Avian community assembly rules: The foliage-gleaning guild.

The theory of animal community organization has been dominated by general models based on the Lotka-Volterra equations. The predictions of these models are difficult to test in particular situations. Moreover, a great deal of ecological information is incommensurate with the data requirements of these models. A different approach to community organization addresses the "ecosystem assembly problem". This problem is defined to be that of constructing an algorithm which assembles a subset of a species pool in a specified environment.A model of ecosystem assembly, based on generative grammars as used in theoretical linguistics, is described. It was constructed from and validated with data collected by D.H. Morse on a guild of foliage-gleaning birds inhabiting spruce forests on islands off the coast of Maine. The data were divided into two groups. One group, from the years 1967-1970, was used for model construction; the second group, from 1971-1975, was used to validate the model.The model has two major components. One component inserts species onto islands according to the microhabitat used by each species and the resources available on each island. A second component deletes those inserted species from islands on which they were not observed to occur during 1967-1970. This component is composed of deletion rules that remove species depending on (a) their sizes and resource requirements, (b) the sizes and resource requirements of other species present in the ecosystem, and (c) the structure of the vegetation on the islands. Model validation was performed by comparing the predicted distributions of species against observed distributions not used in model construction. Model accuracy for the later data (1971-1975) was slightly higher than for the earlier data (1967-1970), approximately 88% and 84%, respectively.The behavior of the model was investigated with several simulations. These included the effects of the removal of certain deletion rules and the effects of the application of the rules without regard to their order. Other simulations demonstrated the application of the model to the prediction of the effects of habitat manipulation and the removal of particular species from the species pool.