Inside or out? Modeling climate effects on western Atlantic bluefin tuna distribution

The rapidly changing ocean environment is driving shifts in fish distribution, fishing effort, catch, and catch rates. Previous assessments of western Atlantic bluefin tuna have struggled to reconcile conflicting trends in US and Canadian catch per unit effort (CPUE) indices in the northwest Atlantic, likely due to changes in bluefin distribution in recent decades. The most recent stock assessment addressed this by using a time series of the Atlantic Multidecadal Oscillation to modulate catchability for these indices and account for changes in fish availability. This approach improved model diagnostics and performance, but several challenges remained. We explored an alternative approach, accounting for changes in fish distribution due to environmental factors outside of the model. Using a Vector Autoregressive Spatio-Temporal Model (VAST), we created a US-only and a joint US-Canada rod and reel CPUE index that integrates several environmental indicators to model trends in abundance across a larger spatial scale. A comparison of approaches suggests improved or similar stock assessment model performance and similar model results in terms of spawning stock biomass and fishing mortality estimates. These results indicate that environmentally driven changes can be addressed within or outside the assessment model. For western Atlantic bluefin tuna, we recommend the joint US/Canada VAST index, which best reconciles conflicting CPUE trends across jurisdictional boundaries. Such indices should be integrated into assessment models after confirming adequate spatial coverage, sound diagnostics, and model stability. We recommend spatiotemporal models like VAST when data allow, and exploring environmental relationships outside the model before internal inclusion.