• ecoregion;
  • gradient;
  • nutrients;
  • periphyton;
  • regression tree model

We attempted to identify spatial patterns and determinants for benthic algal assemblages in Mid-Atlantic streams. Periphyton, water chemistry, stream physical habitat, riparian conditions, and land cover/use in watersheds were characterized at 89 randomly selected stream sites in the Mid-Atlantic region. Cluster analysis (TWINSPAN) partitioned all sites into six groups on the basis of diatom species composition. Stepwise discriminant function analysis indicated that these diatom groups can be best separated by watershed land cover/use (percentage forest cover), water temperature, and riparian conditions (riparian agricultural activities). However, the diatom-based stream classification did not correspond to Omernik’s ecoregional classification. Algal biomass measured as chl a can be related to nutrients in habitats where other factors do not constrain accumulation. A regression tree model indicated that chl a concentrations in the Mid-Atlantic streams can be best predicted by conductivity, stream slope, total phosphorus, total nitrogen, and riparian canopy coverage. Our data suggest that broad spatial patterns of benthic diatom assemblages can be predicted both by coarse-scale factors, such as land cover/use in watersheds, and by site-specific factors, such as riparian conditions. However, algal biomass measured as chl a was less predictable using a simple regression approach. The regression tree model was effective for showing that ecological determinants of chl a were hierarchical in the Mid-Atlantic streams.