76
F. Recknagel
requirements for its own reproduction) from its reservoir to the reservoir of its
trading partner. Though this is a very simple rule, with no bidding between agents,
it does lead to intricate, rational trading interactions as the system evolves. Trades
that provide resources needed for reproduction increase the reproduction rate,
assuring that agents with such rational trading conditions become common
components of the population.
Mating: While an agent can reproduce asexually, simply making a copy of each
of its chromosomes when it has accumulated enough resources (letters), there is
also a provision for recombination of chromosomes. When agents come into
contact and do not engage in combat, the mating condition of each agent is
checked against the mating tag of the other. As with trade, mating is only executed
as a bilateral action. Both agents must have their mating conditions satisfied for
recombination to take place. If this happens, then the agents exchange some of
their chromosome material, as with crossover under the genetic algorithm.
AA characterised by these simply defined capabilities provide for a rich set of
variations illustrating the key kernel properties of complex adaptive systems. They
were originally developed and applied for the study of complex adaptive
economic systems (Holland and Miller 1991) such as stock markets (Wan and
Hunter 1997) and businesses (Lin and Pai 2000). However modified versions of
Echo have meanwhile been used to simulate spatial dynamics of species or
populations represented by individuals strictly based on causal knowledge (Booth
1997; Schmitz and Booth 1997; Kreft, Booth and Wimpenny 1998). These
examples are based on the assumption that local emergence or submergence of
individuals is driven by interrelationships between well-defined individuals and
their environment. Such an individual-based approach seems to be relevant to
terrestrial ecosystems like forests (Schmitz and Booth 1997) where spatial
spreading of individual tree species as an outcome of competitive success is of
major interest.
AA simulation of aquatic ecosystems requires a different
approach as normally neither individual nor spatial aspects are relevant, nor are
adequate data available.
5.3
Individual-Based Adaptive Agents
Individual-based modelling aims at naturally and easily simulating effects of
complex ecological interactions such as individual variation, spatial processes, and
cumulative stress. The concept was introduced by Huston, DeAngelis and Post
(1988) who argued that in ecosystems "amplifying effects can arise from spatial
non-uniformities and variations in the environmental conditions that each
organism experiences, such as moisture and light for plant seedlings, or variable
habitat and patchy distributions of preys for animals. Amplifying effects can also
result from differences among individuals that are properties of the organisms
themselves such as size, age, physiologie al eharaeteristics, and genetic variation".
Even though the coneept appeared to be plausible and applicable especially to the
distinct spatial and heterogeneous nature of terrestrial ecosystems, Railsback
F. Recknagel
requirements for its own reproduction) from its reservoir to the reservoir of its
trading partner. Though this is a very simple rule, with no bidding between agents,
it does lead to intricate, rational trading interactions as the system evolves. Trades
that provide resources needed for reproduction increase the reproduction rate,
assuring that agents with such rational trading conditions become common
components of the population.
Mating: While an agent can reproduce asexually, simply making a copy of each
of its chromosomes when it has accumulated enough resources (letters), there is
also a provision for recombination of chromosomes. When agents come into
contact and do not engage in combat, the mating condition of each agent is
checked against the mating tag of the other. As with trade, mating is only executed
as a bilateral action. Both agents must have their mating conditions satisfied for
recombination to take place. If this happens, then the agents exchange some of
their chromosome material, as with crossover under the genetic algorithm.
AA characterised by these simply defined capabilities provide for a rich set of
variations illustrating the key kernel properties of complex adaptive systems. They
were originally developed and applied for the study of complex adaptive
economic systems (Holland and Miller 1991) such as stock markets (Wan and
Hunter 1997) and businesses (Lin and Pai 2000). However modified versions of
Echo have meanwhile been used to simulate spatial dynamics of species or
populations represented by individuals strictly based on causal knowledge (Booth
1997; Schmitz and Booth 1997; Kreft, Booth and Wimpenny 1998). These
examples are based on the assumption that local emergence or submergence of
individuals is driven by interrelationships between well-defined individuals and
their environment. Such an individual-based approach seems to be relevant to
terrestrial ecosystems like forests (Schmitz and Booth 1997) where spatial
spreading of individual tree species as an outcome of competitive success is of
major interest.
AA simulation of aquatic ecosystems requires a different
approach as normally neither individual nor spatial aspects are relevant, nor are
adequate data available.
5.3
Individual-Based Adaptive Agents
Individual-based modelling aims at naturally and easily simulating effects of
complex ecological interactions such as individual variation, spatial processes, and
cumulative stress. The concept was introduced by Huston, DeAngelis and Post
(1988) who argued that in ecosystems "amplifying effects can arise from spatial
non-uniformities and variations in the environmental conditions that each
organism experiences, such as moisture and light for plant seedlings, or variable
habitat and patchy distributions of preys for animals. Amplifying effects can also
result from differences among individuals that are properties of the organisms
themselves such as size, age, physiologie al eharaeteristics, and genetic variation".
Even though the coneept appeared to be plausible and applicable especially to the
distinct spatial and heterogeneous nature of terrestrial ecosystems, Railsback