Ecological niche models and the geography of biological invasions: a review and a novel application - Invasive Plants: Ecological and Agricultural Aspects - page 31

to date [5, 11, 19] reported substantial overlap between the predicted distribu-
tion in the invaded region and the actual distribution of the invasive species in
its new range. This high degree of overlap between the prediction and the actu-
al invaded distribution suggests 1) that abiotic factors play a dominant role in
structuring the distribution of species and 2) that particular constraints on the
distribution of species are conserved when a species invades a new region.
However, in these retrospective studies, there also were large regions where
the predictions and the invaded ranges did not overlap (i.e., where models pre-
dicted presence of the invader, but where the invader was absent (commission
error) or vice versa (omission error)). The interpretation of these errors is prob-
lematic, because omission errors may represent a difference in which factors
are governing the native and invasive ranges, or may result simply from defi-
ciencies in the model (e.g., the failure to include an important environmental
factor). Moreover, commission errors may reflect suitable habitat that is as yet
unoccupied (because the invader is still spreading into new habitat), or may
again simply reflect deficiencies in the model. Because of these uncertainties,
retrospective analyses alone often are inadequate to test fully for differences
between native and invaded range constraints on the distribution of a species.
Testing the predictability of invasions: the reverse-ENM approach
Another approach to evaluate differences in which factors govern the native
and invaded range of a species is to use the invaded distribution essentially to
back-predict the native distribution. For example, consider the situation where
the two key assumptions of ENM are met without exception - occurrence
points from the native distribution of a species can be used to model perfectly
its ecological niche (i.e., its distribution is not influenced by biotic interac-
tions, and it is not dispersal limited, Forward-ENM in Fig. 1) and the ecolog-
ical niche of the species is conserved when it invades a new region. If this spe-
cies is introduced onto a new continent, and is given enough time to spread to
all available habitats (i.e., reach a distributional 'equilibrium'), our perfect
model should replicate the invaded range of this species. Conversely, because
niche occupancy is the same in the invaded and native ranges, an ecological
niche model based on the invaded range of the invader should duplicate its
native range (Reverse-ENM in Fig. 1).
Now consider the more realistic case in which a species in its native range
occupies only a portion of available habitats because of biotic interactions and
limitations to dispersal. As before, we introduce this species onto a new conti-
nent, and assuming its ecological niche remains stable, allow it to reach equi-
librium. However, this species is limited less by dispersal, or leaves behind
pathogens and predators. As a result, the invader is able to survive and repro-
duce under environmental conditions in its invaded range that were unreach-
able or unsuitable in its native range. In terms of the niche of the species, this
invader occupies a different niche in its invaded range than in its native range.