MODELING GENOTYPE X ENVIRONMENT INTERACTION IN GRAIN YIELD OF WHITE MAIZE HYBRIDS IN MULTIPLE ENVIROMENTS

Abstract

Plant breeding programs aimed at obtaining genotypes with high grain yield and stable in a wide range of environmental conditions face environmental factors that mask potential genotypes. The Genotype x Environment interaction (G × E) might cause the suitability of predicted genotypes to a particular environment to be inaccurate. This study modelled the G × E interaction using different statistical models in a group of hybrids of maize (Zea mays L.) evaluated in tropical environments. Twenty-nine white-endosperm hybrids were evaluated in 15 environments of tropical America, with an alpha-lattice design. Grain yield was first analyzed with a combined analysis of variance. Subsequently, the additive main effect and multiplicative interaction (AMMI) and the site regression (SREG) with analytic factors (FA) model were applied to study and model G × E and to define environments that best discriminate genotypes and allow the grouping of environments and genotypes. The AMMI method pointed out a locality from Guatemala, one from México and one from Nicaragua as the ones with highest G × E; generated four mega-environments; and defined the most stable and good-yielding hybrid. The SREG FA method proved a good predictor since it allowed the identification of four subgroups and grouped environments of different countries with similar features.