Currently, Brachiaria brizantha Stapf is the tropical grass with the largest cultivated area in Brazil, being responsible for a major part of the seed market and for the increase in animal production. There are several superior genotypes available in germplasm banks, collected from the African continent and in need of evaluation. The selection of these genotypes must be based on predicted genotype values of the individuals in function of being perennial plants and of unsexed propagation. The standard procedure for the prediction of genetic values is the individual BLUP (Best Linear Unbiased Predictor), using estimates of variance components achieved by the restricted maximum likelihood (REML) on an individual model. With the aim of selecting B. brizantha materials superior to the commercial cultivars in relation to the agronomic traits, 218 genotypes were evaluated at four experimental fields belonging to Embrapa Gado de Corte in Campo Grande, MS, Brazil. The fields were evaluated at different times, being that field One was evaluated for three years, and the others for two years. The experimental design used was that of randomized blocks in three replications, with Marandu and Xaraés cultivars as witnesses. Genetic parameters estimated for the characters were: accumulation of dry mass (MST), accumulation of foliar dry matter (MSF), dry green matter (MSV), leaf percentage (F(%)), rebudding (REB) and vigor. It was concluded that two years of measurement are enough for selection, adopting a coefficient of determination of 80%.
genetic correlation; tropical forage; heritability; REML/BLUP; repeatability; genotypes
