Chapter 5 . Applications of Adaptive Agents
77
(2001) identified two reasons why individual-based modelling has so far not been
a 'very productive approach to ecology or ecological management': (1) the failure
to encode models in software that allows the behaviour of the model's individuals
to be observed and tested, (2) the use of inappropriate assumptions abound in
individual-based models such as: using model components developed originally
for one set of assumptions to simulate conditions under which those assumptions
clearly are not met; applying relations and parameters developed for one spatial or
temporal scale to other scales that they are not appropriate far; embedding
empirical relations in models that are purported to be mechanistic; confusing
individual- and population-level parameters. However Holland's paradigm on
adaptive agents conceptualised in Echo (Fig. 5.1.) seems to reinforce individualbased modelling as it can be applied to spatially explicit simulation of individuals
of species by single agents.
Gecko (Booth 1997) is an example of an individual-based adaptive agents
simulation system implemented on the Echo framework. In order to simulate
community effects of spatial competition, agents in Gecko are not constrained by
lattice but extent and compete directly for space. In extension of Echo, Gecko
simulates energetics explicitly. "The individual-based components of Gecko
include cross primary production, over a constrained space, and several basic
species types. Organisms are abstracted as spheres on the resource-producing
plane. They have behaviours to acquire food, assimilate food at realistic
efficiencies, and pay metabolic taxes at allometrically specified rates ... Creatures
interact locally, with their neighbourhoods circumscribed by their radii,
determined in turn by their size - the biomass they have amassed. Thus in addition
to having position in two dimensions, Gecko creatures have the spatiotemporal
property of extent" (Booth 1997). Applications of Gecko to a hypothetical
terrestrial food chain consisting of a plant, a herbivorous and a camivorous animal
revealed that the sequential implementation of seven basic rules for agent to agent
interactions during each simulation step allowed to produce qualitatively sound
results for different scenarios (Booth 1997; Schmitz and Booth 1997). Examples
of basic rules considered in this study are: give everyone a chance to interact;
distribute abiotic resources to autotrophic agents at the site; take maintenance tax;
etc. In an attempt to simulate the growth of Eseheriehia eoli from a single cell to a
bacterial colony Gecko was implemented in a deterministic manner based on
equations for cell growth kinetics, diffusion and colony expansion, and produced
qualitatively sound results regarding colony structures for different glucose
concentrations (Kreft, Booth and Wimpenny 1998).
Even though above described examples were rather simple and vague
reflections of real ecosystems the authors bewailed the fact that their achievements
were limited by causal knowledge from the individual to the ecosystem level - a
limitation that is inherent to deductive approaches strictly based on causal
knowledge.
77
(2001) identified two reasons why individual-based modelling has so far not been
a 'very productive approach to ecology or ecological management': (1) the failure
to encode models in software that allows the behaviour of the model's individuals
to be observed and tested, (2) the use of inappropriate assumptions abound in
individual-based models such as: using model components developed originally
for one set of assumptions to simulate conditions under which those assumptions
clearly are not met; applying relations and parameters developed for one spatial or
temporal scale to other scales that they are not appropriate far; embedding
empirical relations in models that are purported to be mechanistic; confusing
individual- and population-level parameters. However Holland's paradigm on
adaptive agents conceptualised in Echo (Fig. 5.1.) seems to reinforce individualbased modelling as it can be applied to spatially explicit simulation of individuals
of species by single agents.
Gecko (Booth 1997) is an example of an individual-based adaptive agents
simulation system implemented on the Echo framework. In order to simulate
community effects of spatial competition, agents in Gecko are not constrained by
lattice but extent and compete directly for space. In extension of Echo, Gecko
simulates energetics explicitly. "The individual-based components of Gecko
include cross primary production, over a constrained space, and several basic
species types. Organisms are abstracted as spheres on the resource-producing
plane. They have behaviours to acquire food, assimilate food at realistic
efficiencies, and pay metabolic taxes at allometrically specified rates ... Creatures
interact locally, with their neighbourhoods circumscribed by their radii,
determined in turn by their size - the biomass they have amassed. Thus in addition
to having position in two dimensions, Gecko creatures have the spatiotemporal
property of extent" (Booth 1997). Applications of Gecko to a hypothetical
terrestrial food chain consisting of a plant, a herbivorous and a camivorous animal
revealed that the sequential implementation of seven basic rules for agent to agent
interactions during each simulation step allowed to produce qualitatively sound
results for different scenarios (Booth 1997; Schmitz and Booth 1997). Examples
of basic rules considered in this study are: give everyone a chance to interact;
distribute abiotic resources to autotrophic agents at the site; take maintenance tax;
etc. In an attempt to simulate the growth of Eseheriehia eoli from a single cell to a
bacterial colony Gecko was implemented in a deterministic manner based on
equations for cell growth kinetics, diffusion and colony expansion, and produced
qualitatively sound results regarding colony structures for different glucose
concentrations (Kreft, Booth and Wimpenny 1998).
Even though above described examples were rather simple and vague
reflections of real ecosystems the authors bewailed the fact that their achievements
were limited by causal knowledge from the individual to the ecosystem level - a
limitation that is inherent to deductive approaches strictly based on causal
knowledge.