Analysis of statistical methods to estimate genotype-by-environment interaction for yield-based selection of new bean cultivars and lines

• Autor(a) principal: Kaneko, Henrique Guilherme Hisashi
• Data de Publicação: 2023
• Outros Autores: Fonseca Júnior, Nelson da Silva, Cirino, Vania Moda
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Semina. Ciências Agrárias (Online)
• Texto Completo: https://ojs.uel.br/revistas/uel/index.php/semagrarias/article/view/47620
• topic: Linear regression; Mixed models; Multivariate analysis; Phaseolus vulgaris L.; Regressão linear; Modelos mistos; Análise multivariada; Phaseolus vulgaris L.

This study aimed to assess the statistical methods available to estimate genotype-by-environment (GxE) for yield-based selection of new bean cultivars and lines, and identify a testing strategy capable of reducing the number of test years and producing reliable results. Fourteen genotypes were tested in 23 environments in Paraná state, Brazil, over three consecutive years. The yield data obtained in each environment were submitted to homogeneity of variance analysis, normality testing, individual and pooled analysis of variance (ANOVA) of all the environments. GxE interaction was studied using linear, multivariate, and mixed model regression. Analyses were performed using data from two and three years of assessment. For linear and mixed model regression, the number of years analyzed did not affect result interpretation. However, in multivariate analysis, genotype behavior varied in the environments studied according to the number of years analyzed. The results obtained indicate that two years of assessment were sufficient to indicate new cultivars adapted to different environments.