Speakers
Description
Monitoring the biodiversity of the Amazon forest poses significant challenges due to its remote location and limited data availability. Remote sensing promises an efficient solution, utilizing advanced sensors to identify biodiversity patterns in vegetation across vast areas. The Spectral Variation Hypothesis (SVH) suggests a correlation between spectral heterogeneity and biodiversity, with unique spectral entities identified as spectral species. We subjected PlanetScope imagery from the dry season to a principal component analysis and K-means clustering to map plant diversity spectrally. The optimum number of clusters was determined using the Silhouette and Elbow methods. Alpha diversity, assessed as Shannon diversity, was estimated at 1 ha grids, while beta diversity mapping at 4 ha grids was achieved through the Bray-Curtis Distance Matrix and Non-Metric Multidimensional Scaling (NMDS).
Field plots revealed a maximum species richness of 129 and species abundance of 1006 individuals per plot. The alpha diversity map highlighted significant spatial patterns, with R² values of 0.45 and 0.57 for k = 40 and k = 81, respectively. Beta diversity characterization using spectral data proved inconclusive, as cross-verification with Detrended Correspondence Analysis and Mantel Test did not yield significant results. Vegetation health and stress were assessed through 69 vegetation indices and principal component analysis. Interestingly, regions with moderate biodiversity exhibited healthier vegetation compared to highly diverse areas, possibly due to similar species clustering into different groups.
Ongoing analysis aims to understand the interaction between spectral diversity and biogenic emissions, providing insights into the ecological dynamics of volatile organic compounds. The study demonstrates that spectral diversity is promising for monitoring and conservation efforts, offering valuable insights into biodiversity patterns in the Amazon forest and beyond.
| Status Group | Postdoctoral Researcher |
|---|
