Comparison of different sire evaluation methods for production and reproduction traits in Frieswal cattle

Abstract

A total 1249 performance records of 470 Frieswal cows, progeny of 71 sires maintained at three Military Farms viz. Guwahati, Dimapur and Bengdubi of Eastern Command during 2000-2012 were utilised for sire evaluation. Sire breeding values were estimated  for lactation length (LL), 305-day milk yield (305-DY) and calving interval (CI) using various methods viz. contemporary comparison (CC) method, least squares (LS), best linear unbiased prediction (BLUP) and average information restricted maximum likelihood (AIREML) in terms of accuracy and efficiency. The age at first calving (AFC) was considered as covariable. LS, BLUP and AIREML models were constructed using univariate models for each trait and multivariate repeatability models (based on all lactation records) involving these three traits.

For LL, BLUP had the least error variance of 1770.43 day² and thus considered as the most efficient method of sire evaluation. For 305-DY, AIREML univariate model (A2) had the lowest error variance of 462054.15 kg² and hence it was judged as the most efficient method. For CI, BLUP model had lowest error variance of 3875.42 day² and thus it was inferred as the most efficient method for sire evaluation. For breeding value estimation of LL, BLUP multivariate model B13 had the highest R² value of 66.74 % and hence it was considered the most accurate model. For 305-DY, BLUP univariate model (B2) was found to be the most accurate model with highest R² value of 57.54%. For CI BLUP multivariate model (B123) had the highest accuracy (R² value of 47.67 %). 305-DY is the primary trait of selection. So, for selection and breeding value estimation, the recommended model is BLUP model. However, inclusion of additional traits will lead to change the R² value.