Adaptability and phenotipic stability in soybean evaluated by the Toler method

The objective of this work was to evaluate 69 soybean genotypes with different maturation cycles in different environments at Mato Grosso and Mato Grosso do Sul States in Brazil using the non-linear regression method of TOLER (1990). This methodology considers the fiting the genotypes in five groups, according to the response pattern A, B, C, D and E. It was observed that the environmental quality estimates had different values, allowing for inferences of worse and better environments. Pedra Preta was the best quality environment, having the greater estimative, that contributed to the increase of the general environment mean. The classification based on yield and on the response pattern showed that genotypes with double desirable response pattern (group A, convex pattern), presented high yield, mainly in the line early cycle FMT-42313. The pattern E genotypes (concave pattern) did not show satisfactory levels of mean yield as MSOY-8550 and Crixás. The response pattern of monosegmented genotypes (B, C and D) showed variable yield, being most of them clustered in group C, having a simple linear response that did not differ greatly from the mean response of the environments. Conquista cultivar had average yield and low variability in distinct environments. The methodology has statistic advantages for the analysis of adaptability and stability among soybean genotypes, since it discriminate genotypes with narrow genetic base in to groups with different response pattern.

Glycine max; non-linear regression; genotype x environment interaction; stability; adaptability