Wilson, D. S. DEB-9707650 Biological communities have often been described as functionally organized units, similar to single organisms in the harmony and coordination of their parts. The evolution of adaptations at any level of the biological hierarchy, however, requires a corresponding process of natural selection at that level. This grant proposal explores the process of community-level selection with laboratory experiments and computer simulations. In the laboratory, a population of isolated soil communities will be established and scored with respect to a phenotypic character (plant growth). The highest and lowest scoring communities will be used as the "parents" to colonize a new set of "offspring" communities. This experimental design is identical to standard artificial selection experiments except that the selected units are communities rather than individuals. If the experiment is successful, it will produce soil communities that are functionally designed to increase (or decrease) plant growth. The computer simulations will allow the same process to be explored in which the interactions among the species can be specified in detail. Artificial selection at the individual level has been employed for centuries to develop strains of plants and animals that are useful to man. The same procedure can potentially be used to evolve multi-species communities that interact as a coordinated team to solve important practical problems, such as promoting plant growth, excluding a target species from a community, or degrading a toxic compound. Artificially selected communities may provide better solutions than artificially selected individuals for the same reason that coordinated teams of people can outperform single individuals. The advantages of "teamwork" are obvious; the problem is how to develop the teams. The proposed research may provide a simple method that so far has been overlooked.
