Melliodendron xylocarpum Handel-Mazzetti,a tree species endemic to China,is of extremely high ornamental value.To understand its response of potential geographical distribution pattern to climate change can shed light on effective protection and reasonable utilization of the species.In this study,we used ENMeval to optimize parameters of Maxent model.The potential distributions during different ages,e.g.the Last Glacial Maximum,mid-Holocene,present,and year 2070 under the scenario of representative concentration pathway 8.5 were simulated by the model based on 137 modern geographical distribution records and 8 bioclimatic variables.Also,the multivariate environmental similarity surface analysis and the most dissimilar variable analysis were used to evaluate the climate anomaly and its key influencing factors,while the percent contribution,permutation importance and Jackknife test were applied to assess the importance of each bioclimatic variable.The results showed that: (1) when the feature combination was linear,quadratic,hinge,product,and threshold,and regularization multiplier being 2,the complexity and degree of overfitting of the model were relatively low.As a result,the accuracy of the model was very high as indicated by the area under the receiver operator characteristic curve for test data up to 0.98530.0055.(2) The highly suitable region for the present distribution was at the junction of Guangxi,Guangdong,and Hunan provinces.During the Last Glacial Maximum,the area of the suitable region was significantly reduced and the suitable region in south China moved southward.In the mid-Holocene,the area of suitable region was increased but no obvious movement occurred.In 2070,the annual mean temperature and maximum temperature of warmest month were the most abnormal variables in low altitude regions,and highly and moderately suitable areas may remain in high altitude regions.(3) Both the temperature and precipitation factors restricted the potential geographical distribution of M.xylocarpum,but the former was more likely to lead to the change of geographical distribution.