Vegetation and Disturbance - Environmental Modelling: Finding Simplicity in Complexity - page 228

acorns and oak seedlings by rabbits can affect natural oak
regeneration (Martins, 2001).

recently burned areas (Cates and Orians, 1975). However,
the grazing pressure on any one patch of vegetation
depends heavily on the other vegetation types available to
the grazing animals within the vicinity. It is clear that the
spatial habitat patterns in the landscape and the relative
location and size of burned areas will affect the distribu-
tion of herbivores (Quinn, 1986; Oom et al ., 2002).

14.1.3 Fireandgrazing interactions

It has been suggested that where multiple disturbances
occur simultaneously they generally have nonadditive
effects on ecosystems (White and Pickett, 1985). These
effects may be either synergistic (greater than additive)
or antagonistic (less than additive). Fire and grazing may
have antagonistic effects over relatively short timescales.
Grazing reduces plant growth and removes available fuel,
thus reducing the intensity of fire, and their combined
effect is thus less than expected from the sum of their
individual impacts. A good example of these relationships
is presented by Turner (1985) who found that the impacts
of multiple disturbances (clipping, grazing and fire) on
vegetation was not as severe as expected based on the
disturbances applied singly. It is also known that woody
species that evolved to recover rapidly following distur-
bance have a greater ability to recover from browsing
than those species that have slower growth rates (Bryant
et al ., 1991).
Early work by Braun-Blanquet described different suc-
cessional pathways following changes in land use with
different regimes of burning and grazing. Naveh (1974)
described the regrowth of Mediterranean plants and rec-
ognized that the presence of heavy grazing pressure could
affect the regeneration processes. Interactions between
fire and herbivory have been described in Mediterranean
environments for dwarf shrub communities of Calicotome
villosa and Sarcopterium spinosum ; the dominance of the
latter species is reduced with the exclusion of goat graz-
ing (Henkin et al ., 1999). Similarly, Calluna heathland
dynamics have been reported to show huge differences
with or without sheep grazing (Ellenberg, 1988). Miles
(1985) clearly showed how the successional transitions
between different vegetation types could be changed and
redirected by different combinations of grazing and fire
frequency (Figure 14.1).
Quinn (1986) produced a detailed review of the
interactions between mammalian herbivory and postfire
resilience in Mediterranean ecosystems. Clear evidence
has been reported on how the selective feeding of postfire
regrowth by grazing animals affects the interspecific plant
competition (Leigh and Holgate, 1979; Hesp et al ., 1983;
Mills, 1983). After a fire, herbivores can either delay the
recovery processes or change the successional outcome
(Quinn, 1986). Grazing animals have a preference for
young palatable vegetation, so they tend to congregate on

14.2 The model simplification:
simulation of plant growth under
grazing and after fire

The representation of such complex systems in computer
models requires considerable simplification. Models must
have clear objectives and these, together with the availabil-
ity of data, will determine the key processes that should
be included and the level of aggregation that will be
possible without unacceptable loss of precision. Models
with different objectives may therefore have very different
structures. These have ranged in the past from very simple
probability-based models of transitions between vegeta-
tion states represented as Markovian processes through to
detailed spatially explicit individual-based models where
each individual plant is represented by a physiological
growth model.
Markov models are the simplest form of succession
model. Markov models can be applied where a patch of
vegetation can be classified into one of a limited number of
possible states and vegetation changes by transitions from
one state to another. If the probability of transition to
another particular state depends only on the current state
then the process is Markovian. These models have been
used successfully by several authors to describe succession
(e.g. Horn, 1975 for forest trees; Gimingham et al ., 1981
for heathland and Rego et al ., 1993 for Mediterranean
garrigue). However, as has frequently been pointed out,
vegetation is not Markovian (Usher, 1992). More complex
models can be developed (e.g. Malanson, 1984 for coastal
sage scrubs dominated by Artemisia californica, Salvia
leucophylla and S. mellifera ) but these models are perhaps
better used as descriptions of past change, or as simple
null-models for detecting non-random processes (e.g.
Hobbs and Legg, 1984; Lippe et al ., 1985).
A simple development of Markov models is to replace
the transition probabilities with rules that express the
circumstances in which a transition between two different
states will occur. In fact, many descriptive successional
models used qualitative information and have been