EVALUATING SPATIAL MISSPECIFICATION IN THE ASSESSMENT AND MANAGEMENT OF FISH STOCKS: A CASE STUDY OF BLACK SEA BASS

Abstract

Spatial structure exists at some level in nearly all natural populations, and the importance of this in fish populations has been recognized for more than a century. However, challenges of data availability and computational techniques have limited inclusion of spatial population structure in stock assessments. This dissertation explored the consequences of failing to account for spatial structure in fisheries management. First, I developed a flexible modeling system that includes an operating model which represents the true population dynamics, an observation model which adds uncertainty to outputs of the operating model to generate pseudo-data for catch and abundance indices, and an estimation model that uses the pseudodata from the observation model to produce indices of the status of an exploited fish species. This model system permits differences in how seasonality and spatial structure are represented in a population. I showed that this modeling system produces accurate and precise estimates when the spatiotemporal structure of the operating model matches that of the estimation model. Next, I used this modeling system to evaluate whether estimates from the estimation model are reliable when the spatiotemporal structure of the operating and estimation models does not match. Unreliable catch data lead to unreliable estimates regardless of the match between spatiotemporal structure of the operating and estimation models, without spatiotemporal variability in population dynamics processes between regions and seasons. When uncertainty in catch data was low, all estimation models yielded accurate and precise estimates of management relevant variables. In contrast, when uncertainty in the catch was high, many model simulations failed to converge, and when available, estimates of variables from such models were biased and imprecise. This suggests that agencies responsible for fisheries management should focus as much on reducing uncertainty in the values of key inputs to models as on getting the seasonal and spatial structure correct. In a final application, I examined the performance of the modeling system when reproduction and movement in a species varies over time, such as has been demonstrated for species like Black Sea Bass (Centropristis striata) under a changing climate. A spatially-explicit estimation model was able to produce accurate and precise estimates in the presence of temporal trends in recruitment when the estimation model was naïve to the recruitment trend. However, model estimates were biased and imprecise in the face of time varying movement between regions when the estimation model was naïve to the movement trend. These results suggest that efforts to include spatial considerations in stock assessments for Black Sea Bass need to be accompanied by similar efforts to understand and estimate rates of movement between regions in this species.