S .. I f'mhicills when Applying the Grid-Based Approach
215
without any way of testing the model, i.e. testing some of the model predictions,
all explanations of natural phenomena deduced from the model are only possihle
explanations; it is impossible to determine which of a set of competing possible
explanations is correct (Grimm 1994).
For some terrestrial systems (e.g., forests, the spread of rabies among red
foxes), the approach used in the first model did indeed result in models producing
testable predictions (Wissel 1992, Jeltsch & Wissel 1994, Jeltsch et al. 1997). Why
did this not work with our first model? To answer this question, the deficiencies of
the first model are discussed below.
Abiotic factors
In many ecological systems the local environment is determined by the biota to
such an extent that it seems reasonable to ignore the spatial heterogeneities of the
abiotic environment as approximation. For example, most grid-based models of
terrestrial systems ignore not only topography but also heterogeneities of edaphic
factors, precipitation, etc.
Obviously, with the Wadden Sea this kind of simplification does not suffice.
Here abiotic factors determine to a large degree the processes which in turn directly or indirectly determine the spatial distribution pattern (settlement, secondary
dispersal, properties of the sediment, etc.). Thus, a grid-based model tailored to the
Wadden Sea has to take into account essential abiotic factors, for example topography.
Pattern
More or less clear patterns in the distribution of the macrozoobenthos can only be
perceived at the spatial scale of entire sandflats (Hertweck 1995; Chap. 3.5,
Fig. 3.5.1). This pattern is partly a simple zonation of more or less clearly distinguished typical associations of species (Dbrjes 1978) and is partly determined by
the complex topography of the Wadden Sea and the complexity of the current
regime interacting with topography (Chaps. 3.3; 3.4; 9). Therefore, in contrast to
rocky shores, there is no simple zonation pattern following some simple gradient in
the intertidal of the Wadden Sea.
An additional problem is that pattern recognition strongly depends on the criteria to define certain associations of macrozoobenthic species (Michaelis & Bbhme
1994). The criteria used by Hertweck (1995; Chap. 3.5) seem to be most suitable
for comparative studies performed by different research projects, because in addition to the dominant species, characteristics of the sediment are also taken into
account (Michaelis & Bbhme 1994).
The scale we aimed at with our model was either the scale of an entire backbarrier tidal tlat or at least of an entire sandtlat whose boundaries are defined by tidal
channels. But after inspecting the spatial distributions produced by the first model
(Fig. 8.2.1), it became apparent that the entire model area must be rather small
(100 x 100 m 2 or smaller), because at larger spatial scales of observation the distribution of dominant macrozoobenthic species is not that scattered as in Fig. 8.2.1,
but there are usually larger patches or zones dominated by one species.
215
without any way of testing the model, i.e. testing some of the model predictions,
all explanations of natural phenomena deduced from the model are only possihle
explanations; it is impossible to determine which of a set of competing possible
explanations is correct (Grimm 1994).
For some terrestrial systems (e.g., forests, the spread of rabies among red
foxes), the approach used in the first model did indeed result in models producing
testable predictions (Wissel 1992, Jeltsch & Wissel 1994, Jeltsch et al. 1997). Why
did this not work with our first model? To answer this question, the deficiencies of
the first model are discussed below.
Abiotic factors
In many ecological systems the local environment is determined by the biota to
such an extent that it seems reasonable to ignore the spatial heterogeneities of the
abiotic environment as approximation. For example, most grid-based models of
terrestrial systems ignore not only topography but also heterogeneities of edaphic
factors, precipitation, etc.
Obviously, with the Wadden Sea this kind of simplification does not suffice.
Here abiotic factors determine to a large degree the processes which in turn directly or indirectly determine the spatial distribution pattern (settlement, secondary
dispersal, properties of the sediment, etc.). Thus, a grid-based model tailored to the
Wadden Sea has to take into account essential abiotic factors, for example topography.
Pattern
More or less clear patterns in the distribution of the macrozoobenthos can only be
perceived at the spatial scale of entire sandflats (Hertweck 1995; Chap. 3.5,
Fig. 3.5.1). This pattern is partly a simple zonation of more or less clearly distinguished typical associations of species (Dbrjes 1978) and is partly determined by
the complex topography of the Wadden Sea and the complexity of the current
regime interacting with topography (Chaps. 3.3; 3.4; 9). Therefore, in contrast to
rocky shores, there is no simple zonation pattern following some simple gradient in
the intertidal of the Wadden Sea.
An additional problem is that pattern recognition strongly depends on the criteria to define certain associations of macrozoobenthic species (Michaelis & Bbhme
1994). The criteria used by Hertweck (1995; Chap. 3.5) seem to be most suitable
for comparative studies performed by different research projects, because in addition to the dominant species, characteristics of the sediment are also taken into
account (Michaelis & Bbhme 1994).
The scale we aimed at with our model was either the scale of an entire backbarrier tidal tlat or at least of an entire sandtlat whose boundaries are defined by tidal
channels. But after inspecting the spatial distributions produced by the first model
(Fig. 8.2.1), it became apparent that the entire model area must be rather small
(100 x 100 m 2 or smaller), because at larger spatial scales of observation the distribution of dominant macrozoobenthic species is not that scattered as in Fig. 8.2.1,
but there are usually larger patches or zones dominated by one species.