Abstract
Seabirds are a highly vulnerable group of marine animals that are intimately linked to the pelagic environment. Interactions with longline fisheries have long been a source of mortality and a contributing factor to population declines, but the environmental drivers behind these interaction events are poorly understood. We used 14 years of fisheries logbook data describing albatross and shearwater interactions in the east Australian pelagic longline fishery together with a suite of environmental variables derived from a data-assimilative ocean model to explore the physical drivers underlying seabird–fishery interactions. We used an ensemble random forest protocol for rare-event modelling. Our findings reveal that season and latitude are the most significant predictors of seabird interaction risk. Albatrosses were at higher risk south of 35°S, shearwaters south of 25°S, and risk was highest in austral autumn and spring for both groups. Environmental variables that indicate mesoscale features were also important predictors, particularly sea surface temperature gradients for shearwaters and sea surface height for both groups. Given the greater predictive power of season and latitude, we recommend application of enhanced spatio-temporal risk-management strategies by industry and managers. Where seabird interactions persist despite effective implementation of existing measures, additional best-practice mitigation measures at high-risk latitudes, coupled with seasonal risk notifications to industry, may further reduce impacts on these highly vulnerable species.