What catch locations tell you about fish densities: Estimating spatial distribution and abundance from catch per unit of effort

Steven Saul

University of Miami, RSMAS

In order to test the performance of estimation methods for relative abundance an agent-based model is being developed for the Florida west coast demersal grouper-snapper fishery. In the model, individual vessels make targeting decisions and redistribute their fishing effort according to a combination of factors including abundance. This operating model therefore requires estimates of the spatial distribution of abundance of species in this complex. Various scenarios are tested to determine how catch per unit of effort information might best be stratified in order to estimate fish abundance on the West Florida Shelf. Deviance analysis is conducted to test various hypotheses about the statistical significance of space in relation to fish abundance. When tested independently in a generalized linear model (GLM), depth strata of 10 feet and NMFS statistical grid were found to be statistically significant however their temporal interactions with month and year were not. An alternative GLM, including the interaction of 60 foot depth strata and three regions, the Florida Pan-handle, Florida Central West Coast (also called the “Middle Grounds”), and the South Florida West Coast also identified region and depth strata as significant. Results of this interaction model are used to spatially simulate reef fish abundance across a finer spatial scale by estimating a Gaussian random field for each region-depth interaction. The resulting matrix is used to initialize spatial abundance, as well as throughout the operating model to distribute annual recruitment.