Abstract:Qinghai-Tibet Plateau is rich in species, and belongs to the climate change sensitive area. Studying the impact of climate change on the potential distribution of species on Qinghai-Tibet Plateau is of great significance for the biodiversity conservation. In this study, we selected the endangered Tibetan medicinal plant Meconopsis integrifolia as the research object, and constructed an ensemble model of Random Forest (RF), Flexible Discriminant Analysis (FDA) and Artificial Neural Network (ANN) using the weighted average algorithm (WAA). At the same time, we also compared the prediction accuracy of the WAA model and different ecological niche models. Finally, we used the WAA model to predict the potential distributions of M. integrifolia under current (averaged 1970-2000) and future (averaged 2041-2060) climate scenarios. For the uncertainty of future climate, we chose two “Shared Socioeconomic Pathways” (SSP2-45 and SSP5-85). The results showed that: (1) the WAA model predictions showed that the AUC of the ensemble model based on RF, FDA and ANN was 0.926, which was 3% higher than that of the RF model with the highest AUC value, and 5% higher than the AUC value for both FDA and ANN models. (2) The WAA model determined that the potential distribution of M. integrifolia was most sensitive to annual precipitation and precipitation of warmest quarter, followed by max temperature of warmest month, and showed lower sensitivity to precipitation of wettest month and isothermality. (3) The current potential distribution areas of M. integrifolia were mainly distributed in southwestern Gansu, eastern and southern Qinghai, western and northwestern Sichuan, northwestern and northeastern Yunnan, and eastern Tibet. (4) The prediction of potential distribution of M. integrifolia on the Qinghai-Tibet Plateau under future climate change showed that under the SSP2-45 scenario in 2050, the size of potential distribution of M. integrifolia was basically the same as that of current, but the potential distribution would spread to the high altitudes and latitudes in the northwest. However, under the SSP5-85 scenario, the size of potential distribution of M. integrifolia would shrink significantly, and the trend of spreading to the northwest high-altitude and -altitude areas was more obvious.