GBLUP-Based Genomic Prediction of Body Measurements and Weight Traits Using High-Density SNP Data in Awassi Sheep

Abstract

This study evaluated genomic variance components, genomic heritability, fixed effects, and the predictive performance of genomic best linear unbiased prediction (GBLUP) for body length, chest depth, heart girth, withers height, and body weight in 553 Awassi sheep genotyped using a high-density single nucleotide polymorphism (SNP) panel. Following quality control, 515,358 SNPs were retained to construct the genomic relationship matrix (GRM). Genomic heritability estimates varied across traits, ranging from 0.221 for body weight to 0.733 for body length, with comparatively high estimates also observed for heart girth and withers height. Birth type, animal age, and sex significantly affected all evaluated traits, whereas parity did not. Repeated five-fold cross-validation demonstrated low-to-moderate within-population genomic predictive ability, with the highest predictive performance observed for body length and withers height and the lowest for body weight. GBLUP consistently improved phenotypic prediction compared with fixed-effects-only models, although reductions in prediction error were modest. Prediction accuracies derived from full-data prediction error variance (PEV) were higher than cross-validated predictive ability. They were therefore interpreted as model-based accuracy measures rather than independent validation results. Overall, the findings demonstrate that genomic information can contribute to the prediction of body measurements and weight traits in Awassi sheep. However, broader implementation in genomic selection programs will require larger, multi-flock reference populations and independent or relationship-blocked validation to establish the robustness and transferability of predictive performance.