Accuracies of Genomic Prediction for Growth Traits at Weaning and Yearling Ages in Yak

SIMPLE SUMMARY: Genomic selection is a new technology in animal breeding after the selection according to the best linear unbiased prediction (BLUP) value and marker assisted selection. Genomic selection has gradually been used in practical applications over recent years following the advent of high-density single nucleotide polymorphism (SNP) chips for livestock and poultry. Yak are a critical species on the Qinghai–Tibet Plateau, which is of great significance to herders. The early selection of yak could save feeding costs and shorten the generation interval. In the present study, we estimated the accuracy of genomic prediction compared with different classical models for yak early growth traits. The results of cross-validation indicated that the average predictive accuracy ranged from 0.147 to 0.391. The average correlation coefficient between prediction and true phenotype was 0.4. ABSTRACT: Genomic selection is a promising breeding strategy that has been used in considerable numbers of breeding projects due to its highly accurate results. Yak are rare mammals that are remarkable because of their ability to survive in the extreme and harsh conditions predominantly at the so-called “roof of the world”—the Qinghai–Tibetan Plateau. In the current study, we conducted an exploration of the feasibility of genomic evaluation and compared the predictive accuracy of early growth traits with five different approaches. In total, four growth traits were measured in 354 yaks, including body weight, withers height, body length, and chest girth in two early stages of development (weaning and yearling). Genotyping was implemented using the Illumina BovineHD BeadChip. The predictive accuracy was calculated through five-fold cross-validation in five classical statistical methods including genomic best linear unbiased prediction (GBLUP) and four Bayesian methods. Body weights at 30 months in the same yak population were also measured to evaluate the prediction at 6 months. The results indicated that the predictive accuracy for the early growth traits of yak ranged from 0.147 to 0.391. Similar performance was found for the GBLUP and Bayesian methods for most growth traits. Among the Bayesian methods, Bayes B outperformed Bayes A in the majority of traits. The average correlation coefficient between the prediction at 6 months using different methods and observations at 30 months was 0.4. These results indicate that genomic prediction is feasible for early growth traits in yak. Considering that genomic selection is necessary in yak breeding projects, the present study provides promising reference for future applications.