Predicting patterns of plant diversity and endemism in the tropics using remote sensing data: A study case from the Brazilian atlantic forest

Abstract

We combine remote sensing (RS) measurements of temperature and precipitation with phylogenetic and distribution data from three plant clades with different life forms, i.e., shrubs and treelets (tribe Miconieae, Melastomes), epiphytes (Ronnbergia-Wittmackia alliance, Bromeliaceae), and lianas (“Fridericia and Allies” clade, Bignoniaceae), to predict the distribution of biodiversity in a tropical hot spot: the Brazilian Atlantic Forest. We assess (i) how well RS-derived climate estimates predict the spatial distribution of species richness (SR), phylogenetic diversity (PD), and phylogenetic endemism (PE) and (ii) how they compare to predictions based on interpolated weather station information. We find that environmental descriptors derived from RS sources can predict the distribution of SR and PD, performing as well as or better than weather station-based data. Yet performance is lower for endemism and for clades with a high number of species of small ranges. We argue that this approach can provide an alternative to remotely monitor megadiverse groups or biomes for which species identification through RS are not yet feasible or available.