Introduction

Microbial mats consist of finely laminated layers of phylogenetically and metabolically diverse microbial communities. Throughout most of Earth's history, the metabolic activity of such mats likely dominated biogeochemical cycling, impacting nutrient and oxygen availability (Hoehler et al., 2001; Anbar and Knoll, 2002; Hayes and Waldbauer, 2006). The abundance of microbially laminated sedimentary strata attests to the ubiquity of such mats and highlights their potential to preserve biosignatures over geologic time (Grotzinger and Knoll, 1999). Although the laminated architecture of microbial mats is thought to result from strong spatial gradients in light intensity and redox in their uppermost few millimeters, there is evidence for ongoing, localized metabolic activity at depth (Visscher et al., 2000; Dupraz et al., 2004; Wieland et al., 2005). Despite many decades of investigation, the relationship between microbial activity (on the micron scale) and the resulting isotopic geochemical profiles remains poorly understood. A refined understanding of how these are linked will improve our ability to understand modern biogeochemical cycling and to interpret the biological significance of geochemical patterns in the rock record.

We have focused our study on sulfur cycling, in particular sulfate reduction, one of the dominant metabolisms within microbial mats (Canfield and Des Marais, 1993; Teske et al., 1998; Baumgartner et al., 2006). Mat-associated sulfate-reducing bacteria (SRB) were initially considered to be strict anaerobes residing below the oxic/anoxic chemocline, which was inhabited by sulfide oxidizers and presumably sulfur disproportionators that catabolize intermediate valence compounds (S0, S2O32−, SO32−) (Habicht et al., 1998). The discovery of abundant SRB and elevated sulfate-reduction rates in the oxic surface layers of mats indicates a more complicated spatial organization of the different sulfur-cycling metabolisms (Frund and Cohen, 1992; Canfield and Des Marais, 1993; Krekeler et al., 1997; Minz et al., 1999a; Sigalevich et al., 2000; Wieringa et al., 2000; Baumgartner et al., 2006). Indeed, among SRB, there is evidence for vertical segregation between different phylogenetic clades in microbial mats (Risatti et al., 1994; Minz et al., 1999b; Decho et al., 2005; Wieland et al., 2005). However, standard geochemical and microbiological techniques lack the spatial resolution to fully document the degree of SRB segregation and its biogeochemical consequences (Risatti et al., 1994), particularly given that lateral variability within these environments often makes one-dimensional (1D) depth studies problematic (Frund and Cohen, 1992).

The ideal geochemical tool for characterizing sulfur cycling in these environments would be able to determine not only the abundance but also the (natural) isotopic composition of relevant sulfur species and their 2D spatial distribution. Isotopic data are essential for separating the different pathways of sulfur cycling (oxidation, reduction and disproportionation), as each of these can be associated with different isotopic fractionations (Habicht et al., 1998). Most importantly, isotopic records can be preserved over geological time, providing the means to relate modern systems to the rock record (Canfield and Teske, 1996; Canfield, 2001; Habicht et al., 2002; Kampschulte and Strauss, 2004; Hurtgen et al., 2005; Fike et al., 2006; Gill et al., 2007). Ideally, to understand the intricacies of microbial interactions involved in sulfur cycling, measurements of sulfur species' concentration and isotopic composition should be made on the spatial scale most relevant to microbial activity: that of the micron-sized microorganisms whose various metabolisms drive the sulfur cycle.

The aim of this study was to develop such a technique using high-resolution secondary ion mass spectrometry (nanoSIMS) to analyze sulfide, coupling high-spatial (submicron) resolution with the ability to simultaneously measure sulfide abundance and isotopic composition. We compared microelectrode and nanoSIMS-acquired microgeochemical data to the spatial distribution of SRB within these mats using catalyzed reporter deposition fluorescence in situ hybridization (CARD-FISH) (Pernthaler et al., 2002). Together, these measurements constitute a paired microbiological and microgeochemical approach that can be applied to correlate the micron-scale variations in microbial community and metabolic structure with the variations in the abundance and isotopic composition of sulfide produced during microbial sulfur cycling.

In this study, we analyzed the well-characterized Microcoleus-dominated benthic microbial mat communities from Exportadora del Sal, SA salt works at Guerrero Negro, Baja California Sur, Mexico. Previous studies of these mats have characterized the bulk microbiological community structure (Risatti et al., 1994; Spear et al., 2003; Ley et al., 2006), large-scale sulfur cycling (Canfield and Des Marais, 1993; Jorgensen, 1994) and daily variations in response to the periodic redox forcing from photosynthesis (Jorgensen, 1994; Visscher et al., 2003; Decker et al., 2005). Microcoleus-dominated mats also support active sulfate reduction by aerotolerant sulfate reducers (Teske et al., 1998) in the photosynthetically active uppermost few millimeters (Canfield and Des Marais, 1993). Analyses in the present study focused on the region around the oxic/anoxic chemocline (∼2 mm depth during the day) to understand the changes in sulfur cycling that occur at and across this important geochemical and biological transition zone.

Materials and methods

Site description and sample collection

Push core samples of laminated Microcoleus mats (∼4- to 5-cm thick, see Supplementary Figure S1) were collected in April 2006 from a hypersaline pond (salinity of ∼90 p.p.t.), known as pond 4 near 5 (P4n5), associated with Exportadora De Sal, SA in Guerrero Negro, Baja California Sur, Mexico. These hypersaline mats are characterized by high levels of sulfate (∼80 mM) throughout the mat and in the overlying water. P4n5 mats are permanently submerged in approximately 1 m of pond water with a seasonal in situ temperature ranging from 14 to 29 °C. A detailed description of the area and mat geochemistry is provided in Des Marais et al. (1989) and Hoehler et al. (2001).

Cores consisting of mat and underlying sediment were transported back to the laboratory and maintained in a tank filled with water from pond 4. Silver-disc incubations and microelectrode measurements were conducted in the laboratory on mats maintained within the core liner (overlain with approximately 6–7 cm of pond water).

Microelectrode profiles of oxygen, hydrogen sulfide and pH

Vertical profiles of oxygen, hydrogen sulfide and pH were measured in intact mat cores under light exposure using microelectrodes (Unisense, Aarhaus, Denmark). The oxygen sensor was a Clark-type amperometric electrode with a built-in reference and a guard cathode (Unisense OX10; Revsbech, 1989) with a tip diameter of 20 μm, a stirring sensitivity of <1% and a 90% response time ⩽1 s. The electrode currents have a linear response to oxygen dissolved in water. A two-point calibration was performed by inserting the oxygen electrode into a calibration chamber filled with water from the sampling site, which was purged with air (100% saturation) or nitrogen (0% saturation) at the in situ temperature. The H2S microsensor used in the analysis was a miniaturized amperometric sensor with an internal reference and a guard electrode (Unisense H2S25; Jeroschewsky et al., 1996; Kuehl et al., 1998). The sensor had a long (5 cm) tapered tip with a tip diameter of 20 μm. The response time was <2 s. Total sulfide concentrations were calculated using pH values measured in tandem with the H2S profiles using a pH microsensor (25 μm tip diameter, Unisense pH25). Calibration of the H2S microsensor was performed with anoxic Na2S standards of known concentrations measured by colorimetric analysis (Cline, 1969). The pH electrode was calibrated using standard pH buffer solutions (VWR).

The electrodes were attached to a micromanipulator and moved vertically into the mat via a computer-controlled motor (Unisense MC18011). Amperometric sensors were connected to a high-sensitivity picoammeter (Unisense PA2000) and the cathodes were polarized against an internal reference. The pH sensor was connected to a high-impedance mV meter (Unisense PHM210). Amperometric signals were amplified and transformed to mV by the picoammeter, and data were collected directly on a computer using the software Profix (Unisense). Measurements were performed in vertical increments of 100 μm for oxygen and 200 μm for H2S and pH. The position of the microsensor in relation to the mat–water interface (0 position) was determined visually using a dissecting microscope (Leica Microscope Systems AG, Wetzlar, Germany). Profiles were performed under full light exposure using a cold light source (McBain Instruments, Chatsworth, CA, USA), photon flux as a combination of ambient day light in the lab and direct exposure of the mat surface to the light source was measured as 270 μM s−1 m−2.

Catalyzed reporter deposition fluorescence in situ hybridization

Samples of mat to be used for fluorescence in situ hybridization were fixed with freshly prepared paraformaldehyde (4%) and incubated for either 2 h at room temperature or overnight at 4 °C. After fixation, mat samples were transferred into 1 × phosphate-buffered saline (PBS) solution (pH 8.0), incubated for 1 h at room temperature and washed another two times with PBS to remove traces of paraformaldehyde (Pernthaler et al., 2004). Washed samples were stored in a 1:1 (v/v) mixture of PBS and ethanol at −20 °C. Whole mat sections were embedded in Steedman's wax (Steedman, 1957) and thin-sectioned into 10-μm slices using a microtome (Leica Microscope Systems AG) according to Pernthaler et al. (2002). To compensate for the autofluorescence and high-background signal inherent in photosynthetic mats, a modified CARD-FISH signal-enhancement procedure was applied to visualize SRB populations. The final optimized protocol included dehydrating the mat sample in 100% ethanol (2 min), followed by two separate incubation steps to inactivate endogeneous peroxidases: 10 min in 0.1 N HCl and 30 min in 0.1% hydrogen peroxide (diluted from 30% with 1 × PBS). Each of these incubations was followed by a second wash step in nanopure water and ethanol dehydration. To increase cell permeability, samples were incubated in TE buffer (10 mM Tris HCl; 1 mM EDTA, pH 9.0) for 3 min in a hybridization microwave oven (Microwave Research Applications Inc., Laurel, MD, USA) at 60 °C set at maximum power. The pre-conditioned mat sections were then washed in nanopure water and hybridized in 300 μl of the CARD-FISH hybridization buffer containing 50% formamide and 1 μl of 50 ng μl−1 horseradish peroxidase-labeled oligonucleotide probe (Biomers, Ulm, Germany) following the protocol of Pernthaler et al. (2004). Hybridizations were carried out in a sealed slide chamber in the hybridization microwave at 46 °C for 30 min. A 15-min post-wash in PBS at room temperature followed the hybridization. Tyramide amplification was found to be optimal using a 1:1000 dilution of the fluorescein-tyramide (Invitrogen, Carlsbad, CA, USA). Amplification included a 15-min incubation at 37 °C. After fluorescent amplification, slides were washed in series using PBS (10 min), nanopure water (2 min) and ethanol (2 min), and then air-dried in the dark prior to viewing. The oligonucleotide probe used in this study was the Desulfobacteraceae-specific probe DSS658 (probebase accession no. pB-00084) (Loy et al., 2007). The probe contained no mismatches with 13 GenBank sequences and a single peripheral mismatch to two additional GenBank sequences previously recovered from the Guerrero Negro mats (Supplementary Table S1). Formamide gradients in increments of 5% were tested during the microwave CARD-FISH hybridization. The optimal formamide concentration was consistently found to be 10% lower in microwave hybridizations than in oven hybridizations for all probes tested. As such, a 50% formamide concentration was used in the hybridization buffer rather than the 60% reported in Manz et al. (1998).

Slides were examined on a DeltaVision RT deconvolution microscope system (Applied Precision, Ithaca, WA, USA) with either a × 20, × 40 or × 60 Olympus objective (PlanApo or UPlanApo). Images were captured and processed using Softworx package (version 3.5.1; Applied Precision) and exported as TIFFs for figure assembly in Adobe Photoshop CS (Adobe Systems Incorporated).

Sulfide capture using silver discs

Dissolved sulfide was captured as silver sulfide (Ag2S) by incubating either silver foil (Figure 1a) or a metallic silver disc (2.5 cm diameter) in the microbial mat (see Supplementary Figure S1). This method is an adaptation of a technique previously used in 35S-radiolabel studies (Visscher et al., 2000; Dupraz et al., 2004; Wieland et al., 2005), where silver foil was coated with 35SO4. After incubation, radiography of the precipitated silver sulfide documented the presence of any 35S in the silver sulfide, which was attributed to sulfate reduction during the incubation (Visscher et al., 2000). In contrast, the approach here does not involve the addition of any isotopic spike to the microbial mat and preserves the natural isotopic composition of sulfide in these systems. Precipitated sulfide is a combination of ambient sulfide present when the disc was inserted and that produced by biological sulfate reduction and subsequently diffused to the disc during the incubation.

Figure 1
figure 1

Sulfide laminations within microbial mats maintained in the laboratory. (a) Millimeter-scale banding visible on silver foil. Glass disc is 2.5 cm. (b) Optical microscopic image of a silver disc. Bright material at top is metallic silver; brown is Ag2S. A transect of nanoSIMS spot analyses (white dots) is visible (boxed). Isotopic results from these points are seen in Figure 2. (c) NanoSIMS optical image showing collection of several 32S ion images (6 μm × 6 μm). The box corresponds to the ion image in (e). Note parallel laminations in sulfide aligned with the mat surface orientation. (d–g) 32S images from mat incubations. (d, e, g) Fine-scale banding parallel to the mat surface, whereas (f) shows irregular patchiness. (e–g) δ34S values overlain on top of their 32S images, mapping the isotopic composition of distinctive 32S regions. No significant deviation in δ34S is observed in (e) and (f), whereas (g) has ∼25‰ variability in localized regions. (h) Image of the precipitated standard, showing small-scale irregularities in 32S and a 9‰ range in δ34S. Color scale for images are as follows (in 32S counts): (d) 0–236; (e) 0–577; (f) 0–7, 627; (g) 0–2, 116; (h) 0–26, 031.

Full size image

Silver-disc incubations were conducted for 12 h on two different P4n5 mat cores under full illumination using a cold light source (McBain Instruments) providing a light intensity of ∼270 μM s−1 m−2. Upon exposure to light, signs of active photosynthesis were observed, including active oxygen production in the upper 1–2 mm (measured by microelectrode) and the formation of gas bubbles on the surface of the mat. The temperature during incubations was ∼23 °C. For mat no. 2, microelectrode measurements were acquired prior to disc incubations and samples for FISH microscopy were collected adjacent to the disc immediately following the incubation under full light exposure. In both mats, less than ∼1‰ of pore water sulfate was reduced, ensuring open system behavior (Habicht and Canfield, 1996).

Visible sulfide precipitation in both of these laboratory experiments first occurred approximately 0.4 mm below the mat surface with trace levels detectable above by the nanoSIMS. The top of this visible sulfide accumulation serves as the origin in Figures 2 and 3. Because of mat-surface irregularities (see Figure 4), there was some uncertainty in the absolute distance between the tops of the sulfide layer and the local mat surface.

Figure 2
figure 2

Example of δ34S profile with depth through the mat. The origin is set to the first visible layer of sulfide at ∼0.4 mm below the mat surface.

Full size image
Figure 3
figure 3

Plot of δ34S and 32S for two different incubation experiments (12 h during the day). Lateral distance is 450 μm in (a) and (b) and 500 μm in (c) and (d). Black dots on (a) and (c) indicate sampling frequency. Plot was made using linear interpolation between measured points. Error bars on averages indicate the standard deviation for measurements for a given depth.

Full size image
Figure 4
figure 4

Image of Microcoleus-dominated subtidal microbial mat from Guerrero Negro (left panel). Note irregularities in mat surface and laminated structure. Microelectrode depth profiles on the right (O2 and H2S), alongside nanoSIMS 32S counts of sulfide precipitated on the silver disc during incubation under active photosynthesis. Error bars for 32S counts represent the standard deviation from adjacent parallel sections (n=10).

Full size image

NanoSIMS analysis and isotopic calibration

The Cameca nanoSIMS 50L is a high-resolution secondary ion mass spectrometer that can collect up to seven ions simultaneously and operate in either cesium ion or oxygen ion mode. The nanoSIMS has previously been used to analyze sulfur isotopes in minerals (Slodzian et al., 2004; Hoppe et al., 2005) and carbon, nitrogen and sulfur distributions in biological samples (Clode et al., 2007; Li et al., 2008). For the present study, data were obtained using electron multiplier multi-collection, operating with the cesium (Cs+) ion source. An ∼40-pA beam was used and the sample surface was pre-sputtered for 120 s prior to analysis. Data were typically collected for 1000 s. 32S and 34S ions were measured simultaneously. This provides a means to calculate δ34S=(Rsample/Rstandard−1) × 1000, reported in permil (‰) relative to the international Vienna Canyon Diablo Troilite standard, where R is the 34S/32S ratio. Reference δ34S values for P4n5 sulfate were obtained by elemental analyzer-isotope ratio mass spectrometry analysis of BaSO4 precipitated from P4n5 water by the addition of a saturated BaCl2 solution.

Instrument mass fractionation was corrected by calibration to silver sulfides of known isotopic composition (Supplementary Figure S2). Two forms of standards were used: acanthite (Ag2S) mineral standards and precipitated silver sulfide. For nanoSIMS analysis, acanthite minerals were subsampled and embedded in epoxy and polished, whereas silver sulfide was precipitated onto silver discs by the addition of AgNO3 to a dissolved sulfide solution (10 mM Na2S) of known isotopic composition (‘precipitated standard’). For conventional analysis, another aliquot of sulfide solution was precipitated with AgNO3. The resulting silver sulfide precipitate was powdered, along with acanthite minerals. These samples were combined with excess V2O5 and combusted to SO2 and analyzed conventionally for δ34S using elemental analyzer-isotope ratio mass spectrometry (Studley et al., 2002). Standard deviations (1σ) were 0.1‰ for known standards and precipitated silver sulfide, and ∼1‰ for acanthite minerals, presumably due to inhomogeneities between subsampled crystals.

The analytic precision of the nanoSIMS for δ34S analysis on sulfide mineral standards ranged between 0.4‰ and 0.8‰ based on Poisson counting statistics with an internal standard error of 0.5‰–1.0‰. The standard deviation (external precision) from replicate runs was 0.6‰–2.0‰ (1σ). The analytic precision for the precipitated standards was 0.4‰ based on Poisson counting statistics with an internal standard error of 0.4‰–0.6‰. However, the external precision of the precipitated standards was ∼3.5‰ (1σ), which likely reflects both the limited abundance of silver sulfide on the disc surface, small topographic irregularities and heterogeneous deposition rates. This method is less precise than measuring bulk homogeneous crystals (that is, acanthite), but sufficient to resolve the >10‰ variations in sulfide δ34S that typically characterize microbial systems (Habicht et al., 1998).

The nanoSIMS beam ionized a considerable fraction of the silver sulfide during analysis of both precipitated standards and mat incubations (Figures 1b and c). Thus, for each measurement, the total sulfur yield decreased with time (see Supplemental Figure S3), precluding a steady-state counting rate for the secondary ion beam, as is routinely achieved for analysis of mineral specimens. However, there was no observed temporal trend in the isotope ratio of sulfide (Supplemental Figure S3), strongly suggesting that the decreasing yield did not cause an isotopic offset in the data. The nanoSIMS-generated data set was also examined for evidence of a quasi-simultaneous arrival (QSA) effect, which has been suggested to generate isotopic fractionations during sulfur isotope (and other) measurements made using electron multipliers on the nanoSIMS (Slodzian et al., 2004). Operating conditions (K<0.0025) indicate that any correction would have been minimal (<2‰), and no evidence for this effect was observed in analysis of >300 data points. As such, no correction was applied (see Supplementary text and Supplemental Figure S4).

Results

Ion images of captured sulfide from the mats

Data were collected using a Cameca nanoSIMS 50L in both ‘image’ and ‘spot’ mode on silver discs incubated in the microbial mats. In image mode, the beam was rastered over a 6 μm × 6 μm square (typically 256 × 256 pixels). The sulfur ion images collected in this manner from mat incubations preserve ultra-high resolution spatial variability that reveals laminations and patchiness in sulfide abundance on the micron-scale (Figures 1d–g), which differ in appearance from the precipitated standard (Figure 1h). The measured sulfur isotopic composition of the mat ion-image analyses ranged between +11‰ and −64‰. In two of the images (Figures 1e and f), there was no significant variability in δ34S within the image, particularly between regions of high and low sulfide abundance, whereas in Figure 1g, variability up to ∼25‰ was observed between some adjacent regions with no variability between others. In comparison, the precipitated standard had a 9‰ range in δ34S (Figure 1h). One of the difficulties in analyzing isotopic compositions of materials at this scale is the challenge of procuring standards that are isotopically homogenous on the submicron scale. Figure 1h demonstrates that our precipitated standard is not isotopically homogenous on the smallest scale we can examine. As such, errors reported here for image analyses (Figure 1) combine the error of the individual image analyses based on Poisson counting statistics, with the estimated 3.5‰ error associated with precipitation on the sulfide disc (see above). In the absence of an isotopically homogenous standard at this spatial resolution, these errors should be considered as an approximation of the true error.

Spot analyses of captured sulfide from the mats

NanoSIMS data were also collected using the instrument in spot mode, rastering the beam over a 6 μm × 6 μm area and spatially integrating the resulting data for increased isotopic precision (at the cost of spatial resolution finer than the spot size). Collection of spot measurements allows for the generation of 1D depth profiles in δ34S. At a spatial resolution of ∼50–100 μm, coherent trends in δ34S are apparent with depth (Figure 2). Enriched δ34S values (∼+5‰) characterized the uppermost sulfide zone associated with the photosynthetically active depth horizon. With increasing depth in the mat, sulfide δ34S dropped rapidly to −25‰ over 2 mm and coherent variations of up to 30‰ were observed down to a depth of 8 mm (Figure 2). Given a bulk starting sulfate composition of +20‰, these data correspond to sulfate–sulfide fractionations of 15‰–55‰.

The collection of 2D depth profiles provides a window into lateral, as well as vertical, variability in mat sulfur cycling (Figure 3). Here, spot measurements were collected every ∼50–200 μm to generate millimeter-scale 2D maps of sulfide concentration and δ34S isotopic composition through two different mats incubated during active photosynthesis. The first-order trend in both incubations was increased sulfide abundance with depth (Figure 3), consistent with microelectrode profiles in these mats (Figure 4) and in similar mat environments (Jorgensen et al., 1979; Revsbech et al., 1983; Canfield and Des Marais, 1993). Superimposed on this trend, several ∼50- to 400-μm thick horizontal laminations in sulfide abundance were observed, as well as similarly sized, localized zones of variable sulfide concentration. In particular, a well-defined band (∼100-μm thick) of increased sulfide abundance was found at a depth of 0.5 mm in mat no. 2 (Figure 3d), within the oxic zone of the mat (Figure 4). Resolution of more closely spaced laminations (as in Figure 1) was not possible when collecting data over such a (relatively) large area. However, given the banding observed in the ion images from Figure 1, the scale of variations in sulfide abundance likely spans the range of 1–400 μm. The combination of laminated and irregular variability in sulfide abundance matches with the spatial patterns of sulfate-reduction rates reported previously in microbial mats and stromatolites (Visscher et al., 2000; Wieland et al., 2005).

The isotopic composition of sulfide also displayed coherent trends with depth in both incubations, decreasing by 30‰–40‰ in the upper 3 mm (Figure 3). δ34S values just below the mat surface were typically enriched (0‰–10‰), corresponding to a relatively small fractionation between sulfate and sulfide (10‰–20‰). Sulfide δ34S decreased with depth to ∼−20‰ over 1–2 mm, approaching the lower end of the periodically aerobic portion of the mat (Figure 4). Although sulfide abundance increased sharply in both incubations at ∼1.5 mm, δ34S increased in mat no. 1 gradually with depth until about 2 mm, whereas δ34S increased in mat no. 2 until ∼1 mm, after which it did not vary significantly. At all depths, horizontal and vertical variation in δ34S of ∼10‰ was found at the scale of ∼50 μm (Figure 3). A similar scale of spatial variability in δ34S and sulfide abundance was observed. The isotopic variability in these spot analyses (Figures 2 and 3) reflects fractionations between sulfate and sulfide of 5‰–55‰.

The nanoSIMS spot analysis of mat sulfide δ34S values range between +15‰ and −35‰ with the greatest variability in the upper 0–2 mm of the mat. The majority of this variability occurs in the upper millimeter of mat, where sulfide concentrations are lowest. The average value of sulfide from below this zone (that is, where most of the sulfide occurs) is −23.5‰±4.8‰ (n=157). This value agrees with bulk mat δ34S (−26.8‰) as determined by elemental analyzer-isotope ratio mass spectrometry by precipitating sulfide from the upper 5 cm of the P4n5 mat with excess AgNO3 (Supplemental Figure S2). These values are also in agreement with the known variability of sulfide (−19‰ to −27‰) in the mats from Guerrero Negro (B Thamdrup and S Severmann, unpublished data).

SRB spatial distribution

Previous work has shown that sulfur-cycling organisms form a major metabolic and phylogenetic component of microbial mats in Guerrero Negro and similar environments (Canfield and Des Marais, 1993; Jorgensen, 1994; Risatti et al., 1994; Krekeler et al., 1997; Teske et al., 1998; Minz et al., 1999a, 1999b; Spear et al., 2003; Sorensen et al., 2005; Ley et al., 2006). Whole-cell FISH surveys of microbial assemblages within the mat, focusing on sulfate reducers and other known monophyletic sulfur-metabolizing guilds, allow for micron-scale mapping of the spatial distribution of these organisms (for example, Minz et al., 1999a; Wieland et al., 2003; Tonolla et al., 2005; Baumgartner et al., 2006). The spatial relationships of mat-associated SRB populations (Figure 5) have not been previously reported within these highly structured microbial communities at this resolution and continuous physical extent (>12 mm2).

Figure 5
figure 5

Epifluorescence microscopy image of Guerrero Negro mat thin section hybridized with the Desulfobacteraceae-specific probe, DSS658. (a) Composite low-magnification ( × 20) image. For all panels: DSS 658 probe (green); DAPI, a general DNA stain (blue); and photopigment autofluorescence (red) indicating the presence of phototrophs. Note, cyanobacteria appear magenta because of an overlay of DAPI staining and autofluorescence. (b–d) High-magnification ( × 60) images; scale bar is 10 μm. (b) Coccoid SRBs in zone b (refer to (a)); note proximity to photosynthesizing filaments. (c) Coccoid and filamentous SRBs in zone c. A large Microcoleous filament, the major mat-forming population in the mats, can be seen at the bottom right. (d) Filamentous SRB in zone d. Note the change in SRB morphology throughout the (b–d) transect. DAPI, 4′-6-diamidino-2-phenylindole; SRB, sulfate-reducing bacteria.

Full size image

In this study, horseradish peroxidase-labeled CARD-FISH probes were used to target SRB in the Guerrero Negro microbial mats. Samples for FISH were collected directly adjacent to the disc-incubation sites and sampled at the same point in the diel cycle. The oligonucleotide probe DSS658 (Manz et al., 1998) targets several genera within the Desulfobacteraceae family, including Desulfosarcina, Desulfococcus and Desulfonema (92%, 82% and 19% coverage of each genus, respectively; see Supplementary Table S1). SRB targeted by the DSS658 probe were found throughout the mat, concentrated in continuous bands within the top ∼2 mm (oxic portion) of the mat immediately adjacent to a dense layer of phototrophs (indicated by red autofluorescence) and in more dispersed, though still continuous, layers below the chemocline to a depth greater than 3.5 mm (Figure 5a). These SRB bands appeared to be laterally continuous for at least 4 mm. There was a distinct change in morphology from bands dominated by cocci in the uppermost ∼1 mm to filament-dominated bands located below the photic zone (Figures 5b–d). Aggregates of Desulfobacteraceae between 20 and 500 μm in size were also observed in the upper portion of the mat (Figure 5a). These observed spatial trends were consistent throughout repeated hybridization experiments on multiple mat thin sections.

Discussion

Cycling of sulfur within microbially dominated sedimentary systems is complex and is controlled by a combination of both biological and physicochemical factors. For example, the observed fine-scale laminations in sulfide concentration (Figure 1) could arise from different rates of active sulfate reduction (production) and/or differential diffusion of sulfide through mat layers (transport). The laminations within microbial mats are known to possess different diffusivities (Wieland et al., 2001) and this is likely to contribute in some extent to the variations in sulfide abundance. However, our preliminary observations indicate that the amount of sulfide precipitated on the silver surface (as total 32S counts on the nanoSIMS) was proportional to total sulfide as measured by microelectrodes (Figure 4). The horizontal and vertical variabilities in the isotopic composition of sulfide over short distances (20‰ over ∼2 mm) are inconsistent with fractionations resulting from purely differential diffusion and argue for localized bacterial-sulfate reduction and/or other sulfur metabolisms at depth.

Mat-surface irregularity and lateral heterogeneity prevent direct comparison between the distribution of microbial assemblages visualized by CARD-FISH and nanoSIMS data collected immediately adjacent within the mat. However, visual observations indicate that the spacing and thickness of the SRB bands are consistent with the laminations in sulfide abundance and its isotope composition (Figures 1, 2, 3). The genera targeted by the DSS658 probe have been shown to comprise a large proportion of SRB within these mats (Risatti et al., 1994; Minz et al., 1999a) and have been found to contribute to sulfate reduction in a variety of oxic and suboxic environments (Cypionka, 2000). Thus, it is likely that these dense bands of Desulfobacteraceae are in part responsible for many of the macro-scale trends in sulfide deposition on the silver foil. Minz et al. (1999a) noted remarkably similar spatial distributions in Desulfobacteraceae-associated rRNA in the oxycline of a microbial mat from Solar Lake, Egypt using rRNA blot hybridizations at 100-μm depth intervals, suggesting this may be a ubiquitous phenomenon.

In addition to laminated structures, there were zones of variable (‘patchy’) sulfide abundance and isotopic composition on the scale between 1 and 400 μm, which suggest heterogeneous sulfate reduction (Figures 1, 2, 3). Many of the Desulfobacteraceae-affiliated microorganisms in Figure 5 have a filamentous morphology analogous to members of the Desulfonema, an aerotolerant SRB common within the mat environment (Teske et al., 1998). SRB are also found in aggregates with similar scale and spatial distribution as that apparent in the nanoSIMS data (Figures 1 and 3). This observed clumping of SRB, particularly in the oxic zone (Figure 5a), has been reported from other oxic habitats, and is thought to generate microenvironments that favor continued sulfate reduction (Cypionka, 2000; Sigalevich et al., 2000; Wieland et al., 2005). The occurrence of SRB in oxic zones may be a trade-off between oxygen stress and the proximity to photosynthate that can be used during sulfate reduction (Frund and Cohen, 1992; Wieland et al., 2005). Both SRB populations and organic substrate consumption are known to vary with depth, supporting spatially variable sulfate reduction based upon the flow of energy and organic carbon substrates (Frund and Cohen, 1992; Risatti et al., 1994; Sigalevich et al., 2000; Wieland et al., 2005; Musselwhite et al., 2007). Such heterogeneous sulfate reduction naturally explains the patchiness in sulfide abundance and isotopic composition observed in this study.

The fractionations between sulfate and sulfide observed here (∼5‰–55‰) encompass the range of bulk fractionations (∼36‰–42‰) reported previously in hypersaline microbial mats (Habicht and Canfield, 1996). There is considerable overlap between our values and the values typically attributed to bacterial-sulfate reduction based on experiments with cultured organisms (4‰–46‰) (Canfield, 2001 and references therein). This large range reflects the many factors that affect isotopic fractionation during sulfate reduction, including sulfate-reduction rates, type of electron donor and carbon source, incomplete or complete oxidation of organic substrates, temperature and species-specific effects (Jorgensen, 1994; Habicht and Canfield, 1996; Habicht et al., 1998; Canfield, 2001; Porter et al., 2007).

To date, however, bacterial-sulfate reduction has not been shown capable of generating fractionations as large as some of these observed in this study, although theoretical studies (Brunner and Bernasconi, 2005) and pore water data (Wortmann et al., 2001) suggest it may be possible. The canonical interpretation for fractionations greater than 46‰ is the presence of sulfur disproportionation, whereby intermediate valence sulfur species (for example, S0, S2O3) are converted into sulfate and sulfide, resulting in increased isotopic fractionation between co-occurring sulfate and sulfide (Canfield, 2001). The nanoSIMS isotopic data presented here suggest the presence and localized metabolic activity of disproportionators distributed within these mat environments. Sulfur cycling is mediated through diverse microorganisms and metabolisms that can affect both the concentration and isotopic composition of sulfur species, including sulfate reducers, sulfur disproportionators and chemolithotrophic and phototrophic sulfide oxidizers—all of which are likely to exist in the mats at Guerrero Negro (for example, Ley et al., 2006). Ongoing work is aimed at extending the CARD-FISH analyses to include different sulfate-reducing populations as well as examining the spatial relationships and abundance of these different sulfur-cycling metabolic guilds within the mats.

Conclusions

We have adapted the silver-plating strategy previously used for sulfate-reduction rate measurements (for example, Visscher et al., 2000) to capture sulfide from microbial mats and analyze its abundance and isotopic (δ34S) composition across the oxycline using nanoSIMS. Captured sulfide records 2D variability in the abundance and isotopic composition of pore water hydrogen sulfide and offers key advantages over conventional measurements for studying sulfate reduction across steep oxic/anoxic transition zones (Wieland et al., 2005). We observe fine-scale (1–400 μm) laminations and localized zonation in sulfide abundance and isotopic composition with fractionations as high as 55‰, suggestive of active sulfur disproportionation. CARD-FISH revealed both spatially coherent (that is, laminated or aggregated) and disseminated sulfate reducers at depth throughout the microbial mats. Together, these data point toward spatially variable microbial metabolic activity generating micron-scale variations in both sulfide concentration and its isotopic composition throughout the mats. This paired high-resolution study describes 2D micron-scale variation in both geochemistry and microbial populations, providing an essential link between active microbial activity and the resulting isotopic signatures at a scale relevant to the microorganisms.