Landscape filters and species traits: Towards mechanistic understanding and prediction in stream ecology

A heuristic framework for understanding and predicting the distribution and categorical abundance of species in stream communities is presented. The framework requires that species be described in terms of their functional relationships to habitat selective forces or their surrogates, which constitute 'filters' occurring at hierarchical landscape scales (ranging from microhabitats to watersheds or basins). Large-scale filters are viewed as causative or mechanistic agents that constrain expression of local selective forces or biotic potential at lower scales. To join a local community, species in a regional pool must possess appropriate functional attributes (species traits) to 'pass' through the nested filters. Biotic interactions are also a potential filter on local community composition, and they are invoked at the lower hierarchical levels, after species have passed through the physico-chemical habitat filters. Potential landscape filters and their associated selective properties are identified, as are prospective species traits (for invertebrates and fish) that correspond with filters. A categorical niche model is used to illustrate how relative abundances of species in local communities might be predicted from habitat data collected at different scales. The framework emphasizes a biologically based approach to understanding and predicting species distribution and abundance and local community composition by explicitly considering environmental constraints imposed at different scales. As such, it can complement non-mechanistic, correlative approaches to community prediction that often lack generality. Operationalizing the framework will require additional research to specify more clearly 1) the degree to which habitat features at different scales are linked functionally or statistically, 2) what species traits are possessed by strongly interactive species (e.g., keystones) and which habitat filters most strongly constrain the distribution of these species, and 3) the functional significance of a range of species traits and the extent to which these traits are correlated and hence respond in concert to the presence, or modification, of a particular filter. Multi-scale, mechanistic understanding of species-environment relations will likely contribute to better predictions about large scale problems, such as the establishment and spread of exotic species or alterations in community composition with changing land use or climate.