RESEARCH ARTICLE Restoration of macroinvertebrates, fish, and habitats in streams following mining subsidence: replicated analysis across 18 mitigation sites Tim Nuttle1,2 , Marisa N Logan1 , David J Parise1 , David A Foltz1 , Joshua M Silvis3 , Mark R Haibach1 Human activities have led to declines in stream functioning and stream restoration seeks to reverse this trend Longwall coal mining, an underground full-extraction method, can cause surface subsidence, affecting streams by creating a series of deep pools that trap sediment, reduce habitat diversity, and impair macroinvertebrate and fish communities Mining effects on streams must be mitigated to maintain the functions, values, and foreseeable uses of streams Gate cutting is a procedure that alleviates pooling by reestablishing the stream grade, accompanied by procedures that stabilize the channel, restore substrates, and enhance in-stream and riparian habitats We evaluated effectiveness of gate cuts at restoring streams affected by subsidence pooling at 18 independent restoration sites over two mines in southwestern Pennsylvania, U.S.A At each site, sampling stations were established to monitor effects of mining subsidence and its restoration on macroinvertebrates, fish communities, and habitats We tested for effects of sequential interventions (subsidence and restoration) on biological and habitat variables in a replicated before–after design, controlling for potentially confounding temporal effects (sample month and antecedent effective precipitation) All biological indices and substrate-related habitat indices declined following subsidence but improved following restoration Macroinvertebrate indicex and taxa richness, substrates, and riparian vegetation continued to improve with time following restoration Whereas other studies have concluded that biological communities may take many years to respond to restoration, these results indicate that where macroinvertebrate and fish communities are altered by subsidence pooling, they can be effectively restored using gate cuts to pre-mining levels within relatively short time periods Key words: before–after design, benthic macroinvertebrates, coal mining, fish, gate cut, habitat, mitigation, sedimentation, stream restoration, subsidence, substrate Implications for Practice • Alleviation of pooling can restore streams to presubsidence condition within short time frames when restoration activities target sources of degradation • Biological recovery of streams is associated with flushing of sediments and improved substrates following restoration • Benthic macroinvertebrate, substrate, and riparian indices continue to improve with time post-restoration • Clearly defined performance criteria allow objective assessment of restoration effectiveness • Results should be generalizable to other interventions that restore stream gradients and riffle-pool sequences, such as removing small dams, weirs, restrictive culverts, and low-water road crossings Introduction Historically, rivers and streams have been substantially altered by human activities and have seen a continual decline in quality during recent decades (Hassett et al 2005) Approximately Restoration Ecology 44% of the streams in the United States are considered impaired or polluted (USEPA 2004) while only 2% are considered to be relatively natural or high quality (Benke 1990) These impairments are caused by many anthropogenic factors Some of the most common are urbanization, dams, farming, and livestock grazing (Hassett et al 2005) Consequently, over the last several decades, there have been a steadily increasing number of stream restoration projects and research studies addressing their effectiveness (e.g Bernhardt et al 2007; Palmer et al 2010) Author contributions: TN, MH, JS conceived and designed the research; ML, DP managed data collection; ML, DP, DF collected field data; TN performed statistical analyses; TN, ML, DP, DF, JS, MH prepared and edited the manuscript Ecological Services Practice, Civil & Environmental Consultants, Inc, 333 Baldwin Road, Pittsburgh, PA 15205, U.S.A Address correspondence to T Nuttle, email tnuttle@cecinc.com PA Operations Engineering, CONSOL Pennsylvania Coal Company LLC., 1000 Consol Energy Drive, Canonsburg, PA 15317, U.S.A © 2017 The Authors Restoration Ecology published by Wiley Periodicals, Inc on behalf of Society for Ecological Restoration This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited doi: 10.1111/rec.12502 Supporting information at: http://onlinelibrary.wiley.com/doi/10.1111/rec.12502/suppinfo Stream restoration following mining subsidence In addition to the aforementioned practices, mining can also impair streams and mitigation for these negative effects is a focus of restoration efforts Effects of surface coal mining, especially mountaintop mining, on streams has attracted most attention in the Appalachian Region of North America (e.g Palmer & Hondula 2014) However, underground mining, used to extract coal in deeper coal seams, is also prevalent in this region Longwall mining (Peng 2006) has been used in many regions globally including Illinois, Ohio, Pennsylvania, Utah, and West Virginia in the United States (Hobba 1981; Sidle et al 2000), as well as Australia (Holla 1997), Great Britain (Bell 1988), Germany (Bell et al 2000), and Ukraine (Palchik 2003) In particular, longwall mining is used extensively throughout the Pittsburgh coal seam in Pennsylvania, Ohio, and West Virginia, where this study takes place Therefore, this study addresses the effects of underground longwall mining on streams and how those effects are mitigated through stream restoration Unlike conventional room and pillar mining, longwall mining removes coal along a strip of the coal seam called a panel, without leaving support for the overburden above the panel (Peng 2006) When the undermined area collapses behind the advancing longwall, subsidence of the surface occurs Adjacent panels (each circa 450 m wide in the Pittsburgh coal seam) are separated by gates (circa 80 m wide), which are left supported to facilitate the transport of coal, personnel, air, and electrical power between the surface and the mine; gates therefore not subside In arid regions and headwater systems, fracturing of bedrock following subsidence can lead to stream dewatering, and subsequent negative effects for life in the stream (Tonsor et al 2014; Commonwealth of Australia 2015) However, low-gradient streams in humid areas like Pennsylvania are more likely to be affected by pooling caused when streams flow across a series of subsided panels and unsubsided gates, where the gates function as dams (Fig 1; Dawkins 2003; Peng 2006) The Pittsburgh coal seam in the study areas is generally 2.1 m thick and 120–180 m below streambed elevation Water depths in affected streams increased typically 60–90 cm mid-panel, where surface subsidence is greatest (in contrast to Bell et al 2000, who reported subsidence up to m over longwall mines in Germany) Subsidence pooling affects streams by trapping fine sediment, reducing habitat variability, and altering the natural flow regime (Poff et al 1997; Tonsor et al 2014) Fine sediment reduces the availability of habitat for benthic macroinvertebrates by filling interstitial space between cobble and large gravel (Bednarek 2001) Subsidence pooling can occur for hundreds of meters in low-gradient streams, thereby replacing diverse riffle-pool sequences with long, slow-flowing pools, negatively affecting the original macroinvertebrate and fish communities, similar to effects of small dams (Bednarek 2001; Tonsor et al 2014) Potential adverse effects include altered sediment transport and increased deposition of fine sediments in subsidence pools and less riffle habitat for benthic macroinvertebrates and riffle-spawning fish (Doyle et al 2005) Although several publications mention possible negative effects of subsidence pooling on in-stream ecological functions (e.g Dawkins 2003; Iannacchione & Tonsor 2011; Tonsor et al 2014; Commonwealth of Australia 2015), and anecdotally many of these effects are obvious, there appear to be few if any quantitative studies on this subject available in the peer-reviewed literature Nevertheless, the need for mitigation to reverse these impacts is legally established in a variety of federal and state laws, regulations, and policies (see below) Making use of existing monitoring data, we evaluated success of stream restoration in meeting biological and habitat objectives at 18 sites (replicates) within two of the most productive underground coal mines in the world, the Enlow Fork and Bailey mines in southwestern Pennsylvania We compared macroinvertebrate, fish, and habitat data from the same sites during pre-mining, post-subsidence, and post-restoration periods We hypothesized that biological parameters would decline following subsidence, associated with declines in habitat and substrate quality, then improve following restoration as substrates improve Restoration Objectives Technical guidance from the Pennsylvania Department of Environmental Protection (PADEP 2005a) requires mitigation when subsidence causes the pool depth to increase by one foot (circa 30 cm) or more, regardless of any observed biological effects Most stream restoration projects in the United States (Palmer et al 2007) and elsewhere (e.g Jähnig et al 2011) lack clear objectives (though this may be improving, see Palmer et al 2014) Furthermore, even when objectives are clear, for many stream restoration projects across the United States, monitoring data are insufficient or not relevant to evaluating those objectives (Bernhardt et al 2007) However, for subsidence-impacted streams in Pennsylvania, objectives and monitoring requirements are clear According to agency guidance (PADEP 2005a) and site-specific restoration permit conditions, streams are generally considered to be mitigated once pooling is alleviated, hydrodynamics are restored, banks are stable with permanent vegetative cover, and an index of biotic integrity (IBI) based on macroinvertebrates (see the Methods section) is comparable to the pre-mining average These performance criteria are indicators that the stream is maintaining its functions, values, and foreseeable uses as required by state and federal laws and regulations (U.S Clean Water Act, 33 Code of Federal Regulations Part 332; Pennsylvania Clean Streams Law, 35 Pennsylvania Code 691.1; Pennsylvania Subsidence Control and Water Supply Replacement: 25 Pennsylvania Code 89.141) Thus, while the criterion triggering the need for mitigation is based on changes in geomorphology, determination of restoration success is based on PADEP’s and U.S Army Corps of Engineers’ (who coevaluate stream restoration projects) consideration of geomorphic, hydrodynamic, and biological criteria We investigated streams where subsidence pooling was observed to have occurred and where mitigation restoration had been performed These restorations employed the most common method to mitigate for subsidence pooling—excavation of surface material within the stream channel over the gate to restore free-flowing conditions in the stream overlying the panel (Fig 1) This procedure, called a gate cut, also necessitates Restoration Ecology Stream restoration following mining subsidence Pre-mining Panel Gate Subsidence pooling Panel Gate PANEL GATE Post-subsidence Elevation (relative, m) Subsidence Top of bank Water level Stream bed 0 GATE CUT Distance along stream (m) 1000 Figure Top left: configuration of panels and gates (underground) to streams and location of subsidence; note the extremely sinuous channel pattern, indicating instability Bottom left: representative stream profile following subsidence with reference to panel, gate, and subsidence locations Right: before and after subsidence at the location of top left areal image; note the channel is deeply incised and banks appear unstable establishing stable channel morphology and restoring instream and riparian habitat in the affected area (Fig 2; Iannacchione & Tonsor 2011; Haibach et al 2012) The stream restoration designs incorporated stable channel dimensions and profiles; in-stream structures for grade control, bank stabilization, and habitat (e.g vanes, toe wood, geo-lifts and brush layering, etc.); and native riparian plantings The exact manipulations employed at each site depend on site-specific condition of the stream, agency objectives, and landowner preferences and constraints However, the dominant activity at each site is channel excavation to restore stream grade While influenced by the natural channel design approach (Rosgen 1996, 2006), restorations on sites investigated here draw on the designers’ experience in this specific system to meet the aforementioned objectives within a permit-mandated 5-year time period Methods to assess compliance, regarding effects on streams, with Consol Pennsylvania Coal Company’s Bailey (Greene County, Pennsylvania) and Enlow Fork (Washington County, Pennsylvania) mine permits and subsequent stream restoration permits As noted above, criteria to evaluate subsidence effects and their mitigation are established in agency guidance (PADEP 2005a) However, for the purpose of this study, we complement the required macroinvertebrate data with additional data on fish communities, substrate size distribution, and USEPA habitat assessments (Barbour et al 1999) The selected metrics include measures of both structural (e.g taxa richness, population size, particle size) and functional attributes (IBI, taxonomic composition, stability and sedimentation indices) of the stream within a particular reach (Palmer et al 2014) We evaluated overall effectiveness of gate cuts by statistically comparing multiple biological and habitat parameters between pre-mining, post-subsidence, and post-restoration periods across 18 restoration sites Overview This study involves a replicated before–after (BA) analysis involving sequential interventions (subsidence and restoration) at 18 independent sites Data used in this study were collected as part of mine-wide biological monitoring programs designed Restoration Ecology Selection of Sites and Timing of Monitoring Biomonitoring stations were established on perennial streams in a location that was determined to be representative over that underground mine panel Where possible, stations were Stream restoration following mining subsidence ENLOW FORK MINE SITES OVERVIEW Site 8: Pre-Mining: 2008 Pre-mining/ Post-subsidence Post-Restoration BAILEY MINE SITES Pre-mining/Post-subsidence/ Post-restoration STATION PAIR (SITE 8) SHOWING SAMPLING DESIGN Site 8: Post-Subsidence: 2009 Pre-mining/ Post-subsidence Post-Restoration Pre-mining/Post-subsidence/ Post-restoration Site 8: Post-Restoration: 2012 Figure Sample station/site locations within Enlow Fork and Bailey mine permit areas in southwestern Pennsylvania, U.S.A Bottom center: example of a station pair showing restoration reach, both monitoring stations, mine panels, and gates (underground) Right: photos of this site showing subsidence-induced pooling and post-restoration habitat improvements placed between the downstream end and center of the mine panel, the area most susceptible to subsidence Pre-mining sampling was initiated generally between and years prior to mining (2004 to 2014) Most stations were sampled at least four times in consecutive years before mining Because monitoring criteria and methods changed with adoption of the statewide guidance (PADEP 2005a), data from prior to 2005 are excluded; therefore, one of the sites we investigated lacks pre-mining data (site 6) Post-mining sampling was generally initiated months to a year after mining passed under the biomonitoring station Because the need to restore a reach was determined by changes in water surface elevation and pool depth and length, and not by biological criteria, restoration of some streams began before any post-mining biological sampling was conducted; however, most stations have data from several post-mining sampling events Because all restoration sites were located in reaches where subsidence actually occurred, we henceforth refer to these post-mining sampling events as post-subsidence Restoration work typically began 1–2 years after mining, though in some cases delays occurred due to negotiations with landowners or other factors Post-restoration monitoring was initiated between months to year after stream restoration was complete, and generally two monitoring events occurred per year Therefore, almost all restoration sites in this study were monitored at least four times within the first years following restoration, though for recent restorations fewer samples were available Figure S1 (Supporting Information) depicts timing of sampling for each site during pre-mining, post-subsidence, and post-restoration periods as well as the dates of undermining and restoration at each site Generally, pre-mining, post-subsidence, and post-restoration samples are all collected in the same reach, over the same underground mine panel in the same location However, sometimes the exact location of sampling had to be changed, usually due to changes in panel alignment that occurred after pre-mining stations were established and the need to monitor within the restoration reach itself To assure that pre-mining/post-subsidence and post-restoration samples were representative of the same stream reach, we include only restoration sites where the pre-mining/post-subsidence and post-restoration stations were over the same mine panel and thus in close proximity (