A new selection index percent emphasis method using subindex weights and genetic evaluation accuracy

The current international standard methodology to quantify trait percent emphasis in selection indexes is based on a simple multiplication of the relative contribution of each trait's economic value (converted to absolute value) and its genetic standard deviation. This method does not reflect the actual selection emphasis applied when the index is used in practice. The economic value does not reflect selection effort when traits differ considerably in their accuracy of evaluation, and no account is taken for either favorable or antagonistic correlations among traits. A new emphasis method adjusted by both accuracy and genetic correlation is proposed. Genetically highly correlated traits are grouped into subindexes by applying a hierarchical clustering method to the genetic correlation matrix. Then each trait's subindex emphasis is calculated within its subindex group, with a weighting included for trait accuracy. Finally, each subindex emphasis is converted to a full index emphasis according to the conventional relative emphasis of its corresponding subgroup. The method can also be applied to sets of breeding values and their economic weights. When applied to a New Zealand sheep breeding selection index where trait genetic correlations are distinct across subindex groups, the new method shrank the emphasis on low-heritability traits, including survival, from 51% to 19%; and expanded that on growth traits from 30% to 49%, better reflecting the selection pressure applied in reality. When genetic correlations across traits were similar, clustering became difficult. Accounting for accuracy affected traits' within-subindex group rankings, whereas the clustering to account for correlations affected all traits within a subgroup equally. Accounting for differences in trait accuracy when describing percent emphasis within selection indexes gives a more practical indication of the likely outcome of selecting on the index. Accounting for correlations among traits when defining percent emphasis made a significant difference only in a subset of case study examples.