An individual-based model to evaluate the effect of fisher behavior on reef fish catch per unit effort

Fisher's decisions can influence the effectiveness of management measures as such decisions can allow fishers to dissipate the benefits of regulation. Furthermore, these decisions determine the spatial and temporal locations of fishery-dependent observations. In many stock assessments such observations are used to infer the abundance of fish populations and are essential in our efforts to understand the dynamics of fish stocks. In order to understand some of the ways that fisher behavior can influence our perception of abundance, an individual-based spatial model of the reef fish fishery off of the West Coast of Florida was developed. The spatial distribution of fish in the simulation was generated by combining three sets of estimates: stock size from stock assessment models, fishery-dependent catch per unit of fishing effort estimates of relative abundance, and fishery-independent estimates of spatial autocorrelation in relative abundance derived from video survey data. Simulated spatial distributions of fish abundance were used to develop a biased random walk algorithm that represented movement of reef fish from inshore nursery habitat to offshore reef habitat in two squared kilometer grids for the West Florida Shelf. Time-at-large, release location, and input movement speed from conventional tagging data were used to parameterize the biased random walk algorithm in the simulation. In order to represent fishing behavior across space and time in the simulation model, three primary decisions were modeled: when to fish, where to fish, and when to return to port. Discrete choice models were fit using logbook data from the fishery, in conjunction with data on vessel characteristics, weather, fish price, regulations, and the price of fuel in order to represent these decisions. Results from the simulation model demonstrate that catch per unit effort (CPUE) is not always proportional to stock biomass and that CPUE standardization using our current statistical methodologies is not always effective at extracting relative biomass trends from CPUE. Furthermore, although most stock assessments assume the spatial distribution of biomass across a management area to be stationary, simulation results show that changes in the spatial distribution of biomass can result from harvest occurring in particular locations and not randomly. Such interaction between the spatial distributions of fishing effort and biomass can lead to situations where relative abundance indices derived from fishery dependent data provide a biased perception of trends in stock biomass.