SPECIAL
ISSUE
Five (or so) challenges for species
distribution modelling
Miguel B. Arau´ jo
1,2
and Antoine Guisan
3
INTRODUCTION
Species distribution modelling is central to both fundamental
and applied research in biogeography. Over the past 10 years
species distribution models have become commonplace in
studies of biogeography, conservation biology, ecology, palaeo-
ecology and wildlife management. Model fitting is usually
based on pattern-recognition approaches, whereby associations
between geographic occurrence of a species and a set of
predictor variables are explored to allow or support statements
of the mechanisms governing species’ distributions. These
models allow estimation of species’ ecological requirements,
although the degree to which causal relationships between
species distributions and the predictor variables are unveiled
depends on the adequacy of the predictors used for model
building. In spite of the widespread use of niche-based species
distribution models (e.g. Guisan & Thuiller, 2005), important
conceptual, biotic and algorithmic uncertainties still need to be
investigated if models are to make important contributions for
conservation and biogeographical research. As with climate
change research (e.g. Smith, 2002), the species’ distribution
modelling community needs to deepen the ongoing debate,
where the strengths and limitations of available approaches are
investigated, fuelled by more rigorous assessments of the
sensitivity of model outcomes to initial assumptions and
parameters (Whittaker et al., 2005). In this essay, we identify
five high-priority areas of inquiry for niche-based species
distribution modelling: (1) clarification of the niche concept;
(2) improved designs for sampling data for building models;
(3) improved parameterization; (4) improved model selection
and predictor contribution; and (5) improved model evalua-
tion. The challenges discussed in this essay do not preclude the
need for development of other areas of species distribution
modelling research, but are critical for allowing this science to
move forward.
CHALLENGE 1: CLARIFICATION OF THE NICHE
CONCEPT
The foundations of niche modelling are deeply rooted in
Hutchinson’s (1957) fundamental and realized niche concepts.
Even though a majority of modellers would subscribe to
Hutchinson’s framework, there are conflicting views about
what the models truly represent. For example, recently
Sobero´ n & Peterson (2005) concluded that niche models
provide an approximation to the species’ fundamental niche.
Others have regarded models as providing a spatial represen-
tation of the realized niche (e.g. Austin et al., 1990; Guisan &
Zimmermann, 2000; Pearson & Dawson, 2003) on the grounds
that the observed species’ spatial distributions that are utilized
1
Department of Biodiversity and Evolutionary
Biology, National Museum of Natural Sciences,
CSIC, 28006, Madrid, Spain,
2
Centre for
Macroecology, Institute of Biology,
Universitetsparken 15, DK-2100 Copenhagen,
Denmark,
3
Department of Ecology and
Evolution, University of Lausanne, 1015
Lausanne, Switzerland
*Correspondence: Miguel Arau´ jo, Department
of Biodiversity and Evolutionary Biology,
National Museum of Natural Sciences, CSIC,
C/Gutie´rrez Abascal, 2, 28006, Madrid, Spain.
E-mail: maraujo@mncn.csic.es
ABSTRACT
Species distribution modelling is central to both fundamental and applied
research in biogeography. Despite widespread use of models, there are still
important conceptual ambiguities as well as biotic and algorithmic uncertainties
that need to be investigated in order to increase confidence in model results. We
identify and discuss five areas of enquiry that are of high importance for species
distribution modelling: (1) clarification of the niche concept; (2) improved
designs for sampling data for building models; (3) improved parameterization;
(4) improved model selection and predictor contribution; and (5) improved
model evaluation. The challenges discussed in this essay do not preclude the need
for developments of other areas of research in this field. However, they are critical
for allowing the science of species distribution modelling to move forward.
Keywords
Fundamental niche, model evaluation, niche models, parameterization, realized
niche, sampling, variable selection.
Journal of Biogeography (J. Biogeogr.) (2006) 33, 1677–1688
ª 2006 The Authors www.blackwellpublishing.com/jbi 1677
Journal compilation ª 2006 Blackwell Publishing Ltd doi:10.1111/j.1365-2699.2006.01584.x

to estimate species–climate relationships are constrained by
non-climatic factors (e.g. Arau´ jo & Pearson, 2005). Conflicting
interpretations of the niche component being represented by
models arise from ambiguities in the original formulation of
the fundamental and realized niche concepts, and from
difficulties in translating Hutchinson’s conceptual framework
into niche modelling, particularly at large biogeographical
scales.
Hutchinson defined the fundamental niche (N) as the
‘n-dimensional hypervolume’ where a species S
1
, in the
absence of competition with species S
2
, is able to persist
indefinitely. The realized niche of species S
1
(N
0
1
) is the part of
the fundamental niche (N
1
) where the species is not absent due
to competition with species S
2
. Following Hutchinson’s
notation, N
0
1
would be the subset of N
1
that does not contain
N
2
:
N
0
1
¼ N
1
 N
2
\ N
1
 N
2
; where S
1
survives: ð1Þ
And, reciprocally,
N
0
2
¼ N
2
 N
1
\ N
1
 N
2
; where S
2
survives: ð2Þ
There are a number of difficulties in adopting such a strict
Hutchinsonian framework in the context of niche modelling.
First, the word niche became firmly entangled with the notion
of interspecific competition (Chase & Leibold, 2003). How-
ever, there is increasing evidence that positive biotic interac-
tions (e.g. mutualism and facilitation) may be as important as
negative interactions for species survival (e.g. Callaway et al.,
2002; Yamamura et al., 2004; Travis et al., 2005). The
entanglement of the niche concept with competition partly
reflects the disproportionate weight given to competition in
Hutchinson’s original writings, but also to ambiguity as to how
other types of interactions would fit in the fundamental and
realized niche framework. Two statements from his ‘conclu-
ding remarks’ (1957) may, however, shed some light on
Hutchinson’s thinking regarding how to integrate biotic
interactions into the niche framework:
(1) it will be apparent that if this procedure [defining the
hypervolume] could be carried out, all…variables, both
physical and biological, being considered, the fundamental
niche of any species will completely define its ecological
properties. (p. 416)
(2) Interaction of any of the considered species [defining the
realized niche] is regarded as competitive…All species other
than those under consideration are regarded as part of the
coordinate system. (p. 417)
It is reasonable to interpret these statements as indicating
that Hutchinson saw biotic interactions other than competi-
tion as comprising the n-dimensional hypervolume defining
the fundamental niche.
It follows from our reading of Hutchinson that limiting
factors (e.g. temperature and presence of mutualist species)
and resource factors (e.g. energy and presence of prey) should
be part of the coordinate system characterizing the funda-
mental niche. Some authors have already implemented this
idea by incorporating distributions of interacting species as
predictor variables within niche models (e.g. Leathwick &
Austin, 2001; Anderson et al., 2002; Gutie´rrez et al., 2005).
One consequence of including both positive and negative
interactions within the niche framework is that the clear-cut
dichotomy between fundamental and realized niches becomes
artificial and its usefulness debatable. If positive interactions
were considered part of the fundamental niche (as implied by
Hutchinson’s statements), then why should negative interac-
tions alone be associated to the realized niche? If the rationale
– as we interpret it – is that the fundamental niche is defined
by the resources and limiting factors required for species’
persistence, and that the realized niche is defined by the
constraints preventing the exploitation of resources, should the
absence of mutualists or facilitators (thus preventing the use of
resources) be included as part of the factors defining the
realized niche? Ambiguities concerning the role of biotic
interactions within the niche framework need to be resolved in
order to allow appropriate integration of these neglected issues
into niche models.
Additional difficulties arise from the utilization of the
Volterra–Gause principle (Hutchinson, 1957) to justify
distinction between fundamental and realized niches. The
principle postulates that two species utilizing, and limited
by, a common resource cannot coexist in the same physical
space. This observation is equivalent to stating that the
realized niches of two co-occurring species do not intersect
and that the species distribution can be restricted by a
superior competitor. However, the relevance of this principle
is likely to be contingent on the spatial grain of the analysis
and the type of organism being considered. Hutchinson was
mainly concerned with small species’ ranges and niches at
the scale of the community, but species distribution models
are often fitted at regional to continental scales. The
characterization of niches is made from gridded species’
occurrence records on maps; the size of the grids can be
large (1–50 km) and species competing for the same
resource may well co-occur within the same grid since they
can shift positions in order to avoid competition. In practice
this means that species with intersecting modelled realized
niches can co-occur in geographical space. Even if geo-
graphical units were small enough to detect patterns of
competitive exclusion, species competing for the same
resources could still reach local equilibrium. Examples
include the case of species with lower degrees of compet-
itiveness that establish in randomly vacant portions of
geographical space (Hutchinson’s ‘fugitive’ species), or
species that use the same resources, at the same place, but
in different periods of time (e.g. diurnal vs. nocturnal
species). Therefore, co-occurring species can have intersect-
ing realized niches and coexist in space and time (see also
Amarasekare, 2003).
Given the discussion above, it is worth asking whether the
distinction between fundamental and realized niches is useful
for niche modelling. A possibility is to discard the fundamental
and realized niche concepts altogether, accepting that any
characterization of the niche is an incomplete description of
M. B. Arau´ jo and A. Guisan
1678 Journal of Biogeography 33, 1677–1688
ª 2006 The Authors. Journal compilation ª 2006 Blackwell Publishing Ltd