The objective of this work was to assess the efficiency of different methods of environmental stratification, the representation level of the evaluation places, and the adaptability and stability of soybean genotypes, through yield trials in the Paraná and Santa Catarina States, Brazil, in the agricultural years of 2000 to 2003, totaling 15 environments. For environmental stratification, the Lin's traditional method and factor analysis, allied to percentage of the simple portion (SP%) of genotype vs. environment interaction (GxE), were utilized. Adaptability and stability of the genotypes were determined by linear and bisegmented regression models. Twenty one soybean genotypes were evaluated on randomized complete blocks design, comprising three replications, and the analyzed variable was grain yield. RB 605 genotype is highly adapted and presents high yield. According to both methods, Palotina and Brasilândia do Sul localities can be reduced to one trial location. The factor analysis allied to SP% of GxE interaction is more selective to environmental stratification in relation to the Lin's traditional method.
Glycine max; genotype x environment interaction; predictability
