Análise genética de curvas de crescimento de bovinos da raça Brahman por métodos de duas etapas e de análise conjunta

Abstract

The objective of this study was to use two methods of genetic analysis of growth curve (two-step method and method based on a Bayesian hierarchical approach called in this study as joint analysis) under the non-linear model that best describes the growth of Brahman cattle, with ages ranging from birth to 24 months. Four nonlinear models were fitted and compared using the coefficient of determination (R2), residual standard deviation (RSD), Akaike information criterion (AIC) and Bayesian information criterion (BIC). The estimation of the variance component of the curve parameters by the two-step method was performed considering the restricted maximum likelihood method (REML), by using the BLUPF90 software. For the joint analysis method, Gibbs Sampling and Metropolis-Hasting were implemented in the R software. Comparative analysis based on the goodness of fit criteria revealed that the Gompertz model turned out to be the most suitable model. The heritability values estimated by the two-step method were 0.41, 0.09 and 0.65 for parameters a, b, and k, respectively. The corresponding values of the joint analysis method were 0.50, 0.51 and 0.52 for the parameter a, b and k, respectively. The heritability estimates tended to be higher by the joint analysis method than those obtained by the two-step method, except for the parameter k. The obtained results reveal that the two-step method presented a residual variance superior of the obtained by using the joint analysis method. Then, less precise estimates of the genetic parameters can be obtained with this first method.