Estratégias de seleção e divergência genética em linhagens e estudo de gerações em soja

Abstract

Breeding of autogamous plants is an ongoing process that includes important steps for successful development of new cultivars. They are the choice of the parents, obtaining the segregant populations and conducting them in the generation advance until the selection of superior lineages. The genetic basis of soybean germplasm in Brazil is short, making it necessary to know the genetic diversity among the available genotypes to propose hybrid combinations. Estimates of genetic parameters are also important because they allow to make inferences about the genetic variability of the studied genotypes, assisting in the selection stage of quantitative traits in both the early generations and the advanced lineages. The selection of superior genotypes is the most important phase of breeding and can be performed based on different selection strategies. The objective of the present work is to evaluate the genetic diversity, the measures of correlation between characters and evaluate selection strategies in soybean lineages; and carry out the study of generations in two F2 segregating populations. Two field experiments were carried out at the Capim Branco experimental farm, in Uberlândia - MG. In the first experiment, 20 soybean strains and two control cultivars were evaluated in a randomized complete block design with three replicates. Initially, the genetic parameters, the correlations between characters and the genetic gains obtained by direct, indirect and index-based selection were evaluated. It was verified the existence of genetic variability among the studied genotypes for characters related with cycle, plant height, number of nodes, number of pods and grain yield, and even though there is correlation between the characters evaluated, especially a negative correlation of high magnitude between the character related to the cycle and the character of grain yield. The selection strategy that presented the most favorable results for the joint selection of characters was the selection index of Mulamba and Mock (1978). Then, genetic dissimilarity was estimated between the evaluated genotypes and subsequent grouping by different hierarchical optimization methods based on canonical variables, in order to identify the diversity among the genotypes. There was a variation between the dissimilarity coefficients among the genotypes, highlighting the control of the experiment as the most divergent. The clustering hierarchical optimization and canonicals variables methods agreed on dividing the genotypes into three distinct groups. In the second experiment, parental and F2 generations were evaluated in two segregating populations, originating from the crosses between Emgopa-316 with UFUS 6901 and Emgopa-316 with UFUS 7415. Estimates of genetic variance and heritability oscillated between the two populations but identified the possibility of selection of individuals with genetic gains oscillating between -6,55% and 24,21% in FP-A population and between -12,41% and 51,06% in FP-B population for the characters evaluated.