Estimates of Biogenic Methane Production Rates in Deep Marine Sediments at Hydrate Ridge, Cascadia Margin

MATERIALS AND METHODS

RESULTS

QPCR to determine methanogen biomass in HR sediments.Using DNA extracted from HR sediment amended with a known quantity of methanogen cells (see the supplemental material), we determined the minimum detection limit for the enumeration of cells with mcrA. Results at the level of 100 cells/g were only intermittently obtained (for about 38% of reactions). Therefore, the practical limit of detection for the assay was conservatively set at 1,000 methanogen cells/g sediment. We expect that our QPCR assay does not distinguish between methanogens and anaerobic methane-oxidizing archaea (ANME) that also possess the mcrA gene (see the supplemental material); however, we did not sample near the sulfate-methane interface where ANME cells would dominate. In using the term “methanogen cells/g,” we acknowledge that some of the detected genes may have come from ANME cells.

Most of the HR sediments that we analyzed yielded methanogen biomass estimates that were below the assay detection limits (Fig. 1). Only 25% of the samples could be confirmed as having any methanogens present by QPCR (Fig. 2). No increase in headspace methane was detected in any of the incubated sediment samples; thus, we could not detect cultivable methanogens in any of the samples by using conventional cultivation-based determinations. Also, most of the samples that had detectable levels of methanogens came from shallow sediments on HR. For example, of the 23 samples collected between 1 and 44 m below seafloor (mbsf), 35% of the QPCR analyses indicated the presence of methanogens. In contrast, of the 25 samples collected from areas deeper than 44 mbsf, only 8% showed evidence of more than 1,000 methanogens/g of sediment. Also, as shown in Fig. 2, practically all of the methanogen numbers estimated for the HR sediments were well below (i.e., <1%) the reported total numbers of cells in the same sediments determined as a part of leg 204 (28).

• Open in new tab
• Download powerpoint

FIG. 1.

(a) Representative QPCR plot comparing fluorescence to cycle number for Methanocaldococcus jannaschii standards and selected samples obtained from HR. (b) Melt curve analysis for the same standards and samples showing the comparison of changes in fluorescence (representative of denatured DNA due to sequence variability) over time as temperature was increased. Standards are amplified from sediments spiked with known concentrations of M. jannaschii cells and are expressed in terms of methanogen cells/g sediment. The analysis of 17 samples from depths of 1.0 to 5.0 mbsf are depicted, and eight of those samples yielded enough fluorescence in the QPCR reactions to be considered positive for the presence of the mcrA gene.

• Open in new tab
• Download powerpoint

FIG. 2.

Numbers of methanogens in HR sediments plotted against depth (mbsf). Methanogens were enumerated by QPCR using primers directed at the mcrA gene in DNA extracted from the sediments. Symbols shown are means of triplicate QPCR runs performed on a single DNA extract of sediment samples acquired from Ocean Drilling Program sites 1244, 1245, and 1251. Two such extractions were conducted on each sample. The y axis, at 10³ methanogens/g, represents the limit of detection for the QPCR assay. Symbols do not distinguish between samples obtained from the different cores. Dashed lines represent the approximate numbers of total cells in the sediments for sites 1244 and 1245 (line A) and site 1251 (line B) (28).

In Fig. 3, methanogen numbers determined by QPCR are also shown with respect to three of the boreholes from which samples were obtained. These data indicate that no apparent difference with respect to numbers of methanogens in samples obtained from the flank of HR (sites 1244 and 1245) and those from the foot of HR, where hemipelagic sediments were sampled (site 1251) existed. Collectively, the samples from these three sites constituted 39 of the 48 samples that we characterized, and while most of these samples from areas deeper than 44 mbsf failed to show the presence of methanogens, a few deeper sediments exhibited high numbers of methanogens. In some cases, the higher methanogen numbers that were found at considerable depth appeared within a few meters of striking geological features, such as the bottom-simulating reflector (BSR) (site 1251) and horizon A (site 1245). Among the eight samples collected from sites 1249 and 1250 at the southern summit of HR, where active methane venting occurs, only one sample (from 43.68 mbsf at site 1250) showed evidence of any methanogens (estimated at 41,000 methanogens/g).

• Open in new tab
• Download powerpoint

FIG. 3.

Methanogen numbers (cells per g sediment) as determined by QPCR to detect the mcrA gene at three of the six sites on and near HR. A plan view (A) and a cross section derived by geophysical three-dimensional seismic survey (55) (B) show sites 1245, 1244 (both on the flank of HR), and 1251 (at the foot of HR), from which samples were acquired. Methanogen numbers for each of these three sites are plotted relative to depth in mbsf (C). One sample from site 1245 and two samples from site 1251 yielded higher mcrA gene numbers than all of the other samples, and the values for these genes are shown (converted to methanogen numbers). For each sample, three replicate QPCR analyses were performed for each of two subsamples. The depths of the sulfate-methane interface (SMI), the BSR, and the presence of horizon A (HA; only at site 1245) are marked at the appropriate depths and shown in red. At each location, the zone of hydrate stability extends from the seafloor to the depth of the BSR. Seismic reflections A, B, B′, Y, Y′, U, and AC in panel B are stratigraphic anomolies (55).

As noted above, most often there was no detectable amplification of the mcrA gene in repeated measurements of a given HR sample. For these samples, we assumed that methanogens were present at less than 1,000 cells per g of sediment (our limit of detection). However, we also observed that measurements made for subsamples collected from a single sample depth sometimes ranged from no cells detected to several thousand cells detected per g (data not shown). Controls with no DNA and those containing DNA extracted from nonmethanogenic archaea consistently gave negative results in our QPCR assays.

Methanogenic rates determined from BRR-sustained cells.To estimate the levels at which methanogens produce methane when existing at maintenance level conditions that simulate the subsurface, a methanogen that was isolated from a hydrate formation (Methanoculleus submarinus [40]) was sustained in the laboratory in a BRR. Unlike most laboratory cultivation systems, a BRR functions by maintaining cells within a fixed volume and with a fixed supply of the growth-limiting substrate. During the operation of the BRR, M. submarinus cells increased in number until controlled by the concentration of the growth-limiting substrate (Fig. 4), resulting in a characteristic growth curve (59, 60). At this point (after 500 h in Fig. 4), the cells in a BRR were deemed to be sustaining themselves at low specific activities; aliquots of the cell suspension were removed from the BRR, dissolved hydrogen and methane were removed from the aliquots, and then short-term accumulation of methane was measured. Methane production rates (at 24°C) during these 40-h incubations were measured in triplicate and calculated from linear regression coefficients derived from plots of headspace methane concentration against time (R² values of 0.9525, 0.9956, and 0.9989; for pooled data, the R² value was 0.8168). Based on the rates derived from these regression curves, compared with curves derived for sterilized controls, the mean methane production rate was estimated to be 1.7 × 10⁻¹⁷ mol methane/M. submarinus cell/day, or 0.017 fmol methane/M. submarinus cell/day.

• Open in new tab
• Download powerpoint

FIG. 4.

Results of BRR experiments to determine maintenance levels of activity of methanogens. (A) Methanoculleus submarinus cell numbers in samples obtained from the BRR are plotted against time of reactor operation. Stable cell numbers after 500 h suggest that the cells in the BRR have maintenance levels of activity. (B) Methane accumulation during incubation of BRR-starved M. submarinus cells in serum bottles. Specific maintenance level activity of M. submarinus was determined from these rates and enumerations of the cells in the BRR.

DISCUSSION

As for other microbes derived from environmental samples, an accounting of the numbers of methanogens present can be difficult to obtain by using cultivation-based methods. Traditional MPN analyses based on the ability of methanogens to grow and produce detectable levels of methane are labor-intensive and are often unsuccessful in cultivating most of the target cells. Similarly, fluorescence in situ hybridization approaches are labor-intensive. In competitive or MPN PCR, the measurement can be affected by biases of endpoint analysis, in which different amounts of final PCR amplicons are obtained from similar starting quantities (3). MPN PCR is also labor- and time-intensive. QPCR is sensitive, specific, and rapid (8) and has been used to enumerate archaea in water, soils, and sediments (30, 44, 58). QPCR reactions are run in replicate, with reaction and analysis times of hours; for our work, approximately three hours was required to analyze 72 samples. Since QPCR results are viewed in real time as the reaction proceeds, the bias produced by endpoint analysis is eliminated.

Our use of QPCR directed at a specific functional gene for population enumeration assumes that the DNA is derived from a living cell. DNA in dead cells or outside of cells typically has a short half-life (27), and recent investigations indicate that extracellular DNA in the upper 1 cm of marine sediments only has a ca. 9.5-year residence time (18). Because all of our samples came from HR sediments whose ages range from hundreds to millions of years, it would appear that the DNA that we extracted was most likely derived from living cells.

While many methods of detection for QPCR exist, the use of Sybr green I offers some unique advantages when working with environmental samples. The use of Sybr green I permits detection of DNA despite some variation in the internal region of the amplicon and also permits selected primers to be used for QPCR without the additional time and cost of developing and optimizing probes. Although Sybr green I will fluoresce with nonspecific PCR primer artifacts (and this can reduce the efficiency of the QPCR reaction), such data can be discriminated by using a melting curve analysis (see Fig. S3 in the supplemental material) (30), permitting this QPCR assay to attain sensitivities as low as 100 cells/g and reliably detect 1,000-cells/g levels.

The QPCR assay targeted the α subunit of the MCR gene generally considered to be present in all methanogens (19, 22, 24, 39), with most methanogen isolates bearing one or two MCR-encoding regions (52). We assumed that one mcrA gene copy existed per cell.

Recent studies have shown that the mcrA gene is also present in archaeal representatives (ANME) of consortia that perform anaerobic methane oxidation or “reverse methanogenesis” (25, 26, 44), and we expect that the mcrA primers that we designed may also recognize this gene in the ANME cells (see the supplemental material). In marine sediments, active anaerobic methane oxidation is restricted to a narrow band where both sulfate and methane are available, the so-called sulfate-methane interface (4). However, we expect that, as for most microbes, ANME cells may be found in sediments away from where chemical or physical conditions are optimal for their growth or metabolic activity. Previously, genes from ANME cells were not detected in clone libraries constructed of 16S rRNA genes amplified from samples acquired at sites 1244, 1245, and 1251 on HR, while methanogen clones were noted at low levels in site 1251 samples (28). While none of the samples used in our study came from the sulfate-methane interface, our enumerations of methanogens may include ANME cells. An understanding of the interstitial water chemistry in the sediments, specifically the concentrations of methane and sulfate, can help to infer whether detection of an abundance of mcrA genes indicates cells associated with anaerobic methane oxidation or those associated with methanogenesis.

For most of the HR samples that we examined, we could not detect methanogens. However, our estimates suggest that, in the samples in which the mcrA gene was detected, the numbers of methanogens were typically ≤1% of the total cell numbers in the sediments (Fig. 2) (28) as well as of the total cell numbers reported to occur in numerous marine sediment samples collected elsewhere (47). The findings were consistent with observations that the numbers of archaeal cells, albeit not specifically methanogens, decrease rapidly in the upper 40 m of the sediment column in ocean margin locations (28, 54).

The overall estimated numbers of methanogens in these sediments were low, but some unexpectedly high values were measured at greater depth in the sediments (Fig. 2 and 3). Our sampling densities (numbers of samples) near prominent geological features such as horizon A (site 1245) and the BSR were insufficient to permit accurate measurement of methanogen biomass at such locations, although some of the highest methanogen numbers that we detected were within several meters of these geological features. Horizon A is a three-meter-thick volcanic-ash-rich stratum that is highly transmissive, conveying fluids from deeper in the sediments, and the BSR is at the lower limit of the hydrate stability zone, a location where three phases of methane (free gas, in the dissolved state, and in a solid phase) may occur within a short vertical sequence (55a).

Despite evidence that much of the methane that occurs in sediments along coastal margins is biogenic (33), the detection of methanogens in such locations has been difficult. The few methanogens detected in the hydrate-rich sediments of HR is consistent with findings from studies examining 16S rRNA gene clone libraries from sediment samples collected in the vicinity of hydrates (32, 51). Even investigations that included 16S rRNA gene clone libraries consisting of thousands of phylotypes have noted few sequences that bear a resemblance to known methanogens (28). Recently, the detection of methanogens in deep Nankai Trough sediments was possible by targeting mcr genes as opposed to 16S rRNA genes (43).

We also noted a heterogeneous distribution of methanogens in our samples, wherein subsamples from the same sediment material (i.e., the same depth and the same borehole) frequently yielded results that were different with respect to the methanogen numbers. This suggested a patchy distribution of methanogens, as has been found in other subsurface environments (7, 9).

By using the BRR to sustain M. submarinus cells under simulated starvation, a maintenance level rate of methane production of ca. 0.017 fmol methane/M. submarinus cell/day was estimated. These cells were starved of substrate, and methane was removed from the system. Therefore, the production of methane by cells from the reactor was likely thermodynamically unconstrained compared to the methane production of methanogens existing in marine sediments where they are surrounded by high levels of methane. M. submarinus can be thermodynamically impaired at high methane partial pressures (11), and though it was originally isolated from hydrate-bearing sediments, we do not know how representative of methanogenic communities as a whole this isolate is. Factors such as the rate of substrate introduction and quantity of substrate added likely differ between the BRR and the subsurface. As with other methods, the BRR method may overestimate or underestimate true rates depending on the degree to which it simulates in situ conditions.

The methanogenic rate that we estimated is lower than previously reported methanogenic rates, including 31.5 and 108.8 to 135 fmol methane/cell/day from environments such as lake sediments (36) and anaerobic reactors (37), respectively. Considering the Arrhenius law, which predicts a doubling of enzymatic activity for every 10°C increase in temperature, the rate determined for cells from the BRR at 24°C would correspond to about 0.004 fmol methane/M. submarinus cell/day if corrected for seafloor temperatures of ca. 4°C. Understanding the impacts of pressure, substrate concentration, product concentration, and the capabilities of communities of microbial cells (in comparison to the conditions when making measurements for an isolate) would further hone our understanding of microbial activities in deep sediments.

To evaluate the determined rates of methanogenesis relative to other microbial rate estimates, the rate of 0.004 fmol methane/M. submarinus cell/day was converted to g methane carbon produced/g cell carbon/h and compared to rates of metabolic-product formation by other microbes existing at various metabolic states as summarized by Price and Sowers (50). Using calculated or published conversion factors (i.e., 16 fg methane/fmol and 170 fg/cell [2]; 50% of the dry mass of a cell consists of carbon [1]), the activity of M. submarinus cells in the BRR was estimated to be 5.9 × 10⁻⁶ g methane carbon produced/g cell carbon/h. This rate was consistent with rates demonstrated for different cell types grown at the same temperature and exhibiting maintenance levels of activity (50). The M. submarinus maintenance level rate of metabolism is similar to rates estimated for sulfate-reducing and methanogenic communities in deep sediments off the Peru margin (ca. 1 × 10⁻⁵ g carbon/g cell carbon/h) (48).

By combining maintenance level methanogenesis rates derived from the BRR with methanogen biomass values for HR samples as determined by QPCR targeting the mcrA gene, an estimate of the microbial methane created at in these marine sediments can be made. Most of the samples yielded methanogen estimates that were less than the 1,000-cells/g sediment detection limit for the QPCR assay, corresponding to <4.25 fmol methane/g sediment/day (i.e., <0.004 fmol methane/cell/day × 1,000 cells/g at seafloor temperature). As noted above, this rate is based on the detection limit for the QPCR assay, and many of the samples that we analyzed were below the limit of detection for this assay. For this reason, and because some of the cells that were detected may have been ANME cells, actual in situ rates may be considerably lower than this rate. Nevertheless, with respect to previously published methanogenesis rate estimates for other seafloor locations where hydrates are important, our calculated rate for HR sediments is lower than rates reported for Cascadia margin and Blake Ridge sediments by at least three orders of magnitude and as many as nine orders of magnitude, respectively (see reference 15 for details).

Using factors that convert mass, time, and density and volume units into more practical terms (i.e., 10¹² fmol equal 1 mmol, 365 days equal 1 year, and a sediment density of 2.6 g/cm³ of sediment equals 10⁶ cm³/m³), the rate that we determined for most of the samples can be expressed as <4.03 × 10⁻³ mmol methane produced/m³ sediment/year. For comparison, the modeled microbial methane production rates for deep (>100-mbsf) and shallow (a few meters below seafloor) sediments on HR are <0.1 and ca. 10 mmol methane produced/m³ sediment/year, respectively (12). Based on higher levels of mcrA gene detected in the shallower sediments, our rates of methanogenesis could also scale upwards in the shallower sediments. The detected mcrA genes could also belong to ANME cells, even though we did not sample at the sulfate-methane interface.

The methanogenic rates that we estimated for these sediments should take into account the amount of methane that can be biologically generated in the sediments if the sediments are the sole supply of microbial reductants and oxidants. In other words, if it is assumed that the total organic carbon (TOC) buried in the sediments is the source of the methane, then the estimated rates of methanogenesis must be consistent with both the amount of TOC available in the sediments and the time over which it can be metabolized. Generally, the TOC constitutes 1.5% of the weight of the HR sediment mass (63), equivalent to 15 mg/g. If 10% of the organic carbon in sediment TOC is available for conversion to methane (13), then as much as 1.5 mg methane/g sediment might be generated from the TOC present. For one of the deeper samples that we analyzed, coming from 375 mbsf, the sediment is estimated to be Pliocene-Pleistocene (1.6 million years of age) (55). Assuming that the methanogens in this sample were continuously active and producing on the order of 4.25 fmol methane/g sediment/day for approximately 1.6 × 10⁶ years, this equates to 2.5 × 10⁻³ mmol methane/g of sediment. Conversion of the methane mass term to milligrams (using 16 mg methane/mmol) yields a methanogenic productivity value of about 0.04 mg methane/g of sediment, a factor of 38 lower than the 1.5 mg methane/g sediment that might be possible according to the amount of TOC assumed to be present. In this closed-system scenario, a methanogenic rate term sustained over geologically relevant time periods (i.e., thousands to millions of years) at a rate much higher than the one we measured would exhaust the sediment TOC that is available for conversion to methane.

Some parts of HR may act as a closed system without outside input of fluids, and in these locations, the rates we estimated are low enough to sustain methanogenic activity over long periods in the sediments, given the amount of organic carbon available. However, this view of HR as being closed is imperfect, as it assumes that the sediments would experience the gradual depletion of microbial reductants and oxidants. In fact, much of the HR subsurface is not closed and instead is a gas- and hydrate-rich environment that has the attributes of two distinctive gas-fed systems with respect to methane supply: a “focused, high-flux” system and a “distributed, low-flux” system (41, 62). To consider how the rates that we estimated compare to the active flux of methane out of HR, we can use the areal extent of the southern side of HR (3.7 × 10⁷ m²) (5) and an arbitrary depth of 100 m into the sediments to derive a volume of 3.8 × 10⁹ m³ as a zone of active methanogenesis for this location. Then, using our base rate of methanogenesis (<4 × 10⁻³ mmol/m³/year) hypothetically occurring uniformly within that volume, we derive <1.5 × 10⁷ mmol of gas/year created by methanogens. This term is 68% of the total amount of 2.2 × 10⁷ mmol of methane C/year estimated to be released into the overlying water from the southern side of HR (5). Thermogenic methane, a deeper biogenic zone, and dissolution of hydrates may all contribute to the difference in these approximate determinations.

None of our samples came from the fractures or horizons that represent the focused, high-flux strata. However, the low-flux system that exists elsewhere at HR would place methanogens and other microbes under conditions that are intermittently open to fluid movement. This could permit the periodic infusion of small, anaerobically oxidizable molecules (hydrogen, carbon monoxide, acetate, formate, and ammonia), along with reducible molecules (sulfate, carbon dioxide, and nitrate), to the sediments. At this time, it is difficult to determine whether other electron acceptor/donor couples (e.g., hydrogen/bicarbonate) in the sediments are present in concentrations high enough to permit methanogenesis, although such compounds are present (38). In a dynamic HR setting, the intermittent flux of appropriate molecules might allow sufficient activity to sustain chemolithotrophs, including methanogens, and the creation of more-complex organic compounds that may, in turn, support heterotrophs. Under these circumstances, there might be considerable fluctuation in the in situ rates of microbial activities.

Determining the concentrations of key electron acceptor/donor couples in the sediments will be important for completing our understanding of the biological activities in these systems. How such concentrations in the subsurface at HR change subtly over time by diffusion of fluids or dramatically through the opening and closing of fractures in the sediments will lead to better conceptual models of microbial activity. As such changes in the subsurface occur, thermodynamic conditions will determine whether the activity of methanogens and other microbes can proceed. This thermodynamic constraint has been demonstrated for methanogenic isolates from other marine environments wherein the cells require higher concentrations of energy-yielding substrate (hydrogen) to remain active in the presence of high levels of methane (11).