Clustering Durum Wheat Genotypes in Multi-Environmental Trials of Rain-Fed Conditions

Abstract

AbstractFor durum wheat genotypes evaluation in multi-environmental trials (MET), measured seed yield is the combined result of effects of genotype (G), environment (E) and genotype by environment GE interaction. The GE interaction structure can be identified if the data are stratified into homogeneous subsets through cluster analysis. A combined analysis to assess GE interactions of 20 durum wheat genotypes across 14 environments was undertaken. The combined analysis of variance for E, G and GE interaction was significant, suggesting differential responses of the genotypes in various environments. Four cluster methods, which differ in the dissimilarity indices depending on the regression model or ANOVA model, were used. According to dendograms of regression methods there were 10 different genotypic groups based on G (intercept) and GE (line slope) sources and 3 different genotypic groups based on GE (line slope) sources. Also, the dendograms of ANOVA methods indicated 11 different genotypic groups based on G and GE sources and 13 different genotypic groups based on GE sources. The above mentioned genotypic groups were determined via F-test as an empirical stopping criterion for clustering. Due to the high values of regression’s determination coefficient which ranged from 92.6 to 99.4, using of the linear regression-based clustering was more practical. The genotypes clustering based on similarity of linear regression parameters or ANOVA model indicated that there were considerable variations among durum wheat genotypes and there are different with each other in response to environmental changes. Such an outcome could be regularly applied in the future to clattering durum wheat genotypes and other crops based on regression or ANOVA models in the Middle East and other areas of the world.