Aim

Due to the socioeconomic and environmental damages caused by invasive species, predicting the distribution of invasive plants is fundamental for effectively targeting management efforts. A habitat suitability model (HSM) is a powerful tool to predict potential habitat of invasive species to help guide the early detection of invasive plants. Despite numerous studies of the predictors used in HSMs, there is little consensus about the most appropriate predictors to use in creating ecologically realistic predictions from HSMs.

Location

The contiguous United States.

Methods

We explore 220 invasive terrestrial plant species' existing HSMs constructed with consistent modelling algorithms, background generation methods, predictor resolution, and geographic extent, and calculate the relative importance of predictors for each species. We sort predictors into eight groups (topography, temperature, disturbance, atmospheric water, landscape water, substrate, biotic interaction, and radiation) and compare the importance of predictor groups by plant lifeforms and phylogenetic relatedness.

Results

Human modification and minimum winter temperature were generally the two highest performing individual predictors across the species studied. The highest-performing predictor groups were disturbance, temperature, and atmospheric water. Across lifeforms, there were minimal differences in the influences of predictor groups, although woody plant models exhibited the largest differences in predictor importance when compared with non-woody plant models. Additionally, we found no significant relationship between the importance of predictor groups and phylogenetic relatedness.

Main Conclusions

This study has implications for informing predictor selection in invasive plant HSMs, leading to more reliable and accurate models of invasive terrestrial plants. Our results emphasize the need to critically select predictors included in HSMs, with special consideration to temperature and disturbance predictors, to accurately predict habitat of invasive plant for detection and response of invasive plant species. With more accurate predictions, managers will be better prepared to address invasive species and reduce their threats to landscapes.