Statistical methods to study phenotypic adaptability and stability in soybean

• Autor(a) principal: Silva, Waldir Camargos Júnior e
• Data de Publicação: 2006
• Outros Autores: Duarte, João Batista
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: Pesquisa Agropecuária Brasileira (Online)
• Texto Completo: https://seer.sct.embrapa.br/index.php/pab/article/view/7101
• topic: Glycine max; GE interaction; yield adaptation; yield stability; Glycine max; interação GxA; adaptação produtiva; estabilidade de rendimento

The objective of this work was to evaluate statistical methods of genotype by environment interaction (GE) analysis, emphasizing phenotypic adaptability and stability. Observed data of soybean grain yield were used, which were obtained from variety trials carried out in seven environments in Goiás State. Twenty eight genotypes were evaluated in each trial, and four of them were commercial varieties. The methods evaluated were Traditional, Plaisted & Peterson, Wricke, Finlay & Wilkinson, Eberhart & Russell, Verma, Chahal & Murty, Toler, AMMI (additive main effect and multiplicative interaction), Hühn, Annicchiarico and Lin & Binns. In order to study the association among the methods, the correlation of Spearman was used. Plaisted & Peterson and Wricke methods were always strongly associated, contraindicating their concomitant use. The Annicchiarico and Lin & Binns methods also showed high correlation between themselves producing, therefore, very similar genotypic classifications. Thus, the simultaneous use of these procedures is not also recommended. The use of one of them, however, is counseled. In the case of methods based exclusively on regression coefficients, it should be considered their association to other method, based in the variance of the GE interaction, or the introduction of statistical measures as the variance of the regression deviations. The joint use of Eberhart & Russell method and AMMI analysis is another recommendation, because their significant correlations with most of the other methods, and of a low correlation between them.