Genome-wide association and genomic prediction of thermal tolerance in olive flounders (Paralichthys olivaceus): A validation study

Global warming poses a significant challenge to aquaculture, with increasing water temperatures adversely affecting the growth, reproduction, and survival of aquatic species. Thus, developing thermally-tolerant strains is essential for sustaining aquaculture in a changing climate. This study aimed to identify single-nucleotide polymorphisms (SNPs) associated with thermal tolerance in olive flounders ( Paralichthys olivaceus ) and evaluate genomic prediction models for breeding thermally-resilient strains. A thermal challenge experiment was conducted with 899 fish, during which survival data were recorded under temperatures ranging from 20 ± 0.3 ° C to 31 ± 0.3 ° C. Genotyping using a 70 K SNP chip yielded 55,752 high-quality SNPs from 765 samples after filtering. Genome-wide association study (GWAS) identified 204 significant SNPs across eight chromosomes (chr 5, 14, 17, 18, 19, 21, 22, and 24), with annotation linking 141 genes to thermal response mechanisms. Validation in an independent population confirmed 35 common significant SNPs. Among the ten genomic prediction models tested, random forest (RF) and Bayesian B (BB) models achieved the highest prediction accuracies, with the BB model demonstrating superior reliability for estimating genomic estimated breeding values (GEBV). Further, cross-validation indicated that most fish with high GEBV survived the thermal challenge. These findings underscore the potential of genomic prediction in developing thermally-tolerant olive flounder strains, offering a robust strategy to enhance aquaculture sustainability amid ongoing climate change. • A genome-wide association study and genomic prediction were conducted to assess thermal tolerance in olive flounders. • Validation study identified 35 common significant SNPs associated with thermal tolerance. • The selected thermal tolerance traits exhibited high heritability. • Random Forest and BayesB models achieved the highest accuracies for predicting thermal tolerance. • The BayesB model outperformed the GBLUP model in cross-validation, with higher gEBV linked to improve thermal resistance.