On bird species diversity and remote sensing — utilizing lidar and hyperspectral data to assess the role of vegetation structure and foliage characteristics as drivers of avian diversity

Avian diversity has long been used as a surrogate for overall diversity. In forest ecosystems, it has been assumed that vegetation structure, composition, and condition have a significant impact on avian diversity. Today, these features can be evaluated via remote sensing. This study examined how structure metrics from lidar data and narrowband indices from hyperspectral data relate with avian diversity. This was assessed in four deciduous-dominated woods with differing age and structure set in an agricultural matrix in eastern England. The woods were delineated into cells within which metrics of avian diversity and remote sensing based predictors were calculated. Best subset regression was used to obtain best lidar models, hyperspectral models, and finally, the best models combining variables from both data sets. The aims were not only to examine the drivers of avian diversity, but to assess the capabilities of the two remote sensing techniques for the task. The amount of understorey vegetation was the best single predictor, followed by foliage height diversity, reflectance at 830 nm, anthocyanin reflectance index 1, and Vogelmann red edge index 2. This showed the significance of the full vertical profile of vegetation, the condition of the upper canopy, and potentially tree species composition. The results thus agree with the role that vegetation structure, condition, and floristics are assumed to have for diversity. However, the inclusion of hyperspectral data resulted in such minor improvements to models that its collection for these purposes should be assessed critically.