Accuracy of genomic prediction using two admixed crossbred populations

Abstract Text: Accuracy of predicting crossbred performance within and across Montbéliard x Charolais and Holstein x Charolais claves was studied. Accuracies obtained with BLUP and GBLUP were compared within population. GBLUP showed higher accuracies ranging from 0.386 to 0.475. Genomic prediction within one crossbred population tended to give the highest accuracies. Genomic prediction across crossbred populations had the lowest accuracy as compare to genomic prediction within one crossbred population, with a decrease from 12.8 to 27.2%. This decrease in accuracy was due to different effect of the Charolais genes when the dam is Montbéliard or Holstein. For crossbred populations having one common parental breed, genetic correlation can be estimated from pedigree information. This estimation was suggested as an indication to assess the possibility to perform genomic prediction across crossbred populations and the gain in accuracy when pooling different crossbred populations into one reference population.

Keywords: crossbred, admixed, genomic selection