Abstract
The predictions of community matrix theory do not agree with the observed patterns of coexistence of grassland and woodland sparrows. Predictions based on the random co—occurrence of species match the observed patterns about as well as do the predictions of community theory. Predictions of both the community theory model and the random co—occurrence model are improved substantially by the addition of thresholds that account for the minimum food requirements of sparrows of different sizes. The observed patterns of species coexistence appear consistent with a model based on resource abundance and aggressive interference; however, they are also consistent with coexistence based on interspecific partitioning of seeds by size.