It has been proposed that the structure of ecosystems is controlled by a few physical and biological processes operating at distinct spatial and temporal scales. These processes are thought to create landscapes that are hierarchically structured, with resources that are available to species of different sizes interacting with the landscape on different scales. This discontinuous resource distribution should result in the species assemblage of a region having characteristics, such as body mass, that are also discontinuously distributed. Using Breeding Bird Survey data, I examined 6 avian species assemblages occupying mutually exclusive ecoregions in the western United States of America. I tested several hypotheses: (1) avian body masses within a landscape are discontinuously distributed, (2) non-indigenous species will have body masses that place them relatively far from discontinuities in the body mass spectrum, and (3) declining native species will have body masses that place them near discontinuities in the body mass spectrum. Analysis of the species assemblages shows that the body mass distributions are consistent across landscapes, with declining species more likely, and introduced species less likely, to be near discontinuities. The possibility that body masses may provide predictive ability about the potential success or failure of a population suggests a novel approach to identifying at-risk native species and for assessing the establishment potential of non-indigenous species.