Ec..otogicat Applications, 22(3), 2012, pp. 870-879 © 2012 by the Ecological Society of America

Geologic processes influence the effects of mining on aquatic ecosystems

TRAvtS S. SCHMlUT",2,4 WtLL.lAM H, CLEMENTS,3 RICHARD B, WANTY,' PHILtP L. VERPLANCK,' STANLEY E, CHURCH,]

CARMA A. SAN JUAN,l DAVIU L. FEy,l BARNABY W. ROCKWELL,l En H. DEWtTT,] AND TERRY L. KLEIN l

I u.s. Geologic Survey, Crustal Geophysics and Geochemistry Science Cenler, Denver, Colorado 80225 USA 2u.S, Geologic Survey, Colorado Water Science Center, Fort Collins Science Center, Fort Collins, Colorado 80526 USA 3Colorado Slate University, Departmenl of Fish, Wildli!e, and Conservation Biology, Fort Collins, Colorado 80525 USA

Abstract. Geologic processes strongly influence water and sediment quality in aquatic ecosystems but rarely are geologic principles incorporated into romine biomonitoring studies. We tcst if elevatcd concentrations of metals in water and sediment are restricted to streams downstream of mines or areas that may discharge mine wastes, We surveyed 198 catchmeJ1ls classified as "historically mined" or "unmincd," and bascd on mineral-deposit criteria, to determine whether water and sediment quality were influenced by naturally occurring mineralized rock, by historical mining, or by a combination of both. By accounting for different geologic sources of mctals to the environment, we were able to distinguish aquatic ecosystcms limited by metals derived from natural processes from those due to mining. Elevated concentrations of metals in water and sediment wcre not restricted to mined catchments; depauperate aquatic communities were found in unmined catchments. The type and intensity of hydrothermal alteration and the mineral deposit type were important determinants of water and sediment quality as well as the aquatic community in hoth mined and unmined catchments. This study distinguished the effects of different rock types and geologic sources of metals on ceosystems by incorporating basic geologic processes into reference and baseline site selection, resulting in a refined assessment. Our results indicate that biomonitoring studies should account for natural sourccs of metals in some geologic environments as contributors to the effect of mines on aquatic ecosystems, recognizing that in mining-impacted drainagcs there may have been high pre-mining background metal concentrations.

Key words: aqualic communities: bio/l1oniroring: geology; mewls: /I1ining; Rocky Mounrain (USA) streams; sediment qualily; water quality.

INTRODUCTION (75 to 45 million years ago), a period of mountain building in western North America that involvcdGlobal anthropogenic metals inputs (atmospheric cmplacement of plutons and associated sulfidc mineral dcposition and industrial discharge) to aquatic ecosys­deposits of various types (e.g., Twcto and Sims 1963). tems exceed that due to natural processcs (Nriagu and Metals from these sulfidc minerals have been released

Pacyna 1988). At local and regional scales, the into aquatic ecosystcms, but the relative contributions of

contributions of natural vs. anthropogcnic metal loading weathering of unmined hydrotherrnally altercd rock vs.

into aquatic ecosystems are poorly constraincd. In historical mining is largely unknown.

Colorado, USA, over 13 840 km of stream and river Hydrothermal alteration can be an integral part of the

havc been designated as impaired (Clean Water Act's ore-fomling process in such a way that different suites of

section 303(d) list in 2008; available online)5 because of minerals are created (and locally extracted by mining) in

elevated metals (aluminum, cadmium, copper, lead, areas with different types of hydrothermal alteration.

iron, selenium, and zinc). Gencrally, this dcgradation Mines can be located in areas where hydrothermal

is assumed to result from the mineral-extraction alteration has resultcd in a mineral dcposit of sufficicnt

economy that began in central Colorado in 1859 metal concentration that it is cconomic to extract metals (Chronic and Chronic 1972). However, central Colorado from the rock. However, mines or mineral deposits of was heavily mineralized during the Laramide orogeny certain deposit types are commonly surrounded by a

more diffuse halo of altered rock that also contains Manuscript received 4 May 20 II; revised 13 October 20 II; concentrations of metals likely above crustal abundance.

accepted 28 October 2011. Corresponding Editor: K. N. Because of these ore-formation processes it is possibleEshleman. that elevated concentrations of metals in bedrock

4 Present address: Fort Collins Science Center, Fort Collins, Colorado 80526 USA. E-mail: tschmidt@usgs.gov influenced water and sediment quality in streams prior

5 http://www.epa.gov/waters/ir/index.html to mining (Tooker 1963, Tweto 1968, Wanty et al. 2002,

870

872 Ecological Applications TRAVIS S. SCHMIDT ET AL. Vol. 22, No.3

immature soils, as well as unconsolidated overburden, are found intermixed at lowcr clcvations and along strcams.

The sample locations in this study are from small catchments (first- and second-order streams) at high altitude, ranging from 2330 to 3550 m above sea level. Catchments were selected that were underlain by a single predominant lithology (rocks of similar geochemistry and mode of formation). Drainage area boundaries (herein called "catchments") werc dcfincd with digital elevation models (DEM; 30 x 30 m resolution) using thc ArcGIS 9.2 geographic information system software (ESRI 2006, Gesch 2007). Catchment boundaries along with geologic maps and databascs of minc locations (sec Geologic mapping, below) were used to classify each sample location (see Site classification, below) prior to going into thc field (Fig. I).

Gcochcmical and benthic macroinvertebrate samples were collected from these locations during base-flow conditions in the summers (July-August) of 2004-2007 (Fig. I; Church et al. 2009, Wanty et al. 2009). Geochemical samples wcrc collccted from streams of various sizes (i.e., watershed area, sitc elevation, discharge), but biological data collection was confined to a narrower range of stream sizes to decrease the role variable habitat plays in structuring the aquatic inver­tebrate community. Although all field studies are influenced by confounding variables such as habitat, neither small-scale (i.e., percentage cover, substrate size) nor large-scale habitat factors (i.e., watershed area, elevation, percentage watershed vegetation) were strong predictors of mean biological responses in this data set (Schmidt et al. 2010). As a result, 198 locations were sampled for gcochemistry (n = 263 samples for water and sediment) and 94 locations for benthic invertebrates (n = 110 samples, five replicates each) were included in this analysis. Thc differences between the number of samples and that of site locations are interannual replicates included to add annual variability of the benthic communitics and geochemistry into the analysis. Duplicates of geochemical samples also were collected from -10% of sample locations.

Geologic mapping

Databases from the State of Colorado (Sares et al. 2008) and the Federal Government (Mineral Resource Data System, USGS 2005) were used to determine dcposit type and disturbance by mining (Fig. I). The number of mine openings and a qualitative estimate of the volume of mine tailing in a catchment were derived from these databases as a description of the extent of disturbance by mining within a catchment. Catchments were characterized as "mined" if publically available data indicated that a commodity from thc sitc was produced. Catchments not mined but with disturbances such as adits and prospect pits were elassified as "unmined" because no commodity was produced and it was likely the extent of disturbance was minimal.

Hydrothermal alteration was mapped and character­ized using mineral maps derived from analysis of Advanced Spacebornc Thcrmal Emission and Reflection Radiometer (ASTER) remote-sensing data (e.g., Rock­well 2009) and verified at the local scale with Airborne Visible/Infrared Imaging Spectrometer (AVI RIS) data. Hydrothermal alteration identified using the ASTER data was classified into quartz-sericite-pyrite (QSP) and propylitic on the basis of spectrally identificd mincral asscmblagcs. Ecologically these two types of alteration are important because they are dcscriptivc of acid­generation potential. QSP alteration contains pyrite in host rock with little acid-ncutralizing capacity resulting from the alteration process. As the pyrite wcathers it releases sulfuric acid, and drainages from these areas often have pH values below 7. In contrast, propylitic alteration emplaces acid-neutralizing minerals (c.g., calcite, chlorite, and epidote) into host rocks that weather to release drainage that is often near-neutral in pH. The mapping of hydrothermal alteration using remote-sensing data is possible only where tile ground is not covered by vegetation, so some watersheds in the study area are hydrothermally altered but not detected by thesc methods. For more dctailed information on the gcological mapping and for the characteristics of principal hydrothermal mineral-dcposit types found in the study area see Church et al. (2009).

Geochemical analysis

Water samples were collected using methods de­scribed in Wilde et al. (1998) to meet the requirements of the biotic ligand model (HydroQual 2007). Routine water-quality parameters (temperature, conductivity, and pH) were measured in the field using a Horiba D­24 combination meter (Wilde et al. 1998). Mctcrs wcrc calibrated at the beginning of each day with certificd standards, and checked periodically throughout thc day. All water samples were filtered through an Acrodisc Premium 25-mm syringe fil ter with 0.45-l-lm nylon membrane (Pall Corporation, Port Washington, Ncw York, USA) in the field and stored at 4°C until analyzed. Water samples for dissolved organic carbon (DOC) were filtered in the field through a 0.70-l-llll glass-fiber filter, acidified with concentrated hydrochloric acid (12 mol/L) to a pH < 2, and stored in baked amber-glass bottles. Dissolved trace-metal samples were acidified in the field with concentrated ultrapure nitric acid (13 moll L) to a pH < 2 and stored in polyethylene bottles. Samples for anions were collected and stored under refrigeration in polyethylene bottles. Streambed sedi­ment samples were collected from surfaces of fluvial sediment deposits in pool or low-velocity areas along the sampled reach using a plastic scoop. The samples were composited, wet-sieved through a 2-mm stainless-stcel screen witll ambient stream water, and collected into a plastic pan. In the laboratory these samples were prepared for chemical analysis by air-drying and dry­sicving them through a l50-l-lm stainless-steel scrcen.

Eculogical Applications 874 TRA VIS S SCHMIDT ET AL.

All catchments n = 263, 110

L'hl'~It 0 oglc re erence . . Hydrothermally altered n=62,15 ~

Unmined'·2 Mined'? n'~ 139,66 n = 62, 29

~ 1 PMV3 Porphyry3 PMV

n = 25, 10 n = 15, 7 ~

Propyliticly asp altered) altered3

n=27,13 n·c 14,7

FIG. 2. Breakdown of the catchment classification, with sample size (n) given for the geochemical (first number) and aquatic invertebrate community (second number). Unmined catchments were subdivided into lithologic reference and those influenced by hydrothermal alteration. Hydrothermally altered catchments were further subdivided by mineral-deposit type. Mined sites from a single deposit type (polymetallic veins, PMV) were reclassified into two distinctly different styles of hydrothem1al alteration: propylitic and QSP (quartz-sericite­pyrite). The superscript numbers define groups of data that were used to test different hypotheses: (I) Data used to test the hypothesis that elevated concentrations of metals in water and sediment and depauperate aquatic communities are restricted to historically mined catchments, with results presented in Fig. 3. (2) Data used to test the second hypothesis, that geologic processes (i.e., hydrothermal alteration and mineral deposit type) control stream geochemistry and aquatic communities, with results presented in Fig. 4. (3) Data used to test the second hypothesis, that geologic processes (i.e., hydrothermal alter­ation and mineral deposit type) control stream geochemistry and aquatic communities, with results presented in Fig. 5.

Probable-effects conccntrations arc empirically derived thresholds abovc which harmful effects are likely to be observed.

Site classification

To test our first working hypothesis, that clevated concentrations of metals in water and sediment and depaupcratc aquatic invertebrate communities are re­stricted to historically mined catchments, all samples were classified by the presence (n = 62 samples for geochemistry, n = 29 samples for biology) or absence (n = 20 I, 81 samples, respcctively) of disturbance by mining. To test our second working hypothesis that geologic processes (i.e., hydrothermal alteration and mineral-deposit type) control stream geochemistry and aquatic communities, we reclassified our samples based on the presence of hydrothermal alteration, mineral­deposit type, and mining (Fig. 2, Table I). Lithologic reference (n = 62, 15 samples, rcspcctively) catchments are areas where hydrothermal altcration was not detected in the AYIRISjASTER images and were not mined. Hydrothermally altered (n = 139, 66 samples, respectively) catchmcnts arc arcas with detectable hydrothermal alteration at the earth's surface, but wcrc not mined. Because not all hydrothermal alteration results in the formation of a mineral deposit, not all

Vol. 22, NO.3

hydrothcrmally altered sites wcre mined or classified into different mineral deposit types (Fig. 2, Table I). Thc effects of different styles of hydrothermal alteration (propYlitic vs. QSP) on unmined polymetallic vein deposits were indistinguishable and werc lumped to­gether into one catcgory, background polymctallic vein deposits (n = 25, 10 samples, rcspectively). Porphyry mineralizcd sites, which were not mincd, were classified as background porphyry deposits (n = 15, 7 samples, respectively). Because few mincd and unremediated porphyry style deposits were found within the study area, the effect of mining and remediation on porphyry­style dcposits was not evaluated.

Statistical analysis

MUltiple working hypotheses were evaluated requir­ing tests between groups of data classified based on disturbance (i.e., presencc or absence of mining), hydrothermal alteration (propylitic and QSP), and mineral-deposit type (polymetallic vein and porphyry). Because normality and homogencity of variance could not be achieved in most log-transformed data, differ­ences in means were tested using nonparametric procedures. A Wilcoxon rankcd sum test (P < 0.05) or Kruskal-Wallis ANOYA by ranks and Dunn's mUltiple-comparison test were used to evaluate differ­ences among groups (Siegel and Castellan 1988, Sokal and Rohlf 2003). Boxp]ots were used to depict the results for casc of interpretation. All statistical compar­isons were developed using R software version 2.7.2 (R Development Core Team 2008).

RESULTS

Samples were collected from diverse lithologies resulting in a wide spectrum of aqueous and sediment geochemical characteristics (Table 2). Disturbance by mining was extensive, with up to 146 inactive mine openings (estimate of the number of mines within a catchment), and >700000 m3 of historic mine tailings (a qualitative visual cstimate of the amount of earth disturbed and deposited on the surface) were observed in some catchments. The prcsence of mineralized rock and historic mining on stream water and sediment caused somc strcams to have very low pH « 4), and concentrations of metals in water and sediment in excess of 200 and 50 times, respectively, the thresholds thought safc for aquatic organisms (Table 2).

Water (Fig. 3A) or sediment (Fig. 3B) from mined (n = 62) catchments were statistically more toxic than in unmined (n = 201) catchments. The average EPT (Ephemeroptera, Plecoptera, Tricoptcra) index (Fig. 3C) in unmined catchments (n = 81) was statistically highcr than in mined catchments (n = 29). Howcvcr, water or sediment cxceeded toxic thresholds in 31 of 20 I unmined (Fig. 3A, B) catchments, and the lowest EPT index (0.2; Fig. 3C) was observed in an unmined catchment. Elevated metal concentrations in water and sediment likely created an environment that supportcd

-----------

876 Ecologica] Applications TRA VIS S. SCHMIDT ET AL. Vol. 22, NO.3

1000

100

Cii m 10 :: !B'c :l

'x()

0.1 0 ~

0.01

0.001

100

C Q)

E '6 10 Q) lJ)

.lB'c :l

'x()

0 ~

0.1

16

14

ro 12

~ 10ci

..s 8

~ "0 .!:: 6

I-a.. 4 ill

2

b

•• A a

W~ 2783

I P<0.001

• _____ .a _______

• I

B a W= 1637 IP<O.OOl •

C W= 1657.5 a

P= 0.001 b

•

•

sediment. We demonstrate here that hydrothermally altered rock alone in a catchment also may result in elevated concentrations of metals in water and sediment (Church et al. 2009, Verplanck et al. 2009, Wanty ct al. 2009). Novel to our work is that risks to aquatic life were quantified and it was determined that metals released naturally from weathering of tile hydrother­mally altered rock limit the EPT (Ephemeroptera, Pleeoptera, Tricoptera) index. Because historical mines

1000 A

b a

••7.' = 50.9100 df = 2 ICii

m P<0.00110::

+c!B

'c :l

'x()

0.1

I­0.01

0

• • 0.001

100 B a

I'E 7.' = 82.9 •Q) df ~ 2E '6 10 P< 0.001 b Q) lJ) I!B

C:::J u'x

'c --+-----•

0 I­

0.1

16 C b c14 a • ro 12

2 x

10ci ..s x CD

8 '1.: ~ 17.23

"0 6 dl = 2.!::

P<0.001 4Ii:

W

2

0

Lithologic Hydrothermally Mined reference altered, unmined

Disturbance

FIG. 4. Effect of hydrothermal alteration and mining on water and sediment toxicity and metal-sensitive aquatic invertebrate communities. Chi-square statistics are the result of a Kruskal-Wallis one-way ANOVA by ranks. Different lowerca~e letters indicate significant difference~ based on Dunn's multiple-comparison te~t (P ::; 0.05). Sec Fig. 3 for boxplot descriptions.

0""----......,-------.-----Unmlned Mined

Disturbance

FIG. 3. Effect of mining on water and sediment toxicity and metal-sensitive aquatic invertebrate communitie~. Different lowercase letler~ indicate significant differences at P ~ 0.05 based on the Wilcoxon rank-sum statistic (W). Toxic units are the summed concentrations of Cd, Cu, and Zn, normalized by procedures described in Me/hods: De/ermination of toxicity units for water and sediment. The EPT (Ephemeroptera, Plecaptera, and Tricoptera) index is the total number of mayfly + stonefly + caddisfly taxa 0 bserved a t a site. Boxplots depict the data distributions where the top and bottom represent the 75th and 25th percentiles and the dividing line i~ the 50th percentile or the median value. The whiskers extend to the 5th percentile (bottom) and 95th percentile (top) of the data distribution. Dots represent outIiers beyond the 5th and 95th percentiles.

878 Ecol<>gical ApplicationsTRA VIS S. SCHMIDT ET AL. Vol. 22, No.3

For cxample, fivc sample locations could not be classified based on mineral-deposit criteria and were not included in comparisons of mineral-deposit type or style of alteration. This inability to classify sites is likely less problematic as compared to our inahility to determine thc total extent of hydrothermal alteration and mineral deposit on thc surface and subsurface of catchments. This latter limitation could explain a substantial amount of the variance in the scdiment and watcr chemistry observed within rock type.

The influcnce of hydrothcrmal alteration, mineral deposits, and mining on aquatic ecosystems in thc United States and the world is significant. Over 200000 abandoned mines are located in the contiguous 48 U.S. states (therc is no count of thc number of abandoned mines globally), one half of which are hard rock mines like those investigated here (Ferderer 1996). Ludington et a!. (1996) estimated that only half of thc extractable mincral commoditics found domestically are exploitcd, while estimating that 50% of the unexploited resources may influcnce surface ecosystems. Mineral deposits similar to those studied hcre can he found all over the world, with a substantial fraction of the U.S. poly­metallic vein and porphyry dcposits found in Arizona, Colorado, Montana, Nevada, and New Mexico (USGS 2009, Verplanck ct al. 2009). Although many of these deposits are unmined, future exploitation of thesc types of mineral deposits could have implications for the health of regional ecosystems (Ludington et al. 1996). Because unmined hydrothermally altered areas are poorly documentcd in the literature, a hetter under­standing of the spatial distribution and ecological cffects of mineral deposits would greatly increase our ability to evaluate ecological risks associated with mining globally.

ACKNOWLEDGMcNTS

Jill Baron and Bob Eppinger provided insightful comments on an earlier version of this manuscript. Funding was providt:d by the USGS and tht: U.S. EPA STAR program (grant number R829640). Travis S. Schmidt was funded by a USDA National Net:ds Fellowship and a USGS Mendenhall Postdoctoral Fellowship. Katy Mitchell, Monique Adams, Michael W. Anthony, James G. Crock, Luke McEachron, Caitlin Balch­Burnell, Sage Bt:lls, Jamie Brennan, Pete Cadmus. Joe Connt:ry, Sigrid Gustafson, Mike Hill, Alicia Louie, Sarah Russell, Dan Scott, David Schnake, David Shaffer, Therese Tepe, Jamie Thayer, and Adam Thompson provided valued project support. Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government.

LlTERI\TURE ClTcD

Canfield, T. 1., N. E. Kemble, W. G. Brumbaugh, F. J. Dwyer, C. G. Ingersoll, and J. F. Fairchild. 1994. Use of benthic invertebrate community structure and the sediment quality triad to evaluate metal-contaminated sediment in the Upper Clark-Fork River, Montana. Environmental Toxicology and Chemistry 13:1999-2012.

Carlisle, D. M., and W. H. Clt:ments. 2005. Leaf liller breakdown, microbial respiration and shreddcr production in metal-polluted streams. Freshwater Biology 50:380-390.

Chronic, J., and H. Chronic. 1972. Prairit: peak and plalt:au. A guidt: to the geology of Colorado. Colorado Geological Surwy Bulletin 32. Colorado Geological Survey, Dcnver, Colorado, USA.

Church, S. E., D. L. Ft:y, T. L. Klt:in, 1'. S. Schmidt, R. B. Wanty, E. H. DeWitt, B. W. Rockwt:ll, and C. A. San Juan. 2009. Environmental effects of hydrotht:rmal alteration and historical mining on water and scdimt:nt quality in ct:ntral Colorado. Pages 85-95 in R. M. T. Webb and D. J. Semmens, editors. Planning for an uncertain future­Monitoring, integration, and adaptation. Proceedings of thc Third Interagency Conlerence on Research in the Watersheds (ICRW), Estes Park, Colorado, USA, 8-11 Scptember 2008. Scientific Investigations Rcport 2009-5049. U.S. Geological Survey, Reston, Virginia, USA. http://pubs.usgs.gov/sir/ 2009/5049/

Clt:mt:nts, W. H., D. M. Carlisle, J. M. Lazorchak, and P. C. Johnson. 2000. Heavy metals structure benthic communities in Colorado mountain streams. Ecological Applications 10:626-638.

Cuffney, T. S., M. D. Bilger, and A. M. Haigler. 2007. Ambiguous taxa: efrt:cts on thc characterization and inter­pretation of invcrtcbratt: assemblages. Journal of tht: North American Bt:nthological Society 26:286 307.

du Bray, E. A. 1996. Prt:liminary compilation of dt:seriptive geot:nvironmcntal mineral deposit models. U.S. Geological Survey. Open-File Report 95-831.

ESRI [Environmental Systems Research Institute]. 2006. ArcGIS 9.2. Environmental Systems Research Institult:, Redlands, California, USA. hllp://csri.com/software/arcgis/ indcx.html

Fcrdert:r, D. A. 1996. National overvit:w of abandont:d minc land sitt:s utilizing tht: minerals availability system (MAS) and geographic information systems (GIS). U.S. Geological Survey. Open-Filt: Report 96-549. http://pubs.usgs.gov/ol; 1996/ofr-96-0549/ofr-96-549.html

Geseh, D. B. 2007. Tht: national elevation datast:l. Pages 99­118 in D. Maune, editor. Digital elt:vation model technolo­gies and applications: the DEM users manual. Second edition. American Society for Photogrammetry and Remote St:nsing, Bcthesda, Maryland, USA.

Ht:y, M. H. J954. A new review of the chlorites. Mineralogical Magazine 30:277-292.

HydroQuaL 2007. Biotic ligand model. Windows interfact:, version 2.2.3. User's guidt: and refcrence manual. Hydro­Qual, Mahwah, New Jersey, USA. http://www.hydroqual. com/blm

Lenat, D. R .. and M. T. Barbour. 1994. Using benthic macroinvertt:brate community structurt: for rapid, cost­effective, water quality monitoring: rapid bioassessment. Pages 187- 215 in S. L. Loeb and A. Spacie, editors. Biological monitoring of aquatic systems. Lewis, Boca Raton, Florida, USA.

Ludington, S. D., D. P. Cox, and R. B. McCammon. 1996. Data base for a national mineral-resource assessment of undiscovered deposits of gold, silver, copper, lead, and zinc in the conterminous United States. U.S. Geological Survey. Open-File Report 96-96. hllp://pubs. usgs.gov/of/I 996/ of96-096/

MacDonald, D. D., C. G. Ingt:rsoll, and T. A. Berger. 2000. Development and evaluation of consensus-based scdiment quality guidelines for freshwater ecosystems. Archives of Environmental Contamination and Toxicology 39:20-3 J.

Merritt, R. W., and K. W. Cummings. 1996. An introduction to the aquatic insects of North Amt:rica. Third edition. Kendall and Hunt Publishing Company, Dubuque, Iowa, USA.

Moulton, S. R., II, J. L. Carter, S. A. Grotheer, T. F. Cuffney, and T. M. Short. 2000. Methods of analysis by the U.S. Geological Survey national water quality laboratory: pro­cessing, taxonomy, and quality control of benthic macroin­