Stream macroinvertebrate communities across a gradient of natural gas development in the Fayetteville Shale

Science of the Total Environment 530–531 (2015) 323–332

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Stream macroinvertebrate communities across a gradient of natural gas development in the Fayetteville Shale Erica Johnson a, Bradley J. Austin b, Ethan Inlander c, Cory Gallipeau c, Michelle A. Evans-White b, Sally Entrekin a,⁎ a b c

Department of Biology, University of Central Arkansas, 201 Donaghey Ave., Conway, AR 72035, United States 601 Science and Engineering, Department of Biological Sciences, University of Arkansas, Fayetteville AR, 72701, United States The Nature Conservancy, Ozark Highlands Office, 38 West Trenton Blvd., Suite 201 Fayetteville, AR 72701, United States

H I G H L I G H T S • • • •

Gas well pad density and proximity related positively to turbidity and chlorophyll a. Gas metrics also related positively to macroinvertebrate densities. Filtering and gathering invertebrate densities related positively to gas activity. Natural gas activities may be altering macroinvertebrate community structure.

a r t i c l e

i n f o

Article history: Received 10 March 2015 Received in revised form 7 May 2015 Accepted 7 May 2015 Available online 3 June 2015 Editor: D. Barcelo Keywords: Anthropogenic stressors Small streams Macroinvertebrates Unconventional oil and gas

a b s t r a c t Oil and gas extraction in shale plays expanded rapidly in the U.S. and is projected to expand globally in the coming decades. Arkansas has doubled the number of gas wells in the state since 2005 mostly by extracting gas from the Fayetteville Shale with activity concentrated in mixed pasture-deciduous forests. Concentrated well pads in close proximity to streams could have adverse effects on stream water quality and biota if sedimentation associated with developing infrastructure or contamination from fracturing fluid and waste occurs. Cumulative effects of gas activity and local habitat conditions on macroinvertebrate communities were investigated across a gradient of gas well activity (0.2–3.6 wells per km2) in ten stream catchments in spring 2010 and 2011. In 2010, macroinvertebrate density was positively related to well pad inverse flowpath distance from streams (r = 0.84, p b 0.001). Relatively tolerant mayflies Baetis and Caenis (r = 0.64, p = 0.04), filtering hydropsychid caddisflies (r = 0.73, p = 0.01), and chironomid midge densities (r = 0.79, p = 0.008) also increased in streams where more well pads were closer to stream channels. Macroinvertebrate trophic structure reflected environmental conditions with greater sediment and primary production in streams with more gas activity close to streams. However, stream water turbidity (r = 0.69, p = 0.02) and chlorophyll a (r = 0.89, p b 0.001) were the only in-stream variables correlated with gas well activities. In 2011, a year with record spring flooding, a different pattern emerged where mayfly density (p = 0.74, p = 0.01) and mayfly, stonefly, and caddisfly richness (r = 0.78, p = 0.008) increased in streams with greater well density and less silt cover. Hydrology and well pad placement in a catchment may interact to result in different relationships between biota and catchment activity between the two sample years. Our data show evidence of different macroinvertebrate communities expressed in catchments with different levels of gas activity that reinforce the need for more quantitative analyses of cumulative freshwater-effects from oil and gas development. © 2015 Elsevier B.V. All rights reserved.

1. Introduction

⁎ Corresponding author. E-mail addresses: [email protected] (E. Johnson), [email protected] (B.J. Austin), [email protected] (E. Inlander), [email protected] (C. Gallipeau), [email protected] (M.A. Evans-White), [email protected] (S. Entrekin).

http://dx.doi.org/10.1016/j.scitotenv.2015.05.027 0048-9697/© 2015 Elsevier B.V. All rights reserved.

Natural gas and oil extraction using horizontal drilling coupled with hydraulic fracturing, currently unconventional methods (UOG), has expanded rapidly across the U.S. and is quickly becoming a more common land use in regions of the U.S. that have historical had little resource extraction (Lave and Lutz, 2014; US DOE/EIA, 2013). Natural gas and oil well extraction in shale basins has been shown to be close

324

E. Johnson et al. / Science of the Total Environment 530–531 (2015) 323–332

to streams and pose multiple environmental threats to streams that include water flow alteration, sedimentation, and surface and groundwater contamination (Williams et al., 2007; Kargbo et al., 2010; Entrekin et al., 2011). The installation of infrastructure needed for natural gas extraction disturbs the landscape and can reduce core forest and riparian areas that decrease habitat and vegetation to buffer nearby streams from runoff (Drohan et al., 2012; Moran et al., 2015). Stream hydrology is often more flashy in catchments with altered riparian areas that can simplify stream habitat by sediment transport that scours the stream bed (Walsh et al., 2005; Roy et al., 2006) and increases sediment deposition, both of which reduce habitat for aquatic biota (Frissell et al., 1986; Wood and Armitage, 1997; Poff et al., 2006b). Associated infrastructure such as pipelines and roads for transporting equipment, fracturing fluids, or moving gas or oil also fragments landscape, increases impervious surfaces and increases the probability for cumulative effects from sedimentation, nutrient leaching, or contamination to receiving streams (Souther et al., 2014). Sediment and nutrients are primary pollutants in streams in the U.S. most often associated with row crop agriculture, cattle grazing, and urban land development (Ryan, 1991; United States Environmental Protection Agency, USEPA, 2009). Increased sediment and associated nutrients can also occur from development associated with natural gas activity; however, there are few published studies quantifying cumulative impacts (but see Olmstead et al., 2013; Brittingham et al., 2014). Stream turbidity often correlates positively with sedimentation and has been shown to positively correlate with gas well density in stream catchments across the Fayetteville Shale in Arkansas (Burton et al., 2014; Entrekin et al., 2011). Stream water turbidity could also increase as a result of cumulative activities associated with natural gas well development similar to increases in streams draining agricultural and urban landscapes with elevated sediment concentrations (Walsh et al., 2005; Drohan et al., 2012). Direct measurements of sediment from recently placed gas pads quantified soil erosion on-pare to a construction site (Williams et al., 2007). Sedimentation to streams also occurs with new road construction and pipeline construction placed near or across streams when proper erosion control structures are not used (Walsh et al., 2005). More studies examining multiple sources of sedimentation at the landscape level are required for predicting water quality alterations associated with infrastructure development associated with oil and gas extraction. Total dissolved solids, metals, and organic compounds may also increase in streams in close proximity to UOG activities from accidental spills and leaks associated with the process of hydraulic fracturing and subsequent production (Preston et al., 2014; Rozell and Reaven, 2012). The probability of accidental spills is uncertain but will increase with greater activity (Rahm and Riha, 2014). Contamination events from UOG will be mostly acute where it is unlikely an event would be detected in most small streams unless monitoring is occurring or biological communities are examined before and after an event. Streams that have experienced chronic stress from construction activities or acute contamination events will have altered biological communities that reflect these stressors over several generations (Weigel, 2003; Burton et al., 2014). Macroinvertebrate community composition represented as functional composition and tolerance to stressors provides metrics to assess changes in trophic status and organization from catchment-scale stressors (Merritt et al., 2008; Barbour et al., 1999). For example, a majority of collector-gathering macroinvertebrates would indicate an abundance of fine benthic organic matter and likely high frequency and intensity of habitat disturbance (Boulton et al., 1992; Resh et al., 1988; Whiles and Wallace, 1992). More collector-filterers would indicate greater delivery of suspended organic sediment and more scrapers are indicative of greater benthic primary production. Density-weighted tolerance of a community provides a more comprehensive indication of overall degradation to water quality that could be a result of contamination from myriad catchment-level alterations. Differences in macroinvertebrate communities and individual taxa in similar streams and

rivers allow scientists and managers to predict effects of catchmentlevel alterations integrated over time (Merritt et al., 2008; Poff et al., 2006a). Our primary objective was to identify differences in aquatic macroinvertebrate communities in receiving streams draining catchments with recent and on-going UOG extraction activities embedded in a landscape of pasture and forest. We predicted greater tolerant taxa and collectors and fewer sensitive taxa, such as shredders represented by Ephemeroptera, Plecoptera, and Trichoptera (EPT) in catchments with more UOG. 2. Methods 2.1. Study area We sampled 10, 4th–6th order streams in north-central Arkansas in the Arkansas River Valley with catchments that ranged from 14 to 84 km2 (Fig. 1). Sites were selected to achieve a gradient of gas well densities (Table 1). Catchment area was calculated for each site using ArcHydro Tools 9 version 1.3 (an ArcGIS extension). For each catchment, gas well data points were accessed from the Arkansas Oil and Gas website (ftp://www.aogc.state.ar.us/GIS_Files/) and well density was calculated as the sum of spud, active, and plugged wells divided by the catchment area (Table 1). Gas well densities across all catchments ranged from 0.2 to 2.2 wells per km2 in May 2010 and 0.6 to 3.6 wells per km2 in May 2011. Land use was estimated for each catchment and dominated by forest and pasture (Table 1). The total length of paved and unpaved roads within each catchment was divided by catchment area to calculate density. Natural gas pads were digitized from USACE (U.S. Army Corps of Engineers) aerial photography from June 2009 at 0.3 m resolution as well as 2009 and 2010 USDA (U.S. Department of Agriculture) NAIP (National Agriculture Imagery Program) aerial photography at 1 m resolution. Where there were active gas wells from the AOGC data for these years, but pads were not visible on the 2009 or 2010 aerial photography (i.e. pads from 2011, 2012, 2013), a standard pad size of ~ 1 ha was used to generate pad polygon. The distance from gas pads to a stream was measured as the path water would flow using ArcHydro Tools 9 version 1.3 (ArcGIS, ESRI, Redlands, CA). Flowpath distances were then inverted and summed to calculate the inverse flowpath length (IFPL) of gas pads as an index of the total proximity of gas pads to streams in a catchment. The gas variables (gas well and pad density and IFPL) were updated annually. Land cover and road variables could not be updated annually. 2.2. Macroinvertebrate benthic sampling Macroinvertebrates were sampled two years in spring from May 7–9 in 2010 and again from May 16–17 in 2011. At each of the 10 ten streams, we delineated 200 meter upstream reaches and used a random number generator to identify sampling locations within each reach. Macroinvertebrates were sampled in five pools with a 650 cm2 d-frame kick net (250 μm mesh), standardized to 1 m with three sweeps along the stream bottom (Snyder et al., 2002). Macroinvertebrates were also sampled in five riffles using a 32-cm diameter Hess sampler (250 μm mesh, Delong and Brusven, 1998). All macroinvertebrates were preserved in 95% ethanol. In the laboratory, macroinvertebrates in each sample were separated into 1-mm and 250-μm size classes using stacked sieves. Macroinvertebrates N1 mm were sorted by eye, while macroinvertebrates b 1 mm but N250 μm were subsampled using a sample splitter with an adequate subsample having at least 100 individuals (Waters, 1969) and sorted with using a dissecting microscope. Chironomids were classified as Tanypodinae predators or non-Tanypodinae and all other invertebrates were identified to genus using Merritt et al. (2008), Stewart et al. (1993), Thorp and Covich (2001), and Wiggins (1996). All individuals

E. Johnson et al. / Science of the Total Environment 530–531 (2015) 323–332

325

Fig. 1. Ten study streams across north-central Arkansas in the Fayetteville Shale play. Each study site was a 200 meter reach upstream from point. Insets are U.S. lower 48 states with Arkansas highlighted and Arkansas major rivers and gas wells in the Fayetteville Shale. Study streams and sample locations are the main feature with stream abbreviations and major stream drainages labeled in full.

were counted and measured to the nearest 1-mm. Macroinvertebrate densities were compared as habitat-weighted means, weighting the macroinvertebrates found in riffles and pools according to the available habitat present when streams were sampled (Grubaugh et al., 1997; Roy et al., 2003). Variables included Shannon's diversity, taxa richness, total density, and abundance of selected species traits calculated for

each sample and averaged within each stream in 2010 and 2011. Functional feeding groups (FFGs) were assigned using Merritt et al. (2008). Tolerance values were assigned from the Environmental Protection Agency Rapid Bioassessment Protocol and multiplied by density to express density-weighted tolerance of the macroinvertebrate community (Barbour et al., 1999). Tolerance values for the southeast are

326

E. Johnson et al. / Science of the Total Environment 530–531 (2015) 323–332

Table 1 Landscape scale variables for each site in the study. Separate data are provided for 2010 and 2011 for variables that changed between sample periods, namely variables associated with natural gas activity (well density, IFPLa, well rate of installation, and pipeline density). PAST = pasture, URB = urban, FOR = forest. Site

Area (km2)

% Slope

% PAST

% URB

% FOR

Clear Creek (CL) Tenmile Creek (TM) Black Fork Creek (BK) Fourteenmile Creek (FM) Clifty Creek (CY) Long Branch Creek (LB) Pine Mountain Creek (PM) Hogan Creek (HG) East Fork Point Remove (EPR) Sunny Side Creek (SS)

19 29 32 37 58 12 84 55 68 14

6 5 6 8 6 10 7 8 8 6

38 48 50 42 47 13 39 23 24 40

2 5 2 2 1 1 5 3 2 1

51 43 39 51 43 66 45 63 64 49

a

Well density (no./km2)

IFPLa

2010

2011

2010

2011

2010

2011

0.21 0.34 0.41 0.54 0.57 0.67 1.15 1.44 1.62 2.29

0.79 0.59 0.69 0.89 0.97 0.67 1.6 1.78 2.32 3.64

0.04 0.25 0.01 0.41 1.18 0.02 3.75 1.48 2.04 0.27

0.06 0.8 1.26 0.84 1.21 0.03 3.98 1.48 2.08 0.28

0.65 2.05 2.02 3.77 7.08 1.51 21.08 16.6 23.2 7.6

1.928 2.78 3.35 5.32 9.17 1.57 23.1 17.28 25.92 8.4

Installation rate (no./yr)

Pipeline (m/km2)

Unpaved road (m/km2)

1686.02 0 204.63 78.66 0 630.48 849.97 1221.08 891.93 1456.71

614.74 629.35 750.84 715.87 1047.67 567.7 715.72 973.75 1067.75 1307.15

IFPL = well pad inverse flow path length.

based on general sensitivity of taxa to organic pollutants (Lenat, 1993) similar to Hilsenhoff (1987).

were processed for turbidity and AFDM using the methods as above for sediment collected using siphon samplers.

2.3. Benthic organic matter

2.5. In-stream habitat

Organic matter from Hess and d-frame kicknet samples was placed in paper bags to dry at 60 °C in spring 2010. After drying, coarse benthic organic matter (CBOM ≥1 mm) and fine benthic organic matter (FBOM ≥250 μm and b1 mm) were sorted as wood or non-wood. Subsamples of CBOM and FBOM were weighed, ashed in a muffle furnace at 500 °C for 4 h, dried for 24 h, and re-weighed to obtain ash free dry mass (AFDM, Benfield, 2006). In spring 2011, five FBOM and CBOM benthic cores were taken along the 200 m stream reach. FBOM (all size fractions ≤1 mm) was frozen until processed and CBOM (≥ 1 mm) was placed in paper bags and dried in a 60 °C oven for at least 24 h. The same processing methods were used for CBOM in 2010 for 2011. Frozen FBOM samples were thawed at room temperature, bottles were shaken, and a subsample was taken with a 60 mL syringe. Subsample volume was recorded, filtered onto an ashed, pre-weighed Whatman™ glass-fiber filter (46 μm mesh and 47 mm diameter), ashed in an aluminum tin, dried at 60 °C for at least 24 h, weighed, ashed, wetted, dried, and re-weighed for AFDM (Benfield, 2006).

We used the point transect method to quantify benthic habitat in summer 2010 and summer 2011 within each stream reach following methods from Gordon et al. (2004). Habitat transects were established every 10 m along the 200 m reaches in each of the 10 sites, for a total of 20 transects in each reach. At each perpendicular transect, substrate type (Wentworth, 1922) and water depth were recorded every 0.5–1.0 m in the wetted channel depending on the channel width. The proportion of each substrate was calculated for each stream reach. Percent canopy cover was quantified at the 0, 100 and 200 meter mark using a spherical densitometer mid-channel in each stream reach (Minshall and Rugenski, 2006).

2.4. Suspended sediment sampling during storms 2.4.1. Siphon samples Siphon samplers were used to passively collect water (Graczyk et al., 2000) during the rising limb of storms on May 10th and 24th of 2010 and February 26th and April 28th of 2011. Two samplers were anchored to a riparian tree or rebar at 10 cm and 30 cm above estimated base flow in each study reach. Each siphon sampler contained a one liter Nalgene™ bottle that was collected following a storm, stored on ice, and returned to the laboratory. In the laboratory, turbidity was measured to the nearest Nephelometric Turbidity Unit (NTU) using a Hach 2100p turbidimeter. Samples were frozen until processed for suspended organic matter. Suspended organic matter concentrations (g/L) were quantified by thawing samples, filtering a known volume of subsample onto ashed and pre-weighed Whatman™ glass-fiber filter (46 μm mesh and 47 mm diameter). Filters were at 60 °C for 24 h, weighed, combusted at 500 °C for at least 4 h, wetted, dried, and reweighed for AFDM. 2.4.2. Grab samples Grab samples for suspended sediment were taken on the rising limb of storms on April 16th, June 4th, and July 21st of 2009 and on February 13th and May 10th of 2010. Each sample was collected with a one liter Nalgene™ bottle following a storm, placed on ice, and frozen. Samples

2.6. Statistical analysis The rate of gas well installation within a catchment was calculated as the slope of the regression line for active and spud wells over time (no./ yr). All suspended sediment concentrations taken during storms were averaged prior to the date of macroinvertebrate sampling for each stream. Principal components analyses (PCA) were used to select landscape variables to relate to macroinvertebrate metrics. Because gas well activity variables were all correlated, we chose well pad inverse flow path length (IFPL), well density, and pasture to represent land use and land cover most likely to explain variation in macroinvertebrate metrics. Well density was chosen because it had the least correlation with other gas activities and IFPL was chosen because it had the lowest correlation with gas well density. Pasture was chosen as the dominant land cover. Average macroinvertebrate density within streams was compared across study sites with one-way analysis of variance (one-way ANOVA) followed by Tukey post-hoc test if model results were less than α = 0.05. Our 3 chosen landscape variables were used in Pearson correlations with macroinvertebrate variables in 2010 and 2011. Instream habitat variables were also correlated with landscape variables. Finally, select macroinvertebrate Pearson correlations with habitat variables are shown in scatterplots. Normal distribution was examined for all variables using normal quantile plots and Shapiro–Wilk goodness of fit test. Residual plots were examined to detect unequal variance. Density data were log10 transformed and percentage data were arcsine square root transformed if patterns for normality or equal variance were not met. Macroinvertebrate community structure across the 10 streams was analyzed in 2010 with non-metric multidimensional scaling (NMDS). NMDS could not be presented for 2011 macroinvertebrate community

E. Johnson et al. / Science of the Total Environment 530–531 (2015) 323–332

Fig. 2. Principal components analyses illustrating measured landscape-level differences across streams in both 2010 (open circles) and 2011 (closed circles).

data as no solution could be resolved even after data transformation. All macroinvertebrate densities were log10 (x + 1) transformed and 32 taxa that were absent in more than six streams were removed from the analysis. Sorenson's distance measures were used to calculate sample distances relative to each other (McCune et al., 2002). Taxa correlation coefficients greater than 0.6 had a p b 0.05 and were considered important in NMDS groupings. NMDS axes were also correlated with landscape variables. PCA and NMDS were run in PC-ORD (version 6.9 MjM software). All other statistics were conducted in JMP® 10.2.0.

3. Results 3.1. Landscape differences across streams and between years The extent of land cover and intensity of land use differed across stream catchments. In particular, gas activities (i.e. well installation rate, pad area, IFPL, pipeline density, and well density) and density of unpaved roads spanned a gradient. Catchment size (12–84 km2) and to a lesser extent forest (39–66%) and pasture (13–50%) also varied across catchments. Gas well activity increased in all catchments from 2010 to 2011, with the exception of Long Branch Creek (Table 1, Fig. 2). The first three PCA axes were significant according to the broken-stick model (McCune et al., 2002). PC axis 1 explained 41%, PC axis 2 explained 28%, and PC3 explained 17% of the variation among streams. All gas well activity metrics and forest cover loaded on the negative side of axis one and pasture on the positive side. Forest cover, watershed slope, and pipeline density loaded on the positive side of axis two. Density of unpaved roads and gas well density loaded on the Table 2 Principal components analysis correlation coefficients for landscape-scale habitat variables from 2010 to 2011. Independent variable 2

Catchment area (km ) Gas well density (no. km−2) Gas well installation rate (no. yr−1) Well pad area (km2) Inverse flowpath length Density of pipelines (m km−2) Watershed slope (%) Forest (% cover) Pasture (% cover) Density of unpaved roads (m km−2) Urban (% cover)

Axis 1 (39%)

Axis 2 (28%)

Axis 3 (17%)

−0.71 −0.63 −0.96 −0.96 −0.72 −0.41 −0.52 −0.53 0.54 −0.45 −0.15

−0.59 0.16 −0.26 −0.28 −0.60 0.43 0.71 0.76 −0.74 0.07 −0.70

−0.18 0.66 0.01 −0.03 −0.19 0.28 −0.38 −0.30 0.39 0.83 −0.40

327

Fig. 3. Log10 mean daily average discharge for a regional river (Cadron Creek near Guy, AR) (USGS gage # 07261000) to show hydrologic variation over the study period where gray windows represent typical period when streams flow based on ten-year patterns. Arrows indicate approximate date macroinvertebrates were sampled.

positive side of axis 3 (Fig. 2, Table 2). Axis loading scores for well installation rate, well density, and IFPL were greater than pipeline density. Well density was least correlated with other gas metrics and IFPL represented other gas metrics and was least correlated with gas well density; therefore, they were selected as indicators of gas activity and correlated with macroinvertebrate variables (Fig. 2, Table 2). Pasture represented the other dominant land cover that could explain differences in macroinvertebrates and was used in correlations (Fig. 2, Table 2). Hydrology differed between the two years and may explain the temporal differences in macroinvertebrate community structure and their relationship to habitat across catchment in each year. In 2010, mean daily discharge, from a nearby USGS gaging station (#07261000), shows flow beginning in October 2009 with many relatively low intensity storms that maintained flow leading up to scheduled sampling in May 2010. However, in 2011, streams were mostly dry from late June 2010 through late January 2011 and sampling had to occur only several weeks after an intense storm when streams were flowing (Fig. 3). 3.2. Macroinvertebrate community structure across streams and years Macroinvertebrates in 53 distinct taxonomic groups were identified in 2010 and 60 in 2011 across streams. In both years, oligochaetes and chironomids represented the majority of taxa density (77%) and total density differed across streams (p b 0.001, Table 3) in 2010 but not in 2011 (p N 0.05). Macroinvertebrate taxa richness ranged from 10 to 24 across streams in both years and did not differ among the 10 streams in 2010 (X = 13 ± 2 (SD), p = 0.070), but richness did differ in 2011

Table 3 Average habitat-weighted macroinvertebrate taxa richness and density (±1 SE) for each stream (n = 10) in 2010 and 2011. Stream

CR TM BF FM CL LB PM HN EPR SS Average ±SE

2010

2011 2

No. taxa

Density (no./m )

No. of taxa

Density (no./m2)

11 ± 6 10 ± 3 17 ± 5 14 ± 4 13 ± 4 14 ± 4 13 ± 4 13 ± 7 15 ± 5 17 ± 3 14 ± 5

13,618 ± 13,499 4584 ± 3485 12,889 ± 13,347 9838 ± 7879 2620 ± 2211 5836 ± 4307 31,616 ± 31,598 23,792 ± 14,572 22,982 ± 29,051 11,602 ± 7238 13,938 ± 12,719

11 ± 2 14 ± 2 20 ± 2 17 ± 2 14 ± 2 15 ± 2 14 ± 1 16 ± 1 18 ± 1 24 ± 1 16 ± 1

3960 ± 1034 5288 ± 2033 13,279 ± 2988 5145 ± 2448 2938 ± 916 1744 ± 282 8289 ± 2169 1874 ± 367 22,746 ± 6838 4102 ± 611 6915 ± 2055

328

E. Johnson et al. / Science of the Total Environment 530–531 (2015) 323–332 Table 4 Taxa spearman rank correlations with non-metric multidimensional scaling ordination axes 1 and 2. r N 0.6. p-Value is less than 0.05 and are shown in bold. Taxa

2010

Acari Baetis Caecidotea Caenis Ceratopogonidae Cheumatopsyche Corydalus Diplectrona Haploperla Hexatoma Hyallela Non-Tanypodinae Oligochaeta Perlesta Progomphus Prosimulium Psephenus Seratella Stenelmis adult Stenelmis larvae Stenonema Tanypodinae

Fig. 4. Functional feeding group relative abundances in each stream (represented by stream abbreviation) in A) 2010 and B) 2011 where col = collector.

(p b 0.0001, Table 3). Macroinvertebrate communities in all streams were relatively tolerant and ranged from 4.5 to 6.1. However, densityweighted tolerance ranged an order of magnitude from 629 to 10,975 (#/m2) across streams in 2010. Communities in 2011 had a greater range in average tolerance values from 4.5 to 7.6 with less variation in density-weighted tolerances (range: 3786–10,663 (#/m2)) across streams. Collector-gatherers were the most abundant functional feeding group (FFG) representing an average of 78% and 65% in 2010 and 2011, respectively. Gatherers were mostly tolerant taxa such as nonpredaceous Chironomidae (tolerance value (tv = 6)), oligochaetes (tv = 5), and Baetis (tv = 5) and Caenis (tv = 6) mayflies. In 2010, predators were the second most abundant functional group at 12% (range 6– 25%) and dominated by Tanypodinae (tv = 7), Acari, and Ceratopogonidae (tv = 6.5) and only 8% of the total in 2011 represented mostly by the same taxa. In 2011, scrapers were second most dominant at

Axis 1

Axis 2

−0.22 −0.49 0.72 0.33 −0.04 −0.70 −0.43 −0.26 −0.05 0.36 0.61 −0.20 0.13 −0.44 0.70 −0.40 0.62 −0.61 −0.24 −0.41 0.08 0.20

0.78 0.42 0.14 0.77 0.81 0.68 0.75 −0.28 −0.32 0.02 0.13 0.90 0.45 0.77 0.11 0.66 0.28 0.59 0.84 0.88 0.62 0.85

21% (range 6–88% across streams) of the total dominated by Stenelmis beetles (tv = 5.4). In contrast, scrapers were only 4% (range 2–6%) of the total in 2010 and dominated by Stenelmis riffle beetles and Stenonema mayflies (Fig. 4A and B). Collector-filterers were only 5% (range 2–13%) of the total among streams in 2010 and 4% in 2011 and were dominated by Prosimulium blackflies (tv = 2.6) in both years, and also Cheumatopsyche (tv = 6.6) netspinning caddisflies in 2010 (Fig. 4). Shredders were the least abundant in both years representing an average of 1% (range 0–6%) in 2010 and 3% in 2011 represented mostly by Caecidotea isopods in 2010 and also Tipula in 2011 (Fig. 4). NMDS analysis separated distinct macroinvertebrate communities in two-dimensional space (final stress 5.16) with Axis 1 explaining 1% and Axis 2 explaining 88% of variation in the communities across streams in 2010. Macroinvertebrates separated along Axis 1 by the abundance and occurrence of the isopod, Caecidotea and amphipod, Hyalella. Many more macroinvertebrates separated along Axis 2, including collectorgatherer chironomid midges, riffle beetle larvae (Stenelmis), and Caenis and Stenonema mayflies (Fig. 5, Table 4). Table 5 Pearson correlations among macroinvertebrate metrics and landscape variables across all study streams in 2010 and 2011. Correlations where p b 0.05 are shown in bold.

Fig. 5. Non-metric multidimensional scaling representing log10 density of all dominant taxa (sampled in greater than 6 of 10 streams) collected spring 2010. Taxa with an r N 0.6 are shown in appropriate locations on the side of NMDS bi-plot. Refer to Table 6 for all taxa correlations with axes.

Macroinvertebrate variables

2010 Well IFPL density

Pasture Well IFPL density

Pasture

Habitat-weighted density Diversity Taxa richness Col-filterer density* Col-gatherer density Scraper density* EPT density E density P density T density Non-Tanypodinae density Tolerance-weighted density NMDS axis 1 (1.3%) NMDS axis 2 (88%) EPT taxa richness

0.46 0.01 0.52 0.34 0.45 0.52 0.55 0.79 0.42 0.37 0.29 0.53

0.84 −0.29 −0.07 0.90 0.86 0.57 0.69 0.62 0.44 0.73 0.79 0.70

−0.29 0.18 0.01 −0.31 −0.26 −0.38 −0.41 −0.31 −0.67 −0.30 −0.05 −0.29

0.17 0.46 0.78

−0.56 −0.24 0.41 −0.25 0.32 −0.30

*

log10 transformed.

2011

−0.06 0.54 0.52 0.25 −0.13 −0.03 0.60 0.74 0.46 −0.15 −0.16 −0.07

0.21 −0.15 0.04 0.27 0.39 0.11 0.09 −0.04 0.35 0.07 0.53 0.17

0.34 −0.37 0.01 −0.12 0.46 0.18 0.40 0.30 −0.30 0.41 0.30 0.31

NA NA 0.46

NA NA 0.49

NA NA −0.08

E. Johnson et al. / Science of the Total Environment 530–531 (2015) 323–332

329

Table 6 Pearson correlations among in-stream and landscape variables across all study streams in 2010 and 2011. Correlations where p b 0.05 are shown in bold. In-stream variables

Turbidity Suspended organic matter Suspended inorganic matter Silt (% cover) Fine benthic organic matter Gross primary production Chlorophyll a

2010

2011

Well density

IFPL

Pasture

Well density

IFPL

Pasture

0.19 −0.38 −0.36 −0.36 −0.26 −0.33 0.56

0.69 −0.40 0.11 0.25 −0.37 −0.66 0.89

−0.01 −0.47 −0.23 0.12 −0.41 −0.25 −0.12

−0.11 −0.03 −0.41 −0.59 0.08 NA 0.48

−0.33 −0.48 −0.17 −0.08 0.70 NA 0.47

0.42 0.29 0.09 0.38 −0.26 NA 0.05

3.3. Macroinvertebrate metrics correlated with landscape and habitat In 2010, habitat-weighted and tolerance-weighted macroinvertebrate densities were significantly positively correlated with IFPL (Table 5). Collector gathering and filtering taxa density, Ephemeroptera, Plecoptera, and Trichoptera (EPT), Ephemeroptera, and Trichoptera density were also positively correlated to IFPL. Plecoptera density was negatively correlated with pasture cover (Table 5). Number of EPT taxa was positively correlated with gas well density (Table 5). In 2011, there were few significant macroinvertebrate correlations with landscape variables; however, Ephemeroptera density was positively related to gas well density (Table 5). Although not statistically significant, of the chosen landscape variables, axis one was most negatively correlated with IFPL (Table 5). None of our 3 selected landscape variables were significantly correlated with NMDS axis 2 either, but well installation rate was weakly positively correlated (r = 0.61, p = 0.06).

Few in-stream variables correlated with landscape variables. In spring 2010, turbidity and chlorophyll a were positively correlated and gross primary production negatively correlated with IFPL (Table 6). In 2011, streams with greater well density had less silt cover; while streams with greater IFPL had more fine benthic organic matter (Table 6). No instream variables correlated with pasture cover in either year. In 2010, NMDS axis 1 was positively correlated with gross primary production whereby streams with greater GPP had more shredding isopods and amphipods and fewer scraping mayflies and filtering caddisflies (Fig. 6A). NMDS axis 2, represented mostly by an increase in tolerant collector-gathering and scraper taxa and riffle beetles, was not correlated with any measured in-stream variables, but showed a weak negative decline as suspended inorganic sediment increased (Fig. 6B). Macroinvertebrate taxa richness declined in streams with more benthic silt cover and Ephemeroptera density increased as chlorophyll a on rocks increased (Figs. 6C and D).

Fig. 6. Selected macroinvertebrate metrics related to in-stream habitat. Non-metric multidimensional scaling axes represent communities in 2010 and their greatest correlation with a measured stream habitat variables.

330

E. Johnson et al. / Science of the Total Environment 530–531 (2015) 323–332

4. Conclusions Total macroinvertebrate density, represented by mostly relatively tolerant gatherers and filterers, increased across a gradient of landscape alterations associated with gas well development in 2010, but not in 2011. In fact, 2011 mayfly density and taxa richness increased with catchment well density. In both years, well pads were constructed, wells drilled and fractured, and gathering lines installed. However, differences in the amount of activity across catchments coupled with differences in hydrology may explain reasons for the different annual patterns. In 2010, well pad inverse flow path length (IFPL) and gas well installation rate were the activities driving differences across sites and related to more organisms that were primarily short-lived generalist taxa like chironomids, oligochaetes and Caenis mayflies. Surprisingly, taxa that were indicative of stream catchments with more gas well activity were also scrapers, predominantly riffle beetles and Stenonema mayflies. Chlorophyll a and turbidity were related to greater gas activity in 2010 and seemed to drive the differences in macroinvertebrate density and community structure. Land clearing and riparian alteration associated with gas activities could have increased sediment and nutrients to streams, thereby increasing net primary production that was correlated to differences in macroinvertebrate densities across the gradient of gas activity. In 2011, only mayfly density positively correlated with catchments having a greater density of gas wells and gross primary production. Stream catchments in the Fayetteville Shale with more gas activity can also have more ponds that could retain sediment and reduce in-stream deposition (Hart, 2014). We speculate that a very dry winter and spring followed by record-breaking floods resulted in powerful stream flows that scoured the benthos and highly disturbed all macroinvertebrate communities. Average stream base flow was not related to macroinvertebrate metrics and more detailed hydrology was not available. Stream catchments with more gas activity also tend to have more retention ponds that may reduce extreme discharge during high-intensity storms. 5. Discussion Gas well development commonly results in more unpaved road construction and clearing land for pads and pipelines (Lave and Lutz, 2014). Across our study catchments, gas activity was more concentrated in the most central portion of our study area as build out occurred near “sweet spots” representing productive extraction sites as indicated by Moran's I index of spatial autocorrelation among gas activity metrics. Gas wells were installed within our study catchments at a rate of ~2 to 25 wells per year since 2005 (data from AOGC). Catchments with faster rate of installation also had greater well, unpaved road and well pad IFPL. It is important to note that while gas activity variables were updated for each catchment each year, land cover data and pipeline data were not updated, thus limiting interpretation between concurrent changes in gas activity and land cover (but see Moran et al., 2015). Pipeline data access is difficult, particularly for gathering lines, whereas land cover is generally updated on five year cycles. Nevertheless, field observations suggest a small change in forest and pasture cover relative to gas activity. Forest fragmentation within a catchment undoubtedly increased from pads, roads, and pipelines, but the total forest loss was ~2% based on an analysis of landscape changes across the Fayetteville Shale gas field (Moran et al., 2015). Land disturbance associated with gas activity was predicted to increase transported sediment into receiving streams or increase the probability of contamination associated with well fracturing (Preston et al., 2014). While the macroinvertebrate community structure reflected a disturbance gradient from gas activity, relationships between measured in-stream variables and landscape alterations were few and tended to be weaker. Turbidity and chlorophyll a were the most consistent positive correlates with gas activity and macroinvertebrate metrics. Canopy cover did not increase along this same gradient,

but nitrate concentrations did, which suggests nutrient enrichment from land clearing or fertilizing reseeded land around pads and pipelines postconstruction were the most likely reasons for greater chlorophyll α (Austin et al. 2015). We also measured specific conductivity one time in each stream and season as an indication of possible chemical contamination. Not surprising based on our sampling regime, we found no indication of elevated conductivity (range: 29–70 μs/cm) although it was positively correlated with gas well density in 2011 (Pearson r = 0.57, p = 0.05) that may be from dissolved nutrients (see Austin et al. 2015). Transported suspended sediment and turbidity were also often related to well density across streams when analyzed within a storm (Entrekin et al., 2011, unpublished data). Ideally, suspended sediment would be quantified regularly at both base flow and on the rising limb of all storms (Dodds and Whiles, 2004). Unfortunately, we were unable to achieve such a comprehensive sampling regime. Another possible reason why variation in in-stream habitat was not always explained by catchment activities was the differences in scale (Frissell et al., 1986; Townsend and Hildrew, 1994; Lammert and Allan, 1999; Allan, 2004). Landscape alterations occurred over months and years, changes in stream habitat occurred over days, weeks, and months. More frequent sampling of suspended sediment was likely necessary to relate landscape and habitat changes. Fortunately, macroinvertebrate community structure reflects longer term alterations in the habitat, water quality and food resources from longer-term landscape level changes (e.g. Roy et al., 2003). Landscape activities associated with gas development in this study seemed to have a similar effect on macroinvertebrate communities as other land disturbance where more short-lived generalist taxa were present in the catchments with greater disturbance (e.g., Hagen et al., 2006; Woodward et al., 2012). Macroinvertebrate density was extremely variable across streams similar to other studies examining land use alterations. For comparison, macroinvertebrate density in similar sized streams in the Etowah River drainage in Georgia ranged from 145 to 5894 no./m2 compared to 2260–31,616 no./m2 in this study (Roy et al., 2003). The average macroinvertebrate density in this study (13,938 no./m2) was much lower than macroinvertebrate density in a tallgrass prairie stream in northeast Kansas (23,843 no./m2) with less landscape alterations (Stagliano and Whiles, 2002). Although our study streams were surrounded by forest, pasture and UOG development activities, the community was dominated by chironomids and oligochaetes whereas predators and shredders dominated in Kansas, which again suggests that either deposited fine organic matter or algae as primary food resource in this study (Stagliano and Whiles, 2002). Land use in north-central Arkansas had been pasture for grazing cattle for decades. In more recent decades, pasture land is mostly hay fields with low density of cattle (http://www.nass.usda.gov/). Pasture cover in a catchment was only related to lower stonefly density in 2010. Stonefly density and richness were low in all streams likely from a long history of altered stream quality from past land use practices (Harding et al., 1998). Despite a true reference condition and pre-gas development data, this study suggests that land disturbance from gas development affected stream communities in 2010; and to a lesser extent in spring 2011. When trying to parse out land use effects on aquatic communities, hydrologic regime is critical to understand, but often difficult to quantify in ungagged streams without direct manipulations (Bunn and Arthington, 2002). It is not unusual for hydrology to affect macroinvertebrate communities differently depending on substrate composition and hydrology also likely mediated the effects of land use on receiving streams (Holomuzki and Biggs, 2000; Poff et al., 2006b). Habitat and macroinvertebrate data from these streams prior to gas drilling activity would have provided useful information to understand the magnitude of effect of gas activity on stream benthic habitat. Macroinvertebrate communities in this study do resemble communities found in systems with long-term disturbance that could make these streams less vulnerable to gas development activities (Lamouroux et al., 2004; Harding et al., 1998; Richards et al., 1997; Paukert et al., 2011).

E. Johnson et al. / Science of the Total Environment 530–531 (2015) 323–332

Even relatively low-impact pasture activities and likely in-stream habitat modifications associated with past land use have resulted in macroinvertebrate communities reflective of disturbed and altered ecosystems with relatively low taxa richness and diversity as has been documented by other studies (Miller and Golladay, 1996; Harding et al., 1998). Macroinvertebrate diversity in this study was ~1.4 compared to 2.7 in forested Oregonian streams (Herlihy et al., 2005). Macroinvertebrate taxa richness was also low in this study, averaging 14 ± 3 compared to an average of 43 taxa found in urbanized streams within the Etowah River drainage in north-central Georgia that had similar range in catchment size (11–125 km2) (Roy et al., 2003). Variation in macroinvertebrate richness and diversity was not really explained by landscape level activities, rather richness and diversity increased when silt decreased and coarse substrate (gravel, pebble, and cobble) increased. Substrate composition of the stream channels explained significant variation as has been found in other studies (Grubaugh et al., 1997; Quinn and Hickey, 1990; Whiles and Wallace, 1992). A decrease in silt cover and subsequent increase in mayfly density and richness could be from more ponds in the catchments retaining runoff particularly during intense storms (Hart, 2014). We speculate that gas activity in 2010 coupled with moderate and earlier flow in spring resulted in detectable alterations of a more stable macroinvertebrate community. In contrast, the storm that occurred three weeks prior to spring 2011 sampling may have had a much greater effect on in-stream habitat and macroinvertebrate community structure than any one landscape activity (Wallace, 1990). Macroinvertebrate community structure in streams with the most gas activity was similar to macroinvertebrate indicators of moderate eutrophication in 10 streams in Ireland with a gradient of nutrient concentrations (Woodward et al., 2012). Elmids, baetid mayflies, and Ephemerellidae were shared indicators between this study and theirs. However, non-predatory chironomids and mites were also important indicators of landscape alteration in this study and were likely not quantified in the pan-European analysis of leaf decomposition. It is not surprising that plurivoltine taxa were most responsive to landscape alterations. Because of their fast lifecycle, they often are excellent indicators of changes occurring over a short term. It will be important to also monitor the large and longer lived taxa sensitive to changes in water quality such as Cheumatopsyche, Perlesta, and Polycentropus in this study (Lamouroux et al., 2004). Chlorophyll a and gross primary production indicate moderately eutrophic streams that may explain ephemerellids and elmids as macroinvertebrate indicators identified in 2010 (Austin et al., 2015). Although it was predicted streams draining catchments with well density at ~4 km2 to be greatly altered; benthic sediment, gross primary production, and macroinvertebrate community structure and tolerance values point to systems where added resources and degraded water quality are increasing density of intermediate and tolerant taxa (Woodward et al., 2012). In Arkansas, the Fayetteville Shale is predicted to provide natural gas for 30 more years adding 500 wells per year (Arthur et al., 2008; Arkansas State Water Plan, 2014); therefore, it is imperative that potential ecological effects are studied. Currently, wells are placed close to streams, which could alter surface water quality and biota (Entrekin et al., 2011). In this study, there were over 1000 natural gas wells installed across 10 catchments, which resulted in patterns of biological change that could be due at least in part to landscape alteration that results in sedimentation and associated nutrient enrichment. Low statistical power and correlative analysis limit our ability to establish cause and affect relationships. However, gas well activities were the overarching landscape-level variable related to macroinvertebrate community structure. Closer examination of landscape–hydrology–habitat relationships, in particular activities associated with natural gas extraction, is needed to protect water quality. Proactive development that sites pads farther from stream channels and slows development in catchments with species of concern would also reduce the potential for negative impacts to receiving streams and provide time for pre-

331

impact assessment in plays where complete build-out has occurred (e.g. Davis and Robinson, 2012). Acknowledgments Funding for this research came from the State Wildlife Grants Program grants #T31-03 and T33-01. University of Central Arkansas College of Natural Science and Math provided additional support to Erica Johnson. We would like to thank Adam Musto, Julie Kelso, Katie Rose, Chris Fuller, Amanda Bates, Josh Bregy, Allyn Fuell, Oliver Herbst, Jordan Johnson, Will Mauer, Lindsey Martindale, Chelsea Miller, Daniel Sniegowski, Loren Stearman, and Richard Walker for their help in the laboratory and in the field. We also thank collaborators Ginny Adams, Reid Adams, Brian Haggard, Steve Filipek, and Lindsey Lewis. References Allan, J., 2004. Landscapes and riverscapes: the influence of land use on stream ecosystems. Annu. Rev. Ecol. Evol. Syst. 35, 257–284. Arkansas State Water http://www.arwaterplan.arkansas.gov/plan/ArkansasWaterPlan/ 2014AWPWaterPlan/AWP%20Update%202014_Summary.pdf. Arthur, J.D., Bohm, B., Coughlin, B.J., Layne, M., 2008. Hydraulic Fracturing Considerations for Natural Gas Wells of the Fayetteville Shale. ALL Consulting. Barbour, M.T., Gerritsen, J., Snyder, B.D., Stribling, J.B., 1999. Rapid Bioassessment Protocols for Use in Streams and Wadeable Rivers: Periphyton, Benthic Macroinvertebrates and Fish. Second edition. U.S. Environmental Protection Agency; Office of Water, Washington, D.C. (EPA 841-B-99-002). Benfield, E., 2006. Decomposition of leaf material. In: Hauer, F., Lamberti, G. (Eds.), Methods in Stream Ecology. Elsevier, Amsterdam, pp. 711–720. Boulton, A.J., Peterson, C.G., Grimm, N.B., Fisher, S.G., 1992. Stability of an aquatic macroinvertebrate community in a multiyear hydrologic disturbance regime. Ecology 73, 2192–2207. Brittingham, Margaret, et al., 2014. Ecological risks of shale oil and gas development to wildlife, aquatic resources and their habitats. Environ. Sci. Technol. 48 (19), 11034–11047. Bunn, S.E., Arthington, A.H., 2002. Basic principles and ecological consequences of altered flow regimes for aquatic biodiversity. Environ. Manag. 30 (4), 492–507. Burton, G.A., Basu, N., Ellis, B.R., Kapo, K.E., Entrekin, S., Nadelhoffer, K., 2014. Hydraulic “fracking”: are surface water impacts an ecological concern? Environ. Toxicol. Chem./SETAC 33 (8), 1679–1689. Davis, J.B., Robinson, G.R., 2012. A geographic model to assess and limit cumulative ecological degradation from Marcellus Shale exploitation in New York, USA. Ecol. Soc. 17 (2), 25. Delong, M., Brusven, M., 1998. Macroinvertebrate community structure along the longitudinal gradient of an agriculturally impacted stream. Environ. Manag. 22, 445–457. Dodds, W.K., Whiles, M.R., 2004. Quality and quantity of suspended particles in rivers: continent-scale patterns in the United States. Environ. Manag. 33 (3), 355–367. Drohan, P., Brittingham, M., Bishop, J., Yoder, K., 2012. Early trends in landcover change and forest fragmentation due to shale-gas development in Pennsylvania: a potential outcome for the northcentral Appalachians. Environ. Manag. 49 (5), 1061–1075. Entrekin, S., Evans-White, M., Johnson, B., Hagenbuch, E., 2011. Rapid expansion of natural gas development poses a threat to surface waters. Front. Ecol. Environ. 9, 503–511. Frissell, C., Liss, W., Warren, C., Hurley, M., 1986. A hierarchical framework for stream habitat classification: viewing streams in a watershed context. Environ. Manag. 10, 199–214. Gordon, N., McMahon, T., Finlayson, B., Gippel, C., Nathan, R., 2004. Stream Hydrology: An Introduction for Ecologists. 2nd edition. John Wiley & Sons, Sussex, England. Graczyk, D., Robertson, D., Rose, W., Steuer, J., 2000. Comparison of water-quality samples collected by siphon samplers and automatic samplers in Wisconsin. United States Geological Survey, United States Geological Survey Fact Sheet FS-067-00. USGS, Middleton, WI. Grubaugh, J., Wallace, B., Houston, E., 1997. Production of benthic macroinvertebrate communities along a southern Appalachian river continuum. Freshw. Biol. 37, 581–596. Hagen, E., Webster, J., Benfield, E., 2006. Are leaf breakdown rates a useful measure of stream integrity along an agricultural landuse gradient? J. N. Am. Benthol. Soc. 25, 330–343. Harding, J., Benfield, E., Bolstad, P., Helfman, G., Jones, E., 1998. Stream biodiversity: the ghost of land use past. Proc. Natl. Acad. Sci. 95, 14,843–14,847. Hart, R., 2014. Simulated effects of existing and proposed surface-water impoundments and gas-well pads on streamflow and suspended sediment in the Cypress Creek watershed. Arkansas: U.S. Geological Survey Scientific Investigations Report, pp. 2014–5057 (36 pp.). Herlihy, A.T., Gerth, W.J., Li, J., Banks, J.L., 2005. Macroinvertebrate community response to natural and forest harvest gradients in western Oregon headwater streams. Freshw. Biol. 50, 905–919. Hilsenhoff, W.L., 1987. An improved biotic index of organic stream pollution. Great Lakes Entomol. 20, 31–39. Holomuzki, J.R., Biggs, B.J.F., 2000. Taxon-specific responses to high-flow disturbance in streams: implications for population persistence. J. N. Am. Benthol. Soc. 19, 670–679.

332

E. Johnson et al. / Science of the Total Environment 530–531 (2015) 323–332

Kargbo, D.M., Wilhelm, R.G., Campbell, D.J., 2010. Natural gas plays in the Marcellus shale: challenges and potential opportunities. Environ. Sci. Technol. 44, 5,679–5,684. Lammert, M., Allan, J., 1999. Assessing biotic integrity of streams: effects of scale in measuring the influence of land use/cover and habitat structure on fish and macroinvertebrates. Environ. Manag. 23 (2), 257–270. Lamouroux, N., Dolédec, S., Gayraud, S., 2004. Biological traits of stream macroinvertebrate communities: effects of microhabitat, reach, and basin filters. J. N. Am. Benthol. Soc. 23, 449–466. Lave, R., Lutz, B., 2014. Hydraulic fracturing: a critical physical geography review. Geogr. Compass 8 (10), 739–754. Lenat, D.R., 1993. A biotic index for the southeastern United States: derivation and list of tolerance values, with criteria for assigning water quality ratings. J. N. Am. Benthol. Soc. 12, 279–290. McCune, B., Grace, J., Urban, D., 2002. Analysis of Ecological Communities. MjM Software Design, Gleneden Beach, Oregon. Merritt, R., Cummins, K.W., Berg, M., 2008. An Introduction to the Aquatic Insects of North America. 4th edition. Kendall Hunt, Dubuque, IA. Miller, A.M., Golladay, S.W., 1996. Effects of spates and drying on macroinvertebrate assemblages of an intermittent and a perennial prairie stream. J. N. Am. Benthol. Soc. 25, 670–689. Minshall, G.W., Rugenski, A., 2006. Riparian processes and interactions. In: Hauer, F., Lamberti, G. (Eds.), Methods in Stream Ecology. Elsevier, Amsterdam, pp. 741–744. Moran, M.D., Cox, B., Wells, R.L., Benichou, C.C., McClung, M.R., 2015. Habitat Loss and Modification due to Gas Development in the Fayetteville Shale. Environ. Manage. 1–9. Olmstead, S.M., Muehlenbachs, L.A., Shih, J.-S., Chu, Z., Krupnick, A.J., 2013. Shale gas development impacts on surface water quality in Pennsylvania. Proc. Natl. Acad. Sci. 110 (13), 4962–4967. Paukert, C.P., Pitts, K.L., Whittierb, J.B., Olden, J.D., 2011. Development and assessment of a landscape-scale ecological threat index for the Lower Colorado River Basin. Ecol. Indic. 11, 304–310. Poff, N.L.R., Olden, J., Vieira, N., Finn, D., Simmons, M., Kondratieff, B., 2006a. Functional trait niches of North American lotic insects: traits-based ecological applications in light of phylogenetic relationships. J. N. Am. Benthol. Soc. 25, 730–755. Poff, N.L.R., Bledsoe, B., Cuhaciyan, C., 2006b. Hydrologic variation with land use across the contiguous United States: geomorphic and ecological consequences for stream ecosystems. Geomorphology 79, 264–285. Preston, T.M., Chesley-Preston, T.L., Thamke, J.N., 2014. A GIS-based vulnerability assessment of brine contamination to aquatic resources from oil and gas development in eastern Sheridan County, Montana. Sci. Total Environ. 472, 1152–1162. Quinn, J.M., Hickey, C.W., 1990. Magnitude of effects of substrate particle size, recent flooding, and catchment development on benthic invertebrates in 88 New Zealand rivers. N. Z. J. Mar. Freshw. Res. 24, 411–427. Rahm, B.G., Riha, S.J., 2014. Evolving shale gas management: water resource risks, impacts, and lessons learned. Environ. Sci.: Processes Impacts 16 (6), 1400–1412. Resh, V.H., Brown, A.V., Covich, A.P., Gurtz, M.E., Li, H.W., Minshall, G.W., Reice, S.R., Sheldon, A.L., Wallace, J.B., Wissmar, R.C., 1988. The role of disturbance in stream ecology. J. N. Am. Benthol. Soc. 7, 433–455. Richards, C., Haro, R., Johnson, L., Host, G., 1997. Catchment and reach scale properties as indicators of macroinvertebrate species traits. Freshw. Biol. 37, 219–230. Roy, A., Rosemond, A., Leigh, D., Paul, M., Wallace, J., 2003. Habitat-specific responses of stream insects to land cover disturbance: biological consequences and monitoring implications. J. N. Am. Benthol. Soc. 43, 292–307.

Roy, A., Freeman, M., Freeman, B., Wenger, S., Ensign, W., Meyer, J., 2006. Importance of riparian forests in urban catchments contingent on sediment and hydrologic regimes. Environ. Manag. 37, 523–539. Rozell, D.J., Reaven, S.J., 2012. Water pollution risk associated with natural gas extraction from the Marcellus shale. Risk Anal. 32 (8), 1382–1393. Ryan, P., 1991. Environmental effects of sediment on New Zealand streams: a review. N. Z. J. Mar. Freshw. Res. 25, 207–221. Snyder, C.D., Young, J.A., Lemarié, D.P., Smith, D.R., 2002. Influence of eastern hemlock (Tsuga canadensis) forests on aquatic invertebrate assemblages in headwater streams. Can. J. Fish. Aquat. Sci. 49, 262–275. Souther, S., Tingley, M.W., Popescu, V.D., Hayman, D.T., Ryan, M.E., Graves, T., Terrell, K., 2014. Biotic impacts of energy development from shale: research priorities and knowledge gaps. Front. Ecol. Environ. 12 (6), 330–338. Stagliano, D.M., Whiles, M.R., 2002. Macroinvertebrate production and trophic structure in a tallgrass prairie headwater stream. J. N. Am. Benthol. Soc. 21, 97–113. Stewart, K., Stark, B., Stanger, J., 1993. Nymphs of North American stonefly Genera (Plecoptera). 2nd edition. University of North Texas Press, Denton, Texas. Thorp, J.P., Covich, A.P., 2001. Ecology and Classification of North American Freshwater Invertebrates. 2nd edition. Academic Press, San Diego, California. Townsend, C.R., Hildrew, A.G., 1994. Species traits in relation to a habitat templet for river systems. Freshw. Biol. 31 (3), 265–275. United States Department of Energy/Energy Information Agency, 2013. Annual Energy Outlook (244 pp.). United States Environmental Protection Agency (USEPA), 2009. National Water Quality Inventory: Report to Congress: 2004. Reporting Cycle Report No. EPR841-R-08-001. Office of Water, Washington, DC. Wallace, J.B., 1990. Recovery of lotic macroinvertebrate communities from disturbance. Environ. Manag. 14, 605–620. Walsh, C., Roy, A., Feminella, J., Cottingham, P., Groffman, P., Morgan, R., 2005. The urban stream syndrome: current knowledge and the search for a cure. J. N. Am. Benthol. Soc. 24, 706–723. Waters, T., 1969. Subsampler for dividing large samples of stream invertebrate drift. Limnol. Oceanogr. 14, 813–815. Weigel, B.M., 2003. Development of stream macroinvertebrate models that predict watershed and local stressors in Wisconsin. Development 22 (1), 123–142. Wentworth, C.K., 1922. A scale of grade and class terms for clastic sediments. J. Geol. 30, 377–392. Whiles, M., Wallace, J.B., 1992. First year benthic recovery of a headwater stream following a 3 year insecticide induced disturbance. Freshw. Biol. 28, 81–91. Wiggins, G.B., 1996. Larvae of the North American Caddisfly Genera (Trichoptera). 2nd edition. University of Toronto Press, Toronto, Canada. Williams, H.F., Havens, D.L., Banks, K.E., Wachal, D.J., 2007. Field-based monitoring of sediment runoff from natural gas well sites in Denton County, Texas, USA. Environ. Geol. 55, 1,463–1,471. Wood, P., Armitage, P., 1997. Biological effects of fine sediment in the lotic environment. Environ. Manag. 21, 203–217. Woodward, G., Gessner, M.O., Giller, P.S., Gulis, V., Hladyz, S., Lecerf, A., …, Chauvet, E., 2012. Continental-scale effects of nutrient pollution on stream ecosystem functioning. Science 336 (6087), 1438–1440.