Climate change vulnerability and uncertainty reduction in projections for Dicksonia sellowiana (Pres.) Hook

Abstract

This work had the objective of investigate the impacts of climate changes with different climate settings on the spatial distribution of Dicksonia sellowiana (Pres.) Hook. The survey was carried out by using spatial distribution modeling and methods in order to reduce the uncertainty of these predictions. The software Openmodeller was used to run the modeling with five spatial distribution algorithms. The modeling included the climate model HadCM3 of green house gases emission for the year 2050 for two CO2 emission scenarios: optimistic (B2) and pessimistic (A2). The area under the curve (AUC statistics) and the Kappa statistics resulted high values for all algorithms tested meaning high values of accuracy. Regarding only the species habitats available all algorithms have shown efficient distribution models and the projections were significantly different between the predictions for the future climate and for the current climate. Both models have shown reduction of 50% of the species potential area . Even with the high values of accuracy the variability of predictions for the future spatial distribution models confirm the need of consensus methods to reduce the uncertainty associated with this type of modeling.