Cost and consistency of different approaches to risk assessment for invasive species

Background/Question/Methods

Non-native species are regularly moved across the globe through vectors such as the aquarium, pet, live food, ornamental plants, and bait trades. These trades generate benefits for importers, retailers, and consumers, but many of these species become invasive and cause large impacts. Risk assessment tools can be used to predict whether a given species will become invasive, based on its traits, environmental tolerances, and invasion history. Many different approaches have been used to develop risk assessment tools, but there has not yet been a rigorous test of which approach yields the best outcomes. Here, we define two important qualities and test the performance of four different risk assessment tools. First, we look at the cost per species assessed, measured as the hours required to conduct an assessment. Second, we look at the consistency of different tools, measured as the proportion of time that different assessors reach the same conclusion about a species. To conduct these analyses we chose four risk assessment tools that span the range of available approaches. Each tool was used to assess the same set of 63 fishes, and each assessment (i.e., tool X species) was independently (i.e., by different assessors) performed at least twice. 

Results/Conclusions

The four risk assessment tools were all developed for fishes, and range from what we call statistical (i.e., relationships between species traits and invasiveness are developed using statistical tools), to questionnaire (i.e., the tool asks many questions about a species, scores the answers, and uses the sum of those scores to predict invasiveness), to rapid (i.e., if a species has a climate match and history of invasion it is predicted to be invasive). Our results show that the rapid risk assessment, which is also the simplest, takes an average of 1.17 hours to complete. The statistical tool required more time (2.19hours/assessment), and the two questionnaire risk assessment we tested had quite different outcomes (1.44 and 2.86 hours). Differences among all tools were significant (one-way ANOVA p<<0.01). The statistical and rapid were most consistent, with approximately 75% of all species assessed multiple times receiving the same outcome from different assessors. Questionnaire tools had lower consistency. Our interpretation of these data is that the most rapid tool, which is very basic and requires little data, may be the best approach to policy given its low cost and relatively high consistency.