8.5 Methodological Conclusions
223
system: storms, ice winters and spatfalls. Although with our assumptions we determined a "normal" distribution pattern, surveys carried out in the years indicated
and presented in Fig. 8.4.4 would lead to completely different results. This emphasizes the major problem of empirical benthic research of how to infer underlying
mechanisms generating distribution from snapshots of the spatial distribution.
Even in our modeL where the mechanisms are, in contrast to reality, very simple
and always the same, the distribution patterns differ enormously from year to year
because they ret1ect not only more or less deterministic, continuous processes but
also the historv of events which determine distributions.
This contingency of actual distribution patterns would also be retlected by a
virtual biofacies survey on the central parts of the hypothetical Swinnplate. Similar
to the real relief casts of Hertweck (1995), a sequence of different benthic communities would show up (i.e., Lanice, M.vtilus, or Arenicola).
The time series in Fig. 8.4.4 show that the dynamics of Lanice is determined by
ice winters. After ice winters the population needs a few years to restore the predisturbance abundances. My til us, on the other hand, is affected by all three kinds of
random events. Spatfalls initiate the dynamics. Storms reduce abundance and after
eight years mussel beds begin to grow old and eventually die out. Likewise, ice
winters reduce the abundance of Mytilus but at the same time may induce spatfalls.
Note again that all these results retlect model assumptions (as is always the case
with models). The aim with TOPOGRID was to run an initial model and to demonstrate what the assumptions of a more realistic, grid-based model would have to
look like, and how these would lead to certain spatial patterns.
8.5
Methodological Conclusions
Applying the grid-based approach of ecological modelling to the macrozoobenthos
of the Wadden Sea proved to be much more difficult than anticipated in view of
the potential of this approach in terrestrial systems. The reasons for these difficulties became clear with the first model. TOPOGRID shows how a grid-based model
which is adapted to the Wadden Sea would have to look. TOPOGRID is, however,
still a demonstration mainly because of our ad hoc assumption about the dependence of major model parameters on topographic height.
Nevertheless, demonstrations offer the chance to go beyond the empirical restrictions of small spatial and temporal scales and to at least think on larger scales.
Still, the ultimate goal of modelling is the production of models that are testable.
However, this goal cannot be achieved in one step.
Our knowledge about local processes is still far from sufficient. At this scale,
i.e. at the scale of tens to hundreds of metres, grid-based models similar to our first
model may help gain a better understanding of local spatial and temporal dynamics. The model of Lanice (Chap. 5.4) is another example of models of this kind.
Models and empirical studies would, however, have to be designed in conjunction
with one another. The results of small-scale field studies, experiments and models
could thus be aggregated to stochastic rules and be fed into large-scale models like
TOPOGRID. To compare model results and reality, large-scale surveys and longterm studies of abundance and occurrence of benthic species are needed.
223
system: storms, ice winters and spatfalls. Although with our assumptions we determined a "normal" distribution pattern, surveys carried out in the years indicated
and presented in Fig. 8.4.4 would lead to completely different results. This emphasizes the major problem of empirical benthic research of how to infer underlying
mechanisms generating distribution from snapshots of the spatial distribution.
Even in our modeL where the mechanisms are, in contrast to reality, very simple
and always the same, the distribution patterns differ enormously from year to year
because they ret1ect not only more or less deterministic, continuous processes but
also the historv of events which determine distributions.
This contingency of actual distribution patterns would also be retlected by a
virtual biofacies survey on the central parts of the hypothetical Swinnplate. Similar
to the real relief casts of Hertweck (1995), a sequence of different benthic communities would show up (i.e., Lanice, M.vtilus, or Arenicola).
The time series in Fig. 8.4.4 show that the dynamics of Lanice is determined by
ice winters. After ice winters the population needs a few years to restore the predisturbance abundances. My til us, on the other hand, is affected by all three kinds of
random events. Spatfalls initiate the dynamics. Storms reduce abundance and after
eight years mussel beds begin to grow old and eventually die out. Likewise, ice
winters reduce the abundance of Mytilus but at the same time may induce spatfalls.
Note again that all these results retlect model assumptions (as is always the case
with models). The aim with TOPOGRID was to run an initial model and to demonstrate what the assumptions of a more realistic, grid-based model would have to
look like, and how these would lead to certain spatial patterns.
8.5
Methodological Conclusions
Applying the grid-based approach of ecological modelling to the macrozoobenthos
of the Wadden Sea proved to be much more difficult than anticipated in view of
the potential of this approach in terrestrial systems. The reasons for these difficulties became clear with the first model. TOPOGRID shows how a grid-based model
which is adapted to the Wadden Sea would have to look. TOPOGRID is, however,
still a demonstration mainly because of our ad hoc assumption about the dependence of major model parameters on topographic height.
Nevertheless, demonstrations offer the chance to go beyond the empirical restrictions of small spatial and temporal scales and to at least think on larger scales.
Still, the ultimate goal of modelling is the production of models that are testable.
However, this goal cannot be achieved in one step.
Our knowledge about local processes is still far from sufficient. At this scale,
i.e. at the scale of tens to hundreds of metres, grid-based models similar to our first
model may help gain a better understanding of local spatial and temporal dynamics. The model of Lanice (Chap. 5.4) is another example of models of this kind.
Models and empirical studies would, however, have to be designed in conjunction
with one another. The results of small-scale field studies, experiments and models
could thus be aggregated to stochastic rules and be fed into large-scale models like
TOPOGRID. To compare model results and reality, large-scale surveys and longterm studies of abundance and occurrence of benthic species are needed.