Sire evaluation based on predicted first lactation 305-day milk yield using monthly test day milk yield values of Karan Fries cattle

Abstract

The present investigation was carried out using data of first lactation monthly (MTDY) test day milk yield records of 1345 Karan Fries cows sired by 84 bulls from 1984–2006 maintained at the NDRI, Karnal. The breeding values of Karan Fries sires were estimated for predicted first lactation 305-day by step-wise backward multiple regression using monthly test day milk yields. The breeding values were estimated by 4 methods, viz. least squares method (LSQ), simple regressed least squares method (SRLS), contemporary comparison method (CC) and best linear unbiased prediction (BLUP). The effectiveness of LSQ, SRLS and BLUP methods of sire evaluation was compared on the basis of rank correlations, error variance, coefficients of determination (R2) and coefficients of variation (CV). The rank correlation for LSQ with SRLS (0.92–0.99) was highest followed by LSQ with BLUP (0.89–0.91) and SRLS with BLUP (0.81– 0.91) which were very close. The lowest rank correlation was observed in LSQ with CC (0.10–0.16) and SRLS with CC (0.10–0.17). It was observed that all the rank correlations between CC with BLUP were negative. The relative efficiency of LSQ and SRLS in comparison with BLUP was more than 95%. The coefficient of determination (R2) for LSQ and SRLS was nearly equal (16.30% to 17.58%) and BLUP had lower R2 (13.10% to 14.75%) than LSQ and SRLS. The narrow range of coefficient of variation (CV) for LSQ, SRLS and BLUP (23.5% to 24.5%) and small deviation from CV of unadjusted data (25.35%) indicated that all the 3 methods were almost equally stable. The BLUP method which had lower error variance was found to be more efficient than the LSQ and SRLS methods using the 305-day milk yield predicted by regression equation considering fourth, sixth and ninth MTDYs.