Like economies, unregulated ecosystems have a tendency to compete towards a single species monopolizing the system. Paul has mentioned this in his blog posts as “going monoculture”. Here are a couple of "rules of thumb" I've been pondering of late.
RULE#1
If multiple species are competing in the same ecological niche, one will always outcompete the others, and take over… unless there is a check to keep the diversity in balance. One way to diversify the overall ecology is to differentiate areas of the habitat to provide local advantage to different species.
RULE#1: Ecological stability is proportional to habitat diversity.*
Habitat diversity can also be a means to provide the sanctuary discussed in Rule#2. This is why it's good to have both aquatic and a terrestrial biomes in the habitat. There are other reasons for having both as well, but this serves to prevent a single species from taking over both biomes. Having a soil based terrestrial biome provides surface area and sanctuary for bacteria and other subterranean life. Having an aquatic biome provides a reservoir for the water in the system, in addition to providing thermal mass, and supporting a volumetrically efficient autotroph habitat. In other words it allows a lot of algae to remove minerals from the water and simultaneously produce oxygen.
Another approach might be to have different climates, where different species might have the local edge in resource competition. In addition, a thermal gradient provides an engine to transport water (hydrologic cycle). Habitat diversity is about gradients; gradients in accessibility/media, light, temperature, nutrients, and humidity.
*This may not be precise in a mathematical sense; indeed it's likely a sigmoidal response (as with most things in nature) rather than linear.
RULE#2
Typically going mono is exemplified by one species of plant overgrowing the habitat. If a species of herbivorous animals is introduced to check plant growth, they will soon decimate the plants, and eat themselves into a starvation induced extinction. The best way to prevent this is to introduce an apex predator, preferably an obligate carnivore. However, unless there are sanctuaries, the herbivores will end up decimated, and the predators will be in the starvation induced extinction position, leading ultimately back to scenario 1.
So this implies the second rule:
Rule #2: The food chain must have an apex predator whose prey has some sanctuary.
Now I had mentioned that this apex predator be an obligate carnivore. There are a couple of possible exceptions.
The first is to use an herbivore or omnivore and instead provide sanctuary for the plants. An example of this might be to have grass growing up through a mesh that is impervious to the gnawing of grazers above. Another method might be to rotate grazing, as Paul discusses in one of his posts. That is less feasible when applied to populations, versus an individual animal.
So that's the first exception. The other one that has me musing, is to use humans. Of course we aren’t obligate carnivores, even on a ketogenic diet (as I currently am). So what checks are to keep us from consuming ourselves into a starvation induced extinction, in a closed habitat (such as, ahem, on Earth)?
• Sanctuary? It is difficult to imagine a sanctuary for our food, devisable by humans that can’t be overcome by humans ingenuity.
• Habitat rotation? Again, I suspect most schemes would be surmountable by human ingenuity
• Rationality? Tragedy of the commons, hello.
• Societal Rules? Maybe.
• Robot Overlords? Well…
A third exception, devised by nature, is another form of predator that consumes its prey incrementally rather than by discrete individuals. Yep, I'm talking about parasites. The beauty of this solution is that such a species links both ends of the food chain making it a loop. However, I've never been a fan of having mosquitos (or ticks, tapeworms, vampire bats) included in my ideal habitat design. Given the choice between mosquitos and Robot Overlords, however… tough call. I'd really like to think we could make some Societal Rules workable.