ABSTRACT

Rapid and accurate monitoring of biodiversity is a major challenge in biodiversity conservation. Obtaining data using unmanned aerial vehicles (UAV) provides a new direction for biodiversity monitoring. However, studies on the relationship between UAV data and biodiversity are limited. In this study, we used a machine learning algorithm to evaluate the effectiveness of UAV-light detection and ranging (LiDAR) and UAV multispectral data for estimating three α-diversity indices in tropical forests located in Hainan, China. We obtained 126 biodiversity-related metrics (68 from multispectral and 58 from LiDAR) based on the UAV data and three α-diversity indices from 62 sample plots at two sites. We used the recursive feature elimination algorithm to filter significant metrics. We found that both multispectral and LiDAR data can be used to predict α-diversity. The coefficient of determination (R²) values of multispectral data (LiDAR data) for the species richness, Shannon index, and Simpson index were 0.69, 0.70, and 0.57 (0.72, 0.63, 0.44), respectively. LiDAR data were more accurate than multispectral data for predicting species richness, whereas multispectral data were more accurate than LiDAR data for predicting the Shannon and Simpson indices. Given the best result obtained with a single datum, the accuracy (R²) of the combination of the two data types for species richness and Shannon and Simpson indices increased by 0.05, 0.05, and 0.06, respectively, indicating that the prediction accuracy of the α-diversity index can be improved by integrating different remote sensing data. Additionally, the most important multispectral metrics used to predict α-diversity were related to vegetation index and texture metrics, whereas the most important LiDAR metrics were related to canopy height characteristics. Our research results indicate that UAV data are effective for predicting the α-diversity index of Hainan tropical forest on a fine scale. UAV data may help local biodiversity workers to identify vulnerable areas.