TOPOGRAPHIC MODELS OF VASCULAR PLANT RICHNESS IN THE SOUTHERN APPALACHIAN HIGH PEAKS

(1) Data from a vascular plant survey of the southern Appalachian high peaks, U.S.A., were used to model the relationship between five floristic categories of species richness (total, high elevation, northern, endemic and rare species), six topographic variables (area above 1680 m, area above 1830 m, maximum elevation, number of peaks, inter-mountain distance and latitude) and the number of community types. (2) The best model of species richness varied with each richness category: total richness was best modelled by number of communities, high elevation species by area, northern species by area, endemic species by number of communities and rare species by maximum elevation. (3) In simple regressions, there was no clear preference among model formulations or species richness transformations. Elevation, number of peaks and area were all relatively important predictors, while isolation was unimportant. In multiple regression models, linear formulations were almost always the stongest. For total species richness, the best multiple regression model was linear and the most important predictor was the number of peaks. (4) Rare species richness increased faster with area or elevation than any other richness category (slope coefficients for log (rare species).sbd.log (area) models were 0.30 for the whole data set and 0.49 for the eight largest areas). (5) Distance between mountains almost always had a positive, though usually minor, effect on species richness, thus casting doubt on the appropriateness of the MacArthur and Wilson model for this data set. We hypothesize that distance between mountain areas is probably a poor measure of ecological isolation and that historic extinctions have shaped the species-area relations we observed.