Predicting Bird Diversity with Lidar-derived Vegetation Structure in an African Savanna

Vegetation structural complexity and the diversity of animal communities are closely linked in vegetated ecosystems. These structure-diversity relationships have the potential to be used to predict biodiversity at large spatial scales using remote sensing data. However, structure-diversity relationships may not be generalizable across different ecosystems or even across ecotypes within a single ecosystem. To understand how structure-diversity relationships vary within the tree-grass mosaic of a savanna environment, we evaluated how bird diversity relates to vegetation structure at multiple scales and across environmental gradients in an East African savanna- the Selenkay Conservancy in southern Kenya. We obtained detailed characterizations of vegetation structure using Light Detection and Ranging (lidar) from Unoccupied Aerial Vehicle (UAV) surveys, and related vegetation structure metrics to bird diversity metrics (Shannon diversity and species richness) collected at 50 sites spread inside and outside of the Selenkay Conservancy. We compared structure-diversity relationships across environmental gradients, including soil type (red and black soils) and protected status (inside and outside the conservancy). We also compared structure-diversity models at multiple scales, testing how relationships changed with scale. We found significant structure-diversity relationships with improved performed at larger spatial scales (≥ 50 m radius or 0.79 ha circular plots). Models of Shannon diversity performed better than those of species richness. While most structure-diversity relationships only applied to specific soil types, certain models showed the potential to be generalized across soil types, explaining ~55-59% of the variance. We found that strong relationships exist between vegetation structure and bird diversity in savannas. While most structure-diversity relationships were only applicable to specific soil types, several vegetation metrics were able to track bird diversity across the entire landscape, performing well in both red and black soil sites. These results demonstrate the potential to use airborne remote sensing to monitor biodiversity across savanna environments.