Predicting the impact of climate change on potential habitats of fir (Abies) species in Japan and on the East Asian continent

Abstract

Statistical models such as the Generalized Additive Model (GAM) and Multivariate Adaptive Regression Splines (MARS) used to predict potential habitats or distributions of wild plants, here referred to as species distribution models (SDMs), are useful tools for large-scale and long-term impact assessments. There have been fewer studies that used SDMs to predict the impact of climate change on potential habitats of plants in Asia than in western countries. Although spatial data on plant species distributions and environmental factors such as climatic parameters are necessary for modeling, the species distribution data usually constrain such modeling studies. We built models predicting the potential habitats of four Abies species ( A. mariesii , A. veitchii , A. homolepis , and A. firma ) endemic to Japan using species distribution data comprising detailed location and elevation records extracted from the Phytosociological Relevé Database (PRDB). We also created a model predicting the habitats of Abies nephrolepis , native to East Asia, using approximately 20-km mesh distribution data digitized from a map of its distribution in China. The potential habitat projections for the four Japanese Abies species at a ca. 1-km resolution under current and future climate conditions indicated that three species presently found at high elevations on mountains would lose large areas of habitat in Japan. The projection for A. nephrolepis at ca. 20-km resolution indicated that this species would gain more habitat area in the north but lose habitat in lowlands relative to its current distribution in northeastern China. Abies species occurring in non-habitats after climate change would decline gradually and be replaced by other plant species. Although a potential habitat projection of a plant on the East Asian continent can be performed by SDMs using low-resolution plant distribution data, SDMs based on more detailed distribution data covering the entire distribution of a species are necessary for practical conservation plans.