Linking variability in species composition and MODIS NDVI based on beta diversity measurements

Finding an effective method to quantify species compositional changes in time and space has been an important task for ecologists and biogeographers. Recently, exploring regional floristic patterns using data derived from satellite imagery, such as the normalized difference vegetation index (NDVI) has drawn considerable research interests among ecologists. Studies have shown that NDVI could be a fairly good surrogate for primary productivities. In this study, we used plant distribution data in the North and the South Carolina states to investigate the correlations between species composition and NDVI within defined ecoregions using Mantel test and multi-response permutation procedure (MRPP). Our analytical approach involved generating compositional dissimilarity matrices by computing pairwise beta diversities of the 145 counties in the two states for species distribution data and by computing Euclidian distances for NDVI time series data. We argue that beta diversity measurements take the pairwise dissimilarities into consideration explicitly and could provide more spatial correlation information compared with uni- or multi-dimensional regressions. Our results showed a significant positive correlation between species compositional dissimilarity matrices and NDVI distance matrices. We also found for the first time that the strength of correlation increased at a lower taxonomic rank. Same trends were discovered when incorporating variability in phenological patterns in NDVI. Our findings suggest that remotely sensed NDVI can be viable for monitoring species compositional changes at regional scales.