Potential of LiDAR for species richness prediction at Mount Kilimanjaro

To mitigate the negative effects of biodiversity loss, monitoring of species and functional diversity is an important prerequisite for focused management plans. However, sampling of biodiversity during field campaigns is labor- and cost-intensive. Therefore, researchers often use proxies extracted from three-dimensional and high-resolution airborne LiDAR (Light Detection and Ranging) data of the vegetation for predicting biodiversity measures (e.g. species richness or diversity).

This study aims at (i) assessing the suitability of LiDAR observations to map species richness across 17 taxonomic groups and four trophic levels at Mount Kilimanjaro and (ii) differentiating the predictive power of LiDAR-derived structural information from what is already explained by elevation, thereby comparing the prediction potential across taxa and trophic levels.

The field data for this study were collected across 59 plots along an elevation gradient of about 4000 meters at the southern slopes of Mount Kilimanjaro using established methods to sample the selected groups of organisms. The prediction is accomplished with three consecutive steps: (1) Species richness of each taxon is estimated using Partial Least Square Regression (PLSR) with only elevation and its square as independent variables. (2) The residuals of this model are then predicted using the LiDAR-derived variables and PLSR. (3) This third model is subsequently compared to a model that uses the same LiDAR-derived variables and PLSR to predict species richness directly rather than its residuals. This procedure allows to analyze the impact of elevation versus structure on each taxon. Furthermore, the standardized study design allows to compare the predictability of species richness across the selected groups of organisms.

Results of this study show that most taxa can be best predicted by elevation, even though in most cases the structural models perform almost equally. As expected, results of the model performances of trophic levels indicate, that herbivores are influenced more by structure than decomposers and generalists.