Performance of whole genome prediction for growth traits in a crossbred chicken population

In the past decades, crossbreeding has been widely used to improve productivity in plant and animal husbandry. With the rapid implementation of genomic selection (GS) in these industries and a decrease in the cost of genotyping, genomic prediction (GP) with data from crossbred populations is an emerging research interest. Using a crossbred population derived from a cross between White Recessive Rock (WRR) and Xinghua (XH) chickens (n = 473), the predictive ability and selection differential of conventional best linear unbiased prediction (BLUP) and 3 GP methods (GBLUP, RKHS, and BayesB) were compared. All chickens were genotyped by a 60K SNP chip. Twenty traits containing body weight (BW) at 1 to 90 d of age, breast muscle weight (BMW), leg muscle weight (LMW), wing weight (WW), and average daily gain (ADG) of different periods were analyzed. The accuracy of GP was higher than that of conventional BLUP for 18 out of 20 investigated traits. The average selection differential on BW selected with GP methods was greater than that from conventional BLUP, with a proportion selected varied between 5 and 30%. Overall, the GP methods outperformed conventional BLUP for both predictive ability and selection effect in the tested crossbred chicken population. Using genomic data from crossbred populations could potentially benefit the decision making for the purpose of marketing or breeding within crossbred population.