População-alvo de ambientes e modelagem da interação de genótipos por ambientes em feijão-comum

• Ano de defesa: 2025
• Autor(a) principal: Cruz, Demila Duarte da Mata
• Orientador(a): Resende, Rafael Tassinari
• Banca de defesa: Resende, Rafael Tassinari , Torga, Paula Pereira , Menezes, Renato de Carvalho
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Goiás
• Programa de Pós-Graduação: Programa de Pós-graduação em Genética e Melhoramento de Plantas (EA)
• Departamento: Escola de Agronomia - EA (RMG)
• País: Brasil
• Palavras-chave em Português: Phaseolus vulgaris L.; TPEmap function; Ambientômica; Modelos lineares mistos
• Palavras-chave em Inglês: Envirotyping; Linear mixed models
• Área do conhecimento CNPq: CIENCIAS AGRARIAS::AGRONOMIA
• Link de acesso: https://repositorio.bc.ufg.br/tede/handle/tede/14735

Environmental variability associated with global climate change imposes significant challenges to agriculture. In this context, understanding the genotype × environment (G × E) interaction becomes essential to develop adapted and stable cultivars with high performance under diverse conditions. Therefore, this study aimed to evaluate the influence of G × E interaction on the productivity of common bean (Phaseolus vulgaris L.) through the analysis of multi-environment trials (METs) conducted by Embrapa Rice and Beans between 2011 and 2018. A total of 424 experimental trials were analyzed, including 241 carioca-type and 183 black-type trials, totaling 87 genotypes evaluated across 71 locations. The Target Population of Environments (TPE) was characterized using the TPEmap function, which, based on geographic polygons, allows parametrizations such as pixel size, buffers, and concavity. This approach enabled the identification of the spatial pattern of experiments and facilitated the integration of genotypic, environmental, and spatial data in envirotyping studies. The estimation of genetic and environmental effects was performed using REML/BLUP mixed linear models, complemented by multivariate analyses such as GGE Biplot and missing data imputation through principal component analysis (PCA). Environmental stratification and the definition of mega-environments allowed grouping locations with similar characteristics based on the presence or absence of G × E interaction and highlighted the influence of geographic distance on phenotypic responses. The analysis of cultivar stability and adaptability was based on a ranking derived from the PRVG, MHPRVG, Lin and Binns, Wi, and FW indices, considering their specificities, and revealed distinct behavioral patterns across growing seasons. The results underscore the fundamental role of integrated statistical modeling and spatial analysis in cultivar recommendation, guiding common bean breeding strategies in the face of environmental heterogeneity.