Applying mixed model selection tools for tropical maize

Maize is fundamental for global food security and a versatile industrial commodity. Its genetic improvement is crucial and models that explore the Genotype by Environment Interaction are essential for developing stable and high yield cultivars in climate change perspectives. This study aimed to explore the GEI and identify superior maize hybrids for Brazilian cropping regions. We analysed grain yield data from 117 hybrids evaluated in 47 environments using a Factor Analytic linear mixed model. This approach efficiently handles unbalanced multi-environment trial data and models the genetic covariance between environments. A model with four latent factors was selected, explaining over 75% of the genetic variance observed. The analysis revealed a complex GEI pattern, with pairwise genetic correlations between environments ranging from -0.762 to 0.987, indicating that 28% of the interaction was of the crossover type, leading to significant rank changes among genotypes. For hybrid selection, factor analytic selection tools was employed, providing the parameters of overall performance, stability and reliability to select the best genotype. Hybrid H01 was identified as a promising candidate, demonstrating a balanced combination of performance and high reliability. The FA model associated with FAST proved to be a powerful tool for guiding cultivar recommendation.