe total rice area and 59% of total rice production. From the initial commercialization of hybrid rice technology in 1976 to 1997, hybrid rice increased China's total accumulated rice yield by 0.3 billion tonnes, sufficient to supply the amount of food grains consumed by the 1.2 billion population for one year. According to FAD, hybrid rice covered an area of 14.6 million ha in China in 1990, accounting for 10% of the world total area, but accounted for 20% of global rice production. In China the average yearly increase of rice production by hybrid rice is 2,400 million kg, sufficient to feed 120 million people, a feat that has prevented great social turmoil. A popular saying in China is " If you want money, ask Deng Xiaoping; if you want a belly full of rice, ask Yuan Longping." Hybrid rice has become the "Miracle Rice" of China (Yuan 1996). Professor Don Paarlberg, in his book Toward a Well-fed World (1988), commented: How did it happen'? Especially how did it happen during the time of insulation from the West, that supposed fountainhead of agricultural science? It is a story of the triumph of peaceful pursuits over violent uprisings and a lesson to those who think that a centrally directed country necessarily stifles the creative urge. It demonstrates the rising scientific competence of an Asian country. It illustrates the utility of agricultural science, the caprice of chance, and the accomplishment of a dedicated man ... Yuan Longping has bought China valuable time with which to bring down the rate of population growth. As agricultural science advances, the threat of famine retreats. Yuan led toward a well-fed world. He has also taught a valuable lesson for those remaining few who need it-that scientific achievement in agriculture has moved beyond the Western nations that first produced it.

6

J.

LI AND Y. XIN

BREEDING STRATEGIES

Professor Yuan is never content with his achievements. From statistical data on the yield potential of hybrid rice since the 1980s, he found that even though hybrid rice had a higher yield advantage over conventional inbreds, the yield of three-line system hybrid rice reached a yield plateau. Therefore, he developed a "three-phase strategy" and described it in a paper in the journal Hybrid Rice in early 1987. The main idea of this breeding strategy can be outlined as follows: (1) to enhance the heterosis level, proceed from intervarietal heterosis to intersubspecific heterosis to distant heterosis; (2) to simplify the methodology, proceed from the three-line or CMS system to the two-line or P(T)GMS (photosensitive or thermosensitive genic male sterility) line system to the oneline or apomixis system. According to his prediction, each of these phases should mark a new breakthrough in rice breeding and result in an even larger increase in yield if attained. The Chinese National Hi-Tech Plan adopted this strategy and established the National Two-line Hybrid Rice Research Program. Under the leadership of Yuan, the two-line system hybrid rice was declared to be successfully developed in China by utilizing P(T)GMS gene(s) and wide compatibility gene(s) in 1996 after nine years of cooperative research by hundreds of rice scientists from 23 research institutes and universities. The newly developed two-line system hybrids, especially intersubspecific two-line hybrids, can outyield the three-line system hybrids by over 15% and have been released for commercial production in China (Yang and Yuan 1995). The total area under two-line hybrid rice was 0.3 million ha in 1997. More recently, Yuan and his assistants developed some super-high-yielding two-line intersubspecific hybrids, which could yield more than 13 t/ha with a daily yield of 100 kg/ha under average cultivation conditions. He recently advanced a new concept concerning super-high-yielding breeding of hybrid rice by a combination of ideotype breeding and heterosis breeding. Most rice breeders in China are following this new idea. FIGHTING THE FUTURE WORLD HUNGER

As an agricultural scientist, Longping Yuan's concerns go beyond China's food supply and extend to the enormous problem of world hunger. Acting on his belief that science should know no boundaries between countries, he shares his knowledge, experience, ideas, and

1. DEDICATION

7

valuable breeding materials with scientists outside China to help develop hybrid rice. He has been invited to be the chief consultant by FAO on hybrid rice projects in India, Vietnam, Myanmar, and Bangladesh, and he and his colleagues also have trained more than 150 rice scientists from 15 countries, including India, Vietnam, Thailand, and Mexico, on hybrid rice technology at the China National Hybrid Rice Research and Development Center (CNHRRDC). With technical assistance from China and IRRI and financial support from UNDP and FAO, hybrid rice is developing rapidly in India and Vietnam. In 1996, the area under hybrid rice was 0.1 million ha in Vietnam and 60,000 ha in India and on average the hybrid rice cultivars yielded 1-2 t/ha more than the local pure line cultivars. Projects for developing hybrid rice have been initiated in Myanmar and Bangladesh, and a cooperative hybrid rice program has been established between RiceTec, Inc. of Texas and the China National Hybrid Rice Research and Development Center (CNHRRDC). HONORS, AWARDS, AND PROFESSIONAL AFFILIATIONS

Longping Yuan has been awarded many national and international prizes and has been acclaimed both nationally and internationally as "the father of hybrid rice." The main national awards include the Ho Leung Ho Lee Foundation Prize, Beijing, China (1994); Meritorious Scientist, Hunan Provincial Government, Changsha, Hunan, China (1992); and the first extraordinary-class National Invention Prize, Beijing, China (1981). The international awards include the Fukui International Koshihikari Rice Prize, Fukui, Japan (1998), the Distinguished Pioneer Scientist in Heterosis, "International Symposium on the Genetics and Exploitation of Heterosis in Crops," Mexico (1997); NIKKEI Asian Prize Award, Tokyo, Japan (1996); World Food Security and Sustainability Medal awarded by FAO, Quebec, Canada (1995); Alan Shawn Feinstein World Hunger Award for Research and Education, Brown University, U.S.A. (1993); the Rank Prize for Agronomy and Nutrition, London, England (1988); UNESCO Science Prize in recognition of the outstanding contribution in the field of science and technology for development, Paris, France (1987); and the WIPO Gold Medal for the outstanding inventor (1985). Professor Longping Yuan is now Director General of China National Hybrid Rice Research & Development Center (CNHRRDC), Honorary President of Hunan Academy of Agricultural Sciences, China, a member

8

J.

LI AND Y. XIN

of Chinese Academy of Engineering (CAE), and guest professor of Hunan Agricultural University, Wuhan University, and Huazhong Teacher's University. He serves as Vice Chairman of China's Society of Crop Science, Chairman of Hunan Agronomy Society, Deputy Director of Hunan Provincial Natural Science Foundation, Vice Chairman of National Expert & Consultant Group for Field Crop Production of China Ministry of Agriculture and Vice Chairman of Science & Technology Association of Hunan Province. A WISE AND GOOD MAN

Professor Longping Yuan, the consummate plant breeder, has outstanding theoretical achievements to his credit. He has released 16 rice lines or hybrids, and has published more than 80 papers and six books. The book Hybrid Rice Breeding and Cultivation (Hunan Science and Technology Press, 1988) has been recognized as an authoritative reference on hybrid rice technology within China, and two others, A Concise Course on Hybrid Rice (Hunan Science and Technology Press, 1986) and Technology of Hybrid Rice Production (FAG, 1996), are considered to be primers for hybrid rice technology both within and outside China. His lucid, simple, and meticulous writing style has facilitated the wide dissemination of his ideas and theories. Professor Yuan is also a good teacher and a kind friend. In spite of the demands on his time that his leadership role imposes (he has directed the national hybrid rice program for about 30 years), he has been able to continue his breeding research and to train assistants and graduate students to develop a strong sense of responsibility, dedication to research, and ability to accomplish a task independently. He has been very effective in obtaining governmental support for research and for organizing cooperative research programs. Under his leadership, the former Hunan Hybrid Rice Research Center (HHRRC) has become the China National Hybrid Rice Research and Development Center (CNHRRDC) with the support of the Chinese government. His progress has been a result of an uncanny capability to think, plan, and analyze problems and initiate work by providing keen insight into forthcoming problems and opportunities for new approaches. His name has became a household word in China and he has become a magnet to Chinese farmers, attracting their support, appreciation, and gratitude. He continues to direct the national hybrid rice breeding program and carries out his research with interest and enthusiasm. His present goals are to replace most of the existing three-line

9

1. DEDICATION

hybrid rice cultivars with two-line hybrids and to deliver hybrid rice technology throughout the world to help eliminate hunger in the next century. Despite his enormous achievements, Professor Longping Yuan remains a modest man, always willing to listen and to learn, considerate, sympathetic, and easily approached. It is a honor to dedicate this volume of Plant Breeding Reviews to him, for he represents the aspirations of plant breeders everywhere.

A LIST OF LONGPING YUAN'S RELEASED RICE LINES AND HYBRIDS Parental lines or hybrids

Year

Collaborators

1972 1985 1986 1987 1991 1993 1994 1996 1996 1996

M. Y. Sun X. L. Deng Z.G.Song X. H. Luo et al. Z.G.Song X.L.Deng X. Q. Li et al. H. Q. Yin et al. Y. C. Yan

1974 1984 1984 1986 1993 1994

M. Y. Sun Z. G. Song X. L. Deng Z. G. Song X. L. Deng

PARENTAL LINES

Er-Jiu-Nan No.1 A(B) Ce64 Ce49 Ce48 Pei Ai 64s 438 647 Ce64s Xiang 125s Lin-Lun HYBRIDS

Nan-You No.2 V64 V48 V49 V438 V647

LITERATURE CITED Paarlberg, D. 1988. Toward a well-fed world. Iowa State Univ. Press, Ames. p. 115-120. Xie, C. J. 1996. "Meritorious scientist" Longping Yuan (in Chinese), China Agr. Publ. House. Yang, S. Q., and J. Yuan. 1995. Hybrid pioneer to scale greater heights: Prof. Yuan stress research. Window, October 13,1995. p. 26-27. Yuan, J. 1996. New breakthrough in "miracle rice": Two-line hybrid-next stage of development. Window, August 9,1996. p. 27.

10

J. LI AND Y. XIN

LONGPING YUAN PUBLICATIONS Yuan, 1. P. 1966. A preliminary report on male sterility in rice (in Chinese). Sci. Bul. 4:32-34. Yuan, 1. P. 1972. An introduction to the breeding of male sterile lines in rice (in Chinese), Proc. China's 2nd Workshop on Genetics. Yuan, 1. P. 1973. Breeding for cytoplasmic male sterile lines via wild abortive cytoplasm (in Chinese). Hunan Agr. Sci. 4:1-4. Yuan, 1. P. 1976. Hybrid rice (in Chinese). Agr. Publ. House, Beijing. Yuan, 1. P. 1977. Key techniques for high production of hybrid rice seeds. Yichuan Yuzhong 1:4-5. Yuan, 1. P. 1977. The execution and theory of developing hybrid rice. Scientia Agr. Sinica 1:27-31. Lin, S. c., and 1. P. Yuan. 1980. Hybrid rice breeding in China. In: Innovative approaches to rice breeding: Selected papers from the 1979 Int. Rice Research Conference. IRRI, P.O. Box 933, Manila 1099, Philippines. Virmani, S. S., 1. P. Yuan, and G. S. Khush. 1981. Current status of hybrid rice research. IRRI Rice Research Conf. P.O. Box 933, Manila 1099, Philippines. Yuan,L. P., and M. Y. Sun. 1984. A new rice hybrid: V64. Correspondence of Agr. Sci. Tech. (in Chinese), 5: 1-2. Yuan, 1. P. 1985. Breeding for super-high-yielding hybrid rice (in Chinese). Hybrid Rice 3:1-8. Yuan, L. P. 1985. A concise course in hybrid rice. Hunan Sci. Tech. Press, Changsha, China. Yuan, 1. P. 1985. Hybrid rice in China. IRRI, P.O. Box 933, Manila 1099, Philippines. Yuan, 1. P., S. S. Virmani, and G. S. Khush. 1985. Wei You 64: An early duration hybrid for China. lnt. Rice Res. Newsl. 10:11-12. Virmani, S. S., 1. P. Yuan, B. Suprihatno, P. J. Jachuck, and H. P. Moon. 1985. Collaborative project on hybrid rice. IRRI, P.O. Box 933, Manila 1099, Philippines. Yuan, L. P. 1986. Hybrid rice in China. Chinese J. Rice Sci. 1:8-18. Yuan, 1. P. 1986. Current status of hybrid rice research and development. Int. Symp. on Hybrid Rice, Changsha, Hunan, China. Yuan, 1. P. 1986. Potential for the improvement of crops by means of apomixis (in Chinese). Crop J. 3: 3-4. Yuan, L. P. 1986. Heterosis utilization in rice. Chapter 8. In: Chinese rice science. Agr. Publ. House, Beijing. Li, Y. C., and 1. P. Yuan. 1986. Genetic analysis of fertility restoration in male sterile lines ofrice. In: Rice genetics. IRRI, P.O. Box 933, Manila 1099, Philippines. Yuan, 1. P. 1987. Scope for commercial exploitation of hybrid vigor in rice. Proc. Int. Symp. "Rice farming system: New directions," Egypt. Yuan, 1. P. 1987. Strategy of hybrid rice breeding. Hybrid Rice 1:1-3. Yuan,L. P., and H. X. Chen. 1988. Breeding and cultivation of hybrid rice. Hunan Sci. Tech. Press, Changsha, China. Yuan, 1. P. and S. S. Virmani. 1988. Organization of a hybrid rice breeding program. In: Hybrid rice. IRRI, P.O. Box 933, Manila 1099, Philippines. Yuan, 1. P., and S. S. Virmani. 1988. Status of hybrid rice research and development. In: Hybrid rice. IRRI, P.O. Box 933, Manila 1099, Philippines. Yuan,1. P. 1989. Commercial exploitation of hybrid vigor in rice. In: Rice farming systems: New directions. IRRI, P.O. Box 933, Manila 1099, Philippines. Yuan, L. P. 1989. Hybrid rice. US Patent 4827664, May 9,1989. Yuan,1. P., S. S. Virmani, and C. X. Mao. 1989. Hybrid rice: achievements and outlook. In: Progress in irrigated rice research. IRRI, P.O. Box 933, Manila 1099, Philippines.

1. DEDICATION

11

Yuan, L. P. 1989. "Apomictic" rice 84-15 is in need of being further verified scientifically. Hybrid Rice 4:3, 2. Yuan, L. P. 1989. An investigation on agricultural research and rice breeding in Japan. Hybrid Rice 5: 4-6. Luo, X. H., and L. P. Yuan. 1989. Selection of wide compatibility lines in rice. Hybrid Rice 2: 35-38.

Yuan, L. P. 1990. Progress of two-line system hybrid rice breeding. Sci. Agr. Sin. 3:1-6. Yuan, L. P., Y. C. Li, and H. D. Deng. 1990. Progress of studies on rice twin seedlings. Apomixis Newsl. 2:42-44. Li, Y. c., and L. P. Yuan. 1990. Studies on genetics of twin seedlings in rice (OI}'za sativa L.). Acta Agron. Sin. 2:176-182. Yuan, L. P., Y. C. Li, and H. D. Deng. 1991. Progress of studies on rice twin seedlings. p. 136-138. In: X. X. Guo (ed.), Progress of studies on rice apomixis in China (1980-1991). Sichuan Publ. House Sci. Tech., Chengdu. Yuan, L. P., Y. C. Li, and H. D. Deng. 1991. Studies on rice twin seedlings: the third report. p. 139-141. In: X. X. Guo (ed.), Progress of studies on rice apomixis in China (1980-1991). Sichuan Publ. House Sci. Tech., Chengdu. Yuan, L. P. 1991. Outlook on the development of hybrid rice breeding. p. 205-211. In: Prospects ofrice farming for 2000. Zhejiang Publ. House Sci. Tech., Hangzhou, China. Yuan, L. P., and C. X. Mao. 1991. Hybrid rice in China-techniques and production. p. 128148. In: Y. P. S. Bajaj (ed.), Rice. (Biotechnol. Agr. Forestry. 14). Springer-Verlag, Berlin. Yuan, L. P. (ed.). 1992. Current status of two-line hybrid rice research. Agr. Press, Beijing, China. Yuan, L. P. 1992. The technological strategy for the development ofT(P)GMS lines. Hybrid Rice 1:1-4. Yuan, L. P. 1992. Advantage of and constraints to use of hybrid rice varieties. Proc. Int. Workshop on Apomixis in Rice (IWAR), Changsha, Hunan, China. Yuan, L. P. 1992. Recent breakthroughs in hybrid rice research and development in China. Int. Rice Comm. Newsl. 41:7-13. Zhang, Z. G., X. G. Lu, and L. P. Yuan. 1992. Considerations on the evaluation and selection of critical temperature for fertility transformation in photoperiod sensitive male sterile rice. Hybrid Rice 6:29-32. Yuan, L. P. 1993. China's experiences in the development of hybrid rice research programme. p. 22-28. In: B. R. Barwale (ed.J, Hybrid rice: Food security in India. Macmillan, Madras, India. Yuan, L. P. 1993. Progress of two-line system in hybrid rice breeding. p. 86-90. In: K. Muralidharan and E. A. Siddiq (ed.), New frontiers in rice research. Directorate of Rice Research, Hyderabad, India. Yuan, L. P. 1993. Development and prospects of hybrid rice breeding. p. 136-144. In: C. B. You (ed.), Biotechnology in agriculture. (Curr. Plant Sci. Biotechnol. Agr., vol. 15). Kluwer Academic Publishers, Dordrecht. (Proc. First Asia-Pacific Conference on Agr. Biotechnol., 1992 Beijing, China). Li, X. Q., and L. P. Yuan. 1993. A study on the development of T(P)GMS lines with low critical sterility temperature in rice. Hybrid Rice 1:10-11. Yuan, L. P. 1994. Raising the yield ceiling and grain quality of hybrid rice to meet the consumers' need and improve its economic efficiency. p. 147-152. In: J. S. Chen (ed.), Establishment of agriculture with high yield potential, good quality, and high profit. China Agr. Press, Beijing. Yuan, L. P. 1994. Purification and foundation seed production of T(P)GMS lines in rice. p. 1-1. National Seminar on Two-line System Hybrid Rice, Yangzhou, China (Abstr.).

12

J. LI AND Y. XIN

Yuan, L. P. 1994. Purification and production of foundation seed of rice PGMS and TGMS lines. Hunan Agr. Res. News!. 2:2-3. Yuan, L. P. 1994. Development and perspective of hybrid rice research in China. Hunan Agr. Res. News!. 1: 7-8. Yuan, L. P., and S. S. Virmani. 1994. Increasing yield potential in rice by exploitation of heterosis. p. 1-6. In: S. S. Virmani (ed.), Hybrid rice technology: New developments and future prospects. Selected papers from Int. Rice Research Conference. IRRI, P.O. Box 933, Manila 1099, Philippines. Yuan, L. P., Z. Y. Yang, and J. B. Yang. 1994. Hybrid rice research in China. p. 143-147. In: S. S. Virmani (ed.), Hybrid rice technology: New developments and future prospects. IRRI, P.O. Box 933, Manila 1099, Philippines. Yuan, L. P. 1995. Current status of hybrid rice in China and future strategies for 21st century. p. 31-33. In: M. I. Ahmed and B. C. Viraktamath (eds.), Hybrid rice seed production technology: Theory and practice. Directorate of Rice Research, Rajendranagar, Hyderabad, India. Yuan, L. P., and X. Q. Fu. 1995. Technology of hybrid rice production. Food and Agriculture Organization of the United Nations, Rome. Yin, H. Q., L. P. Yuan, and H. J. Yin. 1995. The development of utilization of the aromatic T(P)GMS line (Xiang 125s). Hunan Agr. Sci. 1:8-9. Xiao, J. H., J. M. Li, L. P. Yuan, and S. D. Tanksley. 1995. Dominance is the major genetic basis of heterosis in rice as revealed by QTL analysis using molecular markers. Genetics 140:745-754. Yuan, L. P. 1996. Prospects for yield potential in rice through plant breeding. Hunan Agr. Res. News!. 4:1-2. Yuan, L. P. 1996. The potential for increasing China's grain yield in view of plant breeding. Hybrid Rice 4:1-2. Yuan, L. P. 1996. Breeding strategies for development of intersubspecific hybrid rice. Hunan Agr. Res. News!. 2:1-3. Yuan, L. P. 1996. Hybrid rice in China. p. 51-54. In: M. I. Ahmed (ed.), Hybrid rice technology. Directorate of Rice Research, Hyderabad, India. Yuan, L. P. 1996. Prospects for hybrid rice breeding in China. Paper for 3rd Int. Symp. on Hybrid Rice. Hyderabad, India. Yuan, L. P., and J. M. Li. 1996. Outlook on the research of two-line system hybrid rice in the eighth five-year plan (1991-1995): I. Basic aspects of research on two-line system hybrid rice. Hunan Agr. Res. News!. 1:1-3. Yuan, L. P., and J. M. Li. 1996. A review on the research of two-line system hybrid rice in the eighth five-year plan period (1991-1995): II. Practice on two-line system hybrid rice breeding. Hunan Agr. Res. News!. 2:3-4. Xiao, J., J. Li, L. Yuan, and S. D. Tanksley. 1996. Identification of QTLs affecting traits of agronomic importance in a recombinant inbred population derived from a subspecific rice cross. Thl?or. App!. Genet. 92:230-244. Xiao, J. H., J. M. Li, L. P. Yuan, S. R. McCouch, and S. D. Tanksley. 1996. Genetic diversity and its relationship to hybrid performance and heterosis in rice as revealed by PCRbased markers. Theor. App!. Genet. 92:637-643. Xiao, J. H, S. Grandillo, S. N. Ahn, S. R. McCouch, S. D. Tanksley, J. M. Li, and L. P. Yuan. 1996. Genes from wild rice improve yield. Nature 384:223-224. Xiao, J. H., J. M. Li, L. P. Yuan, and S. D. Tanksley. 1996. Dominance as the major genetic basis of heterosis in rice. p. 327-340. In: Proc. 3rd Rice Genetics Symp. (16-20 October 1995), G. S. Khush (ed.), Rice genetics III. IRRI, P.O. Box 933, Manila 1099, Philippines.

1. DEDICATION

13

Liao, F. M., and L. P. Yuan. 1996. A discussion on the genetic purification of critical sterility temperature in rice T(P)GMS lines. Hybrid Rice 6:1-4. Yuan, L. P. 1997. Exploiting crop heterosis by two-line system hybrids: Current status and future prospects. p. 1-7. In: Proc. Int. Symp. on Two-line System Heterosis Breeding in Crops, Changsha, Hunan, China, September 6-8,1997. Yuan, L. P. 1997. Current status and developing prospects in two-line hybrid rice research in China. Res. Agr. Modern 1:1-3. Yuan, L. P. 1997. Strategy for the development of super-high-yielding hybrid rice. Hybrid Rice 6:1-6. Yuan, L. P., X. J. Wu, Y. C. Yan, and X. H. Luo. 1997. A strategy for developing wide spectrum compatibility rice lines. Sci. Agr. Sin. 4:1-8. Yuan, L. P. 1997. Exploiting crop heterosis by two-line system hybrids: current status and future prospects. Hunan Agr. Res. Newsl. 4:1-7. Yuan, L. P. 1997. Hybrid rice technology in China. Paper for China's Second Annual Meeting on Int. Agr., Beijing, China. Xiao, G. Y., and L. P. Yuan. 1997. Effects of water temperature on male sterility of the thermo-sensitive genic male sterile (TGMS) rice lines under the simulated low air temperature conditions appeared occasionally in high summer. Chinese J. Rice Sci. 11:241-244. Yan, Y. C., and L. P. Yuan. 1997. The development of a wide-spectrum WCV, Lin-Lun, in rice. Hybrid Rice 1:7-10. Deng, Q. Y., X. Q. Fu, and L. P. Yuan. 1997. On fertility stability of the P(T)GMS lines and their identification technology. p. 76-85. In: Proc. Int. Symp. on Two-line System Heterosis Breeding in Crops, Changsha, Hunan, China, September 6-8,1997. Wu, X. J., and L. P. Yuan. 1998. A study on population improvement on the percentage of exserted stigma in thermo-sensitive genic male sterile rice. 1. The effect of different methods on the improvement in percentage of exserted stigma. Acta Agron. Sin. 1:68-73. Li, R. H., C. G. Xue, L. P. Yuan, Y. Q. He, C. Q. Sun, S. B. Yu, X. H. Li, X. K. Wang. 1998. Differentiation and classification of parental Jines and favorable genic interactions affecting F1 fertility in distant crosses of rice ((JIyza sativa L.). Theor. App!. Genet. 96:526-538. Yuan, L. P. 1998. Hybrid rice breeding for super high yield. XVIIlth Int. Congr. of Genetics, Beijing, China, August 10-15,1998.

2 Hybrid Rice: Genetics, Breeding, and Seed Production liming Li and Longping Yuan* China National Hybrid Rice Research & Development Center, Changsha,China,410125 1. Introduction

A. China's Achievements B. Hybrid Rice Technology Outside China C. "Bottlenecks" and Potential Solutions II. Heterosis in Rice A. Concept of Heterosis B. Performance of Heterosis C. Genetic Basis of Heterosis D. Prediction of Heterosis E. Approaches for Utilization of Heterosis III. Male Sterility in Rice A. Morphology, Cytology, and Histology of Male Sterile Lines B. Physiological and Biochemical Basis C. Genetic Basis IV. Breeding for Three-line System Hybrid Rice A. Breeding ProctJdure B. Development of A and B Lines C. Development of R Lines D. Development of Elite Hybrid Combinations E. Breeding for Rice Hybrids with Resistance to Insect Pests and Diseases F. Breeding for Rice Hybrids with High Grain Quality *Gratitude is expressed to Donald Wallace, Jules Janick, Tom Thai, Shannon Painter, Yunbi Xu, and Xiaojin Wu for their critical reading, editing, and encouragement; to Cao Xiaobin for his help in searching the Chinese literature; to Susan R. McCouch for her substantiallab support; and to Jinhua Xiao, Yeyun Xin, Xiqin Fu, and Xinqi Li for their generous help. We also extend our appreciation to the Rice Biotechnology program of the Rockefeller Foundation and the hybrid rice breeding programs from the previous Chinese National Science & Technology Committee and China's Ministry of Agriculture for their financial support. Plant Breeding Reviews, Volume 17, Edited by Jules Janick ISBN 0-471-33373-5 © 2000 John Wiley & Sons, Inc. 15

J.

16

LI AND L. YUAN

V. Breeding for Two-line System Hybrid Rice A. Considerations B. Development of T(P)GMS Lines C. China's Progress D. Breeding for Two-line System Rice Hybrids Using Chemical Emasculators VI. Wide Compatibility and Utilization of Intersubspecific Heterosis A. Classification in Rice B. Phenomenon of Wide Compatibility C. Genetics of Wide Compatibility Traits D. Development of WCVs E. Utilization of Intersubspecific Heterosis VII. Hybrid Rice Seed Production A. China's Success B. Key Techniques C: Specifics for CMS Line Multiplication D. Purification of Parental Lines VIII. Future Prospects A. Breeding of Diverse Parental Lines B. Molecular Breeding C. Apomixis Breeding D. Hybrid Seed Production E. Socioeconomic Impact Literature Cited

LIST OF ABBREVIATIONS ABA

ACC

ADH ADV AFLP A line AVG B line BT

CGR CHA

eMS

CSP DES DW EAT EI EMS FASS

abscisic acid l-amino-eyclopane-l-earboxylie acid alcohol dehydrogenase alkali digestion value amplified fragment length polymorphism a cytoplasmic male sterile line aminoethoxy vinylglycine a maintainer line in the three-line hybrid system Boro-Taichung 65 (type male sterile cytoplasm or line) crop growth rate chemical hybridizing agent cytoplasmic male sterility critical sterility point diethyl sulfate Dong-pu wild rice effective accumulated temperature ethyleneimine ethyl methane sulfonate fertility alteration sensitive stage

2. HYBRID RICE: GENETICS, BREEDING. AND SEED PRODUCTION

G GA GCA HPGMR

HL IAA I-KI

IP IRRI

IRTP LW MH NARS

NEU NMS PCR PGMS

QTL RFLP R line SCA s line STS TGMS

T(P)GMS WA WC

WCG WCV

17

Gambiaka-type male sterile cytoplasm or line gibberellic acid general combining ability Hubei Photoperiod Sensitive Genic Male Sterile Rice Hong-Lian type male sterile cytoplasm or line indoleacetic acid iodine-potassium iodine Indonesia Paddy rice International Rice Research Institute international rice testing program Long-An wild rice maleic hydrazide national agricultural research service N-ethy1-N-nitrosourea nuclear male sterility polymerase chain reaction photoperiod sensitive genic male sterile line in the twoline hybrid system quantitative trait loci restriction fragment length polymorphism restorer line specific combining ability photoperiod or temperature sensitive male sterile line sequence tagged sites temperature sensitive or thermo-sensitive genic male sterile line in the two-line hybrid system temperature sensitive or photoperiod-sensitive genic male sterile line in the two-line hybrid system wild~abortive, a male sterile cytoplasm or line wide compatibility, which can produce F 1 hybrids with normal male fertility both to most of indica and to most of japonica rice cultivars wide compatibility gene cultivar which has wide compatibility

I. INTRODUCTION

The commercial production of hybrid rice in China represents one of the most successful breeding efforts of the twentieth century. Heterosis breeding in rice has been reviewed by Chang et al. (1973), Davis and Rutger (1976), Virmani and Edwards (1983), Kim and Rutger (1988), and

18

J. LI AND 1. YUAN

Virmani (1994a, 1996). This review emphasizes hybrid rice breeding and seed production in China. It includes the three-line, two-line, and oneline breeding approaches (see Sections IV, V, VIII). Documentation on heterosis in rice (Oryza sativa L.) has a long history. Jones (1926) first indicated its existence and it was subsequently reported by Ramiah (1935), Idasumi (1936), Kadam et al. (1937), Capinpin and Singh (1938), Ramiah and Rangaswamy (1941), Brown (1953), Oka (1957), Sen and Mitra (1958), Pillai (1961), Namboodri (1963), Rao (1965), Purohit (1972), Saini and Kumar (1973), Sivasubranian and Menon (1973), Saini et al. (1974), Singh et al. (1977), Singh and Singh (1977, 1979), Singh et al. (1980, 1984), Yoshida and Fujimaki (1985), respectively. Producing commercial F 1 hybrid seed by hand emasculation is impractical in rice. Thus, development of male sterile lines is essential in order to exploit rice heterosis. Some male sterile lines from the japonica subspecies were developed in the 1960s, including 'Fujisaka 5 A' (Katsuo and Mizushima 1958; Watanabe et al. 1968) and 'Taichung 65 A' (Shinjyo and Omura 1966). Erickson, the first U.S. researcher of rice cytoplasmic male sterility, determined that both 'Bir-Co' and O. glaberrima contained the cytoplasm that facilitated male sterility, based on crosses with the California japonica rice cuItivars 'Calrose', 'Caloro', and 'Colusa' (Erickson 1969; Carnahan et al. 1972). The male sterile cytoplasm in 'Taichung Native l' also resulted in 'Pankhari 203A' (Athwal and Virmani 1972). However, these male sterile lines have never been put into large-scale commercial production. A. China's Achievements

China was the first country to produce hybrid rice for commercial use. Research on male sterile rice was initiated in 1964 (Yuan 1966). However, rice heterosis was not successfully exploited until after the discovery of the wild abortive (WA) male sterile cytoplasm in the wild species (0. rufipogon Griff or O. sativa f. Spontanea) at Hainan Island in 1970 (Li 1977). The first set of genetic tools (a male sterile or A line, a maintainer or B line, and a restoring or R line) for the three-line system of hybrid rice production was developed in 1973 (Yuan and Virmani 1988). With the establishment of the three-line technology for hybrid rice seed production, the first hybrid rice combinations were put into commercial production in China in 1976. Since then, the area under hybrid rice production has increased from 2.1 million ha in 1977 to 10.9 million ha in 1987 and to 15.3 million ha in 1997. Hybrid rice normally has a yield advantage of 20-30% over non-hybrid rice cultivars (Lin and Yuan 1980; Shen 1980). From 1976 to 1997 hybrid rice enabled China

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

19

to increase rice production by more than 312 million t to feed its everincreasing population. Recently, hybrid rice has yielded about 6.6 t/ha compared with 5 t/ha for conventional cultivars. In 1994 hybrids were grown on 15.7 million ha, 50% of the total rice area and 57% of China's total rice production. Record yields of 11.2 t/ha from a single hybrid crop on a: large scale (1,000 hal and 17.1 t/ha in a small plot (0.1 hal have been reported (Bai and Luo 1996). The double cropping record for hybrid rice is 23.3 t/ha. Furthermore, hybrid rice requires about 4% less labor, and 2% less draft animal services while yielding 19% more than conventional modern cultivars (Lin 1994). High hybrid seed yield has been important for hybrid rice production. Recent average seed yield in China has been 2.4 t/ha. To further reduce costs, many new cytoplasmic male sterile (CMS) lines with high outcrossing efficiency have been developed, thus raising hybrid rice seed production. The current land area ratio among A line multiplication, F 1 seed production, and F 1 commercial cultivation is 1:50:5000. The highest recorded F 1 hybrid rice seed production yield was 7.4 t/ha on a small plot (0.2 hal by Zixing Seed Company in Hunan Province in 1993 (Yuan 1996; Mao et al. 1998). B. Hybrid Rice Technology Outside China China successfully commercialized hybrid rice technology in the 1970s and obtained the first patent on this technology in the United States in 1989 (Yuan 1989). As a result of China's success in hybrid rice production, the International Rice Research Institute ORRI) revived its hybrid rice work in 1979 (Lin and Yuan 1980; Int. Rice Res. Inst. 1980; Yuan and Virmani 1988). Many other countries initiated research on hybrid rice during the period from the 1970s to 1990s, including Japan (Murayama 1973; Murayama et al. 1974; Kato et al. 1994), the United States (Rutger and Shinjyo 1980; Mackill and Rutger 1994), India (Mohanty and Mohapatra 1973; Maurya and Singh 1978; Mallick et al. 1978; Panawar et al. 1983; Devarathinam 1984; Parmasivian 1986; Anandakumar and Sreerangasamy 1986; Prakash and Mahadevappa 1987; Virmani 1993; Siddiq 1994; Barwale 1994; Siddiq et al. 1994; Rangaswamy et al. 1994), Thailand (Chitrakon et al. 1986; Chitrakon 1987), Korea (Kim and Heu 1979; Koh 1987; Moon 1988; Choi 1991; Moon et al. 1994), Vietnam (Nguyen et al. 1985, 1994; Pham et al. 1991; Yin 1993; Nguyen 1994; Li 1995), Indonesia (Suprihatno 1986; Subandi et al. 1987; Suherman 1989; Suprihatno et al. 1994), the Philippines (Lara et al. 1994), Myanmar (K. L. Zhou, pers. commun.), Brazil (Neves et al. 1994), Egypt (Maximos and Aidy 1994), Colombia (Munoz 1992,1994),

20

J.

LI AND L. YUAN

Malaysia (Mohamad et al. 1987; Osman et al. 1988; Guok 1994), Iran (Dorosti 1997; Sattari 1997), Pakistan (Cheema and Awan 1985; Cheema et al. 1988; Ali and Khan 1998), Mexico (Armenta-Soto 1988), Bangladesh (Julfiquar 1998), Sri Lanka (Rothschild 1998), as well as international research institutes (Virmani et al. 1991; Virmani 1994b; Taillebois 1991, 1994) and private companies such as RiceTec, Inc. in the United States, and Mahyco Seed Company, Pioneer Overseas Corporation, and Hybrid Rice International in India. Hybrid rice technology has also attracted the attention ofthe FAO, which started its hybrid rice program following the recommendations of the 16th Session of the International Rice Commission (IRC) held at the International Rice Research Institute in 1985 (Trinh 1992, 1993, 1994; McWilliam et al. 1995). Technical support for hybrid rice technology has been provided to countries such as India, Vietnam, and Bangladesh from the International Rice Research Institute and China. The China National Hybrid Rice Research & Development Center (the former Hunan Hybrid Rice Research Center) has held six international courses on hybrid rice production technology and trained more than 150 rice scientists from various countries including India, Vietnam, Thailand, and Colombia (CNHRRDC 1997). India's hybrid rice project was started in the late 1980s, and its potential for the development and commercialization of hybrid rice is encouraging. Since 1991 India's research network has involved 12 research centers. Over 400 hybrids were developed and evaluated between 1990 and 1994. The best 35 hybrids exceeded the yield of the best check by over 1 t/ha. Several hybrid cultivars released to farmers, including 'APRH1' (IR58025A x Vajram), 'APRH2' (IR58025A x MTU9992), 'DRR-l' (IR58025A x IR40750), 'DRR-2', 'DRR-3', 'MGR-1' (IR62829A x IR10198) and 'KRH1' (IR58025A x IR9761), have performed admirably (Table 2.1). Other noteworthy hybrids include 'CoRH1', which was developed in Tamil Nadu (Rangaswamy et al. 1994); 'CNHR 3' (IR62829A x Ajaya), which was released for dry season cultivation in West Bengal, India; the salt-tolerant hybrid 'TNRH16' (IR58025A x C20R), which recorded a grain yield of 5t/ha, 20% over the check; and 'C043' (Ali et al. 1998). Pioneer Overseas began breeding hybrid rice in Hyderabad, India in 1988 and released 'PHB31' in 1993. Other private sectors such as E.I.D. Parry Ltd. are also involved in the development and commercialization of hybrid rice technology. India's current hybrid seed yield is about 1.5-2.0 t/ha for its standardized hybrid seed production package. A total of 1,300 t of hybrid rice seed was produced for 60,000 ha of the cultivated area under hybrid rice in 1996 (Ahmed et al. 1997a,b; Ahmed 1997). India aims to have two million ha of hybrid rice by the beginning of the 21st century (Trinh 1993). The present challenge facing India is the

Table 2.1.

,..., N

Released rice hybrids by Indian public sectors. Source: Ahmed et al. 1997a.

Year

Hybrid

Parentage

Growth duration (days)

Yield in farm trial (t1ha) Hybrid

Check

Advantage over check (%)

Check cultivar

1994

APHR-1

IR58025A x VAJRAM

130-135

7.14

5.27

35.4

Chaitanya

1994

APHR-2

IR62829A x MTU9992

120-125

7.52

5.21

44.2

Chaitanya

1994

MGR-1

IR62829A x IR10198

110-115

6.08

5.23

16.2

IR50

1994

KRH-1

IR58025A x IR9761

120-125

6.02

4.58

31.4

Mangala

1995

CNRH-3

IR62829A x AJAYAR

125-130

7.49

5.45

37.4

Khitish

1996

DRRH-1

IR58025A x IR40750

125-130

7.30

5.50

32.7

Telia Hamsa

1996

KRH-2

IR58025A x KMR 3R

130-135

7.40

6.10

21.3

Jaya

J.

22

LI AND L. YUAN

successful transfer of technology for hybrid rice seed production in order to achieve practical results for farmers. In 1983 Vietnam started research on hybrid rice at Hau Giang in the Mekong River Delta (Nguyen et al. 1995). Several rice hybrids from the International Rice Research Institute showed 18-45% yield advantage over Vietnam's best local inbreds (Table 2.2) at Cuu Long Delta Rice Research Institute (CLRRI). The Chinese rice hybrids are highly adaptable to the northern mountainous area near China. Some rice hybrids such as 'Shan-You 63', 'Shan-You Gui 99', 'Shan-You Guang 12' and 'BoYou 64' were introduced directly from China to northern Vietnam yielding 6.5-8.5 t/ha, 13-14% higher than the local check 'CR203'. Some farmers obtained up to 10 t/ha in Dien Chou (Nhge An Province) and at Phu Xuyen (Ha Tay Province). Some Chinese hybrids yielded up to 14.0 t/ha at Dien Bien (Lai Chau Province), 12.0 t/ha in Hoa An (Cao Bang Province), and 12.6 t/ha in Van Quan (Lang Son Province). The area under hybrid rice production in the Red River Delta of Vietnam reached 40,000 ha in 1993 and 86,000 ha in 1996 (Hoan et al. 1998). But rice hybrids from China are not adapted to the tropical conditions in the Mekong River Delta where IRRI-bred rice hybrids and parental lines can grow well. By the turn of the century, Vietnam plans to cover about 0.5 million ha with hybrid rice (Pingali et al. 1997). Japan has studied hybrid rice since the 1950s and the Ministry of Agriculture, Forestry, and Fisheries initiated a hybrid rice program in 1983. The first three-line rice hybrid 'Hokuriku-ko 1', developed in 1985, outyielded the check inbred by about 20% (Yasuki et al. 1997). Zen-Noh (the National Federation of Agricultural Cooperative Association) and several private companies such as RAMM Hybrid International CoTable 2.2. Yield performance in Vietnam of some experimental rice hybrids from IRRI. Source: Nguyen et al. 1995. Year 1989/90

Season Dry

Yield (t/ha)

% of Check

54752A x IR64R IR54752A x IR64R IR54752A x OM80R

7.5 7.2 6.7

131* 125* 118*

OMBO

Hybrid

Check

aMBO

OM80

1990

Wet

25A x IR29723R IR62829A x IR29723R

7.6 6.7

143* 126*

MTL58 MTL58

1990/91

Dry

29A x IR29723R IR58025A x IR29723R

6.1 6.0

123* 122*

MTL61 MTL61

1992

Wet

25A x IR52287R

6.7

131*

IR64

1992/93

Dry

25A x IR3235BR

6.8

145*

IR64

*Significantly higher than check at 5% level.

23

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

operation, Kirin Brewer Co., Ltd., and Sumitomo Chemical Co. are also developing and testing rice hybrids (Kato et al. 1994). The Philippines released the first rice hybrid 'IR64616H', registered as 'PSB Rc26H' and named 'Magat' hybrid, in 1994. Another hybrid 'IR68284H' showed standard heterosis of 16-27% across seasons. More hybrids from PhilRice, the International Rice Research Institute, and Cargill are now being evaluated in test nurseries (de Leon et al. 1998). Due to the increasing world population and its requirement for more food, especially in developing countries, the FAO considers the use of hybrid rice technology to be essential for the next 10 years. To meet this goal, the FAO is organizing a task force for Latin America and Caribbean countries. Similarly, FAO is providing financial support to some Southeastern Asian countries, including India, Vietnam, Myanmar, and Bangladesh. It is expected that hybrid rice will be important in fighting world hunger for the next several decades. C. "Bottlenecks" and Potential Solutions

The current Chinese hybri<;l rice cultivars are primarily from the threeline hybrid rice system, but the yield level of these hybrids have reached a plateau since the 1980s (Yuan 1994d, 1997a; Fig. 2.1). An additional 20

-. ~

.d

= Q

:=

!

~

~

8000

Yield

18

7000

......................................

16

// ..................

14 12

.'

6000 5000

Area

10

4000

t=ll

~

8

6::

6

;

3000

'2

~ ~ "0

13

>:

2000 4 1000

2 0

~

--J

(Jl

~

--J 0>

;c; --J --l

0

~

--J CD

~ --J

;C;

co 0

~ ~

~

CD I\J

~

(0

(0

(,)

~

(Jl

CD

CD

CD

~

co

0>

~

CD --J

~ CD CD

~ CD

~

(0

0

~ ~

~ (0 I\J

~ (0 (,)

~

(0 ~

~

(0 (Jl

Year Fig. 2.1.

The leveling out of yield and planting area of hybrid rice in China (1976-1995).

J.

24

LI AND 1. YUAN

threat is that more than 85% of all A lines belong to the "WA" type. This single cyto-sterility system may be vulnerable to the development of destructive pests or diseases. For attaining a higher yield potential from rice heterosis, Yuan (1987) put forward the following three breeding approaches for rice heterosis breeding: (1) three-line method or CMS system; (2) two-line method or T(P)GMS system (the thermo-sensitive or photoperiod-sensitive genic male sterility); and (3) one-line method or apomictic system. The goal is to enhance heterosis, and improve each of the breeding approaches at the following three levels: (1) intercultivar hybrids; (2) intersubspecific hybrids; and (3) distant hybrids (interspecific or intergeneric hybrids). These strategies will be detailed in the following sections. II. HETEROSIS IN RICE

A. Concept of Heterosis

In 1776 Koelreuter published his work on plant hybridization after noting an excessive luxuriance in his Nicotiana hybrids. A hundred years later Darwin (1877) described the hybrid vigor of plants in his book "The Effects of Cross and Self Fertilization in the Vegetable Kingdom." He stated: ... the first and most important conclusion which may be drawn from the observations given in this volume, is that cross-fertilization is generally beneficial and self-fertilization injurious.

At almost the same time (1865) Mendel observed hybrid vigor in his pea hybrids. The term "heterosis" was first coined in a lecture at Gottingen, Germany by Shull in 1914. It referred to "the increased vigor, size, fruitfulness, speed of development, resistance to disease and to insect pests, or to climatic rigors of any kind, manifested by crossbred organisms as compared with corresponding inbreds, as the specific results of unlikeness in the constitutions of the uniting parental gametes" (Shull 1952; Zirkle 1952). Heterosis was first exploited in the 1930s with the large-scale production of hybrid corn, which provided an important impetus for other crops (Pingali et al. 1997). However, unlike in the easily emasculated maize, the inability to emasculate the seed parent had been the primary barrier for the utilization of heterosis in many cross-pollinated and selfpollinated species. The onion research conducted by Jones and Clarke (1943) provided a solution to this problem. They identified male steril-

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

25

ity in the onion cultivar 'Italian Red' (Jones and Emsweller 1936) in 1925, developed the CMS system for hybrid onion production, and revealed the genetic mechanism of eMS in onion (Janick 1989). The breeding strategy for hybrid onion was used by rice scientists in developing three-line hybrid rice. B. Performance of Heterosis

Heterosis is apparent in many morphological and physiological traits. For rice, three main categories of heterosis can be observed. 1. Vegetative Heterosis. Normally F 1 rice hybrids have higher growth rate

and greater vegetative vigor.

Early and Higher Tillering Capacity. When cultivated as a single crop, the rice hybrids 'Nan You 2' and 'Nan You 6' started to develop tillers 12 days after seeding, 6-8 days earlier than their male parental lines. At the Hunan Teacher's College, the largest number of tillers per ha of 'Nan You 2' reached 4.24 million, 0.29-1.25 million more than its parental lines 'Er-Jiu-Nan lA' and 'IR24', and the check conventional cultivar 'Guang Xuan 3' (Yuan and Chen 1988). The growth rate and biomass of the hybrid were greater than those of the parental lines under both high and moderate temperatures. Wider and Deeper Root Distribution and Higher Nutritional Absorption. Root number per plant of 'Nan-You 3' was 121.3% higher than for the conventional rice cultivar 'Guang-Liu-Ai 4', for the same seeding rate (Li et al. 1982). Yichun Agricultural Research Institute in Jiangxi Province, China also reported that the root system of the hybrid 'ShanYou 2' reached 22 cm average length (the longest being 30 cm) and 24 cm average width (the widest being 34 cm) at maturity, compared with 5-9 cm oflength and 9-10 cm of the width at the same stage in the conventional cultivar 'Yi Chun Ai l' (Yuan and Chen 1988). The root system of the hybrids was also larger than that of their parents (Bai and Xiao 1988; Lu et al. 1988). Taller, More Sturdy Culm and Higher Lodging Resistance. 27 of 29 hybrids had positive heterosis for plant height. Guangxi Academy of Agricultural Sciences in China found that wall thickness between the 1st to the 6th internode of the hybrid 'Shan-You 2' was much greater than that of 'Bao-Xuan 3'. Thus, the hybrid rice has higher lodging resistance even though its plants are taller than their parents (Yuan and Chen 1988).

J. LI AND L. YUAN

26

Greater Leaf Area. The leaf area per plant of the hybrid 'Nan You 2' at heading and maturity were 6914 cm 2 and 4124 cm 2 , respectively, compared with 4354 cm 2 and 2285 cm 2 for the male parent 'IR24' (Li et al. 1982). Significantly positive heterosis and heterobeltiosis for flag leaf area was also described in most hybrids (Singh 1997). Superior Physiological Performance. The rice hybrid 'Nan-You 2' had higher photosynthetic efficiency but lower respiration and photorespiration intensity (Lin and Yuan 1980). Greater capacity for synthesis of chlorophyll and faster quenching rate of the chlorophyll fluorescence of the seedling leaves, and higher photosynthetic rate of the flag leaves at the primary heading stage were also observed in hybrid rice as compared to rice inbreds (Li, Wang, and Liu 1990). 2. Reproductive Heterosis and Growth Duration. Hybrid rice generally has higher rice yield. This is due to a larger panicle, more spikelets or longer growth duration.

Larger Panicles, More Spikelets and Higher 1,OOO-grain Weight. The panicle-spikelet structure of China's most popular rice cultivars or hybrids of the last 30 years was studied. The yield increase by 31.3-98.5% of semidwarfrice cultivars in the 1960s, compared with the taller cultivars of the 1950s, was primarily due to an increase of panicle number by 67.5-77.7%. There was little difference in number of spikelets and grain weight. In contrast, the yield increase of hybrid rice by 11.2-32.1 % in the 1970s, compared to the semidwarf rice of the 1960s, came from increase in spikelet number per panicle by 18.0-30.9% and in the 1,000-grain weight by 9.2-12.0% (Table 2.3, Chinese Academy of Agricultural Sciences and Hunan Hybrid Rice Research Center Table 2.3. The panicle-spikelet structure of hybrid rice and conventional rice in a single crop in China. Source: Chinese Academy of Agricultural Sciences & Hunan Hybrid Rice Research Center 1991.

Year

Type of cultivar

Effective panicle number (million/ha)

Number of spikelets per panicle

1,000-grain weight (g)

Yield (kg/ha)

1962-1963

Tall

1.62-2.31

83.6-113.1

25.0-26.6

3465-5580

1964-1965

Semidwarf

2.88-3.87

85.8-113.5

23.5-25.1

6780-7320

1976-1979

Hybrid

2.37-2.99

112.3-133.9

26.0-28.1

7530-9675

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

27

1991). Zeng et al. (1979) also found the 1,000-grain weight of 23 rice hybrids to be superior to the better parent, and the 1,000-grain weight of 31 of 34 rice hybrids was higher than the mid-parent value.

Higher Yield. Various reports (Sun and Cheng 1994) indicated that heterosis for rice yield ranged from 2-157%. Manuel and Palanisamy (1989) also reported that all nine traits measured in 15 hybrids showed heterosis, with the highest yield heterosis being 46%. China's Jiangxi Academy of Agricultural Sciences found 28 out of 29 rice hybrids to exhibit superior yield to their better parent or heterobeltiosis. The yield gain of 18 hybrids was statistically significant. All 29 hybrids exhibited heterosis over the local check cultivars. The average yield heterosis was 35.5% (Yuan and Chen 1988). Longer Growth Duration. The growth duration of rice hybrids are highly correlated with the ecotype of their parental lines. Most data indicate negative heterosis in days to flowering (Namboodri 1963; Dhulappanavar and Mensikai 1967; Purohit 1972; Chang et al. 1973; Mallick et al. 1978; Singh et al. 1980; Lin and Yuan 1980; Fujimaki and Yoshida 1984). The inter-subspecific hybrids have longer growth duration than the intercultivar hybrids. Song et al. (1990) reported that the growth duration of seven out of nine indica-japonica rice hybrids was 15-28 days longer than 'Shan-You 63', a late hybrid rice check. 3. Heterosis for Resistance to Adverse Environmental Conditions. Rice

hybrids have exhibited good resistance to some diseases, insect pests, drought (Yab and Chang 1976; Tian et al. 1980), low temperature, poor soil fertility, high salt content (Akbar and Yabuno 1975), deep water (Singh 1983), and other adverse conditions (Lin and Yuan 1980). Therefore, hybrid rice can be grown between 50 N and 18°N and, in South China, at altitudes up to 1500 III (Chen 1985). Researchers at the Hunan Agricultural College tested the resistance to rice blast of 224 rice hybrids with their parental lines. Gfthe 224 hybrids, 102 showed.dominance for resistance and 15 showed incomplete dominance (Yuan and Chen 1988). It was also reported that all 140 hybrids under three nitrogen levels (0, 60 and 120 kg/hal showed yield heterosis in both dry and wet seasons (Young and Virmani 1990). In the IRTP nurseries, including locations in India, Malaysia, the Philippines, and Vietnam during 1980-1986, the average standard heterosis in the tested rice hybrids was 108-117% under different environmental conditions (Sun and Cheng 1994). 0

28

J. LI AND L. YUAN

C. Genetic Basis of Heterosis

Bruce (1910) explained heterosis as the combined action of favorable dominant or partially dominant factors. Gustafsson (1946), Hull (1945), Castle (1946) and others emphasized interallelic action as the basis of heterosis (Hayes 1952). For practical hybrid rice breeding, the explanation of "gene interaction" was proposed. It is assumed that rice heterosis arises from the overall effects of three types of gene interactions: allelic gene interaction, non-allelic gene interaction, and the interaction between the nuclear and cytoplasmic gene(s) (Yuan and Chen 1988). 1. Interaction of Allelic Nuclear Genes

Dominance Effects. The dominance hypothesis was first suggested by Davenport (1908). Based on the "dominant complementary" hypothesis, Jones (1917) explained heterosis as the integration of beneficial dominant genes from both parents of F 1 hybrids, and the inhibition of harmful recessive genes by the dominant beneficial genes. In a recent molecular analysis of rice heterosis using RFLP markers, for 82% of 37 significant QTL the heterozygotes were superior to the respective homozygotes. There was no correlation between most traits and overall genome heterozygosity. Some recombinant inbred lines in the Fa population had phenotypic values superior to the F 1 for all of the traits evaluated. Moreover, this molecular study did not show evident digenic epistasis and suggested that dominance complementation, instead of overdominance, is the major genetic basis of heterosis in rice (Xiao et al. 1995). Yield heterosis ofIR58025A and IR62829A hybrids resulted from the complementation of traits between parents (Vijayakumar et al. 1997). But this hypothesis does not take into account non-allelic gene interaction or that quantitative traits such as yield are governed by polygenes with additive effect, i.e. there is neither dominance nor recessiveness. Over-dominance Effects. Shull (1908) proposed over-dominance as the basis of heterosis. This hypothesis stated that the heterozygote was superior to the two homozygotes for the same gene. Therefore, an F 1 individual having the greatest number of heterozygous alleles will be most vigorous compared to the two parents. Brewbaker (1964) explained overdominance as the effects of (1) supplementary allelic action; (2) alternative pathways; (3) optimal amount; and (4) hybrid substance. A recent study of the molecular basis for heterosis using QTL analysis for seven agronomic traits of maize also suggested that over-dominance played a role in the heterosis observed (Stuber et al. 1992). Although this hypoth-

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

29

esis was preferred, especially before the 1970s, it does not explain why some traits in rice hybrids are inferior to their parental lines. Many researchers no longer think that over-dominance makes large contributions to heterosis (Crow 1997, 1999). 2. Interaction of Non-allelic Nuclear Genes. In maize breeding, signifi-

cant amounts of epistasis may exist in certain specific combinations, but the magnitude is small (Sprague 1983). Yu et al. (1997) reported that there was little correlation between marker heterozygosity and trait expression, but digenic interactions frequently existed in the F 3 progeny derived from 'Zhen-Shan 97 x Minhui 63'. This suggested that epistasis significantly affected the performance of heterosis in rice. 3. Interaction between Nuclear Gene(s) and Cytoplasmic Gene(s). In plants, all three genetic sources-nuclear, mitochondrial, and chloroplast genomes-are involved in heterosis (Gillham 1978; Kirk and Tilney-Bassett 1978; Srivastava 1983) and, in some cases, the cytoplasmic contributions are critical (Wagner 1969; Srivastava 1972). The reciprocal F 1 crosses in some rice hybrids have shown different levels of heterosis. Furthermore, the same nuclear genome in different cytoplasmic backgrounds has shown different heterosis levels. The cytoplasm, therefore, must play some role in rice heterosis. A study on the effect of eight rice cytoplasms on 12 traits indicated that all eight cytoplasms negatively affected most traits such as plant height, panicle length, number of kernels, number of effective tillers, number of panicles, seed setting percentage, 1,OOO-grain weight, grain weight per plant, yield, and heading date (Sheng 1987). D. Prediction of Heterosis

Heterosis is a complicated phenomenon that is influenced by both genotype and environment. There is no single method that can accurately predict heterosis; however, the following genetical and biochemical methods have been suggested. 1. Genetic Diversity. Genetic diversity can be estimated using the following three methods: (1) geographic origin, (2) multivariate analysis using Mahalanobis D2 statistics (Mahalanobis 1936; Ram and Panwar 1970; Vairavan et al. 1973; Maurya and Singh 1977; Rao et al. 1981; Julfiquar et al. 1985; Vaidyanath and Reddy 1985), and (3) isozyme and RFLP polymorphism (Schwartz and Laughner 1969). Genetic diversity or distance has been reported to be highly correlated with the level of

30

J.

LI AND L. YUAN

heterosis (Maurya and Singh 1978; Xu and Wang 1981; Li and Ang 1988; Zhang et al. 1987). For 43 rice hybrids, yield potential and heterosis had significant positive correlation with genetic distance for 15 indica-indica crosses and 6 japonica-japonica crosses, but no correlation for 22 indica-japonica crosses. The genetic distance method, therefore, seems to be predictive for intra-subspecific heterosis, but not for intersubspecific heterosis (Xiao et al. 1996a). Other workers have found either no direct correlation between heterosis and genetic distance (Cress 1977; Khalique et al. 1977; Peng et al. 1991; Xie 1993) or that it was only a weak indicator of heterosis (Liu et al. 1997b; Liu and Wu 1998). 2. Combining Ability. Crosses with great heterosis are more likely obtained when at least one of the parents has high GCA effects (Peng and Virmani 1990). However, prediction of heterosis on the basis ofGCA may not always hold true (Srivastava and Seshu 1983; Kumar and Saini 1981). 3. Isozymes. There have been two main isozyme methods used to pre-

dict heterosis.

Esterase Isozyme Complementary Band. Esterase isozyme is a comparatively dependable biochemical indication for predicting heterosis. If an F 1 hybrid has the specific band(s) from both parental lines, i.e. dominant complementary band, there will be heterosis for this combination (Xiao and Liu 1981; Shi et al. 1988a,b). But it still seems difficult to predict the existence of heterosis because the existence of the complementary band(s) is not always coincident with the performance of heterosis. Therefore, some researchers do not agree that heterosis can be predicted by the complementary isozyme band(s) (Peng et al. 1988).

Isozyme Difference Index. The isozyme difference index uses the number of isozymes that show difference in band(s) among the F 1 and its parental lines. A difference in at most six isozymes and at least two isozymes in the F 1 hybrids was reported from 12 rice hybrids and eight isozymes. The F 1 hybrid would show heterosis if its isozyme difference index was more than four (Zhu and Zhang 1987). This method is weak and requires further evidence. 4. Mitochondrial Complementation. This method was first proposed by

McDaniel and Sarkissian (1966). The concept is that heterosis might be estimated based on the oxidization activity level of mitochondria in the F 1 and both parental lines at the seedling stage. However, mitochondria are derived from the female parent only.

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

31

A modification of the mitochondrial complementation method is the homogenate complementation method. The homogenate complementation value is positively correlated with the heterosis. Homogenate complementation measures the phenomenon of the oxidation activity of the F 1 being higher than that of the mean value of the two parental lines. Yellowish young plants from both parental lines were provided homogenate mixed at the ratio of 1:1. It was speculated that the complementation of the two homogenates was due to the interaction between the mitochondria and the supernatant of both parental lines (Yang 1991a). 5. ATP Content. ATP content of parental lines in yellowish rice seedling tips is correlated with heterosis of rice. The ATP content of the parental lines of 15 rice hybrids with heterosis was more than 2.0 x 10-6 mM/g of homogenate. The prediction of heterosis was higher than 90% (Yang et al. 1990; Yang 1991b). This method uses a small amount of tissue and takes only a short time, but it requires further study. 6. Enzyme Activity. Some researchers reported that the activities of some enzymes (such as esterase, nitrate reductase, and superoxide dismutase) are correlated with the level of heterosis. To improve the accuracy of the prediction, other enzymes should be tested (Liang 1991; Liang et al. 1991; Xiao et al. 1991; He 1990). 7. Performance of Parents. In general, high-yielding parents produce a larger proportion of high-yielding hybrids than do low-yielding parents (Virmani 1994a).

E. Approaches for Utilization of Heterosis 1. Approaches

Utilization ofIntercultivar Heterosis. Most current rice hybrids are intercultivar crosses, which can yield 20-30% more than improved semidwarf conventional rice inbreds. China's intercultivar hybrid yields have been plateaued at this level for years, due to the narrowing parental germplasm diversity (Luo and Yuan 1990). Utilization of In tersubspecific Heterosis. In the 1950s the idea was proposed to develop rice inbreds by means of indica-japonica crossing (Yang 1959). Some high-yielding indica-japonica inbreds were released in China, such as 'Ai-Jing 23', 'Er-Wan 5' and 'Liao Jing 5', 'Milyang' system

32

J. LI AND L.

YUAN

rice cultivars in Korea, and 'Chogoku 91' in Japan. 'C57', the first restorer line for China's three-line system japonica hybrid rice, used restorer gene(s) transferred from indica to 'Jing-Ying 35', a japonica cultivar. Although indica-japonica rice hybrids have strong heterosis, normally yielding 30-50% more than the intercultivar rice hybrids, four main obstacles exist: low seed set, excessive plant height, excessive growth duration, and unfilled kernels (Wang et al. 1991). Discovery of "wide compatibility" (WC) gene(s) by Ikehashi (1982) provides a much easier use of the intersubspecific indica-japonica heterosis. More details will be discussed in Section VI. Intersubspecific heterosis from crosses between an indica and a japonica cultivar cannot be exploited directly, owing to the genetic divergence being too large and poor adaptability to tropical conditions (Yang 1990b; Virmani 1994b). Another concern is that eating and cooking qualities of typical indica-japonica crosses segregate, and therefore are not acceptable to most rice consumers in China (]. S. Zou, pers. commun.). Yuan (1991a,b) suggested the alternative ofusingjavanica as a parent. Genetic divergence is larger between javanica and indica or between javanica and japonica types than for intercultivar crosses. Both indica-javanica and japonica-javanica hybrids have shown stronger heterosis than the intercultivar rice hybrids. These crosses have fewer problems than the typical indica-japonica intersubspecific crosses.

Utilization of Distant Heterosis. Many agricultural scientists have tried to transfer target genes or traits to rice from maize, sorghum, bamboo, and other distant plant species. For example, marker-assisted selection can now be used to transfer the desired gene(s) from wild rice or other distant species to cultivated rice (Yuan 1996; Xiao et al. 1996b; Tanksley and McCouch 1997). Once inbred rice has been improved using biotechnology, genetic sources for stronger heterosis may be found, especially using NMS (nuclear male sterility), which is not limited by cytoplasmic function. 2. Methodology

Three-line System. Currently most commercial rice hybrids are threeline system rice hybrids with intercultivar heterosis. Breeders are now trying to transfer wide compatibility gene(s) to the parental A, B, or R lines. Some intersubspecific rice hybrids using the three-line system have been successfully developed in China. Because the breeding and seed production procedures are complicated, labor-intensive and costly,

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

33

in the long run, the three-line system will be replaced by less complicated systems.

Two-line System. There are two techniques that have been used: chemical emasculation and the T(P)GMS system. Chemical emasculation of plants was reported as early as the 1950s and China began using chemical emasculation to produce rice hybrids in the 1970s. Some highyielding rice hybrids such as 'Gan-Hua 2' were successfully released. The technique will be presented in more detail in Section V. The T(P)GMS system refers to the thermo-photoperiod sensitive genic male sterility. T(P)GMS lines can be used for male sterile line multiplication and F 1 seed production under different temperature or daylength regimes. This system is genetically controlled by nuclear gene(s) and thus there is no negative effect from the cytoplasm and no risk of unilateral cytoplasmic breakdown. There is more opportunity to develop elite rice hybrids using the T(P)GMS system than using the three-line system. Omission of the B line used to maintain male sterility for the three-line system reduces the seed cost. Also, it is easier to combine T(P)GMS gene(s) with the WC gene(s). One-line System (or Apomictic System): Apomixis is asexual reproduction via seeds. It results in no deterioration of heterosis with year-afteryear seed production because no genetic segregation occurs. The idea for heterosis fixation by apomixis was proposed in the 1930s, but the only prominent example has been for Buffelgrass (Bashaw 1980a,b). To bypass the need for hybrid rice seed production each year, Zhao (1977) and Yuan (1987) proposed utilization of apomixis in rice. Apomixis breeding for fixing rice heterosis will be described in Section VIII. III. MALE STERILITY IN RICE Cytoplasmic male sterility in rice has been reported by many scientists (Nagai 1926a,b; Ishikawa 1927; Miyazawa 1932; Takezaki 1932; Ramanujam 1935; Hara 1946; Jones 1952; Weerarathe 1954; Sampath and Mohanty 1954; Katsuo and Mizushima 1958; Yuan 1966; Athwal and Virmani 1972; Hoff and Chandrapanya 1973; Razzaque 1975; Trees and Rutger 1978; Mahadevappa and Coffman 1980; Rutger and Shinjyo 1980). The first CMS line was developed by Shinjyo and Omura (1966) in Japan using 'Chinsurah Boro II' cytoplasm. The CMS line 'Er-Jiu-Nan 1 A' was the first to be put into commercial production in China in the early 1970s.

J. LI AND 1. YUAN

34

A. Morphology, Cytology, and Histology of Male Sterile Lines 1. Morphological Features. The male sterile rice A line [male sterile cytoplasm (S), recessive nuclear gene (fj)] appears morphologically similar before heading to its maintainer B line [normal cytoplasm (N), recessive nuclear gene (fj)]. After heading, sterility can be recognized from various morphological features involving the anthers and flowering behaviors (Table 2.4). Most of the morphological features of the T(P)GMS lines or gametophytic male sterile lines are almost the same as those of the sporophytic male sterile lines. 2. Cytological Features. Laser and Lerstern (1972) summarized the cytological studies conducted between 1925 and 1972 that analyzed the pollen abortion resulting in male sterility in crops. Rice male sterile lines are classified by the stages of pollen abortion: pollen-free abortion, uninucleate stage abortion, binucleate stage abortion, and the trinucleate stage abortion type.

Table 2.4. The morphological differences between sporophytic A and B lines. Source: Yuan 1985; Sun and Cheng 1994.

Morphological feature

Maintainer line

Sporophytic indica male sterile line

Plant height

Taller than the A line

Shorter than the maintainer line

Tillering capacity

Lower than the A line

Higher, and longer tillering stage

Heading date

Earlier than the A line

3-5 days later than the B line

Panicle

Normal heading

Shorter neck, panicle basal part enclosed in the leaf sheath for the dwarf sporophytic indica type

Flowering behavior

Concentrated flowering time and shorter glume opening time

Diffused and longer flowering time

Anther shape

Plump, golden in color

Empty, slender, thin, milky-white or yellowish in color

Anther dehiscence

Dehiscent

Indehiscent

Pollens

Round and dark-brown when stained with I-KI

I. Irregular in shape and unstained with I-KI; or II. Round and unstained;or III. Round and light brown in color

Fertility

Normal

Self incompatible

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

35

Pollen-free Type Abortion. In this type, pollen abortion occurs before the uninucleate stage. This type has mainly three patterns of abnormal pollen development: sporogenous cells; pollen mother cells; and abnormality after tetraspore formation (Hunan Teachers' College 1972). Uninucleate Stage Abortion. In the uninucleate stage of 'V20A' the percentage of pollen abortion is 96.7% (Rao and Xie 1983). Pollen abortion of WA-type male sterile lines normally occurs at the uninucleate stage; the pollen grain has a dissolved nucleus or nucleoli to some degree, in addition to a collapsed thin cell wall, a vague germination aperture, and a very small or condensed protoplast content (Sun Yat SenUniv. 1976). For the two-line system, some pollen grains of the T(P)GMS lines abort at the uninucleate stage with withered pollen grains, very little cytoplasm, and disappearance of the nucleus (Wu and Wang 1990). The pollen grains that have aborted by the uninucleate stage are irregular in shape (often triangular under the microscope). Uninucleate-type abortion is also referred to as typical abortion. Binucleate Stage Abortion. In this type the reproductive nuclei and nutritive nuclei of most pollen grains start to collapse only at the binucleate stage, such as in the 'Hong-Lian' type A lines. Chiang et al. (1981) reported that 80.3% of the pollen grains of'Hong-Lian' male sterile lines aborted at the binucleate stage, as compared to 12.8% that aborted at the uninucleate stage. For this abortion type, part of the pollen mother cells vacuolize and damaged nuclei are without distinguished cell walls. Some cells form protoplasmic masses. Two or three pollen mother cells connect at their nucleoli in irregular ways. The aborted pollen grains are mostly spherical, hence, the term spherical abortion. Trinucleate Stage Abortion. BT-type 'Taichung 65A' has no distinguishable abnormality in pollen development before the trinucleate stage. At the binucleate or trinucleate stage, size of the nucleoli is reduced in only a few cells with some nuclear membrane collapse (Sun Yat Sen Univ. 1976). At the anaphase of the reproductive karyokinesis, the chromatin grains disappear at the later stage in 'Nong-Jin 2 A'. In 'Fu-You lA', many micronucleoli are scattered in the cytoplasm through nuleolar budding and then disappear. In some cases, reproductive nuclei form two sperms of different size, and some nutritive nuclei of equal size (Teng 1982). Some starch has already been produced and the pollen grains stain brown using I-KI solution, but a lighter brown than for normal pollen grains; hence, this is called stained pollen abortion.

36

J.

LI AND L. YUAN

Pollen abortion can occur at any time from sporogenous cells to the trinucleate stage. Different abortion types occur, not only in different male sterile lines, but also in different flowers of the same rice plant, or even different anthers of the same floret (Pan et al. 1982). The four abortion types mentioned above were classified for practical purposes as: restorermaintainer relationship (WA type-uninucleate abortive; HL type-binucleate abortive; BT type-trinucleate abortive) and cytoplasm-nucleus relationship (sporophytic male sterile system-uninucleate abortive; gametophytic system-binucleate and trinucleate abortive). 3. Histological Features. The following histological abnormalities were found in the stamens of male sterile rice plants:

Abnormality of the Anther Wall. The anther pulling force of some rice male sterile lines is not strong enough to open a dehiscence cavity, as occurs in the normal rice plant. In some cases the dehiscence cavity is not formed or is formed on only one side of the anther. Consequently, anthers fail to dehisce. For example, although there is a strong pulling force in the anthers of some male sterile lines such as in WA-type 'NanTai 13 A', no dehiscence cavity is formed on either side of the anther, thus the pulling force cannot open the anther wall to bring about dehiscence (Chou 1978; Pan and He 1981a,b). Abnormality of the Intercellular and Tapetal Cells. The tapetal cells encircle sporogeneous cells and provide nourishment to the reproductive cells. Abnormal development or damage of intercellular or tapetal cells often results in the pollen abortion of rice male sterile lines (Rao 1988). Excessive proliferation of tapetal cells in HL-type 'Hua-Ai 15 A' causes tapetal periplasmodia to form, pushing the pollen mother cells to the center of the anther, which results in the dissolution of the pollen mother cells. In the case of the WA-type 'Hua-Ai 15A,' fibrocytes of the dermal layer become deformed, thus damaging the tapetal cells and causing pollen abortion (Xu 1979). In 'Er-Jiu-Nan 1 A' and some other WA-type A lines, vacuolization of the intercellular cells and abnormal increase of the radial thickness of the middle lamella cells by the uninucleate stage push the tapetal cells towards the center of anther cells. These cells are distinctively thin with many vacuoles and light-colored cytoplasm. By the binucleate stage, the pollen cells are completely aborted with complete vacuolization and withering of the intercellular cells. At this point the secondary tapetal walls in the intercellular cells can be easily observed under the microscope (Pan 1979; Pan and He 1981a,b). The tapetal and endothecial abnormalities are also the main

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

37

histological causes of male sterility in 'V20A' and 'Pragathi A' (Yogeesha and Mahadevappa 1995). In some three-line system and two-line system male sterile lines, dissolution of tapetal cells is abnormally delayed. For example, the tapetal cells of 'Guang-Xuan 3 A' remain intact even at the trinucleate stage, before pollen abortion occurs (Guangxi Teachers' College 1975). The T(P)GMS line, 'Nong-Ken 58s', under daylength of 14 h, maintains an intact structure of tapetal cells with rich cytoplasm, nuclei, and a small number of vacuoles, plastids and other organelles. Even though metabolism in tapetal cells is active and there is no appearance of pollen dissolution, the grains still start to abort (Li et al. 1993). However, for the same line, 'Nong-Ken 58s' under long daylengths, the intercellular cells and the inner tangential wall of the tapetal cells started to dissolve at meiosis. At the end of the uninucleate stage, the intercellular cells are almost completely collapsed, and the cytoplasm forms a cytoplasmic mass due to the dissolution of the tapetal cell wall (Wang and Tong 1992). This abnormal tapetum development and effect on the male sterility are also observed in other T(P)GMS lines (Zhang et al. 1994a).

Abnormality of Filaments and Connective Vascular Bundles. The filaments and the connective vascular bundles are the channels that transport water and nutrients in anthers. Their development directly affects the quantity of nutrients available to an anther. Filament vessel degeneration has been observed in the wild abortive and the pollen-free abortive types of male sterile lines. The degree of degeneration is correlated with the pollen abortion percentage (Pan 1979). For HL-type 'Hua-Ai 15 A', abnormal development occurs in the connective vascular bundles and the tracheary cells. Vessels of the connective vascular bundles develop poorly, with an enlarged annular space. In some cases the vessels and vessel cavities are damaged. The junction at which vessel cells meet becomes disconnected and the cells become fibrillous with loose connections, disorderly arrangement, and degenerated function. Poor development of connective vascular bundles occurs at the uninucleate stage and binucleate stage, with poor differentiation of phloem and xylem in the WA-type male sterile lines. The cells wrinkle and shrink. The vascular bundles are not highly visible owing to the degenerated and disordered cells. The extent of development of the vascular bundles is negatively related to the degree of pollen abortion (Sun Yat Sen Univ. 1976). There are different abnormalities of the vascular bundles in different rice male sterile lines, and even of male sterile lines with the same nuclear background but different cytoplasmic sources. Vascular bundles of the fertile anthers of 'Nong-Ken 58s' are

38

J.

LI AND L. YUAN

similar to those of the normal rice cultivars, yet abnormal development of the vascular bundles and thin-wall cells of the sterile anthers of 'Nong-Ken 58s' exist at some stages (Wang and Tong 1992). 4. Classification of Male Sterility. More than 600 rice male sterile lines have been developed in China. Classification systems were proposed in the late 1970s and 1980s (Zhu 1979; Li 1980a; Wan et al. 1988). At present, the taxonomic system (Table 2.5) suggested by Wan et al. (1988) is often used. It is a five-step taxonomic key based on (1) the inheritance of male sterility (sporophytic male sterility or gametophytic male sterility); (2) the pollen abortion stage [uninucleate (typical abortion), binucleate, (spherical abortion), or trinucleate (stained abortion)]; (3) the restorer-maintainer relationship; (4) nucleus substitution type, such as wild-cultivar, cultivated-wild; and (5) cultivar differentiation based on the cytoplasmic source.

B. Physiological and Biochemical Basis 1. Transportation and Metabolism. Abnormality of vascular bundles restricts transportation and metabolism of nutrients, which was revealed by an experiment using 32P. In this experiment, 32p was transported to panicles in large amounts in 'Nong-Ken 58s' under conditions favoring fertility, in the sequence: panicle> flag leaf> the leaf beneath the flag leaf. Under conditions favoring sterility, the 32p amount was small in the panicles, and the sequence above reversed compared to conditions for fertility (He et al. 1992). The three-line system male sterile lines had similar mechanism for transportation and metabolism to 'Nong-Ken 58s'. Researchers at Sun Yat Sen University reported that absorptivity of 32p, 14C and 35S in the WA-type male sterile lines 'Er-Jiu-Ai A', 'Zhen-Shan 97A' and in BT-type 'Taichung 65 A' was weaker in their anthers, panicle branches, and the vascular bundle system of the glumes in comparison to the corresponding B lines. Interestingly, the same absorptivity occurred in the ovules of A and B lines. It is concluded that there is a metabolism barrier in anthers of the male sterile lines, whereas the metabolism is normal in their ovules. For 'Nong-Ken 58s', a marked decline of photochemical activity of the chloroplasts, such as lower PS II photochemical activity and less chlorophyll b in the chloroplasts, induced by long daylength, may reduce the available photosynthetic products and result in male sterility (Tang et al. 1994). ATP content is highly related to the fertility performance ofT(P)GMS lines. For example, the ATP content of 'Er-Yi lOSs' is much lower in male sterile plants at the early uninuclear stage than for fertile plants,

Table 2.5.

The taxonomic system for the classification of the three-line system rice male sterility. Source: Wan et al. 1988.

Fertility inheritance Sporophytic pollen abortion

Gametophytic pollen

Pollen abortion stage Uninucleate (typical) abortion

Binucleate (spherical)

Restorermaintainer types DW

wild x cultivated

WA

wild x cultivated

w

Dong-pu wild rice (DW), Tian-Dong long Awn wild rice Hai-Nan wild abortive rice (WA), LiuZhou red-awned wild rice

cultivated x wild

Chao-Yang 1, Lian-Tang-Zhao

indica x japonica

Sheng-Qi, Nan-Guang-Zhan

japonica x indica

Gui-Hua-Huang

indica x indica

Dissi

japonica x japonica

Zhao-Tong-Bei-Zi-Gu

LW

wild x cultivated

Long-An wild rice, Guang-Xi wild rice

indica x indica

Indonesia Paddy rice 6, Gu Y-12

G

cultivated x wild

Jin-Nan-Te 43

indica x indica

Gambiaka

wild x cultivated

Hai-Nan red-awn wild rice

indica x japonica

Tian-Ji-Du

HL

abortion

co

Cytoplasmic source

IP

abortion Trinucleate (stained) abortion

Nucleus substitution

BT

indica x indica

Jing-Quan-Nuo

indica x japonica

Chinsurah Boro II

40

J. LI AND 1. YUAN

only 1/2-1/7 of that of fertile plants and 1/9-1/10 of that of conventional inbred rice (Deng et al. 1990). It was also found that ATP in the anthers of 'An-Nang s-l' and 'Heng-Nong s-l' under high temperature was decreased, and the respiration rate of the floscules and anthers declined gradually (Chen et al. 1994). 2. Protein and Amino Acid Content. The protein content in anthers of

'Er-Jiu-Nan lA' is lower than that of 'Er-Jiu-Nan 1 B' and the restorer line 'IR661' (Shanghai Plant Physiology Research Institute 1977). The content of free histones in male sterile lines is also much lower than in their corresponding B lines from meiosis stage to the trinucleate stage (Dai et al. 1978). Xu et al. (1992) indicated that 'Zhen-Shan 97B' produced five bands of soluble chloroplast proteins and six bands of the water-soluble components, but 'Zhen-Shan 97A' did not. In T(P)GMS system, studies demonstrated that, besides the change of soluble protein content at developmental stages, there were different patterns of protein bands between 'Nong-Ken 58s' and 'Nong-Ken 58' or between some TGMS lines and their ancestral lines, and that some specific protein bands for T(P)GMS lines were present (Shu et al. 1989; Wang, Xiao and Liu 1990; Bai and Tan 1990; Peng and Wang 1991; Huang, Tang, and Mao 1994; Wang et al. 1997b). Study of specific proteins can promote understanding of the mechanism for male sterility. However, currently there is no convincing evidence of a relationship between specific polypeptides and the male fertility performance of rice. In fact, some results from protein studies are contradictory. For the three-line system, the higher amino acid content of A than of B or R lines indicates that synthesis of protein is slower than protein degradation, or that a barrier exists for protein synthesis in the male sterile lines. In some cases, the relative amounts of amino acids differ among the rice male sterile lines. For example, one study demonstrated that proline content was lower in male sterile plants compared to their maintainer lines, while the asparagine content was higher in the male sterile lines than in their respective maintainer lines (Shanghai Plant Physiology Research Institute 1977). Out of 17 amino acids examined, the relative amounts of proline and alanine were most related to pollen abortion in 'Nong-Ken 58s' and 'V20A' compared to 'Nong-Ken 58' and 'V20B' (Xiao et al. 1987). The tendency for declining proline content also occurred for chemically emasculated male sterile and CMS lines. Proline content in B or R line anthers is three to six times more than in the sterile anthers (Raj and Siddiq 1986; Yu et al. 1991). It seems that reduced proline content hinders carbohydrate metabolism and decreases

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

41

the content of other amino acids. Nutritional disorders are a general cause of pollen abortion in male sterile rice lines. 3. Activity of Enzymes. Activity of peroxidase in rice plants of pollenfree sterile lines is stronger than in fertile plants. Peroxidase activity in 'V20A' is stronger than that in the fertile anthers at the early uninucleate stage, but it decreases with pollen abortion and reaches the lowest at the binucleate stage. The increase of peroxidase activity in 'Nong-Ken 58s' is similar to that in 'V20A', but the lowest activity of peroxidase in the sterile anthers of 'Nong-Ken 58s', under long daylength, is at the trinucleate stage, only 37.5% of the activity at the uninucleate stage (Chen and Xiao 1987). The V-max value of ribulose-1,5-bisphosphate carboxylase of male sterile lines is higher than the restorer line or F 1 hybrids (Wei et al. 1994). The activity of other enzymes such as ADH (alcohol dehydrogenase), phosphorylase, RuBPcase, catalase, superoxide dismutase, glycolic acid oxidase, S-adenomethionine decarboxylase, and arginine decarboxylase may also affect the fertility in 'Nong-Ken 58s' (Zhu and Yang 1992). 4. Growth Regulator. Decrease ofIAA by oxidases has been observed to

hinder metabolism during sporophytic development in anthers and bring about pollen abortion (Huang et al. 1984; Yang, Zhu, and Tang 1990). GA and ABA also influence male sterility of 'Nong-Ken 58s' (Yang, Zhu, and Tang 1990; Nakajima et al. 1991; Zhang and Zhou 1992). The study on 'Norin PL12' shows that the presence of GA 4 / 7 is closely related to the expression of the TGMS gene and subsequent development of pollen and anthers (Honda et al. 1997). A significant negative correlation was observed between the ethylene release rate of young panicles and the corresponding pollen fertility in 'Nong-Ken 58s'. Application of AVG (aminoethoxy vinylglycine) causes the fertility of 'Nong-Ken 58s' under long daylength, but the fertility level decreases sharply in 'Nong-Ken 58s' under short daylength treated with ACe (Li, Luo, and Qu 1996). Other studies revealed that ethylene biosynthesis was correlated with the performance of fertility in T(P)GMS lines (Luo et al. 1990). 5. Products of Cytoplasmic and Nuclear Genes. The zymograms of restriction enzymes applied to mtDNA can be strikingly different between rice male sterile lines and maintainer lines. The gene structures for subunit I and subunit II of cytochrome in mitochondria differ between A and B lines, but no difference is found in a Hind III zymogram of ctDNA of

42

]. LI AND L. YUAN

both lines (Kadowaki et al. 1986; Liu et al. 1988; Sakamoto et al. 1990). In contrast, a special dsRNA (double-stranded RNA) was found with 18 kb molecular weight in mitochondrial nucleic acid of the BT-type CMS lines and of 'Nong-Ken 58s', but not of their maintainer lines and 'NongKen 58' (Zhang and Wang 1990; Wang et al. 1990). Some scientists believe the mitochondrial DNA modifications may support the hypothesis of the mitochondrial inheritance of eMS in rice (Mignouna et al. 1987). But, the relationship between the differences between mtDNA or ctDNA and male sterility requires further study. As for the products of mtDNA and ctDNA translation in rice male sterile lines, there is evidence for large differences in polypeptides with male sterility, a result of genes from the mitochondria, chloroplast, and nucleus (Liu 1986; Xu et al. 1992). There have been few reports on the relationship between male sterility and nuclear gene products in rice. That a specific protein causes male sterility in rice has not been confirmed. The abnormal transcripts of the atp6 gene produced in the antisense direction may be involved in cytoplasmic male sterility (Kadowaki et al. 1990; Akagi et al. 1994). CMS anthers were also found to have lower insoluble polysaccharides, proteins, and RNA content than their maintainer lines (Yogeesha and Mahadevappa 1995). Generally the metabolic level of rice male sterile lines are lower than for maintainer lines, especially in the production of starch, proteins, and change of enzyme activity. Male sterility in the three-line system is related to mitochondria and chloroplasts, so mtDNA, ctDNA, and their metabolism have been closely studied. 6. Other Biochemical Factors. Several other biochemical factors have been implicated in male sterility. These include a weaker oxygen scavenger system, high content of H 20 2 and 02; lower efficiency of oxidative phosphorylation, a high level of lipid peroxidation in the anthers of CMS lines or T(P)GMS lines, and a higher level of aspartic acid in sterile anthers (Raj and Siddiq 1986; Chen and Liang 1991, 1992; Liang and Chen 1993).

c. Genetic Basis Despite a great deal of research on the genetic basis of male sterility in rice, the interaction between the nuclear and cytoplasmic gene(s) remains unclear. Several hypotheses have been proposed to explain the mechanism of male sterility.

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

43

1. Nuclear Male Sterility (NMS). In some cases, nuclear male sterility is conditioned only by one recessive gene and female fertility is unimpaired (Kaul 1988). If male sterility is conditioned by recessive gene(s), a maintainer cannot be identified. In contrast, restorer cannot be developed if male sterility is governed by dominant gene(s). 2. Cytoplasmic-Nuclear Male Sterility. In this case, male sterility is controlled by an interaction of cytoplasmic and nuclear gene(s). Only in this type can a maintainer (B) line and a restorer (R) line be found and developed to attain a complete three-line system. The cytoplasms controlling male sterility in rice are described in Table 2.6.

3. Other Hypotheses

.Cytoplasmic Male Sterility. The male sterility is controlled by the cytoplasmic gene(s) alone, so it is impossible to find a restorer for the cytoplasmic male sterile lines (Edwardson 1956). Some researchers do not accept this explanation of male sterility. "Relationship" Theory. Male sterility is considered to be a quantitative trait, resulting from segregation in the F 1 generation and continuous distribution of male sterility in the F 2 generation. Moreover, male sterility is influenced by environmental factors such as temperature. Male sterility is assumed to be generated by the lack of coordination of the genetic factors between the two parental lines (Pei 1980). 4. Genes Controlling Male Sterility in Rice

Genes Controlling Male Sterility in CMS Lines. Male sterility of WA-type has been hypothesized to be controlled by a single recessive gene (Wang 1980), by two recessive genes (Gao 1981), or by multiple minor genes (Fu and Wang 1988). It is commonly thought that male sterility of the BTtype male sterile lines is controlled by one pair of recessive genes (Shinjyo 1984; Sheng 1994). Genes Controlling the Male Sterility in T(P}GMS Lines. The segregation ratios of the F 2 and Bel show that the photoperiod-sensitive genic male sterility in PGMS lines such as 'Nong-Ken 58s' is controlled by a single gene (Lu and Wang 1986; Jin et al. 1987; Lei and Li 1989; Zhang et al. 1990; Zhang and Zhu 1991; Lin et al. 1996) or by two major recessive genes (Sheng 1992; Shao et al. 1993; Shao and Tang 1993; Wan and Ma 1996; Yang 1997). Zhang et al. (1994b) studied a PGMS line, '31001s',

*'*'-

Table 2.6.

Cytoplasmic sources for inducing male sterility in rice. Source: Kinoshita 1997.

Cytoplasm

Name

Maintainer

Restorer

Reference

cms-bo ems-id ems-TA cms-CW ems-WA

Chinsurah boro II cytoplasm Lead rice cytoplasm TA820 (Tadukan) cytoplasm Chinese wild rice cytoplasm Wild abortive cytoplasm

Taichung 65 etc. Fujisaka 5 Norin 8 Fujisaka 5 IR24 etc.

RfI-a RfI-a Rf2 Rf2

cms-HL ems-ok (ems-jpj ems-ARC ems-GAM

Red awned wild cytoplasm Akebono cytoplasm

IR54753A etc. Lien-Tong-Tao

Rfak Rfak

ARC13829-16 cytoplasm Gambiaca cytoplasm

IR54755 Chao Yang 1 etc.

IR42 etc. IR58 etc.

cms-sp

MS5 77A cytoplasm

IR42 etc.

cms-UR89

UR89F cytoplasm

Taichung 65

RfI-b Rfl-b

ems-URI02

UR102F cytoplasm

Taichung 65

Rfl-c Rfl-C

cms-UR104

UR104F cytoplasm

Taichung 65

Rfl-d RfI-d

cms-URI06

UR106F cytoplasm

Taichung 65

Rfl-e RfI-e

cms-UR27 cms-54257 cms-Khiaboro cms-IR 66707A

UR27F cytoplasm 54257 cytoplasm Khiaboro cytoplasm Oryza perennis Acc 104823 cytoplasm

Taichung 65

Shinjyo 1969, 1975 Watanabe 1971 Kitamura 1962b, 1962c Katsuo & Mizushima 1958 Lin & Yuan 1980 Kadowaki et al. 1988 Lin & Yuan 1980 Yabuno 1977 Sakamoto et al. 1990 Virmani etal. 1989 Lin & Yuan 1980; Kadowaki et al. 1988; Virmani et al. 1989 Kadowaki et al. 1988; Virmani et al. 1989 Kadowaki et al. 1988; Shinjyo 1990 Kadowaki et al. 1988; Shinjyo 1990 Kadowaki et al. 1988; Shinjyo 1990 Kadowaki et al. 1988; Shinjyo 1990 Shinjyo 1990 Ling et al. 1989 Nagamine et al. 1995 Dalmacio et al. 1992, 1995

Akibare IR64

Rf3 Rf3, Rf4 Rf4

2. HYBRID RICE: GENETICS. BREEDING. AND SEED PRODUCTION

45

using molecular markers. Two chromosomal regions each containing a photoperiod-sensitive genic male sterility locus designated as pmsl (on chromosome 7) and pms2 (on chromosome 3) were detected. The effect of pmsl is 2-3 times larger than that of pms-2, and dominance is nearly complete at both loci, but pmsl is not the locus relevant to the fertility difference between 'Nang-Ken 58s' and 'Nong-Ken 58' (Wang et al. 1997a). A recessive PGMS gene in 'Nong-Ken 58s', designated as ms Ph , has been also reported to be linked with gh-l and st-2 on chromosome 5 (Qian et al. 1995). This was confirmed by Lin et al. (1996). Using primary trisomies analysis, it was found that one gene for the male sterility in 'Nong-Ken 58s' was linked with d-l on chromosome 5, with a recombination value near 28.41 (Zhang et al. 1990). In breeding practice, continuous distribution of photoperiod-sensitive male sterility in the progeny of the primary generations is indicative of the effect of modifying genes on photoperiod-sensitive male sterility (Mei et al. 1990; Xue and Deng 1991). TGMS genes of '5460s' and 'H89-1' were designated as tmsl and tms2, respectively (Sun et al. 1989; Maruyama et al. 1991a; Kinoshita 1992). The gene tms-l of '5460s' was identified in China. TGMS1.2, located about 6.7 eM from the TGMS gene tms-l, is on chromosome 8 (Wang et al. 1995a, 1996). However, another study indicated that two major recessive genes controlled the thermosensitive male sterility in '5460s' (Wan and Ma 1996). The male sterility of two other Chinese TGMS source materials, 'An-Nong s-1' and 'Heng-Nong s-1', was speculated to be controlled by two unmapped recessive genes (Zhou et al. 1991a; Jiang et al. 1993; Wu and Yin 1992; Wan and Ma 1997) or one recessive gene on chromosome 8 in 'An-Nong s-1' (B. Wang, pers. commun.). The pollen fertility and spikelet fertility of 'Norin PL12' or 'H89l' and 'IR32364TGMS' are controlled by a single recessive gene. Complementation tests revealed that these two genes were different. The TGMS gene in 'Norin PL12' has been designated as tms-2 and the TGMS gene in 'IR32364TGMS' has been designated as tms-3(t) (Maruyama et al. 1991a; Borkakati and Virmani 1996). RFLP analysis revealed that the tms-2 was located between R463A and R1440 on chromosome 7 (Yamaguchi et al. 1997). A recent study indicated that four RAPD markers were linked with tms3(t) (Subudhi et al. 1997). The study used bulked segregant analysis of the F 2 population between 'IR32364TGMS' and 'IR68'. The gene tms3(t) was mapped to the short arm of chromosome 6. The TGMS gene for the Indian TGMS line 'SA2' has been designated as tms4. It was reported that the 0.7-kb amplicon ofOPA 12 and 1.9-kb amplicon ofOPS 1 were specific to the TGMS trait of 'SA2' (Reddy et al. 1998b). Gene mapping for reverse TGMS lines is

46

J. LI AND L.

YUAN

under way in China (B. Wang, pers. commun.). Besides the major genets), TGMS lines could be controlled by some modifying genes since individual F 2 progeny from the same population shows different fertility levels (Reddy et al. 1998b). Allelic relationship analysis revealed that the genes controlling male sterility in 'Nong-Ken 58s' are allelic to those of its derivative lines '7001s', 'N5088s', and 'M105-9s' (Yang 1997). The genes controlling the male sterility of'Pei-Ai 64s', 'An-Nong s-1', 'Heng-Nong s-1', and '5460s' are non-allelic, but the genes of 'Xin-Guang s' and 'Pei-Ai 64s' are allelic (Luo et al. 1996). The allelic relationships among most current Chinese T(P)GMS lines are indicated in Fig. 2.2 (Sun and Cheng 1994). Genes Controlling the Male Sterility in NMS Lines. The male sterility of one rice mutant is found to be controlled by a single recessive gene, ms9, in linkage group 6 with two marker genes, Ur (undulate rachis) and Cl (Clustered spikelet), the order being Cl-ms-9-Ur (Sato and Shinjyo 1991). Suh et al. (1989) reported four NMS genes: ms-ir36(t) from IR36ms; ms-m67(tj from 'Milyang 67ms'; ms-m77(tj from 'Milyang 77ms', and ms-m55(tj from 'Milyang 55ms'. Later Suh et al. (1991) revealed that the gene ms-m67{t) was linked with lax (lax panicle), eg (extra glume), d-l0 (dwarf-l 0) and A (anthocyanin activator) in linkage group III, with recombination values of 0,13.7,23.6 and 34.0%, respectively. The map position of ms-m67(t) is eg-ms-m67{tJ-d-l0-A, and it is completely linked with lax. The NMS gene of ms-m77{t) is linked to mp1 (multiple pistil-l), with a recombination frequency of 14.9%. For male sterile rice mutants derived by chemical or irradiation induction, different recessive genes are involved (Ko and Yamagata 1987; Fujimaki and Hiraiwa 1986). The genes responsible for nuclear male sterility, including environment-conditioned nuclear male sterility, are listed in Table 2.7.

IV. BREEDING FOR THREE-LINE SYSTEM HYBRID RICE

The genetic basis of the three-line hybrid rice breeding system are a male sterile line (a CMS or A line), a maintainer line (B line), and a restorer line (R line).

A. Breeding Procedure The procedure for the three-line system hybrid rice breeding can be divided into two phases, parental line development and heterosis

8912s

311ilS~

7001s

~

M105s

WD1s

~

/ W6154s Er-Yi-l05s

...

japonica-type

allelic; Fig. 2.2.

*'"

'1

/

/

~---....

non-allelic;

...

I

Pei-Ai64s

~I ~-Nongs-D

indica-type

c:::::::::::> indica source T(P)GMS

o

The allelic relationship among T(P)GMS lines developed in China. Source: Sun and Cheng 1994.

...

japonica source T(P)GMS

..I::O:l

Table 2.7.

Genetic male sterility in rice. Source: Kinoshita 1997.

Gene msl {stl

Name

Derivation

Chromosome

Fukukame

spontaneous

6

sterile-2 sterile-3 sterile-4 sterile-5 sterile-6

md-strain Bufumochi Fujiminori Otori Bufumochi

spontaneous spontaneous spontaneous spontaneous spontaneous

sterile-7 sterile-8 sterile-9a sterile-9b sterile-l0 sterile-ll sterile-12 sterile-13 sterile-14 sterile-15 sterile-16 sterile-17

K1:Koshihikari K2 :Koshihikari RTI-3a T65 E-l :Etsunan 77 T-1:Toyonishiki T-2:Toyonishiki T-3:Toyonishiki S-32:Sasanishiki S-40:Sasanishiki S-55 :Sasanishiki S-59:Sasanishiki S-81 :Sasanishiki

EI EI spontaneous EI EI EI EI EI y-ray EI

male sterile-l

Source

GROUP A (ms2-ms6) ms2 ms3 ms4 ms5 ms6

male male male male male

male male male male male male male male male male male male

male male male male

sterile-IS sterile-19 sterile-20a sterile-20b

MSZ 7:Nihonmasari MSZ8:Nihonmasari MS15:Nihonmasari MSZ9:Nihonmasari

3 7 6

Ko & Yamagata 1987, 1989 Sato & Shinjyo 1991

9

5

EI

y-ray

GROUP C (ms18-ms45) ms18 ms19 ms20 ms20

Hara 1946 Shibuya 1973

GROUP B (ms7-ms17) ms7 msB ms9 ms9 mslO msll ms12 ms13 ms14 ms15 ms16 ms17

Reference

EI y-ray EI y-ray

2 Fujimaki et al. 1977; Haraiwa & Tanaka 1980; Fujimaki & Haraiwa 1986; Tamaru 1994

ms21-ms23 ms24 ms24 ms25-ms30 ms31-ms45

male sterile 21-23 male sterile-24a male sterile-24b male sterile 25-30 male sterile 31-45

MS:Nihonmasari MS4:Nihonmasari MS9:Nihonmasari MS5-10:Nihonmasari MS 11-25:Nihonmasari

y-ray y-ray y-ray y-ray E1

M201 M101 M201 Calady Earirose M201 M1D1 M1D1 M2D1 M2D1 M1D1 M2D1 M101 M201 M1D1 M2D1 Calrose 76 M201 M1D1 Caloro

streptomycin y-ray EMS spontaneous spontaneous EMS y-ray y-ray EMS EMS y-ray EMS y-ray EMS y-ray EMS anther culture E1 y-ray spontaneous

GROUP D (ms46-ms63) ms46 ms47, ms48 ms49 ms50 ms50 ms51 ms52 ms53 ms53 ms54-ms56 ms57 ms58 ms59 ms59 ms60 ms60 ms60 ms61 ms62 ms63

male sterile-46 male sterile-4 7,48 male sterile-49 male sterile-50a male sterile-50b male sterile-51 male sterile-52 male sterile-53a male sterile-53b male sterile 54-55 male sterile-57 male sterile-58 male sterile-59a male sterile-59b male sterile-50a male sterile-6Db male sterile-50 male sterile-51 male sterile-52 male sterile-53

Trees & Rutger 1978; Mese et al. 1984; Hu & Rutger 1992

GROUP E: (msIR36-msm77(t))

~

CD

msIR36 msIR36 msIR36

ms-1R36/1R3 6 5495ms/Line5495 5683ms/Line5683

E1 spontaneous spontaneous

Singh & 1kehashi 1981; Suh et al. 1989, 1991

CJ1 0

Table 2.7.

(Continued)

Gene msIR36 msm55(t) msm67(t) msm77(t)

Source

Name Milyang 54ms/ Milyang 54 Milyang 55ms/ Milyang 55 Milyang 67ms/ Milyang 67 Milyang 77ms/ Milyang 77

Derivation

Reference

spontaneous spontaneous spontaneous spontaneous

GROUP F (ms1(t)-ms2(t)

Kaul 1986 Takeda 1987

tmsl

Co40 x Vaigai F 1 Co40 x Vaigai F 1 open hull male sterile 5460:IR54

spontaneous

8

tms2 tms3(t)

Reimei IR32364

y-ray y-ray

6

msl(t) ms2(t) oms

Chromosome

anther culture anther culture spontaneous

Sun et al. 1989; Yang et al. 1992; Wang et al. 1995a,1996 Maruyama et al. 1991a Borkakati & Virmani 1993; Subudhi et al. 1995, 1996,1997

GROUP G (pms1, pms2) pmsl

Nang-Ken 58s

spontaneous

7

pms2

Nang-Ken 58s

spontaneous

3

pms(t)

M201

EMS

Zhu & Yu 1989; Wang et al. 1991; Shao & Tang 1992, 1995; Zhang et al. 1993a Oard et al. 1991; Oard & Hu 1995

51

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

evaluation (Fig. 2.3). This breeding procedure is flexible, and some routine steps may be omitted or repeated, depending on the performance of parental lines or of the hybrids (Yuan 1985). 1. Development of Parental Lines

Source Nursery. This nursery is for plant source materials to be used in breeding the three parental lines. All germplasm except A, B, and R lines should be grown each with 10-20 plants and one seedling per hill or pot. The A and B lines should be planted in isolated plots. To attain better synchronization of flowering time, the A lines should be seeded at regular intervals for testcrossing. Testcross Nursery. This nursery is to determine the fertility of the F1 hybrids and to screen for Rand B lines. Ten to 20 plants are usually Source Nursery Male sterilc plants Matcmalmaterials

x

t

Paternal materials

x

MSline

Restorer line

~

Combining Ability Evaluation Nursery

,..-----=;..--~ I

Replicated Trial

Regional Trial

~ Fig. 2.3.

Release to farmers

/

Farmer's Field Evaluation

The breeding procedures for hybrid rice. Source: Yuan 1985.

52

J.

LI AND 1. YUAN

grown in one row for each testcross with the check cultivar between every 10-20 F1 hybrids. If the F 1 is found to be male sterile, it can be used for the development of an A line by successive backcrossing. When the F 1 shows both normal fertility and good performance overall for desired traits, the male parent can be used to develop R lines through the re-testcrossing. Crosses for which the male parents have both poor restoring ability and poor maintaining ability are generally discarded.

Re-testcross Nursery. This nursery is to confirm the restoring ability and to evaluate heterosis. More than 100 plants are grown and compared with leading commercial hybrids or cultivars. Backcross Nursery. The objectives in this nursery are to develop A and B lines. The male sterile plants should be grown with the recurrent

male plants in pairs. The goal is to achieve stable male sterility and uniform agronomic traits in a population of about 1,000 plants. When this is achieved, the male sterile line and its corresponding male parent (B line) are ready for commercial evaluation. 2. Heterosis Evaluation

Combining Ability Evaluation Nursery. Already developed A (or R) lines are tested with existing R (or A) lines. Each combination is planted with a single seedling per hill, about 500 plants per plot with replicates and the standard commercial cultivar or leading rice hybrid as the check. Replicated Trials. The promising selected hybrids are evaluated in replicated trials. The best-performing for agronomic traits will be selected for the regional trial. Three or four replications with a check cultivar or hybrid and 20 m 2 of plot size in each replication are needed. Usually one or two years is used for the replicated trial. Regional Trials. This trial is to evaluate the adaptability of rice hybrids to different regions and their yield potential. Evaluation in farmers' fields and study of seed production techniques may be conducted simultaneously. B. Development of A and B Lines 1. Breeding Objectives. The development of A and B lines is supremely

important in the three-line system. In addition to yield and product quality of the F 1 hybrid, these lines affect seed yield of the hybrid. Elite

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

53

A lines require: (1) stable performance of male sterility: with no risk of recovering the self fertility after generations of backcrossing or due to different ecological conditions; (2) easy restorability for male fertility: with ease of identifying or developing R lines for which the (A x R) F 1 has normal seed set with minimal influence by environmental conditions; (3) high outcrossing potential: with little enclosure of the panicle within the flag leaf sheath, daily blooming time essentially synchronized with the normal Band R lines, a long floret opening duration, a large floret opening angle, a high stigma exsertion proportion, short and narrow flag leaf, and thus a high yield potential for F 1 seed production and for A line multiplication; (4) good grain quality and resistance to diseases and insect pests; and (5) good combining ability. The B line is a nuclear "twin" of the A line but with a different cytoplasm. Characters such as superior combining ability, outcrossing characteristics, and grain quality are also the breeding objectives for developing B lines (Yuan and Chen 1988). 2. Observation of Male Sterility in Rice. Male sterility can be the result of pollen sterility, indehiscent anthers, anther abortion, pistilloidy of the anthers, or other causes (Rutger and Shinjyo 1980). Constant observation of the expressed level of male sterility is obligatory for both hybrid rice breeding and for hybrid seed production. Male sterility is observable only after heading using the following three methods (Yuan 1985):

Visual Inspection. For sporophytic male sterile lines, an obvious panicle enclosure in the flag leaf sheath is a good indication of male sterility. During anthesis or shortly thereafter, color and plumpness of anthers can be visually observed to determine completeness of the male sterility. Brightly yellow and plump anthers indicate male fertility. The flowering panicles can be shaken to check for anther dehiscence. If pollen grains come from the anthers, the male sterility is incomplete. Bagging Panicles. The most accurate estimation of the male sterility is the seed set two-plus weeks after bagging the panicles with glassine paper bags, after heading starts but prior to any anther dehiscence. In some cases the male sterility will be overestimated, because increased temperature in the glassine bag under the hot sun may increase male sterility. Microscopic Observations. Anthers should be sampled from different parts of the rice panicle, smashed into fine pieces in I-KI solution using tweezers, and observed for male sterility under a light microscope. The

54

J. LI AND L. YUAN

fertile (well-stained) pollen should not exceed 1 % of the pollen grains for sporophytic male sterile lines. 3. Sources of A Lines. There are two main ways of creating male sterile rice plants: (1) mutation (chemical or irradiation) or natural outcrossing that maintains nuclear male sterility, and (2) distant hybridization. In the case of natural outcrossing, it is generally difficult to find a maintainer line. The second method is distant hybridization. From more than 660 CMS lines, there are about 64 cytoplasmic sources, among which 22 sterile cytoplasms were from wild rice, 38 from indica, one from O. glaberrima, and three from japonica (Sheng 1994). Upon identification of a male sterile plant, confirmation of its genetic basis is required. If, for example, this male sterile plant is testcrossed with a fertile plant of the same or a different cultivar, the Fls are fertile, and the Fzs have a segregation ratio of 3 fertile to 1 sterile, then the male sterility of this male sterile plant is controlled by a single recessive gene. 4. Breeding Approaches for A and B Lines

Nuclear Substitution. Nuclear substitution has proven effective in developing a stable male sterile line from an ancestral male sterile plant even though it is difficult to identify a maintainer in some cases. For most wild-cultivated, indica-japonica, indica-indica, and japonica-japonica crosses in China, the source male parental lines are backcrossed recurrently for several generations until segregation of male sterility and other agronomic traits is no longer observed. Male sterility is difficult to stabilize for some of the original male parental lines. When this occurs, a different rice cultivar should be selected as the recurrent parental line. Interspecific Crossing. The origin of O. glaberrima seems to be more primitive than that of O. sativa. A stable male sterile line may be developed by crossing and backcrossing with O. glaberrima as the female and O. sativa as the male. On the contrary, for O. sativa cytoplasm combined with an O. glaberrima nucleus, the substitution line does not set seed even though the pollen grains stain normally with I-KI solution (Sano 1985). Erickson (1969) obtained 100% male sterility in the F I , BC I and BCz by means of nuclear substitution between three rice cultivars from O. glaberrima and three japonica cultivars. Similarly, 0.25% selfing occurred in BC l of the backcross between'Sakotira', a rice cultivar from West Africa, and an indica cultivar AC 5636 (Swaminathan et al. 1972). It seems difficult to develop a restorer for most A lines developed from interspecific crosses.

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

55

Crossing Between Wild and Cultivated Rice. It is easiest to develop male sterile lines from wild-cultivated crosses. China's first and most popular WA-type A lines, 'Er-Jiu-Nan 1 A' (Fig. 2.4), 'V20A', and 'ZhenShan 97A', were all developed from backcrosses with a wild abortive plant at Ya-Xian on Hainan Island. This made possible the breakthrough in the commercialization of hybrid rice technology in China. In addition, Hong-Lian (HL) type A lines were developed from backcrosses with Hainan red-awn wild rice as the female parent. 'IR66707A', which had the cytoplasm of O. perennis 'ACC 104823' and the nuclear background of 'IR64', was found to be stable with almost complete (93-100%) male sterility in crosses with 10 restorers having the WA cytoplasm (Dalmacio et al. 1995). Primitive types such as wild rice are normally employed as the female to develop male sterile lines (Yang and Lu 1989). Crossing Between indica and japonica Rice. Many A lines were developed from indica-japonica crosses. Based on their cytoplasmic sources,

male wild abortive plant /

F

female X X

F/X / /

DCIFI

DC2Ft

12/1971

Er-Jiu-Nan I

06/1972

.® ~®

Er-Jiu-Nan I

X

Er-Jiu-Nan I

Er-Jiu-Nan 1A

03/1971

Er-Jiu-Nan t

X

BC3/X BC4/X Fig. 2.4.

6044

MonthlYear

~®

*® *®

10/1972

02/1973

Er-Jiu-Nan 1

06/1973

Er-Jiu-Nan I

09/1973

Er-Jiu-Nan 1B

The breeding procedure of Er-Jiu-Nan lA. Source: Lin and Yuan 1980.

56

J. LI AND L. YUAN

these A lines can be classified into five types: (1) Bora type. Included are 'Chinsurah Bora II' x 'Taichung 65' (Shinjyo and Omura 1966), 'Lead Rice' x 'Taichung 65' (Watanabe et al. 1968) and 'Chun 190' x 'HongMao-Ying' (Li 1980); (2) Yunnan upland indica type. These include 'ErShan-Da-Bai-Gu' x 'Hong-Mao-Ying', 'Er-Shan-Da-Bai-Gu' x 'Ke-Qing 3' (Li 1980b); (3) Southeast Asian indica type. These include 'IR24' x 'XiuLing', 'Tetep' x 'Norin 8' (Yuan and Chen 1988); (4) Chinese indica landrace type. This type includes 'Tian-Ji-du' x 'Fujisawa 5', 'Lian-TangZao' x 'Li-Ming'; and (5) Late indica or indica waxy rice type. It is represented by 'Jing-Quan-Nuo' x 'Nan-Tai-Jing'. Most japonica A lines were developed from indica-japonica crosses. The basic principles are summarized as follows: It seems easier to develop A lines from indica-japonica crosses with indica rice as female and japonica type as male in the backcross breeding. It is better to select the more primitive type of indica rice as the cytoplasmic source and the more advanced japonica cultivar as the nuclear background. It is more effective in A line breeding to test the performance of male sterility in the reciprocal crosses. If there is much difference in male sterility of the reciprocal crosses, the probability of successful development of an A line will be increased because the male sterility in some indicajaponica crosses may originate from the lack of coordination between the nucleus of the two parental lines, and the fertility will be restored along with the nuclear substitution by several generations of backcrossing. The pollen abortion type of the male sterile plants of indica-japonica crosses belongs to the trinucleate abortion type. This differs from sterility of the WA-type A lines. This type may have some portion of the pollen stained with I-KI solution, so a more dependable procedure is to determine the selfing seed set by bagging panicles. Crosses Between Geographically Distantly Related Rice Cultivars or Cultivars of Different Ecotypes. In Yunnan province of China different japonica ecotypes are grown at different elevations. The Yunnan Agricultural University developed the Dian 4-type and Dian 6-type A lines from the cross between two japonica cultivars: 'Zhao-Tong-Bei-Zi-Gu' from high elevation and 'Ke-Qing 3' from low elevation. Scientists at the Hunan Academy of Agricultural Sciences have determined the frequency of male sterility for different types of crosses. Male sterility is 100% in wild-japonica crosses, 85% in wild-indica crosses, and 4.0% in indica-japonica crosses. The order of frequency of producing male sterile plant(s) is: wild-cultivated> indica-japonica > geographically distant hybridization> mutation induced by physical or chemical factors> intercultivar crosses. Breeding efficiency can be improved for the development of indica A line through wild-cultivated

2. HYBRID RICE: GENETICS. BREEDING. AND SEED PRODUCTION

57

crosses and japonica A line through indica-japonica crosses. Experience in China indicates that a closer relation between the cytoplasmic and nuclear donor parents produces gametophytic male sterility, making it harder to obtain a stable CMS line. The more distantly related the cytoplasm and substituted nucleus, the more likely the male sterile lines and their maintainer lines will be obtained (Zhang 1985).

Male Sterility by Mutation. A few natural mutants for male sterility in rice were identified by intensive screening, as exemplified by the male sterile rice material "424" (Yuan and Chen 1988). Male sterile mutants were induced by irradiation (X ray, yray, neutron, and laser) or chemical mutagens such as EMS, DES, and NED. The Mei-Xian Agricultural Research Institute of Guangdong Province in China identified 30 male sterile plants from 5908 F 3 segregants from 'Zhen-Zhu-Ai 11', 'GuangNong 1', and 'Guang-Xuan 3', each of which had been exposed to 25,000-30,000 R CooD y ray. Natural and induced mutants for male sterility are generally thought to be nuclear mutations, making it difficult to develop a maintainer or restorer. Backcross Breeding of Male Sterility. Transfer of male sterility into new lines is needed to increase the diversity of the A and B lines. The restorer-maintainer relationship and pollen abortion of A lines by backcross breeding are basically the same as the source A lines. Stable A lines can be developed by backcrossing for four or more generations. The first step is to select a new rice cultivar with the desired trait(s) and then testcross it with the source A line and observe the F 1 to estimate the maintenance of male sterility. The second step is to backcross and select segregants with complete male sterility, good blooming characteristics, high stigma exsertion rate, and most of the agronomic traits of the male parent. Crossing and backcrossing with the male parent are continued, with selection for male sterility at each step. Transfer Breeding ofMaintainer Lines. Most rice cultivars with superior target trait(s) have minimal, if any, sterility-maintaining ability. Thus it is necessary to develop new maintainer lines through transfer breeding. For transfer breeding of maintainer lines (Fig. 2.5), the source maintainer line used as the female parent should be closely related to the target cultivar of the intended new maintainer line. If it possesses the same agronomic traits and similar pedigree, there will be less segregation, so fewer generations will be needed to stabilize maintaining ability. Agronomic traits similar to the target cultivar (i.e. the male parent) should be selected during the backcross with the plant containing the male sterile cytoplasm.

J. LI AND 1. YUAN

58

Crossing

x

B line (Nft)

New cultivar or line (SFFor NFF)

+

Backcrossing

X

+

F1 (NFt)

New cultivar or line

(SFF or NFF)

BC1F1

(NFF, NFt)

Selfing

.....,1,-<8>

r--

Testcrossing

Sft X NFF

i

I

t Sft (sterile)

• X

Nft

Backcrossing

N Ft

Sft X Nft

t

SFt (fertile) (segregation)

+

Sft (sterile)

SFF(NFF)

~

• X

SFF(NFF)

NFF, NFt

Selfing

...------*<8>

t

,

Testcrossing Sft X NFF

SFt (fertile)

t

Sft X NFt

t

I

Sft

t

(fertile)

l

JNff

SFt

(sterile)

(segregation)

S

,

Sft X NFt

SFt (fertile)

Backcrossing

--:J.

f

f

(sterile)

The maintainer line from the new cultivar or line

=sterile cytoplasm; N =fertile cytoplasm; f =sterile nuclear gene; F =fertile nuclear gene

Fig. 2.5. al. 1982.

The procedure of transfer breeding for a maintainer line in rice. Source: Li et

Other Methods. Protoplast fusion was used to transfer cytoplasmic male sterility from two CMS lines (MTC-5A and MTC-9A) into a fertile japoniea cultivar, 'Sasanishiki'. The CMS, which can be restored by 'MTClOR' with a single dominant gene Rf-l, was successfully expressed in the cybrid plants. The gene was stably transmitted to their progenies through at least eight generations (Akagi et al. 1995). China has developed many A and B lines, but only a few, such as 'ZhenShan 97A' and 'V20A', have been in large-scale commercial production for years (Table 2.8). These much-used A lines have stable sterility, easy restoration, and good combining ability, but poor grain quality or poor stress resistance (Zhou 1994). Outside China only a few eMS lines such as 'IR58025A', 'IR62829A', 'IR64608A', 'PMSIA', 'PMSBA', and 'PMSIOA', have demonstrated commercial potential on a large scale (Virmani 1994b).

Table 2.8.

::.n

c.o

The characteristics of several major indica eMS lines and their area of production. Source: Zhou 1994. Accumulated area (mil. hal (1988-92)

% total area under hybrid rice

Nan-You 2 Nan-You 6

0.0 (Large area in 1970s)

0.0

Good combining ability. stable sterility. poor grain quality and moderate resistance

Shan-You 63

43.6

65.9

V20A

Good combining ability. stable sterility. poor grain quality and moderate resistance

V64 V6

12.0

18.1

TeA

Good outcrossing rate and combining ability, incomplete sterility

Te-You 63

0.2

0.3

Bo-Bai A (or Eo A)

Very high outcrossing rate, good grain quality, but incomplete sterility

Eo-You 64

2.20

3.3

DA-type

Xie-Qing-Zao A

Good grain quality, but poor restoring ability, incomplete sterility. Fjs senstive to low temperature

Xie-You 63 Xie-You 64

1.6

2.4

D-type

D Shan A

Good combining ability and restorability, but poor grain quality and incomplete sterility

D-Shan-You 63

5.1

7.6

ID-type

II-32 A

Good combining ability. high outcrossing rate, incomplete sterility

II-You 63 II-You 64

0.2

0.4

You-l A (or U-l A)

Good combining ability. high outcrossing rate. incomplete sterility. Fjs sensitive to low temperature

You I 63 You I 64

0.1

0.1

Type of cytoplasm

eMS line

WA-type

Er-Jiu-Nan lA

Good combining ability. stable sterility, poor grain quality and resistance

Zhen-Shan 97A

Major characteristics

Representative hybrid(s)

60

J.

LI AND L. YUAN

5. Main Features ofWA-type A Lines. The WA-type A lines are currently employed over the largest area of three-line hybrid rice production in China. This type is the most stable and hence has been introduced to several other countries, such as Vietnam. All WA-type A lines have been developed from the wild abortive (WA) source rice plant. Discovered in 1970, it has very strong tillering, slender culm, narrow leaf and sheath, long purple awns, black seeds, and a long seed dormancy duration. Its anthers are slender and slightly yellow. They usually do not dehisce, but some dehiscence and seed set are observed when the average temperature is over 30°C for several continuous days. This source WA rice plant was a heterozygote, i.e., there was segregation in the F 1 when it was crossed with indica or japonica cultivars. It is quite probable that this wild abortive rice originated from natural crossing between Hainan redawned wild rice (as the female) and the local late cultivated rice cultivar (as the male). The Hunan Rice Heterosis Utilization Research Cooperative Group in China reported that of 731 rice cultivars, 624 showed good maintaining ability and 18 had partial maintaining ability to the wild abortive rice. All 345 japonica cultivars had maintaining ability (Li et al. 1982). Hundreds ofWA-type A lines (such as 'V20A', 'Zhen-Shan 97A', and 'V41A') were developed by crossing to rice cultivars with good maintaining ability known from testcross nursery, and backcrossing to selected progeny that expressed complete male sterility, normal blooming traits, and agronomic traits similar to the male parent. In general, four to five backcross generations are needed to develop a new WA-type indica A line. Male sterility in the crosses of (wild abortive rice x japonica) can be stabilized much faster than for the crosses of (wild abortive rice x indica). The percentage of male sterile plants in the backcrossing F 1 can reach 100% for the crosses of (wild abortive rice x japonica).

Male Sterility. The arrow-shaped and slender anthers of the WA-type A lines have a milky white color. Most pollen grains abort at the uninucleate stage and some at meiosis and the binucleate stage, so the pollen grains have irregular shape and cannot be stained with I-KI. The male sterility of the WA-type A lines is stable with little influence from temperature and other environmental conditions. Heterosis Utilization. WA-type A lines are mostly developed from semidwarf Chinese rice cultivars, so their complementation with other ecotypes and geographically distant genotypes results in strong heterosis for growth vigor, yield potential, grain quality, and resistance to adverse conditions. However, the WA-type A lines need further improvement for

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

61

resistance to diseases and insect pests, as most WA-type A lines have weak or no resistance to diseases such as rice blast and leaf sheath blight. For hybrid rice, the effect on disease resistance from the female parent is normally larger than from the male parent, indicating some cytoplasmic effect. Therefore, the introduction of various resistance genes to the WA-type A and B lines seems important. In addition, there are striking differences in the restorability of different WA-type A lines. The male sterility of 'Zhen-Shan 97 A' is most easily restored. There is a tendency of the WA-type A lines with easier restoration of male sterility to have wider adaptability to variations in environmental conditions such as high or low temperatures. C. Development of R Lines 1. Breeding Objectives

Strong Restoring Ability. Good R lines should have the ability to restore normal pollen and seed set in the Fl' The F 1 hybrid should have good adaptation to low or high temperature, good pollen shedding, and high seed set (>80%) even under adverse environmental conditions. Good Combining Ability and Other Traits. In addition to good general combining ability and specific combining ability, a reliable restorer line should have resistance to diseases, adaptability, and high grain quality. Good Outcrossing Characteristics. For superior yield of hybrid rice seed production, a restorer line should have a large quantity of pollen, long blooming duration, good pollen shedding, strong tillering capacity, slightly longer growth duration, and be taller than A lines. 2. Source and Distribution of Restorer Gene(s). The Ping-Xiang Agricultural Research Institute of Jiangxi Province in China found all 16 wild rice species (0. rufipogon) collected from the Hainan Island completely restored the male fertility of the wild abortive rice. Similarly, 'Boro II' restored the male fertility of the BT-type A lines. This indicates that the nuclear genome of the rice germplasm that provides the male sterile cytoplasm is an important source of the restorer gene(s). Restorer gene(s) can be screened by testcross. Cultivars closely related to the germplasm that provides the male sterile cytoplasm may have restorer gene(s). Most restorer lines of the WA-type A lines, for example, are cultivars from IRRI-bred materials or with origin close to wild rice, or late-season indica cultivars found at low latitudes. Mutation

J.

62

LI AND L. YUAN

induction by chemicals or by irradiation may create new restorer line with lower plant height, earlier maturity, or stronger restoring ability. The Zhejiang Academy of Agricultural Sciences in China screened 1,500 rice cultivars for restorer lines during 1978-1982. They found 55.5% of rice cultivars from South Asian countries at low latitudes had restoring ability for the WA-type A lines, and 20.5% were indica cultivars from Southern China and Korea. All early-season indica rice cultivars from China's Yangtze Valley and all japonica cUltivars from Northern China, Japan, and Korea lacked restoring ability for the WAtype A lines. The current popular restorer lines for the WA-type A lines, such as 'IR24' and 'IR26', have 'Peta', which has strong restoring ability, in their pedigree. New R lines can be developed through transfer breeding, examples being 'Milyang 46' from the cross [(Tongil x IR24) x (IR1317 x IR24)] and 'IRlll0-78' from the cross (Peta x Taichung Native 1). The evolution sequence was considered to be from wild rice to late indica, to early indica, to late japonica, and to early japonica. It appears that the closer rice cultivars are to wild rice, the more likely they are to have the restorer gene(s) for the male sterile WA-type A lines. 3. Breeding Methods. China has released hundreds of R lines since the

1970s, but only about nine are currently used in large-scale commercial production as shown in Table 2.9 (Xie et al. 1996). R line breeding is essential to further enhance yield levels of hybrid rice. Table 2.9. al. 1996.

China's main commercialized indica restorer lines in 1994. Source: Xie et

R line

Number of hybrid combinations

Area (1,000 hal

% of total area under hybrid rice

Ming-Hui 63

8

6,130

47.7

Ce64

4

1,757

13.7

Milyang 46

4

904

7.0

Gui 99

2

490

3.8

Ming-Hui 77

3

477

3.7

Duo-Xi 1

1

311

2.4

Ce 49

4

216

1.7

CDR22

1

215

1.7

903 Total

1

205

1.6

28

10,705

82.0

63

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

Testcross and Selection for R Lines. In preliminary testcrossing nurseries, at least 10 F 1 plants are required for the initial evaluation of seed set, plant type, other agronomic traits, and resistance to adverse conditions. In the re-testcross nurseries, at least 50-100 plants should be grown of each re-testcross F 1 to evaluate the yield and other agronomic traits. Only a few rice hybrids are selected from the re-testcrossing nursery. These hybrids should be grown on production nurseries to evaluate the productivity of the hybrid and consumer acceptance. Cross Breeding for R Lines. Testcrossing is insufficient in R line breeding for improving resistance, early maturity, and acceptable grain quality. Cross breeding is required. In the single cross breeding method, the restorer gene(s) are transferred to a new rice cultivar through single crossing. The segregating progeny that have restoring ability are selected for the improved traits using the pedigree method. The single crossing can consist of the following combinations: RxR: The frequency of restoring gene(s) will be high, so only the agronomic traits needing improvement require selection. The restoring ability will need to be tested only in the later generations. The new R lines will have the complementary restoring gene(s) and improved traits. BxR or RxB: Because the targeted traits segregate with and without the restoring ability, the selection of improved traits in combination with restoring ability is a tedious effort. The smallest number of plants needed for testcrossing can be determined using the following formula, m ~ log a/log P where m = the smallest number of plants for testcrossing, P = the probability for restorer gene(s), a = the probability for losing the restorer gene(s). This formula was proposed by Wang (1983b; Table 2.10). AxR: The iso-cytoplasmic R lines developed with this method can easily coordinate nuclear and cytoplasmic contents. The genetic diversity will be decreased by selecting new R lines from the cross between the Table 2.10. The smallest number of plants for testcrossing in generations with probability of 95% and 99% for including the restoring gene(s). Source: Wang 1983b. Number of plants for testcrossing Number of F2 restoring gene(s) 95% 99%

F:1 99%

F4

95%

99%

F(j

F" 95%

99%

95%

99%

95%

1

16

11

10

7

8

6

8

5

6

5

2

71

47

29

20

22

15

19

13

17

12

3

286

191

85

56

53

35

43

28

38

25

64

J.

LI AND L. YUAN

A and R line, so the heterosis of rice hybrids using the new R lines may decrease. This lowered heterosis can be overcome by increasing the genetic diversity between the R and A lines. Multiple cross breeding methods can combine advantages, such as good resistance, good grain quality, and early maturity, from each parent into a new restorer line. The early maturing R line 'Z6-Zai-Zao' was developed from the cross [(IRZ6 x Zai-Ye-Qing 8) x Zao-Hui 1]. The backcross method for developing new R lines with the Rf-l restorer gene was also demonstrated by Fujii et al. (1991).

4. Identification of Fertility Restorer Genes. Fertility restoration for the WA-type A lines was reported to be controlled by a single dominant gene (Shinjyo 1969; Wang 1980), but later researchers found that two R genes controlled restoration (Gao 1981; Zhou et al. 1983; Yang and Lu 1984; Sohu and Phu11995; Shen et al. 1996a; Ganesan and Rangaswamy 1997; Kumari et al. 1998). There are two pairs in 'IRZ4'; R1 derives from 'Cina', a Chinese late indica, and R2 derives from 'SL017' (Li and Yuan 1985). For five restorer lines ('IRZ4', 'IRZ9Z73', 'IR5474Z', 'IR9761' and 'ACR11353') there are four restoring gene loci (Ramalingam et al. 1995). There are reports of the restoration of male sterility governed by three or four genes (Huang et al. 1987) or multiple genes (Pei 1980; Fu and Wang 1988; Yang and Chen 1990). A recessive restoration gene "r" was also identified (Wang 1983a; Lei 1983). In contrast, the inheritance of fertility restoration in BT, Dian 1, and Hong-Lian types is gametophytic. The genotypes of the BT system are S(rr) for BT-C (MS line), N(rr) for maintainer, S(RR) for restorer BT-A, and N(RR) for restorer BT -x (Shinjyo 1984). The fertility of Dian 1 type 'Hua-Jing 14A', 'Tu-Dao 4A', and BT type 'Nong-Jin ZA' is controlled by a recessive gene (rr) and the male sterility genes in Dian 1 and BT types are allelic. The fertility of Hong-Lian type 'Hua-Ai 15A' was also reported to be controlled by major gene(s) and possibly also some mod- . Hying factors (Hu and Li 1985). Rf-l for BT-type eMS lines (cms-ba) is located on chromosome 10 and linked with pgl and fl (pgl-Rf-l-fl) (Shinjyo 1975). Yoshimura and coworkers located the Rf gene for cms-ba on chromosome 7 using the translocation method (Virmani 1996). The Rf-2 gene for cms-ld cytoplasm was recently located on chromosome 2, using primary trisomic and linkage tester lines (Shinjyo and Sato 1994). Virmani and Shinjyo (1988) reported that at the Rf-llocus there are at least fOUf multi-alleles: Rf-l a (Rf1), Rf-1 b , Rj-l c and Rf-1 d . Rf-l, Rf-2, and Rj-ak derive from 'Chinsurah Boro II', 'Fujisaka 5', and '0. glaberrima W0440' (Shinjyo 1969; Shinjyo

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

65

and Watanabe 1977; Yabuno 1977). Two independent dominant genes controlling fertility restoration were reported, with the stronger gene RfWA -1 and the weaker gene Rf-WA -2 being located on chromosome 7 and chromosome 10, respectively. 'IR26', 'IR36', 'IR53', and 'IR9761-19-1' all possessed the same restorer genes, whereas 'IR42' and 'IR2797-105-2-23' each had different restorer genes (Raj and Virmani 1988; Bharaj et al. 1991,1995). Rf-1 and Rf-2 are generally considered as independent genes, but Li and Zhu (1988) reported that the two restoration genes of 'IR24' and 'IR26' were linked with the recombination being 38.26% (F z) or 37.56% (BC l ). For Rf-3, six RAPD markers were found to be associated with this gene. Three markers, i.e. OPK05-800, OPUI0-I100 and OPW01350, were mapped on chromosome 1. Using RFLP technology, three markers (RG532, RG140, and RG458) were also found to be closely linked with Rf-3. At the RG532 locus, different alleles were found to restore the male sterility between the two eMS lines, 'Zhen-Shan 97A' and 'IR58025A' (Zhang et al. 1997a). Interval mapping technique showed eight QTLs to be associated with fertility restoration; the two major genes Rfi-3 and Rfi4 were located on chromosome 3 and 4, respectively, and were responsible for 49.6% and 35.4% of the phenotypic variation (Li et al. 1996b). A partial restoration gene Ifr(t) was also identified in 'T65T' (Sano et al. 1992; Teng and Shen 1994a). Restorer genes have been reported from time to time to be modified by other genes (Govinda and Siddiq 1984). For example, an inhibitor gene identified in 'IR17492A' modified the activity of a restoration gene (Govinda and Virmani 1988). The identified restorer genes are described in Table 2.11.

D. Development of Elite Hybrid Combinations The key for breeding of elite rice hybrids is the development and selection of the parental lines. The following principles for the selection of parental lines have proven to be helpful: 1. High Genetic Diversity. Within limits, higher genetic diversity will result in increased heterosis. The estimated genetic diversity can be based on the pedigree relationships, geographical sources, and ecotypes (Xu and Wang 1980; Singh et al. 1984; Subramanian and Rathinam 1984; Yuan and Chen 1988).

2. Complementary Traits. The current Chinese rice hybrids have complementation for agronomic traits, resistance to adverse conditions, and grain quality derived from each parent.

O'l O'l

Table 2.11. Gene

Rfl

Fertility restoring genes in rice. Source: Kinoshita 1997. Origin

Cytoplasm

cms-bo

Chromosome

Reference

Chinsurah bora II 10

(Rfl-a,b,c,d)

Shinjyo 1975, 1990; Sato et a1. 1985; Virmani & Shinjyo 1988; Fukuta et a1. 1992; Yu et a1. 1995; Akagi et a1. 1996; Yokozeki et a1. 1996; Ichikawa et a1. 1997 Watanabe 1971; Shinjyo & Sato 1994 for Rf3 and Rf4: Lu & Zhang 1986; Bharaj et a1. 1991,1995; Teng & Shen 1994b; Zhang et a1. 1996

Rf2

cms-Id

Fukuyama

2

Rf3 (R2, Rf2)

cms-WA

IR24 etc.

1

Rf4 (Rl, Rfl, RfWA1) RfWA2 Rf2,3,4,5 (QTL)

cms-WA

IR24 etc.

7

cms-WA cms-WA

IR36 DHfrom ZYQ/JX

10

Bharaj et a1. 1995

2,3,4,5

Zhu et a1. 1996

Rf5(t} Rf(t) Rfa,b,c Rfa',b',c',d' Rfak (Rfjp) Ifr

cms-WA

1

Shen et a1. 1993a, 1996a,b Maekawa 1982 Maekawa 1982 Yabuno 1977 Sana & Eiguchi 1991; Sana et a1. 1992

cms-bo cms-bo ems-ok cms-bo

Fert. Revertant from II32A (WA-type) H406 Hl03 Akebono partially sterile mutant from Taichung 65CMS

67

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

3. High Yielding Ability. Both parental lines should have elite agronomic traits, since in most cases the performance of F 1 hybrids was correlated with the mean value of the two parental lines for many traits, such as the number of spikelets per panicle, 1,000-grain weight, growth duration, plant height, and the number of productive panicles.

4. Good Combining Ability. A rice cultivar with excellent performance

itself may not be a good parental line in hybrid rice breeding. The important factor is the combining ability, including general combining ability (GCA) and specific combining ability (SCA). China has developed hundreds of A lines, but only a few, such as 'V20A', 'Zhen-Shan 97 A', and 'Xie-Qing-Zao A', are used in large-scale commercial production, owing to their superior combining ability, as shown in Tables 2.12 and 2.13 (Xie et al. 1996). E. Breeding for Rice Hybrids with Resistance to Insect Pests

and Diseases 1. Resistance to Diseases and Insect Pests. The Hunan Agricultural College evaluated the resistance to rice blast in 224 rice hybrids. They found four types of inheritance for rice blast resistance in the Fls: 102 hybrids showed dominant resistance; 31 had recessive resistance; 15 had intermediate resistance between the A line and the R line; and the rest had a different resistance type in the F 1 than in the parents. The resistance to bacterial blight performed similarly to the rice blast resistance in the Fl' Table 2.12. China's main commercialized indica male sterile lines in 1994. Source: Xie et al. 1996.

A line Zhen-Shan 97A V20A Xie-Qing-Zao A BoA Gang-type A D-type A II-32 A Long-Te-Pu A others Total

Number of hybrid combinations 12 8

7 5 3 2 2

2 5 46

Area (1,000 hal

% of total area under hybrid rice

7,221 1,307 838 724 722 399 331 298 256 12,096

56.2 10.2 6.5 5.6 5.6 3.1 2.6 2.3 2.0 94.1

J.

68

LI AND L. YUAN

Table 2.13.

China's main commercialized indica hybrid rice combinations in 1994. Source: Xie et aL 1996. Area (1,000 ha)

% of total

Shan-You 63

4,455

34.7

Shan-You 64

772

6.1

V64

515

4.0

Gang-You 12

471

3.7

Shan-You Gui 99

451

3.5

D-You 63

360

2.8

Bo-You 64

323

2.5

Shan-You 46

323

2.5

Shan-You Duo-Xi 1

311

2.4

II-You 63

297

2.3

V46

291

2.3

Te-You 63

258

2.0

V77

258

2.0

Xie-You 46

255

2.0

Gang-You 22

215

1.7

Xie-You 63

207

1.6

Bo-You 903

205

1.6

9,967

77.7

Rice hybrid

Total

area

It was proposed that the R line was important to breed resistance to brown planthopper. For example, the rice hybrid ('V20A' x 'IR26') and its R line, 'IR26', were resistant to brown planthopper, whereas 'Nan-You 2' (,Er-Jiu-Nan lA' x 'IR24') and 'IR24' had no resistance to brown planthopper (both 'Er-Jiu-Nan lA' and 'V20A' had no resistance to brown planthopper). But, it should be noted that there is a cumulative effect for resistance to brown planthopper and, hence, introduction of resistance gene(s) into male sterile lines can improve the resistance in hybrid rice to brown planthopper and other insect pests or diseases (Yuan and Chen 1988). 2. Breeding Techniques

Selection of Source Materials. Most IRRI-bred indica cultivars have strong resistance to diseases and insect pests, such as rice blast, bacterial blight, and planthopper. Some cultivars introduced from the Inter-

2. HYBRID RICE: GENETICS, BREEDING. AND SEED PRODUCTION

69

national Rice Research Institute, Sri Lanka, and India and some Chinese landraces also have good resistance to brown planthopper.

Testcross Method. If the F 1 between an A line and germplasm with good resistance also shows good maintaining ability of male sterility, successive backcrossing to this germplasm can be used to develop a highly resistant A and B line. 'Zhen-Shan 97A' and its B line are examples developed using this method. If the F 1 shows normal fertility, the tested germplasm is also a potential R line. For example, 'IR9761-19-1', with good resistance to rice blast and bacterial blight, was testcrossed with the WA-type A lines and found to have very good restoring ability and combining ability but was segregating. Some segregants were selected for paired testcrossing in two generations. Thus, the restorer 'Ce64-7' was bred with strong resistance to rice blast, bacterial blight, brown planthopper, and leaf hopper. Cross Breeding Method. Cross breeding can combine strong resistance with other important traits from different rice lines into a new R line. Single Cross Breeding. 'Ming-Hui 63', a restorer line highly resistant to rice blast, was developed from a single cross between 'Gui 630', a larger-grain type restorer, and 'IR30' which has multiple resistance. Multiple Cross Breeding. A multi-resistant and early-maturing R line, '26-Zhai-Zao', was developed from the multiple crossing between 'IR26' (with multiple resistance), 'Zhai-Ye-Qing 8' (highly resistant to rice blast), and 'Zao-Hui l' (good resistance and large panicle size). Recurrent Backcross Breeding. When R or B lines have good combining ability but little resistance to diseases and insect pests, these R or B lines can be used as recurrent parents in backcrosses with a donor to obtain new B or R lines with the advantages of both parents (Yuan and Chen 1988). Breeding for resistance to insect pests and diseases is more extensively covered by Virmani (1994a). F. Breeding for Rice Hybrids with High Grain Quality Since the 1980s, China has been attempting to develop rice hybrids with superior grain quality. For example, aromatic rice hybrids 'XiangYou 63', 'Xing-Xiang-You 77', and 'Xing-Xiang-You 80' were developed with good quality and high yield potential (Zhou and Liao 1995, 1997; Chen et al. 1997a). A survey of 500 households in China showed that for special occasions, such as entertaining guests and celebrating festivals, 25.4% preferred inbreds, 39.8% preferred hybrids, and 34.8% reported

70

J.

LI AND L. YUAN

having no preference. This result indicated that the cooking and eating quality of hybrid rice is acceptable to Chinese consumers (Virmani 1994a). Of 47 rice hybrids tested, most of the indica hybrid rice cultivars had high amylose content, hard gel consistency, intermediate gelatinization temperature and kernel elongation. Most japonica rice hybrids had low amylose content, soft gel consistency, and low gelatinization temperature (Tang 1987). The protein content of most of 30 rice hybrids was between the two parents, near to the mid-parent value and showed positive incomplete dominance. But, 20% of 30 hybrids were superior to their better parent (Liu, Sun, and Cai 1990). A hybrid 'L301A x R29' was developed in 1985 with first-grade grain quality (long grain, alkali digestion value of 2.0, 23% amylose content, 70% of milled rice yield, 57% of head rice yield, and 0-1 chalkiness) (Yuan and Virmani 1988). Other CMS lines such as 'Xing-Xiang A', 'Jin-23 A', and 'Di-Gu A' have been developed with excellent grain quality and good combining ability for grain quality traits (Zhou 1994). Even though the quality of hybrid rice is generally better than for early-season conventional cultivars and near to that of late-season or single-cropping conventional cultivars, most of the Chinese rice hybrids in the 1970s and early 1980s had poor grain quality. Improvement of grain quality is essential for further commercialization of hybrid rice in most developed countries. In Iran, development of hybrid rice technology has been hindered by segregation of the gel consistency and gelatinization temperature of hybrid rice (Dorosti 1997). Major grain quality traits in hybrid rice are (1) milling and head rice recovery; (2) size, shape, and appearance; and (3) cooking and eating characteristics (Khush et al. 1988). The rice grains of the F 1 rice hybrid are actually F 2 seeds. Therefore, both parental lines should have similar good-quality traits for the hybrid to have good quality. In breeding practice, the following principles should be observed (Yuan and Chen 1988): 1. Selecting A and R lines with Reduced Chalkiness and Elite Appear-

ance. The milling-quality traits are controlled by both seed genotype and maternal genotype. Hybrids with higher head rice recovery can be obtained if the parents are selected carefully. If either parent has a higher tendency for grain breakage, the F1 hybrids will normally give lower head rice recovery than the better parent. White centers and white bellies are controlled by a single recessive or dominant gene (United States Dept. Agr. 1963; Chalam and Venkateswarlu 1965; Nagai 1958) or by polygenes (Nakatat and Jackson 1973; Somoto and Hamamura 1973; Somrith et al. 1979). 'V20A' and 'Zhen-Shan 97A' are early-season indica in the Yangtze Valley of China. Both have a large percentage of undesirable chalky grains and chalky area. Most Chinese-bred R lines for japonica hybrid rice have

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

71

this chalkiness due to indica-japonica crosses in their pedigree. To improve the rice quality concerning grain chalkiness, both parental lines should be free from chalkiness (Khush et al. 1988; Yuan and Chen 1988). For example, the chalky area of the indica hybrid ('L301A' x 'IR29') is 6.5% because the chalky area of both parents is less than 100/0. Grain width and length/width ratio were affected not only by maternal additive x environment, but also by direct additive x environment effects (Shi et al. 1998; Chen et al. 1998). Grain length is governed by a single gene, or two or three genes, or polygenes (Ramiah et al. 1931; Bollich 1957; Ramiah and Parthasarathy 1933; Mitra 1962; Chang 1974; Somrith et al. 1979). Grain width and weight are controlled by polygenes (Ramiah and Parthasarathy 1933; Nakatat and Jackson 1973; Chang 1974; Lin 1978). Because the grain size and shape are also determined by the size of lemma and palea, which are governed by the genetic composition of the female parent, the female parent should have the desirable grain size and shape. The length and shape of F 1 grains generally are between those of the parents. Therefore, to develop medium-grain hybrids, parents having long and short grain may be used, but to produce long-grain hybrids, both parents must have long, slender grains. Parents with similar endosperm appearance should be selected to avoid segregation for physical appearance among the grains. 2. Selecting A and R Lines with Elite Cooking and Eating Quality. Shi et al. (1998) reported that endosperm additive and dominance effects accounted for 74.6% of total genetic effects for amylose content, followed by cytoplasmic and maternal effects. Amylose content is also reported to show both dosage and maternal effects, and monogenic inheritance (Kumar and Khush 1986, 1987, 1988; Kumar et al. 1987, 1994). Cooking and eating quality traits such as amylose content, tenderness, and cohesiveness of cooked rice for hybrids are between those of the parents (Bollich and Webb 1973; Ghosh and Govindaswamy 1972; McKenzie and Rutger 1983; Seetharaman 1959; Stansel 1966). In hybrid grains the heterogeneity for amylose content, gelatinization temperature, and gel consistency does not reduce cooking and eating qualities (Khush et al. 1988). In practice, if one parental line has high amylose content, the other parental line should have medium or low amylose content for an indica rice hybrid with an amylose content of about 220/0. Parents with intermediate amylose content and gelatinization temperature and low intrapopulation variation should be crossed to obtain hybrids that have a uniform texture and cooking time (Yuan and Chen 1988). The aromatic trait, whose principal component is 2-acetyl-l-pyroline, was reported to be controlled by a single recessive gene (Sood and

72

J. LI AND L. YUAN

Siddiq 1978; Bollich et al. 1992), two genes (Ramiah and Rao 1953; Zhou 1994), or two or three complementary genes (Tomar and Nanda 1983; Reddy and Sathyanarayanaiah 1981). For good post-cooking aroma and grain elongation, both parents must perform well for both of these traits (Bong and Singh 1993; Zhou 1994).

V. BREEDING FOR TWO-LINE SYSTEM HYBRID RICE

For simplification of the procedure for hybrid rice seed production, two-line system hybrid rice has been extensively studied to eventually replace the existing three-line system. The two-line system should include the T(P)GMS system and the chemical-emasculation system. In this section, the two-line system is referred to as the T(P)GMS system, unless otherwise indicated. The application of the chemical emasculation techniques will also be discussed in this section. A. Considerations 1. Advantages. There are four principal advantages to the two-line breeding.

Simplicity and Effectiveness. Since a maintainer line for the three-line system hybrid rice is not needed, multiplication of T(P)GMS lines is much easier and does not require synchronization of both parental lines in the multiplication plots as in the three-line system. In China, the average seed yield of T(P)GMS lines is 3 to 5 t/ha as compared to 2 t/ha for A line multiplication. Moreover, nuclear genes for male sterility of T(P)GMS lines are much easier to transfer than CMS genes, because they are unaffected by cytoplasmic genets). Removal of the Restriction of Restorer Genes. In the three-line breeding system, the F 1 hybrids between a male sterile line and most rice cultivars of the same subspecies show male sterility or partial fertility (seed set less than 30-50 0jo). Only a small percentage of rice cultivars are restorers of A lines. Therefore, the potential to develop superior rice hybrids by using newly-developed rice cultivars is limited in the threeline system. On the contrary, the male sterility ofT(P)GMS lines is controlled by recessive nuclear genets), so the Fls between a T(P)GMS line and cultivars of the same subspecies show normal fertility. The male

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

73

sterility of 'W6154s' and 'An-Nong s-l' can be restored to normal fertility by 97.6% and 99.3% of indica cultivars, respectively, and that of 'Nong-Ken 58s' can be restored by more than 96.6% of japonica cultivars (He and Yuan 1993). Thus the potential for successful development of elite hybrids is greatly increased.

Easier Utilization of Intersubspecific Heterosis in indica-japonica Crosses. It is easier to introduce wide compatibility gene(s) into T(P)GMS lines than into A lines. Therefore the combination of T(P)GMS gene(s) and WC (wide compatibility) genes will make intersubspecific heterosis breeding more feasible. Overcoming Negative Effects of the Male Sterile Cytoplasm. Yield potential may be fully tapped owing to the absence of negative effects from the male sterile cytoplasm of the three-line system (Wang and Tang 1990; Young and Virmani 1990). For example, a two-line system hybrid 'Liang-You-Pei-Te' (Pei'ai 64s x Teqing) yielded more than 7.5 t/ha (maximum yield 10.4 t/ha) in Hunan, China during the late season of 1991-94 and 9.0 t/ha in single crop plus an average yield of 2.3-3.0 t/ha of the ratooning crop. The highest yield recorded for this hybrid was 17.1 t/ha in Yongsheng County, Yunnan Province of China (Bai and Luo 1996). 2. Disadvantages. The dependency of male sterility on temperature requires much attention from breeders and agronomists concerning F 1 seed production and multiplication of the T(P)GMS lines. Consistent seed production requires that the climatic data of the seed production region should be analyzed in great detail. The scenario for the seed production should be set forth on the basis of local climatic conditions. More cautious and stricter evaluation is required for the commercialization of newly bred T(P)GMS lines. Once a T(P)GMS line is registered, its core seed production procedure should be followed for each generation to keep the CSP at the same level as when first registered in order to minimize the risk of unsuccessful F 1 seed production.

B. Development of T(P)GMS Lines

Six usable T(P)GMS lines have been released in China (Yuan 1997a, Table 2.14). The results of these releases provide a basic genetic tool for developing two-line rice hybrids, a further breakthrough for commercialization of hybrid rice.

J. LI AND L. YUAN

74

Table 2.14. Some commercially used rice P (T)GMS lines developed in China. Source: Yuan 1997a. CSpx (0C)

Sterility type

Hunan

23.5

HTy

Anhui

24.0

LDHP

Nongken 58s

Hubei

24.0

LDHT

Nongken 58s

1995

Hunan

24.0

HT

Annong s

indica

1994

Hunan

23.5

HT

Annong s

indica

1995

Guangdong

23.0

HT

Nongken 58s

P(T)GMS lines

Subspecies

Year identified

Pei'ai 64s

javanica

1991

7001s

japonica

1989

5088s

japonica

1992

810s

indica

Xiang 125s GD 2s

Development province

Gene source Nongken 58s

xCSP = critical sterility point; YHT = high temperature; ZLDHT = long day length, high temperature.

1. Discovery ofT(P)GMS Sources. Kaul (1988) reviewed the male steril-

ity conditioned by temperature, photoperiod, or other unknown environmental factors in his book Male Sterility in Higher Plants. He estimated that in about 44 % of research reports the major environmental factor influencing male sterility was temperature, in 12% it was photoperiod, and in the remaining 44 % it was unknown environmental factors. Rick (1948) reported that temperature affected the male sterility in tomato (Lycopersicon esculentum). Rundfeldt first reported the effect of photoperiod on the male sterility of cabbage (Brassica oleracea var. capitata), and also reported that the male sterility of three mutants, one each from cabbage, pepper, and tomato, was sensitive to both temperature and photoperiod. The cabbage mutant was male sterile in the summer and male fertile in winter, whereas the other two mutants from pepper and tomato were male fertile in summer and male sterile in winter (Rundfeldt 1960; Kaul1988). Male sterility of sorghum (Sorghum vulgare) was reported to be conditioned by photoperiod (Barabas 1962). Environment-conditioned male sterility has now been reported in pepper and tomato (Martin and Crawford 1951), cabbage (Rundfeldt 1960), sorghum (Barabas 1962; Tang et al. 1997), wheat (Jan 1974; Zhou et al. 1997), barley (Ahokas and Hockett 1977), sesame (Brar 1982), pea (Kaul 1988), rape (B. napus) (Xi et al. 1997), soybean (Wei et al. 1997), and rice (Shi 1981; Maruyama et al. 1991a; Oard et al. 1991). 'Nong-Ken 58s', the first rice source material for the development of T(P)GMS lines, was discovered in the Mian-Yang County of Hubei Province in China by Shi in 1973 (Shi 1981,1985; Shi and Deng 1986). Three male plants of this source material appeared to be physically iden-

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

75

tical to the male fertile plants, but were seven to ten days earlier during the initial heading stage. During 1980-1981 sequential plantings showed that photoperiod, not daily average temperature, from seeding to 15 days before initial heading was correlated with sterility performance. The correlation coefficient between daylength and male sterility was 0.91, indicating that daylength was the main factor inducing male sterility in 'Nong-Ken 58s'. 'Nong-Ken 58s' and the derived T(P)GMS lines were designated as "late japonica long-daylength-sensitive genic male sterile rice" in 1983, and as "Hubei Photoperiod-sensitive Genic Male-sterile Rice" (HPGMR) in 1985. In 1987 L.P. Yuan proposed that all T(P)GMS lines be affixed with "s" for simplification and for differentiation from the" A" line of the three-line system (2hu and Yang 1992). In the 1980s, additional TGMS materials were discovered in China, including '5460s', 'An-Nong s-1', and 'Heng-Nong s-1'. Two so-called reverse-TGMS materials, 'Dian-Xun -1' and 'IVA', were also reported for which a range of high temperatures can promote fertility (Jiang 1988; Peng et al. 1993; Jiang et al. 1997). Recently a new rice germplasm, 'YiDS', has been reported to show male sterility under short photoperiod and low temperature, which may be useful for double cropping of hybrid rice seed production in southern China (Wan et al. 1997). Outside of China, Maruyama et al. (1991a) developed a TGMS line, 'Norin PL12' or 'H89-1', from 'Reimei' by irradiation with 20 Kr of gamma rays. Another T(P)GMS line, 'X88', was also reported in Japan (Lu 1994). Oard et al. (1991) reported an environmentally influenced male sterile material from the M7 generation of M201 treated with EMS, with its conditional male sterility controlled by two nuclear genes with epistatic effects. Rutger et al. developed an environmental-conditioned genic male sterile material from a japonica cultivar, 'Calrose-76', using tissue culture. This material showed male sterility under daylength of 15 h, but male fertility under 12 h. These two T(P)GMS materials are still under study, because one produces too many selfing seeds under conditions that promote the sterility, and the other produces insufficient seed under short daylength (Mackill1995). Another mutant was recently identified as a putative photosensitive genic male sterile and is currently under study (Rutger 1997). The International Rice Research Institute developed the TGMS line 'IR32364-20-1-3-2B' by irradiation mutation breeding (Lu 1994). India has identified several TGMS strains, such as 'SM3', 'SM5', 'F61', 'JP2', 'JP8-1-A-12', 'JP8-8-ls', 'ICI0', 'ID24', 'JP1', 'JP24A', 'UPRI95-140', and 'SA2(F43)', among which 'JP8-1-A-12', 'F-61', and 'SA2(F43)' have a low critical sterility-inducing temperature and 'JP24A' belonged to the reverse TGMS group (Ali et al. 1995; Satyanarayana et al. 1995; Reddy et al. 1998a; Li and Pandey 1998). Three

76

J.

LI AND L. YUAN

TGMS lines, 'VN-Ol', 'VN-02', and 'TG-162', were identified in Vietnam (Minh et al. 1997), but these materials still need to be characterized in detail. 2. Responses to Temperature and Daylength

Photoperiod Sensitivity. Two photoperiodic reactions function simultaneously in PGMS rice. One affects the growth by delaying or promoting panicle differentiation and heading. The second affects development and determines the male sterility (Yuan et al. 1993). Fertility Alteration Sensitive Stage (FASS). The FASS varies among different T(P)GMS lines. The FASS of 'Nong-Ken 58s' is from the secondary branch primordium differentiation to the pollen mother cell formation stage, the most sensitive being the pistil and stamen formation stage (Yuan et al. 1988). The FASS of 'Shuang 8-14s' developed from 'Nong-Ken 58s' is different from that of 'Nong-Ken 58s' (Zhu and Yu 1987). Critical Daylengths for Fertility Alteration. The critical daylength is the shortest daylength inducing fertility of T(P)GMS lines. This variant depends on the different genetic backgrounds and ecological conditions. The critical daylength of 'Nang-Ken 58s' and 'E-Yi 105s' is 13.75 to 14.00 h, and their fertility alteration is not an abrupt change (Zhang et al. 1987). Under low temperature in the summer of 1989 'Nong-Ken 58s' and 'N5047s' showed male fertility even with the daylength of 14.17 h at Hangzhou, China (30°05' N). The fertility alteration is unaffected by daylength in some indica TGMS lines such as 'An-Nong s-l', '5460s', 'Heng-Nong s-l', and W6154s (Cheng et al. 1990). The fertility alteration stage also varies with different regions. For example, the critical daylength of 'Nong-Ken 58s' is 12.37 h at Hainan Island, but 13.67 h in Fujian Province of China. 'Nong-Ken 58s' is even fertile all year round under natural conditions in Guiyang of China (Lu 1992a,b). Critical Light Intensity. It was reported that the lowest light intensity that induces sterility was 50 Ix (Zhang et al. 1987). But the critical light intensity inducing male sterility is altered by the temperature. Under high tempera'ture, light intensity as low as 100 Ix can induce complete male sterility, whereas under medium or low temperature the critical light intensity inducing complete male sterility should be over 100 Ix (Liang et al. 1990). Light vs Dark Period. For the FASS of 'Nang-Ken 58s', light interruption with 50 Ix light intensity for 1 h or 200 Ix for 5-15 min could induce male sterility under short daylength and 27°C (Zhang et al. 1987). The effect of light interruption under short daylength was affected by

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

77

temperature. To induce male sterility under high temperature, the light intensity could be low and the light treatment duration could be short (Liang et al. 1990). The dark duration is more crucial than the light period for inducing male sterility. If the dark duration is 10 h or less, the male sterility will be induced. But, if it is 11 h or more, the male fertility will be induced. Short duration (1-3 h) of dark under long daylength causes no obvious effect on the fertility (Lu and Yuan 1991). Red Light vs. Infra-red Light. During the FASS, if red light is present during the dark period under short daylength, the fertility is decreased. If immediately after the red light, the far-red light is turned on the fertility will recover to the level that results from only far-red light illumination. Male sterility is determined by the last-illuminated light. Pure red light incompletely converts fertility as compared with far-red light, and only red light and blue light together can induce complete fertility conversion, which cannot again be reversed by far-red light (Yang and Zhu 1990). It is concluded that in addition to phytochrome, the blue light receptor, cryptochrome, also affects the regulation of male sterility of T(P)GMS lines. Interaction Between Temperature and Photoperiod. Fertility of the socalled PGMS lines is also affected by temperature during the differentiation of the floret primordium, indicating that photoperiod and temperature interact at certain levels (Wu et al. 1993; Zhang et al. 1994c). Transmission Among Main Culm and Tillers. During the sensitive or photoperiod induction stage, the photoperiod signal cannot be transmitted between the main culm and tillers, among tillers, or between the first crop and the ratooning crop (Zhu and Yang 1992).

Temperature Sensitivity. Temperature affects fertility alteration for all T(P)GMS lines (He et al. 1987; Li et al. 1989a; Sun et al. 1991; Xue and Chen 1992). Expression of photoperiod-sensitive male sterility can be altered significantly by changes in the mean daily maximum or minimum temperature during the daylength treatment (Xue and Zhao 1990). Critical Sterility Point (CSP). Different T(P)GMS lines have different critical sterility-inducing temperatures, referred to as the critical sterility point (CSP). For example, the CSP is 28.1-29°C for '5460s' and 24.2-26.5°C for 'An-Nong s-1' (Cheng et al. 1990; Yang 1990a; Chen et al. 1993). The CSP index evaluates the risk during commercial seed production for the two-line system. Molecular mapping using RFLP markers showed that the low-temperature sensitivity to male sterility or low CSP was likely controlled by three independent genomic regions, among which two were on chromosome 1 and one on chromosome 12 (Li et al. 1997).

78

J. LI AND L. YUAN

Temperature Sensitive Stage. In general, the temperature sensitive stage for fertility alteration of indica T(P)GMS lines occurs at about the meiotic division of the pollen mother cells, Le. 10-15 days before heading. It requires three to seven days to induce male fertility with low temperature (Chen et al. 1993). For japonica T(P)GMS line 'Nong-Ken 58s', the temperature sensitive stage is from the second branch primodia differentiation stage to the microspore uninucleate stage, the most sensitive period being from the pistil and the stamen primodia formation stage, which is longer than for the indica T(P)GMS lines, until the meiosis of the pollen mother cells (Zhang et al. 1992). Temperature Sensitive Part of a T(P)GMS Plant. In T(P)GMS plants, the young panicle is the critical organ that is sensitive to low temperature (Zhou et al. 1993a; Xu and Zhou 1996). The "cold water irrigation" method for effective production of T(P)GMS lines with low CSPs was invented by Xiaohe Luo and has been practiced extensively in China's two-line hybrid rice production (Hunan Hybrid Rice Research Center 1992; Chu et al. 1997). Similarly, the warmer water can prevent the se1£ing of T(P)GMS lines caused by low air temperature in the F 1 hybrid seed production (Xiao and Yuan 1997). The Effect of Temperature or Daylength in the Vegetative Stage. The close correlation between the photoperiodic response of heading and male sterility in '7001s' suggests that the genes responsible for heading and photoperiod-sensitive male sterility are not independently inherited (Tang and Shao 1997). The two photoperiodic reactions proposed by Yuan et al. (1993), seemed to interact with each other, and the temperature or daylength in the vegetative stage also influenced fertility alteration in T(P)GMS lines. Higher temperatures or shorter daylengths will lower the CSP (Zhang et al. 1992, 1993b). Model of Fertility Alteration by Interaction Between Temperature and Daylength. Based on the effect of temperature and daylength on fertility, a fertility alteration model was proposed for practical use (Zhang, Lu, and Yuan 1992). The key points are as follows: (1) When temperature is higher or lower than the physiological limit, rice grows poorly and develops abnormal pollen. These temperature limits are 10-15°C for lower limits and 35-41°C for higher limits (Suzuki 1978; Satake and Yoshida 1978). But Chinese studies indicated that the lower physical temperature limit was higher than 10-15°C. (2) If the temperature of a PGMS line is lower than the upper physiological limit but higher than the CSP, this PGMS line shows male sterility even under short daylength. (3) If the temperature of a PGMS line is higher than the lower

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

79

physiological limit and lower than the CSP, this PGMS line will show male fertility or semi-sterility even under long daylength. (4) Only within the daylength-sensitive temperature range, i.e. from the CSP to the critical temperature inducing male fertility, will the PGMS line show daylength sensitivity (Fig. 2.6). The daylength and temperature effects are complementary, Le. if temperature increases then the critical daylength inducing sterility will be shortened. Confirmed from practice is that no TGMS or PGMS materials tested under different daylength and temperature regimes show male sterility affected only by photoperiod (Deng et al. 1997). Therefore, T(P)GMS is employed in most cases of this review instead of TGMS or PGMS. 3. Development of New T(P)GMS Lines. Genic male sterile lines have been developed primarily by chemical or irradiation mutation (Fujimaki et al. 1977; Ko and Yamagata 1980,1987; Singh and Ikehashi 1981; Fujimaki and Hiraiwa 1986) although some T(P)GMS lines have been identified as spontaneous mutations (Suh et al. 1989). Most breeding methods used for inbred cultivars can be employed to develop T(P)GMS lines, including the pedigree method, cross breeding, mutation, and tissue culture.

Pedigree Method. Most Chinese source T(P)GMS materials such as 'Nang-Ken 58s', '5460s', and 'An-Nang s-1' are selected from the source male sterile plant(s). For example, in the case of 'An-Nong s-l' one male sterile plant was discovered in the F5 population of the cross ((Chao 40B x H285) x 6209-3] in 1987. 'An-Nang s-l' was developed from this plant using the pedigree method. Cross Breeding. Currently most Chinese-bred japonica T(P)GMS lines, such as 'N5047s' and '7001s', were developed using single cross breeding. If a japonica source T(P)GMS line is to be transferred to an indica rice cultivar, segregation among the progeny is difficult to stabilize. Therefore, the multiple cross or recurrent backcross breeding should be used (Li 1992). The indica T(P)GMS line 'W6154s' was developed from the triple cross [(Nong-Ken 58s x CS253-2-3-2) x Zhen-Shan 97]. The T(P)GMS line '8902s' was bred from the backcross F 3 progeny of the cross (Shuang 8-14s x Zhen-Shan 97) using 'Zhen-Shan 97' as the recurrent parent. A T(P)GMS line may also be developed from interspecific or intersubspecific crosses. For example, 'Heng-Nong s-1' was developed from the cross (long-awned wild rice x R1083) and 'Xin-Guang s' was developed from an indica-japonica cross. Mutation Breeding. Some scientists have developed T(P)GMS lines using irradiation or chemical mutation (Kato et al. 1990; Rutger and

CXl

o

Nong-Ken ·588

Fig. 2.6.

An-Nong 8-1

physiological upper temperature limit

>36 0

c

>3S 0 c

the critical temperature inducing male sterility

>32 0

c

>260

the critical temperature inducing male fertility

>24 0

c

>260 c

physiological lower temperature limit

>1 SoC

>1S0c

c

Model offertility alteration by temperature-photoperiod interaction. Source: Yuan 1992b; Zhang, Lu, and Yuan 1992.

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

81

Schaeffer 1990; Oard et al. 1991). The International Rice Research Institute developed 'IR32364s' from 'IR32364-20-1-3-2B' using gamma ray irradiation. Tissue Culture. Rutger et al. developed a T(P)GMS line from 'Calrose 76' using tissue culture. This T(P)GMS line showed male sterility under long daylength in California, but male fertility under short daylength in Hawaii. Another TGMS line, 'T-Shan-s', was identified using tissue culture of explants from the mature embryo of 'D-Shan B' (Huang et al. 1992). Other studies have shown the possibility of developing T(P)GMS lines using anther culture (Liu 1995; Niu et al. 1997; Hong et al. 1997).

Approaches to Improve Breeding Efficiency. Breeding practice in China has shown that the frequency of indica T(P)GMS plants is low for crosses with japonica T(P)GMS line as the donor (Lu et al. 1989). The average frequency of T(P)GMS plants in the F 3 generation from crosses between 'Shuang 8-14s', a japonica T(P)GMS line, and japonica cultivars was 6.5%, while in crosses between 'Shuang 8-14s' and indica cultivars it was 0.8%. Among these indica cultivars, the frequency of T(P)GMS plants for early-season indica cultivars in the Yangtze Valley of China was the lowest, 0.1 %, while that ofIRRI-bred cultivars was the highest, 1.5% (Zhu and Yu 1987). It is proposed that rice cultivars with low or weak photoperiod sensitivity and thermosensitivity should be selected as recipient parents (Lu 1992a,b). Ecological breeding is one way to increase the selection pressure for developing new T(P)GMS lines. This involves the use of long daylength and low temperature to identify and select the male sterile plants, and the use of short daylength and high temperature to increase the efficiency of multiplication of the T(P)GMS line. In China, this method has enhanced the efficiency of T(P)GMS line breeding (Lu et al. 1994). Photoperiodic response of plant development is positively correlated with photoperiod-sensitive male sterility, but negatively correlated with temperature-sensitive male sterility. Thermoperiodic response for plant development shows no correlation with photoperiod-sensitive and temperature-sensitive male sterility. Therefore, selection of plants with photoperiodic response would be preferable for the development of photoperiod-sensitive male sterile lines (Chen and Wan 1993). 4. Evaluation ofT(P)GMS Lines. China has set the following eight criteria for the acceptance of new T(P)GMS lines (Yuan 1990; Lu et al. 1994): (1)

the tested population size should be larger than 1,000 plants; (2) these plants should express uniform agronomic traits; (3) the percentage of

82

J. LI AND L. YUAN

male sterile plants should be 100%, and pollen sterility of the male sterile plants should be over 99.5%; (4) the fertility alteration should be obvious; (5) the seed set percentage should be more than 30% during fertility induction; (6) the duration of complete male sterility should last for at least 30 days; (7) the outcrossing seed set percentage should be higher than that of 'V20A', 'Zhen-Shan 97A', or 'Liu-Qian-Xin A'; and (8) the CSP should be between 23-23.5°C, or even lower. In addition, the ideal T(P)GMS lines should have little sensitivity to low temperature for the duration of their male sterility or F 1 seed production, as well as to high temperature for the duration of the male fertility or T(P)GMS line multiplication (Yuan 1992b). Methods to evaluate T(P)GMS lines include field or phytotron evaluation. Field plantings involve sequential plantings at different locations. During the season that the rice plant can grow, 20-30 plants should be grown every 10-15 days. The initial heading date, pollen sterility, and selfing seed set data should be collected and analyzed to evaluate and compare the T(P)GMS lines. However, under natural conditions, long (or short) daylength and high (or low) temperature always coincide, so the individual effect of the temperature and the daylength cannot be resolved. Therefore, it is impossible to distinguish between TGMS and PGMS, and to determine the temperature sensitivity of the T(P)GMS lines only using the ecological evaluation of sequential plantings. The phytotron can be used to dissect the effects of daylength and temperature on the fertility alteration of T(P)GMS lines. For each phytotron treatment, 10-15 plants should be evaluated using the same traits as those measured for the ecological evaluation by sequential plantings. Variable temperature in a day is more dependable and currently used for the evaluation ofT(P)GMS lines (Wang et al. 1994; Deng et al. 1996). Scientists at the China National Rice Research Institute (CNRRI) studied 101 T(P)GMS lines under nine controlled regimes consisting of three photoperiods (15.0, 14.0, and 12.5 h) x three average temperatures (30.1, 24.1, and 23.1°C) and found that 96% ofT(P)GMS lines could be divided among three types based on variance analysis of the seed set: (1) PGMS characterized by significant P (photoperiod) and P x T interaction effects but a non-significant T (temperature) effect on fertility; (2) TGMS characterized by a significant T effect and a non-significant P effect on fertility; and (3) P-TGMS with only a significant P x T interaction effect on fertility. Among the japonica T(P)GMS lines studied, 32.3% were PGMS, 9.7% were TGMS, and 51.6% were P-TGMS. In contrast, among indica T(P)GMS lines studied, none were PGMS, 61.4% were TGMS, and 35.7% were P-TGMS lines (Cheng et al. 1996).

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

83

5. Maintaining Temperature Sensitivity ofT(P)GMS. Individual plants of a T(P)GMS line differ in their response to temperature and daylength. Seed set of plants with higher CSP is higher than for the plants with lower CSP. Consequently, there is "CSP drift" toward higher temperature in subsequent generations. Thus, the T(P)GMS line cannot be put into commercial production, because of the increasing risk of unacceptable F1 hybrid seed production using the conventional method for inbred multiplication. For example, the CSP of 'Pei-Ai 64s' was 23.3°C in 1991, and it increased to 24-25°C in 1993 using the conventional multiplication method. To solve the problem of CSP increase in the generation advance, the "core seed production" procedure was proposed in 1994 (Yuan 1994a,b,c). The procedure consists of: (1) selection of individual plants; (2) treatment of the individuals with low temperature or low temperature and long daylength; (3) selection of plants with low CSP; (4) ratooning of the selected plants to obtain selfed seeds; (5) sequential development of core seeds, breeder's seeds, and foundation seeds; and (6) F 1 seed production. China's experience has demonstrated that the core seed production procedure maintains low CSP for commercial hybrid rice production (Deng and Fu 1998). A similar procedure for the production of nucleus and breeder's seed of TGMS lines has been proposed (Virmani et al. 1997). Multiplication of TGMS lines can also be carried out using the thin layer cell culture technique with 50% callus induction and 100% plant regeneration (Nhan et al. 1997). Other approaches for multiplication and maintenance of genetic stability in rice such as micropropagation or ratooning multiplication deserve further study.

C. China's Progress

The discovery by Shi (1981, 1985) of 'Nang-Ken 58s' and increased understanding of the phenomenon of "wide compatibility" in rice (Ikehashi 1982) provided the genetic tools needed to achieve development of the two-line system for hybrid rice and the utilization of intersubspecific heterosis. With these two genetic tools, China initiated a collaborative research program on the two-line hybrid rice system in the 1980s. After more than ten years of nationwide collaborative study, both the physiological and genetic mechanisms of T(P)GMS are basically understood. Ten two-line system hybrid rice cultivars that normally outyield the three-line system hybrid rice cultivars by 5-10% have been released and put into commercial production (Table 2.15; Yuan 1997a).

J.

84

LI AND 1. YUAN

Table 2.15. Two-line system hybrid combinations certified and registered. Source: Yuan 1997a. Province certification

Combinations

Year registered

Pei'ai 64s x Teqing

Hunan

1994

Pei'ai 64s x 288

Hunan

1996

Pei'ai 64s x Yuhong 1

Hunan

1997

7001s x Xiushui 04

Anhui

1994

7001s x Wanhui 9

Anhui

1994

7001s x 1514

Hubei

1995

5088s x R187

Hubei

1995

7001s x Shuangjiu

Anhui

1997

Pei'ai 64s x Shanqing 11

Guangdong

1996

Shuliangyou 1

Sichuan

1996

The area under the two-line hybrid rice system has increased progressively as shown as Table 2.16 (Yuan 1997a). The area of two-line hybrid rice production in China is expected to reach 1.3-1.5 million ha in the year 2000. D. Breeding for Two-line System Rice Hybrids Using Chemical Emasculators When sprayed on rice plants at specific developmental stages, a gametocide or chemical hybridizing agent (CHA) can emasculate the plants, thus resulting in male sterility while maintaining normal female fertility (Lasa and Bosemark 1993). A rice cultivar having a superior specific Table 2.16. The planting area and yield of twoline hybrid rice in China from 1993 to 1998.

Year

Planting area (x 1,000 hal

Yield (kg/ha)

1993

27

7170

1994

67

7005

1995

73

7215

1996

200

7190

1997

270

7150

1998

437

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

85

combining ability is used as the male parent for producing the F 1 hybrid seed. Many scientists have reported that ethephon [(2-chloroethyl) phosphonic acid] induces male sterility of crops (Rowell and Miller 1971; Bennett and Hughes 1972; Perez et al. 1973; Hughes 1976; Parmar et al. 1979; Chan and Cheah 1983). More than 50 chemical emasculators have been identified for more than 40 crops, but most also damage the pistil or cause abnormal flowering (Wang et al. 1981, 1991b,c; Yu et al. 1991). Some of these chemical emasculators are listed in detail by Kaul (1988). For rice, male sterility is induced in 'PR106A' using 0.4% EMS for 48 h at 10°C (Minocha and Gupta 1988; Minocha et al. 1991). Aswathanarayana and Mahadevappa (1991,1992) reported that 800 ppm ofGA, 8000 ppm of ethephon, 0.02% maleic hydrazide (MH), and 0.8% 2,4dichlorophenoxyacetic acid (2,4-D) induced a high level of male sterility in rice. Kitaoka et al. (1991) reported that the male sterility reached 95% or more using isourea at 3 kg/ha + ethephon at 5000 ppm or by alternatively using isourea at 10 kg/ha + ethephon at 2500 ppm. Other chemicals such as RH531 and DPX [3-(p-chlorophenyl-6-methoxy-3-triazine-2,4-(lH, 3H)dione] have also been tested in rice (Perez et al. 1973; Long et al. 1973; Zhangxing and Chunnong 1980). In the 1970s and 1980s, China developed chemical emasculators using arsenate, such as Male Gametocide 1 (zinc methyl arsenate, CH:lAs0 3 Zn) and Male Gametocide 2 (sodium methyl arsenate, CH;lAsO;{Na z). While they caused excellent emasculation of rice, they were very toxic to the environment. Later, non-arsenic chemical emasculators such as N312, HAC123, CRMS-1, and 13(a pyridazinone derivative) were developed (Luo et al. 1988; Zhong et al. 1997). Currently India has identified some less toxic gametocides including ethyl 3' methoxy oxanilate and ethyl 4' fluorooxanilate (Siddiq 1994; Siddiq et al. 1994). In China the most popular hybrids produced by chemical emasculation were 'Gang-Hua 2', 'GangHua x Qing-Lan', and 'Qing-Hua x Fu-Gui', which were collectively grown on 60,000 ha in Guangdong and Jiangxi province during the mid1970s to early 1980s. The yield data by chemically emasculated hybrids from 1982 to 1985 shows that the yield increase over the check threeline hybrids ranges from 7.8-18.2% (Shao and Hu 1988). Unfortunately, the application ofthe two-line system hybrid rice using chemical emasculators was basically not successful due to the apparent contradiction between the seed purity and the seed yield. 1. Key Techniques for Chemical Emasculation. Theoretically, 100% of male sterility will result in 100% pure hybrid rice seed. Unfortunately, the highest levels of male sterility due to chemical emasculation also

86

J. LI AND L. YUAN

cause increased female sterility, or otherwise decrease the outcrossing potential of the female parental lines and, consequently, the F 1 seed yield. To some extent, lower levels of male sterility can achieve both high yield and high purity in the F 1 seed. The usual criteria for chemical emasculation in China are: 95% male sterility, 30% outcrossed seed set, 85% seed purity and seed production of 1.5 t/ha. Key techniques are careful selection of female and male parental lines and precise application of emasculators. The stamen of the female parental line, but not the pistil, should be very sensitive to the chemical emasculator. Furthermore, the blooming characteristics should be affected little by the chemical emasculator. Erect leaves and uniformity of the developmental stages between tillers and plants result in more effective emasculation. In addition to short plant height, a high stigma exsertion rate and large floret opening angle of the female parental line contribute to high yield of hybrid seed. Possession of favorable agronomic traits, yield performance, and grain quality by the female parent leads to better performing F 1 hybrids. The male parental line should have high specific combining ability, large panicles, large pollen load, long blooming period, be taller than the female, and have growth duration close to that of the female parental line. Finally, uniform application of the chemicals at the correct stage is essential to F1 hybrid seed production. 2. Considerations of Chemical Emasculation

Advantages. Less time is needed to develop hybrids and a broader germplasm is available for maximizing rice heterosis. Development of a male sterile line does not require several generations. The two-line system via T(P)GMS lines eliminates the maintainer line, but these lines must be intensively evaluated prior to any F1 hybrid breeding, and stability of the conditional male sterility must be maintained. In comparison, chemical emasculation requires far less work, as only the identification of an effective chemical emasculator is required prior to breeding of the rice hybrids. Chemical emasculation also circumvents the genetic vulnerability of cytoplasmic-nuclear male sterile lines in the three-line system. As a result of prior progress on inbred breeding, heterosis can be further exploited. For example, the F 1 hybrid between 'Gang-Zi-Zhan' and 'IR661' showed strong heterosis. But, both parental lines were difficult to develop into A lines. Chemical emasculation made this strong heterosis commercially available in China (Guangdong Crop Heterosis Utilization Research Cooperative Group 1977, 1981). Finally, in the chemical emasculation method, segregation for male sterility in the F2 genera-

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

87

tion does not occur. For a rice hybrid between intervarietal cultivars with close growth duration, chemical emasculation provides heterosis not only for the F 1 but also for the F 2 generation of some combinations such as 'LiHua-Da-Zhen' (Li et al. 1986, 1989b; Tu and Hu 1989).

Disadvantages. There are several problems with chemical emasculation. Effective chemical emasculators are still needed. For example, while etheflon induces pollen sterility it also has a phytotoxic effect on the panicle length and spikelet size (Shamsi et al. 1996). Chemical emasculators containing arsenic, which used to be employed in China's hybrid rice seed production, have a number of limitations, not the least of which is harmful residues left following application of the chemicals. It has been shown that the arsenic content in rice stalks reaches 4.19 mg/kg due to using arsinyl at 345-435 g/ha for emasculation (Liu et al. 1983). A similar study also indicated that the arsenic residue in the rice grains was 2 mg/kg and that in rice stalk was 9 mg/kg when 390 g/ha of arsinyl was applied to the female parent of the rice hybrid combination 'Hong-Yang-Ai 2' to achieve 90% purity. Furthermore, the effective concentration for emasculation is narrow. Too small a dosage of arsenic emasculators does not emasculate completely and too large harms the pistils. To complicate the problem, the required application time and dosage varies among different rice cultivars. The compromise between seed production yield and seed purity is also a problem. The best time for application of arsenic emasculators is about 10 days after the meiosis stage until the pollen filling stage, but the development rate is not as uniform among plants and tillers as required, therefore, the chemical emasculation effectiveness varies. Arsenic is harmful to the female parental line. A shortened rice culm, panicle enclosure within the flag leaf sheath, glume closure, damaged pistils, and even decreased F I seed germination can result from arsenic chemical emasculators. Not only the stamens but also the pistils are sensitive to chemical emasculators in most rice cultivars. A complete lack of F 1 hybrid rice seed can be a result of severe damage to the pistils. This disadvantage strongly decreases the probability of exploiting heterosis in the rice germplasm. The interaction between emasculators and climate can be a problem as well. It usually requires 5-6 h for a chemical emasculator to take effect. Rain soon after the application of a chemical emasculator will decrease the effectiveness of the emasculation. If the chemical is applied again, the effective dosage is difficult to determine. Hybrid rice seed cannot be produced if rainy weather lasts for the duration of effective application of the chemical emasculator. A few other climatic conditions

88

J. LI AND L. YUAN

such as temperature, wind, and humidity will also affect the application effectiveness (Zhong et al. 1992).

VI. WIDE COMPATIBILITY AND UTILIZATION OF INTERSUBSPECIFIC HETEROSIS

A. Classification in Rice There are 20 rice species, of which O. sativa and O. glaberrima are the two cultivated rice species. Ecotypes have differentiated within these subspecies. Classification is based on morphological and ecological features of O. sativa (Cheng 1993; Gu 1988). Kato classified Oryza sativa as indica and japonica by analyzing the morphological appearance, affinity, and serological reactions. Some scientists think there should be a third group, the tall and long-panicled Indonesian cultivars designated as javanica by Kornicke (1885), in addition to the indica group and japonica groups (Morinaga 1954; Chang et al. 1991). The javanica rice corresponds to "Group Ie" of Terao and Mizushima (1944), "B plant type" of Matsuo (1952), and the "tropical insular groups IIa" and "lIb" of aka (1958). aka later designatedjavanica as "tropicaljaponica" (aka 1983). Together with the aus cultivars from Bangladesh and eastern India, which have high genetic affinity with both the indica and japonica cultivars, Indonesian cultivars were considered to be an intermediate type by Morinaga and Kuriyama (1958). Chang et al. (1991) suggested that the javanica group should also include the bulu and upland or hill rice. Currently, rice classification is generally based on the morphological, biochemical, and genetical features. 1. Morphological Classification. Morphological traits such as grain shape, apiculus hair length, and phenol reaction are used to classify rice. Cheng (1985) proposed a morphological index for classification in rice (Table 2.17), the accuracy of which can be as high as 95%. For the morphological classification method, one of the most important indices is hybridization compatibility. Seed set of crosses between indica and japonica generally ranges from 0 to 30% except for the cross between the aus ecotype of indica and japonica cultivars. Seed set of specific crosses between indica and japonica still varies over a wide range. The seed setting percentage of the F1 between Chinese indica and Chinese japonica cultivars varies from 1.3% to 80.3%, and some rice cultivars from Yunnan of China have good compatibility with both indica and japonica cultivars, although the seed setting percentages of

89

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

Table 2.17. The grading system for rice cultivars using morphological index method.>: Source: Cheng 1985. Grading Variable Apiculus hair Length Uniformity Hardness Phenol reaction

0

3

2

1

4 Very long Very variable

Very short Very uniform Very hard. erect

Long Uniform

Medium Intermediate

Long Variable

Hard

Medium

Soft

Very soft

Black

Light black

Grey

Stained on edges

Unstained

Panicle internode length (1"1 & 2'1(1)

<2cm

2.1-2.5 em

2.6-3 em

3.1-3.5 cm

> 3.5 cm

Glume color when heading

Green & white

White & green

Yellow & green

Light green

Green

Leaf pubescence

Very high

High

Intermediate

Slight

None

3.5-3.1

3.0-2.6

2.5-2.1

< 2.0

Length-width ratio of spikelets > 3.5

ZNote: Indica cultivars should have a sum of grades for all items ranging from 0 to 8, and the japonica varieties from 18 to 24. If the sum falls between 9 and 13 or between 14 and 17 the eultivars are biased toward indica or japonica. respectively.

most are below 50% (Yu and Lin 1962). Based on their hybridization compatibility, some aus and buIu cultivars are classified as the "intermediate type" (Ikehashi and Araki 1987). Mathematical approaches such as principal component analysis have been employed for classifying rice based on morphological traits (Zhou et al. 1988). 2. Biochemical and Genetical Classification. Glaszmann (1987) indicated that most japonica rice cultivars had the isozyme alleles Acp-1 2, Cat-1 2, Est-3 1, and Pgc-1 2 • Later Est-X for esterase isozyme was found, for which the alleles Est-XlO, Est-X 11, Est-X 13, and Est _X14 differentiated among indica, japonica, aus, and wild rice (Cai et al. 1992). RFLP can also be applied for classification of rice (Tanaka et al. 1989; Zheng et al. 1990; Kawase et al. 1991). The same results should arise from different classification methods, but some RFLP analysis results differ from those of morphological classification. For example, cultivars 'Sipule' and 'Ketan Nangka' were classified as indica and japonica, respectively, by the morphological classification method, but as different subspecies using RFLP markers (Tanaka et al. 1989; Zheng et al. 1990). Consequently, integration

90

J. 11 AND L. YUAN

and comparison of the different classification methods provide a scientific basis for rice classification. B. Phenomenon of Wide Compatibility The F 1 seed set of intersubspecific crosses was usually below 30%, while for crosses between different ecotypes of the same subspecies it was over 70% (Carnahan et al. 1972). There are exceptions, however. Some cultivars such as the ones from the aus (indica type) or buJu (japonica type) showed high F 1 seed set when crossed with indica as well as japonica (Terao and Mizushima 1939; Morinaga and Kuriyama 1958; ,Heu 1967). Normal F 1 seed set was also found in some indicajaponica crosses such as crosses between 'Ai-Zi-Zhan l' (indica) and Taiwanese japonica cultivars, between 'You-Mang-Zao-Sha-Jing' (japonica) and indica cultivars, and between a rice cultivar derived from an indica-japonica cross and indica or japonica cultivars (Min 1986). The "intermediate type" was studied as early as in the 1930s, but Ikehashi (1982) first proposed the term "wide compatibility." Wide compatibility is the phenomenon of the F 1 seed set being normal in crosses between some intermediate-type rice cultivars and both indica and japonica cultivars. These cultivars are called wide compatibility varieties (WCVs). The controlling gene is called the wide compatibility gene (WCG). Discovery of wide compatibility in rice provided the opportunity to overcome the reproductive barrier exhibited in the F 1 generation of crosses between the indica and japonica cultivars, and thereby to use the strong heterosis of intersubspecific crosses. This has received much attention from rice scientists. C. Genetics of Wide Compatibility Traits 1. Hypotheses to Explain Semi-sterility in indica-japonica Crosses. The

following hypotheses were developed to explain the semi-sterility of the F1 generation of indica-japonica crosses:

Duplicate Gametophytic Lethal Hypothesis. The existence of duplicate gametophytic lethal gene(s) has been hypothesized with the duplicate loci being independent (Oka 1953, 1988). If a gamete has the double recessive combination it will abort during its development. These lethal genes could affect both the male and female, or the male gametes only. Chromosome Aberration. Yao et al. (1958) hypothesized that, in the gametic development of the F 1 between indica and japonica, one of the

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

91

homologous chromosomes became aberrant resulting in pollen abortion (Chandraratna 1964; Zhou 1978).

Lack of Coordination between Cytoplasm and Nucleus. Turbin considered that coordination did not occur between the cytoplasm and nucleus of indica and japonica, so the gametes and zygotes from indica-japonica crosses could not develop normally and the pollen aborted (Zhou 1978). Allelic Interaction. The hypothesis is that the gamete lethality originates from the allelic interaction in the F1s between indica and japonica. Suppose the genotype of indica is Fi/Figand japonica is FiJFig, and the F1genotype Fi/Figis gamete-abortive. The F1 will be fertile if Fn/Fng is crossed with Fig/Fig or FVFi g (Kitamura 1961, 1962a,b,c). This hypothesis was later confirmed and developed into the "wide compatibility" theory by Ikehashi and Araki (1986) using the triple cross method. Multigenetic Inheritance. The semi-sterility in the F1 of indica-japonica crosses was reported to be controlled by multi-gene(s) or specific compatibility genes. At least six loci were involved in determining the semisterility of indica-japonica crosses (Zhang and Lu 1989; Shen and Xu 1992). Genetic Recombination. Seven ancestral parents of 'T984', a WCV with a wide spectrum of compatibility, were tested and no wide compatibility was identified in any of the ancestors. This suggests the hypothesis that the wide compatibility of 'T984' arises through genetic recombination (Xiong et al. 1993). 2. Chromosomal Identification ofWCG. The first WCG, 5 11 5 was identi-

fied on chromosome 6 (lkehashi and Araki 1986; Araki et al. 1988). Wide compatibility of other Chinese-bred WCVs (e.g. 'Lun-Hui 422' and '02428') was also determined to be controlled by 5 115 on chromosome 6, but, the loci were ordered differently (Gu et al. 1991; Gu, You, an,d Pan 1991; Gu et al. 1992; Lu and Pan 1992). It was reported that 5 11 5 was between Wx and C, Le. WX-5 11 5 -C in studies by Ikehashi and Araki (1986) and Liu et al. (1992). Conversely, Sll5 was linked with Wx and Gin the order WX-C-S1l 5-RG213-RG64 (Wang et al. 1994; L.S. Liu, pers. commun.). Zheng et al. (1992) confirmed that there was a locus between C and RG138 that controlled the seed set of the indica-japonica F1s, using six RFLP markers on chromosome 6 to analyze the 'Pecos' population. Crosses between WCVs with 511 5 and some rice cultivars do not show normal fertility and linkage with C, so additional loci are assumed to

92

J. LI AND L. YUAN

control wide compatibility (Xu et al. 1989). A genome-wide mapping of a three-way rice cross [(02428 x Nanjing 11) x Balilla] showed that three loci conferred significant effects on hybrid sterility, the major locus on chromosome 6 was 5 n s and the two minor loci on chromosomes 2 and 12 could cause partial sterility even in the presence of 5 n s (Liu et al. 1997a; Zhang et al. 1997c). Two more WCGs have been identified, 5 n 7 in 'Dular' and 5 y in 'Penuh Baru II'. 5 n 7 is located between Rc and Est-9 on chromosome 7 (Ikehashi 1991; Ikehashi et al. 1991, 1994; Yanagihara et al. 1992), and is linked with RZ488 and RG511 in 'Aus 373' O.S. Zou, pers. commun.). Another WCG, 5 n a in 'Akihikari' is about 11.2 cM from Cat on chromosome 6 (Wan et al. 1993). The hybrid sterility of crosses between Chinese indica and japonica cultivars is mainly controlled by the 5-5 locus, whereas the hybrid sterility of aus cultivars crossed to indica, japonica, or javanica cultivars is controlled by allelic interaction among the sterility loci 5-5, 5-7, 5-9 and 5-15 (Wan and Ikehashi 1997). Six WCG loci (5-5, 5-7, 5-8, 5-9, 5-15, and 5-16) have been identified on chromosomes 6, 4,6, 7, 12, and 1, respectively (Virmani 1996). D. Development ofWCVs 1. Screening for WCVs. Six WCVs were identified in a group of 74 rice

cultivars from Indonesia, India, Bangladesh, and the Philippines, i.e. 'Padi Bujang Penedak', 'Aus 373', 'Dular', 'Calotoc', 'CPSLO-17', and 'Ketan Nangka' by Ikehashi and Araki (1984). Thereafter, more and more WCVs have been identified by rice scientists at the International Rice Research Institute and in China (Table 2.18; Zhu and Yang 1992; Min 1990; Xiong et al. 1989, 1990; He and Yuan 1993; Zhang et al. 1988). Japanese rice scientists pointed out that the aus cultivars of Bengal, the buiu cultivars of Java, cultivars from Nepal and other Himalayan tracts, and the landraces of tropical Asian countries gave fertile F 1 plants when crossed with both indica and japonica rice cultivars (Morinaga and Kuriyama 1955,1958; Morinaga 1968; Oka 1988). Three main sources of WCVs were suggested by Luo, Ying, and Wang (1991):

Primitive indica and japonica. In the region of rice origin there may exist primitive indica or japonica types with incomplete differentiation, such as nuda from Yunnan of China or Southeastern Asian countries, javanica from Indonesia, and aus from the Indian subcontinent. M.H. Gu (pers. commun.) also indicated that the Yunnan rice landrace was an important source for wide compatibility genes in addition to the ausand javanica-type rice cultivars.

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

93

Table 2.18. Recently screened rice cultivars with wide compatibility. Source: Zhu and Yang 1992; Sun and Cheng 1994. Universities or institutes

Country

WCVs

Beijing Agricultural University

P40, Chang-Mao-Nuo, Ai-Zi-Zhan, Bai-Lian-Dao-Gu

China National Hybrid Development Center

Pei-C312, CY85-41, CY85-43, Pei-C 116, Pei-Cl18, Pei-Ai64, Lun-Hui422, Pei-Ai64s, Lin-Lun, lin-PeL CB-l, AB78

China National Rice Research Institute

T984, T986, Pecos, Chugoku 91, L201, Gogo Serah, Nggonemal, Tanggalasi, Senatus Madumi, Nova 76, Newbonnet, Bluebonnet, Bluebelle, Changnot

Guizhou Academy of Agricultural Sciences

Bai-Ke-Jing-Dao, Huang-Ke-Jing-Dao

Huazhong Agricultural University

Hao-MeL Lemont, 822, 0046, B5580A 1 -15

Fujian Academy of Agricultural Sciences

Vary Lava 1312

Jiangsu Academy of Agricultural Sciences

02428, Guang-Kang-Jing 2

San-Ming Prefectural Agricultural Research Institute of Fujian Province

SMR, 68-83, CR44-38. BJ8. IR4-114-3-2-1, g4025-2, g4135-1

Sichuan Agricultural University

CA527, CA529, CA537, CA544, Lemont. Bellemont, Jian-12

Wuhan University

MCP231-2. MCP231-4. MCP231-6, MCP231-7. 69 series. 8925s, 8926s

Zhejiang Agricultural University

T8340. Er-Jiu-Feng. IR58, Xin-Guang s, Xiu-shui 117

Japan

Tropical Agricultural Research Institute

Aus 373, Dular. CPSL017, Calotoc, Ketan Nangka, Tykuchern, Kuchem, NK4, DV149. KaladumanL DV52, AS35. Lepudumai. Padi Bujang Pendek. Norin PL9

Philippines

IRRI

BPI76, N22. Moroberekan, PBMN I, Fossa HV, Palawan. Lambayeque 1

China

Intermediate Type Between indica and japonica. During the evolution of cultivated rice, intermediate types between typical indica and typical japonica rice have arisen. These intermediate types may have wide compatibility.

94

J. LI AND L.

YUAN

Progeny from indica-japonica Crosses. 'T984' was developed from multiple crosses between '300' (a cultivar from an intersubspecific cross), 'IR26' (indica) and 'C57' (a japonica R line from an intersubspecific cross). The progeny from indica-japonica crosses are similar to the intermediate type. Some rice cultivars from the United States, Korea, and the Indian 'CR' cultivar system belong to this type and may have wide compatibility. Besides the screening method, anther culture was also successfully employed to develop WCVs (Chen et al. 1997b; Yang et al. 1997). 2. Evaluation of WCVs. Selection of testers is fundamental to evaluating WCVs. Ikehashi and Araki (1984) first selected 'IR26' and 'IR50' as indica testers, and 'Nihonbare' and 'Akihikari' as japonica testers. WCVs should have over 90% pollen fertility and over 75-80% seed set when crossed with testers. Later they suggested four tester cultivars: 'Achar Bhog', 'Ketan Nangka', 'IR36', and 'Taichung 65' or 'Akihikari'. The International Rice Research Institute evaluated WCVs using 'Akihikari', 'Toyonishiki', and 'Taichung 65' as japonica testers, and 'IR36', 'IR50', and 'IR64' as indica testers (Gu et al. 1991; Gu, You, and Pan 1991). The China's National Two-line System Hybrid Rice Research Cooperative Group selected first the following as japonica testers: 'You-Mang-ZaoSha-Jing' (early-season from Shanghai, China), 'Banilla' (mid-season from Italy), and 'Akihikari' (mid-season from Japan); the first selected indica testers were: 'Nan-Te-Hao' (early-season from Jiangxi of China), 'Nan-Jing 11' (mid-season from Jiangsu of China), and 'IR36' (mid-season from IRRI). If the pollen fertility and seed set of the crosses between the cultivar being tested and all six testers are over 70%, it is classified as a first-rate WCV. If the pollen fertility and seed set are over 70% in the crosses with only five testers, the cultivar is considered a second-rate WCV. In later practice the order of wide-compatibility testing ability of the six testers was proven to be: 'Nan-Te-Hao' > 'Nan-Jing 11' > 'IR36' for the indica testers, and 'Banilla' > 'Akihikari' > 'You-Mang-Zao-Sha-Jing' for the japonica testers (Gu et al. 1991; Gu, You, and Pan 1991). Therefore, the Cooperative Group chose 'Nan-Jing 11' and 'IR36' as indica testers, and 'Akihikari' and 'Banilla' as japonica testers (Gu 1992). Min (1990) suggested a statistical standard for the evaluation of the seed set oftestcrossing F1 s be established, rather than an absolute value because environmental conditions affect the F 1 seed set. The concept "spectrum of wide compatibility" was proposed for breeding practice, using only WCVs having both high and a wide spectrum of compatibility to breed for intersubspecific hybrids (Min 1990; Yuan et al. 1997).

2. HYBRID RICE: GENETICS. BREEDING, AND SEED PRODUCTION

95

E. Utilization of Intersubspecific Heterosis Intersubspecific hybrids are expected to increase the genetic diversity of parental lines, to improve on some undesirable traits of indica, and to add tolerance in adverse conditions. Some promising indica-japonica hybrids were developed with 10-50% yield increase over the checks, and are under different levels of field trials. Some, such as 'Pei-Ai64s x E32', 'Kanto Kou 1', and 'Ouu Kou 1', are ready for release to farmers (Ikehashi et al. 1994; Yuan 1998a).

1. Problems from Using indica-japonica Heterosis. Problems encountered during the initial use of intersubspecifc heterosis are described below. Most of these have been overcome in recent years.

Low Seed Set. The compatibility of the WCVs still varies between different subspecies and even different cultivars. For 133 intersubspecific crosses between 12 WCVs and 12 testers, the average seed set percentages were 68.1 % for 33 indica-japonica crosses, 71. 7% for 45 indicajavanica crosses, and 77.3% for 60 japonica-javanica crosses. The sequence of fertility was: indica-japonica crosses < indica-javanica crosses < japonica-javanica crosses, which was the reverse of the sequence for the level of heterosis (Yang and Li 1989). QTL mapping of reproductive barriers in indica-japonica hybrids indicated that nine QTLs on chromosomes 1, 3,4,5,7,8, and 12 increased sterility and only one QTL (stj-6) at chromosome 6 increased fertility (Liu et al. 1997c). Superiority to Parental Lines in Plant Height and Growth Duration. The plant height of the intersubspecific crosses is normally greater than for the parents. This is not a problem if the parental lines are selected to have allelic semidwarf gene(s). Similarly, by selecting the proper parental lines with short or even medium growth duration, the growth duration problems can be overcome. There was no cytoplasmic effect on the growth duration of the indica-japonica crosses (Li 1990, 1991a; Sun et al. 1993). Poor Grain Filling. Poor grain filling was found to be caused by a number of factors, including senescence, sink-source problems, and nutrient and water barriers. Senescence of the intersubspecific cross (W6154s x AB240) arises from the WCG donor, 'CPSL017' (Zhu and Liao 1990). Root senescence may also be one of the causes for poor grain filling (Chen, Deng, and Ma 1992). The problems of coordination between source and sink are due to the long growth duration of intersubspecific crosses. As a result, the panicle size or sink is large, while the source is

96

J.

LI AND L. YUAN

relatively limited (Lu et al. 1992; Li and Ren 1994). The growth duration of the hybrid 'W6154s x AB240' was 144 days and panicle size was 301 spikelets per panicle, so most grains were not completely filled. The seed set was only 40.3% with three separate filling stages on a panicle. The lack of sink-source coordination was confirmed by removing leaves and panicles (Zhu and Liao 1990; Chen et al. 1991). Nutrient and water transportation barriers of the rice plant were also observed to have a negative impact on grain filling. The poor flow of photosynthetic products to the panicles of intersubspecific hybrid rice is the main cause for poor grain-filling (Zhu et al. 1997). Experiments showed that in intersubspecific crosses only 64.5-75.4% of 14C assimilation product was transported to the panicle while 15-19% remained in the flag leaves during spikelet formation. Furthermore, 14C was not detected in the roots of the progeny. In contrast, 80.0% of 14C assimilation product was transported to the panicles and 1.2% to the roots of progeny from the intercultivar crosses (Chen, Deng, and Ma 1992). 2. Strategy for Utilization of Intersubspecific Heterosis. The intersubspecific heterosis level tends to be: indica-japonica F1 > indica-javanica F 1 > japonica-javanica F 1 > indica-indica F 1 > japonica-japonica F1 (Yuan 1990, 1992a,c; Zeng et al. 1997; Zhang et al. 1997b). In comparison with the three-line system indica intercultivar hybrid 'Shan-You 63', the indica-japonica F1 hybrids 'Zao-Xian-Dang x 02428' and '3037 x 02428' had 7.4-41.0% greater yield, 49.4-52.4% more spikelets/unit area, 13.0-15.8% more biomass/unit area, and a 6.4-14.9% higher economic coefficient. They also showed higher photosynthetic efficiency (Gu et al. 1989; Lu et al. 1991). Heterosis of 140-170% was observed in the crop growth rate (CGR) during the first 30 days after transplanting, 110-125 % heterosis in yield and higher tolerance of low temperature in the indica-japonica hybrids (Kabaki et al. 1992). Unfortunately, in crosses between typical indica and typical japonica cultivars, though the vegetative heterosis is large, it is not coordinated with the reproductive heterosis and thus is difficult to utilize (Yang 1990b). It was suggested that low seed set of indica-japonica hybrids under low temperatures possibly originated from the complementation of two pairs of genes from both parents (Li et al. 1996a). Furthermore, the indica-japonica hybrids generally have poor grain quality due to the segregation of quality traits in hybrid grains (Khush and Aquino 1994). To overcome these problems, Yuan (1991a,b) proposed the following breeding strategy for intersubspecific heterosis utilization: (1) Instead of typical indica or typical japonica cultivars, javanica cultivars, biased indica or biased japonica rice cultivars should be selected as parental lines. It has been proven that

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

97

heterosis is best exploited in crosses between the U.S. rice WCVs or intermediate-type cultivars and indica or japonica rice (Yuan 1995; Li et al. 1995). IRRI's study also suggested that hybrids derived from crosses (indica x temperate japonica) had lower heterosis while hybrids from crosses (indica x tropical japonica) had higher yield than the ones from indica-indica and crosses (tropical japonica x tropical japonica) (Virmani 1994a; lnt. Rice Res. lnst. 1995). (2) Plant height should be about 110 em. This height facilitates larger panicles without subsequent lodging. (3) In addition to dominance, the over-dominance and additive effect should both be utilized in intersubspecific heterosis breeding. (4) The panicle size of the intersubspecific F 1 should be 20% larger than the current modern inbred cultivars and three-line system hybrids, with more primary branches on the panicles. (5) High photosynthetic efficiency and a high ratio of grain weight/leaf area should be achieved. A higher ratio of grain weight to leaf area indicates a more effective transformation from vegetative heterosis to reproductive heterosis in intersubspecific crosses. (6) To overcome poor grain filling ofintersubspecifc crosses, the grains of both parental lines should be plump. A recent study revealed that the well-filled grains of the indica-japonica F i s of 'Ce03 x Yang-Dao 4' and 'Lun-Hui 422 x 3037' resulted from the selected parents having superior grain plumpness. In contrast, poor grain filling occurred in the intersubspecific cross 'PC311 x IR36', in which both parents had poor grain plumpness. These results support that selecting parents with superior filling contributes to the success of interspecific rice hybrids (Q.S. Zhu, pers. commun.). (7) To ensure good grain quality, indica parental lines can be crossed with the long-grain javanica cultivars, and japonica parental lines with short-grain javanica cultivars. To obtain high production of hybrid rice seed, indica cultivars are generally developed as the male sterile line and japonica as the R line, because the flowering time of indica cultivars is earlier than for japonica cultivars under natural environments. However, some breeders have suggested utilizing intersubspecific heterosis with japonica as the male sterile line and indica as the R line to solve the grain quality problems (Li and Wu 1993). WCVs can also be used in three-line hybrid rice breeding. The progeny from crosses with many WCVs such as 'CPSL017', 'Calotoc', 'Ketan Nangka', and '02428' have unfavorable plant type or senescence. These WCVs are ofless value in the development of a new restorer line with wide compatibility (Zhang and Deng 1990; Zhang, Xie, and Chen 1992). Restoring gene(s) and wide compatibility gene(s) do not appear to be genetically linked (Cui et al.1993; Yan and Xue 1995). Hence, a parental line with both restoring gene(s) and wide compatibility gene(s) can be developed

J. 11 AND L. YUAN

98

through genetic recombination (Liao et al. 1991; Luo et al. 1994). For example, it is relatively easier to develop a restorer with wide compatibility from 'Lun-Hui 422', an improved WCV (Luo and Yuan 1989). VII. HYBRID RICE SEED PRODUCTION

A. China's Success

The success of hybrid rice in China is due to the successful development of parental lines and hybrids, and, even more important, the high yield of hybrid rice seed production. During the last 20 years there have been large gains in the total hybrid rice seed production and in hybrid productivity (Table 2.19). China's hybrid rice seed production can be generally divided into the following three developmental phases: Phase 1 (1973-1980): This phase was the early developmental stage of the techniques of hybrid rice seed production and multiplication. The average seed production was 0.45 t/ha. Table 2.19. Area, production, and productivity of hybrid rice (PI) seed production in China, 1976 to 1994. Source: Li and Yuan 1996. Year

Area (ha)

Production (t)

Productivity (t/ha)

1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994

85,126 200,613 270,120 218,434 172,353 110,400 154,600 138,800 104,733 87,667 100,533 154,067 135,827 171,866 191,987 124,733 139,389 105,959 117,111

23,365 72,822 128,847 118,282 119,268 73,857 140,531 179,052 148,145 145,045 200,563 309,674 221,058 336,170 431,970 280,898 333,976 234,593 261,392

0.274 0.363 0.477 0.542 0.692 0.699 0.909 1.290 1.415 1.654 1.995 2.010 1.627 1.956 2.250 2.252 2.396 2.214 2.232

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

99

Phase 2 (1981-1985): This phase was the establishment stage during which techniques for seed production and A line multiplication were perfected. Yield of hybrid rice seed production increased to 0.91 t/ha in 1982 and to 1.65 t/ha in 1985. Phase 3 (1986-present): This has been the exploratory stage for super-high-yielding techniques of hybrid seed production. The average seed production level per hectare was 2.25 tin 1991 and 2.4 t/ha in 1996 (Yuan 1998b), with the highest being 7.4 t/ha (Yuan 1996; Mao et al. 1998). B. Key Techniques 1. Choice of Favorite Climatic Conditions. Conditions favorable for normal flowering are a daily average of 24-28°C, 70-80% relative humidity, 8-10°C difference between the day and night mean temperatures, and sunny days with a breeze. Flowering should occur when the seasonal high temperature has ended and the low temperature season has not yet started (Xu and Li 1988). 2. Ensuring Flowering Synchronization. Heading date of the A line should be one or two days earlier than that of the R line. Currently the one-date-seeding technique for the R line is practiced for high-yielding hybrid seed production in China. The advantages are a large amount of effective pollen, a high effective spikelet ratio of AIR, a high pollen density of the R line, and more pollen grains on each stigma.

Methods for Determining the Seeding Interval for the Parental Lines. Three methods are primarily used to determine the difference in seeding date that synchronizes the A and R lines (Hunan Hybrid Rice Research Center 1993). For all three methods, the first sowing date of the R line is taken as the reference date. In the growth duration method, prior data concerning the difference in duration from seeding to initial heading between the A and R line are checked and used to determine the proper seeding date of both parents. This method is simple and easy to apply. In regions where temperature varies greatly during the vegetative growth period, however, the earlyseeded R line will have a different growth duration each year. For a seeding date of the A line adjusted only according to the growth duration, there will sometimes be a great discrepancy in the synchronization of flowering. Therefore, this method is only used in seasons or regions where temperature fluctuation is small.

100

J.

LI AND L. YUAN

The leaf number is used to determine the appropriate seeding dates since the leaf number is relatively stable in rice and the leaf count is a good record of the physiological age for rice plants. The leaf number on the main culm is used to determine the difference in seeding date between the two parents. More than 10 seedlings are required for reliable observation, and observations must be recorded every three days. A "three-ratings" criterion (i.e. 0.2 for just emerged, 0.5 for half emerged, and 0.8 for almost opened leaf) for quantifying the leaf age has been adopted in China's hybrid seed production practice (Yuan 1985). Counting starts when the first complete leaf emerges on the main culm. In the effective accumulated temperature (EAT) method, the EAT from $eeding to initial (10%) heading is relatively stable within a cultivar, but does differ with the seeding date. For rice plants, 12°C is generally used as the lower temperature limit and 27°C as the upper limit. The formula employed to calculate the EAT follows: A = I(T - H - L) where A is the EAT of the specific time duration (OC); T is the daily mean temperature (OC); H is the temperature above the upper limit (27°C), computed for only the days that the daily mean temperature is greater than 27°C; L is the lower limit temperature (12°C), computed for only the days that the daily mean temperature is greater than 12°C, and the accumulation of the temperature is carried out from the beginning to the end of a specific growth stage. When the EAT from seeding to initial heading is available for both the A and R lines, the seeding date for the parental line with a shorter growth duration may be determined based on the EAT difference. The EAT of a cultivar varies by region, therefore it is best to use locally recorded temperatures. At the beginning ofthe commercialization of hybrid rice in China, the EAT and the leaf number method were widely adopted to determine the seeding interval. Later practices demonstrated that the EAT method sometimes was unreliable because it relied on forecasted temperatures and there can be a difference in sensitivity to changes in temperatures between parental lines. One parental line may not change its growth duration in response to the EAT change. The leaf number method is more accurate, but the growth duration and leaf growth rate will vary for the spring seasons of different years because the temperature varies considerably between the spring seasons of different years in some regions such as Hunan, China. Predictions of the three methods described above are closely correlated and hence can be used complementarily to determine the seeding interval. Generally, the leaf number is used as the main method and the other two are used to provide supporting information, especially for China's early-season hybrid seed production. The growth duration method is effective for the single-cropping or late-season hybrid

101

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

seed production because the temperature is much less variable for different years. A recent study in India has confirmed that the leaf number is a reliable parameter for determining the seeding intervals (Viraktamath et al. 1998; Vijayakumar et al. 1998). Predicative regression formulas are available in China based on prior EAT, leaf number, and growth duration data. Prediction of Flowering Stage. The leaf growth rate of both parental lines of the seed production plots should be observed every few days to predict heading date. For hybrid rice seed production, even if the seeding interval between both parents is accurately determined, synchronization in flowering still might not be attained because of variations in temperature and/or differences in field management. The most effective and widely used method for prediction of heading date is by examining the developmental stages of young panicles. Based on their morphological features, the young panicles are classified into eight developmental stages (Table 2.20).

Table 2.20.

Stages of young panicle development in rice. Number of days before heading

Duration of days Stage

Female

Male

Features

Female

Male

I II

2 2-3

2 3-4

25-27 22-24

30-32 27-30

III

3-4

4-5

18-21

22-26

IV

5

6-7

18-25

19-22

V

3

3

12-15

16-19

VI

2

2

9-11

12-15

VII

6-7

7-9

8-9

9-11

VIII

2

2

Differentiation of first bract primordium Differentiation of primary branch primordium Differentiation of secondary branch primordium; young panicle is about 1 mm long and covered with white hairs Differentiation of stamen and pistil; appearance of glumes, young panicle is 0.5-1.0 cm long Formation of pollen mother cells; floret about 1-3 mm long, young panicle 1.5-5.0 cm long From prophase I of meiosis to formation of tetrad; floret about 3-5 mm long and young panicle about 5-10 cm long Filling phase of pollen; floret and panicle reach full length and color turns to green Mature pollen; panicles are to emerge shortly

2

2

102

J.

LI AND 1. YUAN

Adjustment ofFlowering Stage. Current A lines in China normally have a long blooming duration. The heading date of an A line should be one to two days earlier than the R line, in order to synchronize the peak anthesis duration of the two parental lines. Plots with predicted poor synchronization should have their heading date adjusted as early as possible, because earlier adjustment is more effective. Adjustment after Stage IV is only minimally effective. The requirement for perfect synchronization of flowering of the two parental lines means that: (1) a development of the pollen parent should be one stage earlier than the seed parent during Stage I, II, and III of the young panicle development; (2) the seed and pollen parents should be at the same stages during the three middle stages, i.e. Stage IV, V and VI; and (3) the seed parent should be slightly earlier than the pollen parent during the last two stages, i.e. Stage VII and VIII. Two adjustment measures are the flowering enhancing and flowering delay method. The flowering delay method gives a more effective adjustment and therefore it is generally used as the major method. Proper use of the two methods usually enables adjustments when difference in heading dates are five days or less. Larger differences in heading date cannot be adequately adjusted. One major flowering delay method is application of nitrogen fertilizer (120-150 kg/ha for the A line or 30-40 kg for the R line). Granular nitrogen may be applied to the deep root system of the faster-developing parent to delay the heading date approximately four to five days. If the labor force does not permit this, the plots to which the nitrogen will be applied can be drained to expose the mud surface. After one to two days for absorption of the nitrogen, the plot can be again flooded. The effect of this nitrogen application method cannot last long because some of the fertilizer may be washed away. For plots with poor synchronization and too much nitrogen, additional nitrogen should not be reapplied heavily to delay the development of young panicles. For Chinese rice hybrids, the young panicle development of most A lines can be promoted by drying, while that of most R lines will be delayed by drying. If the drying method cannot be employed, the rate of young panicle development speed may be slowed by cutting some roots. To promote young panicle development by about two days, solutions such as 12 g GA 3 plus 60 g KH zP0 4 can be sprayed on the leaves of the later parent. In some cases cutting leaves and roots and removing early flowering panicles of the male can effectively adjust the flowering date (Feng 1984). Lingaraju et al. (1998) reported that, by spraying GA 3 at 60 ppm at the full boot leaf stage, the flowering of 'IR58025A' can be advanced by three

2. HYBRID RICE: GENETICS. BREEDING. AND SEED PRODUCTION

103

to five days, and with an application of urea or phosphorus at 20 kg/ha, flowering can be delayed by two days. 3. Population Establishment for High-yielding Hybrid Seed Production. Population establishment for high-yielding hybrid seed production involves row ratio, planting density, and row orientation.

Row Ratio. The row ratio is the ratio of the male: female row numbers in hybrid seed production plots. The row ratio is adjusted according to the growth duration, growth vigor, the pollen load, and the plant height of the R line. The principles for determining the row ratio include maximization of the number of A line rows based on the pollen supply of the R line and maximization of the width of each A line row in order to reduce the shading of the A line by the R line, thus improving the microclimate of the field for growth and normal flowering of the A line. A row ratio of 1:8 to 1:10 or 2:14-16 (for early- or mid-maturing R lines) or 2:18-20 (for late-maturing R lines) is widely used at present in indica hybrid seed production, and 1:6 or 2:8-10 in japonica hybrid seed production. If the R line has adequate pollen, the row ratio may be increased even more (Li and Yuan 1996). Outside China it has been reported that for F1 hybrid seed production the best row ratios are: 1:6, 2:4, 2:8, 2:10, 2:12, and 3:10 in different seasons (dry or wet) or at different locations (Sahai and Chaudhary 1985; Sahai et al. 1987; Sharma and Virmani 1994; Singh et al. 1997; Prabagaran and Ponnuswamy 1997a; Singh et al. 1998). Planting Density. In China, about 45,000 hills/ha are generally needed for the R line. The plants are transplanted with two or three seedlings per hill and a spacing of 15 cm and 200-250 cm from one row of the restorer to the next, with the A line rows in between. For seed production plots planted with double rows of the R line, the double rows are spaced at about 17-20 cm with spacing 20-35 cm between the R line plants. The A line is transplanted, two seedlings per hill, at a spacing of 12 x 13.3 cm for a density of approximately 300,000 hills/ha of the A line plants (Huang et al. 1994). Row Orientation. Row orientation should be nearly perpendicular to the direction of the prevailing wind during the heading stage. This enhances the cross-pollination (Wan 1989).

104

J. LI AND L. YUAN

4. Improving Outcrossing Potential

Development of Male Sterile Lines with High Outcrossing Traits. Development of good outcrossing traits in male sterile lines is essential for high yield of hybrid rice seed. A lines with higher exsertion of the stigma, especially a double-sided exsertion rate, usually have higher outcrossing potential (Xu and Shen 1988; Tian 1991; Elsy et al. 1998). More detail on the outcrossing mechanism in rice is described by Virmani (1994a). GA 3 Application. GA 3 plays an important role in China's hybrid rice seed production. It can be used to adjust the physiological and biochemical metabolism of rice plants as described below. Its stimulation of elongation of the juvenile cells provides its major role by: (1) enlarging the angle between the flag leaf and the main culm by 15-20 degrees, (2) enhancing elongation of the three uppermost internodes, (3) increasing the panicle exsertion from the flag leaf sheath, (4) increasing the angle of opening of glumes when flowering, (5) increasing the stigma exsertion of the female parent, and (6) increasing the 1,000-grain weight. Because of its effect on panicle exsertion, the dosage of GA 3 applied in China was increased from 7.5-45 g/ha in the 1970s to 60-90 g/ha in the early 1980s to 180-270 g/ha by the late 1980s and 150-180 g/ha for the early 1990s. Some farmers use up to 300 g/ha (Duan and Ma 1992). Currently, GA 3 application guidelines recommend 150-180 g/ha when using a knapsack sprayer (Liu 1997) and 135 g/ha when using an ultralow-volume (ULV) sprayer. A typical spraying schedule is as follows: Knapsack sprayer: total GA 3 = 150-180 g/ha 1st spray: 40 ppm (30 g in 750 L/ha) 2nd spray: 80-100 ppm (60-75 gin 750 L/ha) 3rd spray: 80-100 ppm (60-75 gin 750 L/ha) ULV sprayer: total GA 3 = 135 g/ha 1st spray: 667 ppm (15 g in 22.5 L/ha) 2nd spray: 2667 ppm (60 g in 22.5 L/ha) 3rd spray: 2667 ppm (60 g in 22.5 L/ha) The time of emergence of 1-5 % panicles is the best stage for GA 3 application, but GA 3 can be applied until 10% panicle emergence. The spraying is best done from 7:00 to 11:00, with the next treatment administered between 15:30 to 19:00. GA 3 should not be applied during blooming or at noon. The spraying interval should be as follows: (1) Three applications of GA 3 on consecutive days, if started at 1-5% panicle emergence. The application should be from 7:00 to 11:00, from 15:30 to 19:00 of the same day, and from 7:00 to 11:00 of next day. (2) Two appli-

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

105

cations of GA 3 from 7:00 to 10:00 of two consecutive days if started at 10% of panicle emergence, (3) One application of GA 3 if sprayed at 30% panicle emergence. GA 3 has no effect after 50% of panicle emergence. Recent studies showed that application in the early morning gave higher seed yield than application in the late afternoon (Tong and Ma 1997) and the best time for the GA 3 application was two equal splits in consecutive days (dosage = 100g/ha) at 15-20% panicle emergence (Prabagaran and Ponnuswamy 1997b). GA 3 is dissolved in 75-90% alcohol (1 g in 30 to 50 ml) one week ahead of spraying, with 5 to 8 g neutral detergent!activator added per gram ofGA 3 • Plots are re-sprayed if there is rain within 6 h after a spray. Spray is applied with 3 to 5 em of standing water on the field and 30 to 37.5 kg of urea/ha or 2% urea is applied with GA:l if early senescence appears. The daily mean temperature during GA 3 application should be over 25°C for the best effect. The GA] dosage should be doubled if the daily temperature is 22°C. GA 3 is very costly outside China, so ULV sprayers should be used for GA] application. It was reported that the dosage of GA 3 can be reduced to 15-45 g/ha (lnt. Rice Res. Inst. 1992; Huang et al. 1994; Ahmed et al. 1997b). But the ULV sprayer should not be used if the wind velocity is above 3 m/sec. Because of the expense of GA 3 , substitutes are currently being sought. Mangiferin or 1.5-2.0% urea or 1.5% boric acid is as effective as GA:l in increasing hybrid seed set. A young leaf extract of Albizia amara may also be an alternative treatment to GA: l in hybrid rice seed production (Prasad et al. 1988; Singh and Sahoo 1997; Ponnuswamy and Prabagaran 1997).

Sllpplementary Pollination. Shaking the R line panicles using ropepulling or rod-driving during anthesis can greatly assist in the release of pollen grains from the anthers. This process is even more effective on calm days than on breezy days. When seed production plots are irregular in shape or uneven in topography, and where there is sufficient manual labor, the rod-driving method (using a bamboo stick to stir the canopy layer of the R or B lines) is recommended (Virmani and Sharma 1993). Under other conditions, the rope-pulling method is practiced. With this rope-pulling method, panicles of the R line are shaken by pulling a long nylon rope (about 4 mm in diameter) and walking against the wind at a speed of 1 to 1.5 m/sec. The rope should run parallel to the parental rows. These supplementary pollination procedures are generally conducted in the morning when the A line is flowering. If the R line is flowering but the A line is not in the morning, these procedures should not be

106

J. LI AND L. YUAN

used. In the afternoon, when the R line is still blooming, supplementary pollination should be continued even if the A line has closed its glumes. Generally, the supplementary pollination is repeated at intervals of 15 to 30 min three to five times daily until no pollen remains on the R line. It is not needed when the wind is stronger than a moderate breeze (Virmani and Sharma 1993). A recently established technique of supplementary pollination seems more efficient for increasing the outcrossing rate (P. J. Huang, pers. commun.). This method emphasizes that the best time to conduct supplementary pollination is at the peak of pollen shedding of the R line, instead of at 30-minute intervals. To predict the peak stage of pollen shedding, observations should be made every ten minutes and the number of florets blooming during this interval should be recorded. The R line is considered to be at the beginning of the anthesis peak stage when the average number of blooming florets per panicle is more than five within ten minutes. The best time for the supplementary pollination is within 30 minutes thereafter. The rod-driving method is more effective than the rope-pulling method because it creates a more even pollen distribution. Under supervision by a technician, the highest hybrid seed yield can be attained by performing the supplementary pollination simultaneously at multiple sites within a sizable seed production area. This will create a well-distributed pollen "fog" (Huang and Tang 1990). 5. Ensuring Purity

Isolation. To ensure purity of the hybrid seed, the hybrid seed production plots should be strictly isolated in space and time. An isolation distance of more than 100 m is generally necessary for (A x R) hybrid seed production. No other cultivars should be grown within this area during the same season except the pollen parent. The required isolation distance seems to vary with location and season (Prasad and Virmani 1989). IRRI scientists reported that at least 22-31 m of isolation distance was needed (Sharma et al. 1987; Muker and Sharma 1991). Generally, time isolation requires a period of 20 days, Le., unwanted cultivars within the 100 m distance from the seed parent should flower at least 20 days earlier or later than that of the pollen parent. Under some conditions, topographical features such as hills, woods, rivers, or tall crops (e.g. maize, sugarcane, and sorghum) that cover more than 30 m might provide necessary isolation. To produce a small amount of seed for the replicated yield trials or other purposes, an isolation cloth or plastic sheet of at least 2 min height is normally used as a barrier to prevent unwanted pollination.

107

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

Roguing. Parental lines are easily contaminated during the hybrid seed production, so it is necessary to thoroughly rogue the fields. Purity of hybrid rice seed used for commercial production in China must be over 98%. This requires the purity of the R line and A lines to be more than 99% (Table 2.21). Roguing should be done two or three times, including before heading, at the initiation of heading, and before harvest. Any maintainer and semisterile plants in the A line rows, and all other off-type plants in both the male and female rows should be completely removed from the field. Characteristic features used to guide roguing include: (1) off-type plants-This is based on color of the leaf sheath and leaf, size of leaf blades, growth and developmental status, plant type, plant height, and growth duration; (2) maintainer florets phenotype-The basal part of the maintainer's panicles normally exerts out of the flag leaf sheath and the anthers of the maintainers should be yellow, plump, and completely dehiscent after anthesis; (3) anther phenotype-Anthers of semi-sterile plants are slightly larger than sterile anthers of the A line, yellowish in color, and partly dehiscent after flowering; the nondehiscent anthers of semi-sterile plants become dark yellow several hours after anthesis. 6. Field Management

Raising Productive Seedlings. For more productive seedlings with multiple tillers, the seed parent should be seeded sparsely and evenly at a seeding rate less than 150 kg/ha. When these seedlings have two leaves, 70-100 kg/ha ofurea should be applied to promote tillering. This topdressing should be repeated seven days before transplanting. Spraying 40-60 g of MET (Multi-effect Triazole or Paclobutrazol) with 900 kg/ha water when seedlings have 1.1 leaves controls seedling height and promotes tillering. The seedlings of A lines are generally transplanted Table 2.21.

The minimum standards for nucleus seeds (NS) and foundation seeds (FS). Source: Yuan 1985. Seed grade

Purity

Cleanness

Germination

Moisture

Sterility

Restoring

(%)

(%)

(°lc,)

(%)

(%)

(%)

A line

NS FS

100 >99.8

>99.8 >99.5

>93 >93

<13.5 <13.5

100 100

B line

NS FS

100 >99.8

>99.8 >99.5

>98 >98

<13.5 <13.5

R line

NS FS

100 >99.8

>99.8 >99.5

>98 >98

<13.5 <13.5

Line

>85 >85

108

J. LI AND L. YUAN

at about six leaves. At present, more and more farmers are practicing the two-step method. This method includes raising R line seedlings at a seeding rate of 1500 kg/ha and transplanting seedlings (heel-in) temporarily with 2-3 seedling/hill at leaf number 2.5 and 10 x 13.3 em spacing. The final transplanting of early- or mid-maturing R lines should be conducted at the seventh to eighth leaf number, and late-maturing R lines at the eighth to ninth leaves.

Water and Fertilizer Management. All field management after seeding aims to increase the total floret number and to assure the flowering synchronization. After planting, the soil moisture should cause the parental line seeds to germinate in a short time for better subsequent synchronization. When the third leaf of the female parent appears, nitrogen should be applied to promote the plant growth. Tillering fertilizer should be applied to produce more tillers at the early tillering stage. The A line should have 3.1-3.5 million tillers/ha within 30-32 days after emergence. This facilitates production of 2.5 million effective panicles/ha. Shortly before panicle differentiation, healthy plants should have leaves with slightly light green color. If the color is too light, some N, P, and K fertilizers should be applied to promote conversion of the vegetative growth to reproductive development, which helps the formation of large panicles. However, too much nitrogen will likely make the upper three leaves droopy and fragile and have negative impact on the spread of the pollen and flowering synchronization. The best N:P:K ratio is 2:1:1.5 for high-yielding hybrid seed production. Application of this fertilizer at the later stages enhances acceptivity of the pollen grains by the A line and also increases the pollen shedding percentage of the R line. Therefore, 60-90 kg/ha of urea and 45-60 kg/ha of KCl should be applied at Stage V of the young panicle development, along with four applications of KH zP0 4 and boron fertilizer shortly before or after heading at respective dosages of 15 kg/ha and 1.5 kg/ha. Before heading, water should be kept on the plot. If the plot dries out, even temporarily, at the peak tillering time, poor synchronization will result from the different or reverse reaction of the plant development of the A and R lines. After the grain-filling stage begins, the seed production plots should be irrigated at intervals to fill the grains and protect the plants from diseases. C. Specifics for CMS Line Multiplication The techniques of A line multiplication are basically the same as for hybrid seed production, but there are some differences. Stricter isolation

2. HYBRID RICE: GENETICS. BREEDING, AND SEED PRODUCTION

109

is required. At least 500 m of isolation is needed to ensure maintenance of purity during multiplication of the CMS line. Seeding intervals vary with genotypes. An A line and its maintainer line are like twins and do not differ greatly in their growth duration. However, sporophytic indica A lines are four to six days later in heading than their B lines, so these A lines should be seeded earlier. The first seeding of the B line should be when the A line has 1.5 leaves. The second seeding of the B line should be when the A line has 2.5 to 3 leaves. The gametophytic japonica A lines are similar to their corresponding B lines in days to heading, so in these cases the first seeding of the B line is at the same time as the A line. The second seeding of the B line should be when the leaf number of the A line is 1.5 to 2.0 leaves, about five to seven days after the first seeding. Regardless of their seeding dates, both the A and B lines are transplanted on the same day. There is little difference in plant height between the A line and the corresponding B line. Due to its later seeding, the B line is inferior to the A line in tillering capacity and growth vigor. Therefore, their row ratio should be smaller than for the F 1 seed production. The widely adopted B/A row ratio for CMS line multiplication plots is 1:3 or 2:5. However, 2:6-10 was reported to be the optimum row ratio for CMS line multiplication (Ahmed et al. 1997b; Singh et al. 1998). To maximize the seed yield of the multiplication plots requires promoting the growth of the B line. The B line generally needs to be transplanted with soil-intact seedlings to shorten duration of the transplanting shock. A quick-releasing fertilizer should also be applied to the B line. An ideal population infrastructure for CMS line multiplication is as follows:

A line Basic tillers: 1.8-2.1 million/ha Maximum tillers: 4.5 million/ha Productive panicles: 3.0-3.6 million/ha Total florets: 300-360 million/ha B line

Basic tillers: 0.4-0.6 million/ha Maximum tillers: 2.1 million/ha Productive panicles: 1.2-1.4 million/ha Total florets: 105-120 million/ha

Panicle ratio: B:A

= 1:3

Floret ratio: B:A = 1:2.5-3

110

J. 11 AND L. YUAN

D. Purification of Parental Lines 1. Deterioration of Parental Lines. Deterioration of parental lines leads to deterioration of F 1 hybrids as a result of decreases in seed set and uniformity. Male sterile lines may deteriorate because of either segregation for plant type, maturity and other traits, or due to a breakdown in sterility. Other traits may contribute to the deterioration of an A line including poor fertility restorability, decreased combining ability, increased proportion of unopened florets, reduced proportion of stigma exsertion, less desirable flowering traits, and diffused anthesis time. Similarly, B and R lines may deteriorate and cause maintaining and restoring abilities to become weaker. Their combining ability may decrease with insufficient pollen supply and reduced pollen shedding.

2. Causes of Admixture of Parental Lines. Admixture of parental lines may be due to pollen contamination from outcrossing, mechanical mixture during harvest and postharvest procedures such as threshing, drying, cleaning, transportation, and storage, or genetic variation present in the parental populations.

3. Purification Method of AIB/R Lines: Several methods are available for purification of A, B, and R lines. The simplest and most effective method with regard to practical utility involves the use of testcross, identification, and multiplication nurseries, during the following four steps:

Selection of Elite Plants. Elite individual plants of the three parental lines are selected based on agronomic traits, sterility, and resistance to disease and pests. Testcross and Backcross Nurseries. Paired crosses are made and the individual eMS line plants selected are testcrossed and backcrossed to the R line and to the B line. The number of paired crosses depends on available labor. In general, a minimum of 50 pairs of (A x B) are required, with each pair producing more than 100 hybrid seeds. Likewise, 50 pairs are required for (A x R) combinations, but each should give more than 200 hybrid seeds. Identification Nursery. Three nurseries are used for identification. (1) The sterility identification nursery must be a well-isolated plot with the A line and its B line planted in pairs in the plot. At the initial heading stage, the male sterility of every plant of the A line should be evaluated. If the A line has uniform traits, good flowering behavior, necks

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

111

that are not enclosed or only slightly enclosed, and the percentage of male sterile plants as well as the degree of sterility is 100%, these lines are retained with their corresponding B line. (2) Heterosis evaluation nursery consists of about 100 F 1 plants for each pair of (A x R) crosses. The F 1 plants are evaluated for heterosis including growth vigor, tillering capacity, percentage of productive tillers, seed set rate, uniformity, resistance to stress conditions, and grain yield. The best of these (A x R) crosses are then selected. (3) In the R line evaluation nursery, about 100-200 plants of the R line for each of the F 1 s are grown in isolation. The R lines are evaluated for purity, uniformity, flowering behavior, and the performance of the R lines and of their hybrids. The best male families (R lines) are chosen and allowed to set seed.

Bulk Multiplication. Seeds of the selected A and B lines are separately harvested in bulk and sown in isolation for the core seed production of the A and B lines. The selected restorer families are also harvested in bulk and the seeds of each are sown in another isolated plot for core seed production of an R line. VIII. FUTURE PROSPECTS

World population increased from two billion in 1930 to five billion in 1987 and is likely to reach eight billion in 2020 (Beachell 1989). Over the last 30 years the population in Asian rice-growing countries, where more than 90% of the world rice crop is produced and consumed, has increased by 60%. Fortunately, rice production in these countries has doubled due to the spread of modern inbred and hybrid rice cultivars. Although the rice research community is proud of this remarkable progress in increasing productivity, there is no reason for complacency. Rice production has to increase 1. 7% annually to meet the growing demands, despite the fact that the rice-growing area continues to decline. Since 1989, the global rice production has plateaued at the level of 520 million 1. Annual increase of rice production was only 1.8% during the 1985-1993 period, compared to 2.8% during 1975-1985, and 3.6% for the prior decade (Hossain 1996). It is clear that the food crisis will reemerge in some rice-growing countries without new technical breakthroughs for rice production. Hybrid rice technology developed in China must be transferred to other countries, even as China continues to develop improved methods for producing hybrid rice in the face of its own increasing population. The following are the main objectives for future investigations of hybrid rice production.

112

J. LI AND L. YUAN

A. Breeding of Diverse Parental Lines 1. Further Utilization of Rice Germplasm. Currently, only 30/0 of 1,000

accessions of the germplasm of hybrid rice have been used commercially (Luo and Yuan 1990; Ying 1994). Fortunately, disruptive genetic vulnerability has not occurred in China's hybrid rice. But there is a definite risk of genetic vulnerability due to pleiotropy, like the southern maize leaf blight epidemic during the 1970s in the United States (Tatum 1971). This risk exists because about 93% of the A lines used in commercial hybrid production in China belong to the WA-type. As an example ofthe potential dangers facing the current commercial hybrid rice, 'Shan-You 63', one of the most popular rice hybrids, has already lost its resistance to rice blast resulting in drastic reduction in yield in mountainous regions. Diversification to broaden the rice germplasm base seems essential to solving this problem. Already a series of hybrids such as 'Gang-You 22', 'Shan-You-Duo-Xi 1', and 'K-You 3' have been released to replace 'Shan-You 63' in China in recent years (Wang 1996). To address this problem, Virmani et al. (1986) considered identification and use of additional sources of cyto-sterility critical to preventing genetic vulnerability of the three-line system for hybrid rice to disease or insect epidemics. Some new cytoplasmic sources for male sterility have also been identified, such as 'V20B' (an A line with different male sterile cytoplasm from WA-tye A lines), 'Kalinga' (Pradhan et al. 1990), CMS-ARC (Virmani and Dalmacio 1987; Virmani et al. 1989), 'IR66707A' from O. perennis (Acc. 104823) (Dalmacio et al. 1992, 1993, 1995), O. glumepetala and 'IR62829B' (Int. Rice Res. Inst. 1995). Three new and diverse CMS sources -one is from O. rufipogon and two are from O. nivara-were recently identified using substitution backcrossing and the embryo rescue technique. Among them 'RPMS1' and 'RPMS2' showed gametophytic male sterility with a restorer reaction different from WA-type CMS lines (Hoan 1993; Hoan et al. 1997a,b, 1998). In addition, two novel lines were produced from BT-type CMS sources using the asymmetric protoplast fusion technique (Blackhall et al. 1998) and, in India, some diversified CMS sources were identified from crosses O. nivara (105343) x C045, O. barthii (100934) x IR50, and O. nivara (101508) x IR64 (Rangaswamy and Jayamani 1998). Besides the diversification of cyto-sterility sources, a recent study has revealed that introgression of gene(s) of agronomic importance from wild rice species such as O. rufipogon into the parental lines for hybrid rice might further enhance the productivity of hybrid rice (Xiao et al. 1996b, 1998; Tanksley and McCouch 1997). 2. Further Improvements of Adaptability to Environmental Stresses. CMS lines derived from the Chinese WA sterile plant, such as 'V20A(B)' and 'Zhen-Shan 97A(B)', are not adaptable to the tropics or subtropics

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

113

(Suprihatno et aL 1994; Yogeesha and Mahadevappa 1996). Dependable and diverse parental lines need to be developed for tropical environments. Since rice hybrids generally have better drought resistance as well as better yield performance under drought conditions, upland hybrid rice should be studied and developed for the increasing proportion of non-irrigated rice land within and outside China. 3. "New Plant Type" and Super-high-yielding Hybrid Rice Breeding. Integration of heterosis and superior plant type should further increase yield potentiaL If the new plant type, in association with direct seeding, can increase non-hybrid rice yield by 20-25%, and reach 13t/ha, using the new plant type in hybrid rice breeding might give an additional 20-25% yield advantage and increase this yield to 15 t/ha of grain (Yang 1987; Khush and Aquino 1994; Khush 1995; Khush and Peng 1996; Pingali et al. 1997). Breeding for "new plant type (NPT)" was initiated at the International Rice Research Institute in 1989. The donor germplasms are mostly from bulu cultivars, belonging to javanica or tropical japonica. Breeding objectives for this new plant type were described as follows: (1) low tillering capacity of only three to four tillers when directed seeded; (2) 200-250 grains per panicle; (3) no unproductive tillers and harvest index of about 0.6; (4) sturdy stems; (5) dark green, thick, and erect leaves; (6) a vigorous root system; (7) 90 cm height; and (8) 100-130 day growth duration (Peng et al. 1994; Khush 1995). Based on the NTP design, some promising lines such as 'IR65598-112-2' have been developed. However, there are still a number of constraints, including low biomass production, poor grain filling, pest susceptibility, and early flag leaf senescence (Khush, Peng, and Virmani 1998). Yuan (1997b, 1998a) proposed the criterion for super-high-yielding hybrid rice of 100 kg/ha yield per day, with the model plant type being: 100 cm plant height with a 70 cm culm length, long, erect narrow, V-shape, and thick uppermost three leaves, a moderately compact plant type and moderate tillering capacity, 5 g of panicle weight and about 2.7 million panicles/ha, a 6-6.5 leaf area index (LAI) ofthe upper three leaves, and 0.55 for the harvest index. The two-line hybrid 'Pei-Ai64s x E32' has proven this new strategy for superhigh-yielding hybrid rice breeding. This hybrid had a growth duration of 130 days, yielded 13.3 t/ha on a total area of 0.24 ha at three locations in Jiangsu Province of China in 1997, and reached a daily gain of 100 kg/ha, the criterion for super-high-yield breeding (Yuan 1998a). B. Molecular Breeding Recent advances in rice biotechnology, particularly molecular mapping and genetic transformation, have opened new avenues in hybrid rice breeding.

114

J. 11 AND 1. YUAN

1. Marker-assisted Breeding. A PCR-mediated method for selecting Rf-

1 restorer lines recently used the close linkage between the Rf-l locus and a particular PCR-amplified fragment (Ichikawa et al. 1997). It has been proposed that a new molecular marker (OSRRf) closely linked to the nuclear restorer gene (Rf-l) be applied not only in the development of new R lines and B lines, but also for the purity management of the hybrid rice seeds (Akagi et al. 1996; Fujimura 1996). A defective atp6 gene conferring the male sterility in BT type CMS lines was reported to be located on the mitochondrial genome, so the sequence of this gene was to design PCR primers for detection of CMS lines. For the two-line system, a RAPD marker linked with the PGMS of'Nong-Ken 58s' and an AFLP marker (AF3) closely with tmsl in '5460s' have also been identified (Wang et al. 1995b; B. Wang, pers. commun.). The use of microsatellite markers has also been suggested as a method for screening CMS resources and removing unfavorable alleles and heterozygotic patterns from parental lines (Liu and Wu 1996; Liu et al. 1997b). It is predicted that the MAS (marker-assisted selection) technique will assist in the determination of genetic diversity, identification and accumulation of heterotic gene(s), and improvement of plant traits. The ability of this technique to detect the presence or absence of any number of alleles of interest in one screening is particularly attractive to rice breeders. 2. Other Potential Biotechnologies. Tissue culture and genetic transfor-

mation techniques will be more extensively employed in future hybrid rice breeding. Protoplast fusion enables the direct transfer of CMS into elite rice breeding lines as well as the development of alloplasmic lines having cytoplasms from various wild species and related genera. Plant Genetic Systems in Belgium has pioneered the research and development of a genetically engineered "male sterility gene" for producing hybrid rice. The genes for nuclear male sterility (barnase) and fertility restoration gene (bastar) were cloned and transferred to crops including tobacco and rape (Lasa and Bosemark 1993). Two more organizations, Paladin Hybrids in Canada, and ICI in the UK, have also patented genetically engineered dominant nuclear male sterility systems (Cutler 1991). These systems provide an opportunity to develop robust systems for two-line hybrid rice (Mariani et al. 1990 and 1992; Brar et al. 1994; Jefferson and Nugroho 1998).

C. Apomixis Breeding Since Navashin and Karpechenko in the 1930s demonstrated the value of apomixis in fixing F 1 heterosis, apomixis has been an attractive

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

115

research subject, because through the use of apomixis it is possible to develop true-breeding hybrids, produce high-quality pure seed without isolation requirements, and enhance the selection of a variety of more highly adapted genotypes (Solntzeva 1978). Apomixis, which has been criticized as a threat to the seed industry and ignored by the seed business in a number of developing countries (Koltunow et al. 1995), could be protected potentially by intellectual property rights and a patenting system or potentially through the engineering of technical tricks in apomixis breeding (Savidan 2000). If successful, the development of apomictic rice would also enable resource-poor farmers in developing countries to adopt high-yielding hybrid rice technology (Khush, Brar, and Bennett 1998). 1. Definition and Classification. "Apomixis," practically defined (Nogler 1984a) as asexual reproduction through seeds, can be classified into the following three major categories: (1) apospory; (2) diplospory; and (3) adventitious embryony (Savidan 2000). Apomixis can be either obligate or facultative. Apomixis is the predominant mode of reproduction in the obligate type, while apomixis is combined with sexual reproduction to some extent in the facultative type. Breeding facultative apomicts generally is more difficult than breeding obligate apomicts (Bashaw 1980a). Apomixis exists in more than 300 plant species from 35 families (Bashaw 1980a; Hanna and Bashaw 1987), including a number of fruit crops, and in the wild relatives of some important agronomic crops. Apomixis is also common in the grasses and in several polyploid plant species. Among the major cereals, maize, wheat, and pearl millet have apomictic relatives.

2. Inheritance. Genes governing apomixis may be governed by a single

dominant gene as in Panicum maximum (Savidan 1983), Ranunculus (Nogler 1975, 1984b), Poa pratensis (Matzk 1991), Brachiaria (Lutts et al. 1994; do Valle et al. 1994; do Valle and Savidan 1996; Miles and Escandon 1997), Amelanchier (Campbell and Wright 1996), F 1 maizeTripsacum hybrids (Leblanc et al. 1995), Tripsacum (Grimanelli et al. 1997), and Citrus (Parlevliet and Cameron 1959; Iwamasa et al. 1967). Apomixis has also been reported to be controlled by two or three genes in Parthenium argentatum (Powers 1945), Bothriochloa spp. (Harlan et al. 1964), Pennisetum ciliare (Taliaferro and Bashaw 1966; Gustine et al. 1989), and Poa pratensis (Funk and Han 1967) or a group of genes (Gerstel et a1. 1953; Savidan 1982). Apomixis is generally thought to be controlled by dominant genets), but some studies have indicated that a recessive gene or genes control apomixis in Paspalum notatum (Burton

116

J. LI AND L. YUAN

and Forbes 1960; Asker 1970), Panicum maxicum (Hanna et al. 1973), and Paa pratensis (Akerberg and Nygren 1959; Grazi et al. 1961). In 1977 Carman suggested that apomixis did not originate from specific apomictic genes or alleles but from asynchronously-expressed duplicate genes controlling female development. Apomixis probably is so complex that it cannot result from a single-locus mutation. The whole process might be divided into individual components and the mutations affecting each of those may be identified separately (Savidan 2000). Environmental factors such as temperature, photoperiod, and a number of chemical agents such as MH and gibberellic acid (den Nijs and van Dijk 1993), also affect the expression or stability of apomixis, especially facultative apomixis. 3. Breeding Approaches. Various apomixis breeding procedures were

proposed by Taliaferro and Bashaw (1966) and Bashaw and Funk (1987) for bufflegrass (Cenchrus ciliaris) based on a two-gene model, Pernes et al. (1975) and Hanna (1995) for guineagrass based on a one-dominantallele model, and Hanna (1995) based on a recessive-gene model. Several modified procedures for Brachiaria, Tripsacum, Paspalum, and Citrus are detailed in Savidan's review (2000). Peacock also proposed a synthetic lethal system in screening apomictic mutants. The first apomictic cultivar was the Japanese forage cultivar 'Natsukaze' (Sato et al. 1990). 4. Breeding Apomictic Rice. The basic requirements for fixing rice heterosis using apomixis should involve: (1) embryo development from nucellar cells or 2n embryo sac cell without meiosis; (2) obligate type of apomixis; (3) dominant inheritance involving one or a few gene(s); (4) normal endosperm development; and (5) stable expression of apomixis over environments (Sun and Cheng 1994). Major strategies for developing apomictic rice are (1) screening germplasm of tetraploid wild species as a source of apomixis and transferring the apomictic trait to rice cultivars; (2) inducing apomictic mutants in rice through mutagenesis (seeds and fertilized egg cells can be induced with gamma rays, X rays, EMS, and NED; and (3) use of molecular approaches. In the screening of 108 accessions of tetraploid Oryza species for apospory (multiple embryo sac development) and 86 accessions for diplosory (based on callose detection), including five related genera, no evidence of apomixis was found (Brar et al. 1995). Rutger (1992) also reported a similar negative result after screening 547 accessions of related wild rice species having the AA genome. Since screening and confirming rice apomixis from both rice cultivar and wild relatives failed, the most promising approach would be transfer of apomixis in

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

117

other grasses via protoplast fusion, direct DNA transfer, and genetic transformation. Molecular markers have been identified in progeny of maize (Tripsacum) and in crosses of sexual and apomictic wild species of Pennisetum. The ORSTOM-CIMMYT project is attempting to transfer apomixis from Trisacum dactyloides to maize through wide hybridization (Savidan et al. 1994). Two markers, one RFLP marker (UGT197) and one RAPD marker (OPC-04), were reported to be linked to apospory in Pennisetum (Ozias-Akins et al. 1993). In maizeTripsacum F 1 hybrid, five markers, Le. umc28, csu68, umc62, umc71, and CD0202, were also found linked to apomixis (Leblanc et al. 1995; Grimanelli et al. 1997). Cloning of gene(s) for apomixis from apomictic plant species including Tripsacum, Pennisetum, Brachiaria, and Cenchrus is presently under way. Once such genes become available, they could be introduced into elite breeding lines of rice using transformation technology. But genetic engineering of apomixis requires a detailed understanding of the genetic basis and the molecular mechanisms that control megasporogenesis, megagametogenesis, fertilization, and seed development. For the selection of progeny of rice apomicts, Virmani (1994a) proposed the following indicators: (1) identical maternal progeny from plants of cross-pollinated species, or progeny of F 1 crosses; (2) limited or no genetic variation in the F2 population of a cross between two distinct parents; (3) recessive genotypes from a cross of parents with recessive genes pollinated with a parent possessing a dominant marker gene; (4) unusually high seed fertility in aneuploids, triploids, and wide crosses normally expected to be sterile; (5) aneuploid chromosome number or structural heterozygosity remaining constant from parent to progeny; and (6) multiple seedlings per seed, multiple stigma, multiple ovules per floret, and double or fused ovaries. There are reports of mutants possessing twin seedlings per seed (Yuan et al. 1990; Sharma and Virmani 1990) and multiple pistillate ovaries (Suh 1988). A recent study by Shi et al. (1996a,b) showed that no apomictic phenomena existed in the rice line '84-15', which was once presumed to be an apomictic rice, as many papers from other crops referred to occasional, spontaneous, or induced haplo- or diplo- parthenogenesis as apomixis (Chen et al. 1988, 1992; Chen 1989; Asker and Jerling 1992). Besides '8415', China identified some rice materials with abnormal sexual reproduction process, including SAR-l (Zhou et al. 1991b, 1993b), HDAR (Cai et al. 1991; Yao et al. 1997), Cl00l (Guo et al. 1991; Wu et al. 1991), APIAPIV (Li, Deng, and Yuan 1990; Li and Yuan 1990); PDER (Ye et al. 1995), 322B (Huang 1988), PYl and PJ5 (Liu, Chen, and Zheng 1990), W3338 (Luo, Zhou, and Wang 1991), CDAR (Yan et al. 1991), and ABF (Zhao et al. 1992). However, cytoembryological studies indicated that

118

J. LI AND 1. YUAN

these materials were of little use in fixing rice heterosis (Sun and Liu 1996). All egg cells are produced via meiosis in SAR-1, C1001, and APIV. HDAR only has low-frequency abnormality of megasporogenesis and occurrence of double embryo sacs. A recently identified material 'TAR' (3n 36) is under study. In this material the meiosis of the megaspore was hindered and an unreduced embryo sac was produced via mitosis, in which embryos were formed via egg cell parthenogenesis, and endosperms were formed via pseudogamy of polar nuclei (Sun and Liu 1996). More and more new genetic tools and techniques may contribute to rice breeders' ability to handle apomixis, including tissue culture, embryo rescue, somatic embryogenesis, the Herr-clearing technique, and biochemical or molecular marker-aided selection, thus enhancing the breeding for apomixis. Either the discovery of practically usable source material(s) from the huge rice gene pool or introduction of apomixis gene(s) from other plant species to rice through biotechnologies will make this "utopian scheme" (Hermsen 1980) for using apomixis to fix rice heterosis come true.

D. Hybrid Seed Production 1. Mechanical Seed Production. Simplification of the labor-intensive

and complicated procedures of China's hybrid rice seed production practices should be explored. F 1 seed production in China utilizes the planting of the male sterile line and a pollinator line in alternative rows requiring much labor. Separation of selfing R line seeds and the F 1 seeds on the basis of difference in color, size, or other traits after planting the pollinator and male sterile plants as mixed seed has been proposed. A technique for mechanical separation of hybrid and inbred seed using a photoelectric seed-sorting apparatus and hull color has been patented, and in the future may provide the necessary hybrid seed purity (Barabas 1974; Kato et al. 1994; Suzuki et al. 1990). In addition, use of a femalesterile pollinator and incorporation of a herbicide-sensitive gene into a pollinator were suggested to facilitate mechanical harvest of the hybrid seed (Maruyama et al. 1991b). Herbicide sensitivity to chemicals such as bentazon (3-[1-methyethyl]-[IH]-2,1,3-benzothiodiazin-4-[3H]-one2,2-dioxide) has been introduced into parental lines to destroy the pollen parent before it sets seed, thereby eliminating R line seed contamination of the F 1 seed (MoTi 1984). There was also a report in China that herbicide-resistance genes were successfully introduced into parental lines of hybrid rice via particle bombardment transformation to protect the F 1 seed producing plants from herbicide injury (Huang et al. 1998).

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

119

Exploration of possible mechanization of other processes involved in hybrid rice seed production is under way, such as GA 3 application by an airplane and modified techniques of pollen collection, storage, and spraying (Li et al. 1996c). 2. Monitoring Purity of Hybrid Rice Seed. In the 1980s it was proposed

that some isozyme markers such as esterase isozymes might be used to monitor seed purity (Yan 1987; Glaszmann et al. 1987; Shi et al. 1988a,b; Li 1991b). Since the 1980s a practically usable technique incorporating isozyme markers and seed scanning has proven to be helpful in monitoring purity of hybrid rice seed at the Hunan Hybrid Rice Research Center. A PCR technique was recently used to determine seed purity of 'Shan-You 63' by using the P18 primer to amplify a specific band of 0.8 kb from the restorer line 'Ming-Hui 63' and thereby separate the true hybrids (Qian et al. 1996). Based on the low cost and effectiveness, B. Wang (pers. commun.) proposed that STS and AFLP markers also can be used to monitor the hybrid rice seed purity. It is predicted that seed purity can be monitored more accurately and economically by using molecular markers. E. Socioeconomic Impact

Apart from the technological aspects, the success of hybrid rice technology in China is due primarily to its profitability and government support. Other countries with a high labor-land ratio and a high proportion of irrigated area, such as India, Indonesia, the Philippines, Sri Lanka, and Vietnam, are likely to have the highest potential demand for hybrid rice technology. The availability of market opportunities for hybrid rice will not be a limiting factor for the private sector if government policies are made less restrictive and unfair competition from the public sector is eliminated. Hybrid rice could have the same catalytic effect on the hybrid rice seed industry that hybrid maize had on the seed industry in North America (Sehgal 1994). To popularize the use of hybrid rice, both the NARS and private sectors should identify target areas for hybrid rice cultivation or seed production. An effective hybrid seed production and distribution system must be established to stabilize the price of certified hybrid seed at a reasonable level and maintain hybrid seed purity in the long term. For example, the "seed-producing village" has been proved to be an effective practice for hybrid rice seed production both in China and India. However, policies for hybrid seed production and distribution of quality hybrid seeds are still on the drawing board in most countries other than China. With the successful establishment of

120

J. LI AND L. YUAN

a commercialization system for hybrid rice in countries, including China, it is probable that 18-20 million ha will be planted each year in the next three to five years. That would mean an increase of more than 18-20 million t of rice production, which would be worth US$ 1,800-2,000 million per year for the whole rice world.

LITERATURE CITED Ahmed, M. 1. 1997. Development and use of hybrid rice technology-Lessons learned from the Indian experience. Int. Rice Comm. News!. 46:75-83. Ahmed, M. 1., K. Krishnaiah, and E. A. Siddiq. 1997a. Enhancement of yield heterosis for sustenance of hybrid rice technology in India. p. 78-79. In: J. G. Coors, S. Pandey, M. V. Grinkel, A. R Hallauer, D. C. Hess, K. R Lamkey, A. E. Melchinger, G. Srinivasan, and C. W. Stuber (eds.), The genetics and exploitation of heterosis in crops. Proc. Int. CIMMYT Symp., 17-22 August 1997. ASA, CSSA, SSSA, Madison, WI, CIMMYT. Mexico. (Abstr.) Ahmed, M. 1., S. Singh, C. H. M. Vijayakumar, and M.S. Ramesha. 1997b. Standardization of row ratios and GA3 application in hybrid rice seed production in India. p. 122-123. In: J. G. Coors, S. Pandey, M. V. Grinkel, A. R Hallauer, D. C. Hess, K. R. Lamkey, A. E. Melchinger, G. Srinivasan, and C. W. Stuber (eds.), The genetics and exploitation of heterosis in crops. Proc. Int. CIMMYT Symp., 17-22 August 1997. ASA, CSSA, SSSA, Madison, WI, CIMMYT. Mexico. (Abstr.) Ahokas, H., and Hockett, E. A. 1977. Male sterile mutants of barley. IV. Different fertility levels of msg9ci (cv Vantage) an ecoclinal response. Barley Genet. News!. 7:10-11. Akagi, H., M. Sakamoto, C. Shinjyo, H. Shimada, and T. Fujimura. 1994. A unique sequence located downstream from the rice mitochondrial atp6 may cause male sterility. Current Genet.25:52-58. Akagi, H., T. Taguchi, and T. Fujimura. 1995. Stable inheritance and expression of the CMS traits introduced by asymmetric protoplast fusion. Theor. App!. Genet. 91:563-567. Akagi, H., Y. Yokozeki, A. Inagaki, A. Nakamura, and T. Fujimura. 1996. A codominant DNA marker closely linked to the rice nuclear restorer gene, Rf-1, identified with interSSR fingerprinting. Genome 39:1205-1209. Akbar, M., and T. T. Yabuno. 1975. Breeding for saline-resistant varieties to rice. III. Response of F 1 hybrids to salinity in reciprocal crosses between Jhona 349 and Magnolia. Japan. J. Breed. 25:215-220. Akerberg, E., and A. Nygren. 1959. Poa pratensis, trivalis, palustris, compressa und verwandte Arten. p. 392-418. In: H. Kappert and W. Rudorf (eds.), Ziichtung der Futterpflanzen. Ali, A. J., M. Rangaswamy, R Rajagopalan, S. E. N. Mohamed, and T. S. Manickam, 1998. TNRH16: A salt-tolerant rice hybrid. Int. Rice Res. Notes 1998(2):22. Ali, J., E. A. Siddiq, F. U. Zaman, M. J. Abraham, and 1. Ahmed. 1995. Identification and characterization of temperature sensitive genic male sterile sources in rice (Oryza sativa L.). Indian J. Genet. Plant Breed. 55:243-259. Ali, S. S., and M. G. Khan. 1998. Hybrid rice research in Pakistan. Int. Rice Res. Notes 1998(2):17. Anandakumar, C. R, and S. R Sreerangasamy. 1986. Heterosis and inbreeding depression in rice. Oryza. 23:96-101.

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

121

Araki, H., K. Toya, and H. Ikehashi. 1988. Role of wide-compatibility genes in hybrid rice breeding. p. 79-83. In: Hybrid rice (Proc. Int. Symp. on Hybrid Rice, October 6-10,1986, Changsha, China). Int. Rice Res. Inst., P.O. Box 933,1099 Manila, Philippines. Armenta-Soto, J. L. 1988. Agronomic performance of hybrid rice in northern Mexico. p. 291. In: Hybrid rice (Proc. Int. Symp. on Hybrid Rice, October 6-10, 1986, Changsha, China). Int. Rice Res. Inst., P.O. Box 933, 1099 Manila, Philippines. Asker, S. 1970. Apomixis and sexuality in the PotentjJla argentea complex. II. Crosses within the complex. Hereditas 66:189-204. Asker, S. K, and 1. Jerling. 1992. Apomixis in plants. CRC Press, Boca Raton, Florida. Aswathanarayana, S. c., and M. Mahadevappa. 1991. Method of application of gametocides in hybrid seed production of rice. J. Maharashtra Agr. Univ. 1991(2):198-200. Aswathanarayana, S. C., and M. Mahadevappa. 1992. Effects of gametocides in hybrid seed production ofrice. J. Maharashtra Agr. Univ. 1993(1):14-16. Athwal, D. S., and S. S. Virmani. 1972. Cytoplasmic male sterility and hybrid breeding in rice, p. 615-620. In: Rice breeding. Int. Rice Res. Inst., P.O. Box 933, 1099 Manila, Philippines. Bai, D. 1., and X. H. Luo. 1996. Liangyou Peite, a new two-line hybrid rice released in China. Int. Rice Research Notes 1996(1):42-43. Bai, S. N., and K. H. Tan. 1990. The discovery of the specific protein in HPGMR rice. p. 245. The 5th Symp. of Chinese Society of Plant Physiology. (Abstr.). Bai, S. N., and Y. H. Xiao. 1988. A study of the root growth and respiration of hybrid rice. Acta Agron. Sinica 1988(1):53-59. Barabas, Z. 1962. Observation of sex-differentiation in sorghum by use of induced male sterile mutants. Nature 195:257-259 Barabas, Z. 1974. U.S. Patent Number: 3842538. Barwale, B. R. 1994. Hybrid rice: Food security in India. Macmillan India Ltd. Bashaw, E. C. 1980a. Apomixis and its implication in crop improvement. p. 45-68. In: W. R. Fehr and H. H. Hadley (eds.), Hybridization of crop plants. Am. Soc. Agron., Madison, WI. Bashaw, K C. 1980b. Registration of Nueces and Llano Buffelgrass. Crop Sci. 20:112. Bashaw, K c., and C. R. Funk. 1987. Breeding apomictic grasses. In: W. R. Fehr (ed.), Principles of cultivar development: Vol. 2. Crop species. Macmillan, New York. Beachell, H. M. 1989. The need for a global rice research system. p. 315-324. In: Progress in irrigated rice research. Int. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines. Bennett, M. D., and W. G. Hughes. 1972. Additional mitosis in wheat pollen induced by Ethrel. Nature 240:566-568. Bharaj, T. S., S. S. Bains, G. S. Sidhu, and M. R. Gagneja. 1991. Genetics of fertility restoration of wild abortive cytoplasmic male sterility in rice Oryza sativa 1. Euphytica 56:199-204. Bharaj, T. S., S. S. Virmani, and G. S. Khush. 1995. Chromosomal location offertility restoring genes for 'wild abortive' cytoplasmic male sterility using primary trisomics in rice. Euphytica 83:169-173. Blackhall, N. W., H. Akagi, T. FujimakL J. P. Jotham, M. R. Davey, J. B. Power, and E. C. Cocking. 1998. Diversification of cytoplasmic male sterility systems through somatic cell hybridization. In: S.S. Virmani, KA. Siddiq, and K. Muralidharan (eds.), Advances in hybrid rice technology (Proc. 3rd Int. Symp. Hybrid Rice, Hyderabad, India, October 1996). Int. Rice Res. Inst., P.O. Box 933, 1099 Manila, Philippines. Bollich, C. N. 1957. Inheritance of several economic quantitative characters in rice. Diss. Abstr. 17:1638. Bollich, C. N., J. N. Rutger, and B. D. Webb. 1992. Developments in rice research in the United States. Int. Rice Comm. News!. 41:32-35.

122

J. LI AND L. YUAN

Bollich, C. N., and B. D. Webb. 1973. Inheritance of amylose in two hybrid populations of rice. Cereal Chern. 50:631-636. Bong, B. B., and V. P. Singh. 1993. Cooking quality characteristics of hybrid rice. Indian J. Genet. Plant Breed. 53:8-13. Borkakati, R, and S. S. Virmani. 1993. Inheritance of a thermosensitive genic male-sterile mutant of indica rice. Rice Genet. Newsl. 10:92-94. Borkakati, R P., and S. S. Virmani. 1996. Genetics of thermosensitive genic male sterility in rice. Euphytica 88:1-7. Brar, D. S., R M. Elloran, M. C. Lanuang, V. S. Tolentino, and G. S. Khush. 1995. Screening wild species of rice for apomixis. Philippine J. Crop Sci. 20:6. Brar, D. S., T. Fujimura, S. R McCouch, and F. J. Zapata. 1994. Application of biotechnology in hybrid rice. p. 51-62. In: S. S. Virmani (ed.), Hybrid rice: New developments and future prospects, Int. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines. Brar, G. S. 1982. Male sterility in sesame. Indian J. Genet. Plant Breed. 42:23-27. Brewbaker, J. L. 1964. Agricultural genetics. Prentice-Hall, Englewood Cliffs, NJ. Brown, F. B. 1953. Hybrid vigor in rice. Malaysian Agr. J. 36:226-236. Bruce, A. B. 1910. The mendelian theory of heredity and the augmentation of vigour. Science 32:627-628. Burton, G. W., and I. Forbes. 1960. The genetics and manipulation of obligate apomixis in common bahiagrass (Paspalum notatum Flugge). p. 66-71. In: Proc. the VIIIth Int. Grassl. C~ngr., Univ. of Reading, U.K. CaL D. T., P. F. Ma, H. X. Guan, J. L. Yao, Z. A. Wang, and H. Zhu. 1991. Study of high frequency apomixis in rice. J. Huazhong Agr. Univ. 10:223-227. CaL H. W., X. Q. Wang, K. S. Cheng, and Y. Z. Zhang. 1992. Classification of Asian rices by esterase isozymes. Southwest China J. Agr. Sci. 1992(5):19-24. Campbell, C. S., and W. A. Wright. 1996. Apomixis, hybridization, and taxonomic complexity in eastern north American Amelanchier (Rosaceae). Folia Geobotania & Phytotaxonomica 31 :345-354. Capinpin, J. N., and P. Singh. 1938. A study of varietal crosses and hybrid vigor in rice. Philippine Agr. 27:255-277 Carman, J. G. 1997. Asynchronous expression of duplicate genes in angiosperms may cause apomixis, bispory, tetraspory, and polyembryony. BioI. J. Linn. Soc. 61:51-94. Carnahan, H. L., J. R Erickson, S. T. Tseng, and J. N. Rutger. 1972. Outlook for hybrid rice in the USA. p. 603-607. In: Rice breeding, Int. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines. Castle, W. E. 1946. Genes which divide species or produce hybrid vigor. Proc. Nat. Acad. Sci. (USA) 32:145-149. Chalam, G. V., and J. Venkateswarlu. 1965. Introduction to agricultural botany in India. Vol. I. Asia Publish. House, New Delhi. Chan, Y. W., and C. H. Cheah. 1983. Evaluation of etherel as a selective gametocide in the breeding ofrice (Oryza sativa L.). SABRAO J. 15:149-155. Chandraratna, M.F. 1964. Genetics and breeding of rice. Longman, London. Chang, T. M. 1974. Studies on the inheritance of grain shape of rice. J. Taiwan Agr. Res. Inst. 23:9-15. Chang, T. T., C. C. Li, and O. Tagumpay. 1973. Genotypic correlation, heterosis, inbreeding depression and transgressive segregation of agronomic traits in a diallel cross of rice (Olyza sativa L.) cultivars. Bot. Bul. Acad. Sinica (Taipei) 1973(2):83-93. Chang, T. T., Y. B. Pan, Q. R Chu, R Peiris, and G. C. Loresto. 1991. Cytogenetic, electrophoretic, and root studies of javanica rices. p. 21-32. In: Rice genetics II. Proc. 2nd Int. Rice Genetics Symp., Int. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines.

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

123

Cheema, A. A., and M. A. Awan. 1985. Heterosis, gene action and combining ability study of yield and some of the yield components of four-parent diallel cross in rice. Pak. J. Sci. Ind. Res. 28:175-178. Cheema, A. A., M. A. Awan, G. R. Tahir, and M. Aslam. 1988. Heterosis and combining ability studies in rice. Pakistan J. Agr. Res. 1988(1):41-45. Chen, C. L., Z. C. Deng, P. F. Ma. 1992. The characteristics of photosynthesis and transportation and distribution of 14C photosynthates in hsien keng rice hybrid combination. J. Huazhong Agr. Univ. 11:312-316. Chen, J. G., G. Q. Song, S. H. Qu, and y. Zhou. 1998. Direct and maternal genetic effects on grain quality characters in early hybrid rice. Chinese J. Rice Sci. 1998(2):79-84. Chen, J. S. 1989. Identification of apomictic inheritance using dominant marker genes. Genet. Manipulation Plants 1989(1):19-23. Chen, J. S., Q. S. Jia, L. Yin, and W. J. Cao. 1992. Study on fixing heterosis in hybrid rice. p. 69-82. In: J. S. Chen (ed.), Collection of papers on breeding of apomictic rice. China Agr. Sci. Tech. Press. Chen, J. S., Y. Lin, and C. W. Jun. 1988. A study of apomixis in rice. Genet. Manipulations Crops News!. 1988(2):3-10. Chen, L. B., X. Z. Li, and G. Q. Zhou. 1993. Effects of temperature on the sterility of rice in photo-sensitive and thermosensitive male sterile lines. Acta Agron. Sinica 1993(1):47-54. Chen, L. B., G. Q. Zhou, and X. X. Huang. 1994. Effects of temperature and photoperiod on fertility and physiological activities of rice Annong S-1 and Hengnong S-1. Acta Bot. Sinica 36(suppl.):119-123. Chen, L. Y., G. T. Li, G. H. Liu, and C. L. Xiao. 1997a. Breeding of the newly-developed three-line hybrid rice Xingxiangyou 80. J. Hunan Agr. Univ. 23:316-319. Chen, P., and Y. H. Xiao. 1987. A preliminary study on the peroxidase activity of the anthers during the process of pollen abortion in HPGMR. J. Wuhan Univ. (HPGMR): 39-42. Chen, S. G., S. H. Cheng, Z. X. Sun, H. M. Si, S. K. Min, and C. S. Ying. 1997b. Selection of 064A, a WA type male sterile line of rice with wide compatibility. Sci. Agr. Sinica 1997(2):21-26. Chen, X. B., X. J. XU. and R. Q. Liao. 1991. Study on nutritional physiology of two-line system hybrid rice: 1. Characteristics of sink and source and flow of photosynthetic products in two-line system hybrid rice. Hunan Agr. Sci. 1991(1):7-9. Chen, X. F., and C. Y. Liang. 1991. Accumulation of hydrogen peroxide and lipid peroxidation in male-sterile anthers of Oryza sativa. Acta Phytophysiol. Sinica 17:44-48. Chen. X. F., and C. Y. Liang. 1992. Energy and activated oxygen metabolisms in anthers of Hubei photoperiod-sensitive genic male-sterile rice. Acta Bot. Sinica 34:416-425. Chen. X. H., and B. H. Wan. 1993. Studies on the relationship between photoperiodic response. thermoperiodic response and photoperiod-temperature sensitive male sterility in rice. Acta Agron. Sinica 1993(3):282-287. Chen, Y. W. 1985. Hybrid rice in China. J. Hunan Sci. Tech. Univ. 1985(1):1-9. Cheng, K. S. 1985. A statistical evaluation of the classification ofrice cultivars into hsien and keng subspecies. Rice Genet. Newsl. 2:46-48. Cheng, K. S. 1993. The evaluation of indica and japonica subspecies in Oryza sativa L. Yunnan Sci. Tech. Press, Kunming, China. Cheng, S. H., H. M. Si, L. S. Zhuo, and Z. X. Sun. 1996. Classification of environmentally induced genetic male sterile lines of rice based on their fertility responses to photoperiod and temperature. J. Agr. Sci. 127:161-167. Cheng, S. H., Z. X. Sun, S. K. Min, Z. M. Xiong. and C. S. Ying. 1990. Studies on response to photoperiod and temperature of the photoperiod-sensitive genic male sterile rice

124

J. LI AND L. YUAN

(PGMSR): 1. Observation of fertility in PGMSR under natural conditions at Hangzhou (30 0 05'N). Chinese J. Rice Sci. 1990(4):157-162. Chiang, C. L., J. Y. Zao, and Z. W. Hu. 1981. Cytological studies on pollen abortion 'V20A' and Honglian A in rice. Dept. Biology, Hunan Teachers' College, Changsha, China. Chinese Academy of Agricultural Sciences and Hunan Hybrid Rice Research Center, 1991. The development of hybrid rice in China. Agr. Press, Beijing, China. Chitrakon, S. 1987. Hybrid rice: Status and future in Thailand. Research Monitor (Thailand) 1987(1):4-10. Chitrakon, S., P. Khambanonda, and P. Senawong. 1986. Hybrid rice: Status and future in Thailand. Thai Agr. Res. J. 1986(2):149-157. Choi, J. H. 1991. Status and prospects of hybrid rice research in the Democratic People's Republic of Korea. Int. Rice Comm. Newsl. 40:56-60. Chou, S. T. 1978. Cytological observations on tissue structure and opening inner factors of anther of male fertile and male sterile lines in rice. Acta Genet. Sinica 5:125-132. Chu, C. S., X. F. Lu, K. M. Yao, X. M. Zong, C. S. Zhou, J. M. Li, and J. B. Liu. 1997. Study on applying "cold fertility" method to low-temperature sensitive variety of sterile strain Pei'ai64s. J. Nanjing Inst. Meteorology 20:425-433. CNHRRDC, 1997. Summary report on the International Training Courses on Hybrid Rice, China National Hybrid Rice Research & Development Center (CNHRRDC), Changsha, Hunan, China. Cress, C. E. 1977. Heterosis of the hybrid related to gene frequency differences between two populations. Genetics 53:86-94. Crow, J. F. 1997. Dominance and overdominance. In: J. G. Coors, S. Pandey, M. V. Grinkel, A. R Hallauer, D. C. Hess, K. R Lamkey, A. E. Me1chinger, G. Srinivasan, and C. W. Stuber (eds.), The genetics and exploitation of heterosis in crops: Proc. Int. CIMMYT Symp., ASA, CSSA, SSSA, Madison, WI (in press), 17-22 August 1997, CIMMYT, Mexico. p. 10 (Abstr.) Crow, J. F. 2000. The rise and fall of overdominance. Plant Breed. Rev. 17: Chapter 5. Cui, J., Z. Y. Yang, and Y. Gao. 1993. The relationship between wide compatibility and restorability in rice. Chinese J. Rice Sci. 1993(1):11-16. Cutler, K. 1991. Canadian seed firm offers" Male sterility technology for hire." Seed Ind. 42:4-21. Dai, Y. R, Z. R Sun, Y. R Xu, M. Z. Zhao, J. H. Zong, and Z. Z. Ling. 1978. Comparative investigations on some respiratory enzymes and free histones in anthers from male sterile and maintainer lines of rice. Acta. Genet. Sinica 5:227-234. Dalmacio, R, D. S. Brar, T. Ishii, L. A. Sitch, S. S. Virmani, and G. S. Khush. 1993. Identification and transfer of a new source of cytoplasmic male sterility from wild species Oryza perennis into rice. Philipp. J. Crop Sci. 1993(suppl.1):40 (Abstr.). Dalmacio, R, D. S. Brar, T. Ishii, L. A. Sitch, S. S. Virmani, and G. S. Khush. 1995. Identification and transfer of a new cytoplasmic male sterility source from Oryza perennis into indica rice (0. sativa). Euphytica 82:221-225. Dalmacio, R, D. S. Brar, L. A. Sitch, S. S. Virmani, and G. S. Khush. 1992. Oryza perennis: a new source of cytoplasmic male sterility in rice. Rice Genet. Newsl. 9:108-110. Darwin, C. 1877. The effects of cross and self fertilization in the vegetable kingdom. D. Appleton, New York. Davenport, C. B. 1908. Degeneration, albinos and inbreeding. Science 28:454-455. Davis, M. D. and J. N. Rutger. 1976. Yield ofF 1 , F z and F 3 hybrids ofrice (Oryza sativa L.). Euphytica 25:587-595. Deng, J. X., W. F. Uu, and Y. H. Xiao. 1990. The synthesis of nuclear acids and proteins and ATP content in anthers of HPGMR at the developmental stage in rice. J. Wuhan Univ. (Ser. Natural Science) 1990(3):85-88.

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

125

Deng, Q. Y., and X. Q. Fu. 1998. Studies of fertility stability of P(T)GMS rice. III. Drift of critical temperature inducing male sterility and its controlling technology. J. Hunan Agr. Univ. 1998(1):8-13. Deng, Q. Y., X. Q. Fu, and L. P. Yuan. 1997. On fertility stability of the P(T)GMS lines and their identification technology. p. 76-85. Proc. Int. Symp. on Two-line System Heterosis Breed. in Crops, Changsha, Hunan, China, September 6-8,1997. Deng, Q. Y., A. H. Ou, X. Q. Fu, and Q. R Zhu. 1996. A preliminary study on the method for identifying the practical photo- and thermo-sensitive genic male sterile rice. Hybrid Rice 1996(2):23-27. Devarathinam, A. A. 1984. Study of heterosis in relation to combining ability and per se performance in rainfed rice. Madras Agr. J. 71:568-572. Dhulappanavar, C. V., and S. W. Mensikai. 1967. Study of heterosis in rice. Mysore J. Agr. Sci. 1:117-122. Dorosti, H. K H. 1997. Evaluation of heterosis among Iran's rice cultivars. In: J. G. Coors, S. Pandey, M. V. Grinkel, A. R Hallauer, D. C. Hess, K R Lamkey, A. E. Melchinger, G. Srinivasan, and C. W. Stuber (eds.), The genetics and exploitation of heterosis in crops. Proc. Int. CIMMYT Symp., ASA, CSSA, SSSA, Madison, WI (in press), 17-22 August 1997, CIMMYT, Mexico. p. 342-343 (Abstr.). Duan, X. M., and H. S. Ma. 1992. Effects of gibberellic acid application on seed yield and quality of hyhrid rice. Seed Sci. Tech. 20:209-214. Edwardson, J. R 1956. Cytoplasmic male sterility. Bot. Rev. 22:696-738. Elsy, C. R, M. Rangaswamy, and K N. Ganesan. 1998. Significance of stigma exsertion in enhancing outcrossing in male sterile lines. lnt. Rice Res. Notes 1998(1):25. Erickson, J. R 1969. Cytoplasmic male sterility in rice (Oryza sativa L.). Agron. Abstr. p. 6. Feng, Y. Y. 1984. Regulation of flowering date in seed production of hybrid rice. GlIangdong Agr. Sci. 1984(6):5-9. Fu, A. J., and H. Wang. 1988. Genetical study on rice male sterility. J. Hunan Agr. Univ. 1988(1):1-6. Fujii, K, T. Kato, A. Shumiya, T. Ito, and S. Kudo. 1991. A rapid and laborsaving method of introducing fertility restoring gene Rf-1 in hybrid rice breeding and its application to general back cross hreeding. Res. Bul. Aichi Ken Agr. Res. Cent. p. 27-39. Fujimaki, H., and S. Hiraiwa. 1986. Genetic analysis of male sterile lines induced hy artificial mutation. Japan. J. Breed. 36: 401-408. Fujimaki, H., S. Hiraiwa, K Kushihuchi, and S. Tanaka. 1977. Artificially induced male sterile mutants and their usages in rice breeding. Japan. J. Breed. 27:70-77. FlIjimaki, H., and H. Yoshida. 1984. Development of male sterile-restoring system for hybrid rice production. 2. Variation of days to heading ofhalf-sih F 1 hyhrids with a common female parent "Akihikari". Japan. J. Breed. 36(suppl.1):196-197. Fujimura, T. 1996. Application of molecular markers for rice breeding in Japan. Plant Genome IV, San Diego, CA. FlIkuta, Y., M. Yano, and K Fukui. 1992. Linkage analysis of the restorer gene (Rf-1) in rice using restriction fragment length polymorphism markers. Japan. J. Breed. 42 (suppI.1):164-165. Funk, C. R., and S. J. Han. 1967. Recurrent interspecific hybridization; aproposed method of breeding Kentucky bluegrass Poa pratensis. New Jersey Agr. Expt. Sta. Bul. 818:3-14. Ganesan, K N., and M. Rangaswamy. 1997. Genetics offertility restoration for WA source of CMS lines in rice. Madras Agr. J. 84:42-43. Gao, M. W. 1981. A preliminary analysis of genotype of hybrid hsien rice with wi Id rice cytoplasm. Acta Genet. Sinica 8:66-74. Gerstel, D. U., B. L. Hammond. and C. Kidd. 1953. An additional note on the inheritance of apomixis in guayule. Bot. Gaz. 115:89-93.

126

J. LI AND L. YUAN

Ghosh, A. K., and S. Govindaswamy. 1972. Inheritance of starch iodine blue value and alkali digestion value in rice and their genetic association. Riso. 21:123-132. Gillham, N. W. 1978. Organelle heredity. Raven, New York. Glaszmann, J. C. 1987. Isozymes and classification of Asian rice. Theor. Appl. Genet. 74:21-30. Glaszmann, J. c., O. Edralin, and S. S. Virmani. 1987. Isozyme markers to monitor seed purity in indica hybrid rice. Int. Rice Res. Newsl. 1987(5):17. Govinda, R. K., and E. A. Siddiq. 1984. Genetics of fertility restoration and biochemical basis of male sterility-fertility restoration system in rice. Rice Genetics Newsl. 1:103-104. Govinda, R. K., and S. S. Virmani. 1988. Genetics of fertility restoration of WA type cytoplasmic male sterility in rice. Crop Sci. 28:787-792. Grazi, F., M. Umaerus, and E. Akerberg. 1961. Observations on the mode ofreproduction and the embryology of paa pratensis. Hereditas 47:489-541. Grimanelli, D., O. Leblanc, E. Espinosa, E. Perotti, D. Gonzalez de Leon, and Y. Savidan. 1997. Mapping diplosporous apomixis in tetraploid Tripsacum: one gene or several genes? Heredity 80:33-39. Gu, F. L., C. G. Lu, M. L. Lu, H. Y. Zhu, and J. S. Zou. 1989. Biological basis of highyielding capacity in indica-japanica hybrid rice. Jiangsu Agr. Sci. 1989(suppl.):31-38. Gu, M. H. 1988. The inheritance of wide compatibility gene (s) and its utilization in rice breeding. J. Jiangsu Agr. College 1988(2):19-26. Gu, M. H. 1992. Current situation and perspective for wide compatibility study in Oryza sativa, p. 23-29. In: L. P. Yuan (ed.), Current status of two-line hybrid rice research. Agr. Press, Beijing, China. Gu, M. H, X. B. Pan, Z. X. Chen, and W. C. Yang. 1991. A study on compatibility of standard wide compatibility testers of rice in China. Sci. Agr. Sinica 1991(6):27-32. Gu M. H., A. Q. You, and X. B. Pan. 1991. Genetic analysis of wide compatibility varieties ofrice (Oryza sativa L.). p. 768-769, In: G. S. Khush (ed.), Rice genetics II. Proc. 2nd Int. Rice Genetics Symp. Int. Rice Res. Inst., P.O. Box 933, 1099 Manila, Philippines. Gu, M. H., A. Q. You, and X. B. Pan. 1992. Genetic analysis on allelic relationship of wide compatibility genes among several WCV (Oryza sativa L.), p. 259-267. In: L. P. Yuan (ed.), Current status of two-line hybrid rice research. Agr. Press, Beijing, China. Guangdong Crop Heterosis Utilization Research Cooperative Group. 1977. Heterosis utilization by way of chemical emasculation in rice. Sci. Agr. Sinica 1977(2):47-50. Guangdong Crop Heterosis Utilization Research Cooperative Group. 1981. Chemical emasculation and heterosis utilization in rice. Agr. Press, Beijing, China. Guangxi Teachers' College. 1975. Cytological studies in Guan-Xuan 3 A, "WP" type of rice. p. 147-150. In: Proc. of China's National Symp. for Cooperative Research on Hybrid Rice, November 1975, Changsha, China. Guo, X. X., X. Y. Zeng, G. Liu, T. He, F. L. Kong, Z. X. Li, K. Luo, and D. Q. Wu. 1991. Preliminary study on apomixis of a rice line CI00l. Southwest. China J. Agr. Sci. 1991(2):10-16. Guok, H. P. 1994. Hybrid rice research in Malaysia, p. 207-211. In: S. S. Virmani (ed.), Hybrid rice: New developments and future prospects. Int. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines. Gustafsson, A. 1946. The effect of heterozygosity on variability and vigour. Hereditas 32:263-286. Gustine, D. L., R. T. Sherwood, and J. Gounaris. 1989. Regulation of apomixis in buffelgrass. p. 411-412. In: Proc. XVIth Int. Grassland Congr., Nice, Association FraIl<;:aise pour la Production Fourragere.

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

127

Hanna, W. W. 1995. Use of apomixis in cultivar development. Adv. Agron. 54:333-350. Hanna, W. W., and W. C. Bashaw. 1987. Apomixis: Its identification and use in plant breeding. Crop Sci. 27:1136-1139. Hanna, W. W., J. B. Powell, J. C. Millot, and G. W. Burton. 1973. Cytology of obligate sexual plants in Panicum maximum Jacq. and their use in controlled hybrids. Crop Sci. 13:695-697. Hara, S. 1946. Linkage between factors for sterility and anthocyanin pigmentation in rice plant. Japan. J. Genet. 21 :32. Haraiwa, S., and S. Tanaka. 1980. Induction of male-sterile mutation in rice. Gamma Field Symp. 19:103-115. Harlan, J. R., M. H. Brooks, D. S. Borgaonkar, and J. M. J. de Wet. 1964. The nature and inheritance of apomixis in Bothriochloa and Dichanthium. Bot. Gaz. 125:41-46. Hayes, H. K. 1952. Chapter 3. Development of the heterosis concept. p. 49-65. In: J. W. Gowen (ed.J, Heterosis. Iowa State College Press, Ames. He, H. H., and S. e. Yuan. 1993. Research and applications of two-line system hybrid rice. Jiangxi Sci. Tech. Press, Nanchang, China. He, H. H., Z. G. Zhang, and S. C. Yuan. 1987. Response on development and fertility changes induced by light under different temperature conditions in HPGMR. J. Wuhan Univ. (HPGMR): 87-93. He, Z. e. 1990. The role of the activity of superoxide dismutase in the prediction of rice heterosis. Hubei Agr. Sci. 1990(11):10-12. He, Z.e., Y. H. Xiao, and S. Y. Feng. 1992. Distribution ofP:l2 in the plants in HPGMR (58s). J. Wuhan Univ. (Ser. Natural Science) 1992(1):127-128. Hermsen, J. 1980. Breeding for apomixis in potato, pursuing a utopian scheme. Euphytica 29:595-607. Heu, M. H. 1967. Studies on the growth duration and hybrid sterility in remote cross breeding of cultivated rice. J. Korean. Soc. Crop. Sci. 4:31-71. Hoan, N. T. 1993. Diversification of male sterile cytoplasm and exploitation of hybrid vigor using in vitro techniques in cultivated rice (Oryza sativa L.). Ph.D. Diss. Anghra Pradesh Agr. Univ. Hoan, N. T., N. N. Kinh, B.B. Bong, N. T. Tram, T.D. Qui, and N.V. Bo. 1998. Hybrid rice research and development in Vietnam. In: S. S. Virmani (ed.), Advances in hybrid rice technology (Proc. 3rd Int. Symp. Hybrid Rice, Hyderahad, India, October 1996). Int. Rice Res. Inst., P.O. Box 933, 1099 Manila, Philippines. Hoan, N. T., N. P. Sarma, and E. A. Siddiq. 1997a. Diversification of male sterile cytoplasm with the aid of embryo rescue technique in inter-specific hybridization for rice. p. 393-394. In: J. G. Coors, S. Pandey, M. V. Grinkel, A. R. Hallauer, D. C. Hess, K. R. Lamkey, A. E. Melchinger, G. Srinivasan, and C. W. Stuber (eds.), The genetics and exploitation of heterosis in crops. Proc. Int. CIMMYT Symp., 17-22 August 1997. ASA, CSSA, SSSA, Madison, WI, CIMMYT. Mexico. (Abstr.) Hoan, N. T., N. P. Sarma, and E. A. Siddiq. 1997b. Identification and characterization of new sources of cytoplasmic male sterility in rice. Plant Breed. 116:547-551. Hoan, N. T., N. P. Sarma, and E. A. Siddiq. 1998. Wide hybridization of CMS in rice. Int. Rice Res. Notes 1998(1):5-6. Hoff, B. J., and D. Chandrapanya. 1973. Inheritance of two male sterile characters in rice. Am. Soc. Agron. p. 7 (Abstr.) Honda, I., S. Iwasaki, K. Sudo, H. Kato, K. Maruyama, M. Hasegawa, I. Yamaguchi, N. Murofushi, T. Yanagisawa, and N. Takahashi. 1997. Semi quantification of gibberellins in the anthers of thermosensitive genetic male sterile rir:e (Oryza sativa L. cv. PL12). BiascL, Biatechnol. Biachem. 61:1748-1750.

128

J. LI AND L. YUAN

Hong, L. F., J. P. Liu, B. Q. Liu, D. F. Hu, and Z. W. Mao. 1997. 1647s-a photosensitive genic male sterile japonica rice (PGMS) developed via anther culture. Acta Agr. BorealiSinica 1997(1):41-45. Hossain, M. 1996. Recent developments in the Asian rice economy: challenge for rice research, p. 17-33. In: R E. Evenson, RW Herdt, and M. Hossain (ed.), Rice research in Asia. CAB Int. Oxford & IRRI, Int. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines. Hu, J. G., and Z. B. Li. 1985. A preliminary study on the inheritance of male sterility of rice male sterile lines with four different kinds of cytoplasms. J. Huazhong Agr. Col. 1985(2):15-22. Hu, J., and J. N. Rutger. 1992. Pollen characteristics and genetics of induced and spontaneous genetic male-sterile mutants in rice. Plant Breeding 109:97-107. Huang, C. S., T. H. Tseng, and C. G Chen. 1987. Genetic analysis of fertility restoration in cytoplasmic male sterile indica rice. J. Agr. Res. China. 1987(2):137-150. Huang, D. N., S. Q. Zhang, R Xue, Z. H. Hua, X. B. Xie, and X. L. Wang. 1998. A new method to identify and improve the purity of hybrid rice with herbicide resistant gene. Chinese Rice Res. Newsl. 1998(1):1. Huang, G. S., Y. Y. Hu, and P. Li. 1992. Preliminary study ofthermosensitive genic malesterile lines by somatic culture and breeding in indica rice. J. Sichuan Agr. Univ. 1992(1):1-6. Huang, H. Z., S. L. Lou, H. C. Wang, and R M. Chen. 1984. Auxin destruction and male sterility in rice. J. Xiamen Univ. (Ser. Natural Science) 1984(1):82-87. Huang, P. J., K. Maruyama, H. L. Sharma, and S. S. Virmani. 1994. Advances in hybrid rice seed production technology. p. 63-70. In: S. S. Virmani (ed.), Hybrid rice: New developments and future prospects. Int. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines. Huang, P. J., and S. D. Tang. 1990. A supplementary pollination technique for hybrid rice seed production. Int. Rice Res. Newsl. 1990(5):5-6. Huang, Q. L., X. H. Tang, and J. L. Mao. 1994. Photoperiod and temperature response of a photoperiod-sensitive male-sterile japonica rice line 7001s and their effect on fertility and anther protein. Acta Agron. Sinica 1994(2):156-160. Huang, Q. L., X. H. Tang, and J. L. Mao. 1996. Effects of temperature on pollen fertility and anther protein in thermosensitive genetically male-sterile rice. Acta Physiol. Sinica 1996(1):69-73. Huang, R H. 1988. Preliminary report on the multiple embryos in rice. 1988(2):10-14. Hughes, W. G. 1976. Ethrel induced male sterility in wheat (Triticum aestivum L.). p. 111-114. In: A. Janossy and F.G.H. Lupton (eds.), Heterosis in plant breeding (Proc. 7th Congr. EUCARPIA, Budapest, June 24-29, 1974). Elsevier Sci. Pub. Co., Amsterdam. Hull, F. H. 1945. Recurrent selection for specific combining ability in corn. J. Am. Soc. Agron.37:134-145. Hunan Hybrid Rice Research Center. 1992. Annual report. Changsha, Hunan, China. Hunan Hybrid Rice Research Center. 1993. Manual for International Hybrid Rice Training Course. Changsha, Hunan, China. Hunan Teachers' College. 1972. Morphological and physiological analysis on pollen development in male-sterile rice. Proc. of China's National Symp. for Cooperative Research on Hybrid Rice, October 1972, p. 81-88, Changsha, China. Ichikawa, N., N. Kishimoto, A. Inagaki, A. Nakamura, Y. Koshino, Y. Yokozeki, M. Oka, S. Samoto, H. Akagi, K. Higo, C. Shinjyo, T. Fujimura, and H. Shimada. 1997. A rapid PCR-aided selection of a rice line containing the Rf-1 gene which is involved in restoration of the cytoplasmic male sterility. Molec. Breed. 3:195-202.

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

129

Idasumi, Y. 1936. Investigations in heterosis of rice plant (first report). Proc. Crop Sci. Soc. Japan 8:504-515. Ikehashi, H. 1982. Prospects for overcoming barriers in the utilization of indica-japonica crosses in rice breeding. Oryza 19:69-77. Ikehashi, H. 1991. Genetics of hybrid sterility in wide hybridization in rice (Oryza sativa L.). p. 113-127. In: Biotechnology in agriculture and forestry. Springer-Verlag, Berlin. Ikehashi, H., and H. Araki. 1984. Varietal screening for compatibility type revealed in F j fertility of distant crosses in rice. Japan. J. Breed. 34:304-312. Ikehashi, H., and H. Araki. 1986. Genetics of F j sterility in remote crosses of rice. p. 119-130. In: Rice genetics. Int. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines. Ikehashi, H., and H. Araki. 1987. Screening and genetic analysis of wide-compatibility in F j hybrids of distant crosses in rice, Oryrza sativa. Tech. Bul. Trop. Agr. Res. Center (Japan) 1987(23):1-79. Ikehashi, H., H. Araki, and S. Yanagihara. 1991. Screening and analysis of wide compatibility loci in wide crosses of rice. p. 33-43. In: Rice Genetics II. Int. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines. Ikehashi, H., J. S. Zou, H. P. Moon, and K. Maruyama. 1994. Wide compatibility gene(s) and indica-japonica heterosis in rice for temperate countries. p. 21-31. In: S. S. Virmani (ed.), Hybrid rice: New developments and future prospects. Int. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines. Int. Rice Res. Inst. 1980. Hybrid rice work revived at IRRI. IRRI Reporter. P.O. Box 933, Manila 1099, Philippines. Int. Rice Res. Inst. 1992. Program Report for 1991. P.O. Box 933, Manila 1099, Philippines. Int. Rice Res. Inst. 1995. Program Report for 1994. P.O. Box 933, Manila 1099, Philippines. Ishikawa, J. 1927. Studies on the inheritance of sterility in rice. J. College Agr. Hokkaido Univ.20:79-201. Iwamasa, M., I. Ueno, and M. Nishiura. 1967. Inheritance of nucellar embryony in Citrus. Bul. Hort. Res. Sta. Japan. Ser. B. 1967(7):1-8. Jan, c. C. 1974. Genetic male sterility in wheat (Triticum aestivum L.): expression, stability, inheritance and practical use. Ph.D. Thesis. Univ. California, Davis. Janick, J. (ed.). 1989. Classic papers in horticultural science. Prentice-Hall, Englewood Cliffs, NJ. p. 544-553. Jefferson, R. A., and S. Nugroho. 1998. Molecular strategies for hyhrid rice: male sterility and apomixis. In: S. S. Virmani, E. A. Siddiq, and K. Muralidharan (eds.), Advances in hybrid rice technology (Proc. 3rd Int. Symp. Hybrid Rice, Hyderabad, India, Octoher 1996). Int. Rice Res. Inst., P.O. Box 933,1099 Manila, Philippines. Jiang, Y. M. 1988. Studies on the effect of high temperature on fertility of the male sterile line in Dian-type hybrid rice. J. Yunnan Agr. Univ. 1988(2):99-107. Jiang, Y. M., Y. Rong, and G. X. Tao. 1997. A new type of thermo-sensitive genic male sterility bred by hybridization. p. 188-192. In: Proc. Int. Symp. on Two-line System Heterosis Breed. in Crops, Changsha, Hunan, China, September 6-8,1997. Jiang, Z. S., Q. G. Xu, Y. W. Dong, and C. L. Kang. 1993. Fertility change and heredity in indica photoperiod sensitive genic male sterile rice "An-Nong s-1." J. Hunan Agr. College 19:217-222. Jin, D. M., Z. B. Li, and J. M. Wan. 1987. The use and value of the gene for photosensitive male sterility in rice breeding. Sci. Agr. Sinica 1987(1):6-12. Jones, D. F. 1917. Dominance of linked factors as a means of accounting for heterosis. Genetics 2:466-479. Jones, H. A., and A. E. Clarke. 1943. Inheritance of male sterility in the onion and the production of hybrid seed. Proc. Am. Soc. Hort. Sci. 43:189-194.

130

J. LI AND L. YUAN

Jones, H. A., and S. L. Emsweller. 1936. Development of the flower and macrogametophyte of Allium cepa. Hilgardia 10:415-428. Jones, J. W. 1926. Hybrid vigor in rice. J. Am. Soc. Agron. 18:423-428. Jones, J. W. 1952. Inheritance of natural and induced mutation in caloro rice and observations on sterile caloro types. J. Hered. 43:81-85. Julfiquar, A. W. 1998. Hybrid rice: status and future in Bangladesh. Int. Rice Res. Notes 1998(2):17-18. Julfiquar, A. W., S. S. Virmani, and A. 1. Carpena. 1985. Genetic divergence among some maintainer and restorer lines in relation to hybrid breeding in rice (Oryza sativa L.). Theor. Appl. Genet. 70:671-678. Kabaki, N., H. Hasegawa, H. Yamaguchi, and T. Kon. 1992. Growth and yield ofjaponicaindica hybrid rice. Bul. Hokuriku National Agr. Expt. Sta. 1992(34):111-139. Kadam, B. S., G. G. Patil, and V. K. Patankar. 1937. Heterosis in rice. Indian J. Agr. Sci. 7:118-125. Kadowaki, K., T. Ishige, and S. Suzuki. 1986. Differences in the characteristics of mitochondria DNA between normal and male sterile cytoplasms of japonica rice. Japan. J. Breed. 36:333-339. Kadowaki, K. 1., T. Suzuki, and S. Kazama. 1990. A chimeric gene containing the 5' portion of Atp6 is associated with cytoplasmic male-sterility of rice. Mol. Gen. Genet. 224:10-16. Kadowaki, K., K. Yazaki, T. Osumi, K. Harada, and M. Nakaghara. 1988. Mitochondrial DNA polymorphism in male-sterile cytoplasm of rice. Theor. Appl. Genet. 75:234-236. Kato, H., et al. 1990. Temperature response and inheritance of a thermosensitive genic male sterility in rice. Japan. J. Breed. 40 (suppI. 1):352-369. Kato, H., K. Maruyama, and H. Uchiyamada. 1994. Hybrid rice research in Japan, p. 149-156. In: S. S. Virmani (ed.), Hybrid rice: New developments and future prospects. Int. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines. Katsuo, K., and U. Mizushima. 1958. Studies on the cytoplasmic difference among rice varieties, Oryza sativa L. 1. On the fertility of hybrids obtained reciprocally between cultivated and wild varieties. Japan. J. Breed. 8:1-5. Kaul, M. L. H. 1986. Male-sterility in doubled haploid rice. Rice Genet. Newsl. 3:109. Kaul, M. L. H. 1988. Male sterility in higher plants. Springer-Verlag, Berlin and Heidelberg. Kawase, M., N. Nishimoto, T. Tanaka, A. Yoshimura, S. Yoshimura, K. Saito, A. Saito, M. Yano, N. Takeda, T. Nagamine, and M. Nakagahra. 1991. Intraspecific variation and genetic differentiation based on restriction fragment length polymorphism in Asian cultivated rice, Oryza sativa L. p. 467-473. In: Rice genetics II. Int. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines. Khalique, M. A., O. 1. Joarder, and A. M. Euus. 1977. Heterosis and combining ability in a diallel cross of rice (Oryza sativa L.). Bangladesh J. Agr. Sci. 1977(2):137-145. Khush, G. S. 1995. Breaking the yield frontier ofrice. GeoJournal 35:329-332. Khush, G. S., and R. C. Aquino. 1994. Breeding tropical japonicas for hybrid rice production. p. 3:j-36. In: S. S. Virmani (ed.), Hybrid rice technology: New developments and future prospects, Int. Rice Res. Inst., P.O. Box 933, 1099 Manila, Philippines. Khush, G. S., D. S. Brar, and J. Bennett. 1998. Current status of research on apomixis in rice and prospects for its use in heterosis breeding. In: S. S. Virmani (ed.), Advances in hybrid rice technology (Proc. 3rd Int. Symp. Hybrid Rice, Hyderabad, India, October 1996). Int. Rice Res. Inst., P.O. Box 933,1099 Manila, Philippines. Khush, G. S., 1. Kumar, and S. S. Virmani. 1988. Grain quality of hybrid rice. p. 201-215. In: Hybrid rice (Proc. Int. Symp. on Hybrid Rice, October 6-10,1986, Changsha, China). Int. Rice Res. Inst., P.O. Box 933,1099 Manila, Philippines.

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

131

Khush, G. S., and S. Pengo 1996. Breaking the yield frontier ofrice. In: Proc. Workshop on Raising Yield Potential in Wheat: Breaking the Barriers. CIMMYT, Mexico. Khush, G. S., S. Peng, and S. S. Virmani. 1998. Improving yield potential by modifying plant type and exploiting heterosis. p. 150-170. In: J. C. Waterlow, D. G. Armstrong, L. Fowden, and R. Riley (eds.), Feeding a world population of more than eight billion people: A challenge to science. Oxford, U.K. Kim, C. H., and J. N. Rutger. 1988. Heterosis in rice. p. 39-54. In: Hybrid rice (Proc. Int. Symp. on Hybrid Rice, October 6-10,1986, Changsha, China). Int. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines. Kim, K. H., and M. H. Heu. 1979. A study on heterosis in cross between semi-dwarf rice cultivars. Korean J. Breed. 1979(2):127-132. Kinoshita, T. 1992. Report of the committee on gene symbolization nomenclature and linkage groups. Rice Genet. Newsl. 9:2-4 Kinoshita, T. 1997. Gene symbols and information on male sterility. Rice Genet. Newsl. 14:13-22. Kirk, J. T. 0., and R. A. E. Tilney-Bassett. 1978. The plastids: Their chemistry, structure, growth and inheritance. Elsevier/North-Holland Biomedical Press, Amsterdam. Kitamura, E. 1961. Genetic studies on hybrid sterility in the cross between japonica and indica type of rice (in Japanese). Rec. Adv. Breed. Res. 2:53-63. Kitamura, E. 1962a. Genetic studies on sterilities observed in hybrids between distantly related varieties of rice, Oryza sativa L. (in Japanese with English summary). Bul. Chugoku Agr. Expt. Sta. Ser. A8:141-205. Kitamura, E. 1962b. Studies on cytoplasmic sterility of hybrids in distantly related varieties of rice, Oryza sativa L. I. Fertility of the F 1 hybrids between strains derived from a certain Philippine x Japanese variety crosses and Japanese varieties. Japan. J. Breed. 12:81-84. Kitamura, E. 1962c. Studies on cytoplasmic sterility of hybrids in distantly related varieties of rice, Oryza sativa L. II. Analysis of nuclear genes in Japanese varieties controlling cytoplasmic sterility. Japan. J. Breed. 12:166-168. Kitaoka, M., K. Ota, and M. Ogawa. 1991. Synergistic effects of Ethephon and Isourea on male sterility of rice. Japan. J. Crop Sci. 60:349-354. Ko, T., and H. Yamagata. 1980. Studies on the utility of artificial mutations in plant breeding: XII. Induction and gene analysis of male sterile mutants in rice. Japan. J. Breed. 30:367-374. Ko, T., and H. Yamagata. 1985. Studies on the induction of male sterile strains in rice: VIII. Linkage relationships of male sterility genes to marker genes. Japan. J. Breed. 35:86-87. Ko, T., and H. Yamagata. 1987. Studies on the utility of artificial mutations in plant breeding: XVI. Gene analysis of male sterility induced in rice. Japan. J. Breed. 37:192-198. Ko, T., and H. Yamagata. 1989. Studies on the induction of male sterile strains in rice. 9. Linkage groups of few male-sterility genes. Japan. J. Breed. 39 (suppi. 1):234-235. Koh, J. C. 1987. Studies on combining ability and heterosis in F 1 hybrids using cytoplasmic genetic male sterile lines of rice (Oryza sativa L.). Res. Rep. Rural Dev. Adm. Korea. 1987(2):1-21. Koltunow, A. M., R. A. B. Bicknell, and A. M. Chaudhury. 1995. Apomixis: molecular strategies for the generation of genetically identical seed without fertilization. Plant Physioi. 108:1345-1352. Kornicke, F. U. W. 1885. Handbuch des Getreidebaues, V. 2. Die Sorten und der Anbau des Getreides. Berlin, Germany. Kumar, I., and G. S. Khush. 1986. Genetics of amylose content in rice (Oryza sativa L.). J. Genet. 65:1-11.

132

J. LI AND 1. YUAN

Kumar, I., and G. S. Khush. 1987. Genetic analysis of different amylose levels in rice. Crop Sci. 27:1167-1172. Kumar, I., and G. S. Khush. 1988. Inheritance of amylose content in rice. Euphytica. 38:261-269. Kumar, I., G. S. Khush, and B. O. Juliano. 1987. Genetic analysis of waxy locus in rice. Theor. Appl. Genet. 73:481-488. Kumar, I., K. Maruyama, and H. P. Moon. 1994. Grain quality consideration in hybrid rice. p. 123-130. In: S. S. Virmani (ed.), Hybrid rice technology: New developments and future prospects, lnt. Rice Res. lnst., P.O. Box 933, 1099 Manila, Philippines. Kumar, I., and S. S. Saini. 1981. Diallel analysis in rice combining ability for various quantitative characters. Genet. Agr. 35:243-252. Kumari, S. 1., M. Mahadevappa, and R. S. Kulkarni. 1998. Fertility restoration studies in four WA CMS lines ofrice. lnt. Rice Res. Notes 1998(1):9. Lara, R. J., I. M. Dela Cruz, M. S. Ablaza, H. C. Dela Cruz, and S. R. Obien. 1994. Hybrid rice research in Philippines. p. 173-186. In: S. S. Virmani (ed.), Hybrid rice: New developments and future prospects. lnt. Rice Res. lnst., P.O. Box 933, Manila 1099, Philippines. Lasa, J. M., and N. O. Bosemark. 1993. Chapter 14. Male sterility. p. 213-228. In: M. D. Hayward, N. O. Bosemark, and I. Romagosa (eds.), Plant breeding: Principles and prospects. Chapman & Hall, London. Laser, K. D., and N. R. Lerstern. 1972. Anatomy and cytology of microsporogenesis in cytoplasmic male sterile angiosperms. Bot. Rev. 38:425-454. Leblanc, 0., D. Grimanelli, D. Gonzalez de Leon, and Y. Savidan. 1995. Detection of the apomictic mode ofreproduction in maize-Tripsacum hybrids using maize RFLP markers. Theor. Appl. Genet. 90:1198-1203. Lei, J. C. 1983. The recessive restoration gene in WA-type male sterile lines in rice and its application in breeding. Fujian Agr. Sci. Tech. 1983(5):16-19. Lei, J. X., and Z. B. Li. 1989. A study on hereditary regularity of Hubei Photo-sensitive Genetic Sterile Rice. Hybrid Rice 1989(2):39-43. de Leon, J. c., E. D. Redofia, I. A. dela Cruz, M. F. Ablaza, F. M. Malabanan, R. J. Lara, and S. R. Obien. 1998. Hybrid rice in the Philippines: progress and prospects. In: S. S. Virmani, E. A. Siddiq, and K. Muralidharan (eds.), Advances in hybrid rice technology (Proc. 3rd lnt. Symp. Hybrid Rice, Hyderabad, ~ndia, October 1996). lnt. Rice Res. lnst., P.O. Box 933,1099 Manila, Philippines. Li, C. Q., and S. F. Ang. 1988. Genetic distance and heterosis in japonica rice. p. 257. In: Hybrid rice (Proc. lnt. Symp. on Hybrid Rice, October 6-10, 1986, Changsha, China). lnt. Rice Res. lnst., P.O. Box 933, Manila 1099, Philippines. Li, D. H., B. S. Luo, and Y. L. Qu. 1996. Ethylene biosynthesis of photoperiod-sensitive genic male-sterile rice and fertility alteration. Acta Phytophysiol. Sinica. 22:320-326. Li, D. M., S. R. Liang, Q.L Xuan, and X. Y. Qin. 1989a. A study on utilization of Hubei photosensitive genic sterile rice in south China. Hybrid Rice 1989(1):27-31. Li, H. B., J. Wang, A. M. Liu, K. D. Liu, Q. F. Zhang, and J. S. Zou. 1997. Genetic basis of low-temperature-sensitive sterility in indica-japonica hybrids ofrice as determined by RFLP analysis. Theor. Appl. Genet. 95:1092-1097. Li, H. B., Q. Zhang, A. M. Liu, J. S. Zou, and Z. M. Chen. 1996a. A genetic analysis oflowtemperature-sensitive sterility in indica-japonica rice hybrids. Plant Breed. 115:305-309. Li, J. M. 1990. A preliminary study on the inheritance of growth duration in the progenies of indica-japonica crosses in rice (Oryza sativa 1.) (I). Hybrid Rice 1990(5)32-35.

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

133

Li, J. M. 1991a. A preliminary study on the inheritance of growth duration in the progenies of indica-japonica crosses in rice (Oryza sativa L.) (II). Hunan Agr. Sci. 1991(4)12-13. Li, J. M. 1995. Hybrid rice in Vietnam. Hybrid Rice 1995(1):43-44. Li, J. M., and L. P. Yuan. 1996. Hybrid rice seed production in China. October 29-November 1, 1996. Grupo Tecnico de Trabajo de Hibridos de Arroz Para America Latina y el Caribe (GRUTHA), Ibaqu8, Colombia. Li, P., Y. Wang, and H. Liu. 1990. Physiological bases of high yielding heterosis in indicatype F 1 hybrid rice. Sci. Agr. Sinica 1990(5):39-44. Li, P., K. D. Zhou, Y. Chen, P. He, R. D. Li, and L. H. Zhu. 1996b. RFLP mapping of genes conferring fertility restoration of wild cytoplasmic-genetic male sterility in rice (Oryza sativa subsp. indica). Acta Genet. Sinica. 23:357-362. Li, P. H. 1977. How we studied hybrid rice. Acta Bot. Sinica 19:7-10. Li, R. B., and M. P. Pandey. 1998. Genetics of the thermosensitive genic male sterility in rice. Int. Rice Res. Notes 1998(2):9-10. Li, R. Q., J. B. Wang, and X. M. Wang. 1993. Ultrastructural changes in microsporogenesis and pollen development of photoperiod sensitive genic male-sterile rice under different daylength. Chinese Rice J. 1993(2):65-70. Li, X. F., D. W. Hu, Y. H. Jiang, and F. Y. Huang. 1989b. Utilization of rice F 2 hybrids bred by chemical castration technique. Guangdong Agr. Sci. 1989(4):5-7. Li, X. F., F. Y. Huang, and Y. H. Jiang. 1986. Discussion on the utilization of F 2 generation in the hybrid combination Gang-Hua-Qin-Lan. Hybrid Rice 1986(3):12-13. Li, X. Q. 1992. The use of successive backcrosses in transfer of dual-purpose sterile lines. Hybrid Rice 1992(1):29-31. Li, X. Z., and C F. Ren. 1994. A study on grain plumpness of inter-subspecific hybrid rice (Oryza sativa L.). J. Southwest Agr. Univ. 16:456-460. Li, X. Z., G. Q. Zhou, M. L. Xu, L. B. Chen, C. D. Huang, and H. Q. Huang. 1996c. Effects of mechanical collection and pollination of rice pollen on the vigor and seed set in hybridization. Acta Agron. Sinica 1996(3):353-357. Li, Y. C, H. D. Deng, and L. P. Yuan. 1990. The research progress on rice apomixis. Hybrid Rice 1990(1):29-32. Li, Y. C, and L. P. Yuan. 1985. Genetic analysis offertility restoration in male sterile lines of rice. p. 617-632. In: Rice genetics. Proc. Int. Rice Genetics Symp. Int. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines. Li, Y. c., and L. P. Yuan. 1990. Studies on genetics of twin seedlings in rice (Oryza sativa L.). Acta Agron. Sinica 1990(2):176-182. Li, Z. B. 1980a. A preliminary discussion about the classification of male sterile lines of rice in China. Acta Agron. Sinica 1980(1):17-26. Li, Z. 8., Y. H. Xiao, Y. G. Zhu, R. Q. Li, C. L. Liu, and J. M. Wan. 1982. Study and practice of hybrid rice. Shanghai Sci. Tech. Press, Shanghai, China. Li, Z. 8., and Y. G. Zhu. 1988. Rice male sterile cytoplasm and fertility restorations. p. 85-102. In: Hybrid rice (Proc. Int. Symp. on Hybrid Rice, October 6-10,1986, Changsha, China). Int. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines. Li, Z. C., X. K. Wang, J. P. Wang, and Y. H. Huang. 1995. Studies on the position of American rice varieties in heterotic utilization in F 1 hybrid of rice. Acta Agr. Boreali Sinica 1995(1):22-28. Li, Z. J. 1991b. Evaluation on seed purity ofT(P)GMS system hybrid rice variety, KS-9/ZhiLong. Seed. 1991(3):1-3. Li, Z. Y. 1980b. Dian-type hybrid rice. Yunnan People's Press.

134

J. 11 AND L. YUAN

Li, Z. Y., and J. L. Wu. 1993. A new combining pattern in utilization of heterosis between rice subspecies. Hybrid Rice 1993(1):6-8. Liang, C. Y. 1991. Research status of physiological and biochemical basis for rice heterosis and the prediction of heterosis. p. 211-214. In: Papers on basic research of rice breeding techniques. China Sci. Tech. Press, Beijing, China. Liang, C. Y., and X. F. Chen. 1993. Polyamine contents and antioxidative metabolism in anthers of cytoplasmic male-sterile rice lines and their corresponding maintainer lines. Acta Phytophysiol. Sinica 19:82-88. Liang, C. Y., M. W. Huang, and X. J. Pan. 1991. A preliminary research on the relationship between the esterase and yield heterosis in hybrid rice. p. 279-284. In: Papers on basic research of rice breeding techniques. China Sci. Tech. Press, Beijing, China. Liang, J. H., B. X. Wang, S. C. Yuan, and H. L. Zeng. 1990. The critical light intensities inducing two photoperiodic relations in HPGMR. J. Huazhong Agr. Univ. 1990(4):363-368. Liao, C. L., Z. H. Huang, K. Y. Ni, Y. S. Liu, and G. N. Shang. 1991. Selection for general restorers in hybrid rice using special accumulative model. Southwest China J. Agr. Sci. 1991(5):1-5. Lin, J. Y. 1994. Impact of hybrid rice on input demand and productivity. Agr. Econ. 10:153-164. Lin, M. J. 1978. Genetic study of the relationship between the lowland rice varieties and the traditional upland rice varieties. M.S. Thesis (Univ. Philippines). Lin, S. c., and L. P. Yuan. 1980. Hybrid rice breeding in China. p.33-51. In: Innovative approaches to rice breeding. Int. Rice Res. Inst., P.o. Box 933, Manila 1099, Philippines. Lin, X. H., G. X. Yu, D. P. Zhang, Y. F. Xie and F. L. Qin. 1996. Location of one PGMS gene in Nongken 58s on chromosome 5 ofrice. J. Huazhong Agr. Univ. 1996(1):1-5. Ling, D. H., C. Y. Liang, R. Ma, and M. F. Chen. 1989. Somaclonal male-sterile mutants in indica rice. p. 737-740. In: S. Iyama and G. Takeda (eds.), Proc. 6th Int. Congr. SABRAO. Lingaraju, S., B. V. Chandra, V. Bhaskar, and S. Gangadhariah. 1998. Adjusting flowering of a CMS line in hybrid rice seed production. Int. Rice Res. Notes 1998(2):28. Liu, A. M., Q. F. Zhang, and H. B. Li. 1992. Location of a gene for wide compatibility in the RFLP linkage map. Rice Genet. Newsl. 9:134-136. Liu, J. P. 1995. The development of japonica T(P)GMS lines by way of anther culture. Beijing Agr. Sci. 1995(4):22-24. Liu, K. D., J. Wang, H. B. Li, C. G. Xu, A. M. Liu, X. H. Li, and Q. F. Zhang. 1997a. A genome-wide analysis of wide compatibility in rice and the precise location of the S5 locus in the molecular map. Theor. Appl. Genet. 95:809-814. Liu, L. J., and G. X. Xue. 1995. Studies on specific proteins of the photoperiod sensitive genic male-sterile genes of rice. Acta Agron. Sinica 21:251-253. Liu, P. 1997. Optimum use of gibberellins for seed production in japonica hybrid rice. China Rice 1997(3):22. Liu, Q. K., C. S. Wang, P. L. Chen, and B. Chen. 1983. Studies on the arsentic residue of "male-gametocide No.2" used to induce male-sterility in the production of hybrid rice. Acta Agr. Univ. Zhejiangensis 1983(4):331-337. Liu, X. c., M. A. S. Maroof, J. L. Wu, and L. Pan. 1997b. SSR characterization of heterozygotic patterns of parents for predicting heterosis in rice breeding. p. 72. In: J. G. Coors, S. Pandey, M. V. Grinkel, A. R. Hallauer, D. C. Hess, K. R. Lamkey, A. E. Melchinger, G. Srinivasan, and C. W. Stuber (eds.), The genetics and exploitation ofheterosis in crops. Proc. Int. CIMMYT Symp., 17-22 August 1997. ASA, CSSA, SSSA, Madison, WI, CIMMYT. Mexico. (Abstr.)

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

135

Liu. X. c., and J. L. Wu. 1996. A novel approach for discovery and exploitation of new male sterile cytoplasm resources for hybrid rice. Chinese Rice Res. News!. 1996(3):12. Liu, X. c., and J. L. Wu. 1998. SSR heterogenic patterns of parents for marking and predicting heterosis in rice breeding. Molec. Breed. 4:263-268. Liu, X. D., Q. F. Chen, and S. K. Zheng. 1990. Preliminary study on multiple seedlings in rice. J. Fujian Agr. College 1990(2):131-137. Liu, Y. B., Y. W. Sun, and Q. P. Cai. 1990. Analysis on heterosis in protein content of hybrid rice. Acta Agr. Univ. Jiangxiensis (Special issue for quality rice breeding). p. 9-19. Liu, Y. S., S. M. Wang, Y. Z. Wang, and D. Y. Zuo. 1988. Structural variance analysis of mitochondia COl and COIl genes in normal and cytoplasmic male-sterile cultivars of rice. Acta. Genet. Sinica 15:348-354. Liu, Y. S., L. H. Zhu, J. S. Sun, and P. He. 1997c. Mapping quantitative trait loci for reproductive barriers occming in hybrid between indica and japonica rice. Acta Bot. Sinica. 39:1099-1104. Liu, Z. C. 1986. Chloroplast genome translated products and cytoplasmic male sterility. Acta Genet. Sinica 13:201-206. Long, J. D., M. B. Weed, D. J. Fitzgerald, and G. E. Barrier. 1973. Chemical control of pollination. Proc. Am. Soc. Agron. Madison. p. 69. (Abstr.) Lu, c., and X. G. Pan. 1992. Inheritance of wide compatibility in rice cultivars 02428 and 8504. Chinese J. Rice Sci. 1992(3):113-118. Lu, C. G., F. L. Gu, M. L. Lu, and J. S. Zou. 1991. Studies on physiological basis of highyield of indica-japonica hybrid rice Yayou 2. Jiangsu J. Agr. Sci. 1991(1):15-19. Lu, K. Y., and S. C. Yuan. 1991. Effects of dark-duration in photoperiod for inducing fertility transformation in photoperiod-sensitive genic male sterile rice. J. Huazhong Agr. Univ.1991(3):238-241. Lu, X. G. (ed.). 1992a. Current status on the research of photoperiod or temperaturesensitive male sterility in rice. Agr. Press, Beijing, China. Lu, X. G. 1992b. Considerations on the development of T(P)GMS lines in rice. Hi-tech Newsl. 1992(5):1-6. Lu, X. G. 1994. Retrospects on selection and breeding of photo-thermo-sensitive geneic male sterile rice lines in China. Hybrid Rice 1994(3-4):27-30. Lu, X. G., T. M. Mu, G. C. Fang, M. H. Zhou, and J. X. Zhou. 1989. Studies on the selection and utilization of indica type photoperiod-sensitive genic male sterile lines. I. Effectiveness of controlling the PGMS genes into the indica type rice varieties. Hybrid Rice 1989(5):29-32. Lu, X. G., and T. L. Wang. 1986. Research and utilization of HPGMR. II. Observations on the genetical behavior of fertility performance. Hybrid Rice 1986(4):6-9. Lu, X. G., Z. G. Zhang, K. Maruyama, and S. S. Virmani. 1994. Current status of two-line method of hybrid rice breeding. p. 37-49. In: S. S. Virmani (ed.), Hybrid rice: New developments and future prospects. In1. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines. Lu, X. Y., F. C. Kung, X. K. Li, F. E. Xiao, and Z. M. Luo. 1992. Physiological factors for high proportion of empty-unfilled grains in two-line intersubspecific hybrid rice. J. Hunan Agr. College 18:509-515. Lu, Y. G., and G. Q. Zhang. 1986. Breeding for an isogenic restorer line ot'CMS Zhen-Shan 97 A. Rice Genet. Newsl. 3:90-91. Lu, Z. T., S. Q. Tang, Z. M. Xun, S. K. Min, and Z. D. Sheng. 1988. Analyses of yield components in conventional and hybrid rice. Zhejiang Agr. Sci. 1988(4): 156-158. Lua, B. S., W. B. Li, Y. L. Qu, D. H. Liu, and H. S. Li. 1990. A preliminary investigation of

136

J.

LI AND L. YUAN

fertility alteration in Hubei photoperiod sensitive genic male-sterile rice. J. Huazhong Agr. Univ. 1990(1):7-12. Luo, C. S., and L. P. Yuan. 1990. Status of research in and utilization of hybrid rice germplasm. Crop Gen. Res. 1990(1):5-9. Luo, L. J., c. S. Ying, S. K. Min, X. Q. Yu, H. W. Mei, Y. P. Wang, L. B. Guo, and D. B. Zhong. 1994. Wide compatible and restoring ability of Zhong 413 and its utilization in hybrid rice improvement. Chinese J. Rice Sci. 1994(4):193-199. Luo, L. J., c. S. Ying, and X. K. Wang. 1991. Sources and selection for universally compatible cultivars of rice. Crop Gen. Res. 1991(1):14-15. Luo, M., K. D. Zhou, and X. D. Wang. 1991. Selection effect and cytoembryological studies of rice twin seedlings. p. 65-73. In: K. D. Zhou (ed.), Study on rice apomixis. Chengdu, Sichuan Sci. Tech. Press, China. Luo, X. H., and L. P. Yuan. 1989. The development of WCVs in rice. Hybrid Rice 1989(2):35-38. Luo, Y. c., C. Q. Li, S. H. Wang, and D. Z. Wang. 1996. A study on the allelism ofT(P)GMS genes. Proc. Annual Hainan Meeting of China's National "863" Program. p. 163-166. Luo, Z. M., A. P. Zhong, and Q. X. Zhou. 1988. Effects of a new nontoxic chemical gametocide on rice. p. 147-156. In: Hybrid rice (Proc. Int. Symp. on Hybrid Rice, October 6-10,1986, Changsha, China). Int. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines. Lutts, S., J. Ndikumana, and B. P. Louant. 1994. Male and female sporogenesis and gametogenesis in apomictic Brachiaria brizantha, Brachiaria decumbens and F 1 hybrids with sexual colchicine-induced tetraploid Brachiaria ruziziensis. Euphytica 78:19-25. Mackill, D. J. 1995. Applying genetics to rice improvement in California's temperate environment. Int. Rice Commission News!. 44:62-65. Mackill, D. J., and J. N. Rutger. 1994. Public sector research on hybrid rice in the United States. p. 235-239. In: S. S. Virmani (ed.), Hybrid rice: New developments and future prospects. Int. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines. Maekawa, M. 1982. Studies on genetical differences between distantly related rice varieties. Mem. Fac. Agr. Hokkaido Univ. 1982(2):146-177. Mahadevappa, L. M. M., and W. R Coffman. 1980. Karnatka male sterile rice KMS 1. Int. Rice Res. Newsl. 1980(3):3. Mahalanobis, P. C. 1936. On the generalized distance in statistics. Proc. Nat. Int. Sci. India. B2:49-55. Mallick, E. H., H. N. Ghosh, and P. Bairagi. 1978. Heterosis in indica rice. Indian J. Agr. Sci. 48:384-387. Manuel, W. W., and S. Palanisamy. 1989. Heterosis and correlation in rice. Oryza 26:238-242. Mao, C. X., S. S. Virmani, and 1. Kumar. 1998. Technological innovations to lower the costs of hybrid rice seed production. In: S. S. Virmani, E. A. Siddiq, and K. Muralidharan (eds.), Advances in hybrid rice technology (Proc. 3rd Int. Symp. Hybrid Rice, Hyderabad, India, October 1996). Int. Rice Res. Inst., P.O. Box 933, 1099 Manila, Philippines. Mariani, c., V. Gossele, M. de Beuckeleer, M. de Block, RB. Goldberg, W. de Greef, and J. Leemans. 1992. A chimeric ribonuclease-inhibitor gene restores fertility to male sterile plants. Nature (Lond.) 357:384-387. Mariani, C., M. D. Beuckeber, J. Truettver, J. Leemans, and R. B. Goldberg. 1990. Induction of male sterility in plants by a chimeric ribonuclease gene. Nature 347:737-741. Martin, J. A., and J. H. Crawford. 1951. Several types of sterility in Capsicum frutesceus. Proc. Am. Soc. Hort. Sci. 57:335-338. Maruyama, K., H. Araki, and H. Kato. 1991a. Thermosensitive genetic male sterility

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

137

induced by irradiation. p. 227-232. In: Rice genetics II. Int. Rice Res. Inst., P.o. Box 933, Manila 1099, Philippines. Maruyama, K., H. Kato, and H. Araki. 1991b. Mechanized production of F 1 seeds in rice by mixed planting. Japan. Agr. Res. Quarterly 24:243-252. Matsuo, T. 1952. Genecological studies on the cultivated rice. BuI. NatI. Inst. Agr. Sci. Japan D3:1-111. Matzk, F. 1991. Studies towards novel methods in breeding of the apomictic species Poa pratensis L. p. 191-192. In: A. P. M. den Nijs and A. Elgersma (eds.), Fodder crops breeding: achievements, novel strategies and biotechnology. Pudoc, Wageningen, Netherlands. Maurya, D. M., and D. P. Singh. 1977. Genetic divergence in rice. Indian J. Genet. 37: 395-402. Maurya, D. M., and D. P. Singh. 1978. Heterosis in rice. Indian J. Genet. Plant Breed. 38: 71-76. Maximos, M. A., and 1. R Aidy. 1994. Hybrid rice research in Egypt. p. 227-233. In: S. S. Virmani (ed.), Hybrid rice: New developments and future prospects. Int. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines. McDaniel, R G., and 1. V. Sarkissian. 1966. Heterosis complementation by mitochondria. Science 152:1640-1642. McKenzie, K. S., and J. N. Rutger. 1983. Genetic analysis of amylose content, alkali spreading and grain dimension in rice. Crop Sci. 23:306-313. McWilliam, J. R, H. Ikehashi, and S. K. Sinha. 1995. Progress in hybrid rice in India. Int. Rice Commission News!. 44:80-87. Mei, G. Z., X. M. Wang, and M. Q. Wang. 1990. Genetical analysis of the photoperiod sensitive male sterility of Nongken 58s and its derivatives. J. Huazhong Agr. Univ. 9:400-406. Mese, M. D., L. E. Azzini, and J. N. Rutger. 1984. Isolation of genetic male-sterile mutants in a semidwarf japonica rice cultivar. Crop Sci. 24:523-525. Mignouna, H., S. S. Virmani, and M. Briquet. 1987. Mitochondrial DNA modifications associated with cytoplasmic male sterility in rice. Theor. App!' Genet. 74:666-669. Miles, J. W., and M. 1. Escandon. 1997. Further evidence on the inheritance of reproductive mode in Brachiaria. Can. J. Plant Sci. 77:105-107. Min, S. K. 1986. Cross breeding between indica and japonica. China's Rice Science. Agr. Press, Beijing, China. Min, S. K. 1990. A study on the standard of evaluation for WCVs in rice. J. Zhejiang Agr. 1990(1):1-6. Minh, H. T., T. D. Quy, N. T. Nhan, and 1. Q. Du. 1997. Development ofTGMS lines with diversified source of sterility. p. 152-153. In: J. G. Coors, S. Pandey, M. V. Grinkel, A. R Hallauer, D. C. Hess, K. R Lamkey, A. E. Melchinger, G. Srinivasan, and C. W. Stuber (eds.), The genetics and exploitation of heterosis in crops. Proc. Int. CIMMYT Symp., 17-22 August 1997. ASA, CSSA, SSSA, Madison, WI, CIMMYT. Mexico. (Abstr.) Minocha, J. 1., and R K. Gupta. 1988. Induction of male sterility in rice using chemical mutagens. Mutation Breed. NewsI. 1988(32):5-6. Minocha, J. L., J. S. Sidhu, and R K. Gupta. 1991. Mutagen-induced male sterility in rice. p. 752-753. In: Rice Genetics II, 2nd Int. Rice Genetics Symp. Int. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines. Mitra, G. N. 1962. Inheritance of grain size in rice. Curro Sci. 31:105-106. Miyazawa, B. 1932. On the two cases of semisterility in Oryza sativa. BuI. Miyazaki College Agr. Forestry 4:193-197.

138

J. LI AND L. YUAN

Mohamad, 0., H. P. Guok, and O. Othman. 1987. Hybrid rice breeding in Malaysia. Malays. Appl. BioI. 1987(1):129-138. Mohanty, H. K., and K. C. Mohapatra. 1973. Diallel analysis of yield and its components in rice. Indian J. Genet. Plant Breed. 33:264-270. Moon, H. P. 1988. Hybrid rice research in South Korea. p. 290. In: Hybrid rice (Proc. Int. Symp. on Hybrid Rice, October 6-10,1986, Changsha, China). Int. Rice Res. Inst., P.O. Box 933, 1099 Manila, Philippines. Moon, H. P., M. H. Heu, and C. H. Kim. 1994. Hybrid rice research in the Republic of Korea. p. 217-226. In: S. S. Virmani (ed.), Hybrid rice: New developments and future prospects. Int. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines. Mori, K. 1984. Inheritance of susceptible mutant in rice plant to herbicide, bentazon. Japan. J. Breed. 34(suppl.):421-422. Morinaga, T. 1954. Classification ofrice varieties on the basis of affinity. Japan. J. Breed. 4:1-14. Morinaga, T. 1968. Origin and geographical distribution of Japanese rice. Japan. Agr. Res. Quart. 18:1-5. Morinaga, T., and H. Kuriyama. 1955. ]aponica type in the subcontinent ofIndia and Java. Japan. J. Breed. 5:149-153. Morinaga, T., and H. Kuriyama. 1958. Intermediate type of rice in the subcontinent ofIndia and Java. Japan. J. Breed. 8:253-259. Muker, H. S., and H. L. Sharma. 1991. Isolation distance for producing hybrid rice seed. Int. Rice Res. Newsl. 1991(1):6. Munoz, D. B. 1992. Hybrid rice research in Colombia. Arroz 1992(379):20-28. Munoz, D. B. 1994. Hybrid rice research in Colombia. p. 242-248. In: S. S. Virmani (ed.), Hybrid rice: New developments and future prospects. Int. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines. Murayama, S. 1973. The basic studies on utilization of hybrid vigor in rice. 1. The degree of heterosis and its phenomenon. Japan. J. Breed. 23:22-26. Murayama, S., T. Omura, and K. Kimazato. 1974. Basic studies on utilization of hybrid vigor in rice. IV. Heterosis under different cultural conditions. Japan. J. Breed. 24:287-290. Nagai, 1. 1926a. Studies on the mutations in Oryza sativa L. II. On awned sterile, compactpanicled and dwarf mutants. Japan. J. Bot. 3:55-56. Nagai, 1. 1926b. Studies on the mutations in Oryza sativa 1. III. On paleaceous sterile mutant. Japan. J. Bot. 3:67-84. Nagai, 1. 1958. ]aponica rice: its breeding and culture. Yokendo, Tokyo. Nagamine, T., K. Kadowaki, A. Goto, and T. Ajioka. 1995. A new source of cytoplasmic male-sterility found in Bangladesh Bora rice. Rice Genet. Newsl. 12:232-235. Nakajima, M., 1. Yamaguchi, S. Kizawa, N. Murofushi, and N. Takahashi. 1991. Semiquantification of Ga-1 and Ga-4 in male-sterile anthers of rice by radioimmunoassay. Plant Cell Physiol. 32:511-513. Nakatat, S., and B. R. Jackson. 1973. Inheritance of some physical grain quality characteristics in a cross between a Thailand and Taiwanese rice. Thai J. Agr. Sci. 6:223-235. Namboodri, K. M. N. 1963. Hybrid vigor in rice. Rice News Teller 11:92-96. Neves, P. C. F., E. M. Castro, P. H. N. Rangel, and 1. P. Yokoyama. 1994. Hybrid rice research in Brazil. p. 249-252. In: S. S. Virmani (ed.), Hybrid rice: New developments and future prospects. Int. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines. Nguyen, C. T. 1994. Progress in the hybrid rice programme in Viet Nam. Int. Rice Comm. Newsl. 43:23-28.

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

139

Nguyen, V. L., B. B. Bong, and J. C. Mohan. 1985. Evaluation ofF] hybrids on the Cuu Long Delta, Vietnam. Int. Rice Res. News!. 1985(3):19. Nguyen, V. L., H. T. Minh, and N. V. Suan. 1994. Hybrid rice research in Vietnam. p. 187-193. In: S. S. Virmani (ed.), Hybrid rice: New developments and future prospects. Int. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines. Nguyen, V. 1., V. S. Nguyen, and S. S. Virmani. 1995. Current status and future outlook on hybrid rice in Vietnam. p. 73-80. In: G. 1. Denning and V. T. Xuan (eds.), Vietnam and IRRI: A partnership in rice research. Int. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines. Nhan, N. 1'., H. T. Minh, N. T. Hoai, and V. T. Thuy. 1997. Using the thin layer cell culture technique for multiplication of TGMS lines in rice. p. 160-161. In: J. G. Coors, S. Pandey, M. V. Grinkel, A. R. Hallauer, D. C. Hess, K. R. Lamkey, A. E. Melchinger, G. Srinivasan, and C. W. Stuber (eds.), The genetics and exploitation of heterosis in crops. Proc. Int. CIMMYT Symp., 17-22 August 1997. ASA, CSSA, SSSA, Madison, WI, CIMMYT. Mexico. (Abstr.) den Nijs, A. P. M., and G. E. van Dijk. 1993. Chapter 15. Apomixis. p. 229-245. In: M. D. Hayward, N. O. Bosemark, and I. Romagosa (eds.), Plant breeding: Principles and prospects. Chapman & Hall, London. Niu, J., Y. P. Li, and Z. X. Niu. 1997. Studies on the breeding of japoniea photoperiodsensitive male sterile line via anther culture. p. 164-169. In: Proc. Int. Symp. on Twoline System Heterosis Breed. in Crops, Changsha, Hunan, China, September 6-8,1997. Nagler, G. A. 1975. Genetics of apospory in Rant/neulus aurieomus. IV. Embryology of VI and F4 backcross offspring. Phytomorphology 25:485-490. Nagler, G. A. 1984a. Gametophytic apomixis. p. 475-518. In: B.M. Johri (ed.), Embryology of angiosperms. Springer-Verlag, Berlin. Nagler, G.A. 1984b. Genetics of apospory in apomictic Ranunculus aurieomus. V. Conclusions. Bot. Helvet. 94:411-422. Oard, J. H., and J. Hu. 1995. Inheritance and characterization of pollen fertility in photoperiodically sensitive rice mutants. Euphytica 82:17-23. Oard, J. H., J. Hu, and J. N. Rutger. 1991. Genetic analysis of male sterility in rice mutants with environmentally influenced levels of fertility. Euphytica 55:179-186. Oka, H. 1953. The mechanism of sterility in the intervarietal hybrid. Japan. J. Breed. 3:217-224. Oka, H. 1957. Phylogenetic differentiation of the cultivated rice. XVIII. Sterility and vigor of lines derived from hybrids between distant varieties of rice in crosses with parental varieties. Japan. J. Breed. 7:7-11. Oka, H. I. 1958. Intervarietal variation and classification of cultivated rice. Indian J. Genet. Plant Breed. 18:79-89. Oka, H. I. (ed.). 1983. The indica-japoniea differentiation of rice cultivars: A review. SABRAO, Malaysia. Oka, H. I. 1988. Origin of cultivated rice. Japan Scientific Societies Press, Tokyo. Osman, M., H. P. Guok, and O. B. Omar. 1988. Hybrid rice breeding in Malaysia. p. 289. In: Hybrid rice (Proc. Int. Symp. on Hybrid Rice, October 6-10,1986, Changsha, China). Int. Rice Res. Inst., P.O. Box 933, 1099 Manila, Philippines. Ozias-Atkins, P. E. L. Lubbers, W. W. Hanna, and J. W. McNay. 1993. Transmission of the apomictic mode of reproduction in Pennisetum: Co-inheritance of the trait and molecular markers. Theor. Appl. Genet. 85:632-638. Pan, K. Q. 1979. Cytological and histological investigations on pollen abortion in male sterile line of wild rice. Acta Genet. Sinica 1979(2):211-215.

140

J.

LI AND 1. YUAN

Pan, K. Q., and 1. Q. He. 1981a. The relationship between pollen abortion and tapetum in male-sterile line of rice 424, 131. J. South China Agr. College 1981(3):39-45. Pan, K. Q., and L. Q. He. 1981b. Cytological and anatomical observation of stamens of a rice hybrid with "wild pollen-abortive" cytoplasm and its three parental lines. Guangdong Agr. Sci. 1981(1):18-20. Pan, K. Q., Z. X. Wang, and Q. R. He. 1982. Cytological studies on pollen abortion ofTianrei 409 A in rke. Yunnan Agr. Sci. 1982(2):28-30. Panawar, D. V. S., R S. Paroda, and A. Singh. 1983. Heterosis in rice. Indian J. Genet. 43:363-369. Parlevliet, J. K, and J. W. Cameron. 1959. Evidence on the inheritance ofnucellar embryony in Citrus. Proc. Am. Soc. Hort. Sci. 74:252-260. Parmar, K. S., E. A. Siddiq, and M. S. Swaminathan. 1979. Chemical induction of male sterility in rice. Indian J. Genet. 39:529-541. Parmasivian, K. S. 1986. Estimation of heterosis in hybrids and segregating pattern in F 2 generation for apiculus pigmentation in spikelets of rice (Oryza sativa L.). Madras Agr. J. 73:573-578. Pei, X. S. 1980. A new view on inheritance of male sterility in plant. Hunan Agr. Sci. Tech. 1980(3):1-12. Peng, G. F., Z. Y. Li, W. H. Li, and C. X. Zhang. 1993. Studies on the fertility of the temperature sensitive male sterile line 913 in Dian-type hybrid rice and its application. J. Yunnan Agr. Univ. 1993(4):343-346. Peng, J. Y., J. C. Glaszmann, and S. S. Virmani. 1988. Heterosis and isozyme divergence in indica rice. Crop Sci 28:561-563. Peng, J. Y., and S. S. Virmani. 1990. Combining ability for yield and four related traits in relation to hybrid breeding in rice. Oryza 27:1-10. Peng, J. Y., S. S. Virmani, and A. W. Julfiquar. 1991. Relationship between heterosis and genetic divergence in rice. Oryza 28:129-133. Peng, S., G. S. Khush, and K. G. Cassman. 1994. Evolution of the new plant ideotype for increased yield potential. p. 5-20. In: K. G. Cassman. (ed.), Breaking the yield barrier. Int. Rice Res. Inst., P.O. Box 933, 1099 Manila, Philippines. Peng, Y. F., and Y. R Wang. 1991. Studies on starch and protein contents of HPGMR during pollen abortion. Acta Scientiarum Naturalium Universitatis Sunyatseni 1991(4):92-99. Perez, A. T., T. T. Chang, H. M. Beachell, B. S. Vergara, and A. P. Marciano. 1973. Induction of male sterility in rice with Etherel and RH-531. SABRAO Newsl. 1973(2):133-139 Pernes, J., R Rene-Chaume, J. Rene, and Y. Savidan. 1975. Schema d'amelioration genetique des complexes agamiques du type Panicum. Cah ORSTOM, ser. Bio. 10:67-75. Pham, T. M., M. H. Vu, C. B. Bui, V. L. Nguyen. 1991. Hybrid rice yield trials in the Mekong Delta. Int. Rice Res. Newsl. 1991(5):8. Pillai, M. S. 1961. Hybrid vigor in rice. Rice News Teller 1961(1):15-17. Pingali, P. 1., M. Hossain, and R V. Gerpacio. 1997. Asian rice bowls: The returning crisis? p. 208-233. CAB Int. and Int. Rice Res. Inst., Oxon, U.K. Ponnuswamy, A. S., and S. R Prabagaran. 1997. A sustainable substitute to gibberellic acid for hybrid rice seed production. Madras Agr. J. 84:384-385. Powers, L. 1945. Fertilization without reduction in guayule (Parthenium argentatum Gray) and a hypothesis as to the evolution of apomixis and polyploidy. Genetics 30:323-346. Prabagaran, S. R, and A. S. Ponnuswamy. 1997a. Determination of ideal row ratio for hybrid rice seed production. Madras Agr. J. 84:230-231. Prabagaran, S. R, and A. S. Ponnuswamy. 1997b. Determination of optimum growth stage

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

141

for gibberellic acid application in hybrid rice seed production. Madras Agr. J. 84:231-232. Pradhan, S. B., S. N. Ratho, and P. J. Jachuck. 1990. Development of new cytoplasmic genetic male sterile lines through indica-japonica hybridization in rice. Euphytica 51:127-130. Prakash, B. G., and M. Mahadevappa. 1987. Evaluation of some experimental rice hybrids for field performance and standard heterosis. Olyza 24:75-78. Prasad, M. N., and S. S. Virmani. 1989. Optimum distance of isolation for hybrid rice seed production. Int. Rice Res. News!' 1989(3):4-5. Prasad, M. N., S. S. Virmani, and A. D. Gamutan. 1988. Substituting urea and boric acid for gibberellic acid in hybrid rice seed production. Int. Rice Res. News!' 1988(6):9-10. Purohit, D. C. 1972. Heterosis in rice. Madras Agr. J. 59:335-339. Qian, Q., H. Chen, Z. X. Sun, and L. H. Zhu. 1996. The study on determining true and false hybrid rice II You 63 using RAPD molecular markers. Chinese J. Rice Sci. 1996(4):241-242. Qian, Q., X. D. Zhu, D. L. Zen, Z. M. Xiong, and S. K. Min. 1995. Linkage analysis ofHubei photoperiod-sensitive genic male sterile locus in rice. Acta Agr. Zhejiangensis 7:429-433. Raj, K. G., and E. A. Siddiq. 1986. Biochemical characterization of normal and male sterile anthers in rice Oryza sativa L. Indian J. Genet. Plant Breed. 46:541-549. Raj, K. G., and S. S. Virmani. 1988. Genetics of fertility restoration of WA type cytoplasmic male sterility in rice. Crop Sci. 28:787-792. Ram, J., and D. V. S. Panwar. 1970. Interspecific divergence in rice. Indian J. Genet. 1970(1):1-10. Ramalingam, J., N. Nadarajan, C. Vanniarajan, and P. Rangaswamy. 1995. Genetics of fertility restoration and its allelic relationship in rice (Olyza sativa L.). J. Gen. Breed. 49:265-268. Ramanujam, S. 1935. Male sterility in rice. Madras Agr. J. 23:190. Ramiah, K. 1935. Rice genetics. Proc. Assoc. Econ. Bio. Coimbatore 3:51-61. Ramiah, K., S. Jobiraz, and S. D. Mudaliar. 1931. Inheritance of characters in rice. Part IV. Mem. Dept. Agr. India. Bot. Ser. 18:229-250. Ramiah, K., and N. Parthasarathy. 1933. Inheritance of grain length in rice (Olyza sativa L.). Indian J. Agr. Sci. 3:808-819. Ramiah, K., and K. Rangaswamy. 1941. Hybrid vigor in rice (Oryza sativa L.). Indian J. Genet. Plant Breed. 1:4-12. Ramiah, K., and M. B. V. N. Rao. 1953. Rice breeding and genetics. Sci. Monogr., Indian Council Agr. Res., New Delhi. Vol. 19. Rangaswamy, M., and P. Jayamani. 1998. Diversifying cytoplasmic sources in rice. Int. Rice Res. Notes 1998(2):5. Rangaswamy, M., M. N. Prasad, S. R. S. Rangasamy, S. S. Virmani. E. A. Siddiq, T. B. Ranganathan, W. W. Manual, K. Thiyagarajan, P. Jayamani, V. Palanisamy, K. Angamuthu, A. S. Ponnusamy, G. J. Martin, P. Thangamani, and R. Velusamy. 1994. CORH1: The first rice hybrid for Tamil Nadu, India. Int. Rice Res. Notes 1994(3):19. Rao, A. V., A. S. R. Prasad, S. T. Krishina, D. V. Seshu, and T. E. Srinivasan. 1981. Genetic divergence among some brown planthopper resistant rice varieties. Indian J. Genet. 1981(2):179-185. Rao, G. M. 1965. Studies on hybrid vigor in intervarietal hybrid of rice (Olyza sativa L.). Andhra Agr. J. 12:1-12. Raa, Y. S. 1988. Cytohistology of cytoplasmic male sterile lines in hybrid rice. p. 115-128.

142

J. LI AND 1,. YUAN

In: Hybrid rice (Proc. Int. Symp. on Hybrid Rice, October 6-10,1986, Changsha, China). lnt. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines. Rao, Y. S., and S. C. Xie. 1983. Cytological studies on pollen abortion in 'V20A' offsprings of anther culture of rice. Department of Biology, Hunan Teachers' College, Changsha, China. Razzaque, C. A. 1975. Genetics of male sterility in rice. Indian J. Genet. Plant Breed. 34:303-308. Reddy, O. U. K, E. A. Siddiq, J. Ali, A. J. Hussain, and M. 1. Ahmed. 1998a. Fertilityaltering conditions of promising thermo-sensitive genic male sterile lines in rice. Int. Rice Res. Notes 1998(2):6. Reddy, O. U. K, E. A. Siddiq, J. Ali, A. J. Hussain, and M. I. Ahmed. 1998b. Genetics of thermosensitive genic male sterile lines in rice. Int. Rice Res. Notes 1998(1):10. Reddy, O. U. K, E. A. Siddiq, J. Ali, A. J. Hussain, P. Arti, M. I. Ahmed, and N. P. Sharma. 1998b. Identifying molecular markers for the gene(s) governing thermosensitive genic male' sterility in rice. Int. Rice Res. Notes 1998(2):10-11. Reddy, R. P., and K Sathyanarayanaiah. 1981. Inheritance of aroma in rice. Indian J. Genet. 40:327-329. Rick, C. M. 1948. Genetics and development of nine male sterile tomato mutants. Hilgardia 18:599-633. Rothschild, G. H. 1,. 1998. IRRI's role and vision for hybrid rice. In: S. S. Virmani, E. A. Siddiq, and K. Muralidharan (eds.), Advances in hybrid rice technology (Proc. 3rd Int. Symp. Hybrid Rice, Hyderabad, India, October 1996). Int. Rice Res. Inst., P.O. Box 933, 1099 Manila, Philippines. Rowell, P. 1,., and D. G. Miller. 1971. Induction of male sterility in wheat with 2chloroethylphosphonic acid (Ethrel). Crop Sci. 11:629-631. Rundfeldt, H. 1960. Untersuchungen zur Ziichtung des kopfkohls (B. oleracea L. var. capitata). Z. Pflanzenziicht. 44:30-62. Rutger, J. N. 1992. Searching for apomixis in rice. p. 36-39. In: Proc. Apomixis Workshop. U.S. Dept. Agr., Beltsville, MD. Rutger, J. N. 1997. Rice genetics and germplasm enhancement. Research Ser.-Arkansas Agr. Expt. Sta. 1997(456):22-27. Rutger, J. N., and G. W. Schaeffer. 1990. An environmental sensitive genetic male sterile mutant in rice. p. 25. In: Proceedings of Twenty-third Rice Technical Working Group in USA, February 26-28, 1990. Biloxi, Mississippi. Rutger, J. N., and C. Shinjyo. 1980. Male sterility in rice and its potential use in breeding. In: Innovative approaches to rice breeding: Selected papers from 1979 International Rice Research Conference. Int. Rice Res. Inst., P.O. Box 933,1099 Manila, Philippines. Sahai, V. N., and R. C. Chaudhary. 1985. Standardization of parameters for seed production techniques in hybrid rice. Seeds & Farms 11:9-14. Sahai, V. N., S. Saran, and R. C. Chaudhary. 1987. Hybrid rice research in Bihar, India. Int. Rice Res. Newsl. 1987(2):23. Saini, S. S., and I. Kumar. 1973. Hybrid vigor for yield and yield components in rice. Indian J. Genet. Plant Breed. 1973(2):197-200. Saini, S. S., I. Kumar, and M. R. Gagneia. 1974. A study on heterosis in rice (Oryza sativa 1,.). Euphytica 23:219-224. Sakamoto, W., K Kadowaki, and T. Yabuno. 1990. Analysis of mitochondria DNAs from Oryza glaberrima and its cytoplasmic substituted line for Oryza sativa associated with cytoplasmic male sterility. Japan. J. Genet. 1990(1):1-6. Sampath, S., and H. K Mohanty. 1954. Cytology of semi-sterile rice hybrids. Curro Sci. 23:182-183.

2. HYBRID RICE: GENETICS. BREEDING, AND SEED PRODUCTION

143

Sano, Y. 1985. Interspecific cytoplasm substitutions of an indica strain of Oryza sativa and Oryza glaberrima. Euphytica 34:587-592. Sano, Y, and M. Eiguchi. 1991. A nuclear gene inducing fertility restoration in cytoplasmic male-sterile rice. Rice Genet. Newsl. 8:113-115. Sano, Y., M. Eiguchi, H. Y. Hirano, and M. Yamada. 1992. A nuclear gene modifying instability of fertility restoration in cytoplasmic male-sterile rice. Genet. Res. Camb. 60:195-200. Satake, T., and S. Yoshida. 1978. High temperature-induced sterility in indica rice at flowering. Japan. J. Crop Sci. 47:6-7. Sato, H., N. Shimizu, H. Nakagawa, and K. Nakajima. 1990. A new registered cultivar "Natsukaze" of guineagrass (Panicum maximum). Japan. Agr. Res. Quarterly 23:196-201. Sato, S., L. Sakamoto, and S. Nakasone. 1985. Location ofEf-1 for earliness on Nishimura's seventh chromosome. Rice Genet. Newsl. 2:59-60. Sato, S., and C. Shinjyo. 1991. Genetical studies on genic male sterile mutant obtained through breeding of isogenic translocation line of Taichung 65 in rice Oryza sativa L. Japan. J. Breed. 41:451-460. Sattari, M. 1997. Identification heterotic hybrid for exploitation of hybrid vigor in Iranian rice cultivars. p. 344-345. In: J. G. Coors, S. Pandey, M. V. Grinkel, A. R. Hallauer, D. C. Hess, K. R. Lamkey, A. E. Melchinger, G. Srinivasan, and C. W. Stuber (eds.), The genetics and exploitation of heterosis in crops. Proc. Int. CIMMYT Symp., 17-22 August 1997. ASA, CSSA, SSSA, Madison, WI, CIMMYT. Mexico. (Abstr.) Satyanarayana, P. V., I. Kumar, and M. S. S. Reddy. 1995. A new source of thermosensitive genic male sterility for 2-line hybrid rice breeding. Int. Rice Res. Notes 1995(1):10. Savidan, Y. 1982. Nature et heredite de l'apomixie chez Panicum maximum Jacq. Travaux et documents ORSTOM, Paris. 153:1-159. Savidan, Y. 1983. Genetics and utilization of apomixis for the improvement of guineagrass (Panicum maximum Jacq.). p. 182-184. In: J. A. Smith and V. W. Hayes (eds.J, Proc. Int. Grassl. Congr., 14th, Lexington, KY, June 15-24, 1981. Westview Press, Boulder. CO. Savidan, Y. 2000. Apomixis: Genetics and breeding. Plant Breed. Rev. 18: chapter 2. Savidan, Y., O. Leblanc. and J. Berthaud. 1994. Transferring apomixis to maize. In: G. S. Khush (ed.), Apomixis: exploiting hybrid vigor in rice. Int. Rice Res. Inst., P.O. Box 933, 1099 Manila, Philippines. Schwartz, D., and W. J. Laughner. 1969. A molecular basis for heterosis. Science. 166:626-627. Seetharaman, R. 1959. Inheritance of iodine value in rice and its association with other characters. Diss. Abstr. 20:856. Sehgal, S. M. 1994. Capabilities and limitations of the global seed industry in hybrid rice development. p. 257-266. In: S. S. Virmani (ed.), Hybrid rice technology: New developments and future prospects. Int. Rice Res. Inst., P.O. Box 933, 1099 Manila, Philippines. Sen, P. K., and G. N. Mitra. 1958. Inheritance of photoperiodic reaction in rice. Nature (Lond.) 182:119. Shamsi, M. A., B. K. Murthy, D. S. Manyam, and M. V. B. Rao. 1996. Effect of chemical gametocides on pollen sterility and seed set in rice (Oryza sativa L.). Indian J. Genet. Plant Breed. 56:100-103. Shanghai Plant Physiology Research Institute. 1977. Index of physiology and biochemistry in hybrid rice. Hunan Agr. Sci. Tech. 1977(1):40-51 Shao, K. X., and D. W. Hu. 1988. Chemical emasculators for hybrid rice. In: Hybrid rice (Proc. lnt. Symp. on Hybrid Rice, October 6-10,1986, Changsha. China). Int. Rice Res. Inst., P.O. Box 933, 1099 Manila, Philippines.

144

J. 11 AND L. YUAN

Shao, Y., and X. H. Tang. 1992. Analysis ofrestorer genes for photoperiod sensitive malesterile rice. Rice Genet. Newsl. 9:52-54. Shao, Y., and X. H. Tang. 1993. Studies on the relationship of allelic sterile genes between photoperiod-sensitive sterile rice and BT-type sterile rice. J. Shanghai Agr. College 1993(4):273-277. Shao, Y., and X. H. Tang. 1995. Allelic relation between photoperiod-sensitive genic male-sterility and cytoplasmic-genic male-sterility ofBT type in rice. Rice Genet. Newsl. 12:231-232. Shao, Y., X. H. Tang, Q. L. Huang, and G. Z. Pan. 1993. Studies on the sterile genes and its restoring genes in photoperiod-sensitive male sterile rice. Acta Agr. Shanghai 1993(3):22-29. Sharma, H. L., H. Singh, and D. P. Joshi. 1987. Minimum isolation distance for hybrid rice production. Int. Rice Res. Newsl. 1987(2):24. Sharma, H. L., and S. S. Virmani. 1990. Some varietal sources of twin seedlings trait in rice. Rice Genet. Newsl. 7:134-135. Sharma, H. L., and S. S. Virmani. 1994. Optimum pollen load and male-female row ratio for hybrid rice seed production. p. 294. In: S. S. Virmani (ed.), Hybrid rice technology: New developments and future prospects. Int. Rice Res. Inst., P.O. Box 933, 1099 Manila, Philippines. Shen, J. H. 1980. Rice breeding in China. In: Rice improvement in China and other Asian countries. Int. Rice Res. Inst., P.O. Box 933, 1099 Manila, Philippines. Shen, Y., Q. CaL M. Gao, and X. Wang. 1996a. Isolation and genetic characterization of a fertility-restoring revertant induced from cytoplasmic male sterile rice. Euphytica 90:17-23. Shen, Y., Z. Guan, H. Qiang, J. Lu, K. Zheng, and X. Wang. 1996. Identification of RAPD markers linked to fertility restoring gene of rice cytoplasmic male-sterile line. Rice Genet. Newsl. 13:115-117. Shen, Y. W., Q. H. Cai, and M. W. Gao. 1993a. Preliminary characterization of fertilityrestoring mutants induced from cytoplasmic male-sterile rice. Rice Genet. Newsl. 10:90-92. Shen, Y. W., Q. H. Cai, and M. W. Gao. 1993b. A new thermosensitive radiation-induced male-sterile rice line. Rice Genet. Newsl. 10:97-98. Shen, Z. T., and Y. B. Xu. 1992. Genetic analysis for hybrid sterility of intersubspecific crosses in rice. Chinese J. Rice Sci. 1992(3):105-112. Sheng, X. B. 1987. Genetic effects of cytoplasm on agronomic characteristics in hybrid rice. Chinese J. Rice Sci. 1987(3):155-170. Sheng, X. B. 1992. Genetics of photoperiod sensitive genic male sterility of Nong-Ken 58s (Oryza sativa). Chinese J. Rice Sci. 1992(1):5-14. Sheng, X. B. 1994. The review and prospect of hybrid rice male sterility genetics research in China. Hybrid Rice 1994(3-4):58-62. Shi, C. H., G. L. Chen, J. Zhu, R. C. Zang, and J. G. Wu. 1998. Analysis of genetic effects and heritabilities for amylose content of indica rice in different environments. Chinese Rice Res. Newsl. 1998(2):10-12. Shi, G. C., P. C. Ni, and J. X. Song. 1996a. The identification of apomixis in rice variety 84-15. Sci. Agr. Sinica. 1996(5):56-61. Shi, G. c., P. C. Ni, J. X. Song, X. J. Mu, and T. B. Liang. 1996b. A cytoembryological study on rice 84-15, a rice line presumed to be apomictic. Acta Agron. Sinica. 22:657-660. Shi, M. S. 1981. The preliminary report on the breeding and application of a late-season japonica natural dual-purpose male sterile line in rice (in Chinese). Hubei Agr. Sci. 1981(7):1-3.

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

145

Shi, M. S. 1985. Discovery of and preliminary studies on a photosensitive recessively male sterile rice. Sci. Agr. Sinica 1985(2):44-48. Shi, M. S., and J. Y. Deng. 1986. The discovery, determination and utilization of the Hubei China photosensitive genic male-sterile rice Oryza-sativa-ssp-japonica. Acta Genet. Sinica 13:107-112. Shi, S. Y., A. M. Liu, Q. H. Yi, and Z. K. Zhong. 1988a. Application of isozyme analysis to selection for restorer lines in hybrid rice. Jiangsu J. Agr. Sci. 1988(3):9-14. Shi, S. Y., Q. H. Yi, J. R. Jiang, and A. M. Liu. 1988b. Using isozymes to determine purity of hybrid rice seed. p. 274. In: Hybrid rice (Proc. Int. Symp. on Hybrid Rice, October 6-10,1986, Changsha, China). Int. Rice Res. lnst., P.O. Box 933, 1099 Manila, Philippines. Shibuya, N. 1973. Studies on causal genes of partial male-sterility in rice and their utilization in testing cool temperature tolerance. Bul. Yamagata Univ. Agr. Sci. 6:571-625. Shinjyo, C. 1969. Cytoplasmic genetic male sterility in cultivated rice (Oryza sativa L.). II. The inheritance of male sterility. Japan. J. Genet. 44:149-156. Shinjyo, C. 1975. Genetic studies of cytoplasmic male sterility and fertility restoration in rice (Ol}'za sativa L.). Sci. Bul. College Agr. Univ. Ryukyus 22:1-57. Shinjyo, C. (ed.). 1984. Cytoplasmic male sterility and fertility restoration in rice having genome A. p. 321-338. In: Biology of rice. Japan Sci. Soc. Press, Tokyo. Shinjyo, C. 1990. Identification of male sterile cytoplasms and fertility restoring genes derived from wild species. p. 106-111. In: N. Iwata (ed.), Exploitation and identification of rice gene resources (Report of Grant-in-Aid for Sci. Res.). Shinjyo, c., and T. Omura. 1966. Cytoplasmic male sterility in cultivated rice (Ol}'za sativa L.). I. Fertility ofF 1 , F2 and offsprings obtained from their mutual reciprocal backcrosses; and segregation of completely male sterile plants (in Japanese). Japan. J. Breed. 16 (suppl.):179-180. Shinjyo, c., and S. Sato. 1994. Chromosomal location of fertility-restoring gene Rf-2. Rice Genet. Newsl. 11 :93-95. Shinjyo, c., and Y. Watanabe. 1977. Allelism test of two genes showing fertility-restoring effect for male sterile cytoplasm of Lead rice and their inheritance. Japan. J. Breed. 27:70-71. Shu, Q. F., S. S. Wang, and Q. M. Yi. 1989. Two-dimensional gel electrophoresis of endosperm proteins in male sterile rice. Hereditas (Beijing) 1989(3):5-8. Shull, G. H. 1908. The composition of a field of maize. Am. Breed. Assoc. Rep. 4:296-301. Shull, G. H. 1952. Chapter 2. Beginnings of the heterosis concept. p. 14-48. In: J. W. Gowen (ed.), Heterosis. Iowa State College Press, Ames. Siddiq, E. A. 1994. Hybrid rice in India. Int. Rice Comm. Newsl. 43:29-31. Siddiq, E. A., P. J. Jachuck, M. Mahadevappa, F. U. Zaman, R. V. Kumar, B. Vidyachandra, G. S. Sidhu, I. Kumar, M. N. Prasad, M. Rangaswamy, M. P. Pandey, D. V. S. Panwar, and I. Ahmed. 1994. Hybrid rice research in India. p. 157-171. In: S. S. Virmani (ed.), Hybrid rice: New developments and future prospects. Int. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines. Singh, G. B., A. S. Randhawa, G. S. Magnat, and G. S. Sidhu. 1997. Studies on relative seed set and seed yield in different CMS lines of rice. J. Res. (Punjab Agr. Univ.) 1997(2):125-130. Singh, N. B., H. G. Singh, and P. Singh. 1977. Heterosis and combining ability for quality component in rice. Indian J. Genet. 37:347-352. Singh, R. 1997. Studies on heterosis for certain physiological and biochemical parameters in rice. p. 164-165. In: J. G. Coors, S. Pandey, M. V. Grinkel, A. R. Hallauer, D. C. Hess, K. R. Lamkey, A. E. Melchinger, G. Srinivasan, and C. W. Stuber (eds.), The genetics

146

J. LI AND L. YUAN

and exploitation of heterosis in crops. Proc. Int. CIMMYT Symp., 17-22 August 1997. ASA, CSSA, SSSA, Madison, WI, CIMMYT. Mexico. (Abstr.) Singh, R., and S. K. Sahoo, 1997. Artificial and natural parameters influencing outcrossing of cyto-sterile lines in rice. p. 162-163. In: J. G. Coors, S. Pandey, M. V. Grinkel, A. R. Hallauer, D. C. Hess, K. R. Lamkey, A. E. Melchinger, G. Srinivasan, and C. W. Stuber (eds.), The genetics and exploitation of heterosis in crops. Proc. Int. CIMMYT Symp., 17-22 August 1997. ASA, CSSA, SSSA, Madison, WI, CIMMYT. Mexico. (Abstr.) Singh, R. J., and H. Ikehashi. 1981. Monogenic male sterility in rice: induction, identification and inheritance. Crop Sci. 21:286-289. Singh, R. P., and R. R. Singh. 1979. Heterosis in rice. 0IYza 17:119-122. Singh, S., B. C. Viraktamath, C. H. M. Vijayakumar, M. S. Ramesha, and M. I. Ahmed. 1998. Standardizing hybrid rice seed production practices. Int. Rice Res. Notes 1998(1):22. Singh, S. P., and H. G. Singh. 1977. Heterosis in rice. 0IYza. 15:173-175. Singh, S. P., H. G. Singh, and A. K. Singh. 1984. Hybrid vigor for yield and its components in rice. Genet. Iber. 36:53-59. Singh, S. P., R. P. Singh, and R. V. Singh. 1980. Heterosis in rice. OIyza 18:109-113. Singh, U. S. 1983. Hybrids excel standard cultivar. Deep Water Rice 1:1. Sivasubranian, S., and M. Menon. 1973. Heterosis and inbreeding depression in rice. Madras Agr. J. 60:1139-1144. Sohu, V. S., and P. S. Phul. 1995. Inheritance of fertility restoration of three sources of cytoplasmic male sterility in rice. J. Gen. Breed. 49:93-96. Solntzeva, M. P. 1978. Apomixis and hemigamy as one ofits forms. Proc. Indian Nat. Sci. Acad. Part B. 44,78. Somoto, S., and K. Hamamura. 1973. Inheritance of grain characters in rice plant. Hokuriku Crop Sci. 8:11-13. Somrith, B., 1'. T. Chang, and B. R. Jackson. 1979. Genetic analysis of traits related to grain characteristics and quality in two crosses of rice. IRRI Res. Pap. Ser. 35. Song, X. F., Z. M. Xiong, and S. K. Min. 1990. Biological characteristics of competitive heterosis in indica/japonica intersubspecific rice hybrids. p. 129-134. In: Selected papers on the development of P(T)GMS rice lines and utilization of intersubspecific heterosis. China's Ministry Agr., Beijing. Sood, B. C., and E. A. Siddiq. 1978. A rapid technique for scent determination in rice. Indian J. Genet. 38:268-271. Sprague, G. F. 1983. Chapter 2. Heterosis in maize: Theory and practice. p. 47-70. In: R. Frankel (ed.), Heterosis: Reappraisal of theory and practice. Springer-Verlag, Berlin. Srivastava, H. K. 1972. Mitochondrial complementation and hybrid vigor. Indian J. Genet. Plant Breed. 32:215-228. Srivastava, H. K. 1983. Chapter 10. Heterosis and intergenomic complementation: mitochondria, chloroplast, and nucleus. p. 260-286. In: R. Frankel (ed.), Heterosis: Reappraisal of theory and practice. Springer-Verlag, Berlin. Srivastava, M. N., and D. V. Seshu. 1983. Combining ability for yield and associated characters in rice. Crop Sci. 23:741-744. Stansel, J. W. 1966. Influence of heredity and environment on endosperm characteristics ofrice (Oryza sativa 1.). Diss. Abstr. 27:48b. Stuber, C. W., S. E. Lincoln, D. W. Wolff, 1'. Helentjaris, and E. S. Lander. 1992. Identification of genetic factors contributing to heterosis in a hybrid from two elite maize inbred lines using molecular markers. Genetics 132:823-839. Subandi, B., B. Suprihatno, D. A. Sihombing, and S. Suemardhi. 1987. Research and development of hybrid variety of maize and rice in Indonesia. Indonesian Agr. Res. Dev. J. 1987(1-2):13-18.

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

147

Subramanian, S., and M. Rathinam. 1984. Heterosis in rice. Madras Agr. J. 71:402-405. Subudhi, P. K., R. P. Borkakati, S. S. Virmani, and N. Huang. 1995. Identification ofRAPD markers linked to rice thermosensitive genetic male-sterility gene by bulk segregant analysis. Rice Genet. News!. 12:228-231. Subudhi, P. K., R. P. Borkakati, S. S. Virmani, and N. Huang. 1996. Inheritance and molecular mapping of the thermosensitive genetic male-sterility gene in rice. p. 601-606. In: Rice Genetics II. Int. Rice Res. Inst., P.O. Box 933, 1099 Manila, Philippines. Subudhi, P .K., R .P. Borkakati, S. S. Virmani, and N. Huang. 1997. Molecular mapping of a thermosensitive genetic male sterility gene in rice using bulked segregant analysis. Genome 40:188-194. Suh, H. S. 1988. Multiple pistillate male sterile rices for hybrid seed production. p. 181-187. In: Hybrid rice (Proc. Int. Symp. on Hybrid Rice, October 6-10,1986, Changsha, China). Int. Rice Res. Inst., P.O. Box 933,1099 Manila, Philippines. Suh, H. S., J. D. Kim, M. H. Heu, and G. S. Chung. 1991. Linkage relationships between genetic male sterile and marker genes in rice. p. 121-125. In: Rice Genetics II. Proc. 2nd Int. Rice Genetics Symp., Int. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines. Suh, H. S., J. C. Koh, and S. K. Lee. 1989. Identification of four genes for genetic male sterility in rice. p. 317-320, In: S. Iyama and G. Takeda (ed.), Breeding research: The key to the survival of the earth. Proc. 6th Int. Congr. of SABRAO. Suherman, 0.1989. Performance of hybrid rice in Indonesia. Int. Rice Res. News!. 1989(4):9. Sun, J. S., and Y. S. Liu. 1996. Current status and potential of utilization of apomixis in rice breeding. China Rice 1996(6):1-2. Sun Vat Sen Univ. 1976. The study of biological characters of "three lines" in crops: III. The comparative research to the formation and embryological characters of pollen cells, metabolic block and development of the vascular bundle in anther septum of several MS types in rice. Acta Genet. Sinica 3:119-127. Sun, Y. W., J. Q. Li, and Y. B. Liu. 1993. Heterosis and combining ability for the days to heading of the 'two-line system' intersubspecific hybrid rice. Chinese J. Rice Sci. 1993(2):105-108. Sun, Z. X., and S. H. Cheng. 1994. Hybrid rice breeding: from three-line system, two-line system to one-line system (in Chinese). China Agr. Sci. Tech. Press, Beijing, China. Sun, Z. X., S. H. Cheng, S. K. Min, Z. M. Xiong, C. S. Ying, and H. M. Si. 1991. Studies on response of photoperiod and temperature to the photoperiod sensitive genic male sterile rice (PGMSR): II. Identification of japonica PGMSR in growth chamber. Chinese J. Rice Sci. 1991(2):56-60. Sun, Z. X., S. K. Min, and Z. M. Xiong. 1989. A temperature-sensitive male-sterile line found in rice. Rice Genet. News\. 6:116-117. Suprihatno, B. 1986. Hybrid rice: its prospects and problems in Indonesia. Indonesian Agr. Res. Dev. J. 1986(3-4):51-58. Suprihatno, B., B. Sutaryo, and T. S. Silitonga. 1994. Hybrid rice research in Indonesia. p. 196-206. In: S. S. Virmani (ed.), Hybrid rice: New developments and future prospects. Int. Rice Res. Inst., P.O. Box ~l33, Manila 1099, Philippines. Suzuki, 1., Y. Vagi. H. Hashimoto, A. Shimizu, and T. Kato. 1990. The development of a sorting machine for rice seeds. Japan. J. Breed. 40(suppl 2):374-375. Suzuki. S. 1978. Anther and pollen abnormalities induced by cold treatment and their varietal differences in rice plants. Japan. J. Breed. 28:21-32. Swaminathan, M. S., E. A. Siddiq, and S. D. Sharma. 1972. Outlook for hybrid rice in India. In: Rice breeding. p. 609-G13. Int. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines. Taillebois, J. 1991. F 1 hybrid rice: prospects for Mediterranean climates. Int. Rice Comm. News!. 40:33-35.

148

J. LI AND L. YUAN

Taillebois, J. 1994. Hybrid rice research at CIRAD/IRAT. p. 253-256, In: S. S. Virmani (ed.), Hybrid rice: New developments and future prospects. Int. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines. Takeda, K. 1987. Open hull male sterile mutant in rice. Rice Genet. Newsl. 4:78-79. Takezaki, y. 1932. Inheritance of a few sterile plants. I. Male sterile rice. Agr. and Hortic. Japan 7:387-397. Taliaferro, C. M., and E. C. Bashaw. 1966. Inheritance and control of obligate apomixis in breeding bufflegrass, Pennisetum ciliare. Crop Sci. 6:473-476. Tamaru, N. 1994. Breeding studies on genetic male-sterility and hybrid sterility in rice (01)1za sativa 1.). Mem. Fac. Agr. Hokkaido Univ. 1994(2):203-256. Tanaka, T., M. Kawase, A. Saito, M. Yano, N. Kishimoto, K. Saito, A. Yoshimura, H. Takeda, T. Nagamine, and M. Nakagahra. 1989. Intraspecific variation on restriction fragment length polymorphism of nuclear DNA in rice, Oryza sativa L. p. 599-602. In: S. Iyama and G. Takeda (ed.), Breeding research: The key to the survival of the earth. Proc. 6th Int. Congr. SABRAO, Tsukuba-machi, Japan. Tang, C. Q., Z. Tong, D. C. Peng, and Q. D. Zhang. 1994. Photosynthesis characteristics of chloroplasts in photoperiod-sensitive genic male sterile rice (Nongken 58s Rice). Acta Phytophysiol. Sinica 20:257-262. Tang, S. X. 1987. A study on the cooking and eating qualities of Chinese hybrid rice (01)1za sativa 1.). Sci. Agri. Sinica 1987(5):17-22. Tang, W. G., G. 1. Yang, G. 1. Li, T. 1. Ye, and Z. B. Yin. 1997. Breeding of the dual-purpose genic male sterile line in sorghum and its utilization. p. 206-214. In: Proc. Int. Symp. on Two-line System Heterosis Breed. in Crops, Changsha, Hunan, China, September 6-8,1997. Tang, X. H., and Y. Shao. 1997. Relationship between the rice heading sensitivity and the male sterile sensitivity to photoperiod in view of developmental genetics. Acta Bot. Sinica. 39:618-622. Tanksley, S. D., and S. R. McCouch. 1997. Seed banks and molecular maps: unlocking genetic potential from the wild. Science 277:1063-1066. Tatum, 1. A. 1971. The southern corn leaf blight epidemic. Science 171:1113-1116. Teng, 1. S., and Z. T. Shen. 1994a. Genetic studies on fertility restoration in cytoplasmic male-sterile rice. Hybrid Rice 1994(3-4):63-66. Teng, 1. S., and Z. T. Shen. 1994b. Inheritance of fertility restoration for cytoplasmic malesterility rice. Rice Genet. Newsl.11:95-97. Teng, 1. T. 1982. Cytological studies on development of sperm and pollen abortion in Nong Jin 2, Fu You 1 A, and B of rice. p. 130-136. In: Selections of graduate theses, Dept. Biology, Hunan Teachers' College, Changsha, China. Terao, H., and U. Mizushima. 1939. Some considerations on the classification of 01)1za sativa 1. into subspecies, so-called japonica and indica. Japan. J. Bot. 1939(3):213-258. Terao, H., and U. Mizushima. 1944. On the affinity of rice varieties cultivated in East Asia and America. Bur. Agr. Expt. Sta., Ministry Agr. Commerce, Japan. Tian, c., X. Cheng, and Z. Liang. 1980. Several views on popularization of Xian (indica) hybrid rice. Yunnan Agr. Sci. Tech. 1980(2):12-18. Tian, D. C. 1991. Cultivation for outcrossing rice. Sichuan Sci. Tech. Press, Chengdu, China. Tomar, J. B., and J. S. Nanda. 1983. Inheritance of aroma in rice (OJYza sativa L.). Madras Agr. J. 70:491-494. Tong, H. J., and H. S. Ma. 1997. Cultivation techniques for increasing the outbred seed setting percentage in seed production of hybrid rice. China Rice 1997(5):14-17. Trees, S. c., and J. N. Rutger. 1978. Inheritance of four genetic male steriles. J. Hered. 69:270-272. Trinh, T. T. 1992. The FAO hybrid rice programme. Int. Rice Comm. Newsl. 41:51-59.

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

149

Trinh, T. T. 1993. New developments in hybrid rice. Int. Rice Comm. Newsl. 42:28-34. Trinh, T. T. 1994. FAO's contribution to hybrid rice development. p. 267-274. In: S. S. Virmani (ed.), Hybrid rice: New developments and future prospects. Int. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines. Tu, Z. P., and D. W.Hu. 1989. Using chemical male gametocides in hybrid rice breeding in China. p. 237-243. In: Progress in irrigated rice research: Selected papers and abstracts from the Int. Rice Res. Conference, September 21-25,1987, Hangzhou, China. United States Department of Agriculture. 1963. Rice gene symbolization and linkage groups. U.S. Dept. Agr. Res. Ser. 34-38:1-56. Vaidyanath, K., and G. M. Reddy. 1985. Studies on genetic divergence in the genus Olyza. J. Cytol. Genet. 20:59-68. Vairavan, S., E. A. Siddiq, V. Arunachalam, and M. S. Swaminathan. 1973. A study on the nature of genetic divergence in rice from Assam and northeast Himalayas. Theor. Appl. Genet. 43:213-221. do Valle, C. B., C. Glienke, and G. O. C. Leguizamen. 1994. Inheritance of apomixis in Brachiaria, a tropical forage grass. Apomixis Newsl. 1994(7):42-43. do Valle, C. B., and Y. Savidan. 1996. Genetics, cytogenetics, and reproductive biology of Brachiaria. p. 147-163. In: J. W. Miles, B. 1. Maass, and C. B. do Valle (eds.), Brachiaria: Biology, agronomy, and improvement. CIAT-EMBRAPA, Cali, Colombia. Vijayakumar, C. H. M., M. 1. Ahmed, B. C. Viraktamath, and M. S. Ramesha. 1997. Selection of parents and prediction of heterosis in rice. p. 18-19. In: J. G. Coors, S. Pandey, M. V. Grinkel, A. R. Hallauer, D. C. Hess, K. R. Lamkey, A. E. Melchinger, G. Srinivasan, and C. W. Stuber (eds.), The genetics and exploitation of heterosis in crops. Proc. Int. CIMMYT Symp., 17-22 August 1997. ASA, CSSA, SSSA, Madison, WI, CIMMYT. Mexico. (AbstL) Vijayakumar, C. H. M., M. S. Ramesha, B. C. Viraktamath, S. Singh, and M. 1. Ahmed. 1998. Analyzing leaf number in parental lines for hybrid rice seed production. Int. Rice Res. Notes 1998(2):27. Viraktamath, B. C., C. H. M. Vijayakumar, M. 1. Ahmed, S. Singh, and M. S. Ramesha. 1998. Leaf number: a reliahle parameter for determining seeding intervals between parental lines in hybrid rice seed production. Int. Rice Res. Notes 1998(2):25-26. Virmani, S. S. 1993. Progress in hybrid rice research in India. Int. Rice Comm. News!. 42:35-38. Virmani, S. S. 1994a. Heterosis and hybrid rice breeding. Springer-Verlag, Berlin and Heidelberg. Virmani, S. S. 1994b. Prospects of hybrid rice in the tropics and subtropics. p. 7-19. In: S. S. Virmani (ed.), Hybrid rice: New developments and future prospects. Int. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines. Virmani, S. S. 1996. Hybrid rice. Adv. Agron. 57::~77-462. Virmani, S. S., and R. D. Dalmacio. HJ87. Cytogenic relationship between two cytoplasmic male sterile lines. Int. Rice Res. Newsl. 1987(1):14. Virmani, S. S., and 1. B. Edwards. 1983. Current status and future prospects for breeding hybrid rice and wheat. Adv. Agron. 36:145-214. Virmani, S. S., R. K. Govinda, C. Casal, R. D. Dalmacio, and P. A. Aurin. 1986. Current knowledge of and outlook on cytoplasmic-genetic male sterility and fertility restoration in rice. p. 633-647. In: Rice genetics. Proc. Int. Rice Genetics Symp. Int. Rice Res. Inst., P.O. Box 9~B, Manila 1099, Philippines. Virmani, S. S., R. V. Kumar, and C. CasaL 1989. Cytoplasmic diversity among CMS lines of rice. Rice Genet. Newsl. 6:127-129. Virmani, S. S., and H. L. Sharma. HJ93. Manual for hybrid rice seed production. Int. Rice Res. Inst., P.O. Box 933, 1099 Manila, Philippines.

150

J. LI AND L. YUAN

Virmani, S. S., and C. Shinjyo. 1988. Current status of analysis and symbols for male sterile cytoplasms and fertility restoring genes. Rice Genet. Newsl. 5:9-15. Virmani, S. S., B. C. Viraktamath, and M. T. Lopez. 1997. Nucleus and breeder seed production of thermosensitive genic male sterile lines. Int. Rice Res. Notes 1997(3): 26-27. Virmani, S. S., and B. H. Wan. 1988. Developments ofCMS lines in hybrid rice breeding. p. 103-114. In: Hybrid rice (Proc. Int. Symp. on Hybrid Rice, October 6-10,1986, Changsha, China). Int. Rice Res. Inst., P.O. Box 933, 1099 Manila, Philippines. Virmani, S. S., J. B. Young, H. P. Moon, I. Kumar, and J. C. Flinn. 1991. Increasing rice yields through exploitation of heterosis. IRRI Res. Pap. Ser. 156. Wagner, R. P. 1969. Genetics and phenogenetics of mitochondria. Science 163:1026-1031. Wan, B. H., X. H. Chen, Y. P. Lu, and K. Q. Liang. 1997. Studies and utilization of rice germplasm with short photoperiod and low temperature inducing male sterility. p. 59-63. In: Proc. Int. Symp. on Two-line System Heterosis Breed. in Crops, Changsha, Hunan, China, September 6-8,1997. Wan, B. H., D. M. Li, and Q. L. Xuan. 1988. Classification of male sterile cytoplasms in rice by their nuclear-cytoplasm interactions. p. 262-263, In: Hybrid rice (Proc. Int. Symp. on Hybrid Rice, October 6-10,1986, Changsha, China). Int. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines. Wan, B. H., and H. F. Ma. 1996. Genetical study on three T(P)GMS germplasms in rice. p. 127-131. In: China National Hybrid Rice Research and Development Center (ed.), Proc. Annual Hainan Meeting China's National Two-line Hybrid Rice Program Plan, Hainan, China. Wan, B. H., and H. F. Ma. 1997. Studies on the inheritance of three types germ plasm in photo-thermo sensitive genic male sterile rice. J. South China Agr. Univ. 1997(2):1-6. Wan, C. C. 1989. Techniques on hybrid rice seed production and cultivation. Agr. Press. Beijing, China. Wan, J. M., and H. Ikehashi. 1997. Identification of two types of differentiation in cultivated rice (Oryza sativa L.) detected by polymorphism of isozymes and hybrid sterility. Euphytica 94:151-161. Wan, J. M., S. Yanagihara, H. Kato, and H. Ikehashi. 1993. Multiple alleles at a new locus causing hybrid sterility between a Korean indica variety and a javanica variety in rice (Oryza sativa 1..). Japan. J. Breed. 43:507-516. Wang, B., Y. N. Li, X. W. Zhang, L. Hu, and J. Z. Wang. 1990. Double-stranded RNA and male sterility in rice. Theor. Appl. Genet. 79:556-560. Wang, B., W. W. Xu, J. Z. Wang, M. L. Weng, M. Y. Li, R. C. Yang, and H. T. Nguyen. 1996. Tagging and mapping the thermosensitive genic male-sterile gene of rice with molecular markers. p. 596-600. In: G. S. Khush (ed.), Rice genetics III. Int. Rice Res. Inst., P.O. Box 933, 1099 Manila, Philippines. Wang, B., W. W. Xu, J. Z. Wang, W. Wu, H. G. Zheng, Z. Y. Yang, J. D. Ray, and H. T. Nguyen. 1995.a. Tagging and mapping the thermo-sensitive genic male-sterile gene in rice (Oryza sativa 1..) with molecular markers. Theor. Appl. Genet. 91:1111-1114. Wang, C. 1.., and Y. G. Tang. 1990. Studies on the genetic effects of sterile cytoplasm in hybrid rice (Oryza sativa L. subsp. sinica). Acta Agron. Sinica 1990(4):335-341. Wang, F. P., M. H. MeL C. G. Xu, and Q. F. Zhang. 1997a. PMS1 is not the locus relevant to fertility difference between the photoperiod-sensitive male sterile rice Nongken 58s and normal rice Nongken 58. Acta Bot. Sinica. 39:922-925. Wang, J. J" Y. B. Xu, and Z. T. Shen. 1991. Exploring the major restraints to the utilization of indica-japonica hybrids in rice. Sci. Agr. Sinica 1991(1):27-33. Wang, J. X., L. Zou, L. P. Xu, Y. T. Cai, and L. S. Liu. 1994. Preliminary identification ofWCG in rice. p. 78-79, National Seminar Two-line System Hybrid Rice, Yangzhou, China.

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

151

Wang, J. Z., B. Wang, Q. F. Xu, D. C. Yang, and Y. G. Zhu. 1995b. Tagging rice photoperiodsensitive genic male sterile gene by RAPD analysis. Acta Genet. Sinica 22:53-58. Wang, K. X. 1983a. Preliminary study of the genotypic model of the maintainer and restorer of WA type male sterile cytoplasm. Fujian Agr. Sci. Tech. 1983(4):22-24. Wang, S. H., C. Q. Li, K. N. Xu, Y. C. Luo and D. Z. Wang. 1994. The effects of ecological conditions on the fertility of photoperiod (thermo)-sensitive genic male sterile rice. J. Anhui Agr. Sci. 1994(2):119-123. Wang, S. L. 1980. Inheritance of R genes in rice and methods for selection of new R lines. Hunan Agr. Sci. Tech. 1980(4):1-4. Wang, S. L. 1983b. The development of R lines: Testcross breeding method in early generations. Hunan Agr. Sci. 1983(2):1-4. Wang, S. L. 1996. Summary on China's three-line hybrid rice breeding during 1991-1995. p. 219-220. In: China National Hybrid Rice Research and Development Center (ed.), Proc. Annual Hainan Meeting China's National Two-line Hybrid Rice Program Plan, Hainan, China. Wang, T., and Z. Tong. 1992. Changes in anther microstructure of the photoperiodsensitive genic male sterile rice. Acta Agron. Sinica 1992(2):132-136. Wang, T., Z. Tong, T. Y. Kuang, and P. S. Tang. 1997b. Analysis of N-terminal sequence of a 41kDa protein from a photoperiod-sensitive genic male-sterile mutant of rice. Acta Bot. Sinica 39:979-982. Wang, T., Y. H. Xiao, and W. F. Liu. 1990. Study on changes of leaf protein in HPGMR during photoperiodic induction. J. Huazhong Agr. Univ. 9:369-374. Wang, X., R. F. Que, Z. R. Jiang, and M. F. Xu. 1981. Physiological effects of Ethrel on the induction of male sterility of rice. Acta Plant Physiol. Sinica 7:11-18. Wang, X. M., M. Q. Wang, G. Z. Mei, H. Y. Wu, W. J. Duan, and W. J. Wang. 1991a. Photoperiod-conditioned male-sterility and its inheritance in rice. p. 217-226. In: Rice Genetics II. Int. Rice Res. Inst., P.O. Box 933,1099 Manila, Philippines. Wang, X., M. Y. Yu, L. X. Tao, and F. D. Yao. 1991b. The male sterility in rice plants induced by CRMS. I. Effect ofCRMS on male sterility in rice plants. Chinese J. Rice Sci. 1991(2):49-55. Wang, X., M. Y. Yu, L. X. Tao, and F. D. Yao. 1991c. The male sterility in rice plants induced by CRMS. II. Physiological effects of CRMS on the inducing male sterility in rice. Chinese J. Rice Sci. 1991(3):104-108. Wang, Z. Q., Y. H. Xiao, and W. F. Liu. 1991. Study on the changes in leaf protein in NongKen 58s induced by photoperiod. Hereditas (Beijing) 1991(5):1-3. Watanabe, T., S. Sakaguchi, and M. Kudo. 1968. Male-sterile rice with the cytoplasm of Lead rice from Burma. Japan. J. Breed. 18 (Suppl 2):77-78. Watanabe, Y. 1971. Establishment of cytoplasmic and genetic male-sterile lines by means of indica-japonica cross. Oryza. 8(Suppl. 2):9-16. Weerarathe, H. 1954. Hybridization technique in rice. Trop. Agr. (Colombo) 110:93-97. Wei, B. G., G. C. Sun, and J. W. Chang. 1997. Photoperiod sensitivity of a photoperiod-sensitive male sterile line in soybean. p. 114-120. In: Proc. Int. Symp. on Two-line System Heterosis Breed. in Crops, Changsha, Hunan, China, September 6-8,1997. Wei, J. c., R. L. Wang, and G. Y. Cheng. 1994. Studies on the kinetic properties of ribulose1, 5-bisphosphate carboxylase from F 1 hybrid rice. Acta Phytophysiol. Sinica. 20:55-60. Wu, B. J., M. T. Xie, Y. P. Chen, H. Jiang, and X. X. Guo. 1991. Cytologic and embryological studies on apomicts discovered from male sterility line CI001 in rice (Oryza sativa L.). Sci. China. Ser. B. 34:823-831. Wu, C. M., J. M. Cao, and X. L. Fu. 1993. Fertility alteration and flowering behaviors of photoperiod sensitive genic male-sterile rice (PGMR) in the cold regions. Chinese J. Rice Sci. 1993(2):113-116.

152

J. LI AND 1. YUAN

Wu, H. Y., and X. M. Wang. 1990. Effects of length of photoperiod on microsporogenesis in Nongken 58s. J. Huazhong Agr. Univ. 9:464-465. Wu, X. J., and H. Q. Yin. 1992. Genetics and stability of thermo-sensitive genic male sterile rice. Chinese J. Rice Sci. 1992(2):63-69. Xi, D. W., W. J. Cheng, D. 1. Yi, D. R. Yu, Z. 1. Ning, X. X. Deng, and M. Li. 1997. Genetic analysis on the TGMS line in rape of B. napus Xiangyou 91s. p. 215-220. In: Proc. Int. Symp. on Two-line System Heterosis Breed. in Crops, Changsha, Hunan, China, September 6-8, 1997. Xiao, G. Y., and 1. P. Yuan. 1997. Effects of water temperature on male sterility of the thermo-sensitive genic male sterile (TGMS) rice lines under tbe simulated low air temperature conditions appeared occasionally in high summer. Chinese J. Rice Sci. 1997(4):241-244. Xiao, J. H., S. Grandillo, S. N. Ahn, S. R. McCouch, S. D. Tanksley, J. M. Li, and 1. P. Yuan. 1996b. Genes from wild rice improve yield. Nature 384:223-224. Xiao, J. B., J. M. Li, S. Grandillo, S. N. Ahn, 1. P. Yuan, S. D. Tanksley, and S. R. McCouch. 1998. Identification oftrait-improving QTL alleles from a wild rice relative, Oryza rufipagan. Genetics 150:899-909. Xiao, J. H., J. M. Li, 1. P. Yuan, S. R. McCouch, and S. D. Tanksley. 1996a. Genetic diversity and its relationship to hybrid performance and heterosis in rice as revealed by PCRbased markers. Theor. Appl. Genet. 92:637-643. Xiao, J. H., J. M. Li, 1. P. Yuan, and S. D. Tanksley. 1995. Dominance is the major genetic basis of heterosis in rice as revealed by QTL analysis using molecular markers. Genetics 140:745-754. Xiao, Y. H., P. Chen, and W. F. Liu. 1987. A comparative analysis of free amino acids of HPGMR during its abortion of anthers. J. Wuhan Univ. (HPGMR):7-16. Xiao, Y. H., and W. F. Liu. 1981. Studies on esterase isozymes of hybrid rice and its three parental lines. Hubei Agr. Sci. 1981(11):1-4. Xiao, Y. H., G. F. Sun, and W. F. Liu. 1991. Research on prediction of rice heterosis and screening for elite rice hybrids on the basis of the activity of nitrate reductase. p. 261-267. In: Papers on basic research of rice breeding techniques. China Sci. Tech. Press, Beijing, China. Xie, F. M. 1993. Hybrid performance and genetic diversity as revealed by RFLPs in rice (Oryza sativa 1.). Ph.D. Diss. Texas A & M Uniy., College Station. Xie, H. A., J. M. Luo, J. T. Zhang, S. G. Zhang, and J. X. Zhang. 1996. Progress on the development of restorer lines and elite hybrids in rice. p. 221-227. In: China National Hybrid Rice Research and Development Center (ed.). Proc. Annual Hainan Meeting China's National Two-line Hybrid Rice Program Plan, Hainan, China. Xiong, Z. M., S. H. Cheng, and 1. Y. Cao. 1989. The status and prospects in the research and utilization of heterosis in crosses between indica and japanico in rice. Rice Abstr. 1989(5):1-4. Xiong, Z. M., S. K. Min, S. H. Cheng, and L. Y. Cao. 1990. Identification and evaluation of wide compatibility in rice cultivars T984 and Pecos. Chinese J. Rice Sci. 1990(1): 9-14. Xiong, Z. M., S. K. Min, X. D. Zhu, J. 1. Wang, and 1. Y. Cao. 1993. Discussions on genetics of wide compatibility by pedigree analysis. Chinese J. Rice Sci. 1993(2):101-104. XU, J., and 1. Wang. 1980. A preliminary study on heterosis and combining ability of rice. Yichuan (Hereditas, Beijing) 2:17-19. XU, J. F., and 1. Y. Wang. 1981. The genetic divergence and the heterosis in rice. Anhui Agr. Sci. (Special issue, Rice Quantitative Genetics). p. 65-77. Xu, M. L. and G. Q. Zhou. 1996. Studies on thermo-sensitive part of Pei'ai 64s relating to its fertility expression. Hybrid Rice. 1996(2):28-30.

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

153

Xu, R. L., X. H. Jiang, S. Y. Shi, A. M. Liu, and Q. H. Yi. 1992. Comparative studies on the constitutions of protein polypeptides between a WA type cytoplasmic male sterile rice and its maintainer. Acta Genet. Sinica 19:446-452. Xu, S. H. 1979. Cytological studies on pollen abortion of Honglian Hua-ai male sterile lines in rice. Nat. Sci. J. 1979(2):79-85. Xu, S. J., and B. H. Li. 1988. Managing hybrid rice seed production. p. 157-163. In: Hybrid rice (Proc. lnt. Symp. on Hybrid Rice, October 6-10,1986, Changsha, China). Int. Rice Res. Inst., P.O. Box 933, 1099 Manila, Philippines. Xu, Y. B., and Z. T. Shen. 1988. Receptivity of exserted stigmas. Int. Rice Res. Newsl. 1988(3):7-8. Xu, Y. B., J. J. Wang, and Z. T. Shen. 1989. Study on intersubspecific compatibility in rice. Chinese J. Rice Sci. 1989(3):113-118. Xue, G. X. and P. Chen. 1992. The effects of temperature on the fertility-transition ofPGMS rices induced by different photoperiods. Acta Agron. Sinica 1992(6):429-438. Xue, G. X., and J. Y. Deng. 1991. Studies on the Hubei photoperiodic sensitive nuclear male sterile rice Oryza-sativa-ssp-japonica: modification of minor genes on the phenotypic values of progenies of photoperiodic sensitive male sterile rice. Acta Genet Sinica. 18:59-66. Xue, G. X., and J. Z. Zhao. 1990. A preliminary study on the critical daylength evoking the photoperiodic sensitive male sterility of rice and their responses to other environmental factors. Acta Agron Sinica 1990(2):112-122. Yab, T. c., and S. C. Chang. 1976. Heterosis of rice hybrids under wetland and dryland conditions. SABRAO J. 1976(1):35-40. Yabuno, T. 1977. Genetic studies on the interspecific cytoplasm substitution lines of japonica varieties of Oryza sativa L. and O. glaberrima Steud. Euphytica 26:451-463. Yamaguchi, Y., R. Ikeda, H. Hirasawa, M. Minami, and A. Ujihara. 1997. Linkage analysis of thermosensitive genic male sterility gene, tms-2 in rice (Oryza sativa L.). Breed. Sci. 47:371-373. Yan, J. Q, and Q. Z. Xue. 1995. Genetic analysis of wide-compatibility and fertility restoration of cytoplasmic male sterility in rice (Oryza sativa L.). Acta Genet. Sinica. 22 :361-3 71. Yan, Q. C. 1987. Identification of the genuineness and the purity of hybrid rice (Oryza sativa ssp. indica) and its parental lines by electrophoretic analysis of esterases. Seed Sci. Tech. 15:645-649. Yan, S. G., Y. Song, and J. Zhang. 1991. Preliminary report on Chengdu apomictic rice. p. 101-105. In: K. D. Zhou (ed.), Study on rice apomixis. Chengdu, Sichuan Sci. Tech. Press, China. Yanagihara, S. H., H. Kata, and H. Ikehashi. 1992. A new locus for multiple alleles causing hybrid sterility between an Aus variety and Javanica varieties in rice (Oryza sativa L.). Japan. J. Breed. 42:793-801. Yang, D. c., and Y. G. Zhu. 1990. The photoreversibility ofred/far-red light for the pollen fertility of HPGMR (Nongken 58s). J. Huazhong Agr. Univ. 18:461-463. Yang, D. c., Y. G. Zhu, and L. J. Tang. 1990. The contents of four hormones in the leaves of HPGMR and the fertility conversion. J. Huazhong Agr. Univ. 9:394-399. Yang, D. H. and F. Y. Chen. 1990. A study on heredity of hybrid rice with wild abortive cytoplasm male sterility and restoring characteristics. J. Guangxi Agr. College 1990(2):61-69. Yang, F. Y. 1991a. The prediction of heterosis via homogenate complementation method in hybrid rice. p. 251-255. In: Papers on basic research of rice breeding techniques. China Sci. Tech. Press, Beijing, China. Yang, F. Y. 1991b. Rice heterosis and the ATP content in the yellowish tips ofits parental lines. p. 246-250, In: Papers on basic research of rice breeding techniques. China Sci. Tech. Press, Beijing, China.

154

J. 11 AND 1. YUAN

Yang, F. Y., Q. R. Xing, W. W. Chen, and J. P. Zhang. 1990. ATP concentration in parental seedlings in relation to heterosis of hybrid rice. Acta Bot. Sinica 32:622-625. Yang, J. D. 1990a. The pollen fertility transformation of a temperature-sensitive male sterile line, 5460TS, under natural conditions in Fuzhou. J. Fujian Agr. College 1990(3):245-251. Yang, K. H. and Z. B. Li. 1989. A preliminary study on rice wide compatibility and its utilization in hybrid rice breeding. J. Huazhong Agr. Univ. 17:302-310. Yang, R. c., K. Liang, N. Wang, and S. Chen. 1992. A recessive gene in indica rice 5460s for thermosensitive genic male-sterility. Rice Genet. Newsl. 9:56-57. Yang, R. C., and H. R. Lu. 1984. Preliminary analysis of restorer genes in the restorer line IR24. Acta Agron. Sinica 1984(2):81-86. Yang, R. c., and H. R. Lu. 1989. Study on effects of nucleo-cytoplasmic interaction in Oryza 1. Sci. Agr. Sinica 1989(2):56-63. Yang, S. R. 1959. Research on the indica/japonica crossing. Agr. J. 1959(4):3-6. Yang, S. R. 1987. New trends in breeding rice for very high yields-combination of ideal plant morphology with favorable vigor. J. Shenyang Agr. Univ. 1987(1):1-5. Yang, Z. P. 1997. Inheritance of photoperiod sensitive genic male sterility and breeding of photoperiod sensitive genic male-sterile lines in rice (OIyza sativa 1.) through anther culture. Euphytica 94:93-99. Yang, Z. P., M. H. Chen, X. M. Deng, and Y. F. Wang. 1997. Rice anther culture: a tool for breeding of widely compatible restorer lines for cytoplasmic male sterility. Acta Agr. Shanghai 1997(1):1-6. Yang, Z. Y. 1990b. Evaluation and utilization of usable and unusable heterosis in F I hybrids between indica and japonica rice subspecies. Chinese J. Rice Sci. 1990(2):49-55. Yao, J. 1., D. T. Cai, P. F. Ma, and H. Zhu. 1997. Embryological studies on apospory in rice HDAR. Chinese J. Rice Sci. 1997(2):113-117. Yao, S. Y., M. T. Henderson, and N. E. Jodon. 1958. Cryptic structural hybridity as a probable cause of sterility in intervarietal hybrids of cultivated rice, Oryza sativa 1. Cytologia 23:46-55. Yasuki, U., O. Hisatoshi, S. Hiroyuki, and O. Hiroshi. 1997. Utilization of environmentsensitive genic male sterility in hybrid rice breeding. p. 193-197. In: Proc. Int. Symp. on Two-line System Heterosis Breed. in Crops, Changsha, Hunan, China, September 6-8,1997. Ye, X. 1., Z. 1. Chen, and Y. Q. Li. 1995. The characteristics of apomictic PDER rice variety and development of diploid embryo sac. J. Trap. Subtrop. Bot. 1995(3):36-42. Yin, H. Q. 1993. Progress and recommendations in hybrid rice in northern Vietnam. Int. Rice Comm. Newsl. 42:39-41. Ying, C. S. 1994. Rice germplasm resources in China. Chinese Rice Res. Newsl. 1994(2):9-10. Yogeesha, H. S., and M. Mahadevappa. 1995. Histological and histochemical causes of male sterility in two cytoplasmic-genic male sterile lines of rice (Oryza sativa L.). Indian J. Genet. Plant Breed. 55:58-64. Yogeesha, H. S., and M. Mahadevappa. 1996. Adaptability of CMS and maintainer lines of rice (Oryza sativa 1.). Seed Res. 1996(2):98-103. Yokozeki, Y., A. Nakamura, A. Inagaki, N. Ichikawa, Y. Koshino, H. Araki, H. Shimada, T. Tsuchiya, C. Shinjyo, and T. Fujimura. 1996. DNA markers of a fertility restoration gene Rf-1 in rice, Oryza sativa 1. Breeding Science 46(suppl. 1):73. Yoshida, H., and H. Fujimaki. 1985. Development of male sterile-restoring system for hybrid rice production. 4. Standard heterosis appearing in yield components of half-sib F I hybrids with a common female parent "Akihikari". Japan. J. Breed. 36(suppl.1): 98-99.

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

155

Young, J. B., and S. S. Virmani. 1990. Heterosis in rice ov~r environments. Euphytica 51:87-93. Yu,1. Q., and Q. Lin. 1962. A study on the relationship of cutivated rice varieties in China. Acta Agron. Sinica 1:233-258. Yu, M. Y., X. Wang, 1. X. Tao, and F. D. Yao. 1991. The male sterility in rice plant induced by CRMS: III. Effect of CRMS on free proline content in anthers of rice and its possible relation to abortion of anther. Chinese J. Rice Sci. 1991(4):169-174. Yu, S. B., J. X. Li, C. G. Xu, Y. F. Tan, Y. J. Gao, X. H. Li, Q. F. Zhang and M. A. Maroo£. 1997. Importance of epistasis as the genetic basis of heterosis in an elite rice hybrid. Proc. Nat. Acad. Sci. (USA). 94:9226-9231. Yu, Z. H., S. R. McCouch, T. Kinoshita, S. Sato, and S. D. Tanksley. 1995. Association of morphological and RFLP markers in rice (Oryza sativa 1.). Genome. 38:566-574. Yuan, 1. P. 1966. A preliminary report on male sterility in rice (in Chinese). Sci. Bul. 1966(4):32-34. Yuan, L. P. 1985. A concise course in hybrid rice. Hunan Sci. Tech. Press, Changsha, China. Yuan, L. P. 1987. The strategy on hybrid rice breeding. Hybrid Rice 1987(1):1-3. Yuan, L. P. 1989. Hybrid rice US Patent-4827664 (May 9 1989). Yuan, L. P. 1990. Progress of two-line system hybrid rice breeding. Sci. Agr. Sinica 1990(3):1-6. Yuan, L. P. 1991a. Breeding for intersubspecific heterosis in rice. Program Report. China National Hybrid Rice Res. Develop. Center. Yuan, L. P. 1991b. Outlook on the development of hybrid rice breeding. p. 205-211. In: Prospects of rice farming for 2000. Zhejiang Pub!. House Sci. Tech., Hangzhou, China. Yuan, L. P. (ed.). 1992a. Current status of two-line hybrid rice research. Agr. Press, Beijing, China. Yuan, 1. P. 1992b. The technological strategy for the development of T(P)GMS lines. Hybrid Rice 1992(1):1-4. Yuan, 1. P. 1992c. Recent breakthroughs in hybrid rice research and development in China. Int. Rice Commission News!. 41:7-13. Yuan,1. P. 1994a. Purification of T(P)GMS lines of rice and their production of foundation seeds. Hybrid Rice. 1994(6):1-3. Yuan, 1. P. 1994b. Purification and foundation seed production of T(P)GMS lines in rice. National Seminar on Two-line System Hybrid Rice, Yangzhou, China. p. 1. (Abstr.) Yuan, L. P. E}94c. Purification and production of foundation seed of rice PGMS and TGMS line. Hunan Agr. Res. Newsl. 1994(2):2-3. Yuan, L. P. 1994d. Increasing yield potential in rice by exploitation of heterosis. p. 1-6. In: S. S. Virmani (ed.), Hybrid rice: New developments and future prospects. Int. Rice Res. Inst., P.O. Box 933, Manila 1099, Philippines. Yuan, L. P. 1995. Current status of hybrid rice in China and future strategies for 21st century. p. 31-33. In: M. I. Ahmed, and B. C. Virakatamath (eds.), Hybrid rice seed production technology: Theory and practice. Hyderabad, India. Yuan, 1. P. 1996. Prospects for hybrid rice breeding in China. October 29-November 1, 1996. Grupo Tecnico de Trabajo de Hibridos de Arroz Para America Latina y el Caribe (GRUTHA), Ibaqu8, Columbia. Yuan, L. P. 1997a. Exploiting crop heterosis by two-line system hybrids: Current status and future prospects. p. 1-7. In: Proc. Int. Symp. on Two-line System Heterosis Breed. in Crops, Changsha, Hunan, China, September 6-8,1997. Yuan, L. P. 1997b. Strategy for the development of super-high-yielding hybrid rice. Hybrid Rice 1997(6):1-6. Yuan, L. P. 1998a. Hybrid rice breeding for super high yield. XVIIIth Int. Congr. Genet., Beijing, China, August 10-15, 1998.

156

J. LI AND L. YUAN

Yuan,1. P. 1998b. Hybrid rice breeding in China. In: S. S. Virmani, E. A. Siddiq, and K. Muralidharan (eds.), Advances in hybrid rice technology (Proc. 3rd lnt. Symp. Hybrid Rice, Hyderabad, India, October 1996). lnt. Rice Res. lnst., P.O. Box 933,1099 Manila, Philippines. Yuan, 1. P., and H. X. Chen. 1988. Breeding and cultivation of hybrid rice. Hunan Sci. Tech. Press, Changsha, China. Yuan, 1. P., Y. C. Li, and H. D. Deng. 1990. Progress of studies in rice twin seedlings. The 4th annual meeting ofthe Rockefeller Found. lnt. Program Rice Biotechnol., May 9-12, 1990. lnt. Rice Res. lnst., P.O. Box 933,1099 Manila, Philippines. Yuan, 1. P., and S. S. Virmani. 1988. Status of hybrid rice research and development. p. 7-24. In: Hybrid rice (Proc. lnt. Symp. on Hybrid Rice, October 6-10,1986, Changsha, China). lnt. Rice Res. lnst., P.o. Box 933, Manila 1099, Philippines. Yuan, 1. P., X. J. Wu, Y. C. Yan, and X. H. Luo. 1997. A strategy for developing wide spectrum compatibility rice line. Sci. Agr. Sinica 1997(4):1-8. Yuan, S. c., Z. G. Zhang, H. H. He, H. L Zen, K. Y. Lu, J. H. Liang, and B. X. Wang. 1993. Two photoperiod-reactions in photoperiod-sensitive genic male-sterile rice. Crop Sci. 33:651-660. Yuan, S. c., Z. G. Zhang, and C. Z. Xu. 1988. Studies on the critical stage offertility change induced by light and its phase development in HPGMR. Acta Agron. Sinica 1988(1): 7-13. Zeng. S. J.. C. Y. Liang, J. K. Liang, and B. Y. Chen. 1997. Studies on the heterosis of twoline hybrid rice. 1. Performance of main agronomic characters. J. Trop. Subtrop. Bot. 1997(4):56-61. Zeng, S. X.. Z. W. Lu, and X. Q. Yang. 1979. Studies on the heterosis ofF j hybrids in rice and its relation to the parents. Acta Agron. Sinica 1979(3):23-24. Zhang, A. F., B. S. Shen, X. K. Dai, M. H. Mei, M. A. Saghai Maroof, and Z.B. Li. 1993a. An RFLP-based genetic analysis of photoperiod sensitive male-sterility in rice. Rice Genet. News!. 10:94-97. Zhang. D. P., X. N. Deng, G. X. Yu. X. H. Lin. Y. F. Xie, and Z. B. Li. 1990. Chromosomal location of the photoperiod sensitive male sterile gene in Nongken 58s. J. Huazhong Agr. Univ. 18:407-419. Zhang, G., T. S. Bharaj, Y. Lu, S. S. Virmani, and N. Huang. 1997a. Mapping of the Rf-3 nuclear fertility-restoring gene for WA cytoplasmic male sterility in rice using RAPD and RFLP markers. Theor. Appl. Genet. 94:27-33. Zhang, G., Y. Lu, and N. Huang. 1996. Molecular mapping of fertility-restoring gene Rf-3 in rice. lnt. Rice Res. Notes 1996(2-3):16-17. Zhang, G. Q., and Y. G. Lu. 1989. Genetic studies on the hybrid sterility in cultivated rice: 1. Diallel analysis of the hybrid sterility among isogenic F j sterile lines. Chinese J. Rice Sci. 1989(3):97-101. Zhang, H. L., and Y. D. Deng. 1990. Progress in the development of three-line system intersubspecifc hybrid rice between indica and japonica. Hybrid Rice 1990(2):29-31. Zhang, J. H., Z. Tong, Z. K. Chen. and F. Zhou. 1994a. Ultrastructural studies on microsporogenesis in the photoperiod-sensitive genic male-sterile rice. Acta Bot. Sinica. 6:129-133. Zhang, J. Z., 1. J. Fang, and Y. W. Yuan. 1987. Study on the application of genetic distance to the selection of Honglian type restorers in hybrid rice. Guangdong Agr. Sci. 1987(1): 1-3. Zhang, N. G., and X. Zhou. 1992. Relation between endogenous acidic phytohormones (IAA. GA1+4. ABA) and fertile transformation of photoperiod sensitive genic malesterile rice (PGMR)-Nongken 58s. J. Nanjing Agr. Univ. 1992(3):7-12.

2. HYBRID RICE: GENETICS, BREEDING, AND SEED PRODUCTION

157

Zhang, P. J., S. Ming, C. w. Xu, L. S. Yang, Y. S. Bai, C. Q. Sun, and X. K. Wang. 1997b. Heterosis and combining ability of hybrid rice and its relation to japonica-indica index of parents. Rice Genet. Newsl. 14:34-36. Zhang, Q. F., B. Z. Shen, X. K. Dai, M. H. MeL M. A. S. Maroof, and Z. B. Li. 1994b. Using bulked extremes and recessive class to map genes for photoperiod-sensitive genic male sterility in rice. Proc. Nat. Acad. Sci. (USA). 91:8675-8679. Zhang, Q. F., K. D. Liu, G. P. Yang, M. A. A. Shaghai, C. G. Xu, and Z. Q. Zhou. 1997c. Molecular marker diversity and hybrid sterility in indica-japonica rice crosses. Theor. Appl. Genet. 95:112-118. Zhang, S. Q., S. H. Cheng, and L. Y. Cao. 1988. The compatibility of F 1 hybrid between indica and japonica. Chinese J. Rice Sci. 1988(2):94-96. Zhang, S. Q., X. B. Xie, and H. F. Chen. 1992. A preliminary research and progress in the development of three-line system intersubspecific crosses. Presentation Annual Meeting of Zhejiang "8812" Research Plan, Hangzhou, China. Zhang, T. B. 1985. Male sterility and cytoplasmic regulation of gene action in rice: A brief account of studies. SABRAO J. 1985(1):87-90. Zhang, X., and B. Wang. 1990. A newly discovered double stranded RNA associated with cytoplasmic male-sterile rice. Acta Genet. Sinica 17:289-293. Zhang, X. G., and Y. G. Zhu. 1991. The inheritance of sterility in Hubei photoperiod sensitive nuclear male-sterile rice. Heriditas (Beijing) 1991(3):1-3,37. Zhang, Z. G., X. G. Lu, and L. P. Yuan. 1992. Reflections on identification of light and temperature properties of fertility transformation of PGMR. Hybrid Rice 1992(6): 29-32. Zhang, Z. G., S. C. Yuan, and C. Z. Xu. 1987. The influence of photoperiod on the fertility changes of Hubei photoperiod-sensitive genic male-sterile rice (HPGMR). Chinese J. Rice Sci. 1987(3):137-144. Zhang, Z. G., H. L. Zeng, Y. Z. Li, S. C. Yuan, and D. P. Zhang. 1992. Influence of photoand thermo-conditions during vegetative growth of PGMS on its fertility alteration. Hybrid Rice 1992(5):34-36. Zhang, Z. G., H. L. Zeng, Y. Z. Li. S. C. Yuan, and D. P. Zhang. 1993h. Influence on the fertility alteration of photo- and thermoconditions during vegetative growth phase in PGMR J. Huazhong Agr. Univ. 1993(1):84-86. Zhang, Z. G., H. L. Zeng, J. Yang, S. C. Yuan, and D. P. Zhang. 1994c. Conditions inducing fertility alteration and ecological adaptation of photoperiod-sensitive genic malesterile rice. Field Crops Res. 38:111-120. Zhangxing, Z., and H. Chunnong. 1980. A study of male sterility induced rice (Dryza sativa L.) by 3-(p-chlorophenyl) 6-methoxy-s-triazine-2,4-(1H,3H) Dione. J. Hangzhou Univ. (Zhejiang, China) 1980(4):81-87. Zhao, B. R, H. D. Deng, and Y. Q. Li. 1992. Preliminary report on the embryological study of apomixis in rice twin seedlings. Wuhan Bot. Res. 1992(3):213-218. Zhao, S. X. 1977. How to fix heterosis'? Genetics & Breed. 1977(:~):21-22. Zheng, K. L., B. Shen, H. R Qian, and J. L. Wang. 1992. Tagging genes for wide compatibility in rice via linkage to RFLP markers. Chinese J. Rice Sci. 1992(4):145-150. Zheng, K. L., B. Shen, F. Yu, C. Z. Zhao, X. F. Qi, and X. M. Xu. 1990. Restriction fragment length polymorphism in rice. Chinese J. Rice Sci. 1990(4):145-149. Zhong, W. G., C. G. Li, Z. D. Chen, and A. N. Yang. 1997. Efficacy of chemical gametocide 13 in emasculation ofrice. Jiangsu J. Agr. Sci. 1997(3):191-192. Zhong, W. G., A. N. Yang, and J. S. Zou. 1992. A preliminary report on the effect of temperature conditions on the chemical emasculation by Emasculator No.2. Jiangsu Agr. Sci. 1992(4):4-5.

158

J. LI AND L. YUAN

Zhou, C. S., J. B. Liu, K. Y. Wu, and X. N. Li. 1993a. Studies of the cold water irrigation technique for multiplication of low critical temperature TGMS rice. Hybrid Rice 1993(2):15-16. Zhou, G. F., C. M. Guo, C. Q. Yin, and X. B. Sheng. 1991a. Preliminary observations on the fertility alteration and genetic behavior of the indica dual-purpose TGMS line, Heng-Nong s-1. Hunan Agr. Sci. 1991(4):10-12. Zhou, H., J. G. Glaszmann, K. S. Cheng, and X. Q. Shi. 1988. A comparison of classification of cultivated rice. Chinese J. Rice Sci. 1988(1):1-7. Zhou, K. D., H. W. Li, and Y. Cheng. 1994. Realization of fine grain quality certainly being the way to develop hybrid rice. Hybrid Rice 1994(3-4):42-45. Zhou, K. D., X. D. Wang, M. Luo, K. M. Gao, S. L. Zhou, Z. B. Yan, P. Li, and F. Chen. 1991b. Advance in study on Sichuan apomixis rice. J. Sichuan Univ. 28:382-387. Zhou, K. D., X. D. Wang, M. Luo, K. M. Gao, S.L. Zhou, Z. B. Yan, P. Li, F. Chen, and G. M. Zhou. 1993b. Initial study on Sichuan apomixis rice (SAR-1). Sci. China. Ser. B. 36:420-429. Zhou, K. L. 1994. Breeding of CMS lines in indica hybrid rice. Hybrid Rice 1994(3-4): 22-26. Zhou, K. L., and F. M. Liao. 1995. Xiangyou 63, a quasi-aromatic hybrid rice with good quality and high yield. lnt. Rice Res. Notes 1995(4):9-10. Zhou, K. L., and F. M. Liao. 1997. Xingxiangyou 77, a high-yielding and fine-quality semi-aromatic hybrid rice. Int. Rice Res. Notes 1997(3):22-23. Zhou, M. L., Q. Y. Tang, and J. M. He. 1997. Effects of photoperiod on fertility of ecological male sterile (EMS) wheat. p. 107-113. In: Proc. Int. Symp. on Two-line System Heterosis Breed. in Crops, Changsha, Hunan, China, September 6-8,1997. Zhou, S. L. 1978. Rice science. Agr. Press, Beijing, China. Zhou, T. L., J. H. Shen, and F. C. Yeo 1983. A genetic analysis on the fertility of hsien type hybrid rice with wild rice cytoplasm. Acta Agron. Sinica 1983(4):241-247. Zhu, L., C. Lu, P. Li, L. Shen, Y. Xu, P. He, and Y. Chen. 1996. Using double haploid populations of rice for quantitative trait locus mapping. p. 745-748. In: G. S. Khush (ed.), Rice genetics III. Int. Rice Res. Inst., P.O. Box 933, 1099 Manila, Philippines. Zhu, Q. S., Z. J. Zhang, J. C. Yang, X. Z. Cao, Y. Z. Lang, and Z. C. Wang. 1997. Sourcesink characteristics related to the yield in intersubspecific hybrid rice. Sci. Agr. Sinica. 1997(3):52-59. Zhu, Y. C., and F. M. Liao. 1990. Research progress on heterosis utilization in two-line system intersubspecific crosses. Hybrid Rice 1990(3):32-34. Zhu, Y. G. 1979. Studies on male sterile lines of rice with different cytoplasms. Acta Agron. Sinica 1979(4):29-38. Zhu, Y. G., and D. C. Yang. 1992. Study and application on photoperiod-sensitive genic male-sterile rice. Wuhan Univ. Press, Wuhan, China. Zhu, Y. G., and J. H. Yu. 1987. A study on HPGMR breeding in indica. J. Wuhan Univ. (HPGMR):139-144. Zhu, Y. G., and J. H. Yu. 1989. Stability and inheritance of photoperiod-sensitive genic male-sterility in rice. p. 745-748. In: S. Iyama and G. Takeda (eds.), 6th Int. Congr. SABRAO. Tsukuba-machi, Japan. Zhu, Y. G., and W. G. Zhang. 1987. Studies on isozymes with heterosis in the seedlings of hybrid rice. Acta Agron. Sinica 1987(2):89-96. Zirkle, C. 1952. Chapter 1. Early ideas on inbreeding and cross breeding. p. 1-13. In: J. W. Gowen (ed.), Heterosis. Iowa State College Press, Ames.

3 Rose: Genetics and Breeding Serge Gudin * Universite d'Aix-Marseille III Service 442 Avenue Escadrille Normandie-Niemen 13397 Marseille. Cedex 20 France

I. Introduction II. Systematics A. Taxonomy of Rosa B. Horticultural Classification III. Cytogenetics and Genetics of Rosa A. Genomic Analysis B. Conventional Genetic Analysis IV. Breeding Objectives A. Cut Flowers B. Garden Roses V. Breeding Criteria and Selection Procedures A. Selection for Disease and Pest Resistance B. Prediction of Flower Productivity C. Prediction of Postharvest Longevity D. Predicting Rootstock/Cultivar Compatibility E. Adaptation VI. Breeding Technology A. Sexual Reproduction 1. Pollen and Pollination 2. Seed Maturation and Germination B. Interspecific Hybridization 1. Ploidy Level Manipulation 2. Embryo Rescue 3. Protoplast Fusion *1 thank Mr. A. Meilland, Prof. N. Zieslin, Mr. J. Mouchotte, Mr. J. Meynet, Dr. S. ReyndersAloisi, Dr. D. Zhang, Dr. J. Janick, and Ms. A. Coulon for their assistance and suggestions.

Plant Breeding Reviews, Volume 17, Edited by Jules Janick ISBN 0-471-33373-5 © 2000 John Wiley & Sons, Inc. 159

S. GUDIN

160 C. Mutation Breeding 1. Radiation and Chemical Mutagens 2. Somaclonal Variation

D. Biotechnology 1. Molecular Markers 2. Genetic Transformation VII. Conclusion Literature Cited

I. INTRODUCTION

Roses have been grown and admired since the origins of civilization (Wylie 1955a). There is evidence that roses were cultivated 5,000 years ago by the ancient civilizations of China, western Asia and northern Africa (Shepherd 1954). The oldest known representation of a rose was discovered, as part of a fresco, by the archaeologist A. Evans in Knossos, Crete (Testu 1984), which dates from the sixteenth century B.C. (Shepherd 1954) and was identified in 1926 by C. C. Hurst as R. x richardii Rehd 1 (Testu 1984). Before they became the most popular garden plant, roses were admired for their petals, and as a source of perfume and edible hips. Moreover, in antiquity, they decorated the tombs of Greece and China, symbolized secrecy in Rome, virtue in the far East, and silence in Egypt (Rowley 1966). Considerable information about roses in antiquity can be found in the writings of the Greek historian Herodotus (490-420 B.C.), the Greek philosopher Theophrastus (372-287 B.C.) and the Roman naturalist Pliny (23-79). Pliny records that the Romans already grew roses in glasshouses that could be heated by hot water in wintertime, in Praeneste, Leporia, and Paestum, and that 32 remedies were based on roses and were derived from flowers, hips, and scent. Maia and Venard (1976) conclude that the roses that were grown in Europe until the thirteenth century were strictly summer roses, such as R. moschata, R. gallica, R. alba, and R. damascena var. bifera Hort. non Regel, which has sporadic recurrent flowering and was known in southern Europe since the fourteenth century. There is some evidence that R. chinensis was grown in Italy from the beginning of the sixteenth century. This species and R. x odorata (Andr.) Sweet, which brought the recurrent flowering character to modern cultivars, were introduced in England at the end of the eighteenth century and beginning of the nine-

lAuthorities for binomial of Rosa spp. not indicated in the text may be found in Tables 3.2 and 3.4.

161

3. ROSE: GENETICS AND BREEDING

teenth century (Maia and Venard 1976; Testu 1984). The huge popularity of specialized rose gardens and rose breeding is a feature of the nineteenth and twentieth centuries. Rosa is the major economically important genus of ornamental horticulture. Indeed, the area of cut-rose production worldwide is expanding, with a remarkable progression within developing countries (Table 3.1). In 1994, the area devoted to rose production was about 5,362 ha for the 18 leading producing countries. The production of cut roses is carried out with a relatively low cost of greenhouse investment ($20-50/m2 ) in developing tropical countries, such as Colombia, Ecuador, Kenya, or Zimbabwe or with a very high cost ($200-300) in developed Northern Hemisphere countries, such as the Netherlands, Scandinavia, or North America (Zieslin 1996). There is also a very substantial market for garden rose bushes in developed countries. Approximately 20 million plants are sold annually in the United Kingdom alone (Roberts and Lewis 1996). Millions of miniature pot roses are now sold every year (Borch et al. 1996), mainly produced in northern European countries and North America. Although no miniature roses were listed in the American Rose Section Selection Handbook in 1959, 13 were listed in 1969 (Abler 1990). Roses are also used in the perfume industry and for their medicinal and culinary qualities (Shepherd 1954). During World War II,

Table 3.1. Cut rose greenhouse production areas for 1985 and 1994 from 18 countries (Pertwee 1995, Meilland Star Rose, unpublished). Continent (country) Africa (Kenya, Zimbabwe, Zambia, Tanzania, Malawi)

Production area (ha) 1985

1994

13

238

Asia (Israel, Japan)

492

658

Central America (Mexico)

:10

60

Europe (Netherlands, Italy, France, Spain, Germany, Belgium, United Kingdom)

2439

3149

North America (USA)

380

357

South America (Colombia, Ecuador)

500

900

162

S. GUDIN

500-700 t of rose hips were gathered annually, as the British Ministry

of Health recognized them as a major source of vitamin C. Rose hip marmalades or infusions are well known in Europe and North America, as are rose petal sweets and pastries in Asia. The literature on rose genetics is scarce and fragmented. In spite of the economic importance of roses relative to other ornamentals, they are of minor importance when compared to major food crops and consequently less well studied. Furthermore, rose breeding is mainly carried out by highly competitive private companies and their applied genetic knowledge is proprietary and unpublished (De Vries and Dubois 1996; Gudin 1995; Gudin and Mouchotte 1996). This review summarizes the actual available knowledge about rose systematics, genetics, and breeding and proposes some perspectives especially focused on breeding. II. SYSTEMATICS A. Taxonomy of Rosa Rose species are found throughout the colder and temperate regions of the Northern Hemisphere, from the Arctic Circle to the subtropics (Marshall 1973; Maia and Venard 1976; Zieslin and Moe 1985), including USA (New Mexico), Iraq (Marshall 1973), Ethiopia, Bengal, and southern China (Zieslin and Moe 1985). Two millenia before Linnaeus, Theophrastus used a binomial denomination for roses, calling those with double corollas "rhoda hekatontaphylla" (roses with one hundred leaves), not to be confused with R. ceniifolia. Linnaeus stated in 1753 that "the species of Rosa are very difficult to determine and those who have seen few species can distinguish them more easily than those who have examined many" (Shepherd 1954). Indeed, the number of recognized species varies according to the classification systems and this genus represents a problem for the taxonomists because of its important polymorphism and the detailed manner in which its characters are described (Maia and Venard 1976). Because intensive hybridization has been practiced during the last two centuries, it is almost impossible to distinguish between pure species, hybrids, garden forms with latinized names, and cultivars, many of which are synonyms (Zieslin and Moe 1985). Cytology studies should contribute substantially to rose taxonomy, but rose chromosomes are small and difficult to observe through classical procedures (Rowley 1961; Roberts and Short 1979; Lata 1982; Ma et al. 1996). Rose cytological studies have a long history (Berninger 1992). The first investigations concerned the meiotic behavior in the section Caninae

3. ROSE: GENETICS AND BREEDING

163

(Tackholm 1922). Hurst (1925) was the first to bring valuable information about chromosome counts and morphology in the genus Rosa and provided evidence of the relation between genomes (haploid sets of 7 chromosomes or "septets") and morphological traits. Thus, a tentative classification of rose species and botanical hybrids was proposed and five chromosomes sets (genomes) were proposed (Hurst 1927). However, Hurst's classification system was based on the assumption that the species belonging to a single group were interfertile, which was not verified by further hybridization studies, mainly those of Lewis and Basye (1961). They and others (Erlanson 1929, 1930, 1933, 1934; Ratsek et al. 1939, 1940; Khara 1944) demonstrated high sterility levels between species within Hurst's groups. Hurst's classification was questioned by other investigators (Raby 1937; Gustafsson 1944; Wylie 1954; Lewis 1962; Shahare and Shastry 1963; Salafia 1965; Klasterska 1969; Klasterska and Klastersky 1974; Roberts 1977; Meenakshi 1977; Koncalova and Klastersky 1978). Although no exhaustive key for identification is yet available (Jacob et al. 1996), according to cytotaxonomists such as Darlington and Wylie (1955) or Maia and Venard (1976), the most satisfactory classification available is that of Rehder (1940), itself adapted from Crepin (1889), who based his analysis on purely morphological traits. Rehder's classification has been supported by other types of systematic approaches, such as the one of Ueda and Tomita (1989), using the morphometric analysis of pollen exine patterns. According to Rehder (1940), the genus Rosa is represented by approximately 120 species and is divided into four subgenera, among which three (HuIthermia, Platyrhodon and Hesperhodos) only include one or two species. Subgenus Eurosa is represented by 115 species divided into 10 sections, according to the insertion mode of the stipules, the style lengths, the inflorescence types, the numbers of leaflets, and thorn shapes. The 10 sections of Eurosa, along with their number of species, chromosome count, geographical distribution, and main species, are presented in Table 3.2. The studies of Ratsek et al. (1939) and Lewis and Basye (1961) on pollen viability of inter- and intrasectional hybrid indicate that the sections Cinnamomeae and Carolinae should be considered as one (Darlingtoin and Wylie 1955; Maia and Venard 1976; Zieslin and Moe 1985). Hurst (1941) and Wylie (1954) published important bibliographical and cytological works that investigate the origins of modern garden roses and indicated that all known cultivars at that time originated from only 10 species: R. canina, R. chinensis, R. foetida, R. gaIIica, R. gigantea, R. moschata, R. multiflora, R. phoenicea Boiss., R. rugosa, R. wichuraina.

164

S. GUDIN

Table 3.2. Classification of sub-genus Eurosa adapted from Rehder (1940); Darlington and Wylie (1955); Maia and Venard (1976).

Section BANKSIAE

BRACTEATAE CANINAE

Number of species

2n chromosome number

2

14

Eastern Asia

R. banksiae Ail., R. cymosa Tratt.

14

2 2~~

Geographical distribution

Main species

Asia

R. bracteata Wendl.

28-42

Europe, eastern Asia. North Africa

R. canina L.

North America

R. carolina 1.,

CAROLINAE

2

28

CHINENSES

2

14

R. foliosa Nutt. Eastern Asia

(Indicae) CINNAMOMEAE

46

14-56

North America, Asia

R. chinensis Jacq, R. gigantea Colett ex Crep. R. rugosa Thunb., R. nuktana Pall.,

R. acicuJaris Lindl. 4

28

Ethiopia, Europe, western Asia

R. gaWca 1.. R. damascena Mill., R. centifolia L.

14

Eastern Asia

PfMPINELLffOLIAE

10

14-28

Asia, southern Europe

SYNSTYLAE

23

14

Western Asia

R. laevigata Michx R. sericea Lindl., R. foetida Herm.. R. xanthina LindL, R. hugonis Hemsl. R. moschata Herrm., R. wichuraiana Crep., R. sempervirens 1., R. multiflora Thunb. ex MUff.

GALLICAE

LAEVIGATAE

Most of these species are diploids (2n = 14). R. gallica and R. foetida are tetraploids, and R. canina is pentaploid (Darlington and Wylie 1955). These species originate from three main geographical regions: eastern Asia, western Asia, and Europe (mainly the Mediterranean region). Three more species are often reported as ancestors of modern cultivars but, in fact, they are thought to be hybrids. They include R. damascena, which could have arisen from R. gallica x R. moschata (for the "summer" Damask rose group) or from R. gallica x R. phoenicea (for the "autumn"

3. ROSE: GENETICS AND BREEDING

165

Damask rose group, also called R. bifera Pers.); R. alba L., which could have arisen from R. canina x R. gallica; and R. centifolia, which could have arisen from (R. ruba Blackw. x R. moschata) x R. alba (Maia and Venard 1976). Thus, three Eurosa sections do not appear in the recognized origins of modern cultivars: Banksiae, Laevigatae, and Bracteatae. In fact, some old varieties belonging to these sections are known; some have been used in interspecific crosses, such as R. banksiae x R. laevigata that gave rise to 'Silver Moon', R. bracteata x R.laevigata that gave rise to R. x leonida Moldenke, and R. bracteata x (R. gigantea x R. gallica) that gave rise to 'Mermaid' (Testu 1984). However, species of these sections have indeed not been used for breeding for many decades. This is also true for the three sub-genera, Platyrhodon, Hulthemia, and Hesperhodos, although some crosses between sub-genera were made during the nineteenth century, such as R. roxburghii x R. odorata (R. roxburghii being the only species of Platyrhodon) and R. clinophylla Wend!. x Hulthemia persica (Michx.) Bornm. (Testu 1984). Pioneering work being done in North American university programs (Buck 1960; Marshall 1973) has led to the incorporation of unexploited species in breeding, such as R. spinosissima L. (section Pimpinellifoliae) , or R. fedtschenkoana Regel, R. blanda Ait., R. nitida Willd., R. woodsii Lindl., R. arkansana Porter, R. acicularis Lindl. (section Cinnamomeae). The cross R. banksiae x R. laevigata has been reinvestigated (Basye 1990). Studies are under way that should soon extend and clarify Rosa botanical classification. Thus, new descriptive attempts are being made using methods such as flow cytometry for ploidy level determination (Demilly et al. 1994; Jacob et al. 1996), randomly amplified polymorphic DNA (RAPD) analysis of DNA samples (Reynders-Aloisi and Bollereau 1996; Moreno et al. 1996), statistical analysis of floral phenolic compounds (Jay et al. 1996; Raymond et al. 1994; Raymond et al. 1995), characterization of volatile compositions (Antonelli et al. 1997), computerized canonical discriminant analysis and cluster analysis of phenotypic data (Teyssier et al. 1996; Grossi et al. 1999). Furthermore, present efforts of the scientific community in the harmonization of descriptions should also contribute to this effort (Gandelin and Mistou, 1994; Mistou et al. 1996).

B. Horticultural Classification

Because botanical classification is difficult and because the origins of old hybrids are poorly known, a completely satisfactory horticultural classification is difficult to achieve. The American Rose Society (ARS) was formally designated in 1955 as the International Registration Authority

166

S. GUDIN

for Roses by the International Society for Horticultural Science. Later, in 1981, the ARS was designated to serve as the registrar for the Communaute Internationale des Obtenteurs de Plantes Ornementales et Fruitieres a Reproduction Asexuee (CIOPORA) (Haring 1986). The ARSapproved horticultural classification is presented in Table 3.3. This classification is mainly based on the grouping of botanical characters, but not exclusively as demonstrated by the use of some denominations such as "Miscellaneous Old Garden Roses," "Species," "Miniature," "Shrub." In fact, there are almost as many horticultural classifications as there are authors. Maia and Venard (1976) have, for instance, proposed a short classification of cultivated roses mainly based on historical chronology (Table 3.4). Differences appear in the two presented classifications. Thus, Noisette, Bourbon, Hybrid China and Portland appear in the "main modern roses" section in the classification of Maia and Venard whereas they appear in the "old garden roses" one in ARSapproved classification. Furthermore, although most ofthe plant names are the same in both classifications, some new ones appear in the Maia and Venard classification, such as "Recurrent Flowering Hybrids" or "Pernetiana. "

Table 3.3. Horticultural classification of roses, as approved by the ARS, and adapted from Haring (1986). Main classes

Associated groups

OLD GARDEN ROSES

Alba, Ayrshire, Bourbon, Boursault, Centifolia, China, Climbing Bourbon, Climbing China, Climbing Hybrid Perpetual, Climbing Moss, Climbing Tea, Damask, Eglanteria, Gallica, Hybrid Alba, Hybrid Bracteata, Hybrid Bourbon, Hybrid Canina, Hybrid China, Hybrid Foetida, Hybrid Multiflora, Hybrid Perpetual, Hybrid Sempervirens, Hybrid Setigera, Hybrid Spinosissima, Moss, Miscellaneous Old Garden Roses, Noisette, Portland, Species, Tea

MINIATURE ROSES

Climbing Miniature, Miniature

SHRUBS

Hybrid Blanda, Hybrid Hugonis, Hybrid Laevigata, Hybrid Macounii, Hybrid Macrantha, Hybrid Moyesii, Hybrid Musk, Hybrid Nitida, Hybrid Nutkana, Hybrid Rugosa, Hybrid Suffulta, Kordesii, Shrub

HYBRID TEAS, GRANDIFLORA,

Climbing Floribunda, Climbing Grandiflora, Climbing Hybrid Tea, Climbing Polyantha, Floribunda, Grandiflora, Hybrid Tea, Large-flowered Climber, Polyantha, Rambler

FLORIBUNDAS, POLYANTHAS, RAMBLERS, CLIMBERS

167

3. ROSE: GENETICS AND BREEDING

Table 3.4. Horticultural classification of cultivated roses, adapted from Maia and Venard (1976). Main classes

Associated groups and species

OLD GARDEN ROSES

Musk, Gallica, Damask, Centifolia, Alba

CHINA ROSES

R. semperflorens Curtis, R. indica odorata Andr.= R. indica fragrans Thory, R. indica sulphurea Thory

MAIN MODERN ROSES

Noisette, Bourbon, Hybrid China, Portland, Recurrent Flowering Hybrid, Hybrid Tea, Pernetiana, Polyantha, Hybrid Polyantha

Some group denominations which have not been introduced above are defined below. 1. Noisette. This group results from the historic cross made by J. Champ-

ney in 1802 between R. moschata and 'Old Blush', itself probably resulting from R. chinensis f. spontanea Rehd. & Wils. x R. gigantea (Testu 1984). In the F 2 of this cross, resulting from selfing, P. Noisette selected the first plant in 1814; it gave its name to the group, although it was later found that this rose type had little variation and was often sterile. It has a compact habit with inflorescences grouped in clusters, constant double corollas, and recurrent flowering through the summer. The introduction of the yellow color by hybridization with R. indica sulphurea led to the yellow flowered Tea and Noisette roses (Maia and Venard 1976). In some other classifications, it appears as a "Tea-Noisette" group but corresponds, in fact, to a sub-group of the Noisettes, characterized by lack of winterhardiness (Testu 1984). 2. Bourbon. This group originates from natural hybridizations between the autumn Damask rose and 'Old Blush' that occurred in the 1800s on Bourbon island (now called La Reunion). The first Bourbon roses were triploid. Later on, tetraploids appeared, probably resulting from unreduced gamete formation (Maia and Venard 1976). Bourbons are characterized by incurved silky petals, semi-double or double corollas, cup- or flat-shaped and are not all remontant or summer and fall blooming (Testu 1984). 3. Hybrid China. This group results from crosses, first made in 1815, between R. gallica and R. indica odorata (a Chinese clone introduced by A. Hume in 1809 and supposedly resulting from the cross R. chinensis x R. gigantea). The first cultivars were triploid and sterile but,

168

S. GUDIN

as in Bourbons, tetraploids appeared later. China Hybrids are sensitive to cold, and are summer flowering (Maia and Venard 1976). 4. Portland. This group originates from crosses between R. galliea, the autumn Damask rose and, later on, R. semperflorens (a Chinese rose introduced by G. Slater in 1792 and close to R. ehinensis, according to C. C. Hurst) (Maia and Venard 1976). The Portlands are tetraploids (Maia and Venard 1976) and resemble the Bourbons, but they have shorter peduncles. They all are semi-double or double and fragrant. Most of them lack vigor and are very thorny (Testu 1984). 5. Remontant Hybrids. These hybrids resulted from crosses first made in 1843 between China Hybrids and Portlands. Later, crosses with Bourbons also led to cultivars assigned to this group. In fact, in spite of its name, this group is characterized by a low level of remontance, i.e. autumn blooming is weak. However, hardy and vigorous hybrids are tetraploid (Maia and Venard 1976).

6. Hybrid Tea. This important group originated from the triploid 'La France', obtained by L. Guillot in 1867. Although the pedigree of 'La

France' is uncertain, it combines characters of Tea roses originating from R. gigantea (and so called because oftheir tea fragrance) and R. galliea. As in Bourbons and China Hybrids, tetraploids later appeared and all Hybrid Teas are now tetraploid (Maia and Venard 1976). Various types of Hybrid Teas are now referred to in the cut-flower trade as Intermediate (medium bud size and stem length, 45-55 em), Sweetheart (small buds and short stems, <45 em), and Sprays (inflorescences, mostly corymb-like clusters that originate from Polyantha cultivars, sometimes crossed with small-size Hybrid Teas). Hybrid Tea in this trade classification is devoted to cultivars with large flower buds and long stems (>55 em). 7. Pernetiana. This group originates from crosses made by A. PernetDucher in 1900 between Hybrid Teas and R. foetida cv. Persian Yellow, from which the yellow color was introduced. 8. Polyantha. This group developed at the same time as the Hybrid Teas. It originates from natural hybridizations between R. multiflora (described under the name of R. polyantha Sieb. & Zucc.) and a dwarf mutant of R. ehinensis known as R.lawraneeana Sweet. These cultivars are diploid and remontant (Maia and Venard 1976) and characterized by inflorescences organized in clusters.

3. ROSE: GENETICS AND BREEDING

169

9. Hybrid Polyantha. This group results from the Polyanthas crossed with Hybrid Teas. The first cultivars were triploid, but tetraploids appeared later. This is a complex group where genes from species, such as R. rugosa, R. moschata, and R. wichuraiana, were introgressed (Maia and Venard 1976). Some refer to this group as "Tea-Polyantha" (Testu 1984).

10. Other Groups. Other names appearing in the ARS classification correspond to horticultural utilizations and/or morphological groups, because their genetical origin is either wide or quite uncertain. Miniature Roses. These correspond to a group of dwarf cultivars most commonly grown as a pot plant. Their botanical origin is R. chinensis var. minima, resulting from R. chinensis x R. gigantea crosses (Testu 1984).

Shrubs. These correspond to a group of mostly undefined hardy and vigorous bushy cultivars that are used mainly for landscaping. Grandifloras. These correspond to Hybrid Teas with large flower buds. The name is commonly used in the U.S.A. Floribundas. These correspond to a sub-group of Hybrid Polyanthas characterized by clusters of relatively large flowers. The name was first applied by J. H. Nicolas, a research manager of Jackson and Perkins Company (Shepherd 1954). Ramblers and Climbers. These apply to cultivars with corresponding growing habits. Understocks. These constitute a classification group for rootstocks. Leemans (1966) provides a comprehensive list of species and cultivars used as understocks. Perfume Roses. A classification of roses for extraction of perfumery products is also available (Padhye 1982). Primarily four species are used for extraction of rose attar (or oil): R. damascena, R. ceniifolia, R. alba, and R. gallica (Singh and Malik 1982).

III. CYTOGENETICS AND GENETICS OF ROSA

A. Genomic Analysis

Hurst (1925) showed that five genomes-A, B, C, D, E-could be distinguished in different rose species, according to some determining morphological characters, such as type of inflorescence or range of style length. These species could be diploids, auto- or allotetraploids and, in the section Caninae, pentaploids (AABDE). In this latter case, only the

170

S. GUDIN

ABDE female gametes and A male gametes are functional, which leads to a predominantly maternal type inheritance (Berninger 1992). In the pollen mother cell of pentaploid species of the section Caninae (5n = 35), the 7 bivalents and the 21 univalents are arranged on the equatorial plate, the bivalents separate first and migrate toward the poles, followed later by random distribution of the univalents. During second division, chromosomes derived from the bivalents divide and migrate to the poles. The univalents, unable to divide, do not migrate to the poles. As a consequence, nuclei of various sizes are formed and contain 7 chromosomes of bivalent origin, with 0-21 univalents. Only gametes with 7 chromosomes are functional. During female gametogenesis, the unpaired chromosomes are not arranged on the equator but migrate toward the micropylar pole, where they are joined by the 7 bivalents resulting from first division. There are, therefore, 28 chromosomes at one pole and 7 at the other. The second regular division results in two large nuclei and two small ones; the 28 chromosome nucleus differentiates in the embryo sac. Roberts (1975), by bringing evidence of meiosis similarities between R. nanothamnus Bouleng. (characterized by 7 bivalents and 14 univalents at pachytene) and members of the section Caninae, demonstrated that this species could be placed in this section. In addition, apomixis has also been described (Crane and Lawrence 1956), as in the closely related genus Rubus (Haskell 1960), but is probably of rare occurrence (Maia and Venard 1976), although it has recently been described in many pentaploid numbers of the section Caninae (Wisseman and Hellwig 1997). F1 hybrids between section Caninae species were characterized by apomictic reproduction, but it disappeared following crosses with other section species. Gustafsson (1944) therefore concluded that the apomixis character was recessive. Lewis and Basye (1961) studied the meiotic behavior of F 1 hybrids resulting from crosses between species belonging to different sections of Rehder's classification (1940). Some of these hybrids were highly sterile despite the regularity of meiosis. Sterility was explained by the formation of inviable gametes resulting from crossing-over with "incompletely" homologous chromosomes during prophase I of meiosis. Maia and Venard (1976) examined the meiotic behavior of many diploid, triploid, and tetraploid species, using phase contrast microscopy. While all the diploid species they studied were found to have regular meiosis, the meiosis was irregular in many hybrids. At the tetraploid level, meiosis was regular in R. pisocarpa Gray (section Cinnamomae) but irregular in its hybrids. In triploids, 0 to 6 trivalents were formed at metaphase 1. The univalents migrated randomly toward both poles and divided at anaphase 1. High sterility was

3. ROSE: GENETICS AND BREEDING

171

reported, and many cellular fusions occurred during different stages of meiosis. This phenomenon of "cytomixy" had previously been described in Prunus (Salesses 1970). This tendency of meiotic irregularity leads to aneuploidy (Rowley 1961). Furthermore, B chromosomes, although rare, also occur in roses (Jones 1975), and one or two B chromosome fragments have been observed (Lata 1982), resulting in high pollen sterility. B. Conventional Genetic Analysis Technical factors make the rose a difficult model system for genetic analysis. Among them are high heterozygosity (see Rowley 1966; Berninger 1992; Gudin and Mouchotte 1996), ploidy levels (Berninger 1992; Jacob et al. 1996), plus well-known difficulties in sexual reproduction, from pollination to seed germination (Buck 1960; Gudin 1995; Gudin and Mouchotte 1996). Thus, hybridizations between modern cut rose or garden cultivars sometimes result in hip set as low as 25% (Gudin and Arene 1991), only 4 achenes per hip (Gudin et a1. 1991a), and 18% seed germination (Gudin et a1. 1990). A number of characters in roses have been shown to be under simple genetic control. Male sterility, which is frequently observed in R. setigera Michx., is inherited as a dominant (Lewis and Basye 1961), as are the moss character (De Vries and Dubois 1984), dwarfness (Dubois and De Vries 1987), and resistance to blackspot disease (Von Malek and Debener 1998). Recurrent flowering is controlled by a single recessive (Semeniuk 1971; Svejda 1977; De Vries and Dubois 1978). Berninger (1992) suggested that thornlessness is determined by a recessive, although our observations support multigenic control. In the closely related genus Rubus, thornlessness is due to a dominant gene (Norton and Skirvin 1997). Winterhardiness is multigenic control and may be controlled by very few or closely linked genes (Svejda 1979). The inheritance of flavonoid pigments (pelargonidin, cyanidin, kaempferol, and quercitin) appears to be multigenic and additive (De Vries et al. 1974, 1980; Marshall et a1. 1983). Experiments of plant regeneration from callus (Arene et aI., 1993) have shown that some somaclonal variants had systematically associated characters, such as a fixed number of petals (10) and dwarfness or a fixed number of petals (5) and an orange color. These associations suggest the existence of linkage between the genes responsible for these characters. Recent studies by N. Zieslin (pers. commun.) indicate that an abnormal development of rose flowers known as phyllody, which consists of

172

S. GUDIN

a leafy aspect of the center of the flower (Mor and Zieslin 1992), is genetically controlled. This group also demonstrated the involvement of a gene RAG, homologous to AG of Arabidopsis thaliana (L.) Heynh. (Yanofsky 1995), in the control of phyllody development. The gene expression at early stages of rose flower development is strongly affected by climatic and stress conditions in the root environment.

IV. BREEDING OBJECTIVES A. Cut Flowers

Traditional breeding objectives include greater productivity, especially under lower temperatures, postharvest vase life, and tolerance to pests and diseases (Halevy 1986; Berninger 1992; De Vries and Dubois 1996; Van den Berg 1996). Increased fragrance in cut roses is now desired by consumers (Van de Pol et al. 1986; De Vries and Dubois 1996), and is now under selection by breeders. Gudin (1995), however, has indicated that high and early expression of "real rose" fragrance is physiologically incompatible with good postharvest life although some rose fragrances such as anis are not linked to poor vase life, as typified in 'Meipasty'. Postharvest longevity is increasingly important as rose production moves increasingly to developing countries of the subtropics (Zieslin 1996). Adaptation to complex flower transport networks are of primary importance. Thus, thornlessness, because of easier manipulation, is also highly desirable and thornless stems are more easily bundled by mechanical bunching, which is now widespread in northern Europe. Recently, some cultivars have been released, such as 'Meicofum', 'Meicobuis', 'Meileoduyv', 'Olijbrau', or 'Meinalpir', which change color throughout flower opening. This type of color evolution, combined with the characteristic rose shape changes during flower bud opening, could create a new fashion. Some other desirable traits are related to changes in growing technology. The development of soilless cultivation and the required degree of plant nutrition control (Brun et al. 1996; Brun and 5ettembrino 1996; Cabrera et al. 1996) suggest an advantage for cultivars that perform well on their own roots and can therefore be efficiently propagated by cuttings, which are less expensive than grafts. Cultivars with reliable bottom break sprouting are preferred because of the popularity of the "bending" technique associated with low plant height management (Kool 1996; Tsujita and Blom 1996; Dominguez 1997; Morisot 1997).

3. ROSE: GENETICS AND BREEDING

173

B. Garden Roses

There is a growing demand for fragrant cultivars in garden roses just as there is in cut roses. As this character is already present in the existing garden commercial range, some breeders are diverting their efforts toward the introduction of new kinds of fragrances or blends such as anis, lemon, peach, pear, apple, raspberry, or coumarin scents 0. Mouchotte, pers. commun.). However, the fragrance character, at least for "real rose" fragrance, appears to be genetically linked to undesirable disease sensitivity, such as to blackspot or botrytis blight (Gudin 1995). Ancient flower shapes, similar to R. centifolia, are again fashionable. This explains the success of such commercial product lines as "English Roses" or "Romantica"®. Moreover, there is a growing market for landscape roses, not only because of the interest of the landscape professionals (highway companies, town or landscape architects) but also because of amateur gardeners. Indeed, this kind of roses is usually selected in order to be as "carefree" as possible. Therefore, genotypes must be relatively tolerant to diseases and pests and require less pesticides than traditional garden roses, especially in northern Europe, where environmental sensitivity is increasing (De Vries and Dubois 1996). Because the "landscape rose" concept is growing, an all-season decorative effect is sought. This includes good and esthetic hip set. Some breeders are attempting to develop "multi-use" cultivars that could be grown in the garden with attractive hips, and produce cut flowers, and perhaps be vigorous on their own roots (]. Mouchotte, pers. commun.). V. BREEDING CRITERIA AND SELECTION PROCEDURES A. Selection for Disease and Pest Resistance

New in vitro and in vivo selection procedures can aid in identifying sources of resistance in rose germplasm. Building on previous investigations of Knight and Wheeler (1978), Walker et al. (1996) succeeded in producing in vitro infective conidia of DipJocarpon rosae Wolf (the fungus responsible for blackspot disease) and used cell cultures and filtrates of various roses to evaluate resistance. Wisniewska-Grzeszkiewicz and Wojdyla (1996) classified 120 rose cultivars, according to their resistance level, by studying in vitro the occurrence of stem canker (caused by Coniothyrium juckeJii Sacc.), botrytis blight on petals (caused by Botrytis cinerea Pers. ex Fr.), and blackspot disease of leaves. The same gernlplasm was evaluated by controlled in vivo infections with Sphaerotheca

174

S. GUDIN

pannosa (Wallr.) Lev. (responsible for powdery mildew) and Peronospora sparsa Berk. (responsible for downy mildew). Voisin et al. (1996) studied the host resistance of 32 different rose rootstocks to a nematode (Meloidogyne hapla Chitwood), using a high inoculum pressure test. Reynders-Aloisi et al. (1996, 1998) analyzed the reaction of various rootstocks and botanical species following inoculation with different strains of Agrobacterium tumefaciens (E.F.Sm.:;tTown.) (responsible for the crown gall disease). B. Prediction of Flower Productivity De Vries (1976) demonstrated that the time from seedling emergence to flowering can be useful to predict flower production. De Vries et al. (1978) developed a method for screening seedlings at low irradiance in order to select for low blind shoot occurrence, a major determinant of yield (De Vries et al. 1978).

C. Prediction of Postharvest Longevity Barthe et al. (1991a,b) demonstrated that vase life can be efficiently and objectively monitored by measuring pH, conductivity, and osmolarity of the petal cell sap. The pH, measured at a well-defined stage of flower opening, was strongly correlated to vase life (Gudin 1995; Gudin and Mouchotte 1996). This kind of screening could be applied early in the selection process, and was easier to standardize and more reliable than vase-life tests, such as those described by Van Doorn et al. (1991) and Gudin (1992b). Moreover, the physiological indicator method was less influenced by the plant-growing environment conditions than vase-life tests, which are noticeably influenced by preharvest temperature (Gudin 1992b). This early indicator test was used during the selection of some cut-rose cultivars with outstanding postharvest abilities, such as 'Meitanet,' 'Meitinor', 'Meiyacom', and 'Meipasty', recently released by Meilland Star Rose.

D. Predicting Rootstock/Cultivar Compatibility Rose bushes are mainly propagated by grafting, but rootstock compatibility varies (Van de Pol et al. 1988; De Vries and Dubois 1989). Zieslin et al. (1987) described an in vitro method using callus that may predict compatibility levels more quickly than traditional grafting tests.

3. ROSE: GENETICS AND BREEDING

175

E. Adaptation

With the increasing development of worldwide production of cut roses, multi-location trials, as highly recommended by Austin (1988), have increased in importance. A world selection network has been organized since the 1980s. Thus, Meilland Star Rose sends propagation material of new and potential cut-flower cultivars at an early stage of evaluation to 13 testing agencies distributed all around the world, covering all the existing range of rose-growing climatic and cultural conditions, from soilless, artificially lighted, CO 2 supplemented, and well-heated cultivation under glass in the Netherlands, to cultivation in soil under simple unheated plastic structures in the high altitudes of the savanna of Bogota, Colombia. International trials permit evaluation of new cultivars on different root systems, according to regional habitats. Plants can thus be observed from cuttings or as grafts using various rootstocks such as R. indica var. Major, R. canina, R. multiflora, cvs. Natal Briar and Dina. VI. BREEDING TECHNOLOGY

A. Sexual Reproduction Breeding efficiency can be greatly increased by improving the organization of pollination, and seed handling (Gudin 1995; Gudin and Mouchotte 1996). 1. Pollen and Pollination. A one-month cold treatment (4°C) of rose plants just after cut-back increased hip and seed sets after self- or crosspollination as a result of improved pollen germination and female fertility (Gudin 1992a). Pollen germination is improved at 23-30°C with 60-65% relative humidity (Gudin et a1. 1991c), by the choice of pollination season, May-June in the south of France (Gudin et a1. 1991a), and by choosing female parental cultivars with a stigmatic exudate of low pH, approximately 5 (Gudin and Arene 1991). A putrescine treatment of the stigmata increased pollen germination and female fertility (Gudin and Arene 1992). 2. Seed Maturation and Germination. The innermost layer of the pericarp in the rose achene, the endocarp, is the most impermeable layer. Endocarp thickness and, therefore, seed germinability, was influenced by temperature during achene development (Gudin et a1. 1990, 1991b). The reduction of endocarp thickness seems to be related to a

176

S. GUDIN

temperature-regulated increase in the rate of embryo development and may be explained by nutritional competition between the embryo and this seed coat layer, as previously reported in tomato (Nitsch 1951) and barley (Norstog 1961). The endocarp thickness is dependent upon genotype (Gudin et al. 1990). Low-density seeds were shown to have low germination rates because of micro-cracks in their pericarp that result in bad protection of the embryo (Gudin et al. 1992). An optimal seed postmaturation treatment involved stratification in darkness for 1 month at 23°C and then 2 months at 4°C (Gudin et al. 1990). Yambe and Takeno (1992) demonstrated an impressive improvement of germination in R. multiflora after a seed treatment with macerating enzymes.

B. Interspecific Hybridization 1. Ploidy Level Manipulation. Most modern rose cultivars are tetraploids

(Shahare and Shastry 1963). However, it has long been known (Palmer et al. 1966; Saunders 1970) that some wild species, mainly diploids, represent a potential source of high resistance to diseases, especially to blackspot. Therefore, different strategies were developed in order to exploit this germplasm. The first crosses between these diploid species and tetraploid cultivars led to sterile triploids (Wylie 1955b). Attempts were made to double the chromosome numbers of the species either by colchicine treatments of seedling shoots (Semeniuk and Arisumi 1968) or culturing seedlings in vitro in a liquid medium (Roberts et al. 1990). There were few positive results, because of the high frequency of cytochimeras (Roberts et al. 1990). Byrne et al. (1996) have used amphidiploidy for the development of blackspot-resistant rose germplasm. A natural fertile amphidiploid, R. kordesii, combines the diploid genomes of R. rugosa and R. wichuraiana (Wulff 1951). The amphidiploid strategy had been used earlier by Svejda (1977) in a breeding program for increased winterhardiness. Partially fertile hybrid roses have been obtained from crosses between amphidiploids and tetraploid cultivars (Basye 1990, 1992). Our haploidization attempts by anther, microspore, or ovule culture, carried out in the late 1980s, were unsuccessful. Meynet et al. (1994) succeeded in haploidizing the tetraploid 'Sweet Promise' (syn. 'Sonia Meilland'®) through parthenogenesis induced by using irradiated pollen and in vitro culture of immature embryos. The resulting diploids are able to produce flowers and pollen (Meynet et al. 1996) and some are fertile (J. Meynet, pers. commun.). This technique could lead to a better understanding of basic rose genetics, because tetraploidy of cultivated roses

3. ROSE: GENETICS AND BREEDING

177

makes conventional genetic analysis difficult. Gene transfer from diploid species to haploidized cultivars can be envisaged. Furthermore, haploidized plants could be ploidized by colchicine treatment (Semeniuk and Arisumi 1968; Roberts et al. 1990) in order to obtain tetraploids with an increased level of homozygosity. Agitation of excised nodes in colchicine solution or tissue culture of shoots on a medium with colchicine is an efficient technique for chromosome doubling (Ma et al. 1997). Use of the spindle inhibitor oryzalin needs to be attempted. 2. Embryo Rescue. Embryo rescue in the Rosaceae was first reported by Tukey (1933) on cherry, which permitted the incorporation of earliness. This technique has been used in roses (Gudin 1994; Marchant et al. 1994; Meynet et al. 1994). Progenies have been obtained from two crosses that traditionally fail because of early abortion (Gudin 1994). The embryos had either to be rescued very early in ovulo (heart-shape stage, about 0.25 mm long) or, isolated at later stages (Gudin 1994). Early embryo rescue has also been used in the haploidization method using pollination with irradiated pollen (Meynet et al. 1994). An original interspecific hybrid, R. rugosa x R. foetida, has been successfuly produced with embryo rescue (Gudin and Mouchotte 1996). 3. Protoplast Fusion. Matthews et al. (l991) demonstrated a protoplast to plant system in roses. Recently, Mottley et al. (l996) regenerated putative somatic intergeneric hybrid plants involving R. hybrida cv Frensham and blackberry or cherry. Morphology and RAPD analysis suggest that the plants correspond to genetic variants of the rose cultivar. Protoplasts of the diploid hybrid, R. persica Michx. x R. xanthina, have been self-fused and tetraploid plants were obtained. Protoplast fusion could become a reliable method to obtain hybrids from difficult or impossible sexual crosses, but it is more complicated than early embryo rescue. Embryo rescue is probably the technology of choice to obtain interspecific hybrids and should be also investigated on intergeneric crosses. Rose hips have developed, for instance, following pollination with strawberry pollen at Interplant BV, Netherlands (Roberts and Lewis 1996).

C. Mutation Breeding 1. Radiation and Chemical Mutagens. There has been considerable

work done on induced mutations in roses, using ethylmethanesulphonate (Kaicker 1982), ionizing radiations (Chan 1966; Streitberg

178

S. GUDIN

1966a,b; Gupta and Shukla 1970, 1971; Nakajima 1973; Lata 1975; Lata and Gupta 1975; Dommergues 1976; Broetjes and Van Harten 1978; Gupta and Datta 1982; Gupta et al. 1982; Kaicker and Dhyani 1985; Smilanksy et al. 1986), sometimes combined with in vitro culture (Walther and Sauer 1985). Mutation breeding has been used by breeding companies such as Kordes-Sohne, Rosen-Tantau (Walther and Sauer 1985), and Selection Meilland, but this technique has led to only a few cultivars (Maluszynski et al. 1992). Up to 1974, only 27 cultivars were registered as resulting from induced mutations, as compared to 35 alstroemerias, 187 chrysanthemums, and 34 dalhias (Sigurbjornsson and Micke 1974). This is a poor result considering that almost 10% ofthe 18,000 cultivars listed by Haring (1986) derive from spontaneous sports. Induced mutation breeding is now infrequently used. During the First International Symposium on Roses at Rehovot, Israel, 7-12 July 1985 (Acta Hort., vol. 189,1986), only one communication was made on the subject and, during the Second International Symposium on Roses at Antibes, France, 20-24 February 1995 (Acta Hort., vol. 424, 1996), there was none. Broertjes and Van Harten (1978) have suggested that the lack of results of induced mutation is a consequence of the long time it takes from mutagenic treatment up to the detection of mutants and their large-scale propagation. Up to now, induced mutation studies only attempted to induce color variants of commercialized cultivars, similar to those obtained by spontaneous mutation (Kaicker 1982). 2. Somaclonal Variation. Hill (1967) was the first to observe morpho-

genesis of shoot primordia on rose calluses. Tweddle et al. (1984) obtained shoot induction and Valles and Boxus (1987) were the first to report complete plant regeneration. Since then, many authors reported successful regeneration either through organogenesis or somatic embryogenesis (Lloyd et al. 1988; Burger et al. 1990; Leffering and Kok 1990; De Wit et al. 1990; Rout et al. 1991; Noriega and Sondahl1991; Matthews et al. 1991; Moyne et al. 1993). The explants and media used by these authors have been reviewed by Roberts et al. (1995). Arene et al. (1993) obtained successful regeneration through both organogenesis and somatic embryogenesis from various explants, including stem, leaf, root segments, zygotic embryos, anthers, ovules, petals, sepals, and receptacles. Regenerants, unlike those from cuttings or from direct in vitro regeneration (without callus formation), often display somaclonal variation. Of these variants, 21 % derived from calluses developed on vegetative explants and 70% from calluses developed on zygotic embryos. The variants differed from the original cultivar in the number, shape, and color of petals, or growth habit. Yokoya et al. (1996) demonstrated

3. ROSE: GENETICS AND BREEDING

179

that plant regeneration from callus can be successful with many rose cultivars. Exploitation of somaclonal variation through callus culture might become a source for new cultivars if this method is combined with strategic and efficient in vitro selection pressures (Gudin and Mouchotte 1996).

D. Biotechnology 1. Molecular Markers.

Selection for Quantitative Traits. Published reports about the use of molecular marker assisted selection in roses are not yet available. However, Debener and Mattiesch (1996) have demonstrated the usefulness of RAPD markers for the construction of a chromosome linkage map, using crosses between R. multiflora derived genotypes that differed in a range of floral and vegetative characters. Furthermore, Debener et al. (1997) have used RAPD in order to achieve a parentage analysis in interspecific crosses between different wild rose species. Cultivar Identification. Varietal identification is now benefiting from the use of molecular markers, including restriction fragment length polymorphism, RFLP (Rajapakse et al. 1992; Hubbard et al. 1992), isozyme (Grossi et al. 1997, and 1998), and RAPD (Torres et al. 1993). Isozymes were less able to characterize different cultivars than RAPD markers (Cubero et al. 1996) and RAPD is now the most widely used technique (Ballard et al. 1996; Debener et al. 1996; Debener and Mattiesch 1996; Gallego and Martinez 1996; Moreno et al. 1996; Reynders-Aloisi and Bollereau 1996; Debener and Mattiesch 1998). DNA microsatellites have also been investigated (Vainstein and Ben-Meir 1994) and although they are expensive, results have been extremely precise. Vainstein et al. (1995) demonstrated that the probability of two offspring from the crossing of similar rose genotypes having identical DNA fingerprints is as low as 2 x 10-H• An application study of the amplified fragment length polymorphism (AFLpTM) technique is under investigation (D. Zhang and S. Reynders-Aloisi, pers. commun.). Reynders-Aloisi and Bollereau (1996) have stressed that some rose species identification and classification based on morphology is still unsatisfactory, so that the molecular approaches should be of great help in solving cultivar and species identification problems. 2. Genetic Transformation. The first published report of successful transformation in Rosa was that of Van der Mark et al. (1990) who transformed several clones of R. canina using Agrobacterium rhizogenes as

180

S. GUDIN

a vector in vivo, subsequently confirmed by Walker (1996). Firoozabady et al. (1994) co-cultivated embryogenic callus of R. hybrida cv Royalty with A. tumefaciens and A. rhizogenes and obtained transgenic plants regenerated from somatic embryos or newly formed shoots. They introduced the rose chalcone synthase gene that reduced the color intensity of the original cultivar. Souq et al. (1996) co-cultivated embryogenic tissues of R. hybrida cv. Deladel with A. tumefaciens and obtained two types of transgenics, one incorporating a sequence of a gene modifying morphology and the other incorporating a rose chalcone synthase antisense gene. The morphological variants (which may be somaclonal variants) included dwarfs, with wrinkled leaves, increased branching, rare or abnormal sexual organs, numerous small thorns, very reduced flower diameter, modified shape or with greatly reduced numbers of petals. Antisense gene transformants had reduced petal color intensity, unlike Petunia and Chrysanthemum, where total suppression of chalcone synthase synthesis resulted in completely white flowers (Courtney-Gutterson 1994). Although these results clearly show that rose transformation is feasible, they demonstrate that many steps must be controlled before the technique can be efficiently used in rose breeding. The regenerated transgenic plants must exhibit the newly introduced trait and should not be chimeric. Furthermore, it is clear that indirect regenerating systems may be a source of somaclonal variants (Arene et al. 1993; Souq et al. 1996). The integration site of the inserted gene probably has an important influence on gene expression (Souq et al. 1996). Recently, Van der Salm et al. (1998) have obtained a ROL gene transformant of the rootstock 'Moneyway', whose rooting abilities have been changed. Even more recently, Marchant et al. (1998) have obtained a kitinase expressing transformant of R. hybrida with increased tolerance to the blackspot disease. VII. CONCLUSION

The present-day rose breeder can choose among several options to traditional hybridizing and selection. This has been made possible by the development of many new biotechnology techniques. Some of these new techniques, however, have not been perfected in Rosa, as compared to other plants, and further research is needed before they can be used with full efficiency. Careful identification of the improvement strategy, according to the objective, is of primary importance. For example, the search for a blue rose, which is ongoing in a private biotechnology company (Dalling 1991), justifies the use of transformation technology because the pigment responsible for a genuinely blue flower

3. ROSE: GENETICS AND BREEDING

181

color (delphinidin) is unknown in the genus Rosa. The genome portion responsible for expression had to be obtained from Petunia (Dalling 1991). However, in this example, even if transformation and regeneration of transformed plants are successfully handled, interfering factors such as co-pigmentation, vacuolar pH (Buxton and Darbishire 1929; Lawrence 1932; Robinson 1935; Scott-Moncrieff 1936), and gene silencing (Starn et al. 1997) may affect the final result. A tremendous pool of variation is still available in Rosa. Although close to 200 rose species have now been described worldwide, only about 10 have been used to create the enormous available cultivar diversity (Maia and Venard 1976). Tremendous improvement potential is still available through traditional hybridizing, but "sophisticated" techniques, such as embryo rescue, protoplast fusion, and ploidy level manipulation, may be required to better exploit this variation. Increased knowledge of the basic genetics of the genus is still required. Nevertheless, plant breeding still remains a dynamic force in the rose industry, with 25 to 30 companies and numerous amateur breeders involved.

LITERATURE CITED Abler, E. 1990. Secrets of the miniature rose. Dgicom Publ., Lake Forest, 11. Antonelli, A., C. Fabbri, M. E. Giogioni, and R. Bazzocchi. 1997. Characterization of 24 old garden roses from their volatile compositions. J. Agr. Food Chem. 45:4435-4439. Arene, 1., C. Pellegrino, and S. Gudin. 1993. A comparison of the somaclonal variation level of Rosa hybrida 1. cv Meirutral plants regenerated from callus or direct induction from different vegetative and embryonic tissues. Euphytica 71 :83-90. Austin, R. B. 1988. New opportunities in breeding. HortScience 23:41-45. Ballard, R, S. Rajapakse, A. Abbott, and D. Byrne. 1996. DNA markers in rose and their use for cultivar identification and genome mapping. Acta Hart. 424:265-268. Barthe, P., V. Vaillant, and S. Gudin. 1991a. Definition of indicators of senescence in the rose: Effect of the application of plant hormones. Acta Hart. 298:61-68. Barthe, P., V. Vaillant, and S. Gudin. 1991b. pH of cell sap and vacuolar pH during senescence of the rose petal. Acta Hart. 298:135-139. Basye, R. 1990. An amphidiploid of Rosa banksiae and R. Jaevigata induced by colchicine. Am. Rose Annu. 75:82-88. Basye, R 1992. The future of the rose. 1992. Am. Rose Annu. 77:60-63. Berninger, E. 1992. Le rosier de serre pour fleurs a couper. p. 490-504. In: A. Gallais and H. Bannerot (eds.), Amelioration des especes vegetales cultivees. INRA, Paris. Barch, K., M. H. Williams, and 1. Hoyer. 1996. Influence of simulated transport on postharvest longevity of three cultivars of miniature potted rose. Acta Hart. 424:175-180. Broertjes, C., and A. M. Van Harten. 1978. Application of mutation breeding methods on the improvement of vegetatively propagated crops. Elsevier Sci. Publ. Co., Amsterdam. Brun, R, 1. Chazelle, Y. Tarere. and S. Voisin. 1996. Nycthemeral uptake of water and nitrate during the growth of rose plants in hydroponic culture. Acta Hart. 424:59-66. Brun. R., and A. Settembrino. 1996. Production and quality of Rosa hybrida cv Sonia in

182

S. GUDIN

rockwool at three levels of electrical conductivity combined with two levels of relative humidity. Acta Hort. 424:115-118. Buck, G. J. 1960. Progress report on breeding hardy everblooming roses. Am. Rose Annu. 45:95-99.

Burger, D. W., 1. Liu, K. W. Zary, and C. I. Lee. 1990. Organogenesis and plant regeneration from immature embryos of Rosa hybrida 1. Plant Cell Tissue Org. Cult. 21:147-152. Buxton, B. H., and F. V. Darbishire. 1929. On the behaviour of anthocyanins in varying hydrogen-ion concentrations. Proc. John Innes Hort. Inst. 21:71-103. Byrne, D. H., W. Black, Y. Ma, and H. B. Pemberton. 1996. The use of amphidiploidy in the development of blackspot resistant rose germplasm. Acta Hort. 424:269-272. Cabrera, R. I., R. Y. Evans, and J. 1. Paul. 1996. The uptake of nitrate and ammonium by greenhouse roses. Acta Hort. 424:53-58. Chan, A. P. 1966. Chrysanthemum and rose mutations induced by X-rays. Proc. Am. Soc. Hort. Sci. 88:613-620. Courtney-Gutterson, N. 1994. The biologist's palette: Genetic engineering of anthocyanin biosynthesis and flower colour. p. 93-124. In: B. E. Ellis (ed.), Genetic engineering of secondary metabolism. Plenum Press, New York. Crane, M. B., and W. J. C. Lawrence. 1956. The genetics of garden plants. Macmillan and Co. Ltd., St. Martin's Press, New York. Crepin, F. 1889. Sketch of a new classification ofroses. J. Royal Hort. Soc. 11:217-228. Cubero, J. I., T. Millan, F. Osuna, A. M. Torres, and S. Cobos. 1996. Varietal identification in Rosa by using isozyme and RAPD markers. Acta Hart. 424:261-264. Dalling, M. 1991. Color it blue. Am. Rose Mag. Nov. p. 32-33. Darlington, C. D., and A. P. Wylie. 1955. Chromosome atlas of flowering plants. George Allen and Unwin Ltd., London. Debener, T., C. Bartels, and 1. Mattiesch. 1996. RAPD analysis of genetic variation between a group of rose cultivars and selected wild rose species. Molecular Breed. 2:321-327. Debener, T., C. Bartels, and W. Spethmann. 1997. Parentage analysis in interspecific crosses between rose species with RAPD markers. Gartenbauwissenschaft 62:180-184. Debener, T., and 1. Mattiesch. 1996. Genetic analysis of molecular markers in crosses between diploid roses. Acta Hort. 424:249-252. Debener, T., and 1. Mattiesch. 1998. Effective pairwise combination of long primers for RAPD analyses in roses. Plant Breed. 177:1-5. Demilly, D., B. Billy, M. Lemaire, and Y. Dattee. 1994. Denombrement du genre Rosa, collection de la roseraie. Colloque du centenaire de la roseraie de Val-de-Marne. Les avancees scientifiques et techniques dans Ie genre Rosa. (Recueil des actes). p. 85-95. De Vries, D. P. 1976. Juvenility in hybrid tea-roses. Euphytica 25:321-328. De Vries, D. P., and 1. A. M. Dubois. 1978. On the transmission of the yellow flower colour from Rosa foetida to recurrent flowering hybrid tea-roses. Euphytica 27:205-210. De Vries, D. P., and 1. A. M. Dubois. 1984. Inheritance of the recurrent flowering and moss characters in F} and F 2 Hybrid Tea x R. centifolia muscosa (Aiton) Seringe populations. Gartenbauwissenschaft 49:97-100. De Vries, D. P., and 1. A. Dubois. 1989. Variation ofthe shoot production of Rosa hybrida "Sonia" as induced by different Edelcanina rootstock clones. Gartenbauwissenschaft 53:211-215.

De Vries, D. P., and 1. A. M. Dubois. 1996. Rose breeding: Past, present, prospects. Acta Hort. 424:241-248. De Vries, D. P., 1. A. Dubois, and 1. Smeets. 1978. Hybrid tea-roses under controlled light conditions. 3. Flower and blind shoot production in the glasshouse of seedlings selected far flowering or flower bud abortion at low irradiances in a growth room. Neth. J. Agr. Sci. 26:399-404.

3. ROSE: GENETICS AND BREEDING

183

De Vries, D. P., F. Garretsen, L. A. Dubois, and H. A. Van Keulen. 1980. Breeding research on rose pigments. II. Combining ability analyses of variance offour flavonoids in F 1 populations. Euphytica 20:115-120. De Vries, D. P., H. A. Van Keulen, and J. W. De Bruyn. 1974. Breeding research on rose pigments. 1. The occurrence of flavonoids and carotenoids in rose petals. Euphytica 23:447-457. De Wit, J. c., H. F. Esendam, J. J. Honkanen, and U. Tuominen. 1990. Somatic embryogenesis and regeneration of flowering plants in rose. Plant Cell Rep. 9:456-458. Dominguez, M. 1997. Tecnica del pulm6n en rosal. Horticultura 123:15-17. Dommergues, P. 1976. Mutagenese induite chez Ie rosier. Hort. Franc;aise. 74:1-3. Dubois, L. A. M., and D. P. De Vries. 1987. On the inheritance of the dwarf character in Polyantha x Rosa chinensis minima (Sims) Voss F 1 populations. Euphytica 36: 535-539. Erlanson, E. W. 1929. Cytological conditions and evidence for hybridity in North American wild roses. Bot. Caz. 57:443-506. Erlanson, E. W. 1930. The occurrence and cytology of triploid, aneuploid and partially sterile individuals among roses in the Section Cinnamomeae. Proc. 5 Int. Bot. Congr., Cambridge. p. 223-224. Erlanson, E. W. 1933. Chromosome pairing, structural hybridity and fragments in Rosa. Bot. Gaz. 94:551-566. Erlanson, E. W. 1934. Experimental data for a revision of the North America wild roses. Bot. Gaz. 96:197-259. Firoozabady, E., Y. Moy, N. Courtney-Gutterson, and R. Robinson. 1994. Regeneration of transgenic rose (Rosa hybrida) from embryogenic tissue. BioTechnologie 2:609-613. Gallego, F. J., and L. Martinez, 1996. Molecular typing of rose cultivars using RAPDs. J. Hort. Sci. 71:901-908. Gandelin, M. H., and M. N. Mistou. 1994. Harmonisation des descriptions de varietes de rosier au plan national et international. Colloque du centenaire de la roseraie du Valde-Marne. Les avancees scientifiques et techniques dans Ie genre Rosa. (Recueil des actes). p. 113-133. Grossi, c., O. Raymond, and M. Jay. 1997. Isozyme polymorphism of Rosa spp. and cultivar identification. Euphytica 98:11-19. Grossi, C., O. Raymond, and M. Jay. 1998. Flavonoid and enzyme polymorphisms and taxonomic organization of Rosa sections: Carolinae, Cinnamomeae, Pimpinellifoliae and Synstylae. Biochemical Systematics and Ecology. 26:857-871. Grossi, c., O. Raymond, C. Sanlaville-Boisson, and M. Jay. 1999. Rosa taxonomy and hierarchy of markers defined by ACT-STATIS. Z. Naturforsch. 54:25-34. Gudin, S. 1992a. Influence of bud chilling on subsequent reproductive fertility in roses. Scientia Hort. 51:139-144. Gudin, S. 1992b. Effect of preharvest growing temperatures on the development of cut roses. Postharvest BioI. Technol. 2:155-161. Gudin, S. 1994. Embryo rescue in Rosa hybrida L. Euphytica 72:205-212. Gudin, S. 1995. Rose improvement, a breeder's experience. Acta 11ort. 420:125-128. Gudin, S., and L. Arene. 1991. Influence of the pH of the stigmatic exudate on male-female interaction in Rosa hybrida L. Sex. Plant Reprod. 4:110-112. Gudin, S., and L. Arene. 1992. Putrescine increases effective pollination period in roses. HortTechnology 2:211-213. Gudin, S., L. Arene, and C. Bulard. 1991a. Influence of season on rose pollen quality. Sex. Plant Reprod. 4:113-117. Gudin, S., L. Arene, and A. Chavagnat. 1992. Relation entre imbibition, densite, taux de remplissage et faculte germinative chez l'akene de Rosa hybrida L. Agronomie. 12:123-126.

184

S. GUDIN

Gudin, S., L. Arene, A. Chavagnat, and C. Bulard. 1990. Influence of endocarp thickness on rose achene germination: Genetic and environmental factors. HortScience 25:786-788. Gudin, S., L. Arene, A. Chavagnat, and C. Bulard. 1991b. Structure and water absorption of different parts of the rose achene pericarp during imbibition. BioI. Plantarum 33: 262-267. Gudin, S., L. Arene, and C. Pellegrino. 1991c. Influence of temperature and hygrometry on rose pollen germination. Adv. Hort. Sci. 5:96-98. Gudin, S., and J. Mouchotte. 1996. Integrated research in rose improvement. A breeder's experience. Acta Hart. 424:285-292. Gupta, M. N., and S. K. Datta. 1982. Effects of gamma rays on roses-sprouting, growth and induction of somatic mutation. Ind. Rose Annu. 2:42-54. Gupta, M. N., P. Nath, and S. K. Datta. 1982. Effects of gamma irradiation on Rosa Damascena Mill. Ind. Rose Annu. 2:55-61. Gupta, M. N., and R. Shukla. 1970. Mutation breeding of garden roses. Studies on induction of somatic mutations after gamma irradiation of bud-wood. Progr. Hort. 2:69-83. Gupta, M. N., and R. Shukla. 1971. Effects of gamma irradiation on some scented roses. Japan. J. Breed. 21:129-136. Gustafsson, A. 1944. The constitution of the Rosa canina complex. Hereditas 30:405-428. Halevy, A. H. 1986. Rose research: Current situation and needs. Acta Hort. 189:11-20. Haring, P. A. 1986. Modern roses 9. American Rose Society, Shreveport. LA. Haskell, G. 1960. Role of the male parent in crosses involving apomictic Rubus species. Heredity 40:101-113. Hill, G. P.1 967. Morphogenesis of shoot primordia in cultured stem tissue of a garden rose. Nature 216:596-597. Hubbard, M., J. Kelly, S. Rajapakse, A. G. Abbott, and R. E. Ballard. 1992. Restriction fragment length polymorphisms in rose and their use for cultivar identification. HortScience 27:172-173. Hurst, C. C. 1925. Chromosomes and characters in Rosa and their significance in the origin of species. Experiments in Genetics 37:534-550. Hurst, C. C. 1927. Differential polyploidy in the genus Rosa L. Verhandlungen des V. Internationalen Kongresses fur Vererbungswissenschaft. p. 867-906. Hurst, C. C. 1941. Notes on the origin and evolution of our garden roses. J. Royal Hort. Soc. 66:73-82. Jacob, Y., C. Teyssier, S. C. Brown, and S. Reynders-Aloisi. 1996. Use ofcytometry for the rapid determination of ploidy level in the genus Rosa. Acta Hort. 424:273-278. Jay, M., J. Reynaud, O. Raymond, and C. Grossi. 1996. Les composes phenoliques et la caracterisation des niveaux taxonomiques inferieurs du monde vegetal. Acta Bot. Gallica 143:481-491. Jones, R. N. 1975. B chromosome systems in flowering plants and animal species. Int. Rev. Cytol. 40:1-18. Kaicker, U. S. 1982. Mutation breeding in roses. Ind. Rose Annu. 2:35-42. Kaicker, U. S., and D. Dhyani. 1985. Induced mutations in roses. Mutation Breed. Newsletter 25:6-7. Kihara, H. 1944. Chromosome conjugation in a rose hybrid, Rosa multiflora x R. rugosa. Japan. ]. Genet. 20:55-58. Klasterska, I. 1969. Cytology and some chromosome numbers of Czechoslovak roses I. Folia Geobot. Phytotax. Praha. 4:175-188. Klasterska, I., and I. Klastersky. 1974. Cytology and some chromosome numbers of Czechoslovak roses II. Bot. Notiser. 127:328-341.

3. ROSE: GENETICS AND BREEDING

185

Knight, C., and B. E. ]. Wheeler. 1978. Evaluating resistance of roses to blackspot. Phytopath. Z. 91:218-229. Koncalova, M. N., and I. Klastersky. 1978. Cytology and chromosome numbers of some Czechoslovak roses III. Folia Geobot. Phytotaxon. 13:67-87. KooL M. T. N. 1996. System development of glasshouse roses. Ph.D. Thesis. Univ. Wageningen, The Netherlands. Lata, P. 1975. Effects of ionizing radiation on roses. II: Meiotic studies on rose cultivar "Pink Parfait" and its gamma induced mutants. Cytologia 40:633-638. Lata, P. 1982. Inheritance of B chromosomes in garden roses. Genetica 58:51-54. Lata, P., and M. N. Gupta. 1975. Effects of ionizing radiations on roses: Meiotic studies on control and gamma ray induced mutants of rose cultivar "Montezuma." Cytologia 40:213-219.

Lawrence, W.

J. C.

1932. Interaction of flavones and anthocyanins. Nature 129:834-

835.

Leemans, J. A. 1966. Porte-greffes pour rosiers. Haarlem, Netherlands. Leffering, L., and E. Kok. 1990. Regeneration of rose via leaf segments. Prophyta 8:245. Lewis, W. H. 1962. Monograph of Rosa in North America. IV. R. x dulcissima. Brittonia 14:572-579.

Lewis, W. H., and R. E. Basye. 1961. Analysis of nine crosses between diploid Rosa species. Proc. Am. Soc. Hort. Sci. 78:573-579. Lloyd, D., A. V. Roberts, and K. C. Short. 1988. The induction in vitro of adventitious shoots in Rosa. Euphytica 37:31-36. Ma, Y., D. H. Byrne, and J. Chen. 1997. Amphidiploid induction from diploid rose interspecific hybrids. HortScience 32:292-295. Ma, Y., M. N. Islam-Faridi, C. F. Crane, D. M. Stelly, H. J. Price, and D. H. Byrne. 1996. A new procedure for making metaphase chromosome slides in roses. HortScience 31 :855-857.

Maia, N., and P. Venard. 1976. Cytotaxonomie du genre Rosa et origine des rosiers cultives. p. 7-20. In: Travaux sur rosiers de serre. FNPHP, Antibes. Maluszynski, M., B. Sigurbjornsson, E. Amano, L. Sitch, and O. Kamra. 1992. Mutant varieties-Data bank FAO/IAEA data-base. Part II. Mutation Breed. Newslett. 39:14-33. Marchant, R., M. R. Dazey, J. A. Lucas, C. J. Lamb, R. A. Dixson. and J. B. Power. 1998. Expression of kitinase transgen in rose (Rosa hybrida L.) reduce development of blackspot disease (DiplocQIpon rosae Volf). Molecular Breeding. 4:187-194. Marchant, R., J. B. Power, M. R. Dazey, and J. Chartier-Hollis. 1994. Embryo rescue, for the production of F j hybrids, in English rose. Euphytica 74:187-193. Marshall, H. H. 1973. Rose breeding for the prairies. Canada Agr. 18:24-25. Marshall, H. H., C. G. Campbell, and L. M. Collicutt. 1983. Breeding for anthocyanin colors in Rosa. Euphytica 32:205-216. Matthews. D., J. Mottley, I. Horan, and A. V. Roberts. 1991. A protoplast to plant'system in roses. Plant Cell, Tissue Org. Cult. 24:173-180. Meenakshi, K. 1977. Chromosome number of some garden roses. Curr. Sci. 46:589-617. Meynet, J., R. Barrade, A. Duclos, and R. Siadous. 1994. Dihaploid plants of roses (Rosa x hybrida) obtained by parthenogenesis induced using irradiated pollen and in vitro culture of immature seeds. Agronomie 2:169-175. Meynet, J., E. Botton, J. Eychene. and F. Aime. 1996. Optimization of a method for the haploidization of cultivated roses. Acta Hort. 424:399-401. Mistou, M. N., M. H. Gandelin, and L. Poetto. 1996. National rose group: census of varieties and harmonization of descriptions. Acta Hart. 424:297-300.

186

S. GUDIN

Mor, J., and N. Zieslin. 1992. Phyllody malformation in flowers of Rosa x hybrida cv Motrea: Effects of rootstocks, flower position, growth regulators and season. ]. Expt. Bot. 43:89-93. Moreno, M. T., A. M. Torres, T. Millan, J. Armada, and J. I. Cubero. 1996. Use of molecular markers in taxonomic studies of Rosa sp. Acta Hort. 424:293-296. Morisot, A. 1997. La "pelate" au "canduite japonaise" du rosier: moins de rendement, plus de qualite. PHM-Rev. Hort. 386:22-25. Mottley, J., K. Yokoya, D. Matthews, J. Squirrell, and J. E. Wentworth. 1996. Protoplast fusion and its potential role in the genetic improvement of roses. Acta Hort. 424:389-392. Moyne, A. 1., F. Souq, 1. H. Yean, S. C Brown, M. Boulay, and B. S. Sangwan-Norreel. 1993. Relationship between cell ploidy and regeneration capacity of long term Rosa hybrida cultures. Plant Sci. 93:159-168. Nakajima, K. 1973. Induction of useful mutations of mulberry and roses by gamma rays. p. 105-116. In: Induced mutations in vegetatively propagated plants. IAEA, Vienna. Nitsch, J. P. 1951. Growth and development in vitro of excised ovaries. Am. J. Bot. 38:566-577. Noriega, C, and M. R. Sandahl. 1991. Somatic embryogenesis in hybrid tea roses. Biotechnology 9:991-993. Norstog, K. 1961. The growth and differentiation of cultured barley embryos. Am. J. Bot. 48:876-884. Norton, M. A., and R. M. Skirvin. 1997. Somaclonal variation among ex vitro "Thornless Evergreen" trailing blackberries: The morphological status of selected clones after seven years of field growth. J. Am. Soc. Hort. Sci. 122:152-157. Padhye, V. S. 1982. Fragrance in roses. Ind. Rose Annu. 2:100-112. Palmer, J. G., P. Semeniuk, and R. N. Stewart. 1966. Roses and blackspot. I. Pathogenicity to excised leaflets of Diplocarpon rosae from seven geographic locations. Phytopathology 56:1277-1282. Pertwee, J. 1995. Production and marketing of roses. 2nd ed. Pathfast Publ., Frinton on Sea, U.K. Raby, J. 1937. The cytology ofrose stocks. Gard. Chron. 102:145. Rajapakse, S., M. Hubbard, J. W. Kelly, A. G. Abbott, and R. E. Ballard. 1992. Identification of rose cultivars by restriction fragment length polymorphism. Scientia Hort. 52:237-245. Ratsek, J. C., W. S. Flory, Jr., and S. H. Yarnell. 1940. Crossing relations of some diploid and polyploid species of roses. Proc. Am. Soc. Hort. Sci. 38:637-654. Ratsek, J. C., S. H. Yarnell, and W. S. Flory, Jr. 1939. Crossing relations of some diploid species of roses. Proc. Am. Soc. Hort. Sci. 37:983-992. Raymond, 0., J. P. Biolley, and M. Jay. 1994. Exploitation statistique de la distribution des composes phenoliques floraux en vue de la comprehension du processus de domestication chez Ie rosier. Colloque du centenaire de la roseraie du Val-de-Marne. Les avancees scientifiques et techniques dans Ie genre Rosa. (Recueil des actes). p.35-67. Raymond, 0., J. P. Biolley, and M. Jay. 1995. Fingerprinting the selection process of ancient roses by means of floral phenolic metabolism. Biochemical systematics and Ecology. 23:555-565. Rehder, A. 1940. Manual of cultivated trees and shrubs. Macmillan, New York. p. 426-451. Reynders-Aloisi, S., and P. Bollereau. 1996. Characterization of genetic diversity in genus Rosa by randomly amplified polymorphic DNA. Acta Hort. 424:253-260. Reynders-Aloisi, S., G. PelloH, and A. Bettachini. 1996. Analysis of the behaviour ofvar-

3. ROSE: GENETICS AND BREEDING

187

ious rootstocks and botanical species of the genus Rosa towards Agrabacterium tumefaciens. Acta Hort. 424:215-220. Reynders-Aloisi, S., G. Pelloli, A. Bettachini, and C. Poncet. 1998. Tolerance to crown gall differs among genotypes of rose rootstocks. HortScience 33:296-297. Roberts, A. V. 1975. The nature and taxonomic significance of the system of inheritance in Rosa nanothamnus. Bot. J. Linn. Soc. 71 :59-66. Roberts, A. V. 1977. Relationship between species in the genus Rosa, section Pimpinellifoliae. Bot. J. Linn. Soc. 74:309-318. Roberts, A. V., and R Lewis. 1996. Rose phenotypes: questions of diversity and development. Acta Hort. 424:313-319. Roberts, A. V., D. Lloyd, and K. C. Short. 1990. In vitro procedures for the induction of tetraploidy in a diploid rose. Euphytica 49:33-38. Roberts, A. V.. and K. C. Short. 1979. An experimental study of mitosis. J. BioI. Ed. 13:195-198. Roberts, A. V., K. Yokoya, S. Walker, and J. Mottley. 1995. Somatic embryogenesis in Rosa spp. p. 278-289. In: S. Jain, P. Gupta and R Newton (eds.). Somatic embryogenesis in woody plants, vol. 2. Kluwer Academic Publishers, Amsterdam. Robinson, R 1935. Chemistry of the anthocyanins. Nature 135:732-736. Rout, G. R, B. K. Debata. and P. Das. 1991. Somatic embryogenesis in callus cultures of Rosa hybrida L. cv Landora. Plant Cell, Tissue Org. Cult. 27:65-69. Rowley, G. 1961. Aneuploidy in the genus Rosa. J. Genet. 57:253-257. Rowley, G. D.1966. The experimental approach to rose breeding. Scientia Hort. 18:131-135. Salaria, P. 1965. Chromosome numbers in cultivated roses. Hort. Srinagar. 1:128-168. Salesses, G. 1970. Sur Ie pbenomEme de cytomixie chez des hybrides triploYdes de prunier: consequences genetiques possibles. Ann. Amelior. Plantes. 20:383-388. Saunders, P. J. W. 1970. The resistance of some cultivars and species of Rosa to DiploeQI'pan rasae Wolf causing blackspot disease. Rose Annu. 118-128. Scott-Moncrieff, R. 1936. A biochemical survey of some mendelian factors for flower color. J. Genet. 32:117-169. Semeniuk, P. 1971. Inheritance of recurrent blooming in Rosa wichuraiana. J. Hered. 62:203-204. Semeniuk, P., and T. Arisumi. 1968. Cokhicine-induced tetraploid and cytochimeral roses. Bot. Gaz. 129:190-193. Shahare, M. L., and S. V. S. Shastry. 1963. Meiosis in garden roses. Chromosoma 13:702-710. Shepherd, R E. 1954. History of the rose. Macmillan, New York. Sigurbj6rnsson, B., and A. Micke. 1974. Philosophy and accomplishments of mutation breeding. p. 303-343. In: Polyploidy and induced mutations in plant breeding. IAEA, Vienna. Singh, B., and R S. Malik. 1982. Rose growing for rose oil. Ind. Rose Annu. 2:97-99. Smilansky, Z., N. Dmiel, and N. Zieslin. 1986. Mutagenesis in roses (cv. Mercedes). Environ. Expt. Bot. 26:279-283. Souq, F., P. Coutos-Thevenot, H. Yean, G. Delbard, Y. Maziere, J. P. Barbe, and M. Boulay. 1996. Genetic transformation of roses, two examples: One on morphogenesis, the other on anthocyanin biosynthetic pathway. Acta Hart. 424:381-388. Starn, M., J. N. M. Mol, and J. M. Kooter. 1997. The silence of genes in transgenic plants. Ann. Bot. 79:3-12. Streitberg, H. 1966a. Rosenzikhtung mit hilfe der R6ntgenbestrahlung. Arch. Gartenbau. 14:81-88.

188

S. GUDIN

Streitberg, H. 1966b. Schaffung wirtschaftlich wertvoller sporsvarianten bei rosen durch behandlung mit Rontgenstrahlen. Z. Pflanzenzticht. 55:165-182. Svejda, F. 1977. Breeding for improvement of flowering attributes of winterhardy Rosa kordesii Wulff hybrids. Euphytica 26:703-708. Svejda, F. 1979. Inheritance of winterhardiness in roses. Euphytica 28:309-314. Tackholm, G. 1922. Zytologische studien tiber die Gattung Rosa. Acta Hort. Berg. 7:97-381. Testu, C. 1984. Les roses anciennes. Flammarion, La Maison Rustique, Paris. Teyssier, c., S. Reynders-Aloisi, and Y. Jacob. 1996. Characterisation of a collection of botanical rose trees by phenotypic analysis. Acta Hort. 424:302-308. Torres, A. M., T. Millan, and J. 1. Cubero. 1993. Identifying rose cultivars using random amplified polymorphic DNA. HortScience 28:333-334. Tsujita, J., and T. Blom. 1996. Arching method of rose production. Roses Inc. Bul. April:33-37. Tukey, H. B. 1933. Artificial culture of sweet cherry embryos. J. Heredity 24:7-12. Tweddle, D., A. V. Roberts, and K. C. Short. 1984. In vitro culture ofroses. p. 529-530. In: Nowak, L. Havel, and J. Dolozel (eds.), Plant tissue and cell culture: Application to crop improvement. Czechoslovak Academy of Sciences, Prague. Ueda, Y., and H. Tomita. 1989. Morphometric analysis of pollen exine pattern in roses. J. Japan. Soc. Hort. Sci. 58:211-220. Vainstein, A., and H. Ben-Meir. 1994. DNA fingerprinting analysis of roses. J. Am. Soc. Hort. Sci. 119:1099-1103. Vainstein, A., H. Ben-Meir, A. Zuker, A. A. Watad, G. Scovel, A. Ahroni, and M. Ovadis. 1995. Molecular markers and genetic transformation in the breeding of ornamentals. Acta Hort. 420:65-67. Valles, M., and P. Boxus. 1987. Regeneration from Rosa callus. Acta Hort. 212:691-696. Van de Pol, P. A., H. W. M. Fuchs, and N. F. Van de Peppel. 1988. "Multic" kan spectaculaire produktie verhoging geven. Vakblad v.d. Bolemisterij 24:42-45. Van de Pol, P. A., F. T. Schiet, and J. E. R. Frijters. 1986. A preliminary investigation of odor change in cut and intact roses in various stages of maturation. Acta Hart. 189:215-220. Van den Berg, G. A. 1996. Rose factories, fiction or future? Acta Hort. 424:405-411. Van der Mark, F., J. P. Pijnacker-Hordijk, G. A. 1. Varega, D. P. de Vries, and J. J. M. Dons. 1990. In vivo transformation of clonal Rosa canina rootstocks with Agrobacterium rhizogenes. J. Genet. Breed. 44:263-268. Van der Salm, T. P. M., C. J. G. Van der Toorn, C. H. Hanisch ten Cate, W. M. Van der Krieken, and J. J. M. Dons. 1998. The effects of exogenous auxin and ROL genes on root formation in Rosa hybrida L. "Moneyway". Plant Growth Reg. (In press.) Van Doorn, W. G., G. Groenewegen, P. A. Van de Pol, and C. E. M. Berkholst. 1991. Effects of carbohydrate and water status on flower opening of cut Madelon roses. Postharvest BioI. Technol. 1:45-57. Voisin, R., J. C. Minot, D. Esmenjaud, Y. Jacob, G. Pelloli, and S. Aloisi. 1996. Host suitability of rose rootstocks to the root-knot nematode Meloidogyne hapla using a high inoculum-pressure test. Acta Hort. 424:237-240. Von Malek, B., and T. Debener. 1998. Genetic analysis of resistance to blackspot (Diplocarpon rosae) in tetraploid roses. Theor. Appl. Genet. 96:228-231. Walker, S. 1996. A discussion of the consumer acceptance of genetic transformation in roses. Acta Hort. 424:389-391. Walker, S., Z. Mandegaran, and A. V. Roberts. 1996. Screening roses for resistance to Diplocarp on rosae. Acta Hart. 424:209-214. Walther, F., and A. Sauer. 1986. In vitro mutagenesis in roses. Acta Hort. 189:37-46.

3. ROSE: GENETICS AND BREEDING

189

Wisniewska-Grzeszkiewicz, H., and A. T. Wojdyla. 1996. Evaluation of rose cultivars to fungal diseases susceptibility. Acta Hort. 424:223-236. Wissemann, V., and F. H. Hellwig. 1997. Reproduction and hybridisation in the genus Rosa, section Caninae (Ser.) Rehd. Bot. Acta 110:251-256. Wulff, H. D. 1951. Rosa kordesii, eine neue amphidiploide rose. Ztichter 21:123-132. Wylie, A. P. 1954. The history of garden roses. J. Royal Hort. Soc. 79:555-571. Wylie, A. P. 1955a. The history of garden roses, Part 2. J. Royal Hort. Soc. 79:8-24. Wylie, A. P. 1955b. The history of garden roses, Part 3. J. Royal Hort. Soc. 80:77-87. Yambe, Y., and K. Takeno. 1992. Improvement of rose achene germination by treatment with macerating enzymes. HortScience 27:1018-1020. Yanofsky, M. F. 1995. Floral meristem to floral organs: genes controlling early events in Arabidopsis flower development. Ann. Rev. Plant Physiol., Plant Mol. BioI. 46:167-188. Yokoya, K., S. Walker, and V. Sarasan. 1996. Regeneration ofrose plants from cell and tissue cultures. Acta Hort. 424:333-338. Zieslin, N. 1996. Influence of climatic and socio-economical factors on mode of cultivation and research of rose plants. Acta Hort. 424:21-22. Zieslin, N., H. Gavish, and M. Ziv. 1987. Growth interactions between calli and explants of rose plants in vitro. Plant Sci. 49:57-62. Zieslin, N., and R. Moe. 1985. Rosa. p. 214-225. In : A. H. Halevy (ed.J, Handbook offlowering, vol. 4 . CRC Press, Boca Raton, FL.

4 White Clover Molecular Genetics Derek W. R. White, Derek R. Woodfield, and Brigitta Dudas New Zealand Pastoral Agriculture Research Institute, Private Bag 1100B, Palmerston North, New Zealand Richard 1. S. Forster, and David 1. Beck .The Horticulture and Food Research Institute of New Zealand, Private Bag 92169, Auckland, New Zealand I. Introduction II. Transgenic Approaches to White Clover Improvement A. Transgenic Plants 1. Tissue Culture 2. Agrobacterium-mediated Transformation 3. Transgene Expression 4. Transgene Inheritance B. Transgenic Breeding Strategies 1. Modified Synthetics with Progeny Testing 2. Backcrossing into Cultivars 3. Direct Transformation of Inbred Lines C. Genetic Engineering for Resistance to Viral Diseases 1. Coat Protein-mediated Resistance 2. Resistance Using Mutated Viral Genes 3. Gene Silencing and Virus Resistance D. Transgenic Plants for Pest Resistance E. Potential for Improved Nutritive Value III. Marker-Assisted Selection A. Map-Based Selection B. Genetic Diversity and Parental Selection IV. Conclusions Literature Cited

Plant Breeding Reviews, Volume 17, Edited by Jules Janick ISBN 0-471-33373-5 © 2000 John Wiley & Sons, Inc. 191

192

D. WHITE, D. WOODFIELD, B. DUDAS, R. FORSTER, AND D. BECK

I. INTRODUCTION

White clover (Trifolium repens 1.), a perennial legume species, is an important component of many temperate grazed animal production systems. The species is adapted to regions from the Arctic Circle to the subtropics, and from sea-level to altitudes of 6000 m in the Himalaya mountains (Williams 1987a). White clover's indigenous range covers all of Europe, western central Asia and areas of North Africa, and it has become naturalized in China, Korea, Japan, Australasia, North, Central and South America (Gillet 1985; Taylor 1985; Williams 1987a). As a crop, white clover is grown in association with temperate grasses to provide a low-cost, high-protein feed source for grazing livestock. White clover is particularly valued for its capacity to fix atmospheric nitrogen in grazed swards (Boller and Nosberger 1987; Crush 1987), and much of this nitrogen becomes available to support the growth of the companion grasses. White clover has higher nutritive value than the associated grasses with which it is grown, due to lower levels of structural carbohydrate (Ulyatt 1981), higher levels of crude protein (Minson 1990), and a faster rate of passage through the rumen of grazing animals (Rattray and Joyce 1974). In addition, the voluntary daily intake rates (kg dry matter/animal) of white clover by grazing ruminants are higher than for grasses due to a lower length-to-width ratio of fibers in the foliage and smaller quantities of cell wall components (Rogers et al. 1982; Minson 1990). The seasonal growth patterns of white clover and associated forage grasses are complementary, such that when grown together they provide a more consistent feed supply than if grown separately. Growth of white clover occurs later in summer and fall when grass production is reduced. The stoloniferous growth form and phenotypic plasticity of white clover also make it a preferred companion legume in grass swards and enable it to withstand severe defoliation. White clover (2n=4x=32) is predominantly an obligate outcrossing species with disomic inheritance and has a gametophytic se1£incompatibility system based on multiple alleles at the S locus (Williams 1987b). The outcrossing nature of white clover means that natural populations are a heterogeneous mixture of highly heterozygous individuals. This has resulted in high levels of genetic variation both within and between populations, and it is this genetic variability that has been critical to the success of white clover over such a wide environmental range. White clover can hybridize with three species: Trifolium uniflorum 1. (2n=4x=32), T. nigrescens Vivo (2n=2x=16) and T. occidentale Coombe (2n=4x=32), and with greater difficulty with at least three other Trifolium species (T. ambiguum, Bieb T. hybridum, 1. and T. isthmo-

4. WHITE CLOVER MOLECULAR GENETICS

193

carpum Brot.), this later group requiring the use of embryo rescue techniques (Williams et al. 1978). Genetic improvement in white clover yield and pasture content has averaged 0.6% per annum for the past 60 years of conventional breeding (Woodfield and Caradus 1994), which is high relative to the rates reported for other forage species (van Wijk and Reheul 1991; Holland and Bingham 1994), but lower than rates reported for grain crops (Austin et al. 1980; Russell 1991). The main objective in most white clover breeding programs has been to increase production from grazing animals. Clover content in the pasture, sward yield, persistence, and forage quality are all selection criteria. Other breeding goals include resistance to pathogens and insect herbivores, tolerance of low soil phosphorus levels, and eliminating anti-quality factors such as estrogen compounds which may reduce animal fertility. Despite the improvements achieved by traditional methods, many important breeding objectives have been elusive. Only limited progress has been made in the development of germplasm resistant to viral diseases, tolerant to insect damage, and with significant improvement in nutritional value. Molecular genetics has provided the means to create unique variation that is not present within the natural white clover gene pool, and to overcome difficulties of low heritability or the absence of a suitable screening technique. This review describes techniques available for the molecular breeding of white clover and details progress made to date towards applications intended to overcome some of its agricultural deficiencies. The range of possibilities is deliberately restricted to the areas of interest to the authors. A wider range of prospects for the genetic engineering of white clover are described by White (1997).

II. TRANSGENIC APPROACHES TO WHITE CLOVER IMPROVEMENT

A. Transgenic Plants

Recent advances in transformation provide opportunities to introduce cloned genes, of diverse origin, into white clover, and to create novel traits through the use of gene modification. Essential components of this process are: (1) tissue culture protocols for the in vitro regeneration of plants from differentiated tissues, (2) a nlethod for efficient transformation with cloned genes, and (3) appropriate expression of the introduced gene(s), and stable inheritance of the transgene, such that it can be used in a breeding program.

194

D. WHITE, D. WOODFIELD, B. DUDAS, R. FORSTER, AND D. BECK

1. Tissue Culture. As with many legumes, initial difficulties in obtaining regeneration of plants from cell or tissue cultures delayed the development of methods for the transformation of white clover. Despite attempts that used media containing a wide range of phytohormone combinations (auxins and cytokinins), only sporadic shoot formation from callus cultures was obtained (Oswald et al. 1977; Gresshoff 1980; Mohapatra and Gresshoff 1982). In white clover there is a substantial effect of plant genotype on the capacity to regenerate from callus or suspension cultures. This difficulty was at first avoided by selecting genotypes with a high regeneration capacity (White 1984; Bond and Webb 1989; Yamada 1989). However, only about 1 % of white clover genotypes regenerated reliably from callus or suspension culture, and some of these plants had agronomic deficiencies (D. W. R. White, unpubl.). Another difficulty experienced with this approach was the long intervals in tissue culture prior to plant regeneration. Ideally, plant regeneration should extend to a wide range of elite germplasm so that transgenes can be incorporated directly into genetic material with a proven agronomic performance. If possible, regeneration should be direct, via shoot organogenesis or somatic embryogenesis without an intervening unorganized callus growth stage, in order to avoid tissue culture induced alterations in genotype (Phillips et al. 1994), and, for the purposes of selecting and characterizing genetic transformants, a single cell origin for regenerated plants is preferable. Direct regeneration from the explant to form shoots or somatic embryos also has the potential to reduce the interval in tissue culture and hence make the process of transformation more efficient. One approach has been to identify explants with a high capacity for direct regeneration. Protocols have been developed for direct regeneration by somatic embryogenesis from the hypocotyls or cotyledons of immature zygotic embryos of white clover (Maheswaran and Williams 1985; Parrott 1991). However, both the number of somatic embryos formed and the subsequent plant regeneration frequencies obtained with these explants were relatively low. A protocol described by Bond and Webb (1989) for direct shoot organogenesis from stolon internode segments without an intervening callus stage still relied on the use of a highly regenerative genotype previously selected from cv. Grasslands Huia. To overcome these difficulties, White and Voisey (1994) developed a method for direct shoot organogenesis and rapid plant regeneration from cotyledon explants taken from germinating white clover seedlings. The cotyledon stalks of three-day-old germinated seedlings have proved to be a particularly responsive tissue for plant regeneration. Although there was some indication of a residual genotype influence on regenera-

4. WHITE CLOVER MOLECULAR GENETICS

195

tion frequency, 88 to 100% of cotyledon explants from seedlings, of eight different cultivars, produced shoots on a medium containing a combination of a-napthaleneacetic acid (NAA) and 6-benzylaminopurine (BA). Regeneration was generally prolific, with an average of 16 to 23 shoots per cotyledon, and regenerated plants could be transferred to soil within 6 to 8 weeks after cultures were initiated. Regenerated plants were similar to their cultivar of origin in both morphology and growth rates, and flowered and set seed when intercrossed. Studies using scanning electron microscopy and histological sectioning established that shoots formed from cotyledons originated from single cells (White and Voisey 1994). After 2 to 3 days of culture, individual epidermal cells, predominantly on the adaxial surface of the cotyledon stalk proximal to the site of explant excision, began dividing to form de novo shoot meristems without an intervening callus phase. These shoots develop into rooted plantlets within 4 to 5 weeks of culture. It may be possible to increase the frequency of direct plant regeneration from seedling cotyledons by selective breeding for this trait. However, any attempt to do so needs "to be related to the intended transgenic breeding strategy (see below). If an outcrossing strategy is to be used, then prior selection of genotype combinations that produce highly regenerative progeny might limit the types of breeding material available for transformation. Conversely, if an inbreeding strategy was to be used for the development of transgenic lines, then selection for tissue culture regeneration capacity may be required. 2. Agrobacterium-mediated Transformation. Agrobacterium-mediated transformation is the established method for introducing cloned genes into white clover. The first protocol for white clover transformation, described by White and Greenwood (1987), used stolon (stem) internode segments of a highly regenerable genotype (White 1984) as the target tissue. Following co-cultivation with Agrobacterium (strain LBA4404) containing a binary vector incorporating a gene conferring antibiotic resistance (neomycin phosphotransferase II, nptIl) , transgenic callus lines were selected on kanamycin-containing medium. While this process was relatively efficient for the selection of transgenic cell lines, prolonged culture periods of 8 to 12 months were required before regenerated transgenic plants could be established in soil. Also, a portion of the transgenic plants displayed abnormalities in growth or development. However, the method was successfully used to introduce genes encoding the following proteins into white clover: a sulphur amino acid-rich seed storage protein (Ealing et al. 1994), a Bacillus thuringiensis (B t) 8-endotoxin active against an insect pest of white clover in New

196

D. WHITE, D. WOODFIELD, B. DUDAS, R. FORSTER, AND D. BECK

Zealand, Wiseana eervinata (Walker) (Voisey et al. 1994), and a viral coat protein (Dudas et al. 1993). Subsequently, the method for directly regenerating plants from the cotyledons of three-day-old germinating seedlings (White and Voisey 1994) was adapted for Agrobaeterium-mediated transformation (Voisey et al. 1994). Using this method, we have produced transgenic white clover plants from a wide range of cultivars: Grasslands Huia, Grasslands Tahora, Grasslands Kopu, California Ladino, Regal Ladino, Milka, and Haifa. When T-DNA vectors containing a 35S Cauliflower mosaic virus (CaMV) promoter:nptII gene fusion as a kanamycin resistance selectable marker were used, transformation frequencies ranged from below 1% up to about 20%. Recently Larkin et al. (1997) described a protocol which used the cotyledons of imbibed mature seeds as the target tissue, together with vectors containing a CaMV 35S promoter:bar gene fusion, and phosphinothricin (PPT) herbicide resistance selection, for the Agrobaeterium-mediated transformation of white clover. With this method, transgenic plants were recovered from over 50% of treated cotyledons. However, when a binary vector containing a nopaline synthase promoter:nptII gene fusion was used, together with kanamycin selection, the frequency of transformation achieved was an order of magnitude lower than with PPT. Hence the type of selectable marker used has a substantial effect on the transformation frequency of white clover. The direct shoot formation, Agrobaeterium-mediated method oftransformation produces plants that can be transferred to soil within 10 to 12 weeks of the Agrobaeterium co-cultivation. 3. Transgene Expression. As yet there is only limited information about transgene expression in white clover plants. In most cases the objective has been to obtain constitutive transgene expression, and for this purpose flanking CaMV 35S promoter and oes or nos 3' termination sequences have been used. The pattern of expression of a ~-glucuronidase (uidA) gene, flanked by CaMV35S promoter, and oes 3' termination sequences, was examined in transgenic white clover plants using histochemical methods (Voisey et al. 1994). GUS activity was detected in most tissues examined: leaves, stolons, nodes, petioles, trichomes, and roots, but not in mature pollen. The specific activity of ~-glucuronidase in the leaves of three primary transgenic white clover plants varied over an eight-fold range, indicating considerable variation in transgene expression between different transformants. Abundant steady state levels of transgene specific mRNA were also detected in the leaves of white clover plants transformed with a CaMV35S promoter-pea albumin

4. WHITE CLOVER MOLECULAR GENETICS

197

1 gene fusion, although only low levels of transgene encoded protein were detected (Ealing et al. 1994). So far the expression patterns of only three other promoters have been analyzed in transgenic white clover plants. Larkin et al. (1997) transformed white clover with a uidA reporter gene fused to the auxinresponsive promoter of the soybean GH3 gene (Hagen et al. 1991), and examined transgenic plants for the pattern of expression, response to exogenous auxin, and gravistimulation. High levels of expression were detected in the leaves, roots and particularly the stolons of untreated plants. This differs from the low levels of activity obtained in untreated transgenic tobacco (Hagen et al. 1991). In transgenic white clover (as in transgenic tobacco) GH3:GUS activity increased with the addition of NAA, but not in response to other types of phytohormones. Localized expression was detected in the outer cortex adjacent to where lateral roots formed and on the non-elongating side of roots subjected to gravistimulation. Pittock et al. (1997) transformed both tobacco and white clover with a tobacco basic chitinase promoter:uidA fusion and determined that in the two species the transgene had similar environmental and developmental patterns of expression. It has been proposed that the chitinases exist in plants as a defense against chitin-walled pathogens (Schmidt et al. 1994), however, the tobacco basic chitinase gene is also under developmental control (Neale et al. 1990). During root morphogenesis, the transgene was expressed in primary root meristems and in lateral root meristems, but only after they emerged from the primary root epidermis. There was no expression in developing nodules on transgenic white clover, but there was transient expression among inner cortex cells following induction with a nodulation-competent Rhizobium strain. These differences in nodule and lateral root expression were taken as evidence that these organs may differ in the mechanism used for the initiation of cell division. As expected, wounding (mechanical or due to aphid feeding) resulted in localized expression of the tobacco basic chitinase-uidA transgene in the leaves of transgenic white clover and tobacco. The expression pattern of the promoter of the tobacco rootspecific, RB7 aquaporin gene (Yamamoto et al. 1991) has been examined in transgenic white clover (D. W. R. White, unpubl.). Although the RB7 promoter:uidA gene fusion gave a predominantly root-specific pattern of expression, some GUS activity was also detected in stolon vascular and parenchyma tissues adjacent to nodes undergoing root meristem formation. These examples illustrate that promoter expression patterns in transgenic white clover may differ from those obtained in transgenic tobacco. In order to increase the options for transgene expression in

198

D. WHITE, D. WOODFIELD, B. DUDAS, R. FORSTER, AND D. BECK

white clover, there is a need to test the function of further heterologous promoter sequences and to identify promoters from endogenous tissuespecific genes. 4. Transgene Inheritance. A high proportion of transgenic white clover plants have single copy T-DNA inserts. Of over 150 transgenic white clover plants analyzed for T-DNA copy number by Southern hybridization, 66% had single copy insertions (D. W. R. White, unpubl.). Similarly, seven of twelve transgenic plants analyzed by Larkin et al. (1997) had a single transgene. Results reported by Ealing et al. (1994) and Voisey et al. (1994) demonstrated that when transgenic plants containing a single T-DNA were outcrossed with non-transgenic white clover plants, progeny segregated 1: 1 for the presence and absence of the transgene. Inheritance of a CaMV35S promoter:uidA transgene as a single dominant locus was obtained through four generations of outcrossing to non-transgenic plants, and a breeding strategy (modified synthetics with progeny testing, see below) to obtain transgene expression in all members of an outcrossing population has been successfully evaluated (Scott, Woodfield and White, 1998). These results indicate that a transgenic plant containing a beneficial trait conferred by a single transgene insert could be used as the basis of a cultivar breeding program. B. Transgenic Breeding Strategies

The development of transgenic technologies has not eliminated the need to use conventional plant breeding methods (Bingham 1983). There is still a requirement to (1) screen a population of primary transformants to identify those with appropriate expression levels of the inserted gene(s), (2) examine the progeny of these initial transformed plants for stable inheritance and expression of the transgene(s), and (3) identify genetic backgrounds that enhance the function of the transgene(s) and incorporate important agronomic traits such as forage yield, persistence, and seed production. As in conventional plant breeding programs, these steps must be accomplished before parent plants can be selected to develop a cultivar. However, the process is considerably aided for transgenic plants by the precision that comes from molecular analysis of the introduced gene. For the breeding of transgenic cultivars, it is preferable to identify primary transgenic plants that have a single, intact, T-DNA insertion, because the presence of multiple T-DNA loci or repeat T-DNA copies at

4. WHITE CLOVER MOLECULAR GENETICS

199

a single locus, in transgenic plants, can lead to transgene silencing (Flavell 1994). The presence of a single T-DNA insertion should also make subsequent genetic analysis less complex, since only null (0), hemizygous (1), and homozygous (2) T-DNA loci need to be considered. Development of transgenic cultivars in cross-pollinating species such as white clover is more difficult than in species where clonal, inbred, or hybrid cultivar development methods are used. Generally, it would be useful to obtain expression of the foreign gene(s) in all progeny of a transgenic cultivar, and this can be accomplished by producing parents that are homozygous for the transgene(s). However, the traditional approach to white clover breeding is to identify elite genotypes that have good general combining ability and then to intercross these plants to accumulate favorable genes while maintaining high levels of heterozygosity (Williams 1987b). Therefore, there is a need to develop alternative breeding strategies for outcrossing plants that result in uniform expression of transgenes without causing inbreeding depression (Conner and Christey 1994). These strategies should also attempt to minimize the number of generations of intermating required for the development of a variety. Two alternative methods were developed by Woodfield and White (1996) to obtain homogeneous transgene expression in synthetic cultivars: a modified synthetic breeding strategy with progeny testing, and backcrossing into existing cultivars (with or without progeny testing). A longer-term option would involve the direct transformation of inbred lines and the subsequent deployment of transgenes in F 1 hybrid cultivars. These breeding strategies are summarized below. 1. Modified Synthetics with Progeny Testing. The modified synthetic approach involves introgression of a single transgene into a random mating population. In this breeding strategy, a primary transgenic plant containing a single T-DNA insertion would be outcrossed to at least 16 non-transgenic plants from the target population. The SOCYo of the F 1 progenies containing a single copy of the transgene would then be intermated to provide F z populations segregating for 0 (null), 1 (hemizygous), or 2 (homozygous) copies of the T-DNA allele. The next step is to identify those plants homozygous for the transgene. Progeny testing is required at this stage to identify homozygous individuals, ifno clear phenotypic distinction between heterozygous and homozygous genotypes exists. Transgenic F z plants would be test-crossed to a non-transgenic plant, and progeny assessed for presence of the transgene. All test-cross progeny from homozygous F z transgenic plants should contain the transgene. At least 11 plants per F z family need to be screened to have a 9S o/0

200

D. WHITE, D. WOODFIELD, B. DUDAS, R. FORSTER, AND D. BECK

probability of identifying one homozygous plant (Sedcole 1977). If simple phenotypic screening is possible, then this stage can be accomplished rapidly and at low cost. However, if the transgenic trait can only be identified by detailed molecular or biological analysis, this stage of the breeding strategy may be costly. The confirmed homozygotes would then be intercrossed to produce a synthetic cultivar that should be homogeneous for two copies of a single transgene. This modified synthetics method of transgenic plant breeding controls the dosage of the transgene (and the linkage group in which it is located) but maintains a high level of heterozygosity in the remainder of the genome. This is likely to give more uniform expression of the transgene in the population than a standard breeding approach, but less uniform expression than in clonal, inbred, or hybrid crops where the genetic background is more constant. There is some potential for inbreeding depression to occur because the linkage group in which the T-DNA is inserted will remain homozygous. Monitoring for the presence of deleterious genes (possibly recessive) linked to the insert therefore becomes essential. This transgenic breeding strategy is simple and complements existing breeding practices in cross-pollinated crops, but it slows cultivar development because a progeny testing step is required. However, progeny testing for agronomic performance could also be conducted concurrently with the transgenic progeny testing to capitalize on the delay. 2. Backcrossing into Cultivars. Backcrossing can transfer specific gene(s) into an otherwise elite genetic background. In transgenic breeding, the donor or non-recurrent parent plant containing a single T-DNA insert is crossed with an elite population of non-transgenic plants (recurrent parents), followed by several generations of backcrossing to the recurrent parents. At least five generations of backcrossing are required to regain approximately 97% of the recurrent parent's genetic background. In cross-pollinated crops, it is essential to backcross to different genotypes of the recurrent parent population in each successive generation to avoid inbreeding depression. Backcrossing would maintain the transgene in a heterozygous form. Therefore, the transgene will be present in only 50% of the progeny in each generation and identification of the transgenic progeny will be required at each step for subsequent backcrossing. In the final step, transgenic plants (all T-DNA insert heterozygotes) would be intercrossed to produce a population segregating for 0 to 2 copies of the transgene locus. Progeny testing, the identification of transgene homozygotes, as in the modified synthetic approach above, and their intermating, would be required to develop a backcross-derived population with all individuals homozygous for the transgene.

4. WHITE CLOVER MOLECULAR GENETICS

201

The main advantage of this backcrossing strategy is that a single transgene could be deployed into multiple genetic backgrounds, distinctly different from the primary transformant. The major disadvantage would be the additional 5 to 6 generations of backerossing and progeny screening required. 3. Direct Transformation of Inbred Lines. Both severe inbreeding depression and some hybrid vigor have been reported in white clover (Williams 1987b; Michaelson-Yeates et al. 1997), however, the identification of self-compatible genotypes in white clover (Yamada et al. 1989) opens up the prospect of hybrid breeding approaches for the development oftransgenic cultivars. Hybrid cultivars have the foll~wing potential advantages over synthetic methods for transgene deployment: (1) fewer generations of multiplication would be required, (2) there is optimal use of non-additive gene action, and (3) maximum uniformity of the population and transgenic trait should occur. Furthermore, the higher yield and greater environmental stability of hybrids should provide improvements over traditional synthetic varieties (Stuber 1994). Pollination control mechanisms are required for the development of hybrid varieties. In the absence of a genetic male sterility system, the gametophytic self-incompatibility of white clover is the most practical mechanism for pollination control. Breeding schemes for producing hybrids using self-incompatibility have been described by Taylor (1985), Williams (l987b), and Yamada et al. (1989). Transformation of partial inbreds (S2 or S:{) would allow transgenic trait stabilization during the subsequent inbreeding to S5 or Su, Inbreds with optimal transgene action could be combined during hybrid production to obtain multiple transgenic traits in the final cultivar. C. Genetic Engineering for Resistance to Viral Diseases

Viruses have significant adverse effects on white clover yields (Pratt 1967; Barnett and Gibson 1975; Harville and Derrick 1978; Forster and Morris-Krsinich 1985; McLaughlin et al. 1992; Ralph 1992; Helms et al. 1993). Furthermore, infection by viruses can cause dramatic reductions in shoot and root growth, nodulation, nitrogen fixation, soluble protein and crude fiber content, seed yields and seed germination rates, and can increase susceptibility to attack by other pathogens. Infected white clover plants can also provide reservoirs of viruses that can infect other crops. Among these viruses, white clover mosaic potexvirus (WCIMV) is the most widespread, worldwide pathogen of white clover. In New Zealand

202

it has

D. WHITE, D. WOODFIELD, B. DUDAS, R. FORSTER, AND D. BECK

also been recorded in red clover, subterranean clover, and in other species such as broad bean and sweet pea (Fry 1959). In white clover, the economic inlportance of WCIMV in New Zealand pastures was first recognized in the 1950s (Fry 1959). In the study by Fry (1959), the virus was found in all districts surveyed, usually in more than 50% of white clover plants in mature pastures. Furthermore, the virus was shown to depress growth of white clover in pastures by 24 %, and to depress growth of subterranean clover and red clover by 36% and 55%, respectively. In a number of other studies in New Zealand, Australia, and North America, WCIMV infection of white clover has been shown to decrease dry matter production by up to 56%, nodulation and nitrogenase activity by up to 70%, and seed yields by up to 90% (Fry 1959; Tapio 1970; Barnett and Gibson 1977; Guy et al. 1980; Scott 1982; Forster et al. 1984; Garrett 1990,1991; Potter 1993). In a recent study of the incidence ofWCIMV in New Zealand pastures, white clover samples were collected from 14 young pastures « 4 years old) and from 46 mature pastures (> 4 years old) in the North Island (Dudas et al. 1998). Our survey results confirm earlier studies (Fry 1959; Forster et al. 1983) that WCIMV is widely established. In the younger pastures, incidence ranged from 3% to 48%, and the average incidence was 21 %. In the mature pastures, incidence ranged from 20% to 95%, and the average incidence was 69%. The virus was also found in samples collected from the roadside in the vicinity of eight of the sampled pastures. Virus incidence ranged from 29% to 69% and the average incidence was 52%. A study of the effects of WCIMV on the growth of a modern white clover cultivar, Grasslands Tahora, found that virus infected plants produced 36.5% less dry matter than virus-free plants. Our results show that under current agricultural practices, this virus has the potential to exert demonstrable negative effects on productivity of New Zealand pastures (Dudas et al. 1998). In our studies with WCIMV, the proportion of infected plants that produced foliar symptoms in both naturally infected and inoculated plants was always low. This emphasizes the unreliability of disease diagnosis by visual inspection and shows that yield decreases are associated with, for the main part, symptomless infection. The lack of correlation between symptoms and yield loss indicates that the strategy proposed by Martin et al. (1990) to breed for lack of WClMV symptoms ("tolerance") in red clover is inappropriate for control of WClMV in white clover. Furthermore, breeding for reduced virus titre may also be inappropriate; Scott (1982) showed a lack of correlation between WCIMV virus titre and yield loss in both red and white clover.

4. WHITE CLOVER MOLECULAR GENETICS

203

Viruses cannot be eradicated readily using chemicals, and the use of chemical pesticides to control insect, fungal, or nematode vectors of viruses in forage legumes is economically unviable. Traditional methods of control have involved, wittingly or unwittingly, selection for tolerance or breeding for resistance. Sources of tolerance or resistance to viruses have been identified in several forage species, including Trifolium repens L., T. pratense L., T. subterraneum L., and Medicago sativa L. (Crill et al. 1970a,b; Barnett and Gibson 1975; Taylor et al. 1986; Gibson et al. 1989; Martin et al. 1990; Pesic-Van Esbroeck and Hiruki 1990; McLaughlin and Fairbrother 1993). Resistance has also been identified in a number of other Trifolium species, such as T. ambiguum Bieb. (Barnett and Gibson 1975; Alconero 1983a,b; Pemberton et al. 1989; Anderson et al. 1991a). However, the genetic basis for the resistance in these other species has not been determined and the resistance is not always readily transferable to other forage species. Embryo rescue technology has been used to develop interspecific hybrids between T. repens L. and T. ambiguum Bieb., and between other Trifolium species (Williams and Verry 1981; Pederson and McLaughlin 1989; Anderson et al. 1991a). Although the virus resistance identified in T. ambiguum Bieb. was apparently inherited in F z progeny from these hybrids (Pederson and McLaughlin 1989), the prospects for introgressing this form of resistance into white clover may be doubtful (Anderson et al. 1991b). Current difficulties include low fertility of hybrids and in bridging crosses (Hussain and Williams 1997) and problems in obtaining recombination between the T. repens L. and T. ambiguum Bieb. genomes. By contrast, molecular genetics offers new opportunities for the control of forage legume virus diseases through the ability to overcome species-specific barriers, the ability to develop multigene resistance, avoidance of linkage drag, and the ability to manipulate levels and sites of expression. Examples of resistance to representatives of most groups of plant viruses have been reported using transgenes of viral or non-viral origin (Grumet 1994). Virus-resistant transgenic plants have also been tested successfully under field conditions. The most successful and widely used application of this technology is based on the concept that expression of genes from a pathogen's own genome in a host plant can be used to interrupt the infection cycle (Sanford and Johnston 1985). Since 1986, resistance has been reported to a wide range of viruses in plants engineered to express a number of plant virus genes or RNA sequences. The examples presented here will concentrate on results obtained in our laboratory using WCIMV, transgenic Nicotiana benthamiana L. (a model system), and transgenic T. repens L.

D. WHITE, D. WOODFIELD, B. DUDAS, R. FORSTER, AND D. BECK

204

6.0 kb

m7GPPPG--jt-

14_7_k_D_a

'---_~

Fig. 4.1. Genome organization of white clover mosaic potexvirus (WCIMV). The five open reading frames encode for proteins required for: replication (147 kDa), cell-to-cell and systemic movement (triple gene block proteins, 26 kDa, 13 kDa, and 7 kDa), and the coat protein.

The WCIMV virus (Fig. 4.1) has a plus (messenger) sense RNA genome of approximately 6.0 kb, a 5' cap structure, a 3' tract of poly [AJ, and five open reading frames (indicated by the size of the protein product they encode). These open reading frames code for proteins that are required for replication (147 kDa) and cell-to-cell and systemic movement [26 kDa, 13 kDa, and 7 kDa open reading frames (which comprise the triple gene block)], plus the coat protein (Forster et al. 1988, 1992; Beck et al. 1990,1991). 1. Coat Protein-mediated Resistance. CP-mediated resistance is the most widely reported form of genetically engineered resistance against plant viral infection (Fitchen and Beachy 1993). This form of resistance is achieved by incorporating a viral gene sequence, capable of expressing the CP of the virus, into the genome of a transgenic plant. Resistance is conferred against the virus from which the CP gene was derived (homologous resistance), or to other related viruses (heterologous resistance) (Beachy et al. 1990). The first report of increased resistance to viral infection in transgenic plants expressing a virus CP was by Powell-Abel et al. (1986). Nicotiana tabacum 1. plants were transformed with a chimeric gene encoding the tobacco mosaic virus (TMV) CP coding region. Transgenic plants expressing the TMV CP were demonstrated to be resistant, or exhibited delayed symptom development, when inoculated with relatively low levels ofTMV inoculum (0.4-2.0 /lg ml- I ). Plants accumulating relatively high levels (>0.01 % total soluble protein) of TMV CP were, in general, more resistant than those transgenic plants that expressed the transgene at low levels (Powell et al. 1990). Plants expressing a modified version of the CP gene in which the initiation codon was disrupted were not resistant, indicating that resistance was conferred by the CP rather than the RNA (Powell et al. 1990).

4. WHITE CLOVER MOLECULAR GENETICS

205

Since the first report in 1986, CP-mediated resistance has been demonstrated for viruses from most plant virus groups. The mechanism(s) of action by which the CP transgenes interfere with the infection process are not well understood. Several different stages of the viral infection process appear to be affected by the expression of the viral CP in transgenic plants. These include interference in uncoating, replication, and spread (Register and Beachy, 1988; Osbourn et al. 1989; Wisniewski et al. 1990).

The success of the CP gene of other viruses to provide virus protection suggested that the same strategy would be useful for engineering resistance to WCIMV. A total of 15 N. benthamiana L. transgenic plant lines designed to express the WCIMV CP were produced following Agrobacterium-mediated transformation. These transgenic lines were morphologically indistinguishable from non-transgenic control lines. The levels of CP expression for the transgenic lines varied from 0.005-0.06% of total SDS soluble protein. The transgenic lines were shown to be resistant to infection by both virions and viral RNA. The most resistant lines were protected from systemic infection at high (250 llg virions ml- 1 ) inoculum challenge, with less than 10% of the inoculated plants becoming systemically infected. Less resistant lines showed only a delay in development of systemic infection (Beck et al. 1993). Results for transgenic Nicotiana benthamiana L. plants engineered to express the WCIMV CP indicated that the major effect of the WCIMV CP transgene on virus infection was a block on long-distance movement of the virus. The successful demonstration that CP-mediated resistance against WCIMV infection worked using a model systenl (N. benthamiana L.) (Beck et al. 1993) led to the transformation of T. repens L. cv. Tahora with the WCIMV CP gene. Although the transformation frequency was lower than usual (4/900; 0.44%), 4 plant lines were eventually established. These plants were shown by Southern blot analysis to contain three copies of the CP gene. Expression of the WCIMV CP in these transgenic T. repens L. lines was confirmed by northern blot, DAS-ELISA and western blot analysis. Expression of the transgene varied from a low of less than 5 ng CP mg- 1 total soluble protein «0.0005%), to 50 ng mg- 1 total soluble protein (0.005%). Fig. 4.2 shows the results of inoculation of two "high" expressing plant lines with two levels of virus inoculum. While some resistance was shown for both levels of inoculum, the major effect of the transgene was to delay systemic infection. The level of expression of the WCIMV CP in the "high" expressing T. repens L. lines was approximately that of the lowest expressing N. benthamiana L. lines, which similarly showed a delay to systemic infection by WCIMV.

N

o

5.0 J.lg/ml WCIMV

0.05 IJg/ml WCIMV

O"J

+ __+ .-

~ en C ro

0.. "0

•

/

I / /

60

/

o

I

/

a

/ I

I I

I

I

)

I I I

20

I I

I I

/

80

I I

I

60

I I

I I

I

/

40

I I

I

I

I I

~ /

/

rI

"

I

I

I

40

o

..-:fY"

~/

cr'" I

?

100

..-,..... ..0

/

/

/

I

t5

.£:

~~

80

Q)

~

./0

+ --- + --+

+

100

/

o

20

I

II

I I

I

o

9

20

30

36

48

50

o

9

20

Control - 0 -

WCP5

36

48

50

o

dpi

dpi - +-

30

0-

WCP7

+-

Control - 0-

WCP5

-0

WCP7

Fig. 4.2. Influence of WCIMV-coat protein transgene expression on resistance to WClMV. Transgenic white clover plants expressing a WClMV-CP gene (WCP5 and WCP7) were delayed in systemic infection (days post inoculation), when inoculated with either low (0.05 Ill/ml) or high (5.0 lll/ml) levels ofWCIMV.

4. WHITE CLOVER MOLECULAR GENETICS

207

Once a T. repens L. line is identified with a level of expression of the WCIMV CP in excess of 0.05% total soluble protein, that line is expected to be highly resistant to systemic infection. 2. Resistance Using Mutated Viral Genes. The hypothesis that a mutated protein may be able to compete with and disrupt the function of a wild type protein has been termed "dominant negative" (Herskowitz 1987). Work performed in our laboratory provided a demonstration of the relatively broad resistance that can be achieved using resistance genes derived from mutated viral movement genes (Beck et al. 1994). The genome of the potexvirus, carlavirus, furovirus, and hordeivirus groups (Morozov et al. 1987; Forster et al. 1988), as well as in those of several unclassified plant viruses (Kanyuka et al. 1992; Sumi et al. 1993) contain a set of three genes known as the triple gene block. These three genes coordinate along with the coat protein, cell-to-cell and long distance transport of viral RNA or particles (Lough et al. 1998). In all viruses that contain a triple gene block, the central open reading frame encodes a small protein (12-14 kDa) with conserved structural motifs (two hydrophobic domains separated by a hydrophilic domain containing conserved amino acids) (Morozov et al. 1989). The WCIMV 13 kDa gene was mutated to encode a protein product (13*) with six of the conserved hydrophilic amino acids in the central hydrophilic domain replaced by hydrophobic amino acids (Beck et al. 1994). These mutations were designed not to disrupt the two hydrophobic domains but were predicted to have an effect on the conserved hydrophilic domain. Transgenic N. benthamiana L. plants were produced by Agrobacterium-mediated transformation using a binary plant transformation vector containing the mutated WCIMV 13 kDa gene (13*) gene. These plants were selfed and seed (R 1 generation) were collected. Seedlings from 10 lines of transgenic plants designed to express the mutated 13 kDa movement protein were shown to be resistant to systemic infection by the homologous virus (WCIMV strain 0), whereas all plants from controllines became systemically infected. Three 13 * transgenic lines were further analyzed and shown to be resistant to systemic infection when challenged by inoculation with three different WCIMV strains (0, M, or J) or with WCIMV strain 0 RNA. In addition, some plants of the three 13* transgenic plant lines were found to be resistant to systemic infection with two other members of the potexvirus group, potato virus X (PVX) and narcissus mosaic potexvirus (NMV), and the carlavirus potato virus S (PVS), but not to TMV (which does not encode triple gene blocklike proteins).

208

D. WHITE, D. WOODFIELD, B. DUDAS, R. FORSTER, AND D. BECK

Sepparen et al. (1997) mutated a similar region of the PUX 12K gene

and tested this mutated gene as a potential broad range resistance gene

in transgenic N. tabacurn plants. These plants were resistant to two potexviruses and two carlaviruses. Transgenic white clover plants expressing the 13 * resistance gene have been produced. These clover plants demonstrate resistance against WCIMV (Dudas et al., unpublished). Further studies will determine if the WCIMV resistant white clover plants are also resistant to other viruses expressing triple gene block proteins. The phenomenon of posttranscriptional gene silencing (reviewed by Grant 1999; Wassenegger and Pelissier 1998) as a mechanism to explain transgene-induced virus resistance in plants was first proposed by Dougherty and co-workers (Lindbo and Dougherty 1992a,b; Lindbo et al. 1993a,b; Smith et al. 1994). These workers concluded that the cellular mechanism(s) leading to sequence specific degradation of the transgene mRNA are also involved in conferring virus resistance. The mechanism proposed by Dougherty and others to explain virus resistance in transgenic plants is consistent with observations that our laboratory has made using transgenic N. bentharniana plant lines designed to express the WClMVreplicase gene (Beck et al. 1996). Twenty independent R1 generation plant lines were analyzed for resistance to virus infection by homologous virus challenge. All R1 lines demonstrated some resistance to infection, with one line showing immunity. Nuclear runoff anslysis of the rate of transcription in resistant vs susceptible plants showed that the level of production of the transgene of mRNA was similar. However, Northern analysis of the steady state level of transgene mRNA showed the resistant plants accumulated 10 fold less mRNA. This evidence suggests that the transgene mRNA is being degraded in the cytoplasm at a turnover rate higher in the resistant line relative to the susceptible. A process which degrades viral RNA sequences identical to the transgene mRNA sequences in the cytoplasm has been hypothesized to result in virus resistance (Baulcombe 1996; English et al. 1996; Ruiz et al. 1998). 3. Gene Silencing and Virus Resistance.

D. Transgenic Plants for Pest Resistance

White clover is less troubled by insect pests than are other forage legumes such as red and alfalfa. The stoloniferous growth habit of white clover reduces reliance on a single vulnerable tap root, enabling com-

4. WHITE CLOVER MOLECULAR GENETICS

209

pensation for any damage due to insects. Nevertheless, in some regions, white clover pests can be extremely destructive. For example, the grass grub, Costelytra zealandica (White), (Coleoptera: Scarabaeidae), can be extremely damaging to white clover in permanent pastures in New Zealand. A diverse range of other pests, including species feeding on seeds and flowers, foliage and roots, have been recorded from different areas (Gaynor and Skipp 1987). Pests such as aphids and leafhoppers may also be important in instances where they vector viruses and phytoplasmas. Traditionally, pesticide chemical applications have been the main method of control, particularly in high-value seed crops. However, rising costs and human health and environmental considerations have impacted on the use of pesticides. Other strategies include cultural control through changes in grazing and cropping management. The use of biological control has been successfully demonstrated for several important pests, but additional control strategies are often required for longterm effective control (Burgess and Gatehouse 1997). The development of cultivars that can better tolerate or resist attack offers advantages that are economical and environmentally acceptable. Resistance to nematode pests has been recognized in other Trifolium spp. and it may be transferable to white clover (Williams and Barclay 1972). Further, a number of toxins and feeding deterrents have been identified against the grass grub (Gaynor and Skipp 1987). However, naturally occurring resistance to other pests is unknown or only poorly characterized. A transgenic approach to resistance offers several advantages over other approaches to control. A major advantage over conventional plant breeding is the ability to introduce specific characteristics without altering desirable traits such as yield, nutritional status, and digestibility. Transgenics also offer accessibility to additional forms of resistance not available in the genus Trifolium. For instance, several classes of genes, most notably toxins from Bacillus thuringiensis and protease inhibitors from various organisms, have been identified with activity against insect and nematode pests. Another potential advantage is the ability to deploy several genes to reduce the chances of selecting insect populations able to overcome single resistance genes. To date, most progress in developing transgenic crop plants has involved the use of toxins from B. thuringiensis. This gram-positive soil bacterium produces crystalline inclusions during sporulation that contain insecticidal and nematicidal proteins called O-endotoxins. The ingestion of these inclusions by feeding insects, with subsequent solubilization in the insect midguts, releases the 8-endotoxins that, upon proteolytic activation, exhibit a highly specific insecticidal activity. The

210

D. WHITE, D. WOODFIELD, B. DUDAS, R. FORSTER, AND D. BECK

8-endotoxins bind to specific high-affinity receptors in the midgut epithelium of susceptible insects, and are believed to form lytic pores in the gut membrane. Rapid cessation of feeding occurs following ingestion by susceptible insects, ultimately leading to death. Initial attempts to express B. thuringiensis 8-endotoxins in plants resulted in low expression levels (Vaeck et al. 1987; Fischhoff et al. 1987). However, this problem was overcome through the use of synthetic genes constructed to avoid AT-rich regions potentially associated with premature transcription termination, aberrant mRNA splicing, mRNA instability, or inefficient codon usage (Perlak et al. 1991; Murray et al. 1991). An alternative approach, involving expression of unmodified B. thuringiensis 8-endotoxin genes in chloroplasts, has also been shown to be successful (McBride et al. 1995). Protease inhibitors from a variety of sources, including mammals and plants, have been shown to affect insect digestive proteolysis and affect development (Burgess and Gatehouse 1997). Burgess and Gatehouse (1997) have proposed that, in addition to the direct role of protease inhibitors in the inhibition of digestive enzymes, these compounds may also exert other physiological effects analogous to those observed in mammals following the ingestion of certain protease inhibitors. Several inhibitors of serine and cysteine proteases engineered into transgenic plants have been shown to provide enhanced protection against predation by insect pests (Hilder et al. 1987; Johnson et al. 1989; McManus et al. 1994; Wolfson and Murdock 1995). E. Potential for Improved Nutritive Value

Although the crude protein content of white clover is at adequate levels (20-24% of dry weight), the digestible carbohydrate content (5-20%) is typically insufficient for optimal animal production (Beever 1993). An optimal feed for ruminant digestion would require a non-structural carbohydrate content (or equivalent energy source) of 30% of dry weight. The imbalance between protein and carbohydrate in white clover results in inefficient use of feed nitrogen by grazing cattle and sheep. Protein is degraded in the rumen by microbial action to form volatile fatty acids (VFAs), acetate and propionate, which are utilized for animal metabolism. If there is insufficient energy available in the rumen, then the excess ammonia from protein degradation is not all used for microbial growth, but instead is taken up by the animal and has to be detoxified via conversion into urea. Up to 40% of feed nitrogen can be lost as urea, due to the imbalance between protein and digestible carbohydrate in white clover (Ulyatt et al. 1988).

4. WHITE CLOVER MOLECULAR GENETICS

211

Alterations that increase the soluble carbohydrate content of the plant would improve the nutritional value of white clover by reducing nitrogen wastage during rumen digestion. However, the partitioning and storage of soluble carbohydrates in higher plants is under complex physiological and developmental regulation. While the level of sucrose could be increased in transgenic plants by overexpression of sucrose phosphate synthase, the level of starch would be reduced so that there would be no net increase in soluble carbohydrate levels (Galtier et al. 1993). An alternative approach may be to create a novel carbohydrate biosynthesis capacity for white clover, which is independent of the existing regulatory processes that control soluble carbohydrate partitioning in leaves. For example, normally non-fructan-storing potato plants, transformed with either a levansucrase gene from Bacillus subtilis (sacB) or . a fructosyltransferase gene from Streptococcus mutans (fij), accumulated very large (MW> 2 x 10° D) bacterial levans (~:2-6 fructans) to high levels, up to 30% dry weight in leaf tissues (van der Meer et al. 1994). Neither the normal non-structural carbohydrate content of the leaves nor the production of foliage by the transgenic plants was affected. Because fructans normally accumulate in the vacuole of fructan-storing plants, the bacterial fructan (levan) biosynthesis genes were fused to a DNA sequence encoding a vacuolar targeting signal. This modification was critical, as expression of the enzyme elsewhere in the plant cell was lethal. Since the level of levansucrase activity detected in transgenic plants was very low, the accumulation of high concentrations of novel fructan may be because the plants cannot degrade or mobilize levans. Before transgenic white clover plants expressing levansucrase are produced in an attempt to increase the soluble carbohydrate content, there is a need to conduct rumen digestion experiments to determine whether bacteriallevans would have a significant impact on feed quality. III. MARKER ASSISTED SELECTION

Genetic gains in white clover to date have been achieved through conventional breeding and selection (Woodfield and Caradus 1994); however, the advent of molecular markers and detailed genetic maps offers methods for dramatically enhancing the rate of gain. Marker-assisted selection relies on identifying markers closely linked to genes of agronomic importance and then accumulating specific combinations of markers (Stuber 1991; Paterson 1996). White clover breeders have attempted to do this historically by identifying morphological (e.g. white

212

D. WHITE, D. WOODFIELD, B. DUDAS, R. FORSTER, AND D. BECK

chevron leaf markings) and biochemical markers (e.g. cyanogenic loci Li and Ac) which are associated with improved population performance. One such linkage that has been used in white clover is associated with the loci involved in the production of hydrogen cyanide (HCN). Most white clover populations contain a mixture of cyanogenic and acyanogenic plants (Williams 1987a). Cyanogenic plants liberate HCN when the leaf is damaged, providing variable levels of protection against invertebrate pests (Williams 1987a). In temperate environments, Caradus and Williams (1989) reported that populations with higher forage yield and persistence contained higher frequencies of cyanogenic genotypes, while acyanogenic genotypes exhibit better adaptation in colder environments (Williams 1987a). The success of acyanogenic cultivars in the USA has also been widely reported and acyanogenesis has been an intrinsic selection parameter since the early 1900s (Gibson and Cope 1985; Williams 1987a). Polymorphic cDNAs have been isolated for linamarase (Li) and these can be used to map this cyanogenic locus (Dunn et al. 1988; Oxtoby et al. 1991). Isozymes have been successfully used to confirm that white clover behaves genetically as a disomic tetraploid (Michaelson-Yeates 1986; Williams et al. 1997) and to identify populations in ecological studies of white clover persistence in grazed swards (Prins et al. 1989; Sawada and Yamauchi 1994). Phosphogluco-isomerase-l, phosphoglucoisomerase-2, esterase, aconitase, and shikimate dehydrogenase have all been studied in white clover. The advent of molecular markers such as RFLPs, AFLPs, RAPDs, and microsatellites in the past decade has increased the scope and precision with which marker-assisted selection can be pursued. Marker-assisted selection is most useful for low-heritability traits both in theory (Paterson et al. 1991) and in practice (Schneider et al. 1997). Quantitative trait loci (QTL) mapping in white clover using molecular markers has been initiated in New Zealand, the USA, and Wales for a range of lowheritability traits, including persistence, forage quality, forage yield, winterhardiness and drought tolerance (M. Abberton, pers. commun.). Molecular markers are also being used in white clover for cultivar identification, genetic diversity, and evolutionary studies to determine its putative parental species (W. Williams, pers. commun.). A. Map-Based Selection

Map-based selection methods have not been attempted yet because no reliable genetic linkage map exists for white clover. Wright et al. (1996) created an F 1 mapping population in white clover and analyzed both the

4. WHITE CLOVER MOLECULAR GENETICS

213

population and parents with DNA clones that had previously been mapped in alfalfa. A total of 85 of 120 alfalfa probes screened produced some scorable data from the parents, but only 17 of 45 probes gave scorable data on the mapping progeny. This resulted in the initial linkage map only containing five linkage groups (Wright et al. 1996). The high levels of polymorphism exhibited for most alfalfa cDNA probes and the presence of identical alleles segregating in each genome were complicating factors (W. Parrott, pers. commun.). Similar difficulties were also reported in genetic studies involving isozymes (Michaelson-Yeates 1986; Williams et al. 1997). Efforts to overcome these difficulties in both New Zealand and Wales involve (1) using recently developed inbred parents in mapping populations to reduce the level of polymorphism and (2) white clover derived probes. More than 80 white clover RFLP and AFLP markers that provide scorable (low copy number) data have been identified in New Zealand (A. Griffiths, pers. commun.). In addition, a number of cDNAs for known white clover genes, such as alcohol dehydrogenase (Ellison et al. 1990a), ribulose bisphosphate carboxylase small subunit (Ellison et al. 1990b), and linamarase (Oxtoby et al. 1991), are available for use in mapping. Initially the RFLP, RAPD, and AFLP markers that have already been developed in white clover will be used to establish linkage maps; however, microsatellites are likely to become increasingly important for both mapping and marker-assisted selection. Microsatellites are potentially the most powerful genome-mapping tool available because they can be locus-specific and genome-specific. Microsatellites have already been successfully used for mapping in both diploid and autotetraploid alfalfa (Diwan et al. 1996) and in hexaploid wheat (Korzan et al. 1997). In allopolyploids, such as wheat, they do not amplify loci from homoeologous chromosomes, a feature that would simplify mapping in white clover (Zhao and Ganal 1996). Once developed, white clover microsatellite markers also have potential uses for the identification and PVR protection of improved cultivars, something that at present relies on morphological observations (Zhao and Ganal 1996). B. Genetic Diversity and Parental Selection

Most outcrossing species carry a high genetic load of deleterious recessive alleles due to the high levels of heterozygosity, and are either unable to be inbred or suffer from severe inbreeding depression (Jones and Bingham 1995; Yamada et al. 1989; Michaelson-Yeates et al. 1997). In alfalfa, which is also an outcrossing species, genetic dissimilarities based on 244 restriction fragments were poorly correlated with diploid prog-

214

D. WHITE, D. WOODFIELD, B. DUDAS, R. FORSTER, AND D. BECK

eny performance but highly correlated with the performance of their isogenic tetraploid progeny (Kidwell et al. 1994a). At the tetraploid level, higher yield was associated with low levels of RFLP homozygosity and high levels ofRFLP heterozygosity (Kidwell et al. 1994b). Given the presence of multiple alleles at each locus in alfalfa, this equates to minimizing the potential for deleterious recessive alleles (genetic load) to be expressed and thus reduce agronomic performance. Kidwell et al. (1994a,b) suggested that genotypes selected for maximum genetic dissimilarity based on molecular markers could be used to improve agronomic performance in a heterozygous, heterogeneous polyploid species such as white clover. While the number of alleles at each locus in white clover is less than that in alfalfa, the degree of inbreeding depression and heterosis suggest that molecular markers could be used to select highly heterozygous parents to create synthetic cultivars with maximum heterozygosity at marker loci. The success of this strategy depends on (1) the presence of detectable polymorphisms among parental genotypes to ensure a reasonable range of genetic diversity estimates from which to select the most diverse parents and (2) positive correlation of progeny performance and RFLP-based diversity measurements. High levels of polymorphism have been detected within and among accessions of white clover (W. Williams, pers. commun.). The level of polymorphism has also been assessed within two very narrow gene pools: the parents oftwo existing synthetic white clover cultivars, Grasslands Sustain (14 parents) and Grasslands Prestige (10 parents). The genetic diversity estimates were very high, ranging from 0.18 to 0.85 among cv Grasslands Sustain parents and from 0.13 to 0.93 among cv Grasslands Prestige parents (D. R. Woodfield, unpubl. data). All possible single-cross progenies among these parents are currently in the third year of evaluation to determine the association between progeny performance and genetic diversity measurements. Selection of parents for synthetic cultivars in white clover has been based solely on phenotypic performance (Woodfield and Caradus 1994); however, ifthe association is positive, molecular markers will be used to develop heterozygous hybrids with improved performance. IV. CONCLUSIONS

Although the application of the techniques of molecular genetics to the breeding of white clover is still at an early phase, and has yet to result in the development of new cultivars, most of the components required are already in place. Production of transgenic plants is now routine

4. WHITE CLOVER MOLECULAR GENETICS

215

practice and one of a number of possible strategies for the development of transgenic cultivars in an outbreeding population has successfully been evaluated. Patterns of transgene expression, stability, and inheritance appear to be similar to those experienced with transgenic plants of other species. However, because information about the genetic composition of white clover is sparse, opportunities to use molecular markers and to manipulate the expression of endogenous genes for plant improvement purposes are still limited. While there are some opportunities to adopt molecular genetic information from other more intensively studied plants, particularly related legumes like Medicago sativa, there is a need to further characterize patterns of gene expression in white clover. Studies with model transgenic plant systems indicate that there are good prospects for the development of transgenic white clover plants resistant to viral diseases and insect pests. These plants should improve the reliability of forage feed supply for animal production. Using gene modification to improve the quality of white clover for animal production may be more complex because most of the plant material is initially digested in the rumen, and it is rumen biomass that ultimately becomes the diet of cattle and sheep. Therefore, there is the additional challenge of modifying the composition of white clover so that either this rumen degradation process is optimized, or essential nutrients are protected from degradation in the rumen and made available for subsequent digestion. The application of a molecular genetic approach to the breeding of white clover is about to enter an exciting phase in which substantial improvements to traits controlled by single genes will accelerate the rate of genetic gain. In the future, there is the possibility of determining the genetic basis of complex traits, like persistence and digestibility, of using molecular markers and molecular linkage maps in conventional breeding, and also for the manipulation of multigene combinations-all developments that will provide further opportunities to enhance the value of white clover as a forage crop.

LITERATURE CITED Alconero, R. 1983a. Viruses infecting six species of perennial clover (Trifolium spp.) in field evaluations of plant introductions and cultivars. Plant Dis. 67:1270-1271. Alconero, R. 1983b. Response of Trifolium hybridum germplasm to mechanical inoculation with bean yellow mosaic and clover yellow vein viruses in the greenhouse. Plant Dis. 67:1272-1273.

216

D. WHITE, D. WOODFIELD, B. DUDAS, R FORSTER, AND D. BECK

Anderson, J. A., S. A. Ghabrial, and N. L. Taylor. 1991a. Natural incidence of peanut stunt virus infection in hybrid populations of Trifolium ambiguum x T. repens. Plant Dis. 75:156-159. Anderson, J. A., N. L. Taylor, and E. G. Williams. 1991b. Cytology and fertility of the interspecific hybrid Trifolium ambiguum x T. repens and backcross populations. Crop Sci. 31:683-687. Austin, R. B., J. Bingham, R. D. Blackwell, L. T. Evans, M. A. Ford, C. L. Morgan, and M. Taylor. 1980. Genetic improvements in winter wheat yields since 1900 and associated physiological changes. J. Agr. Sci., Camb. 94:675-689. Barnett, O. W., and P. B. Gibson. 1975. Identification and prevalence of white clover viruses and the resistance of Trifolium species to these viruses. Crop Sci. 15:32-37. Barnett, O. W., and P. B. Gibson. 1977. Effect of virus infection on flowering and seed production of the parental clones of tillman white clover (Trifolium repens). Plant Dis. Repr. 61 :203-207. Baulcoinbe, D. C. 1996. RNA as a target and an initiator of post-transcriptional gene silencing in transgenic plants. PI. Mol. BioI. 32:79-88. Beachy, R. N., S. Loesch-Fries, and N. E. Turner. 1990. Coat protein-mediated resistance against virus infection. Annu. Rev. Phytopath. 28:451-474. Beck, D. L., R M. Eagles, E. Balmori, D. Voot, T. J. Lough, and R L. S. Forster. 1996. Gene silencing associated with resistance to virus infection. p. 149-157. In: D. Grierson, G. W. Lycett, and G. A. Tucker (eds.), Mechanisms and applications of gene silencing. Nottingham Univ. Easter Seminar Series. Nottingham Univ. Press, Nottingham. Beck, D. L., R. L. S. Forster, M. W. Bevan, K. A. Boxen, and S. C. Lowe. 1990. Infectious transcripts and nucleotide sequence of cloned eDNA of the potexvirus white clover mosaic virus. Virology 177:152-158. Beck, D. L., P. J. Guilford, D. V. Voot, M. T. Anderson, and R L. S. Forster. 1991. Triple gene block proteins of white clover mosaic potexvirus function as transport proteins. Virology 183:695-702. Beck, D. L., C. J. Van Dolleweerd, B. Dudas, D. W. R. White, and R L. S. Forster. 1993. Coat protein-mediated protection against white clover mosaic virus and potato virus X in tobacco. Proc. XVII Int. Grass. Conf. p. 1173-1175. Beck, D. L., C. ]. Van Dolleweerd, T. ]. Lough, E.. Balmori, D. M. Voot, M. T. Anderson, I. E. W. O'Brien, and R. L. S. Forster. 1994. Disruption of virus movement confers broad spectrum resistance against systemic infection by plant viruses with a triple gene block. Proc. Nat. Acad. Sci. (USA) 91:10310-10314. Beever, D. E. 1993. Ruminant animal production from forages: present position and future opportunities. p. 158-164. In: M. J. Baker (ed.). Grasslands for our world. SIR, Wellington, New Zealand. Bingham, E. T. 1983. Molecular genetic engineering vs. plant breeding. Plant Mol. BioI. 2:222-224. Boller, B. c., and J. Nosberger. 1987. Symbiotically fixed nitrogen from field-grown white and red clover mixed with ryegrasses at low levels of 15N-fertilisation. Plant & Soil. 104:219-226. Bond, ]. E., and K. ]. Webb. 1989. Regeneration and analysis of plants from stolon segments of Trifolium repens (white clover). Plant Sci. 61:119-125. Burgess, E. P. J., and A. M. R. Gatehouse. 1997. p. 229-258. In: B. D. McKersie and D. C. W. Brown (eds.), Biotechnology and the improvement offorage legumes. CAB Int., Wallingford, UK. Caradus,]. R, and W. M. Williams. 1989. Breeding for legume persistence in New Zealand.

4. WHITE CLOVER MOLECULAR GENETICS

217

p. 523-537. In: Persistence of forage legumes. G.c. Marten et al. (eds.). Proc. trilateral workshop, Hawaii. Am. Soc. Agronomy, Madison, WI. Conner, A. J., and M. C. Christey. 1994. Plant breeding and seed marketing options for the introduction of transgenic insect-resistant crops. Biocontrol Sci. & Technol. 4: 463-473. Crill, P., D. J. Hagedorn, and E. W. Hanson. 1970a. Incidence and effect of alfalfa mosaic virus on alfalfa. Phytopathology 60:1432-1435. Crill, P., E. W. Hanson, and D. J. Hagedorn. 1970b. Resistance and tolerance to alfalfa mosaic virus in alfalfa. Phytopathology 61 :369-371. Crush, J. R. 1987. Nitrogen fixation. p. 185-201. In: M. J. Baker and W. M. Williams (eds.), White clover. CAB Int., Wallingford, UK. Diwan, N., J. H. Bouton, G. Kochert, A. A. Bhagwat, and P. B. Cregan. 1996. Mapping of simple sequence repeats (SSR) DNA markers in diploid and tetraploid alfalfa. Abstract poster 198, Plant Genome IV, San Diego. Dudas, B., R. L. S. Forster, C. J. Van Dolleweerd, and D. W. R. White. 1993. Transformation of white clover with a gene encoding a white clover mosaic virus coat protein. p. 1040-1041. In: Proc. XVII Int. Grasslands Congr., Palmerston North, New Zealand. Dudas, B., D. R. Woodfield, P. M. Tong, M. F. Nicholls, G. R. Cousins, R. Burgess, D. W. R. White, D. L. Beck, T. J. Lough, and R. L. S. Forster. 1993. Estimating the agronomic impact of white clover mosaic virus on white clover performance in the North Island of New Zealand. New Zealand J. Agr. Res. 41:171-178. Dunn, M. A., M. A. Hughes, and A. L. Sharif. 1988. Synthesis of the cyanogenic betaglucosidase, linamarase, in white clover. Arch. Biochem. & Biophys. 260:561-568. Ealing, P. M., K. R. Hancock, and D. W. R. White. 1994. Expression of the pea albumen 1 gene in transgenic white clover and tobacco. Transgenic Res. 3:344-354. Ellison, N. W., P. L. Yu, and D. W. R. White. 1990a. Nucleotide sequence of a white clover alcohol dehydrogenase eDNA. Nucl. Acids Res. 18:4913. Ellison, N. W., P. L. Yu, and D. W. R. White. 1990b. Nucleotide sequence of a white clover ribulose bisphosphate carboxylase small subunit gene. Nucl. Acids Res. 18:4914. English, J. J., E. Mueller, and D. C. Baulcombe. 1996. Suppression of virus accumulation in transgenic plants exhibiting silencing of nuclear genes. PI. Cell. 8:179-188. Finnegan, J., and D. McElroy. 1994. Transgene inactivation: plants fight back! Bio/Technology 12:883-888. FischhotJ, D. A., K. S. Bowish, F. J. Perlak, P. G. Marrone, S. M. McCormick, J. G. Niedermeyer, D. A. Dean, K. Kusano-Kretzmer, E. J. Mayer, D. E. Rochester, S. G. Rogers, and R. T. Fraley. 1987. Insect tolerant tomato plants. Bio/Technology 5:807-813. Fitchden, J. H., and R. N. Beachy. 1993. Genetically engineered protection against viruses in transgenic plants. Annu. Rev. Microbiol. 47:739-763. Flavell, R. B. 1994. Inactivation of gene expression in plants as a consequence of specific sequence duplication. Proc. Nat. Acad. Sci. (USA) 91 :3490-3496. Forster. R. L. S., D. L. Beck, P. J. Guilford. C. J. Van Dolleweerd, and M. T. Anderson. 1992. The coat protein of white clover mosaic potexvirus has a role in facilitating cell to cell transport in plants. Virology 191:480-484. Forster, R. L. S., M. W. Bevan, S-A. Harbison, and R. C. Gardner. 1988. The complete nucleotide sequence of the potexvirus white clover mosaic virus. Nucl. Acids Res. 16:291-303. Forster, R. L. S., D. V. Faulds, P. J. Guilford, T. L. Scott, D. R. Musgrave, H. S. Easton, and K. Steele. 1984. Viruses of forage legumes. DSIR Plant Diseases Division Summary of Research Report, Auckland, New Zealand.

218

D. WHITE, D. WOODFIELD, B. DUDAS, R FORSTER, AND D. BECK

Forster, R 1. S., and B. A. M. Morris-Krsinich. 1985. A distinct strain oflucerne Australian latent virus in white clover in New Zealand. Ann. Appl. BioI. 107:449-454. Fry, P. R 1959. A clover mosaic virus in New Zealand pastures. New Zealand J. Agr. Res. 2:971-981. Galtier, N., C. H. Foyer, J. Huber, T. A. Voelker, and S. C. Huber. 1993. Effects of elevated sucrose-phosphate synthase activity on photosynthesis, assimilate partitioning, and growth in tomato (Lycopersicon escuJentum var UC82B). Plant Physiol. 101:535-543. Garrett, R 1990. The effect of some viruses on the yield of subclover and white clover. Australian Dept. of Agriculture and Rural Affairs, Crop Protection Bul. Issue. 13:6-8. Garrett, R 1991. Impact of viruses on pasture legume productivity. In: N. P. McFarlane (ed.). Proc. White Clover Conference 1991, Victoria, Australia. Conference Proc. 45. Gaynor, D. 1., and R A. Skipp. 1987. Pests. p. 461-492. In: M. J. Baker and W. M. Williams. (eds.), White clover. CAB Int., Wallingford, UK Gibson, P. B., O. W. Barnett, G. A. Pederson, M. R McLaughlin, W. E. Knight, J. B. Miller, W. A. Cope, and S. A. Tolin. 1989. Registration of Southern regional virus resistant white clover germplasm. Crop Sci. 29:241. Gibson, P. B. and W. A. Cope. 1985. White clover. Agronomy 25:471-490. Gillet, J. M. 1985. Taxonomy and morphology. p. 7-69. In: N. M. Taylor (ed.), Clover science and technology. Agronomy 25. Grant, S. R 1999. Dissecting the mechanisms of posttranscriptional gene silencing: divide and conquer. Cell. 96:303-306. Gresshoff, P. M. 1980. In vitro culture of white clover: callus, suspension, protoplast culture and plant regeneration. Bot. Gaz. 141:157-164. Grumet, R 1994. Development of virus resistant plants via genetic engineering. Plant Breed. Rev. 12:47-79. Guy, P., A. Gibbs, and K. Harrower. 1980. The effect of white clover mosaic virus on nodulation of white clover (Trifolium repens 1. cv. Ladino). Austral. J. Agr. Res. 31:307-311. Hagen, G., A. Kleinschmidt, and T. J. Guilfoyle. 1991. Auxin-induced expression of the soybean GH3 promoter in transgenic tobacco plants. Plant Molec. BioI. 17:567-579. Harville, B. G., and K S. Derrick. 1978. Identification and prevalence of white clover viruses in Louisiana. Plant Dis. Rep. 62:290-292. Helms, K, W. J. Muller, and P. M. Waterhouse. 1993. National survey of viruses in pastures of subterranean clover. I. Incidence of four viruses assessed by ELISA. Austral. J. Agr. Res. 44:1837-1862. Herskowitz, I. 1987. Functional inactivation of genes by dominant negative mutations. Nature 329:219-222. Hilder, V. A., A. M. R. Gatehouse, S. E. Sherman, R. F. Barker, and D. Boulter. 1987. A novel mechanism for insect resistance engineered into tobacco. Nature. 330:160-163. Holland, J. B., and E. T. Bingham. 1994. Genetic improvement of alfalfa cultivars representing different eras of breeding. Crop Sci. 34:953-957. Hussain, S. W., and W. M. Williams. 1997. Development of a fertile genetic bridge between Trifolium ambiguum m. Bieb. and T. repens 1. Theor. Appl. Gen. 95: 678-690. Johnson, R, J. Narvaez, G. An, and C. Ryan. 1989. Expression of proteinase inhibitors I and II in transgenic tobacco plants: effects on natural defense against Manduca sexta larvae. Proc. Nat. Acad. Sci. (USA) 86:9871-9875. Jones, J. S., and E. T. Bingham. 1995. Inbreeding depression in alfalfa and cross-pollinated crops. Plant Breed. Rev. 13:209-233. Kanyuka, K V., V. K Vishnichenko, K E. Levay, D.Y. Kondrikov, E. V. Ryabov, and S. K. Zavriev. 1992. Nucleotide sequence of shallot virus X RNA reveals a 5'-proximal cistron

4. WHITE CLOVER MOLECULAR GENETICS

219

closely related to those of potexviruses and a unique arrangement of the 3' proximal cistrons. J. Gen. Virol. 73:2553-2560. Kidwell. K. K.. E. T. Bingham, D. R. Woodfield. and T. C. Osborn. 1994a. Relationships among genetic distance. forage yield and heterozygosity in isogenic diploid and tetraploid alfalfa populations. Theor. Appl. Gen. 89:323-328. Kidwell, K. K., D. R. Woodfield, E. T. Bingham. and T. C. Osborn. 1994b. Molecular marker diversity and yield of isogenic 2x and 4x single-crosses of alfalfa. Crop Sci. 34:784-788. Korzan, V.. M. W. Ganal, and M. S. Roser. 1997. Genetic mapping of microsatellite markers in the wheat (Triticum aestivum) genome. Abstract poster 57, Plant and Animal Genome V, San Diego. Larkin. P. J., J. M. Gibson. U. Mathesius, J. J. Weinman. E. Gartner, E. Hall. G. J. Tanner. B. G. Rolfe. and M. A. Djordjevic. 1997. Transgenic white clover. Studies with the auxinresponsive promoter, GH3. in root gravitropism and lateral root development. Trans. Res. 5:325-335. Lindbo, J. A.. and W. G. Dougherty. 1992a. Pathogen-derived resistance to a potyvirus: immune and resistance phenotypes in transgenic tobacco expressing altered forms of a potyvirus coat protein nucleotide sequence. Mol. Plant-Microbe Int. 5:144-153. Lindbo, J. A.. and W. G. Dougherty. 1992b. Untranslatable transcripts of the tobacco etch virus coat protein gene sequence can interfere with tobacco etch virus replication in transgenic plants and protoplasts. Virology 189:725-733. Lindbo, J. A.. L. Silva-Rosales. and W. G. Dougherty. 1993a. Pathogen derived resistance to potyviruses: working but why? Seminars in Virology 4:369-379. Lindbo, J. A.. L. Silva-Rosales. W. M. Proebsting. and W. G. Dougherty. 1993b. Induction of a highly specific antiviral state in transgenic plants: implications for regulation of gene expression and virus resistance. Plant Cell 5:1749-1759. Lough. T. J., K. Shash. B. Xoconostle-Cazares, K. R. Hofstra, D. L. Beck, E. Balmori, R. L. S. Forster and W. J. Lucas. 1998. Molecular dissection of the mechanism by which potexvirus triple gene block proteins mediate cell-to-cell transport of infectious RNA. MPMI. 11:801-814. Maheswaran. G., and E. G. Williams. 1985. Origin and development of somatic embryoids formed directly on immature embryos of Trifolium repens in vitro. Ann. Bot. 56:619-630. Martin, P. H.. B. E. Coulman. and J. F. Peterson. 1990. Genetics of tolerance to white clover mosaic virus in red clover. Crop Sci. 30:1191-1194. McBride. K. E.; Z. Svab. D. J. Schaaf. P. S. Hogan, D .M. Stalker. and P. Maliga. 1995. Amplification of a chimeric Bacillus gene in chloroplasts leads to an extraordinary level of an insecticidal protein in tobacco. Bio/Technology 13::362-365. McLaughlin. M. R.. and T. E. Fairbrother. 1993. Selecting subclover for resistance to clover yellow vein virus. Phytopathology 83:1421 (Abstr.). McLaughlin. M. R.. G. A. Pederson. R. R. Evans. and R. I. Ivy. 1992. Virus diseases and stand decline in a white clover pasture. Plant Dis. 76:158-162. McManus. M. T., D. W. R. White. and P. G. McGregor. 1994. Accumulation of a chymotrypsin inhibitor in transgenic tobacco can affect the growth of insect pests. Transgenic Res. 3:50-58. Michaelson-Yeates, T. P. T. 1986. Phosphoglucoisomerase variation in Trifolium repens. Genetica. 70:53-58. Michaelson-Yeates. T. P. T., A. Marshall. M. T. Abberton. and I. Rhodes. 1997. Selfcompatibility and heterosis in white clover (Trifolium repens L.). Euphytica 94:341-348.

220

D. WHITE, D. WOODFIELD, B. DUDAS, R. FORSTER, AND D. BECK

Minson, D. J. 1990. Forage in ruminant nutrition. Academic Press, San Diego. Mohapatra, S. S., and P. M. Gresshoff. 1982. Ecotypic variation of in vitro plantlet formation in white clover (Trifolium repens). Plant Cell Rep. 1:189-192. Morozov, S. Y., V. V. Dolia, and J. G. Atabekov. 1989. Probable reassortment of genomic elements among elongated RNA-containing plant viruses. J. Mol. Evol. 29:52-62. Morozov, S. Y., L .I. Lukasheva, B. K Chernov, K G. Skryabin, and J. G. Atabekov. 1987. Nucleotide sequence of the open reading frames adjacent to the coat protein cistron in potato virus X genome. FEBS Lett. 213:438-442. Murray, K E., T. Rocheleau, M. Eberle, C. Stock, V. Sekar, and M. Adang. 1991. Analysis of unstable RNA transcripts of insecticidal crystal protein genes of Bacillus thuringiensis in transgenic plants and electroporated protoplasts. Plant Mol. BioI. 16:1035-1050. Neale, A. D., A. J. Wahlethner, M. Lund, H. T. Bonnett, A. Kelly, D. R. Meeks-Wagner, W. J. Peacock, and E. S. Dennis. 1990. Chitinase, ~-1,3-glucanase,osmotin, and extensin are expressed in tobacco explants during flower formation. Plant Cell 2:673-684. Osbourn, J. K, J. W. Watts, R. N. Beachy, and T. M. A. Wilson. 1989. Evidence that nucleocaspid dissassembly and a later step in virus replication are inhibited in transgenic tobacco expressing TMV coat protein. Virology 172:370-373. Oswald, T. H., A. E. Smith, and D. V. Phillips. 1977. Callus and plantlet regeneration from cell cultures of Ladino clover and soybean. Physiol. Plant 39:129-134. Oxtoby, K, M. A. Dunn, A. Pancoro, and M. A. Hughes. 1991. Nucleotide and derived amino acid sequence of the cyanogenic beta-glucosidase (linamarase) from white clover (Trifolium repens L.). Plant Mol. BioI. 17:209-219. Parrott, W. A. 1991. Auxin-stimulated somatic embryogenesis from immature cotyledons of white clover. Plant Cell Rep. 10:17-21. Paterson, A. H. 1996. DNA marker-assisted crop improvement. p. 71-83. In: A. H. Paterson (ed.), Genome mapping in plants. R. G. Landes Co., Georgetown, Texas. Paterson, A. H., S. Damon, J. D. Hewitt, D. Zamir, S. K Rabinwitch, E. S. Lincoln, K S. Lander, and S. D. Tanksley. 1991. Mendelian factors underlying quantitative traits in tomato: comparison across species, generations and environments. Genetics 127:181-197. Pederson, G. A., and M. R. McLaughlin. 1989. Resistance to viruses in Trifolium interspecific hybrids related to white clover. Plant Dis. 73:997-999. Pemberton, I. J., G. R. Smith, and M. R. McLaughlan. 1989. Evaluation ofarrowleafclover for tolerance to bean yellow mosaic virus. Phytopathology 79:230-234. Perlak, F. J., R. L. Fuchs, D. A. Dean, S. L. McPherson, and D. A. Fischhoff. 1991. Modification of the coding sequence enhances plant expression of insect control protein genes. Proc. Nat. Acad. Sci. (USA) 88:3324-3328. Pesic-van Esbroeck, Z., and C. Hiruki. 1990. Hypersensitive reaction in alfalfa (Medicago sativa) induced by alfalfa mosaic virus. Plant Pathol. 39:524-529. Phillips, R. L., S. M. Kaeppler, and P. Olhoft. 1994. Genetic instability of plant tissue cultures: Breakdown of normal controls. Proc. Nat. Acad. Sci. (USA) 91:5222-5226. Pittock, c., J. J. Weinman, and B. G. Rolfe. 1997. The activity of a basic chitinase promoter in transgenic white clover provides insights into plant development and symbiosis. Austral. J. Plant Physioi. 24:555-561. Potter, L. R. 1993. The effects of white clover mosaic virus on vegetative growth and yield of clones of S.100 white clover. Plant Pathol. 42:797-805. Powell, P. A., P. R. Sanders, N. Turner, R. T. Fraley, and R. N. Beachy. 1990. Protection against tobacco mosaic virus infection in transgenic plants requires accumulation of coat protein rather than coat protein RNA sequences. Virology 175:124-130.

4. WHITE CLOVER MOLECULAR GENETICS

221

Powell-Abel, P. A., R. S. Nelson, B. De, N. Hoffmann, S. G. Rogers, R. T. Fraley, and R. N. Beachy. 1986. Delay of disease development in transgenic plants that express the tobacco mosaic virus coat protein gene. Science. 232:738-743. Pratt, M. J. 1967. Reduced winter survival and yield of clover infected with clover yellow mosaic virus. Can. J. Plant Sci. 47:289-294. Prins, E., P. M. Sanders, and T. B. Lyons. 1989. Use of electrophoretic techniques to identify the proportion of an improved white clover cultivar ('Grasslands Kopu') in a mixed sward. New Zealand J. Agr. Res. 32:515-520. Ralph, W. 1992. Viruses contribute to clover decline. Rural Research (Australia) 155:4-10. Rattray, P. V., and J. P. Joyce. 1974. Nutritive value of white clover and perennial ryegrass. 4. Utilization of dietary energy. New Zealand J. Agr. Res. 17:401-408. Register, J. c., and R. N Beachy. 1988. Resistance to TMV in transgenic plants results from interference with an early event in infection. Virology 11111:524-532. Rogers, G. L., R. H. D. Porter, and 1. Robinson. 1982. Comparison of perennial ryegrass and white clover milk production. p. 213-214. In: K. L. Macmillan and V. K. Taufa (eds.), Dairy production from pastures. Proc. N.Z. and Aust. Soc. Anim. Prod. Ruiz, M. T., O. Voinnet, and D. C. Baulcombe. 1998. Initiation and maintenance of virusinduced gene silencing. PI. Cell. 10:937-947. Russell, W. A. 1991. Genetic improvement of maize yields. Adv. in Agronomy. 46:245-298. Sanford, J. c., and S. A Johnston. 1985. The concept of parasite-derived resistance: deriving resistance genes from the parasite's own genome. J. Theor. BioI. 113:395-405. Sawada, H., and K. Yamauchi. 1994. Identification of white dover (Trifolium repens L.) clones using isozymes. J. Jap. Soc. Grassl. Sci. 39:488-496. Schimdt, E. D., A. J. de Jong, and S. C. de Vries. 1994. Signal molecules involved in plant embryogenesis. Plant Molec. BioI. 26:1305-1313. Schnieder, K. A, M. E. Brothers, and J. D. Kelly. 1997. Marker-assisterl selection to improve drought resistance in cornman bean. Crop Sci. 37:51-60. Scott, A., D. Woodfield, and D. W. R. White. 1998. Allelic composition and genetic background effects on transgene expression and inheritance in white clover. Molec. Breeding.4:479-490. Scott, S. W. 1982. Tests for resistance to white dover mosaic virus in red and white clover. Ann. Appi. BioI. 100:393-398. Sedcole, J. R. 1977. Number of plants necessary to recover a trait. Crap Sci. 17;667-668. Seppanen, P., R. Puska, J. Honkanen, L. G. Tyulkina, O. Fedorkin, S. Y. Morozov, and J. G. Atabekov. 1997. Movement protein-derived resistance to triple gene blockcontaining plant viruses. J. Gen. Viral. 78:1241-1246. Smith, H. A, S. L. Swaney, T. D. Parks, E. A Wernsman, and W. G. Dougherty. 1994. Transgenic plant virus resistance mediated by untranslatable sense RNAs: expression, regulation, and fate of nonessential RNAs. Plant Cell. 6:1441-1453. Stuber, C. W. 1991. Biochemical and molecular markers in plant breeding. Plant Breed. Rev. 9;37-61. Stuber, C. W. 1994. Heterosis in plant breeding. Plant Breed. Rev. 12:227-251. SumL S., T. TsuneyoshL and H. Furutani. Hl93. Novel rod-shaped viruses isolated from garlic, Allium sativum, possessing a unique genome organization. J. Gen. Viral. 74:1879-1885. Tapio, E. 1970. Virus diseases of legumes in Finland and in the Scandinavian countries. Ann. Agr. Fenniae 9:1-97. Taylor, N. L. 1985. Clovers araund the world. p. 1-6. In: Clover science and technology. N.L. Taylor (ed.), Agranomy 25.

222

D. WHITE, D. WOODFIELD, B. DUDAS, R. FORSTER, AND D. BECK

Taylor, N. 1., S. A. Ghabrial, S. Diachun, and P. 1. Cornelius. 1986. Inheritance and backcross breeding of the hypersensitive reaction to bean yellow mosaic virus in red clover. Crop Sci. 26:68-74. Ulyatt, M. J. 1981. The feeding value of herbage: can it be improved. New Zealand Agr. Sci. 15:200-205. Ulyatt, M. J., D. J. Thomson, D. E. Beever, R T. Evans, and M. J. Haines. 1988. The digestion of perennial ryegrass (Lolium perenne cv. Melle) and white clover (Trifolium repens cv. Blanca) by grazing cattle. British J. Nutr. 60:137-149. Vaeck, M., A. Reynearts, H. Hafte, S. Jansens, M. De Beuckeleer, C. Dean, M. Zabeau, M. Van Montagu, and J. Leemans. 1987. Transgenic plants protected from insect attack. Nature 328:33-37. van der Meer, 1. M., M. J. M. Ebskamp, R G. F. Visser, P. J. Weisbeek, and S. C. M. Smeekens. 1994. Fructan as a new carbohydrate sink in transgenic potato plants. Plant Cell 6:561-570. van Wijk, A. J. P., and D. Reheul. 1991. Achievements in fodder crops breeding in maritime Europe. Proc. 16th Meeting Fodder Crops Section of Eucarpia. p. 13-18. Voisey, C. R, D. W. R White, B. Dudas, R D. Appleby, P. M. Ealing, and A. G. Scott. 1994. Agrobacterium-mediated transformation of white clover using direct shoot organogenesis. Plant Cell Rep. 13:309-314. Voisey, C. R, D. W. R White, P. G. McGregor, P. J. Wigley, and C. N. Chilcott. 1994. Transformation of white clover with B.t. genes. p. 75-83. In: R J. Akhurst (ed.), Proc. 2nd Canberra Bacillus thuringiensis Meeting, CSIRO, Australia. Wassenegger, M. and T. Pelissier. 1998. A model for RNA-mediated gene silencing in higher plants. PI. Mol. BioI. 37:349-362. White, D. W. R 1984. Plant regeneration from long-term suspension cultures of white clover. Planta 162:1-7. White, D. W. R 1997. Potential of biotechnology to alter pasture yield and quality. p. 441-454. In: R A. S. Welch, D. J. W. Burns, S. R Davis, A. 1. Popay, and C. G. Prosse, (eds.), Milk composition, production and biotechnology. CAB Int., Wallingford, U.K. White, D. W. R, and D. Greenwood. 1987. Transformation of the forage legume Trifolium repens 1. using binary Agrobacterium vectors. Plant Mol. BioI. 9:461-469. White, D. W. R, and C. Voisey. 1994. Prolific direct plant regeneration from cotyledons of white clover. Plant Cell Rep. 13:303-308. Williams, E. G., and 1. M. Verry. 1981. A partially fertile hybrid between Trifolium repens and T. ambiguum. New Zealand J. Bot. 19:1-7. Williams, E. G., W. M. Williams, D. W. R White, and K. K. Pandey. 1978. Interspecific hybridization in Trifolium: Progress report. Clover and Special Purpose Legumes Res. 11: 52-57. Williams, W. M. 1987a. Adaptive variation. p. 299-321. In: M. J. Baker and W. M. Williams (eds.), White clover. CAB Int., Wallingford, U.K. Williams, W. M.. 1987b. Genetics and breeding. p. 343-419. In: M. J. Baker and W. M. Williams (eds.), White clover. CAB Int., Wallingford, U.K. Williams, W. M., and P. C. Barclay. 1972. The effect of clover stem eelworm on the establishment of pure swards of white clover. N.Z. J. Agric. Res. 16:77-80. Williams, W. M., K. M. Mason, M. 1. Williamson. 1998. Genetic analysis of shikimate dehydrogenase allozymes in Trifolium repens 1. Theoret. and Appl. Genetics 96:859-868. Wisniewski, 1. A., P. A. Powell, R S. Nelson, and R N. Beachy. 1990. Local and systemic spread of tobacco mosaic virus in transgenic tobacco. Plant Cell. 2:559-567. Wolfson, J. 1., and 1. 1. Murdock. 1995. Potential use of protease inhibitors for host plant resistance: a test case. Enviro. Entomol. 24:52-57.

4. WHITE CLOVER MOLECULAR GENETICS

223

Woodfield, D. R., and J. R. Caradus. 1994. Genetic improvement in white clover representing six decades of plant breeding. Crop Sci. 34:1205-1213. Woodfield, D. R., and D. W. R. White. 1996. p. 125-130. In: D. R. Woodfield (ed.), White clover: New Zealand's competitive edge. Agron. Soc. New Zealand Special Publ. 11. Wright, S., J. H. Bouton, W. A. Parrott, and G. Kochert. 1996. Comparative mapping of white clover and alfalfa. Proc. 14th Trifolium Conf. p. 11. Yamada, T. 1989. Selection of a highly-regenerative genotype of white clover (Trifolium repens L.) and plant regeneration from protoplasts derived from this genotype. Euphytica 44:181-186. Yamada, T., H. Fukuoka, and T. Wakamatsu. 1989. Recurrent selection programs for white clover (Trifolium repens L.) using self-compatible plants. 1. Selection of selfcompatible plants and inheritance of a self-compatibility factor. Euphytica 44:167-172. Yamamoto, Y. T., C. G. Taylor, G. N. Acedo, C-l. Cheng, and M. A. Conkling. 1991. Characterization of cis-acting sequences regulating root-specific gene expression in tobacco. Plant Cell. 3:371-382. Zhao, X. and M. W. Ganal. 1996. Applications of repetitive DNA sequences in plant genome analysis. In: A. H. Paterson (ed.), Genome and mapping in plants. R. G Landes Co., Austin, Texas. p 111-125.

5 The Rise and Fall of Overdominance James F. Crow* Genetics Department University of Wisconsin Madison, Wisconsin 53706

I. Introduction II. Early History III. Alternative Explanations of Heterosis A. The Reappearance of Overdominance B. The Apogee of Overdominance C, Doubts about Overdominance D. Dominance Takes the Field E. Is There a Role for Overdominance and Epistasis'? IV. Why Are Hybrids So Good'? V, Physiological and Molecular Studies VI. Is the Hybrid Breeding Method Best'? A. Maize B, Other Crop Species VII. Conclusions Literature Cited

Nature tells us, in the most emphatic manner, that she abhors perpetual self-fertilization. Darwin (1862) The first and most important conclusion which may be drawn from the observations given in this volume, is that cross-fertilization is generally beneficial and self-fertilization injurious. Darwin (1876) *1 should like to dedicate this review to George Sprague, a leader in the field for many years, who has so often been right. His death on November 24, 1998, ended an outstanding career. His wisdom, along with a large measure of common sense, will be missed. An earlier draft of this paper was read by Charles Stuber, Howie Smith, Donald Duvick, Kendall Lamkey, Wyman Nyquist, Arnel Hallauer, Jerry Kermicle, and Oliver Nelson. I am indebted to all of them for catching errors and for useful comments. I also want to thank Forest Troyer and Charles Gardner for supplying data for Figs. 5.1 and 5.2, respectively.

Plant Breeding Reviews, Volume 17. Edited by Jules Janick ISBN 0-471-33373-5 © 2000 John Wiley & Sons, Inc. 225

J. CROW

226

I. INTRODUCTION

Hybrid maize (Zea mays L.) is one of the greatest triumphs in agriculture. Since the introduction of hybrid seed in the 1930s, the yield in the United States has increased about 5-fold. Data compiled by A. F. Troyer (Fig. 5.1) show that, after a long plateau from 1866 until about 1935, the introduction of double-cross corn led to an annual increase of 1.04 bushel per acre (65 kg ha- 1 ). Then, after single crosses became practical, the increase was considerably faster, 1.71 bushels per acre (107 kg ha- 1), with no apparent slowing of the rate of improvement in recent years. Some of this increase was due to improved management and environment;, for example, synthetic nitrogen fertilizer also increased during much of this period. Yet the major improvement was in genetic quality. By saving seed from an earlier period and growing plants from these seeds in a later environment, the environmental changes during that time interval could be assessed. Numerous studies indicate that the

140

120

100

a.

80

~ (/)

:::J

co

A

80

b=1.71

(/)

Q)

100

single cross

l!? 0

Ci>

6

Civil War to 1997

120

«

140

U.S. Corn Yields & Kinds of Corn

/'A

60

60

,;

40

40

open-pollinated

JAl "tIA •JIt. IJo.Ai6a ~ JdJICMj,A A' W /', ll6. l'\A AfrA "6 A

20

20

6

b=0.02 0 1860

1880

1900

1940

1920

1960

1980

0 2000

Year Fig. 5.1. Maize yield in bushels per acre in the United States. The periods dominated by open-pollinated, double-crosses, and single-crosses are indicated. along with regression coefficients (bu.lacre). Redrawn with permission of A. Forest Troyer, from data compiled by USDA. Reprinted from Genetics.

5. THE RISE AND FALL OF OVERDOMINANCE

227

genetic contribution to the improved maize yields was between 33% and 89%, with a mean of 71 % (Duvick 1992). Similar results are beginning to appear from the tropics (Fernandes and Franzon 1997; Alliprandini et al. 1998). Despite the great importance of heterosis, its genetic basis has been a matter of uncertainty and changing views from the beginning. And, although there is now a general consensus that additive and dominance effects are the main components of the genetic variance, the possible additional roles of overdominance and epistasis are still uncertain. In this article, I review the explanations of heterosis as they have changed with the shifting winds of prevailing opinion. Fortunately, the progress under various systems of selection has not waited for a resolution of the issue. Increasing performance of hybrid maize continues unabated. Not surprisingly, this review will deal almost entirely with maize. That's where the action has been and still is. I have also, quite gratuitously, seized this opportunity to record my own vacillations over the years.

II. EARLY HISTORY The puniness of inbred plants and animals and the strength and luxuriance of hybrids has been a part of conventional wisdom and folklore since classical antiquity. The Greeks, and especially the Romans, recognized and utilized the hardiness and vigor of mules. Zirkle (1952) cites a number of striking examples of such ancient wisdom, along with some equally striking examples of human credulity and predilection for myths. In 1766, K6lreuter described the greater luxuriance of species hybrids in many plant genera. Naudin (1865) reported that 24 of 35 crosses in assorted plant genera showed a size increase. The most spectacular of these was in Datura, where the hybrids were twice as tall as the parents. The nineteenth-century literature is full of examples. There is evidence that the Native Americans, to whom we owe the development of maize, were aware of the value of hybrids (Collins 1909; Goldman 1998) When Charles Darwin undertook a project, he did his customarily thorough job-exhaustive may be a more appropriate word. His 490-page book, published in 1876, The Effects of Cross and Self-fertilization in the Vegetable Kingdom, is filled with examples, and his final chapter begins with the second quotation at the beginning of this article. The first quotation is from his earlier (1862) book on fertilization in orchids. Significantly for our purpose, Darwin gave considerable attention to maize. As

228

J. CROW

pure description, his 1876 book is a masterpiece, and the data are fully convincing. Distrustful of his own statistical skills, he called on his cousin, Francis Galton, to judge what we would now call the statistical significance of his carefully controlled experiments. Yet, because of the prevailing ignorance of the nature of inheritance, Darwin's work is incomplete. An acceptable explanation had to await the rediscovery of Mendel's laws in 1900. Had Darwin known of Mendel's work, he might well have had some deeper explanatory insights. Incidentally, Mendel (1865) also noticed heterosis. Hybrids between tall and short cultivars of peas considerably exceeded the height of the tall parents. But of course this didn't become generally known until the rediscovery of Mendel's work in 1900. By the latter part of the nineteenth century, maize breeding was becoming a major research endeavor in the midwestern United States. An early leader was W. J. Beal, a student of the great Harvard botanist, Asa Gray. Beal, clearly inspired by Darwin's book, which he reviewed, started experiments of his own (Beal 1880; Zirkle 1952). He found that the yield of hybrids between two strains that had been geographically isolated for many years exceeded their parents in yield by 50%. Significantly, Beal developed the system of detasseling alternate rows, permitting the unambiguous identification of inbred and hybrid seed, and initiated what became the universal breeding practice, called top crossing. Curiously, the most influential post-Mendelian research on hybrid performance did not occur in the U.S. Corn Belt, but on the much less maize-friendly East Coast. G. H. Shull did his path-breaking studies at Cold Spring Harbor, on Long Island, New York, while at the same time E. M. East was doing almost identical experiments at Connecticut State College. In 1908 Shull wrote a paper with the curiously mundane title, "The composition of a field of maize," which gave not the slightest hint of the importance of its contents. Shull noted that inbred lines showed general weakening and reduced yield, but that regardless of the number of generations of inbreeding, the hybrids recovered immediately. Furthermore, the hybrid performance often exceeded that of the randomly mating cultivars from which the inbreds had been derived. Shull also noted that the hybrids were remarkably uniform, a finding of great importance in later years after mechanical harvesting had become routine. In his next paper, Shull (1909) outlined the way in which hybrid vigor could be used, foreseeing much of what has since become commonplace. East's (1908a,b) results were very similar. He observed the deterioration caused by inbreeding and the high quality of hybrids, but he didn't

5. THE RISE AND FALL OF OVERDOMINANCE

229

see the practical value of crossing inbred lines. He disliked the idea of using weak inbreds as parents. Yet, on hearing Shull's report in January 1908, East wrote to him, saying: "Since studying your paper, I agree entirely with your conclusion, and wonder why I have been so stupid as not to see the fact myself." His papers were generous to Shull and added a great deal of corroborating data. Shull and East differed strongly on one point, however. East believed that the performance of inbred lines was so poor that Shull's method was impractical. The weak inbreds would not produce enough seed to be successful commercially. He suggested that farmers should produce their own hybrid seed by crossing open-pollinated cultivars (Nelson 1993). Shull was unconvinced. The disagreement was strong, and they differed on other questions as well; but the two men agreed not to let their differences show in public debate, nor to let personalities intrude. They remained true to their word. In his talk at the Iowa State College Heterosis Conference in 1950, Shull's (1952) generosity to East was apparent to all and inspiring to hear. The practical use of hybrid maize languished until D. F. Jones (1918, 1922) took the decisive step. While still a graduate student, Jones suggested using double crosses, that is, crossing two unrelated hybrids. In this way the seed would be grown on high-yielding hybrid plants, and the problem of insufficient and expensive seed was circumvented. The double-cross hybrids were somewhat more variable than single crosses, but were much less variable than open-pollinated cultivars. Curiously, Shull had himself produced double-cross hybrids as early as 1911 and found them to be fully comparable to single crosses (Shull 1952). Yet, for some reason, he didn't see this as a solution to the problem raised by East. The double-cross idea, simple in its concept but revolutionary in its consequences, had to wait for a young graduate student some seven years later. With the double-cross method, maize breeding research returned to the Midwest. Shull moved to Princeton University and stopped much of his maize research. Among other things, he founded the journal, Genetics, and devoted a great deal of time and energy to it (Crow 1998a). East moved to Harvard, where he worked on a variety of materials of which maize was only one. The development of hybrid maize was left to others. For more on the early history of maize, see East and Jones (1919), Crabb (1947), Shull (1952), Wallace and Brown (1988), and Goldman (1998). The deepest theoretical analysis in the early period was not based on corn, but on Guinea pigs. Sewall Wright (1922a,b) studied an extensive series of crosses within and between long-time inbred lines. His

230

J. CROW

statistical analyses are remarkable; in particular, his adjustment for the environmental extremes of Guinea pig housing in the vicinity of Washington, D.C.-poorly heated winter quarters and absence of airconditioning in the notoriously hot, humid summers. Wright demonstrated the deterioration with inbreeding of almost every trait, including important ones related to fitness. He also noted the immediate recovery on crossing different inbreds. In particular, he could predict quantitativelythe performance of backcrosses and Fzs from knowledge ofinbreds and F i s. This was consistent with any level of dominance, but suggested that directional epistasis was not a major influence. The word, "heterosis," was first introduced by Shull (1914). Despite alternative explanations, he made clear that the word was intended to be purely descriptive, a synonym for hybrid vigor, and obviating such awkward expressions as "the stimulus of heterozygosis" (Shull 1948). Almost from the beginning, there were two different interpretations, and to them I now turn. III. ALTERNATIVE EXPLANATIONS OF HETEROSIS

The two competing hypotheses to explain heterosis were dominance and overdominance. The dominance hypothesis assumes that, in hybrids, detrimental recessives from one parent are masked by dominant alleles from the other. The overdominance hypothesis assumes that, at least at some loci, heterozygotes are superior to either homozygote. Overdominance has had a variety of names: stimulation of heterozygosis, superdominance, overdominance, single gene heterosis, cumulative action of divergent alleles, and sometimes simply heterosis. Actually, superdominance has priority (Fisher 1918), but overdominance has prevailed and I shall use it. Also, following Shull, I shall use heterosis simply as a descriptive term. Although the two hypotheses have been treated as opposites, they are not mutually exclusive. Both are consistent with inbreeding decline and heterosis, and distinguishing between the two is not easy. C. B. Davenport (1908) was the first to point out that recessive mutations usually have a weakening effect, and that this could explain the deteriorating consequences of inbreeding. He was thinking of a small number of genetic factors with individually large effects, in contrast to later views which have given more emphasis to polygenic inheritance. Shortly afterward, Keeble and Pellew (1910) found a particularly clear explanation of heterosis in garden peas. They crossed two cultivars that differed in two ways. One had a dominant gene for long internodes

5. THE RISE AND FALL OF OVERDOMINANCE

231

while the other had a dominant factor for more internodes. The hybrids had both long internodes and more of them, and hence were considerably taller than their parents, each of which was deficient in one or the other respect. Although only two loci were involved, the authors mentioned that the principle ought to hold for larger numbers. The same year, Bruce (1910) presented a more formal argument. He used p and q to stand for dominant and recessive allele frequencies in one strain and P and Q for the frequencies in the other. With random mating and linkage equilibrium, the proportion of the various multilocus genotypes in the two strains will then be given by the appropriate terms in the expansion of (p2 DD + 2pq DR + q2 RR)JJ and (P2 DD + 2PQ DR + Q2 RR)JJ, in which D and R stand for the dominant and recessive alleles and n for the number of loci. If these two populations are crossed, the mean number of homozygous recessive loci is nqQ, whereas the average for the two parent populations is n(q2 + Q2)/2. The expression (q2 + Q2)/2 can be written as qQ + (q- QF/2, which, unless q = Q, is larger than qQ. Hence, there will be fewer recessive homozygotes in the hybrid population. Or, as Bruce noted, this follows from the fact that the geometric mean is smaller than the arithmetic. Bruce then said: "If, now, it can be assumed that dominance is positively correlated with vigor, we have the final result that the crossing of two pure breeds produces a mean vigor greater than the collective mean vigor of the parent breeds." He went on to say, with admirable restraint, "I am aware that there is no experimental evidence to justify the assumption that dominance is correlated with a 'blending' character like vigor; but the hypothesis is not an extravagant one, and may pass until a better takes the field." I have two comments on Bruce's paper. First, although his algebra implied that the allele frequencies would be the same at all loci, this is not a necessary assumption. His argument still holds with unequal frequencies. Second, the same algebra that he employed to show a deficiency of recessive homozygotes in the hybrids could be used to show an excess of heterozygotes. He could as well have used this to argue for overdominance, although he would probably have regarded this as an "extravagant" hypothesis. In any case, his analysis is traditionally regarded as one of the strong early arguments for the dominance hypothesis. There were two early objections to the dominance hypothesis. The first is that, if the hypothesis is correct, it should be possible to select individuals that are homozygous for all the beneficial alleles and, therefore, to produce inbreds as good as hybrids. The second objection is that in the F2 generation, the dominant and recessive genotypes would be distributed according to the expansion of (3/4 + 1/4)n, where n is the

232

J. CROW

number of loci. Thus the distribution should be skewed, and no such skewness was observed. These objections were removed when Jones (1917) pointed out that, if a beneficial dominant were linked to a detrimental recessive, the association might not be readily broken up by recombination and the pair of linked loci would mimic a single overdominant locus; later, this was called pseudo-overdominance or associative overdominance. Thus, with linkage, the consequences of the dominance hypothesis were essentially the same as those postulating superior heterozygotes. Later, Collins (1921) emphasized that when n is large the skewness largely disappears, even without linkage. Furthermore, getting all the favorable alleles into one homozygous strain is extremely improbable unless the number of factors is very small. And all the evidence pointed to multiple factors, at least for yield. From that time on, the dominance hypothesis was strongly favored. Experience from a variety of organisms, especially Drosophila, showed emphatically that virtually all detectable mutations are harmful and that most of them are recessive. Furthermore, those harmful mutations that persist in the population will be overwhelmingly recessive, since harmful dominants are quickly eliminated by natural selection. So, to repeat Bruce's words, the "hypothesis is not an extravagant one." The view was clearly stated by Wright (1922b): Given the Mendelian mechanism of heredity, and this more or less perfect correlation between recessiveness and detrimental effect, and all the longknown effects of inbreeding-the frequent appearance of abnormalities, the usual deterioration in size, fertility, and constitutional vigor in the early generations, the absence of such decline in anyone or all these respects in particular cases, and the fixation of type and prepotency attained in later generations-are the consequences to be expected.

The alternative assumption of a stimulating effect of heterozygosity goes back to both Shull (1908,1911) and East (1908a,b). They assumed that there was a physiological stimulus to development which increased with the diversity of the uniting gametes. In the earlier writings, especially those of Shull, the effect was not necessarily assumed to be Mendelian. But later, this view came to be associated with intra-locus heterozygote superiority, or overdominance. A clear exposition of this view was given later by East (1936). He postulated a series of alleles, each of which has a positive function, and with these functions being to some extent cumulative. Thus the heterozygote would have something like the sum of the two homozygous effects. East

5. THE RISE AND FALL OF OVERDOMINANCE

233

argued that, as the functions became more and more divergent, they became more nearly additive in the heterozygote. I do not know whether East was intending to support the generality of overdominance, which was not widely accepted at that time, or was simply providing a rationale for such few overdominant loci as might occur. I think that probably the latter is correct. East accepted the linkage hypothesis earlier when Jones formulated it (East and Jones 1919). Furthermore, Hayes (1952), who knew East well, believed that he "continued also to accept the previous explanation that heterosis was dependent on the cumulative effect of dominant or partially dominant linked genes." From the time of Jones (1918,1922) and Collins (1921) until the middle to late 1940s, the dominance hypothesis was generally accepted. It depended on a well established fact, namely, the almost universally observed population correlation between recessivity and deleterious effect. Furthermore, with a large number of factors, some of them undoubtedly linked, the hypothesis seemed eminently reasonable. The overdominance hypothesis, in contrast, depended on a form of gene action that was thought to be rare at most. There were hardly any convincing examples of overdominant loci. Stadler (1939) noted that with some R alleles in maize the heterozygote had a larger pigmented area than either homozygote. But examples of important traits in which the heterozygote at a specific locus exceeds the better homozygote were essentially unknown. That overworked example of heterozygote superiority in fitness, sickle-cell hemoglobin, was not yet understood. Distinguishing between true overdominance and close repulsion linkage was usually impossible. Not surprisingly, the dominance hypothesis held sway. A. The Reappearance of Overdominance

Doubts about the adequacy of the dominance hypothesis began to appear in the 1940s. For one thing, mass selection for maize yield had been surprisingly ineffective. The yield of open-pollinated corn had hardly improved, despite selection for many decades. I should note, however, that most early selection efforts used open-pollinated cultivars, with no control over the pollen parent. One person, Fred Hull (1945, 1946), was increasingly impressed (and depressed?) by this failure, and began to entertain the possibility of overdominance. Others offered bits of evidence. Jones (1945) pointed out the possibility that some deleterious mutations may be beneficial in heterozygotes and Gustafsson (1947) and Brieger (1950) argued for heterozygote superiority.

234

J. CROW

Hull (1945) first introduced the word overdominance and urged its importance. Hull was a relative of J. 1. Lush, America's leading animal breeder, and this undoubtedly helped spread the gospel. Whoever was most influential, by the mid-1940s there was growing discussion of overdominance in maize-breeding circles and the word had caught on. Fisher (1918) had earlier used the word superdominance for heterozygote superiority, and I have preferred this word on grounds of priority and as being more consistent with Latin origins. It is curious that Hull (1945) coined a new word rather than using Fisher's, given Hull's great admiration for Fisher and familiarity with his papers in which superdominance was used. Hull's first argument noted the failure of mass selection to produce substantial increases in yield. He thought, quite correctly, that with complete dominance, and especially with partial dominance, selection should be effective. Yet none of the selection programs seemed to be making much progress. By the time of Hull's 1945 paper, hybrid maize was already producing substantially higher yields than standard cultivars and the yields were increasing steadily. Hull's second argument was naively simple. He noted that in most cases, the hybrid yield was more than the sum of the inbred parents. This would not be possible, he said, with simple dominance. The validity of this argument depends on the genes' acting in a strictly additive way, whereas they could perhaps be acting multiplicatively, or there could be epistasis. Furthermore, inbred lines lacked the homeostatic mechanisms of hybrids and were notoriously variable. Nevertheless, the model of simple additivity between loci provided a reasonably good fit to the data. Neal (1935) studied inbreds, along with 2-, 3-, and 4-way hybrids and their selfed progeny. The fit to additivity was excellent. As expected, the yield of selfed progeny of double-cross hybrids regressed 1/4 of the way to the average of the inbred parents, actually 25.8%. Curiously, Neal's three-way crosses gave a yield reduction of 36.6%, closer to 37.5% than his erroneous expectation of 33%. (When confronted with this, Neal laughingly said that at least his honesty was intact; he could not be accused of fudging the data in the direction of his expectation.) Stringfield (1950) reported similar results. Although he found evidence for epistasis in the double-cross hybrids, which did not yield quite as well as single crosses, the differences were small. If there was much epistasis, positive and negative effects approximately canceled. Jenkins's (1934) successful method for predicting double-cross hybrid yields from single-cross hybrid data assumed no epistasis.

5. THE RISE AND FALL OF OVERDOMINANCE

235

Hull's third argument carried more weight. He noted what he called the "Fisher equilibrium" (Fisher 1922), the maintenance of a stable polymorphism under overdominance. His argument utilized what he called "constant parent regression." The regression of the F1 on the value of one parent, when the other parent was held constant, had different expectations for dominance and overdominance. In particular, when the constant parent was high-yielding, the regression could curve and be negative for some values. This idea could be criticized on the grounds that any study involving inbreds had a large uncertainty and because different genes may be contributing to inbred and hybrid yield. Nevertheless, Hull's arguments made a substantial impact on the maize-breeding community. I first heard of overdominance as a hypothesis to be taken seriously at a statistics course at North Carolina State College in the summer of 1946. Accepting the sufficiency of the dominance hypothesis, I asked myself how much increase in yield would be expected if all deleterious recessive alleles could be eliminated. On the dominance hypothesis, this would represent the maximum increase in hybrids compared with the population from which the hybrids were derived. I reasoned that, because of a long history of selection, yield could be equated to fitness. Then, at equilibrium, the reduced fitness from recessive homozygotes should equal the genomic mutation rate. If qi is the frequency of a recessive allele at the i-th locus and Si is the deleterious effect of the recessive homozygote, then at equilibrium the number of genes eliminated by selection is siqi , which at equilibrium is balanced by the mutation rate Jli' Equating yield to fitness, the reduced yield from homozygous recessives is Sjqj or simply Jlj • Summing over all relevant loci the reduction is the total mutation rate per (haploid) genome at these loci. Since the total mutation rate for all loci, per genome was then believed to be about 5% or less, I thought that this represented the maximum increase in yield in hybrids compared to the open-pollinated cultivars from which the inbreds were derived; the actual increase would likely be somewhat smaller. This was considerably less than the 15% or more that was observed. I concluded that the overdominance theory might be correct after all. I discussed this idea with R. A. Fisher, who was the star attraction in the summer statistics course, and he agreed (after he had quickly verified it for himself). Later, he used the same idea. In his book (Fisher 1949, p. 118-119), he said: "... it would appear that the total elimination of deleterious recessives would make less difference to the yield of cross-bred commercial crops than the total mutation rate would suggest. Perhaps no

236

J. CROW

more than a 1 percent improvement could be looked for from this cause. Differences of the order of 20 percent remain to be explained." My results were published in September 1948 (Crow 1948). Fisher did not refer to this paper and my guess is that his 1949 book was written before my paper appeared, and he could well have forgotten our brief discussion in 1946. In any case, we both agreed that purging of deleterious recessives during inbreeding was insufficient to account for the large increase in hybrid performance. Later, and fortunately before submitting my 1948 paper, I learned that I had reinvented the wheel. The "mutation load" theory that I used had in fact been presented some years earlier by Haldane (1937). (If one is going to be scooped, it is much more pleasant if it is done by a person of such eminence as Haldane!) Incidentally, it was Sewall Wright who called my attention to the Haldane paper. Fisher apparently was not familiar with it. Earlier, Sprague and Tatum (1942) had coined the expressions general combining ability and specific combining ability and had shown how to measure them. In the 1940s three breeding systems designed to maximize the value of hybrids were advocated. The first involved selection for combining ability with a heterogeneous tester population (Jenkins 1940). This would select for general combining ability (Sprague and Tatum 1942). The second, proposed by Hull (1945), selected for specific combining ability by using a uniform tester, either an inbred or an Fl' The third, proposed by Comstock et al. (1949), was reciprocal recurrent selection. In this system, each population was selected for combining well with the other. The first breeding system (heterogeneous tester) should work well with partial or complete dominance. The second (uniform tester) was suitable if overdominance was important. The third (reciprocal recurrent selection) was designed to be effective with incomplete-, complete-, or overdominance, and therefore appeared to be the method of choice, given the uncertainty as to the dominance level.

B. The Apogee of Overdominance In 1950, the Iowa State College organized a five-week summer conference devoted to heterosis, and the papers were published two years later as a book (Gowen 1952). Many breeders had doubts about the adequacy of the dominance hypothesis and overdominance was in the air. Hull was at the Heterosis Conference and presented his by-thenstandard arguments along with additional data (Hull 1952). The arguments, some of which I have mentioned above, included: mass selection

5. THE RISE AND FALL OF OVERDOMINANCE

237

had been a failure; recombinants of parents of elite hybrids yielded little more than the first cycle; hybrids produced more than twice the yield of inbreds; the results of constant parent regression were consistent with overdominance; and selection for specific combining ability was effective. Although none of these was individually fully convincing, collectively they made a large impact on those in attendance. Hull had proposed earlier (1945) that recurrent selection for specific combining ability be given a more extensive test and this suggestion was repeated. Although there were skeptics, my memory is that overdominance was preferred by the majority of those at the conference, at least among those who had much to say. Overdominance was the Zeitgeist. Comstock and Robinson (1952) had devised three breeding experiments, of which the third (Experiment III) was the most informative as a way to measure dominance. In this breeding scheme, F z plants are backcrossed to the two inbred lines from which the hybrids were derived. From these data, Comstock and Robinson (1952) were able to estimate the degree of dominance. As a measure they used the symbol, a. This is 0 when the heterozygote is exactly intermediate between the two homozygotes (no domina~lce), is 1 for complete dominance, and is greater than 1 for overdominance. Their estimate, growing naturally out of analysis of variance, is a weighted average of a 2, the weights being the squared differences between the homozygotes. The measure of dominance, then, is the positive square root, noted by a. For yield, the reported value of a was 1.6. This is clearly in the overdominance range. In agreement with other workers, they found lower estimates of a for other traits, usually individual components of yield; these were often in the partial dominance range. . Dickerson (1952) reached similar conclusions, based mainly on swinebreeding data. He, too, noted the regularity of inbreeding decline and the ineffectiveness of selection in partially inbred lines. Heritability estimates were uniformly low, especially for total performance. He argued that the degree of dominance was greater and the heritability lower for total performance than for its individual components, which sometimes showed negative correlations with each other. The inability of selection to offset the decline in performance from mild inbreeding "casts doubt on the assumption that epistasis or ordinary dominance (between none and complete) can account for the major influence of inbreeding on performance in swine." Either overdominance or close repulsion linkages must be involved, he thought. My argument was essentially the mutation-load argument, presented in 1948 (Crow 1952). I'll not repeat it here, but will give instead a somewhat different formulation (Crow 1998b). The original population is

J.

238

CROW

assumed to be in Hardy-Weinberg proportions with p and q standing for the frequencies of the dominant A and recessive a alleles. One inbred is derived from the randomly mating population and is fixed for either A or a with probability p' and q'. The primes allow for the allele frequency to change during inbreeding. The other inbred line is assumed to be fixed for the dominant allele; this assumes that every recessive allele is covered and thus gives the maximum hybrid performance. The quantities, sand hs, are the reduction in yield, regarded as fitness, of recessive homozygotes and heterozygotes. Thus, h is a measure of dominance. Mean fitness

Fitness or frequency Genotype Relative fitness Frequency Random mating, R Inbred, I Other inbred Hybrid, H

AA 1

p2 p'

Aa 1- hs

aa l-s

2pq

q2 q'

W R = 1 - 2pqhs - q2 s W/= 1- q's

0 0

WH = 1- q'hs

1

0 0

p'

q'

The difference between the hybrid and random mating fitnesses is ~

= WH -

WR

= hsq + (1- 2h)sq2 + hs(q -

q').

[1]

Assume that the randomly mating population was at equilibrium between selection for A and mutation from A to a at a rate Il per generation. Then the equilibrium frequency of allele a is given by s(1 - 2h)q2 + hS(1 + Il)q - Il = 0

[2]

(Crow 1986, p. 86). Substituting [2] into [1] leads to ~

= 1l(1 -

hsq) + hs(q - q').

[3]

With complete dominance (h = 0), ~ = u, the same value as given originally. At the time of the Heterosis Conference, I was not yet aware of the ubiquity of partial dominance, so only complete recessivity was considered. With partial dominance, hsq Il (Crow 1986, p. 87). If q' = q, then ~ 1l(1- Il) Il, and the increase is the same as with complete dominance. It is likely, however, that q' has decreased somewhat by selection during inbreeding. If q' = 0, then ~ Il + hsq 21l. But this extreme is Z

Z

Z

Z

Z

5. THE RISE AND FALL OF OVERDOMINANCE

239

unlikely at more than a few loci, and the increase is probably closer to 1.1 than to 21.1. At the 1950 Heterosis Conference, I concluded (Crow 1952) that with complete dominance the maximum increase in hybrid yield compared with that of the open-pollinated ancestor is the genomic mutation rate, ~Il. Estimates at that time placed the value of ~1.1 at about 5%, probably less. The argument just given would say that the value with partial dominance is unlikely to be much greater, perhaps 6 or 7%, although I did not think of this in 1950. It should be clear, however, that this is an argument for overdominance as an explanation of the excess of hybrids over open-pollinated cultivars. There is no such limit on the increase of hybrids over inbreds. I concluded that the dominance hypothesis is fully adequate to explain inbreeding decline and recovery on hybridization (Crow 1952). At the Heterosis Conference, the overdominance theory attracted a great deal of interest. Dickerson (1952) suggested breeding systems that would maximize increase in performance with overdominance. Comstock et al. (1949) and Comstock and Robinson (1952) were advocating reciprocal recurrent selection. Several breeders started selection programs to capitalize on overdominance, or used reciprocal recurrent selection as a hedge against any level of dominance. These included not only maize breeders, but also swine and poultry breeders. Dobzhansky (1952) presented evolutionary arguments for overdominance. He distinguished between two forms of heterosis. The first was "mutational heterosis," due to sheltering of deleterious mutations by their dominant alleles. The second was "balanced heterosis," due to overdominance. In his words, "Balanced heterosis is an evolutionary contrivance that permits maintenance in a population of a multiplicity of genotypes that may be adaptive in different ecological niches which the population occupies." This was an early stage in Dobzhansky's drift toward his later, more extreme view. Rather than emphasizing that a small minority of overdominant loci can make a disproportionate contribution to the population variance, he later came to believe that overdominance was typical of the majority of loci and was obsessed by this idea for the rest of his life. A persistent criticism of the overdominance hypothesis was that it required a type of gene action for which convincing examples were hard to find. There were examples, such as the R alleles in maize that I mentioned earlier. Self-sterility alleles in plants were cited as examples, but these seemed special and more likely to involve frequencydependent selection than overdominance. Somewhat later, Brewbaker

240

J.

CROW

(1964, p. 81-85) described different kinds of gene action that can in principle lead to overdominance and gave some examples. One of these was a locus in flax conferring resistant to rust. Whereas each homozygote was resistant to a particular strain, the heterozygote was resistant to both. Another cited example was the blood-group locus where the AB genotype has both A and B antigens. East (1936) had given a possible mechanism, but, in general, examples were rare and sometimes contrived. On the other side, it was pointed out that only a small fraction of overdominant loci can make a major contribution to the variance of a randomly mating population or hybrids between inbred lines. This is because at classicallod, where the recessive (or nearly recessive) alleles are at a frequency determined by mutation-selection balance, the frequency of such alleles is very low. Overdominant loci, in contrast, are likely to have frequencies in an intermediate range, and each locus makes a much larger contribution to the variance (Crow 1952). So, since only a small fraction of the loci need be overdominant for these to dominate the genetic variance, the paucity of convincing individual cases seemed less important. But it was clearly a weak point in the argument. Not everyone at the Heterosis Conference hopped on the overdominance bandwagon. For example, Hayes (1952) had this to say: "Dominance or partial dominance seems of great importance as an explanation of hybrid vigor. In some cases there may be extra vigor correlated with the heterozygous condition of pairs of alleles." Sprague (1952) emphasized the uncertainty of current knowledge, and suggested that reciprocal recurrent selection might be the best hedge, since it should work well for any level of dominance. Later, Gardner et al. (1953) presented more data using Experiment III as a measure of dominance. The value of a for yield ranged from 1.31 to 2.14, all in the overdominance range. Yet, the authors were careful to point out that this might be pseudo-overdominance, not the real thing. Among other things, they emphasized that continuing the experiment for future generations with an opportunity for linkage equilibration at each cycle could distinguish between the two situations, and such experiments they began.

C. Doubts about Overdominance In the mid 1950s and early 1960s, several observations caused the climate of opinion to change. For one thing, mass selection in randomly mating cultivars began to be effective. The reasons are not completely clear. I think, however, that the main factor was simply more efficient selection. Early selection was largely based on ears, and the pollen was

5. THE RISE AND FALL OF OVERDOMINANCE

241

usually uncontrolled. Control of pollination had now become the rule. Another factor was control over the environment. By the early 1950s, R. A. Fisher's statistical methods and experimental designs had become routine. With better knowledge of the parentage by controlled pollination, with replicates and proper randomization, and with analysis of variance and covariance, selection was far more efficient than it had been. Once it was effectively utilized, additive variance for performance in maize turned out to be substantial. In 1956 Sprague reported the first results of an experiment he had devised earlier (Sprague and Russell 1956), which made clearly different predictions for the two hypotheses. Two maize populations were each selected for several generations for improved performance in hybrids with an inbred tester. With overdominance the two strains would tend to accumulate alleles complementary to the tester and similar to each other; the result would be decreased yield in the selected strains and in crosses between them. With partial or complete dominance, each of the populations should show increased yield, as should crosses between them. This is what was found. Sprague presented his early results in 1956 at an International Symposium in Tokyo and Kyoto (Sprague and Russell 1956). I heard his talk and this marked the beginning of my doubts about overdominance. Later generations rendered the conclusion completely convincing (see Sprague 1983). My own mutation load argument was open to the criticism that some ofthe assumptions might not be met. Equating yield with fitness and the assumption of mutation-selection equilibrium could be questioned. Also, whether the starting populations from which inbreds were derived were a single randomly mating population was questioned (see, for example, Hallauer and Miranda 1981, p. 350). But, more important from my standpoint, the argument fell flat with new and better estimates of the mutation rate. By the early 1960s, evidence for a higher rate of polygenic mutations in Drosophila had appeared. From long-time mutationaccumulation experiments, Mukai (1964) estimated the genomic mutation rate for viability genes several times greater than earlier estimates. These experiments were later repeated, with consistent results (Mukai et al. 1972). At the same time, analysis of variance for viability in Drosophila showed a large additive component and no evidence for any measurable contribution from overdominance. Furthermore, the mean persistence of mutant genes in the population was consistent with partial dominance, not overdominance (for a review, see Crow 1993). So, by the late 1950s I had abandoned the overdominance hypothesis. In the 1960s came increasing evidence for mutation rates in maize considerably higher than the 5% per genome that I had assumed. Both

J.

242

CROW

long-time inbreds and doubled monoploids (Chase 1949, 1952) showed considerable new genetic variability (Sprague et al. 1960; Russell et al. 1963). The inbred lines produced 2.8 detectable mutations per trait per generation per 100 gametes, the doubled haploids, 4.5. These are undoubtedly underestimates, since inconspicuous mutations would be missed. Adding the rates for all traits gives a value far in excess of 5 %. At about the same time, Experiment III of Comstock and Robinson (1952) was continued for successive generations, but before each round of mating there was an opportunity for recombination to occur. This should diminish the linkage disequilibrium and decrease the effect of repulsion-phase linkages. The results were decisive. Gardner (1963) found, as Comstock and Robinson (1952) had earlier, that the initial value of dominance, 0, was in the overdominance range-about 1.4. By the next generation it was barely above 1, indicating slight overdominance, but in the third and fourth cycles it was well into the partial dominance range (see Fig. 5.2). Moll et al. (1964) did a very similar experiment, with concordant results. The conclusion was clear: the apparent overdominance of the early generations was pseudooverdominance due to linkage disequilibrium. At the same time, methods of selection had become more and more effective. Inbred lines were much improved and, although not as good as

1.6 1.4 1.2

li

1.0 0.8 0.6 0.4 0.2 0.0 2

4

6

8

10

12

14

16

Generation Fig. 5.2. The change of average dominance, 0, in Experiment III as each cycle permits randomization of linkage disequilibrium. Dominance, a, is measured on a scale in which o is no dominance (heterozygote equals the mean of the homo zygotes), 1 is complete dominance, and >1 is overdominance. Data from C. O. Gardner, pers. commu

5. THE RISE AND FALL OF OVERDOMINANCE

243

hybrids, they were as good as the hybrids of an earlier period. Experiments designed to measure additive variance showed considerable amounts (summarized in Hallauer and Miranda 1981). Mass selection experiments became effective when the parentage was properly controlled and the experimental design and statistical controls were adequate to minimize environmental effects. The overdominance bubble had burst.

D. Dominance Takes the Field Since the 1960s, the evidence has strongly favored the dominance hypothesis. Even during the heyday of overdominance, it was realized that dominance is sufficient to account for inbreeding decline and recovery on crossing (Crow 1952). But increasing evidence, as summarized above, argues that additive and dominance effects are also the major factors in population variability and heterosis. See also Jinks (1983). We were back to the 1920s in our viewpoint. Sprague (1983) summarized the then state of knowledge. Studies have shown that additive and dominance gene effects are generally much greater than other types of gene effects. Additive effects are precisely those which respond to selection. Specially designed experiments have shown that both overdominance and epistasis exist, but neither has been shown to be important at the population level. ... Thus, as far as the maize breeder is concerned, a pragmatic solution to the dominanceoverdominance controversy has been reached. Additive and dominance effects provide a satisfactory model for the heterosis and for the rather remarkable progress achieved through breeding. Genetic variance estimates for populations under selection indicated little decrease in variability, thus giving assurance of further substantial progress.

Despite uncertainty about dominance levels, selection has been effective from the beginning. But with improved methodology, selection has become more efficient. Almost every system of selection works, but of course some work better than others. Many have tested reciprocal recurrent selection and found it to be effective (e.g. Betran and Hallauer 1996; Menz and Hallauer 1997). Coors (1998) has recently analyzed and summarized the results of many experiments by many workers, especially experiments designed to maximize heterosis. Selection schemes that are based on additive variance have been effective. Yet improvement of interpopulation crosses, which use dominance efficiently, has been faster. Despite the fact that selection cycles may take several generations, the rate of change per year is about 22% faster than simpler selection schemes (Coors 1998). Selection for combining ability is clearly effective.

244

J.

CROW

The general equation for selection of quantitative traits is:

where G is the gain per generation, K is the selection differential, k is the standardized selection differential, (J~ is the additive variance, (J2p is the total (phenotypic) variance, and n is the number of generations per selection cycle. For an explication, see Nyquist (1991, p. 272-278). This is of course modified by the form of selection. I wonder if, in one respect, this formula may perhaps be more accurate than its simple form would suggest. Kimura (1965) showed that if the recombination value is larger than the selective difference, which is true for most pairs of loci, the population, after several generations of selection, attains an approximately constant level of linkage disequilibrium; he called this quasi-linkage equilibrium. In this state, the variance due to linkage disequilibrium and that due to epistasis cancel almost exactly. Thus the progress under selection is given by the additive variance alone, with no appreciable contribution from epistatic components. For a discussion, see Crow and Kimura (1970, p. 195-204, 217-224). Some maize cultivars have been selected continuously for many generations. It seems likely that the Kimura principle holds. If so, this can provide a stronger justification for the convenient custom of ignoring epistasis in selection experiments. E. Is There a Role for Overdominance and Epistasis?

With the now-solid evidence that the great bulk of genetic variance is additive and dominance, is there now any contribution from epistasis and overdominance? I think it likely that the best hybrids, although they depend mainly on additive and dominance variance, may be getting an extra boost from epistasis. An experiment of Lamkey et al. (1995) bears on this point. Crosses of maize inbreds B73 or B84 with Mo17 produce particularly good hybrids. The experiments consisted of making F1 hybrids of B73 and B84, the F 2 , and the two backcrosses, which were then tested in crosses to Mo17. With no epistasis, there are certain expected relationships; for example, a backcross should be the average of the F 1 and the recurrent inbred. The data differed significantly from these expectations and, in particular, strains in which a parental genome remained intact were higher than expected. These, along with other comparisons, provided strong evidence for directional epistasis. Very similar results are reported by Wolf and Hallauer (1997) for B73 and Mo17. Others who reported evidence for epistasis were Bauman (1959),

5. THE RISE AND FALL OF OVERDOMINANCE

245

Sprague and Thomas (1967), Otsuka et al. (1972), Stuber et al. (1973), and Keeratinijakal and Lamkey (1993). Current methods of improvement, relying as they do on modifying inbred lines by introgression from closely related strains, offer a chance to maintain favorable epistatic combinations that improve the yield of inbreds and also that of the hybrids. So, even though the bulk of high performance is due to additive and dominance effects, it is quite likely that the very best hybrids are getting a small, but important boost from epistasis. That extra bit may be what makes them outshine their rivals. Evidence for overdominance is weaker. But perhaps the same idea applies. There is very little evidence for overdominance at the population level. Yet, the elite crosses may be getting an extra increment from a few overdominant loci. But, distinguishing between overdominance and pseudo-overdominance with very tight linkage is usually impossible by conventional methods. The best hope is with molecular methods, although with individually small effects this will be difficult. After all the years since Shull's discovery in 1908, there are still very few convincing examples of overdominance that might serve as models for heterosis. Scandalios et al. (1972) noted intra-allelic complementation and discussed this as a possible basis for heterosis. Yet the fact remains that, despite all the complexities uncovered by molecular analysis (see below), good models for single-gene heterosis are elusive. But see Brewbaker (1964) for some possible examples of overdominance. In 1983 Sprague said: "It appears reasonable to assume that the types of gene action and interaction involved in the expression of heterotic traits will be at least as varied and complex as those which have appeared for quantitative traits." The remarkable thing is that, despite our ignorance of the niceties of gene action and interaction, the methods of selection, inbreeding, and hybridization have been so extremely effective. IV. WHY ARE HYBRIDS SO GOOD?

Starting with the widespread use of double-cross hybrids, maize yields in the United States grew rapidly. By 1960, inbred lines were good enough that single crosses quickly replaced double crosses. As I mentioned earlier, current yields are five times what they were before hybrids took over (Fig. 5.1). Meanwhile, randomly mated cultivars have gotten better, but the best of them still are not as good as the best hybrids. With complete dominance, and especially with partial dominance, mass selection should be effective. With control of parentage and good

246

J.

CROW

experimental design, progress should be fairly rapid, even for a trait with low heritability. Improved selection has certainly worked well for individual traits, and more recently, also for yield. Long ago, Fisher (1949, p. 119 ff) wrote this: It may be noted first that "hybrid corn" has been an immense success in

that species, in which thousands of inbred lines have been produced. In species in which so far only a few inbred lines have become available, success has not been conspicuous.

When many homozygous lines are available the conditions for effective selection seem to be at their best. (i) We are selecting the actual genotype to be used for production, not merely an ancestor of it. (ii) Owing to the reliability of breeding performance achieved by inbreeding, lots of all sizes will be available for testing. Many crosses get no further than a first inspection, but promising crosses may be tested in quantity with all the precision which modern experimental design makes possible. (iii) Any special advantage, limited perhaps by locality or by industrial use, remains a permanent property of the hybrid, which will reappear when it is made up. It can later be produced in the quantity appropriate to the special role it is to play. Every careful determination of quality is a permanent contribution to the optimal utilization of the material. Fisher of course did not know of all the experimental work that has been done to optimize the making of good hybrids. In particular, he didn't anticipate that future progress in increasing performance would entail improving existing inbreds rather than starting anew with a great multiplicity of new ones. Within the working framework, there is room for a wide variety of selection programs. Testing can be early or late. Improved inbred lines are important for their own sake. New sources of germ plasm can be brought in. Agronomic characters can be improved. The list is endless. But the general framework is established. A striking increase was realized when single crosses became practical. There are several possible reasons. One is efficiency of selection. Cockerham (1961) showed that with additive effects selection is twice as effective with single crosses as with double crosses; three-way crosses are intermediate. With dominance, the discrepancy is greater. An additional factor, for which there is evidence (see above), is that epistatic factors may be playing a role. Favorably interacting genes in an inbred remain intact in the single-cross hybrids, but are broken up in double crosses. For all these, and undoubtedly other reasons, single crosses have produced a major change in the steady rate of yield improvement.

5. THE RISE AND FALL OF OVERDOMINANCE

247

V. PHYSIOLOGICAL AND MOLECULAR STUDIES Naturally, heterosis has attracted a great deal of attention from physiological and developmental geneticists. Mangelsdorf (1952) emphasized physiological bottlenecks, the removal of which, perhaps by a better allele, can improve the physiological process. The problem was how to remove one bottleneck without introducing others, so as to maintain the metabolic balance. At the enzyme level, alleles are additive, but metabolic flux is expected to be nonlinear, leading to various levels of dominance and epistasis. Kaeser and Burns (1983), in a highly influential paper, formulated a quantitative relationship between metabolic flux and enzyme levels as determined by gene dosage. Thus dominance is a consequence of nonlinearity in the relation of metabolic flux to enzyme level. For a recent update of the Kaeser-Burns theory, see Keightley (1997). Rhodes et al. (1993), influenced by Kaeser and Burns, measured the amounts of a large number of amino acids and metabolic flux in different maize hybrids. A small difference in metabolic flux can lead to substantial heterosis, despite additivity at the enzyme level. Rhodes et al. detected flux alterations by changes in the pool size of nitrogenous solutes, and variability in these was correlated with hybrid grain yield. They also suggested that quantifying genetic distance by metabolic profiling may be a way to quantify genetic distances and thereby predict promising hybrids. So far it has not been practical to associate yield changes with specific loci, but as individual loci become identified by molecular markers, this limitation will be overcome. Then the association of a specific locus with a specific solute pool and ultimately with grain yield may be possible, along with a deeper understanding of a possible metabolic basis for heterosis. The Kaeser-Burns theory is one way to go. This is a time of rapid progress in molecular understanding. No doubt, heterosis will be a beneficiary of this exciting new technology. Eventually, probably soon, we shall have the complete DNA sequence of maize, although it may follow after rice or other plants with less DNA. Sequence knowledge is certain to be valuable, especially in identifying useful qualitative traits. How practical it will be for polygenic traits is less certain, although eventually it will surely make an impact. Without waiting for full sequence information, much can be learned from expressed sequence tags. Distinction between overdominance and pseudo-overdominance will surely be one benefit of new techniques. A specific use of molecular markers has been to assign strains to appropriate heterosis groups (Stuber 1994). At the moment there is great interest in identifying quantitative trait loci. Stuber et al. (1992) analyzed two elite inbred lines and their

248

J.

CROW

hybrids, along with intercrosses and backcrosses. The experiment was extensive, using 76 chromosome markers that covered some 90 to 95% of the genome. The experimental design included replicates in six environments in three states. Almost every chromosome was found to contain at least one quantitative trait locus (QTL) with a significant effect on grain yield. The experiments were designed to maximize the ability to detect QTLs affecting heterosis. Collectively, the identified QTLs contributed a substantial fraction to the heterosis in the cross. The analysis revealed very little genotype x environment interaction, which is surprising given the range of environments tested. In addition almost all the QTLs appeared to show overdominance. But, we must remember that the size of the units identified was large, so it is not possible to distinguish between overdominance and pseudo-overdominance. An analysis by Cockerham and Zeng (1996) argued that the seeming overdominance is due mainly, if not entirely, to repulsion linkages, as would be expected. In particular, one seemingly overdominant QTL has been resolved into two loci in repulsion linkage (Graham et al. 1997). But finding QTLs is only the first step (Xu 1997). It should be possible to close in on the active sites and see ifthey are few or many. It seems certain that molecular analysis will eventually demonstrate how much of a role, if any, overdominance plays. If this is appreciable, the relevant loci, which are almost certain to be few, can be identified. More immediately promising, I suspect, is the identification of epistatic combinations which may make the best hybrids still better. The maize genome has turned out to be very complicated. Transposable elements are clearly playing a role (see, for example, Jayaram and Peterson 1990). Duplications, methylation, and transpositions all add to the complication. The polyploidization in the ancestry of maize means that a number of loci are redundant, or partially so. I mentioned earlier Stadler's (1939) noting that some mutants at the R locus show a form of overdominance. This locus has now been sequenced and includes virtually every known complication-paramutation, translocations, inverted sequences, imprinting, epigenetic silencing, amplification, and transposition-and it is highly polymorphic in cultivated strains (Kermicle 1996). There is no reason to think that this locus is unique. Finally, cytoplasmic male sterility (Rhoades 1931, 1933) was widely employed until the disastrous Helminthosporium epidemic of 1970, but surely there is a further role for mitochondrial genetics. Maize, of course, is the species in which transposable elements were first discovered and exploited by Barbara McClintock. The variability in colors and color patterns in both ears and plants has been known from the beginning. Some of the Native Americans who domesticated maize

5. THE RISE AND FALL OF OVERDOMINANCE

249

clearly selected for colorful and variegated ears. They may well have been selecting indirectly for transposons. These have now been discovered in one species after another, but maize would appear to be excessively loaded with such elements. Are they an impediment to yield improvement, or a possible source of useful variation? Rasmusson and Phillips (1997) argue that plant breeders can achieve substantial improvements by using new molecular discoveries. For example, gene amplification and transposable elements can enhance genetic variability. Methylation is associated with changes in mutation rates. Rasmusson and Phillips believe that these and similar sources can supply additional selectable genetic variability that can be utilized to enhance performance. How immediately useful this kind of molecular trickery will be is in doubt. That it will be tried is surely not in doubt. That such techniques will ultimately be useful, I am sure. But we know that essentially all mutations are harmful from the standpoint of fitness, and therefore probably from the standpoint of yield. They might be useful for new and improved qualitative traits, however. At the moment, the best source of additional quantitative variability for improving inbred lines is genes introduced from other strains, often those very closely related to the one being improved. And, at the moment, the rate of improvement seems to be undiminished. So, the value of newer techniques for generating genetic variability for yield improvement remains to be demonstrated. There are two contrasting objectives. One is, having obtained a particularly good inbred, to maintain it with as few changes as possible. Alternatively, if one wants to improve the line further, new variability is needed. For the first, it would be best to reduce mutability, perhaps by choosing strains with a minimum of methylation and the fewest transposons. For the second objective, genetic variability is needed. There is already a great deal of genetic variability in maize, usually more than the capacity of breeders to evaluate and utilize (Paterniani 1990). Whether new molecular sources will prove useful remains to be seen; but the promise is there.

VI. IS THE HYBRID BREEDING METHOD BEST?

A. Maize It has been suggested, especially by those with a particular political viewpoint (Berlan and Lewontin 1986; Kloppenberg 1988), that hybrid

maize is a calculated device for enriching seed producers and equipment

250

J.

CROW

manufacturers at the expense of maize growers, who must buy expensive seed each season. They argue that if the same amount of effort and resources had gone into within-population selection the results might have been just as good. The stronger the evidence for dominance and especially for partial dominance, the more plausible is such a viewpoint. It is fair to say that no such extensive breeding program has been carried out with randomly mating cultivars as has been done for hybrids. If dominance and partial dominance are the whole story, then it should be possible eventually to produce populations and possibly even inbreds equal to the best hybrids. But, if so, it might take a very long time. Recently there has been much improvement in selected open-pollinated populations, but they are still well behind the best hybrids. Duvick (1996) has provided a thoughtful view. "There are theoretical reasons, backed by minimal amounts of data, to believe that with equivalent starting materials and equivalent inputs of time, people, and technology, the breeding methods for population improvement that are now available might be as effective as the inbred-hybrid method for making certain kinds of genetic improvement." But he goes on to say that the methods now available to the breeder were not available in the 1930s when the hybrid maize programs began. The controlling argument was that advanced by Fisher (1949), namely that with the inbred-hybrid procedure, when a particularly happy gene combination occurs it can be identified, tested in replicates, and multiplied to whatever extent is desired. And there were the additional benefits of uniformity and control over qualitative traits. Selection methods are now much better. Methods such as the best linear unbiased prediction (BLUP) can now be used to maximize selection efficiency (Bernardo 1996). This permits optimum use of individual merit and information from relatives, all done under conditions of optimum field designs. Extensive replication is now de rigueur, and has been for some time. We are now in a period when there is rapid improvement of maize in many areas of the globe. Hybrid programs and efficient intra-population selection regimes are available at the start, and can therefore be compared directly. Tropical climates present a particular challenge. Selection for high yield under conditions of drought, poor soil, and no fertilizer is not the same as producing high-yielding hybrids for Iowa. It may well turn out that selection within locally adapted land races will be a better procedure than the hybrid method. The important point is that all selection schemes should be available and the one optimum for the particular locale can be adopted (Pandey and Gardner 1992). It

5. THE RISE AND FALL OF OVERDOMINANCE

251

remains to be seen how closely the development of maize in tropical climates will mimic what happened in the American corn belt. B. Other Crop Species

This discussion has been confined to maize. This is for good reason, since most of the experimental evidence about heterosis has come from this species. Yet many other crops show substantial heterosis; others do not. Sorghum, oats, millet, rape seed, soybeans, cowpeas, sunflowers, wheat, rice and cotton, along with trees and vegetable crops are all being studied (Pandey and Gardner 1992). There is every reason to expect that the results will not be the same for all species. Whether the maize paradigm is best remains to be seen. For example, doubled haploids have not had much impact on maize breeding, but this may not be true for other species (Raina 1997). In the summer of 1997, another Heterosis Conference was held, this time in Mexico City (Coors et al. 1998). It is remarkable that there has been so much progress in breeding better strains in one crop after another. A combination of existing genetic knowledge, new breeding methods (such as apomixis in some species), better management practices geared to tropical climates and different social systems, can go a long way in keeping up with the ever-increasing worldwide need for more food. Among those crops in which heterosis is large, the question of overdominance arises. Such things as differences in amount of selffertilization and in ploidy may dictate different selection regimens. Although the role of overdominance will eventually be analyzed, it is a good prediction that there will be pragmatic solutions and rapid progress in the near future, as has been the case with maize, and that genetic understanding will come later. VII. CONCLUSIONS

The deteriorating effects of inbreeding and the vigor of hybrids has been known since classical antiquity. Case after case was reported, especially in plants, before and throughout the nineteenth century. A genetical interpretation, however, had to wait until the rediscovery of Mendel's laws in 1900. Two main hypotheses appeared early and, although not mutually exclusive, have provided alternative explanations. The overdominance

252

J. CROW

hypothesis assumes the existence of loci at which the heterozygote is superior to either homozygote. Hybrids, being more heterozygous than inbreds, therefore have better performance.The dominance hypothesis is based on the observed correlation between recessiveness and deleterious effect. Inbreeding produces deterioration by exposing recessive genes through homozygosity; hybrids have increased performance as dominant alleles from each parent conceal deleterious recessives from the other. In the early 1900s, the time of the earliest inbreeding and hybridization experiments in maize, the overdominance hypothesis was preferred. A principal reason was the failure of selection to offset the deteriorating effects of inbreeding. By 1920 it was generally realized that with multiple factors and linkage, such a result was to be expected and the dominance hypothesis became the favored one. Another reason was the continued failure to find good examples of overdominant loci. In the 1940s there was another switch of opinion. The failure of selection to improve randomly mated populations and the results of experiments to measure dominance pointed toward overdominance, and by the time of a heterosis conference held at Iowa State College in 1950, overdominance was in the air (Gowen 1952). A final switch began in the late 1950s and early 1960s, when mating systems that had previously indicated overdominance showed only partial dominance after a chance for randomization by recombination. At the same time, some experiments designed to make a direct comparison pointed toward dominance and partial dominance. Additive and dominance components dominated the analysis of variance. This evidence has been strengthened in the ensuing years, and the dominance hypothesis, in essentially the original form of the 1920s is now accepted. This does not, however, rule out the possibility that a small contribution may come from some overdominant loci and that these may give a small, but important boost to the best hybrids. There is stronger evidence for epistasis, and it has been demonstrated in several experiments. The amount is small, but it is quite likely that the best· single-cross hybrids have a small, but important increase from favorable interactions in the inbred parents.

LITERATURE CITED Alliprandini, L. F., A. P. Duarte, and R. A. B. Kanthock. 1998. Genetic gain in commercial maize in summer and autumn-winter crops in the Paranapanema River Valley, Brazil, 1992-1997. Maydica 43:55-64. Bauman, L. F. 1959. Evidence of non-allelic gene interaction in determining yield, ear height, and kernel row number in corn. Agron. J. 51:531-534.

5. THE RISE AND FALL OF OVERDOMINANCE

253

Beal, W. J. 1880. Indian corn. Rept. Michigan State Board Agr. 19:279-289. Berlan, J. P., and R. Lewontin. 1986. The political economy of hybrid corn. Monthly Rev. 38:35-47. Bernardo, R. 1996. Best linear unbiased prediction of maize single-cross performance. Crop Sci. 36:50-56. Betnin, F. J., and A. R. Hallauer. 1996. Hybrid improvement after reciprocal recurrent selection in BSSS and BSCBl maize populations. Maydica 41:25-33. Brewbaker, J. L. 1964. Agricultural genetics. Prentice-Hall, Englewood Cliffs, NJ. Brieger, F. G. 1950. The genetic basis of heterosis in maize. Genetics 35:420-445. Bruce, A. B. 1910. The Mendelian theory of heredity and the augmentation of vigor. Science 32:627-628. Chase, S. S. 1949. Monoploid frequencies in a commercial double cross hybrid maize, and its component single cross hybrids and inbred lines. Genetics 34:328-332. Chase, S. S. H.15~. Monoploids in maize. p. 389-399. In: J. W. Gowen (ed.), Heterosis. Iowa State College Press, Ames. Cockerham, C. C. 1961. Implications of genetic variances in a hybrid breeding program. Crop Sci. 1:47-52. Cockerham, C. c., and Z.-B. Zeng. 1996. Design III with marker loci. Genetics 143: 1437-1456. Collins, G. N. 1909. The importance of broad breeding in corn. USDA Bureau of Plant Industry Bu!. 141, Part 4:33-42. Collins, G. N. 1921. Dominance and the vigor of first generation hybrids. Am. Nat. 55:116-133. Comstock, R. E., and H. F. Robinson. 1952. Estimation of the average dominance of genes. p. 494-516. In: J. F. Gowen (ed.), Heterosis. Iowa State College Press, Ames. Comstock, R. E., H. F. Robinson, and P. H. Harvey. 1949. A breeding procedure designed to make maximum use of both general and specific combining ability. Agron. J. 41:360-367. Coors, J. G. 1998. Selection methodologies and heterosis. In: J. G. Coors, S. Pandey, M. V. Ginkel, A. R. Hallauer, D. C. Hess, K. R. Lamkey, A. E. Melchinger, G. Srinivasan, and C. W. Stuber (eds.), The genetics and exploitation of heterosis in crops: Proc. Int. CIMMYT Symp. ASA, CSSA, SSSA, Madison, WI. Coors, J. G., S. Pandey, M. V. Ginkel, A. R. Hallauer, D. C. Hess, K. R. Lamkey, A. E. Melchinger, G. Srinivasan, and C. W. Stuber (eds), 1998. The genetics and exploitation of heterosis in crops: Proc. Int. CIMMYT Symp. ASA, CSSA, SSSA, Madison, WI. Crabb, A. R. 1947. The hybrid corn makers: prophets of plenty. Rutgers Univ. Press, New Brunswick, NJ. Crow, J. F. 194ft Alternative hypotheses of hybrid vigor. Genetics 33:477-487. Crow. J. F. 1952. Dominance and overdominance. p. 282-297. In: J. W. Gowen (ed.), Heterosis. Iowa State College Press, Ames. Crow, J. F. 1986. Basic concepts in population, quantitative, and evolutionary genetics. W. H. Freeman, New York. Crow, J. F. 1993. Mutation, mean fitness, and genetic load. Oxford Surveys in Evo!' Bio!' 9:3-42. Crow, J. F. 1998a. 90 years ago: The beginning of hybrid maize. Genetics 148:923-928. Crow, J. F. 1998b. Dominance and overdominance. In: J. G. Coors, S. Pandey, M. V. Ginkel. A. R. Hallauer, D. C. Hess, K. R. Lamkey, A. E. Melchinger, G. Srinivasan, and C. W. Stuber (eds.), The genetics and exploitation of heterosis in crops: Proc. Int. CIMMYT Symp. ASA, CSSA, SSSA, Madison, WI. Crow, J. F., and M. Kimura. 1970. An introduction to population genetics theory. Harper and Row, New York. Reprinted by Burgess Int. Group, Edina, MN.

254

J. CROW

Darwin, C. 1862. On the various contrivances by which British and foreign orchids are fertilised by insects, and on the good effects of intercrossing. John Murray, London. Darwin, C. 1876. The effects of cross and self fertilization in the vegetable kingdom. John Murray, London. Davenport, C. B. 1908. Degeneration, albinism and inbreeding. Science 28:454-455. Dickerson, G. 1952. Inbred lines for heterosis tests? p. 330-351. In: J. W. Gowen (ed.), Heterosis. Iowa State College Press, Ames. Dobzhansky, Th. 1952. Nature and origin of heterosis. p. 218-223. In: J. W. Gowen (ed.), Heterosis. Iowa State College Press, Ames. Duvick, D. 1992. Genetic contributions to advances in yield of U.S. maize. Maydica 37:69-79. Duvick, D. 1996. Plant breeding, an evolutionary concept. Crop Sci. 36:539-548. East, E. M. 1908a. Inbreeding in corn. Rept. Conn. Agr. Expt. Sta. for 1907. p. 419-428. East, E. M. 1908b. Hybridization methods in corn breeding. Rep. Am. Breed. Assoc. 6:63-72. East, E. M. 1936. Heterosis. Genetics 21:375-397. East, E. M., and D. F. Jones. 1919. Inbreeding and outbreeding: Their genetic and sociological significance. Lippincott, Philadelphia, PA Fernandes, J. S. c., and J. F. Franzon. 1997. Thirty years of genetic progress in maize (Zea mays 1.) in a tropical environment. Maydica 42:21-27 Fisher, R A. 1918. The correlation between relatives on the supposition of Mendelian inheritance. Trans. Roy. Soc. Edinb. 52:399-433. Fisher, R. A. 1922. On the dominance ratio. Proc. Roy. Soc. Edinb. 52:321-341. Fisher, R A. 1949. The theory of inbreeding. Oliver and Boyd, Edinburgh. Gardner, C. O. 1963. Estimates of genetic parameters in cross fertilizing plants and their implications in plant breeding. p. 225-252. In: W. D. Hanson and H. F. Robinson (eds.), Statistical genetics and plant breeding. Special Pub. 982, NAS-NRC, Washington, D.C. Gardner, C. 0., P. H. Harvey, R. E. Comstock, and H. F. Robinson. 1953. Dominance of genes controlling quantitative characters in maize. Agron. J. 45:186-191. Goldman, 1. 1. 1998. From out of old fields comes all this new corn: An historical perspective on heterosis in plant improvement. p. 1-12. In: K. R. Lamkey and J. E. Staub (eds.), Concepts and breeding of heterosis in crop plants. CSSA Special Pub., Madison, WI.

Gowen, J. W. 1952. Heterosis. A record ofresearches directed toward explaining and utilizing the vigor of hybrids. Iowa State College Press, Ames. Graham, G. 1., D. W. Wolff, and C. W. Stuber. 1997. Characterization of a yield quantitative trait locus on chromosome five of maize by fine mapping. Crop Sci. 37:1601-1610. Gustafsson, A. 1947. The advantageous effect of deleterious mutations. Hereditas 33:573575. Haldane, J. B. S. 1937. The effect of variation on fitness. Am. Nat. 71:337-349. Hallauer, A. R, and J. B. Miranda. 1981. Quantitative genetics and maize breeding. 2nd edition, 1988: Iowa State Univ. Press, Ames. Hayes, H. K. 1952. Development of the heterosis concept. p. 49-65. In: J. W. Gowen (ed.), Heterosis. Iowa State College Press, Ames. Hull, F. H. 1945. Recurrent selection for specific combining ability in corn. J. Am. Soc. Agron.37:134-145. Hull, F. H. 1946. Overdominance and corn breeding where hybrid seed is not feasible. J. Am. Soc. Agron. 38:1100-1103. Hull, F. H. 1952. Recurrent selection and overdominance. p. 451-473. In: J. W. Gowen (ed.), Heterosis. Iowa State College Press, Ames.

5. THE RISE AND FALL OF OVERDOMINANCE

255

Jayaram, c., and P. A. Peterson. 1990. Anthocyanina pigmentation and transposable elements in maize aleurone. Plant Breed. Rev. 8:91-137. Jenkins, M. T. 1934. Methods of estimating the performance of double-crosses in corn. J. Am. Soc. Agron. 26:199-204. Jenkins, M. T. 1940. The segregation of genes affecting yield of grain in maize. J. Am. Soc. Agron. 32:55-63. Jinks, J. L. 1983. Biometrical genetics of heterosis. p. 1-46. In: R Frankel (ed.), Heterosis. Springer-Verlag, Berlin, Heidelberg. Jones, D. F. 1917. Dominance of linked factors as a means of accounting for heterosis. Genetics 2:466-479. Jones, D. F. 1918. The effects of inbreeding and crossbreeding upon development. Conn. Agr. Expt. Sta. But. 107. Jones, D. F. 1922. The productiveness of single and double first generation corn hybrids. J. Am. Soc. Agron. 14:242-252. Jones, D. F. 1945. Heterosis resulting from degenerative changes. Genetics 30:527-542. Kaeser, H., and J. A. Burns. 1983. The molecular basis of dominance. Genetics 97:639-666. Keeble, F., and C. Pellew. 1910. The mode of inheritance of stature and of time of flowering in peas (Pisum sativum). J. Genet. 1:47-56. Keeratinijakal, V., and K. R Lamkey. 1993. Responses to reciprocal recurrent selection in BSSS and BSCBI maize populations. Crop Sci. 33:73-77. Keightley, P. D. 1997. A metabolic basis for dominance and recessivity. Genetics 143:621-625.

Kermicle, J. L. 1996 Epigenetic silencing and activation of a maize r gene. Epigenetic mechanisms of gene regulation. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Kimura, M. 1965. Attainment of quasi linkage equilibrium when gene frequencies are changing by natural selection. Genetics 52:875-890. Kloppenburg, J. R 1988. First the seed. Cambridge Univ. Press, Cambridge. K6lreuter, J. G. 1766. VorHiufigen Nachricht von einigen das Geschlecht der Pflanzen betreffenden Versuchen und Beobachtungen. Leipzig. Lamkey, K. R, B. J. Schnicker, and A. E. Melchinger. 1995. Epistasis in an elite maize hybrid and choice of generation for inbred line development. Crop Sci. 35:1272-1281. Mangelsdorf, A. J. 1952. Gene interaction and heterosis p. 320-329. In: J. W. Gowen (ed.), Heterosis. Iowa State College Press, Ames. Mendel, G. 1865. Versuche iiber Pflanzen-Hybriden. Naturf. Ver. in Brunn Verh. 4:3-47. Menz, M. A., and A. R Hallauer. 1997. Reciprocal recurrent selection of two tropical corn populations adapted to Iowa. Maydica 42:239-246. Moll, R H., M. F. Lindsey, and H. F. Robinson. 1964. Estimates of genetic variances and level of dominance in maize. Genetics 49:411-423. Mukai, T. 1964. Spontaneous mutation rate of polygenes controlling viability. Genetics 50:1-19.

Mukai, T., S. 1. Chigusa, L. E. Mettler, and J. F. Crow. 1972. Mutation rate and dominance of genes affecting viability in Drosophila melanogaster. Genetics 72:335-355. Naudin, C. 1865. Nouvelles recherches sur l'hybridite dans les vegetaux. Nouv. Arch. Mus. Hist. Nat. Paris 1:25-174. Neal, N. P. 1935. The decrease in yielding capacity in advanced generations of hybrid corn. J. Am. Soc. Agron. 27:666-670. Nelson, O. E. 1993. A notable triumvirate of maize geneticists. Genetics 135:937-941. Nyquist, W. E. 1991. Estimation of heritability and prediction of selection response in plant populations. Crit. Rev. Plant Sci. 10:235-322.

256

J. CROW

Otsuka, Y., S. A. Eberhart, and W. A. Russell. 1972. Comparisons of prediction formulas for maize hybrids. Crop. Sci. 12:325-331. Pandey, S., and C. O. Gardner. 1992. Recurrent selection for population, variety, and hybrid improvement in tropical maize. Adv. Agron. 48:1-87. Paterniani, E. 1990. Maize breeding in the tropics. CRC Crit. Rev. Plant Sci 9:126-154. Raina, S. 1997. Doubled haploid breeding in cereals. Plant Breed. Rev. 15:141-186. Rasmusson, D. c., and R. L. Phillips. 1997. Plant breeding progress and genetic diversity from de novo variation and elevated epistasis. Crop Sci. 37:303-310. Rhoades, M. M. 1931. Cytoplasmic inheritance of male sterility in Zea mays. Science 73:340-341.

Rhoades, M. M. 1933. The cytoplasmic inheritance of male sterility in Zea mays. J. Genet. 27:71-93.

Rhodes, D., G. C. Ju, W.-J. Yang, and Y. Samaras. 1993. Plant metabolism and heterosis. Plant Breed. Rev. 10:53-91. Robinson, H. F., R. E. Comstock, and P. H. Harvey. 1949. Estimates of heritability and the degree of dominance in corn. Agron. J. 41:353-359. Russell, W. A., G. F. Sprague, and L. H. Penny. 1963. Mutations affecting quantitative characters in long time inbred lines of maize. Crop Sci. 3:175-178. Scandalios, J. G., E. H. Uu, and M. A. Campeau. 1972. The effects of intragenic and intergenic complementation on catalase structure and function in maize: a molecular approach to heterosis. Arch. Biochem. Biophy. 153:695-705. Shull, G. H. 1908. The composition of a field of maize. Rep. Am. Breeders' Assoc. 4:296-301.

Shull, G. H. 1909. A pure-line method in corn breeding. Rept. Am. Breeders' Assoc. 5:51-59.

Shull, G. H. 1911. The genotypes of maize. Amer. Nat. 45:234-252. Shull, G. H. 1914. Duplicate genes for capsule form in Bursa bursa-pastoris. Z. indo Abst. Vererb. 12:97-149. Shull, G. H. 1948. What is "heterosis"? Genetics 33:439-446. Shull, G. H. 1952. Beginnings of the heterosis concept. p. 14-48. In: J. W. Gowen (ed.), Heterosis. Iowa State College Press, Ames. Sprague, G. F. 1952. Early testing and recurrent selection. In: J. W. Gowen (ed.), Heterosis. Iowa State College Press, Ames. Sprague, G. F. 1983. Heterosis in maize: Theory and practice. p. 47-70. In: R. Frankel (ed.), Heterosis. Springer-Verlag, Berlin, Heidelberg. Sprague, G. F., and W. A. Russell. 1956. Some evidence on type of gene action involved in yield heterosis in maize. p. 522-526. Proc. Int. Genet. Symp., Tokyo and Kyoto. Sprague, G. F., W. A. Russell, and L. H. Penny. 1960. Mutations affecting quantitative traits in selfed progeny of doubled monoploid maize stocks. Genetics 45:855-865. Sprague, G. F., and L. A. Tatum. 1942. General vs. specific combining ability in single crosses of corn. J. Am. Soc. Agron. 34:923-932. Sprague, G. F., and W. I. Thomas. 1967. Further evidence of epistasis in single and threeway cross yields of maize. Crop. Sci. 7:355-356. Stadler, L. J. 1939. Some observations on gene variability and spontaneous mutation. Sprague Memorial Lectures, Michigan State College. p. 1-16. Stringfield, G. H. 1950. Heterozygosis and hybrid vigor in maize. Agron. J. 42:145-152. Stuber, C. W. 1994. Heterosis in plant breeding. Plant Breed. Rev. 12:227-251. Stuber, C. W., S. E. Lincoln, D. W. Wolff, T. Helentjaris, and E. S. Lander. 1992. Identification of genetic factors contributing to heterosis in a hybrid from two elite maize inbred lines using molecular markers. Genetics 132:823-839.

5. THE RISE AND FALL OF OVERDOMINANCE

257

Stuber, C. W., W. P. Williams, and R. H. Moll. 1973. Epistasis in maize (Zea mays 1.). Crop Sci. 13:195-200. Wallace, H. A., and W. 1. Brown. 1988. Corn breeding and its early fathers. Rev. ed. Iowa State College Press, Ames. Wolf, D. P., and A. R. Hallauer. 1997. Triple testcross analysis to detect epistasis in maize. Crop Sci. 37:763-770. Wright, S. 1922a. The effects of inbreeding and crossbreeding on guinea pigs. 1. Decline in vigor. II. Differentiation among inbred families. U.S. Dept. Agr. Tech. Bul. 1090. Wright, S. 1922b. The effects of inbreeding and crossbreeding on guinea pigs. III. Crosses between highly inbred families. U.S. Dept. Agr. Tech. Bul. 1121. Xu, Y. 1997. Quantitative trait loci: Separating, pyramiding, and cloning. Plant Breed. Rev. 15:85-139. Zirkle, C. 1952. Early ideas on inbreeding and crossbreeding. p. 1-14. In: J. W. Gowen (ed.), Heterosis. Iowa State College Press, Ames.

6

An Informal History of Statistics N. W. Simmonds* 9 McLaren Road Edinburgh, EH9 2BN Scotland

I. Introduction A. General B. Outline II. Statistics-l and Political Arithmetic A. The Start, Sir John Sinclair B. Political Arithmetic C. Life Assurance and Annuities III. Probability A. Beginnings of Probability B. First Generalizations C. The Nature of Probability D. Later Refinements, Inference IV. Error Theory and the Central Limit A. The Gauss-Laplace Synthesis, Normality B. The Emergence of Statistics-2 C. Statistical Physics V. Twentieth-Century Developments A. Agricultural Research and R. A. Fisher B. Recent Decades, Extension and Development C. Statistics and Plant Breeding VI. Conclusions VII. Biographical Sketches Literature Cited

* Acknowledgements: My best thanks are due to the following for helpful comments (not all incorporated) on the outline of statistical history: D. J. Finney, T. Leonard, H. D. Patterson; and to the following for helpful discussion of applications to and relations between plant and animal breeding: D. S. Falconer, W. G. Hill, R. A. Kempton, W. Powell, and W. Spoor.

Plant Breeding Reviews, Volume 17, Edited by Jules Janick ISBN 0-471-33373-5 © 2000 John Wiley & Sons, Inc. 259

260

N. SIMMONDS

I. INTRODUCTION

A. General The object of this review is to present a general, informal essay on the history of statistics as an element of biology, agriculture, and plant breeding. There is little or no mathematics because the subject is both beyond my competence and in any case unnecessary for the purpose. I am an agricultural botanist and plant breeder by profession, not a statistician, and some readers may feel that I have done less than full justice to recent mathematical and, more especially, computing developments. This may be true, but I believe that all the more important events took place before World War II, especially in the hands of Sir Ronald A. Fisher. This belief dominates the structure of this review. A critical survey of the most recent statistics has yet to be written. The essay was partly provoked by a profound liking for the subject, stretching now over 50 years, and a growing appreciation of its importance; statistics, as the word is used by scientists, is now fundamental for biology and agriculture, and it starts at least to impinge on all other empirical sciences. There is also the question of the curious history of the very word itself, which I hope to explain below and which will probably be unfamiliar to most readers. I found, in the course of wide reading, now spread over many years, that it was often hard to find terse biographies of eminent probabilists and statisticians and they were not collected in one place. So I concocted a biographical list for ready reference and present it below (Section VII) in the hope that readers will find it useful. The literature on the subject is vast and I had to be very selective in choosing what to cite. No comprehensive general bibliography is available. In practice, although I scanned many original sources, I have chosen nearly always to cite secondary sources, especially those easy of access and broad in scope. Some 18 works covered a good deal of the subject and they are cited, with an indication of content, in Table 6.1. The references in the columns of the table are to the relevant sections of this article. Other references are given in the text in the usual way. Full treatment of the mathematics of the subject will be found in Hald (1990, 1998).

There are also seven weighty works which I have consulted frequently but which are not conveniently cited in the text. They are as follows. Two biographical compilations have been useful, both of mathematicians in general but including many statisticians, namely Bell (1953) and Gillispie (1970-1980); the latter (16 volumes) is especially comprehen-

261

6. AN INFORMAL HISTORY OF STATISTICS

Table 6.1.

Index of leading references. Section

References

lIB

lIC

IIIA

IIIB

IIIC

IIID

IVA

IVB

IVC

VA VB

Adams 1974 Daston 1988 Gigerenzer et a1. 1989 Hacking 1975, 1990 Hald 1990, 1998 Harter 1974-75 Kendall and Plackett 1977 Kruger et a1. 1987a Krugeretal.1987b Mackenzie 1981 Maistrov 1974 Owen 1976 Pearson 1978 Pearson and Kendall 1970 Porter 1986,1995 Stigler 1986 Todhunter 1865 Walker 1929

sive and useful. Two encyclopaedias have also been valuable, namely: Kruskal and Tanur (1978) and Kotz et al. (1982-1989), the latter being the more comprehensive. Two editions of one dictionary have also been frequently consulted, namely: Kendall and Buckland (1982) and its latest edition by Marriott (1990). A recent attempt to list systematically the first uses of statistical terms has been made by David (1995). I take all these to be standard sources which are not or hardly cited in the text and run across the entire subject. B. Outline

The contents are complex, so a sketch of them here may be helpful. The essay attempts to outline the history of statistics, especially as a fundamental element of modern biology. The subject started and persists to this day, in part, as "political arithmetic," which is composed of sets of public data about society, usually assembled from censuses and surveys. This subject is, although quantitative, essentially non-mathematical and I call it statistics-i. By contrast, "probability" as a serious mathematical activity developed from the mid-1600s, at first as a practical aid to dicing and gambling. Theory grew and an uneasy duality developed

262

N. SIMMONDS

between the two views of probability, namely: the frequentist, objective, aleatory probability-l and, by contrast, the subjective, personal, epistemological probability-2. An uneasy ambivalence persists, despite much disputation over the past 200 years. From about 1800, probability theory developed into an aid to empirical science, especially in the form of error theory, which sought to exploit the normal curve and its associated mathematics. Two great mathematicians, Laplace and Gauss, were leading exponents and astronomy, geodesy, and physics were all profoundly influenced. The object of error theory was to eliminate errors, rather than to understand them. But, from later in the 1800s, a great English school of practical statisticians, initiated by Francis Galton, sprang up, devoted to biological and other scientific applications of this, a new study, which I call statistics2. Mathematical theory in the early 1900s was deficient for the needs, but the subject was effectively constructed in modern form by the genius of Fisher. A great deal of very powerful computing has been added in recent years. Probability theory per se, an abstract mathematical pursuit, was effectively developed by a great Russian school from about 1850 onwards; but statistics-2, essentially the treatment of numerical scientific data under uncertainty, owes little to it in practice (whatever ought to have happened in theory). Biology, however, has been transformed and agricultural research, the subject in which Fisher worked for years, led the field. The principles, however, are universal and as applicable to medicine and industry (for example) as to agriculture. I attempt in Section VC to place genetics and plant breeding in their appropriate statistical contexts.

II. STATISTICS·l AND POLITICAL ARITHMETIC A. The Start, Sir John Sinclair

The word statistics started to take on the first of its modern meanings about 1790 in the usage of Sir John Sinclair of Ulbster, the great agricultural improver, man of public affairs, Member of Parliament, and first Lord Thurso (1785). Karl Pearson (E. S. Pearson, 1978) relates how Sir John purloined the word from the German statistik, meaning statecraft in a verbal, non-quantitative sense. The German use itself came from the earlier Italian and it had been adopted by the G6ttingen School of Achenwall. However, Sinclair, in his comprehensive first Statistical Account of Scotland (1791-1799), the first great work of political arithmetic, deliberately, and to the fury of the G6ttingen school, "annexed" the

6. AN INFORMAL HISTORY OF STATISTICS

263

word in his new sense because he hoped that it would attract more "public attention" in Britain. It did and that was the start of what I call statistics-1, political arithmetic. Sir John could hardly have foreseen that the word would acquire yet another meaning by the end of the century. As late as 1806-1807, "the dying school of Achenwall passionately protested against the brainless bungling of the number-statisticians, those slaves oftables, the skeleton makers" (Pearson, 1978, quoting Karl Pearson's vivid words). To set Sinclair in the wider context, a general historical synopsis of statistics at large is given in Fig. 6.1 which covers all the main themes to be explored in this essay. I shall refer to it repeatedly. Sinclair

....---... •

..:.

• • --------------f I

1800 ....~~....~

•

•• . . _-""'!!.!!"".---oooIl ..

Quetelet

•

•

1900

~

t------~_t

2000 ......_ ......

Edgeworth

I I

W_ald

......

.:..._B_Ol..:1_ey--'

Fig. 6.1. Outline of the history of statistics in its two meanings. On the right, Statistics1 is Sir John Sinclair's original Political Arithmetic; at the bottom, center, enclosed in heavy dotted lines, is the contrasted usage, Statistics-2, familiar to scientists as the treatment of numerical data under uncertainty.

264

N. SIMMONDS

invented the term and was the first to recognize the social importance of orderly political arithmetic, but he did not start the study of statistics. As Fig. 6.1 shows, both probability and quantitative information about society had been going for over 100 years. But Sinclair named the study and crystallized the trend; Karl Pearson was surely right to give him the credit, accorded to him by few other writers. There is an excellent biography of this very picturesque figure by Rosalind Mitchison (1962). B. Political Arithmetic

For the general literature, see column 1 of Table 6.1. Sinclair's intention in undertaking the Statistical Account of Scotland was to create a comprehensive census of the country. It relied wholly upon correspondence with the Parish clergy, contained much that was good but certainly had unreliable patches (Mitchison 1962). However, whatever the intentions, no census is ever perfect and Sinclair's was probably not much worse than many successors. It presaged what quickly became in the 19th century a veritable torrent of numbers. The United States had already had its first census, in 1790. The political need for information on populations, ages, illnesses, taxes, wages, diet, housing, and so on had been perceived by the more intelligent autocrats of the 18th century, such as Frederick the Great and Napoleon. Only with good information could armies be constructed, fed, armed, and managed. The autocrats had to make do with some pretty dubious estimates based on what might now be regarded as unsatisfactory sampling procedures. Thus Laplace and the great chemist Antoine Laurent Lavoisier (1743-1793) were both involved in estimates of the political arithmetic of France as an aid to Napoleon's military ambitions; that particular (though enlightened) despot knew very well whereon an army marched. The political need for censuses had been perceived in Britain in the mid-1700s, but achievement had been, characteristically, blocked by the House of Lords. So censuses, in the form of attempted complete enumerations, became widely established by the 1830s (Westergaard 1932). Censuses were often associated with Societies, for example the Statistical Society of London in 1833 and the American Statistical Association in 1839. Official bureaucracy sometimes took over, however, as in Prussia, which had established a huge official statistical office by the 1820s. International collaboration followed. Nine international congresses were held in the period 1853-1878 and the general principle remained clear enough: complete but politically neutral enumeration. Opinions were not permitted: it was for statisticians simply to present the facts and for politicians, scholars, and civil servants to use the data.

6. AN INFORMAL HISTORY OF STATISTICS

265

The history of the Statistical Society of London well exemplifies these principles (Bonar and Macrosty 1934; Hilts 1978). It arose out of the 1833 meeting of the British Association for the Advancement of Science held at Cambridge, and Charles Babbage (1791-1871) and Thomas Malthus (1766-1834) were founding members. It became the Royal Statistical Society in 1877 and helped to promote the International Statistical Institute about 1900. Its motto included a sheaf of wheat and the words Aliis exterendum (let others thresh out). Virtually all the data recorded were counts rather than measurements of any kind, for obvious practical reasons. Many of the early proponents hoped for quantitative analysis, leading to a sort of "social physics"; Quetelet was among their number. But deep analysis was perceived by thoughtful observers from the start to be quite impossible. Dreams of serious mathematics quickly faded and statistics-l has rarely been more than a mass of numbers, some means and indices, and a few standard errors and correlations. Bonar and Macrosty (1934) even labeled one of their chapters The Age of Mathematics, quite meaninglessly. Political arithmetic is indeed socially essential but should not be over-rated scientifically. Dubious data derived from censuses have all too often been "over-interpreted" and pushed to yield conclusions far beyond what they can reasonably bear. But some political arithmetic has been of profound value, particularly perhaps in the field of public health. Even nowadays, there are some commentators prepared to believe that objective truths can be found at the bottom of the well of numbers. They disregard the general confounding of factors involved in any data classification, especially of non-random samples, even though there were thoughtful observers well over 100 years ago who perceived the traps of doing so. Quetelet's colleague Baron de Keverberg was one and the great French economist Antoine Augustin Cournot (1801-1877), was another. Huff (1973) gives an admirably pithy and witty account of the public misuse of seemingly quantitative data, but scepticism such as that of Huff is still regrettably rare. Furthermore, there is often cause to doubt the accuracy of the numbers themselves and the objectivity of those who frame the questions. Mere numbers acquire their own magic, as newspapers and television nowadays regularly demonstrate. No wonder then that Gigerenzer et al. (1989) were able to write that "numbers rule the world"; and often unreliable, even meaningless, numbers at that, they might have added. Herein lies the explanation of the well-known aphorism "Lies, Damned Lies and Statistics," attributed by Mark Twain in his autobiography to Benjamin Disraeli. Other attributions are sometimes heard. Whatever the source, it vividly illustrates the scepticism that ought to be applied to

266

N. SIMMONDS

numbers derived from any public or official source. All too often, "statistics" (official, so they must be true) are selected to validate assertions rather than to investigate quantities or relationships. Nor are scientific data always quite so pure, accurate, and objective as they are sometimes represented to be. Reasonably satisfactory political arithmetic can only be generated by well-constructed sampling procedures which, at their best, are based on the principles of scientific experimentation. Though a few pioneers, such as Edgeworth and Bowley, saw the point a century ago, sampling has begun to usurp the place of censuses only in this century. The USA led in the 1920s and others followed, the UK as late as the 1940s. But competent sampling is very much a part of statistics-2 and has nothing to do with conventional statistics-l, which remains what it always was, little more than a catalog of uncertain numbers, means, and indexes. As Fig. 6.1 shows, statistics-l went well back before Sinclair, even though he marked the start of the modern movement. First steps are usually linked to the early exponents of demography and life expectations, notably John Graunt using the London Bills of Mortality in the mid-17th century, his successor William Petty (1623-1687), Edmund Halley (1656-1742, a close friend of Isaac Newton), then a veritable flood of Dutch, French, and British workers interested in lives, diseases, and annuities. This, however, is to intrude on the next section. Only in the later 1700s and the 1800s did attention turn to massive enumerations of populations for political ends. Long lists of proponents of statistics-l are given by Pearson and Kendall (1970), Fitzpatrick in Kendall and Plackett (1977), and Karl Pearson (E. S. Pearson 1978). Statistics-l survives to this day and was exemplified in one of its better manifestations in Arthur Bowley's book (1945); Bowley (prominent 1910-1930) was a competent mathematician but no hint of this emerges in his later work, wherein the notions of Aliis exterendum still ruled. From all this it follows that two parallel but distinct tracks appear on the right-hand side of Fig. 6.1. Statistics-l proper and demography-andlife-table studies (admittedly not sharply distinguishable). A historically significant figure, but one of little practical consequence, appeared in the mid-19th century, namely Alphonse Quetelet. He was a Belgian astronomer who espoused the normal curve as the explanation of nearly everything; in fact he used a version of the binomial, but rather inaccurately according to Stigler (1986), and hoped for the emergence of "social physics"; he also wrote of the "average man" as though he were some sort of ideal, rather than a numerical artifact. Though promoted in Britain and elsewhere by the very influential John Herschel (1850), Quetelet was unimportant in himself; however, he did popularize the

6. AN INFORMAL HISTORY OF STATISTICS

267

idea of statistics-l and influenced Galton, to the ultimate benefit of statistics-2. The "average man," incidentally, was borrowed from the great French biologist, G. 1. 1. Buffon (1707-1788), one of the earlier experimental probabilists.

c.

Life Insurance and Annuities

The distinction between annuities and life tables and statistics-1 is not really sharp because both drew largely upon publicly available numerical data and treated at least some very similar questions. Good collective references are cited in Table G.1 column 2, and historical connections are roughly indicated in Fig. 6.1. Life tables have generally been treated empirically; the mathematics came later. Essentially, a life table is simply a table of expectations of mortality set against age at the time in question. Since a complete enumeration of the contents of a table would take as long as the longest life, it follows that most of the early tables, at least, were patched up from bits of data with gaps filled in by guesswork. Hald (1990) presents Graunt's table in detail. All life tables are specific to the population from which they are drawn and no valid generalizations are permissible, although they are often made. There is a broad contrast between rich and poor societies illustrated in Fig. 6.2; the latter have shorter lives and higher juvenile mortality. Eminent early exponents were De Witt and Huygens in Holland and Graunt, Petty, and Halley in Britain, all in the mid-1600s. Graunt's table (1662) ofthe London Bills of Mortality due to the Plague were especially influential, though much misused, for example by linear interpolation. Halley in 1694 exploited Neumann's famous Breslau table of 1693, but really good tables had to wait for 100 years and the foundation of the Equitable Life Assurance Society. The Society for Equitable Assurances on Lives and Survivorships (1762), founded by the mathematician James Dodson, later became (and still is) the Equitable Life Assurance Society which adopted the much improved Northampton tables (1780) developed by Richard Price (1723-1791, Thomas Bayes's friend). Right down to the present, empirical tables have ruled and most insurance business is still done by using them; life (i.e. mortality) tables for life insurance, other tables for their various purposes. Hardly surprising, the medical profession, as several authors have pointed out, rather disliked the determinism inherent in the life table and the relative impotence of medical intervention thereby implied. From an early stage, however, mathematicians interested themselves in life tables, for example Abraham de Moivre, Daniel Bernoulli, and Nicholas Struyck. Karl Pearson (E. S. Pearson 1978) lists many of the

268

N. SIMMONDS Survival (%)

100

80

,..................

,, ,, ,, ,, ,

~~ ...... ......I......-

......

...... ......

......

...... ~ \

,, ,, Rich , country ,, Poor ,, " country ,, ,, ,, ,, ,, ,, ,, ,, ,, , ... ,, ,, ..., ,, ,, ,, ,, ,... ,, ,, ,

\ \ \ \ \

\

\

\

\

60

\

40

20

\

\

\

\

\

\

\

\

\

\

\

\

-or--....

O-+-.....- .... o 20

-..~...,.-

40

60

\

\

\

.....- .....-,.......,r-\

80

100

Age

Fig. 6.2. Two life (= mortality) curves; generalized constructions to illustrate two contrasted patterns of survival, in poor and rich countries, respectively. Any curve is specific to that country at that time. Drawn as curves, though the underlying data will usually be discontinuous yearly tables. The vertical line drawn at age = 40 shows the very different survival rates expected for the two populations (about 30 and 90%, respectively).

earlier mathematical exponents. De Moivre eked out a poor living in London calculating actuarial and gambling problems in a London coffee house; when Newton was asked a question on such a problem, he advised the enquirer to "go to de Moivre; he understands these things better than I do." The later outcome was a definite school of actuarial mathematics which only really came to prominence in this century. Its methods are essentially deterministic, as were the methods of the first

6. AN INFORMAL HISTORY OF STATISTICS

269

exponents in the 1600s. Insurance is not at heart a probabilistic business, relying as it does upon the means of large numbers, but nearly always with a component dependent upon subjective judgment. Annuities do, however, go back a long way but were for a long time based on flat rates almost independent of life expectations. Life insurance was more of a gamble or wager than a calculation of risks and discounted returns. Annuities were, in effect, loans and were often favored by governments trying to pay for ventures (sometimes ill-judged ones). Civil servants, however, were generally neither mathematicians nor accountants and often got the sums wrong, so it was not until the middle 1800s that reasonable rates for annuities were arrived at. Even then, UK Friendly Societies (Mutual Life Insurance Companies in the USA), which are only now, in the 1990s, in terminal decline, worked to the great benefit of their members on no clear mathematical basis. Friendly Societies are owned by members, not by shareholders. And that, I think, is the essential story. Statistics-1 and Life Assurance ran in close parallel for about 200 years because they shared public data and have only fairly recently diverged. Actuarial mathematics exists as a serious and recondite study in its own right, but practical insurance still relies largely upon empirical tables and graphs, albeit far better ones than were formerly available.

III. PROBABILITY A. Beginnings of Probability

Florence David's book (1969) is a key source for information on the early days of probability and useful general works are listed in Table 6.1 column 3. Serious probability sprang from gambling in the mid-1600s. Early gambling was certainly known in pre-Christian times and there are fragments (some rather disapproving) in the Bible and Talmud. The Greeks, although excellent geometers, had no serious arithmetic and no probability. There was probably some early Indian work, yet unexplored. Gaming was known and astragali were used under various names. They are metacarpal bones from the hind legs of ruminants, sheep being favored. The derived osselot, dice (made of bone or stone), and the drawing oflots were all used anciently (David 1969). Some religious meanings were attached to random choices of events, but there seems to have been no notion of randomness, of the fundamental probability set (FPS), or of the definition of chance. The last is neatly illustrated by the "values" attached to the four possible throws of the

270

N. SIMMONDS

astragalus; those conventionally scoring 4,3,1,6 had rounded probabilities of 40,40,10,10 (David's figures agreeing quite well with Maistrov's and my own experimental ones). The "best" throw ("Venus") was one of each kind, which was certainly not the least probable combination. On the other hand, the best of the early dice (Egyptian) were quite good, though biased ones were known, too. Playing cards were a later (medieval) invention, perhaps because good cardboard was previously unavailable. In the 1500s and early 1600s, real beginnings were apparent in the recognition by Girolamo Cardano (1501-1576) and Galileo Galilei (1564-1642) of the fundamental probability set (FPS), the set of equally likely outcomes generated by suitable mechanical devices such as dice, the favorites of contemporary gamblers. Both solved elementary problems by enumerating possibilities and counting the "successes," as was to be the practice for close on 100 years. The connection with gaming was clear from the title of Cardano's book, Liber de Ludo Aleae (pre1576, pub. 1663). Gambling indeed dominated the early years, and the tradition persisted down to the 1800s and even recent times in references (especially by French writers) to drawing colored balls from urns, in the fashion of lotteries. In the mid-1600s, Fermat and Pascal, great mathematicians both, made a serious start, but always by enumerating and counting. Their famous correspondence, largely based on advice offered to the noted gambler, Chevalier de Mere, dated from the 1650s but was not published until 1679. They were able to go well beyond what even the informed and thoughtful gambler such as de Mere could achieve by deeper understanding and use of the Pascal triangle (by no means a new discovery) for more complex cases. They were early joined by the excellent Dutch physicist and mathematician Christian Hugyens who formalized much of the early work in the first textbook of probability, De Ratiociniis in Ludo Aleae (1657). Hacking (1975) is inclined to attach much weight also to a little-known work, Logic or the Art of Thinking (1662), probably by the French writers Pierre Nicole (1625-1695) and Antoine Arnauld (1612-1694) who, he says, first firmly identified probabilities as ratios. Gambling being what it is, gamblers expect at least occasional pay-offs and the notion of an "expectation" turns up in Huygens and was to persist in various forms down to the present, where we generally mean probability x cash payoff. The "expectation" of life of a new-born infant, current from the time of Graunt, did not have quite the same character, of course. And Daniel Bernoulli was later to convert "moral expectations" into the first "utilities," beloved of modern economists and operational researchers.

6. AN INFORMAL HISTORY OF STATISTICS

271

Kendall in Pearson and Kendall (1970) remarked that theory lagged millenia behind the practice of gambling and some intuitive knowledge of it. He was inclined to attribute the lag to a lack of the idea of "randomness." Indeed, this remains a difficult idea right down to the present, as will emerge later. The mathematical exploitation of the properties of the FPS depend upon it, yet it is not simple conceptually and never provably achieved in practice. The several meanings attached to the word "probability" will be explored later (Section HIC). For the moment it will suffice to note that, universally in the 1600s, the idea was purely "frequentist" or "aleatory," that is, it attached to limiting values of numerical ratios. I later call this meaning probability-l, to distinguish it from an alternative view which treats it as a subjective, epistemological notion. It is hard to believe that subjective views did not exist in the 1600s, but they are certainly not conspicuous. The work comes from that which is approvable or acceptable to authority, according to Hacking (1975). Historically, a frequentist probability was the same as a "chance"; de Moivre wrote under the title The Doctrine of Chances, later made to sound more respectable by Laplace as The Calculus of Probabilities. A "hazard" was another word for the same thing; it came from the Arabic al-zhar and neatly illustrates, yet again, the connection with gambling. The word "hazard" nowadays certainly has some connotation of risk. B. First Generalizations

Three important generalizations were achieved in the decades following the year 1700. Hald (1990) gives mathematical analyses of them and other important references are listed in Table 6.1, column 4. First, James (Jakob) Bernoulli (1654-1705) gave his well known "Golden Theorem" in a posthumous work of 1713, Ars Conjectandi. It was the first "limit theorem" in probability and many would date the serious mathematics of the subject from there. He considered strictly discontinuous probabilities of the general form P = p/(p+q) estimated from a long run of draws with replacement from a population containing a proportion p of one kind of counter, q of another, such that (p+q) = 1. He started with p=q=0.5 and went on to generalize for all p and q. He attempted to estimate posterior values of p and q and to put bounds on them. He had no means of estimating errors, however, and could only assert that the ratio approached closely to P in the limit. More narrowly, he stated that the deviation from the "true" value of P could be made as small as one pleased by increasing the number of observations in the run. Long runs and large numbers were therefore implied and the "Golden Theorem"

272

N. SIMMONDS

has often been called "The Law of Large Numbers." This may sound like an unremarkable achievement now, but it was the beginning and cost Bernoulli a great deal of deep thought. Next, in rough historical sequence, came Pierre de Montmort (16781719) whose Essai d'Analyse sur Ies Jeux d'Hazard gave the earliest plain illustration of the central limit concept. Again, like Bernoulli, he was dealing strictly with discontinuous probabilities, not with continuous functions, even though continuity was approached with large numbers. He was interested in cards and other games, but it was from dice (yet again) that the important idea emerged. He showed, numerically, in considering "the problem of points" that successive sums of larger and larger numbers of dice throws moved quickly from uniformity, through triangular to what we would now recognize as the normal form (Fig. 6.3). Like de Moivre, however, he had no ideas as to continuous probability distributions and it was left to later mathematicians (notably Laplace) to make the connection. Mathematically his idea was close to that of de Moivre because both were concerned with the limits of binomial distributions with large n. Later workers showed, what was unknown at the time, that nearly any initial distribution ultimately led to a normal form, albeit sometimes rather slowly; indeed this is itself the central limit concept. In statistical practice in this century, random normal distributions were regularly calculated by using such sums until mathematically more sophisticated methods adapted to computers took over. The third, and perhaps the most important figure of the three, was Abraham de Moivre (1667-1754), a Huguenot emigre from France who settled in London, did excellent mathematics, and became a close friend of Isaac Newton. He wrote several books, of which the Treatise of Annuities and the Doctrine of Chances were, by common consent, the most important. He also wrote a crucial pamphlet in 1733, at first given as an appendix to a paper and later appended to the last (posthumous) edition of the Doctrine of Chances (1756); David reprints the last in her book (1969). Essentially, he expanded the binomial distribution with large n and partially integrated the result, to give the first approximation to the normal probability integral (though it was not called that). His treatment was of a discontinuous probability, not of a continuous function, and Karl Pearson was wrong to have identified de Moivre as the source of the normal curve. Many authors have pointed out that de Moivre was concerned simply to expedite the calculation of probabilities (usually applied to gaming or life assurance, of course). Many authors have also pointed out that a crucial step in his argument required the use of the

273

6. AN INFORMAL HISTORY OF STATISTICS

Large Number of Throws of Specified Number of Dice N= 1

Frequency

2

100

o

2

3

4

5

6

4

6

8

10

12

N=4

~ 8

4

12

16

20

24

N=8

150 100 50 0

8

16

24

32

40

48

Total scores

Fig. 6.3. The central limit concept as demonstrated by Montmort (1713) (Hald 1990, p. 210, copied by courtesy of the author and of Messrs Wiley). Throws ofN = 1, 2,4,8 dice, showing the transition from the fundamental probability set through a triangular distribution to near-normal. Ordinates are frequencies, abscissas are total scores of the dice. In practice, 10 dice provide a close approach to normality. Montmort was concerned simply with discontinuous gambling problems, not at all with continuity and the central limit theorem. This is an exact calculation, not a simulation; thus ordinates are absolute numbers (scaled down for convenience in the third and fourth entry, n = 4,8).

274

N. SIMMONDS

well-known approximation to the factorial derived independently by James Stirling (1692-1770), which works remarkably well even for quite small n.

De Moivre integrated his curve by a combination of numerical methods (after Newton) and "mechanic quadrature." But he recognized that the tails were difficult and avoided them. De Moivre was no statistician, but he was a great mathematician and expressed a healthy scepticism about "luck": he derided "the superstition that there is in play such a thing as luck, good or bad." C. The Nature of Probability

The nature of probability seemed orderly and simple only in the 1600s when it could be regarded as a simple outcome of the FPS. Since then, it has become ever more complex, and Hacking (1975), in a key work on the subject, has written of a "confused duality," while other authors think that ideas betray an ever-increasing confusion, to some of a character more theological than scientific. Fortunately, science gets along quite well against a background of theological confusion, even if metaphysicians and mathematicians cannot make their minds up. Science always was more concerned with sensible coherence than with logic or metaphysics. The purpose of this section is to set out the principal issues, though avoiding for the moment treatment of the writer and the questions that have caused much of the trouble, namely, Thomas Bayes and his treatment of inference and indifference. They appear below. Historically, the leading distinction normally made is between objective and subjective views. The objective view, which I shall call probability-1, insists on approximation to numerical ratios, so may be called frequentist or, by reference to history, aleatory; frequentist views are properly formed only on the basis of experimental data already recorded. Some authors refer to frequency estimates (always ratios in the range 0 to 1) as "chances" (as de Moivre did, for example). By contrast, a subjective probability is a degree of belief (again in the range 0 to 1); such a view is intuitive or epistemological in nature, may be formed prior to experience, and would be regarded by some as the only proper meaning of the word. I propose to call this probability, the subjective kind, probability-2, when necessary to distinguish it from the objective probability-1.

6. AN INFORMAL HISTORY OF STATISTICS

275

Virtually all working scientists are staunch (some would say unthinking) frequentists but would necessarily take the view that anyone experiment may tend to support or negate a hypothesis but can never strictly prove or disprove it; they therefore adopt indisputably subjective interpretations (while avoiding simple logical dichotomies between "true" and "false"). Fisher himself (e.g. in his text, 1925-1970 and 1956) was certainly a frequentist in practical statistical matters (Box 1978), as was von Mises (1957). The last, however, was a physicist who regarded small samples as meaningless and hung his theory on a theoretically attractive but unidentifiable "collective" (Kollektiv). Hogben (1957) tended to the same view as that of von Mises and regarded Mendelism and physics as the only real successes of frequentist probabilities. Among mathematicians of the 1800s, G. Boole, J. S. Mill, and J. Bertrand were frequentists and so also was Karl Pearson (though accused by some of being a crypto-subjectivist). Of eminent subjectivists, B. de Finetti, H. Jeffreys (1948), J. M. Keynes (1929), and 1. J. Savage (1954) are among the most frequently mentioned. John Venn (1866) was labelled a "staunch frequentist" by one commentator but recognized acute logical difficulties in holding that position and I read him as having been quite ambivalent. In sum, Hacking's "confused duality" still rules. A sensible empirical view would seem to be that a frequency basis (probability-l) is always to be preferred if available but that a subjective opinion (probability-2) is inevitably reached from any experiment, even if fiducial or confidence limits of estimates of parameters are available. So an objective enquiry leads to a subjective judgment, followed cyclically by another round of objective enquiry. And, however rigorously held subjective views may be, no reasonable subjectivist would deny the strength and cogency of good frequency data. In short, this view leads to a universal ambivalence which seems to me to be both an unavoidable and a rational position. Extremes are simply not tenable. Logically, the difficulty, indeed the impossibility, of defining "randomness" satisfactorily precludes philosophical confidence. Von Mises's "collective" was, it seems, wrecked by the impossibility of defining both it and "randomness" simultaneously and independently. In a sense, the ambiguity goes back to the very beginnings of probability. The FPS is itself a subjective concept which can only be accepted, not critically tested. I "know" that the chance of a die falling six is onesixth, whatever my doubts as to manufacturers of dice in real life. On this view, early gambling experiments may have seemed strictly frequentist but rested on subjective assumptions. The two probabilities

276

N. SIMMONDS

have never been absolutely distinct and Hacking (1975, 1990) and others were probably mistaken to have drawn so sharp a distinction. There are several other words which deserve at least passing mention here, though they are not to be deeply explored. First, there is "likelihood," which is a frequency function but not a probability; it has been much used in estimating statistical parameters because, in some circumstances, maximum likelihoods (or normally their logarithms) define estimating functions. The idea goes far back (to Daniel Bernoulli, for example) but was really exploited only in this century, especially by R. A. Fisher (Edwards 1974). It is now perhaps the dominant method of estimation, and genetic linkage is a classic (perhaps the classic) example. "Fiducial inference" is another idea exploited by Fisher, especially in the context of small samples; but it is held by some to be simply wrong and has met with rather less than universal approbation, especially by the Neymann-Pearson school. The latter exploited the not unrelated "confidence" idea, which was part of a decision-making system that sought to define what experimental conclusions should be accepted as well as those to be rejected. These subjects have sometimes generated rather more heat than light. D. Later Refinements, Inference

The main references for this section are gathered in Table 6.1 column 6. It is convenient to defer consideration of the work of one especially influential writer until later, so I shall first outline the general history ofprobability and only later come to Thomas Bayes and his ideas on inference. By 1800 the general view of working scientists was that scientific probabilities were frequentist in character and the notion of considering continuous as well as discrete distributions had taken hold. However, there were then mathematical reservations about the "respectability" of probability theory. This is no longer a substantial issue, because the theory went forward, especially in the hands of Russian mathematicians, for over 100 years in parallel with the practical development of workaday statistics applied to science (Fig. 6.1). The following summary is largely drawn from Maistrov (1974). The important figures from the mid-1800s onwards were P. L. Chebyshev and his great pupil, A. A. Markov; they were followed by A. M. Lyapunov, S. N. Bernstein, A. Y. Khinchin and, above all, Andrei Kolmogorov. The last axiomatized gambling, numbers, the "central limit" concept, and other practical considerations out of sight and at last placed probability on a respectable mathematical footing. Kolmogorov also named "branching processes," in 1947. These went back to Francis Galton and later to Markov, around the turn of the century (Fig. 6.1). Markov's work was probably more influential in practical statistics

6. AN INFORMAL HISTORY OF STATISTICS

277

than that of all the other Russian probabilists together. In truth, refinements of the "central limit theorem" and axiomatics based on sets did not matter much, but Markov was immensely influential; his work on serial stochastic processes led into time series, random walks, chains, and branching processes. These topics have found significant applications in statistics (Fig. 6.1) but have been even more important in specialized corners of applied mathematics such as operational research, queue theory, and economics (for example, the behavior of markets). Curiously, Markov was anticipated in one of these problems by Francis Galton who, in 1873, proposed to H. W. Watson the extinction-ofsurnames problem. This was not finally settled until the 1930s when both Fisher and J. B. S. Haldane had tackled it for implications in population genetics (Kendall in Kendall and Plackett 1977). In summary, probability ideas worked well enough and were widely applied to scientific problems by the early 1800s but were mathematically far from watertight. They have since been axiomatized into a state of respectability, mostly by a succession of distinguished Russian mathematicians, with the present outcome that statistics remains immensely practical but unrespectable (to mathematicians), while the main stream of "pure" probability has separated from applications. Only Markov, in the past 100 or so years, can be said to have had substantial scientific impact. We come now to Thomas Bayes and his ideas on inference and shall conclude that his arguments, though certainly tricky, have generated literature beyond their true importance. Bayes (1702-1761) wrote but one substantial paper, which appeared in the Philosophical Transactions and is conveniently reprinted, with Richard Price's commentary, in Pearson and Kendall (1970). Price was founder of what became the Equitable Life Assurance Society (Section IIC) and was an important figure in his own right; he put a religious gloss on Bayesian arguments. "Bayesian," incidentally must be one of the more widely used eponyms in the history of science, and all the odder for the fact that its meaning is not wholly clear. Bayes's paper contains a "theorem" and a "scholium," but some commentators refer mysteriously to a "proposition." Following Kendall and Buckland (1982) and Marriott (1990), the theorem is a straightforward statement of the rule for combining probabilities multiplicatively, whereas the "scholium" rests on subjective estimates of prior probabilities. To this extent, then, Bayes was a subjectivist, indeed one of the earliest. Professor D. J. Finney draws my attention to an agreeably acerbic quotation from the writings of M. G. Kendall: "On other occasions, I have lamented that Bayesian statisticians do not stick closely enough to the pattern laid down by Bayes himself:

278

N. SIMMONDS

if they would only do as he did and publish posthumously, we should all be saved a lot of trouble." Subjective prior distributions are held by many to be necessary for inference, and Bayes has usually been said to propose the view that, when in doubt (which is usually), choose uniform priors (= equiprobability = equipossibility = indifference). This principle, often called "the equal distribution of ignorance," leads to acute logical problems and paradoxes, which make Bayesian ideas unacceptable to most working scientists and to many statisticians. It is (I think quite reasonably) argued that, if one is completely ignorant about the system being investigated, it is better to do the experiment, draw whatever inferences seem reasonable (from that experiment alone, without making prior assumptions), and then do more experiments, rather than indulge in metaphysical speculation. It is generally agreed that, when prior information is available, then Baysian requirements decline as the prior probability approaches unity. This is but a simple extension of the view expressed above that we are always at least somewhat ambivalent about probabilities. The frequentist result simply generates the subjective view that helps to define the next (frequentist) experiment. The common-sense approach just advocated (which would probably be held intuitively by most scientists) is not universal. There are heavy theoretical works advocating Bayesian methods for universal application (Lee 1989; Dale 1991; Bernardo and Smith 1994). Lindley (1965,1985) has long been an exponent of applications to the partial exclusion of purely frequentist ideas. Most of the subjectivists mentioned in the preceding section would be thought to hold Bayesian views. Furthermore, Bayesian components are usually (but not always) conspicuous in attempts to convert statistics into any form of decision-making system. By contrast, thoughtful and critical arguments about inference that arise from considering the paradoxes, as well as diverse, sometimes contrasting, viewpoints, are given by Barnard and Cox (1962), one ofthe most rewarding publications on this confusing subject. That Bayes himself had doubts about the philosophical uncertainties of his position is well argued by Stigler (1986). However, most scientists are unregenerate frequentists even if they admit (as most do) a more or less cyclical subjective element. Thus R. A. Fisher (e.g. 1956) was one such and a robustly anti-Bayesian view was propounded by Finney (1968). However, some authors have held that Bayes can be dragged into almost any argument about inference, and Karl Pearson was even accused by some of being a crypto-Bayesian, while Fisher's fiducial argument has been labelled Bayesian-in-disguise. In practice, I do not believe it matters much (Kendall and Buckland 1982) and think that working scientists and statisticians can simply accept the empir-

6. AN INFORMAL HISTORY OF STATISTICS

279

ical proposition of ignoring Bayes most of the time, but adopt his arguments if they are found to be necessary and usable for the problem in hand.

IV. ERROR THEORY AND THE CENTRAL LIMIT A. The Gauss-Laplace Synthesis, Normality Important general references are given in Table 6.1 column 7. We have seen that the earlier work of de Moivre and Montmort treated discontinuous probabilities with no reference to possible continuity but yet arrived at the central limit idea, in the former's case developing a binomial formula that was recognizably "normal." The first proposal for a continuous distribution of probabilities came from Thomas Simpson (see Hald 1998), better known as an able general mathematician, teacher, writer of textbooks, and controversialist. He seems to have started from the old observation that small "errors" were commonest and clustered round a center, conventionally the mean, large ones rare and bounded in range. Simpson simply used (in 1757) the idea of an isosceles triangular distribution of errors and showed that it was easily integrated (in two halves, of course). In fact, Simpson's triangle describes the normal curve quite well (Fig. 6.4) and the integral also agrees well. His work was later rediscovered but never made practical impact. Somewhat later, in the 1770s, Daniel Bernoulli deduced what was in effect a normal curve but did not exploit it and, instead, explored several symmetrical and continuous substitutes such as the semicircle and semi-ellipse. They led to terrible mathematical difficulties and got nowhere. By the 1780s both Laplace and Gauss had arrived at a more or less modern form of the central limit idea. Besides a central measure with which to estimate the parameter of interest, some measure of dispersion which shall be, in some useful sense, optimal, is also required. Both Laplace and Gauss accepted what had been implicit since the time of Galileo, namely the use of the arithmetic mean as the central measure. Laplace used probability concepts to convert discontinuity, probably incorporating the idea of "elementary errors" and certainly adopting Bayesian "prior" notions. Gauss seems to have been more direct and he used the idea of least squared deviations (MLS, Method of Least Squares), as a measure of dispersion, from about 1795. A. M. Legendre (17521833), in 1806, also suggested least squares, but without any probabilistic basis, and quarrelled with Gauss over priority. As Stigler (1986) says, there was at least an element of circularity in Gauss's arguments but there was no doubt that the MLS idea was highly appropriate both

280

N. SIMMONDS Frequency

0.4

0.3

0.2

0.1

-2

-1

o

+1

+2

Deviations in standard measure

Fig. 6.4. Thomas Simpson's triangular distribution of errors, shown to be a good approximation to a normal curve (plotted by solid triangles). The first continuous probability distribution. Ordinate Yare frequencies, abscissa X are deviations, in standard deviations, about a mean of zero. The integral of Simpson's triangle agrees very well with the normal integral but the fit breaks down towards the end because of discontinuity at X == O. Treated as two curves discontinuous at X == 0, the fit is excellent, and has the merit of not going to ± infinity. The X axis is arbitrarily defined as ± 2.3 and Y is calculated so as to make the area under the triangle unity, as required.

to the use of the mean and of normality. The normal curve answers many empirical needs pretty well but, as such, was of little practical use; it was simply a distribution of errors and practical exploitation demanded integration. The curve could not be integrated analytically, but Christian Kramp (1760-1826), by 1800, had calculated a good table numerically. Gauss then had the basis, in his great Thearia Matus of 1809, to make probabilistic statements about sets of measurements. This was, in effect, the start of error theory and the ultimate source of modern statistics.

6. AN INFORMAL HISTORY OF STATISTICS

281

It is perhaps worth noting in passing that all the earlier theory related to generalized large samples, even though real samples must often have been small or very small. But the methods to treat small samples correctly had to wait another century. Furthermore, the normal curve goes to infinity in both tails, which is certainly unreal, even though it is a workable approximation. Again, error theory was concerned, as its name implies, with errors, that is, with observational and instrumental errors. So data were usually replicate observations rather than replicate treatments such as would now normally be required for acceptable experiments. Error theory was a start but it was still a long way from statistics-2. Laplace and Gauss were two of the greatest mathematicians in history and their rather contrasted approaches are interesting. Laplace was very much the formal mathematician; Gauss, though perfectly capable of profound rigor when required, had already taken up the astronomy, physics, and geodesy which were to occupy much of his later life, and it looks as though he adopted normality and the MLS simply as workable procedures that yielded both tractable algebra and sensible science. He went further, of course, and expounded the so-called "normal equations" and hence opened up regression, correlation, covariance, and even multivariate problems. The additive elements of the normal equations are those of the analysis of variance, which was not to be explored for another century; to fit additive constants (as practiced by many working scientists and statisticians today) is simply to calculate the regression parameters that underly the normal equations. Plant breeders use them in the form of "combining abilities." Harter (1974-1975) has given a massive review of the MLS with, unfortunately, an unusable bibliography. Tippett (1937) in one of the earlier classical but modern textbooks well explains the MLS, and many other matters, with good examples. Though the notion of using simple deviations from the mean had proponents and intuitive attractions, it led to difficulties with signs and magnitudes, to much more complicated algebra than the MLS, and was not compatible with normality. Stigler gives an interesting account of Roger Boscovich's early efforts (1757) to fit a regression on the "figure of the earth" (degree of oblateness) by simple deviations and of Laplace's involvement in the enterprise. Normality certainly works, but it is partly a common-sense convenience, rather than a Law of Nature, and one may suspect that Laplace and Gauss differed in their views of this matter; Gauss was wearing his physicist's hat. It has been said that the "normal law" must be right because mathematicians say it is theoretically correct and scientists have established it empirically. Neither statement is true. Normality works pretty well much of the time, is robust against "reasonable" departures, is mathematically

282

N. SIMMONDS

tractable, and is very well understood. In any case, if it goes wrong, other parametric methods, transformations, and non-parametric techniques may be available to help to rescue the researcher in trouble. Several other derivations of the normal curve were recorded in the earlier 1800s, for example those of the American teacher, Robert Adrain. None was influential because, by then, Laplace and Gauss dominated the scene. It is clear that sheer convenience at least partly guided the choice of normality. It might just be that the complex derivations, almost contortions, that some mathematicians went through could have been replaced by a remarkably simple semi-empirical argument, as follows. Errors are plainly clustered symmetrically around a center and there are few extremes. The distribution curve must be more or less bi-sigmoid (not "bell-shaped," which is three-dimensional). Simple inspection of the distribution of errors suggests the equation (Bradley 1968): dy/dx

= -Cxy

which is easily integrated to give y

= k· exp (_~CX2)

which is a skeletal version of normal. One wonders whether this simple argument did not weigh (though unadmitted) with mathematicians in the 1700s. An odd feature of the early history of the subject is that there was so much argument about the form of the error distribution without empirical investigation. A few months work with diverse data would surely have sufficed to make better progress than decades of mathematics. In practice, the first good empirical tests came as late as 1818, when Bessel successfully fitted normal curves to astronomical data. One might have expected the indefatigable Buffon to have done experiments, as he threw needles and rolled dice. There have, of course, been innumerable fittings ever since Bessel and many instances of poor agreement have emerged; hence the later appreciation of the need for transformations, other special techniques, and Karl Pearson's interest in non-normal forms. As emphasized above, the normal curve works pretty well but not always. Even today, justifying normality presents problems, especially in smaller samples. Error theory was in regular use in the earlier 1800s. Its function was to help to give precision to means, and to attach standard errors to them as measures of uncertainty. No one seems to have been interested in the deviations per se and there is no evident reason why anyone should have

283

6. AN INFORMAL HISTORY OF STATISTICS

been interested. The purpose was synoptic, not analytic. The focus began to change around the middle of the century, when Quetelet started to regard anthropometric deviations (e.g. of heights and chest measurements) as interesting in their own right. Quetelet's arithmetic was shaky, however, and he worked entirely with binomial approximations (Stigler 1986). No matter, he saw the point that biology needed statistics and fired Galton's interest later in the century (Fig. 6.1). He also interested Herschel, who wrote a remarkable essay in 1850. Herschel praised Quetelet's two books and, more importantly, gave his own derivation of the normal curve (much cited but thought by some to be obscure). More interesting, though, was the fact that this was Herschel's one excursion into statistics (he was a great astronomer) and he drew the excellent analogy of the aimer at a target (stochastikos) generating a bivariate normal distribution with his arrows (Fig. 6.5). Clearly, bivariate normality, viewed

Xl = 0.09

:I:

•

0.326

+2

i••••• •

• • • •• • •• • I .~

0

-2

.1

•I

•

-4

-4

-2

•

X z = 0.01

0

:f:

0.293

+2

John Herschel's (1850) demonstration of a bivariate normal distribution from considering the two-dimensional distribution of arrows aimed at a target. "Stochastikos" was an aimer, at a target or at the truth. The example is mine, constructed by dropping a dart, from a height of 3 m, at a target 30 m 2 marked with a central star. The units on both axes X1 and X2 are in standard measure. Both means are very near zero and there is a (nonsignificant) hint of a positive correlation. There are two obvious outliers but both variates give linear quantile-quantile plots. Fig. 6.5.

284

N. SIMMONDS

thus, can be distorted obliquely by correlation, which was precisely the problem investigated by Auguste Bravais (1811-1863) and C. Schols (1849-1897), who anticipated Galton (Mackenzie 1981), to be discussed in the next section. Bravais was a French naval scientist and Schols a Dutch engineer. The main point here is that Herschel had an excellent insight into bi-normality and that the necessary mathematics to deal with correlation was already available. A minor figure whose name is well known as an eponym rather than because of the owner's distinction, was S. D. Poisson, a pupil of Laplace and a capable mathematical physicist who flourished in the early 1800s. He gave the Poisson Series for the special case of the binomial (p + q)n, when p is much greater than q and n is large. The series has been very useful indeed in treating diverse biological problems and questions posed by operational research. By a happy misjudgment, 1. von Bortkiewicz (1868-1931) (well known for his data on deaths from mule [horse?] kicks in the Prussian army) called the Poisson Series the "law of small numbers," perhaps for fear it might become confused with Bernoulli's theorem, "the law of large numbers."

B. The Emergence of Statistics-2 I have treated statistics-1 above as a simply crude political arithmetic, the outcome of censuses, occasionally subject to some speculative mathematics, more often not. One of the few people in Britain who really tried to make something of statistics-1 in the later 1800s (Fig. 6.1) was Francis Edgeworth, a capable self-taught mathematician who did good work but to little practical effect. I remarked above on Arthur Bowley's conscientious adherence to the Aliis exterendum principle, then and indeed much later. The science of statistics passed into other hands. The situation around the 1840s was that statistics-l was burgeoning, Quetelet had written two books and tried ineffectually to develop "social. physics," the astronomers and geodesists used error theory to excellent effect to refine their calculations, and normal theory was about to be transferred, very successfully, to statistical physics (Section IVC). But Quetelet was soon to disappear, despite having had some transient influence, and error theory was to be transformed into something more interesting and important, namely statistics-2. This is well described as the analysis of quantitative scientific data under uncertainty; that is, nearly all the time, for most science is quantitative and virtually all is uncertain, in at least some degree. Disraeli was surely right to be sceptical but his aphorism, "Lies, damned lies and statistics," referred only to statistics-l; statistics-2 is

6. AN INFORMAL HISTORY OF STATISTICS

285

quite a different matter; it is a serious aspect of science to which DisraeH's remark is inappropriate. But in the public mind, statistics is still statistics-i. Statistics-2 is virtually unknown outside professional circles and Disraeli's aphorism stands, alas. By far the weightiest agent of change was the great Victorian polymath, Francis Galton. He touched many sciences, mostly biological, and influenced, even transformed, everything he touched. His definitive biography is that of Karl Pearson (1914, 1924, 1930a,b) and an excellent modern one is that of Forrest (1974). His scientific position has been treated by many writers, notably Cowan (1977) and Mackenzie (1981). He was a cousin of Charles Darwin but contributed little or nothing to evolutionary theory as such, although his genetics, his finest achievement among many, ought, one feels, to have been influential. He came to genetics, not as a Mendelian, but as a biometrician, from his interest in anthropometry. He concerned himself with quantitative similarities and differences between relatives, arrived at "regression" (earlier, "reversion") in 1877, at "correlation" in 1888, and invented the term "normal" in 1889. He understood the idea of continuity of the germplasm (in the form of his "stirp") by 1875, several years before Weismann did (1883) and recognized identical twins for what they were, monozygotic. He made statistical mistakes but his quantitative genetics is usually regarded as a fundamental start, even though it was flawed and he had no notion of what Mendel was doing about the same time. In effect, he started quantitative genetics by investigating genetic regressions and heritability. The first genetic regression was Galton's analysis of an experiment on sweet peas (1877), which is still a useful teaching example. It is unfortunate that Darwin had no understanding of what Galton was up to, so evolutionary theory had to wait for decades before R. A. Fisher, Sewall Wright, and J. B. S. Haldane finally reconciled genetics and evolution. The lack of genetics was the principal defect of Darwin's treatment, however correct he was about natural selection. Statistically, Galton had profound instincts but no mathematics. Indeed, his treatment of normal distributions now seems very oldfashioned, since he mostly worked from ranks, "ogives," and medians. But his crucial step was to treat divergences from the simplest expectation as though they mattered and were not merely "errors" to be got rid of. Truly, he treated scientific data under uncertainty, the mark of statistics-2. Thus Galton, in effect, inverted error theory and invented biometrical genetics and then statistics-2 in the process. At first, Galton still meant statistics-l when he wrote of statistics; when he died in 1911 statistics-2 was already well launched. A short quotation from his Natural Inheritance (1889) is illuminating: "Some people hate the very

286

N. SIMMONDS

name of statistics but I find them full of beauty and interest ... their power of dealing with complicated phenomena is extraordinary. They are the only tools by which an opening can be cut through the formidable thicket of difficulties that bar the path of those who pursue the Science of man." The crucial step was perhaps Galton's treatment of genetic regressions and correlations or, more generally, bivariate normality. The anticipation by the French naval scientist, Bravais, and the Dutch engineer, Schols, of Galton in treating correlation, but from an entirely different standpoint, has often been remarked (e.g. by Mackenzie 1981). Though Galton emphasized the genetic elements in human characteristics in his genetic studies, he was well aware that both nature and nurture were determinants, and often interactive. This well-worn phrase in fact goes back to Shakespeare, though modern usage comes from Galton's twin studies (Forrest 1971, p. 132) and has all too often been wrongly taken to mean that Galton rejected environmental effects on human characteristics. Much has been made by some writers (Cowan 1977; Mackenzie 1981) of Galton's interest in eugenics (a word he invented in 1883). Eugenics was an interest he shared with Karl Pearson and, much later, with R. A. Fisher. However, I could not detect in Galton's writings what some authors have claimed, namely an overwhelming eugenic motivation. He seems to me merely to express views that were a fairly natural outcome of his social concerns and genetic understanding. Eugenics is now, of course, politically incorrect, and I believe Galton's several detractors were mostly promoting special views for the sake of generating socially and academically acceptable theories. Galton, Pearson, and Fisher did indeed all have eugenic interests, but they were subordinate to their overwhelming concerns with statistics and genetics. J. F. Box (1978) wrote that R. A. Fisher and fellow-geneticists resigned from the Eugenics Society about 1930 because of the domination of that society by sociologists, to the disbenefit of a rational genetic basis for the subject. However, genetical ideas and statistics intruded ever more strongly into medical research from 1932 onwards. Eugenics nowadays is effectively dead, despite mounting medical appreciation of a large genetical component in human variation. A vocal anti-eugenics movement notwithstanding, some resurgence of eugenic discussion is probably in prospect. Peel (1998) has presented a thoughtful summary ofthe history and present status of eugenics. Galton started statistics-2 but much better mathematicians carried it on, notably Karl Pearson, Udny Yule, and later, Fisher and E. S. Pearson (the son of Karl). In the later 1800s there was a marvellous flowering of statistical activity in England, allied to biological research (Fig.

6. AN INFORMAL HISTORY OF STATISTICS

287

6.1). This is the more extraordinary when one recalls that Galton himself, a few years earlier, had tried vainly to interest no fewer than seven mathematicians in the matter. At all events, a great school developed in University College, London, largely under the stimulus of Karl Pearson who travelled, taught, wrote books, philosophized, founded Biometrika, ran a major table-calculating enterprise, and did a vast amount of statistical research himself. He was another polymath and a giant, albeit a rather quarrelsome one. He thought he was about to revolutionize evolutionary theory by the use of correlations and fancy curve-fitting, but did not. However, he developed multivariate ideas, the use of the chisquared (X 2 ) statistic and invented the word "biometry" (Fig. 6.1) which is still much used (though often called "biometrics"); it betrays the essentially biological source of statistics-2 but is now of far from universal utility because of the way in which statistics-2 has permeated the other sciences and industrial development work, too. Karl Pearson was a prime mover in the unseemly Mendelian-Biometrical controversy of the early 1900s. In retrospect, it was unfortunate, even fatuous, and Galton should really have been on both sides (but maintained something like neutrality). A very important feature of Karl Pearson's work was his recognition of the fundamental importance of small-sample theory and his encouragement of the excellent W. S. Gossett, who published as "Student" and was in the employ of Guinness. Gossett's talents were later to flower under the benign influence of R. A. Fisher. In summary, statistics-2 burst out around 1900, as mostly an English enterprise, centered around London and derived from the primary stimulus of Francis Galton but developed by the powerful school of Karl Pearson. Error theory had been partly inverted, from the elimination of errors to the treatment of them, but the transition from statistics-1 was yet incomplete; statistics-2, already strongly marked by biological research (as indicated by the word biometry), was yet to be transformed by the greatest statistical scientist of them all, R. A. Fisher. C. Statistical Physics

In the 1860s and 1870s, physics, hitherto regarded as an exactly deterministic science, was profoundly affected by statistics. It became clear that some processes involving large numbers of particles and the transfer of energy were susceptible of interpretation only by the use of statistical arguments [references in Table 6.1 column 9, of which Porter (1986) is the most useful]. The acutest problem was tackled by James Clerk-Maxwell, of Edinburgh, later of London and Cambridge Universities. He treated the

288

N. SIMMONDS

movement of energy in gases as a three-dimensional normal problem of the distribution of velocities. It was already clear that the Second Law of Thermodynamics was in some difficulty in relation to what seemed to be an inexorable increase in entropy in any system and the apparent impossibility of defining a direction of time. Maxwell's "Demon" provided no more than an exceedingly improbable escape from the difficulty. Ludwig Boltzman, working in Austria about the same period, reached rather similar conclusions. Strict determinism no longer seemed possible in at least one important area of physics, a conclusion which could be assimilated in the scientific context by appeal to statistics but which caused endless difficulties in philosophical and religious arguments. Quantum mechanics later took up versions of similar statistical ideas. As I remarked above (Section lIIC), Hogben (1957) tended to regard statistical physics, along with genetics, as the only substantial successes for objective, numerical probabilities, even though Clerk-Maxwell had had to make some fairly critical assumptions about the nature of the gas molecules and their independence.

V. TWENTIETH-CENTURY DEVELOPMENTS

A. Agricultural Research and R. A. Fisher The period 1900-1930 saw a statistical revolution, no less, in which the beginnings of statistics-2 were transformed into an extremely potent new science. This transformation was, in effect, the work of one man, R. A. Fisher (1890-1962), later Sir Ronald Fisher, FRS. He wrote modern statistics almost single-handedly, did profound population genetics, and was surely one of the giants of modern science. And, not only were his researches many and profound, but he wielded a prodigious and generally benign influence on others, colleagues and students. The key period was 1900-1930, but Fisher's influence spread over his entire adult lifetime. Some important collective literature is cited in Table 6.1 column 10. Joan Fisher-Box gave an authoritative biography (Box 1978). Fisher's publications were catalogued by J. H. Bennett (1971-1974) and exegeses of his work and influence have been presented by, especially but among many, D. J. Finney (ed.) (1964), Finney and Yates (1981), and Yates (1951). Fisher read mathematics at Caius, Cambridge, and was first Wrangler in 1912; although he always claimed to be a scientist (which he was), he was undoubtedly a first-class mathematician with a profoundly practical bent.

6. AN INFORMAL HISTORY OF STATISTICS

289

Fisher's early career, a mixture of business and schoolmastering, was catastrophic, but his talents were recognized by Sir E. J. Russell, Director of the then small Rothamsted Experimental Station, who appointed him to analyze past field trials, some of them later treated as long-term rotation experiments, analyses which were to become classics of their kind. But, before going to Rothamsted, Fisher was already deeply involved in his lifetime studies, having collaborated with Student (W. S. Gossett) in small-sample theory (exact distributions of t and r), and having been much involved (not always harmoniously) with Karl Pearson. In 1918, he reconciled Mendelian genetics and biometrical theory in a classic paper (still a profound one) in the Transactions of the Royal Society of Edinburgh; it had been refused by the Royal Society of London. Thus his genetic interests were already well developed, and were to last during his life, through to the Balfour Chair of Genetics in Cambridge from 1943 onwards. (Parenthetically, I had the privilege of attending his first course in that post; not many survived to the end). Fisher would have been a major scientist even if he had only done genetics. Along with Sewall Wright and J. B. S. Haldane, three giants, he helped to found modern evolutionary, neo-Darwinian theory. At Rothamsted, in 14 short years before he followed Karl Pearson as Galton Professor of Eugenics in University College, London, Fisher wrote modern statistics and revolutionized not only agricultural, but nearly all, scientific research. Sir Harold Jeffreys remarked in 1953 that "the standard of presentation of results in agriculture is better than in any of the so-called exact sciences." Fisher's great biometrical genetic paper of 1918 was succeeded by the equally profound Mathematical Foundations of Theoretical Statistics in 1922 and his great book on evolution, The Genetical Theory of Natural Selection, in 1929. His Statistical Methods for Research Workers, the fundamental textbook on the subject, appeared first in 1925 and was ill-reviewed but went on into 14 editions (to 1970). Meanwhile, he invented the analysis of variance, defined the fundamental needs of good field experiments (replication, randomization, and local control, with randomization as perhaps the most important), and recognized the statistically fundamental nature of consistency, sufficiency, and efficiency. With F. Yates and D. J. Finney he explored profoundly the design of experiments, from complete randomized blocks to very diverse confounded, incomplete, lattice, and other designs. The older tradition, of the 1800s, had been the rigorously "controlled" twotreatment design on which an enormous amount of thought and effort had been expended [see Stigler (1986) for a good review]. Fisher turned over the theory completely by showing that complex (Le., factorial)

290

N. SIMMONDS

experiments were not only more informative but more economical. In a classic and oft-repeated statement, he wrote: "No aphorism is more frequently repeated in connection with field trials than that we must ask Nature few questions, or, ideally, one question at a time. The writer is convinced that this view is wholly mistaken. Nature, he suggests, will best respond to a logical and carefully thought-out questionnaire; indeed, if we ask her a single question, she will often refuse to answer until some other topic has been discussed." The analysis of variance is fundamental to a great deal of statistics-2 but, useful as it often is, is thought by some to over-emphasize main effects at the expense of interactions. Though in a sense subordinate to design, it is, nevertheless, the fundamental way of looking at a great many bodies of data and the basis of much of modern statistics-2. Put simply, comparisons between treatments are often of intrinsic interest, interactions may be important, and factorial experiments promote (relatively) efficient estimation of error from high order interactions. In addition to all these achievements, Fisher also explored profoundly the concept oflikelihood (from 1917), wrote books about the design of experiments and statistical inference, explored the related ideas of information and invariance, transformed the concept of analysis of variance into efficient methods of sampling and survey (which have similar underlying structures), stated what is now a truism, that real "parameters" of populations are estimated by "statistics," introducing the useful Latin/Greek notation to help to make the distinction. He developed diverse transformations to help to cope with departures from normality and at least began to explore curve-fitting and multivariate problems, for which he commanded the mathematics but for which no one at the time had sufficient computing power. He also introduced the idea of the null hypothesis and hypothesis testing. Some observers thought he over-emphasized hypothesis-testing, but if it were indeed a misjudgment at least it helped to concentrate the minds of scientists upon the crucial question of how to design and interpret experiments. He later corrected any possible misjudgment by declaring what is now surely the strong consensus view of scientists and statisticians, that the business of statistics is to estimate parameters within limits in which any one estimate may be asserted to lie, at a stated level of probability. Confidence limits, introduced by Neymann and Pearson, were, he thought, theoretically inadmissible; but it should be said that they are widely used and that not all theoreticians accept the fiducial argument. Fisher was strongly anti-Bayesian, but some have sought Bayesian elements in the fiducial concept. At all events, contemporary statistics pays (or should pay) much attention to estimation of parameters and probability

6. AN INFORMAL HISTORY OF STATISTICS

291

limits, and very little to testing null hypotheses which have, indeed, been grossly misused (Gigerenzer et al. 1989). Finney (1988-89) gives a robustly modern view of this subject and much else that was developed by Fisher; he does not look kindly upon hypothesis-testing in general and multiple-range testing in particular. Fisher always insisted that he was a working scientist and that his profound mathematics generally followed from practical calculations on his famous "Millionaire" calculating machine: numbers first, mathematics later. Indeed, he was something of a fanatic for accurate data-checking. His eyesight was very bad, to which has been attributed his frequent use of multi-dimensional geOlnetric imagination as often as algebra. (Certainly, as a lecturer, his blackboard was a deplorable example of how not to do it!) But that he was a profound mathematician would never be disputed; and some have remarked how often practical requirements have stimulated mathematicians to some of their finest work, a rule well exemplified by R. A. Fisher. B. Recent Decades, Extension and Development Though the general shape of statistics had been well enough established by R. A. Fisher by the 1940s, several areas of work have seen substantial recent developments, none wholly new and all with pre-WWII roots, some indeed a century or more old. Several go well beyond the root-area of statistical development, namely agriculture. Different students would probably write different lists; mine contains the following eight items, all having large American contributions to their development. The American elements are not all surprising in view of the overwhelming strength of the U.S. effort in the general areas of scientific research and statistical teaching, research, and application in the post-WWII years. Several useful general references are given in Table 6.1 column 11. Owen's collection of essays (1976) is especially useful because, despite the generality of the title, it is almost all recent and hardly looks outside the USA. First, in industrial experimentation the problem commonly is to devise improved working practices without disrupting the smooth operation of a plant that is already going reasonably well, even if suboptimally. The solution that generally emerged was a succession of small multi-factorial trials, each applying small adjustments to main effects, so that violent disruptions were avoided. But the small adjustments allowed progressive improvements over complex response surfaces. Several cycles could thus achieve large changes with little effect on contemporary operations. The underlying ideas are distinctly sequential,

292

N. SIMMONDS

though the experimental structures are perfectly Fisherian. Box (Box and Draper 1969) has been a leading exponent, and see also John (Chapter 1 in Owen 1976). Box coined the useful term EVOP for evolutionary operation. Second, to an increasing extent, surveys have become an economical substitute for enumeration and real advances have been made in this area. Indeed, samples often have positive advantages of both speed and accuracy over censuses. The needs first became apparent in the public and economic sectors, for example in census and economic returns. Methods were later extensively exploited by industrialists and pollsters (psephologists), though not without some social concerns on the part of those who were troubled that numbers could be "cooked" and that published results were likely to affect the results that they purported to describe or predict. The earliest modern ideas came from Udny Yule and Arthur Bowley in England in the earlier years of this century; Yule was quite clear as to the merits of stratified-random procedures well before his time. The Rothamsted school under R. A. Fisher and F. Yates advanced the subject in the 1930s, regarding the problems as being essentially matters of defining statistically acceptable structures (which were often reminiscent of the analysis of variance). For some purposes, of course, simple large, well-randomized samples were appropriate. Since the late war, there have been vast developments, largely American in origin and intensely practical in outlook. Some day, censuses may become obsolete, but no statistical refinement can cope with bad data. Useful essays are given by Hansen and Madow (Chapter 4) and Changed (Chapter 14) in Owen (1976). Third, "decision theory" was a phrase that was rarely or never heard before the past few decades. Though the historical roots go deep, the notion has only grown since the 1930s that statistics was a method of decision-making, not merely an aid to numerical analysis under uncertainty. The starting point usually cited was the Neyman-Pearson approach developed during the 1930s, which sought not only to reject a defective hypothesis but also to identify a "correct" one. The idea offers a number of inferential obstacles which have, indeed, not yet been resolved; can statistics ever identify certain truth rather than high probability? Abraham Wald (1950) was a key theorist in this area (as he was also in sequential work) and he was joined by John von Neumann and Oskar Morgenstern with their "game theory" of the late 1940s, which became a weighty component of political and economic discourse. Of necessity, decision theory incorporated loss/risk functions and subjective probabilities. Even Bayes, Laplace, and Gauss have been invoked as early subjective decision-makers. Economic applications include utility,

6. AN INFORMAL HISTORY OF STATISTICS

293

a subject which goes back to Daniel Bernoulli's "moral expectation" and his classical consideration of Sempronius's insurance problem of 1738 (Maistrov 1974). Thus decision theory has been a significant bridge between statistics and operational research. That the last is enormously useful and well-established is not in doubt, but there must still be questions about the connections between statistics and decision-making. Probably, statistics, if wisely used, should be regarded as a fundamental component of sensible decision-making rather than as a system of decision-making per se. Useful reviews are given by Neyman (Chapter 7) (a very full discussion) and Ferguson (Chapter 16) in Owen (1976); Wald's book (1950) on the subject is standard and Raiffa's (1968) book gives a readable introduction, enhanced by a diagrammatic taxonomy of statistics from a subjectivist viewpoint (Fig. 10.1). Fourth, sequential analysis was an important war-time development from the pre-WWII quality-control procedures that were widely used to keep running checks on the products of industry, a long-established practice, considerably refined in recent years. The decision idea is essential; should one stop now and adjust the process or proceed? Some chart form was and still is commonly adopted and no essentially new statistics was needed. Sequential analysis, however, goes further and did demand some new theory, provided essentially by Abraham Wald (1947); in suitable areas of application, economies of cost of the order of 50% could be achieved simply by deciding to stop runs of observations at the first moment permitted by theoretical considerations. Stopping rules presented (and still do present) certain difficulties, however. Continuous rather than batch-wise experimentation is implied. However, some areas of research simply do not lend themselves to the idea, for example most agricultural research, in which cropping seasons or animal life-times are unavoidably pre-determined. Other areas, however, have benefited greatly, especially perhaps clinical research in medicine, in which to stop an experiment is usually ethically attractive. Darling (Chapter 18 in Owen 1976) gives a survey of sequential studies. Fifth, stochastic processes have a long history going back to the mid19th century and Markov's work around 1900. Galton posed one of the earlier difficult problems (that of survival of surnames which was not properly settled until the 1930s). The essence of the matter is a chain of events in which each one is dependent upon its predecessor, usually by way of some probabilistic rule, with a random element added (for example, a time-trend with "noise," or a step in a random walk). Autocorrelations (which go back to Udny Yule) may be involved and timeseries are often implied [see Brillinger, Chapter 12 in Owen (1976)]. Cyclical processes and recurrent patterns also emerge, as in economics,

294

N. SIMMONDS

epidemiology, and ecology. Many recent developments are heavily dependent upon computing power, especially if Monte Carlo simulations are employed. There are strong sequential and industrial overtones but also a huge scientific and mathematical literature. Large computer "models" abound. Stochastic processes have vast applications in demography, population genetics, epidemics, queue theory, economics, timeseries, industrial reliability, and renewal. They could probably be classified, along with computing, as the most important post-WWII development of the subject. Much such work can probably be better characterized as "operational (operations) research" than statistics. Puri (Chapter 19 in Owen 1976) gives a useful review in the medical research context. Sixth, as mentioned above, normality sometimes fails and the failure cannot be evaded by transformation. However, evasion of the consequences of non-normality is not the sole raison d'etre oftransformations; they can be abused, both in choice and interpretation. If transformation is either impossible or inadmissible, then non-parametric methods, which appeal only to ranks and orders, rather than to measurable quantities, may be useful. Sometimes, indeed, such methods can approach conventional normal statistics quite closely as to accuracy and reliability [see Doksum Chapter 11 in Owen (1976); and Bradley (1968)]. It could be that, enthusiasts notwithstanding, non-parametric methods have yet been somewhat under-exploited. But some caution is still required; non-parametric methods may be quick and quite acceptable for significance testing but at the expense of good parameter estimation, which is usually far more important. Seventh, computers have made their mark on recent statistics. The trend has been two-fold: towards the development of large, powerful "packages" (such as GENSTAT, GLIM, and SPSS), aggregates of "programs," with very wide ranges of competence; and second, towards the small, specialized package, hardly more than a single "program," designed to do a specific task very efficiently. Not only has sheer computing power permitted calculations which would not have been possible (though well understood mathematically), a few decades ago, but it has also allowed, even encouraged, the development of numerical "models," the construction of graphics (of varying quality), the assembly of huge" data bases," the use of iterative methods of estimation, and the adoption of multivariate techniques. Not all these activities have been well done and computers, alas, have all too often encouraged bad calculations by ill-informed researchers or even by over-enthusiastic statisticians. They seem to have been particularly abused in "econometrics," which encourages the practice of assigning causal connections to

6. AN INFORMAL HISTORY OF STATISTICS

295

what are really only multiple correlations, a process against which welltaught young biologists are sternly warned. Computers, therefore, have had their uses but there has been, and still is, a damaging downside. They are terribly open to abuse. especially by those who do not know what packages can and cannot do. Among a formidable catalogue of sensible and not-so-sensible statistical practices, Finney (1988-89) lists the misuse of computers highly. They have been responsible for many horrors and will no doubt be responsible for more. Daft numbers of decimal digits and wildly over-stated probabilities are by no means unknown. But, wisely used, computers are immensely powerful and potent tools, perhaps especially in the area of heavy iterative numerical estimation. I believe we witness something of a dichotomy between those who believe that statistics has been revolutionized and the moderate sceptics (such as Finney) who agree as to their immense value and power but persist in harboring doubts as to their misuse. One specialized aspect of computer use deserves at least passing mention, namely: generation of "random numbers." Virtually all modelling activities depend upon the use of random numbers and derived random normal deviates. Perhaps one ought to say "pseudo-random" numbers because, logically, "randomness" defies practical definition: it can, however, be closely approached and so the word "random" in practice means "what good random number generators write." These last derive from profound, mathematical number theory allied to cryptography. They do not generate "truly" random numbers but merely guarantee that a quasirandom sequence repeats only at vastly long intervals. It is an agreeable reflection that the decimal digits of pi (It), though perfectly deterministic, would serve (rather inconveniently) as random numbers, as judged by numerous empirical tests. Eighth and finally, the place of professional statisticians and textbooks deserves comment (Finney 1988-89). Many institutes and laboratories now have professional statisticians in-house and there is no doubt that their skills should be invoked throughout, from design, through execution to analysis and presentation. Fisher wrote of the "marriage of design and analysis," which really was practiced at Rothamsted, even if other places have not been quite so effective. At all events, the ideal is clear, even if sometimes unobserved, as evidenced by the very many bits of bad statistics that intrude into the published literature. Textbooks, likewise, pose problems; many researchers are reasonably competent but do need good texts, even if they know enough to know when to stop and ask for professional guidance. Statistical Methods for Research Workers revised up to 1970 is fundamental but is not an easy book; several early works remain good (I pick out Tippett 1937,

296

N. SIMMONDS

especially). In more recent decades, a veritable flood of texts has appeared, all too many of them rather poor, some downright bad, a few good or very good. An estimable and widely used one has been Snedecor and its descendants (which emerged directly from Fisher's influence) (e.g. Snedecor and Cochrane 1989). The amateur can only learn to understand when he gets out of his depth and seeks professional advice when he does.

C. Statistics and Plant Breeding Statistics-2 arose, as we have seen, from Galton's perception that biological "departures" were not merely errors to be got rid of but were intrinsically interesting. Galton's first concerns were anthropocentric but were soon widened to quantitative genetics, which he interpreted mainly as a problem of genetic regressions. By the time Mendel's work became widely known early in this century, Galton had in effect retired. Mendelian genetics, then and later, provided many examples for the development of frequency statistics, including the general use of the chisquared (X 2 ) test, in the development of which both Karl Pearson and Fisher participated. Hogben (1957) regarded Mendelian genetics as one of the very few examples of really successful practical applications of statistics because it rested on large rather than on small samples. By the early 1900s, linkage was recognized but was not well interpreted until the great surge of Drosophila research in the 1920s. From there on, Mendelian genetics flourished, always a quantitative science, it is true, but making no significant demands on statistics beyond ratio-testing and estimation and, under Fisher (doing human genetics), some rather more refined parameter fitting. As far as plant breeding is concerned, Mendelian genes are "easy," but they are not often responsible for economic characters. Nearly all the time, the plant breeder is dealing with quantitative characters polygenically determined: hence the importance of biometrical or quantitative genetics. There were predecessors, it is true, but the key early paper was that of Fisher (1918), who first formally reconciled Mendelism with quantitative inheritance; he and Sewall Wright at Chicago covered much the same ground in quantitative genetics and the mathematical theory of evolution during subsequent decades. Both influenced breeders' thinking in animal and crop improvement. The total body of understanding of quantitative genetics, from the early years of this century down to the present time, has been very considerable and has been the subject of a major anthology thoughtfully edited and annotated by Hill (1984). In general, animal breeding was

6. AN INFORMAL HISTORY OF STATISTICS

297

well ahead of plant breeding in this connection. Lush (1937) was a pioneer, a follower of Sewall Wright, and more concerned with response to selection than with formal variance-splitting as a means of analyzing genetic variance. In Britain, Lush was followed by Falconer (four editions, 1960, 1981, 1989, 1996) whose book has been very influential. Plant breeders interested in the subject more than 20 years ago had, perforce, to read the animal literature because there was very little devoted to plants. I myself read Falconer in the 1960s when there was no book about plants to turn to. The reason for the leading historical position occupied by animal breeding research is that animals are few and expensive and uniformly kept. It is worthwhile to go to a lot of trouble to get and use information about individuals, especially if genotype x environmental (G x E) effects can be practically ignored in uniform environments. By contrast, plants are generally numerous and cheap and subject to many and extreme environmental influences. We know that there will be much waste, that selection is inefficient, and that G x E effects will make trials difficult, sometimes downright deceptive. Kempthorne (1973) presented the standard text on quantitative genetics, equally applicable to animals and plants, and derived directly and explicitly from Fisher's 1918 paper, the first formal partition of additive from other kinds of genetic variance. Kempthorne acknowledged his debt to Fisher very plainly and stated that he had no intention of providing guidance to practical breeders of whatever organisms. Fisher's only venture into the quantitative genetics of crops was an inconsequential excursion into higher-order statistics which are, of course, weakly estimated and were never heard of again. So the paper by Fisher, Immer, and Tedin (1932) has been widely cited but was a dead-end. Instead, plant breeders followed their animal colleagues in concentrating on variance-splitting and response to selection, though treating both subjects with becoming caution. The more important works devoted to the genetic theory of plant breeding are, I think, the books ofWricke and Weber (1986) and Mayo (1987); these are both later editions of works which first appeared some years earlier. Wricke and Weber is largely devoted to formal variancesplitting; Mayo's book goes wider and makes it clear that crops are diverse, including inbreeders and outbreeders, annuals and perennials, clonal and seed-propagated kinds. Mayo also recognizes the merits of the purely statistical device of using "combining abilities" which derive directly from the analysis of variance of factorial structures (and so trace back to Fisher yet again). They had been recognized for some years but had been but little exploited. "General combining ability" (eGA) is the

298

N. SIMMONDS

additive fraction of the genetic variance attributable to parents; "specific combining ability" (SCA) is the non-additive remainder. This is a simple and robust distinction that carries no covert genetic assumptions as to gene action. The use of combining abilities is surely growing, even though they do not lead directly to the calculation of heritabilities and response to selection. However, GCA values (which are first-degree statistics) give a good basis for judgment of the genetic worth of potential parents, whether for single crosses or polycrosses. The plant breeding literature of quantitative genetics is therefore fairly recent; a notably fine specific work is that of Hallauer and Miranda (1981) on maize. The sheer quality and volume of maize research comes out very well; of no other crop can it be said, I believe, that the practice has really been deeply influenced by genetic understanding. An interesting feature of this book is the fact that the authors bring out the essentially factorial structure of variance (genetic components, environmental ones, and interactions) as being parallel to the Fisherian concept of factorial design for all field experiments (e.g., fertilizer main effects, interactions, and error). Really good maize research goes back several decades before Hallauer and Miranda to, for example, the pioneering studies of Comstock and Robinson, who proposed orderly mating designs and methods of analyzing them. I argue, therefore, that quantitative genetics has certainly contributed to plant breeding in recent years but that, unlike animal breeding, it has not transformed it. Several other activities have had deeper effects. First, trials, in all their aspects, have been vastly improved in recent years. The older complete randomized blocks, Latin Squares, and balanced lattices have been, in effect, partly superseded by modern, more flexible and efficient designs, which are compact and well adapted to large numbers of entries. They are commonly unbalanced (i.e., non-orthogonal) and examples are the alpha-lattices and row-and-column designs which can now readily by adopted because the means of generating and analyzing them are at hand: namely computers. Further, large sets of trials pose particular problems which were being tackled by Yates at Rothamsted before WWII but have really only entered standard currency in recent years. Authors in Kempton and Fox (1997) give useful summaries both of designs and of management of sets of trials. Second, and more generally, computers, as indicated in the preceding section (VB), permit calculations which would never have been accessible before. Thus, huge iterative estimations of parameters, multivariate methods, spatial analyses, and selection indices have all become feasible though not universally used. Computers also permit, even encourage, the adoption of large data-bases which are increasingly

6. AN INFORMAL HISTORY OF STATISTICS

299

used by plant breeders to keep control of parents and crossing schedules; data management tasks are not the least of the computer's uses. The scientific calculations mentioned above would not have posed any mathematical difficulties to Fisher, but they simply could not have been done in pre-computer days. Third, biotechnology, in its various manifestations, is all too often said to be revolutionizing plant breeding. "Molecular biology usually fails to live up to its headlines," as one recent author said. It has certainly achieved prominence but up to about 1996 had little effect on actual outcomes. Since then, however, several cultivars of diverse crops have been developed by ll1eans of transgenes to be resistant to specific herbicides (especially glyphosate). Whether these transgenes will be effective in the longer term remains to be seen; analogy with disease resistances would suggest the emergence of new strains of old weeds or of new weeds resistant to the herbicide in question. Time will tell. Meanwhile, however, there is certainly a social reaction (in Europe perhaps rather than in the USA) against the use of crop biotechnology. There are no evident statistical implications. The ultimate prospects for transformed cultivars look attractive now but the future is unforseeable. A longer-term problem must surely be that only single transgenes are manipulable, so that polygenic characters, the very "stuff" of plant breeding after all, are yet inaccessible. Various biochemical diagnostics are much more serious, both now and as having potential for the future. Quite diverse biochemical characters, from isozymes (which pre-date biotechnology anyway) to DNA fragments have become prominent. They may be used as such, simply as taxonomic markers, to identify cultivars or strains of diseases, for example. Or their use may be linked to quantitative trait loci (QTLs) or, again, overlapping DNA fragments may be sought as a means of attacking phylogenetic questions under the general title of"genomics." Genomics has two components, namely the structural sequencing of DNA and the determination of gene function. Statistically, the methods used for sequencing are those appropriate to linkage analysis, which has been well understood for decades. Purely taxonomic enquiry does not require sequencing but appeals rather to multivariate techniques, including the construction of dendrograms. Functional analysis has yet hardly begun in crops but may be expected to illuminate at least some plant breeding processes in the course of time. The use of QTLs to facilitate selection in crops adapted to their use do pose somewhat harder problems because a QTL seems to be as often a segment as a locus; their use is well summarized by Falconer (1996). Very full marker-mapping is needed for their effective exploitation.

300

N. SIMMONDS

Fourth and finally, G x E interactions have had a great deal of study or, at least, a large body of publication, in the past 20 years, though recognized to exist for far longer. They are universally found whenever and wherever sought and can be examined by quite various statistical techniques, from analysis of variance to diverse regressions and multivariate methods; the only consistent outcome has been no outcome. They occur universally but erratically, are rarely repeatable, confound the analysis and interpretation of trials, and have been the subject of many fruitless studies. So far, G x E effects must be thought of as "noise" in the system which imposes the need for numerous trials rather than few and always confounds interpretation. (It is not helpful to use disease resistance in diverse environments as an example of an interpretable G x E effect; it often is but is better thought of quite differently, simply as disease resistance). The general argument that emerges from this discussion is, I think, that plant breeding has picked up statistical techniques as they presented themselves and used them in context, often very effectively. It certainly uses statistical concepts to choose parents and crosses, to select among segregating progeny, to design and manage trials systems, and to reach sensible conclusions from the results (though avoiding any suggestion that statistics per se offers a decision-making system). Plant breeding, then, borrowed from statistics but cannot be said to have contributed significantly to the subject. In support of this contention, I scanned the 650odd references in Hald's great book (1998); only one looked as though it had been evoked by a plant breeding question and that was Fisher, Immer, and Tedin (1932), a dead end if ever there were one. VI. CONCLUSIONS

There are two kinds of statistics; the first, which I call statistics1, was first identified about 1800 as political arithmetic, mostly public data arising from the censuses, which were widely adopted during the 1800s. However, political arithmetic went back to the mid-1600s and its earlier history was closely intertwined with that of life assurance and annuities, which had very chequered histories. Statistics-l persists to this day. 2. An entirely different strand of enquiry emerged in the mid-1600s as probability, a mathematical pursuit that carne directly from gambling/dicing. Early results were based on the fundamental probability set and the counting of suitable enumerations of it. But mathematical theory soon emerged as a trend to generalization, evident by the early 1700s. Already, there were uneasy doubts

1.

6. AN INFORMAL HISTORY OF STATISTICS

3.

4.

5.

6.

301

about the distinctions between objective/frequentist probabilities (probability-l) and the subjective/epistemological kind (probability-2). Further, problems of inference under uncertainty and ofthe use of prior probabilities, especially associated with the name of Thomas Bayes, obtruded. An uneasy objective/subjective duality persists to this day, but it may well be that an ambivalent view is more appropriate than explicit choice; we are probably all both objective and subjective on occasion, often cyclically. Later in the 1700s, two great mathematicians, Laplace and Gauss, transformed discrete probability theory into consideration of continuous scientific observations and created error theory. This was applied with great success to the practical problems of astronomy, geodesy, and physics. The normal (=gaussian) curve became established as the standard method of treating uncertainty, even though it was neither theoretically nor empirically secure. However, it worked pretty well, was numerically integrable, was robust, and algebraically tractable. The normal frequency distribution has dominated statistics-2 ever since, though many complementary procedures have been invented. The error theorists sought to eliminate "errors" and were not interested in departures as such. During the 1800s, biologists started to become interested in error theory and fastened on the departures as intrinsically interesting. Francis Galton, in the later 1800s, was the prime mover. He came to statistics from his genetical studies and essentially generated the great English school that practiced the analysis of scientific, especially biological, data under uncertainty, a field I call statistics-2. The shape of statistics-2 was emerging especially in University College, London, in the early 1900s and was transformed over the following 30-40 years by the genius of R. A. Fisher. Fisher was a very fine and intensely practical mathematician who effectively wrote the modern subject single-handedly (though not without some controversy). Fisher worked for 14 years at Rothamsted Experimental Station, which accounts for the fact that agricultural research, primarily, and biological research more generally, dominated statistics-2 since its inception. Fisher was a great scientist as well as a distinguished mathematician, an important geneticist as well as statistician. Probability as a purely mathematical study was forwarded by a strong school of Russian researchers in the later 1800s and early 1900s; it was placed on an abstract, axiomatic basis that is theoretically acceptable but has had little or no practical impact,

302

N. SIMMONDS

except in the person of A. A. Markov, a leading worker on stochastic processes. 7. Post-WWII developments in the later 20th century have mostly been extensions of Fisher's work. Perhaps the weightiest of them has been the rise of computing, enabling the use of calculations previously inaccessible numerically, though well enough known to Fisher's brilliant mathematics. VII. BIOGRAPHICAL SKETCHES

Note on sources: Bell (1953) gives biographies of all the more prominent mathematicians, including several early probabilists. Gillispie (1970-1980) give excellent lives of nearly all eminent scientists, mathematicians included. Short biographies explicitly of probabilists and statisticians are presented by Kruskal and Tanur (1978) and some by Kotz, Johnson, and Read (1982-1992). I do not generally cite these works, though I have consulted and used them. Some extra sources are cited, though. The list follows, in alphabetical order; others would no doubt have chosen slightly different lists of entries. Bayes, Thomas (1702-1761)

Born in London, son of Joshua, of a clerical family, Thomas a cleric, too. Many years in Tunbridge Wells. Died London. Competent mathematician, matriculated in Edinburgh about 1719-21; his later training and reason for election to the Royal Society (1742) mysterious. One minor work, perhaps two, in life. One very important posthumous paper edited by his friend Richard Price (1723-1791), also a cleric and a pioneer of life insurance. His paper was: An Essay Towards Solving a Problem in the Doctrine of Chances (Phil. Trans. Roy. Soc. 53, 370-418, 1763), with a substantial exegesis by Price). "Bayesian" is one of the most widely used eponymous adjectives in the history of science. No-one doubts the "theorem"; the conditional assumption implying an "equal distribution of ignorance" poses the difficulty. Sources: Pearson (1978); Dale (1991). Bernoulli, Daniel (1700-1782)

Born in Groningen, Netherlands, moved in 1705 to Basel; education Strasbourg, Heidelberg, and Basel. Visited Venice and S1. Petersburg (1725-1733) and traveled widely in Europe. Nephew ofJames Bernoulli.

6. AN INFORMAL HISTORY OF STATISTICS

303

Settled as Professor in Basel until 1776, teaching medicine and later physics. Several prizes from the Paris Academy. Despite his medical background, he was a formidable general mathematician, especially in calculus and mechanics, but also did some competent human physiology. In statistics, he invented a new method (1770-1771) for deriving the normal distribution as limit of a binomial but did not go beyond de Moivre. He also touched on maximum likelihood and examined the idea of "elementary errors" as approaching normality, but missed the central limit idea later exploited by Laplace. The idea of utility derived from study of the 51. Petersburg Paradox (in De Mensura Sortis, 1738) and was developed in the context of shipping insurance. He also did much work on life tables. Sources: Hald (1990); Sheynin in Kendall and Plackett (1977). Bernoulli, James (Jakob, Jacques) (1654-1705)

Born in Basel and studied there (theology degree, 1676); traveled widely in Europe over the years. Settled in Basel 1683, professor of physics and mathematics. Senior member of a quarrelsome family; brother of Nicholas, whose son Nicholas (1687-1759) was also an eminent probabilist; brother also of John (1667-1748) (father of Daniel, 1700-1782), both eminent mathematicians. Parallel to the Bach family and equally gifted. He worked on differential equations and calculus of variations, mechanics, series, probability. His Ars Conjectandi (1713), written in the late 17th century, was nearly complete at his death, but publication was delayed by family squabbles until 1713. Contains the "Golden Theorem," the first limit theorem in probability. Interested in application to games and to civil and moral affairs but a theorist, not a practical statistician. Source: Hald (1990). Bernoulli, Nicholas (junior) (1687-1759)

Born in Basel and studied there. Professor of mathematics in Padua (1716), then Basel (in logic and law). Traveled widely in Europe and returned to Basel 1722. Nephew of James Bernoulli and edited the Ars Conjectandi. Much substantial work in several branches of mathematics, including probability in which, according to Hald (1990), he has been widely undervalued. Author of De Usu Artis Conjectandi in Jure (1709), an important work on life tables, annuities, and legal applications. Close associate of de Montmort in 1710-1712. Source: Hald (1990).

304

N. SIMMONDS

Bowley, Arthur Lyon (1869-1957) Born in Bristol, educated at Christ's Hospital and Trinity, Cambridge (wrangler and prizeman). Taught in several English schools. Joined London School of Economics at its founding in 1895 and retained roots there for life. Later taught also in Reading and Oxford. First London chair of economics, 1919. Retired 1936 but remained active. His studies were in mathematical statistics applied to social studies, economics (e.g. national income), and sampling techniques. Leader in survey work by 1915. Many public functions and duties. FBA, CBE, Knight (1950). "A competent but rather old-fashioned mathematician" according to the Dictionary of National Biography. Sources: Maunder in Kendall and Plackett (1977). Condorcet, Marie-Jean-Antoine-Nicholas, Marquis de (1743-1794)

Born and studied in Paris. Committed suicide, probably by poison, to avoid execution under the Terror. Of distinguished family, young prodigy. Liberal views and mistakenly supported Robespierre. Friend of Richard Price. Active in the Academie (permanent secretary 1776 onwards). Life after 1787 mostly political. General mathematician, with some work on probability (lotteries, voting, elections, testimony, inference). Considerable output but little weight. Later wrote upon history. Source: Pearson (1978). Edgeworth, Francis Ysidro (1845-1926) Born in Ireland, educated Trinity College, Dublin, read classics; then Balliol, Oxford 1867. London (called to Bar 1877). Professor King's College 1888, then All Soul's Oxford, 1891-1922. Of a distinguished family, distant cousin of Francis Galton and relative of the novelist Maria. Se1£taught mathematician with interests and teaching experience of logic and economics as well as statistics. Associate of and extended Galton's ideas. Associate also of eminent economists B. Jowett, W. P. Jevons, A. Marshall, J. M. Keynes. Shy and reserved, sometimes obscure but influential. Fellow of the British Academy and President of Royal Statistical Society. His work was mostly socio-economic statistics but securely mathematical, touching on essentials of analysis of variance and multivariate analysis. Founder of quantitative economics. Sources: Pearson (1978); Kendall in Pearson and Kendall (1970).

6. AN INFORMAL HISTORY OF STATISTICS

305

Fermat, Pierre de (1601-1665)

Born in Beaumont-de-Lomagne, France, educated in Toulouse, Orleans, Bordeaux. Family of wealthy merchants. Undistinguished life as parliamentary lawyer in Toulouse from 1631; perhaps a sinecure. Very distinguished amateur mathematician, especially in number theory and analytic geometry, beginnings of calculus. Probability, mostly in correspondence with Pascal (1654), inspired by the gaming problems of de Mere, a very minor part of his oeuvre. Source: Hald (1990). Fisher, Ronald Aylmer (1890-1962)

Born in London, educated at Harrow and Caius College, Cambridge, wrangler 1912. Taught in schools, 1915-1919. Rothamsted Experimental Station, 1919-1933. University College, London, 1933-1943. Balfour Professor of Genetics, Cambridge, 1943-1957. CSIRO, Adelaide, Australia, 1959-1962. Very short-sighted, perhaps contributing to geometric and intuitive "style." Of great charm but could be peppery, rude, and controversial. Gift for polemic. Large family. Eugenic interests (following Galton), which have been exaggerated by some writers. ScD 1926, FRS 1929, Knight 1952. Very many academic honors. He was trained in mathematics and physics but was the greatest mathematical statistician of this century, probably of all time. Primarily a practical statistician, with a deep understanding of numbers and data. Very expert calculator. Vast interests in statistics, including small sample theory, exact tests, analysis of variance, significance testing, inference problems, trials design. In addition he was a very distinguished geneticist, primarily responsible for the reconciliation of Mendelian theory with biometry (1918), the bases of quantitative genetics, human genetics (especially blood groups), evolutionary genetics, of which, with Sewall Wright and J. B. S. Haldane, he was a major pioneer. Several major books, especially Statistical Methods for Research Workers (1925 onwards), The Genetical Theory of Natural Selection (1930), Design ofExperiments (1935), Statistical Methods and Scientific Inference (1956). Sources: Finney (1964); Box (1978); Kendall in Pearson and Kendall (1970); Yates (1951). Galton, Francis (1822-1911)

Born in Birmingham, educated in Birmingham, London, and Trinity College, Cambridge, 1838-1844. Traveled widely in Africa, 1845-1852. In 1854 settled for rest of life in London. Of prosperous and distinguished

306

N. SIMMONDS

family, related to Wedgewoods, Darwins, and Huxleys (cousin of Charles Darwin). "Establishment" figure, courteous, correct, and reserved. Numerous committees, both public and scientific. Several medals of learned societies (1853-1910), Knight 1909. He was professionally very diverse. Vast range of scientific interests, truly a "polymath," but with no real mathematical skill, despite training. Geography, exploration, meteorology, statistics, quantitative genetics, twin studies, eugenics (invented word), anthropometry, fingerprints, psychology, mechanically inventive (though no technician). Effective founder of modern statistics and quantitative genetics. Founded Eugenics Record Office (which became the Galton Laboratory of University College, London, 1906). Four major books (1869-1889), a very dull autobiography and scores of papers. In statistics, he promoted Quetelet's ideas, normal distributions, regression and correlation (invented words), inspired the interest of Karl Pearson and his school in the serious mathematics on which he, Galton, had intuition but no technical skill. Sources: Forrest (1974); Karl Pearson (1914, 1924, 1930a,b); Cowan (1977); Mackenzie (1981). Gauss, Carl Friedrich (1772-1855) Born in Braunschweig, Germany, and educated locally. Worked at G6ttingen (1795), Braunschweig (1798), and back to G6ttingen (1807). Of poor family, precocious mathematical genius. Largely solitary, few direct associates and founded no "school," but carried on a vast correspondence. Astronomical activities in Braunschweig were very productive and led towards "error theory." Unstable family life; rather a cold personality. Diverse squabbles about priority in which his famous notebooks usually proved him to have been first. Honored by many foreign academics (Paris, London, Russia). One of the three greatest mathematicians in history (along with Archimedes and Newton), achieving vast generalizations in number theory, algebra, geometry, and probability. Held strongly empirical scientific views and was no narrow theorist; extraordinary numerical skills, fruitfully applied to astronomy, geodesy, and. physics. The normal equations and "error theory" were among his principal statistical achievements. Preceded Legendre in least squares approach. Practiced probability and statistics through much of his active life. Source: Gillispie et al. (1970-1980). Gossett, William Sealey ("Student") (1876-1937) Born in Canterbury, educated at Winchester and New College, Oxford (Natural Sciences, 1899). Sent first by Guinness, 1906, to study under

6. AN INFORMAL :HISTORY OF STATISTICS

307

Karl Pearson at University College, London; then spent whole career with the company, first as technical scientist responsible for field and laboratory work, later as Head Brewer (1935) in London. Field experience with barleys was the basis of his statistical interests. Pioneer (with R. A. Fisher) of small-sample theory (t-test and distribution 1908), analysis of variance, development of modern field experiments; pioneer also of Monte Carlo methods, used by him to estimate errors of small samples in advance of theoretical treatment. Source: E. S. Pearson in Kendall and Pearson (1970). Graunt, John (1629-1674)

Born in London and died there. Family of drapers and a respected citizen. Little formal education. Lost property in fire of 1666. Religious difficulties. Impoverished and died poor. Founder-member of Royal Society, 1662. Friend and associate of William Petty, a more prominent and successful man than Graunt but a lesser one. No formal science or mathematics but had strong statistical intuition as to construction, interpretation, and use of life tables. Based on the London Bills of Mortality, published from 1604 onwards and extrapolated to give a rough life table (treated as a frequency distribution long before the idea was recognized elsewhere). His table, despite roughness and inaccuracies, was used for many years. Wrote the first text on life tables: Natural and Political Observations Made upon the Bills ofMortality (1662), immensely influential. Interests in human sex-ratio, attempts to estimate population sizes and causes of death; invented histogram. Sources: Pearson (1978); Hald (1990). Halley, Edmund (1666-1742)

Born in London, educated at St. Paul's School and Queen's, Oxford but took no degree. Traveled widely on astronomical studies. Prosperous family. After return from travels, 1678, elected FRS and active in the Society for many years (Secretary 1713 onwards). Later Professor of Astronomy, Oxford, and latterly Astronomer-Royal, Greenwich. Close associate of Newton. Foreign Member of the French Academy of Sciences. Died at Greenwich. Interests included mathematical astronomy, geophysics, meteorology, and linguistics. Statistical work almost solely on life tables, using Caspar Neumann's (1614-1715) Breslau data of 1693. Distrusted the Graunt table (justifiably, despite its general adoption). His work was regarded by Karl Pearson as the real foundation of "vital statistics." Sources: Pearson (1978); Hald (1990).

308

N. SIMMONDS

Huygens, Christian (1629-1695)

Born and died in The Hague, studied at Leiden and Breda. Visited Paris, 1666-1681. Of distinguished and cultivated family. Wide correspondence and travels. FRS 1663. Member of Paris Academy, 1666. His name has been variously spelled Christiaan/Christian, Huygens/Huyghens, none obviously "correct." His work covered a vast range of mathematics, physics, astronomy, probability. His Tractatus de Ratiociniis in Ludo Aleae (1657, 1660) was the first systematic text on probability, entirely on the traditional gaming theory. Includes the "duration of play" problem. Probability work important but almost trivial in comparison in relation to his main scientific achievements. Source: Gillispie (1970-1980).

Laplace, Pierre-Simon, Marquis de (1749-1827)

Born in Normandy, near Caen, probably studied there, resident thereafter in Paris from 1771 onwards. Of peasant small-farming family. Held chair at Ecole Militaire, Paris from 1771, Ecole Normale 1794. Persona grata with Napoleon and has been (perhaps unfairly) criticized for surviving the Terror and Napoleon unscathed, even ennobled. One of the greatest mathematicians and theoretical physicists, with very many foreign distinctions. His work was vast and i~fluential, mostly in analysis and celestial mechanics, as well as much on probability and combination of observations. He was referred to as "the Newton of France." Numerous memoirs and two major books on probability. The first, a purely verbal treatise, was probably an impossible task (Karl Pearson), the latter (mathematical) was one of the most important works ever published on probability, but was complicated, difficult, and sometimes even confused (Karl Pearson). The books were Essai Philosophique sur le Probabilit6 (1812,1814), which eschewed algebra, and Theorie Analytique des Probabilit6s (1812), the difficult and profound one. Sources: Pearson (1978); Gillispie (1970-1980). Legendre, Adden-Marie (1752-1833)

Born and died in Paris, educated in Paris. Taught at Ecole Militaire, 1775-1780. Of well-to-do family. Some public duties on weights and measures and decimal system, a formidable organizer of heavy calculations. Public office, 1813-1833. Member of Academy, 1783. Mathematician, mostly in analysis and algebra, celestial mechanics, number theory. Noted pedagogue and author of several textbooks. Published

6. AN INFORMAL HISTORY OF STATISTICS

309

method ofleast squares (MLS) (1806) and quarreled with Gauss over priority. MLS was an algebraic, not a probabilistic, technique, but Gauss had seen the point, though Legendre did not. Source: Gillispie (19701980). Lexis, Wilhelm (1837-1914)

Born near Aachen, Germany, educated in Bonn and Heidelberg. Visited Paris. Taught in several German universities, especially Freiburg, also Breslau and Gottingen. Mathematician interested in quantitative economics but disillusioned [cf. A. A. Cournot (1801-1877)] and abandoned study. Social/population studies, especially homogeneity and stability of parameters over time. Examined within- and between-sample variability; approached analysis of variance and chi-squared, both developed by others years later. Source: Kruskal and Tanur (1978). Markov, Andrei Andreivich (1855-1922) Born in Ryazan, Russia, died in Leningrad, educated University of St. Petersburg, pupil ofP. 1. Chebyshev. Professor at St. Petersburg 1886 and stayed there. Of prosperous family, suffered poor health all his life. Some political turbulence under the Revolution. Academician 1886. Came to probability fairly late in life. Important mathematician in number theory and analysis. Central limit theorem, long used since Laplace, made rigorous in late 1800s. Important book 1900, later translated into German (1912) under title Wahrscheiniichskeitsrechnung. "Markov Chain" work (1906 onwards), developed from central limit theorem studies, was purely mathematical, not regarded as applicable. "Markov Processes" in general followed, with the concept of stochastic dependence, the antithesis of Huygens's classic "absence of after-effect." Source: Kruskal and Tanur (1978). Moivre, Abraham de (1667-1754) Born in Vitry, France, educated at Sedan, Saumur, and Paris. Emigrated to England in 1688 and lived in London thereafter. Of Huguenot family evicted from France after Edict of Nantes (1685). In London, soon entered Royal Society circles (FRS 1697), and a friend of Halley and Newton. Private tutoring and insurance consulting in Slaughter's coffee house. Tried for but never attained academic post. Had several prominent squabbles and controversies (e.g. with T. Simpson). Member of Berlin Academy and widely well regarded. Died in poverty of "somnolence." Very distinguished

310

N. SIMMONDS

mathematician (equations, series, trigonometry) and came to probability quite late in life. A theoretician rather than a practical statistician. Fifteen papers in Phil. Trans. and several important books: The Doctrine of Chances: or Method ofCalculating the Probability ofEvents at Play (1718, 1738, 1756); Annuities upon Lives (1725,1743, 1750, 1752, later incorporated in the preceding Doctrine ofChances); Miscellanea Analytica de Seriebus et Quadraturis (1703, also incorporated in late edition of The Doctrine). The bibliography is very confusing. His greatest achievement, the normal approximation to the binomial distribution, appeared in a paper (Approximatio of 1733) copied into later editions of The Doctrine of Chances. Despite Karl Pearson's claim, his normal approximation was intenqed only to facilitate binomial calculations and was not a continuous normal curve, his partial integration of it notwithstanding. Sources: Pearson (1978); Hald (1990). Montmort, Pierre Remond de (1678-1719) Born in Paris and died there. Visited England several times and had wide personal contacts in Europe (e.g. with Halley and Taylor in London). FRS 1715. Of noble family. Lived mostly as country gentleman in France. Died of smallpox. Rather wide mathematical interests but not weighty. In probability wrote the influential Essai d'Analyse sur les Jeux d'Hazard (1708). The "problem of points" of dice throwing led him to a spectacular illustration of the approach to a normal curve generated by accumulating uniform distributions, but the result was unappreciated for nearly a century (see Figure in Hald 1990, p. 210). Source: Hald (1990). Neyman, Jerzey (1894-1981) Born in Bendery, Russia, educated in Kharkov University. Taught in Kharkov and Poland until 1934. University College, London, 1934-1938. University of California, Berkeley, 1938-1981. Visited London and Paris, 1926-1928. Long association with Egon S. Pearson, son of Karl. Very distinguished; several honorary doctorates, medals, and memberships of academies. Primarily a mathematician with very wide interests, including philosophy of science. In statistics, estimation and decision theory; much conflict with R. A. Fisher. Sources: Owen (1976); Kotz et al. (1982-1992). Pascal, Blaise (1623-1662) Born in Clermont-Ferrand, France, died in Paris. Infant prodigy not only in mathematics. Much illness from 1640. Controversially religious.

6. AN INFORMAL HISTORY OF STATISTICS

311

Moved to Rouen from Paris, 1640. Associate of Mersenne. Mathematics, especially geometry, invented mechanical calculator, some physics. Interest in probability fired by Chevalier de Mere's gambling questions. Famous correspondence with Fermat, 1654. "Pascal's triangle" named after him but had been invented several times centuries earlier. However, he did exploit it effectively in his calculations on gaming and wrote Traite du Triangle . .. Matiere (1654,1665). Source: Hald (1990).

Pearson, Karl (1857-1936) Born in London, educated at University College School, King's College. Cambridge (wrangler), and Germany. Called to London Bar 1882 but did not practice. Various posts in London University, 1884 onwards. Galton Professor in University College, London, 1911-1933. Energetic and productive but polemical and quarrelsome; "admired and feared rather than loved," according to the Dictionary of National Biography. Active in many fields, including philosophy, politics, social affairs and history, besides mathematics. FRS 1895. Several medals and foreign fellowships. Taught general applied mathematics in University College. Developed major statistical researches from late 19th century, following and exploiting Galtonian ideas. Curve fitting, correlation surfaces, contingency, chi-squared, history of statistics (E. S. Pearson 1978). Founder and editor of Biometrika and author of Grammar of Science (1892). Ran "statisticallaboratory" in Galton laboratory of University College, responsible for massive working tables of statistical functions. Interested in evolution but obstructed rather than promoted the subject. Took prominent part in the great Mendelian-Biometrical controversy, in opposition to Bateson. Invented terms "standard deviation" and "random walk" (1905). Wrote the great Galton biography (Pearson 1914-1930). Sources: essay by Eisenhart in Gillispie (1970-1980); Haldane in Pearson and Kendall (1970).

Poisson, Simeon-Denis (1781-1840) Born in Pithivier, France, educated at Ecole Polytechnique and stayed there. Of modest family. Prodigy, but wide interests. Supported by his teacher, Laplace. Post in the Ecole Polytechnique, 1802. Wide external functions, some at exalted political levels. Much derided and criticized for superficiality. Came to statistics late in life. Mathematical physicist who wrote excellent textbooks. Social statistics, parallel to and in touch with Quetelet. Little professional impact but developed the Poisson approximation to the binomial with small p, since widely used in many

312

N. SIMMONDS

contexts but general only from the late 19th century. The "Poisson series" is one of the better known statistical eponyms; he would hardly be remembered otherwise. Sources: Stigler (1986); Gillispie (1970-1980).

Quetelet, Alphonse (Lambert-Alphonse-Jacques) (1796-1874) Born in Ghent, Belgium and studied there. Taught in Brussels University, 1819 onwards. Visited Paris repeatedly. Died in Brussels. A mathematician and astronomer by training but with strong literary and social bent. Much involved in public affairs (with de Keverberg), for example censuses, of which he was a pioneer. International connections in public statistical work. Member of Belgian Academy, 1820. Many decorations and fellowships. Very productive writer and popularizer, including three books on statistics. A pioneer of biological measurements and promoted the normal curve in a descriptive rather than an analytical role (e.g. physical measurements of soldiers). Developed the "average man" idea (1835 onwards), which was influential and later taken up by Galton. But his normal curve, though borrowed from the astronomers, was actually still a binomial and statistically trivial. Touched "elementary errors" ideas about 1845. Source: Stigler (1986).

Simpson, Thomas (1710-1761) Born in Market Bosworth. Moved to Nuneaton, then Derby, then London. Royal Military Academy, Woolwich, teaching maths for many years (1743-1761). Son of a weaver, of poor family. Largely self-taught. Some controversy (e.g. quarrel with de Moivre) and accusations of plagiarism. FRS 1745. Competent general mathematician who understood superiority of continental mathematics over English. Distinguished teacher, 11 books (including several standard texts). Eminent probabilist (annuities) and statistician (first continuous probability functions applied to error theory, 1755, 1757). His triangular distributions were wrong but remarkably good approximations and he saw clearly what he was doing. Wrote Nature and Laws of Chance (1740); Doctrine of Annuities and Reversions . .. Rates of Interest (1742). Sources: Pearson (1978); Stigler (1986); Hald (1990).

Sinclair, Sir John (1754-1835) Born in Thurso, Scotland, educated at Edinburgh University, Glasgow University (law), and Trinity College, Oxford. Barrister 1782; Baronet. First Lord Thurso, 1786; Privy Councillor, 1810. Robust health, tem-

6. AN INFORMAL HISTORY OF STATISTICS

313

perate, "respectable," hard-working, perhaps a little dull, lifelong attachment to his native Caithness in the extreme north of Scotland. Worked tirelessly in very diverse public affairs, as one of the great "agricultural improvers"; MP (1780-1811); political economy, controversy, and writing; President of the first Board of Agriculture; promoted paper currency; initiated and largely ran the Statistical Account of Scotland (1790-97, in 21 volumes). Sources: Mitchison (1962); Pearson (1978). Struyck, Nicholas (1687-1769)

Born in Amsterdam, of Burgher fanlily. Educated and lived in Holland as teacher and author. FRS and Member of French Academy. Appreciation of his work hindered by having been written in Dutch. Worked on astronomy and accountancy as well as statistics. Probability, especially population statistics and annuities. Author of Calculation ofthe Chances in Play . .. Lotteries and Interest (1716) and other books on life statistics 1740, 1763. Developed new life tables from annuity results. Critical about quality of data and methods of analysis. Sources: Pearson (1978); Hald (1990). Witt, Jan de (1625-1672)

Born in Dordrecht, Netherlands, educated in Leiden and Anger. Died in The Hague. Lawyer in The Hague, politically eminent but murdered by mob. Competent mathematician in geometry and conic sections, but his more important work was on annuities applied to public funding. General approach was by way of expectations of present values based on life tables and interest rates; essentially the modern approach, far ahead of its time. Collaborated with Jan Hudde (1628-1704). Author of Value ofLife Annuities in Proportion to Redeemable Annuities (1671). Source: Hald (1990). Yule, George Udny (1871-1951)

Born near Edinburgh, educated at Winchester and University College, London; later studied physics and engineering in Germany. Joined Karl Pearson's department in London, 1893, and continued in University College until 1909. Cambridge lecturer, 1912-1931. Took up statistics under Pearson but soon broke away to follow his own very original bent. FRS; CBE, 1918. Rather wide interests, especially in correlation and skew distributions. Hints of analysis of variance. Early recognition that Mendelism and biometry were not incompatible (1902, pre-Fisher 1918). Important work on contingency and chi-squared and was a

314

N. SIMMONDS

pioneer of time series (Markov processes, 1921). Numerous papers and two books (standard text, 1911; Statistical Study ofLiterary Vocabulary, 1944). Sources: Kendall in Pearson and Kendall (1970).

LITERATURE CITED Adams, W. J. 1974. The life and times of the central limit theorem. Kaedmon, New York. Barnard, G. A., and D. R. Cox, (eds.). 1962. The foundations of statistical inference, a discussion. Methuen, London. Bell, E. T. 1953. Men of mathematics. 2 vols. Penguin, London. Bennett, J. H. (ed.). 1971-1974. Collected papers ofR. A. Fisher. Adelaide Univ. Press, Adelaide. Bernardo, J. M., and A. F. M. Smith. 1994. Bayesian theory. Wiley, Chichester. Bonar, J., and H. W. Macrosty, (eds.). 1934. Annals of the Royal Statistical Society, 1834-1934. Royal Statistical Society, London. Bowley, A. 1. 1945. An elementary manual of statistics. 6th ed. Macdonald & Evans, London. Box, G. E. P., and N. R. Draper. 1969. Evolutionary operation. A statistical method for process improvement. Wiley, New York and London. Box, J. F. 1978. R. A. Fisher. The life of a scientist. Wiley, New York. Bradley, J. V. 1968. Distribution-free statistical tests. Prentice-Hall, Englewood Cliffs, NJ. Cowan, R. S. 1977. Nature and nurture: The interplay of biology and politics in the work of Francis Galton. Studies in the History of Biology 1:133-208. Dale, A. L. 1991. A history of inverse probability from Thomas Bayes to Karl Pearson. Springer, New York. Daston, 1. 1988. Classical probability in the enlightenment. Princeton Univ. Press, Princeton. David, F. N. 1969. Games, gods and gambling: A history of probability and statistical ideas. Griffin, London. David, H. A. 1995. First(?) occurrence of common terms in mathematical statistics. Am. Statistician 49:121-133. Edwards, A. W. F. 1974. The history of likelihood. Int. Stat. Rev. 42:9-15. Falconer, D. S. 1960,1981, 1989, 1996. Introduction to quantitative genetics. 4 editions. Oliver and Boyd, Edinburgh. Finney, D. J. (ed.). 1964. In memoriam Ronald Aylmer Fisher, 1890-1962. Biometrics 20:237-373. Finney, D. J. 1968. Statistics for mathematicians, an introduction. Oliver & Boyd, Edinburgh. Finney, D. J. 1988-89. Was this in your statistics textbook? Expt. Agr. 24:153-161; 24:343-353; 24:421-432; 25:11-25; 25:165-175; 25:291-311. Finney, D. J., and F. Yates. 1981. Statistics and computing in agricultural research. p. 219-236. In: G. W. Cooke (ed.), Agricultural research, 1931-1981. Agricultural Research Council, London. Fisher, R. A. 1925-1970, 14 eds. Statistical methods for research workers. Oliver & Boyd, Edinburgh. Fisher, R. A. 1930. The genetical theory of natural selection. Oxford Univ. Press, London. Fisher, R. A. 1947. The design of experiments. Oliver & Boyd, Edinburgh.

6. AN INFORMAL HISTORY OF STATISTICS

315

Fisher, R A. 1956. Statistical methods and scientific inference. Oliver & Boyd, Edinburgh. Fisher, R A., F. RImmer, and O. Tedin. 1932. The genetical interpretation of statistics of the third degree in the study of quantitative inheritance. Genetics 17:107-124. Forrest, D. W. 1974. Francis Galton: the life and work of a Victorian genius. Elek, London. Gigerenzer, G., Z. Switjink, T. Porter, J. Beatty, and L. Kruger. 1989. The empire of chance. Cambridge Univ. Press, Cambridge. Gillispie, C. C. (ed.). 1970-1980. Dictionary of scientific biography, vols. 1-16. Scribner, New York. Hacking, 1. 1975. The emergence of probability. Cambridge Univ. Press, Cambridge. Hacking, 1. 1990. The taming of chance. Cambridge Univ. Press, Edinburgh. Hald, A. 1990. A history of probability and statistics and their applications before 1750. Wiley, New York. Hald, A. 1998. A history of mathematical statistics from 1750 to 1930. Wiley, New York. Hallauer, A. R, and J. B. Miranda. 1981. Quantitative genetics in maize breeding. Iowa State Univ. Press, Ames. Harter, H. L. 1974, 1975. The method ofleast squares and some alternatives. Int. Stat. Rev. 42:147-174,235-264; 43:1-44,125-190,269-278. Herschel, J. 1850. Quetelet on probability. Edinburgh Rev. 92:1-57. Hill, W. G. (ed). 1984. Benchmark papers in genetics. 15. Quantitative genetics 1 and 2. Van Nostrand, Reinhold, New York. Hilts, V. L. 1978. Aliis exterendum or the origins of the Statistical Society of London. Isis 69:21-43. Hogben, L. 1957. Statistical theory. Allen & Unwin, London. Huff, D. 1954, 1973. How to lie with statistics. Gollancz, London; Penguin, Harmondsworth. Jeffreys, H. 1948. Theory of probability. 2nd ed. Oxford Univ. Press, Oxford. Kempthorne, 0.1973. An introduction to genetic statistics. Wiley, New York. Kempton, R. A., and P. N. Fox, 1997. Statistical methods for plant variety evaluation. Chapman and Hall, London. Kendall, M. G., and L. Buckland, W. R. 1982. A dictionary of statistical terms. 4th ed. Longman, London, (d. Marriott 1990). Kendall, M. G., and R. L. Plackett (eds.). 1977. Studies in the history of probability and statistics, vol. 2. Griffin, London. (d. Pearson and Kendall 1970). Keynes, J. M. 1929. A treatise on probability. Macmillan, London. Kotz, S., N. L. Johnson, and C. B. Read (eds.). 1982-1989. Encyclopaedia of statistical sciences, vols. 1-10. Wiley, New York. Kruger, L., L. Daston, and M. Heidelberger (eds.). 1987. The probabilistic revolution, 1. Ideas in history. MIT Press, Cambridge, MA. Kruger, L., G. Gigerenzer, and M. S. Morgan. 1987. The probabilistic revolution, 2. Ideas in the sciences. MIT Press, Cambridge, MA. Kruskal, W. H., and J. M. Tanur (eds.). 1978. Int. encyclopaedia of statistics, vols. 1 and 2. Macmillan Free Press, London. Lee, P. M. 1989. Bayesian statistics: An introduction. Oxford Univ. Press and Edward Arnold, New York. Lindley, D. V. 1965. Introduction to probability and statistics from a Bayesian viewpoint. Cambridge Univ. Press, Cambridge. Lindley, D. V. 1985. Making decisions. Wiley, London. Lush, J. L. 1937. Animal breeding plans. Iowa State Univ. Press, Ames.

316

N. SIMMONDS

Mackenzie, D. A. 1981. Statistics in Britain, 1865-1930. The social construction of scientific knowledge. Edinburgh Univ. Press, Edinburgh. Maistrov,1. E. 1974. Probability theory: A historical sketch. Academic Press, New York. Marriott, F. H. C. 1990. A dictionary of statistical terms. Longman, London; Wiley, New York (5th ed.). (d. Kendall and Buckland 1982). Mayo, O. 1987. The theory of plant breeding. 2nd ed. Clarendon Press, Oxford. Mises, R. von. 1928-1957. Probability, statistics and truth. Allen & Unwin, London. Mitchison, R. 1962. Agricultural Sir John. The life of Sir John Sinclair of Ulbster, 1754-1835. Bles, London. Owen, D. B. (ed.). 1976. On the history of statistics and probability. Dekker, New York and Basel. Pearson, E. S. (ed.). 1978. The history of statistics in the 17th and 18th centuries against the changing background of intellectual, religious and scientific thought. Lectures by Karl Pearson given at University College London During the Academic Sessions 1921-1933. Griffin, London. Pearson, E. S., and M. G. Kendall (eds.). 1970. Studies in the history of statistics and probability. vol. 1. London. (d. Kendall and Plackett 1977). Pearson, K. 1914,1924, 1930a,b. The life and letters of Francis Galton. 4 vols. Cambridge Univ. Press, Cambridge. Pearson, K. 1978. (See Pearson, E. S., 1978). Peel, R. A. (ed.). 1998. Essays in the history of eugenics. The Galton Institute, London. Porter, T. M. 1986. The rise of statistical thinking. Princeton Univ. Press, Princeton. Porter, T. M. 1995. Trust in numbers. Princeton Univ. Press, Princeton. Raiffa, H. 1968. Decision analysis. Addison-Wesley, Reading, MA. Savage, 1. J. 1984. The foundations of statistics. Wiley, New York. Snedecor, G. W., and W. G. Cochran, 1989. Statistical methods. 8th ed. Iowa State Univ. Press, Ames. Stigler, S. M. 1986. The history of statistics. The measurement of uncertainty before 1900. Belknap Press, Cambridge, MA. Tippett, 1. H. C. 1937. The methods of statistics. Williams & Norgate, London. Todhunter, 1. 1865. A history of the mathematical theory of probability from the time of Pascal to that of Laplace. Macmillan, Cambridge. Venn, J. 1866. The logic of chance. Macmillan, London. Wald, A. 1947. Sequential analysis. Wiley, New York and London. Wald, A. 1950. Statistical decision functions. Wiley, New York. Walker, H. M. 1929. Studies in the history of statistical method. Williams & Wilkins, Baltimore. Westergaard, H. 1. 1932. Contributions to the history of statistics. King, London. Wricke, G., and W. E. Weber. 1986. Quantitative genetics and selection in plant breeding. De Gruyter, Berlin and New York. Wright, S. 1968, 1969, 1977, 1978. Evolution and the genetics of populations. Univ. of Chicago Press, Chicago and London. Yates, F. 1951. The influence of statistical methods for research workers on the development of the science of statistics. J. Am. Stat. Assoc. 46:19-34.

Subject Index Volume 17 A

Apomixis, rice, 114-116

B Biography, Yuan, Longping, 1-13 Breeding: Hybrid, 225-257 Rice, 15-156 Rose, 159-189 Statistics, 296-300 White clover, 191-223

c Clover, molecular genetics, 191-223 Cytogenetics, rose, 169-171

F Forage, breeding, white clover, 191-223 G

Genetics: Overdominance,225-257 Rice, hybrid, 15-156 Rose, 171-172 White clover, 191-223 Grain breeding, rice, 15-156 H

Heterosis: Overdominance, 225-257 Rice, 24-33

Hybrid and hybridization, see also Heterosis: Overdominance,225-257 Rice, 15-156 I

Insect and mite resistance, white clover, 209-210 Interspecific hybridization, rose, 176-177 Intersubspecific hybridization, rice, 88-98 M

Maize: Hybrid breeding, 249-251 Overdominance,225-257 Male sterility, rice, 33-72 Molecular markers: Rice, 113-114 Rose, 179 White clover, 212-215

a Ornamental breeding, rose, 159-189 Overdominance, 225-257 Q Quantitative genetics: Overdominance, 225-257 Statistics, 296-300

317

SUBJECT INDEX

318

R

Trifolium, see Clover, White Clover

Rice, hybrid breeding, 1-15, 15-156 Rosa, see Rose Rose breeding, 159-189

v

s

w

Seed, rice production, 98-111, 118-119 Somaclonal variation, rose, 178-179 Statistics, history, 259-316

White clover, molecular genetics, 191-223

T Taxonomy, rose, 162-169 Transformation: Rice, 179-180 White clover, 193-211

Virus disease, white clover, 201-209

Cumulative Subject Index (Volumes 1-17) A

Adaptation: blueberry, rabbiteye, 5:351-352 durum wheat, 5:29-31 genetics, 3:21-167 testing, 12:271-297 Alfalfa: inbreeding, 13:209-233 in vitro culture, 2:229-234 somaclonal variation, 4:123-152 unreduced gametes, 3:277 Allard, Robert W. (biography), 12:1-17

Almond: breeding self-compatible, 8:313-338

transformation, 16:103 Alstroemaria, mutation breeding, 6:75

Aneuploidy: alfalfa, 10:175-176 alfalfa tissue culture, 4:128-130 petunia, 1:19-21 wheat, 10:5-9 Anther culture: cereals, 15:141-186 maize, 11:199-224 Anthocyanin pigmentation, maize aleurone, 8:91-137 Antifungal proteins, 14:39-88 Antimetabolite resistance, cell selection, 4:139-141, 159-160 Apple: genetics, 9:333-366 rootstocks, 1:294-394

Apple transformation, 16:101-102 Apomixis: Reproductive barriers, 11:92-96 Rice, 17:114-116 Apricot transformation, 16:102 Arachis, in vitro culture, 2:218-224 Artichoke breeding, 12:253-269 Avena sativa, see Oat Azalea, mutation breeding, 6:75-76 B

Bacillus thuringensis, 12:19-45 Bacterial diseases: apple rootstocks, 1:362-365 cell selection, 4:163-164 cowpea, 15:238-239 raspberry, 6:281-282 soybean, 1:209-212 sweet potato, 4:333-336 transformation fruit crops, 16:110 Banana: Breeding, 2:135-155 transformation, 16:105-106 Barley: anther culture, 15:141-186 breeding methods, 5:95-138 doubled haploid breeding, 15:141-186

gametoclonal variation, 5:368-370 haploids in breeding, 3:219-252 photoperiodic response, 3:74, 89-92,99

vernalization, 3:109 Bean (Phaseolus): breeding, 1:59-102 319

320

Bean (Phaseolus) (cont'd) breeding mixtures, 4:245-272 breeding (tropics), 10:199-269 heat tolerance, 10:149 in vitro culture, 2:234-237 photoperiodic response, 3:71-73, 86-92,16:102-109 protein, 1:59-102 Biochemical markers, 9:37-61 Biography: Allard, Robert W., 12:1-17 Bringhurst, Royce S., 9:1-8 Burt,on, Glenn W., 3:1-19 Draper, Arlen D., 13:1-10 Duvick, Donald N., 14:1-11 Gabelman, Warren H., 6:1-9 Hallauer, Arnel K, 15:1-17 Harlan, Jack K, 8:1-17 Jones, Henry A., 1:1-10 Munger, Henry M., 4:1-8 Ryder, Edward J., 16:1-14 Sears, Ernest Robert, 10:1-22 Sprague, George F., 2:1-11 Vogel, Orville A., 5:1-10 Weinberger, John H., 11:1-10 Yuan, Longping, 17:1-13 Birdsfoot trefoil, tissue culture, 2:228-229 Blackberry, 8:249-312 mutation breeding, 6:79 Black walnut, 1:236-266 Blueberry: breeding, 13:1-10 rabbiteye, 5:307-357 Bramble transformation, 16:105 Brassica, see Cole crops Brassicaceae: incompatibility, 15:23-27 molecular mapping, 14:19-23 Brassica napus, see Rutabaga Breeding: alfalfa via tissue culture, 4:123-152 almond,8:313-338 apple, 9:333-366 apple rootstocks, 1:294-394 banana, 2:135-155 barley, 3:219-252; 5:95-138

CUMULATIVE SUBJECT INDEX

bean, 1:59-102; 4:245-272 biochemical markers, 9:37-61 blackberry, 8:249-312 black walnut, 1:236-266 blueberry, rabbiteye, 5:307-357 carbon isotope discrimination, 12:81-113 cassava, 2:73-134 cell selection, 4:153-173 chestnut, 4:347-397 chimeras, 15:43-84 chrysanthemum, 14:321-361 citrus, 8:339-374 coffee, 2:157-193 coleus, 3:343-360 competitive ability, 14:89-138 cowpea, 15:215-274 cucumber, 6:323-359 diallel analysis, 9:9-36 doubled haploids, 15:141-186 durum wheat, 5:11-40 exotic maize, 14:165-187 fescue, 3:313-342 forest tree, 8:139-188 gene action 15:315-374 genotype x environment interaction, 16:135-178 grapefruit, 13:345-363 grasses, 11:251-274 guayule, 6:93-165 heat tolerance, 10:124-168 herbicide-resistant crops, 11:155-198 heterosis, 12:227-251 homeotic floral mutants, 9:63-99 hybrid,17:225-257 hybrid wheat, 2:303-319; 3:169-191 induced mutations, 2:13-72 insect and mite resistance in cucurbits, 10:199-269 isozyrnes, 6:11-54 lettuce, 16:1-14 maize, 1:103-138, 139-161; 4:81-122; 9:181-216; 11:199-224; 14:139-163, 165-187, 189-236

321

CUMULATIVE SUBJECT INDEX

molecular markers, 9:37-61 mosaics, 15:43-84 mushroom, 8:189-215 negatively associated traits, 13:141-177

oat, 6:167-207 oil palm, 4:175-201 pasture legumes, 5:237-305 pearl millet, 1:162-182 perennial rye, 13:265-292 plantain, 2:150-151; 14:267-320 potato, 3:274-277; 9:217-332; 16:15-86

proteins in maize, 9:181-216 quality protein maize (QPM), 9:181-216

raspberry, 6:245-321 recurrent restricted phenotypic selection, 9:101-113 recurrent selection in maize, 9:115-179; 14:139-163

rice, 17:15-156 rose, 17:159-189 rutabaga, 8:217-248 sesame, 16:179-228 sugar cane, 16:272-273 soybean, 1:183-235; 3:289-311; 4:203-243

soybean nodulation, 11:275-318 soybean recurrent selection, 15:275-313 spelt, 15:187-213 statistics, 17:296-300 strawberry, 2:195-214

supersweet sweet corn, 14:189-236

sweet cherry, 9:367-388 sweet corn, 1:139-161; 14:189-236

sweet potato, 4:313-345 tomato, 4:273-311 triticale, 5:41-93; 8:43-90 Vigna, 8:19-42 virus resistance, 12:47-79 wheat, 2:303-319; 3:169-191; 5:11-40; 11:225-234; 13:293-343

wheat for rust resistance, 13:293-343

white clover, 17:191-223 wild rice, 14:237-265 Bringhurst, Royce S. (biography), 9:1-8

Broadbean, in vitro culture, 2:244-245

Burton, Glenn W. (biography), 3:1-19

c Cajanus, in vitro culture, 2:224 Carbohydrates, 1:144-148 Carbon isotope discrimination, 12:81-113

Carnation, mutation breeding, 6:73-74

Cassava, 2:73-134 Castanea, see Chestnut Cell selection, 4:139-145,153-173 Cereal breeding, see Grain breeding. Cherry, see Sweet cherry transformation, 16:102 Chestnut breeding, 4:347-397 Chickpea, in vitro culture, 2:224-225

Chimeras and mosaics, 15:43-84 Chinese cabbage, heat tolerance, 10:152

Chromosome, petunia, 1:13-21, 31-33 Chrysanthemum: breeding, 14:321-361 mutation breeding, 6:74 Cicer, see Chickpea Citrus, protoplast fusion, 8:339-374 Clonal repositories, see National Clonal Germplasm Repository Clover: in vitro culture, 2:240-244 molecular genetics, 17:191-223 Coffea arabica, see Coffee Coffee, 2:157-193 Cold hardiness: breeding nectarines and peaches, 10:271-308

wheat adaptation, 12:124-135

CUMULATIVE SUBJECT INDEX

322

Cole crops: Chinese cabbage, heat tolerance, 10:152 gametoclonal variation, 5:371-372 rutabaga, 8:217-248 Coleus, 3:343-360 Competition, 13:158-165 Competitive ability breeding, 14:89-138 Controlling elements, see Transposable elements Corn, see Maize; Sweet corn Cotton, heat tolerance 10:151 Cowpea: breeding, 15:215-274 heat tolerance, 10:147-149 in vitro culture, 2:245-246 photoperiodic response, 3:99 Cryopreservation, 7:125-126,148-151,167 buds, 7:168-169 genetic stability, 7:125-126 meristems, 7:168-169 pollen, 7:171-172 seed,7:148-151,168 Cucumber, breeding, 6:323-359 Cucumis sativa, see cucumber Cucurbitaceae, insect and mite resistance, 10:309-360 Cytogenetics: alfalfa, 10:171-184 blueberry, 5:325-326 cassava, 2:94 citrus, 8:366-370 coleus, 3:347-348 durum wheat, 5:12-14 fescue, 3:316-319 Glycine, 16:288-317 guayule, 6:99-103 maize mobile elements, 4:81-122 oat, 6:173-174 pearl millet, 1:167 perennial rye, 13:265-292 petunia, 1:13-21, 31-32 rose, 17:169-171 rye, 13:265-292 Saccharum complex, 16:273-275

sesame, 16:185-189 triticale, 5:41-93; 8:54 wheat, 5:12-14; 10:5-15; 11:225-234 Cytoplasm: cybrids, 3:205-210 molecular biology of male sterility, 10:23-51 organelles, 2:283-302; 6:361-393 pearl millet, 1:166 petunia, 1:43-45 wheat, 2:308-319 D

Dahlia, mutation breeding, 6:75 Diallel cross, 9:9-36 Disease and pest resistance: antifungal proteins, 14:39-88 apple rootstocks, 1:358-373 banana, 2:143-147 blackberry, 8:291-295 black walnut, 1:251 blueberry, rabbiteye, 5:348-350 cassava, 2:105-114 cell selection, 4:143-145,163-165 citrus, 8:347-349 coffee, 2:176-181 coleus, 3:353 cowpea, 15:237-247 durum wheat, 5:23-28 fescue, 3:334-336 herbicide-resistance, 11:155-198 host-parasite genetics, 5:393-433 induced mutants, 2:25-30 lettuce, 1:286-287 potato, 9:264-285 raspberry, 6:245-321 rutabaga, 8:236-240 soybean, 1:183-235 spelt, 15:195-198 strawberry, 2:195-214 virus resistance, 12:47-79 wheat rust, 13:293-343 Doubled haploid breeding, 15:141-186 Draper, Arlen D. (biography), 13:1-10

CUMULATIVE SUBJECT INDEX

Drought resistance: durum wheat, 5:30-31 soybean breeding, 4:203-243 wheat adaptation, 12:135-146 Durum wheat, 5:11-40 Duvick, Donald N. (biography), 14:1-11 E

Elaeis, see Oil palm Embryo culture: in crop improvement, 5:181-236 oil palm, 4:186-187 pasture legume hybrids, 5:249-275 Endosperm: maize, 1:139-161 sweet corn, 1:139-161 Endothia parasitica, 4:355-357 Evolution: coffee, 2:157-193 grapefruit, 13:345-363 sesame, 16:189 Exploration, 7:9-11, 26-28, 67-94 F

Fabaceae, molecular mapping, 14:24-25 Fescue, 3:313-342 Festuca, see Fescue Floral biology: almond,8:314-320 blackberry, 8:267-269 black walnut, 1:238-244 cassava, 2:78-82 chestnut, 4:352-353 coffee, 2:163-164 coleus, 3:348-349 fescue, 3:315-316 guayule, 6:103-105 homeotic mutants, 9:63-99 induced mutants, 2:46-50 pearl millet, 1:165-166 pistil in reproduction, 4:9-79 pollen in reproduction, 4:9-79

323

reproductive barriers, 11:11-154 rutabaga, 8:222-226 sesame, 16:184-185 sweet potato, 4:323-325 Forage breeding: alfalfa inbreeding, 13:209-233 fescue, 3:313-342 perennials, 11:251-274 white clover, 17:191-223 Forest crop breeding: black walnut, 1:236-266 chestnut, 4:347-397 ideotype concept, 12:177-187 quantitative genetics, 8:139-188 Fruit, nut, and beverage crop breeding: almond,8:313-338 apple, 9:333-366 apple rootstocks, 1:294-394 banana, 2:135-155 blackberry, 8:249-312 blueberry, 13:1-10 blueberry, rabbiteye, 5:307-357 cherry, 9:367-388 citrus, 8:339-374 coffee, 2:157-193 ideotype concept, 12:175-177 genetic transformation, 16:87-134 grapefruit, 13:345-363 mutation breeding, 6:78-79 nectarine (cold hardy), 10:271-308 peach (cold hardy), 10:271-308 plantain, 2:135-155 raspberry, 6:245-321 strawberry, 2:195-214 sweet cherry, 9:367-388 Fungal diseases: apple rootstocks, 1:365-368 banana and plantain, 2:143-145, 147 cassava, 2:110-114 cell selection, 4:163-165 chestnut, 4:355-397 coffee, 2:176-179 cowpea, 15:237-238 durum wheat, 5:23-27 host-parasite genetics, 5:393-433

324

Fungal diseases (cont'd) lettuce, 1:286-287 raspberry, 6:245-281 soybean, 1:188-209 spelt,15:196-198 strawberry, 2:195-214 sweet potato, 4:333-336 transformation, fruit crops, 16:111-112 wheat rust, 13:293-343 G

Gabelman, Warren H. (biography), 6:1-9 Gametes: almond, self compatibility, 7:322-330 blackberry, 7:249-312 competition, 11:42-46 forest trees, 7:139-188 maize aleurone, 7:91-137 maize anthocynanin, 7:91-137 mushroom, 7:189-216 polyploid, 3:253-288 rutabaga, 7:217-248 transposable elements, 7:91-137 unreduced, 3:253-288 Gametoclonal variation, 5:359-391 barley, 5:368-370 brassica, 5:371-372 potato, 5:376-377 rice, 5:362-364 rye, 5:370-371 tobacco, 5:372-376 wheat, 5:364-368 Garlic, mutation breeding, 6:81 Genes: action, 15:315-374 apple, 9:337-356 Bacillus thuringensis, 12:19-45 incompatibility, 15:19-42 incompatibility in sweet cherry, 9:367-388 induced mutants, 2:13-71 lettuce, 1:267-293

CUMULATIVE SUBJECT INDEX

maize endosperm, 1:142-144 maize protein, 1:110-120, 148-149 petunia, 1:21-30 quality protein in maize, 9:183-184 rye perenniality, 13:261-288 soybean, 1:183-235 soybean nodulation, 11:275-318 sweet corn, 1:142-144 wheat rust resistance, 13:293-343 Genetic engineering: bean, 1:89-91 fruit crops, 16:87-134 host-parasite genetics, 5:415-428 maize mobile elements, 4:81-122 transformation by particle bombardment, 13:231-260 virus resistance, 12:47-79 Genetic load and lethal equivalents, 10:93-127 Genetics: adaptation, 3:21-167 almond, self compatibility, 8:322-330 apple, 9:333-366 Bacillus thuringensis, 12:19-45 bean seed protein, 1:59-102 blackberry, 8:249-312 black walnut, 1:247-251 blueberry, 13:1-10 blueberry, rabbiteye, 5:323-325 chestnut blight, 4:357-389 chimeras, 15:43-84 chrysanthemums, 14:321 clover, white, 17:191-223 coffee, 2:165-170 coleus, 3:3-53 cowpea, 15:215-274 durum wheat, 5:11-40 forest trees, 8:139-188 fruit crop transformation, 16:87-134 gene action, 15:315-374 herbicide resistance, 11:155-198 host-parasite, 5:393-433

CUMULATIVE SUBJECT INDEX

incompatibility, 15:19-42 incompatibility in sweet cherry, 9:367-388 induced mutants, 2:51-54 insect and mite resistance in Cucurbitaceae, 10:309-360 isozymes, 6:11-54 lettuce, 1:267-293 maize aleurone, 8:91-137 maize anther culture, 11:199-224 maize anthocynanin, 8:91-137 maize endosperm, 1:142-144 maize male sterility, 10:23-51 maize mobile elements, 4:81-122 maize mutation, 5:139-180 maize seed protein, 1:110-120, 148-149 male sterility, maize, 10:23-51 mapping, 14:13-37 maturity,3:21-167 markers to manage germplasm, 13:11-86 metabolism and heterosis, 10:53-59 molecular mapping, 14:13-37 mosaics, 15:43-84 mushroom, 8:189-216 oat, 6:168-174 organelle transfer, 6:361-393 overdominance, 17:225-257 pearl millet, 1:166, 172-180 perennial rye, 13:261-288 petunia, 1:1-58 photoperiod,3:21-167 plantain, 14:264-320 potato ploidy manipulation, 3:274-277; 16:15-86 quality protein in maize, 9:183-184 quantitative trait loci, 15:85-139 reproductive barriers, 11:11-154 rice, hybrid, 17:15-156 rose, 17:171-172 rutabaga, 8:217-248 sesame, 16:189-195 soybean, 1:183-235

325

soybean nodulation, 11:275-318 spelt, 15:187-213 supersweet sweet corn, 14:189-236 sweet corn, 1:139-161; 14:189-236 sweet potato, 4:327-330 temperature, 3:21-167 tomato fruit quality, 4:273-311 transposable elements, 8:91-137 triticale, 5:41-93 virus resistance, 12:47-79 wheat gene manipulation, 11:225-234 wheat male sterility, 2:307-308 wheat molecular biology, 11:235-250 wheat rust, 13:293-343 white clover, 17:191-223 yield, 3:21-167 Genome: Glycine, 16:289-317 Poaceae, 16:276-281 Genotype x environment, interaction, 16:135-178 Germplasm, see also National Clonal Germplasm Repositories; National Plant Germplasm System acquisition and collection, 7:160-161 apple rootstocks, 1:296-299 banana, 2:140-141 blackberry, 8:265-267 black walnut, 1:244-247 cassava, 2:83-94, 117-119 chestnut, 4:351-352 coffee, 2:165-172 distribution, 7:161-164 enhancement, 7:98-202 evaluation, 7:183-198 exploration and introduction, 7:9-18,64-94 genetic markers, 13:11-86 guayule, 6:112-125 isozyme, 6:18-21

CUMULATIVE SUBJECT INDEX

326

Germplasm (cont'd) maintenance and storage, 7:95-110,111-128,129-158,15 9-182; 13:11-86 maize, 14:165-187 management, 13:11-86 oat, 6:174-176 pearl millet, 1:167-170 plantain, 14:267-320 potato, 9:219-223 preservation by tissue culture, 2:265-282 rutabaga, 8:226-227 sesame, 16:201-204 spelt, 15:204-205 sweet potato, 4:320-323 triticale, 8:55-61 wheat, 2:307-313 Gesneriaceae, mutation breeding, 6:73 Gladiolus, mutation breeding, 6:77 Glycine, genomes, 16:289-317 Glycine max, see Soybean Grain breeding: barley, 3:219-252, 5:95-138 doubled haploid breeding, 15:141-186 ideotype concept, 12:173-175 maize, 1:103-138, 139-161; 5:139-180; 9:115-179, 181-216; 11:199-224; 14:165-187 oat, 6:167-207 pearl millet, 1:162-182 rice, 17:15-156 spelt, 15:187-213 transformation, 13:231-260 triticale, 5:41-93; 8:43-90 wheat, 2:303-319; 5:11-40; 11:225-234, 235-250; 13:293-343 wild rice, 14:237-265 Grape, transformation, 16:103-104 Grapefruit: breeding, 13:345-363 evolution, 13:345-363

Grass breeding: breeding, 11:251-274 mutation breeding, 6:82 recurrent selection, 9:101-113 transformation, 13:231-260 Growth habit, induced mutants, 2:14-25 Guayule, 6:93-165 H

Haploidy, see also unreduced and polyploid gametes apple, 1:376 barley,3:219-252 cereals, 15:141-186 maize, 11:199-224 petunia, 1:16-18,44-45 potato, 3:274-277; 16:15-86 Harlan, Jack R. (biography), 8:1-17 Heat tolerance breeding, 10:129-168 Herbicide: resistance, cell selection, 4:160-161 resistant crops, 11:155-198 tolerance, transformation fruit crops, 16:114 Heterosis: gene action, 15:315-374 overdominance, 17:225-257 plant breeding, 12:227-251 plant metabolism, 10:53-90 rice, 17:24-33 Hordeum, see Barley Honeycomb selection, 13:87-139 Host-parasite genetics, 5:393-433 Hyacinth, mutation breeding, 6:76-77 Hybrid and hybridization, see also Heterosis barley, 5:127-129 blueberry, 5:329-341 chemical, 3:169-191 interspecific,5:237-305 overdominance, 17:225-257 rice, 17:15-156 wheat, 2:303-319

327

CUMULATIVE SUBJECT INDEX

I

Ideotype concept, 12:163-193 In vitro culture: alfalfa, 2:229-234; 4:123-152 barley, 3:225-226 bean, 2:234-237 birdsfoot trefoil, 2:228-229 blackberry, 8:274-275 broadbean, 2:244-245 cassava, 2:121-122 cell selection, 4:153-173 chickpea, 2:224-225 citrus, 8:339-374 clover, 2:240-244 coffee, 2:185-187 cowpea, 2:245-246 embryo culture, 5:181-236, 249-275 germplasm preservation, 7:125,162-167 introduction, quarantines, 3:411-414 legumes, 2:215-264 mungbean, 2:245-246 oil palm, 4:175-201 pea, 2:236-237 peanut, 2:218-224 petunia, 1:44-48 pigeon pea, 2:224 pollen, 4:59-61 potato, 9:286-288 sesame, 16:218 soybean, 2:225-228

Stylosanthes, 2:238-240 wheat, 12:115-162 wingbean, 2:237-238 zein, 1:110-111 Inbreeding depression, 11 :84-92 alfalfa, 13:209-233 cross pollinated crops, 13:209-233 Incompatibility: almond, 8:313-338 molecular biology, 15:19-42 pollen, 4:39-48

reproductive barrier, 11:47-70 sweet cherry, 9:367-388 Incongruity, 11:71-83 Industrial crop breeding, guayule, 6:93-165 Insect and mite resistance: apple rootstock, 1:370-372 black walnut, 1:251 cassava, 2:107-110 clover, white, 17:209-210 coffee, 2:179-180 cowpoa,15:240-244 Cucurbitaceae, 10:309-360 durum wheat, 5:28 maize, 6:209-243 raspberry, 6:282-300 rutabaga, 8:240-241 sweet potato, 4:336-337 transformation fruit crops, 16:113 white clover, 17:209-210 Interspecific hybridization: blackberry, 8:284-289 blueberry, 5:333-341 citrus, 8:266-270 pasture legume, 5:237-305 rose, 17:176-177 rutabaga, 8:228-229

Vigna, 8:24-30 Intersubspecific hybridization, rice, 17:88-98 Introduction, 3:361-434; 7:9-11, 21-25 Ipomoea, see Sweet potato Isozymes, in plant breeding, 6:11-54

J Jones, Henry A. (biography), 1:1-10

Juglans nigra, see Black walnut K

Karyogram, petunia, 1:13 Kiwifruit transformation, 16:104

CUMULATIVE SUBJECT INDEX

328

supersweet sweet corn,

L

14:189-236

Lactuca sativa, see Lettuce Legume breeding, see also Oilseed, Soybean: cowpea, 15:215-274 pasture legumes, 5:237-305 Vigna, 8:19-42 Legume tissue culture, 2:215-264 Lethal equivalents and genetic load, 10:93-127

Lettuce: genes, 1:267-293 breeding, 16:1-14 Linkage: bean, 1:76-77 isozymes, 6:37-38 lettuce, 1:288-290 maps, molecular markers, 9:37-61 petunia, 1:31-34 Lotus: hybrids, 5:284-285 in vitro culture, 2:228-229 Lycopersicon, see Tomato M

Maize: anther culture, 11:199-224; 15:141-186

anthocyanin, 8:91-137 breeding, 1:103-138, 139-161 carbohydrates, 1:144-148 doubled haploid breeding, 15:141-186

exotic germplasm utilization, 14:165-187

hybrid breeding, 17:249-251 insect resistance, 6:209-243 male sterility, 10:23-51 mobile elements, 4:81-122 mutations, 5:139-180 overdominance, 17:225-257 protein, 1:103-138 quality protein, 9:181-216 recurrent selection, 9: 115-179; 14:139-163

transformation, 13:235-264 transposable elements, 8:91-137 unreduced gametes, 3:277 Male sterility: chemical induction, 3:169-191 coleus, 3:352-353 lettuce, 1:284-285 molecular biology, 10:23-51 pearl millet, 1: 166 petunia, 1:43-44 rice, 17:33-72 sesame, 16:191-192 wheat, 2:303-319 Malus spp, see Apple Malus xdomestica, see Apple Malvaceae, molecular mapping, 14:25-27

Mango transformation, 16:107 Manihot esculenta, see Cassava Medicago, see also Alfalfa in vitro culture, 2:229-234 Meiosis, petunia, 1:14-16 Metabolism and heterosis, 10:53-90 Microprojectile bombardment, transformation, 13:231-260 Mitochondria genetics, 6:377-380 Mixed plantings, bean breeding, 4:245-272

Mobile elements, see also transposable elements: maize, 4:81-122; 5:146-147 Molecular biology: comparative mapping, 14:13-37 cytoplasmic male sterility, 10:23-51

herbicide-resistant crops, 11:155-198

incompatibility, 15:19-42 molecular mapping, 14:13-37 molecular markers, 9:37-61, 10:184-190; 12:195-226; 13:11-86; 14:13-37 quantitative trait loci, 15:85-139 somaclonal variation, 16:229-268 soybean nodulation, 11:275-318

329

CUMULATIVE SUBJECT INDEX

somaclonal variation, 4:123-152;

transposable (mobile) elements,

5:147-149

4:81-122; 8:91-137 virus resistance, 12:47-79 wheat improvement, 11 :235-250 Molecular markers, 9:37-61 alfalfa, 10:184-190 clover, white, 17:212-215 fruit crops, 12:195-226 mapping, 14:13-37

plant genetic resource management, 13:11-86 rice, 17:113-114 rose, 17:179 somaclonal variation, 16:238-243 white clover, 17:212-215 Monosomy, petunia, 1:19 Mosaics and chimeras, 15:43-84 Mungbean, 8:32-35 in vitro culture, 2:245-246 photoperiodic response, 3:74, 89-92

Munger, Henry M. (biography), 4:1-8

Musa spp., see Banana, Plantain Mushroom, breeding and genetics, 8:189-215

Mutants and mutation: alfalfa tissue culture, 4:130-139 apple rootstocks, 1:374-375 banana, 2:148-149 barley, 5:124-126 blackberry, 8:283-284 cassava, 2:120-121 cell selection, 4:154-157 chimeras, 15:43-84 coleus, 3:355 cytoplasmic, 2:293-295 gametoclonal variation, 5:359-391 homeotic floral, 9:63-99 induced, 2:13-72 maize, 1:139-161,4:81-122; 5:139-180

mobile elements, see Transposable elements mosaics, 15:43-84 petunia, 1:34-40 sesame, 16:213-217

sweet corn, 1:139-161 sweet potato, 4:371 transposable elements, 4:181-122; 8:91-137

tree fruits, 6:78-79 vegetatively-propagated crops, 6:55-91

zein synthesis, 1:111-118 Mycoplasma diseases, raspberry, G:253-254 N

National Clonal Germplasm Repository (NCGR), 7:40-43 cryopreservation, 7:125-126 genetic considerations, 7:126-127

germplasm maintenance and storage, 7:111-128 identification and label verification, 7:122-123 in vitro culture and storage, 7:125 operations guidelines, 7:113-125 preservation techniques, 7:120-121

virus indexing and plant health, 7:123-125

National Plant Germplasm System (NPGS), see also Germplasm history, 7:5-18 information systems, 7:57-65 operations, 7:19-56 National Seed Storage Laboratory (NSSL), 7:13-14, 37-38, 152-153

Nectarines, cold hardiness breeding, 10:271-308

Nematode resistance: apple rootstocks, 1 :368 banana and plantain, 2:145-146 coffee, 2:180-181 cowpea, 15:245-247 soybean, 1:217-221 sweet potato, 4:336

330

Nematode resistance (cont'd) transformation fruit crops, 16:112-113 Nicotiana, see Tobacco Nodulation, soybean, 11:275-318

o Oat, breeding, 6:167-207 Oil palm: breeding,4:175-201 in vitro culture, 4:175-201 Oilseed breeding: oil palm, 4:175-201 sesame, 16:179-228 soybean, 1:183-235; 3:289-311; 4:203-245; 11:275-318; 15:275-313 Organelle transfer, 2:283-302; 3:205-210; 6:361-393 Ornamentals breeding: chrysanthemum, 14:321-361 coleus, 3:343-360 petunia, 1:1-58 rose, 17:159-189 Ornithopus, hybrids, 5:285-287 Orzya, see Rice Overdominance, 17:225-257 Ovule culture, 5:181-236

p Papaya transformation, 16:105-106 Parthenium argentatum, see Guayule Paspalum notatum, see Pensacola bahiagrass Passionfruit transformation, 16:105 Pasture legumes, interspecific hybridization, 5:237-305 Pea: flowering, 3:81-86,89-92 in vitro culture, 2:236-237 Peach: cold hardiness breeding, 10:271-308

CUMULATIVE SUBJECT INDEX

transformation, 16:102 Peanut, in vitro culture, 2:218-224 Pear transformation, 16:102 Pecan transformation, 16:103 Pennisetum ameritanum, see Pearl millet Pensacola bahiagrass, 9:101-113 Pepino transformation, 16:107 Peppermint, mutation breeding, 6:81-82 Perennial grasses, breeding, 11:251-274 Perennial rye breeding, 13:261-288 Petunia spp., genetics, 1:1-58 Phaseolin, 1:59-102 Phaseolus vulgaris, see Bean Phytophthora fragariae, 2:195-214 Pigeon pea, in vitro culture, 2:224 Pistil, reproductive function, 4:9-79 Pisum, see Pea Plant introduction, 3:361-434; 7:9-11, 21-25 Plant exploration, 7:9-11, 26-28, 67-94 Plantain breeding, 2:135-155; 14:267-320 Plastid genetics, 6:364-376, see also Organelle Plum transformation, 16:103-140 Poaceae: molecular mapping, 14:23-24 Saccharum complex, 16:269-288 Pollen: reproductive function, 4:9-79 storage, 13:179-207 Polyploidy, see also Haploidy alfalfa, 10:171-184 alfalfa tissue culture, 4:125-128 apple rootstocks, 1:375-376 banana, 2:147-148 barley, 5:126-127 blueberry, 13:1-10 gametes, 3:253-288 isozymes, 6:33-34 petunia, 1:18-19 potato, 16:15-86

331

CUMULATIVE SUBJECT INDEX

reproductive barriers, 11:98-105 sweet potato, 4:371 triticale, 5:11-40 Population genetics, see Quantitative Genetics Potato: breeding, 9:217-332 gametoclonal variation, 5:376-377 heat tolerance, 10:152 mutation breeding, 6:79-80 photoperiodic response, 3:75-76, 89-92

ploidy manipulation, 16:15-86 unreduced gametes, 3:274-277 Protein: antifungal, 14:39-88 bean, 1:59-102 induced mutants, 2:38-46 maize, 1:103-138, 148-149;

Recurrent restricted phenotypic selection, 9:101-113 Recurrent selection, 9:101-113, 115-179; 14:139-163

soybean, 15:275-313 Red stele disease, 2:195-214 Regional trial testing, 12:271-297 Reproduction: barriers and circumvention, 11:11-154

pollen and pistil, 4:9-79 Rhododendron, mutation breeding, 6:75-76

Rice, see also Wild rice: anther culture, 15:141-186 doubled haploid breeding, 15:141-186

gametoclonal variation, 5:362-364 heat tolerance, 10:151-152 hybrid breeding, 17:1-15, 15-156 photoperiodic response, 3:74,

9:181-216

Protoplast fusion, 3:193-218 citrus, 8:339-374 mushroom, 8:206-208 Prunus: amygdalus, see Almond avium, see Sweet cherry Psophocarpus, in vitro culture, 2:237-238

89-92

Rosa, see Rose Rose breeding, 17:159-189 Rubus, see Blackberry, Raspberry Rust, wheat, 13:293-343 Rutabaga, 8:217-248 Ryder, Edward J. (biography), 16:1-14

Q Quantitative genetics: forest trees, 8:139-188 genotype x environment interaction, 16:135-178 overdominance, 17:225-257 statistics, 17:296-300 trait loci (QTL), 15:85-139 Quantitative trait loci (QTL), 15:85-138

Quarantines, 3:361-434; 7:12,35 R

Rabbiteye blueberry, 5:307-357 Raspberry, breeding, 6:245-321

Rye: gametoclonal variation, 5:370-371 perennial breeding, 13:261-288 triticale, 5:41-93

s Saccharum complex, 16:269-288 Salt resistance: cell selection, 4:141-143 durum wheat, 5:31 Sears, E.R. (biography), 10:1-22 SecaIe, see Rye Seed: apple rootstocks, 1:373-374 banks, 7:13-14, 37-40, 152-153 bean, 1:59-102

CUMULATIVE SUBJECT INDEX

332

Seed (cont'd) lettuce, 1:285-286 maintenance and storage, 7:95-110,129-158,159-182

maize, 1:103-138, 139-161, 4:81-86

pearl millet, 1:162-182 protein, 1:59-138, 148-149 rice production, 17:98-111, 118-119 soybean, 1:183-235, 3:289-311 synthetic, 7:173-174 vari~gation, 4:81-86 wheat (hybrid), 2:313-317

Selection, see also Breeding cell, 4:139-145,153-173 honeycomb design, 13:87-139 recurrent restricted phenotypic, 9:101-113

recurrent selection in maize, 9:115-179; 14:139-163

Sesame breeding, 16:179-228 Sesamum indicum, see Sesame Solanaceae: incompatibility, 15:27-34 molecular mapping, 14:27-28 Solanum tuberosum, see Potato Somaclonal variation, see also Gametoclonal variation alfalfa, 4:123-152 isozymes, 6:30-31 maize, 5:147-149 molecular analysis, 16:229-268 mutation breeding, 6:68-70 rose, 17:178-179 transformation interaction, 16:229-268

utilization, 16:229-268 Somatic embryogenesis, 5:205-212; 7:173-174

oil palm, 4:189-190 Somatic genetics, see also Gametoclonal variation; Somaclonal variation: alfalfa, 4:123-152 legumes, 2:246-248

maize, 5:147-149 organelle transfer, 2:283-302 pearl millet, 1:166 petunia, 1:43-46 protoplast fusion, 3:193-218 wheat, 2:303-319 Sorghum: photoperiodic response, 3:69-71, 97-99

transformation, 13:235-264 Southern pea, see Cowpea Soybean: cytogenetics, 16:289-317 disease resistance, 1:183-235 drought resistance, 4:203-243 in vitro culture, 2:225-228 nodulation, 11:275-318 photoperiodic response, 3:73-74 recurrent selection, 15:275-313 semidwarf breeding, 3:289-311 Spelt, agronomy, genetics, breeding, 15:187-213

Sprague, George F. (biography), 2:1-11

Sterility, see also Male sterility, 11:30-41

Starch, maize, 1:114-118 Statistics, history, 17:259-316 Strawberry: red stele resistance breeding, 2:195-214

transformation, 16:104 Stress resistance: cell selection, 4:141-143,161-163 transformation fruit crops, 16:115 Stylosanthes, in vitro culture, 2:238-240

Sugarcane: and Saccharum complex, 16:269-288

mutation breeding, 6:82-84 Sweet cherry, pollen-incompatibility and selffertility, 9:367-388 transformation, 16:102 Sweet corn, see also Maize:

333

CUMULATIVE SUBJECT INDEX

endosperm, 1:139-161 supersweet (shrunken2), 14:189-236

Sweet potato breeding, 4:313-345; 6:80-81 T

Tamarillo transformation, 16:107 Taxonomy: apple, 1:296-299 banana, 2:136-138 blackberry, 8:249-253 cassava, 2:83-89 chestnut, 4:351-352 chrysanthemum, 14:321-361 clover, white, 17:193-211 coffee, 2:161-163 coleus, 3:345-347 fescue, 3:314 Glycine, 16:289-317 guayule, 6:112-115 oat, 6:171-173 pearl millet, 1:1 K~-164 petunia, 1: 13 plantain, 2:136; 14:271-272 rose, 17:162-169 rutabaga, 8:221-222 Saccharum complex, 16:270-272 sweet potato, 4:320-323 triticale, 8:49-54 Vigna, 8:19-42 White clover, 17:193-211 wild rice, 14:240-241 Testing: adaptation, 12:271-297 honeycomb design, 13:87-139 Tissue culture, see In vitro culture Tobacco, gametoclonal variation, 5:372-376

Tomato: breeding for quality, 4:273-311 heat tolerance, 10:150-151 Toxin resistance, cell selection, 4:163-165

Transformation:

alfalfa, 10:190-192 cereals, 13:231-260 fruit crops, 16:87-134 mushroom, 8:206 rice, 17:179-180 somaclonal variation, 16:229-268 white clover, 17:193-211 Transpiration efficiency, 12:81-113 Transposable elements, 4:81-122; 5:146-147; 8:91-137

Tree crops, ideotype concept, 12:163-193

Tree fruits, see Fruit, nut and beverage crop breeding Trifolium, see Clover, White Clover Trifolium hybrids, 5:275-284 in vitro culture, 2:240-244 Trisomy, petunia, 1:19-20 Triticale, 5:41-93; 8:43-90 Triticum: Aestivum, see Wheat Turgidum, see Durum wheat Triticosecale, see Triticale Tulip, mutation breeding, 6:76

u United States National Plant Germplasm System, see National Plant Germplasm System Unreduced and polyploid gametes, 3:253-288; 16:15-86

Urd bean, 8:32-35

v Vaccinium, see Blueberry Vegetable breeding: artichoke, 12:253-269 bean, 1:59-102; 4:245-272 bean (tropics), 10:199-269 cassava, 2:73-134 cucumber, 6:323-359 cucurbit insect and mite resistance, 10:309-360

CUMULATIVE SUBJECT INDEX

334

Vegetable breeding (cont'd) lettuce, 1:267-293; 16:1-14 mushroom, 8:189-215 potato, 9:217-232 16:15-86 rutabaga, 8:217-248 tomato, 4:273-311 sweet corn, 1:139-161;

Weinberger, John A. (biography), 11:1-10

Wheat: anther culture, 15:141-186 chemical hybridization, 3:169-191

cold hardiness adaptation, 12:124-135

14:189-236

cytogenetics, 10:5-15 doubled haploid breeding,

sweet potato, 4:313-345 Vida, in vitro culture, 2:244-245 Vigna, see Cowpea, Mungbean in vitro culture, 2:245-246;

15:141-186

drought tolerance, 12:135-146 durum, 5:11-40 gametoclonal variation,

8:19-42

Virus disease: apple rootstocks, 1:358-359 clover, white, 17:201-209 coleus, 3:353 cowpea, 15:239-240 indexing, 3:386-408,410-411, 423-425

in vitro elimination, 2:265-282 lettuce, 1:286 raspberry, 6:247-254 resistance, 12:47-79 soybean, 1:212-217 sweet potato, 4:336 transformation fruit crops, 16:108-110

white clover, 17:201-209 Vogel, Orville A. (biography), 5:1-10

w Walnut (black), 1:236-266 Walnut transformation, 16:103

5:364-368

gene manipulation, 11:225-234 heat tolerance, 10:152 hybrid, 2:303-319; 3:185-186 in vitro adaptation, 12:115-162 molecular biology, 11 :235-250 photoperiodic response, 3:74 rust interaction, 13:293-343 triticale, 5:41-93 vernalization, 3:109 White clover, molecular genetics, 17:191-223

Wild rice, breeding, 14:237-265 Winged bean, in vitro culture, 2:237-238

z Zea mays, see Maize, Sweet corn Zein, 1:103-138 Zizania palustris, see wild rice

Cumulative Contributor Index (Volumes 1-17) Abdalla, 0.5., 8:43 Acquaah, G., 9:63 Aldwinckle, H.S., 1:294 Anderson, N.D., 10:93, 11:11 Aronson, A.I., 12:19 Ascher, P.D., 10:93 Ashri, A., 16:179 Basnizki, J., 12:253 Beck, D.L., 17:191 Beineke, W.F., 1:236 Bingham, KT., 4:123, 13:209 Binns, M.R, 12:271 Bird, R McK., 5:139 Bjarnason, M., 9:181 Bliss, F.A., 1:59; 6:1 Boase, M.R, 14:321 Borlaug, N.K, 5:1 Boyer, C.D., 1:139 Bravo, J.K, 3:193 Bressan, RA., 13:235, 14:39 Bretting, P.K., 13:11 Broertjes, c., 6:55 Brown, I.W.S., 1:59 Brown, S.K., 9:333,367 Burnham, GR, 4:347 Burton, G.W., 1:162,9:101 Byrne, D., 2:73 Campbell, K.G., 15:187 Cantrell, RG., 5:11 Carvalho, A., 2:157 Casas, A.M., 13:235 Choo, T.M., 3:219 Christenson, G.M., 7:67 Christie, B.R, 9:9

Clark, R.L., 7:95 Clarke, A.K, 15:19 Clegg, M.T., 12:1 Condon, A.G., 12:81 Cooper, RL., 3:289 Cornu, A., 1:11 Costa, W.M., 2:157 Cregan, P., 12:195 Crouch, J.H., 14:267 Crow, J.F., 17:225 Cummins, J.N., 1:294 Dana, 5.,8:19 De Jong, H., 9:217 Deroles, S.c., 14:321 Dhillon, B.S., 14:139 Dickmann, D.I., 12:163 Dodds, P.N., 15:19 Draper, A.D., 2:195 Dumas, G, 4:9 Duncan, D.R, 4:153 Echt, C.S., 10:169 Ehlers, J.D., 15:215 Evans, D.A., 3:193; 5:359 Everett, L.A., 14:237 Ewart, L.G, 9:63 Farquhar, G.D., 12:81 Fasoula, D.A., 14:89, 15:315 Fasoula, V.A., 13:87, 14:89, 15:315 Fasoulas, A.C., 13:87 Fazuoli, L.G, 2:157 Fear, GD., 11:1 Ferris, RS.B., 14:267 Flore, I.A., 12:163 335

336

Forsberg, RA., 6:167 Forster, RL.S., 17:191 French, D.W., 4:347 Galiba, G., 12:115 Galletta, G.J., 2:195 Gmitter, F.G., Jr., 8:339, 13:345 Gold, M.A., 12:163 Gradziel, T.M., 15:43 Gressel, J., 11:155 Gresshof, P.M., 11:275 Grombacher, A.W., 14:237 Grosser, J.W., 8:339 Grumet, R, 12:47 Gudin, S., 17:159 Guimaraes, C.T., 16:269 Gustafson, J.P., 5:41, 11:225 Guthrie, W.D., 6:209 Hall, A.K, 10:129, 12:81, 15:215 Hall, H.K., 8:249 Hallauer, A.R, 9:115, 14:1,165 Hamblin, J., 4:245 Hancock, J.F., 13:1 Hancock, J.R, 9:1 Hanna, W.W., 13:179 Harlan, J.R, 3:1 Hasegawa, P.M. 13:235, 14:39 Hillel, J., 12:195 Hunt, L.A., 16:135 Hutchinson, J.R, 5:181 Hymowitz, T., 8:1; 16:289 Janick, J., 1:xi Jayaram, Ch., 8:91 Johnson, A.A.T., 16:229 Jones, A., 4:313 Jones, J.S., 13:209 Ju, G.c., 10:53 Kang, H., 8:139 Kann, RP., 4:175 Karmakar, P.G., 8:19 Kartha, KK, 2:215,265 Kasha, KJ., 3:219 Keep, K, 6:245

CUMULATIVE CONTRIBUTOR INDEX

Kleinhofs, A., 2:13 Knox, RR, 4:9 Kollipara, K.P., 16:289 Konzak, C.F., 2:13 Kononowicz, A.K, 13:235 Krikorian, A.D., 4:175 Krishnamani, M.RS., 4:203 Kronstad, W.K, 5:1 Lamkey, K.R, 15:1 Lavi, D., 12:195 Layne, RKC., 10:271 Lebowitz, RJ., 3:343 Levings, III, C.S., 10:23 Lewers, KR, 15:275 Li, J., 17:1,15 Liedl, RK, 11:11 Lin, C.S., 12:271 Lovell, G.R, 7:5 Lukaszewski, A.J., 5:41 Lyrene, P.M., 5:307 McCoy, T.J., 4:123; 10:169 McCreight, J.D., 1:267; 16:1 McDaniel, RG., 2:283 McRae, D.H., 3:169 Maheswaran, G., 5:181 Marcotrigiano, M., 15:43 Maizonnier, D., 1:11 Martin, F.W., 4:313 Medina-Filho, H.P., 2:157 Miller, R, 14:321 Morrison, RA., 5:359 Mowder, J.D., 7:57 Mroginski, L.A., 2:215 Murphy, A.M., 9:217 Mutschler, M.A., 4:1 Myers, 0., Jr., 4:203 Namkoong, G., 8:139 Neuffer, M.G., 5:139 Newbigin, K, 15:19 Ortiz, R, 14:267; 16:15 Palmer, RG., 15:275

337

CUMULATIVE CONTRIBUTOR INDEX

Pandy, S., 14:139 Parliman, B.J., 3:361 Paterson, A.H., 14:13 Pedersen, J.F., 11:251 Perdue, RE., Jr., 7:67 Peterson, P.A., 4:81; 8:91 Porter, RA., 14:237 Proudfoot, K.G., 8:217

SIeper, D.A., 3:313 Smith, S.E., 6:361 Sodas i Company, R, 8:313 Sobral, B.W.S., 16:269 Sandahl, M.R, 2:157 Stevens, M.A., 4:273 Stoner, A.K., 7:57 Stuber, C.W., 9:37; 12:227

Raina, S.K., 15:141 Ramage, RT., 5:95 Ramming, D.W., 11:1 Ray, D.T., 6:93 Redei, G.P., 10:1 Reimann-Phillipp, R, 13:265 Reinbergs, E., 3:219 Rhodes, D., 10:53 Richards, RA., 12:81 Roath, W.W., 7:183 Robinson, RW., 1:267; 10:309 Ron Parra, J., 14:165 Roos, E.E., 7:129 Rowe, P., 2:135 Russell, W.A., 2:1 Rutter, P.A., 4:347 Ryder, E.J., 1:267

Tai, G.c.c., 9:217 Talbert, L.E., 11:235 Tarn, T.R, 9:217 Tehrani, G., 9:367 Thompson, A.E., 6:93 Towill, L.E., 7:159, 13:179 Tracy, W.F., 14:189 Tsai, c.Y., 1:103

Samaras, Y., 10:53 Sansavini, S., 16:87 Saunders, J.W., 9:63 Sawhney, RN., 13:293 Schaap, T., 12:195 Scott, D.H., 2:195 Seabrook, J.E.A., 9:217 Sears, E.R, 11:225 Shands, Hazel L. 6:167 Shands, Henry L. 7:1,5 Shannon, J.C., 1:139 Shattuck, V.I., 8:217, 9:9 Shaun, R, 14:267 Sidhu, G.S., 5:393 Simmonds, N.W., 17:259 Singh, RB., 15:215 Singh, RJ., 16:289 Singh, S.P., 10:199 Singh, Z., 16:87

Ullrich, S.E., 2:13 Van Harten, A.M., 6:55 Varughese, G., 8:43 Vasal, S.K., 9:181, 14:139 Veilleux, R, 3:253; 16:229 Villareal, RL., 8:43 Vogel, K.P., 11:251 Vuylsteke, D., 14:267 Wallace, D.H., 3:21, 13:141 Wan, Y., 11:199 Weeden, N.F., 6:11 Wehner, T.C., 6:323 Westwood, M.N., 7:111 Whitaker, T.W., 1:1 White, D.W.R, 17:191 White, G.A., 3:361; 7:5 Widhalm, J.M., 4:153,11:199 Widmer, RE., 10:93 Widrlechner, M.P., 13:11 Wilcox, J.R, 1:183 Williams, E.G., 4:9; 5:181,237 Williams, M.E., 10:23 Wilson, J.A., 2:303 Woodfield, D.R, 17:191 Wright, G.c., 12:81 Wu, L., 8:189

338

CUMULATIVE CONTRIBUTOR INDEX

Xin, Y., 17:1,15 Xu, Y., 15:85

Yopp, J.H., 4:203 Yun, D.-J., 14:39

Yan, W., 13:141 Yang, W.-J., 10:53

Zimmerman, M.J.a., 4:245 Zohary, D., 12:253