Genetic Parameters for Milk Yield and Lactation Length of Crossbred Buffaloes from Brazil by Bayesian Inference

Abstract

The objective of the study was to estimate heritability and repeatability for milk yield (MY) and lactation length (LL) in buffaloes using Bayesian inference. The Brazilian genetic improvement program of buffalo provided the data that included 628 females, from four herds, born between 1980 and 2003. In order to obtain the estimates of variance, univariate analyses were performed with the Gibbs sampler, using the MTGSAM software. The model for MY and LL included direct genetic additive and permanent environment as random effects, and contemporary groups, milking frequency and calving number as fixed effects. The convergence diagnosis was performed with the Geweke method using an algorithm implemented in R software through the package Bayesian Output Analysis. Average for milk yield and lactation length was 1,546.1±483.8 kg and 252.3±42.5 days, respectively. The heritability coefficients were 0.31 (mode), 0.35 (mean) and 0.34 (median) for MY and 0.11 (mode), 0.10 (mean) and 0.10 (median) for LL. The repeatability coefficient (mode) were 0.50 and 0.15 for MY and LL, respectively. Milk yield is the only trait with clear potential for genetic improvement by direct genetic selection. The repeatability for MY indicates that selection based on the first lactation could contribute for an improvement in this trait.