• TABLE OF CONTENTS
HIDE
 Front Cover
 Title Page
 Table of Contents
 Foreword
 Acknowledgement
 Yield variability in cereals: Concluding...
 Part I. Overview of issues
 Part II. Evidence and cause of...
 Part III. Approaches to reducing...
 Abstracts of workshop papers
 Workshop participants
 Back Cover














Title: Summary proceedings of a workshop on cereal yield variability
CITATION THUMBNAILS PAGE IMAGE ZOOMABLE
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/UF00085383/00001
 Material Information
Title: Summary proceedings of a workshop on cereal yield variability
Physical Description: Book
Publisher: International Food Policy Research Institute,
Copyright Date: 1986
 Record Information
Bibliographic ID: UF00085383
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: 14242614 - OCLC

Table of Contents
    Front Cover
        Page i
        Page ii
    Title Page
        Page iii
        Page iv
    Table of Contents
        Page v
        Page vi
        Page vii
        Page viii
    Foreword
        Page ix
        Page x
    Acknowledgement
        Page xi
    Yield variability in cereals: Concluding assessment
        Page 1
        Page 2
        Page 3
        Page 4
        Page 5
        Page 6
        Page 7
        Page 8
        Page 9
        Page 10
        Page 11
        Page 12
    Part I. Overview of issues
        Page 13
        Page 14
        Page 15
        Page 16
        Page 17
        Page 18
        Page 19
        Page 20
        Page 21
        Page 22
        Page 23
        Page 24
        Page 25
        Page 26
        Page 27
        Page 28
        Page 29
        Page 30
        Page 31
        Page 32
        Page 33
        Page 34
        Page 35
        Page 36
        Page 37
        Page 38
        Page 39
        Page 40
        Page 41
        Page 42
        Page 43
        Page 44
        Page 45
        Page 46
        Page 47
        Page 48
        Page 49
        Page 50
        Page 51
        Page 52
        Page 53
        Page 54
        Page 55
        Page 56
        Page 57
        Page 58
        Page 59
        Page 60
        Page 61
        Page 62
        Page 63
        Page 64
        Page 65
        Page 66
        Page 67
        Page 68
        Page 69
        Page 70
        Page 71
        Page 72
        Page 73
        Page 74
        Page 75
        Page 76
        Page 77
        Page 78
        Page 79
        Page 80
        Page 81
        Page 82
        Page 83
        Page 84
        Page 85
        Page 86
        Page 87
        Page 88
        Page 89
        Page 90
        Page 91
        Page 92
        Page 93
        Page 94
        Page 95
        Page 96
        Page 97
        Page 98
        Page 99
        Page 100
        Page 101
        Page 102
        Page 103
        Page 104
        Page 105
        Page 106
        Page 107
        Page 108
        Page 109
        Page 110
        Page 111
        Page 112
    Part II. Evidence and cause of change in yield variability and yield correlations
        Page 113
        Page 114
        Page 115
        Page 116
        Page 117
        Page 118
        Page 119
        Page 120
        Page 121
        Page 122
        Page 123
        Page 124
        Page 125
        Page 126
        Page 127
        Page 128
        Page 129
        Page 130
        Page 131
        Page 132
        Page 133
        Page 134
        Page 135
        Page 136
        Page 137
        Page 138
        Page 139
        Page 140
        Page 141
        Page 142
        Page 143
        Page 144
        Page 145
        Page 146
        Page 147
        Page 148
        Page 149
        Page 150
        Page 151
        Page 152
        Page 153
        Page 154
        Page 155
        Page 156
        Page 157
        Page 158
        Page 159
        Page 160
        Page 161
        Page 162
        Page 163
        Page 164
        Page 165
        Page 166
        Page 167
        Page 168
        Page 169
        Page 170
        Page 171
        Page 172
        Page 173
        Page 174
        Page 175
        Page 176
        Page 177
        Page 178
        Page 179
        Page 180
        Page 181
        Page 182
        Page 183
        Page 184
        Page 185
        Page 186
        Page 187
        Page 188
        Page 189
        Page 190
    Part III. Approaches to reducing yield variability
        Page 191
        Page 192
        Page 193
        Page 194
        Page 195
        Page 196
        Page 197
        Page 198
        Page 199
        Page 200
        Page 201
        Page 202
        Page 203
        Page 204
        Page 205
        Page 206
        Page 207
        Page 208
        Page 209
        Page 210
        Page 211
        Page 212
        Page 213
        Page 214
        Page 215
        Page 216
        Page 217
        Page 218
        Page 219
        Page 220
        Page 221
        Page 222
        Page 223
        Page 224
        Page 225
        Page 226
        Page 227
        Page 228
        Page 229
        Page 230
        Page 231
        Page 232
    Abstracts of workshop papers
        Page 233
        Page 234
        Page 235
        Page 236
        Page 237
        Page 238
        Page 239
        Page 240
        Page 241
        Page 242
        Page 243
        Page 244
        Page 245
        Page 246
        Page 247
        Page 248
        Page 249
        Page 250
        Page 251
        Page 252
        Page 253
        Page 254
        Page 255
        Page 256
        Page 257
        Page 258
        Page 259
        Page 260
        Page 261
        Page 262
        Page 263
        Page 264
        Page 265
        Page 266
        Page 267
        Page 268
        Page 269
        Page 270
        Page 271
        Page 272
    Workshop participants
        Page 273
        Page 274
        Page 275
        Page 276
        Page 277
    Back Cover
        Page 278
        Page 279
Full Text





Summary Proceedings of a Workshop
on Cereal Yield Variability







Summary Proceedings of a Workshop
on Cereal Yield Variability

edited by Peter B. R. Hazell


International Food Policy Research Institute (IFPRI)
Washington, D.C.
Deutsche Stiftung for Internationale Entwicklung/
Zentralstelle fur Erndhrung und Landwirtschaft (DSE/ZEL)
Feldafing



































Published in 1986 by the

International Food Policy Research Institute
1776 Massachusetts Avenue, N.W.
Washington, D.C. 20036
U.S.A.


Library of Congress Cataloging-in-Publication Data

Summary proceedings of a workshop on cereal yield
variability.

Includes bibliographies.
1. Grain--Congresses. 2. Crop yields--Congresses.
I. Hazell, P. B. R. II. International Food Policy
Research Institute. III. Deutsche Stiftung fur
International Entwicklung. IV. Title: Cereal yield
variability. V. Title: Yield variability.
SB188.2.S86 1986 633.1 86-21078 DOK 1343 A/a
ISBN 0-89629-306-8 IT 73-013-85













Contents


FOREWORD John W. Mellor and Erhard Krusken

ACKNOWLEDGMENTS

1. YIELD VARIABILITY IN CEREALS: CONCLUDING ASSESSMENT

Lloyd T. Evans

PART I. OVERVIEW OF ISSUES

2. INTRODUCTION

Peter B. R. Hazell

3. CLIMATIC CHANGES AND YIELD VARIABILITY

Timothy R. Carter and Martin L. Parry

4. GENETIC ASPECTS OF YIELD VARIABILITY

John H. Holden


5. YIELD VARIABILITY AND THE TRANSITION OF THE NEW
TECHNOLOGY

H. K.Jain, M. Dagg, and T. A. Taylor


6. YIELD VARIABILITY AND INCOME, CONSUMPTION, AND FOOD
SECURITY

David E. Sahn and Joachim von Braun






PART II: EVIDENCE AND CAUSE OF CHANGE IN YIELD
VARIABILITY AND YIELD CORRELATIONS


WHEAT


7. CIMMYT PRESENTATION: YIELD STABILITY IN BREAD WHEAT 115

Arthur Klatt, W. H. Pfeiffer, and H. J. Braun


8. SUMMARY AND ASSESSMENT OF THE WHEAT PAPERS 127

C. James Peterson


9. REPORT OF THE WORKING GROUP ON WHEAT 131

Roger B. Austin


RICE


10. IRRI PRESENTATION: YIELD STABILITY AND MODERN RICE 133
TECHNOLOGY

John C. Flinn and D. P. Garrity


11. SUMMARY AND ASSESSMENT OF THE RICE PAPERS 147

Randolph Barker


12. REPORT OF THE WORKING GROUP ON RICE 153

Gerald A. Carlson


MAIZE


3. CIMMYT PRESENTATION: YIELD STABILITY OF CIMMYT MAIZE 157
GERMPLASM

H. N. Pham, S. R. Waddington, and J. Crossa


14. SUMMARY AND ASSESSMENT OF THE MAIZE PAPERS 161

Jock R. Anderson







15. REPORT OF THE WORKING GROUP ON MAIZE 165

Donald Winkelmann


SORGHUM AND MILLETS

16. ICRISAT PRESENTATION: YIELD VARIABILITY IN SORGHUM 167
AND MILLET

Thomas S. Walker and John R. Witcombe


17. SUMMARY AND ASSESSMENT OF THE SORGHUM AND MILLET PAPERS 177

Jock R. Anderson


18. REPORT OF THE WORKING GROUP ON SORGHUM AND MILLET 179

John R. Witcombe


BARLEY


19. SUMMARY AND ASSESSMENT OF THE BARLEY PAPERS 181

M. S. Mekni

20. REPORT ON THE WORKING GROUP ON BARLEY 185

G. Fischbeck


21. CROP VARIETIES AND YIELD VARIATION: A SYNTHESIS 187

Michael H. Arnold


PART III: APPROACHES TO REDUCING YIELD VARIABILITY


BREEDING


22. SUMMARY AND ASSESSMENT OF THE PAPERS ON BREEDING 193

Donald N. Duvick


23. REPORT OF THE WORKING GROUP ON BREEDING 199

C. James Peterson







INPUT MANAGEMENT


24. SUMMARY AND ASSESSMENT OF THE PAPERS ON 203
INPUT MANAGEMENT

Robert W. Herdt


25. REPORT OF THE WORKING GROUP ON INPUT MANAGEMENT 209

J. C. Headley


FARMING SYSTEMS


26. SUMMARY AND ASSESSMENT OF THE PAPERS ON FARMING SYSTEMS 211

Ulrich von Poschinger-Camphausen


27. REPORT OF THE WORKING GROUP ON FARMING SYSTEMS 215

Wolfgang Vogel


PUBLIC POLICY


28. SUMMARY AND ASSESSMENT OF THE PAPERS ON PUBLIC POLICY 217

Hartwig de Haen


29. REPORT OF THE WORKING GROUP ON PUBLIC POLICY 227

John R. Tarrant


30. APPROACHES TO REDUCING YIELD VARIABILITY: A SYNTHESIS 231

Winfried von Urff


ABSTRACTS OF WORKSHOP PAPERS 235


WORKSHOP PARTICIPANTS 273
















Foreword


John W. Mellor and Erhard Krusken



Many countries have achieved impressive rates of growth in
national foodgrain production in recent years. Much of this growth
can be attributed to new technologies and the increased use of modern
inputs, such as fertilizers. At the same time, the variability of
world foodgrain production around trend also increased as measured by
the variance or the coefficient of variation of production. This
increased variability is reflected in increased market and price
instability, which poses difficult problems for farmers and poor
consumers alike. It also increases the size of emergency food stocks
that need to be carried by governments to ensure that consumption
does not fall precipitously below trend.
Research by the International Food Policy Research Institute
(IFPRI) on countries and crops shows that in most cases increases in
yield variability and, more importantly, a loss in offsetting
patterns of variation (increased correlations) in crop yields between
regions are the predominant sources of the increase in production
variability.
There has been a tendency by some researchers to attribute this
increased yield variability to improved seed/fertilizer based
technologies. Some researchers have also argued that plant breeders
should focus less on maximizing average yields and more on reducing
yield sensitivity to environmental stress. Such recommendations may
prove costly for future growth in foodgrain production, and they
cannot be warranted before more thorough and quantitative studies of
the sources of increased variability have been undertaken.
In view of the importance of this issue to national breeding
programs and to the international agricultural research centers, the
Deutsche Stiftung fUr Internationale Entwicklung (DSE) and IFPRI
convened an interdisciplinary workshop for an intensive four-day
discussion of a broad range of issues associated with increasing
yield variability. There were about 60 participants, comprising
biologists, social scientists, and policy makers, and with particu-
larly strong representation from the centers of the Consultative
Group on International Agricultural Research (CGIAR).
Workshop participants discussed the relationship between changes
in yield variability and yield correlations and such causal factors
as changes in agricultural technology, weather, irrigation, input
availability, and related variables. They also discussed the
consequences of increasing yield variability, including its effect on
different types of farmers and on poor urban and rural consumers.








Participants were asked to make specific recommendations for agricul-
tural research policy in the fields of plant breeding, farming
systems, and management of irrigation, fertilizers, and pesticides,
and to address the need for changes in national and international
agricultural policies.
This volume summarizes the workshop discussions and includes an
assessment of findings prepared by a review panel chaired by Lloyd T.
Evans. It also contains summaries and abstracts of thirty-five
papers prepared specifically for the workshop. Selected papers from
the workshop plus papers commissioned to fill important gaps will
subsequently be published as a separate book by IFPRI and DSE. It is
our hope that these proceedings will stimulate debate and further
research on the important topic of yield variability and that it will
lead to improved policies and agricultural research priorities for
coping with yield risks in the future.


John W. Mellor Erhard Krusken
(IFPRI) (DSE)















Acknowledgments


The publication of these Proceedings marks the end of a three-
year planning and organizational effort by me and Klaus Klennert. I
am greatly indebted to Klaus for his dedicated support, and to all
those who helped us in our planning task. We are especially grateful
to Randolph Barker, G. Fischbeck, John W. Mellor, H. Schmutterer,
Joachim von Braun, Ulrich von Poschinger-Camphausen, Winfried von
Urff, Thomas S. Walker, Trevor Williams, and Donald Winkelmann, who
served at various times on a planning advisory committee.
The success of the workshop owes much to the thorough adminis-
trative arrangements made by the staff of DSE's Food an Agriculture
Development Center in Feldafing, where the meetings were held.
Special thanks go to Erhard Krisken, Klaus Klennert, and Mrs. Dorre
for arranging the workshop, and to George Bauer and Uwe Krappitz who
provided valuable assistance with visual aids.
Preparation for this volume is possible only because of the
efforts of the many participants who wrote down their comments and
reported on working group sessions. Lloyd T. Evans chaired a review
panel and performed the miraculous task of synthesizing the findings
and recommendations of the workshop in a cohesive statement.
Finally, financial support provided by the Ministry for Economic
Cooperation of the Federal Republic of Germany is gratefully acknowl-
edged.

Peter B. R. Hazell













Yield Variability in Cereals:
Concluding Assessment*

Lloyd T. Evans


My experience has been more with the instability of data than
with data on instability. Variability in cereal yields and produc-
tion constitutes an important problem in relation to world agricul-
ture and food supplies, even if it turns out that there is no clear
evidence that the relative variability has been changed by modern
varieties and agronomic practices. Whatever the causes of the
variability, research on them should lead to more effective national
and international management.
Several earlier analyses of the problem (for example, Mehra
1981; Barker et al. 1981) suggest that some aspects of the new
agricultural technology resulted in greater variability in yield, but
Hazell's subsequent work (1982, 1985) shows that the increased
variability in world cereal production is due not only to greater
yield variability but also to a reduction in the offsetting patterns
of variation in yields between crops and regions. It is timely,
therefore, that IFPRI and DSE have provided this opportunity for a
comprehensive consideration of the issues, and with a particular
focus on cereal yields. Feast or famine in many parts of the world
depends on cereal yields, and their variation has been the subject of
a range of analyses (for example, Thompson 1969, 1975; Luttrell and
Gilbert 1976; Stanhill 1976; Church and Austin 1983; Gales 1983).
The answer that emerges from this workshop is complex, varying
by crop, by country, and by stage of technological change, but what
is clear is that, whereas in many cases there has not been an
increase in the coefficient of variation for production with the new
technology, there has nevertheless been an increase in covariance
across regions, countries, and crops. In fact some of the practices
that decrease variance at the farm level may increase covariance at
the regional level.
Hazell's analysis for this workshop suggests that the probabil-
ity of a 5 percent shortfall in total cereal production may have
doubled in recent years. Even though the effect of such shortfalls
on per capital consumption may be buffered (Sahn and von Braun,
Chapter 6), this increased probability constitutes a major problem in
need of further research. At this stage of the analysis, however,
the "green revolution" should not be blamed for the problem, a

*The panel consisted of Jock R. Anderson, Nazmi Demir, Lloyd T.
Evans (chairman), G. Fischbeck, Eduardo Venezian, and Donald
Winkelmann.










conclusion supported by the fact that the coefficients of variation
for world production of both wheat and rice, the flag bearers of the
green revolution, have in fact decreased (Hazell, Chapter 2).

SOME METHODOLOGICAL PROBLEMS

Before we dissect the anatomy of cereal production and yield
variability, some methodological problems should be mentioned.
In general, as production and yield levels rise, we should
expect the absolute variance also to rise, more or less in propor-
tion. For some issues, such as the management of surpluses and
international trade and their effects on prices, it is the absolute
variance that poses the problem and provides the relevant measure.
But for comparisons between crops, varieties, countries, environ-
ments, and levels of input, it is the size of the variance relative
to yield or production -- the coefficient of variation (cv) -- that
is most useful. However, even the cv of the many cvs presented at
the workshop would be high! The answers we get may also depend on
how the variances are estimated, whether in relation to trends or
expectations, whether by farmers or government, and whether detrended
or pooled (Bindlish et al., Workshop Paper 3).
The point should also be made, as it was by Jock Anderson, that
instability and risk are not the same thing, and that real risks may
not be related directly to variance. A random variable of interest
to a decision maker at the policy or farm level can increase in both
variance and cv and yet be less risky than the previous situation.
But the judgement of such changing riskiness depends on more compre-
hensive comparisons of distribution functions and on assumptions
about utility preferences. The price of relevance may therefore be a
more comprehensive estimation of the uncertainties involved.
There are also major problems estimating variances before and
after the technological change or, more correctly, during the earlier
and later phases of such change. Given the extent of climatic
variability and its impact on cereal yields, consecutive periods of
10 to 15 years are barely enough to yield reliable evidence of
trends. Yet longer posttechnology periods are often not available.
The choice of boundary years can affect conclusions (compare Anderson
and Findlay, Workshop Paper 1), and the basis and reliability of much
of the statistical data can itself change in the course of the two
periods.
But beyond these problems there is the need to supplement the
statistical data with local knowledge and assessments -- to read the
newspapers, as Randy Barker put it -- in order to know "when a new
irrigation scheme came on stream; or what year the pumps failed; or
when fertilizers were scarce; or when moth and rust corrupted; or
when extensive marginal areas came in or out of production; or when
crop locations changed."
With such complications in mind, there was no general agreement
with the proposition that the cv of cereal production had been
increased by the new technology (or the green revolution). Nor is
there clear evidence that, in aggregate, variability in cereal
consumption per capital had increased (Sahn and von Braun, Chapter 6).










CLIMATIC CHANGE

A major source of yield variance in all cases, but especially in
cereals grown in more arid areas, is the variability in crop weather.
Carter and Parry (Chapter 3) conclude, however, that there is no
indication that recent changes in cereal yield variability can be
ascribed to climatic change.
Interannual variations -- such as those associated with the El
Nifo/Southern Oscillation phenomenon or with the sub-Saharan droughts
of 1972, 1977, and 1983/84 -- have certainly influenced global cereal
production and variability. But for changes in variance and covar-
iance in recent years it is necessary to look elsewhere for the
causal factors, even though long-term climatic changes associated
with rising atmospheric carbon dioxide levels are likely to have
important implications for cereal production in the future.

THE INTERACTION OF GENETIC AND AGRONOMIC IMPROVEMENT

In the few cases in which the relative contributions of genetic
and agronomic improvement to yield increase have been estimated,
plant breeding and agronomy have contributed about equally overall,
although the proportions differ with the stage of advance (Evans
1984).
Varietal improvement often acts as the catalyst beginning the
process, the Trojan horse for the new technology, but both genetic
and agronomic improvement are needed for sustained increases in
yield.

SOME CHARACTERISTICS OF GENETIC IMPROVEMENT

Adaptability. Shortening the life cycle of cereal crops and reducing
their sensitivity to seasonal signals, such as day length, allow
crops to perform more evenly across a range of sites, latitudes, and
climates, thereby increasing their adaptability. So too does wider
tolerance of soil conditions.

Hardiness. Another important characteristic is the ability of cereal
varieties to withstand drought, cold, heat, and other climatic
insults, especially at the most sensitive stages of the life cycle.
Such hardiness is sometimes highly specific, sometimes general.
Specific resistance to extremes of heat or cold has been improved in
many crops -- for example, rice and millet in Japan to cold -- and
although the changes may seem small in a physiological sense, they
may be of considerable significance in reducing downside variability.
Hybrids may exhibit a more general hardiness in that, although they
may be no more productive than inbreds under optimal conditions, they
may perform substantially better than their parents at both high and
low temperatures (McWilliam and Griffing 1965; McWilliam et al.
1969).

Reduced Vulnerability. The incorporation in cereal varieties of
genetic resistance -- wide or narrow -- to the current biotypes of










pests and diseases is a major preoccupation of plant breeders and a
major contribution toward yield stability. In general, reduced
vulnerability has been easier to achieve than resistance to climatic
stresses.

Responsiveness. Another desirable characteristic is the ability of a
variety to give a return of greater yield or improved quality on
favorable conditions or higher inputs.

Competitiveness. Competitiveness is desirable especially in marginal
environments or where weed problems are serious.
All of these genetic improvements can influence the variability
of yield (responsiveness especially on the upside, hardiness and
reduced vulnerability on the downside). However, they are not always
compatible with one another, and trade-offs between them must be
decided by the plant breeder -- for example, between hardiness and
responsiveness or, especially in the case of tall versus dwarf
selections, between competitiveness and responsiveness.

STABILITY

Another characteristic more talked of than understood is
stability. In the ecological literature, stability is a quite
complex concept, implying not that the plant community is unmoving
but rather that its response to change is muted by negative feedback
reactions and that it tends to return to its former equilibrium after
disturbance (Rindos 1984). The conventional wisdom is that complex-
ity begets stability, but in fact complex communities such as rain
forests can be quite fragile, while monocultures such as crops can be
quite stable.
Plant breeders' usage of the term and their ways of estimating
stability are so varied, it is difficult to know what they mean by
it. In general, but by no means always, they mean stability of
performance across years at a particular site, a matter of real
concern to farmers and policy makers. All the characteristics
mentioned previously, except responsiveness, contribute positively to
such stability.
Stability came into fashion among plant breeders with the
widespread use of the method of analysis used by Yates and Cochran
(1938), Finlay and Wilkinson (1963), and Eberhart and Russell (1966).
Their method is a convenient and seductive way of presenting multi-
location, multivariety trial results of global plant breeding
programs.
But this form of analysis can be misleading (especially when
only the regressions and not the data points are presented); it can
obscure valuable site-specific adaptation; it is too open to the
selective presentation of data; it does not tell us how to deal with
trade-offs among mean yield, regression slope, and the variance from
that; and it tends to be rather unhelpful at the low mean-yield sites
characteristic of on-farm conditions, especially in developing
countries. The regression (or b value) is not a fixed varietal
characteristic, as Peterson et al. (Workshop Paper 22) show for
Kharkov wheat, where it has fallen progressively over the years










(presumably because it is estimated in relation to the average for
all the varieties in the trial and because the responsiveness of the
new entries has risen over the years).
It is clearly time to explore other forms of data assessment,
such as multivariate and cluster analysis and stochastic dominance
(compare Witcombe, Workshop Paper 35), which could overcome these
failings. We hope the international agricultural research centers
will give more attention to these in the future. In the meantime,
however, there are three points to make in the context of the usual
form of analysis.

Breeding for Responsiveness

Ignoring for the moment the problems of crossovers in varietal
performance at the poorer sites and the need to minimize vulnerabil-
ity to pests and diseases, the major element of yield improvement is
increased responsiveness -- that is, the ability to reward favorable
sites or years and high inputs with progressively higher yields per
crop, or per day in the tropics.
Inevitably, such emphasis will accentuate the upside variability
problems, but it would be unrealistic to expect plant breeders to
place a moratorium on such improvements or to expect farmers not to
take advantage of them. Indeed, improvements should be welcomed as
enlarging the overall potential food supply, whatever management
problems they may create.

Breeding for Marginal Conditions

There is less agreement about breeding for the most marginal,
lowest yield sites. Varietal improvement under these conditions is
difficult and will be slower, less certain, and more costly in terms
of plant breeding effort. But it can be achieved, as examples for
most cereals indicate. Improved tolerance for drought and heat
stresses (U.S. maize hybrids), greater tolerance of adverse soil
conditions (IRRI rice), more efficient performance in low-nutrient
conditions (Mahsuri rice), greater resistance to Striga or downy
mildew, and many other characteristics have already improved cereal
performance under poor conditions. Even a small improvement may
substantially affect adoption (finger millet in India).
Yet various factors tend to discourage plant breeding for poor
environments. Gains are not spectacular and may be seen as having
little effect on the variability of cereal production. Many farm
conditions may be even poorer than the poorest experimental test
sites, and their conditions may also be inherently more variable and
diverse, leading to greater site specificity. Government policies
for varietal testing and release may discourage such work, as may
policies restricting the allocation of fertilizers to such areas.
We recommend, therefore, that the IARCs give more attention to
this complex of problems and recognize that such work may require
fuller analysis of on-farm constraints in these areas and that
agronomic improvements may be a necessary prelude to genetic improve-
ment.










Crossovers in Performance

For many CIMMYT wheat varieties (Pfeiffer and Braun, Workshop
Paper 23) and U.S. maize hybrids (Duvick, Workshop Paper 6), superi-
ority in favorable conditions appears to be associated with superi-
ority (or at least not inferiority) at the poorest test sites. But
there are also cases where a clear crossover in relative performance
occurs, as may be seen in the original analysis of barley varieties
by Finlay and Wilkinson (1963), which particularly reflects cultivar
differences in length of life cycle vis-a-vis length of the growing
season. Walker and Witcombe's data for pearl millet (Workshop Paper
35) provide another significant example, as does that extracted from
CIMMYT yield trials by Laing and Fischer (1977). Thus it could be
well worthwhile to select varieties that are superior only under poor
conditions. Most plant breeders feel that such efforts would not
make best use of their time, but more breeding work dedicated to such
objectives could be justified.

GENETIC VULNERABILITY

The extent to which the genetic base of modern cereal varieties
and hybrids influences the downside risks is difficult to assess.
Outbreaks of pests and diseases have had an effect, sometimes
disastrous, throughout recorded history. Problems still occur --
for example, with downy mildew on millet in India (compare Walker and
Witcombe, Chapter 16) -- but major disasters, like the earlier stem
rust epidemics in North American wheat crops, have been contained in
recent years. Southern corn leaf blight on T-cytoplasm maize hybrids
was pandemic in 1970, but within a year the genetic base was changed
enough to deal with the pathogen. Other problems loom as possible
threats, such as the lack of resistance in IR-36 rice to brown plant
hopper biotype 2, and in some CIMMYT varieties to leaf rust, or the
widespread cytoplasmic uniformity of IRRI rices (Coffman and Har-
grove, Workshop Paper 5), but replacement varieties are already in
reserve. Breeding for pest and disease resistance is now so sophis-
ticated that rapid turnover of varieties in time substitutes for many
traditional varieties used at one time.
However, the fact that several wheat and rice varieties, such as
Bezostaia wheat in Eastern Europe and IR-36 rice in Asia, are grown
on more than 10 million hectares inevitably means that their sudden
failure would raise the covariance in yield, as could their similar
response to weather conditions common to a large region. This
element of covariance may, however, decline in the future as plant
breeding -- whether public, private, or in the IARCs -- evolves
toward greater emphasis on regional and local adaptation.
Three other points should be made. First, the breadth of the
genetic base is not simply proportional to the number of varieties in
present use: many varieties and hybrids may be closely related. On
the other hand, modern varieties often bring together an extremely
wide range of genotypes in their ancestry. Second, there is, by and
large, no direct trade-off between comprehensive resistance to pests
and diseases on the one hand and yield potential on the other,
although there may be a trade-off between resistance to climatic










stress and yield potential. Yield advance may, however, be slower
when selection for many pest and disease resistances has to be
practiced. Third, in some crops (pearl millet, for example) genetic
uniformity may be more limiting to the improvement of adaptability
and hardiness than to the reduction of vulnerability.


AGRONOMIC INPUTS

Agronomic inputs are as significant as genotype to cereal
production and stability, yet they received far less attention at the
workshop, where breeders outnumbered agronomists.
In general, it seems likely that variability in yield is
exacerbated during the early stages of more widespread and heavier
use of a particular input, but then falls as its use becomes uniform
and as its rate of application approaches the response asymptote.
For example, variability of wheat crops in the Punjab fell as tube
well irrigation became more extensive (Mehra 1981), but limited,
uneven, and unreliable irrigation of dry season rice crops in the
Philippines increased variability (Rosegrant, Workshop Paper 26).
With nitrogenous fertilizer application to wheat and barley,
Hanus and Schoop (Workshop Paper 12) found that the yield variance
changed little as yields rose in response to heavier applications, so
that the cv fell markedly. However, at the heaviest application
rates, the variance rose as diseases increased, and it was reduced
only by the application of fungicides.
Such interactions between inputs on yield variability deserve
more attention, because the development of modern agriculture has
involved a sequence of inputs, each of which has successively rescued
the yield response from the asymptote for earlier inputs, and the
variances to each might interact in a complex way, as Austin and
Arnold (Workshop Paper 2) indicate in their binomial model.
Conflicting forces may be at work as agriculture becomes more
intensive. Variability tends to fall as agronomic control of the
environment becomes more complete, as in the case of wheat in Western
Europe and the Punjab. But selection for higher yield is dependent
on enhanced agronomic support for the crop, and when this is unreli-
able the higher yielding varieties may be vulnerable to greater
variation. In general, however, there may be considerable scope for
the reduction of variability by more flexible, better informed, and
more diversified and specific use of inputs.


DIFFERENCES AMONG CROPS

The considerable differences in variability among the major
cereal crops, evident in Hazell's overview (Chapter 2), probably
reflect differences in growing conditions rather than differences
among species.
Intensive irrigation and deep bunding probably account for rice
having the lowest cv of all cereals in both periods. Likewise, an
increasing proportion of the world's wheat crop is grown under
irrigation and with high inputs. At the other end of the scale are
the millets, grown in marginal conditions and with low and variable










rainfall. Moreover, there is a tendency in many arid regions for
maize to push sorghum and for sorghum to push millet toward the least
favorable environments. Likewise, barley may be pushed by wheat into
more marginal environments, as Fischbeck (Workshop Paper 7) shows,
and this could account for the rise in its cv.
The rise in the variability of maize production is more
puzzling. The closer synchronization of maize plantings across the
U.S. corn belt, described by Duvick (Workshop Paper 6), may have
contributed to the substantial rise in the cv of maize yields in the
United States. But the cv of U.S. maize production has not in-
creased, and the greater variability in world maize production in
recent years appears to be associated with those countries where
maize production is growing rapidly. Another factor contributing to
greater variability in some crops may be growing them under both
intensive, irrigated conditions and marginal, dry land conditions,
the relative proportions of which may change from year to year (as
with wheat in several countries).

FARMING SYSTEMS

Research in farming systems also merits more attention, for its
holistic approach to the problem of production variability and for
its potential in reducing downside variability and risk. Greater
diversification of crops, varieties, and practices has been empha-
sized, but such practices as more flexible operations (like reduced
tillage), better fertility maintenance, and better on-farm storage
could also reduce variability. Research more clearly targeted on
these objectives is needed, particularly in high-risk environments.

COVARIANCE: THE TOGETHERNESS PROBLEM

Even if some doubt remains as to whether increased variance of
cereal production and yield is significant, or is applicable only to
certain countries and crops, or reflects only a transitional stage,
there is clear agreement that increased covariances across crops,
regions, and nations merit further analysis. However, there is no
agreed notion of how this covariance should be apportioned among
weather patterns, input supplies, varietal responses, agronomic
practices, irrigation, and price signals.
The increasing synchronization of crop life cycles across
countries and regions made possible by better weather.forecasting,
mechanization, inputs, and varietal homogeneity, as with maize in the
United States (Duvick, Workshop Paper 6), may contribute to covari-
ance. But whereas synchronization of the crops in a region may make
them all susceptible to extremes of heat or cold or drought at
particular stages, it may also spread the risk of losses from birds
or rodents, as with rice crops in Asia. And even when irrigation,
better information, or availability of inputs help to reduce varia-
tion at the farm level, they may increase covariance at the regional
level.
Indeed, such covariation is surely to be expected more and more
as agriculture becomes better informed, its infrastructure improves,










and it becomes more responsive to opportunities on a global scale.
In that context, it seems more effective to focus attention on
policies aimed at mitigating the adverse socioeconomic effects of
covariation rather than expecting plant breeders and agronomists to
solve the problem. While many other aspects of the variability
problem merit further research, the highest priority should be given
to socioeconomic policy research aimed at reducing the adverse
effects of cereal production variability.

POLICY ASPECTS

As farmers become more responsive, as trade grows, and as the
probability of shortfalls increase for some countries, variability
will increase and policy problems will grow. There is no one optimal
solution. Policies will vary with the magnitude of variability, the
level of the economy, the extent of urbanization, the amount of
poverty, the grain storage capacity, whether a country imports or
exports cereals, and so on.
Evidence of changes in the variability of cereal production in
the centrally planned economies, especially the U.S.S.R. but also
countries such as Egypt and Syria, suggests that central planning
does not overcome the problem. However, national policies to assure
input supplies (such as improved varieties, fertilizers, and irriga-
tion water), to distribute them more widely, and to stabilize prices
should help reduce variability, as would policies encouraging
regional and crop diversification. Hazell's (1982) analysis of the
covariance problem in India suggests that it could be advantageous to
distribute production among crops and states in a more risk-efficient
way -- for example, through the distribution of public investment in
irrigation or agricultural research -- but such an approach could be
inconsistent with other public objectives.
In general, policies to ameliorate the effect of covariances
rather than to reduce them seem likely to be more effective. Crop
insurance is not effective, nor is consumption credit. But well-
managed food-for-work schemes and flexible, well-targeted food
subsidies can deal with downside variability in rural and urban
areas, respectively. In this context, long-term research on house-
hold data, which could indicate how poor families are buffered
against production variability, merits sustained support by the
CGIAR.

IN SUMMARY

Although variability in cereal yield is largely weather driven,
climatic change is not the likely cause of recent changes in varia-
bility. And although plant breeding and improved agronomy have
probably enhanced upside variability in cereal yields, they may have
decreased downside swings. Thus the major components of the problem
are the stronger covariances across crops and regions. Genetic and
agronomic improvement may have contributed to these, but the main
factors are probably better information about weather, crop, and







10


market prospects, better infrastructure, and a more responsive
agricultural sector.
Policies for diversification may reduce the supply side of the
problem to some degree, but those to ameliorate the social effects of
such variability are likely to be more effective and should be given
priority in further research.











REFERENCES

Barker, R.; Gabler, E. C.; and Winkelmann, D. 1981. "Long Term
Consequences of Technological Change on Crop Yield Stability."
Food Security for Developing Countries. Edited by Alberto
Valdds. Boulder, Colo.: Westview Press.

Church, B. M., and Austin, R. B. 1983. "Variability of Wheat Yields
in England and Wales." Journal of Agricultural Science Cambridge
100:201-204.

Eberhart, S. A., and Russell, W. A. 1966. "Stability Parameters for
Comparing Varieties." Crop Science 6:36-40.

Evans, L. T. 1984. "Physiological Aspects of Varietal Improvement."
In Gene Manipulation in Plant Improvement, pp. 121-146. Edited
by J. P. Gustafson. New York: Plenum.

Finlay, K. W., and Wilkinson, G. N. 1963. "The Analysis of Adapta-
tion in a Plant Breeding Program." Australian Journal of Agricul-
tural Research 14:742-754.

Gales, K. 1983. "Yield Variation of Wheat and Barley in Britain in
Relation to Crop Growth and Soil Conditions A Review." Journal
of Science and Food Agriculture 34:1085-1104.

Hazell, Peter B. R. 1982. Instability of Indian Foodqrain Produc-
tion. Research Report 30. Washington, D.C.: International Food
Policy Research Institute.

S 1985. "Sources of Increased Variability in World Cereal
Production Since the 1960s." Journal of Agricultural Economics
36:145-159.

Laing, D. B., and Fischer, R. A. 1977. "Adaptation of Semi-dwarf
Wheat Cultivars to Rainfed Conditions." Euohvtica 26:129-139.

Luttrell, C. B., and Gilbert, R. A. 1976. "Crop Yields: Random,
Cyclical or Bunchy?" American Journal of Agricultural Economics
58:521-531.

McWilliam, J. R., and Griffing, J. B. 1965. "Temperature-dependent
Heterosis in Maize." Australian Journal of Biological Science
18:569-583.

McWilliam, J. R.; Latter, B. D. H.; and Mathison, M. J. 1969.
"Enhanced Heterosis and Stability in the Growth of an Interspe-
cific Phalaris Hybrid at High Temperature." Australian Journal
of Biological Science 22:493-504.

Mehra, S. 1981. Instability in Indian Agriculture in the Context of
the New Technology. Research Report 25. Washington, D.C.:
International Food Policy Research Institute.







12


Rindos, D. 1984. The Origins of Agriculture: An Evolutionary
Perspective, p. 325. New York: Academic Press.

Stanhill, G. 1976. "Trends and Deviations in the Yield of the
English Wheat Crop During the Last 750 Years." Aqroecosvstems
3:1-10.

Thompson, L. M. 1969. "Weather and Technology in the Production of
Corn in the U.S. Corn Belt." Aqronomics Journal 61:453-456.

1975. "Weather Variability, Climatic Change and Grain
Production." Science 188:535-541.

Yates, F., and Cochran, W. G. 1938. "The Analysis of Groups of
Experiments." Journal of Agricultural Science 28:556-580.








Part I

Overview of Issues
















2


Introduction

Peter B. R. Hazell

There are two levels of concern about possible increases in the
variability of cereal production:
* increased risks for farmers, which may make new technologies less
attractive for adoption;
* increased instability in national food production, which acts to
destabilize domestic prices, national income, and the food consump-
tion of the poor.
The purposes of the workshop are to establish whether there has
been a significant increase in the variability of cereal production
in recent years, and, if so, why this has occurred and what, if any-
thing, should be done about it. This chapter has the more modest
objective of organizing the major issues and questions in, one hopes,
a systematic way. It also has a definitional content, seeking to
clarify and relate some of the differing concepts of instability (or
stability) that biological and social scientists use in thinking
about these issues.

CONCEPTS OF YIELD VARIABILITY

In preparing this workshop, it became apparent early that plant
breeders and economists work with very different concepts of yield
variability, and this can all too easily lead to misunderstandings
and unnecessary disagreements. These differences do not invalidate
the approach of either discipline. Rather, they reflect differences
in the clientele of breeders and economists and differences in the
sources of variability in yield data measured at different levels of
aggregation.
Breeders are primarily concerned with providing farmers with
high-yielding varieties that also offer acceptable levels of risk.
Thus breeders tend to focus on reducing downside yield risks and on
selecting varieties that will perform well for farmers over time.
Their analyses are based on yield data collected in experimental
plots or in farmers' fields.
Economists are more concerned with national food problems, which
can be brought on by both high and low yields. Low national yields
may result in food shortages or high food prices for the poor, where-
as high yields may result in unacceptably low prices for farmers and
excessive government-owned food stocks. Both upside and downside
risks are therefore important to economists, and they work with










regional or national yield data, which embody much more diverse
sources in their variability.
This chapter develops the relationship between sources of varia-
bility and the level of aggregation at which yields are measured. It
needs to be noted here, though, that there can also be important
differences in the type of yield distributions observed at the farm
and national levels. Experimental plot and farm yields are often
skewed (Day 1965), whereas national yields tend to be more symmetric.
This is to be expected, since the latter are a weighted sum of many
individual farm yields, and many of these are only weakly correlated.
In fact, to the extent that many farm yields are independently
distributed, the central limit theorem suggests that national yields
could be approximately normally distributed even if farm yields are
highly skewed.
When yield distributions are symmetric, then measures of
variability that focus only on downside risks give the same results
as comparable measures that focus on upside risks. The variance or
the coefficient of variation of yields are then satisfactory measures
of variability for a wide range of purposes. However, if yield
distributions are skewed, then other measures of variability, such as
the semivariance, or the probability of yield falling below some
specified level, may be more relevant.

VARIABILITY IN WORLD CEREAL PRODUCTION SINCE THE 1960s

Total cereal production for the world (excluding the People's
Republic of China) grew at an average yearly rate of 2.7 percent
between 1960/61 and 1982/83. The average yield during this period
grew by 2.0 percent per year, and the total gross cropped area
allocated to cereals grew by 0.7 percent per year.
This growth in aggregate production has been accompanied by a
widening band of variability around the trend. An encouraging
feature is that each trough in production has been consistently
higher than all previous downturns. However, as growth in consump-
tion has kept pace with the growth in production, a more realistic
indicator of the trend in world food security is the probability with
which aggregate production can fall 5 percent or more below trend.
Hazell (1985) has calculated that this probability increased from 3.5
percent in the period 1960/61 to 1970/71 to 6.8 percent in the period
1971/72 to 1982/83.
Table 2.1 shows the changes in the mean and variability of world
cereal production by crop between 1960/61 to 1970/71 and 1971/72 to
1982/83. The data were obtained from the U.S. Department of Agricul-
ture and have been detrended through regression analysis (see Hazell
1985). The People's Republic of China is excluded from the analysis
because of the poor quality of data available during the 1960s and
because of the upheavals of the "Great Leap Forward" and the "Cul-
tural Revolution." Given the importance of China in world cereal
production, it would not make sense to measure change in production
variability using a base period contaminated in this way. A separate
analysis of China is provided by Stone and Zhong, Workshop Paper 28.
Between the two periods, total world cereal production increased by
37 percent, or by 305 million tons. Increases in wheat and maize






Table 2.1--Changes in the mean and variability of world cereal production, 1960/61 to 1970/71
and 1971/72 to 1982/83



Coefficient of Variation
Average Production of Production F Ratio
Cereal 1960/61 1971/72 1960/61 1971/72 Area
to 1970/71 to 1982/83 Change to 1970/71 to 1982/83 Change Production Sown Yield

(1,000 metric tons) (percent)

Wheat 253,454 352,982 39.3 5.46 4.83 -11.5 1.52 0.34** 1.64

Maize 210,074 317,303 51.0 3.29 4.41 34.0 4.08** 1.65 4.17**

Rice 119,971 155,031 29.2 3.97 3.80 -4.3 1.52 2.45* 0.88

Barley 95,283 150,997 58.5 4.81 7.50 55.9 6.18*** 3.13** 3.28**

Millet 19,705 21,381 8.5 7.78 7.66 -1.5 1.14 2.22 0.69

Sorghum 40,159 53,386 32.9 4.75 5.70 20.0 2.55* 1.08 2.10

Oats 49,033 47,595 -2.9 11.30 5.35 -52.6 0.21*** 0.07*** 4.42**

Other 41,404 35,231 -14.9 4.57 9.33 104.2 2.95** 0.36* 3.61**

Total 829,087 1,133,908 36.8 2.76 3.36 21.7 2.78* 2.22 2.69*


Source: U.S. Department of Agriculture.

Note: Data do not include the People's Republic of China.


*statistically significant at the 10 percent level of confidence.
**statistically significant at the 5 percent level.
***statistically significant at the 1 percent level.











production accounted for one-third each of this total increase, while
rice accounted for 12 percent of the total increase, barley for 18
percent, and sorghum and millet for the rest. The production of oats
and "other" cereals (rye and mixed grains) declined modestly between
the two periods.
The coefficient of variation of total world cereal production
increased from 2.8 percent to 3.4 percent between the two periods, an
increase of 22 percent. At the same time, the variance of total
cereal production increased by 178 percent. The F ratio of 2.78 is
significant at the 10 percent confidence level. Both area and yield
variability also increased, although only the F ratio for yields is
statistically significant at the 10 percent confidence level.
Table 2.1 also shows that, despite sizeable increases in world
wheat and rice production, this growth was not accompanied by a
significant increase in production variability. In fact, the coeffi-
cients of variation declined -- from 5.5 percent to 4.8 percent for
wheat and from 4.0 percent to 3.8 percent for rice. The coefficient
of variation of rice production would have declined even more in the
second period had there not been a significant increase in area
variability.
In contrast, the production variability of coarse grains--
maize, barley, and other cereals (rye and mixed grains) -- increased
significantly. Except for barley, yield variability was the primary
source of this increase. The variability of oat production declined
significantly, but this was because of a sharp decline in area
variability. The variability of oat yields increased sharply.
Table 2.2 shows the changes in the mean and variability of total
cereal production for the 34 most important cereal producing coun-
tries. There is little observable relationship between a country's
performance in increasing cereal production and the changes in
production variability. The correlation across countries between the
percentage change in average production and the change in the
coefficient of variation of production is -0.15. This coefficient is
not significantly different from zero at the 10 percent confidence
level.
Production variability has increased most in France, Brazil,
Mexico, Turkey, Italy, Spain, and South Korea. Increases in yield
variability were particularly large in the United States, France,
Poland, Spain, and South Korea. Area variability increased most in
BraZil, the Philippines, Italy, and Japan. There was a significant
decline in production variability in Nigeria and Egypt, which
originates from significant declines in both yields and area vari-
ability.
Table 2.3 shows the coefficients of variation of production by
crop and country. Production is most variable in Australia and South
Africa; the coefficients of variation for total cereals are about 20
percent, and they are even higher for individual crops. Production
is also relatively unstable in Canada and the U.S.S.R. The least
risky countries are those that predominantly grow rice, presumably
because much of the crop is irrigated. These countries include
Indonesia, Thailand, Bangladesh, and Japan.
Table 2.3 also shows that while coefficients of variation of
wheat and rice production declined for many countries, as well as at
the global level, there are some important exceptions. Wheat produc-










tion became considerably more variable in Mexico, Turkey, Bangladesh,
Poland, Italy, Spain, and Czechoslovakia, and rice production became
considerably more variable in the United States, France, Mexico, and
South Korea. Similarly, there are many countries where the variabil-
ity of coarse grains went down between the two periods, even though
global variability increased.
Tables-2.4 and 2.5 show the coefficients of variation of yields
and of areas sown, by crop and country. The patterns of yield
variability closely parallel the patterns of variability in produc-
tion. Yields are most variable in Australia and South Africa and
least variable for rice growing countries in Asia. The coefficients
of variation of areas sown are typically much smaller than the
coefficients for yield variability -- by a factor of one-half. The
coefficient of variation for the global area sown to cereals was 1.4
percent in the second period. This compares with a coefficient of
variation for the average yield of 3.4 percent. Both coefficients
increased between the two periods, but it does seem that increased
yield variability must have been the most important source of the
increase in production variability.
The importance of increased yield variability is confirmed by a
statistical decomposition analysis of the increase in the variance of
world cereal production between the 1960/61 to 1970/71 period and the
1971/72 to 1982/83 period (Hazell 1985). Of the total variance
increase, 26.4 percent is due to increases in the variances of
individual cereal yields within countries and a further 69.5 percent
is due to increases in yield covariances (Table 2.6).
Within most crops, increased yield variances account for the
lion's share of the contribution to the variance of total cereal
production. For example, when summed over countries the increased
production variances for wheat account for 7.61 percent of the
increase in the variance of total cereal production. Of this,
5.27/7.61 = 69.3 percent is due to increased yield variances.
Similarly, the yield variance shares for other crops are maize 124
percent, rice 36 percent, millet 57 percent, sorghum 77 percent, and
total cereals 77 percent.
Changes in yield covariances are much more important than
changes in yield variances for the variability of world cereal
production. However, part of the increase in the yield covariances
is itself a direct consequence of increased yield variances. Part of
it may also be due to autonomous changes in yield correlations and to
interactions between changes in yield variances and changes in yield
correlations.
Using an additional decomposition analysis, I found that only 6
percent of the 69.5 percent increase in the variance of total cereal
production arising from changes in yield covariances is directly
attributable to changes in yield variances (Hazell 1985). Some 52
percent of the increase is attributable to changes in yield correla-
tions alone, and the remaining 42 percent is due to interaction
effects. Of the correlation increases, the predominant ones are
between the yields of the same or different crops in different
countries. Increases in the intercrop yield covariances within
countries were nearly all attributable to increased yield variances.












Table 2.2--Changes in the mean and variability of total cereal production by
major countries, 1960/61 to 1970/71 and 1971/72 to 1982/83


Coefficient of Variation
Average Production of Production
Country 1960/61 to 1971/72 to 1960/61 to 1971/72 to
1970/71 1982/83 Change 1970/71 1982/83 Change


United States
U.S.S.R
India
Canada
France
Indonesia
Brazil
Argentina
Mexico
Turkey
Australia
Thailand
Germany, F.R.
Bangladesh
Poland
Romania
United Kingdom
Italy
Pakistan
South Africa
Yugoslavia
Burma
Japan
Vietnam
Hungary
Spain
Philippines
Nigeria
Czechoslovakia
Germany, D.R.
Iran
Bulgaria
South Korea
Egypt
Rest of world
(excluding
China)
Total world
(excluding
China)


(1,000 metric tons)
181,982 265,022
138,436 180,952
74,753 104,000
29,991 40,033
27,456 41,085
13,464 20,341
16,500 26,149
17,186 23,764
10,487 15,571
12,932 18,363
12,618 17,445
8,555 13,255
16,030 22,211
10,544 12,861
8,373 13,135
11,602 17,360
12,442 16,754
14,219 16,680
7,668 13,179
7,499 11,999
11,397 15,069
4,933 6,537
14,565 11,393
6,011 7,326
7,342 12,115
9,291 13,676
4,295 7,005
7,793 8,491
6,189 9,688
4,606 7,147
4,955 6,508
5,429 7,706
5,266 6,227
5,789 7,109


(percent)
45.6
30.7
39.1
33.5
49.6
51.1
58.5
38.3
48.5
42.0
38.2
54.9
38.6
22.0
56.9
49.6
34.7
17.3
71.9
60.0
32.2
32.6
-21.8
21.9
65.0
47.2
63.1
9.0
56.5
55.2
31.3
41.9
18.3
22.8


98,481 117,747 19.6


829,087 1,133,908


36.8


6.83
12.16
7.65
17.07
6.01
6.09
5.19
11.80
7.03
7.06
19.54
7.82
9.13
7.21
9.21
10.87
8.73
3.44
10.23
20.37
9.98
9.88
6.01
8.99
10.08
8.09
5.51
11.68
11.73
11.29
8.29
10.27
5.97
4.95


(percent)
6.64
14.26
5.42
10.66
9.19
5.15
8.87
14.04
11.10
9.71
23.15
8.40
5.96
5.03
9.29
9.87
8.34
5.68
3.15
19.69
5.18
7.68
9.31
5.59
6.05
13.86
5.43
5.05
7.54
6.40
9.24
7.47
10.77
2.67


-2.8
17.3
-29.2
-37.6
52.9
-15.4
70.9
19.0
57.9
37.5
18.5
7.4
-34.7
-30.2
1.0
-9.2
-4.5
65.1
-69.2
-3.3
-48.1
-22.3
54.9
-37.8
-40.0
71.3
-1.5
-56.7
-35.7
-43.3
11.4
-27.3
80.4
-46.1


3.19 2.80 -12.2


2.76 3.36 21.7


Source: U.S. Department of Agriculture.

*statistically significant at the 10 percent level.
**statistically significant at the 5 percent level.
***statistically significant at the 1 percent level.















Probability of a 5%
F Ratio Shortfall Below Trend
Area 1960/61 to 1971/72 to
Production Sown Yield 1970/71 1982/83


1.97 1.24 8.23*** 23.3 22.6
2.35* 1.28 1.69 34.1 36.3
0.97 0.65 0.92 25.8 17.9
0.69 0.22*** 0.44* 38.6 31.9
5.26*** 1.58 4.30** 20.3 29.5
1.62 0.74 2.89* 20.6 16.6
7.25*** 4.30** 2.47* 16.9 28.8
2.72* 1.04 2.12 33.7 35.9
5.58*** 3.99** 3.40** 23.9 32.6
3.80** 3.98** 3.45** 23.9 30.2
2.66* 1.65 1.67 39.7 41.3
2.76* 3.00** 2.01 26.1 27.4
0.82 3.24** 0.59 29.1 20.1
0.72 0.20*** 1.05 24.5 16.1
2.52* 0.12*** 4.00** 29.5 29.8
1.83 0.80 2.17 32.3 30.5
1.66 0.33** 1.77 28.4 27.4
3.72** 5.50*** 0.66 7.4 18.9
0.28** 0.44* 0.27** 31.2 5.6
2.40* 2.63* 1.99 40.1 40.1
0.47 0.74 0.57 30.9 16.9
1.06 0.45 1.77 30.5 25.8
1.45 4.27** 1.58 20.3 29.5
0.58 1.26 0.41* 28.8 18.7
0.98 0.35** 1.39 39.9 20.3
6.37*** 0.68 7.73*** 26.2 35.9
2.56* 6.87*** 0.77 18.1 16.1
0.22*** 0.16*** 0.14*** 33.4 25.5
1.01 0.07*** 1.62 33.4 25.5
0.78 1.18 0.65 33.0 21.8
2.15 1.00 3.88** 27.4 29.5
1.05 3.55** 0.72 31.2 25.1
4.62** 0.96 7.76*** 20.1 32.3
0.44* 0.23** 0.37* 15.6 3.1


1.10 0.47 0.75 5.9 3.8


2.78* 2.22 2.69*


3.5 6.8








Table 2.3--Coefficients of variation of production by cereal crop and country, 1960/61 to 1970/71 and
1971/72 to 1982/83


Other Total
Country Wheat Maize Rice Barley Millet Sorghum Oats Cereals Cereals

(percent)


United States
1960/61 to 1970/71
1971/72 to 1982/83


U.S.S.R.
1960/61 to 1970/71
1971/72 to 1982/83


India
1960/61 to 1970/71
1971/72 to 1982/83


Canada
1960/61 to 1970/71
1971/72 to 1982/83


France
1960/61 to 1970/71
1971/72 to 1982/83


Indonesia
1960/61 to 1970/71
1971/72 to 1982/83


Brazil
1960/61 to 1970/71
1971/72 to 1982/83


8.1 9.0 9.3 5.4
8.7 8.3 14.6 12.7


11.8 9.8 16.9 6.8
14.2 12.7 35.4 6.6


15.9 24.0 7.0 15.3 18.7 44.1 10.2 12.2
15.1 18.7 7.1 21.7 35.9 10.8 21.4 14.3


17.2 10.4 8.9 15.7
8.6 8.9 9.4 14.7



26.9 10.3 17.3
13.7 12.0 19.1


14.9 10.0
13.7 11.4


- 7.6
- 5.4


14.8 27.6 17.1
11.1 10.5 10.7


12.9 22.4 11.3 9.7 16.4 7.2 16.1 6.0
9.3 21.6 28.6 8.4 -9.6 10.8 13.6 9.2



S 21.7 5.1 6.1
14.7 4.4 5.1


58.8 5.3 10.4 16.5
34.0 10.9 8.0 44.3


- 7.2
28.6 17.2


-5.2
-8.9





Argentina
1960/61 to 1970/71
1971/72 to 1982/83


Mexico
1960/61 to 1970/71
1971/72 to 1982/83


Turkey
1960/61 to 1970/71
1971/72 to 1982/83


Australia
1960/61 to 1970/71
1971/72 to 1982/83


Thailand
1960/61 to 1970/71
1971/72 to 1982/83


Germany, F.R.
1960/61 to 1970/71
1971/72 to 1982/83


Bangladesh
1960/61 to 1970/71
1971/72 to 1982/83


Poland
1960/61 to 1970/71
1971/72 to 1982/83


Romania
1960/61 to 1970/71
1971/72 to 1982/83


31.1 13.8 20.8 38.9 23.3 37.2 24.0 41.1 11.8
19.4 22.8 10.5 28.4 27.1 24.4 21.1 44.6 14.0


9.9 10.4 10.0 8.9 24.1 29.5
18.9 14.1 19.7 29.4 18.3 33.4


7.2 9.4 14.5 9.8 6.2
11.4 6.0 11.5 10.5 13.1


7.0
11.1


5.8 11.5
4.0 4.8


25.0 13.3 8.4 30.1 22.7 57.8 25.0 25.6 19.5
26.7 22.3 9.6 27.5 28.4 24.1 30.3 43.6 23.2



5.2 8.9 7.8
S 19.3 6.9 26.2 8.4



10.3 39.1 9.7 9.1 10.1 9.1
5.5 19.3 5.0 -12.6 7.1 6.0


17.1 6.7 24.6 14.0
29.7 5.3 7.9 9.8



7.0 25.6 12.8 13.6
11.8 90.0 16.3 39.8


7.2
5.0



10.1 7.3 9.2
10.0 9.3


19.7 10.5 21.4 13.1 31.5 26.9 10.9
12.1 12.7 19.5 15.7 22.0 6.7 9.9
(continued)








Table 2.3--(continued)


Country



United Kingdom
1960/61 to 1970/71
1971/72 to 1982/83


Italy
1960/61 to 1970/71
1971/72 to 1982/83


Pakistan
1960/61 to 1970/71
1971/72 to 1982/83


South Africa
1960/61 to 1970/71
1971/72 to 1982/83


Yugoslavia
1960/61 to 1970/71
1971/72 to 1982/83


Burma
1960/61 to 1970/71
1971/72 to 1982/83


Japan
1960/61 to 1970/71
1971/72 to 1982/83


Other Total
Wheat Maize Rice Barley Millet Sorghum Oats Cereals Cereals


(percent)


14.0
7.3



15.2 9.0
12.9 10.6



13.1 10.7
5.6 15.4



S 29.2
34.6



14.3 18.8
13.2 15.5



9.9
8.4



5.2 19.5
8.9 22.2


~ t




11.8
10.3






- a
- t






16.9
8.8



19.4


13.3
10.6



12.0
23.8 9.2



6.5
10.1



46.0 23.2
34.4 15.7



7.7 11.3
22.4 9.7








10.0
10.1


40.0 8.7
24.4 8.3



10.8 3.4
12.3 5.7



10.2
3.2



20.4
19.7



8.7 10.0
7.8 5.2



9.9
7.7



6.0
9.3





Vietnam
1960/61 to 1970/71
1971/72 to 1982/83


Hungary
1960/61 to 1970/71
1971/72 to 1982/83


Spain
1960/61 to 1970/71
1971/72 to 1982/83


Philippines
1960/61 to 1970/71
1971/72 to 1982/83


Nigeria
1960/61 to 1970/71
1971/72 to 1982/83


Czechoslovakia
1960/61 to 1970/71
1971/72 to 1982/83


Germany, D.R.
1960/61 to 1970/71
1971/72 to 1982/83


Iran
1960/61 to 1970/71
1971/72 to 1982/83


Bulgaria
1960/61 to 1970/71
1971/72 to 1982/83


8.9
30.8 6.2


9.0
5.6


14.8 10.4 14.7 26.4 15.3 10.1
13.8 10.0 10.6 25.3 13.7 6.1



3.0 11.3 5.3 23.0 46.2 13.5 10.0 8.1
18.2 10.6 3.0 18.7 19.2 16.2 11.4 13.9



5.5 6.1 5.5
7.7 5.0 5.4


11.0 8.4
6.2 10.1


9.7 12.4
14.0 17.3


15.5 12.3
7.4 6.0


14.5
9.5


11.7
5.0


15.8 10.6 11.7
15.6 11.4 7.5


15.7 14.6 13.2 20.3 11.3
5.9 9.8 18.9 37.1 6.4



10.5 13.5 7.8 8.7 8.3
9.3 18.3 8.5 13.8 9.2


14.5 14.9 17.2 13.6
7.6 17.4 14.7 11.5


26.9 16.7 10.3 n
19.9 17.3 7.5
(continued)












Table 2.3--(continued)


Other Total
Country Wheat Maize Rice Barley Millet Sorghum Oats Cereals Cereals

(percent)


South Korea
1960/61 to 1970/71
1971/72 to 1982/83


Egypt
1960/61 to 1970/71
1971/72 to 1982/83


Rest of world
(excluding China)
1960/61 to 1970/71
1971/72 to 1982/83


Total World
(excluding
1960/61 to
1971/72 to


29.2 32.1 8.9 14.2
32.6 21.3 13.0 20.0



12.3 7.1 16.0 11.6
6.3 4.3 6.4 14.0


18.6 6.0
21.1 10.8


4.2
7.6


4.9
2.7


7.9 4.0 3.0 7.1 4.4 9.6 5.7 6.9 3.2
7.9 3.6 4.5 5.1 7.6 5.7 7.5 7.1 2.8




5.5 3.3 4.0 4.8 7.8 4.7 11.3 4.6 2.8
4.8 4.4 3.8 7.5 7.7 5.7 5.3 9.3 3.4


China)
1970/71
1982/83


Source: U.S. Department of Agriculture.





Table 2.4--Coefficients of variation of
1971/72 to 1982/83


yields by cereal crop and country, 1960/61 to 1970/71 and


Other Total
Country Wheat Maize Rice Barley Millet Sorghum Oats Cereals Cereals

(percent)


United States
1960/61 to 1970/71
1971/72 to 1982/83


U.S.S.R.
1960/61 to 1970/71
1971/72 to 1982/83


India
1960/61 to 1970/71
1971/72 to 1982/83


Canada
1960/61 to 1970/71
1971/72 to 1982/83


France
1960/61 to 1970/71
1971/72 to 1982/83


Indonesia
1960/61 to 1970/71
1971/72 to 1982/83


Brazil
1960/61 to 1970/71
1971/72 to 1982/83


4.8 6.6 3.1 5.0 6.5 5.5 6.0 4.0
6.7 9.6 6.4 8.6 12.0 6.9 10.4 8.8



16.2 10.9 4.1 14.7 18.9 14.2 11.0 13.2
13.7 11.3 3.8 17.7 34.1 11.9 15.7 13.4


10.6 8.4 8.0 8.9 13.7 8.3
5.9 7.8 7.9 9.0 10.8 10.4


17.8 7.8
9.7 7.7


12.1
7.2


- 6.2
- 4.6


-9.1 21.1 14.1
6.4 5.6 7.3


7.9 19.3 11.7 8.6 13.1 5.0 6.9 5.6
8.6 10.5 19.1 8.5 10.7 10.8 12.1 8.2



S 5.0 2.5 2.1
4.2 2.8 2.5


16.8 4.7 6.9 13.4
25.2 8.4 5.2 19.0


-7.1
10.8 8.6


-4.6
-6.7
(continued)








Table 2.4--(continued)


Other Total
Country Wheat Maize Rice Barley Millet Sorghum Oats Cereals Cereals

(percent)


Argentina
1960/61 to 1970/71
1971/72 to 1982/83


Mexico
1960/61 to 1970/71
1971/72 to 1982/83


Turkey
1960/61 to 1970/71
1971/72 to 1982/83


Australia
1960/61 to 1970/71
1971/72 to 1982/83


Thailand
1960/61 to 1970/71
1971/72 to 1982/83


Germany, F.R.
1960/61 to 1970/71
1971/72 to 1982/83


Bangladesh
1960/61 to 1970/71
1971/72 to 1982/83


18.5 9.0 5.0 16.5
8.7 14.9 7.4 12.8



7.4 8.2 6.6 5.8
10.5 11.3 7.5 19.7


7.4 8.6 9.7 9.0 5.9
0.3 6.7 6.5 9.8 12.7


9.4 14.8 8.5 16.8 8.6
10.9 12.4 8.6 14.1 9.4


9.7 14.2
15.5 21.2


-5.5
-8.1


-5.2 9.3 6.9
-3.8 3.8 9.1


15.8 9.9 10.2 15.6 17.0 19.2 18.7 15.4 15.2
21.6 8.7 10.9 19.8 16.1 12.4 13.2 16.5 18.5



11.0 7.2 6.4
17.4 6.7 37.3 8.1


9.1 7.7
5.7 8.8


10.6
4.2


- 6.5 9.2 8.8
- 9.0 5.2 5.2


15.2 4.2 9.4 6.0 4.3
8.6 4.0 5.8 7.1 3.9





Poland
1960/61 to 1970/71
1971/72 to 1982/83


Romania
1960/61 to 1970/71
1971/72 to 1982/83


United Kingdom
1960/61 to 1970/71
1971/72 to 1982/83


Italy
1960/61 to 1970/71
1971/72 to 1982/83


Pakistan
1960/61 to 1970/71
1971/72 to 1982/83


South Africa
1960/61 to 1970/71
1971/72 to 1982/83


Yugoslavia
1960/61 to 1970/71
1971/72 to 1982/83


Burma
1960/61 to 1970/71
1971/72 to 1982/83


Japan
1960/61 to 1970/71
1971/72 to 1982/83


5.1 20.6 8.5 6.4 -9.4 3.4 6.2
9.1 18.6 7.4 14.6 -8.2 14.8 9.2


16.9 9.1 15.6 10.0
10.5 9.8 17.7 10.3


13.6 11.9
14.6 3.9


7.8 5.9 -3.4 3.2 6.1
7.5 6.1 -7.5 9.1 6.6



6.5 8.5 9.8 6.7 10.8 4.1 3.9
5.1 2.8 12.7 7.1 9.6 8.6 4.7 2.3



9.8 8.2 11.9 7.6 5.9 7.0 8.6
4.1 3.1 3.9 3.6 4.8 5.4 3.0


17.6 24.1 24.8
11.9 21.9 22.8



11.8 11.1 10.7 10.7
9.1 5.8 11.4 9.2


22.9 24.7 19.8
26.9 16.1 18.2


8.1
10.5


7.9 5.8 9.4
6.9 7.1 4.7


24.9 32.5 6.6 34.9 -- 6.4
8.3 10.0 7.5 47.5 6.6



19.6 6.3 4.0 17.7 9.1 8.2 5.4
10.5 5.7 5.7 8.4 6.8 5.8
(continued)








Table 2.4--(continued)


Other Total
Country Wheat Maize Rice Barley Millet Sorghum Oats Cereals Cereals

(percent)


Vietnam
1960/61 to 1970/71
1971/72 to 1982/83


Hungary
1960/61 to 1970/71
1971/72 to 1982/83


Spain
1960/61 to 1970/71
1971/72 to 1982/83


Philippines
1960/61 to 1970/71
1971/72 to 1982/83


Nigeria
1960/61 to 1970/71
1971/72 to 1982/83


Czechoslovakia
1960/61 to 1970/71
1971/72 to 1982/83


Germany, D.R.
1960/61 to 1970/71
1971/72 to 1982/83


8.0
6.5 5.6


12.5 8.4
10.6 6.7


- 8.2
- 5.0


11.0
11.0


17.1 7.6 7.2
19.1 16.7 4.8


10.6 18.9 4.3 10.2 22.2 11.6 8.6 6.9
15.0 6.2 3.8 16.8 11.4 15.5 11.0 13.2



4.3 7.8 5.9
3.5 4.8 4.0



14.8 8.5 17.5 8.7 9.3
S 3.5 4.0 5.9 4.3 3.8


7.7 14.6
9.1 19.2


12.1
7.6


- 9.6 6.3 8.4
- 12.8 9.4 7.3


9.5 15.0 8.6 9.4 10.5
6.0 7.8 13.6 13.4 6.5







Iran
1960/61 to 1970/71
1971/72 to 1982/83


Bulgaria
1960/61 to 1970/71
1971/72 to 1982/83


South Korea
1960/61 to 1970/71
1971/72 to 1982/83


Egypt
1960/61 to 1970/71
1971/72 to 1982/83


Rest of world
(excluding China)
1960/61 to 1970/71
1971/72 to 1982/83


Total world
(excluding China)
1960/61 to 1970/71
1971/72 to 1982/83


5.5 9.0 9.9 4.4 3.6
5.4 14.6 9.2 13.1 6.1


15.2 13.5 15.1 12.1
6.4 11.1 14.8 7.5


24.1 11.3 10.3
10.7 10.5 5.7


8.0 12.1 8.0 14.2 12.1 -5.6
15.6 17.5 12.6 15.8 13.2 11.0


10.2 14.0 5.8 20.6
4.3 3.5 3.8 5.3


1.8 4.5
4.6 -2.3


6.3 4.2 3.2 4.5 4.4 3.4 4.7 5.1 3.2
4.8 3.4 1.0 6.8 6.2 3.2 5.5 6.1 2.5




5.0 3.0 3.3 4.3 7.3 4.0 3.0 5.5 2.6
4.9 4.6 2.6 6.4 5.8 4.6 5.6 8.0 3.4








Table 2.5--Coefficients of variation of area sown
1971/72 to 1982/83


by cereal crop and country, 1960/61 to 1970/71 and
rN


Other Total
Country Wheat Maize Rice Barley Millet Sorghum Oats Cereals Cereals

(percent)


United States
1960/61 to 1970/71
1971/72 to 1982/83


U.S.S.R.
1960/61 to 1970/71
1971/72 to 1982/83


India
1960/61 to 1970/71
1971/72 to 1982/83


Canada
1960/61 to 1970/71
1971/72 to 1982/83


France
1960/61 to 1970/71
1971/72 to 1982/83


Indonesia
1960/61 to 1970/71
1971/72 to 1982/83


Brazil
1971/72 to 1982/83
1971/72 to 1982/83


9.8 8.1 8.8 6.4
10.6 4.2 11.5 10.6


11.0 7.4 14.1 5.3
7.0 8.0 25.8 5.3


4.4 23.7 4.9 12.8 9.0 -34.4 9.9 2.7
3.7 14.9 5.7 7.5 -5.3 12.7 3.0



8.7 4.5 1.5 8.6 2.7 4.2 1.8
4.0 2.3 1.9 10.7 4.1 2.7 1.4



20.5 6.7 12.7 7.4 16.1 10.8
6.1 7.3 15.5 7.9 7.1 4.9



6.4 11.6 3.5 4.5 21.1 4.4 13.3 1.1
3.4 18.6 20.6 3.2 14.6 3.5 5.8 1.3



17.4 3.7 5.8
13.7 2.3 4.6


57.4 1.9 6.0 19.9 8.4
28.3 4.7 7.5 32.6 26.0 14.9


3.0
4.3





Argentina
1960/61 to 1970/71
1971/72 to 1982/83


Mexico
1960/61 to 1970/71
1971/72 to 1982/83


Turkey
1960/61 to 1970/71
1971/72 to 1982/83


Australia
1960/61 to 1970/71
1971/72 to 1982/83


Thailand
1960/61 to 1970/71
1971/72 to 1982/83


Germany, F.R.
1960/61 to 1970/71
1971/72 to 1982/83


Bangladesh
1960/61 to 1970/71
1971/72 to 1982/83


Poland
1960/61 to 1970/71
1971/72 to 1982/83


Romania
1960/61 to 1970/71
1971/72 to 1982/83


19.7 7.0 17.0 28.3 18.9 31.4 19.3 30.2 9.6
15.0 15.4 7.3 16.4 17.6 16.9 16.9 33.3 9.5



5.1 4.6 6.6 6.7 22.8 24.5 2.7
10.6 3.9 15.5 15.4 14.3 38.1 4.6



0.7 2.1 22.9 1.3 4.4 4.5 4.7 0.7
1.4 1.5 12.4 4.0 11.8 1.9 3.4 1.3



17.3 8.3 7.1 23.0 27.5 36.7 8.9 16.4 9.9
10.8 22.7 10.1 19.0 32.7 24.6 22.6 35.0 9.9


9.9 4.2 -
6.5 5.5 16.3


3.7 35.3 2.8
2.0 14.7 2.5


3.9
4.8


3.8 2.2 1.1
5.8 4.2 1.9


13.8 3.7 21.1 10.8 4.2
29.1 1.9 9.2 6.9 1.7



4.6 16.0 9.9 11.7 6.5 7.8 6.3
5.1 83.2 12.8 32.8 6.0 4.6 1.9


6.1 3.4 26.1 8.0
2.5 4.4 11.4 10.8


25.8 20.3 3.2
20.6 6.2 3.1
(continued)







Table 2.5--(continued)


Other Total
Country Wheat Maize Rice Barley Millet Sorghum Oats Cereals Cereals

(percent)


United Kingdom
1960/61 to 1970/71
1971/72 to 1982/83


Italy
1960/61 to 1970/71
1971/72 to 1982/83


Pakistan
1960/61 to 1970/71
1971/72 to 1982/83


South Africa
1960/61 to 1970/71
1971/72 to 1982/83


Yugoslavia
1960/61 to 1970/71
1971/72 to 1982/83


Burma
1960/61 to 1970/71
1971/72 to 1982/83


Japan
1960/61 to 1970/71
1971/72 to 1982/83


- 11.3
2.8


13.1 41.7 5.0
6.4 22.7 2.6


2.5 3.3 11.0 3.4 5.9 10.8
7.2 2.8 3.1 6.1 19.4 2.6 12.5


5.7 5.7 3.5 6.1 9.7 8.7
2.6 4.3 6.4 15.8 11.6 8.3


14.6 3.8
7.7 5.7


- 31.5 9.9
19.1 10.7


- 3.6
- 2.1


-2.7
4.2


6.8 2.2 18.2 9.1 10.1 4.0 4.6 1.5
5.2 2.8 5.2 8.7 18.6 4.4 5.3 1.5



47.7 37.5 4.8 22.4 5.3
20.7 4.0 3.7 32.5 3.5



18.6 8.5 2.6 6.2 14.9 9.6 2.5
12.2 12.4 6.2 16.2 11.8 7.7






Vietnam
1960/61 to 1970/71
1971/72 to 1982/83


Hungary
1960/61 to 1970/71
1971/72 to 1982/83


Spain
1960/61 to 1970/71
1971/72 to 1982/83


Philippines
1960/61 to 1970/71
1971/72 to 1982/83


Nigeria
1960/61 to 1970/71
1971/72 to 1982/83


Czechoslovakia
1960/61 to 1970/71
1971/72 to 1982/83


Germany, D.R.
1960/61 to 1970/71
1971/72 to 1982/83


Iran
1960/61 to 1970/71
1971/72 to 1982/83


Bulgaria
1960/61 to 1970/71
1971/72 to 1982/83


3.4 3.7
33.3 3.0 3.5


6.0 6.1 12.4
4.7 6.0 6.4



4.2 8.3 3.4 20.1
8.7 5.4 2.5 5.0



3.5 2.7
7.3 3.0


14.4 10.5
21.6 6.7


38.9 3.4 5.1 2.0
19.2 3.8 2.8 1.6



S- 2.2
-4.7


15.5 8.3 5.6 8.0 4.8
S 4.0 11.2 2.3 2.2 1.7



4.3 13.5 3.6 8.7 9.5 4.0
6.2 10.9 7.1 -13.9 5.7 1.0


8.4 10.3
3.8 9.4



5.9 6.7 8.0 6.1
6.0 11.9 3.5 2.2



3.2 6.0 7.3 4.1
2.8 10.6 7.7 10.9


7.5 18.0 4.1
12.2 31.5 3.7



5.5
4.8



6.4 8.1 2.0
13.7 19.3 4.1
(continued)









Table 2.5--(continued) oc



Other Total
Country Wheat Maize Rice Barley Millet Sorghum Oats Cereals Cereals

(percent)


South Korea
1960/61 to 1970/71
1971/72 to 1982/83


Egypt
1960/61 to 1970/71
1971/72 to 1982/83


Rest of world
(excluding China)
1960/61 to 1970/71
1971/72 to 1982/83


Total world
(excluding China)
1960/61 to 1970/71
1971/72 to 1982/83


26.2 28.6 2.9 3.6 15.5
25.0 8.2 0.9 11.9 20.8


6.2 8.5 12.9 21.9
4.8 5.1 3.4 13.8


- 3.4
- 4.0


3.4
5.0


- 5.9
- 2.7


3.5 6.5 2.2 4.1 2.5 8.9 2.2 3.3 2.1
4.5 2.7 4.3 5.3 3.4 3.5 4.8 3.5 1.4




3.4 1.4 1.3 3.5 1.5 2.2 10.1 5.4 1.0
1.9 1.6 1.8 4.7 2.1 2.2 3.0 5.1 1.4







Table 2.6--Components of change in the variance of world cereal production, 1960/61 to 1970/71
and 1971/72 to 1982/83



Source of Change

Change Change
Change Change in Yield in Area Change in Change in
Variance in Mean in Mean Variances & Variances & Area-Yield Interaction Change in
Component Yields Areas Covariances Covariances Covariances Terms Residual Row Sums

(percent)
Crop variances
Wheat 2.06 -2.38 5.27 -0.57 3.57 -0.49 0.15 7.61
Maize 6.67 1.94 17.16 -6.15 -5.01 -1.54 0.73 13.80
Rice 0.11 0.25 0.45 0.12 0.16 0.13 0.05 1.26
Barley 0.43 2.30 1.87 0.86 1.37 4.67 0.96 12.46
Millet 0.01 -0.01 0.04 0.01 0.06 -0.02 0.00 0.07
Sorghum 0.19 0.07 0.57 -0.23 0.12 0.07 -0.05 0.74
Oats 0.83 0.27 0.11 -1.25 -0.54 -1.06 -0.19 -1.85
Other 0.14 -0.15 0.94 -0.14 0.29 -0.77 0.06 0.36

Sum of crop variances
within countries 10.44 2.28 26.40 -7.36 0.01 0.99 1.70 34.45

Intercrop covariances
within countries 0.97 4.48 36.68 -0.94 -9.38 1.89 1.65 35.35

Intercountry covariances
within crops 0.09 1.61 11.49 -3.61 -4.40 -0.98 0.49 4.70

Covariances between
different crops in
different countries 2.75 0.85 21.36 19.13 -28.51 6.43 3.55 25.50

Sum of variances
and covariances 14.24 9.22 95.93 7.22 -42.28 8.33 7.40 100.00


Note: Data do not include People's Republic of China.










CHANGE IN YIELD VARIABILITY AND YIELD CORRELATION

The previous analysis shows that world cereal production has
become less stable since the 1960s, primarily because of increases in
yield variances and yield correlations.
It is also clear that there is no general uniformity to the
patterns of change. The coefficients of variation of wheat and rice
yields declined at the global level (Table 2.4), but the coefficients
of variation increased for coarse grain yields. There are also
important differences among countries, many of which defy the global
patterns of change in the variability of individual cereal yields.
While these contrasting changes complicate the task of the workshop,
they also may be invaluable in identifying the causal factors at
work. We have as much to learn from cases where yield variability
went down as we do from cases where it went up.
I turn now to formulating hypotheses about the possible causes
of change in yield variability and yield correlations. This may be
viewed as an attempt to provide a checklist for the workshop discus-
sions. The presentation is also organized to show the relationship
between the sources of change in yield variability and yield corre-
lations and the level of aggregation at which yields are measured.

Sources of Change: Experimental Plot Data

The least number of possible sources of change in yield vari-
ability, both within and between plots, arise in experimental plot
data, because the experimenter has considerable control over the
inputs (treatments) used and the period and location (environment) in
which the experiment is conducted. th
If yi denotes the yield of the i genotype, then a useful yield
model is as follows:
yi = fi (X, E, u), (2.1)
where
X = a vector of controlled inputs, for example, fertilizers;
E = a vector of environmental variables like weather, altitude,
and soil type (some of these variables are stochastic, in
which case the experimenter can control only for the aver-
age values of such variables);
u = a stochastic residual over which the experimenter has no
control; and
fi = any relevant functional form.
Variability in yi is then due to variability in E and u, but it
will also be conditioned by the choice of genotype (i), the input
levels (X), and the mean values of E. This conditioning is espe-
cially important if, as is usually the case, the genotype interacts
with X and E.
Common breeding techniques involve measuring the yields of
selected genotypes at different locations (E) with varying levels of
treatments (X) at each location. Composite measures of stability are
then calculated across X and E, and these provide a basis for
comparing the stability of different genotypes (see, for example,
Eberhart and Russell 1966; Finlay and Wilkinson 1963).
When controlled in this way, the only systematic source of
difference in the variability of different genotypes is their









inherent biological stability. Much of the available evidence
supports the argument that breeders have been successful in recent
years in reducing plot yield variability across locations. However,
we should probably expect their success to have been greatest where
the relevant range of environments (E) is narrow, for example, for
irrigated paddy rather than upland rice.
Two consequences of this approach to breeding must be mentioned.
First, because most stability tests on specific genotypes are carried
out for only one or two years, there is a strong presumption that
stability across different locations is a good proxy for stability
over time at specific locations. The evidence for this is not very
encouraging (Watson and Anderson 1977; Evenson et al. 1978).
Second, by screening for genotypes that perform well in many
locations at the same time, breeders may inadvertently be increasing
yield correlations between locations, hence between farms or regions.
This need not be a problem for farmers, but it may add to the
variability of national yields.


Sources of Change: Farm Field Data

Once the experimenter is replaced by a business oriented
decision maker, there are additional sources of yield variability
beyond those operating at the experimental plot. This is especially
true of time-series yield data.
The application of inputs (X) is no longer controlled, since the
farmer will adjust the levels used each year in response to price
changes and the availability of supplies. This behavioral component
to farm yields could lead to changes in yield variability over time
if price variability changes or if input supplies become more stable
or more erratic.
The variability of farm-gate prices has increased significantly
in many countries since the early 1970s, and this may be an important
source of the increased yield variability. The effect may have been
amplified by the coterminous and widespread adoption of high-yielding
varieties, which in developing countries also increased farmers'
dependence on modern inputs. There is also some evidence that input
supplies have become more erratic in some developing countries. For
example, electricity supplies for irrigation pumps in India became
more erratic at the same time that farmers became more dependent on
pumping to achieve higher yields with HYVs.
Farmers also change and improve their cultural practices over
time, often in conjunction with the adoption of improved genotypes,
and this may lead to changes in yield variability. For example,
increased planting densities and greater monocropping may lead to
significant increases in yield variability. On the other hand,
improvements in weed and pest control practices or increases in
irrigation may reduce yield variability.
Time series yield data from farmers' fields may also be affected
by changes in weather patterns, and particularly if yield variability
within relatively short time periods is compared. Changes in
genotypes and cultural practices may also interact with changes in
weather, with quite complex consequences for changing patterns of
yield variability.










Sources of Change: National Data
At the national level of aggregation, changes in yield covari-
ances between farms (and between regions) also become important
sources of change in yield variability. In fact they are often the
dominant source of change.
Suppose n farmers grow the same crop, then a simple measure' of
their aggregate yield is
n
Y = 1/n E y,. (2.2)
j=1

The variance of aggregate yield over time is then

V(Y) = 1/n2 [EjV(yj) + E E Cov (yi,yj)]. (2.3)
ij

That is, V(Y) is the sum of individual farm yield variances, E V(yj),

plus the sum of all the yield covariances between farms,

i4j j Cov (yiYj).

Since there are n farms, then there are n yield variances and
(n2 n) yield covariances in equation (2.3). For example, if n =
1,000 then there are 1,000 variances and 999 thousand covariances.
It is not therefore surprising that changes in interfarm yield
covariances are usually the dominant source of change in V(Y).
By definition, a yield covariance can be written as

Cov (yi,Yj) = Pij [V(yi) V(yj)]1, (2.4)

where p. is the correlation coefficient. Part of the change in
yield cPdariances over time is therefore directly attributable to
changes in the variances of farm yields. But part may also be due to
autonomous shifts in the correlations.
Hazell (1984) provides evidence of sharp increases in interstate
correlations for U.S. maize yields since the mid-1960s. Walker
(Workshop Paper 30) shows even stronger trends in interdistrict yield
correlations for sorghum and pearl millet in India.
Interregional (or interfarm) yield correlations may have
increased over time for a number of reasons:
Weather patterns may have become more covariate across regions.
More erratic supplies of farm inputs affect many farmers simulta-
neously, and this may lead to more synchronized patterns of vari-
ability in yields. A good example is the increased irregularity of
electricity supplies for irrigation pumping in India. Large
regions are affected at the same time, and this may lead to a
common decline in farm yields, particularly in drought years.

1This measure does not weight for differences in farm production.










* An increase in price variability at the same time that new geno-
types are adopted has increased farmers' dependence on modern
inputs. A common response to the same price signals will lead
farmers to adjust their use of fertilizers and other yield affect-
ing inputs in the same direction.
* A narrow range of genotypes, which have a common susceptibility to
the same kinds of pest and weather stresses, have been widely
adopted. Dalrymple (1976) and Hargrove, Coffman, and Cabanilla
(1979) provide evidence of the narrowing of the genetic base for
wheat and rice.
* Genotypes that have been screened for stability at a wide range of
locations have been widely adopted.
* There has been a move toward more uniform cultural practices, which
increases the exposure of larger areas of a crop to the same risks
-- for example, practices that lead to a narrower range of planting
dates.
* Irrigated area has been increased. Although irrigation may be
effective in reducing yield variability within fields, it may, by
reducing some climatic influences on yields, lead to more synchro-
nized patterns of variability across locations.
Expansion of cereal production into more marginal lands has been
an important source of increase in the variability of national yields
in some countries, for example, Australia and Brazil. Changes in the
size distribution of farms, such as those incurred through land
reforms or other structural and institutional changes, can also be
important.


CONSEQUENCES OF INCREASED INSTABILITY IN CEREAL YIELDS

Consequences for Farmers

Increased yield risks associated with improved varieties or new
technologies may hinder their widespread adoption by farmers, thereby
limiting growth in national food supplies. There is plenty of
empirical evidence to show that most farmers, and particularly small-
scale farmers in developing countries, act in risk-averse ways when
making resource allocation decisions that affect their income (see,
for example, Binswanger 1980; Dillon and Scandizzo 1978). However,
studies of the relationship between yield risks and the adoption of
specific varieties or technologies show mixed results. For example,
Roumasset (1976, 1979), O'Mara (1971), and Gladwin (1977) found that
risk aversion was not a significant impediment to the adoption of the
improved technologies they studied. Walker (1981) in his study of
the adoption of maize hybrids in El Salvador also found that the risk
attitudes of adopters and nonadopters were about the same. In
contrast, Binswanger (1980), Moscardi (1976), Moscardi and de Janvry
(1977), and Scandizzo (1974) found that risk can be an impediment to
adoption.
These differences may be due to the different technologies and
farming systems studied (for example, irrigated versus rainfed
agriculture). There are still too few studies using comparable
methodologies to permit useful cross-study analyses. But the
conflicting results may also reflect the complexity of the relation-










ship between yield risks and the variability of farm or family
income. Since it is presumably the stability of income (or family
consumption) that concerns farmers most, increased yield risks should
be a problem only if they lead to greater instability in income.
Yield risks are only one of many risks that affect a farmer's
income, and some of these risks may act to offset each other. Within
a crop, higher yield risks may be partly offset by negatively
correlated fluctuations in prices, and the return from the crop may
be much more stable than the variability of prices and yields alone
would suggest. When more than one crop is grown, there is also scope
for low or even negative correlations among the returns of the
different crops, with a resultant stabilizing effect on aggregate
income. Work at ICRISAT (for example, Walker and Jodha 1986)f has
shown that even small-scale farmers in dryland India can be surpris-
ingly efficient in reducing income risks through a variety of
cultural practices (intercropping, spatial diversification, staggered
planting dates, and so on), through off-farm employment, through the
use of credit, and by participating in land leasing arrangements,
which effectively share some of the yield risks with landlords.
Within this rather complex framework, few generalities about the
relationship between yield risks for individual cereals and the
stability of family income seem likely to emerge.

Consequences for Poor Consumers and the National Economy

Yield variability is important at the more macrolevel if it
translates into instability in the supplies of important food or
export crops. We have already seen that much of the increase in
production variability since the 1960s is attributable to increases
in yield variances and covariances. This is true for many individual
countries as well as at a global level. Of course, this link between
yield and production variability will not always be true, and much
depends on how the variability of the sown area behaves.
In principle, one would expect high production years for major
cereals to be good for poor consumers. They should gain from more
plentiful food supplies, from lower prices, and perhaps from in-
creased agricultural employment. The opposite might be expected in
low production years. But as in the case of farm incomes, things can
be more complex than this. Since consumers typically purchase a
number of different food crops, shortages or high prices for one may
simply be offset by substituting other foods whose supplies are more
plentiful or whose prices are lower. There is a surprising lack of
evidence on the relationship between the variability of individual
food supplies and the instability of the incomes and nutritional
intake of the poor. Sahn and von Braun (Chapter 6) have mustered
most of this evidence in their background paper for the workshop.
Production variability can lead to increased price variability
in domestic markets and hence indirectly increase the price risks
confronting farmers. Whether this will worsen their income insta-

2This previously published paper was circulated to workshop
participants.










ability depends on how individual farm yields are correlated with
market prices. If this relationship is negative, then price and
yield risks will tend to offset each other, and income will remain
relatively stable. The opposite will happen for farmers whose yields
are positively correlated with market prices. At the national level
there is likely to be a negative relationship between aggregate yield
and market prices, simply because domestic demand is downward
sloping. But this relationship may be weakened, or even reversed, if
the market is dominated by import or export prices.
In poor agrarian countries, variability in the yields of
important food or export crops can have serious destabilizing
consequences for national income, employment, and the balance of
payments. Even in as rich a country as the United States, it is
quite surprising how much of the nonfarm economy and the banking
system is adversely affected by fluctuations in cereal prices and
yields.

CONTAINING INCREASING YIELD VARIABILITY

Assuming that increasing yield variability is a problem, either
because of increasing variability at the farm level or because of
increasing interfarm (and interregional) yield correlations, what can
be done about it?
There are basically two approaches. One is to directly attack
the cause of increasing yield variability. The other is to accept
increasing yield variability as a necessary consequence of improve-
ments in average production and to attempt to parry, or offset, its
effects through appropriate interventions. Obviously, the relative
costs and benefits of the two approaches need to be considered,
particularly if a direct attack on the problem involves substantial
trade-offs with growth in aggregate production.

Direct Approaches

Any direct approach must obviously be tied to proper identifi-
cation of the causes of increased yield variability. If the problem
lies with genotypes that are inherently too risky, then appropriate
changes in plant breeding priorities may be in order. However, if
the problem is due primarily to economic factors, such as increased
price variability, more erratic input supplies, crop expansion into
more marginal areas, or land reforms and other institutional and
policy changes, then these factors need to be attacked in the policy
arena.
Most likely, improved technologies have aggravated the insta-
bility induced by changes in the economic environment. They may have
directly increased interfarm and interregional yield correlations,
because of the way improved genotypes are selected and because they
increase yield response to (covariate) economic factors. They may
also have indirectly increased yield variability in farmers' fields
by permitting a greater range of yield response to input use.
Perhaps breeders can tackle some of these problems directly, by
giving more thought to yield correlations between locations, by










maintaining a more diverse range of genotypes in farmers' fields, and
by focusing more on instability over time rather than across loca-
tions. Yield stability might also be improved through better
management of fertilizers and pesticides, through improved cultural
practices, and through increased or better managed irrigation. All
these options need to be explored.

Indirect Approaches

If greater yield variability proves to be a barrier to the
adoption of necessary yield increasing technologies, then crop
insurance programs may be appropriate. Unfortunately, past experi-
ence with crop insurance is not encouraging, and the costs of
publicly provided insurance have usually far exceeded their benefits
(Hazell, Pomareda, and Valdds 1986). Nor should the efficiency with
which farmers and traditional village institutions cope with risks be
neglected. Walker and Jodha (1986) have provided a very interesting
paper on these issues. They point out that crop insurance might
simply provide a more costly substitute to existing private risk
sharing arrangements. Improvements to financial institutions might
be a viable approach, particularly an expansion of medium-term
consumer credit so that farmers could borrow money in bad years and
pay it back in good years.
At the national level, increased instability in prices and food
consumption can be contained through buffer stocks. However, IFPRI's
work shows that in most cases it is more cost effective for govern-
ments to use world markets to stabilize domestic consumption, using
the International Monetary Fund's food facility as a source of
funding for food imports when appropriate (Valdes 1981). Interven-
tions can also be targeted, such as food subsidies for the poor,
relief employment, and food-for-work schemes. The efficiency of
these and other interventions are discussed more fully by Sahn and
von Braun (Chapter 6).










REFERENCES

Binswanger, H. P. 1980. "Attitudes Toward Risk: Experimental
Measurement in Rural India." American Journal of Agricultural
Economics 62:395-407.

Dalrymple, Dana. 1976. Development and Spread of High-Yielding
Varieties of Wheat and Rice in the Less-Developed Nations.
Foreign Agricultural Economic Report 95. Washington, D.C.:
U.S. Department of Agriculture.

Day, R. H. 1965. "Probability Distributions for Field Crop Yields."
Journal of Farm Economics 47:713-741.

Dillon, J. L., and Scandizzo, P. L. 1978. "Risk Attitudes of
Subsistence Farmers in Northeast Brazil: A Sampling Approach."
American Journal of Agricultural Economics 60:425-435.

Eberhart, S. A., and Russell, W. A. 1966. "Stability Parameters for
Comparing Varieties." Crop Science 6:36-40.

Evenson, R. E.; O'Toole, J. C.; Herdt, R. W.; Coffman, W. R.; and,
Kauffman, H. E. 1978. "Risk and Uncertainty as Factors in Crop
Improvement Research." IRRI Research Paper Series 15. Manila,
Philippines: International Rice Research Institute.

Finlay, K. W., and Wilkinson, G. N. 1963. "The Analysis of Adapta-
tion in a Plant Breeding Program." Australian Journal of
Agricultural Science 14:742-754.

Gladwin, C. 1977. "A Model of Farmers' Decisions to Adopt the
Recommendations of Plan Puebla." Ph.D. dissertation, Stanford
University.

Hargrove, T. R.; Coffman, W. R.; and Cabanilla, V. L. 1979. "Genetic
Interrelationships of Improved Rice Varieties in Asia," IRRI
Research Paper Series 23. Manila, Philippines: International
Rice Research Institute.

Hazell, Peter B. R. 1984. "Sources of Increased Instability in
Indian and U.S. Cereal Production." American Journal of
Agricultural Economics 36:302-311.

1985. "Sources of Increased Variability in World Cereal
Production Since the 1960s." Journal of Agricultural Economics
36:145-159.

Hazell, Peter B. R.; Pomareda, Carlos; and Valdds, Alberto, eds.
1986. Crop Insurance for Agricultural Development: Issues and
Experience. Baltimore: Johns Hopkins University Press.

Moscardi, E. 1976. "A Behavioral Model for Decision Making Under
Risk Among Small Holding Farmers." Ph.D. dissertation, Univer-
sity of California at Berkeley.










Moscardi, E. R., and de Janvry, A. 1977. "Attitudes Toward Risk
Among Peasants: An Econometric Approach." American Journal of
Agricultural Economics 59:710-716.

O'Mara, G. T. 1971. "A Decision Theoretic View of the Microeconomics
of Technique Diffusion." Ph.D. dissertation, Stanford Univer-
sity.

Roumasset, J. A. 1976. Rice and Risk: Decision Making Among Low
Income Farmers. Amsterdam: North Holland.

1979. "Unimportance of Risk for Technology Design and
Agricultural Development Policy." In Economics and Design of
Small-Farmer Technology, pp. 48-65. Edited by Alberto Valdds,
Grant Scobie, and John Dillon. Ames, Iowa: Iowa State Univer-
sity Press.

Scandizzo, P. C. 1974. Resistance to Innovation and Economic
Dependence in Northeastern Brazil. Working Paper RPO:273/XIV/1.
Washington, D.C., World Bank.

Valdds, Alberto, ed. 1981. Food Security for Developing Countries.
Boulder, Colo.: Westview Press.

Walker, T. S. 1981. "Risk and Adoption of Hybrid Maize in El
Salvador." Food Research Institute Studies 18:59-88.

Walker, T. S., and Jodha, N. S. 1986. "How Small Farm Households
Adapt to Risk." In Crop Insurance for Agricultural Development:
Issues and Experience, pp. 17-34. Edited by Peter Hazell,
Carlos Pomareda, and Alberto Valdes. Baltimore: Johns Hopkins
University Press.

Watson, W. D., and Anderson, J. R. 1977. "Spatial versus Time-Series
Data for Assessing Response Risk." Review of Marketing and
Agricultural Economics 45:80-84.














3

Climatic Changes
and Yield Variability

Timothy R. Carter and Martin L. Parry


Increases in cereal yield variability have been recorded in
certain regions during recent decades. This chapter is intended to
assess the extent to which changes in climate may have contributed to
this increased variability. In attempting to examine this issue,
several difficulties are encountered. There are problems in matching
the scales of analysis (both spatial and temporal), problems in
isolating the significant climatic variables that affect crop yield
variability, and problems with the inadequacies of the data them-
selves.
Three means by which climate can influence crop yield variabil-
ity can be identified: through changes in mean climate, changes in
climatic variability, and changes in cultivated area. We examine the
evidence for these changes by reviewing some of the recent litera-
ture. We conclude that there is no indication, in general, that cli-
matic changes are behind recent increases in cereal yield variabili-
ty, although the role of climate may have been significant in a few
regions (for example, in India, in sub-Saharan Africa, and, very
recently, in Japan). The 1960s stand out in many regions as a period
of generally low climatic variability compared with adjacent periods.
However, the examples here are drawn from a rather ad hoc collection
of sources, which together provide very little convincing evidence
upon which to base any concrete judgement. Further work would re-
quire a more focused approach to the issue, using refined analytical
methods.

ANALYZING CLIMATE EFFECTS

Temporal Variations in Climate and Crop Yield

At any location, measures of weather and climate (such as air
temperature, solar radiation, and precipitation) can be observed or
derived over time scales ranging from microseconds to millennia. The
detection of variations in climate therefore depends critically on
the time frame of reference. Furthermore, the temporal resolution of
climatic data is of great importance in establishing causal relation-
ships between climate and crop yields, for yields too can be analyzed
over a variety of time scales. The range extends from measurements
of the physiological development of crop plants at short intervals
throughout the growing season to harvested yield averaged over
decades or longer periods.











It does not follow, however, that in this investigation, which
concentrates on fluctuations in average annual crop yields, the
climatological analysis necessarily should focus on fluctuations in
average annual climate. It is clear that the yield in a single year
is not simply a function of an average annual climate, but rather the
integrated effect of weather variations on the crop and its environ-
ment both before and during the growing season. Thus it is necessary
to conduct our analyses at several temporal scales of resolution,
although, in practice, considerations of data handling, quality,
length of record, and spatial coverage sometimes restrict our
options.

Spatial Variations in Climate and Crop Yields

Crop yield is an average quality taken to represent a particular
area. On the basis of the size of area used to calculate yields, it
is possible to identify three spatial scales that are of interest in
this study. This enables us to discuss how we can select appropriate
climatic data to match the spatial scale of the yield data.
First, it is important to recognize that a value of regional
crop yield is simply an average that obscures any intrareqional
variations. As yields are commonly measured only over standard
regions (such as counties, districts, or provinces), it is often
difficult to obtain subregional scale data. However, when this is
possible, it is then necessary (for the purpose of climatological
analysis) to obtain data for locations that are representative of
each of the subregions. For example, Huff and Neill (1982), using
district data for five Midwestern states in the United States,
demonstrate that the pattern of variability of July rainfall (the
most important climatic factor influencing yields) matches closely
the pattern of maize yield variability.
Second, one of the explanations cited for increased variability
in national crop yield is that the cultivation techniques and crop
hybrids used have become more spatially homogeneous. It is argued
that with this decrease in regional diversity, countries may be more
vulnerable to simultaneous interregional variations in crop yield
over large geographical areas (for example, see Hazell 1984; Duvick,
Workshop Paper 6). In testing this hypothesis, the effects of
climate need also to be incorporated. Therefore, it is again
necessary to establish a representative data set of regionally
averaged climate from each of the regions for which yields are to be
compared.
The third scale of analysis considers yield variations at remote
locations. The significant crop failures in the Soviet Union in the
1970s and those that are causing devastating impacts in the semiarid
zone of Africa to this day have, for different reasons, awakened
public awareness of the importance and the effects of regional
climatic variations. Over the same period, climatic research has
progressed to the stage where scientists are beginning to identify
global-scale "teleconnections" between climatic events occurring at
geographically remote locations. The implications of this research
are exciting, if only because they suggest causal mechanisms that may
help explain coincident variations in crop yield variability in










regions spatially distant but that together may play a key role in
determining world food trade, prices, and security. These issues are
developed further below.


The Orthodox Classification of Climatic Changes

Four types of climatic change are illustrated in Figure 3.1. In
the context of this paper, we can regard the high-frequency fluctua-
tions depicted here as interannual variations that might be asso-
ciated with year-to-year fluctuations in crop yield. While not
dismissing the possibility of periodic or quasi-periodic variations
in climate, such as are depicted by lines A and B in Figure 3.1, the
time frame of this study (essentially the last few decades) precludes
some of the long-period cycles, which might well resemble a trend
such as that shown by line C in Figure 3.1. The more interesting
climatic variations, from the point of view of crop yield, relate to
shifts in central tendency -- either impulsive, step-like changes
(line B), or gradual trends (line C) -- and to changes in interannual
variability around the average (line D), or, indeed, to a combination
of these.



Figure 3.1. Types of climatic change
Periodic
Variation
A

Impulsive Change of
Central Tendency
VQuasi-Periodic Variation

I |



Downward Trend

C Stable Central Tendencies
3 nw AA. / (Stationary)


Source: Based on Hare 1985.










ISOLATING THE SIGNIFICANT VARIABLES

For a cereal crop, the relative importance of temperature,
light, water, and nutrients varies throughout the crop's development.
During each phase of growth, the crop can be thought of as exhibiting
an optimum tolerance range in response to its environment. If the
weather conditions remain within these tolerance ranges during the
growth of the crop, then these factors should not, by themselves,
impose limitations upon crop yields (although other constraints may,
of course, intervene).
However, the direct effect of climate as a constraint on yields
assumes an importance with the occurrence of anomalous weather events
that fall outside the tolerance range of the crop. These may be of
very short duration (such as night frosts during the flowering phase,
and strong wind, heavy rainfall, or hail before harvest), or occur
over an extended period (such as serious soil moisture deficits
brought on by anomalously low rainfall amounts, or periods of high
rainfall where waterlogging may occur, causing leaching of important
nutrients from the soil or rotting of the crop).
Thus if we are to test whether climate has contributed at all to
increased crop yield variability, a standard climatological analysis
of variability (for example, examining the interannual variance of
air temperature averaged on an annual, seasonal, or monthly basis for
successive decades) may not be wholly appropriate. Instead, we
should first identify the critical tolerance ranges of a crop during
each phase of growth, thus enabling us to evaluate the frequency of
damaging weather events that lie outside these limits. Nonetheless,
this information on its own serves little purpose unless the weather
anomalies can be converted into a measure of effect on crop yield.
For this reason, the use of mathematical models that stimulate crop
responses to climate is of particular value.

The Use of Models

Models, by incorporating various physiological characteristics
of a crop (including tolerance ranges, growth phases, and responses
to environmental conditions) as well as crop management considera-
tions (for example, sowing date, fertilizer applications, and weed
control), can be used to test the sensitivity of crop growth and
yield to any prescribed weather conditions (defined as model input
variables).
For example, a simulation model has been developed by Horie
(forthcoming) to study the sensitivity of Japanese rice yield to
climatic variations. Irrigated rice in northern Japan is particu-
larly susceptible to cold summer damage, which can cause grain
sterility and hence reduced yields. Model runs can help to determine
the nature and timing of the critical weather events (for example,
July-August temperatures and, less importantly, June-September
radiation). Thus, Horie was able to conduct a sensitivity study to
ascertain the tolerance range of crop yield to these critical
anomalies.
Of course, in conducting experiments of this kind, it is assumed
that a model has been satisfactorily verified against actual condi-










tions. If this condition is fulfilled, then models may be quite
effective tools for identifying those particular climatic anomalies
to which a crop is more sensitive, while at the same time holding
other factors (such as crop variety, fertilizer application, and
sowing density) constant. Inevitably these factors are also likely
to change through time, so it is clear that, in order to conduct a
rigorous study of the impact of climate on crop yield variability
over several decades, all of these adjustments need to be considered
and modeled. This task is feasible in theory but only realizable in
a few selected cases, where models and data are available.


Climatic Change as a Change in Risk

It is possible, as noted, to evaluate the frequency of damaging
weather events that lie beyond the tolerance ranges of a crop. Given
long-period data sets, such frequencies can be converted to measures
of probability, forming the basis for assessments of the risk of
climate impact. In areas where a single weather variable is the
dominant yield-determining factor (such as precipitation in most dry
regions and temperature in high latitude or high altitude locations),
and in the absence of a detailed model, it is often instructive to
compute indices of risk of climate-induced crop failure and to
analyze how these risk levels may change between periods. (For
example, see the analysis of failures in oats harvests over three
centuries in southern Scotland by Parry and Carter 1985). As an
important corollary to such studies of risk, if it is assumed that
interannual climatic variability remains unchanged, any impacts from
long-term changes in the mean climate are likely to be felt through
changes in the risk of short-term events (Parry and Carter 1985;
Mearns et al. 1984).

Mapping Spatial Shifts of Crop Suitability

The type of crop cultivated in a particular area is a function
of many physical, economic, social, and political factors, but the
preference for one crop rather than another usually bears some
relation to its suitability for the prevailing climatic conditions of
that region. While boundaries between crop types are difficult to
discern on the ground, since they are really transition zones of
comparative advantage, they can often be located approximately on the
basis of climatic criteria related to crop tolerance ranges.
However, we know that climate is not static, and secular changes
either in the mean or the coefficient of variation will affect the
location of the mapped isopleths. The risk of yield shortfalls will
shift, and the hypothetical cultivation limits (however defined) will
either contract across formerly cultivated land or extend into new
territory. Such shifts have been illustrated by Newman (1980), who
simulated the effect on the location of the U.S. corn belt of
hypothetical changes in mean annual air temperatures of 1t centi-
grade relative to the 1969-78 normal temperatures.










CLASSIFYING THE PATHWAYS OF CLIMATE EFFECTS

The effect of climate on crop yield variability may be transmit-
ted via a number of different routeways.

Changes in Mean Climate

Using the concept of a crop tolerance range, we can illustrate
(Figure 3.2a) how an existing crop is likely to be relatively well
matched to the prevailing climate. In this example the probability
of critical climatic events (falling outside the tolerance range) is
quite low (shaded regions) implying a correspondingly high chance of
a successful crop yield. Figure 3.2b shows how a change in mean
climate, with no attendant change in variability (see also line B in
Figure 3.1), causes the whole distribution to shift relative to the
normal situation. Moreover, if this occurred as an abrupt perturba-
tion, we could expect there to be no initial change in locally grown
crop variety, so the tolerance limits remain fixed.
Thus the shift in mean climate would, ceteris paribus, have a
destabilizing effect on yields, with an increase in probability of
upper-tail (or lower-tail for a shift in the other direction) anoma-
lies disproportionately greater than the decrease in probability of
lower-tail or upper-tail anomalies. However, if the transition to
this changed climatic condition is gradual, it is likely that farmers
would respond by planting a better adapted variety or a completely
new crop, in order to maintain yield stability.


Changes in Climatic Variability

Figure 3.2c depicts the equivalent situation to that in Figure
3.2b, but here it is the variance of the climate that changes (in


Figure 3.2. Change in climate relative to normal

(a) Normal Climate (b) Change in Mean (x x') (c) Change in Variability

Tolerance Range Tolerance Range Tolerance Range
-- -



TI x T T1 xx' T2 T1 T2
Note: Shaded areas represent those parts of the frequency
distribution of a climatic parameter lying outside the
tolerance range (T1 T2) of a particular crop.


1Although the frequency distributions of some climatic variables
such as rainfall or windspeed are often nonnormal, Figure 3.2a
suffices for illustrative purposes.











this example, symmetrically) about a fixed mean (for example, see
line D in Figure 3.1). Naturally our expectation would be of
increased (or decreased) yield variability in response to equivalent
changes in climatic variability. Note, however, that in these
examples, if the normal range of climatic fluctuations had been well
within the tolerance range (as might be the case in the core of a
large grain growing region, for example) then the changes in climate
represented in Figure 3.2b and 3.2c may have only a minor effect.

Changes in Cultivated Area

In a large region, the relationship between the size of cultiva-
ted area and the regional climate can certainly influence crop yield.
In an earlier section we considered the effect of climatic changes on
spatial boundaries of cultivation. Toward the limits of tolerance in
a region, we would expect a crop to be more sensitive to climate and
thus to exhibit higher variability of yields. However, if the
cropped area is adjusted, this could affect regional yields either by
expanding cultivation into more marginal land or by contracting into
the more suitable, low-risk areas.
Aggregating these effects, we can construct a 4 x 3 matrix of
climatic versus area effects on crop yield variability, showing
qualitatively our intuitive expectations of the outcomes as an
increase, as a decrease, or as no appreciable change (Table 3.1).


EVIDENCE FOR CHANGES IN MEAN CLIMATE

Long-term climatic data series for surface conditions (most
appropriate for agricultural applications) are available from
hundreds of meteorological stations over the globe, but their
coverage is extremely variable both in space and time, making it very


Table 3.1--Response of crop yield variability to changes
in regional climate and cultivated area


Cultivated Area
Current Expanded Contracted
Climate Area Area Area

Present climate -
Change in mean climate + ft t
Increase in climatic
variablity t t -
Decrease in climatic
variability 44


Note: t increase, 4 decrease, little change.











difficult to evaluate unbiased averages over large areas. The usual
procedure is to use statistical methods to interpolate the station-
point values to a grid network and from this to compute mean values.
The two variables that have received the most attention are surface
air temperature and precipitation. Variations in these are consid-
ered separately below and at two spatial scales: the global and
zonal scale and the regional scale.

Global Temperature Variations

The majority of the long-period mean annual temperature series
have been constructed for the Northern Hemisphere, extending over
about a century of observations, by, for example, Jones et al. (1982,
in press) and Vinnikov et al. (1980) on mainly land stations; and
Folland et al. (1984) using marine data. All point to a mean
hemispheric warming of around a half degree centigrade from about
1880 until the mid-1980s, punctuated by a cooling phase in the period
1940-65.
Records for the Southern Hemisphere are rather sparse, but some
attempts have been made to produce temperature series. For example,
Hansen et al. (1981) computed trends for the southern latitudes as
part of a global analysis of surface temperatures. While the pattern
of change is different from that in the Northern Hemisphere, a
similar long-term warming tendency is evident.
Several workers have suggested that these trends provide
evidence of a carbon dioxide induced climatic warming, which, when
combined with estimates of the effects of volcanic aerosol loading
and variations in solar activity, can explain a large percentage of
the observed temperature variations (see Hansen et al. 1981, and
Gilliland 1982). However, such claims should be treated with caution
until robust statistical procedures have been developed and applied
to test the significance of model fits (Weller et al. 1983).
This comment underlines the need for continued monitoring of
surface temperature and further investigations into the causes of
observed changes. For example, the important role of the oceans in
regulating the climate cannot be ignored (see also Hansen et al.
1985), nor the observed increases of other (that is, non-C02) "green-
house" gas concentrations in the atmosphere.
Trends in temperature have also been identified on a seasonal
basis. Using the Gruza and Ran'kova (1980) Northern Hemisphere data
set for January and July, Angell and Gruza (1984) report that the
warming up to 1940 and the subsequent cooling until about 1965 show
up strongly in the January data but hardly at all in the July data,
implying that the hemispheric trends are dominated by winter condi-
tions.
Finally, there is a striking contrast between temperature series
from different latitudinal zones in the increase in interannual
variability toward higher latitudes. This has been illustrated by
Angell and Gruza (1984) for five latitudinal zones, and by Kelly et
al. (1982), who remark that the range of variations in the data for
the Arctic zones (65-85'N in their analysis) is three times greater
than the range for the Northern Hemisphere average.










Global Precipitation Changes

Long-period records of precipitation, like temperature, are
quite abundant for land stations, particularly in the Northern
Hemisphere. However, in contrast to temperature, it is much more
difficult to evaluate average precipitation amounts over large areas.
There are several reasons for this.
First, precipitation amounts can be highly variable over space,
especially in dry regions subject to localized and infrequent
convective rainfall, which can distort both individual station
records and area averages, so that special procedures are necessary
to produce large-area averages that are at all representative. A
second problem concerns the large areas, particularly over the
oceans, where data are extremely scant. Third, the effect of
orography on precipitation is more difficult to quantify than on
temperature, and further complicates the compilation of area aver-
ages. Finally, precipitation trends at the global and hemispheric
scale have received little attention from atmospheric scientists, who
have focused on testing their hypotheses about global temperature
changes.
The evidence that is available suggests that in most latitude
bands in the Northern Hemisphere there has been an increase in
January precipitation between 1945 and 1960 and a decrease thereafter
to 1975 (Angell and Gruza 1984). In the July records, there are
differences between zones, although a similar trend to that in
January is strongly evident in the low latitudes. Over the conti-
nents, there is evidence of a long-term increase in January precipi-
tation in northern Asia and in Eurasia, but in July this trend is not
apparent, although there is an interesting "out-of-phase" relation-
ship between long-term variations in July precipitation in America
and those in northern Asia (Angell and Gruza 1984).
Lamb (1977) has expressed 1960-69 annual precipitation over the
globe as a percentage of 1931-60 averages. An equivalent comparison
of the 1970-79 pattern relative to the 1931-60 averages has also been
conducted for the Northern Hemisphere (Lamb 1981), and while there
are some differences between the patterns depicted, there are also
noteworthy similarities. Both show significant negative departures
over subtropical Africa, northern India, China, and much of the U.S.
Great Plains. Positive anomalies are evident in both periods between
northern European U.S.S.R. and eastern Europe, the southern United
States, and western Scandinavia.


Regional Temperature Variations

It can be misleading to associate large-scale climatic change
with changes in regional agricultural production. For instance,
during the hemispheric "cooling" period (1940-65) some areas actually
recorded a warming trend including the Ukraine, a major grain growing
region. Other areas, for example, northern U.S.S.R., Alaska, and
northwestern Canada, recorded significant cooling (Jones and Kelly
1983). The recent "warming" (1965-80), in contrast, has been
strongest (>0.5*C) over northern Scandinavia, most of the U.S.S.R.,
Alaska, northwestern Canada, the southwestern United States, and










northern Africa, with cooling occurring over the Canadian Arctic
islands and northeast Greenland.
Interpretation of regional temperature changes is further com-
plicated by the differences in seasonal temperature trends. Small
changes in annual mean temperature often mask large trends in the
seasonal data. Of course the relevance of such trends for crop pro-
duction depends upon the magnitude of changes and the particular
seasons in which changes occur. For example, in northern Japan, rice
is grown toward the limits of its tolerance to cool temperatures, so
the recent warming trend (since about 1955), which is apparent in
summer but does not show up in either the winter or annual records
(Yoshino forthcoming), may help to explain the rather stable rice
yields about an increasing trend during the period 1957-79 (Uchijima
and Seino, forthcoming). However, four consecutive years (1980-83)
of damage from cool summers have reemphasized that rice production in
Japan is still vulnerable to the effects of climate. Thus although
increasing temperatures would probably favor more stable rice produc-
tion, there still remain cooler episodes embedded within the trend
that can be damaging.
In the Southern Hemisphere, the rise in annual mean maximum
temperatures in much of Australia was recorded between 1946 and 1975,
with the greatest increase occurring in inland southeastern Australia
(Coughlan 1979; Paltridge and Woodruff 1981). While there are
considerable variations between regions, and while it is difficult to
state unequivocally that a general warming has occurred in Australia
(Hobbs forthcoming), there are also clear indications of warming in
New Zealand from the 1860s until the present (Salinger 1979), leading
some to suggest a possible causal link to CO2 induced climatic change
(for example, Pittock 1983). Trends toward increased precipitation
since 1946 (discussed below) would also be consistent with the
current estimates from the general circulation model of CO2 related
climatic change in this region.

Regional Precipitation Changes

The most distinctive and disturbing precipitation trends of this
century have occurred in the semiarid sub-Saharan zones of Africa
over the last 20 years. Even accounting for differences in the
various data sets, all the rainfall series that have been compiled
indicate a similar downward trend in annual rainfall since the 1950s
(Farmer and Wigley 1985). A comparison of these records indicates
how the drying trend intensified after the 1960s, with three waves of
particularly dry conditions peaking in 1972, in 1977, and in 1983
through 1984 (the two driest years this century).
One effect of the sub-Saharan drought has been to produce a
marked spatial shift of the rainfall regime. This is demonstrated in
Figure 3.3, which shows how the isohyets of mean annual rainfall in
Senegal calculated for the 16-year period 1968-83 are displaced by
about 113 kilometers south of the isohyets based on long-term mean
rainfall calculated for 1954-83. The agricultural implications of
such a shift (paralleled across most of the sub-Saharan zone) include
reduced rainfall amounts, a shortened growing season, and a lower
probability of sufficient rainfall for crop production (that is,
increased risk of crop failure; Todorov 1985).











Figure 3.3. Isohyets of mean annual rainfall in Senegal

....--------- 200 mm
.-.. Podor
200-258 Kilometers

St. Louis .---- -------------M400 mm
233-281
81- 1954-83
Linguere Matam
358-418 314-411113 1968-83


^^'.--"""' Diourdel --.
-o482-589 ------ -- 600 mm
Dakar 48-2*-589
329-458\' _---800 m
Kao ack Tambacounda
53n d 7s ae nt 26 1s in s e

Kolda
977-1128
.. Kedougou
---- 1147-1239
may be brought oiguinchorgrazing or by drought itself), by affecting
1151-1347 '-- .----------
-1,400 mI
Source: Based on Todorov 1985.


Despite the many hypotheses put forward to explain the sub-
Saharan drought, the causes are not known. Changes in surface albedo
and soil moisture (characteristics often strongly correlated, which
may be brought on by overgrazing or by drought itself), by affecting
the surface radiation balance, can induce increased subsidence.
These effects have a physical basis and can be replicated using
computer models. However, in the sub-Saharan regions, the available
evidence does not provide any convincing proof that these are major
controlling mechanisms (see Rasool 1984).
The majority (over 70 percent) of precipitation in the Sahelian
zone of the sub-Sahara originates from squall lines (World Climate
Data Program 1985), which are associated with convergence in the
lower troposphere between two upper-level easterly air flows, the
tropical easterly jet to the south and the African easterly jet to
the north. A weakening of these large-scale features can seriously
disrupt the circulation patterns, resulting in enhanced upper-air
convergence and subsidence, with consequent rainfall reductions at
the surface. Weakening such as this occurred in 1982 and 1983 and
may be related to the El Nino/Southern Oscillation (ENSO) phenomenon
(see Figure 3.4; also see appendix to this chapter). However, while
perhaps responsible for periodic enhancement of drought intensity,
the ENSO-related events cannot explain the progressive drying trend
in this region. It has been speculated that in many years, although
the squall lines have been present (controlled as they are by large-
scale effects), they have produced less rainfall because each system
contains less water vapor (a possible result of deforestation, poor
vegetation cover, and dry soils; Kandel, personal communication).










Figure 3.4. Rainfall in the Sahel and the occurrence of the
El Nino/Southern oscillation phenomenon, 1900-83


+1.0




-1.0

1 E1 Nifo

I I I I I I 1 I I I I I I I I I I
1900 1910 1920 1930 1940 1950 1960 1970 1980
Source: Based on World Climate Data Progrm 1985.



Finally, the severity of the present conditions relative to
earlier recorded droughts has prompted some to speculate on the
possible role of atmospheric carbon dioxide and trace gas concentra-
tions in influencing the climate of the region (for example, Farmer
and Wigley 1985). However, temperature and precipitation records
offer few clues about any global trends during the past 30 years that
might be associated with the changes in the sub-Saharan zone.
Evidence for long-term changes in annual precipitation can also
be discerned from the instrumental records of rainfall in Australia.
Much of eastern Australia was wetter in the latter part of the
nineteenth century than in the first half of the twentieth century,
and this may account for the failure of denser settlement in much of
northeastern South Australia, particularly in the 1890s (Hobbs
forthcoming). After 1945/46, an increase in annual rainfall amounts
(concentrated in the spring, summer, and autumn months) of about 10
to 20 percent was observed over the major wheat growing area of
southeast Australia (for example, Pittock 1983). Along with the
temperature increases in some regions (see above), a spatial shift in
the pattern of climate occurred, which might well have affected grain
yield stability in the wheat belt.


EVIDENCE FOR CHANGES IN CLIMATIC VARIABILITY

Changes in Temperature Variability

There is no clear evidence of a link between trends in inter-
annual variability and trends in mean hemispheric temperature over
the last century (van Loon and Williams 1978). More specifically, in
recent years (1961-78) there has been no significant increase in
annual temperature variability in Europe relative to the 1931-60









period (Schuurmans 1984), although there is some evidence of a
negative correlation between period mean winter air temperature and
interannual variability (Schuurmans and Coops 1983).
In the United States Midwest a tendency toward decreased
interannual mean monthly temperature variability between the 1930s
and the 1970s was observed, mainly during the winter months (Chico
and Sellers 1979). In common with a tendency in the Northern
Hemisphere as a whole, however, this does not appear to be related to
trends in mean temperature (van Loon and Williams 1978).
Temperature variability can also be viewed in terms of shorter
time-scale fluctuations and changes in the range of temperature
extremes. For example, over much of the United States and Canada, a
significant decrease (most apparent in the summer and autumn) in the
mean diurnal temperature range (that is, the difference between daily
maximum and minimum temperatures) has been observed during the period
1941-80 (Karl et al. 1984). These results can be considered along-
side a recorded increase of about 1' centigrade in mean spring and
summer temperatures and a decrease of 20 to 40 percent in spring and
summer precipitation in the central and northern United States Great
Plains over a similar period (Karl and Riebsame 1984). Combining all
of these results, it is pertinent to ask two questions:
e What has been the impact on crop production, if any, of these
apparent tendencies during the growing season in the Great Plains
toward increased mean temperatures, decreased diurnal temperature
range, and decreased mean precipitation?
* Furthermore, if (as conjectured by Karl and Riebsame) these trends
could serve as a partial analog of a future CO2 enriched climate,
what might the impacts of continued changes be on the level and
stability of crop production?
As partial response to the first question, there is little doubt
that benign weather conditions contributed to very favorable and
stable crop yields in the Great Plains during the period 1956-73
(Thompson 1975). However, recent events such as the 1984 drought
serve as reminders of the potential effects of the climate on crop
yields, especially if such events are likely to become more frequent
through trends such as those implied above.

Changes in Precipitation Variability

In general, for any latitude zone, interannual precipitation
variability tends to be highest where the mean annual rainfall is
lowest. This relationship implies that, in the absence of irriga-
tion, where crops are grown in dry regions, not only are average
water resources restricted, but they are also highly variable. The
coefficient of variation of annual precipitation trends in semiarid
areas such as the Sahel zone of Africa, northwestern and central
India, northeastern Brazil, and much of Australia usually exceeds 25
percent and often approaches 40 percent at desert margins (Lockwood
forthcoming). In some regions positive annual anomalies may imply
torrential rain and flooding, but in all regions, large negative
departures indicate drought, and this generally implies crop failure.
It has been suggested that no regular secular pattern in the interan-
nual variability of precipitation can be detected for semiarid










regions, except perhaps a very weak two-to-three-year rhythm in some
areas, and a 10-, 20-, or 30-year recurrence interval in others
(Rasool 1984). We examine this assertion further below.
Whereas meteorologists look to the global temperature record for
indications of global-scale climatic change, they use regional
precipitation patterns to provide clues about the mechanisms under-
lying short-term climatic variability. Several global teleconnec-
tions have been noted. For example, there is a strong tendency for
the El Nifo phenomenon to occur simultaneously during -- or within
one year of -- drought in northeast Brazil (Hastenrath 1984), and
ENSO events seem also to be associated with some of the severest
droughts in Australia (including five out of the six that have
occurred since 1950; Hobbs forthcoming). Similarly, there appears to
be a strong negative correlation between the occurrences and inten-
sity of El Niio and the Indian summer monsoon rainfall (Rasmusson and
Wallace 1983). Moreover, there appears to be a significant positive
correlation between the strength of the Indian monsoon and the
strength of the zonal circumpolar westerlies over Eurasia, implying
that when the Indian monsoon is drier than usual there is a tendency
toward blocking and meridional flow in the middle latitudes (Raman
and Maliekal 1985). This relationship supports an independent
observation of increased irregularity of the Indian monsoon in the
1960s and 1970s relative to the period 1925-60, coincident with a
general decline from about 1950 to 1980 in the number of days with
surface westerly winds over the British Isles, a good indicator of
the strength of the zonal circulation (Lamb 1981).
A century-long rainfall series for southeastern Africa reveals
that 22 of the 28 ENSO events during the past 110 years were accom-
panied by below-normal rainfall (1875-1977 mean). There are indica-
tions that the periodic intensification of the sub-Saharan drought
mentioned above was related to the occurrence of ENSO (Figure 3.4),
and ENSO episodes have also been correlated with above-average winter
temperatures in southwestern Canada and the northwestern United
States, and with abnormally wet winters in the Gulf States and
northern Mexico (Rasmusson and Wallace 1983). However, as these
authors point out, there are many regions where there is no evidence
for systematic patterns of anomalies during ENSO events (for example,
in Eurasia and over much of North America), although several of these
areas have recorded unusual weather during an individual episode,
such as the 1982-83 ENSO.
Further work in Brazil (Molion and Nobre forthcoming) and in
Australia (Nicholls and Woodcock 1981; Hobbs forthcoming) explores
the possibility of exploiting the teleconnections between measurable
atmospheric and oceanic effects to allow prediction of droughts
several months in advance of their occurrence. Results seem encour-
aging but verification is difficult.
More important for the purposes of this investigation is the
identification of any trends toward increased or decreased vari-
ability of precipitation over the last few decades. The recent
increased irregularity of the Indian monsoon and the sub-Saharan
drying trend have already been noted, but in other regions there is
little evidence of significant changes in variability. However, the
global teleconnections do show us that simultaneous anomalies of
climate are, in many regions, linked in some way to the ENSO phenome-











non and are not simply coincidental. Furthermore, the ENSO events,
when they occur, are often associated with some of the severest
climatic anomalies. It may be that, while the episodes themselves
are probably no more frequent than previously, the world food system
is more sensitive to the distinctive global anomalies that they
induce.
One characteristic of many precipitation series that hampers the
interpretation of relationships between yield variability and climate
is the clustering of years with anomalously low rainfall. One good
example of this is the record of drought in the United States Great
Plains where the 1930s and 1950s show up as periods of persistent
drought conditions (Warrick 1980). Similar traits of drought
clustering are also evident in the Soviet Union (Rauner 1985).
The probability of simultaneous drought has also been investi-
gated. For example, estimates show that droughts are likely to occur
simultaneously in both the U.S.S.R. (Asian and European) and United
States grain growing regions in about 1 year in 20 or 25 (Rauner
1980). Drought is a major cause of wheat yield variability in each
of these regions, and it is interesting to match this probability (4-
5 percent) with the probability of significant wheat yield shortfalls
(10 percent or more below mean trend yield) occurring simultaneously
in both countries, estimated at about 7-8 percent (Sakamoto et al.
1980). The implications for world food security of such events could
be far reaching, particularly if their likelihood increases as a
result of climatic change.

EVIDENCE FOR CHANGES IN CULTIVATED AREA

It is extremely difficult to judge the extent to which changes
in cropped area have influenced regional crop yield variability. For
instance, of the 800 thousand hectares of paddy rice taken out of
cultivation in Japan from 1970-84, about 20 percent was withdrawn in
Hokkaido, the northernmost and most unstable production area (Yoshino
forthcoming). This contraction into more favorable areas (a policy
for reducing total production) may have contributed to a subsequent
increase in mean yields and to a stabilizing of total production,
although other factors, including improved technology and management
and the weather itself, are also important and may tend to mask any
trends.
In the United States, each year a certain percentage of the
planted area is not harvested, particularly in regions that are
climatically marginal for crop growth. In the Great Plains, unfavor-
able weather conditions are of major significance in influencing the
farmers' decisions to abandon spring and winter wheat crops (Michaels
1985), actions that may have contributed to a stabilization of
harvested yields in this region.
In the U.S.S.R., the expansion in the 1950s of grain cultivation
to drought-prone regions in the new lands of Kazakhstan and western
Siberia, though adding a highly variable component to total U.S.S.R.
grain production, may well have helped, paradoxically, to stabilize
national yields, for good yields in these regions tend to offset poor
yields in the traditional grain growing regions (Sakamoto et al.
1980). Note that the assessment of crop yield in the U.S.S.R., since
it is based on planted area and not harvested area as in the United










States (Tarrant, Workshop Paper 29), may exaggerate the variability
of crop yields relative to that in the United States, although other
differences in yield measurement may negate this disparity.

CONCLUSIONS

It is extremely difficult to evaluate the role of climate in
influencing changes in grain yield variability. There are problems
of
* separating out the role of climate from a host of other factors
that together determine crop yields and yield variability;
e identifying the climatic variables that are significant in in-
fluencing crop yields and matching their spatial and temporal
scales of measurement to those of the crop;
* detecting possible trends in the climatic data that may have
affected crop yield variability; and
* explaining the connection between remote variations in climate and
how these may affect regional grain yield variability and the
stability of world grain production and trade.


Summary of Results

We have attempted to synthesize some of the available knowledge
on how climate can influence crop yield variability. Three cate-
gories of change were identified as potential influences on crop
yield variability: (a) changes in mean climate; (b) changes in
climatic variability; and (c) changes in cultivated area. Evidence
for changes of each of these types (and at a variety of spatial
scales) have been identified. A summary of the first two types of
direct climatic changes for 1940-80 are presented in Figures 3.5 and
3.6. These are not intended as exhaustive lists; the trends (if any)
are depicted on a decade-to-decade basis and are meant to provide an
impression of the appropriate relative magnitude of any changes.
From the material assembled in Figures 3.5 and 3.6 and through-
out the chapter, we conclude that there is no indication, in general,
that climatic changes are behind recent increases in cereal yield
variability, although the role of climate may have been significant
in a few regions (for example, in India, in sub-Saharan Africa, and,
very recently, in Japan). The 1960s stand out in many regions as a
period of generally low climatic variability compared with adjacent
periods. At a global scale there is widespread evidence for a recent
warming trend from about 1965 to the present. Trends in precipita-
tion are more difficult to discern, but there appear to be strong
global teleconnections between certain regional precipitation anomaly
patterns and the El Niio/Southern Oscillation phenomenon.


Further Work

The examples presented here have been drawn from a variety of
sources, but together they still provide very little convincing
evidence upon which to base any concrete judgements. If more compre-
hensive and definitive results are to be obtained, there is need














Figure 3.5. Change in mean climate by continent
and hemisphere, 1940-80


Mean
Area and 0 0
climate variable O S
Global mean
annual temperature
Northern Hemisphere
mean annual temperature
Northern Hemisphere
precipitation 30 800N
Northern Hemisphere
precipitation 0 800N


Sub-Saharan
annual precipitation
Southeast Africa
Nov.-March precipitation


U.S. Great Plains
summer temperature
U.S. Great Plains
summer rainfall


Northeast Brazil
annual precipitation





Northern Japan
mean summer temperature /
India
monsoon rainfall
China annual
precipitation
Southeastern Australia
summer precipitation
New Zealand
mean annual temperature


U.S.S.R annual precipi-
tation European zone





Note: Assessments are approximately on a decadal basis except where
a dashed line is used. Height of bars depicts relative change
and is not to scale.













Figure 3.6. Change in climate variability by region, 1940-80


variable 1940 50 60 70 80
Area and climate 1940 50 60 70 80
variable
Frequency of negative south
Australian index
Sub-Saharan
precipitation variability
Southeast Africa
precipitation variability


Canadian plains
range of temperature
U.S. Great Plains
drought frequency r 'a


Northeast Brazil
precipitation variability



Japan cool sumner
crop damage
India rainfall '
variability



South Island, New Zealand tenr\
perature-based frequency of
potential wine-producing years
North Island, New Zealand
drought frequency


British Isles
days with westerly winds

Note: Assessments are approximately on a decadal basis except where
a dashed line is used. Height of bars depicts relative change
and is not to scale.


* to evaluate the critical tolerance ranges of crops with respect to
climate, estimating these according to the crop varieties grown in
a particular regional environment during the investigation period;
* to conduct direct analyses of climatological data using methods
that will produce information pertinent for studies of crop yield
variability; for example, studies of trends in maximum and minimum
temperatures and evaluations of frequencies of climatic events that
are damaging to crops;
* to use appropriate and validated crop-climate models that can
simulate, by holding other factors constant, the sensitivity of
crop yields to climate over several decades, allowing the user to
identify any climatic variables that may be responsible for changes
in modeled crop yield variability; and
* to monitor changes in cultivated area and abandonment trends
(seasonal or longer) of planted cropland, so that these can be
incorporated into any assessment of crop yield variability.










APPENDIX: THE EL NINO AND THE SOUTHERN OSCILLATION (ENSO)

El Niho, or the Christ child (so called because it generally
develops soon after Christmas), is a southward flowing ocean current,
which brings warm waters to the normally cool coast of Ecuador and
Peru. In recent years, this local oceanic phenomenon has been linked
to a global atmospheric phenomenon known as the Southern Oscillation
(SO). The SO is related to the observation that high atmospheric
pressure over the Pacific Ocean is usually associated with low
pressure over the Indian Ocean. This normal condition is linked, in
turn, to average tropical atmospheric circulation patterns, which are
characterized by three major convective rain generating areas of
rising motion over southeast Asia and the western Pacific, tropical
South America (Amazonia), and Africa (the Congo). Rainfall varies in
the opposite direction to pressure, so that during a reversal
(oscillation) of the pressure field, a high-pressure anomaly develops
over the Indian Ocean (with low-rainfall anomalies over Australia and
south and southeast Asia), and pressure becomes anomalously low over
the equatorial central and eastern Pacific (giving above-average
rainfall in these regions). The effect may be transmitted to other
regions by shifting the location of the equatorial "Walker" circula-
tion cells.











REFERENCES

Angell, J. K., and Gruza, G. V. 1984. "Climate Variability as
Estimated from Atmospheric Observations." In The Global
Climate, pp. 25-26. Edited by J. T. Houghton. Cambridge:
Cambridge University Press.

Chico, T., and Sellers, W. D. 1979. "Interannual Temperature
Variability in the United States Since 1886." Climatic Change
2:139-147.

Coughlan, M. J. 1979. "Recent Variations in Annual Mean Maximum
Temperatures Over Australia." Quarterly Journal of the Royal
Meteorological Society 105:707-719.

Dennet, M. D.; Elston, J.; and Rodgers, J. A. In press. "A Reap-
praisal of Rainfall Trends in the Sahel." Journal of Clima-
tology.

Farmer, G., and Wigley, T. M. L. 1985. "Review of Climatic Trends
for Tropical Africa." In Report to the U.K. Overseas Develop-
ment Administration, p. 33. London: ODA.

Folland, C. K.; Parker, D. E.; and Kates, F. E. 1984. "Worldwide
Marine Temperature Fluctuations 1856-1981." Nature 310:670-673.

Gilliland, R. L. 1982. "Solar, Volcanic, and CO2 Forcing of Recent
Climatic Changes." Climatic Change 4:111-131.

Gruza, G. V., and Ran'kova, E. Ya. 1980. Structure and Variability
of Observed Climate. Northern Hemisphere Air Temperature.
Leningrad: Gidrometeoizdat.

Hansen, J.; Johnson, D.; Lacis, A; Lebedeff, S.; Lee, P.; Rind, D.;
and Russell, G. 1981. "Climate Impact of Increasing Carbon
Dioxide." Science 213:957-966.

Hansen, J.; Russell, G.; Lacis, A.; Fund, I.; and Rind, D. 1985.
"Climate Response Times: Dependence on Climate Sensitivity and
Ocean Mixing." Science 229:857-859.

Hare, F. K. 1985. "Climatic Variability and Change." In Climate
Impact Assessment, SCOPE 27, pp. 37-68. Edited by R. W. Kates,
J. H. Ausubel, and M. Berberian. Chichester: Wiley and Sons.

Hastenrath, S. 1984. "Predictability of North-east Brazil Droughts."
Nature 307:531-533.

Hazell, Peter B. R. 1984. "Sources of Increased Instability in
Indian and U.S. Cereal Production." American Journal of
Agricultural Economics 66:302-311.











Hobbs, J. Forthcoming. "Climatic Patterns and Variability in the
Australian Wheatbelt." In Assessment of Climate Impacts on
Agriculture. Volume 2: In Semi-Arid Regions. Edited by M. L.
Parry, T. R. Carter, and N. T. Konijn. Dordrecht: Reidel.

Horie, T. Forthcoming. "Rice Yield Under Changing Climatic Condi-
tions in Hokkaido Island." In Assessment of Climate Impacts on
Agriculture. Volume 1: In High Latitude Regions. Edited by M.
L. Parry, T. R. Carter, and N. T. Konijn. Dordrecht: Reidel.

Huff, F. A., and Neill, J. C. 1982. "Effects of Natural Climatic
Fluctuations on the Temporal and Spatial Variation in Crop
Yields." Journal of Applied Meteorology 21:540-550.

Hulme, J. 1985. "Secular Climatic and Hydrological Change in Central
Sudan." Ph.D. dissertation, University College, Swansea, U.K.

Jones, P. D., and Kelly, P. M. 1983. "The Spatial and Temporal
Characteristics of Northern Hemisphere Surface Air Temperature
Variations." Journal of Climatology 3:243-252.

Jones, P. D.; Raper, S. C. B.; Bradley, R. S.; Diaz, H. F.; Kelly,
P. M.; and Wigley, T. M. L. In press. "Northern Hemisphere
Surface Air Temperature Variations, 1951-1984." Journal of
Climatology and Applied Meteorology.

Jones, P. D.; Wigley, T. M. L.; and Kelly, P. M. 1982. "Variations
in Surface Air Temperatures: Part I, Northern Hemisphere, 1881-
1980." Monthly Weather Review 110:59-70.

Karl, T. R.; Kukla, G.; and Gavin, J. 1984. "Decreasing Diurnal
Temperature Range in the United States and Canada from 1941
through 1980." Journal of Climatology and Applied Meteorology
23:1489-1504.

Karl, T. R., and Riebsame, W. E. 1984. "The Identification of 10- to
20-year Temperature and Precipitation Fluctuations in the
Contiguous United States." Journal of Climatology and Applied
Meteorology 23:950-956.

Kelly, P. M.; Jones, P. D.; Sear, C. B.; Cherry, B. S. G.; and
Tavakol, R. K. 1982. "Variations in Surface Air Temperatures:
Part II, Arctic Regions, 1881-1980." Monthly Weather Review
110:71-83.

Lamb, H. H. 1977. Climate: Present. Past and Future. Volume 2:
Climatic History and the Future. London: Methuen.

S 1981. "Climatic Changes and Food Production: Observa-
tions and Outlook in the Modern World." GeoJournal 5:101-112.











Lamb, P. J. 1985. "Rainfall in Sub-Saharan West Africa During 1941-
43." In Third Conference on Climate Variations and Symposium
on Contemporary Climate: 1850-2100, pp. 64-67. Washington,
D.C.: American Meteorological Society.

Lockwood, J. G. Forthcoming. "Climate and Climatic Variability in
Low Latitudes." In Assessment of Climate Impacts on Agricul-
ture. Volume 2: In Semi-Arid Regions. Edited by M. L. Parry,
T. R. Carter, and N. T. Konijn. Dordrecht: Reidel.

Mearns, L. 0.; Katz, R. W.; and Schneider, S. H. 1984. "Extreme High
Temperature Events: Changes in Their Probabilities with Changes
in Mean Temperature." Journal of Climatology and Applied
Meteorology 23:1601-1613.

Michaels, P. J. 1985. "Economic and Climatic Factors in 'Acreage
Abandonment' Over Marginal Cropland." Climatic Change 7:185-
202.

Molion, L. C. B., and Nobre, C. A. Forthcoming. "The Climatology of
Drought in N.E.O Brazil and Drought Predictions." In Assessment
of Climate Impacts on Agriculture. Volume 2: In Semi-Arid
Regions. Edited by M. L. Parry, T. R. Carter, and N. T. Konijn.
Dordrecht: Reidel.

Newman, J. E. 1980. "Climate Change Impacts on the Growing Season of
the North American 'Corn Belt'." Biometeorology 7(2):128-142.

Nicholls, N., and Woodcock, F. 1981. "Verification of an Empirical
Long-Range Weather Forecasting Technique." Quarterly Journal of
the Royal Meteorology Society 107:973-976.

Nicholson, S. E. 1983. "Sub-Saharan Rainfall in Years 1976-80:
Evidence of Continued Drought." Monthly Weather Review
111:1646-1654.

Paltridge, G., and Woodruff, S. 1981. "Changes in Global Surface
Temperature from 1880 to 1977 Derived from Historical Records of
Sea Surface Temperatures." Monthly Weather Review 109:2427-
2434.

Parry, M. L., and Carter, T. R. 1985. "The Effect of Climatic
Variations on Agricultural Risk." Climatic Change 7:95-110.

Pittock, A. B. 1983. "Recent Climatic Change in Australia: Implica-
tions for a C02-Warmed Earth." Climatic Change 5:321-340.

Puterbaugh, T. L.; Motha, R. P.; Strommen, N. D.; Decker, W. L.; and
LeDuc, S. K. 1983. "A Regional Comparison of Seasonal African
Rainfall Anomalies." In Proceedings of the Second International
Meeting on Statistical Climatology, pp. 8.5.1-8.5.8. Lisbon:
Institute Nacional de Meteorologia e Geofisica.











Raman, C. R. V., and Maliekal, J. A. 1985. "A 'Northern Oscillation'
Relating Northern Hemispheric Pressure Anomalies and the Indian
Summer Monsoon." Nature 314:430-432.

Rasmusson, E. M., and Wallace, J. M. 1983. "Meteorological Aspects
of the El Niio/Southern Oscillation." Science 222:1195-1202.

Rasool, S. I. 1984. "On Dynamics of Deserts and Climate," In The
Global Climate, pp. 107-120. Edited by J. T. Houghton.
Cambridge: Cambridge University Free Press.

Rauner, Yu. L. 1980. "The Synchronous Recurrence of Droughts in the
Grain Growing Regions of the Northern Hemisphere." Soviet
Geography 21:159-175.

S1985. "Clusters of Drought Years in the Grain Zone of the
U.S.S.R." Soviet Geography 26:73-90.

Salinger, M. J. 1979. "New Zealand Climate: The Temperature Record,
Historical Data, and Some Agricultural Implications." Climatic
Change 2:109-126.

Schuurmans, C. J. E. 1984. "Climate Variability and Its Time Changes
in European Countries, Based on Instrumental Observations." In
The Climate of Europe: Past, Present, and Future, pp. 55-101.
Edited by H. Flohn and R. Fantechi. Dordrecht: Reidel.

Schuurmans, C. J. E., and Coops, A. J. 1983. "Level and Interannual
Variability of Seasonal Mean Temperatures in Europe." In Pro-
ceedings of the Second International Meeting on Statistical
Climatology, 16.2.1-16.2.5. Lisbon: Instituto Nacional de
Meteorologia e Geofisica.

Thompson, L. M. 1975. "Weather Variability, Climatic Change, and
Grain Production." Science 188:535-541.

Todorov, A. V. 1985. "Sahel: The Changing Rainfall Regime and the
'Normals' Used for Its Assessment." Journal of Climatology and
Applied Meteorology 24:97-107.

Uchijima, Z., and Seino, H. Forthcoming. "Climate-Induced Latitu-
dinal Shift of Plant Growth Potential in Japan." In Assessment
of Climate Impacts on Agriculture. Volume 1: In High Latitude
Regions. Edited by M. L Parry, T. R. Carter, and N. T. Konijn.
Dordrecht: Reidel.

van Loon, H., and Williams, J. 1978. "The Association Between Mean
Temperature and Interannual Variability." Monthly Weather
Review 106:1012-1017.










Vinnikov, K. Ya.; Gruza, G. V.; Zakharov, V. F.; Kirillov, A. A.;
Kovyneva, N. P.; and Ran'kova, E. Ya. 1980. "Present-Day
Climatic Changes in the Northern Hemisphere." Meteorloqiva i
Gidrologia 6:5-17.

Warrick, R. A. 1980. "Drought in the Greater Plains: A Case Study
of Research on Climate and Society in the United States." In
Climatic Constraints and Human Activities, pp. 93-123. Edited
by J. Ausubel and A. K. Biswas. Oxford: Pergamon Press.

Weller, G.; Baker, D. J.; Gates, W. L.; MacCracken, M. C.; Manabe,
S.; and Vonder Haar, T. H. 1983. "Detection and Monitoring of
CO2-Induced Climatic Changes." In Changing Climate: Report of
the Carbon Dioxide Assessment Committee, National Research
Council, pp. 292-382. Washington, D.C.: National Academy
Press.

World Climate Data Program. 1985. "The Global Climate System: A
Critical Review of the Climate System During 1982-1984." WCDP
Climate System Monitoring Project. Geneva: World
Meteorological Organization.

Yoshino, M. M. Forthcoming. "The Impact of Climatic Variations on
Agriculture in Japan--Introduction: The Policy and Planning
Issue." In Assessment of Climate Impacts on Agriculture.
Volume 1: In High Latitude Regions. Edited by M. L. Parry, T.
R. Carter, and N. T. Konijn. Dordrecht: Reidel.












4

Genetic Aspects
of Yield Variability

John H. Holden


BREEDING FOR YIELD STABILITY

A characteristic feature of breeding programs in developed
countries has been the high priority accorded to selection for yield
under high-input systems. Crop varieties produced in this way have
been described as high-response varieties, and it has been a feature
of many of them that they have been particularly liable to show
variations in yield response to variations in the environment. Year-
to-year variations in soil nitrogen and soil moisture seem particu-
larly important environmental components of yield stability.
The awareness of genotypic variation in yield stability of crop
varieties comes from two principal sources, namely agricultural expe-
rience and experimental evidence. Finlay and Wilkinson (1963) devel-
oped a method for detecting and measuring differences between geno-
types in sensitivity to the environment. In the barley varieties
they analyzed, they found that high yielders were often unstable in
their expression of this character (high genotype x environment
[G x E] interaction) and that, conversely, low yielders were usually
more stable in yield; that is, they were less responsive to environ-
mental change (low G x E interaction). A third type of variety was
shown to occur having intermediate sensitivity and yield across a
range of environments.
Since the 1960s breeders have given increasing attention to the
development of varieties with high and consistent performance. Breed-
ers have two routes to this goal: (a) purposive breeding from par-
ents selected for their ability to give stable progeny, and (b) se-
lection among advanced lines of progenies not specifically bred for
stability, to estimate G x E interactions (in multisite trials) and
to detect those with the more stable yields. Both methods have been
used and there are now numerous reports in the literature of the pro-
duction of varieties with both improved yield and stability. There
are a number of studies of cereals: for maize -- Cross (1977),
Fakorede and Mock (1978), Francis and Kannenberg (1978), Lee et al.
(1983), Subandi (1979), Toderkan (1980), Toit et al. (1979); for rice
-- Kim et al. (1983), Mahadevappa et al. (1979), Mohanty and Roy
(1974); Morais et al. (1981), Shankare Gowda et al. (1973), Vergara
(1976); for wheat -- Gupta et al. (1977), Heiner (1976), Jatasra and
Paroda (1980), Saulescu et al. (1980); for sorghum -- Rao and Rao
(1978); for barley -- Centro Internacional de Mejoramiento de Maiz y
Trigo (1977); for oats -- Reinbergs (1977).









Thus the breeding of varieties having both enhanced yield and
stability is becoming commonplace and is probably feasible in all
crops. However, the priority accorded to yield stability, among the
other selection criteria in a breeding program, will depend on the
agricultural system for which the variety is intended. In advanced
systems, where the major variables of the crop environment are
largely controlled by the farmer through irrigation, fertilizers,
fungicides, and insecticides, stability achieved by genetic means can
be less important, and yield is frequently an overriding breeding
objective.


PHENOTYPIC ASPECTS OF STABILITY

Numerous attempts have been made to analyze and explain stabil-
ity in terms of physiological, anatomical-morphological, or develop-
mental features of the plant. Various mechanisms have been invoked,
such as higher and more efficient uptakes of major and trace elements
present at low levels in the soil (Mahadevappa et al. 1979) or of
soil moisture under conditions of water stress. In the case of some
maize hybrids, yield stability was attributed to their possession of
a short grain filling period, thereby reducing the time during which
grain development was susceptible to environmental stress (Francis
and Kannenberg 1978).
This diversity of interpretations of stability at the functional
level is a reflection of the fact that yield -- of seed, fruit, root,
or leaf -- is the result of the interaction of the diverse physiolo-
gical activities of the plant and that each of these can act as a
limiting factor on yield when stressed by a particular component of
the environment. Identifying the critical development features or
phases and breeding and selecting for desired expressions of these
characteristics can more effectively lead to genotypes that are
insensitive to environment than simply screening the final selections
from a breeding program for stability of performance in multisite
trials. But whatever the level of precision adopted in a breeding
program, the point to note is that variation in yield can be reduced
by genetic means.


BUFFERING IN POPULATIONS

The genotypically and phenotypically heterogeneous cereal
landraces presented a degree of diversity to the environment such as
the ensuing range of interactions served to buffer the response of
the population as a whole and probably conferred a degree of yield
stability from site to site and year to year.
One of the principal aims of plant breeding since the mid-
nineteenth century has been to reduce the variability in crops, both
inbreeders and outbreeders, and to increase the proportion of
individuals with the desired phenotypes. In this, breeders have been
very successful, raising both uniformity and yield and, in many
cases, crop quality, also. However, in the absence of conscious or
effective selection for yield stability, environmental sensitivity
seems to have been an undesirable consequence of these breeding aims.










There has been a long history of experimentation in cereals with
mixtures of pure lines to test the hypothesis that different compo-
nents of the mixture could exploit different niches or strata of the
crop environment and so raise yield levels above those of the
components grown as pure stands. The aims of this work have usually
been to raise yield levels per se rather than to improve yield
stability, and while some experiments failed to demonstrate any
advantage in either character (Lang et al. 1975), the general trend
seems to be toward positive but small advantages in yield for the
mixtures (Simmonds 1979). Recently, to profit from the buffering due
to heterogeneity, attempts have been made to construct mixtures from
high-yielding lines in order to improve yield stability.
An example of the successful use of population diversity to
increase yield stability concerns resistance to wind dispersed fungal
diseases. Two types of mixture have been proposed: (a) multiline
varieties constructed from a mixture of isogenic or near isogenic
lines and differing in alleles at a major gene locus, for example,
resistance to stem rust in wheat (Borlaug 1959); and (b) mixtures of
established varieties that differ in their resistance to physiologic
races of the pathogen (Wolfe 1977). Both methods have been shown to
delay the rate of development of an epidemic compared to that in
crops consisting of monocultures of the components of the mixtures,
to raise yields above the mean of the components grown singly, to
reduce variation in yield, and to slow the rate of adaptation of the
fungal population to the host resistance.
Despite the clear evidence from experiment of the benefits of
heterogeneity, it seems that the magnitude of these benefits has been
insufficient to offset the disadvantages to the management or
marketing of the crop, for mixtures have not found a significant
place in modern agriculture.

BUFFERING IN INDIVIDUALS

Heterozyqotes

Since heterozygosity of individuals is usually associated with
heterogeneity of populations in outbreeding species, both could
contribute to buffering, and experiments to demonstrate the signifi-
cance of heterozygosity must take this into account.
A comparison of six characters in maize inbreds and hybrids
showed that the inbreds had larger coefficients of variability and
were therefore less stable in phenotype (Shank and Adams 1960). This
type of result is usually attributed to the allelic diversity of the
hybrids, permitting greater flexibility of response to environmental
change and thereby providing greater possibility of maintaining
stability of phenotypic expression.

Homozyqotes

With regard to homozygotes, however, Shank and Adams's experi-
ment in maize also showed that the inbreds differed among themselves
in their stability, and therefore it is necessary to recognize that
attributes of genotypes other than heterozygosity are important when










considering breeding for stability in inbreeding crops. This view is
supported, of course, by much practical experience in wheat, oats,
barley, and rice, which demonstrates unequivocally that some pure
lines are more stable than others and that breeding and selection for
this character is a perfectly feasible objective.

IMPLICATIONS FOR GENETIC RESOURCES

Three kinds of genetic diversity influencing sensitivity to
environmental variation have been identified: in populations, in
heterozygotes, and between homozygotes. It is clear that each is
capable of being manipulated and exploited to produce varieties of
more stable yield. Since modern varieties are the result of a long
process of reduction of genetic diversity, the question may be asked
whether there is evidence that this progressive narrowing of the
genetic base of our crop plants -- desirable as it may be from many
points of view -- will seriously interfere with the selection of more
stable varieties of inbreeders and outbreeders. The answer appears
to be no, in that successes have been recorded in many crops where
the attempt has been made to improve stability. For this reason
breeders may be expected to continue to work with their adapted gene
pools, so long as they can provide genetic gains in yield and
stability, rather than risk disrupting adapted gene complexes by wide
outcrossing.
In the special cases of disease and drought stresses, which for
many crops are the principal causes of yield variation, there is
clearly a need for new sources of variation from outside the gene
pool of advanced cultivars. In these cases, breeders will continue
to explore the wider gene pool of the crop and its wild and weedy
ancestors for exploitable variation. It is in this area that we can
expect crop genetic resources to continue to contribute to both the
raising and stabilization of crop yields.
In conclusion, it should be said that while breeding has a
significant contribution to make in the stabilization of crop yields,
there are limits to what can be achieved in this way. The probabil-
ity of breeding varieties capable of giving high and stable yields
under widely fluctuating environments is low. It has to be remem-
bered that breeders rarely, if ever, select for one character in
isolation and usually try to hold simultaneously the desirable
expressions of several others. For example, it rapidly becomes
impossible to retain palatability and digestibility in forages or
grain quality in cereals when selecting for resistance to increas-
ingly low soil water.
Breeding, therefore, has a part to play in stabilizing yields,
but its contribution relative to complementary approaches will depend
on the crop, the nature of the environmental variables (drought,
diseases, or pests), and the agricultural system under which the crop
is grown.










REFERENCES

Borlaug, N. E. 1959. "The Use of Multibreed or Composite Varieties
to Control Airborne Epidemic Disease of Self-Pollinated Crop
Plants." In "Proceedings of the International Wheat Genetic
Symposium," pp. 12-26, mimeographed.

Centro Internacional de Mejoramiento de Mafz y Trigo. 1977.
"Barley." CIMMYT Review 1976. Mexico.

Cross, H. Z. 1977. "Inter-Relationships Among Yield Stability and
Yield Components in Early Maize." Crop Science 17:741-745.

Fakorede, M. A. B., and Mock, J. J. 1978. "Stability and Adaptation
Responses of Maize Variety Hybrids Developed by Recurrent
Selection for Grain Yields." Maydica 23:89-100.

Finlay, K. W., and Wilkinson, G. N. 1963. "The Analysis of Adapta-
tion in a Plant Breeding Program." Australian Journal of
Agricultural Research 14:742-754.

Francis, T. R., and Kannenberg, L. W. 1978. "Yield Stability Studies
in Short-Season Maize. 1. A Descriptive Method for Grouping
Genotypes." Canadian Journal of Plant Science 58:1029-1034.

Gupta, V. P.; Khehdra, A. S.; and Bains, K. S. 1977. "Concepts in
Stability Analysis." In Genetics and Wheat Improvement. Vol.
5: Genetics of Adaptability and Physioloqy, pp. 143-152. Edited
by A. K. Gupta. New Delhi: Oxford and IBA Publishers.

Heiner, R. E. 1976. "New Developments in Hard Spring Wheats The
Public Sector." Cereal Foods World 21:446.

Jatasra, D. S., and Paroda, R. S. 1980. "Phenotypic Adaptability of
Characters Related to Productivity in Wheat Cultivars."
Industrial Journal of Genetics and Plant Breeding 40:132-139.

Kim, B. H.; Kim, J. M.; and Lee, Y. S. 1983. "Studies on Variation
and Stability of Yield Among Years of Recommended Rice Varieties
in Gyeongnam Area." Research Reports, Office of Rural Develop-
ment 25:144-151. South Korea. (In Korean).

Lang, R. W.; Holmes, J. C.; Taylor, B. R.; and Waterson, H. A. 1975.
"The Performance of Barley Variety Mixtures." Experimental
Husbandry 28:53-59.

Lee, T. C.; Lu, H. S.; and Wan, H. 1983. "Studies on the Yield
Stability of Double Hybrids of Opaque-2 Maize." Journal of
Agricultural Research of China 32:219-227. (In Chinese).

Mahadevappa, M.; Ikehashi, H.; and Ponnamperuma, F. N. 1979. "The
Contribution of Varietal Tolerance for Problem Soils to Yield
Stability in Rice." IRRI Research Paper Series 43:1-15.










Mohanty, H. K., and Roy, A. 1974. "Breeding of Semi-Dwarf Rice
Varieties for High Yield and Yield Stability in Orissa." Crop
Improvement 1:32-35.

Morais, 0. P. de; Silva, J. C.; Vieira, C.; et al. 1981. "Adaptabil-
ity and Yield Stability of'Varieties and Lines of Irrigated Rice
(Orvza sativa L.) in Minas Gerais." Revista Ceres 28:150-159.
(In Portuguese).

Rao, S. A., and Rao, K. V. 1978. "Genotypic Stability of Sorghum
Varieties and Hybrids." Industrial Journal of Agricultural
Science 48:691-695.

Reinbergs, E. 1977. "Oxford Oats." Canadian Journal of Plant
Science 57:991-993.

Saulescu, N. N.; Popa, S.; and Pacurar, I. 1980. "New Romanian
Winter Wheat Varieties and Their Release for Cultivation." Cer.
si Plante Technice, Prod. Vegetala 32:3-12. (In Romanian).

Shank, D. B., and Adams, M. W. 1960. "Environment Variability Within
Inbred Lines and Single Crosses of Maize." Genetics 57:119-126.

Shankare Gowda, B. T.; Mahadevappa, M.; et al. 1973. "Yield Stabil-
ity of Rice Varieties in Uniform Varietal Trials." Industrial
Journal of Agricultural Science 43:449-451.

Simmonds, N. W. 1979. Principles of Crop Improvement, pp. 116-120.
London: Longman.

Subandi. 1979. "Yield Stability of Nine Early Maturing Varieties of
Corn." Contr. Central Res. Inst. Aqr. Boqor #53.

Toderkan, V. G. 1980. "Ecological Adaptability of Different Types of
Maize Hybrids in Respect to Grain Yield in Moldavia." Selek.
Genet. i Tekhnol. Vozdelivaniva Kukuruzv v Moldavii 1980:32-39.
(In Russian).

Toit, A. J. L. du; Eberhart, S. A.; et al. 1979. "Yield Stability of
Maize Cultivars." In Proceedings of the Third South African
Maize Breeding Symposium, 1978, pp. 35-39. Edited by J. G. du
Plessis. Pretoria: Republic of South Africa, Dept. of Agricul-
tural Technical Services.

Vergara, B. S. 1976. "Physiological and Morphological Adaptability
of Rice Varieties to Climate." In Procedures of the Symposium
on Climate and Rice, pp. 67-83. Los Bahos, Philippines:
International Rice Research Institute.

Wolfe, M. S. 1977. "Yield Stability in Barley Using Varietal
Mixtures and Disease Control." Cereal Research Commission
5:119-124.














5
Yield Variability
and the Transition
of the New Technology

H. K. Jain, M. Dagg, and T. A. Taylor


The new agricultural technology based on improved varieties and
the increased use of modern inputs has proved to be highly rewarding.
It has led to significant increases in production growth rates in
many developing countries. Tsutsui and Singh (1985) show that the
growth rate of cereal production and yield has been higher in the
Asia-Pacific region than in the rest of the world in recent years.
They especially draw attention to increased production of wheat,
rice, and coarse grains in these countries.
The effect of the new technology in many of these countries can
also be seen in other ways. Thus Bangladesh, which faced a desperate
food situation at the time of its independence in the early 1970s,
has recorded a major advance in the production of both wheat and rice
and could be well on its way to food self-sufficiency. Indonesia has
more than doubled its rice production in the last ten years.
Thailand has emerged as a major exporter of food, including not only
its traditional export, rice, but also maize, soybeans, and cassava.
Similarly, many Latin American countries have recorded major gains in
their food production. Africa as a continent remains an exception,
although some African countries (for example, Zimbabwe) have made
effective use of the new technology to increase their production of
maize and other crops.
Gains in food production should obviously contribute to an
increasing sense of food security and stability. There is, however,
no consensus on food stability even if it is generally accepted that
the overall food situation in many of these countries has improved.
Some social scientists have pointed out that food stability may, in
fact, have declined in the sense that there is greater variability in
the production and yield of some of the major food crops since the
introduction of the new high-yield technology.
Many of these authors specifically refer to the situation in
India. The link between agricultural growth and variability of
agricultural output first received attention from Indian scientists
like Sen (1967) and Rao (1975). Rao drew attention to the fact that

*This is a summary of Workshop Paper 16. We are grateful to
Philip Pardey for advice with regard to statistical analysis and to
Bob Solinger for computational work.











since variability in yields tends to be far greater than that in
area, productivity based growth has contributed to greater vari-
ability, as suggested also by Barker et al. (1981). Summarizing the
observations of these and other authors, Hazell (1984) has posed the
question whether the yields of crops grown with the new technologies
may be more sensitive to weather and disease. Further, because they
require more input, their yields may be more sensitive to year-to-
year variability in input use arising from frequent price changes or
from supply restrictions.
The concept of food security incorporates in it elements of
stability, so that a country's agriculture may become a dependable
source of food supplies year after year. In traditional agriculture
this has seldom been possible. It would appear from the studies
mentioned above that the new agricultural technology developed in
recent years may also create instability. Obviously, this raises
some basic issues, and it is useful at this stage to review some of
the data that have led to these conclusions.

FOODGRAIN PRODUCTION IN INDIA


Soon after gaining political independence in the 1950s, India
organized programs of agricultural development based essentially on
crop varieties that do not require intensive use of inputs like
chemical fertilizers. In the 1960s India made important policy
decisions to transform its traditional agriculture. A major instru-
ment of the new policy was the high-yielding varieties program, a
program that has continued to grow and that now covers many crops and
areas. India thus provides two fairly well-defined periods, one
associated with traditional technology and one associated with modern
technology.

Previous Studies

A number of authors have examined the effect of the new technol-
ogy in terms of gains in production and yield and expansion of area
(for example, Jain and Singh 1985). Some of them have also studied
its effect on the stability of production and productivity. The
first comprehensive analysis of this kind was carried out by Mehra
(1981), who analyzed data for two periods: 1949/50 to 1964/65 and
1967/68 to 1977/78. Thus Mehra had 16 years of crop production data
for the traditional period and 11 years for the modern period (the
agricultural year in India extends from June to May). She excluded
from her analysis the two crop years of 1965/66 and 1966/67 on the
grounds that these were extreme drought years and could create
possible distortions.
Mehra's main finding is that both the standard deviation and the
coefficient of variation increased for production and yield for most
of the foodgrains in the second period as measured by deviations from
trend. She further observed that the difference in the variability
of foodgrain and nonfoodgrain crops widened during the second period,
indicating that foodgrains fluctuated more in recent years.











These studies have been extended more recently by Hazell (1984),
who reached essentially similar conclusions. Hazell concludes that
the coefficient of variation of production increased during the
second period for all cereal crops except minor millet. Thus crops
like rice, wheat, grain sorghum, maize, pearl millet, and barley
showed increased yield variability during the second period, which
saw the advent of the new technology.

Extending the Variability Analysis

India's high-yielding varieties program has come a long way
since 1977/78, the last year encompassed in the two studies discussed
above. It is now possible to analyze data over a longer span of 35
years, ending with the crop season 1983/84. The 35-year span has
been divided into two periods: a period of 18 years (1949/50 to
1966/67) before the high-yield period, and a period of 17 years
(1967/68 to 1983/84) of high yield. The longer time span should make
for greater reliability in the analysis of the data.
More important, it is no longer necessary to leave out any of
the crop years. Droughts are an integral part of Indian agriculture
and its most important source of instability. Mehra and Hazell were
probably justified in leaving out the two drought years of 1965/66
and 1966/67 from their pre-high-yield period considering the fact
that the high-yield period was only 11 years. However, Indian
agriculture since the early 1970s has encountered several serious
droughts, including those of 1972/73 (characterized as very poor for
agricultural production), 1973/74 (poor), 1974/75 (very poor),
1976/77 (poor), 1979/80 (very poor), and the next three years (all
poor) (see Fertilizer Association of India 1984). Besides, the high-
yield period saw the destabilizing effect of the oil crises of 1973
and 1979. Considering these factors, it seems more justified not to
leave out any of the years from the two periods. It may well be
that, on balance, more adverse conditions have been encountered
during the second period than the first, but this must be ignored for
the purpose of the present analysis until a more satisfactory
quantitative measure of such adversity can be devised.
Table 5.1 summarizes the results of this analysis on India's
important cereal crops: rice, wheat, maize, sorghum, and pearl
millet. These five cereal crops have received the largest attention
in terms of genetic improvement and form the main component of the
high-yielding varieties program. Year-to-year variability in
production, yield, and area was computed for each of the two periods
in the form of standard deviations and coefficients of variation (cv)
calculated around linear trend lines, which were estimated separately
for each of the two periods.
The analysis shows that while the standard deviation of total
cereal production increased between the two periods, relative varia-
bility, as measured by the cv, declined by nearly 17 percent. Rice
and wheat, the two most important foodgrains of India, and grain
sorghum also show a decline in their cvs between the two periods.
However, maize and pearl millet show greater cvs in the second
period. Yield variability follows a similar pattern, and total
cereals, rice, wheat, and sorghum all show smaller cvs for yield
during the second period.









Table 5.1--Production and yield of cereals in India, 1949/50 to 1966/67 and 1967/68 to 1983/84




Production Yield per hectare Area
1949/50 to 1967/68 to Percent 1949/50 to 1967/68 to Percent 1949/50 to 1967/68 to Percent
Crop 1966/67 1983/84 Change 1966/67 1983/84 Change 1966/67 1983/84 Change


(103 tons) (kilograms per hectare) (103 hectares)
Mean
Total cereals 60,080.72 103,353.18 72.02 677.61 1,008.06 48.77 88,123.61 102,257.88 16.04
Rice 29,287.72 45,833.29 56.49 882.11 1,181.00 33.88 32,987.67 38,680.76 17.26
Wheat 9,321.22 29,105.47 212.25 768.00 1,424.59 85.49 11,987.72 20,061.71 67.35
Maize 3,502.33 6,335.35 80.89 837.22 1,085.59 29.67 4,095.89 5,831.71 42.38
Sorghum 8,092.28 10,131.47 25.20 462.06 607.88 31.56 17,422.56 16,728.06 -3.99
Pearl millet 3,545.56 5,403.00 52.39 319.17 456.12 42.91 11,047.72 11,803.88 6.84

Standard deviation
Total cereals 5,062.79 7,241.26 43.03 44.70 57.12 27.79 2,088.88 1,687.45 -19.22
Rice 3,101.60 4,156.81 34.02 82.19 86.18 4.85 679.35 836.35 23.11
Wheat 872.53 2,434.70 179.04 55.90 84.76 51.63 745.94 785.15 5.26
Maize 255.91 715.19 179.47 57.58 115.55 100.68 121.76 136.21 11.87
Sorghum 978.92 1,165.69 19.08 46.50 59.09 27.08 634.75 759.29 19.62
Pearl millet 476.04 1,371.86 188.18 32.54 96.08 195.27 694.60 731.69 5.34

Coefficient of variation (%)
Total cereals 8.43 7.01 -16.84 6.06 5.67 -14.09 2.37 1.65 -30.38
Rice 10.59 9.06 -14.35 9.32 7.30 -21.67 2.06 2.16 4.85
Wheat 9.36 8.37 -10.58 7.28 5.95 -18.27 6.22 3.91 -37.14
Maize 7.31 11.29 54.45 6.88 10.64 54.65 2.97 2.34 21.21
Sorghum 12.10 11.51 -4.88 10.06 9.72 -3.38 3.64 4.54 24.73
Pearl millet 13.43 25.39 89.05 10.20 21.06 106.47 6.29 6.20 -1.43










When contrasted with the results of earlier studies, the
analysis suggests that the introduction of the new agricultural
technology in India is now reaching a stage where the increased
production of some cereals is combined with greater relative stabil-
ity of production. There are exceptions, and these deserve some
consideration. Pearl millet, for example, seems more variable in its
production following the advent of the high-yielding varieties. This
has a relatively simple explanation. Pearl millet in India provides
a classic example of the genetic vulnerability brought about by
increased genetic uniformity. The successful high-yielding varieties
program in this crop was based on four single-cross hybrids, all of
which had a common cytoplasmic male sterile parent (developed at
Tifton, Georgia). This parental line was responsible for the
vulnerability of all four hybrids to downy mildew, which resulted in
a sharp decline in production and yield in the early 1970s. It is
only in the past five years or so that the genetic diversity of the
male sterile lines has been sufficiently increased to increase the
production of pearl millet.
Maize, which also shows an increase in its cv between the two
periods, presents a different situation. It is one of the most
demanding crops in terms of its need for agronomic management and its
sensitivity to climatic conditions. For this reason, maize is now
being shifted from the kharif (rainy season) to the rabi (winter
season), to which it appears to be better adapted. As a native of
Central America, maize has never been at home in India during the
highly volatile monsoon season.

THE TRANSITION FROM TRADITIONAL TO MODERN AGRICULTURE

Most developed countries practicing modern agriculture do not
experience the kind of instability in their agricultural production
so common in developing countries in transition from traditional to
modern farming systems. In a changing agriculture, many factors can
be sources of variability in the early transition years. Some of
these factors are considered here.

Aqronomic Sensitivity

The new agricultural technology is based on genotype-environ-
mental interactions. The agronomic environment based on the use of
costly inputs like chemical fertilizer is particularly important for
the expression of the full genetic potential of the new crop varie-
ties. The introduction of this kind of technology in a developing
country, with its diverse groups of farmers and socioeconomic milieu,
is a completely different process from its gradual evolution in the
developed countries. In the latter countries, the improved agricul-
tural technology evolved over a longer time and in association with
the development of an industrial and service infrastructure, creating
in its wake a wide variety of modern farm inputs and vast purchasing
power.
Obviously, new technology in a developing country is adopted at
different times by farmers, and there is a time lag before it
permeates large sections of the farming community. Enlightened










governments in developing countries could help to bridge this time
gap by creating institutional mechanisms to help the small and the
resource-poor farmers. This is now happening in many developing
countries. Even so, one must expect that the first effect of the
introduction of a high-yield technology in a traditional society will
be considerable variability in crop production and yields.

Response to Environment

The climatic factor in tropical and subtropical environments is
potentially an important source of variability for the new production
technology. With their high-yield potentials, the new varieties ob-
viously have more to lose in stress environments, such as those
resulting from failure of rains. It is important, however, to
recognize the number of counteracting conditions.
First, the high-yielding varieties have been generally recom-
mended for irrigated conditions and for those rainfed lands where the
moisture stress is not too great. Agricultural scientists have yet
to evolve a highly effective production technology for the drier
environments. Second, the high-yielding varieties, even under
nonirrigated conditions, usually receive better agronomic management
than the traditional varieties. A common recommendation is that
farmers apply some chemical fertilizers even under rainfed condi-
tions. The response to fertilizer application in these lands, which
are not only thirsty but hungry, is often very marked. Third,
evidence suggests that some high-yielding varieties are better
buffered against climatic variability than others. Some CIMMYT wheat
varieties have been successfully grown in many countries under
diverse conditions. Similarly, some IRRI rice varieties have been
grown in large areas of Southeast Asia, South Asia, and other parts
of the world. These varieties were bred for wide adaptation through
photoinsensitivity, but they also seem to have an improved buffering
mechanism. Some sorghum hybrids developed in the Indian program give
high yields in years of both normal and poor rainfall. These
varieties appear to show homeostatic properties, which may be a
function of their heterozygosity.

Disease and Pest Epidemics

Some of the early high-yielding varieties of wheat and rice
carrying the Norin 10 and the De-Gee-Woo Gen dwarfing genes, respec-
tively, were found to be highly susceptible to diseases or pests.
Their replacements, developed at the international agricultural
research centers and in some of the national programs, generally show
a greater degree of genetic resistance against pests and pathogens.
However, these high-yielding varieties continue to be a potential
source of production instability. A more basic approach to the
problem is needed.
The broad adaptation characteristics of some of the high-
yielding varieties of wheat and rice developed in recent years have
served a most important purpose. It has been relatively easy to
organize seed multiplication and to bring very large areas in many
countries under the high-yielding varieties program. At the same
time, it would be a mistake to believe that future agriculture in the










developing or even the developed countries could be organized around
a few broadly adapted varieties of this kind. They bring about too
much genetic uniformity, which is not good for production stability.
There is a time and place for such varieties; basically they
help to buy time for solutions of a more lasting nature. These
solutions must be found in diversifying the genetic base of crop
varieties rather than in narrowing it, as has happened in the past 15
to 20 years. Already, in the last 15 years there have been several
serious pest and disease epidemics, all of which can be traced to the
genetic uniformity brought about by the high-yielding varieties. In
Southeast Asia and South Asia in the 1970s, brown planthoppers
attacked many of the dwarf rice varieties developed by the Interna-
tional Rice Research Institute. Sources of resistance to various
biotypes of brown planthopper were discovered in some of the existing
landraces and traditional varieties grown in countries like India.
In 1970, southern corn lead blight broke out in the United States,
appearing first in Florida and subsequently covering most of the corn
belt in the north, with serious side effects on yields. The epidemic
was related to a common genetic base in the male sterile parent of
many of the recommended hybrids, which became a source of suscep-
tibility to Helminthosporium mavdis. The third example is the downy
mildew disease epidemic, which attacked the newly released hybrids of
pearl millet in many parts of India in the early 1970s.
The answer to the emerging problem of genetic uniformity lies in
erecting genetic barriers against the spread of disease and pests.
These barriers must be built around a large number of genetically
diverse varieties, strategically placed in different regions of a
country. Jain (1985) describes this approach as the development of a
multilineal complex of varieties, deriving their resistance from
different sources, and each recommended for a different and limited
area. He refers to more than 30 wheat varieties recommended in
recent years for different parts of India by the scientists of the
Indian Wheat Program. The more important question, of course, is how
does a country build up a complex of genetically diverse varieties of
this kind. The present nature of the relationships between the
international agricultural research centers and national agricultural
programs obviously does not favor such diversification. The pattern
that has emerged in the last 15 years is of the international centers
making many crosses and distributing the resulting segregating
material or advanced breeding lines to a number of national programs
for local selections.
It is relevant to ask whether the development and distribution
of germplasm evolved by the international centers will remain valid.
With the ongoing efforts to strengthen research at the country level,
national breeders should increasingly be able to make crosses of
their own. Then, if national programs undertake their own hybridi-
zation (using parental germplasm available locally and from the
international centers and other national programs), a wider gene pool
will be created for evolving new varieties. Solutions of this kind
must receive attention as the CGIAR System acquires maturity and new
relationships are established between the centers and the national
programs. Centers like IRRI are beginning to move in this direction.
The problem of genetic diversification for the stability of
future agriculture is so fundamental that it must receive urgent











attention both in the national and international programs. Hazell
(1984) refers to a very definite shift toward more highly correlated
yields of important cereal crops among states within India and among
states in the United States, drawing attention to the widespread
cultivation of new varieties with a common genetic base. While we
may take some comfort from the fact that the new agricultural
technology has spread so widely, be it maize hybrids in the United
States or wheat and rice varieties in India, there is no getting away
from the fact of a vastly reduced genetic base for our future
agriculture. The problem is made worse by the patenting of new crop
varieties evolved mostly by private breeders in the western coun-
tries. The commercial incentive is to evolve those varieties whose
seeds can be sold most widely. The legislation on the rights of
private breeders, whose seeds have been introduced into many devel-
oped countries, has serious implications for world agriculture. This
legislation has not received the attention it deserves.

THE TRANSITION TOWARD GREATER STABILITY

The stability of agricultural production as a general propo-
sition appeals to most people. However, it can hardly be a major
objective in an agriculture in an accelerated state of transition
from traditional to modern. In the developed countries, this transi-
tion occurred as a long-term, evolutionary process, and its effects
on production were less disruptive than in the developing countries.
Many of the latter that are now rapidly trying to catch up with the
agriculture of the developed countries are obviously passing through
a phase of greater disruption, which produces considerable vari-
ability in its wake. The variability is the greater because massive
investments in modern farm inputs, which the new technology requires,
cannot be mobilized quickly -- these countries simply do not have the
short-term economic capacity to provide them. For this reason, many
developing countries have organized phased programs of production,
defining annual targets on the basis of planned coverage with the
high-yielding varieties. It is clear that agriculture in many of
these countries will continue to be in a state of flux, as the high-
yielding varieties program is extended to larger areas and to new
groups of farmers.
It does not follow that the new technology will always be a
source of instability. As developing countries continue to improve
their management support for agriculture by creating new institu-
tional mechanisms and infrastructure, the sources of variability
considered above will be reduced. In some countries, this has
already begun to happen. The results from the Indian analysis
presented here indicate that 17 years after the advent of the high-
yielding varieties program a measure of stability is beginning to
emerge.
The coefficient of variation of production has declined for most
crops in the United States in recent years (Hazell 1984). This
observed decline is consistent with the proposition that, as a
country's agriculture becomes more fully modernized, the sources of
variability decline. But the contradictory behavior of maize in the
United States remains of interest. A possible explanation is that










the improved maize technology in the United States has continued to
make rapid strides since the development of the first double-cross
hybrids in the early 1940s. The average yield of maize in Iowa
increased by as much as 20 quintals (2 metric tons) per hectare since
the release of the first hybrids (Duvick 1977). More generally, in
North America maize yields have doubled during the 38-year period
from 1941 to 1979 (Stoskopf 1981, p. 40). Much of this gain in yield
has come through continued selection for a higher harvest index.
Also, while genes played a key role in improving the harvest index of
wheat and rice, progress in maize has extended over a longer period,
with continued selection based on polygenic variability.
Technological innovations that have a major effect on production
obviously become an important source of variability. They should,
however, lead to a new equilibrium in which a higher level of
production is combined with greater relative stability. The time
needed to achieve this new equilibrium is a function of the adoption
rate of the new technology by the farming community. In the devel-
oped countries the transition is typically quite rapid. In the
developing countries it extends over a longer period, largely as a
function both of slow adoption of the recommended agronomic manage-
ment practices and of relatively ineffective price supports. Farmers
will not invest in a technology that is cost intensive unless they
have some assurance of remunerative prices for their produce.

NEW TECHNOLOGY FOR STRESS ENVIRONMENTS

It is clear that improved varieties alone cannot solve the
problems of stress environments, where the heritability of quanti-
tative traits, such as yield, is low. Real stability in the agricul-
ture of developing countries will come only when scientists begin to
address the problems of moisture stress, fertility stress, and stress
arising from pests and pathogens. Perhaps the time has come for the
national programs as well as the international centers to shift from
a purely genetic approach to one that includes research on soils,
water management, and other factors of production. Crop environment
can be enhanced through improved moisture conservation practices,
better soil health, and improved pest control measures. The next 20
years should see agricultural scientists evolving such research
priorities.










REFERENCES

Barker, R.; Gabler, E. C.; and Winkelmann, D. 1981. "Long Term
Consequences of Technology Change on Crop Yield Stability." In
Food Security for Developing Countries, pp. 53-78. Edited by
Alberto Valdds. Boulder, Colo.: Westview Press.

Duvick, D. N. 1977. "Genetic Rates and Gains in Hybrid Maize Yields
During the Past 40 Years." Maydica 22:187-196.

Fertilizer Association of India. 1984. "FAI Annual Review of
Fertilizer Consumption and Production 1983-84." Fertilizer News
29(8):77-140.

Hazell, Peter B. R. 1984. "Sources of Increased Instability in
Indian and U.S. Cereal Production." American Journal of
Agriculture Economics 66:302-311.

Jain, H. K. 1985. "Agriculture of Tomorrow Greater Productivity,
Efficiency, and Diversity." In Biotechnology in International
Agricultural Research. Manila: International Rice Research
Institute.

Jain, H. K., and Singh, D. 1984. "Impact of the New Agricultural
Technology." Agricultural Situation in India: 623-627.

Mehra, Shakuntala. 1981. Instability in Indian Agriculture in the
Context of the New Technology. Research Report 25. Washington,
D.C.: International Food Policy Research Institute.

Rao, C. H. H. 1975. Technological Change and Distribution of Gains
in Indian Agriculture. Delhi: Macmillan.

Sen, S. R. 1967. "Growth and Instability in Indian Agriculture," pp.
1-31. Address to the 20th Conference of the Indian Society of
Agricultural Statistics, January 10-12 (mimeographed).

Stoskopf, N. C. 1981. Understanding Crop Production. Reston, Va.:
Reston Publishing Company.

Tsutsui, H., and Singh, R. B. 1985. "Green Revolution in the Asia-
Pacific In Retrospect and Prospect." Paper prepared for the
International Symposium on World Food Problems, March 27-29.
Ministry of Agriculture, Forestry and Fisheries, Yokohama
(mimeographed).














6 Yield Variability
and Income, Consumption,
and Food Security

David E. Sahn and Joachim von Braun

Variability in world foodgrain production has increased. In
particular, there is evidence of an increase in interregional and
intercrop production variations during the past two decades (Chapter
2). This chapter discusses the effects of increased production
variability (IPV) on food security at the national and household
levels. More specifically, three questions are addressed here:
To what extent does IPV result in increased variability in food
consumption?
Which low-income groups are adversely affected by IPV, and how?
What policy measures could cope with the adverse effects of IPV,
in particular for the poor?
This chapter is concerned specifically with fluctuations in
production and the resulting transitory changes in prices and
incomes. This focus does not suggest that chronic undernutrition
related to persistent deficiency in food consumption is a less
heinous problem. Indeed, increasing food availability, along with
increasing demand for labor and wages, are corollaries to any
agricultural development strategy (Mellor 1976).
In answering the questions posed above, our point of departure
is that malnutrition is closely linked with poverty, and that
poverty, to some extent, is episodic in nature. Households that fall
within the category of extreme poverty one year may well fall outside
it the next year (Srinivasan 1985; Scott 1980), and villages affected
by natural or man-caused disasters in one agricultural cycle may
rebound in the following harvest cycle. By the same token, regions
and countries that might dramatically reduce hunger and poverty at
one time can quickly revert to deficiencies in basic needs. As
stated by Mellor and Desai (1985), "the temporal variations in
poverty include ... substantial intermediate-term undulations that
can cause the number of people in absolute poverty to vary by 50
percent or more."
The causes of these fluctuations in measured poverty differ from
one country to another and from one circumstance to another. The
proportion of the poor in a population is a function of many complex
relationships among exogenous events (like price shocks, deterio-
rating terms of trade), domestic policy changes (like increased price
of tradeables vis-a-vis nontradeables due to a devaluation), stochas-
tic weather-induced events (like a drought), existing technology, and
the country's resource endowment. Our primary concern is with
weather-induced fluctuations in yields, and thus, in cereal produc-
tion.











We begin by examining the effects of instability on market
aggregates. Thereafter, these effects are linked with the food
economy of households. Our hypothesis is as follows: a variety of
factors may increase output fluctuations, which translate into
greater variability in income or prices of food and nonfood commodi-
ties; and income-price variability results in greater fluctuations in
food consumption, representing nutritional risk to the household.


MARKET-LEVEL EFFECTS OF IPV

The process by which IPV affects commodity prices is of particu-
lar importance. Figure 6.1 traces the global and national market
effects and indicates links between production variability and
prices.
In the context of the whole world food economy, changes in
levels of food production translate into price changes. Therefore an
increase in production variability must be of concern, whether it be
in developed or developing countries (Mellor and Johnston 1984).
Given the high share of grain trade in developed countries, fluctua-
tions in export supply or import demand will also exert substantial
effects on world prices, and therefore on food security in developing
countries. Variations in trade volumes and stocks and the extent to
which exporters and importers responded to world-market price
fluctuations in adjusting the volumes of trade were major determi-
nants of the high variability in world prices experienced during the
past two decades (Siamwalla and Valdds 1980). In fact Table 6.1
exemplifies how short-run price variability in the world grain
markets occurs despite small global production fluctuations.


Figure 6.1. Effects of market-level instability on consumption


Nonagricultural
employment
cultural


Consumption

Nutrition










Table 6.1--World cereal production, trade, and price of
wheat during the "food crises"

Year Cereal Production Cereal Trade Wheat Price
(million tons) (million tons) (US$/ton)

1970/71 1,104 110 74
1971/72 1,194 110 70
1972/73 1,161 134 100
1973/74 1,266 142 203
1974/75 1,213 136 204
1975/76 1,239 152 187

Source: U.S. Department of Agriculture, 1982.


As more countries exploit trade opportunities to stabilize
domestic prices, further volatility in international grain markets
may result (Koester 1984; Johnson 1975). As Johnson remarks, "There
has been little recognition of the extent to which one nation or
region achieves stability at the expense of instability to others"
(p. 823). The fact that governments are interested in domestic
stability and are far less concerned with price stability abroad
reduces the likelihood that sovereign nations will work together to
stabilize global supply and prices. At the individual country level,
the effects of IPV on domestic prices and supply can be examined in
the context of two distinct international grain trade regimes.
The first case is the open economy that participates actively in
grain markets. Domestic prices will not be affected directly by IPV
at the national level unless other factors, like foreign exchange
constraints, limit participation in grain markets. Instead, the
effects of IPV will be felt in the coffers of the treasury, which
must have foreign exchange to purchase commodities on international
markets. It is not simply the domestic IPV that will cause the
foreign exchange bill to fluctuate, because the variability in world
prices will have the same effect even if domestic supply and demand
remain constant. However, the work by Valdds and Konandreas (1981)
indicates that variability in the import bill is primarily due to
volume rather than price. This poses a special problem among the
poorer countries, where foreign exchange is clearly a constraint to
food imports.
While foreign exchange constraints may partially explain
fluctuations in imports and supply, most countries choose not to
adhere to free-trade principles for a variety of reasons. Thus
fluctuations in prices become part of a complex web of not only
international prices but also domestic food production, demand, and
price policy. Government decisions regarding imports and storage
further compound the problem of predicting price movements in the
face of greater instability.
The second case is the closed economy trade regime. Many
landlocked African countries resemble the closed economy case that
does not participate extensively in grain trade because of prohibi-




University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs