Modelo hier????rquico bayesiano aplicado na avalia???§???￡o gen????tica de curvas de crescimento de bovinos de corte

摘要

The estimation of the (co)variance components for the parameters of the growth models can be evaluated by many methods. The Bayesian approach is an alternative method of the estimation. A study was performed using simulated and real data from Nelore cattle for estimation of the (co)variance components for the parameters of Von Bertalanffy growth curve, using a bayesian hierarchical model. From the estimated components, the heritabilities for each parameter and genetic and environmental correlations between these parameters were determined. The samples of posterior marginal distributions for the parameters a, R, , u, G, and 2 e were obtained by using Gibbs Sampler algorithm and for the parameters b e k by using the Metropolis-Hastings algorithm. The efficiency of the bayesian inference methodology was verified since estimated parameters were quite close to the simulated ones. The parameters a and k from real data showed heritabilities compatible with the reality indicating they could be used in selection programs.