Vegetation and Disturbance - Environmental Modelling: Finding Simplicity in Complexity - page 230

species and the mechanism of surviving or recolonizing
an area after fire. A rule-based model of Banksia
species, taking into account interspecific competition for
space, provided some evidence that species coexistence
depended on both fire regime and topographic gradients
(Groeneveld et al ., 2002).
Historically, models of forest dynamics have been
developed either at the stand level, with individual trees
grown deterministically using an annual time-step differ-
ence equation, and by simulations calculating the fate of
each individual, or at the level of some integrated unit,
such as an age cohort. This approach was initially devel-
oped in JABOWA by Botkin et al . (1972) and applied
to different forest ecosystems (e.g. FORET by Shugart
and West 1977; BRIND by Shugart and Noble, 1981)
(see Chapter 23). However, despite their widespread use,
this family of individual-based gap-phase models pro-
vided limited explanations for the simulated successional
response and did not account for spatial processes.
Spatial structure modelling has largely been derived
from individual-based ideology, taking into account
nearest-neighbour interactions (Ford, 1975). These inter-
actions could be a result of shading and/or redistribution
of the soil and water resources among the trees. Hybrid
models that combine the initial individual-based mod-
els with process models were a further step forward
(Mladenoff and Baker, 1999), now taking into considera-
tion the whole forest stand-soil system including the role
of ground vegetation in the redistribution of resources
for growth (Chertov et al ., 1999).
Since the earliest days of systems ecology (Odum, 1983),
the modelling of forest ecosystems as whole entities has
been seen as another important required step. As a conse-
quence, biogeochemical process models were developed
to simulate vertical fluxes of carbon, water, and nitrogen
between ecosystem components based on fundamental
physical and ecophysiological relationships. As in the
FOREST-BGC model, individual trees are generally not
identified and the model assumes horizontal homogene-
ity (Running and Coughlan, 1988; Running and Gower,
1991). The consideration of landscapes as heterogeneous
land areas composed of clusters of interacting ecosystems
initiated landscape ecology as a scientific field (Forman
and Godron, 1986). As a result, regional and landscape
models are now considered as indispensable for forestry
and environmental planning and management but, in
spite of some attempts (Waring and Running, 1998), it
seems that the field of forest simulation has been devel-
oped less intensively than stand models of various types
(Chertov et al ., 1999).

Recent modelling applications have shown how both
patch distribution in the landscape and plant disper-
sal capability can affect the patterns of competition and
dynamics of the vegetation (Higgins and Cain, 2002).
Another simulation study has also shown the importance
of the presence of 'stepping stones', or islands of veg-
etation, in the survival and distribution of species with
short-range dispersal ability (Sondgerath and Schroder,
2001). Both growth rate and dispersal range influenced
the outcome of a spatially explicit population model sub-
jected to patch destruction and recovery dynamics (Johst
et al ., 2002).
There have been several different approaches to mod-
elling the use of plant resources by herbivores. Foraging
theory has been used to make many predictions on for-
aging patterns of herbivores (e.g. Charnov, 1976). It
was initially proposed that ungulates tend to maximize
digestible energy intake (Van Soest, 1982), but other
authors claim that food selection cannot be based simply
on 'energetic optimization' theory because very few chem-
ical constituents, which are usually volatile and unstable,
play the major role in plant defence against browsing
animals (Bryant et al ., 1991).
It is, however, generally accepted that the quality of
browsing is as important - or more important - than
the amount available, and it is known that quality
changes with plant species, plant age and environmen-
tal conditions thereby making predictions more difficult.
Nevertheless, there are some successful examples where,
for example, it has been found that the spatial distribution
of female red deer ( Cervus elaphus ) could be modelled by
an interaction of vegetation biomass and quality (Fonseca,
1998), which is as expected from the dependence of
reproductive success on their ability to acquire resources
(Clutton-Brock and Harvey, 1978). Models of resource
selection have been used extensively in predicting the spa-
tial distribution of animals (Manly et al ., 1993). It is also
important to recall that under drought conditions the spa-
tial distribution of ungulates is often strongly constrained
by the availability of drinking water (Western, 1975).
When simulating the long-term effects of the combi-
nation of fire and grazing in heterogeneous mixed forests
of pine and oak, it was concluded that grazing may
encourage the development of pine stands by selective
foraging on oaks (Turner and Bratton, 1987). These pine
stands probably burn more readily than oak forests, thus
enhancing the spread of fire and reducing potential oak
recovery. Thus, over the long term, grazing may inter-
act synergistically with fire resulting in the reduction of
the forest matrix and the spread of fire-resilient shrubs