Table E-l. Model and description (continued) 
Discriminant analysis. Rejmanek and Richardson (1996) use a simple discriminant analysis to determine 
predictors of invasiveness. The model uses 10 life-history characteristics of cultivated pine species and classifies 
them as invasive characteristics or non-invasive characteristics. Results indicate three traits predict invasive species: 
short juvenile period, short interval between large seed crops, and small seed mass. The best predictor for 
herbaceous plants is their latitudinal range. The authors recommend this model as general screening tool for 
detection of invasive, woo dy seed plants. _ 
Discriminant function and principal component analyses.* Mandrak (1989) uses discriminant function analysis 
(DFA) and principal components analysis (PCA) to compare the ecological characteristics of potential invasive 
species to recently invading species to determine potential invaders’ response to climate change. The DFA uses a 
covariance matrix, and the PCA uses a correlation matrix. Analyses show that 27 of the 58 possible invaders 
(46.5%) are considered to be potential invaders of the lower and upper Great Lakes region. Eight potential invaders 
are thermally restricted to the Lower Great Lakes region, however, under climate change, their spread could be 
relatively swift. The author concludes that management practices such as stocking and rehabilitation implemented 
to maintain cool and cold water fisheries may be altered by rapid increases of warmwater species. 
Ecological niche modeling. Peterson (2003) reviews techniques for ecological niche modeling, an approach that 
requires the following assumptions: (1) species distribution is limited by its ecological niche; and (2) a species can 
only disperse to an area with similar ecological characteristics. The results of the author’s review indicate that the 
ecological niche or geographic element of an invasion constrains the distribution potential of a species. The author 
concludes the potential contribution of ecological niche modeling on the prediction of the potential range of an 
invasion has not been fully appreciated. The author also notes that invasive species predictions can be integrated 
with global change predictions. _1 
Ecological niche modeling/Genetic algorithm for rule-set production. Peterson et al. (2003) use the genetic 
algorithm for rule-set production (GARP) to model ecological niches of and predict the geographic distribution of 
four North American invasive plants. GARP models relate ecological traits of areas where a species is located to 
points sampled randomly from the rest of the test area to determine decision rules that best describe those traits 
associated with the species’ presence. Results show that ecological niche models based on the native range of 
species can predict invasions; thus, the authors conclude that GARP approach to modeling has more precise 
predictive power than other approaches. 
Ecological niche model/GARP. Underwood et al. (2004) developed a model using GARP to predict non-native 
species’ environmental niches in Yosemite Valley, considering elevation, slope, and vegetation structure. The 
results demonstrate the predictive potential of GARP for identifying potential invasion sites. The authors conclude 
that similar models can be developed for other national parks and that such models may increase the efficiency of 
fieldwork and monitoring while decreasing cost to managers. 
Ecological niche model/GARP. Peterson and Vieglais (2001) outline a framework for developing projections of to 
identify the risk of invasion by species from a specific region. The authors describe the procedure for modeling 
ecological niches, focusing on GARP and describe several tests of GARP’s accuracy. The framework depends on 
availability of biodiversity data. The authors identify future possibilities for using GARP, including using models to 
help create avoidance strategies based on what activities could result in invasions. They also note that it will be 
important to work with scientists and managers that have valuable biodiversity data. 
Bioclimate envelope model.* Pearson and Dawson (2003) review bioclimate envelope models, discuss limitations, 
and propose that the model can be useful as a first approximation to understand climate impact on biodiversity. 
Bioclimate envelope models consider only climate variables of a species range and no other factors that can affect 
species’ distributions. The authors state that it is not possible to accurately predict biogeographical responses to 
climate change, but that bioclimate models may be the best available guide for making policy decisions. The 
authors recommend a hierarchical modeling framework with climate as a dominant factor on a large, continental 
scale and biotic factors dominant at micro-scales. 
E-4 
