An evaluation of the method of dairy bull dam selection in Kenya

Abstract

5670 lactation records of 1,701 Ayrshire cows from 18 officially milk recording herds, made between 1990 and 2004, were used to evaluate the efficiency of the current bu II dam selection method in Kenya. The data was sorted using the current criteria for selection, which states that; the dams must be registered by the Kenya stud book, must be officially milk recorded by the dairy recording services of Kenya, must have had at least three lactations of at least 199 days and must have been served by artificial insemination. Type and conformation traits are also considered. A univariate animal model with relationships and the Derivative Free - Restricted Maximum Likelihood (DF-REML) procedure was used to predict the breeding values (BVs) of the cows for milk yield. The cows were then ranked on the basis of their BVs. The BVs for milk yield of the top 100 cows ranged from + 550Kg to + 1115Kg. Out of the 113 bull dams included in the study, only 25 were among the top 100 cows. The records were then adjusted for systematic environmental factors and the same procedure repeated to obtain BVs which ranged from + 127Kg to +71 OKg. In this case only 8 of the 113 bull dams included in the study were ranked among the top 100 cows. A rank correlation was computed between the 2 rankings of the BVs for milk yield, and the correlation was found to be + 0.057. Further analysis was done to determine the ranking of the cows according to their BVs for milk yield, using the reduced and unreduced animal model. A correlation coefficient of +0.079 was obtained. These results show that the current bull dam selection method is inaccurate based on analysis using milk yields. Consequently, genetically superior cows were left out in the contract mating scheme. This is expected and is largely the reason for the low rank correlations. This study recommends the use of Best Linear Unbiased Prediction with the unreduced animal. model for use in the evaluation of breeding values.