Climatic Controls on Natural Forest Distribution and Predicting the Impact of Climate Warming: Especially referring to Buna (Fagus crenata) Forests

摘要

This paper illustrates studies on impacts of climate change on buna (Fagus crenata) forests in Japan and discusses issues for further study. The occurrence probability of buna forests was predicted by tree-based models. In the models, the response variable was presence/absence data on actual distribution of buna forests derived from the Third Mesh Vegetation Database (MVDB). Predictor variables comprised four climatic ones, i.e., the Warmth Index (WI), the mean minimum daily temperature of the coldest month (TMC), the precipitation in the warm season (PRS) and that in the cold season (PRW); and five non-climatic variables, i.e., topography, surface geology, soil, slope aspect and inclination. According to the models developed, DWS values, which evaluate the contribution of each predictor variable to a model, were the highest for PRW, followed by WI and TMC. They were low for PRS as well as for the non-climatic variables. The models showed climatic conditions for varying occurrence probability for buna forests, indicating that the variables controlling buna forest distribution and their thresholds vary among regions. Suitable habitats for buna forests, defined as the areas with over 0.5 in occurrence probability, cover 26,220 km~2 under the current climate. The area of suitable habitats is predicted to decrease to 9% of the current area under the CCSR/NIES climate change scenario and 37% under the RCM20 scenario. Under both scenarios, suitable habitats disappear in Kyushu, Shikoku and the Pacific Ocean side of Honshu, and decline greatly even in the Tohoku area of northern Honshu with the most extensive suitable habitats. The vulnerability index, which indicates the degree to which climatic conditions become worse for buna with climate change, is defined as the reciprocal of the occurrence probability, and its map was depicted. Buna forests in the areas which become unsuitable after climate change may gradually decline along with the death of the buna trees. This method is useful for predicting distributions of plants and vegetation on large spatial and temporal scales. The MVDB provides distribution data for vegetation types but not for plant species. It is necessary to construct databases on plant species distributions in Japan in order to predict habitats for various plant species.