Land use impacts on stream community composition and concordance along a natural stress gradient

Most ecosystems are subjected to multiple stressors derived from natural and anthropogenic sources and community responses to human disturbance in naturally stressful habitats may differ from those in more benign habitats. We examined the influence of a natural (geology-driven acidity) vs. human-induced stress (land drainage) and their interaction on the composition and concordance of stream diatom, bryophyte and invertebrate communities. To account for differing drainage impacts in circumneutral (sedimentation) and naturally acid (reduced pH and increased metal concentrations) streams we investigated concordance in three groups of streams: reference (circumneutral and naturally acidic reference), circumneutral (reference and drained) and naturally acidic (reference and drained) streams. We expected diatoms to respond more strongly to anthropogenic acidification and more weakly to sedimentation compared to bryophytes and invertebrates. We expected overall strong concordance among the three taxonomic groups, but especially so in reference streams. All three organism groups had distinct species composition in naturally acidic vs. circumneutral streams. Concordance between communities was overall strong, especially so in the reference streams. All groups responded to drainage disturbance in both types of streams. Invertebrates were slightly less responsive to increased acidification in the naturally acidic streams but were more affected by sedimentation in the circumneutral streams than were the other two groups. The natural stressor affected communities more than the anthropogenic stressors. Naturally stressed communities were affected by an anthropogenic stressor as much as those in more benign habitats, although the additional stressor was similar to the initial stress (further reduction of stream pH). Naturally acid streams may need special concern in bioassessment because models based on circumneutral reference sites will likely produce biased predictions for these streams.