Artificial Ecosystems
The branch of artificial life research that is first explored by Thomas Ray's Tierra investigates a wide range of biological problems using artificial evolving systems with endogenous fitness.
1. Population Dynamics
environmental management for artificial ecosystems
In every evolutionary simulation of endogenous fitness, one has to take the population dynamics of the artificial ecosystem in account. If a populations grows to big, the simulation gets too slow and if a population is too small, the experiment can fail because of extinction. Predator-prey or foodchain ecosystems, in particular,are notably difficult to maintain stable.
Fortunately, there is a large body of explanatory and predictory models available from the theoretical biology literature. Models such as the Lotka-Volterra model can enable us to predict important experimental parameters like the carrying capacity and (in)stability of a particular simulation. This knowledge can be used to enable studies in for example sustained co-evolution in artificial ecosystems.
2. Controlling Population Dynamics through Decoys
environmental management for artificial ecosystems
Predator-prey or host-parasite dynamics can be altered by the presence of other species through several mechanisms. One such mechanism is the "decoy effect," which itself can take a variety of forms. In its simplest form, the third species, which is inedible to the predator, nonetheless interferes with predation because the predator spends time investigating these decoys. The effect of this is to reduce instability in the ecosystem by damping the Lotka-Volterra oscillations. This effect may be considerable in dense ecosystems such as the gut microflora.
Other forms include the prey species creating decoys by autotomy of appendages, such as tails, in the case of certain lizards. Finally, decoys are created by various parasites to avoid elimination by the immune system. In this review the theoretical models and supporting observations for these forms of the decoy effect are discussed, and some refinements on the theory are proposed. In particular, a model for autotomy is extended, and the phenomenon of signaling unpalatability or toxicity is discussed within the context of the decoy effect. It is shown that such signaling will increase predation on edible competitors.
This 'decoy effect' may be very useful in Virtual life experiments to reduce instability in artificial predator-prey ecosystems, in order to enable long-term co-evolutionary experiments.
- Michael H. F. Wilkinson, Decoys in Predation and Parasitism, Comments on Theoretical Biology, 2003.
3. Red queen effect
"in this place it takes all the running you can do,
to keep in the same place"
Since every improvement in one species will lead to a selective advantage for that species, variation will normally continuously lead to increases in fitness in one species or another. However, since in general different species are coevolving, improvement in one species implies that it will get a competitive advantage on the other species, and thus be able to capture a larger share of the resources available to all. This means that fitness increase in one evolutionary system will tend to lead to fitness decrease in another system. The only way that a species involved in a competition can maintain its fitness relative to the others is by in turn improving its design. (Principia Cybernetica web)
When, in a stable evolving population, a mutation is introduced that enables owners to better collect food, the mutant will probably have more offspring than the others in the population (i.e. has higher fitness) and this mutation will spread through the population. During this process the average fitness of mutants will be higher than the fitness of non-mutant creatures in the population. However, after all creatures in the population possess the beneficial mutation, the mutation no longer results in higher fitness.
How, if not by fitness, can we measure evolutionary adaptation and complexification? Menczer and Belew devised the Latent Energy Environments in which they are able to measure the behavioural complexity of the evolving agents through the carrying capacity of the population. However, in our example, it is unclear whether the population of mutant creatures that are good at collecting food will results in higher or lower carrying capacity if food availability remains unchanged.
4. Controlling Environmental Complexity
controlling environmental complexity: environmental complexity as a function of food distribution (availability and spatial distribution)
- co-evolutionary arms races (Krebs, Dawkins) in co-evolving predator-prey ecosystem with natural selection
|
