Spatiotemporal models to improve stock size abundance indices derived from stratified bottom-trawl surveys, with an application to witch flounder

A fundamental part of a fishery stock assessment is an abundance index that reflects stock trends and changes across time. An abundance index is a measurement that is proportional to stock size, and the constant of proportionality is assumed to be the same over time and space. The abundance index can be obtained from the information provided by research vessel surveys of fisheries resources. Changes in survey coverage are a common problem in such surveys, and these changes create uncertainty in interpreting the trend of abundance indices. This is because when there are changes in survey coverage, the changes in survey abundance indices over time may partially reflect changes in the areas surveyed in addition to changes in stock size. Thus, changes in survey coverage over time may lead to errors in fish stock assessment. This thesis provides improved survey indices of stock size using spatiotemporal models that address changes in survey coverage for witch flounder on the east coast of Newfoundland and Labrador. I first estimate the design-based abundance indices to use as a baseline to compare the performance of indices derived from spatiotemporal models. Second, I develop spatiotemporal models and select the best-fitting model via model comparison to estimate the spatial variation of mean catches. These models can predict and interpolate the abundance indices for missing areas and years. These models enable us to fill the data gaps in some survey years to provide more reliable survey indices and longer time-series for the witch flounder stock assessment