HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Related articles in Crop Science Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (9) Citing Articles via Google Scholar Google Scholar Articles by Lillemo, M. Articles by Rajaram, S. Search for Related Content PubMed Articles by Lillemo, M. Articles by Rajaram, S. Agricola Articles by Lillemo, M. Articles by Rajaram, S. Related Collections Wheat Plant and Environment Interactions

CROP BREEDING, GENETICS & CYTOLOGY

Associations among International CIMMYT Bread Wheat Yield Testing Locations in High Rainfall Areas and Their Implications for Wheat Breeding

^a Dep. of Plant and Environmental Sciences, Agricultural Univ. of Norway, P.O. Box 5003, N-1432 Ås, Norway
^b Wheat Program, International Maize and Wheat Improvement Center (CIMMYT), Apdo. Postal 6-641, 06600 Mexico DF, Mexico

^* Corresponding author (morten.lillemo{at}nlh.no).

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
A good understanding of how the target environments for a breeding program differentiate the germplasm with respect to yield is crucial and allows plant breeders to better target their germplasm. To determine the relationships among high rainfall yield testing locations, yield data from 8 yr of CIMMYT's High Rainfall Wheat Yield Trial (HRWYT) were analyzed by shifted multiplicative model (SHMM) and incremental sum of squares (ISS) classification analyses to group sites within and across years. In the cumulative cluster analysis, about half of the sites clustered into a group characterized by increasing temperature toward maturity. The SHMM analysis identified several sites with high overall association with other sites around the world, and which can be considered as good predictors of global yield performance within the high rainfall megaenvironment. These are autumn-sown locations, which fall into the biggest group of the cumulative cluster analysis with increasing temperature during the growing season. On the other hand, remarkably low associations with global yield ranking were shown for Sta. Catalina (Ecuador) and CIMMYT's primary high rainfall yield-testing location at Toluca (Mexico), which in contrast experience decreasing temperatures toward maturity. Although excellent sites for disease screening, this analysis shows that they do not associate well with the world's high rainfall wheat growing areas for yield.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
THE BREAD WHEAT BREEDING PROGRAM of the International Maize and Wheat Improvement Center (CIMMYT) breeds spring wheat lines for all major wheat growing areas in the developing world (Rajaram and van Ginkel, 2001). On a yearly basis, about 2000 advanced lines are distributed to collaborators in more than 60 countries.

To better target germplasm adapted to different environmental conditions, various agroecological zones or megaenvironments have been defined which represent similar biotic and abiotic stresses, cropping system requirements, and consumer preferences (Rajaram et al., 1994; Rajaram and van Ginkel, 2001). One of these major megaenvironments is characterized by average cropping season rainfall above 500 mm. Representative regions include high rainfall sites in West Asia and North Africa (WANA), the central highlands of eastern and central Africa, the southern cone and Andean highlands of South America, and the highlands of central Mexico. The total area in developing countries exceeds 8 million hectares.

Spring bread wheat (Triticum aestivum L.) germplasm targeted to this high rainfall mega-environment is developed by shuttling segregating generations between two contrasting environments in Mexico; a fully irrigated, high-yield potential site located near Ciudad Obregon in north-western Mexico and a high-rainfall, high disease incidence site at Toluca in the Central Mexican Highlands (Braun et al., 1996). Since 1992, advanced bread wheat lines targeted to this megaenvironment have been distributed globally through the HRWYT, following yield testing at Toluca.

The selection of superior genotypes is generally complicated by the presence of genotype x environment (G x E) interactions, whereby the relative yields of genotypes vary across different environments. A useful way to deal with G x E interactions in a breeding program is to characterize the crop environments in terms of the way they influence the relative performance of genotypes. Pattern analysis, as defined by Williams (1976), is the joint use of classification and ordination methods. Many models have been proposed for extraction and interpretation of grouping patterns (see DeLacy et al. (1996a), for review). The shifted multiplicative model (SHMM) is a clustering method that identifies subsets of locations with negligible crossover interaction, i.e., locations that give the same relative ranking of genotypes (Cornelius et al., 1992; Crossa et al., 1993). However, it requires balanced data sets where the same locations and genotypes are repeated over years. Another, highly recommended classification method for multienvironment trials is the incremental sum of squares (ISS) or Ward's strategy (Ward, 1963), which searches to minimize the incremental sum of squares within each group. ISS has a strong clustering property which tends to minimize the growth of large groups and produces groups of relatively even size (DeLacy et al., 1996a). In multienvironmental trials, where the composition of genotypes change from year to year, but many of the same locations are repeated, a cumulative analysis can be performed by averaging the environmental distance matrices over years and then eliminating rows and columns with empty cells (DeLacy et al., 1996b).

DeLacy et al. (1994) used pattern analysis based on ISS to examine associations among environments over time for the International Spring Wheat Yield Nursery (ISWYN), targeted to all the worlds spring bread wheat growing areas. Recently, we have used both SHMM and ISS classification methods to analyze the relationships among international testing sites for the Semi-Arid Wheat Yield Trial (SAWYT) (Trethowan et al., 2001) and the Elite Spring Wheat Yield Trial (ESWYT) targeted to irrigated environments (Trethowan et al., 2003).

Our objective in the present study was to evaluate the associations among test locations where the HRWYT nursery was grown and to attempt to explain the underlying causes of these associations.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
Locations and Genotypes
Genotypes yield-tested in HRWYTs 1 to 8 (1992–1999) were bred in Mexico for high-rainfall environments by shuttle breeding as described by Braun et al. (1996). Segregating materials were shuttled between the Centro de Investigaciones Agricolas del Noroeste (CIANO) (27°23'N and elevation 38 m above sea level) near Ciudad Obregon, a dry, fully irrigated site in northwestern Mexico, and CIMMYT's research station at Atizapan, Toluca, in the central Mexican Highlands (19°16'N and elevation 2640 m above sea level). Leaf rust [caused by Puccinia recondita Roberge ex Desmaz. f. sp. tritici (Eriks. & E. Henn.) D.M. Henderson] and stem rust (caused by P. graminis Pers.:Pers.) are the prevalent diseases at CIANO and stripe rust (caused by P. striiformis Westend.), Septoria blotch (caused by Septoria tritici Roberge in Desmaz.), leaf rust, Fusarium head blight (caused by Fusarium graminearum Schwabe), Barley yellow dwarf virus (BYDV), and intermittent water-logging are common at Toluca. The final yield trials for selecting germplasm to enter the HRWYT nursery were conducted across 2 to 3 yr at the research station in Toluca.

Each year the HRWYT nursery was assembled from seed increased under fungicide application at Mexicali, a disease free site located in northwestern Mexico and distributed globally upon request to international collaborators. Each trial consisted of between 30 and 50 entries and was planted on the basis of local agronomic practices. Two-replicate -lattice designs were used (Barreto et al., 1997). The composition of lines varied from year to year, representing newly developed germplasm, and a local check variety representing the best locally adapted germplasm was included at each site each year. The local check varied among locations and in some instances changed between years at the same location. Genotypes were considered as fixed effects and replicates and subblocks within replicates as random effects. Adjusted means were calculated and used in all subsequent SHMM and cumulative cluster analyzes to examine site clustering or grouping. A total of 187 location years or individual yield trials were used for the analysis and are listed in Table 1. Except for the cumulative cluster analysis, all other statistical analyses of the yield data were performed with SAS 8.1 (SAS, 1999).

SHMM Analysis
The SHMM clustering procedure (Crossa et al., 1993) was used to examine the associations among sites for each of the eight years of HRWYT and to identify groups of sites with reduced COI. The methods were the same as previously outlined in Trethowan et al. (2001)(and 2003). The SHMM model for the mean of the ith genotype (i = 1, 2, ..., g) in the jth site (j = 1, 2, ..., s) (_ij.) can be represented _ij. = ß + ^t_{k=1 k ik jk} + _ij. (Seyedsadr and Cornelius, 1992), where ß is the shift parameter; _k(_{1 2} ... _t) are the singular values (scale parameters) that allow imposition of orthonormality constraints on the singular vectors for genotypes, _1k,...,_gk and for sites, _1k,...,_sk, such that _i²_ik = _j²_jk and _iikik' = _jjkjk' = 0 for k k'; _ik and _jk, for k = 1,2,3,..., are called "primary," "secondary," "tertiary," ..., effects of the ith genotype and the jth site, respectively; _ij. is the residual error.

The distances for all possible pairs of sites were calculated, and dendrograms constructed using the complete linkage (farthest neighbor) clustering method. The third fusion level was selected as an arbitrary cut-off point to determine site clusters, and each site's associations with other sites were calculated as the number of times (in pair vise comparisons) it clustered together with other sites divided by the total number of possible clusters. A summary table was made by adding associations across years, using the procedure described by Trethowan et al. (2001)(and 2003).

Cumulative Cluster Analysis
The cumulative cluster analysis was based on the ISS clustering method recommended by DeLacy and Cooper (1990) and used by Abdalla et al. (1996) and Trethowan et al. (2001)(and 2003). Dissimilarities between sites were measured by squared Euclidean distance (SED), and distance matrices were calculated for each year for all sites with at least two years of data. An across-year distance matrix was constructed by averaging distances from each year where data was available for any site-by-site comparison. Since clustering algorithms require a complete distance matrix with values for all site-by-site comparisons, sites contributing empty cells were subsequently removed to end up with a complete distance matrix. The statistical software package SEQRET (DeLacy et al., 1998) was used for conducting the cumulative cluster analysis. The most representative site for each cluster in the resulting dendrogram was identified as that with the smallest sum of SED to other sites.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
SHMM Analysis
A summary of the dendrogram results for individual years of the SHMM cluster analysis was made to examine the association of various sites with different geographical regions (Table 2). The frequency at which a site clusters together with other sites in different geographical regions is expressed as a fraction of the total number of possible groupings. Because of the inherent uncertainty of associations based on only a few years of data, only sites with data from at least 4 yr are considered in the following discussion, but data from all locations was used to make the summary in Table 2.

View larger version (14K): [in this window] [in a new window]   Fig. 1. Monthly mean temperatures for the five last months of the growing season in Toluca and Sta. Catalina, compared to the temperature profile of three sites showing high association with global yield ranking.

Cumulative Cluster Analysis
Since most locations only planted a few of the eight HRWYTs, a complete distance matrix could only be made for 20 of the 46 locations that planted at least 2 yr of the nursery (Fig. 2) . On the basis of the average SED to other locations, the two most representative sites in the resulting dendrogram were the Iranian yield testing location at Bayecola and Bethlehem in South Africa.

View larger version (39K): [in this window] [in a new window]   Fig. 2. Dendrogram of the relationships among sites sown to the HRWYT in two or more years, constructed from cumulative cluster analysis. The most representative site for each cluster is indicated by a star (*), and the most representative site across all locations is indicated by two stars (**).

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
Generally, there are many external or environmental factors that influence the yield ranking of cultivars from site to site. The most common are latitude, altitude, planting date, cultural management, daylength, temperature, water availability, diseases, and specific abiotic stresses such as low pH.

Both the SHMM analysis and the cumulative cluster analysis point to a common environmental feature of sites with good predictability of global yield performance; they all have increasing temperature toward maturity. However, this study does not provide any data to explain whether there is any direct relationship between temperature profile and the ability to predict global yield performance. Different temperature profiles mostly reflect differences in planting dates, and are also driven by daylength variation, rainfall, and solar radiation. It is likely that this association could also reflect some other underlying characteristics of the sites. Although disease data is scarce for some of the locations with good predictive ability, those locations all have favorable conditions for leaf rust, which is globally among the most important biotic stress factors in wheat production.

Except for Marcos Juarez, which did not occur in the cumulative cluster analysis, all globally good predictors identified from the SHMM summary grouped into group two of the cumulative cluster analysis. The general similarity of these sites and their high association with global performance, as indicated from both analyses, make them good indicators for the identification of germplasm with broad adaptability. On the other hand, Toluca and Sta. Catalina, which were identified from the SHMM analysis as poor predictors of global yield performance, clustered into separate groups of the cumulative cluster analysis, which indicates the more specific adaptability required for these sites.

Sta. Catalina, located in the highlands of Ecuador, has a high incidence of stripe rust, and is characterized by a different, more virulent race composition (Broers and Danial, 1994). There are also indications that the soil at this site is infected with root lesion nematodes (Pratylenchus thornei Sher and Allen; Trethowan, pers. comm.). A similarly low association between Sta. Catalina and other international yield trial sites was also found in the analysis of the ESWYT nursery (Trethowan et al., 2003).

Toluca's low association with global yield performance is also in accordance with earlier findings (Braun et al., 1992). Apart from cooler conditions and shorter days during grain filling with occasional night frosts, Toluca experiences other extremes; about 850 mm of rain falls during the growing season and serious disease epidemics such as stripe rust, Septoria blotch, BYDV, Fusarium head scab, and late-arriving leaf rust occur. Soils are frequently water logged, and crop lodging and preharvest sprouting are constraints.

It is therefore not surprising that the very special conditions at these two locations resulted in a different yield ranking of lines compared with other global test sites which have more favorable conditions for wheat cultivation. Although such extreme locations can be excellent for disease screening (e.g., stripe rust in Sta. Catalina and stripe rust, Septoria blotch and Fusarium head scab in Toluca), it is clear from this analysis that their yield data is not relevant for identification of genotypes with global adaptation. Nevertheless, it is likely that the high rainfall germplasm has benefited from several generations of selection at Toluca, since this ensures photoperiod insensitivity (being part of the Cd. Obregon-Toluca breeding shuttle), and good resistance to the most important diseases and abiotic stresses for high rainfall areas around the world (Braun et al., 1992; Campuzano 1997; Rajaram and van Ginkel, 2001). Further work should aim to identify more representative yield testing locations in Mexico for the high rainfall breeding material.

	CONCLUSIONS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
This is the first study of the relationships among yield testing sites for the high-rainfall germplasm since the partition of the CIMMYT breeding material into megaenvironments and the establishment of the HRWYT nursery. This paper shows that within the high-rainfall megaenvironment, there are at least two subgroups. The largest subgroup consists of autumn- and/or early spring-planted sites characterized by increasing temperature as the crop approaches maturity and generally high association with global yield performance. The second, and smaller subgroup, consists of spring-planted sites with either stable or decreasing temperature in the later stages of development and low association with global yield performance.

The SHMM analysis identified sites that associated well with overall global yield ranking, thereby facilitating the identification of representative or key yield testing locations. The methods applied in this study are not only relevant to international wheat breeding but can be used to analyze any breeding program, regardless of crop species, as long as a sufficient number of yield trials are sown at representative locations throughout the target area.

	ACKNOWLEDGMENTS

 
This study was conducted during the first author's stay at CIMMYT in Mexico, which was made possible by a grant from the Research Council of Norway. The authors are thankful to Tom Payne for providing valuable information about the international yield testing locations, and to Jose Crossa for advise on the use of statistical methods.

Received for publication March 10, 2003.

	REFERENCES

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 

• Abdalla, O.S., J. Crossa, E. Autrique, and I.H. DeLacy. 1996. Relationships among international testing sites of spring durum wheat. Crop Sci. 36:33–40.[Abstract/Free Full Text]
• Barreto, H.J., G.O. Edmeades, S.C. Chapman, and J. Crossa. 1997. The alpha lattice design in plant breeding and agronomy: Generation and analysis. p. 544–551. In G.O. Edmeades et al. (ed.) Developing drought- and low N-tolerant maize. Mexico, D.F., Mexico.
• Braun, H.J., W.H. Pfeiffer, and W.G. Pollmer. 1992. Environments for selecting widely adapted spring wheat. Crop Sci. 32:1420–1427.[Abstract/Free Full Text]
• Braun, H.J., S. Rajaram, and M. van Ginkel. 1996. CIMMYT's approach to breeding for wide adaptation. Euphytica 92:175–183.
• Broers, L.H.M., and D.L. Danial. 1994. Stability of resistance to yellow rust in spring bread wheat: Field observations at three highland locations. p. 91–96. In D.L. Danial (ed.) Aspects of durable resistance in wheat to yellow rust. Ph.D. Thesis. Wageningen Agricultural University. Wageningen, the Netherlands.
• Campuzano, L.F. 1997. Selection systems for wheat (Triticum aestivum L.) in alternate enviroments and in situ. (In Spanish, with English abstract). Ph.D. Thesis. Colegio de Postgraduados, Montecillos, Mexico.
• Cornelius, P.L., M. Seyedsadr, and J. Crossa. 1992. Using the shifted multiplicative model to search for "separability" in crop cultivar trials. Theor. Appl. Genet. 84:161–172.[ISI]
• Crossa, J., P.L. Cornelius, M. Seyedsadr, and P. Byrne. 1993. A shifted multiplicative model cluster analysis for grouping environments without genotypic rank change. Theor. Appl. Genet. 85:577–586.[ISI]
• DeLacy, I.H., K.E. Basford, M. Cooper, J.K. Bull, and C.G. McLaren. 1996a. Analysis of multi-environment trials– an historical perspective. p. 39–124. In M. Cooper and G.L. Hammer (ed.) Plant adaptation and crop improvement. CAB International. Wallingford, UK.
• DeLacy, I.H., K.E. Basford, M. Cooper, and P.N. Fox. 1996b. Retrospective analysis of historical data sets from multi-environment trials–theoretical development. p. 243–267. In M. Cooper and G.L. Hammer (ed.) Plant adaptation and crop improvement. CAB International. Wallingford, UK.
• DeLacy, I.H., and M. Cooper. 1990. Pattern analysis for the analysis of regional variety trials. p. 301–334. In M.S. Kang (ed.) Genotype-by-environment interaction in plant breeding. Louisiana State Univ., Baton Rouge, LA, USA.
• DeLacy, I.H., M. Cooper, K.E. Basford, and P.N. Fox. 1998. The SEQRET package. Computer programs for retrospective pattern analysis. p. 39. In Plant improvement group research Report No 1. School of land and food, The University of Queensland, Brisbane, Australia.
• DeLacy, I.H., P.N. Fox, J.D. Corbett, J. Crossa, S. Rajaram, R.A. Fischer, and M. van Ginkel. 1994. Long-term association of locations for testing spring bread wheat. Euphytica 72:95–106.[ISI]
• Rajaram, S., and M. van Ginkel. 2001. Mexico, 50 years of International Wheat Breeding. p. 579–608 (chapter 22) In A.P. Bonjean and W.J. Angus (ed.) The world wheat book, A history of wheat breeding. Lavoisier Publishing, Paris.
• Rajaram, S., M. van Ginkel, and R.A. Fischer. 1994. CIMMYT's wheat breeding mega-environments (ME). p. 1101–1106 In Proceedings of the 8th International Wheat Genetics Symposium, Beijing, China.
• SAS. 1999. Version 8.1. SAS Institute Inc., Cary, NC, USA.
• Seyedsadr, M., and P.L. Cornelius. 1992. Shifted multiplicative model for nonadditive two-way tables. Commu. Statist. B. Simulation Comput. 21:807–822.
• Trethowan, R.M., J. Crossa, M. van Ginkel, and S. Rajaram. 2001. Relationships among bread wheat international yield testing locations in dry areas. Crop Sci. 41:1461–1469.[Abstract/Free Full Text]
• Trethowan, R.M., M. van Ginkel, K. Ammar, J. Crossa, B. Cukadar, S. Rajaram, and E. Hernandez. 2003. Relationships among twenty years of high yielding international bread wheat yield evaluation environments. Crop Sci. 43:1698–1711.[Abstract/Free Full Text]
• Ward, J.H. 1963. Hierarchical grouping to optimize an objective function. J. Am. Stat. Assoc. 58:236–244.[ISI]
• Williams, W.T. 1976. Pattern analysis in agricultural science. Elsevier, Amsterdam, the Netherlands.

Related articles in Crop Science:

THIS ISSUE IN CROP SCIENCE

Crop Science 2004 44: 1109-1112. [Full Text]  

This article has been cited by other articles:

				Y. Xu and J. H. Crouch Marker-Assisted Selection in Plant Breeding: From Publications to Practice Crop Sci., March 19, 2008; 48(2): 391 - 407. [Abstract] [Full Text] [PDF]

				A. Navabi, R.-C. Yang, J. Helm, and D. M. Spaner Can Spring Wheat-Growing Megaenvironments in the Northern Great Plains Be Dissected for Representative Locations or Niche-Adapted Genotypes? Crop Sci., March 27, 2006; 46(3): 1107 - 1116. [Abstract] [Full Text] [PDF]

				S. B. Blanche and G. O. Myers Identifying Discriminating Locations for Cultivar Selection in Louisiana Crop Sci., February 24, 2006; 46(2): 946 - 949. [Abstract] [Full Text] [PDF]

				M. Lillemo, M. van Ginkel, R. M. Trethowan, E. Hernandez, and J. Crossa Differential Adaptation of CIMMYT Bread Wheat to Global High Temperature Environments Crop Sci., October 27, 2005; 45(6): 2443 - 2453. [Abstract] [Full Text] [PDF]

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Related articles in Crop Science Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (9) Citing Articles via Google Scholar Google Scholar Articles by Lillemo, M. Articles by Rajaram, S. Search for Related Content PubMed Articles by Lillemo, M. Articles by Rajaram, S. Agricola Articles by Lillemo, M. Articles by Rajaram, S. Related Collections Wheat Plant and Environment Interactions