Geoscience Reference
In-Depth Information
Obviously the approach just described has some drawbacks. Without an
adequate description of the cause-effect relationship between the species and
environmental variables, models lose in transferability, in both space and time,
and this limits their predictive capabilities (Levin 1992).
ASSUMPTIONS
All models analyzed extrapolate their results to an entire study area on the
assumption of space independence of the phenomenon observed at a given
place. That is, in the case of both a deductive and an inductive approach, the
species-environment relationship is built on evidence that a certain species
occurs somewhere and that we know the values of the environmental variables
at those locations. Obviously we know only that a species occurs at locations
where it has been observed, only part of these locations have measurements of
the environmental variables, and usually these measurements are collected only
for the limited time range during which the investigation was carried out. Thus,
when building distribution models, evidence collected in a portion of the range
is extrapolated to the entire range of occurrence of a species. In order to do so,
it is assumed that the species-environment relationship used to build the model
is invariant in space and time. Most of the time this is not the case, especially
for species with a wide range and for generalist species. In fact, the higher the
variance of the species-environment relationship, the higher the number of
locations required to provide an adequate ecological profile for the species.
Second, it is generally implicitly assumed that variables that are not
included in the analysis have a neutral effect on the results of the model. That
is, we need to assume either that the species' ecological response to these envi-
ronmental variable is constant or that the response is highly correlated with the
other variables included.
Even though both of these general assumptions are very difficult to test, we
believe that they should be discussed on a case-by-case basis because the result
of their violation is species-specific. Errors may be negligible in certain cases
but can introduce major interpretation problems in other cases.
Biological assumptions
Biological assumptions are direct consequences of the general assumptions dis-
cussed in the previous paragraph. We nevertheless believe that they are proba-
bly the most critical, but have received minimal attention in the literature.
The first assumption, which follows from the general assumption of space
