Homogeneity and heterogeneity of variance components for milk and protein yield at different cluster sizes in Iranian Holsteins

Abstract

This study was conducted to investigate the homogeneity and heterogeneity of variance components for milk yield (MY) and protein yield (PY) at different cluster sizes and to evaluate effect of different data transformation methods on the ranking of elite animals in Iranian Holsteins. Data included 305-day lactation records for MY and PY from the first three lactations that were collected in the period from 1983 to 2014 by the Animal Breeding Center and Promotion of Animal Products of Iran. The mean and standard deviation of 305-day MY and PY for each herd were obtained and used as the basis for clustering the herds. Four different data transformation methods including Logarithmic, Arc sin, Square root and Box-Cox were applied and the data were tested for heterogeneity of variance before and after transformation using Bartlett's test. Genetic parameters and heritability estimates as well as animals’ breeding values (BVs) were estimated by the VCE program, using animal model. The results indicated the heterogeneity of variance in all clusters (P<0.01) before transformation. All methods of data transformation decreased the heterogeneity of variance components for MY and PY in the first three lactations and some of these methods caused removing the heterogeneity of variance. Results showed that the heritability estimates of MY and PY were decreased from the first lactation to third, also heritability estimates of transformed data were slightly higher than the original data ones. Some re-ranking of animals occurred from the transformation, but the Box-Cox method only had slight effects on the overall ranking and rank correlations of animals. The applied data transformation caused a substantial re-ranking of elite sires and dams in the EBV for each cluster. This study provided evidence that the applied transformation methods reduced the bias in the EBVs due to heterogeneous variances and increased the accuracy of bull-dam selection. Therefore, effects of heterogeneous variances need to be accounted for in genetic evaluation of Holstein cows in Iran.