A simple method for correcting the bias caused by genomic pre-selection in conventional genetic evaluation

With the development of genomic evaluation in dairy cattle, an ever increasing number of countries have been pre-selecting genotyped calves as parents of animals of future generations with a much higher selection intensity than in the past. If this genomic pre-selection is ignored in conventional genetic evaluation, proofs will be biased. We developed a simple method to correct the pre-selection bias in conventional genetic evaluation. In order to validate our bias correction method, we simulated genomic pre-selection four years ago with three levels of culling rate and conducted test genetic evaluations with and without adjusting for the pre-selection bias. A total of 655 genotyped Holstein bulls, born in 2004, were treated as if they were genotyped selection candidates four years ago. Their official milk yield proofs from August 2009 German official conventional evaluation were removed from further analyses and these bulls were selected or culled based on their estimated direct genomic values. Results of the test evaluations were compared to the reference bull evaluation with about 24,000 Holstein bulls included. For all levels of culling rate, the simple bias correction method gave much smaller averages and standard deviations of proof biases. In particular, proofs of dams of the culled genotyped bulls were extremely overestimated when the pre-selection was ignored in the test evaluations. Overall, the proposed bias correction method seemed to be very efficient. Some aspects on further development were also discussed. The genomic pre-selection problem exists for international bull comparison, too. The proposed method can be easily implemented in both national and international genetic evaluations.