IJA_2026v16n5

International Journal of Aquaculture, 2026, Vol.16, No.5, 378-396 http://www.aquapublisher.com/index.php/ija 384 intelligent strategies while monitoring water quality, thereby avoiding overfeeding and pond environmental deterioration (Katta et al., 2025; Puangsuwan et al., 2025). Digital monitoring technology can optimize feeding amount, feeding time, and feeding frequency by combining water temperature, dissolved oxygen, weather conditions, growth stage of cultured organisms, and feeding behavior. For example, feeding can be appropriately increased when water temperature is suitable, dissolved oxygen is sufficient, and feeding activity is high; under low oxygen, cooling, heavy rainfall, or abnormal water quality conditions, feeding should be reduced or suspended to avoid increasing the environmental burden of the pond. Deep reinforcement learning has been applied to the joint optimization of feeding rates and water quality in recirculating aquaculture systems, helping reduce feed consumption, improve system stability, and generate economic benefits (Elmessery et al., 2025). Edge artificial intelligence and deep Q-network-based frameworks can further coordinate feeding, aeration, and water exchange, improving feed conversion efficiency and reducing energy consumption (Katta et al., 2025). In ecological pond scenarios, the integration of intelligent aeration and feeding management helps match energy input and feed input with ecosystem carrying capacity, thereby promoting pond aquaculture toward energy-saving, efficient, and environmentally friendly development. 3.3 Disease early warning and production process traceability Disease prevention and control are key and difficult aspects of ecological pond aquaculture management. Pond aquaculture diseases are often closely associated with water quality deterioration, excessive stocking density, improper feeding, abrupt temperature changes, and stress responses of cultured organisms. Traditional disease prevention and control mainly rely on operators observing the activity, feeding, and mortality of fish and shrimp, and treatment measures are often taken only after disease has already occurred, which can easily miss the optimal prevention and control window. Through continuous collection of water quality, meteorological, equipment operation, and production management data, digital monitoring technology can identify environmental abnormal signals before disease occurrence and provide a basis for early disease warning. Relevant studies have shown that by combining continuous water quality data, such as temperature, dissolved oxygen, pH, salinity, and nutrients, with disease surveillance records, machine learning models can identify environmental risk patterns and predict outbreaks of specific parasitic or other diseases (Islam et al., 2024; Xie et al., 2025; Rohini et al., 2026). In practical applications, when the system detects conditions such as continuously low dissolved oxygen, abnormal increases in ammonia nitrogen or nitrite, large pH fluctuations, or sharp changes in water temperature, it can promptly send early-warning alerts to aquaculture operators. Operators can then inspect pond conditions according to the alarm information and take measures such as improving water quality, adjusting feeding, strengthening aeration, applying microbial preparations, or conducting pathogen detection, thereby reducing the risk of disease outbreaks. For example, Xie et al. (2025) integrated multi-year disease outbreak data with high-resolution environmental predictors to construct an early-warning system for cryptocaryoniasis in marine fish, and provided weekly updated spatially resolved prediction information to farmers through a web platform. More general smart aquaculture systems can also apply machine learning or deep learning to image and sensor data to identify abnormal fish behavior or lesion characteristics, providing early warnings for disease intervention and reducing mortality and economic losses (Islam et al., 2024; Rohini et al., 2026). Some digital systems can further combine video monitoring and feeding behavior analysis to assist in judging whether cultured organisms show abnormal activity, reduced feeding, or floating head aggregation, thereby further improving the accuracy of disease risk identification. Digital monitoring can also be used for production process traceability management. By using data platforms to record seedling sources, stocking time, feed input, medication use, water quality changes, equipment operation, disease treatment, and product harvest information, relatively complete aquaculture records can be formed. This not only helps operators summarize production experience and analyze the causes of problems, but also helps standardize aquaculture operations and improve the quality and safety management of aquatic products. AI-IoT-blockchain schemes can record water quality, feeding, growth, and disease-related information in tamper-resistant ledgers, enabling end-to-end traceability from seedling stocking to harvest, and supporting the

RkJQdWJsaXNoZXIy MjQ4ODYzNA==