Assessing ecosystem vulnerability to invasive rusty crayfish (Orconectes rusticus)

Abstract

Despite the widespread introduction of nonnative species and the heterogeneity of ecosystems in their sensitivity to ecological impacts, few studies have assessed ecosystem vulnerability to the entire invasion process, from arrival to establishment and impacts. Our study addresses this challenge by presenting a probabilistic, spatially explicit approach to predicting ecosystem vulnerability to species invasions. Using the freshwater-rich landscapes of Wisconsin, USA, we model invasive rusty crayfish (Orconectes rusticus) as a function of exposure risk (i.e., likelihood of introduction and establishment of O. rusticus based on a species distribution model) and the sensitivity of the recipient community (i.e., likelihood of impacts on native O. virilis and O. propinquus based on a retrospective analysis of population changes). Artificial neural networks predicted that approximately 10% of 4200 surveyed lakes (n = 388) and approximately 25% of mapped streams (23 523 km total length) are suitable for O. rusticus introduction and establishment. A comparison of repeated surveys before vs. post-1985 revealed that O. virilis was six times as likely and O. propinquus was twice as likely to be extirpated in streams invaded by O. rusticus, compared to streams that were not invaded. Similarly, O. virilis was extirpated in over three-quarters of lakes invaded by O. rusticus compared to half of the uninvaded lakes, whereas no difference was observed for O. propinquus. We identified 115 lakes (approximately 3% of lakes) and approximately 5000 km of streams (approximately 6% of streams) with a 25% chance of introduction, establishment, and extirpation by O. rusticus of either native congener. By identifying highly vulnerable ecosystems, our study offers an effective strategy for prioritizing on-the-ground management action and informing decisions about the most efficient allocation of resources. Moreover, our results provide the flexibility for stakeholders to identify priority sites for prevention efforts given a maximum level of acceptable risk or based on budgetary or time restrictions. To this end, we incorporate the model predictions into a new online mapping tool with the intention of closing the communication gap between academic research and stakeholders that requires information on the prospects of future invasions.