ARS | Publication request: Models for Predicting the Risk of Naturalization of Non-Native Woody Plants in Iowa

Submitted to: Journal of Environmental Horticulture
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: December 26, 2003
Publication Date: April 9, 2004
Citation: Widrlechner, M.P., Thompson, J.R., Iles, J.K., Dixon, P.M. 2004. Models for predicting the risk of naturalization of non-native woody plants in Iowa. Journal of Environmental Horticulture. (22):23-31.

Interpretive Summary: Nursery professionals have introduced many useful non-native landscape trees and shrubs, but potential exists for new introductions to escape from cultivation and become agricultural weeds or invade natural plant communities. Our objective was to develop a strategy to assess the possibility that non-native trees and shrubs cultivated in Iowa might escape from cultivation. After collecting information on reproductive characteristics and native distributions for 100 non-native landscape trees and shrubs grown in Iowa, we tested three approaches to predict their ability to escape cultivation and compared the predictions to what is known about these plants' record of escaping cultivation in Iowa. The three approaches included use of a previously published decision tree designed to work across North America, modification of that decision tree to include traits important regionally, and design of a new, regional decision tree including a geographic-risk factor (based on the plants' native distributions) and selected reproductive and other biological traits. Modifications to the original decision tree and our new decision tree were all more powerful than was the original tree (as measured by ability to classify plants based on risk of escape) without reducing accuracy (as measured by frequency of misclassification). These results will be used to design better risk-assessment models for the North Central US, to help land managers and horticulturists understand risks associated with the cultivation of non-native landscape plants and avoid the introduction of new, invasive trees and shrubs.

Technical Abstract: Nursery and landscape professionals have introduced many useful non-native woody plants for managed landscapes, but the potential exists for new introductions to escape from cultivation and become pests. The objective of this study was to develop a comprehensive strategy to assess the risk of naturalization of non-native woody plants in Iowa. We examined life history traits and native distributions of 100 woody plant species, including 28 species known to naturalize in Iowa and 72 other species not known to naturalize in the state. We tested three approaches to predict naturalization of woody plants in Iowa: (1) direct application of a previously developed decision tree designed to predict naturalization on a continental scale within North America; (2) application of the continental model modified to include traits important on a regional scale, and (3) development of a new regional model that included a geographic-risk component along with life-history traits. Our modifications to the continental model and the new regional model all were more powerful than the unmodified, continental model, as measured by their ability to classify species (classification rate) based on risk of naturalization, without reducing accuracy, as measured by the frequency of misclassification (error rate). Classification rates ranged from 65% for the unmodified continental model to 90% for the modified continental model. Error rates varied from 6% for the new regional model to 20% for the unmodified continental model, including horticulturally limiting error rates of 4% to 17%, respectively.