Biology Reference
In-Depth Information
niche has important practical considerations for evaluating the scale to which
model predictions can be extended and for the information collected for developing
models (Thuiller et al. 2005). Because environments are dynamic and heterogene-
ous, factors that influence a species' realized niche can be expected to vary unpre-
dictably, both spatially and temporally. Therefore, a good rule of thumb is to
assemble data on species (e.g., distribution, abundance) and environmental varia-
bles (e.g., elevation, soils) from areas in close geographic proximity to where the
early detection program will be applied. It is also very important that the environ-
ment has not substantially changed since the time when the data were collected.
2.5.3 Postestablishment Prediction Information
for Single Species
Models of species in the spread and equilibrium phases are focused on local scales
(e.g., a reserve, national park, or state forest). At this phase of invasion, nonnative
species have a proven ability to establish themselves and survive regional climatic
conditions. The objective of modeling efforts then becomes predicting where the
species can reproduce, persist, and disperse.
For obvious reasons, developing statistical models for species that are in the
equilibrium phase would not be a good investment of financial or human resources.
Therefore, statistical models are most appropriate for species in the establishment
and, to a lesser degree, the spread phase of invasion. Even then, the usefulness of
these models may be limited. Data might be too sparse for developing models for
species in the establishment phase, because populations are restricted in distribu-
tion and/or abundance. Although species known to be spreading are better suited
for modeling, they may be beyond the point of practical control efforts.
Basic information to gather on species in the establishment and spread phases
are estimates of distribution and abundance (Table 2.3). Predictive models are often
based on presence-absence (incidence) of species in an area, but abundance data
(e.g., cover, density) give a far more ecologically meaningful correlation of the spe-
cies along environmental gradients (Austin 2002; Klinger et al. 2006). Although
incidence-based models have utility, we strongly recommend the use of abundance
data if they can be obtained. Models based on incidence data essentially give equal
weight for species relationships along environmental gradients; a species that
occurs at 10%, 30%, 50%, and 70% values for a given predictor variable provides
the same amount of information at each value (it simply occurs there, but in what
amount we do not know). A species with densities of 10, 40, 60, and 80 at 10%,
30%, 50%, and 70% values for the predictor variable provides much more ecologi-
cal information and has greater predictive value.
Standard environmental data to correlate with species distribution and abundance
patterns include topographic, soil, and land cover variables (Table 2.3). In addition to
these standard environmental variables, invasive species biologists have identified other
variables that are often correlated with the occurrence of invasive plant species (Mack
