Description
Cannibalism is one of the most important behavioral problems in aquaculture and contributes significantly to fingerling mortality and to the productivity and profitability of catfish farming activities worldwide. Among the species cultivated, African catfish shows particularly strong cannibalism in the early stages of its life, driven by hunger-induced aggression, a diversity of stocking sizes, and overpopulation. Traditional mitigation strategies—including manual size grading, feeding optimization, and photoperiod manipulation - have proven ineffective because they cannot deal with the dynamic, multi-factorial nature of cannibalistic triggers. This paper presents a comprehensive review of how Internet of Things (IoT) technologies can reduce cannibalism and stabilize the ecosystem of aquaculture through real-time environmental monitoring, automatic feeding, and closed-loop behavioral control. Drawing on a mixed-method study involving interviews with 12 catfish farmers and a structured survey of 48 IoT experts in Kenya, complemented by a simulation-based computational experiment using Wokwi and ThingSpeak, the research shows that cannibalism is primarily a behavioral risk and not an inherent biological trait and is not a major cause of mortality. This behavioral risk is exacerbated by environmental stressors, in particular low dissolved oxygen (2.45 mg/ml on average, dropping to 0.93 mg/ml under manual controls) and increased ammonia (β = -8.234, p < 0.001), which are the most potent negative predictors in the regression model (R² = 0.846, RMSE = 0.1278). The proposed seven-layer model of the IoT ecosystem including: input sensing, edge control, automation and activation, mitigation mechanisms, connectivity, cloud data management and user notifications; reduced simulated mortality from 59.7 percent to 22.5 percent under controlled conditions and increased survival from 403 to 775 per 1,000 fish. Automated, time-oriented feeding tailored to growth stages and biomass reduced food competition and the associated aggressive behavior, directly addressing the primary trigger for cannibalism identified by the farmers. A standardized beta sensitivity analysis identified dissolved oxygen (SI = 0.637) and ammonia (SI = 0.444) as the most predictive survivability variables, followed by temperature (0.370), pH (0.280), turbidity (0.263), and water level (0.125).
| Title | IoT-Driven Precision Aquaculture: Mitigating Cannibalism and Stabilizing Ecosystems in Catfish Fingerling Production |
|---|