New approaches for sampling and modeling native and exotic plant species richness

We demonstrate multi-phase, multi-scale approaches for sampling and modelin g native and exotic plant species to predict the spread of invasive species and aid in control efforts. Our test site is a 54,000-ha portion of Rocky Mountain National Park, Colorado, USA. This work is based oil previous rese arch wherein we developed vegetation sampling techniques to identify hot sp ots of diversity, important rare habitats, and locations of invasive plant species. Here we demonstrate statistical modeling tools to rapidly assess c urrent patterns of native and exotic plant species to determine which habit ats are most vulnerable to invasion by exotic species. We use stepwise mult iple regression and modified residual kriging to estimate numbers of native species and exotic species, as well as probability of observing an exotic, species in 30 x 30-m cells. Final models accounted for 62% of the variabil ity observed in number of native species, 51% of the variability observed i n number of exotic species, and 47% of the variability associated with obse rving an exotic species. Important independent variables used in developing the models include geographical location, elevation, slope, aspect, and La ndsat TM bands 1-7. These models can direct resource managers to areas in n eed of further inventory monitoring, and exotic species control efforts.