PREDICTION OF RARITIES FROM HABITAT VARIABLES - COASTAL-PLAIN PLANTS ON NOVA SCOTIAN LAKESHORES

Predictive models relating species richness of rare plants to measured habitat variables were developed using data from the shoreline vegetation of lakes in southwestern Nova Scotia. These lakes contain large disjunct populations of many Atlantic Coastal Plain plant species whose main ranges are along the eastern seaboard of the United States. The richness of rare coastal plain herbs was easier to predict than richness of the "background flora" of wide-ranging species from noncoastal plain elements. Multiple-regression models using habitat variables accounted for 83% of the variability in species richness of rare coastal plain species but only 45% of that for the background flora. Richness was best correlated with the two inter-related variables, watershed area and shoreline width. The mechanism underlying this pattern appears to be that flooding kills woody plants, thereby reducing competition from shrubs and creating open expanses of shoreline.