Evaluation of Genotype x Environment Interaction for Yield Attributes in Maize across Multiple Environments

Abstract

Accurate prediction of hybrid performance under variable environments requires models that capture both genetic and climatic drivers. In this study, linear mixed models (LMMs) were applied to grain weight, kernel rows, and kernels per row across multiple environments to evaluate the contribution of environmental covariates in improving model performance. Baseline models considered genotype, environment, and their interaction as fixed or random effects, while extended models incorporated cumulative growing degree days (CGDD) at anthesis and silking. Model evaluation using Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) showed that models including environmental covariates consistently achieved better fit for grain weight and kernels per row, highlighting the role of thermal time in explaining phenotypic variation. In contrast, kernel rows were largely unaffected by temperature-related covariates, indicating that this trait is primarily determined by G×E interactions. These findings provide actionable insights for maize breeders, demonstrating that incorporating environmental covariates can improve hybrid selection and yield prediction under diverse environmental conditions.