Using species traits to predict invasiveness of aquatic plants in the Great Lakes

Background/Question/Methods

Many plant species are transported globally outside of their native ranges. Only a small proportion of the species that are introduced outside of their native ranges become established, and of those, an even smaller number become invasive. Risk assessment tools have the potential to prevent importation of invasive species, averting both ecological and economic harm. An aquatic weed risk assessment, developed by Biosecurity New Zealand and tested and modified for use in the United States, has high accuracy in predicting both invasive and non-invasive species at the U.S. level. 83 aquatic plant species in the Great Lakes were assessed using this risk assessment tool, modified for use in the region. Three a priori categories were defined in order to be able to test the accuracy of the assessment: not established (present in the trade but not naturalized); established, not invasive (naturalized but not causing ecological impacts); established, invasive (naturalized and causing ecological impacts). Multiple score thresholds for distinguishing species in these categories were determined using statistical analyses. We sought the answer to the question: Which decision thresholds provide the most environmental protection and economic benefit in preventing the importation of invasive species and minimizing restriction of non-invasive species?

Results/Conclusions

Because of factors that can contribute to increased invasiveness over time, established species can become more or less invasive. To determine the best way to distinguish invasive from non-invasive species, our analysis was conducted two ways: once with established, not invasive species grouped with not established species, and again with established, non-invasive species grouped with established, invasive species. In the first analysis, a higher threshold of 57 resulted in 91.6% accuracy, but more established species had scores below the threshold, which would result in classification as non-invasive. This is less precautionary from a regulatory perspective because such species have established and could potentially become invasive. In the second analysis, a lower threshold of 35 yielded 84.3% accuracy, with the majority of established and invasive species categorized as invasive, but with a larger number of not established species incorrectly categorized as invasive. This prohibition of non-invasive species is a concern for industries that import plant species. Eventually, it will be the decision of managers and other stake-holders in the region to implement the assessment methods and decision-thresholds that work best for their region. It is hoped that providing multiple options from statistical analyses will optimize the decision-making process for all stakeholders.