Measuring the effect of change in selection indices

Genetic selection goals for dairy cattle, originally aimed at production traits only, have been expanded in stages over the past 30 yr to include up to 12 target traits covering production, functionality, and health and fertility. Each addition to the selection goal often involves the use of additional measured phenotypic variates. The net effect of these additions is usually described as causing change in the relative emphasis on different traits, though there are varying definitions of what this means. This paper suggests that the current definitions of this term may be inappropriate and shows that, as usually used, it tends to overstate the net effect of the changes. A new definition of the relative emphasis of each target trait is proposed. It is defined as the percentage of total economic value of genetic gain in all traits attributable to gain in that particular trait. A useful parallel statistic measures the relative contribution of each phenotypic variate recorded. The result of applying these measures is contrasted with the use of current methods using United States Holstein data.