Combining AMMI and BLUP analysis to select high-yielding soybean genotypes in Benin

Abstract

Thirty soybean [Glycine max (L.) Merr.] genotypes, along with three checks, were evaluated over three seasons across five communes in Benin. The experiments were laid out in an alpha lattice design with three replicates. Additive multiplicative mean interaction (AMMI) and best linear unbiased predictor (BLUP) analysis were combined to assess differential agronomic performance and yield stability among genotypes. There was significant variation (p < 0.001) between genotypes for all traits, with highly significant environmental and genotype × environment interaction (GEI) effects on soybean grain yield (p < 0.001). The likelihood ratio test indicated that both genotype and interaction effects were highly significant (p < 0.001). The low R² (0.21) for GEI reflected the presence of high residual variation in the GEI component, in contrast to the AMMI analysis of variance, which explained a high proportion of the GEI through the first two interaction principal component axes (52%). The very high value of the predictive accuracy (0.89) confirmed the model's reliability in selecting superior genotypes. The low (0.33) genotypic correlation between environments indicated that it was difficult to select superior genotypes for each environment. Based on the superiority index (weighted average absolute scores from BLUP for yield) of BLUP, simultaneous selection led to the identification of Jenguma 2.67 ± 0.06 t ha⁻¹ as the most stable and productive genotype across environments, followed by Favour 2.34 ± 0.08 t ha⁻¹, and Afayak 2.46 ± 0.08 t ha⁻¹. The agronomic performance of soybean in this study suggested great potential for diversifying cotton-based cropping systems in Benin, thereby improving their sustainability. The effect of these soybean genotypes on the productivity of intercrop combinations and sequences of cash crops, such as cotton, is yet to be investigated.