Environmental Engineering Reference
In-Depth Information
Background and Development of the Approach
Early applications of this approach go back to the 1970s. They were a response to
the requirements to include more biological realism and explicit spatial represen-
tations into ecological models (Łomnicki 1988). First models were introduced
by Kaiser (1976; territoriality of dragonflies), Hogeweg and Hesper (1983; social
interaction of bumblebees), Seitz (1984; life stage sequences in a Daphnia popu-
lations), and DeAngelis et al. (1979; development of cohort structure in small-
mouth bass populations) and modelling forest stand dynamics (Botkin et al. 1972;
Shugart and West 1977).
The minimum requirement for an individual-based model is the separate repre-
sentation of individual entities, which can be distinguished in one or more char-
acteristics. These characteristics of the individuals must be separately accessible
and tracking of individual state changes must be possible during simulation. In most
of the application cases, the number of states, and the repertoire to modify the states
depends on the internal conditions. In relevant application cases, individual-based
models attempt to provide a coherent picture of how particular organisms would act
in a particular condition.
To use the full potential of the approach it is also possible to include different levels
of entities. Complex modular organisms like trees can be represented as a set of
individual branches, roots, leaves, fruits, etc. The individual organism is then a
compound instance of sub-units. On the other hand, it is also possible - and sometimes
useful - to operate other compound entities. For example, a representation of an
environment can comprise a spatially differentiated structure where physical and
chemical parameter differ locally and give rise to specific local responses of the
organism's activity. Furthermore, abstract entities like populations can be represented,
either as units with specific parameters like age distribution, biomass spectrum, which
change during simulation, or as an aggregate that integrates over the individuals
included in the model. Such an extended specification of an IBM may thus comprise
configurations in which the components are not basic units but particular components
of ecological systems. In principle, these may range from (sub-)cellular units, plant
modules (Breckling 1996; Eschenbach 2005) to aggregations such as cohorts, social
animals (nests, hives), populations, functional types, or spatial or temporal units of
higher order (K
ohler et al. 2003; Middelhoff et al. in print).
With such an extended understanding of how the IBM approach can be used,
one can see that it is in fact structurally identical,
sensu strictu
, to agent-based
models (ABMs). The term ABM originally emerged in a computational context
with applications in physics as well as applications in social sciences and econom-
ics. ABM often describe robotic aggregates responding to a variable environment,
or they simulate complex behaviour of humans in social networks. From an
ecological perspective, it appears reasonable to use both terms (IBM and ABM)
synonymously. In a similar way, the term multi-agent system or multi-agent
simulations (MAS, e.g. Ferber 1999) bases on the same concept, however empha-
sizes the interactions of a larger number of autonomously acting software agents
and is even more common in technical applications.
€
