Crop Science Journal of Natural Resources and Life Sciences Education
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (16)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Trethowan, R. M.
Right arrow Articles by Hernandez, E.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Trethowan, R. M.
Right arrow Articles by Hernandez, E.
Agricola
Right arrow Articles by Trethowan, R. M.
Right arrow Articles by Hernandez, E.
Related Collections
Right arrow Crop Genetics
Right arrow Plant and Environment Interactions
Published in Crop Sci. 43:1698-1711 (2003).
© 2003 Crop Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA

CROP BREEDING, GENETICS & CYTOLOGY

Associations among Twenty Years of International Bread Wheat Yield Evaluation Environments

R. M. Trethowan*,a, M. van Ginkela, K. Ammara, J. Crossab, T. S. Paynea, B. Cukadarc, S. Rajarama and E. Hernandeza

a Wheat Program, International Maize and Wheat Improvement Center (CIMMYT) Apdo. Postal 6-641, 06600 Mexico DF, Mexico
b Biometrics and Statistics Unit, CIMMYT Apdo. Postal 6-641, 06600 Mexico DF, Mexico
c Monsanto Corporation, RR3 Box 331C, Harbstadt, IN, USA

* Corresponding author (r.trethowan{at}cgiar.org).


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSION
 REFERENCES
 
Understanding the way different environments differentiate cultivars for yield allows the plant breeder to optimize choice of parents, germplasm screening, yield testing, and resource use within the target region. To determine the associations among yield testing environments, wheat (Triticum aestivum L.) yield data from 963 replicated trials sown across a 20-yr period were analyzed by means of pattern analysis and the shifted multiplicative model (SHMM) to group sites within and across years. Pattern analysis identified four primary clusters of sites and four representative locations within these clusters were identified by squared Euclidean distances. Group-1 represented primarily Mediterranean and West Asian locations and South American sites. Group-2 was comprised of generally warmer sites in southern and eastern Asia. Group-3 comprised higher rainfall locations in South America and eastern Africa and Group-4 represented cooler sites in South America and West Asia. The respective key locations for each of the four groups were Sakha, Egypt; Quezaltenango, Guatemala; Londrina, Brazil; and Pirsabak, Pakistan. The four key sites were then used to examine site clusters within each year by SHMM. The sites at Pirsabak and Sakha associated best across all global wheat-growing regions where a combined total of 700 of 1117 (62%) possible clusters with other global wheat locations were realized. This compared with 52% for Quezaltenango and 38% for Londrina. Factors with a primary influence on site clustering were cropping season moisture availability and temperature. Genotype performance at Pirsabak and Sakha can be used to enhance genetic progress in a range of related wheat growing environments thereby improving the effectiveness of global wheat breeding.

Abbreviations: CIANO, Centro de Investigaciones Agricolas del Noroeste • CIMMYT, Centro Internacional de Mejoramiento de Maiz y Trigo (International Maize and Wheat Improvement Center) • COI, crossover interaction • ESWYT, Elite Spring Wheat Yield Trial • GEI, genotype x environment interaction • SHMM, shifted multiplicative model


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSION
 REFERENCES
 
THE BREAD WHEAT BREEDING PROGRAM at the International Maize and Wheat Improvement Center (CIMMYT) develops spring wheat germplasm for the wheat production environments of developing countries. Much of this germplasm is targeted to highly productive, high yield potential areas. These environments are diverse and range from fully irrigated conditions, typical of the Punjab in South Asia, to the rainfed areas of the eastern African highlands. During the past 55 yr, CIMMYT has developed this spring wheat germplasm by shuttling segregating generations between two contrasting environments in Mexico; one located near Ciudad Obregon in northwestern Mexico and the other at Toluca in the Central Mexican Highlands (Braun et al., 1996). CIMMYT bread wheat germplasm bred in Mexico by means of this shuttle is distributed globally in the Elite Spring Wheat Yield Trial (ESWYT).

In an attempt to focus the wheat breeding effort at CIMMYT, the major wheat production zones have been divided into zones of similar agro-ecological adaptation (Calhoun et al., 1994; Rajaram et al., 1994). Abdalla et al. (1997) examined durum (Triticum turgidum L.) wheat yield trials sown at 40 different locations in 1990-1991 and concluded that test sites associated on the basis of latitude and similar production constraints. Others have indicated the importance of identifying and targeting wheat germplasm to specific environments (DeLacy and Lawrence, 1988; DeLacy et al., 1994; Peterson and Pfeiffer, 1989). These studies examined associations among locations by estimating genotype x environment interactions.

Two models have been used to study the effects of GEI on site groupings without crossover interaction (COI). These are the shifted multiplicative model (SHMM) (Cornelius et al., 1992; Crossa et al., 1993, 1996) and the site regression model (SREG) (Crossa and Cornelius, 1997). The SHMM has also been used to cluster genotypes without genotypic rank change (Cornelius et al., 1993). These models allow clustering sites and genotypes into groups with smaller COI. The models have been successfully used to examine site clustering when all locations are sown to the same genotypes (Abdalla et al., 1997; Trethowan et al., 2001). If genotypes vary in comparisons among sites then a combination of classification and ordination analyses, commonly called pattern analysis, can be used (Fox et al., 1985; DeLacy and Lawrence, 1988; Peterson and Pfeiffer, 1989; Abdalla et al., 1996; Trethowan et al., 2001). DeLacy et al. (1994) and Trethowan et al. (2001) used pattern analysis to examine associations among environments across time for CIMMYT spring bread wheat nurseries. These analyses identified global regions or zones of similar adaptation for spring bread wheat germplasm. In the case of DeLacy et al. (1994), the trials spanned the time period 1964 through 1990 and represented germplasm from the International Spring Wheat Yield Nursery (ISWYN). The ISWYN was targeted to spring wheat-growing areas throughout the world, whereas the analysis of Trethowan et al. (2001) concentrated on dry environments only.

The objective of this research was to use yield data recorded on germplasm (ESWYTs 1–20, spanning 1979–1998) developed specifically for highly productive irrigated environments to determine the associations among international trial locations. Representative or key locations within groups of sites that differentiate wheat genotypes in a similar way can then be determined. Plant breeders can use genotype performance at these key sites to improve parental choice and the subsequent evaluation of progeny targeted to the wider production environment.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 CONCLUSION
 REFERENCES
 
Locations and Genotypes
Genotypes yield tested in ESWYT trials spanned the period 1979 through 1998 and were bred in Mexico, by a team led by Dr. S. Rajaram, for highly productive irrigated environments using shuttle breeding as described by Braun et al. (1996). Segregating materials were shuttled between the Centro de Investigaciones Agricolas del Noroeste (CIANO) (27°20'N and elevation 38 m above sea level), an arid-irrigated site in northwestern Mexico, and CIMMYT's high rainfall research station at Toluca in the central Mexican Highlands (19°16'N and elevation 2640 m above sea level).

Each ESWYT, distributed globally on request each year, was assembled from seed increased at the same site in northwestern Mexico. Trial seed was treated with Vitavax, packaged, and distributed each year from Mexico. A two-replicate, randomized complete-block design was generated for each ESWYT before 1990. After this date, two-replicate {alpha}-lattice trial designs (Barreto et al., 1997) were used. Trials were sown locally under conventional agronomic practices and consisted of between 30 and 50 entries. Yield data from each trial were analyzed using SAS (SAS, 1988). Genotypes were considered to be fixed effects and replicates and subblocks within replicates as random effects. Adjusted means were calculated and used in all subsequent SHMM and pattern analyses to examine site clustering or grouping. Sites returning incomplete data or aberrant values were removed, leaving a total of 963 site-years or individual yield trials representing 161 different geographical sites. Details of the sites used in the analyses are recorded in Table 1. Although the ESWYT entries were developed for irrigated conditions, many collaborators also grew them under rainfed conditions.


View this table:
[in this window]
[in a new window]
  		Table 1. Summary of locations used in the analysis of ESWYT trials number 1 to 20 spanning the period 1979–1998, their latitude, altitude, and frequency of occurrence.

 
Site Clustering and Construction of Dendrograms
Pattern Analysis
The pattern analysis used was that recommended by DeLacy and Cooper (1990) and used by Abdalla et al. (1996) and Trethowan et al. (2001). Only those sites sown to two or more ESWYTs were included in the analysis. Following the analysis outlined in Trethowan et al. (2001), dissimilarity between sites within the same year and across different years was measured as squared Euclidean distance (SED). The incremental sum of squares and the agglomerative hierarchical strategy procedure were used for site classification and the SED used as the dissimilarity measure. Pattern analysis uses the site effect to adjust for variability among sites. The resulting data centered by site was used to compute Euclidean distances.

SHMM Analysis for Clustering Sites into Groups without Crossover Interaction
To examine the associations among global yield testing locations, the SHMM clustering procedure for grouping sites without COI (Crossa et al., 1993) was applied to each of the 20 yr of ESWYT. The SHMM model for the mean of the ith cultivar (i = 1,2,...,g) in the jth environment (j = 1,2,...,e), ij, is as follows:

where ß is the shift parameter; {lambda}k ({lambda}1 >= {lambda}2 >= ... >={lambda}t) are singular values of the kth multiplicative term that allow the imposition of orthonormality constraints on the singular vectors for cultivars, {alpha}ik = ({alpha}ik,...{alpha}gk), and sites, {gamma}jk = ({gamma}1k...,{gamma}ek), such that {sum}i{alpha}2ik = {sum}j{gamma}2jk = 1 and {sum}i{alpha}ik{alpha}ik' = {sum}j{gamma}jk{gamma}jk' = 0 for k != k'; ij. is the residual error associated with ij..

The SREG model used by Crossa and Cornelius (1997) for clustering sites without COI is ij. = µj + {sum}tk = 1{lambda}k{alpha}ik{gamma}jk + ij. where µj is the jth site mean.

The SHMM clustering of sites finds subsets of sites with reduced COI by defining the distance between two sites as the residual sum of squares, RSS(SHMM1), after least squares fitting of SHMM with one multiplicative term (SHMM1). Once the pairwise distances have been obtained, a complete linkage cluster analysis is computed and a dendrogram generated. For a pair of sites, SHMM1 can be reparameterized to SREG1, which has one multiplicative term. Consequently, the unconstrained distance for a pair of sites is RSS(SHMM1) = RSS(SREG1).

Dendrograms were constructed and clusters of sites with reduced COI were found. Key sites within each cluster of the pattern analysis were chosen using the sum of the squared Euclidean distances taken from the dissimilarity matrix for each site versus all other sites within each cluster or group obtained from the pattern analysis. The site with the smallest sum of squared Euclidean distances within each cluster became a key site, or the best representative site within the cluster. Dendrograms were summarized by determination of the number of times the key location within each cluster (evident at the third fusion level of the pattern analysis) clustered with other sites and regions. The result was expressed as the total number of realized clusters/the total number of possible clusters. This procedure is similar to that used by Trethowan et al. (2001) to characterize associations among drier yield-testing locations and was considered an alternative to presenting each individual dendrogram.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 CONCLUSION
 REFERENCES
 
Pattern Analysis of Locations Represented in Two or More Years
The dendrogram constructed from pattern analysis of those ESWYT locations repeated in two or more years identified four primary clusters at the third-fusion level (Fig. 1). The geographical distribution of sites within each cluster is summarized in Fig. 2. Group-1 is the largest of the four groups and is comprised primarily of irrigated and rainfed sites from the Mediterranean and West Asian areas (14 of 31 sites) and South American locations (9 of 31). This group also includes CIMMYT's primary irrigated yield-testing location at the CIANO research station located near Cuidad Obregon, Sonora, Mexico. Group-2 is comprised of generally warmer sites in southern and eastern Asia (7 of 9 sites), plus Guatemala and Heilongjiang in northern China. Sites in Group-3 are located in the Southern Cone (4 of 9 sites), East Africa (3 of 9), Ecuador, and China. Group-4 combines three Southern Cone sites with one site each in Greece, Turkey, India, Iran, Pakistan, and South Africa.



View larger version (105K):
[in this window]
[in a new window]
  		Fig. 1. Dendrogram constructed from pattern analysis of the associations among sites sown to the ESWYT in two or more years.

 


View larger version (22K):
[in this window]
[in a new window]
  		Fig. 2. Map of locations clustering within the four groups identified from pattern analysis of ESWYTs repeated in two or more years.

 
Sites were considered to show an association or not on the basis of the way they differentiated genotypes for yield. To examine in detail the associations among the four primary groups, one site was selected from each group that represented a key location on the basis of the sum of the squared Euclidean distances (Table 2). In the instance where two sites had the same low value, the most frequently repeated location was chosen as most representative of the cluster. These four key sites, listed in Table 2, chosen from each of the four primary clusters, were then used for all subsequent comparisons and interpretations.


View this table:
[in this window]
[in a new window]
  		Table 2. Sum of the squared Euclidean distances taken from the dissimilarity matrix of the pattern analysis for each site within each cluster or group of the pattern analysis in Fig. 1.

 
Pattern Analysis of the Association of the Four Key Locations across Years
Pattern analysis was conducted across years on the four key locations identified from the previous section, Sakha (Egypt), Quezaltenango (Guatemala), Londrina (Brazil) and Pirsabak (Pakistan). Pattern analysis was chosen instead of SHMM because genotypes varied across years at the same location. The associations among years from the derived dendrograms are summarized in Table 3. Primary groups are summarized at both the second and third fusion levels where two and three clusters or groupings were observed for each site, respectively. At Sakha, ESWYT trials were sown in 14 of the 20 yr examined. The largest cluster comprised 50% of years at both the second and third fusion levels. Quezaltenango and Londrina gave one primary cluster accounting for 71% of years at the second fusion level and at Pirsabak, the primary cluster at the second fusion level accounted for 80% of years.


View this table:
[in this window]
[in a new window]
  		Table 3. The number of times each of four key sites, identified from the four clusters in Fig. 1, cluster across years at both the 2nd and 3rd fusion levels of the dendrograms constructed from pattern analysis of these locations across time.

 
Association of the Four Key Locations with Global Test Locations Using SHMM
The individual dendrograms generated from SHMM analysis for each of the 20 yr of ESWYT are summarized in Table 4. The dendrograms are summarized in a similar fashion to Trethowan et al. (2001), where the total number of realized clusters at the third fusion level are divided by the total number of possible clusters. Possible clusters are all instances where two sites appear in the same ESWYT or year; these sites may potentially cluster together at the third fusion level. The key locations from each of the four groups identified using pattern analysis are compared with all other global test sites and regions. We chose to summarize the dendrograms as a function of these four key locations because of the very large number of comparisons that would have to be made should every possible site by site contrast be calculated.


View this table:
[in this window]
[in a new window]
  		Table 4. Summary of the degree of association among those sites and regions clustering with four key locations identified from each of the four sites clusters in Fig. 1.

 
Associations with the African Region
Of the four key sites, Sakha and Pirsbak gave the closest association with African sites (75/136 and 58/106 of possible clusters, respectively, or 55% of all possible clusters). Londrina was the least associated key site across all African locations (40%), largely reflecting poorer associations with sites in eastern (25%) and central-southern Africa (36%).

Associations with Asian Sites
Pirsabak (74%), Sakha (66%), and Quezaltenango (62%) were best associated with all Asian locations. Londrina was clearly the least associated key site with the Asian region (44%). Pirsabak (67%) associated best with West Asia, whereas three sites, Pirsabak, Quezaltenango, and Sakha, showed equivalent association with eastern and southern Asian locations. Within South Asia, Pirsabak correlated strongly with other Pakistani sites (30/36) and Indian locations (27/32).

Associations with European Sites
There was little difference between Sakha (61%), Pirsabak (69%), and Quezaltenango (58%) in their association with all European sites. The poorer relationship between Londrina and these locations (41%) reflects a generally poor association with southern European sites (37%). Among northern European sites, Pirsabak (71%) and Sakha (68%) were the best key site predictors and Quezaltenango (50%) the poorest.

Associations with Sites in the Americas
Of the four key locations, Pirsabak and Sakha not only associated best across the Americas (62 and 58%, respectively) but also with the individual regions of North, South and Central America. Surprisingly, the Brazilian site Londrina did not associate at all with other Brazilian sites (0/9) and was poorly associated with all southern (27%), central (17%) and northern (35%) American sites. Quezaltenango was not as strongly related to the American locations as either Pirsabak or Sakha. However, this site was better associated than Londrina. In North America, the strong relationship of Pirsabak and Sakha with regional locations is influenced by the two locations' association with Mexican locations (33/49 and 26/39, or 67%, respectively).


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
CONCLUSION
 REFERENCES
 
Repeatability of Key Locations
Clearly all four key locations demonstrated a high degree of repeatability on the basis of their association across years. Importantly, Pirsabak, which was the most closely associated of all four key sites with global wheat growing areas based on the summaries of the SHMM analyses in Table 4, gave a primary cluster representing 80% of years or one primary year type. The other highly associated key site, Sakha, although producing two equally important year groupings at the second fusion level, showed that at the third fusion level of the dendrogram derived from pattern analysis, half of the ESWYT years fell into the same primary grouping. This high degree of repeatability improves the accuracy of subsequent comparisons of these key sites with other global test sites.

Complete environmental data were not available for all variables for all environments in each year; we were therefore unable to perform a three way clustering analysis to obtain quantitative associations of sites with environmental variables. Instead we used year-wise site clustering summarized across years to identify possible causes of association. Furthermore, the SHMM model did not allow us to incorporate external environmental variables. Table 5 contains long-term weather data for most of the locations represented in the pattern analysis. Almost no rain fell during the wheat-cropping season at Sakha, where an annual average of 3 cm was recorded during the past 11 yr. The ESWYT trials sown at this location were fully irrigated thereby reducing genotype x year interactions attributable to variation in rainfall and average yield was high (6.79 ± 1.96 Mg ha-1, data not shown). Similarly, the site at Pirsabak was irrigated each year, although this was supplemented by additional rainfall (on average 20 cm fell during the cropping season). Average yield (4.23 ± 1.14 Mg ha-1) was similar to the average across all ESWYT locations and years (4.22 Mg ha-1). The site at Quezaltenango is by contrast a rainfed location receiving on average 66 cm of annual cropping season rainfall. Nevertheless, average yield (4.29 ± 1.14 Mg ha-1) was similar to Pirsabak. Temperatures were much lower at Quezaltenango (annual average 12°C) than the other three key sites because of high altitude (2407 m above sea level). Relatively stable temperature and rainfall were primary contributors to the clustering of sites into one primary year grouping (71%) at the second fusion level. ESWYTs sown at Londrina also clustered into one primary year grouping 71% of the time. However, this may relate more to specific soil nutrient problems such as poor P availability (Riede and Campos, 1988) than stability in either growing temperature or rainfall. Average yield at this site was lower and yield variability, measured as standard deviation, higher (3.23 ± 1.66 Mg ha-1) than for the other three key locations.


View this table:
[in this window]
[in a new window]
  		Table 5. Summary of long-term average temperature and precipitation for key locations and some frequently repeated locations sown to the ESWYT (source: www.weatherbase.com; verified 14 April 2003).

 
The Biological Premise for Associations among Locations: Pattern Analysis
The four key clusters observed from pattern analysis of sites repeated at least three times during the 20 yr in which ESWYT was grown are influenced by temperature, rainfall, latitude, altitude, and disease incidence. Sites in the southern hemisphere had planting dates ranging from March–June, whereas spring wheat sowings in the northern hemisphere were generally in the period October–December.

Group-1 contains sites subject to primarily a Mediterranean weather pattern. Seventeen of the 31 sites in the group tended to be wetter earlier in the cropping season with lower rainfall in the grain-filling period; temperatures also rose, sometimes dramatically, toward the end of the crop cycle (Table 5). Most of these sites are scattered around the Mediterranean basin (such as Spain, Algeria, Tunisa, Jordan, and Syria) and extend into Iran. However, South American sites located on or west of the Andes are also characterized by predominately Mediterranean rainfall conditions (Colombia, Peru, and Chile). Five sites within this group represent dry irrigated sites of low to intermediate latitude (ranging from 17°N–31°N); these are Cuidad Obregon in Mexico, sites in Egypt including the key location Sakha, Saudi Arabia, India, and Thailand. Exceptions include sites in southeastern South America, which tend to be drier early in the season and become wetter as the crop matures, and the Kenyan highlands where rainfall is high throughout the cropping season. Stripe rust (Puccinia striformis Westend.) may have influenced clustering within this group because infection was recorded at some locations in some years. Of the 31 sites in this cluster, stripe rust was recorded at 26 sites.

Group-2 locations extend from the Pakistani–Indian Punjab eastwards through Nepal, Bangladesh, central and northeastern China, and Thailand. The low latitude, high altitude key site at Quezaltenango in Guatemala is also represented. All the Asian sites are dry locations that experience little cropping season rainfall. Average cropping season temperatures at these locations are mild to high, ranging from 17 to 24°C. High humidity may also have been a common factor in this cluster. The site at Quezaltenago is a high rainfall location and the trials were conducted without irrigation. The mild average cropping season temperature of 14°C and plentiful cropping season moisture are likely reasons why this site clustered with those in southern and eastern Asia. Regular stripe rust incidence in Guatemala and the occasional occurrence of stripe rust in China, India, and Pakistan may have had some influence on these site associations. However, the lack of regular incidence across years at the Asian locations reduces this likelihood.

The sites clustering in Group-3 combine sites in the Southern Cone of Latin America (4) with Eastern African locations (3) and one site in each of Ecuador and China. These locations are generally higher rainfall sites. Marcos Juarez, Argentina; Quito, Ecuador, and Njoro, Kenya, are the coolest locations with average cropping season temperatures ranging from 12 to 14°C. Stripe rust incidence (if not infection) was high at all locations with all sites recording the occurrence of this disease across years. Low soil pH may also have been a factor in the association among these sites as two of the three Brazilian locations and those in Tanzania and Kenya have acidic soils (Reide and Campos, 1988; Tanner and van Ginkel, 1988). In Ecuador, high rainfall may have led to soil nutrient imbalances that influenced the germplasm rank.

Group-4 combines sites from the Southern Cone of South America (3) with four sites of similar latitude ranging from Greece to Pakistan, one location in India, and one location in South Africa. Many of these sites have cooler cropping seasons with average temperatures close to 12°C (Argentina, Turkey, South Africa, and Greece), although Ocepar, Brazil (18°C) and Kanpur, India (20°C) are exceptions. There are two sub-groups within this cluster characterized by differences in cropping season precipitation. The sites in Pakistan, Argentina, Turkey, and Uruguay are drier early in the season, becoming wetter later (Kanpur is an exception). The sites in South Africa, Greece, and Iran are characterized by higher rainfall early in the season and less moisture during grain maturation.

The Biological Premise for Associations among Locations: SHMM Analysis
The strong association of Sakha (Egypt) and Pirsabak (Pakistan) with sites across the wheat growing areas of the world (Table 4) reflects the relative stability of these production environments as discussed earlier. The association of Pirsabak with African locations is influenced by its relationship with North African locations, particularly those in Egypt (26 of 35 possible clusters realized). Pirsabak is at least as good as Sakha (27/42) in predicting Egyptian locations, and Sakha associates well with Pakistani locations (48/60) reflecting similarities in temperature and growing season moisture between these two regions. Even though the association of Pirsabak with Asian locations is dominated by clustering with India and Pakistan, the significant associations with West Asia (67%) and East Asia (66%) largely reflect similarities in growing season moisture as growing season temperatures vary considerably across Asia. Many of these locations record limited growing season rainfall (Table 5) as wheat is generally grown in the cooler, dry winter months, particularly in Southern and Eastern Asia. This observation is supported by the stronger association of Sakha (the driest and highest yielding key site) with southern (76%) and eastern (76%) Asian locations compared with West Asian sites (48%), which are characterized by higher growing season rainfall.

It is more difficult to understand and explain the dominance of Pirsabak and especially Sakha in associations with European and South American locations as growing season rainfall in these regions is much higher and the occurrence of irrigated trials is significantly lower. The majority of European and South American trial sites were rain fed and average yields across these respective regions were 5.08 ± 1.86 Mg ha-1 and 3.46 ± 1.68 Mg ha-1 (data not shown). Disease incidence was unlikely to influence the clusters with European sites as many of these trials were sprayed with a fungicide and average growing season temperatures tend to be much cooler at European locations compared with either Sahka or Pirsabak. It appears that similarities in productivity between Sakha and Pirsabak and productivity in these regions also reflect similarities in cultivar ranking.

Quezaltenango in Guatemala, like Pirsabak and Sakha, associated well with African, Asian and European locations. This relationship is supported by the association between Guatemalan locations and Pirsabak (5/6) and Sakha (4/6). There is no clear geographical explanation for these associations as Quezaltenago is a low-latitude, high-altitude site with high growing-season rainfall. Nevertheless, these geographical parameters combine to give cool and stable growing-season temperatures, generally nonlimiting soil moisture and relatively low crossover interactions with climatically different wheat-growing regions. The key site at Londrina, Brazil was clearly the poorest predictor of all global wheat-growing sites and regions on the basis of the SHMM analysis. Londrina was least associated with South American locations, particularly other Brazilian sites (0/9), which probably reflects regional differences in soil acidity (Riede and Campos, 1988). Although the site at Londrina is not high in exchangeable aluminum, which is typical of many soils in the region, P availability according to Riede and Campos (1988) is relatively low. Interestingly, this lack of association with other Brazilian locations is not supported by the pattern analysis where two other Brazilian sites cluster with Londrina. Of the four key sites identified from pattern analysis, Londrina was also the lowest yielding and the most variable across years (3.23 ± 1.66, data not shown), which may have influenced regional associations.


   CONCLUSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSION
REFERENCES
 
Nachit et al. (1992) and Peterson and Pfeiffer (1989) found available moisture to be a primary influence on location clustering. This finding is supported by our study as the ESWYT was grown across a range of soil moisture conditions, and cropping season water availability was clearly a primary differentiating factor. In addition to moisture availability, growing-season temperature, determined by a combination of sowing date, latitude, altitude, and rainfall, was a far more influential factor than latitude alone. The influence of temperature (Haji and Hunt, 1999), its effects on maturity date (Peterson and Pfeiffer, 1989; Abdalla et al., 1996) and inferences for location clustering have been noted previously. Others found that diseases rather than climatic variables influence site associations (Brancourt-Hulmel et al., 1999). The observed relationship between irrigated locations and rainfed sites in the current study was influenced by the shuttle breeding method used by CIMMYT (Braun et al., 1996). Adaptation to high rainfall conditions and their associated diseases is captured in the germplasm entering ESWYT through selection of materials at Toluca in the central Mexican highlands. While disease incidence, most notably the Puccinia species and Septoria tritici Roberge in Desmaz., was frequently recorded, the level of infection was seldom high enough to cause significant differentiation of the germplasm, probably because of inherent disease resistance in the ESWYT germplasm. Disease incidence was, therefore, considered to be a secondary rather than primary cause of site association in this study.

The well-watered and stable cropping conditions at Pirsabak and Sakha, and, to a lesser extent, Quezaltenango, correlate significantly with the primary spring wheat-cropping areas of the world. The high degree of relationship between these sites and sites in Mexico where the germplasm was developed (67% of total possible clusters were realized for both Pirsabak and Sakha, when these sites were clustered individually with Mexican locations) indicates that these environments differentiate cultivars in a similar manner. The high repeatability of these two locations from year to year, and similarities among their environmental variables and those of other wheat-growing regions, provide valuable evidence of global performance and adaptation of wheat genotypes bred in Mexico. Historically, wheat cultivars bred in Mexico by CIMMYT have adapted well to a wide range of cropping conditions around the world (Rajaram, 1995). However, it may be possible to enhance the development of suitable wheat cultivars and improve their global adaptation by combining information from Pirsabak and Sakha in future crossing, selection and screening strategies as both these sites associate well with a wide range of different wheat growing locations. Materials selected at these key locations could be expected to perform well in a range of geographically diverse wheat growing environments.

Received for publication October 9, 2002.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSION
 REFERENCES
 




This article has been cited by other articles:


Home page
 Crop Sci.Home page
W. Putto, A. Patanothai, S. Jogloy, and G. Hoogenboom
Determination of Mega-Environments for Peanut Breeding Using the CSM-CROPGRO-Peanut Model
Crop Sci., May 1, 2008; 48(3): 973 - 982.
[Abstract] [Full Text] [PDF]

Home page
 Crop Sci.Home page
J. G. Robins, B. L. Waldron, K. P. Vogel, J. D. Berdahl, M. R. Haferkamp, K. B. Jensen, T. A. Jones, R. Mitchell, and B. K. Kindiger
Characterization of Testing Locations for Developing Cool-Season Grass Species
Crop Sci., May 31, 2007; 47(3): 1004 - 1012.
[Abstract] [Full Text] [PDF]

Home page
 Crop Sci.Home page
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]

Home page
 Crop Sci.Home page
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]

Home page
 Crop Sci.Home page
M. Lillemo, M. van Ginkel, R. M. Trethowan, E. Hernandez, and S. Rajaram
Associations among International CIMMYT Bread Wheat Yield Testing Locations in High Rainfall Areas and Their Implications for Wheat Breeding
Crop Sci., July 1, 2004; 44(4): 1163 - 1169.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (16)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Trethowan, R. M.
Right arrow Articles by Hernandez, E.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Trethowan, R. M.
Right arrow Articles by Hernandez, E.
Agricola
Right arrow Articles by Trethowan, R. M.
Right arrow Articles by Hernandez, E.
Related Collections
Right arrow Crop Genetics
Right arrow Plant and Environment Interactions

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal Vadose Zone Journal
Journal of Natural Resources
and Life Sciences Education		 Soil Science Society of America Journal
Journal of Plant Registrations Journal of
Environmental Quality		 The Plant Genome