Behavioural biases in multispecies commercial fisheries and their impacts on stock assessments
Accurate population estimates are a central aspect in species management, especially for fish stocks subjected to harvest pressures. Commercial fisheries catch landings provide an abundant source of data, but are inherently biased due to fishers actively targeting or avoiding certain species and areas. I investigated how biases in the behaviour of the fishers may impact the overall abundance index through generalized additive mixed models using subsets of data selected based on the inferred behaviour of the fishers for two commercially valuable species on the east coast of Canada. For both haddock (Melanogrammus aeglefinus) and redfish (Sebastes spp), fishing sets targeting the species and sets where the species were caught as bycatch produced different relative abundance indices despite being drawn from the same underlying population. When these indices were used in a virtual population analysis stock assessment for haddock, the resulting spawning stock biomass estimates reflected the biases in the index. Indices produced from bycatch data provided a more robust population estimate than target data, and may be a suitable alternative for when survey data are unavailable.
Multispecies fisheries, Targeting, Standardized catch, GAMM, Virtual population analysis, Bias