Biodiversity loss presents an increasing challenge for managing forest ecosystems. Nonetheless, our ability to predict global changes impacts on vegetation distribution and its attributes is still limited. Biodiversity indices have mostly been based on species richness and abundance, despite the increasing evidence advocating other facets of diversity. Functional diversity (FD), the diversity of organism’s functional traits, have the potential to be more accurate than species-centered approaches because of the continuous nature of functional traits, and the possibility to scale up from organs to higher organizational levels including ecosystems and biomes. Forest canopy traits like leaf mass per area (LMA), leaf nitrogen (LNC), leaf phosphorus (LPC), and leaf water content (LWC) can be use as proxies to understand plant physiology, and the mechanisms related to nutrient uptake and carbon allocation. We used airborne imaging spectroscopy from the Carnegie Airborne Observatory to quantify FD in an Andes-to-Amazon elevation gradient (215-3537 m) across nine 1-hectare plots. We address two questions: (1): how do spectral diversity and FD change along elevation? (2) Is spectral diversity a good proxy for FD across the elevation gradient? We estimated spectral and Shannon diversity using the raw spectra of the visible-to-shortwave infrared (VSWIR). We quantified multidimensional indices using trait distributions of LMA, LNC, LPC, and LWC from airborne imaging. These indices include functional divergence (FDiv), functional evenness (FEve), functional dispersion (FDis) and Rao’s quadratic entropy (RaoQ).
We found that spectral diversity and Shannon diversity decreased with elevation, which explained 40% of the variation in these indices. FEve showed no trends, while FDiv showed a strong decline. The RaoQ and FDis both increased with elevation, and were positively correlated with spectral diversity and Shannon diversity. FDis and RaoQ, represent parameters of the chemical composition turnover of plants along the elevational gradient (e.g., beta diversity), and could be predicted to some extent by Shannon and spectral diversity. Along the elevational gradient, resources become limited (e.g., lower temperature and radiation), and as such we found a reduction in spectral diversity. Increases in FDiv with altitude suggest a greater variety of strategies for nutrient uptake and allocation probably related with more resources at lower elevations. Further testing of the remotely-sensed FD indices against field data is required to make this approach operational, but our findings have important implications for quantifying trait diversity using leaf spectroscopy in highly diverse ecosystems, and assessing changes in FD over space and time.