Landscape indicators and thresholds of stream ecological impairment in an intensively mined Appalachian watershed

Abstract

Coal-mine development is occurring at a rapid rate in the central Appalachians, but few tools exist to assess the consequences of cumulative effects of mining to downstream aquatic resources. We constructed and applied an index of mining intensity (MI) to the Lower Cheat River basin, northern West Virginia. Our objectives were to: 1) determine if the MI could be used to predict stream-water quality and biological conditions, 2) quantify the extent to which geology and the geographic position of mines modulate the effects of mining on in-stream conditions, and 3) identify thresholds of MI that produce quantifiable changes to benthic macroinvertebrate communities. We quantified water chemistry, habitat quality, and benthic macroinvertebrate communities from May 2002 to May 2003 in 39 stream segments randomly distributed across a range of MI, coal geology, elevation, and watershed area. We sampled benthic macroinvertebrates at an additional 41 validation sites in May 2002. The MI was positively correlated with dissolved metals (r = 0.65–0.85) and negatively correlated with ecological condition metrics (r = 0.49–0.78), including total richness, Ephemeroptera, Plecoptera, Trichoptera richness, and the West Virginia Stream Condition Index. Coal geology and distance from mining had a significant interactive effect on benthic macroinvertebrate responses. Streams draining watersheds with Freeport coal geology had significantly poorer water quality and ecological condition than streams draining watersheds with similar MI but with Kittanning coal geology. Mining effects on stream conditions diminished as distance from the nearest mining activities upstream increased to a distance of,10 km. Changepoint analysis provided evidence of threshold effects of mining on benthic macroinvertebrate communities in Freeport coal watersheds but not in Kittanning coal watersheds. Abrupt reductions in ecological condition occurred at MI values as low as 1 to 5% of maximum intensity, and ecological impairment to streams became almost certain at MI .18 to 20%. Our results provide evidence of an interactive effect of landuse intensity, underlying geology, and the spatial arrangement of disturbance on the degree of impairment to receiving water bodies. The thresholds we identified could be used by water-resource managers to protect and restore stream conditions in actively mined watersheds of the central Appalachian region.