In previous subsections, the ecosystems modeled are food chains, not food webs. To branch into food webs, we need more than one species in a trophic level. Constructing such systems might be expected to be tricky. Small competitive advantages can keep competitors from coexisting.
To begin, I tried multiple plant species. The difference between the species was straightforward and profound: though they started from same-size seed, and had the same environment and behaviors, they had different breeding radii. Not slightly different, but very different. In the variation on this experiment shown in Figures 5 and 6, the small plants reproduce at only a quarter the biomass of the large ones. The plants which reproduce at one quarter the biomass, tend to reach reproductive age in one half the time, according to simulation.
Despite the tremendous reproductive advantage of small, these three species coexist indefinitely. Gradually, small is displaced by large, but never forced to extinction in runs of 4000 rounds. In fact, all three species were seen to coexist for up to 80,000 rounds in sample runs.
The small and large plants coexist with grasshoppers and spiders present,
and they coexist without them. Plant biomass gradually shifts
to most in largest, least in smallest. Brief experiments were also done
with plant species of even wider discrepancy in breeding radii, for a biomass
difference of 14
. Grasshoppers and spiders
did not survive the scenario, but the plants coexisted indefinitely.
This is, of course, a simulator artifact, a result of the built-in rules. The interesting question is, Are the rules wrong? The result, that bigger plants, slower to reproduce, do have a competitive advantage---they win the biomass race---but don't thereby kill off the competing small plants, is a consequence of Gecko's geometry. Crowding simply curbs small plants more than big plants, as the same overlap is proportionally more costly to small plants than big ones. This result is reminiscent of a field of wildflowers in spring. A month later, the wildflowers are not extinct, but are invisible in a sea of tall grass. They had their chance with growing conditions when the grass was small, used that chance to reproduce, and their seeds are waiting for next spring.
The obvious next step is to try grasshopper species of different breeding radii. And the result is as expected: the one that reproduces fastest, wins. The others go extinct. I tried a few more subtle species differences between grasshoppers, but if they were subtle enough to coexist, they could also be interpreted as normal variation within a species. For instance, three grasshopper species defined by slight differences in veer can coexist for a short while. With a large enough site supporting a larger population, they might coexist a long while. Which veer won was pure chance. Larger differences in veer had a deterministic winner---the other variants died out. These experiments were not pursued enough to pin down which veer won and why. The Evolution Strategy engine was also applied across mulitple parameters in attempt to find coexisting grasshopper definitions, which although inconclusive, also failed. No experiments to date have been done on multiple carnivore species.
Note that no ``real'' differences in behavior---differences in the basic Gecko parameters and lifecycles---are necessary to model multiple species or niches. One could assign plants to species A, B, and C, and grasshoppers to A-eater, B-eater, and C-eater, for an abstract study. Or one could model substantive differences in behavior, resource transformation, and prey selection.