Correção do efeito ambiental em ensaios de famílias de soja, intercaladas com testemunhas, para predição de ganhos por seleção

One F5 soybean population was evaluated in Vicosa-MG, Brazil, during the 1997/98 growing season. The experimental design was of families without replications, inserted with controls, with replications, allowing the estimation of the environmental component associated to the phenotypic variance of the population under study, and consequently the genetic parameters. The evaluations at family and at individual within the family allowed to decompose the genotypic variance and to estimate the heritability on the narrow sense. The environmental effect correction was based on the replications of controls. The correction of the environment effect favored the selection of the families that originally showed performance a little inferior to the original selection truncation point, but were not favorable by the environmental, causing changes in the family groups selected through the simulation of the same selection strategies on the original and corrected data.