US-Wide Observation-Level Fisheries Meta-Analysis Using Delay Difference Models

Stock assessment models are often tailored to a specific stock, and it’s rarely the case that more than one type of assessment models is applied to a given stock.  Moreover, little has been done to evaluate the performance of a standardized assessment modeling approach in comparison to tailored assessment models. Delay difference models are simpler than statistical catch-at-age models, but still retain a level of biological realism through the inclusion of recruitment, growth, and natural mortality.  Bottom trawl survey data from the National Marine Fisheries Service (NMFS) is now publicly available through the OceanAdapt portal, a collaboration between the Pinsky Lab of Rutgers University and NMFS.  Over forty years of data are available for seven regions in the contiguous United States and Alaska, including a total of 360 marine taxa. Using NOAA trawl survey data, state (or regional) level catch data, and life history parameters, we fit non-spatial delay difference models to multiple stocks in order to conduct an observation-level meta-analysis using the same model for all species nationally.  We further compare the forecasting ability of the delay difference model with the stock assessment model currently in place.