Sheep & Goats

Refining U.S. Suffolk genetic evaluations using updated models and genomics for weight, carcass, and reproductive traits

---

Abstract

The Suffolk is widely used as a terminal sire sheep breed around the world due to its superior growth rate and carcass traits. In the U.S., current National Sheep Improvement Program (NSIP) genetic evaluations for this breed rely on historic models and genetic parameters established over two decades ago, and do not integrate genomic information. Therefore, the main objectives of this study were to propose updated variance component estimates and statistical models for the NSIP genetic evaluation of U.S. Suffolk sheep, incorporating alternative fixed and random effects, and to assess the benefits of incorporating genomics by evaluating predictive performance through a forward validation study. The traits considered included birth weight (BWT; n=51,414), weaning weight (WWT; n=36,303), post-weaning weight (PWWT; n=24,841), post-weaning ultrasound eye muscle (PEMD; n=7,663) and fat (PFD; n=7,470) depth, and number of lambs born (NLB; n=28,994) and weaned (NLW; n=28,994). The pedigree consisted of 75,271 animals. WWT and PWWT were pre-adjusted to a target age at recording within sex. Contemporary groups (CG) with fewer than five animals with records were removed. The best combination of fixed effects was chosen based on ANOVA significance and biological interpretation. Inclusion of covariance between direct and maternal additive genetic effects was tested. Linear Regression (LR) metrics for forward validation (accuracy, bias, and dispersion) were computed using animals recorded in the last three yr (focal individuals) for each trait. Variance components were estimated by fitting the A matrix for pedigree-based and H matrix for genomic-based evaluations (1,954 animals with 41,782 single-nucleotide polymorphisms after quality control). The BLUPF90+ software was used for the analyses and theoretical accuracies (TAcc) were calculated from the prediction error variances. Fitting CG as a random effect, even with flock-year included as a fixed effect, did not improve model fit or predictive performance compared with treating CG as a fixed effect, based on LR metrics. However, including the covariance between direct and maternal effects improved the fit and predictive performance for all body weight traits. Despite the relatively small number of genotyped animals, the inclusion of genomic information increased LR accuracy across all traits while maintaining comparable bias and dispersion values, highlighting the added value of genomic data for the NSIP Suffolk genetic evaluation. Except for NLB and NLW, the re-estimated variance components were significantly different from those previously reported for Suffolk sheep, likely reflecting the inclusion of different fixed and random effects and historic adjustment factors used in earlier analyses (Table 1). Genomic data minimally affected the variance component estimates; however, average TAcc improved across all traits and from 0.32 to 0.34 for NLB with the H matrix, which is substantial given the limited genotyping.

Keywords: 2026

How to Cite:

Rocha, A., Saleem, A., Yazar Gunes, H., de Carvalho, F., Wilson, C., Freking, B., Murphy, T., Burke, J., Lewis, R. & Brito, L., (2026) “Refining U.S. Suffolk genetic evaluations using updated models and genomics for weight, carcass, and reproductive traits”, World Congress on Genetics Applied to Livestock Production Digital Archive 2026(1): 2286658. doi: https://doi.org/10.31274/wcgalp.24083

Rights: 1