Traditional stock assessment techniques assume a spatially and temporally homogeneous stock that is exploited by a single gear. This is a pattern that does not hold true for most species, especially those that target highly migratory species. Dolphinfish, Coryphaena hippurus, is one such species that is heavily commercially exploited by numerous Caribbean countries and forms an important part of the recreation fisheries of the United States, Mexico, and most Caribbean nations. Additionally, dolphinfish are caught as bycatch in large commercial operations including the U.S. and Venezuelan pelagic longline fleets that operate in the western Atlantic and Caribbean region. These various fisheries are heterogeneous in that they are exploiting this species with different gears and at different times during the year. Because dolphinfish does not adhere to the assumptions of traditional stock assessment, a different approach, one that improves the capabilities of a model suitable for a highly migratory species, should be investigated. In order to assess this species and characterize its movement, a definition of stock structure must be developed based on specific physical environmental parameters that delineate where and when dolphinfish will be abundant regionally. The goal of this research is to improve the understanding of the abundance in time and space of a migratory fish species through spatial interpolation and stochastic modeling of environmental attributes and the application of geostatistical analysis for the description of spatial patterns and identification of scales of variability.

A spatially explicit modeling approach will be used to examine the spatio-temporal abundance of dolphinfish according to classifications of physical environmental regimes (temperature, currents, convergence zones, and fronts) as suitable dolphinfish habitat. Three main areas will be focused on in order to address the overarching goal: 1) understanding of how environmental parameters, gear configurations, and species assemblages affect dolphinfish CPUEs in a multivariate context; 2) model tests of generalized linear models (GLMs) and generalized additive models (GAMs) to standardize catch rates, and incorporation of spatial autocorrelation to these indices to test improvement of the fit of the models; 3) development of a surplus production model that incorporates spatial information within the standardized indices of abundance.