In situ measurements of hydrogen sulfide, oxygen, and temperature in diffuse fluids of an ultramafic-hosted hydrothermal vent field (Logatchev, 14°45′N, Mid-Atlantic Ridge): Implications for chemosymbiotic bathymodiolin mussels
The Logatchev hydrothermal vent field (14°45′N, Mid-Atlantic Ridge) is located in a ridge segment characterized by mantle-derived ultramafic outcrops. Compared to basalt-hosted vents, Logatchev high-temperature fluids are relatively low in sulfide indicating that the diffuse, low-temperature fluids of this vent field may not contain sufficient sulfide concentrations to support a chemosymbiotic invertebrate community. However, the high abundances of bathymodiolin mussels with bacterial symbionts related to free-living sulfur-oxidizing bacteria suggested that bioavailable sulfide is present at Logatchev. To clarify, if diffuse fluids above mussel beds of Bathymodiolus puteoserpentis provide the reductants and oxidants needed by their symbionts for aerobic sulfide oxidation, in situ microsensor measurements of dissolved hydrogen sulfide and oxygen were combined with simultaneous temperature measurements. High temporal fluctuations of all three parameters were measured above the mussel beds. H2S and O2 coexisted with mean concentrations between 9 and 31 μM (H2S) and 216 and 228 μM (O2). Temperature maxima (≤7.4°C) were generally concurrent with H2S maxima (≤156 μM) and O2 minima (≥142 μM). Long-term measurements for 250 days using temperature as a proxy for oxygen and sulfide concentrations indicated that the mussels were neither oxygen limited nor sulfide limited. Our in situ measurements at Logatchev indicate that sulfide may also be bioavailable in diffuse fluids from other ultramafic-hosted vents along slow and ultraslow spreading ridges.
Hydrothermal vents along oceanic spreading ridges occur in different types of geological settings. According to the prevailing rock types that characterize the ocean floor, vents have been classified as basalt hosted, ultramafic hosted, felsic hosted, or sediment hosted [Tivey, 2007]. Vents on mid-ocean ridges with ultrafast to intermediate spreading rates occur exclusively in basalt-hosted settings in which the upper oceanic crust is composed entirely of basaltic rocks [Wetzel and Shock, 2000]. In contrast, vents on slow and ultraslow spreading mid-ocean ridges are either basalt or ultramafic hosted. In ultramafic-hosted settings, the upper oceanic crust is composed mainly of mantle-derived peridotite [Cannat et al., 1997; Snow and Edmonds, 2007; Tivey, 2007; Wetzel and Shock, 2000]. Ultramafic systems are now known to occur more often than previously recognized along slow and ultraslow spreading ridges [Beaulieu, 2010; Snow and Edmonds, 2007]. Back-arc basin spreading centers are common in the West Pacific where the host rock composition can be felsic (consisting of andesite, rhyolite, and dacite) as well as basaltic [Martinez et al., 2007; Tivey, 2007]. Finally, hydrothermal settings can also be sediment hosted if they are associated with ridges close to continental margins, such as numerous vent fields along the west coast of North America [Tivey, 2007; Tunnicliffe et al., 1998].
Basaltic, ultramafic, felsic and sedimentary rocks react differently with subsurface seawater, which leads to differences in the fluid composition [German and Von Damm, 2006; Tivey, 2007]. In this paper we focus on basalt- and ultramafic-hosted systems along mid-ocean ridges. High-temperature hydrothermal fluids discharged from basalt-hosted systems have total sulfide concentrations (Stot = H2S + HS− + FeS + Sx2-) as high as 20 mM that can temporarily rise to as high as 30–110 mM during a magmatic event [German and Von Damm, 2006; Lilley et al., 2003; Tivey, 2007]. In contrast, high-temperature fluids from ultramafic-hosted systems such as Rainbow (36°14′N) and Logatchev (14°45′N) on the Mid-Atlantic Ridge (MAR) have high concentrations of dissolved hydrogen (12–19 mM) and methane (2.1–3.5 mM) but relatively lower total sulfide concentrations (0.8–2.5 mM) [Charlou et al., 2002; Douville et al., 2002; Schmidt et al., 2007]. Correspondingly, discrete sampling of Rainbow and Logatchev low-temperature diffuse fluids followed by shipboard analysis indicated only low total sulfide concentrations (Stot at Rainbow: ≤ 22 μM) with the exception of a single measurement of 70 μM free or bioavailable sulfide (Sfree = H2S + HS−) in Logatchev diffuse fluids [Geret et al., 2002; Schmidt et al., 2007]. In situ measurements confirmed the low sulfide content in Rainbow diffuse fluids (Stot < 5 μM) [Le Bris and Duperron, 2010; Schmidt et al., 2008], which suggests that only very little bioavailable sulfide is present in diffuse fluids from ultramafic-hosted vents (see section 2 for definitions of total and bioavailable sulfide). Beyond the study by Schmidt et al.  at Rainbow, which focused only on habitats where the vent shrimps Rimicaris exoculata occur (habitat temperatures between 5°C and 18°C), no in situ sulfide concentrations of fluids from ultramafic-hosted vent sites have been published.
At Logatchev, dense beds of the deep-sea mussel Bathymodiolus puteoserpentis [Maas et al., 1999; von Cosel et al., 1999] cover areas where diffuse fluids emerge from the subsurface. These mussels harbor two types of bacterial symbionts in their gills, methane oxidizers and sulfur oxidizers (also known as methanotrophic and thiotrophic symbionts) [Duperron et al., 2006]. The high methane concentrations in the end-member fluids of this ultramafic-hosted vent [Charlou et al., 2002; Schmidt et al., 2007] indicate sufficient quantities of methane for endosymbiotic methanotrophy. However, the lack of data on in situ concentrations of sulfide at ultramafic-hosted vents raises the question if enough free sulfide (H2S + HS−) is available to support the thiotrophic symbionts of B. puteoserpentis. In this study we combined in situ measurements of dissolved hydrogen sulfide (H2S), oxygen, and temperature over B. puteoserpentis mussel beds using amperometric H2S microsensors [Jeroschewski et al., 1996; Kühl et al., 1998], Clark-type O2 microsensors [Revsbech, 1989] and temperature sensors. We compare the results with data from the ultramafic-hosted Rainbow field as well as basalt-hosted hydrothermal vent systems and discuss their implications for chemosymbiotic bathymodiolin mussels.
2. Material and Methods
2.1. Study Site
The ultramafic-hosted Logatchev hydrothermal vent field is located at 14°45′N and 44°58′W on the Mid-Atlantic Ridge at a water depth of 2910–3060 m (Figure 1). The geological setting of this field has been described in detail by Bogdanov et al. , Gebruk et al. , and Petersen et al. . Here we provide a short summary of the six sites we investigated in this study, Sites 1–5 at Irina II and Site 6 at Irina I.
Irina II is a mound with a basal diameter of about 50 × 25 m and steep slopes rising about 15 m above the surrounding seafloor. Five vertical up to 3 m high smoker chimneys at the top of the mound rather slowly emanate black smoke [Petersen et al., 2009] (Figure 1). Diffuse flow escapes through the chimney walls of beehive and pillar-like sulfide structures. The Irina II mound is densely covered with mussels [Petersen et al., 2009]. A single mussel bed of at least 20 m diameter extends from the base to the top of the central smoker complex [Gebruk et al., 2000]. In addition, the mound is surrounded by smaller mussel beds (up to 3 × 4 m) and mussel patches of 20–30 cm in diameter. These mussel beds are inhabited by polychaetes, gastropods (snails and limpets), alvinocarid shrimps, crabs, and large numbers of brittle stars [Gebruk et al., 2010, 2000; Van Dover and Doerries, 2005]. A single highly active black smoker rises at the southeastern end of the mound [Petersen et al., 2009].
Irina I is characterized by smoking craters, emanating high-temperature fluids, structures that are only known from the Logatchev vent field. Small chimneys occur on the crater rims and black smoke is intensively vented either from the chimneys on the crater rim or from holes in the ground within the craters [Petersen et al., 2009]. Meter-wide bacterial mats grow at the northwestern part of Irina I, but animals are scarce and mainly consist of actinians and hydrozoans in the vicinity of the outer crater rim, while mussels are completely absent [Kuhn et al., 2004].
Four mussel habitats at Irina II were investigated in terms of the chemical energy supply from diffuse venting: (1) a colonized chimney on the top of the main structure (Site 1), (2) the lower end of the main mussel bed covering the western slope of the mound (Site 2), (3) an overgrown sulfide pillar on the main structure (Site 3), and (4) a small mussel patch at the base of the mound (Site 4). Additionally, two sites without mussels were investigated: a chimney wall on the main structure covered with alvinocarid shrimp (Site 5) and the rim of a smoking crater at the Irina I site (Site 6), 230 m southeast of Irina II [Borowski et al., 2008].
2.2. Methods for Measuring Sulfide in Situ
Two basic principles have been used over the last two decades to determine in situ sulfide concentrations in hydrothermal diffuse flow: colorimetry and electrochemistry [see also Le Bris and Duperron, 2010]. Colorimetric detection was initially based on conventional flow analysis [Johnson et al., 1986a; Massoth and Milburn, 1997; Sakamoto-Arnold et al., 1986] and was later replaced by flow injection analysis [Le Bris et al., 2000; Sarradin et al., 1999a; Vuillemin et al., 2009]. Colorimetry measures the total sulfide concentration (Stot), which is the sum of several different sulfide species such as dissolved H2S, HS−, and labile metal sulfides (Stot = H2S + HS− + FeS + Sx2-). At hydrothermal vents the concentration of labile metal sulfides can be quite high [German and Von Damm, 2006]. Although these metal sulfides can be exploited by some free-living sulfur-oxidizing microorganisms they cannot be used by symbiotic sulfur oxidizers of vent invertebrates, which dominate the biomass at deep-sea hydrothermal vents [Van Dover, 2000]. Therefore, total sulfide values do not reflect the sulfide species that are biologically relevant or bioavailable for symbiotic sulfur oxidizers, i.e., H2S and HS− (free sulfide or Sfree).
In contrast to colorimetric methods, electrochemical sensors can distinguish between different sulfide species. They relate an electrochemically induced change of the electric current between a working electrode and a counter electrode, both polarized at different potentials against a reference electrode, to the in situ concentration of the investigated sulfur species. Voltammetric sensors can detect and distinguish between free sulfide (Sfree = H2S + HS−), soluble iron-sulfide clusters (FeSaq), elemental sulfur (S0), polysulfides (Sx2-), thiosulfate (S2O32-), and tetrathionate (S4O62-) [Di Meo-Savoie et al., 2004; Luther et al., 2001a, 2001b, 2008; Mullaugh et al., 2008; Rozan et al., 2000; Waite et al., 2008]. In contrast, amperometric sensors detect sulfide exclusively in its undissociated form as dissolved gaseous hydrogen sulfide (H2S), but the concentration of HS− can be calculated if the pH is measured simultaneously. The principle of H2S detection is based on the ferricyanide/ferrocyanide ([Fe(CN)6]3-/[Fe(CN)6]4-) redox couple. H2S diffuses via a silicone membrane (permeable only for uncharged molecules, high gas permeability) and is oxidized by ferricyanide resulting in the formation of elemental sulfur and ferrocyanide. The latter is electrochemically reoxidized thereby creating a current that is directly proportional to the dissolved H2S concentration [Jeroschewski et al., 1996; Kühl et al., 1998].
The purpose of our study was to examine if free sulfide (H2S + HS−) is available in sufficient concentrations to support the thiotrophic symbionts of B. puteoserpentis mussels. As metal sulfide concentrations are relatively high at ultramafic-hosted hydrothermal systems [Douville et al., 2002; German and Von Damm, 2006], the colorimetric detection of sulfide would likely severely bias sulfide bioavailability results [Le Bris et al., 2003]. In contrast, both electrochemical methods are suitable for identifying the bioavailable portion of sulfides. Since in situ voltammetry was not available, we chose the amperometric approach along with pH detection. Amperometric H2S microsensors have not been used previously at deep-sea hydrothermal vents and were deployed in hydrothermal diffuse flow for the first time in this study.
2.3. Combined in Situ Measurements of Dissolved Hydrogen Sulfide, Dissolved Oxygen, and Temperature
In situ measurements of dissolved hydrogen sulfide, dissolved oxygen, and temperature were performed in 2004 during the Hydromar I cruise with the R/V Meteor (M60/3 [Kuhn et al., 2004]). All measurements were recorded for 2 to 5 min at the six different study sites (Figure 1 and Table 1). The sulfide, oxygen, and temperature sensors were connected to a custom-built in situ instrument (ISI). The ISI is identical to the autonomous microprofiler used previously in sedimented environments [de Beer et al., 2006; Wenzhöfer and Glud, 2002; Wenzhöfer et al., 2000] but here measurements were conducted semiautonomously on four dives using the ROV Quest (MARUM, Bremen, Germany; Table 1). The ISI was fastened horizontally on an extendable drawer underneath the ROV, with the sensor tips protruding from the drawer. To prevent damage, the sensors were surrounded by a protective cage made of titanium bars bolted on robust Teflon discs. The cage allowed seawater and diffuse fluids to pass freely to the sensors (Figures 2a and 2b). Power supply and data exchange between the instrument and the ROV was enabled via a SeaNet Cable/Connector System (Schilling Robotics, Davis, California). Data were recorded every second during the entire dive. During measurements the ROV remained in a fixed position and the ROV drawer with the instrument was horizontally extended so that the sensor tips were positioned as close as possible to the mussel beds (Figure 2b). Due to the technical setup and the horizontal ISI deployment mode, H2S and O2 microsensors were typically positioned between 5 and 13 cm above the beds while the temperature sensor was located between 7 and 11 cm above the mussels. All data were stored in the ISI internal memory as well as in the central ROV database and downloaded for analysis at the end of the dive.
All sensors were calibrated before each dive. Dissolved H2S was measured with amperometric H2S microelectrodes (response time (t90) < 0.5 s; stirring sensitivity < 2%; detection limit ∼ 2 μM) as described by Jeroschewski et al.  and Kühl et al. . The sensors were calibrated according to previously published methods in an anoxic phosphate buffer solution (pH 7.5, 200 mM) by stepwise addition of discrete amounts of a 1.3 M Na2S stock solution [de Beer et al., 2006; Wenzhöfer et al., 2000]. To determine concentrations of total dissolved sulfide [Stot] subsamples were taken, fixed in acidic 5% (w/v) zinc acetate solution, and stored at 4°C in the dark until further analysis. The resulting ZnS precipitate concentration was measured spectrophotometrically at 670 nm using the methylene blue method according to Cline . The H2S concentrations in the subsamples were calculated as described by Jeroschewski et al.  using equation (1) where pK1* is the negative decadal logarithm of the first dissociation constant of the sulfide equilibrium K1 corrected for salinity according to Millero et al. . The salinity of the phosphate buffer was 27‰ as determined with a refractometer.
To account for the varying in situ temperatures at the hydrothermal vent, the calibration was performed at least at two temperatures.
Dissolved O2 was measured with Clark type microelectrodes (response time (t90) < 1 s; stirring sensitivity < 1%) [Revsbech, 1989]. The sensors were calibrated in seawater bubbled with nitrogen gas for anoxic and air for oxic calibrations. The O2 concentrations for saturation were calculated according to Weiss . To account for the sensor's temperature sensitivity, the sensors were calibrated at two to three different temperatures.
Temperature was measured with a Pt100 stainless steel sensor (UST Umweltsensortechnik Geschwenda, Germany) at a resolution of 0.01°C. The sensor was calibrated at eight temperatures between 1 and 78°C by comparison with a commercial digital thermometer.
To relate the microsensor signals to the dissolved gas concentration, a calibration curve was calculated for each calibration temperature using linear regression based on the least squares method. To account for in situ temperatures, regression equations for any virtual temperature were calculated from both the slopes and intercepts computed for the calibration temperatures. From these equations the calibration formulas for temperatures from 2.5 to 27°C were calculated with a resolution of 0.01°C. These formulas were fine-tuned by relating them to in situ background sensor signals. The H2S calibration formulas were corrected for the sensor signal recorded outside the hydrothermal vent field where no H2S occurred. Likewise, the O2 microsensor signals recorded on the ocean floor outside hydrothermal activity were related to the corresponding O2 concentration and the calibration data accordingly adjusted. The O2 concentration of the nonhydrothermal bottom water was determined by Winkler titration on a ROV recovered water sample [Hansen, 1999]. All data sets were carefully inspected for outlying data by applying a filtering algorithm.
2.5. Long-Term Temperature Measurements
For long-term measurements of temperatures above mussel beds, a miniaturized temperature data logger (MTL; described by Pfender and Villinger ) was deployed immediately after the measurements with the ISI. The MTL was placed directly on top of the mussel bed for 130 h (5.5 days) at Site 2 (Figure 1) and monitored the temperature at 10 s intervals with 0.001°C resolution and an absolute accuracy of ±0.1°C.
For long-term measurements of temperature inside mussel beds at high temporal and spatial resolutions, sensor lances were designed. Eight temperature sensors were installed at 4 cm intervals over a total length of 28 cm inside a steel tube and connected to an autonomous data logger. Each sensor was calibrated to a total precision of ±0.005°C. These lances were deployed in 2005 during the Hydromar II cruise (R/V Meteor, M64/2) using the ROV Quest [Lackschewitz et al., 2005]. At Irina II, one such sensor lance was placed horizontally inside a mussel bed at the southeastern end of the mound in close proximity to the small active black smoker and a few meters away from the earlier ISI measurement at Site 2 (Figures 1 and 2c). The sensor lance was inserted approximately 10 cm underneath the top of the mussel bed and had an inclination of approximately 20° so that the tip of the lance was located about 20 cm below the top of the mussel bed. Temperatures were monitored at the 8 sensors simultaneously over a period of 250 days at one minute time intervals, resulting in a total of 2.8 million samples. All data were stored inside the data logger. After 20 months on the seafloor the loggers were recovered in 2007 during the Hydromar III cruise (R/V Maria S. Merian, MSM 04/3 [Borowski et al., 2007]) using the ROV Jason II (Woods Hole Oceanographic Institute, Figure 2d).
During four ROV dives, each of which explored the Logatchev vent field for approximately 6 h, data from 6 sites were collected at time intervals of one second (Tables 1 and 2). The bottom water temperature at approximately 3,000 m water depth outside of the area of hydrothermal influence was 2.6°C with a corresponding mean dissolved oxygen concentration of 232 μM. The highest temperature of 26.8°C was measured at the mussel-free Irina I site close to the rim of a smoking crater (Site 6). Correspondingly, the highest concentrations of dissolved H2S and the lowest O2 concentrations were measured at this site (Figures 3 and 4).
Table 2. Mean, Minima, and Maxima of Distinct in Situ Measurements at Mussel Populations
Ambient mean seawater oxygen concentration was 232 μM.
Minimum and maximum values of dissolved O2 and H2S are defined as the 3% and 97% percentiles to account for the core data of the data set and to eliminate outlying data peaks. Med, median; mean is given ± standard deviation; S1 and S2, H2S sensors 1 and 2.
Compared to ambient seawater, physicochemical conditions approximately 10 cm above mussel beds at Irina II (Site 1–5) were generally characterized by elevated temperatures, with slightly reduced oxygen and increased hydrogen sulfide concentrations (Figure 4 and Table 2). A temperature maximum of 7.4°C concurrent with a maximum of dissolved H2S (156 μM) and a minimum of dissolved O2 (142 μM) was recorded at Site 1A over mussel beds at a chimney base (Figure 5a). Similar conditions were found at Site 2 at the lower edge of an extended mussel bed with temperature and H2S maxima of 6.0°C and 94 μM, respectively (Figure 5b). There was less variability in temperature and oxygen and hydrogen sulfide concentrations at the two other mussel beds investigated at Sites 3 and 4. Above a pillar-like structure densely overgrown with mussels (Site 3) and a small mussel patch (Site 4) temperatures did not exceed 3.9°C and 3.4°C, respectively, while the highest recorded concentrations of dissolved H2S were 64 μM and 66 μM, respectively (Figures 5c and 5d). In comparison, at a chimney wall populated with alvinocarid shrimp but free of mussels (Site 5) the ISI recorded a temperature of 3.3°C and 48 μM dissolved H2S (Figure 5e).
Temperatures recorded with the ISI generally fluctuated noticeably within seconds at all investigated sites (Figures 3 and 5) with local minima close to ambient seawater. Mean temperatures never exceeded 3.0 ± 0.4°C (chimney base at Site 1) and 3.5 ± 0.7°C (mussel bed at Site 2) and were only 2.7 ± 0.1 to 2.8 ± 0.2°C at Sites 3–5. Mean dissolved H2S concentrations ranged from 9 ± 7 μM to 31 ± 13 μM at these sites. The mean O2 concentrations at all sites were between 216 ± 16 μM and 228 ± 9 μM (Figure 4 and Table 2).
3.2. Correlation of Temperature With H2S and O2 Based on ISI Microhabitat Measurements
The H2S microsensors used in this study did not respond ideally to changes in H2S concentrations (i.e., in the order of seconds) and returned only slowly to the baseline after they were removed from the mussel habitats (i.e., in the order of 5 to 45 min depending on the physicochemical dynamics of the investigated site). Nevertheless, their response was fast enough to observe a clear time synchronicity between increases in temperature and dissolved H2S concentrations and to detect positively correlated sulfide-temperature ratios (Figures 3 and 5). H2S-T ratios were in the range of 2 to 60 μM H2S/°C and increased threefold to fourfold when the regression line was forced through 2.6°C, the temperature of the sulfide-free ambient seawater. However, H2S-T ratios had only weak determination coefficients (R2 ≤ 0.31, forced regressions R2 = 0.00 in all but one cases) and were thus only very weakly supported (Figures 3 and 5 and Table 2). Therefore, no reliable sulfide-temperature ratios could be retrieved from the data. The response of the O2 sensors was faster so that their profiles exactly mirrored temperature profiles, as visible in a strongly negative (−6 to −47 μM O2/°C) and linear (determination coefficient R2 = 0.40–0.83) correlation of O2 with temperature (Figure 5). The result of the linear regressions between O2 and temperature (i.e., the O2-T ratio) was variable and depended on the range of temperature variation: the narrower the temperature range the more negative the ratio (Figure 6). For the sulfide pillar at Site 3 (2.6–3.9°C), the mussel patch at Site 4 (2.6–3.4°C), and the chimney wall at Site 5 (2.6–3.3°C), the O2-T ratios were similar and in the range of −47 to −35 μM/°C. The O2-T ratios for the chimney base at Site 1 (2.8–7.4°C), the mussel bed at Site 2 (2.7–6.0°C), and the rim of the smoking crater at Site 6 (2.6–26.8°C) were less negative and ranged from −24 to −6 μM O2/°C (Table 2).
3.3. Long-Term Temperature Measurements
Following the ISI deployment at Site 2, a miniaturized temperature logger (MTL) was placed on top of the mussel bed at the same site (Figure 5b) and recorded the temperature for 5.5 days. During the initial 4 days of the MTL deployment we recorded a mean temperature of 4.1°C ± 0.5°C (total range from 2.8°C to 6.3°C). This is consistent with the ISI temperature recordings at this site (Figure 4). After the fourth day the temperature suddenly dropped to a mean of 3.0°C ± 0.1°C with a maximum of 3.4°C and a minimum of 2.7°C and remained at this temperature until the end of recording (Figure 7). Based on the ISI data set from Site 2 and the resulting oxygen to temperature ratio of −6 μM O2/°C, the oxygen concentration above the mussels at this site was calculated from the MTL temperature recordings to have a mean of 225 ± 4 μM with a minimum of 210 μM at 6.3°C and a maximum of 232 μM at 2.7°C (Figure 7).
After the ISI and MTL deployments, an 8-channel temperature lance was pushed horizontally into a mussel bed in close proximity to the preceding ISI and MTL measurements at Site 2 (Figures 1, 2c, and 8). The sensor lance was inserted approximately 10 cm into the mussel bed at an angle of approximately 20° so that the tip of the lance was located about 20 cm below the surface of the mussel bed. Over a length of 16 cm (corresponding to a vertical depth of ∼10–16 cm) the temperature lance recorded temperature fluctuations between 2.7°C and 15.8°C, however, 96% of all data (3% percentile to 97% percentile) were in the range of 3.0°C to 7.1°C. At the tip of the lance at 28 cm (i.e., 20 cm below the mussel bed surface) a temperature maximum of 18.2°C was recorded and the 3% to 97% percentiles were in the range of 3.3 to 16.0°C. Correspondingly, the mean temperatures increased from 3.8 ± 0.6°C to 4.4 ± 1.1°C 10–16 cm below the mussel bed surface, while higher mean temperatures of 8.3 ± 2.6°C and 10.9 ± 3.4°C were recorded 18 and 20 cm below the mussel bed surface (Figure 9 and Table 3). This indicates that the deeper mussel layers at the tip of the sensor lance were closer to the source of fluid flow and illustrates the small-scale spatial gradients that exist in mussel beds.
Table 3. Mean, Minima, and Maxima of Long-Term Temperature Measurements 10–20 cm Inside a Mussel Bed Over a Horizontal Distance of 28 cm
Vertical depth is approximate and relative to channel T8 at ∼10 cm inside the mussel bed. (The sensor lance had an inclination of about 20°. Therefore, the tip of the sensor lance (channel T1) was deeper inside the mussel bed than the end of the lance (channel T8)). SD, standard deviation.
The ISI recordings at the Irina II site of the Logatchev hydrothermal vent field revealed mean dissolved H2S concentrations between 9 μM and 31 μM, mean dissolved O2 concentrations between 216 μM and 228 μM, and mean temperatures between 2.8 and 3.5°C approximately 10 cm above the mussel beds (Figure 4 and Table 2). These data show that the mussels at Logatchev have simultaneous access to both dissolved H2S and oxygen and indicate that their diffuse flow habitats provide sufficient concentrations of electron donors and acceptors for the mussel symbionts to gain energy through aerobic sulfide oxidation. The coexistence of H2S and oxygen in diffuse flow areas is in agreement with previous observations [Johnson et al., 1986b, 1988b; Luther et al., 2008; Mullaugh et al., 2008; Nees et al., 2009, 2008; Podowski et al., 2009, 2010] and does not contradict sulfide oxidation kinetics per se. The half-life for free sulfide oxidation (H2S + HS−) in air saturated seawater is about 26 h at 25°C and pH 8.0 [Millero et al., 1987] and increases toward lower temperatures and lower oxygen concentrations to about 380 h at 2°C, pH 7.8, and 110 μM O2 [Johnson et al., 1994]. In fact, the reaction of H2S with O2 is thermodynamically unfavorable at all pH so both species can coexist [Luther, 2010]. Clearly, sulfide oxidation kinetics is significantly accelerated in the presence of metal ions which act as strong catalysts. For example, Snavely and Blount  and Chen and Morris  found that metal ions such as Ni2+, Co2+, Mn2+, Cu2+, Fe2+, all of which are abundant in Logatchev end-member fluids [Schmidt et al., 2007], increase the sulfide oxidation rate 45 to 100 fold. However, molecules must meet in order to react and catalysts must be present at exactly the same time and place to catalyze the reaction efficiently. At 2.5 mM H2S in the end-member fluid and metal ions in micromolar concentrations (except for ferrous iron, see below) [Schmidt et al., 2007], sulfide oxidation at Logatchev is apparently not fast enough to cause the complete removal of free sulfide. Thus, metal ion concentrations at Logatchev are not high enough to prohibit the coexistence of H2S and O2 in Bathymodiolus mussel beds. In brief, sulfide oxidation kinetics predicts that H2S and O2 can potentially coexist even in the presence of metal ions. Notably, in Logatchev mussel beds H2S and O2 are not mutually exclusive.
The wealth of dissolved H2S in Logatchev diffuse flow habitats is in contrast to the extremely low free sulfide concentrations at Rainbow. At Rainbow, in situ measurements in diffuse fluids dominated by the shrimp Rimicaris exoculata had total sulfide concentrations (H2S + HS− + FeS) below 5 μM [Schmidt et al., 2008]. This may not be surprising as fluids in the ultramafic-hosted Rainbow hydrothermal system are extremely rich in ferrous iron (Fe2+) and precipitate most of the sulfide [Le Bris and Duperron, 2010]. Indeed, Rainbow high temperature fluids have an Fe-H2S ratio of 24 [Charlou et al., 2002; Douville et al., 2002] and the Fe-H2S ratio exceeds 30 in the shrimp environment [Schmidt et al., 2008]. In the immediate surrounding of mussels, the iron concentration can exceed 100 μM and iron sulfide almost completely dominates sulfide speciation so that free sulfide is below detection levels at Rainbow [Le Bris and Duperron, 2010]. In contrast, ferrous iron concentrations at Logatchev are in the range of those reported from basalt-hosted vent systems on the MAR and the Fe-H2S ratio is about 1 [Le Bris and Duperron, 2010; Schmidt et al., 2007]. This indicates that sulfide (H2S + HS−) may not be fully precipitated by Fe2+. Our data clearly show that dissolved H2S is abundant in Logatchev diffuse fluids and imply that H2S is neither fully precipitated nor otherwise completely removed by ferrous iron.
In contrast to our in situ measurements, shipboard analyses of diffuse fluids after discrete sampling indicated fivefold to twelvefold lower sulfide concentrations in Irina II diffuse fluids. Schmidt et al.  reported a maximum of 6.0 μM free sulfide (H2S + HS−) which corresponds to 2.7 μM dissolved H2S at their reported pH (7.0) and an assumed in situ temperature of about 3.0°C [Jeroschewski et al., 1996; Millero et al., 1988]. This discrepancy between in situ and shipboard measurements after discrete sampling is most likely due to sulfide oxidation and/or precipitation during sample recovery, and has also been observed by others [Le Bris et al., 2006a; Schmidt et al., 2008].
As the measured sulfide species in this study was dissolved H2S our data are not directly comparable to free sulfide (H2S + HS−) or total sulfide (H2S + HS− + FeS) concentrations reported from basalt-hosted systems. However, concentrations of free sulfide can be calculated from dissolved H2S concentrations if the pH, salinity, and temperature are also known [Jeroschewski et al., 1996; Millero et al., 1988]. In situ measurements of pH and salinity were attempted in this study using pH microsensors and a conductivity sensor but failed. Therefore, we can only estimate the free sulfide concentrations of Logatchev diffuse fluids: At mean temperatures of 2.8–3.5°C and an assumed pH range of 6.0 to 8.0 over Bathymodiolus mussel beds [De Busserolles et al., 2009; Sarradin et al., 1999b], mean dissolved H2S concentrations of 9 to 31 μM correspond to mean free sulfide concentrations of 10–400 μM (pH 6.0: 10–35 μM, pH 6.5: 13–43 μM, pH 7.0: 20–69 μM, pH 7.5: 45–150 μM, pH 8.0: 122–408 μM). At basalt-hosted systems, free sulfide concentrations above mussel beds ranged from nondetectable to 87 μM, with mean concentrations of 27 μM [Lutz et al., 2008; Nees et al., 2008] (Table 4). Our estimates therefore indicate that free sulfide concentrations at the Logatchev vent field are in the same range as those reported from basalt-hosted vents.
Table 4. In Situ Sulfide, Oxygen, and Temperature Conditions of Various Hydrothermal Diffuse Flow Habitats in Comparison
H2S + HS− (μM)
H2S + HS− + FeS (μM)
Minimum and maximum values of dissolved H2S and O2 are defined as the 3% and 97% percentiles to account for the core data of the data set and to eliminate outlying data peaks.
The mean temperatures at the Logatchev hydrothermal vent field (Irina II site) ranged from 2.8 ± 0.1°C to 3.5 ± 0.7°C above mussel beds and from 3.8 ± 0.6°C to 10.9 ± 3.4°C inside mussel beds, which is in agreement with temperature means reported from mussel beds at other hydrothermal vent fields. Mean temperatures between 5.7°C and 10.0°C have been reported for B. azoricus from Lucky Strike vent field on the slow spreading MAR [Desbruyères et al., 2001; Vuillemin et al., 2009]. Johnson et al. [1988a, 1994] reported mean temperatures between 3.0°C and 5.5°C above and inside clumps of B. thermophilus from the basalt-hosted Rose Garden vent field on the intermediate spreading Galapagos Rift whereas Moore et al.  found the same species to prosper at average values between 1.6°C and 9.6°C at the fast spreading East Pacific Rise (9°50′N). Mean values between 5.5°C and 10.5°C have been reported from mussel beds in the Lau Back-Arc Basin [Podowski et al., 2009, 2010] (Table 4).
While the range of short-term fluctuations indicates a species' tolerance toward these variations, average values of environmental conditions reflect the general physicochemical situation to which a given species is exposed during part or most of its lifespan and are thus valuable environmental descriptors, particularly for hemisessile organisms such as bathymodiolin mussels. The long-term average temperatures at which bathymodiolin mussels have been regularly found rarely exceed 10°C (Table 4), although occasionally long-term average temperatures between 11°C and 17°C have been reported [Chevaldonné et al., 1991; Desbruyères et al., 2001; Moore et al., 2009]. Interestingly, in respirometry experiments under nonoxygen limiting conditions, bathymodiolin mussels were able to survive sustained exposure to temperatures as high as 18°C [Henry et al., 2008]. This indicates that the distribution of bathymodiolin mussels at hydrothermal vents is influenced more by their oxygen requirements than by their temperature tolerance. Correspondingly, mussels at Logatchev were generally exposed to temperatures below 10°C (Table 4), and extrapolation of temperature to an oxygen concentration of zero using the oxygen-temperature relationship observed at the rim of a smoking crater (Site 6) indicates that diffuse flow habitats at Logatchev become anoxic around this temperature (Figure 3).
4.3. Correlation of Temperature With H2S and O2 Based on Microhabitat Measurements
Johnson et al. [1988a] showed that temperature can be used to estimate sulfide and oxygen concentrations in diffuse vent fluids. Temperatures of 2 to 11°C correlated positively with total sulfide (H2S + HS− + FeS) and negatively with oxygen although the relationship was not always linear due to biological consumption of these compounds [Johnson et al., 1988a]. Le Bris et al. [2003, 2006a, 2006b] confirmed the linear relationship between sulfide and temperature, but pointed out that sulfide to temperature ratios are only constant at a given site and can be highly variable between sites. Temporal and spatial variability of sulfide-temperature ratios has recently been demonstrated using in situ voltammetry [Nees et al., 2009, 2008]. Our data show that sulfide is positively correlated with temperature, however because of the slow response times and signal drifts of our H2S microsensors, we were not able to reliably estimate sulfide-temperature ratios at Logatchev.
Our study is the first in which in situ oxygen concentrations of fluids from an ultramafic-hosted hydrothermal vent were measured. O2 profiles from all sites in the Logatchev vent field consistently mirrored temperature profiles and showed a strongly inverse linear correlation of O2 with temperature. The ratio between oxygen and temperature varied between sites and ranged from −6 to −47 μM/°C. Temporal and spatial variability of oxygen-temperature ratios has also been shown in diffuse fluids of basalt-hosted hydrothermal vents using in situ voltammetry [Moore et al., 2009; Nees et al., 2008]. There are several explanations for this variability. First, bathymodiolin mussel beds laterally divert the flow of diffuse vent fluids, thereby diluting the fluid and its components as it moves through the mussel beds by variable mixing with seawater [Johnson et al., 1994, 1988b]. Second, mussels at Logatchev are likely supplied by multiple sources of diffuse venting from many small cracks in the ocean floor [Petersen et al., 2009] which may also differ in their degree of dilution with ambient seawater. Third, deep ocean currents, tides, and local topography affect the flow of diffuse fluids and thus influence its relative composition over space and time [Chevaldonné et al., 1991; Johnson and Tunnicliffe, 1985; Little et al., 1988].
Although a few studies have examined the relationship between temperatures of diffuse fluids and either their oxygen or their sulfide contents, little in situ data are available on the correlations between the three key variables temperature, oxygen, and sulfide at different vents and their effect on vent communities. Johnson et al. [1988a, 1994] showed that diffuse fluids at Rose Garden on the Galapagos Ridge became anoxic above 11°C. At this temperature sulfide concentrations (Stot) were above 100 μM but never exceeded 200 μM. In contrast, diffuse fluids in the Lau Basin with average temperatures of 7–25°C were well oxygenized (60–150 μM) and contained comparably little sulfide (Sfree = 8–70 μM) ([Podowski et al., 2010] based on 332 survey locations containing chemosymbiotic macrofauna). In our study, Bathymodiolus mussels were generally found at mean temperatures below 6°C, so that these experienced oxygen concentration between 100 and 200 μM (depending on the site) and free sulfide concentrations of 10 to 400 μM (depending on the pH).
At Logatchev, fluids became anoxic at around 10°C. The long-term average of temperatures inside Logatchev mussel beds were generally far below this value (Figures 8 and 9 and Table 3) indicating that over the monitored period of 250 days the mussels were neither oxygen limited nor overexposed to free sulfide. Noticeably, mussels were also not limited by sulfide availability as free sulfide was always present above the mussel beds, i.e., had not been completely chemically oxidized or biologically depleted during the ascent of the vent fluids through the mussel bed. Our data show that the Logatchev mussels have simultaneous access to both free sulfide and oxygen. Their diffuse flow habitats thus provide sufficient concentrations of electron donors and acceptors for the mussel symbionts to gain energy through aerobic sulfide oxidation.
Our data confirm previous studies that physicochemical conditions in hydrothermal diffuse flow habitats can be highly variable at very small temporal and spatial scales. It is becoming increasingly apparent that each individual mussel lives in a temporally changing microenvironment that can differ in its physicochemical properties from the microenvironments of neighboring mussels. Direct in situ characterization of this microhabitat heterogeneity is highly challenging but worthwhile as it can reveal how spatial and temporal gradients in diffuse fluids affect the vent biota and thus contributes to our understanding of geosphere-biosphere interactions at hydrothermal vents.
We gratefully acknowledge the chief scientists Thomas Kuhn, Klas Lackschewitz, and Christian Borowski of the involved research cruises as well as the crews of the research vessels Meteor and Maria S. Merian and the ROV's Quest and Jason II for their invaluable support. We thank Volker Meyer, Paul Färber, and Harald Osmers for their commitment during last minute adaptations of the in situ instrument for use with the ROV Quest. Ursula Werner and Eva Walpersdorf are acknowledged for their introduction to working with microsensors. We appreciate Gabriele Eickert who skillfully constructed the microelectrodes and gave indispensable comments for their deployment in the field. We are grateful to two anonymous reviewers who gave valuable comments for the improvement of this paper. This work was supported by the German Science Foundation (DFG) Priority Program 1144 “From Mantle to Ocean: Energy-, Material- and Life Cycles at Spreading Axes” (publication 61), the DFG Cluster of Excellence “The Ocean in the Earth System” at MARUM, Bremen, and the Max Planck Society.