Linking floristic patterns with soil heterogeneity and satellite imagery in ecuadorian amazonia

Florisitic ground surveys in tropical rain forests are laborious and time consuming, so we tested to what degree reflectance differences visible in Landsat Thematic Mapper (TM) satellite images can be used to predict differences in florisitic composition and species richness among rain forest sites. To gain ecological understanding of the rain forest ecosystem, we also tested to what extent variation in these vegetation characteristics can be explained by edaphic site conditions. The study was conducted in a relatively homogeneous area of Amazonian rain forest in Yasuní National Park, Ecuador. We established 27 transects of 5 m × 500 m within an area of ∼20 km × 25 km to study edaphic and floristic patterns mainly within the tierra firme (non‐inundated) forest. In each transect, soil samples were collected for chemical and textural analyses, and the abundance of each species belonging to two understory plant groups, pteridophytes (ferns and fern allies) and the Melastomataceae, was assessed. Floristic similarity between transect pairs varied widely and ranged from almost no overlap in species composition to very high overlap. The among‐transect floristic similarity patterns of the two plant groups were strongly correlated with each other no matter whether presence–absence or abundance data were used. The floristic similarity patterns were also strongly correlated with the similarity in pixel values of the infrared bands in the Landsat TM satellite image and with the similarity in most of the measured soil variables. Similarity in species richness, on the contrary, was neither correlated with similarity in pixel values nor with similarity in most of the soil variables. We conclude that reflectance patterns in satellite images can be efficiently used to predict landscape‐scale floristic and edaphic patterns in tierra firme rain forest. Predicting patterns in species richness, on the other hand, is not possible in the same straightforward manner. These results have important practical implications for land use and conservation planning as well as for ecological and biodiversity research. Corresponding Editor: C. A. Wessman.