The current rate and extent of global change exceeds our capacity to monitor biodiversity with field-based methods alone, calling for the development of remote sensing methods that can provide rapid estimates of biodiversity across large areas. Imaging spectroscopy provides a particularly valuable approach for detecting plant biodiversity because spectra of plants depend on their biochemical, morphological, anatomical, and architectural traits that structure plant communities and affect entire ecosystems. Here we test the degree to which remotely sensed spectral diversity - the dissimilarity in interactions of canopy plants with solar radiation - is predictive of taxonomic, functional, and phylogenetic plant diversity in ecosystems ranging from Neotropical forest to Arctic tundra. For this work, we use imaging spectroscopy and field data collected by the National Ecological Observatory Network (NEON) across the U.S. We calculate taxonomic diversity from plant species inventories; functional diversity from biochemical and structural foliar traits; phylogenetic diversity from a large phylogeny of seed plants; and plant spectral diversity as spectral variance among image pixels.
Results/Conclusions
We found an overall positive relationship between spectral diversity and most aspects of biodiversity measured on the ground. However, the strength of the spectral diversity - biodiversity relationship depends on ecosystem type, canopy structure, and the methodology used to derive measures of biodiversity from inventory data. Spectral diversity also captured different levels of biodiversity (alpha- or beta-diversity) depending on the plant to pixel size ratio. Casting spectral diversity as variance enabled investigating two important aspects of the spectral diversity-biodiversity relationship which we discuss: Spectral variance partitioning allowed calculating the contribution of individual spectral bands or features to spectral diversity, and it allowed estimating and mapping the contributions of individual image pixels or plant communities to the spectral diversity of a region. In summary, our study provides critical insight into the ecological meaning of spectral diversity and highlights the importance of biodiversity observatory networks for developing novel methods for large scale biodiversity assessments.