A fuzzy logic model with genetic algorithm for analyzing fish stock-recruitment relationships

Abstract

A new fuzzy logic model with a genetic algorithm is developed that overcomes some of the inherent uncertainties in the fish stock-recruitment process. This model is applied to stock-recruitment relationships for the Southeast Alaska pink salmon (Oncorhynchus gorbuscha) and the West Coast Vancouver Island Pacific herring (Clupea pallasi) stocks. In both examples, the annual mean sea surface temperature is used as an environmental intervention in the model. The fuzzy logic model provides the functional relationship between the number of fish spawners and the sea surface temperature that is used to reconstruct the historical fish recruitment time series and also to predict the number of fish that will recruit in the future. Globally optimized genetic learning algorithms are used to find the optimal values of the parameters for the fuzzy logic model. The results from this fuzzy logic model are compared with results from both a traditional Ricker stock-recruitment model and a recent artificial neural network model. These comparisons demonstrate the superior capability of the fuzzy logic model for addressing problems of uncertainty and vagueness in both the data and the stock-recruitment relationship. The fuzzy logic model approach is recommended as a useful addition to the analytical tools currently available for fish stock assessment and management.