Comparison of sire evaluation methods in Holstein Friesian in the central highlands of Ethiopia

Abstract

First lactation records (387) of Friesian-Boran crossbred cows, daughters of 37 Holstein Friesian (HF) sires were analyzed. Holstein Friesian sires were evaluated based on breeding values estimated considering first lactation 305- days milk yield of their progeny using 4 sire evaluation methods, viz. contemporary comparison (CC), least squares method (LS), best linear unbiased prediction (BLUP) and derivative free restricted maximum likelihood (DFREML). The effectiveness of the methods was judged using various criteria, viz. error variance, coefficient of determination, coefficient of variation and rank correlation. The highest (1781.1 kg) and lowest (1766.78 kg) average breeding values were obtained by CC and DFREML methods, respectively. DFREML had the smallest (197061 kg2) error variance and highest (60%) R2 value and was adjudged as the most efficient and accurate method compared with others. The CV values computed by different methods were similar indicating lack of meaningful variation amongst the methods as far as stability is concerned. The rank correlations between the ranks of HF sires evaluated by different methods were high (greater than 0.93), indicating that ranking of sires using any of these methods could yield similar order. The implication of the result is that, based on availability and computational suitability of the method, one can use any of the methods in sire evaluation under the Ethiopian condition.