Univariate and multivariate animal models for evaluation of Sahiwal sires using derivative-free restricted maximum likelihood method

Abstract

Data on 586 daughters of 50 Sahiwal sires maintained at National Dairy Research Institute, Karnal over a period of 47 years (1961–2007) were used to estimate the expected breeding values of sires by univariate and multivariate animal models using DFREML method. The first lactation 305-day or less milk yield (FL305DMY) was considered as the principal trait. For multi-trait evaluation, two sets of data were used. The first set consisted of first three individual monthly part lactation records along with FL305DMY and the second set was with the combination of different cumulative milk yields up to 90-days along with FL305DMY. The overall average expected breeding value for FL305DMY was estimated as 1800.08 kg. The error variance of single trait animal model was lower than multiple trait animal models and the relative efficiencies of multi trait DFREML for data set I and II were 77.08 and 65.52%, respectively. The rank correlation between the breeding values for first lactation 305-day or less milk yield of single trait DFREML and multi trait REML methods were very high (0.9580 and 0.9726 for data set I and II, respectively) and statistically highly significant (P<0.01) indicating that the evaluation of sires by both the models gave similar ranking to the extent of 96%. It was concluded that the single trait DFREML method was more efficient than the multi trait model for estimation of breeding value of Sahiwal sires for first lactation 305-day milk yield.