Interactive effects of water quality, physical habitat, and watershed anthropogenic activities on stream ecosystem health

Ecological degradation of streams remains a major environmental concern worldwide. While stream restoration has received considerable attention, mitigation efforts focused on the improvement of physical habitat have not proven completely effective. Several small-scale studies have emphasized that effective restoration strategies require a more holistic understanding of the variables at play, although the generalization of the findings based on the small-scale studies remains unclear. Using a comprehensive statewide stream monitoring database from West Virginia (WV), a detailed landscape dataset, and a machine learning algorithm, this study explores the interactive impacts of water quality and physical habitat on stream ecosystem health as indicated by benthic macroinvertebrate scores. Given the long history of energy extraction in this region (i.e., coal mining and oil/gas production), investigation of energy extraction influences is highlighted. Our results demonstrate that a combination of good habitat and low specific conductance is generally associated with favorable benthic macroinvertebrate scores, whereas poor habitat combined with water quality conditions typically indicative of high ionic strength are associated with impaired stream status. In addition, streams impacted by both energy extraction and residential development had a higher percentage of impairment compared to those impacted predominantly by energy extraction or residential development alone. While water quality played a more important role in the ecosystem health of streams impacted primarily by energy extraction activities, habitat seems to be more influential in streams impacted by residential development. Together, these findings emphasize that stream restoration strategies should consider interactive effects of multiple environmental stressors tailored to specific sites or site types - as opposed to considering a single stressor or multiple stressors separately. (C) 2017 Elsevier Ltd. All rights reserved.