A Spatially Explicit Fishing Simulation: A Theoretical Assessment of Overexploitation and Fishermen's Ecological Knowledge

Abstract

Common problems in the world’s fisheries include both rampant overexploitation and economic stagnation. In recent decades, managers have sought to alleviate these problems through various regulatory schemes which attempt to ensure sustainability of populations and/or increase profitability for fishermen. In some cases, the structure of these regulations results in conflicts between conservation and the economic incentives of fishermen. In response to these issues, the motivation of this theoretical study is twofold: to assess the usefulness of catch per unit effort (CPUE) as an indicator of fish abundance and to examine the degree to which the characteristics of fish and fishermen influence stock size. These questions are addressed through a spatially explicit fishing simulation which models both the population dynamics of fish and the strategies of fishermen. The modules composing the simulation interact on an ocean space to produce a simulated fishery with 4,000 individual fishing events. The output-data of the simulation, produced by nearly 2,200 simulated fisheries, tracks the ‘true’ fish population size and the fishermen’s estimate of the population size (a proxy for CPUE). A linear regression of the output-data suggests there is no statistically significant relationship between CPUE and population size (p=.096, R2 =.0013). These findings suggest that the spatial nature of fisheries may affect the interpretability of CPUE, and that it is not necessarily a reliable metric for evaluating fish abundance. Additionally, an analysis using generalized additive models (GAMs) suggests that three simulation parameters—fisherman’s memory length (ML), fisherman’s reluctance to explore (rtx), and habitat suitability index sensitivity (HSIS) are statistically significant in explaining population size (p < 2*10-16 for all three parameters), while one parameter—crowding sensitivity (CRS)—is not (p = .0861). Finally, a GAM analysis for the response variable of fishermen’s population estimate shows the same three parameters were statistically significant (p < 2*10-16 for ML, rtx, and HSIS; p = .497 for CRS). This simulation provides a baseline upon which future research, using a similar modelling framework, could evaluate the effects of different regulatory schemes on fisheries.