Multi-scale multi-sensor satellite remote sensing based analysis of forest β-diversity

Abstract:

Satellites provide robust, timely and continuous data to assess biodiversity in remote or protected areas, where direct field observations can be prevented by difficult accessibility. The objective of this study is to extend the concept of remote sensing based assessment of β-diversity at multi-scale by multi-resolution optical satellite data. This study was conducted in a reserved forest of western Himalaya, India, affected by the invasive Lantana camara L. We calculated and compared Rao’s Q and Shannon index at different spatial resolutions (0.5, 5, and 30 m) and scales (window sizes) by using imageries from Pléiades, RapidEye, and Landsat-8 acquired in April 2013, the pre-monsoon season. Rao’s Q index explained diversity more accurately than Shannon index for the three stand densities analyzed. Diversity was better estimated by Rao’s Q index calculated by Pléiades 1A at a resolution of 0.5 m at low stand density. We observed higher correlations of the average coefficient of variation (CV) with Rao’s Q and Shannon indices for areas associated with mixed spectral reflectance caused by overstory and understory vegetation. CV was lower in open areas dominated by L. Camara. These results indicated a strong scale and spatial resolution dependence of Rao’s Q index on remote sensing-derived spectral heterogeneity information. When applied in heterogeneous forest environments, Rao’s Q index could represent a better remote sensing proxy to estimate β-diversity in respect to the conventional Shannon index.