Effects of different strategies for selection of animals as reference population on the accuracy of genomic evaluation for moderate heritability traits in dairy cattle

An important question about genomic evaluation is the effectiveness of using superior animals as reference population, on the accuracy of estimated breeding values of selection candidates. In this research, the accuracy of genomic evaluation is selecting superior animals as reference population (strategy1) was compared to situations in which the animals in reference population were a random sample of population (strategy2) and superior and inferior animals (strategy3). Best linear unbiased prediction method was used to estimate marker effects. The results showed that using only superior animals as reference population would decrease the accuracy of genomic evaluation. If the ratio of animals in the reference group is low (for example 10%) the difference between strategy1 and the other strategies would be more than the situation in which this ratio is high (for example 50%). For example in situation that the generation before validation set (generation four) was used as reference population, the accuracy of strategy1 was about 0.34 lower than strategy3 when 10% of animals were used as reference population but this difference was decreased to 0.04 when 50% of animals were used as reference population. These results showed that genotyping and using some of no superior animals in the reference population, beside to superior animals with high accurate traditional estimated breeding values, could lead to increase in the accuracy of genomic evaluation.