A species distribution model integrating at-sea behaviour, phenology, and population size to inform marine risk exposure for a threatened seabird during breeding

Seabirds face myriad threats on land and at sea resulting in widespread population declines. Predictive species distribution models (SDMs) provide a powerful tool to estimate numbers of birds exposed to risks, and to attribute them to source populations. The Atlantic population of Leach’s Storm-petrel ( Hydrobates leucorhous ) has declined rapidly in recent decades; however, information on their exposure to threats at sea is lacking. We address this knowledge gap for the breeding period using a multi-year GPS tracking dataset ( n = 451 birds, 2016-2023) collected from six colonies which together comprise > 90% of the Atlantic population. As threat exposure can differ according to behaviour, we used a Hidden Markov Model (HMM) to classify behavioural states and define associated risk categories (primarily in-air or on-water). We also summarized results by time of day because attraction to, and disorientation from, artificial light at night (ALAN) poses risks to Leach’s Storm-petrels. We developed a dynamic SDM predicting monthly relative densities at sea (June-September), for the western North Atlantic regional population, accounting for all 16 colonies with > 10,000 individuals. Reflecting geographic patterns in colony size, regional densities were highest east of Newfoundland in deepwater areas off the northern Grand Bank (Orphan Basin and Flemish Pass), and south of the Flemish Cap (Newfoundland Basin). At-sea densities increased through summer, being lowest in June when one parent is incubating, and highest in September when both parents forage at sea. Case studies indicated that relative to risks within Grand Bank oil and gas production areas, commencement of deepwater drilling in the Orphan Basin and Flemish Pass has substantially increased risks for Leach's Storm-petrels, because bird densities in these offshore areas are 2–4 times higher. Within our region, risks from ALAN were consistently estimated to be higher than risks from hypothetical oil spills. Though offshore facilities have been removed from the Scotian Shelf, ongoing bids for exploration licences suggest future risks for specific populations in Nova Scotia. We focused on oil and gas; however, our framework can be applied to other individual or cumulative risks to inform conservation planning for this threatened species.