Abstract: Accurately estimating carbon storage has long been a focus under the context of global climate change. In this study, we collected data from 80 plots in two major Chinese fir plantations production areas in Chinese fir sensitive area to climate change, i.e. Dabie mountainous area and southern Anhui Province, with low human disturbance and different age sequences. Random forest (RF) algorithm was used to construct models that estimate carbon storage \[aboveground carbon storage (C_a), belowground carbon storage (C_b) and total carbon storage (C)\] with candidate independent variables such as stand, site, and climate factors. Based on the optimal model, the sources of the estimation variations in carbon storage (C_a, C_b and C) were quantified. The effects and significance of the influencing factors on carbon storage were evaluated to understand the impacts of climate change on tree growth. The results showed that stand factors were the dominant factors affecting carbon storage (C_a, C_b and C) of Chinese fir plantations, and their relative importance value was over 76%. Among the stand factors, stand basal area (BA) had the greatest influence, with C_a, C_b, and C increasing with the increases of BA. Climate factors were of secondary importance in affecting the carbon storage of Chinese fir plantation, with the relative importance value being less than 17%. Among the climate factors, Precipitation seasonality (Bio15) exerted the greatest influence, with C_a, C_b and C decreasing slightly with the increase of Bio15. Site factors had limited explanatory power, with their relative importance being less than 3%. There were little changes in C_a, C_b and C with the site factors.

Key words: stand factor, climate, site, random forest, carbon storage