Bayesian inference for genetic parameter estimation on growth traits for Nelore cattle in Brazil, using the Gibbs sampler

This data set consisted of over 29 245 field records from 24 herds of regis tered Nelore cattle born between 1980 and 1993, with calves sired by 657 si res and 12 151 dams. The records were collected in south-eastern and midwes tern Brazil and animals were raised on pasture in a tropical climate. Three growth traits were included in these analyses: 205- (W205), 365- (W365) an d 550-day (W550) weight. The linear model included fixed effects for contem porary groups (herd-year-season-sex) and age of dam at calving. The model a lso included random effects for direct genetic, maternal genetic and matern al permanent environmental (MPE) contributions to observations. The analyse s were conducted using single-trait and multiple-trait animal models. Varia nce and covariance components were estimated by restricted maximum likeliho od (REML) using a derivative-free algorithm (DFREML) for multiple traits (M TDFREML). Bayesian inference was obtained by a multiple trait Gibbs samplin g algorithm (GS) for (co)variance component inference in animal models (MTG SAM). Three different sets of prior distributions for the (co)variance comp onents were used: flat, symmetric, and sharp. The shape parameters (nu) wer e 0, 5 and 9, respectively. The results suggested that the shape of the pri or distributions did not affect the estimates of (co)variance components. F rom the REML analyses, for all traits, direct heritabilities obtained from single trait analyses were smaller than those obtained from bivariate analy ses and by the GS method. Estimates of genetic correlations between direct and maternal effects obtained using REML were positive but very low, indica ting that genetic selection programs should consider both components jointl y. GS produced similar but slightly higher estimates of genetic parameters than REML, however, the greater robustness of GS makes it the method of cho ice for many applications.