| Abstract: | Direct tracking methods in combination with remote sensing data allow examination of habitat use by birds during migration. Species that roost communally during migration, such as some swallows, form large aggregations that can attract both avian and terrestrial predators. However, the extent to which they might use patchy habitats that could reduce predation risk during migration is unknown. We tested the hypothesis that Purple Martins (Progne subis) use forest islands (patches of suitable forest habitat surrounded by unsuitable habitat) as roost sites during migration between breeding sites in North America and overwintering sites in South America. We used high‐precision (< 10 m), archival GPS units deployed and retrieved during the 2015 and 2016 breeding seasons, respectively, at 12 colonies located across eastern North America. We found that Purple Martins roosted in forest islands more often than expected based on availability during both spring and fall migration. Despite an apparent association with urban habitats by Purple Martins based on observational and radar data in North America during the fall, the roost locations we identified during spring and fall migration were not more closely associated with urban areas than random locations. The use of forest islands during both spring and fall migration suggest that Purple Martins may use these habitats to reduce predation risk during migration. Our results suggest that some species of birds may use similar habitats as stopover sites during migration and that patches of forest habitat may be important conservation targets for Purple Martins and other species. Identifying habitat use during migration represents an important advance in support of full annual‐cycle conservation of Purple Martins and other migratory species with declining populations. |
