Detecting environmental regimes in fish stock-recruitment relationships by fuzzy logic

A fuzzy logic approach is developed to model and test the impact of environmental regimes on fish stock-recruitment relationships. Traditional methods use environmental variables to classify stock-recruitment data into different membership percentiles followed by fitting the stock-recruitment models for each subset. In contrast, the fuzzy logic approach uses a continuous membership function to provide a rational basis for the classification. Thus, parameter estimation is based on a more logically consistent foundation without resorting to subjective partitions. This new approach is applied to herring stock from the west coast of Vancouver Island (Clupea harengus pallasi) using sea surface temperature as the environmental variable and to Pacific halibut stock (Hippoglossus stenolepis) using the Pacific Decadal Oscillation as the environmental variable. From these applications, the herring stock-recruitment relationships were found to vary significantly during different regimes, whereas this was not the case for halibut. However, in both instances, the fuzzy logic approach demonstrated that density-dependent effects differed between regimes. The fuzzy logic model consistently outperformed traditional approaches as measured by several diagnostic criteria. Because fuzzy logic models address uncertainty better than traditional approaches, they have the potential to improve our ability to understand factors influencing stock-recruitment relationships and thereby manage fisheries more effectively.