PREDICTING WOODY-PLANT SPECIES RICHNESS IN TROPICAL DRY FORESTS: A CASE STUDY FROM SOUTH FLORIDA, USA

Abstract

Tropical dry forests are one of the world's most endangered forest types. Currently there are no comparative data on extent or levels of species richness for remaining forest fragments. This research identifies landscape metrics and spectral indices that can be applied at the stand and patch level to predict woody-plant species richness in tropical dry forests. This study was undertaken in 18 stands of tropical dry forest with nine sites in the Florida Keys and nine sites within an urban–agricultural matrix in mainland Florida, USA. Woody-plant species richness was quantified at the stand level (belt transects totaling 500 m2) and patch level (presence/absence data for 65 native tropical plants ≤2.5 cm dbh) for all study sites. Landsat Enhanced Thematic Mapper Plus (ETM+) satellite images (pixel resolution 30 × 30 m) were used to assess the utility of landscape metrics (forest patch area, nearest neighbor distance, shape index, boundary complexity) and spectral indices (normalized-difference vegetation index [NDVI] for nine pixels and 500 pixels directly over transects, and all pixels in the forest patch area) for predicting stand- and patch-level species richness. The 18 stands of tropical dry forest sampled in this study included 4248 woody plants, representing 71 species. Islands in the Florida Keys had higher levels of woody-plant species richness than mainland sites. There was a significant positive relationship between mean NDVI for the nine pixels over each stand and stand species richness and a significant negative relationship between species richness and standard deviation of NDVI for nine pixels over each stand. The density of evergreen plants explained 66% of the variability in mean NDVI. At the patch level, forest patch area and mean NDVI at the stand, 500-pixel, and patch level were all positively associated with patch species richness. However, combining forest patch area with NDVI significantly improved the prediction of patch species richness. Results from this study support the species–energy theory at the level of a forest stand and patch and suggest that a first-order approximation of woody-plant species richness in stands and patches of tropical dry forest is possible in biodiversity hot spots.