Biology Reference
In-Depth Information
years, and there do appear to be traits that have some generality for predicting
invasiveness (Kolar and Lodge 2001; Rejmanek et al. 2005), especially for particu-
lar taxa (Rejmanek and Richardson 1996; Grotkopp et al. 2002).
A number of models have been developed that attempt to predict the likelihood
of different species becoming invasive if they are introduced in an area (e.g.
Rejmanek and Richardson 1996; Reichard and Hamilton 1997; Pheloung et al. 1999;
Daehler et al. 2004). Some of these models have good predictive ability even outside
geographic areas in which they were developed (Krivanek and Pysek 2006; Pauchard
et al. 2004). A potential limitation is that both the decision tree and statistical models
require a large amount of detailed information which is not always available, such
as species life-history characteristics or environmental conditions. On a more funda-
mental level, the models have often been applied at much larger scales (e.g., coun-
tries, bioregions, or even continents) than the effective scale of most early detection
programs (i.e., local or designated management units). Although they may be useful
for predicting
what
species might become invasive over a large geographic region,
they generally do not predict
where
species are most likely to become established at
a scale appropriate for most early detection programs.
Early detection programs are generally targeted at species early in the coloniza-
tion phase of invasion and implemented at local or, perhaps, regional scales.
However, in some instances, there may be a need to develop an early detection pro-
gram for a large geographic area. In these cases, there is a group of predictive models
known as climatic-envelope models (CEM) that form a bridge between the preintro-
duction models discussed above and postintroduction models. CEMs are based on
general relationships between climate and species biogeographic patterns (Rouget
et al. 2004), and require little if any detailed species life-history information or envi-
ronmental characteristics. Predictions are for large geographic areas, but they have
the flexibility to be applied to species in either preintroduction or postintroduction
phases. Information needed for developing CEMs includes the native and nonnative
ranges of the species, basic climatic data for where the species occurs, productivity
(which rainfall can often be a surrogate for), and evapotranspiration (Table 2.3).
2.5.2.1 Postintroduction Prediction Models for Single Species
Postintroduction predictive models are often developed with preexisting data from
plant surveys and GIS data. The fundamental ecological concept that is the founda-
tion of most predictive modeling studies is the ecological niche (Grinnell 1917;
Hutchinson 1957; MacArthur 1968). A species' fundamental niche is determined
by a large number of abiotic, biotic, and behavioral factors. Where species actually
occur is best conceptualized as its realized niche (e.g., Austin and Meyers 1996).
Although a species could have greater ranges of distribution, biotic interactions
(e.g., competition, predation, pathogens), the lack or limitation of important
resources (e.g., moisture, light), and/or the inability to cross barriers restricts its
actual distribution. Consequently, predictive models are based on data of a species'
realized niche. Differentiation between the fundamental niche and the realized
