Bright nights and dynamic seascape features drive leatherback turtle (Dermochelys coriacea) interactions in an Australian pelagic longline fishery

Abstract

• Western boundary current drives highly variable seascapes in east Australian waters. • We used environmental data from Bluelink ReANalysis (BRAN2020) and fisheries data. • Ensemble Random Forests modelled fishery interaction risk with leatherback turtles. • Risk increased around dynamic seascape features and brighter lunar phases. • Foundation for operational tool for dynamic management of wildlife interactions. Marine wildlife bycatch is a persistent and pervasive issue limiting sustainable fisheries management. Key gaps in knowledge include the spatio-temporal dynamics of where fisheries and marine wildlife overlap, and what drives interactions between them. The Critically Endangered Western Pacific subpopulation of leatherback turtles ( Dermochelys coriacea ) migrate and forage along the east coast of Australia, where they interact with a pelagic longline fishery. This region is characterised by a western boundary current system with a highly dynamic seascape, yet little is known about how leatherback turtles use this pelagic habitat or the drivers of fisheries interaction risk. Here, we use Ensemble Random Forests to examine the influence of surface and subsurface conditions derived from outputs of an openly available data-assimilative ocean model, season, and lunar illumination, on fisheries interactions with leatherback turtles. We found the highest risk areas for leatherback interactions were also the most oceanographically dynamic. Persistent links between leatherback turtle interactions and dynamic ocean features, particularly those associated with the East Australian Current, are explored. We hypothesise that turtles are likely using these features to forage and conserve thermal energy. The risk of interaction also increased during brighter periods of the lunar cycle; a finding likely driven by foraging ecology. These findings provide new information of value to industry and managers and a foundation for testing the utility of data-assimilative ocean models as the basis of decision-support tools for bycatch avoidance.