Abstract
Indices of abundance based on fishery independent survey data are preferred in stock assessments compared to fishery catch rates because they are designed to be representative of stock trends. Fishermen also collect catch data which, when standardized to remove effects of variables other than abundance, can potentially supplement information from a survey index. Updated understanding of the stock structure of Atlantic cod (Gadus morhua) supports shifting from two assessed management units to five populations. This change requires fine scale data, especially for populations with small sample size in the NEFSC Bottom Trawl Survey. I compared methods for producing indices of abundance for cod from vessel logbook data. First, data were divided into spatial management units, assessment units, and population units. Generalized linear models (GLMs) were fit to each unit’s data with covariates including combinations of year, month, depth, vessel horsepower, vessel tonnage, mesh size, and statistical area. An optimal model for each spatial unit was selected, and indices were produced from the transformed year effects. Trends in abundance differed among populations within the management units, and predictive performance improved when the populations were modeled separately. However uncertainty was high for populations with low sample sizes. Next, whole-region models that accounted for population structure with stock area effects were explored to see if sharing information about covariate relationships could decrease uncertainty for data-poor stocks. The resulting indices produced for each population area were similar to those from the individual GLMs. Although uncertainty was reduced in some areas in some years, there was not enough improvement to justify the more complex modeling approach of using the generalized additive mixed model. Ultimately the coarse spatial resolution and distribution of fishing trips among stocks were limiting factors in this analysis. Future work should explore other fishery-dependent datasets and leverage fishermen’s ecological knowledge to improve the application of the data they collect.