Genomic selection studies in sheep using simulation and real data

Abstract

This dissertation evaluates genomic selection strategies and genotype-by-environment modeling to improve genetic gains through sustainable breeding practices in sheep. It integrates a thorough review of sheep breeding programs in Turkey, a forward-looking simulation study of genomic selection in Akkaraman sheep, and an analysis of empirical data on genotype-by-environment interactions for parasite resistance in U.S. Katahdin sheep. The first study summarizes the status of genetic improvement in Turkey, one of the world's largest sheep producers. Akkaraman and Morkaraman are two local breeds that have adapted well to harsh environments. However, the review found that most breeding programs still rely on phenotype-based selection, with limited pedigree recording and no use of genomic selection. The review emphasizes the urgent need to modernize the national breeding infrastructure by developing reference populations, implementing routine genotyping, and employing community-based selection strategies that utilize genomic tools. Building upon these findings, the second study of this dissertation employs a stochastic simulation framework to evaluate the feasibility and long-term genetic consequences of implementing genomic selection in the Akkaraman breed in Turkey. Simulated breeding schemes replicated realistic flock structures, including 2,000 ewes and 200 rams per generation, 50,000 single-nucleotide polymorphism genotypes, and 300 underlying quantitative trait loci. Genomic selection yielded up to 77% more cumulative genetic gain compared to phenotypic selection. However, it also resulted in a rapid reduction of additive genetic variance and a measurable increase in inbreeding over nine cycles. Incorporating ram rotation among sub-populations mitigated these risks by maintaining genetic connectedness and preserving diversity. The simulation demonstrates that genomic selection is both operationally feasible and highly beneficial for Turkish sheep improvement provided that strategies for long-term genetic sustainability are implemented alongside. The third study examined genotype-by-environment interactions for gastrointestinal parasite resistance in Katahdin sheep raised in three ecologically distinct U.S. production zones. The study applied univariate and bivariate single-step maximum restricted likelihood methods to estimate genetic parameters within and between eco-clusters. These clusters were defined using NASA POWER climate data. The study used a dataset of more than 33,000 fecal egg count records and a pedigree of 127,000 animals. It also used imputed genotypes for 32,000 single-nucleotide polymorphism markers. Heritability estimates for log-transformed fecal egg count ranged from 0.17 to 0.34 across eco-clusters. Genetic correlations between eco-clusters ranged from 0.61 to 0.89, indicating the presence of genotype-by-environment interactions and genotype re-ranking, particularly at the weaning stage. These results underscore the importance of environment-specific genetic evaluations for optimizing breeding outcomes of traits influenced by climate and grazing conditions.