Deciphering the optimal genomic selection (GS) strategy of alkalinity tolerance trait in spotted sea bass (Lateolabrax maculatus)

Breeding strains tolerant to saline-alkaline conditions was a promising strategy for advancing the development and utilization of saline-alkali water areas. Spotted sea bass (Lateolabrax maculatus), with characteristics of broad salinity tolerance, was a favorable candidate for saline-alkaline tolerant strain. In this study, genotype data involved three markers (SNPs, InDels and SVs) and phenotype data from 287 individuals was used to construct the optimal genomic selection (GS) system. The results showed that significantly greater prediction accuracy was achieved using the GWAS-information mark selection strategy in comparison to the random selection strategy, particularly under 0.1 k density conditions (mean accuracy: 0.30 vs. 0.02). In terms of model, Bayesian models (mean accuracy:0.33) were more suitable than machine learning models (mean accuracy:0.29) for GS of alkalinity tolerance trait in spotted sea bass. Maximum predictive accuracies were attained with 100 SNPs, 600 InDels, and 12,800 SVs, where the BayesB model demonstrated consistently higher performance, yielding accuracies of 0.47, 0.41, and 0.53, respectively. Additionally, the use of combined markers resulted in higher prediction accuracy than any single marker type alone, particularly "SNP + InDel + SV" combination, which achieved an accuracy of up to 0.71. We have successfully established the GS system for alkalinity tolerance traits in spotted sea bass, providing valuable reference for the breeding of stress tolerance traits in other fish species.