Correlation estimates and assessment of selection strategies in five soybean populations

Abstract

The objectives of the study were to analyze correlation among some important traits for breeding and assess three family selection strategies in soybean populations. Five soybean populations were assessed, three in the F6 generation and two in the F5 generation in Viçosa, MG, in the 1997/98 growing season. The experiment involved non replicated families inserted with replicated controls to allow estimation of the environmental component associated to the phenotypic variance of the studied populations. This arrangement allows the estimation of genetic parameters. The estimates of genetic correlation showed that the number of pods per plant is a good indicator of grain yield. The estimates of the gains from simulation of the three selection strategies showed that the direct selection on yield strategy resulted in greatest gains for the same trait but indirect selection via number of pods per plant resulted in gains close to those of direct selection. Selection based on the Pesek and Baker index enabled grain yield gains and simultaneous reduction in the cycle and plant height, despite the positive correlation of these two traits with grain yield.