International Journal of Marine Science, 2025, Vol.15, No.5, 268-276 http://www.aquapublisher.com/index.php/ijms 272 Figure 2 QTL mapping and LOD profiles of growth-related traits in abalone among 18 linkage groups: (A) total weight, (B) shell length, and (C) shell width. The red solid line indicates the chromosome-wide significance threshold (Adopted from Kho et al., 2021) 5.3 Statistical models and prediction approaches GS employs advanced statistical models to estimate genomic estimated breeding values (GEBVs). A few of the popular models that are often employed include genomic best linear unbiased prediction (GBLUP) having equal marker effects and Bayesian methods (BayesA, BayesB, BayesR) where marker effects are different and prior knowledge is used. More recently, machine learning (ML) and deep learning (DL) techniques—e.g., support vector regression, random forests, convolutional neural networks, and ensemble learning—have also found potential to manage complex non-additive genetic structures and improve prediction precision of low-heritability traits in abalone (Abdollahi-Arpanahi et al., 2020; Wang et al., 2024; Su et al., 2025). 5.4 Research progress and case studies of genomic selection in abalone Recently, research on abalone has demonstrated the utility and value of GS. As an example, GS was employed to improve heat resistance in Pacific abalone with moderate heritability and higher prediction accuracy through Bayesian models than GBLUP. GWAS-identified SNPs also improved prediction accuracy. The results show the potential of GS to speed up genetic improvement for growth and stress resistance characteristics in abalone breeding programs (Liu et al., 2022). While most GS applications in aquaculture remain to be seen, coupled high-density genotyping and advanced prediction models are poised to open the door toward practical implementation in abalone. 5.5 Integration modes of genomic selection with traditional breeding GS may be combined with traditional family and population selection to achieve optimal genetic gain. Breeders can utilize the benefits of phenotypic and genomic information through the use of GEBVs as a second-tier selection aid. Hybrid approaches, such as GWAS-enhanced GBLUP and multi-trait models, improve selection accuracy and efficiency. The application of GS reduces intervals for generations, increases the intensity of selection, and enables the identification of high-quality broodstock at a younger age, complementing and augmenting traditional breeding schemes (Zhang et al., 2023). 6 Key Issues and Challenges in Implementing Genomic Selection in Abalone 6.1 Limitations in phenotypic data quality and quantity Effective genomic selection (GS) for abalone requires large high-quality phenotypic data. Phenotyping is costly,
RkJQdWJsaXNoZXIy MjQ4ODYzNA==