Modeling Shark Bycatch Mitigation Strategies in Longline Fisheries

This dissertation uses various modelling approaches to evaluate methods to reduce shark bycatch in the U.S. Atlantic pelagic longline and Gulf of Mexico (GOM) bottom longline fisheries. Combinations of environmental and gear variables are used to parameterize models to predict where and under what conditions bycatch occurs and propose bycatch mitigation strategies. All work uses NOAA NMFS observer datasets for U.S. commercial longline fleets operating in the Atlantic and/or Gulf of Mexico (GOM). Several statistical models are used to identify environmental conditions, regions and fishing methods that favor high bycatch of the overfished shortfin mako shark, Isurus oxyrinchus, based on the outputs of the delta-lognormal model and quantile regression of the upper quantiles. The results suggest that using the binomial portion of the delta-lognormal model, the probability of positive catch, to define a hot set basis for a “no fish” algorithm. Second, an individual based movement model is used to test three closure scenarios: stationary, seasonal, and moving weekly closures, for their ability to decrease shortfin mako incidental catch while minimizing the impact on the target fishery. Results suggest that any of the tested closures have potential to improve rebuilding when compared to the status quo. While the two moving closures give some reprieve to the population when compared to current fishing practices, the varying success and failure to surpass the stationary closure indicated that the time scale of a moving closure is very important, and a mismatch can dampen the benefits of the closure. The dissertation finishes with consideration of how to mitigate bycatch of 12 commonly caught shark species in the GOM reef bottom longline fishery including: blacknose, nurse, Atlantic sharpnose, scalloped hammerhead, sandbar, smooth dogfish, night, blacktip, silky, tiger, bigeye sixgill, and sevengill sharks. Catch rates of each species are modeled as a function of environmental and gear variables individually and all combined, as well as grouped by similar ecology. Gear and behavior variables were the most consistently retained in the best predictive models across all species and were the only variables with the potential to be used for a single rule that could decrease bycatch across all studied species. Patterns of environmental variables were only consistent across species with similar ecology and habitat. For both the shortfin mako shark and bottom longline examples, we found that environmental conditions and gear configurations can be used to predict shark bycatch well enough to suggest bycatch mitigation strategies that significantly reduce shark bycatch in longline fisheries; however, there are tradeoffs involved in minimizing bycatch of multiple species, and minimizing bycatch while not unduly restricting target species catch.