The objective of this work was to compare random regression models with different structures of residual variances, in the estimate of covariance components and genetic parameters for growth traits in goats. Regression functions using Legendre orthogonal polynomials of the fourth order were used for modeling animal growth trajectory, and polynomial of third order for modeling direct and maternal additive genetic effects, and permanent environmental effect. Different residual variance structures were considered using alternatively step functions (from one to seven different classes of residual variances) or variance functions using ordinary and Legendre polynomials, with orders of fit from the first to fourth order. Models were compared by likelihood ratio test, Akaike's information criterion and Schwarz's bayesian criterion. Models including residual variance functions showed better results than those including classes of variance. The ordinary polynomial of third order showed better results than the other models. Genetic parameters are affected by different residual variance structures in the models, however, genetic parameters estimates using four classes of residual variance, ordinary polynomial and Legendre polynomial of third order are similar.
covariance function; genetic parameter; Legendre polynomial; ordinary polynomial
