International Journal of Aquaculture, 2025, Vol.15, No.2, 76-87 http://www.aquapublisher.com/index.php/ija 85 7.2 Intelligent breeding: AI algorithms and big data decision-making support The development of modern information technology has injected new vitality into traditional breeding. The application of artificial intelligence (AI) and big data in abalone breeding will drive the breeding model from experience-driven to intelligent-driven. Artificial intelligence algorithms can be used to analyze complex breeding data, mine potential laws and optimize decisions. Some studies have introduced deep learning algorithms in GS to capture nonlinear genetic effects and gene interactions that are difficult to characterize in traditional linear models. Preliminary results show an improvement in the prediction accuracy of low hereditary traits. Big data integration analysis can assist breeding decisions. Abalone breeding involves genetic, environmental and management data. For example, by integrating the production big data of different breeding farms over the years, AI algorithms can identify which families perform better in specific environments (such as high temperature and high density ponds), thereby guiding the targeted promotion of the strains and achieving matching of "good varieties and good methods". For example, when using computer vision technology to monitor the feeding and exercise behavior of young abalone in the seedling cultivation stage, and to judge the health status early based on behavioral abnormalities, it will greatly improve the survival rate and selection intensity of breeding populations. If blockchain and other technologies are introduced into breeding data management, it can ensure that the breeding process data is authentic and traceable, making it easier for AI to learn more trustworthy information (Liu et al., 2022). The vision of intelligent breeding is to build a "digital twin" breeding scenario: simulate different assembly schemes and breeding environments in a computer, conduct virtual evaluations of each candidate scheme, and screen out most suboptimal schemes before real-life experiments. This will greatly compress the test cycle and cost. 7.3 Exploration of targeted breeding and personalized marine breeding Faced with the diversified market demands in the future and the complex and changeable breeding environment, abalone breeding is developing towards directional and personalization. The so-called targeted breeding refers to the targeted cultivation of varieties or strains that meet the target based on specific breeding goals and conditions. For example, in order to adapt to the high-temperature sea areas under the background of global warming, new heat-resistant abalone varieties can be selected in a directional manner; in order to meet the demand for super-large abalone in the high-end market, extremely fast-growing and large individual varieties can be cultivated in a directional manner; for the deep sea cage breeding model, abalone types that are resistant to waves and flows and have stronger adhesion can be selected. The implementation of targeted breeding is inseparable from the support of the aforementioned GS, gene editing and AI technologies. Genome selection allows us to accelerate selection at the molecular level for a target trait without waiting for complete multitrait improvement. Based on targeted breeding, the concept of personalized marine aquaculture has gradually emerged. Different farmers and industries may have different preferences for abalone varieties: some focus on growth rate, some emphasize the meat yield and flavor, and some focus on the ornamental shell color. In the future breeding system, breeding plans can be customized for the needs of the main categories. For example, for the leisure agriculture of "yard abalone" can provide ornamental products with gorgeous shell colors but slightly slow growth; for the processed dry abalone industry, it can provide special products with huge individual and dense fleshy texture suitable for sun drying. The cultivation of these personalized lines was difficult to take into account in the past due to resource constraints, and with the help of genomic technology, it was possible to advance multiple breeding programs in parallel (Arai and Okumura, 2013). Acknowledgements We would like to sincerely thank our colleagues for their support and help in the collation and discussion of the data. At the same time, we would like to thank the two review experts for their valuable revisions and suggestions on this article to make the article more perfect. Conflict of Interest Disclosure The authors confirm that the study was conducted without any commercial or financial relationships and could be interpreted as a potential conflict of interest.
RkJQdWJsaXNoZXIy MjQ4ODYzNA==