Reamostragem de um plano experimental multiambientes na seleção de progênies de soja

Abstract

The experimental design plays a crucial role in planning experiments for plant breeding programs. This study aimed to evaluate the effect of progeny’s unbalancing in multi- environment trials using the augmented block design for selection purposes in a soybean breeding program. The experiments were conducted during the 2021/22 growing season in eight locations across the northern and northeastern regions of Brazil. Based on balanced data, scenarios with different levels of unbalancing, ranging from 12.5% to 87.5%, were simulated. The data from all scenarios were analyzed using the mixed model approach with the BLUP (Best Linear Unbiased Predictor), with (ABLUP) and without (IBLUP) a pedigree matrix. The estimation of genetic and environmental variance components was performed using the REML (Residual Maximum Likelihood) method. The efficiency of the experimental designs was assessed based on the estimates of selective accuracy for the progeny mean, Spearman's rank correlation, and the selection coincidence index. The unbalancing resulted in optimized and more efficient experimental designs for resource allocation. The ABLUP analysis outperformed IBLUP in all scenarios. A progressive reduction in selective accuracy, Spearman’s rank correlation, and selection coincidence index was observed as imbalance increased. Intermediate imbalance scenarios proved to be an interesting option for optimizing experimental designs, as they exhibited a low reduction in the evaluated parameter estimates and high resource-use efficiency, especially in the early stages of a breeding program when there are many genotypes and limited genetic material available for testing.