Summary1. Landscape classifications group tracts of land based on similar physico‐chemical attributes that may affect the biological characteristics of streams at local scales. We tested the ability of five landscape models to account for variation in algal and macroinvertebrate biomass, brook trout (Salvelinus fontinalis) growth and macroinvertebrate community composition from 132 riffles in 15 catchments in the Big Horn Mountains, Wyoming.2. A model created by the U.S. Forest Service (FS) combined catchment and ecoregion approaches to classify the landscape. Our model used digital elevations to create a landscape classification for streams (DEM). The last three models were based on: (1) standard ecoregions (Ecoregion), (2) the type of underlying bedrock (Geology) and (3) the geographical distance between sites (Site Proximity model).3. Overall, the Ecoregion and Geology models performed better than the two catchment models (FS and DEM) in predicting local biological characteristics. The Geology model was best at predicting differences in algal and macroinvertebrate biomass, the Site Proximity and Ecoregion models were best at predicting patterns of similarity in macroinvertebrate community composition, and the Site Proximity, Ecoregion, Geology, and FS models, in order from best to worst, accounted for significant variation in brook trout growth. The Site Proximity model performed well because of the effects of spatial autocorrelation. The DEM was consistently one of the worst models at predicting local biological characteristics because it failed to include important attributes (e.g. dominant geology). Calcareous geology was positively associated with greater macroinvertebrate biomass and faster brook trout growth, but it was inversely related to algal biomass.4. None of the models accounted for a large amount of variation in local biological characteristics. Single‐scale, landscape classifications may never accurately predict variation in local biological characteristics because: (1) landscapes show a high degree of spatial heterogeneity, (2) local effects are stronger than landscape attributes and (3) there are too many intervening levels between landscape and local scales in the nested hierarchy of streams. However, landscape classifications did account for significant variation in biological characteristics. Thus, they would be a valuable management tool as part of a multi‐scale, hierarchical technique for classifying stream ecosystems.
Read full abstract