A THEORY OF FAUNAL BUILDUP FOR COMPETITION COMMUNITIES.

Invasion-structured communities have more species than do coevolution-structured communities assembled using the same resource distribution. Species in invasion-structured communities are tightly packed, occupying the upper portion of the resource axis; species in coevolution-structured communities are more widely spaced, and most are located in the lower portion of the resource axis. As a consequence, coevolution-structured communities tend to be more stable than comparable invasion-structured communities, but more open to invasion. Both invasion-structured and coevolution-structured communities have niche separations that are significantly different than would be expected if species were assorted at random. Two-species communities formed by the invasion-only algorithm under asymmetric competition had the majority of their niche separations in the range 0-0.5. All other communities had niche separations that were greater than expected. The most common separations were in the range 1.0-3.5. Thus, while not a common feature of many communities, nicheseparation patterns similar to those described by Hutchinson (1959) appear as an "ensemble" property of many communities. The faunal-buildup graphs formed by the coevolutionary algorithm differ from those formed by the invasion-only algorithm, showing community cycling whenever asymmetric competition is present. Through this cycling behavior the coevolutionary faunal-buildup algorithm provides both a theoretical basis for Wilson's (1959) taxon cycle and a hypothesis explaining the distribution of Anolis lizards in the Lesser Antilles.