用GARP生态位模型预测刺萼龙葵在我国的潜在分布区

Solanum rostratum Dunal. (buffalobur), as a quarantine weed in China, has been spread to North China in recent years. In order to monitor and control this weed effectively, it is necessary to investigate the potential distributional areas of this weed. In this study, both an ecological niche modeling method, called Genetic Algorithm for Rule-set Prediction (GARP) and Geographic Information System (GIS) program were applied to build the ecological niche models, based on the data of S. rostratum in the current distribution in the native and other invaded areas. The Area Under Curve of Receive Operation Curve (ROC) test is 0.816 and indicate good performance of our models. The models were applied to predicting the potential distribution of S. rostratum in China. The results suggested that the potential distribution of S. rostratum is highly extensive, the potential distribution of S. rostratum include all areas of China except Tibet, Qinghai, Hainan province, some part of Xijiang Uigur Autonomous Region, south part of both Guangdong,Guangxi Zhuang Autonomous Region and east part of Taiwan Province. North, East and Center China show high suitability especially. We suggest the related department may pay more attentions to this danger weed because it probably enters a hazard rapid-spread period.