Estratégias experimentais para avaliação de linhagens em programas de melhoramento do feijoeiro

Abstract

Brazil is one of the largest producers and consumers of bean common (Phaseolus vulgaris L.) in the world. Currently, the per capita consumption of this legume in the country is around 12.7 kg/inhabitant/year (EMBRAPA, 2021). The production of bean common in the 2020/21 harvest was 2.8 million tons, in an estimated area of 2.9 million hectares, corresponding to an average productivity of 1489 kg/ha-1. The search for greater efficiency in genetic improvement programs is constant, due to the need to increase the productivity and stability of cultivated species. In order to choose promising strains and obtain reliable results, it is necessary to carry out experiments with high precision. Increasing the number of repetitions is the most general and efficient method for this. However, the indefinite increase in the number of repetitions leads to high testing costs and an experimental area that is too large for field experiments. Therefore, in the experimental design, the number of repetitions must be combined with the judicious choice of an adequate plot size and format. The objective of this work was to determine the ideal plot format and number of repetitions for the evaluation of lines in common bean breeding. The present work consisted of four field experiments, two experiments conducted in the 2020 "water" harvest, and the other two experiments in the 2021 "drought" harvest. A total of 75 carioca-type bean lines were evaluated. regarding grain productivity. In order to verify the effect of the plot format on the evaluation of lines, joint analyzes of variance were performed with data from experiments with different plot formats in the same season. To verify the effect of the number of repetitions, simulations of individual analyzes of variance were performed considering different numbers of repetitions in the experiments with three and six repetitions. It was found in this work that experiments conducted with two replications and with plots of two lines of two meters proved to be sufficient to carry out the accurate selection of lines in bean breeding programs.