Spatial Population Models for Animals - Environmental Modelling: Finding Simplicity in Complexity - page 218

only when there are potential colonists nearby is only slightly
slower, and is considerably quicker than the other two
strategies. It is also cheaper in the sense that it does not incur
costs of reintroduction. Not only can the explicit

representation of space alter a model's dynamic but thinking
spatially in designing monitoring and management programs
can have substantial benefits (see Huxel and Hastings, 1999,
for detailed discussion).

GIS-supported IBM developed to explore the interplay
between habitat loss, tiger ( Panthera tigris ) population
dynamics and human-wildlife conflict in lowland Nepal.
The model also includes a social component in that the
implications of human attitudes towards tigers, especially
in the face of predation of their livestock, are repre-
sented by a variable probability of poisoning. Ahearn
et al . (2001) use their model in an exploratory man-
ner to evaluate the conditions under which tiger-wildlife
conflict is minimized in a multiuse landscape. The inte-
gration of ecological models with socio-economic models
is becoming ever more common as it is recognized that
the protection of biodiversity is more than just an eco-
logical problem (Liu et al ., 2007) (see further discussion
in Chapter 18).
From the description above it may appear that IBMs
represent an (even the?) ideal approach or framework
for modelling spatial population dynamics. However, the
level of detail that such models tend to contain does, of
course, come at a cost. As Levins (1966) observed, allmod-
els are forced to trade off generality, precision and realism.
While, IBMs do offer great flexibility, the cost of this flexi-
bility is a loss of tractability and generality (although note
that not all IBMs are highly detailed - Craig Reynold's
(1987) 'Boids' model provides a counter-example). As
Grimm et al . (1999: 275) note, IBMs are 'hard to develop,
hard to communicate, and hard to analyse.' 1 New meth-
ods have been developed to try to combat some of
these challenges. For example, pattern-oriented mod-
elling (POM - Grimm et al ., 2005) is a strategy that tries
to use observed patterns to hone the structure of com-
plicated ecological models by finding minimal model
structures that reproduce key elements (patterns) of
the phenomenon of interest. Pattern-oriented modelling
emphasizes learning from models; that is, looking for

model outcomes that have not been observed in the field
and using these to generate new hypotheses for future
research. Likewise, new tools have also been developed to
help communicate IBMS; one example is the Overview,
Design concepts, and Details (ODD) protocol proposed
by Grimm et al . (2006).
Finding a model of appropriate complexity hinges on
the question of purpose: why is a specific model being
designed? How will it be used? Just as some questions
demand detailed and realistic models (as indeed some
IBMs are), so there is also a place for 'fast-and-frugal'
models (Carpenter, 2003). Grimm (1999) noted that
most ecological IBMS are designed to address questions
related to specific sites and/or specific taxa. He argued that
this 'pragmatic' approach means that while the individual
modellermight learn a great deal fromthe development of
an IBM, ecology as a whole may not. Of course, the same
could be said of any complicated and site- or taxa-specific
model. Developing approaches that allow the generation
of new theory, rather than just site and system-specific
data, from complicated simulation models remains an
ongoing challenge.

13.3 The research frontier: marrying
theory and practice

The last two decades have seen significant advances in the
development of spatial ecological models, and they now
play a fundamental role in theory generation in ecology.
Due in part to the relative speed with whichmodels can be
developed (one of their inherent virtues), experimental
and empirical testing of much of this theory now lags well
behind the modelling 'frontier' (Agrawal et al ., 2007).
While the need to better link theory and practice is well
recognized, the problem remains that many theoretical
developments remain intractable from a practical stand-
point (Weiner, 1995). Inpart this intractability stems from
the difficulties of designing and replicating large-scale field
experiments, especially spatial ones, capable of evaluat-
ing and refining models. Equally, empiricists have been
slow to embrace the benefits that modelling can provide

1 By 'communicate', Grimm et al . are referring to communication
via formal media such as journal articles; explicitly describing all
of a complicated IBM is difficult. This difficulty is despite the fact
that IBMs are in many ways more intuitive than classical models
because they explicitly model organisms in a way that is analogous
to how we behave and think of ourselves as individuals.