Implementation of genomic selection at national level: impact of pre-selection and biased national BLUP evaluations on international genetic evaluations

Abstract

It has been shown that genomic pre-selection of young bulls leads to biased national BLUP evaluations if information of culled animals is not accounted for. The objective of this study was to assess the consequences of such missing and/or biased national data on international genetic evaluations. Various genomic selection scenarios were simulated in 3 actual populations participating in Interbull evaluations. They were first simulated separately to clearly understand how bias is propagated. Then, they were combined to illustrate a more realistic case. The current international genetic evaluations when national proofs are available for all candidates and supposed to be unbiased were used as a reference. Hence, bias was measured among young sires and by country of origin as the average difference between the current evaluations and the ones obtained in the simulated scenarios. Bias due to missing information on the culling process at national level or due to national biased proofs in one country highly penalized young sires from that country. But it also had an impact on the evaluation of foreign young and older sires. Moreover, it becomes more difficult if not impossible to predict this impact when different sources of bias were combined from different countries: all effects interact with each other. But a change in ranking is certain so that selection efficiency becomes clearly suboptimal.