210
H Gild 8ascd Modelling
killed, Lanice quickly recolonized areas where it had occurred previously. And
secondly, relief casts (Chap. 3.5) suggest that over the course of years and decades,
at a certain location different dominant communities replace each other (e.g.,
communities dominated or characterized by Mytilus edulis, Arenicola marina,
Lanice conchilega).
Besides these results of ELA W A T, the formulation of the grid-based model
TOPOGRID presented below would not have been possible without the survey of
the benthos in the entire backbarrier tidal flat of Spiekeroog performed by
Hertweck (1995; Chap. 3.5; Fig. 3.5.1). Hertweck's map gives an impression of
the patchiness of benthic communities on the scale of the entire backbarrier tidal
flat. Moreover, the maps allow to draw up hypotheses about local characteristics
(e.g., topographic height, proximity to tidal channels) that determine the local type
of community (e.g., areas with low or high density of Lanice tubes, mussel beds).
8.2
The First Model: A Demonstration
The first model presented is - in the positive meaning of the word - a "naive"
attempt to transfer an extremely simple grid-based approach, which has nevertheless been successfully applied to some terrestrial systems (Wissel 1992, leltsch &
Wissel 1994), to the macrozoobenthos of the Wadden Sea. The model is presented
here because its deficiencies highlight the extensions and modification the gridbased approach needs if it is to be applied to the Wadden Sea. However, the model
also has some value of its own because it allows some basic relationships to be
demonstrated. Very often models are nothing but "demonstrations" of possible
mechanisms (Crick 1988).
In the model, the area under consideration is divided into 32x32 grid "cells"
which may be in one of four different states: empty (i.e. not occupied) because of a
disturbance (indicated by "0" in the parameter names); dominated by lugworms
(Arenicola marina. "A"), sand masons (Lanice cOf/chi/ega, "L"), or blue mussels
(Mytilus edulis, "M").
Every year, the following processes occur one after the other in the model: settlement, succession, the dispersal of Mytilus, and disturbance events. Each process
is characterized by a certain set of transition rules, i.e. rules which determine the
transitions between different states of a certain grid cell. Thus, time is not described continuously but in four discrete steps. Each step summarizes the effects
the corresponding processes have within one year. Running the simulation then
means applying the four sets of transition rules representing the processes in the
model one after the other to each grid cell.
The model rules are probabilistic rules. The probabilities of transitions between
states (Table 8.2.1) of a cell do not reflect hard field data but are estimated based
on general knowledge and assumptions. This underpins the "if-then" character of
ecological models emphasized by Wissel (1989). Ecological models do not claim
to map reality with all its details into the computer or a mathematical formula.
Instead, they are designed to explore the consequences of our assumptions (see
also Starfield 1997).
H Gild 8ascd Modelling
killed, Lanice quickly recolonized areas where it had occurred previously. And
secondly, relief casts (Chap. 3.5) suggest that over the course of years and decades,
at a certain location different dominant communities replace each other (e.g.,
communities dominated or characterized by Mytilus edulis, Arenicola marina,
Lanice conchilega).
Besides these results of ELA W A T, the formulation of the grid-based model
TOPOGRID presented below would not have been possible without the survey of
the benthos in the entire backbarrier tidal flat of Spiekeroog performed by
Hertweck (1995; Chap. 3.5; Fig. 3.5.1). Hertweck's map gives an impression of
the patchiness of benthic communities on the scale of the entire backbarrier tidal
flat. Moreover, the maps allow to draw up hypotheses about local characteristics
(e.g., topographic height, proximity to tidal channels) that determine the local type
of community (e.g., areas with low or high density of Lanice tubes, mussel beds).
8.2
The First Model: A Demonstration
The first model presented is - in the positive meaning of the word - a "naive"
attempt to transfer an extremely simple grid-based approach, which has nevertheless been successfully applied to some terrestrial systems (Wissel 1992, leltsch &
Wissel 1994), to the macrozoobenthos of the Wadden Sea. The model is presented
here because its deficiencies highlight the extensions and modification the gridbased approach needs if it is to be applied to the Wadden Sea. However, the model
also has some value of its own because it allows some basic relationships to be
demonstrated. Very often models are nothing but "demonstrations" of possible
mechanisms (Crick 1988).
In the model, the area under consideration is divided into 32x32 grid "cells"
which may be in one of four different states: empty (i.e. not occupied) because of a
disturbance (indicated by "0" in the parameter names); dominated by lugworms
(Arenicola marina. "A"), sand masons (Lanice cOf/chi/ega, "L"), or blue mussels
(Mytilus edulis, "M").
Every year, the following processes occur one after the other in the model: settlement, succession, the dispersal of Mytilus, and disturbance events. Each process
is characterized by a certain set of transition rules, i.e. rules which determine the
transitions between different states of a certain grid cell. Thus, time is not described continuously but in four discrete steps. Each step summarizes the effects
the corresponding processes have within one year. Running the simulation then
means applying the four sets of transition rules representing the processes in the
model one after the other to each grid cell.
The model rules are probabilistic rules. The probabilities of transitions between
states (Table 8.2.1) of a cell do not reflect hard field data but are estimated based
on general knowledge and assumptions. This underpins the "if-then" character of
ecological models emphasized by Wissel (1989). Ecological models do not claim
to map reality with all its details into the computer or a mathematical formula.
Instead, they are designed to explore the consequences of our assumptions (see
also Starfield 1997).