Abstract: The distribution and community composition of forest could be affected by rapid climate change. Understanding the suitable distribution range of dominant species in forest and the influence of climate change is the basis of forest resource protection and utilization. We used species distribution model to analyze the distribution dynamics of Quercus lamellosa Sm., an important tree species in semi-evergreen broad-leaved forest with endemic distribution in the Eastern Himalayas. A total of 54 distribution points of Q. lamellosa and eight environmental variables related to temperature and precipitation were collected. To simulate the potential distribution areas of Q. lamellosa in the past (Last Glacial Maximum, LGM), present, and future (2041-2060, 2081-2100), an ensemble model was developed using five species distribution models (Maximum Entropy Model, Artificial Neural Network, Generalized Linear Model, Generalized Additive Models, and Random Forest). The R and ArcGIS were used to estimate the change of the distribution area and centroid over time. The results showed that among those five models, the Random Forest performed the best and the score of Generalized Additive Models was the lowest. The true skill statistics (TSS) and the area under receiver operating characteristic curve (AUC) were 0.987 and 0.999, respectively, showing good simulation. Dominant environmental factors affecting the potential distribution ofQ. lamellosa were mean annual temperature and temperature seasonality. Currently, Q. lamellosais mainly distributed in the Himalaya-Hengduan Mountains. The current distribution area of Q. lamellosais one third less than that in LGM period, and would further decrease with future climate change. The centroid of theQ. lamellosasuitable area migrated northward with climate warming, and the migration speed was positively correlated with the extent of climate change. The potential distribution range of Q. lamellosais more reliable, as it was obtained by integrating analysis results of multiple models in this study. By demonstrating the distribution dynamics of Q. lamellosasuitable area, our results would provide reference for the maintenance of ecological stability and biodiversity conservation of semi-evergreen broad-leaved forest under the background of climate change.

Key words: Quercus lamellosa, biomod2, ensemble model, suitable area, centroid.