COS 16-1
Growing here not there: The spatial distribution of an endemic Hawaiian plant species, Schiedea globosa
Understanding the connection between habitat characteristics and species distributions is vital to comprehend species persistence in particular locations. While many models exist to explain the distribution of common and invasive species, such models have not been adequately evaluated for rare or uncommon species. We investigated three different species distribution models (using logistic regression, maximum entropy, and boosted regression trees) for Schiedea globosa, an endemic Hawaiian coastal species.Although not endangered like most species in its genus, the species is considered uncommon. Presence and absence data were used with climate and topographic predictors to develop species distribution models (SDM) in order to predict the impact of changing shoreline on suitable habitat in the next 100 years.
Results/Conclusions:
All models were better at predicting presences and not absences. The best model indicated that elevation, aspect, slope and rainfall are significant predictors of S. globosa presence (> 25% deviance explained). Slope and summer rainfall vary across sites and appear to be significant factors affecting occurrences. Of the five populations sampled, two will be critically affected with current recorded presences lost directly due to shoreline change. Results suggest that there will be a much higher loss of suitable habitat on Maui than Oahu due to average sea-level rise. Endemic species often have narrow ecological ranges, thus predictive models are useful to make inferences on species sensitivities to future habitat change.