Bayesian inference in a quantitative genetic study of growth traits in Nelore cattle (Bos indicus)

ABSTRACT

The objective of this study was to estimate (co)variance components and genetic parameters for weights (W) and scrotal circumferences (SC) at 365 and 450 days of age, of Nelore (Bos indicus) cattle, using Bayesian inference in single and multiple-trait animal models. The fitted linear models included, besides the animal and residual random effects, the contemporary group (herd-year-season-sex-management group) and age-of-dam as "fixed effects". The analyses were carried out using a Gibbs sampler implemented through the MTGSAM program. The heritabilities (in parentheses) obtained fitting single-trait models were W365 (0.49), W450 (0.52), SC365 (0.68) and SC450 (0.66). Estimates of means, modes and medians for genetic parameters obtained from marginal posterior distributions were similar for all traits. The W365 and SC365 can be considered as suitable traits to be included as selection criteria in genetic improvement programs, not only because of their relatively high heritabilities but also due to the fact that they can be measured when the animals are relatively young compared to the corresponding traits W450 and SC450. The Bayesian approach appears to be an appropriate alternative for estimating genetic parameters, and has the advantage over point estimation methods of allowing inferences on marginal posterior distributions.