Ecological niche models and the geography of biological invasions: a review and a novel application - Invasive Plants: Ecological and Agricultural Aspects - page 28

are dynamic entities: species can respond to a changing environment both by
local adaptation and by shifting their distributions. Failure to consider these
factors when building models can result in misinterpretation of the niche and,
consequently, in prediction errors when the model is projected onto a new
region.
However, recent enhancements in modeling techniques, and increased
availability of fine-resolution climate, topography and landcover datasets,
have led to improved predictive capabilities. Yet, despite improvements, ENM
always may be criticized for several reasons [2, 8]. First, few exotic species
become invasive, even when introduced into suitable environmental condi-
tions [9]. This suggests that other factors also influence the success of inva-
sive species. Secondly, ENM currently cannot incorporate all factors that
limit the distribution of species, such as dispersal rates (but see [10]) and biot-
ic influences (yet, the possibility of incorporating geographic representations
of biotic factors (e.g., as distributional maps of natural enemies) deserves
attention). Therefore, the current geographic distribution of a potential invad-
er, alone, is insufficient to predict whether the factors that govern the native
range of the species also will govern its distribution in a new geographic and
biological setting.
To validate the ENM approach to predicting invasions, ecologists must
answer two questions: 1) Are the distributions of invaders constrained by the
same factors that constrain their native distributions? 2) If not, are the differ-
ences meaningful at the macro-scales of analysis of ENM? The purpose of this
chapter is to investigate these questions. We first briefly review the use of
ENM to predict species invasions. Second, we discuss the assumptions and
limitations of ENM, investigate why biological invasions may violate these
assumptions, and describe how failure to meet these assumptions can result in
prediction errors. Third, we describe a new application of ENM, 1) to test
whether invasive species are subject to the same distributional constraints in
their invaded range as in their native range, and 2) to develop hypotheses as to
why constraints on the distribution of a species in its native and invaded ranges
may differ. Our new application uses a combination of traditional ENM (which
we term “Forward-ENM”), coupled with a unique application of ENM, name-
ly Reverse-ENM. Reverse-ENM uses a niche model based on occurrence
points from the invaded range to model the potential native range (Fig. 1). The
predictions from Forward-ENM and Reverse-ENM are then compared. We
suggest that differences between the forward and reverse predictions may
reveal whether the distribution of an invader is constrained by the same factors
that constrain its native distribution. Finally, we analyze the invasion of the red
imported fire ant (RIFA, Solenopsis invicta Buren) into the United States as an
initial demonstration of the Reverse-ENM method.