A predictive model for improving placement of wind turbines to minimise collision risk potential for a large soaring raptor

Abstract With the rapid growth of wind energy developments world‐wide, it is critical that the negative impacts on wildlife are considered and mitigated. This includes minimising the number of large soaring raptors, which are killed when they collide with wind turbines. To reduce the likelihood of raptor collisions, turbines should be placed at locations which are least used by sensitive species. For resident or breeding species, this is often delineated crudely through the use of circular buffers centred on nest sites, which assume uniform habitat use around a nest site. Using GPS tracking data together with a digital elevation model we build and cross‐validate a simple generalisable model, to classify the spatial likelihood of wind turbine collisions for resident adult Verreaux's eagles in any landscape where there are known nests. We apply our methods to operational developments in South Africa to validate the model and demonstrate its ability in predicting actual collision mortalities. Our collision risk potential (CRP) model included the variables distance to nest, distance to conspecific nest, slope, distance to slope and elevation. Using our model, rather than a circular buffer, resulted in c . 4%–5% improvement in eagle protection while excluding development from the same amount (but not shape) of area. For an equal level of eagle protection, our model can make c . 20%–21% more area available for wind energy development compared to a circular buffer. Exploring collisions at operational wind farms in South Africa we show that our CRP model correctly predicted 79% of known collisions, while circular buffers (5.2 km radius) only captured 50% of collisions. Synthesis and applications . We show that by using predictive models to account for habitat use instead of simple buffers around a nest, a greater area of land can be made available for wind energy development without increased mortality risk to raptors. Our predictive model can be used to provide robust guidance on wind turbine placement in South Africa in a way which minimises the conflict between a vulnerable raptor species and the development of renewable energy.