Abstract
The goal of this dissertation was to evaluate the feasibility and utility of ecosystem-based fisheries management (EBFM) approaches aimed at managing an abundant forage fish in the northern Gulf of Mexico (GOM): Gulf menhaden (GM) (Brevoortia patronus). This dissertation constitutes an evaluation of the influence of environmental drivers on the dynamics and assessment of GM and the subsequent influence of GM on upper trophic level consumers in the northern GOM ecosystem. This work addressed a number of data gaps in the diet literature and identified remaining uncertainty in those predatory interactions involving GM prey. GM were found to be important in the diets of a number of GOM piscivores, including seatrout, mackerels, and large and small coastal sharks. However, the relative contribution of GM to the diets of croakers, catfish, and highly migratory predators (e.g., tunas and billfish) remains uncertain. Such uncertainties informed construction of ecosystem models, each a distinct yet plausible representation of the feeding behavior of GOM predators. These models were applied to evaluate how uncertainty in predator diet influences our perception of the trophic importance of the GM population to the northern GOM ecosystem. A variety of pelagic and demersal piscivores were shown to respond to changes in menhaden abundance, the largest changes being simulated in those groups for which menhaden have a relatively large contribution to predator diet. However, uncertainty in predator diet prevents identification of the predator(s) most sensitive to menhaden dynamics. Regional managers for the GM stock are therefore recommended to consider ecosystem reference points aimed at reserving adequate menhaden forage for predatory guilds (e.g., all pelagic and/or demersal piscivores). The EwE models developed in this dissertation are believed in need of further refinement (i.e., inclusion of prey switching) before being recommended for use by fishery managers. The relative need to identify environmental drivers of GM stock dynamics was evaluated in simulations that estimate the bias in assessment models that differ in their treatment of environmental variability. Approaches that implicitly account for inter-annual variability in stock-recruitment were shown to perform just as well as those that explicitly include drivers of recruitment dynamics, supporting the continued application of the current GM stock assessment model. This finding largely agrees with the literature, much of which is based on simulations wherein operating models and assessment models are built within the same single-species modeling framework. Conversely, the simulation framework developed for this dissertation links ecosystem models (as operating models) to single-species assessment models, allowing for consideration of complex trophic interactions at a resolution that was not previously possible and believed necessary for a forage fish that interacts with a wide variety of populations.