Chapter 4 . Applications of Evolutionary Computation
65
4.4.7
Modelling Hierarchical Ecosystems
Living systems involve many levels of hierarchical interaction, from the genetic
level through the individual to the community and complete ecosystem (Conrad
and Pattee 1970). Conrad and Pattee argued that a theory of evolution that does
not reflect this structure cannot be expected to be useful in terms of predictions
and models of real systems. They presented a model (EVOLVEI) based on a
population of cell-like organisms subject to a strict mass conservation law. This
limitation of resources induced competitive behavior between individuals in the
population. The system showed that artificiallife models could lead to discoveries
about biological evolution.
Extensions of this work involved the construction of three nested models, each
corresponding to a different layer of biological organization (Rizki and Conrad
1985; Conrad and Rizki 1989; O'Callaghan and Conrad 1992). EVOL VE III
(O'Callaghan and Conrad 1992) included organizations for genetic structure,
organisms and populations, where each of these components was modelIed
independently. The genetic structure contained simple representations of DNA
and an algorithm for abstracting transcription, translation and protein folding.
Organisms had a number of phenotypic traits, including response to light
intensity, rates of energy usage, protection and aggression mechanisms and a life
cycle history . Each trait was coded by a collection of genes that could be mutated
during reproduction to allow variation in future organisms. The ecosystem was
represented as a set of populations and an abiotic environment, where each
population was composed of individual organisms. The system was designed to
allow various types of populations, organisms and genetic structure to be
independently studied. For example, a simple population consisted of producers
and decomposers. Producer organisms used nutrients from the environment and
returned them to the environment with a degraded energy value. In turn,
decomposers used these degraded nutrients and returned them to the environment
as nutrients available to producers after aperiod of time. This allowed the system
to produce a food cycle where mass was conserved.
Demonstrating the link between theoretical and real ecosystems, EVOL VE III
was used to explore relationships between adaptability of populations and the
variability of the environment. Results from the theoretical model suggested that
populations cultured in a constant environment usually dominated those cultured
in a variable environment when both were placed in a variable environment at an
early stage of development. This pattern was verified by laboratory experiments
and indicates the potential predictive value of the model. This work represented a
significant approach and goal of simulation studies using evolution: the systems
must be able to demonstrate behavior that can be used to interpret real ecosystems
and that the results should be verifiable through laboratory experiments. EVOL VE
IV (Brewster and Conrad 1998) was designed to explore the effects of
environmental uncertainty on niche proliferation and the evolution of interspecific
65
4.4.7
Modelling Hierarchical Ecosystems
Living systems involve many levels of hierarchical interaction, from the genetic
level through the individual to the community and complete ecosystem (Conrad
and Pattee 1970). Conrad and Pattee argued that a theory of evolution that does
not reflect this structure cannot be expected to be useful in terms of predictions
and models of real systems. They presented a model (EVOLVEI) based on a
population of cell-like organisms subject to a strict mass conservation law. This
limitation of resources induced competitive behavior between individuals in the
population. The system showed that artificiallife models could lead to discoveries
about biological evolution.
Extensions of this work involved the construction of three nested models, each
corresponding to a different layer of biological organization (Rizki and Conrad
1985; Conrad and Rizki 1989; O'Callaghan and Conrad 1992). EVOL VE III
(O'Callaghan and Conrad 1992) included organizations for genetic structure,
organisms and populations, where each of these components was modelIed
independently. The genetic structure contained simple representations of DNA
and an algorithm for abstracting transcription, translation and protein folding.
Organisms had a number of phenotypic traits, including response to light
intensity, rates of energy usage, protection and aggression mechanisms and a life
cycle history . Each trait was coded by a collection of genes that could be mutated
during reproduction to allow variation in future organisms. The ecosystem was
represented as a set of populations and an abiotic environment, where each
population was composed of individual organisms. The system was designed to
allow various types of populations, organisms and genetic structure to be
independently studied. For example, a simple population consisted of producers
and decomposers. Producer organisms used nutrients from the environment and
returned them to the environment with a degraded energy value. In turn,
decomposers used these degraded nutrients and returned them to the environment
as nutrients available to producers after aperiod of time. This allowed the system
to produce a food cycle where mass was conserved.
Demonstrating the link between theoretical and real ecosystems, EVOL VE III
was used to explore relationships between adaptability of populations and the
variability of the environment. Results from the theoretical model suggested that
populations cultured in a constant environment usually dominated those cultured
in a variable environment when both were placed in a variable environment at an
early stage of development. This pattern was verified by laboratory experiments
and indicates the potential predictive value of the model. This work represented a
significant approach and goal of simulation studies using evolution: the systems
must be able to demonstrate behavior that can be used to interpret real ecosystems
and that the results should be verifiable through laboratory experiments. EVOL VE
IV (Brewster and Conrad 1998) was designed to explore the effects of
environmental uncertainty on niche proliferation and the evolution of interspecific