Unlike smaller raptors, which can readily use flapping flight, large raptors are mainly restricted to soaring flight due to energetic constraints. Soaring comprises of two main strategies: thermal and orographic soaring. These soaring strategies are driven by discrete uplift sources determined by the underlying topography and meteorological conditions in an area. High-resolution GPS tracking of raptor flight allows the identification of these flight strategies and interpretation of the spatiotemporal occurrence of thermal and orographic soaring. In this study, we develop methods to identify soaring flight behaviors from high-resolution GPS tracking data of Verreaux's eagle Aquila verreauxii and analyze these data to understand the conditions that promote the use of thermal and orographic soaring. We use these findings to predict the use of soaring flight both spatially (across the landscape) and temporally (throughout the year) in two topographically contrasting regions in South Africa. We found that topography is important in determining the occurrence of soaring flight and that thermal soaring occurs in relatively flat areas which are likely to have good thermal uplift availability. The predicted use of orographic soaring was predominately determined by terrain slope. Contrary to our expectations, the topography and meteorology of eagle territories in the Sandveld promoted the use of soaring flight to a greater extent than in territories in the more mountainous Cederberg region. Spatiotemporal mapping of predicted flight behaviors can broaden our understanding of how large raptors like the Verreaux's eagle use their habitat and how that links to energetics (as the preferential use of areas that maximize net energy gain is expected), reproductive success, and ultimately population dynamics. Understanding the fine-scale landscape use and environmental drivers of raptor flight can also help to predict and mitigate potential detrimental effects of anthropogenic developments, such as mortality via collision with wind turbines.
Keywords: behavior classification; collision risk; energy landscape; flight; predictive modeling; random forest; soaring; uplift.