Implicações da interação de genótipos com ambientes na recomendação de cultivares de feijoeiro comum: validação de regras e importância de fatores ambientais

• Autor(a) principal: Barros, Matheus Souza de
• Data de Publicação: 2019
• Tipo de documento: Tese
• Idioma: por
• Título da fonte: Repositório Institucional da UFG
• dARK ID: ark:/38995/00130000082z0
• Texto Completo: http://repositorio.bc.ufg.br/tede/handle/tede/10290
• topic: Phaseolus vulgaris; Simulação computacional; Componentes de variância; Modelos mistos; Interação de genótipos com ambientes; Phaseolus vulgaris; Computer simulation; Variance components; Mixed models; Genotypes by environments interaction; GENETICA::GENETICA VEGETAL

The genotypes by environments interaction (GxA) can be defined as the differential phenotypic response of genotypes in different environments. This phenomenon is the main complicating factor in recommending broadly adapted cultivars in common bean and others crops. The value of cultivation and use (VCU) tests are required for registration of new cultivars. These tests are intended to generate agronomic information about the performance of candidate lines for new cultivars in the various cultivation environments. The rules for conducting VCU tests were very restrictive as they require many tests to register the new cultivar. This step in the development process of new cultivars is the most costly for common bean breeding programs for logistical and operational reasons. Because of this, the standard rule has been relaxed since 2010 and was considers the regionalization of Brazil in edaphoclimatic regions. Thus, ten environments are currently accepted for regions I (South) and II (Central), and six environments for Region III (Northeast). Of which three environments are required per sowing season for the season in which the cultivar is to be indicated. The sowing seasons are for region I “waters” (águas) and “drought” (seca); and for region II "waters" and "winter" (inverno). The tests must be conducted for two years. Thus, this work aims to: validate the number of environments (VCU assays) currently accepted for registration of new cultivars, through computer simulations with real data, and; to evaluate environmental factors to determine their relevance to the phenotypic variation of candidate lines. Grain yield data were used for the study. Data were obtained from 406 VCU trials during 17 years of the common bean breeding program of Embrapa Rice and Beans. During this period 101 candidate lines and 19 commercial cultivars were evaluated as control. The trials were distributed among the three edaphoclimatic regions that contribute most for of the common common bean grain production. For the simulation study an algorithm was built to randomly sample the environments in various combinations. The combinations represent several scenarios, which vary in the number of environments. 288646 simulations were performed and the five best classified genotypes were compared, by coincidence, with the five classified in the complete joint analysis. This analysis uses all available environments in each VCU cycle (two years). Nonlinear modeling was used to adjust estimates to the asymptotic curve to obtain the adjusted averages of coincidence. The curve equation was derived to obtain the instantaneous rate of change. For the criterion of determining the minimum number of environments, the mean value theorem was used to estimate the average rate of change (∆dM) between scenarios, where the x value for the average rate represents the minimum number of environments. For the study of environmental factors two approaches were used: the modeling by mixed models to estimate the variance components and; the classical approach to analysis of variance with decomposition of GxA interaction. In addition to these analyzes, the GxA interaction was decomposed into the simple and complex parts. The results of the simulation study indicated high average coincidence between genotypes even in scenarios with few environments. The elevation of the coincidence was progressive until the scenario with eight environments in regions I and II, which represents the point of ∆dM. However, the number of currently accepted environments (ten) for these regions was more appropriate. For region III, the ∆dM occurred 6.25 indicating that the minimum number of environments for this region is capable of detecting the genotypes most adapted to this region. For sowing seasons, three environments resulted in estimates of over 60% of average coincidence, except for the winter season (53.4%). Thus, it is concluded that the number of environments currently accepted for registration of new cultivars is capable of indicating the superior genotypes. The mixed model evaluation of the environmental factors analysis by region indicated that the GxLxExA interaction is the component of variance that contributes most to the total variance, followed by the effect of locations for regions I and II. In region III the effect of sites was the most important of the components. The analysis of variance of the factors and their partial decompositions indicated that in region I that the isolated effects of times and places together with the GxL interaction were more relevant. In region II, GxE interaction was the most significant componet involving genotypes. The isolated environmental components varied in importance between cycles in the region II. The local effect and GxL interaction are the most expressive components in region III. The decomposition of the interaction was predominantly complex in all studied cycles for all regions. It is concluded, therefore, that in region II the environmental factors sowing seasons, years and location were the ones that participated with most of the total variation. GxE was the most significant among the interactions of environmental factors involving genotypes in region II. In region III the main sources of variation for the isolated effects were location and years, in that order. The variance components indicated that the interaction of genotypes with the environmental components were more important for regions I and II, and for region III the location effect was more relevant. The location effect is the isolated variance component that most contributes to the total variation in all regions. The type of complex interaction was predominant among the combined assays in all regions.