Formate, acetate, and propionate as substrates for sulfate reduction in sub-arctic sediments of Southwest Greenland

Volatile fatty acids (VFAs) are key intermediates in the anaerobic mineralization of organic matter in marine sediments. We studied the role of VFAs in the carbon and energy turnover in the sulfate reduction zone of sediments from the sub-arctic Godthåbsfjord (SW Greenland) and the adjacent continental shelf in the NE Labrador Sea. VFA porewater concentrations were measured by a new two-dimensional ion chromatography-mass spectrometry method that enabled the direct analysis of VFAs without sample pretreatment. VFA concentrations were low and surprisingly constant (4–6 μmol L−1 for formate and acetate, and 0.5 μmol L−1 for propionate) throughout the sulfate reduction zone. Hence, VFAs are turned over while maintaining a stable concentration that is suggested to be under a strong microbial control. Estimated mean diffusion times of acetate between neighboring cells were <1 s, whereas VFA turnover times increased from several hours at the sediment surface to several years at the bottom of the sulfate reduction zone. Thus, diffusion was not limiting the VFA turnover. Despite constant VFA concentrations, the Gibbs energies (ΔGr) of VFA-dependent sulfate reduction decreased downcore, from −28 to −16 kJ (mol formate)−1, −68 to −31 kJ (mol acetate)−1, and −124 to −65 kJ (mol propionate)−1. Thus, ΔGr is apparently not determining the in-situ VFA concentrations directly. However, at the bottom of the sulfate zone of the shelf station, acetoclastic sulfate reduction might operate at its energetic limit at ~ −30 kJ (mol acetate)−1. It is not clear what controls VFA concentrations in the porewater but cell physiological constraints such as energetic costs of VFA activation or uptake could be important. We suggest that such constraints control the substrate turnover and result in a minimum ΔGr that depends on cell physiology and is different for individual substrates.

Volatile fatty acids (VFAs) are key intermediates in the anaerobic mineralization of organic matter in marine sediments. We studied the role of VFAs in the carbon and energy turnover in the sulfate reduction zone of sediments from the sub-arctic Godthåbsfjord (SW Greenland) and the adjacent continental shelf in the NE Labrador Sea. VFA porewater concentrations were measured by a new two-dimensional ion chromatography-mass spectrometry method that enabled the direct analysis of VFAs without sample pretreatment. VFA concentrations were low and surprisingly constant (4-6 mol L −1 µ for formate and acetate, and 0.5 µmol L −1 for propionate) throughout the sulfate reduction zone. Hence, VFAs are turned over while maintaining a stable concentration that is suggested to be under a strong microbial control. Estimated mean diffusion times of acetate between neighboring cells were <1 s, whereas VFA turnover times increased from several hours at the sediment surface to several years at the bottom of the sulfate reduction zone. Thus, diffusion was not limiting the VFA turnover. Despite constant VFA concentrations, the Gibbs energies ( G r ) of VFA-dependent sulfate reduction decreased downcore, from Introduction Mineralization of buried organic matter drives microbial activity and element cycles in sub-seafloor sediments (Reeburgh, 1983;Arndt et al., 2013). The amount and quality of this sedimentary organic matter determines the rates of microbial processes and the microbial abundance Røy et al., 2012;Algora et al., 2013;Glombitza et al., 2013;Orcutt et al., 2013). The initial step in the degradation of high molecular weight organic matter in anoxic sediments is the hydrolytic breakdown of polymers by exoenzymes excreted by heterotrophic bacteria. This produces smaller compounds (mono-and dimers) which can be taken up by fermentative bacteria (e.g., Capone and Kiene, 1988). Fermentation does not result in a net oxidation of the organic matter but produces volatile fatty acids (VFAs), small alcohols, amines, CO 2 , H 2 etc. The VFAs are quantitatively important intermediates and are turned over in the terminal steps of organic matter mineralization (e.g., sulfate reduction) by which they serve as electron donors (Middelburg, 1989).
VFA concentrations in sediment porewater thus reflect a balance between VFA generation and consumption, both of which take place in the same sediment zone. Many studies have analyzed the terminal mineralization steps by measuring the turnover of the electron acceptor, in particular of sulfate (Jørgensen, 1982;Burdige, 1993;Orcutt et al., 2013). However, the intermediates are equally important because they link the organic matter decomposition to the terminal oxidation processes. Under steady state conditions the concentrations of the intermediates are low and reflect a close coupling of fermentation and terminal oxidation. However, when environmental conditions change the transition can cause an imbalance with elevated concentrations. For example Hoehler et al. (1998) reported that acetate (and H 2 ) concentrations increased after depletion of sulfate in incubation experiments with coastal marine sediments. The loss of the dominant electron acceptor, sulfate, resulted in an increase in fermentation products because they were initially degraded at a lower rate. After a significant time lag the steady state concentrations of the intermediates were restored by microorganisms under methanogenic conditions.
The role of specific VFAs as electron donors in organic matter mineralization has been studied by radiotracer and inhibitor incubation experiments (Christensen and Blackburn, 1982;Ingvorsen et al., 1984;Shaw and McIntosh, 1990;Finke et al., 2006). The earlier analytical methods for VFAs in marine porewater required sample pretreatment such as derivatisation or distillation and they often lacked sufficient sensitivity to quantify the low in-situ concentrations. We recently developed a very sensitive analytical approach based on two-dimensional ion chromatography-mass spectrometry (2D IC-MS) (Glombitza et al., 2014). This approach enables the direct quantification of several C 1 -C 5 VFAs, lactate and pyruvate from marine porewater without sample pretreatment and with a detection limit of 0.1-0.5 µmol L −1 . We used this 2D IC-MS method to obtain high resolution depth profiles of VFAs in the porewater from 6-m deep sediment cores retrieved from the sub-arctic Godthåbsfjord system in Southwest Greenland as well as outside the fjord on the Greenland shelf in the northeastern Labrador Sea. A pycnocline separating the saline bottom water from the glacial fresh water at the surface prevents the water from mixing and maintains a constant bottom temperature of 2 • C (Mortensen et al., 2011). This stable environment is ideal to investigate how VFA concentrations are balanced at steady state.
The aim of our study was thus to understand the controls on in-situ porewater concentrations of VFAs in relation to the turnover of carbon and energy in the sulfate reduction zone of marine sediments. To relate the in-situ VFA concentrations to the carbon and energy turnover in the sediment, porewater concentrations of VFAs, sulfate, sulfide, and DIC were used to calculate the Gibbs energy of sulfate reduction utilizing individual VFAs as electron donors. We calculated the turnover rates of individual VFAs from experimentally measured sulfate reduction rates. We also evaluated the influence of VFA diffusion in the sediment on the microbial availability of VFAs and discuss the potential for thermodynamic control of in-situ porewater concentrations. A more detailed analysis of microbial activity and diversity in relation to sediment age and properties will be presented in a subsequent study (Jaussi et al., unpublished).

Study Area
The Godthåbsfjord (Figure 1) is a complex, sub-arctic fjord system in southwest Greenland consisting of a main fjord named Godthåbsfjord or Nûp Kangerdlua and several connected side branches, Qôrnup Suvdlua and Umánap Suvdlua, Kapisigdlit Kanderdluat, and the Kangersuneq which receives three glacier outlets from the Greenland Ice Sheet (GIS). The capital of Greenland, Nuuk (64 • 10 ′ N, 51 • 44 ′ W), is located at the mouth of the fjord. The combined area of the Godthåbsfjord is 2013 km 2 with a volume of 525 km 3 . Mean water depth is 260 m with several sills (Mortensen et al., 2011). The fjord system is characterized by the inflow of saline, well-oxygenated water from the Labrador Sea (West Greenland Current water) as a subsurface/bottom current filling the deep basins of the fjord, and the outflow of less saline water, primarily derived from glacier meltwater, which exits the fjord as a surface current (Buch, 2002).
In August 2013, 6-m long gravity cores were retrieved at four stations outside and inside of the fjord system during a cruise with R/V Sanna. Additionally, two shorter cores were retrieved by a Rumohr Corer (Meischner and Rumohr, 1974) at Station 3 and 6 ( Table 1). Station 3 (Figures 1, S3), is located outside of the fjord on the continental shelf in the Labrador Sea. At this site, there is a strong northward flow of the West Greenland Current. Station 5 (Figures 1, S5) is located in the main fjord (Figure 1, a). A number of turbidites were identified in this core. Station 6 (Figures 1, S6) is located in the Kapisigdlit Kanderdluat fjord (Figure 1, d) and Station 8 (Figures 1, S8) is located in close proximity of the northernmost outlet of the GIS in the Kangersuneq (Figure 1, e). This site is today characterized by turbid water and significant sediment influx from the melting glacier. CTD profiles were measured   at all sites and confirmed the presence of oxygenated bottom water at marine salinity (S = 35). In-situ temperatures measured at the bottom of the cores and were similar to the bottom water temperatures (Station 3: 4 • C, Stations 5, 6, and 8: 2 • C).

Porewater Sampling
Samples for porewater analyses were taken from all cores by Rhizon soil-moisture samplers (Seeberg-Elverfeldt et al., 2005) that had been previously cleaned by filtering through 50 mL Milli-Q water and stored in sealed gas-tight bags to minimize contamination by VFAs from air and equipment. This treatment was previously shown in blank samples to reduce contamination by VFAs from the porous membrane of the Rhizon samplers to below detection levels (Glombitza et al., 2014). Porewater samples were obtained by pushing the soil-moisture samplers through 3-mm wide holes drilled in the core liner. The first 2 mL were used for chloride, sulfate, sulfide, and dissolved inorganic carbon (DIC) analysis. The following 1 mL was sampled for VFA analysis in baked (450 • C, 5 h) glass vials and stored frozen at −80 • C for 3 months before analysis. Porewater samples were taken every 5-10 cm in the Rumohr cores and every 25 cm in the gravity cores.

Sampling for Cell Counting
Sediment for cell counts was retrieved with sterile 2.5 mL syringes with cut-off tips through windows cut in the core liners. One cm 3 sediment was transferred into centrifuge tubes with 4 mL filter-sterilized NaCl solution (30 g L −1 ) amended with paraformaldehyde (2% w:v). The sample was shaken to form a homogenous slurry. Samples were stored at 4 • C until analysis in the laboratory.

Samples for Sulfate Reduction Rate Determination
Small sub-cores for sulfate reduction rate determination were taken from sediment cores in sterile 5 mL plastic syringes with a cut-off tip through cut windows immediately after porewater sampling. The windows were cut a few cm away from the porewater sampling positions at the same depths as the porewater samples but carefully avoiding cross contamination. The samples were kept in air-tight bags with AnaeroGen ™ O 2 scrubbers (OXOID, Thermo Scientific; do not leak H 2 ) at the in-situ bottom water temperature (2 • C) until incubation with 35 S-labeled sulfate tracer, immediately after subsampling the core (typically within 2-4 h).

VFA Concentrations in Porewater
VFA concentrations were measured by two-dimensional ion chromatography-mass spectrometry using a Dionex ICS3000 ion chromatograph coupled to an MSQ Plus (Thermo Scientific) mass spectrometer. The method is described in detail in Glombitza et al. (2014). Briefly, by this method the first IC dimension is used to separate the inorganic ions, such as chloride, from the VFAs. The VFAs are trapped on a concentrator column and subsequently separated in the second IC dimension. The column for the first dimension was a Dionex IonPac ™ AS24 and for the second dimension was a Dionex IonPac ™ AS11HC (both Thermo Fisher Scientific).
Prior to IC-MS analysis the samples were defrosted and filtered through disposable Acrodisc R 13 mm IC syringe filters (pore size 0.2 µm) that had been flushed with 10 mL Milli-Q water directly before use. The first 0.5 mL of porewater after filtration was discarded and the next 0.5 mL was used for IC-MS analysis. All samples were measured undiluted and without further treatment.
Detection limits for the individual VFAs are all between 0.1 and 0.5 µmol L −1 . For a detailed discussion of analytical and statistical parameters (detection limits, sensitivity, accuracy, and precision) of all analytes see Glombitza et al. (2014). Because the samples were measured without dilution or pre-treatment other than filtration, potential contamination may only derive from the sampling device (Rhizon), the syringe filter or the sampling vial. This was all evaluated carefully in Glombitza et al. (2014) and detailed instructions for sampling and sample preparation (such as baking the vials and effective cleaning and storage of filters and Rhizon samplers) were presented to minimize the contamination below the analytical detection limit. The background (peak areas deriving from the instrumental analytical procedure itself, i.e., peaks deriving from formate and acetate background in eluent accumulating in the trap column) was determined by repeated blank measurement runs without sample injection every 5-10 measurements and used to correct the peak areas in the sample measurements.

Inorganic Ion Concentrations in Porewater
Sulfate and chloride concentrations were measured by suppressed ion chromatography with an ICS2500 system (Dionex) equipped with an eluent generator (EG50) and KOH eluent generator cartridge (EGC III KOH). The column was a Dionex IonPac ™ AS18 operated at 30 • C. KOH concentration started at 20 mmol L −1 and was raised to 32 mmol L −1 at the end of the analysis run at 15 min. Hydrogen sulfide (sum of H 2 S, HS − and S 2− ) was determined spectrophotometrically at 670 nm (FLUOstar Omega, BMG Labtech GmbH, Orthenberg, Germany) on zincpreserved porewater samples by the methylene blue method (Cline, 1969;Reese et al., 2011). DIC was measured immediately after the cruise on headspace-free porewater samples stored at 4 • C. Samples were transferred to sealed exetainers and acidified with 85% (v:v) phosphoric acid. After 24 h of equilibration time, the produced CO 2 was measured from the headspace of the exetainer by a Delta V ™ isotope ratio mass spectrometer (IRMS, Thermo Scientific).

Cell Abundance
Microbial cells were quantified according to the protocol of Kallmeyer et al. (2008) with a few modifications. Briefly, the slurries were first treated with a filter-sterilized solution of NaCl (30 g L −1 ), detergent mix and methanol and then mixed for 60 min on a Thermomixer comfort R (Eppendorf AG, Hamburg) at 1200 rpm and 18 • C. After a first density centrifugation with filter-sterilized Nycodenz solution (50% w/v) to separate cells from sediment particles, the slurries were further treated mechanically by sonication. The sonication probe (SONOPULS HD 2070, Bandelin Berlin, Germany) was placed directly in the samples and pulsed 3 times for 10 s at the minimum adjustable level (10%) including a 30 s break between the individual pulses. After a second density centrifugation with Nycodenz solution, the pooled cell extracts were filtered through 25 mm black polycarbonate filters (GTBP, 0.2 µm-pore size) and washed with 2 mL of TE buffer (pH 8.0). After drying, the filters were stained with DAPI-CV-mounting solution for 20 min [DAPI stain (1:100), 4:1 CitifluorTM (Citifluor Ldt, England)-Vectashield R mounting medium (Vector laboratories)]. At least 400 cells were counted over a minimum of 12 fields of view in an epifluorescence microscope (Axiovert 200M Zeiss, Germany). An additional acid treatment was included for slurries containing carbonates, before the chemical treatment (Kallmeyer et al., 2008). Those slurries were amended with 500 µL carbonate dissolution mix [aqueous solution with high acidity and moderate pH (4.6) containing 0.43 mol L −1 glacial acetic acid and 0.43 mol L −1 sodium acetate] for minimum 1 h (until CO 2 bubbles were no longer observed), followed by three consecutive washing steps with NaCl (30 g L −1 ).

Sulfate Reduction Rate Determination
Sulfate reduction rates (SRRs) were determined by incubation with 35 S-labeled sulfate in undisturbed sediments samples (Jørgensen, 1978;Røy et al., 2014). Carrier-free (without nonradioactive sulfate) 35 SO 2− 4 (10 µL, 10 kBq µL −1 ) was injected with a Hamilton gas tight syringe in the center of each sub-core. The samples were incubated for 12 h at 2-3 • C in the dark in sealed air-tight bags with AnaeroGen ™ O 2 scrubber (OXOID, Thermo Scientific). To terminate the incubation, the samples were frozen in the bags at −20 • C. In order to calculate sulfate reduction rates, the total reduced inorganic sulfur (TRIS) was separated from sulfate by a cold chromium distillation described in detail in Kallmeyer et al. (2004) including modifications and recommendations by Røy et al. (2014). Before the distillation, the samples were thawed and suspended in 5 mL 20% (w:v) Zinc-Acetate (ZnAc) solution. Na 2 S (200 µL, 0.5 mol L −1 ) was added to the reaction flask as a sulfide carrier. At the end of the distillation, the distillate in the ZnAc trap was transferred into 15 mL scintillation liquid (Ecoscint XR, National diagnostics, Atlanta, GA, USA). The radioactivity of the total sulfur (a TOT ) and in the reduced sulfur fraction (a TRIS ) was measured in a liquid scintillator counter (Packard Tri-Carb 2900 TR liquid scintillation analyzer). Samples were counted for 30 min. Blank samples, which were transferred to ZnAc (20% w:v) before tracer injection, were used to determine the background. Sulfate reduction rates (SRR) were calculated according to Kallmeyer et al. (2004) (Equation 1): where SO 2− 4 is the porewater sulfate concentration, φ is the porosity, a TRIS the radioactivity of the reduced sulfur fraction, a TOT the total sample radioactivity, t the incubation time and 1.06 the correction factor for the estimated microbial isotopic fractionation of sulfur during sulfate reduction (Jørgensen and Fenchel, 1974). Porosity was calculated from the weight loss of 2 cm 3 of wet sediment during drying (100 • C until constant weight). Mean values were calculated when duplicate SRRs were determined at the same depth.

Calculation of the Gibbs Energy
The Gibbs energy of a reaction ( G r ) is constrained by the chemical and physical environment in which the reaction takes place, namely the pressure, the temperature, and the activities of the educts and products. The temperature was 2 • C while the pressure ranged from 4.0 to 6.3 MPa according to the water depth ( Table 1). We used a mean pressure of 5 MPa for calculations. The G r is calculated from the standard Gibbs energy of reaction, G 0 r , according to Equation (2): where R (0.008314 kJ mol −1 K −1 ) is the universal gas constant, T (in K) is the temperature, and Q r is the reaction quotient of the specific reaction. The standard Gibbs energies of reaction, G 0 r , are a function of temperature and pressure. Values of G 0 r (T, p) are calculated using the revised-HKF (Helgeson-Kirkham-Flowers model) equations of state (Helgeson et al., 1981;Tanger and Helgeson, 1988;Shock et al., 1992) and the software package SUPCRT92/OBIGT . Thermodynamic properties were taken from Shock and Helgeson (1988), Shock (1995), and Shock et al. (1997). The standard state thermodynamic properties of the species are listed in the Supplementary Table 1. The reaction quotient Q r is calculated as the product ( ) of the activities of the species in the specific reaction (Equation 3): where a denotes the activities of the reaction participants (educts and products), and ν the stoichiometric coefficient of the i-th educt or product. The activities were approximated by multiplying the measured concentrations of the species by the activity coefficients. Activity coefficients were calculated using an extended version of the Debye-Hückel equation (Helgeson, 1969) and the Geochemists Workbench R (www.gwb.com) software for an ionic strength of I = 0.7 and a temperature of 2 • C (275 K

Results
Formate, Acetate, and Propionate Formate, acetate, and propionate were detected in the porewater of all four stations. Butyrate and valerate were constantly below our detection limit (0.13 and 0.18 µmol L −1 ). Depth profiles of the measured VFA concentrations are shown in Figure 2.
Formate concentrations were nearly constant with depth at Station 3 and 8 with average values of 4-5 µmol L −1 . At Station 6, the concentrations were mostly similar (average of 5 µmol L −1 ), with the exception of 100-300 cmbsf where they were slightly higher and more scattered (average of 9 µmol L −1 ). Concentrations in the porewater of Station 5 were slightly higher and more scattered throughout the whole core with concentrations between 5 and 13 µmol L −1 (average 7 µmol L −1 ).
Acetate concentrations at Station 3 decreased slightly from 9 to 5 µmol L −1 through the upper 100 cm below which the concentrations remained at 5-6 µmol L −1 . At Stations 6 and 8 average concentrations were 6 µmol L −1 with exception of the 100-300 cmbsf interval at Station 6 where acetate concentrations were around 8 µmol L −1 and more scattered. As for formate, the acetate concentrations in the porewater of Station 5 were generally slightly higher and more scattered (2.5-21 µmol L −1 , average 14 µmol L −1 ).
Propionate concentrations were about an order of magnitude lower than formate and acetate in all cores but also rather constant with depth. At Stations 3 and 8, average concentrations were 0.5 µmol L −1 . At Station 6, they were slightly higher (0.7 µmol L −1 ), in particular in the 100-300 cmbsf depth interval with average concentrations of 1 µmol L −1 . Like for formate and acetate, propionate concentrations at Station 5 were generally higher and more scattered with values of 0.7-1.8 µmol L −1 (average 1.3 µmol L −1 ).

Sulfate Reduction Rates
At Station 3, sulfate reduction rates were highest, 35 nmol cm −3 d −1 , near the sediment surface (Figure 3). At this station, the rates showed a >1000-fold decrease downcore and reached 0.01 nmol cm −3 d −1 at the bottom of the core. SRRs at Station 5 were relatively constant with depth (∼0.1 nmol cm −3 d −1 ). At Station 6, SRRs were high near the top and decreased to 0.1 nmol cm −3 d −1 at 100 cmbsf. Below this depth SRRs were below detection. At Station 8, in contrast, SRRs were generally high (around 1 nmol cm −3 d −1 ) throughout the sediment core.

Cell Abundance
Cell numbers at Stations 3 and 6 were typical for costal sediments (Parkes et al., 2000;Kallmeyer et al., 2012) with up to nearly 10 9 cells cm −3 in the uppermost samples (<10 cmbsf) and a decrease down core (Figure 4A). At Stations 5 and 8, cell numbers remained rather constant throughout the cores at 10 8 cells cm −3 .

Discussion
Our study focused on potential controls on in-situ porewater concentrations of formate, acetate and propionate which are key intermediates in subsurface microbial metabolism. We measured VFA concentrations in the porewater of sediment cores from the sub-arctic Godthåbsfjord and the adjacent continental shelf. In order to understand what determines the in-situ concentrations of these metabolic intermediates we measured sulfate reduction rates by radiotracer incubations to estimate the VFA turnover rates. We determined the abundance of cells in the sediments to evaluate the influence of VFA diffusion times between the cells. Additionally, we analyzed the concentrations of sulfate, sulfide, and DIC in the porewater to calculate Gibbs energy for sulfate reduction in order to identify energetic limitations for this metabolic process.

Porewater VFA Concentrations
Porewater VFA concentrations in the sulfate reduction zone were in the low micromolar (formate and acetate) to sub-micromolar (propionate) range. In coastal marine sediments, higher acetate concentrations (several 10 s to 100 µmol L −1 ) have been reported earlier, by using a vacuum distillation technique to separate the VFA from porewater chloride and by subsequent gas chromatographic analysis (e.g., Christensen and Blackburn, 1982). Other studies that applied ion chromatographic analysis after vacuum distillation (Parkes and Taylor, 1983) or derivatisation and subsequent HPLC analysis (Albert and Martens, 1997) reported similar porewater acetate concentrations in the sulfate reduction zone as reported here (Ansbaek and Blackburn, 1980;Balba and Nedwell, 1982;Jørgensen and Parkes, 2010;Vandieken and Thamdrup, 2013).
It is striking that the porewater VFA concentrations show very little variability with depth and that the profiles are very similar between the stations. The slightly higher and more fluctuating VFA concentrations at Station 5 were most likely caused by fluctuating sedimentation, indicated by numerous turbidities throughout the core. The amount and quality (reactivity) of sedimentary organic matter usually control the rates of mineralization in the sediments (Jørgensen, 1982;Røy et al., 2012;Algora et al., 2013;Glombitza et al., 2013). From the observed decrease in measured sulfate reduction rates (Figure 3) it can be concluded that the organic matter mineralization rates are highest at the top 10's of cm of sediment and decrease with depth. Accordingly, we assume a decrease of VFA production rates by fermentation.
In general, the in-situ porewater concentrations measured in all cores were constant and presumably in steady state. As the mineralization rates, and thus the production rates of the individual VFAs, decrease strongly with depth, the VFA concentrations are apparently under strict control by the consumers. There appears to be a lower threshold concentration below which the microbial uptake of VFAs is inhibited. The threshold concentrations are similar for formate (4.0-4.5 µmol L −1 ) and acetate (5.4-5.9 µmol L −1 ), with only slightly lower concentration for formate, whereas the threshold concentration for propionate (0.5 µmol L −1 ) is an order of magnitude lower. As propionate may be further fermented to acetate and H 2 , the low concentration of propionate could be due to control by fermentation instead of the terminal oxidation by sulfate reduction. If the observed VFA concentrations are indeed threshold levels controlled by the uptake by sulfate reducers or other heterotrophic bacteria in the sulfate zone, then concentrations might be different in other sediment zones (e.g., in the underlying zone of methanogenesis). Although the relatively constant VFA concentrations found in the sulfate reduction zone suggest uptake control it is not clear what determines the specific levels of the observed VFA concentrations. Are the apparent threshold concentrations determined by kinetic or energetic properties of the cellular uptake and metabolism, by the chemical properties of VFAs in the porewater, or by other mechanisms?

VFA Turnover Rates
Turnover rates of VFAs by sulfate reduction can be estimated from measured sulfate reduction rates based on the stoichiometry of their reaction with sulfate (formate:sulfate = 4:1, acetate:sulfate = 1:1, propionate:sulfate = 0.57:1, see Equations 6-9, Section Gibbs Energy of Sulfate Reduction) and an estimation of the contribution of the individual VFAs to feed the sulfate reducers. There are only few studies that have estimated the contribution of different VFAs as electron donors for sulfate reduction in marine sediments. To the best of our knowledge, no study has yet estimate the in-situ contribution of formate to sulfate reduction. Sørensen et al. (1981) measured the initial accumulation of acetate, propionate, butyrate, and hydrogen in coastal anoxic sediment during incubation with molybdate (MoO 2− 4 ) added to inhibit sulfate reduction. Accumulation rates of the VFAs and hydrogen were expected to represent the turnover rates of these substrates by sulfate reduction in the un-inhibited sediment. The authors attributed 10% of the sulfate reduction to hydrogen, 40-50% to acetate, 10-20% to propionate and 10-20% to butyrate. In a similar experiment, Christensen (1984) found a mean contribution of 65% acetate, 14% propionate, 8% butyrate, and 6% isobutyrate.
The turnover of acetate in sediment was also determined by incubation experiments with 14 C-labeled acetate added and subsequent measurement of the concentrations of acetate and the decrease of the labeled substrate (Christensen and Blackburn, 1982;Shaw et al., 1984;Shaw and McIntosh, 1990). These experiments showed mostly higher rates than those determined by the molybdate-amendment method and sometimes rates even exceeded the measured sulfate reduction rates or the total mineralization rates estimated from DIC or NH + 4 production (Ansbaek and Blackburn, 1980;Christensen and Blackburn, 1982). It was suggested that a part of the measured acetate was complexed or adsorbed and thus represented an acetate pool which was less available to microorganisms than the free 14 C-labeled acetate (Parkes et al., 1984). Thus, earlier estimates based on the 14 C-incubation method seem to overestimate in-situ turnover rates. Finke et al. (2006) estimated the contribution of individual substrates to sulfate reduction in Arctic fjord sediments using two different methods: (1) inhibition of sulfate reduction by the addition of selenate (SeO 2− 4 ) and measuring the initial increase of substrate concentrations, and (2) incubations with 14 Clabeled substrates that were previously equilibrated in sterilized porewater to achieve potential complexion of the labeled acetate and to mimic in-situ conditions. The turnover rates determined by both methods were similar. A contribution of 40% acetate, 8% propionate, 1.3% butyrate, and 3% lactate to the total sulfate reduction was estimated.
In order to estimate acetate and propionate turnover rates, we have chosen values for the respective contributions to sulfate reduction of 40% acetate and 8% propionate as published in Finke et al. (2006). We consider these data to be realistic estimates based on the discussion above. Furthermore, the relative differences between the published data are small in relation to the main conclusions drawn in the following discussion.
Turnover rates for acetate and propionate were calculated from sulfate reduction rates (Figure 3) as described above and are shown in Figure 5A. Acetate turnover rates were approximately nine-fold faster than those for propionate (which is the ratio between 40 and 8% divided by the stoichiometry 0.57:1). At Station 3, sulfate reduction rates and thus estimated acetate and propionate turnover rates decreased strongly downcore, starting in the uppermost sample with rates of 14 nmol cm −3 d −1 for acetate and 1.6 nmol cm −2 d −1 for propionate. With a decrease of nearly four orders of magnitude, Station 3 covered the widest range of acetate and propionate turnover rates of all four sites. Rates at Station 5 were relatively low in the uppermost samples (acetate = 0.2 nmol cm −3 d −1 , propionate = 0.024 nmol cm −3 d −1 ) and decreased over only one order of magnitude. Station 6 showed a decrease over two orders of magnitude within the upper 100 cmbsf (acetate = 5 nmol cm −3 d −1 , propionate = 0.55 nmol cm −3 d −1 ). Below 100 cm depth, sulfate reduction rates were below detection and VFA turnover rates could not be calculated. Station 8 showed relatively constant VFA turnover rates throughout the core, similar to Station 5 but at rates approximately twice as high (uppermost samples: acetate = 0.49 nmol cm −3 d −1 , propionate = 0.056 nmol cm −3 d −1 ). In these calculations we have assumed a complete oxidation of propionate to bicarbonate. However, if propionate is fermented to acetate and H 2 , as discussed in Sørensen et al. (1981), a proportion of the turnover calculated for acetate would additionally account for propionate turnover.
Relative to the calculated VFA turnover rates due to sulfate reduction, which cover a broad span over four orders of magnitude, the in-situ porewater concentrations are much more constant throughout the cores. This suggests that the in-situ VFA concentrations are not directly controlled by the kinetics of their turnover.

Diffusion and VFA Turnover Time
Diffusion of VFAs could, in theory, represent a bottleneck for the uptake of the substrates if the diffusion time between the cells generating the VFAs (e.g., fermenters) and the cells utilizing these VFAs as substrates (e.g., sulfate reducers) were long compared to the substrate turnover time. The measured VFA concentrations represent the average concentration in the porewater retrieved from a few cm 3 of sediment. On a cell to cell scale, however, the VFAs diffuse from producer cells to consumer cells and this diffusion must be associated with local gradients. To evaluate the potential for diffusion limitation we compared mean diffusion times of VFAs between neighboring cells with the turnover times of the VFA pools.
The diffusion coefficient (D) for acetate in aqueous dilutions was calculated to be 1.95×10 −5 cm 3 s −1 at 2 • C ( Leaist and Lyons, 1984). We used this value of D to calculate the mean diffusion Assuming an even distribution of cells, we calculated the mean distance between cells (L) (Figure 4B) from the cell abundance in the sediment (Figure 4A). At Stations 3 and 6, the average neighboring cell distance increases from 10 µm at the top to 30 µm at the bottom of the cores. The relatively constant cell numbers with depth at Station 5 and 8 result in average neighboring cell distances of 13-20 µm. Calculated diffusion times for acetate between the neighboring cells at Stations 3 and 6 range from 0.03 s at the surface to 0.4 s at the bottom of the core (Figure 4C), while at Stations 5 and 8 mean diffusion times were around 0.1 s. The assumptions of even cell distribution and diffusion only between neighboring cells are obviously over-simplified. However, if we assume a 3-10-fold greater distance between fermenters and sulfate reducers, about 100 µm, the acetate diffusion time would still be below 4 s. With an even distribution of cells this would correspond to 10% sulfate reducers (Hoehler and Jørgensen, 2013) and 1% fermenters, which is probably an underestimate.
By dividing the in-situ acetate and propionate concentrations in the porewater (Figure 2) by their estimated turnover rates (Figure 5A), the turnover times of the acetate and propionate pools were calculated ( Figure 5B). Turnover times at Station 3 increased with depth over a wide range from 10 h to 3-4 years for acetate and 7 h to 3 years for propionate. Turnover times at Station 5 increased from 2 months to 3 years for acetate and from 1 to 8 months for propionate. At Station 6, turnover times increased from 1 day at the surface to 5 months below 100 cmbsf below which sulfate reduction was below the detection limit. Fastest turnover times for acetate and propionate at depth were observed at Station 8.
For the calculation of turnover rates and times we used the estimated contributions of acetate and propionate of 40 and 8%, respectively, published by Finke et al. (2006). Even when taking a higher contribution of these acids, 65 and 14%, as published by Christensen (1984), turnover times would still be in a similar range, e.g., between 7 h and 2 years for acetate and between 5 h and 2 years for propionate in Station 3, thus covering the full range of turnover times represented by our estimates.
The very long turnover times of acetate and propionate, ranging from days to years, are in stark contrast to the calculated mean diffusion times of acetate between cells, which are in the range of seconds. This shows that diffusion is relatively a very fast process and that the microbial cells are surrounded by dilute substrates at constant steady-state concentrations and slow turnover. Hence, it is strongly indicated that diffusion does not limit the turnover of VFAs and that, in turn, the VFA concentrations are not the result of diffusion-limited uptake.
Similar conclusions were obtained for samples from ODP Leg 201, Site 1226. In the deep sediment cores of Site 1226 (up to 400 mbsf) acetate turnover times were estimated to be 24 years (Wang et al., 2010). Here we demonstrate that even in much shallower sediments with shorter VFA turnover times diffusion is not controlling the uptake or concentrations of VFAs.

Gibbs Energy of Sulfate Reduction
The energy that is available to microorganisms by a specific catabolic reaction is expressed by the Gibbs energy of the chemical reaction ( G r ) which is constrained by the chemical environment in which the reaction takes place, namely the concentrations of educts and products, as well as the insitu temperature and pressure. Standard Gibbs energy of the reactions, G 0 r , were calculated using standard state (i.e., activities of all compounds = 1 mol L −1 ) at the in-situ pressure and temperature as described above. For comparison, we also calculated the G 0 r at the reference pressure and temperature conditions ( at reference T and p are mainly due to the temperature difference between 25 (reference temperature) and 2 • C (insitu temperature). The influence of increasing pressure over the in-situ range has only negligible effect on the G 0 r values. Higher in-situ temperatures would thus result in higher Gibbs energy values. However, sub-arctic fjord sediments were exposed to constant low temperatures of 2 • C and do not show seasonal variations. As a result the system can be considered stable over long time periods and the measured concentrations and calculated Gibbs energies reflect a long-term equilibrium. Gibbs energy of organoclastic sulfate reduction under in-situ conditions ( G r ) in the sediment were calculated from Equation (2) using porewater concentrations of VFAs (Figure 2), sulfate, DIC, and hydrogen sulfide (Figure 2). For the calculation of G r from formate and propionate a pH of 7 was assumed. Calculated G r values of in-situ sulfate reduction utilizing different VFAs as electron donors are shown in Figure 6. The data are expressed in kJ (mol VFA) −1 ( G r,i ) according to Equation (4), i.e., per mol formate ( Figure 6A), acetate ( Figure 6B) or propionate (Figure 6C). The G r becomes more negative, corresponding to a larger molar energy yield, with increasing number of carbon atoms in the VFAs. At all stations, the Gibbs energy of sulfate reduction utilizing formate, actetate or propionate becomes less negative down-core, mainly as a result of increasing DIC and decreasing sulfate concentrations. Energy gained from sulfate reduction ranged from −28 to −16 kJ mol −1 for formate, −68 to −31 kJ mol −1 for acetate and −126 to −65 kJ mol −1 for propionate from top to bottom of the cores.
Sulfate reduction at the shelf Station 3 showed a strong decrease in free energy yield with depth, most steeply near the sediment surface and covering the whole range of G r values calculated for the three VFAs (Figure 3). Within the top 0-5 cm at Station 6, G r was similar to Station 3, −28 kJ mol −1 for formate, −67 kJ mol −1 for acetate, and −125 kJ mol −1 for propionate, and the free energy decreased with depth in the upper 100 cmbsf. Below this depth, free energy yields became only slightly less negative with depth as a result of the rather constant concentrations of sulfate and DIC (Figure 2). This coincides with the observations that sulfate reduction rates could not be measured experimentally below this depth. Absolute G r values at Stations 5 and 8 decreased only slightly with depth showing values around −24 kJ mol −1 for formate, −60 kJ mol −1 for acetate and −115 kJ mol −1 for propionate. In these sediments the catabolic reaction products (i.e., sulfide and DIC) have lower concentrations and consequently Gibbs energy of the catabolic reaction was higher (more negative). Propionate concentrations in the sediment were an order of magnitude lower and propionate oxidation had a significantly more negative G r . The oxidation of butyrate was energetically even more favorable and butyrate concentrations were generally below detection limit in the sediment. However, as already discussed, if propionate and butyrate were further fermented to acetate and H 2 their low concentrations in the porewater could be controlled by the fermenters rather than by the sulfate reducers. A recently published approach to evaluate the available energy in subsurface environments is the combination of the G r values with the concentrations of the substrate present in the environment (energy density). By multiplication of G r with the substrate concentrations energy densities can be expressed in J (kg H 2 O) −1 (LaRowe and Amend, 2014; Osburn et al., 2014). This provides a metric which can relate the energetics of potential microbial processes to the environmental conditions or physiologies. Because the VFA concentrations in the porewater were generally constant with depth and also similar between the individual stations, energy density profiles follow the Gibbs energy profiles shown in Figure 6. However, due to the low concentrations of propionate compared to formate and acetate, the energy density provided by acetate was highest in the sediments [

Energy Limitation
It is a fundamental requirement that the catabolic reactions provide sufficient energy to enable the organisms to generate ATP which is the energy currency in living cells (Thauer et al., 1977;Schink, 1997). The synthesis of ATP in cells requires about 50 kJ mol −1 (Thauer et al., 1977). A fraction of the energy invested in ATP generation is lost as heat. This energy dissipation is estimated to be 20 kJ (mol ATP) −1 , resulting in a minimum energy requirement for the synthesis of one mole of ATP of 70 kJ (Schink, 1990). Even under energy limitation (where cells are believed to reduce the heat loss to a minimum) at least 60 kJ (mol ATP) −1 may likely be required (Schink, 1997). ATP synthesis itself is coupled to the transport of charged ions, usually protons but also sodium, across the cell membrane (Mitchell, 1966). The general theory is that three protons cross the membrane for the generation of one molecule of ATP (Ferguson and Sorgato, 1982;Maloney, 1983). This results in a minimum free energy requirement of -20 kJ (mol proton) −1 to maintain a living cell (Schink, 1990(Schink, , 1997. This theoretical concept was tested for hydrogen consuming microorganisms in anoxic sediments from Cape Lookout Bight, i.e., for sulfate-reducing bacteria or, in the absence of sulfate, for methane-producing Archaea (Hoehler et al., 2001). The calculated free energy yields for hydrogen consuming sulfate reduction were found to decrease with depth until they reached an asymptote of −19.1 kJ (mol sulfate) −1 , which is in accordance with the theoretical concept. For hydrogen-driven methanogenesis an asymptote was found at −10.6 kJ (mol CO 2 ) −1 .
A similar asymptote to a certain energy yield for organoclastic sulfate reduction was not observed in our study (see Figure 4A). All calculated G r values per mol sulfate (or per mol acetate) exceeded (i.e., were more negative than) the −20 kJ mol −1 limit. However, hydrogen uptake in cells is different from the uptake of VFAs as hydrogen can freely diffuse though the cell membrane whereas VFA anions require an active uptake. Such an uptake system might require additional energy for the transport of VFAs which could result in a higher energy limit for organoclastic sulfate reduction. Additionally, molecular activation of VFAs (e.g., Schauder et al., 1986) might require additional energy.
To the best of our knowledge the acetate uptake mechanism in microbes from the deep biosphere is unknown and chemostat experiments with pure cultures are required to elucidate such mechanistic details. For acetoclastic sulfate reduction in Desulfobacter spp. Rabus et al. (2006) estimated that up to 2/3 of a mol ATP is synthesized by reducing one mole of sulfate. For an estimated 50 kJ mol −1 needed to generate 1 mole ATP from ADP (Thauer et al., 1977;Schink, 1997) this would result in a minimum energy requirement of −33 kJ (mol sulfate) −1 (or acetate) for acetoclastic sulfate reduction. This is in accordance with findings by Jin and Bethke (2009) in batch reactor experiments with Desulfobacter hydrogenophilus (DSM3380) that consumed acetate and sulfate. In their experiment sulfate and acetate decreased during 6 days of incubation and the available energy leveled off at −33.1 to −42.8 kJ mol −1 . It is striking that this level is similar to what we observe at the bottom of Station 3 (−31 kJ mol −1 ). At this depth, sulfate is depleted to 240 µM (Figure 2) and acetate turnover rates have dropped to 0.01 nmol cm −3 d −1 (Figure 5A). Thus, sulfate reduction might indeed proceed at the energetic limit in the sediment. In contrast, at Station 6, sulfate reduction is already below detection limit at 100 cmbsf (Figure 3). The Gibbs energy of sulfate reduction from acetate is, however, with −49 kJ mol −1 still relatively high, considerably higher than what we observe at the bottom of Station 3 (Figure 6). Furthermore, it is obvious that the low turnover rates observed at Station 5 are not a result of low energy yield of the reactions as G r is significantly above the lowest G r observed at the bottom of Station 3.

Conclusion
The aim of the current study was to understand the in-situ controls in sulfate reducing sediments on the concentrations and turnover of porewater VFAs, which represent key intermediates in the anaerobic microbial food chain. For this purpose, we analyzed porewater concentrations of sulfate, sulfide, DIC, and VFAs, as well as cell abundance and sulfate reduction rates. VFA concentrations were low and surprisingly constant with depth and sediment age, suggesting that the concentrations reflect a tightly controlled steady state. In sediments with very low rates of sulfate reduction, below our experimental detection limit, concentrations seemed to be less well constrained. We suggest that these VFA data represent threshold concentrations below which the microorganisms are unable to take up and metabolize the VFAs.
The observation that the steady-state concentrations of VFAs were similar over a very broad range of turnover rates suggests that they were not kinetically controlled. Calculated turnover times of acetate and propionate by sulfate reduction vastly exceeded diffusion times between the cells in the sediments. Consequently, cells lived in a uniform and constant, dilute solution of VFAs. Diffusion was not limiting the uptake of VFAs and VFA concentrations were not controlled by VFA diffusion. Gibbs energy of VFA dependent sulfate reduction in the sediments decreased with depth throughout the sulfate zone. Propionate was the most energy yielding substrate per mol among the three VFAs. Propionate concentrations were an order of magnitude lower than formate and acetate concentrations. A potential energetic limit of acetoclastic sulfate reduction might be reached at a G r of approximately −30 kJ (mol acetate) −1 at the bottom of the shelf Station 3.
It still remains unclear what exactly controls the VFA concentrations in the sediment porewater. It is likely that physiological constraints, perhaps associated with VFA uptake and activation, determine the in-situ porewater concentrations and that utilization below certain threshold concentrations is not feasible. These concentrations, together with the concentrations of the other substrates and products, determine the minimum G r for organoclastic sulfate reduction in the sediment. Consequently, this minimum energy would be different for different microbial communities (e.g., in sulfate reducing and methanogenic communities) and also different for individual substrates.

Author Contributions
CG, HR, and BJ designed the study; MJ, HR, and BL collected the samples; BL curated the cores and led the sampling; MS led the cruise; CG and MJ performed the laboratory work; CG and MJ wrote the paper; HR and BJ edited the paper; all authors reviewed the manuscript.