Predictive modeling of Ospreys (Pandion haliaetus) nesting in urbanized subarctic Alaska using GIS Open Access data and Machine Learning

Ospreys (Pandion haliætus) are migratory raptors with a global distribution, connecting tropical and subarctic ecosystems along respective flyways. Their populations are poorly managed and, although ospreys are among the most studied raptor, a lack of relevant information remains, while ospreys affect the wider landscape in strong ways. Here, we provide the first open access data and open-source GIS application of an ensemble of five Machine Learning (ML) algorithms (TreeNet, CART, RandomForest, MARS, and Maxent) to analyze the nesting ecology for this species in Alaska, with a special focus on 16 known nests in the municipality of Fairbanks. We used three predictors to determine nest site suitability and produce the first predictive ecological niche model during breeding season (summer). Our model results align with citizen science data, supporting nest site inference. Nest sites are strongly associated with waterbodies, specifically stocked lakes and are usually located near roads on powerline poles. Beyond habitat preferences, we find that ospreys are shot and stressed along the flyway. In Alaska, no specific management beyond the Migratory Bird Act (MBA) exists. The absence of a breeding bird atlas and data, hinders informed conservation planning for anthropogenic climate change, especially in the subarctic breeding hotspot of Fairbanks. Our findings underscore the need for a comprehensive, cross-border, science-based conservation framework that accommodates sustainable activities, climate change adaptation, and species protection.