Background/Question/Methods At the scale of broad climatic zones, the bioclimatic niche in a weed’s native range is an important predictor of its ability to spread in a new region: temperate species generally invade temperate habitats while tropical species tend to invade tropical habitats. However, at finer scales other factors such as soil, disturbance regime and competitors may interact with climate, leading to more complex weed distribution patterns. Species distribution models (SDMs) are a promising approach for predicting the geographic distribution of weeds. SDMs can incorporate course-scale as well as fine-scale environmental data to generate weed distribution predictions. We used European weeds in Hawai‘i that have invaded across a tropical-to-alpine elevation gradient to build SDMs and test their predictions. Further, we used data from a global network on plant invasions into mountains (Mountain Invasion Research Network, MIREN) to examine the generality of results from SDMs developed for Hawaii. Results/Conclusions Climatic variables interacted with habitat characteristics to determine weed distributions, and SDMs can be used to model these interactions and make useful predictions. We suggest three uses of SDMs for invasive species management. First, SDMs identify areas at high risk of invasion so that pro-active measures can be taken to minimize transport and establishment of weeds into those areas. Second, SDMs provide spatial visualizations of invasion patterns and processes across patchy landscapes, allowing managers overlay patterns of infestation and land use categories (pasture, roadsides, natural areas, etc) in order to develop efficient strategies for containment and control. Finally, SDMs, together with experimental studies and mechanistic modeling, can facilitate interactions and knowledge exchange among researchers and managers who are concerned with invasion scenarios associated with climate change and shifts in land use practices. Because SDMs are easily refined as new data on weed occurrences become available, they are easily integrated into an adaptive management framework.
