Mapping α- and β-diversity of mangrove forests with multispectral and hyperspectral images

Mangrove deforestation has rapidly declined by an order of magnitude compared to that reported for the 20th century, but the remaining mangrove ecosystems are still undergoing biodiversity loss and degradation. Laborintensive ground surveys that are usually used for terrestrial plant biodiversity assessment are difficult to conduct in the mangrove forests. Assessing plant biodiversity from space does offer a novel perspective, but existing solutions are mainly focused on estimating alpha- or beta-diversities (i.e., diversity within communities or diversity among communities) alone. This paper applied a novel holistic biodiversity approach that could partition gamma-diversity into alpha- and beta-diversities for plant diversity mapping with operational satellites (WorldView-2, Sentinel-2, and Zhuhai-1) and field plots in the Qinglangang Provincial Nature Reserve, Hainan, China. We compared the resulting outputs of the alpha-diversity from the holistic method (SD alpha) to those by the coefficient of variation (CV) and the Rao's Q index, and the contributions of individual spectral features to alpha- and beta-diversity were also measured. Results indicated that alpha- and beta-diversities accounted for -30% and - 70% of the total diversity in the Reserve, respectively. alpha-diversity derived from the WorldView-2 images showed statistically higher correlations with the observed Shannon's index (R2: 0.20-0.42) compared to that from Sentinel-2 and Zhuhai-1 (R2: 0.03-0.15). beta-diversity derived from WorldView-2 images had the highest accuracy (90.00%), followed by that from Sentinel-2 and Zhuhai-1 (83.33% and 73.33%, respectively). Red-edge and near-infrared spectral features were the most informative features for diversity mapping while shortwave infrared (SWIR) features were also valuable for beta-diversity mapping. Concurrent mapping of alpha- and beta-diversities of mangrove forests represents the first step toward achieving rapid biodiversity monitoring schemes of mangrove forests over a national or global scale.

Automated summary

This paper presents a novel approach to map plant biodiversity in mangrove forests using satellite imagery and field plots. The results show that alpha- and beta-diversities account for different proportions of the total diversity, and the mapping accuracy varies among different satellite images. The red-edge and near-infrared spectral features are found to be informative for diversity mapping.