Use and abuse of fishery models

Recent failures of important fish stocks give mathematical models a poor reputation as tools for fishery management. This paper examines the role of models in fish stock assessment and identifies reasons why they can fail. Starting with laws of arithmetic, models attempt to relate observed data to unknown quantities, such as the stock biomass and abundance. Typically, the number of unknowns greatly exceeds the number of observations, and models must impose hypothetical constraints to give useful estimates. We use the word "fishmetic" (rhymes with arithmetic) to represent uncertainty in the conversion of arithmetic to practical fishery models. Arbitrary assumptions cannot be avoided, even though different choices can greatly influence the outcome of the analysis. We compare the modeling process in fisheries with that in other sciences. World literature also offers useful analogies. Potential reasons for failure suggest possible improvements to the application of fishery models. We recommend that modelers remain skeptical, expand their knowledge base, apply common sense, and implement robust strategies for fishery management. Particularly creative thought must be applied to the problem of translating scientific knowledge into management practice. Comparisons between fish stocks and financial stocks illustrate some possibilities.