Testing the spectral variation hypothesis by using the RAO-Q index to estimate forest biodiversity: Effect of spatial resolution

Forests are the most biodiverse terrestrial ecosystem; the assessment of the species diversity is therefore needed for a better conservation and management of this ecosystem. Remote sensing is a key source for the estimation of species diversity, creating update, standardized and relatively inexpensive data. One of the method used to estimate biodiversity from remote sensing data is the Spectral Variation Hypothesis (SVH) that has been tested by many authors in various ecosystems using different indexes and measures. In this paper the Rao-Q index, used in ecology as a measure of functional diversity has been tested for the first time through the concept of SVH, with optical remote sensing data (of different spatial resolution) to better assess the forest species diversity. For this purpose, for the first time, ten Sentinel 2 (S2) bands (those with resolution of 10 and 20m), all the Landsat 8 (L8) bands and the Normalized Difference Vegetation Index (NDVI), derived from the two satellite, were tested and correlated to the species diversity derived from field data (through the Shannon’s H.). The correlations were different between the two satellite data: the S2 bands achieved better results than the L8 bands but, with the exception of the S2 NIR band (S2 R²=0.592), the correlation were low. Good results showed the NDVI, in particular from the S2 satellite where the correlation achieved a R² of 0.7, underlying that the vegetation index could be a reliable indicator for the SVH, when the spatial resolution is properly selected