Optimization of maize hybrids selection in preliminary environmental trails: comparison of classical and spatial approaches

Introduction: In plant breeding, identification of high-yield genotypes and suitable stability for introducing new cultivars or selection of high potential genotypes in preliminary breeding trials is very important. However, experimental design may not be sufficient to consider field heterogeneity in the plot area. The aim of this study was to evaluate the potential of spatial models to correct spatial trends and improve prediction of genotype values and increase accuracy of selection of genotypes in comparison with classic statistical design.

Materials and methods: In this study, data analysis were conducted based on a two - step procedure. In the first step, the data of each experiment were modified separately using spatial correction models (SpATS, AR1×AR1 and SpATS + AR1×AR1 combined model) to remove the spatial heterogeneity and systematic environmental effects. Then, in the second step, the corrected data entered the Multi-Environment Trials based on factor Analytic (FA) to accurately model the genotype × environment interaction and to identify superior genotypes based on yield and stability. For this purpose, an experiment was conducted with 105 hybrid maize hybrids in α -lattice design with two replications and five incomplete blocks per replication in three stations of Karaj, Kerman and Kermanshah.

Results and discussion: The results showed that the SpATS model effectively eliminate large - scale spatial variations and decreases the error of prediction (RMSE) in comparison with raw data and other models. Comparison of SpATS spatial model with the alpha lattice design showed that the SpATS model significantly improved the accuracy of the yield and ranking of genotypes. After correction of data with SpATS, Multi-Environment Trials were performed based on factor Analytic (FA) to model the interaction of genotype × environment and genotypes with high yield and suitable stability were identified. Kermanshah station with high genetic variance, had more genotype discriminate and was the most suitable station for identification of superior genotypes. The FA2 biplot indicated that genotypes located near the origin, such as H80, H102, H33, and H22, exhibited broad adaptability and high stability. In contrast, genotypes positioned farther from the origin, including H60, H24, H86, H93, and H91, showed stronger genotype × environment interactions and greater dependence on environmental conditions. Genotypes such as H24, H35, H26, H28, and H34 demonstrated the most favorable combination of FA1 and FA2 scores. In addition, the stability index WAASB and the combined performance–stability index WASSBY were employed for genotype selection. Genotypes H24, H30, H26, and H49, with high WASSBY values, exhibited a desirable balance between yield performance and environmental stability. Based on performance and stability indices, despite the superiority of five hybrids (H24, H26, H30, H35, and H49), approximately 20% of the top-performing genotypes were advanced to the next stage of the breeding program in order to maintain genetic diversity and enhance breeding opportunities. Overall, the results suggest that the application of spatial models, particularly SpATS, while accounting for field heterogeneity in maize trials, in combination with FA analysis, improves data analysis and the prediction of genotypic values. This integrated approach provides an efficient and reliable framework for increasing the accuracy of selecting superior genotypes in multi-environment trials.