Application of Adaptive Neuro-Fuzzy Inference System to carrying capacity assessment for cage fish farm in Daya Bay, China

Marine cage fish farming has grown dramatically during the last three decades in coastal counties worldwide, however, its adverse environmental impact has already led to growing concerns. The control of carrying capacity is a key problem for cage fish farming. Based on the fish stock and eco-environment survey data in a cage fish farm area and its vicinity in Dapeng Ao Cove, Daya Bay, South China from June 2001 to October 2004 on a seasonal basis, an Adaptive Neuro-fuzzy Inference System (ANFIS) is used to learn and model the nonlinear relationships among the fish stock, surveyed biological and physicochemical factors. The ANFIS well simulate the effect of fish stock on the environmental factors of dissolved oxygen in water (DO), organic carbon content in sediment (SOC) and sulfide content in sediment (SSC). The computed theoretic maximum carrying capacity values were about 389~545t in different season within the restrict criteria of DO > 5mg/L, SOC <; 3% and SSC <; 600mg/kg. The actual fish stock in the cage fish farm was higher than the theoretic maximum carrying capacity value in most time except for in September and December 2001 and June 2002.