M Smale1

International Maize and Wheat Improvement Center, Mexico, D.F., Mexico

In this chapter, several indicators that social and biological scientists have used to describe varietal and genetic diversity in farm fields are applied to data for bread wheats grown in developing countries. The discussion is based on patterns that can be observed in farm fields rather than those developed from population genetics or molecular measurements. This emphasis reflects both our interest in factors that shape farmers' choice of varieties and the difficulty of assembling genetic or molecular data on such a large scale.

Spatial Diversity

Empirically, there is an inverse relationship between area sown to modern bread wheats in developing countries and the numbers of distinct varieties2 grown per million hectares. South Asia, the Southern Cone, and West Asia, which contain the largest areas planted to bread wheats in the developing world, have the lowest number of crosses per million hectares sown (Table 5.1).3

The percentage of area planted to the top five crosses ranges between 43% in the Southern Cone and 71% for Mexico and Central America and the Andean region. West Asia also has a relatively low concentration of area under leading crosses, which may in part reflect the importance of traditional bread wheat varieties in that region.

While these percentages appear high, it is important to recognize that the concentration of wheat area under modern cultivars is probably less today than in earlier decades of this century for major wheat-producing regions of both the industrialized and developing world (see data in Maclndoe and Brown, 1968; Reitz, 1979; Lupton, 1992; and Thomas, 1995). Since the beginning of

© CAB INTERNATIONAL 1998. Agricultural Values of Plant Genetic Resources

(eds R.E. Evenson, D. Gollin and V. Santaniello) 85

Table 5.1. Indicators of spatial diversity among bread wheats grown in the developing world in 1990.

Sub-Saharan Africa

North Africa

West Asia

South Asia

Mexico and Guatemala

Andean Southern Developing region Cone world3

Number of modern cultivars 39 28 51 64 42 27

Number of crosses from which cultivars are selected 30 23 47 51 36 25

Area in modern cultivars

Modern cultivars as percent of area in bread wheats 86 83 53 93 94 87

Crosses Mha-1 modern cultivars 45 13 5 2 41 145

Top five crosses as percent of area in modern cultivars 64 62 48 59 71 71

93 6

49.8

82 5

36.4

Source: calculated from CIMMYT Wheat Pedigree Management System and data from CIMMYT Wheat Impacts Survey, summarized In Byerlee and Moya (1993).

a Regional numbers of cultivars and crosses do not total to Developing world because the same cultivar or cross may be grown in more than one region. The Developing world category excludes China.

the green revolution the concentration of planted area among leading bread wheats has also changed. The number of cultivars released in developing countries that were derived from the CIMMYT variety Veery is at least twice that of the cultivars derived from the II8156 (Mexipak) cross, but the area planted to all of them is small compared with the area once sown to II8156 alone (Byerlee and Moya, 1993).

Estimates suggest, for example, that the area planted to a single cultivar was high in the Indian Punjab in the late 1950s prior to the green revolution: a tall bread cultivar called C591 may have covered most of the irrigated area and some of the rainfed area. Semi-dwarf wheat varieties generally replaced the tall, modern varieties (such as C591) that were released by the Indian national breeding programme from the early 1900s. Since the late 1960s, the percentage of area in leading cultivars has fluctuated, but if any long-term trend is observable since independence in 1947, it has not been upward. In the Pakistani Punjab, for the shorter period from 1978, the percent of wheat area in the dominant and top five cultivars has been only slightly lower than in the Indian Punjab, and the pattern over time also appears to be cyclical. The cyclical pattern reflects the replacement of older varieties with newer releases, or the decline and rise in popularity among leading cultivars. For both the Indian and Pakistani Punjabs, however, the concentration of planted area among modern wheats still appears to be relatively high (Smale, 1995).

Temporal Diversity

The average age of crosses in farmers' fields, weighted by area planted, is a measure of the temporal diversity of cultivars, or diversity in time (Duvick, 1984). The average age of crosses grown in farmers' fields, weighted by area planted, ranges between about 8 years for Mexico and Guatemala and about 15 years for the North Africa region (Table 5.2). The rapid rate of change among crosses grown in Mexico and Guatemala reflects in part the rapid rate of change in the virulence of rusts in that zone.

As a point of comparison with these figures, Brennan and Byerlee (1991) have estimated the weighted average age of cultivars for a number of specific wheat-producing zones of the industrialized and developing world over several decades. Among the zones they studied, the Yaqui Valley of Mexico had the highest temporal diversity (a weighted average age of only 3.1 years over the 1972-1986 period) and the Punjab of Pakistan had the lowest (a weighted average age of about 11 years over the 1978-1986 period). The Punjab of India had a weighted average age of 5.3 over the 1970-1986 period. Brennan and Byerlee found that the commercialized wheat-producing zones in Brazil, Argentina, the US, Australia, New Zealand and the Netherlands had an average age of 7-10 years. By contrast, Canada has a relatively low level of temporal diversity for an industrialized major wheat producer, ranging from about 10 to 13 years over the past 20 years (see Thomas, 1995).

Table 5.2. Temporal diversity among bread wheats grown in the developing world in 1990.

Weighted average Region age of crosses3

Sub-Saharan Africa 11.3

West Asia 10.6

North Africa 14.7

South Asia 12.8

Southern Cone 9.2

Andean region 13.7

Mexico and Guatemala 8.0

Source: calculated from CIMMYT Wheat Pedigree Management System and data from CIMMYT Wheat Impacts Survey, summarized in Byerlee and Moya (1993). a Weights are percentage area planted to cultivars derived from same crosses.

The weighted average age of crosses has implications for resistance to both known and unknown pathogens. Using data from a number of countries, Kilpatrick (1975) estimated an overall average of 5-6 years' cultivar longevity for leaf and stripe rusts, when resistance is monogenic. Using that estimate alone, the rate of turnover among crosses would be less than desirable for all of the regions of the developing world in 1990. But the longevity of cultivars in terms of rust resistance is very environment specific, and the socially optimal period for cultivar replacement is a function of many economic and biological factors, of which resistance to pathogens is only one (Heisey and Brennan, 1991).4 Generally, there is a need for higher rates of varietal turnover in more favourable production environments, because the conditions that are conducive to high productivity are also conducive to the development of disease.

Diversity Indicators Based on Genealogical Characteristics Latent Diversity

As calculated from the coefficients of parentage, the latent diversity of the top ten cultivars planted in the developing world in 1990 appears to be fairly high, although the average coefficient of diversity varies by geographical region (Table 5.3).5 Among regions of the developing world, the average coefficients of diversity are significantly higher among the top ten lines grown in West Asia, and the Southern Cone of Latin America, than in South Asia, Mexico and Guatemala.

As a point of comparison, the same indicators are presented for three of the

Table 5.3. Latent diversity of the top ten bread wheat crosses grown in regions of the developing world and in selected industrialized nations in 1990.

Average coefficient of diversity weighted Minimum Maximum

Average

by

pairwise

pairwise

Genea-

coefficient

cultivated

coefficient

coefficient

logical

Region/country

of diversity

area

of diversity

of diversity

distance

Developing world

0.78

0.70

0.43

0.98

8.18

Sub-Saharan Africa

0.79

0.77

0.28

0.99

8.29

North Africa

0.79

0.73

0.57

1.00

7.12

West Asia

0.84a

0.80

0.67

0.99

8.11

South Asia

0.72b

0.63

0.35

0.96

7.70

Mexico and Guatemala

0.69b

0.63

0.57

0.88

5.80

Andean region

0.80

0.72

0.41

0.99

7.89

Southern Cone

0.82a

0.80

0.69

1.00

7.78

Selected major industrialized

bread wheat producers

Canada (spring wheats)

0.48c

0.22

0.01

0.80

4.71

Australia (spring wheats)

0.74b

0.72

0.30

0.98

8.63

US (hard red spring wheats) 0.84a

0.79

0.53

1.00

8.71

Source: calculated from CIMMYT Wheat Pedigree Management System and data from CIMMYT Wheat Impacts Survey, summarized in Byerlee and Moya (1993). Notes: coefficient of diversity = 1 — coefficient of parentage. Genealogical distance measured as total branch length of dendrogram constructed from Ward's cluster analysis of coefficients of diversity (see Weitzman, 1992). Average coefficients of diversity with different letters are statistically different, using a non-parametric test. China is excluded from the Developing world category.

four major bread wheat producers of the industrialized world, also for the top ten crosses, and for spring wheats. In Australia average and weighted coefficients are almost equal which implies that the top ten crosses are distributed equally as a percent of national area. Each state of Australia has a different set of leading cultivars, and the environment is more heterogeneous than in the US or Canada. The top ten lines grown in Canada are statistically less diverse than the top ten in any of the developing or industrialized regions considered. The minimum diversity among pairs of crosses is also near zero in Canada, while the maximum diversity is lower than for the other industrialized producers and the developing regions.

An estimate of genealogical distance suggested by the work of Weitzman (1992)6 is also shown in Table 5.3. In comparison with a simple average of the coefficients of diversity for each group of ten cultivars, this indicator represents the sum of the distances of each cultivar from all other cultivars in the set based on the pairwise coefficient of diversity as a measure of distance. Once again, Canada's leading spring wheats appear to be markedly less diverse than those of either the other major industrialized wheat producers or the developing regions. Mexican wheats, grown in a small relatively homogeneous production environment, also appear to be considerably less diverse - a result that is not as clear with a simple average of coefficients of diversity. The top ten bread wheats of West Asia appear among the most diverse for developing regions.

The data demonstrate clearly how the factors affecting the spatial distribution of planted area among cultivars can influence latent diversity. For all developing country regions, weighting by area planted to the cultivars in 1990 reduces the average coefficients of diversity, although not by a very large magnitude. In Canada, weighting by percent of area halves an already low average coefficient of diversity.

The difference between the weighted and unweighted measures of diversity crudely reflects the effects of factors related to varietal adoption, such as seed distribution systems. Farmers will choose to grow the variety that is most attractive to them (in terms of profits or other measures of economic value), but the range of their choice is often limited by the few seed types that are locally available. Policy factors that affect the rate of release of cultivars, and the policy, institutional and behavioural factors that determine the varieties that farmers plant and their rate of varietal replacement, are principal determinants of wheat diversity in farmers' fields. These are generally outside the influence of plant breeders and are those in need of more careful study by social scientists.

In the Indian Punjab, both the average and weighted average coefficients of diversity of the leading cultivars grown in farmers' fields have increased significantly over time since the late 1970s. The movement around the trend line is greater for the weighted average coefficient of diversity, reflecting changes in area planted among leading cultivars. The upward trend is greater for the weighted average than the average coefficient of diversity, and no trend is perceptible over the period for the average coefficient of diversity among all wheats released by the national programme.7

Landrace Use

In a sample of 800 wheats released by breeding programmes in developing countries over the past 30 years, the average number of different landraces per pedigree has continued to increase. This is an important finding. Although we can expect the frequency of landrace use to increase over time as pedigrees grow longer, it is not necessarily true that the number of different landraces also increases. For example, in the early part of this century, plant breeders in many regions of the world used a few landraces from the former Soviet Union, Europe and India extensively (see information summarized in Smale and McBride, 1996). When advanced materials were later exchanged among breeding programmes, the frequency of many of these landraces in the pedigrees of wheat releases increased, but not necessarily the number of different landraces.

Among wheat breeding programmes in developing countries, wild relatives and landraces are entered less frequently in crossing blocks than other germplasm materials - but breeders do use them (in roughly 14% of all crosses), and particularly when they make crosses for biotic resistance, tolerance to abiotic stress or quality (Table 5.4). Other results reported in Rejesus et al. (1998) suggest that turnover of wild relatives and landraces in wheat breeders' crossing blocks is also lower than for other types of materials.

Turnover of landraces in crossing blocks and the representation of land-races among active parental stocks probably reflect closely the way in which they are used and breeders' perceptions of expected returns from their investment. To determine which landraces 'combine' well with modern germplasm and transmit the trait(s) of interest requires several breeding cycles and several hundreds of crosses. Verifying that a desirable trait has been transferred to, and is stable in, the progeny requires further testing. Transferring desirable genes without also transferring deleterious genes represents a further challenge. As Harlan (1992, p. 154) has stated, the plant breeder 'wants the genes not the linkages'.

Landraces are infrequently the direct parents of leading wheat varieties grown in farmers' fields. Gerek 79, a major Turkish wheat variety and one of the top ten wheat varieties grown in the developing world in 1990, is an exception - one of its parents is a Turkish landrace. When new materials are brought into a wheat breeder's programme, most are advanced materials with long pedigrees. Many have similar genealogical backgrounds to materials previously used by the breeder. Some have landrace ancestors that are not found in materials previously used by the breeder. Very few are landraces that have never been used before in wheat breeding. (See pedigrees shown in Smale and McBride, 1996.)

Table 5.4. Type of parent materials used in crossing, by breeding goal, wheat programmes in developing countries in 1994.

Parent material

Percent of crosses, by goal

Yield

Biotic resistance

Abiotic resistance

Quality

All

Wild relatives and landraces

4.7

15.4

22.1

20.9

14.4

Advanced materials

69.0

54.6

51.2

55.1

59.2

CIMMYT International

Nurseries

23.2

26.6

22.3

20.4

23.0

Others

3.1

3.4

4.4

3.6

3.4

Total

100

100

100

100

100

Source: survey conducted for CIMMYT World Wheat Facts and Trends (1996). Note: Includes responses from 70 wheat breeders. Advanced materials Included released varieties and advanced lines from respondent's programme or other national programmes. Others category includes materials from other, sub-national programmes in the respondent's nation, or materials from other international nurseries.

Source: survey conducted for CIMMYT World Wheat Facts and Trends (1996). Note: Includes responses from 70 wheat breeders. Advanced materials Included released varieties and advanced lines from respondent's programme or other national programmes. Others category includes materials from other, sub-national programmes in the respondent's nation, or materials from other international nurseries.

Yield Stability

The yield stability of wheat in the developing world is compared over four decades in Table 5.5. For every region, variation was greater in the decade preceding 1965 (the year that marks the early phase of the green revolution) than in the most recent decade. In regions where the largest proportion of wheat area is planted to modern wheats (South Asia, Mexico/Guatemala, and the Southern Cone of South America) the variation in wheat yields has declined since 1965. In West Asia and North Africa, where modern wheats cover a smaller proportion of area, yield stability has not worsened over the past three decades. Only in the Andean region and sub-Saharan Africa, two regions with very small wheat areas and with distinctive growing conditions, does the variation in wheat yields appear to have increased since 1965. In both of these regions, however, the overall level of variation is quite low.

As explained previously, because most of the year-to-year variation in aggregate yields is caused by differences in weather, use of irrigation, and pathogens, the factors explaining the largest proportion of variation in aggregate yields are probably associated less with plant stature or genotype than with input supply and pricing policy. The balance of general evidence concerning the relationship between mean yields and yield variance in farmers' fields over time suggests that yield stability has increased even as mean yields have increased, from the 1950s through the 1980s, across the world, in major wheat-producing countries of the developing world, and in India (Anderson and Hazell, 1989; Singh and Byerlee, 1990). In particular, Singh and Byerlee (1990) showed that technological variables such as the level of adoption of high-yielding varieties and levels of fertilizer use had no effect on differences in wheat yield stability across countries.

Table 5.5. Yield stability of all wheats grown from 1955 to 1994 in the developing world.

Coefficient yield of variation adjusted for trend (%)

Table 5.5. Yield stability of all wheats grown from 1955 to 1994 in the developing world.

Coefficient yield of variation adjusted for trend (%)

Sub-

Mexico

Saharan

North

West

South

and

Andean

Southern

Africa

Africa

Asia

Asia

Guatemala

region

Cone

1955-

-1964

10.8

13.4

8.7

6.5

12.3

9.8

12.9

1965-

-1974

4.3

10.3

8.0

9.1

7.9

2.4

8.1

1975-

-1984

7.1

12.1

4.0

3.0

5.6

5.6

12.2

1985-

-1994

8.8

11.0

7.5

4.0

5.5

4.8

5.0

Source: constructed from FAO yield data using the Cuddy-Della Valle index (Cuddy and Della Valle, 1978). Note: China is excluded.

Source: constructed from FAO yield data using the Cuddy-Della Valle index (Cuddy and Della Valle, 1978). Note: China is excluded.

Conclusions

The findings summarized here suggest that the percentage of area planted to leading cultivars in major bread wheat-producing zones of the developing world and industrialized world is high, although less so than in earlier periods of this century, when the first products of scientific plant breeding were widely distributed across Europe, Australia, North America and India. Evidence from India also indicates that the concentration of area among the top cultivars is lower now than in the green revolution period. These findings are not inconsistent, however, with the generally held view that the ancient patterns of genetic variation in farmers' varieties have been replaced during the past 200 years by patterns based on modern plant breeding. Further, broad perspectives such as those presented here cannot capture the effects of important changes in the micro-centres of diversity. Such changes must be studied in detail on a case-by-case basis.

Notes

1. Numerous CIMMYT scientists have contributed to the work summarized here.

2. By 'modern', we denote both improved tall and semi-dwarf varieties - or all varieties with known pedigrees that are the products of a scientific breeding programme. We contrast the number of distinct varieties with the number of cultivars, because the same variety can be released under several names. This happens, for example, when national programmes re-release a variety obtained from an international research institution or another national programme under a new name. Many lines can also be selected from one cross. The most precise level of detail for identifying a variety is given by a combination of cross and selection information. In these tables, and in the reported calculations of coefficients of parentage, selections from one cross have been treated as the same cross and called a 'variety'. This slightly overstates the similarity of parentage and understates the diversity.

3. The People's Republic of China is the largest national producer of wheat in the developing world, but the CIMMYT Wheat Impacts Survey (summarized in Byerlee and Moya, 1993) contains wheat cultivar data from only one of its regions. CIMMYT Economics and Wheat Programs are currently engaged in improving the coverage and quality of data on wheat releases and pedigree information for China. Some preliminary findings are reported in Yang and Smale (1996).

4. Recall that historically, some single cultivars dominated the wheat areas of industrialized countries for decades, such as Wilhelmina and Juliana in the Netherlands, the Vilmorin crosses in France, and Federation in Australia (MacIndoe and Brown, 1968; Lupton, 1992). In more recent years, in Canada, the fact that Neepawa occupied over 50% of wheat area for years has contributed to a relatively low measure of temporal diversity (see Thomas, 1995).

5. Souza etal., (1994) have defined (1-COP) as an indicator of latent genetic diversity. In wheat, the coefficient of parentage measures the probability that two cultivars are identical-by-descent for a character (observable or unobservable) that varies genetically and is not expressed as a result of intensive selection by plant breeders.

6. The sum of the branch lengths of the dendrogram constructed from Ward's cluster analysis of pairwise, ultrametric distances. Here, the pairwise distance measures are coefficients of diversity. Any pairwise distance measure that satisfies ultrametric properties can be used as the basis of analysis.

7. Data reported in Smale (1995) show a constant or slightly positive trend among wheats released by the national programme over the last 80 years.

References

Anderson, J.R. and Hazell, P.B.R. (1989) Variability in Grain Yields: Implications for Agricultural Research and Policy in Developing Countries. Johns Hopkins University Press, Baltimore, Maryland.

Brennan, J.P. and Byerlee, D. (1991) The rate of crop varietal replacement on farms: measures and empirical results for wheat. Plant Varieties and Seeds 4, 99-106.

Byerlee, D. and Moya, P. (1993) Impacts of International Wheat Breeding Research in the Developing World, 1966-90. CIMMYT, Mexico.

Cuddy, J.O.A. and Della Valle, P.A. (1978) Measuring instability of time series data. Oxford Bulletin of Economics and Statistics 40, 79-85.

Duvick, D.N. (1984) Genetic diversity in major farm crops on the farm and in reserve. Economic Botany 38(2), 161-178.

Harlan, J.R. (1992) Crops and Man. American Society of Agronomy, Inc., and Crop Science Society of America, Inc., Madison, Wisconsin.

Heisey, P.W. and Brennan, J.P. (1991) An analytical model of farmers' demand for replacement seed. American Journal of Agricultural Economics 73, 1044-1052.

Kilpatrick, R.A. (1975) New Wheat Cultivars and Longevity ofRust Resistance, 1971-5. ARS-NE-4. Agricultural Research Service, US Department of Agriculture, Beltsville, Maryland.

Lupton, F.G.H. (1992) Wheat varieties cultivated in Europe. In: Lupton, F.G.H. (ed.) Agroecologicai Atlas of Cereal Growing in Europe, Vol. 4, Changes in Varietal Distribution of Cereals in Central and Western Europe. Wageningen University, Wageningen, the Netherlands.

Maclndoe, S.L. and Brown, C.W. (1968) Wheat Breeding and Varieties in Australia. Science Bulletin No. 76, 3rd Edn. New South Wales Department of Agriculture, Sydney.

Reitz, L.P. (1979) 60 years of wheat cultivar history in the United States. Annual Wheat Newsletter 25, 12-17.

Rejesus, R., Van Ginkel, M. and Smale, M. (1998). Wheat breeders' perspectives on genetic diversity and germplasm use: findings from an international survey. Plant Varieties and Seeds 9, 129-147.

Singh, A.J. and Byerlee, D. (1990) Relative variability in wheat yields across countries and over time. Journal of Agricultural Economics 41(1), 21-32.

Smale, M. (1995) Ongoing Research at CIMMYT: Understanding Wheat Genetic Diversity and International Flows ofGenetic Resources. CIMMYT World Wheat Facts and Trends Supplement, Part I. International Maize and Wheat Improvement Center, Mexico.

Smale, M. and McBride, T. (1996) Understanding Global Trends in the Use of Wheat Diversity and International Flows of Wheat Genetic Resources. CIMMYT 1995/96 World Wheat Facts and Trends, CIMMYT, Mexico.

Souza, E., Fox, P.N., Byerlee, D. and Skovmand, B. (1994) Spring wheat diversity in irrigated areas of two developing countries. Crop Science 34, 774-783.

Thomas, N. (1995) Use of IARC germplasm in Canadian crop breeding programmes: spillovers to Canada front the CGIAR. Spring bread wheats. Draft prepared for CIDA.

Weitzman, M.L. (1992) On diversity. Quarterly Journal of Economics, 107, 363-404.

Yang, N. and Smale, M. (1996) Indicators of wheat genetic diversity and germplasm use in the People's Republic of China. Draft, Natural Resources Group Working Paper, International Maize and Wheat Improvement Center (CIMMYT), Mexico.

Was this article helpful?

0 0

Post a comment