Environmental Engineering Reference
In-Depth Information
The approach becomes operational only if the computational challenges of such a
model can be met. This was made possible by developments in computer program-
ming and programme processing. The representation of a larger number of indivi-
duals with an interaction potential is feasible only with larger processing and data
management capacities. The advancement in hardware and software development
allowed more resource-demanding applications like Object-Oriented Programming
(OOP). In the late 1960s, the programming language SIMULA (Dahl et al. 1968)
provided the ground for the virtual representation of active agents, which was later
adapted to various programming languages (Smalltalk, Delphi, C++, Java, and
others).
Early IBMs often had a narrow focus and concentrated on single species
investigations (e.g. Kaiser 1976; DeAngelis et al. 1979; Seitz 1984). These early
models applied quasi-automatic transition between the single model-states (e.g.
age, biomass, location). However, they could already illustrate the great potential of
the IBM-approach. It thus added a new perspective to modelling in a close relation
to the specific characteristics of ecological systems (see Chap. 4 on systems
analysis), compared to the homogeneity requirements of variables as they were
used in the classical systems dynamic approaches (Forrester 1968).
Further developments made IBMs applicable for investigations of behavioural
decisions and interactions in social groups. Paulien Hogeweg and co-workers
pioneered this field with their model on social interaction in bumblebee colonies
(Hogeweg and Hesper 1983). A first paradigmatic overview was presented by
Huston et al. (1988).
Facilitating a representation of variable environments, structured populations and
behavioural traits, the modelling of complex life histories emerged (e.g. Wolff 1994;
Colasanti and Hunt 1997). For instance, it became possible to simulate highly resolved
time-energy-budgets as a basis for behavioural decision processes. The model on the
reproduction phase of a robin population is such an example (Reuter and Breckling
1999) and allowed to investigate reproduction success under different environmental
settings. A further development in IBM-methodology involved the number of consid-
ered and interacting species. In this context the inclusion of interaction rules plays a
major role. These rules often refer to trophic relations (e.g. Kaitala et al. 2001), spatial
competition or even to succession processes (Breckling 1990).
An increasing number of models combine sophisticated internal resolution of
organismic processes with the representation of several species and their interac-
tions to analyze e.g. food webs and community dynamics. Examples for this type
are the simulation of plant competition including different herbivores by Parrot and
Kok (2002) and the analysis of regular population cycles of small mammals (Reuter
2005, see Sect.
12.4
). Often the simulated entities are designed to operate in
heterogeneous environments including a spatially explicit habitat structure, season-
ality and varying climate data. The models simulate explicitly designed scenarios
directly, using empirical data involving e.g. GIS-derived maps and assumptions on
local temporal and spatial developments. In the marine context we find successful
attempts extending simulation models to include all relevant trophic levels (so-
called end-to-end models, e.g. Travers et al. 2007). These approaches in marine
