ISOTOPES OF CARBON IN A KARST AQUIFER OF THE CUMBERLAND PLATEAU OF KENTUCKY, USA OGLJIKOVI IZOTOPI V KRAŠKEM VODONOSNIKU CUMBERLANDSKE PLANOTE , KENTUCKY, ZDA Lee J. FLOREA1 Abstract UDC UDK: 546.26.027*14:551.444(736.9) Lee J. Florea: Isotopes of Carbon in a Karst Aquifer of the Cumberland Plateau of Kentucky, USA In this study, the concentration and isotopic composition of dissolved organic carbon (DOC) and dissolved inorganic carbon (DIC) are measured in the karst groundwater of the Otter Creek watershed of the Cumberland Plateau of Kentucky, USA. Comparisons among these data and with the geochemistry of carbonate and gypsum equilibrium reactions reveal that DOC concentration is inversely related to discharge, multiple reaction pathways provide DIC with isotopic enrichment that may be directly related to mineral saturation, and oxidation of reduced sulfur is possible for dissolution. DOC is derived from C3 vegetation with an average 513Cdoc of -27%o. DIC in groundwater is derived from both pedogenic CO2 and HCO-from dissolved carbonate. At input sites to the karst aquifers DIC concentrations are expectedly low, less than 1 mmol/L, in waters that are undersaturated with respect to calcite. At the output of these karst aquifers DIC concentrations reach 3 mmol/L in waters that are at or above calcite saturation. Values of 513CDIC range between -6.3 and -12.4% with CO2 degassing and calcite precipitation at some sites obfuscating a simple relationship between 513CDIC, discharge, and mineral saturation. In addition, concentrations of DIC in sulfur seeps within the watershed range between 2-7 mmol/L with 513CDIC values in some samples skewed more toward the anticipated value of carbonate bedrock than would be expected from reactions with carbonic acid alone. This suggests that the oxidation of reduced sulfur from shallow oilfield brines liberates bedrock DIC through reactions with sulfuric acid. Keywords: dissolved organic carbon, dissolved inorganic carbon, sulfur redox, ion geochemistry, saturation index. Izvleček UDK UDK: 546.26.027*14:551.444(736.9) Lee J. Florea: Ogljikovi izotopi v kraškem vodonosniku Cum-berlandske planote, Kentucky, ZDA V kraški podtalnici zaledja potoka Otter Creek (Cum-berlandska planota, Kentucky, ZDA), smo merili koncentracijo in izotopsko sestavo raztopljenega organskega (DOC) in anorganskega (DIC) ogljika. Podatke smo primerjali z ravnotežno kemijo karbonatov in sadre. Ugotovili smo, da koncentracija DOC pada s pretokom vode, da lahko izotopsko obogatitev DIC vzdolž več reakcijskih poti, neposredno povežemo z nasičenostjo raztopine in da ima pri raztapljanju pomembno vlogo oksidacija reduciranega žvepla. DOC izhaja iz vegetacije tipa C3, s povprečno 513Cdoc -27%. DIC izhaja iz pedogene-ga CO2 in iz raztopljenih karbonatov v obliki HCO- . Dotoki v vodonosnik so nenasičeni na kalcit in imajo izrazito nizk koncentracijo DIC, manj kot 1 mmol/L. Na izvirih so vode v ravnotežju oz. prenasičene na kalcit, koncentracija DIC pa doseže 3 mmol/L. Na mestih, kjer iz vode izhaja CO2 in se izloča kalcit, so izotopske vrednosti DIC med -6,3 in -12,4 %, kar zamegljuje enostavna razmerja med 513CDic, pretokom in stopnjo nasičenja. V žveplenih vodah so koncentracije DIC med 2-7 mmol/L in odstopajo od vrednosti, ki bi jih pričakovali pri raztapljanju karbonatov z ogljikovo kislino, v smeri vrednosti v matični karbonatni kamnini. Očitno se del DIC iz karbonatov sprošča pri reakciji z žvepleno kislino, ki nastaja ob oksidaciji reduciranega žvepla iz plitvih slanic naftnih polj. Ključne besede: raztopjen organski in anorganski ogljik, redukcija žvepla, geokemija, indeks nasičenja. 1 Department of Geological Sciences, Ball State University, e-mail: lflorea@bsu.edu Received/Prejeto: 25.2.2013 INTRODUCTION Carbonate aquifers, hosting 60% of the world's proven petroleum reservoirs, 40% of known gas reserves, and influencing approximately 25% of the world's drinking water (Ford and Williams 2007), are particularly vital to our understanding of the global carbon cycle. A significant fraction of carbon transport through karst is in dissolved form. Dissolved organic carbon (DOC) comprises organic acids and carbohydrates leached out of vegetation and soil. The composition of that DOC depends largely upon the overlying climate and land use. Dissolved inorganic carbon (DIC) includes products of mineral reactions with acidity in water. Of primary importance is the carbonate equilibrium reaction with calcite and dolomite, or hCO^aq) + [(l-x)Ca, xMg]CO3(sol)^ (1-x)Ca(aq) + xMg(aq) + 2HCO3(aq), (1) where the subscripts x and 1-x are proportional to the magnitude of calcite and dolomite within a karst aquifer, respectively. In Equation 1, CO2 sequestered into the aqueous system from the atmosphere, organic oxidation, or microbial respiration in the soil reacts with carbonate bedrock to release carbonate ions into solution. In the pH ranges typical of water in karst aquifers, bicarbonate (HCO--) is the principal dissolved ion. More recently, some emphasis has shifted toward carbonate aquifers with secondary porosity that has partly evolved via the oxidation of reduced sulfur, via the following reaction: HSO4(aq) + 2[(1-x)Ca, xMg]CO3(so[)^ 2(1-x)Ca^ + 2xMg(aq) + SOMaq) + 2HCO3(aq). (2) In such systems, such as the classic example of the karst of the Guadalupe Mountains of New Mexico (Hill 1990), CO2 is not sequestered from the atmosphere and DIC is liberated from the carbonate bedrock alone. STABLE ISOTOPES OF CARBON In nature, carbon occurs as two stable isotopes, 12C and 13C with the abundance of the heavier isotope of approximately 1.1%. Mass-spectrometry can distinguish between these isotopes and compute the enrichment or depletion of the heavier isotope of carbon in a sample as compared the Vienna Pee Dee Belemnite (VPDB), denoted by S13C and calculated using un i i2r iin / vr gl3c = c/ C(s^ple)- C/ C(VPDB) x 1000 (3) 13C / 12C(VPDB) Values of S13C are reported in parts per thousand or 'per mille' (%o - VPDB). Various processes in nature may fractionate the heavier or lighter stable isotope of carbon (Kendall & Caldwell 1998) In vegetation, the photosynthetic process preferentially uptakes the lighter isotope during carbon fixation (Schlesinger 1997; Ehleringer & Cerling 2000). The resulting S13C of organic carbon in vegetation is depleted in the heavier isotope. The magnitude of this fractionation depends upon the nature of the photosynthetic pathway. In humid landscapes, most native plants utilize the C3 pathway, which yields organic matter with S13C values between -23 and -27%. C4-type vegetation, in contrast, is more adapted to arid conditions, fixes less CO2 during photosynthesis than the C3 pathway, and consequently has S13C values between -10 and -14%. In shallow ground water, values of S13C in dissolved organic and inorganic carbon (S13CDOC and S13CDIC) reflect a combination of microbiologic reactions, limestone dissolution, and gas-water exchange processes. When carbon dioxide CO2 produced in soil by oxidation or by biogenic reactions is dissolved into water, the process preferentially selects the heavier isotope. The S13CDIC resulting enrichment is from combined microbial, diffusion, and equilibrium fractionation and may be as much as +6.4% (Clark & Fritz 1997). This pedogenic-derived DIC reacts with carbonate bedrock via Equation 1 to produce a DIC in groundwater from each source, with the stoichiometry in a simple reaction mandating a 50% blend of carbon in the products at mineral saturation. Undersaturated solutions would theoretically have values of S13CDIC that more reflect soil CO2. Oversaturated solutions may be more enriched in bedrock-derived DIC compared to a sample at saturation. The exact nature of S13CDIC in karst groundwater can be significantly more complex than simple mixing and governed by phases of CO2 enrichment or degassing along the flowpath (Marlier & O'Leary 1984), by alternative chemical weathering phenomena like those that release DIC from carbonate bedrock into an aqueous solution without corresponding soil CO2 (e.g., Equation 2), or by weathering of certain silicates that sequesters CO2 from the atmosphere without the addition of bedrock DIC (e.g., weathering of wollastinite, CaSiO3 - Berner et al. 1983). In summary, changes in end-member contributions to DOC and DIC (e.g., vegetative cover and bedrock composition) as well as variations in the hydraulic function of the underlying aquifer system from droughts or storm events may manifest as changes to the values of S13CDIC. This paper considers the spatial and temporal variation of dissolved organic and inorganic carbon from one portion of the karst within the western margin of the Cumberland Plateau in southeast Kentucky as a component of a larger-scale investigation of carbon flux from karst in the Appalachian lowland plateaus. To that end, the geochemistry of water samples from sites considered by Dugan et al. (2012) and Florea (2013a) are used to compute calcite and gypsum saturation indices, calcite saturation ratios, as well as the concentra- tion of the species of inorganic carbon. Complementary to these data are measurements of dissolved organic carbon (DOC) and the stable isotopes of dissolved carbon in both organic and inorganic form (S13Cdoc and S13Cdic) in these same samples. Variations in these data are investigated as they pertain to sampling site and aquifer characteristics, as well as seasonal patterns manifest, in part, as variations in discharge at selected sites. HYDROGEOLOGIC SETTING Data in this study come from water samples collected within the Otter Creek watershed along the western escarpment of the Cumberland Plateau in Wayne County of southeastern Kentucky, USA (Fig. 1). Otter Creek is a third-order tributary of the Cumberland River. Staged incision of the Cumberland River associated with interglacial phases of Plio-Pleistocene glaciation has been the primary control on the geomorphic evolution of the region (Anthony & Granger 2004) and has resulted in tiered cave systems (Simpson & Florea 2009). These caves have developed within middle-Mississippian carbonates of the Slade Formation, including, from oldest to youngest, the St. Louis, Ste. Genevieve, Kidder, and Bangor Limestone members (Ettensohn et al. 1984). These are underlain (J Alpha N Scale: 5 miles CT BCA aeae< 'erC h eek ^ Monticello 'BC1 cAJ T - LH U-' OC3 ) SCS i SSS i HB« DH1 BH O J SC Karst Springs Tufa Springs Sulfur Seeps Ground Water Flow Surface Water Flow Known Oil Reservoir Fig. 1: Study area in Wayne County, KY, USA. Index map of Kentucky illustrates the location of Wayne County. The light gray is the Otter Creek watershed. The Redmond Creek karst aquifer outlined in black includes the sampling sites for Sandy Springs and Stream Cave and are indicated by SS and SC, respectively. Other sampling sites, including karst springs, tufa springs, and sulfur seeps are labled with codes decribed in the text. The axis of the Sunnybrook Anticline is indicated as a black line. Ground-water flow paths are inferred by dye tracing and known cave survey. Oil reservoirs in shallow Mississippian-age strata identified by Abbott (1921) are denoted by gray ovals. by the early-Mississippian calcareous shale of the Salem-Warsaw and Fort Payne Formations and overlain by the late-Mississippian calcareous shale of the Paragon Formation. Regionally, early Pennsylvanian sandstones and conglomerates of the Lee Formation comprise the plateau surface. Relief in the study area exceeds 230 m with ridge tops above 530 m and valley floors below 300 m. The active flow system in the karst aquifers of the Otter Creek watershed comprise dendritic networks of tributary conduits (Palmer 1991) that 'stair step' through the stratigraphy (Crawford 1984) and emerge briefly at the land surface where the flowpath at the base of the Bangor Limestone traverses the Hartselle Formation before sinking again into the Kidder Limestone (Walden et al. 2007). These tributaries coalesce into sinuous, master conduits that may parallel hillside contours (Sasowsky 1994) and discharge at gravity-flow springs at the base of the plateau (Crawford 1984). The karst of the Cumberland Plateau is largely epigene; in other words, the source of acidity is derived from meteoric recharge and driven largely by the carbonate equilibrium reaction presented in Equation 1. Florea (2013b) considers the timing and mode of this recharge using stable isotopes of oxygen and hydrogen (S18O and S2H) and reveals a strong seasonal bias toward winter months when evapotranspiration is reduced. Fig. 2: Time series plots of selected field, ion, and carbon data from Stream Cave (SC - open squares and solid line) and Sandy Springs (SS - open diamonds and dashed line). All y-axes share the dates at the bottom of the figure. In addition to the springs from epigene karst aquifers, preliminary geochemical investigations in the Otter Creek watershed by Dugan et al. (2012) provide a first look at geochemical data of water chemistry from travertine (tufa) springs and sulfur seeps, that are in part spatially controlled by the Sunnybrook Anticline (Fig. 1). This anticline has an amplitude of approximately 30 m, is oriented N-NE, and is parallel to the trend of the Cumberland Escarpment. Waters at the tufa springs can be oversaturated with respect to calcite and where they emerge, calcite precipitates. These springs probably resurge from long, strike-parallel flowpaths on the west flank of the anticline. Similarly, some caves in the region contain significant travertine deposits within active water passages, including rimstone dams and flowstone, suggesting periods of calcite oversaturation linked with chemical changes that lead to mineral precipitation. Florea (2013a) investigates the nature of water chemistry in the Otter Creek watershed, in particular from the Redmond Creek karst aquifer (Fig. 1), and using a comparison of reaction products and principal component analysis concludes that, although dissolution via Equation 1 dominates the chemistry, dissolution via Equation 2, with the sulfur derived from the entrain-ment of shallow brines, is possible at the local scale. This process is particularly important adjacent to sulfur seeps in caves and streams. These documented sulfur seeps are largely concentrated on the east flank of the Sunnybrook Anticline in the direction of the Appalachian Basin (Fig. 1). The presence of these seeps is a manifestation of shallow petroleum reservoirs in lower Mississippian strata that underlie the carbonates that host the karst aquifers (Fig. 2). METHODS SAMPLES AND ANALYTICAL METHODS Samples comprising this study include 16 bi-monthly samples collected in 2010-2011 from two sites in the Redmond Creek karst aquifer, Stream Cave (SC) and Sandy Springs (SS) considered by Florea (2013a and b). Measurements of instantaneous discharge (Q) computed using an acoustic flow meter and standard USGS gauging techniques (e.g., Rantz 1982) accompany these samples. At Stream Cave, water emerges from a hillside spring in the Bangor Limestone, traverses the Hartselle Formation, and sinks within 100 m into the underlying Kid-der Limestone. This site represents one of many inputs to the Redmond Creek karst aquifer. Sandy Springs, in contrast, is the principal outlet for the Redmond Creek karst aquifer. Complementing these data are samples collected in 2011 from three sulfur seeps: Slickford Bridge (SB), Bertram Hollow (BH), and Beaver Creek #1 (BC1). In January of 2012, additional samples from 13 springs in the Otter Creek watershed complete the dataset. These samples come from SS, SB, Otter Creek #3 (OC3), Don Carter Seep (DC), Blowing Cave (BCA), Tonya's Cave (TC), Coal Trace (CT), Herlan Buck Springs (HB), Triple S Cave (SSS), Spelunger Cave Springs (SCS), Dry Hollow #1 (DH1), Dry Hollow #2 (DH2), and Light House Spring (LH). Three of these sites (SB, OC3, and DC) represent sulfur seeps. Nine additional sites (SS, BCA, TC, CT, HB, SSS, SCS, DH1, DH2) are classic, gravity-flow karst springs. The remaining site (LH) is a large tufa spring. With the exception of SC, all springs in this study arise from at or near the base of the St. Louis limestone, the regional base of the primary carbonate sequence in the Carboniferous. Data from the samples are summarized in Tab. 1. These data comprise ionic measurements using HACH field titrations for bicarbonate (HCO-) as well as Ca2+ and Mg2+ concentrations in the 2012 samples, ion chromatography for 2010-2011 samples, and a HACH DR 2800 spectrophotometer for select ions in the 2012 samples. Field measurements of pH and temperature (T) complement these data. Samples for ion analysis were filtered using a 0.45-^m membrane and collected using 250 mL HDPE bottles and kept at 4°C until time of analysis. Samples for cations were preserved using 2 mL of 6N HNO3. Precision and accuracy were not reported from the lab contracted for the ion analyses. Computed charge balance values are reported in Tab. 1 for samples where appropriate. For each filtered sample, a split was stored in a 30mL glass bottle, treated with CuSO4 as an anti-micro-bial agent, and stored at 4°C with a parafilm seal until analysis for S13Cdic. A similar split sample was analyzed for S18O and S2H with results summarized in Florea (2013b). Concurrent with each of the 16 sets of 20102011 data from SC and SS, samples for measurement of DOC and S13CDOC were collected in a 1L HDPE bottle spiked with 1 mL of 12 N HCl to prevent microbial activity. In the lab, these samples were dehydrated and the remaining solids treated with H2SO3 to effervesce CO2 Tab. 1 . Selected field, ion, and. carbon data from sample collection. Cations Anions Carbon Site Sample Name Date Temp pH Chg Bal Ca2* Mg2* K* Na* hco3- Fl- Cl- no3- SO42- DOC s°CDOC sljcDlC (•C) mmol/L mmol/L mmol/L mmol/L mmol/L mmol/L mmol/L mmol/L mmol/L % mg/L - VPDB - VPDB SS NL070710A 07/07/10 16.33 7.95 1.120 0.225 0.026 0.155 1.61 0.008 0.137 0.053 0.088 18 -26.5 -7.9 SS NL072010A 07/20/10 12.54 7.08 0.948 0.196 0.024 0.112 1.51 0.008 0.090 0.038 0.079 14 1.6 -27.8 -12.4 SS NL080310A 08/03/10 14.60 7.82 0.987 0.197 0.024 0.178 1.87 0.008 0.139 0.040 0.099 6 1.7 -26.5 -7.5 SS NL081710A 08/17/10 13.80 7.69 1.259 0.219 0.026 0.092 2.39 0.008 0.087 0.075 0.067 6 1.0 -26.9 -8.0 SS NL083110A 08/31//10 14.55 7.38 1.057 0.195 0.023 0.108 0.007 0.081 0.050 0.063 0.8 -26.5 -8.1 SS NL091410A 09/14/10 14.80 7.46 1.154 0.219 0.024 0.148 2.09 0.007 0.125 0.054 0.094 8 0.6 -27.0 -8.8 SS NL092810A 09/28/10 15.21 7.95 0.413 0.204 0.002 0.206 2.05 0.003 0.205 0.064 0.086 27 0.7 -26.0 -9.2 SS NL101210A 10/12/10 14.34 7.91 1.290 0.230 0.024 0.260 2.61 0.004 0.245 0.070 0.098 3 0.5 -27.8 -7.3 SS NL102610A 10/26/10 14.24 7.93 1.300 0.239 0.025 0.260 2.52 0.004 0.261 0.076 0.124 4 0.5 -26.5 -8.5 SS NL110910A 11/09/10 13.44 7.52 1.053 0.197 0.024 0.163 0.003 0.154 0.059 0.086 1.1 -29.2 -8.6 SS NL112310A 11/23/10 12.18 7.75 0.914 0.168 0.025 0.083 1.83 0.004 0.080 0.042 0.079 3 1.7 -26.5 -9.9 SS NL120710A 12/07/10 0.697 0.136 0.020 0.188 1.61 0.012 0.170 0.032 0.099 4 0.5 -27.1 -8.6 SS NL122110A 12/21/10 10.50 8.34 0.582 0.113 0.017 0.208 0.99 0.002 0.199 0.041 0.117 5 0.7 -27.8 -9.0 SS NL010411A 01/04/11 10.22 7.47 0.523 0.104 0.016 0.179 1.77 0.001 0.172 0.030 0.107 20 1.1 -26.9 -9.7 SS NL011811A 01/18/11 10.58 7.73 0.651 0.151 0.016 0.113 1.39 0.002 0.099 0.022 0.093 1 1.6 -26.2 -8.4 SS NL020811A 02/08/11 9.64 0.557 0.135 0.015 0.130 1.09 0.003 0.112 0.018 0.099 4 0.5 -27.4 -9.3 SC NL070710B 07/07/10 14.27 6.40 0.226 0.102 0.015 0.031 0.30 0.007 0.032 0.020 0.039 23 -7.0 SC NL072010B 07/20/10 18.07 6.93 0.097 0.035 0.014 0.021 0.19 0.007 0.028 0.012 0.033 1 2.1 -8.9 SC NL080310B 08/03/10 17.06 7.47 0.167 0.049 0.015 0.024 0.007 0.029 0.015 0.036 1.3 -8.4 SC NL081710B 08/17/10 16.12 7.07 0.225 0.060 0.017 0.031 0.47 0.007 0.031 0.019 0.035 2 0.2 -8.2 SC NL083110B 08/31/10 16.85 7.31 0.193 0.060 0.014 0.032 0.006 0.034 0.017 0.042 0.2 -27.1 -7.6 SC NL091410B 09/14/10 15.95 7.65 0.147 0.039 0.014 0.017 0.37 0.006 0.041 0.027 0.038 13 1.1 -26.4 -7.3 SC NL092810B 09/28/10 15.50 8.03 0.177 0.049 0.013 0.033 0.38 0.002 0.031 0.011 0.035 1 0.9 -7.5 SC NL101210B 10/12/10 14.14 7.69 0.200 0.049 0.013 0.031 0.45 0.003 0.032 0.016 0.038 3 0.5 -6.3 SC NL102610B 10/26/10 13.53 7.49 0.119 0.033 0.027 0.021 0.31 0.002 0.034 0.024 0.042 13 2.5 -7.7 SC NL110910B 11/09/10 11.89 7.11 0.127 0.032 0.016 0.026 0.002 0.025 0.006 0.039 1.1 -26.5 -6.7 SC NL112310B 11/23/10 11.14 7.47 0.064 0.035 0.024 0.023 0.27 0.002 0.022 0.004 0.039 21 3.0 -28.1 -9.4 SC NL120710B 12/07/10 0.071 0.042 0.012 0.023 0.21 0.003 0.020 0.004 0.047 12 0.7 -7.7 SC NL122110B 12/21/10 5.05 7.08 0.073 0.042 0.011 0.021 0.13 0.001 0.021 0.006 0.056 2 0.7 -8.2 SC NL010411B 01/04/11 4.97 6.97 0.069 0.045 0.011 0.025 0.81 0.001 0.017 0.002 0.054 56 0.9 -8.0 SC NL011811B 01/18/11 3.19 7.19 0.072 0.045 0.010 0.024 0.33 0.001 0.020 0.001 0.051 25 0.7 -26.6 -8.1 SC NL020811B 02/08/11 3.84 7.34 0.071 0.047 0.011 0.023 0.33 0.002 0.017 0.011 0.052 26 0.7 -8.0 SB LF051311B 05/13/11 13.84 6.94 6.849 0.706 0.034 1.036 2.70 0.017 0.988 0.003 3.518 21 -5.7 BH LF051511D 05/15/11 13.09 7.13 3.602 0.486 0.029 0.798 2.10 0.016 0.348 0.015 2.246 13 -4.5 BC1 LF051511A 05/15/11 15.20 6.59 6.084 1.536 0.113 4.794 4.11 0.021 1.893 0.000 4.302 16 -5.6 BCA LF012012B 01/20/12 12.50 8.26 0.970 0.148 2.17 0.052 1 -11.7 TC LF012012C 01/20/12 12.20 7.97 0.798 0.086 1.84 0.135 9 -13.1 CT LF012012D 01/20/12 12.50 7.64 1.180 0.156 1.66 0.687 6 -11.4 SS LF012012E 01/20/12 10.80 7.93 0.571 0.060 1.22 0.125 8 -8.7 HB LF012012F 01/20/12 11.70 7.66 1.170 0.092 2.52 0.073 3 -12.9 SSS LF012012G 01/20/12 11.30 7.80 0.535 0.116 1.74 0.062 18 -11.2 SCS LF012012H 01/20/12 12.20 7.83 0.621 0.060 1.34 0.062 4 -11.3 DH1 LF012012I 01/20/12 10.90 7.93 0.551 0.052 1.19 0.187 13 -10.6 DH2 LF012012J 01/20/12 10.80 7.90 0.607 0.056 1.04 0.198 4 -10.6 SB LF012012K 01/20/12 13.80 7.16 6.679 0.240 3.27 6.870 10 -10.7 OC3 LF012012L 01/20/12 13.40 7.48 8.176 0.336 4.00 7.911 8 -8.9 DC LF012012M 01/20/12 6.90 7.31 7.257 1.219 3.92 8.119 9 -9.3 LH LF0120120 01/20/12 13.80 8.22 1.521 0.140 2.98 0.094 2 -12.2 from the fraction of inorganic carbon in the solid residue. Secondary dehydration provided organic resin for isotopic analysis. Measurements of the stable isotopes of carbon were conducted at the Stable Isotope Laboratory at the University of South Florida. For S13Cdic, 12 mL borosilicate glass vials with septa were filled with 1 mL of 85% phosphoric acid. After a flush of helium gas, 0.2 mL of sample was introduced into the vial with a syringe. The CO2 produced in the reaction between the acid and the sample DIC was analyzed using a Thermo-Finnigan IRMS attached to a Gas Bench II. For S13CdoC, weighed masses of organic resin were loaded into tin capsules and combusted in an Elemental Analyzer. The CO2 produced in the combustion was analyzed in the IRMS. Isotopic values were compared to the VPDB standard using simultaneous measurements from reference samples of Carrara marble and NBS 18 standards for S13CdiC and Fergie-CN for S13CdoC. Measurements of S13CdiC and S13CdoC are accurate to within ±0.1%o. ANALYSIS OF ANALYTICAL RESULTS The saturation index with respect to calcite and gypsum and the calcite saturation ratio for each sample were computed using the Debye Huckle relationship to compute molar activities of ionic species as summarized by Ford and Williams (2007) and White (1988) for the carbonate equilibrium reactions. Additionally for each sample, the partial pressure of CO2 in solution, PCO2, was computed using the temperature-dependent equilibrium equations for pKCO2, pKb pK2, and pKc summarized in Ford and Williams (2007, p. 48) and derived from Garrels and Christ (1965) and Plummer and Blusenberg (1982). Using Henry' Law, the equivalent molar concentration of CO2 in solution was computed from PCO2. The sum of the molar concentrations of CO2 and HCO31- in solution is a measure of the total DIC in solution. Summaries of computed geochemical values are provided in Tab. 2. Tab. 2. Summary of analytical data. Site Sample Name Date SIcai C/Cs SIgfP PCO2 CO2 DIC atm mmoi/L mmoi/L SS NL070710A 07/07/10 0.06 0.99 -2.65 9.86E-04 4.72E-05 1.653 SS NL072010A 07/20/10 -0.90 0.45 -2.75 6.59E-03 3.18E-04 1.828 SS NL080310A 08/03/10 -0.06 0.90 -2.65 1.52E-03 6.89E-05 1.939 SS NL081710A 08/17/10 0.01 0.95 -2.73 2.57E-03 1.20E-04 2.510 SS NL083110A 08/31//10 SS NL091410A 09/14/10 -0.31 0.73 --2.61 3.88E-03 1.75E-04 2.265 SS NL092810A 09/28/10 SS NL101210A 10/12/10 0.27 1.18 -2.57 1.70E-03 7.79E-05 2.688 SS NL102610A 10/26/10 0.28 1.19 -2.46 1.56E-03 7.19E-05 2.592 SS NL110910A 11/09/10 SS NL112310A 11/23/10 -0.17 0.83 -2.77 1.70E-03 8.30E-05 1.913 SS NL120710A 12/07/10 SS NL122110A 12/21/10 -0.02 0.93 -2.77 2.34E-04 1.20E-05 1.002 SS NL010411A 01/04/11 -0.68 0.55 -2.86 3.09E-03 1.60E-04 1.930 SS NL011811A 01/18/11 -0.44 0.67 -2.82 1.34E-03 6.84E-05 1.458 SS NL020811A 02/08/11 SC NL070710B 07/07/10 -2.85 0.10 -3.59 6.57E-03 3.02E-04 0.602 SC NL072010B 07/20/10 -2.86 0.09 -3.99 1.30E-03 5.35E-05 0.243 SC NL080310B 08/03/10 SC NL081710B 08/17/10 -1.99 0.19 -3.64 2.25E-03 9.80E-05 0.568 SC NL083110B 08/31/10 SC NL091410B 09/14/10 -1.68 0.24 -3.77 4.67E-04 2.04E-05 0.390 SC NL092810B 09/28/10 -1.22 0.36 -3.73 1.99E-04 8.80E-06 0.389 SC NL101210B 10/12/10 -1.43 0.30 -3.65 5.06E-04 2.33E-05 0.473 SC NL102610B 10/26/10 -2.01 0.19 -3.81 5.51E-04 2.58E-05 0.336 SC NL110910B 11/09/10 SC NL112310B 11/23/10 -2.35 0.14 -4.11 4.91E-04 2.47E-05 0.295 SC NL120710B 12/07/10 SC NL122110B 12/21/10 -3.00 0.09 -3.88 5.43E-04 3.29E-05 0.163 SC NL010411B 01/04/11 -2.36 0.14 -3.95 4.33E-03 2.63E-04 1.073 SC NL011811B 01/18/11 -2.50 0.13 -3.93 1.05E-03 6.73E-05 0.397 SC NL020811B 02/08/11 -2.36 0.14 -3.94 7.47E-04 4.70E-05 0.377 SB LF051311B 05/13/11 -0.11 0.87 -0.53 1.51E-02 7.03E-04 3.403 BH LF051511D 05/15/11 -0.25 0.78 -0.91 7.76E-03 3.69E-04 2.469 BC1 LF051511A 05/1511 -0.31 0.74 -0.54 5.17E-02 2.31E-03 6.420 BCA LF0120312B 01/20/12 0.44 1.35 -2.99 6.25E-04 3.02E-05 2.200 TC LF0120312C 01/20/12 0.01 0.95 -2.59 1.03E-03 5.04E-05 1.890 CT LF0120312D 01/20/12 -0.23 0.79 -1.76 1.97E-03 9.53E-05 1.755 SS LF0120312E 01/20/12 -0.34 0.72 -2.74 7.46E-04 3.79E-05 1.258 HB LF0120312F 01/20/12 -0.02 0.94 -2.72 2.85E-03 1.41E-04 2.661 SSS LF0120312G 01/20/12 -0.35 0.71 -3.07 1.44E-03 7.22E-05 1.812 SCS LF0120312H 01/20/12 -0.37 0.71 -3.00 1.05E-03 5.11E-05 1.391 DH1 LF0120312I 01/20/12 -0.37 0.71 -2.58 7.28E-04 3.69E-05 1.227 DH2 LF0120312J 01/20/12 -0.41 0.68 -2.51 6.80E-04 3.46E-05 1.075 SB LF0120312K 01/20/12 0.16 1.08 -0.29 1.09E-02 5.08E-04 3.778 OC3 LF0120312L 01/20/12 0.63 1.59 -0.18 6.28E-03 2.96E-04 4.296 DC LF0120312M 01/20/12 0.40 1.33 -0.22 8.48E-03 4.85E-04 4.405 LH LF01203120 01/20/12 0.71 1.68 -2.52 9.43E-04 4.39E-05 3.024 VARIATION OF CARBON IN KARST AQUIFERS Most generally, the results of this study reveal the blend of dissolved carbon that is transmitted though the largely epigene karst of the Cumberland Plateau. These data further reveal information on the source and timing of carbon transport. As a baseline for further research, the data characterize the magnitude of carbon flux in present conditions. Since rates of chemical weathering are sensitive to environmental and climate conditions, reference studies as this provide one important regional context to help understand the potential impacts of changes in land use (Zhang 2011) and atmospheric CO2 (Cao et al. 2012) upon temperature, precipitation, plant community structure, and resultant DOC and DIC in karst aquifers. DOC AND 513CDOC Values of DOC are within the expected range of karst waters as observed by Simon et al. (2007). Values at the aquifer input (SC) are consistently higher (average and standard deviation = 1.11±0.82 mg/L) than at the aquifer output (SS) (average and standard deviation = 0.97±0.47 mg/L). This suggests a conversion of DOC into DIC between aquifer input and output that is at least in part due to microbial activity. DOC values at SC are particularly sensitive to discharge (Fig. 3). During August 2010 when dry conditions reduced the base flow at SC to less than 0.8 L/s, DOC levels were close to zero. In contrast, higher discharge conditions Discharge (L/s) Fig. 3: Analytical measurements of A) $13Cdic, B) S13Cdoc, and C) DOC compared against measured value of discharge with Stream Cave (SC) shown as open squares and Sandy Springs (SS) shown as open diamonds. A linear regression is shown for values of DOC at SC in panel C. Matched logarithmic regressions for values of S13CDIC are shown for SC and SS in panel A. Fig. 4: Calculated values of saturation index for calcite (x-axis) and gypsum (y-axis) at Stream Cave (SC - open squares), Sandy Springs (SS - open diamonds), and sulfur seeps (open circles) from 2010-2011. Also shown are DIC measurements from 2012 samples including those from karst springs (filled diamonds), the tufa spring (filled triangle), and the sulfur seeps (filled circles). Undersaturated and over-saturated fields are identified. in October and November 2010, combined with organic loading from falling deciduous leaves, produced DOC values up to 3 mg/L. No similar relationship between DOC and Q is visible at SS; however, similar timing of high and low values of DOC is clear from the time series data (Fig. 3). Despite scatter, the contribution of C3-type vegetation is clearly demonstrated by the values of 513CDOC observed at both SC and SS (average and standard deviation = -27.0±0.8%o). No relationship between 513CDOC and Q is visible in Fig. 3, suggesting that the source of DOC does not shift either seasonally from agriculture or according to availability of water. This conforms to existing land use in the watersheds for both SC and SS, which are minimally impacted by human activities, being primarily woodland with a history of selected timber harvesting dating back to the early 1900s (Kay Koger, landowner - personal communication). The estimated population density for Redmond Creek is ~ 0.01 ha-1 and is a similar density for the larger Otter Creek watershed. S13CDIC AND DIC CONCENTRATION VERSUS DISCHARGE S13CDIC values from the Otter Creek watershed generally coincide with carbonic acid dissolution of limestone. The concentration of this DIC generally decreases with increasing Q. Data from SC and SS in 2010-2011 weakly suggest that S13CDIC at the aquifer output is inversely proportional with discharge (Fig. 3) indicating that under-saturated waters during higher discharge conditions have excess soil-derived PCO2 and contribute more depleted S13CDIC values than during base flow conditions. Con- versely, base flow conditions are more enriched in 13C, with weak logarithmic regressions for SC and SS that approach -6.5% at a Q = 1 L/s (Fig. 3). S13CDIC values are somewhat dependent upon the DIC concentration at SC and SS, but not among the other sites in this dataset, including tufa springs and sulfur seeps (Fig. 5). DIC in the form of HCO- dominates the water chemistry (Tab. 1 and 2). The concentration of DIC among all samples increases with calcite saturation and, on average, exceeds 2 mmol/L in saturated waters (Fig. 5). Along the flowpath between SC and SS, DIC in CO2 form is converted to HCO3- via the reaction in Equation 1. Careful inspection of data in Tab. 2 demonstrates greater CO2 concentration at SS when compared to SC (average and standard deviations = 1.1x10-4±0.85x10-4 and 0.8x10-4±0.98x10-4, respectively); however, at SC the CO2 comprises a greater fraction of the DIC in the sample (Fig. 5). s13cdic versus calcite saturation Walden (1999), during an investigation of groundwa-ter from the Redmond Creek karst aquifer, used ratios of 87Sr/86Sr to differentiate between conduit and diffuse flow. Her results expectedly found a trend toward values similar to the host limestone along the aquifer flow path. Following this logic, values of SI and C/Cs should in theory approach, or even exceed, saturation along that same flowpath. Data from Tab. 2 demonstrate that this is likely the case with values of SI at SC (average and standard deviation = -2.62±0.59), an input to the Redmond Creek aquifer, consistently lower than at SS (average and standard deviation = -0.56±0.39), the primary output for the Redmond Creek aquifer. -3.6 -4.6 -5.6 -6.6 -7.6 -8.6 -9.6 -10.6 -11.6 -12.6 -13.6 -14.6 A 1 1.5 2 DIC (mmol/L) B o*' * * -3.6 -4.6 ) -5.6 )B -6.6 PD -7.6 > -8.6 i -9.6 ( -10.6 -11.6 -12.6 -13.6 -14.6 C 0 1 2 3 4 5 DIC (mmol/L) y = 0.891n(x) - 6.3 o R- = 0.24 o o -4.00 -3.50 -3.00 -2.50 -2.00 -1.50 -1.00 -0.50 0.00 0.50 1.00 SI calcite „ F y = 5.30x - 5.4 R2= 0.19 y = 16.5x - 5.8 R2= 0.99 y = 4.11x - 7.9 R2= 0.64 t -3.60 -4.60 -5.60 -6.60 -7.60 -8.60 -9.60 -10.60 -11.60 -12 .60 -13.60 -14.60 -5 -4.5 -4 -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5 1" SI gypsum Correspondingly, values of S13CDIC should increasingly reflect DIC derived from the bedrock. With an approximated shift of +6.4%o from CO2 in the soil to CO2 in solution (Clark and Fritz 1997), water with C/Cs = 0 that is in equilibrium with soil CO2 should have a S13CDIC of -20.6%. Theoretically, karst water with a SI = 0 or C/Cs = 1 would have a 50% blend of DIC from the soil and from the bedrock following Equation 1. Without prior knowledge of the bedrock end member, the value of S13CDIC expected at saturation is unknown. However, the data from both SC and SS appear to converge at saturation to a S13CDIC value of -6.3% and -8.1% for SC and SS, respectively (Fig. 5), which is similar to the values observed at base flow conditions in Section 5.3 (Fig. 3). Assuming that Equation 1 is the only reaction pathway for producing DIC, these results would imply a bedrock source of +8.4% for SC and +5.8% for SS, considerably more enriched in 13C than typical values for marine carbonates that are typically close to that of the VPBD standard (e.g., Hoefs 1997). Since S13C of carbonates may vary by several per mille, a comparative investigation of the S13C signal of bedrock is currently underway to quantify the additional enrichment, if present. Fig. 5: Analytical measurements of dissolved inorganic carbon (DIC) at Stream Cave (SC - open squares), Sandy Springs (SS - open diamonds), and sulfur seeps (open circles) from 2010-2011. Also shown are DIC measurements from 2012 samples including those from karst springs (filled diamonds), the tufa spring (filled triangle), and the sulfur seeps (filled circles). From top to bottom: A) Values of S13Cdic at SC and SS are compared against DIC concentration in the form of CO2 (open symbols) and HCO3- (gray-filled symbols). B) Values of S13CDIC are compared against total DIC concentration. C) Values of S13CDIC are compared against C/Cs with logarithmic regressions shown for data from SC and SS. D) DIC concentration in the form of HCO3- are compared against the calcite saturation ratio (C/Cs). E) Values of S13CDIC are compared against the saturation index with respect to calcite; linear regressions are shown for the data from SC and SS. F) Values of S13CDIC are compared against the saturation index with respect to gypsum; linear regressions are shown for the data from SC and SS. 0 3 6 7 One possible source of additional S13CDIC enrichment is evident in the data from the sulfur seeps. These sites are closer to saturation with respect to gypsum (Fig. 4). In 2011 these sites, despite high values of Ca2+ and HCO- in solution (Tab. 1), are undersaturated with respect to calcite (Tab. 2). They are correspondingly enriched with respect to 13C (Fig. 5). This implies additional enrichment in bedrock carbon without a corresponding additional contribution from PCO2 from the soil. Such a process may be driven by the oxidation of dissolved sulfides, which can decrease the pH and drive additional carbonate dissolution via Equation 2. Interestingly, the 2012 data from the sulfur seeps reveal oversaturation with respect to calcite and decreased enrichment of 13C. These samples were collected in higher discharge conditions during the winter months. They follow trends similar to SC and SS and converge on a value S13CDIC of -11% at SI = 0 and C/Cs = 1 (Fig. 5). Collectively these data suggest the potential for lessened oxidation of reduced sulfur at the time of the 2012 samples. The particular reason why the 2012 sulfur seep samples are more depleted in 13C compared to the samples from SS, which are in turn more depleted in 13C than the samples from SC is unclear. Logic suggests that the trend should be the opposite; the potential influence of Equation 2 upon S13CDIC should increase at sites with known influence of sulfur-rich brines, which is certainly the case the sulfur seeps and observed to a lesser degree at sites within the cave that contributes to SS. One possible mechanism that could generate the trends in Fig. 5 is that the average bedrock S13CDIC may vary by several per mille within the contributing watershed for each site. A second possibility is the potential for the oxidation of pyrite within the bedrock within the watershed of SC, known to be present in significant quantities in the Bangor Limestone and the organic-rich shale and coal that overlie the Bangor. A third possibility is the potential for methanogenesis within the deeper flow systems that contribute some percentage of flow to SS and the sulfur seeps where reduction of organics may lead to depleted values of S13CDIC. Finally, the samples from SC, which represent waters that have limited interaction with bedrock, may simply not be in equilibrium with soil CO2. Each of the postulates remains untested at this time. Of the remaining samples during 2012 from all other sites, the values of S13Cdic are equal to, or more depleted than those observed at SS in 2010-2011 (Fig. 5). Five of those sites, CT, SSS, SCS, DH1, DH2, are clustered and similar in range to the data from SS. These five springs have similar physical characteristics (strati-graphic position, discharge, etc.) to SS. Four other sites, BCA, TC, HB, and LH, are depleted in 13C compared to the other samples at the same calcite saturation. These springs may collect waters from aquifers that have somewhat different bedrock isotope composition than SS or each other. The sample from LH is significantly oversatu-rated with respect to calcite and is an active tufa spring. Fractionation of the heavier isotope during CaCO3 precipitation combined with degassing of CO2 enriched in the lighter isotope potentially drives this sample toward depletion. The remaining three depleted sites may also experience similar fractionation from degassing and cal-cite precipitation to varying degrees and underscores the complexity of interpreting a limited number of S13CDIC values from a site. CONCLUSIONS In the Otter Creek watershed of the Cumberland Plateau of Kentucky, USA, the concentration of dissolved organic carbon is less than 3 mg/L, and in the case of some sites is inversely proportional to discharge. C3 vegetation is the source of this DOC, with an average 513CDOC of -27%. DIC concentrations that may exceed 4 mmol/L are inversely proportional to discharge and directly proportional to the saturation of the aqueous solution with respect to calcite and gypsum at some sites. Values of S13CDIC reflect soil and bedrock sources and conform to measurements expected from the carbonate equilibrium reactions. Increasing enrichment of 13C at two sites in this study proceeds with increased saturation of groundwa-ter with respect to calcite. Sulfur seeps follow this same trend at higher base flow conditions. These same sulfur seeps are enriched in 13C at low flow, but undersaturated, thus suggesting that the oxidation of reduced sulfur may enrich these sties in bedrock DIC. Other locations appear to follow independent DIC pathways. Some sites, such as tufa springs may have depleted S13CDIC from CO2 degassing and calcite precipitation. ACKNOWLEDGEMENTS The author acknowledges the major contributions by Bill Walden, Chasity Stinson, and Nick Lawhon during sample collection as well as the conversations, hospitality, and access provided by landowners Tim Pyles and Kay Koger and cavers Rick Gordon, Deb Moore, Harry Gopel, and Eric Weaver. Finally, the author appreciates the collaborations with Jonathan Wynn and Bogdan Onac at the USF Stable Isotope Laboratory. Funding for this work was provided by a WKU Provost Incentive grant, a WKU start-up index, and a Ball State University start-up package. Welcome comments from two anonymous reviewers greatly improved the flow and content of this manuscript. REFERENCES Abbott, W.R., 1921: Oil Field Map-Wayne and Clinton Counties, Kentucky. Anthony, D.M. & D.E. Granger, 2004: A Late Tertiary origin for multilevel caves along the western escarpment of the Cumberland Plateau, Tennessee and Kentucky, established by cosmogenic 26Al and 10Be.-Journal of Cave and Karst Studies, 66, 2, 46-55. Berner, R.A., Lasaga, A.C. & R.M. Garrels, 1983: The carbonate-silicate cycle and its effect on atmospheric carbon dioxide over the past 100 million years.-American Journal of Science, 284, 641-683. Clark, I.D. & P. Fritz, 1997: Environmental Isotopes in Hy-drogeology.- Lewis Publishers, Boca Raton, 328 pp. Crawford, N.C., 1984: Karst landform development along the Cumberland Plateau Escarpment of Tennessee. In: RG LeFleur (ed.), Groundwater as a geomorphic agent. Boston, Allen and Unwin, Inc., pp. 294-338. Dugan, C.R., Florea, L.J. & W.D. Walden, 2012: A Geo-chemical Investigation of Springs within the Otter Creek watershed: Wayne County, Southeastern Kentucky.- Geological Society of America Abstracts with Programs, 44, 4, 26. Ehleringer, J.R., Buchmann, N. & L.B. Flanagan, 2000: Carbon isotope ratios in belowground carbon cycle processes.- Ecological Applications, 10, 2, 412-422. Ettensohn, F.R., Rice C.R., Dever, G.R. Jr. & D.R. Ches-nut, 1984: Slade and Paragon Formations; New stratigraphic nomenclature for Mississippian rocks along the Cumberland Escarpment in Kentucky.-U.S. Geological Survey Bulletin 1605-B, 37 pp. Florea, L.J.l., 2013: Selective recharge and isotopic composition of shallow groundwater within temperate, epigenic carbonate aquifers. journal of Hydrology 489: 201-213. doi 10.1016/j.jhydrol.2013.03.008 Florea, L.J., 2013a: Investigations into the Potential for Hypogene Speleogenesis in the Cumberland Plateau of Southeast Kentucky, U.S.A. In, Proceedings of the 16th International Congress of Speleology, Brno, Czeck Republic, July 2013, pp. 356-361. Ford, D.C. & P. Williams, 2007: Karst Hydrogeology and Geomorphology.- John Wiley & Sons, 562 pp. Garrels, R.M. & C.M. Christ, 1965: Solutions, Minerals, and Equillibria. Harpers' Geoscience Series. Harper and Row, New York. 450 pp. Hill, C.A., 1990: Sulfuric acid speleogenesis of Carlsbad Cavern and its relationship to hydrocarbons, Delaware Basin, New Mexico and Texas.- American Association of Petroleum Geologists Bulletin, 74, 11, 1685-1694. Hoefs, J., 1997: Stable isotope geochemistry.; 4th Ed., Springer, New York, 200 pp. Cao, J.H., Yuan, D.X., Groves, C., Huang, F., Huang H. & Q. Lu, 2012: Carbon fluxes and sinks: The consumption of atmospheric and soil CO2 by carbonate rock dissolution.- Acta Geologica Sinica, 86, 963-972. Kendall, C. & E.A. Caldwell, 1998: Fundamentals of Isotope Geochemistry. In: Kendall, C. & J. J. McDonnell, (eds.), Isotope Tracers in Catchment Hydrology Elsevier Science B.V., Amsterdam, pp. 51-86. Marlier, J.F. & M.H. O'Leary, 1984: Carbon kinetic isotope effects on the hydration of carbon dioxide and the dehydration of bicarbonate ion.- Journal of the American Chemical Society, 106, 5054-5057. Palmer, A.N., 1991: Origin and morphology of limestone caves.- Geological Society of America Bulletin, 103, 1-21. Plummer, L. & N.E. Busenberg, 1982: The Solubilities of calcite, aragonite and vaterite in CO2 - H2O solutions between 0 and 90°C, and an evaluation of the aqueous model for the system CaCO3 - CO2 - H2O.-Geochimica et Geochimica Acta, 46, 1011- 1040. Rantz, S.E., 1982: Measurement and computation of streamflow, Volume 1, Measurement of stage and discharge.- United States Geological Survey Water Supply Paper 2175, 284 p. Sasowsky, I.D. & W.B. White, 1994: The role of stress release fracturing in the development of cavernous porosity in carbonate aquifers.- Water Resources Research, 30, 12, 3523-3530. Schlesinger, W.H., 1997: Biogeochemistry: An Analysis of Global Change.- Academic Press, San Diego, 443 pp. Simon, K.S., Pipan, T. & D.C. Culver, 2007: A conceptual model of the flow and distribution of organic carbon in caves.- Journal of Cave and Karst Studies, 69, 2, 279-284. Simpson, L.C. & L.J. Florea, 2009: The Cumberland Plateau of Eastern Kentucky.- In: Palmer, A.N. & MV Palmer (eds.), Caves and Karst of America. National Speleological Society, pp. 70-79. Walden, K., 1999: Strontium isotopes in Redmond Creek Cave, Monticello, Kentucky.- Honors Thesis, Ohio State University, 66 p. Walden, W.D., Walden, K.M. & L.J. Florea, 2007: The Caves and Karst of Redmond Creek, Wayne County, Kentucky.- National Speleological Society Convention Program Guide, p. 107. White, W.B., 1988: Geomorphology and Hydrology of Karst Terrains. New York, Oxford University Press, 464 pp. Zhang, C., 2011: Carbonate rock dissolution rates in different landuses and their carbon sink effect.- Chinese Science Bulletin, doi: 10.1007/s11434-011 -4404-4.