Assessing the potential impacts of climate change on the alpine habitat suitability of Japanese stone pine (Pinus pumila)

To assess the potential distribution of Pinus pumila, a dominant species of the Japanese alpine zone, and areas of its habitats vulnerable to global warming, we predicted potential habitats under the current climate and two climate change scenarios (RCM20 and MIROC) for 2081-2100 using the classification tree (CT) model. The presence/absence records of P. pumila were extracted from the Phytosociological Relev, Database as response variables, and five climatic variables (warmth index, WI; minimum temperature for the coldest month, TMC; summer precipitation, PRS; maximum snow water equivalent, MSW; winter rainfall, WR) were used as predictor variables. Prediction accuracy of the CT evaluated by ROC analysis showed an AUC value of 0.97, being categorized as "excellent". We designated Third Mesh cells with an occurrence probability of 0.01 or greater as potential habitats and further divided them into suitable and marginal habitats based on the optimum threshold probability value (0.06) in ROC analysis. Deviance weighted scores revealed that WI was the largest contributing factor followed by MSW. Changes in habitat types from the current climate to the two scenarios were depicted within an observed distribution (Hayashi's distribution data). The area of suitable habitats under the current climate decreased to 25.0% and to 14.7% under the RCM20 and MIROC scenarios, respectively. Suitable habitats were predicted to remain on high mountains of two unconnected regions, central Honshu and Hokkaido, while they were predicted to vanish in Tohoku and southwestern Hokkaido. Thus P. pumila populations in these regions are vulnerable to climate change.