Abstract
AIM: Mapping the geographic distribution of non-native aquatic species is a critically important precursor to understanding the anthropogenic and environmental factors that drive freshwater biological invasions. Such efforts are often limited to local scales and/or to single species, due to the challenges of data acquisition at larger scales. Here, we map the distribution of non-native freshwater species richness across the continental United States and investigate the role of human activity in driving macro-scale patterns of aquatic invasion. LOCATION: The continental United States. METHODS: We assembled maps of non-native aquatic species richness by compiling occurrence data on exotic animal and plant species from publicly accessible databases. Using a dasymetric model of human population density and a spatially explicit model of recreational freshwater fishing demand, we analysed the effect of these metrics of human influence on the degree of invasion at the watershed scale, while controlling for spatial and sampling bias. We also assessed the effects that a temporal mismatch between occurrence data (collected since 1815) and cross-sectional predictors (developed using 2010 data) may have on model fit. RESULTS: Non-native aquatic species richness exhibits a highly patchy distribution, with hotspots in the Northeast, Great Lakes, Florida, and human population centres on the Pacific coast. These richness patterns are correlated with population density, but are much more strongly predicted by patterns of recreational fishing demand. These relationships are strengthened by temporal matching of datasets and are robust to corrections for sampling effort. MAIN CONCLUSIONS: Distributions of aquatic non-native species across the continental US are better predicted by freshwater recreational fishing than by human population density. This suggests that observed patterns are driven by a mechanistic link between recreational activity and aquatic non-native species richness and are not merely the outcome of sampling bias associated with human population density.