Genetic correlations and genotype × environment interaction analysis in tropical maize

Genotype × environment interaction (GEI) is one of the major challenges in maize breeding programs because it alters the relative performance of genotypes across environments and complicates the recommendation of cultivars with broad adaptation. In this context, factor analytic models enable the description of the covariance structure of the interaction and provide useful information to characterize the similarity among environments and identify superior genotypes. Thus, the objective of this study was to evaluate GEI through genetic correlations among environments and identify superior hybrids using tools based on factor analytic models. Grain yield data from 242 tropical maize hybrids developed by Embrapa maize breeding program, evaluated in 62 environments across different regions of Brazil between 2022 and 2025, were analyzed. Factor analytic models were fitted, and the five-factor model was selected because it provided the best fit to the data and explained 78.36% of the genetic variance of the interaction. Heritability estimates ranged from 0.27 to 0.78. Genetic correlations between pairs of environments were obtained from the estimated genetic covariance matrix and exhibited wide variation, indicating different levels of similarity in hybrid responses. High genetic correlations were observed mainly among trials conducted at the same location in different years or under similar experimental conditions, such as Janaúba, Londrina, Sinop, Sete Lagoas, and Vilhena, indicating stability in genotype ranking within these groups of environments. In contrast, Chapadinha showed predominantly negative correlations with several environments, suggesting differential hybrid responses and highlighting its potential to maximize genetic discrimination. High similarity was also observed among environments subjected to different fertilization and Azospirillum inoculation management practices in Sete Lagoas, indicating that these practices had less influence on the relative ranking of hybrids than the environmental differences among locations. Additionally, the Factor Analytic Selection Tools (FAST) were used to estimate overall performance, stability, and prediction reliability, enabling the construction of a selection index to identify hybrids that were simultaneously high-yielding, stable, and reliable. Overall, the results demonstrate that factor analytic modeling, combined with genetic correlations and FAST metrics, is an efficient approach for understanding GEI, identifying groups of genetically similar environments, demonstrating that environmental similarity depends on both geographic location and the specific environmental conditions of each growing season, and providing support for the optimization of evaluation networks and the more efficient recommendation of tropical maize hybrids.