Aim Biological invasions pose a major conservation threat and are occurring at an unprecedented rate. Disproportionate levels of invasion across the landscape indicate that propagule pressure and ecosystem characteristics can mediate invasion success. However, most invasion predictions relate to species’ characteristics (invasiveness) and habitat requirements. Given myriad invaders and the inability to generalize from single-species studies, more general predictions about invasion are required. We present a simple new method for characterizing and predicting landscape susceptibility to invasion that is not species-specific.
Location Corangamite catchment (13,340 km2), south-east Australia.
Methods Using spatially referenced data on the locations of non-native plant species, we modelled their expected proportional cover as a function of a site’s environmental conditions and geographic location. Models were built as boosted regression trees (BRTs).
Results On average, the BRTs explained 38% of variation in occupancy and abundance of all exotic species and exotic forbs. Variables indicating propagule pressure, human impacts, abiotic and community characteristics were rated as the top four most influential variables in each model. Presumably reflecting higher propagule pressure and resource availability, invasion was highest near edges of vegetation fragments and areas of human activity. Sites with high vegetation cover had higher probability of occupancy but lower proportional cover of invaders, the latter trend suggesting a form of biotic resistance. Invasion patterns varied little in time despite the data spanning 34 years.
Main conclusions To our knowledge, this is the first multispecies model based on occupancy and abundance data used to predict invasion risk at the landscape scale. Our approach is flexible and can be applied in different biomes, at multiple scales and for different taxonomic groups. Quantifying general patterns and processes of plant invasion will increase understanding of invasion and community ecology. Predicting invasion risk enables spatial prioritization of weed surveillance and control.