Olive flounder (Paralichthys olivaceus) is a major marine aquaculture species in South Korea, with Jeju Island contributing over half of the national production. Although approximately 90% of farms still rely on moisture pellets (MP) made from wild-caught fish, this practice carries increasing risks due to unstable supply chains, rapid disintegration of MP in seawater, and elevated disease susceptibility. In response, the Korean government is promoting a transition to extruded pellets (EP), which offer advantages in storage stability, automation compatibility, and sustainability. However, the growth performance of fish fed EP remains inferior to that of fish fed MP, creating a substantial barrier to industry-wide adoption. This challenge underscores the need for breeding programs aimed at improving growth performance under EP-based feeding conditions.
To this end, we examined the genetic basis and genomic predictability of growth traits under two EP feeding regimes: a commercial high-fishmeal EP (70% fishmeal) and a low-fishmeal EP (35% fishmeal). In both conditions, 1,200 individuals were PIT-tagged and monitored for growth between June and October 2023, with genotyping performed using a custom 70K SNP array. Under the high-fishmeal condition, growth traits exhibited a predominantly polygenic architecture, with no significant SNPs detected by GWAS and a moderate heritability estimate (maximum h² = 0.38 for final weight). In contrast, the low-fishmeal condition revealed several significant SNPs associated with growth traits, suggesting the presence of diet-specific quantitative trait loci (QTL). Genomic prediction accuracies also varied across feeding regimes: the GBLUP model achieved the highest predictive ability (53.8%) under the high-fishmeal condition, while the Bayes B model performed best under the low-fishmeal condition. Moreover, incorporating the top SNPs identified through GWAS increased prediction accuracy by up to 79.5% in the high-fishmeal group.
Using genomic prediction models developed from both feeding-condition groups, we estimated the genomic estimated breeding values (GEBV) of broodstock candidates and selected elite individuals for the production of the F1 generation. To further enhance prediction accuracy for growth traits, a subset of the F1 generation is being established as a new reference population, with phenotypic measurements and genotyping currently in progress. This approach will enable comparative analyses of genetic parameters and model performance across generations and support the integration of F0 and F1 datasets to construct more accurate and robust prediction models tailored to feed-specific growth responses.
In summary, this study delivers key insights into diet-specific genetic responses and provides a practical framework for implementing genomic selection in sustainable fish breeding programs. These findings support the ongoing shift from MP to EP, offering a pathway toward increased productivity and environmental resilience in the Korean olive flounder industry.