Genotype × environmental interactions for wheat yields and selection for widely adapted wheat genotypes

Abstract

Large genotype x environment (g x e) interaction variances for yield relative to those for genotypes have been recognized for wheat cultivars in Queensland. The utility of a linear model to explain these interactions was examined by yield-testing 100 different wheat cultivars at nine different environments, including four locations and three years, in south-eastern Queensland. The linear model was found to explain less than 40% of the total g x e interaction and to give only a general indication of cultivar responses to different environments. Selection strategies to identify widely adapted cultivar involving several parameters (mean cultivar yield over all environments, the g x e interaction for each cultivar and the regression coefficient for each cultivar), singly and in combination, were evaluated. Greater selection differentials were found in most environments when selection was practiced for high mean yield across all environments when the yield of each cultivar in each environment was expressed as a percentage of the environment mean yield.