Methods of estimation of variances and related parameters for growth traits in cattle

Abstract

The Iterative Henderson Simple Method (IHSM), Maximum Likelihood (ML) and Restricted Maximum Likelihood (REML) were evaluated, regarding genetic (g_ii) and residual (r_ii) variances and other genetic parameters estimates from univariate analysis of body weights at weaning and at 12 months of age in cattle. The results of several analysis were compared to those obtained by method 3 of Henderson (H3), traditionally used in the estimation of genetic parameters. IHSM, ML and REML converged with the same number of iterations, although REML demanded more computational time. All methods produced similar results for the residual (co) variances, but REML estimated higher values for g_ii, in both traits and consequently higher values for heritabilities and genetic correlations. Compared to REML, ML was similar and H3 was the most divergent with respect to the genetic parameters estimation.