Calculation of Selection Indexes that Include Highly Correlated Traits

Abstract

In order to estimate breeding values for dairy bulls, the individual test day results of lactation were combined, using selection index theory. The method can in theory allow lactations of varying length (varying numbers of tests) to be combined and also the economic weighting of individual test days to differ. Some of the results of general interest from the aspect of selection index theory were presented. These concern the influence of genetic correlations among individual test days and their heritability on accuracy of index, variation in selection index weights, and expected response. Accuracy of index was not affected when the genetic correlations among the test days were altered or when the phenotypic variances differed. Individual b-values showed a wide variation and were especially affected by changes in the pattern of heritabilities for the individual test days. The percentage distribution of expected genetic response on individual test days was not affected by the wide variation in b-values in some alternatives. Altering the phenotypic variation of single traits had the greatest effect on the percentage distribution. As selection index theory assumes error-free parameters, the choice of parameters in any given situation is of prime importance. The commonly used rTI-value is not efficient for measuring the influence of variations in parameters.