function of ecosystems before they are invaded. Decades can pass between an introduction and the manifestation of its impact. As a result, we recognize that an invasion has occurred only after the ecosystem has changed. Before a predictive theory about the impact of a potentially invasive plant pest can be developed, better characterization of many more ecosystems in the United States is needed.
Methods or systems for predicting the invasive potential of introduced organisms have focused largely on identifying species that have a record of becoming invasive elsewhere and categorizing their invasiveness according to some schema. Expert judgment has been the most commonly used tool, and collective judgment can strengthen the reliability of such species’ assessments. For example, an effort is under way to have members of an expert network rank wildland weeds in the United States against a list of invasiveness categories (Randall et al. 2001). The value of the proposed categories will be determined by the consistency of the experts’ rankings.
Identification of potentially invasive species according to a suite of life-history traits (mostly in plants) has been attempted with some success. For a few taxonomic groups (woody plant species), the prediction of invasiveness on the basis of traits has been shown to be reliable and should be considered in the regulation of plant importation. Wider application of these approaches to other taxonomic groups has had mixed results.
Climatic simulation models (such as CLIMEX) that identify similar climates around the world offer a preliminary screening tool to evaluate the invasive potential of groups of organisms, based on comparisons of climates in a species’ native and potential new ranges. The crucial limitation in using climate-matching as a predictor of establishment in a new range lies in the assumption that climate alone determines a species’ distribution. Species’ distributions are also strongly influenced by the biotic component of an environment.
Risk assessments conducted by APHIS for the purpose of regulating the importation of commodities or managing potential pests incorporate elements of a predictive system. The main limitations of these predictive elements are that they require subjective, qualitative determination of characteristics of nonindigenous species and the environments into which they might be introduced, and they identify risk by subjectively placing species and environment characteristics into likelihood categories. Thus, the manner in which these assessments are conducted reduces the opportunity for their replicability. Alternative methods that incorporate quantitative scenario analysis constitute an improved approach despite the lingering subjectivity in the probability distribution attached to events.
The weakness in prediction among current models of potential invasiveness and of risk assessments does not mean that they are unscientific. But a scientifically based predictive system for invasiveness should meet three criteria: it must