Abstract: In recent years, quarantine weed Solanum rostratum has being spread to North China. Based on the distribution characteristics of S. rostratum in its native and invaded areas, and by using 14 environmental variables, the potential distribution areas of S. rostratum in China were predicted by the Genetic Algorithm for Rule-set Prediction (GARP) niche model combining with Geographic Information System (GIS). The results from jackknife operation indicated that 7 environmental variables could improve the accuracy of the prediction. The one-tailed goodness-of-fit χ²-statistics was employed to test the significance of the model, which indicated that all of our predictions were highly significant. Moreover, the Receive Operation Curve (ROC) test by using leaving-aside data indicated the good performance of our model, and all the known distribution points in China were all correctly predicted. The potential distribution areas of S. rostratum included whole China except Tibet, Qinghai, and south Guangdong and Guangxi, and the Central, North, and East China had high suitability. S. rostratum would probably enter its hazard’s rapid-spread period, and the related departments should pay more attentions to this danger weed.

Key words: Landscape pattern, Landscape change, Taishan Mountains, Anthropogenic disturbances