Temporal variability of in situ methane concentrations in gas hydrate-bearing sediments near Bullseye Vent, Northern Cascadia Margin



[1] To assess the temporal variability in the methane fluxes from marine sediments that overly gas hydrate bearing sediments and the factors that might control its rate, in situ methane concentrations were measured near Bullseye Vent on the Northern Cascadia continental margin. A long-term sampling device collected overlying water and pore-fluid samples from 25 cm above seafloor, at the sediment-water interface (SWI), and 7 cmbsf (centimeters below seafloor) over a 9 month period (August 2009–May 2010). These samples provide a record at ∼4 day resolution of in situ methane, ethane, propane, sulfate, and chloride concentrations, as well as stable carbon isotope ratios of methane (δ13C-CH4) and dissolved inorganic carbon (δ13C-DIC). We show that pore fluids near the SWI are saturated or supersaturated with respect to methane (∼80 mM) and the methane flux from the seabed is variable over time. We hypothesized that regional seismic activity controlled this variable CH4 flux in the Northern Cascadia continental margin setting. However, we found no direct correlation between earthquakes and CH4 flux. We also posited alternative controls on CH4 flux variability, such as storms, regional oceanography and microbial activity. Again, no direct correlation was seen. This study takes first steps toward exploring which physical factors play a role in methane flux from hydrate-bearing sediments.

1. Introduction

[2] Deep-sea sediments contain an enormous amount of methane as gas hydrate, ∼500–2500 Gt [Milkov, 2004], which are crystalline solids of methane gas contained within a water lattice [e.g., Kvenvolden, 1995]. Since methane (CH4) is a powerful greenhouse gas, exhibiting >20 times more heat absorbing potential than carbon dioxide over 100 years [Forster et al., 2007], understanding the factors that control its formation and release from such reservoirs is important. Gas hydrate occurrences include occasional outcrops on the seafloor [e.g., MacDonald et al., 1994] or in shallow marine sediments [e.g., Torres et al., 2002] and thus could be affected by changes in pressure or temperature of overlying water (OLW). Here we seek to better understand the impact of physical events on CH4 fluxes from the seafloor and how these fluxes might impact the sedimentary microbial processes involved in methane cycling.

[3] Microbial processes mediate the flux of CH4 out of marine sediments [Reeburgh, 2007]. Methane is formed either by thermogenic alteration of buried organic matter or microbial methanogesis in the sediment column. Due to low concentrations of methane in the water column, it is constantly diffusing in the sediments upward toward the seafloor. As it diffuses up, it comes into contact with downwardly diffusing sulfate, where it is subject to anaerobic oxidation via bacterial sulfate reduction [Hoehler et al., 1994; Borowski et al., 1996; Boetius et al., 2000; Joye et al., 2004]. Although ∼90% of the CH4 is believed to be consumed before it leaves the sediments [Reeburgh, 1996], some of this methane can bypass this microbial filter and reach the OLWs.

[4] CH4 fluxes may also be controlled by physical processes such as tides and fluid flow rates. For example, in shallow water, sedimentary CH4 fluxes can vary with the tides [Chanton et al., 1989]. In deep water (>600 m), where hydrates exist, tides can also control transient rates of fluid flow into and out of deep water chemosynthetic communities [Tryon et al., 1999; Tryon and Brown, 2001; Torres et al., 2002; Solomon et al., 2008].

[5] Beyond tides, microbial processes, and fluid flow, several studies have made an indirect link between CH4 release and earthquakes. For example, in the Okhotsk Sea, gas flares found in a tectonically active area of seabed were hypothesized to have caused an increase in bottom water CH4 concentrations after the Neftegorsk earthquake in 1998 [Obzhirov et al., 2004]. A similar dataset came from the Cariaco Basin where higher CH4 concentrations in the water column were assumed to be a result of a 1967 earthquake [Kessler et al., 2005]. Mau et al. [2007] measured an increase in bottom water CH4 concentrations offshore Chile after the 2002 earthquake. And finally, a magnitude 5.8 earthquake in the Gulf of Mexico was hypothesized to have caused an increase in CH4 from the sediment to the water column at a gas hydrate site [Lapham et al., 2008].

[6] In light of these studies, we hypothesize that earthquakes have a significant influence on CH4 flux from sediments to the water column in the Northern Cascadia Margin.

1.1. Geological Setting and Site Description

[7] The Northern Cascadia continental margin contains the Juan de Fuca plate that is subducting under the North American plate in the eastern Pacific Ocean, offshore Vancouver Island, Canada. As subduction occurs, the folding and faulting sediments allow organic matter to become buried and undergo microbial degradation. Eventually, CH4 is produced from these primarily biogenic origins [Pohlman et al., 2005; Riedel et al., 2006; Pohlman et al., 2009b]. Over a large area of this margin, the pressure, temperature, and CH4 saturation conditions are appropriate for gas hydrate formation [Spence et al., 2000]. Seafloor gas venting is also known to occur at several sites, including Bullseye Vent [Riedel et al., 2006]. Bullseye Vent has been studied extensively in the previous 10 years including Integrated Ocean Drilling Program (ODP) Expedition 311 and thus the geological setting is extensively documented [Riedel et al., 2006, 2009]. Bullseye vent is also near the ODP 889 node of the north-east Pacific time series underwater networked experiments (NEPTUNE) Canada cabled observatory (Figure 1a) [Barnes and Tunnicliffe, 2008].

Figure 1.

(a) Map of NEPTUNE Canada cabled observatory, with the ODP 889, Bullseye Vent, node circled. (b) Close-up of ODP 889 node site, showing location of Bubbly Gulch, mPFA, and other seafloor instrumentation. Basemap is AUV bathymetry. U1328 = IODP borehole, BBS = Broad band seismometer, JB = NEPTUNE Canada junction box, BPR = Bottom pressure recorder, and mPFA = mini-pore fluid array.

[8] The focus of this study is a seafloor gas venting area (latitude 126.841529°, longitude 48.675127°) that lies ∼1 km to the northeast of Bullseye Vent, called Bubbly Gulch (Figure 1b). Vents occur in 1265 m water depth within a presumed slump scar on the flank of the plateau on which Bullseye Vent occurs [Riedel et al., 2006]. The existence of these vents was discovered by water column acoustic anomalies in echo-sounder data [Vidalie, 2007]. In 2009, a detailed bathymetric survey was conducted of the Bullseye Vent area by Monterey Bay Aquarium Research Institute (MBARI) using an autonomous underwater vehicle (AUV) which was followed with remotely operated vehicle (ROV) dives that further explored Bubbly Gulch.

[9] ROV observations show that the flank of Bubbly Gulch contains a series of small topographic mounds that are ∼10 m across and rise up to 1 m above the surrounding seabed [Paull et al., 2009]. The surface of these mounds is marked with cracks that are several meters in length (Figure 2). In 2009 the cracks were mostly open and gray mud was visible on the crack walls and the side walls were fringed with white bacterial mats. However, the sediment, more than ∼20 cm away from the cracks, was indistinguishable from the brown silty mud characteristic of the region. Neither clams nor authigenic carbonates were seen, but gas bubbles were seen emanating from some of these cracks. Below the sediment surface, a hard layer exists at ∼50 cm depth and free gas is trapped underneath [Paull et al., 2009]. Similar seafloor cracks elsewhere have been attributed to expansion associated with the formation of gas hydrate near the seafloor [e.g., Hovland and Svensen, 2006; Paull et al., 2008].

Figure 2.

Seafloor photographs of Bubbly Gulch, the exact site where the mPFA was deployed in 2009 and retrieved in 2010. For scale, the mPFA gray box stands 75 cm tall. In 2009, (a) the seafloor looks freshly cracked and a thin layer of brown mud sits atop the gray clays (inset). (b) A close-up photo of probe tip on edge of crack. In 2010, (c) the cracks have mostly been filled with sediment and clams are more prevalent. (d) Photo showing position of probe tip in cracks. Photos courtesy of ROV Doc Ricketts, MBARI.

[10] During the 2009 ROV diving expedition to Bubbly Gulch, we deployed a long term in situ pore-water sampler (described below) within one of the seafloor cracks. The purpose of the deployment was to evaluate whether there are changes in the CH4 flux over time and if so, to determine the factors influencing these changes. Here we report on the temporal evolution of in situ CH4 concentrations associated with this time series deployment. Along with concentrations, stable isotopic ratios of both the CH4 and dissolved inorganic carbon (DIC) were measured. Influence of microbial activity on the CH4 concentrations was also addressed by measuring sulfate concentrations.

2. Methods

2.1. In situ CH4 Concentrations

[11] To obtain an “in situ” concentration, either sophisticated tools must be used to measure these concentrations on the seafloor [e.g., Short et al., 2001; Camilli and Duryea, 2009; Zhang et al., 2011; Wankel et al., 2013], or a sample must be collected from the seafloor and maintained at in situ pressures when retrieved [e.g., Dickens et al., 1997; Lapham et al., 2008]. Otherwise, if concentrations are above saturation at 1 atm, CH4 will come out of solution during ascent in the water column [Wallace et al., 2000; Paull and Ussler, 2001]. For this study, we chose to collect samples and maintain in situ pressures so that (1) complimentary analyses could be carried out on the same sample and (2) pore-water samples, and not just bottom water, could be collected.

2.1.1. Description of the Mini-Pore Fluid Array (mPFA)

[12] In situ CH4 concentrations were measured on water samples collected with a mPFA. The mPFA is a modified version of an existing sample collection tool that collects and maintains samples at in situ pressure [Lapham et al., 2008].

[13] The mPFA is essentially a polyvinyl chloride (PVC) box (45 cm × 60 cm × 60 cm) that contains three OsmoSamplers, a high-pressure valve submerged in oil to reduce corrosion, and three 300 m long coils of 0.16 cm outer diameter copper tubing (Figure 3). Each OsmoSampler has eight 2ML1 semipermeable membranes (Durect Co.) that separate a saturated salt solution from deionized water, creating an osmotic potential, which creates the pump [Jannasch et al., 2004]. OsmoSamplers have been successfully used in several gas hydrate related seafloor hydrocarbon seep settings [Tryon et al., 1999, 2002; Solomon et al., 2004; Tryon and Brown, 2004; Lapham et al., 2008]. Sampler pumping rates were calibrated in the laboratory prior to deployment and found to be ∼0.5 mL/d at 4°C. The copper material was used so CH4 could not diffuse out during deployment. The copper coils were filled with degassed, high-purity fresh water (17 MΩ) before deployment. Using 0.16 cm diameter tubing, the samplers were plumbed through the high-pressure valve, to the coil and then to ports along a meter long, PVC, T-handled stick (Figure 3b). The sample ports collected samples from the OLW, at the sediment-water interface (SWI) and at 7 cmbsf (Figure 3b).

Figure 3.

(a) Seafloor picture of the mPFA (gray box) on the seafloor with ROV arm placing the white PVC stick in the mud, which is the sampling probe tip. The white PVC rod in the gray box controls the valve handle. (b) Drawing of the OsmoSampler assembly within the box and the sample intervals along the PVC stick in the mud. Drawing shows just one depth plumbed. Thin tubing connects the sampling ports to the OsmoSamplers within the mPFA box. mPFA Contains three OsmoSamplers: two connected to valve as shown and the third, for OLW, bypasses the valve. Sample is collected through the ports in the OLW, at the SWI, and at 7 cm below seafloor (cmbsf). Seafloor photo courtesy of ROV Doc Ricketts, MBARI.

[14] The mPFA was deployed 7 August 2009, using MBARI's ROV Doc Ricketts (dive 63) on the R/V Western Flyer at Bubbly Gulch (Figure 3a). The high-pressure, multiport, two-position valve was left in a position to allow sample flow to go from the ports, through the valve, to the coil, and back through the valve. On 14 May 2010, the mPFA was retrieved by the ROV Remotely Operated Platform for Ocean Science (ROPOS) (dive 1328) on the R/V John P. Tully in conjunction with NEPTUNE Canada activities. The ROV closed the valve on the seafloor so that the copper coils were isolated and not allowed to degas upon retrieval of the mPFA package ensuring in situ pressures within the coils and in situ CH4 concentrations were maintained. Upon visual inspection of the site, we found the seafloor cracks were still open and there was an active bubble stream, but still no clams were observed. One year later, however, in 2011, Bubbly Gulch was revisited again and we observed that the cracks had been filled in with fresh sediment and live clams were visible, but no gas bubbles emanated (Figure 2b).

[15] Upon retrieval of the mPFA to the dock, the copper tubes were crimped using a crimping tool that was pressure tested to hold up to 13.8 MPa pressure within the crimped copper tubing. The coils were then stored at 4°C until they were subsampled in the laboratory.

2.2. Subsampling Copper Coils

[16] Once in the laboratory, the coils were cut into 4 m long sections. Based on the ∼0.5 mL/d pumping rate determined in the lab and the size of the tubing (0.16 cm outer diameter × 0.4 cm inner diameter), each 4 m section represents 4 days and gives about 2 mL sample volume. The water within each of the 4 m long sections was then expelled into an evacuated 10 mL glass vial with blue butyl rubber septa using a benchtop tube roller and a gas tight adapter to go from the tubing to a luer-lock needle. As each section was rolled, the pressure would build up inside the tubing until it exceeded what the end seal could handle and the water sample broke through the end seal into the evacuated vial. Samples were not exposed to air. A ∼10 µL subsample was then taken from the vial to measure salinity with a handheld refractometer. Toward the end of the 300 m coil, the salinities changed from seawater values (e.g., 35 psu) to fresh water (0 psu), which indicated the end of the time series. Sample vials were frozen upside down until future analysis.

2.3. Analytical Techniques

[17] A 100 µL subsample of the fluid in the vial was diluted 1:125 with milli-Q water and measured for sulfate and chloride concentrations on a Dionex ion chromatograph (ICS-1000). The remaining fluid in the vial was acidified with 100 µL of 40% nitric acid to measure headspace for DIC and CH4 concentrations and stable carbon isotopes, along with ethane and propane concentrations.

[18] CH4, ethane and propane concentrations were measured on a Shimadzu gas chromatograph 2010 with a 50 m POROPLOT capillary column set at 30°C with a 1 mL sample loop. A temperature ramping program was used to elute ethane and propane (10 min at 30°C, ramp at 60°C per min to 150°C and hold for 15 min). To each sample, helium gas was added to the headspace to bring the evacuated vials to 1 atm pressure plus ∼10 mL additional helium to overpressurize the vial for GC injection. The headspace was then equilibrated with the ∼2 mL water sample expelled from the copper sections and 5 mL of this headspace injected to fill the 1 mL sample loop. By comparing the integrated areas with CH4, ethane, and propane standards, headspace concentrations could then be calculated. In situ concentrations reported in millimole are corrected for the helium dilution and Bunsen solubility.

[19] Stable carbon isotopes of CH4 and DIC were measured by directly injecting microvolumes of headspace aliquots into a continuous flow Hewlett-Packard 5890 GC equipped with a 6 m Poroplot Q column at 35°C and a Thermo Delta V (Bremen, Germany) isotope ratio mass spectrometer. Isotope ratios were reported using the standard “del” notation, δ13C (‰) = (R(sample)/R(PDB standard) − 1) ×1000, where R is the ratio of the heavy to light isotope (13C:12C). CH4 headspace aliquots were obtained from vials previously measured for concentrations. Samples were run in duplicate and analytical error was <0.6‰.

2.4. Bottom Water Temperature

[20] Bottom water temperature at Bubble Gulch was obtained from an Antares© recording thermistor affixed to the mPFA box. Measurements were made every 15 min for the deployment period.

2.5. Statistics

[21] Nonparametric Spearman Rank correlations were used to calculate correlations between methane concentrations and earthquake magnitude and distance. Correlations were done using the statistical package in OriginPro 9.0.

3. Results

[22] A 9 month long time series of CH4 concentrations was measured from OLW, at the SWI, and at 7 cmbsf at a gas seep at Bubbly Gulch. The probe stick was on the edge of the seafloor cracks which were filled with sediment (Figure 2b). Upon retrieval, the samplers were visually inspected, and there were no obvious signs of damage (i.e., cracks in samplers, tubing broken or pinched, etc.). Thus, the mPFA was brought back to the lab for subsequent sampling at each depth.

[23] CH4 concentrations in the OLW averaged 0.0014 ± 0.0008 mM (Figure 4a), which is well below saturation at in situ pressures and temperatures (i.e., 65 mM) [Duan and Mao, 2006]. Two CH4 concentration spikes occurred during the 9 months (Figure 4a, two shaded regions), which reached 0.014 mM on 7 February 2010, and 0.004 mM on 23 February 2010.

Figure 4.

(a–c) Methane concentrations (CH4), (d and e) stable carbon isotopes of methane (δ13C-CH4), (f–h) chloride (Cl), and sulfate (SO42−) concentrations, and (i–k) δ13C-CH4 from August 2009 to May 2010 for OLW (first row), SWI (second row), and 7 cmbsf (third row). In Figure 4e, the δ13C-CH4 value in the early record is −34‰ which is off-scale. Shaded regions show the timing of two methane concentration spikes. Please note change in scale between a, b, and c.

[24] Close to and within the sediments, CH4 concentrations reached saturation values. At the SWI, concentrations were around 25 mM at the beginning of the time series, increased to 80 mM around December 2009 and ranged between 60 and 85 mM to the end of the deployment (Figure 4b). In mid-January 2010, concentrations reached as high as 85 mM then decreased to near saturation when the second OLW CH4 spike occurred. At 7 cmbsf, CH4 concentrations were below saturation at the beginning of the time series, until about February 2010 when concentrations reached saturation of 65 mM (Figure 4c), which coincided with the OLW CH4 spike. Concentrations decreased to ∼50 mM for the rest of the time series.

[25] CH4 stable carbon isotopes (δ13C-CH4) were measured for the SWI and pore-water samples. In the beginning of the time series, values for the SWI were about −53‰ and decreased steadily to ∼−64‰ over 4 months until December 2009 (Figure 4d). At this time, the values increased at a much faster rate to about −61‰ and then averaged about −60‰ for the rest of the time series. The δ13C-CH4 values for 7 cmbsf showed a similar trend, with a little more variability (Figure 4e). The values started at around −35‰, decreased to ∼−67‰ until December and then increased steadily with a few peaks over the rest of the record. CH4 isotopes were not measured for OLW because concentrations were too low. Chloride (Cl) concentrations were measured in the bottom waters and in pore fluids (Figures 4f–4h). The first 3 months of the OLW sampler gave inconsistent salinity data, thus we concluded the data was compromised and this part of the OLW time series was eliminated. For the rest of the OLW time series, concentrations were 525 ± 15 mM. At the SWI, concentrations were similar, and averaged 526 ± 10 mM. At 7 cmbsf, values averaged 526 ± 13 mM.

[26] To determine temporal variability in microbial activity, sulfate (SO42−) concentrations were measured in the OLW, at the SWI, and at 7 cmbsf. In the OLW coil, SO42− concentrations were around 28 mM, typical of seawater. The SWI coil shows the SO42− concentrations were initially low, between 2 and 8 mM and then increased to near seawater values of 25 mM around December 2009 (Figure 4g). The timing of this increase did not coincide with the OLW CH4 spikes. However, at the same time of the first CH4 spike, the SO42− concentrations decreased to around 20 mM and then recovered back to 25 mM after about a month. At 7 cmbsf, SO42− concentrations showed even more variability, with three distinct oscillations, each lasting about 1.5 months (Figure 4h). The timing of one of these oscillations corresponded with an OLW CH4 spike.

[27] DIC stable isotope ratios (δ13C-DIC) were measured to determine the influence of microbial processes. DIC is produced when organic matter or CH4 is microbially degraded, but it is also consumed through carbonate reduction methanogenesis to form CH4 [Whiticar, 1999]. The δ13C-DIC value can be helpful in deciphering these processes because DIC produced from organic matter degradation has a δ13C value typically around −22‰ and when produced from CH4, it is much more isotopically depleted in 13C, or more negative [Whiticar, 1999]. For reference, the δ13C-DIC value for OLW is typically around 0‰.

[28] For the OLW samples, the δ13C-DIC values were around 0‰ from November to the end of the time series (Figure 4i). At the SWI, the δ13C-DIC values were much more depleted in 13C to around −40‰ which suggests a substantial contribution in CH4-derived carbon (Figure 4j). However, two spikes in the δ13C-DIC values are observed at and right after the two OLW CH4 spikes (gray shaded regions in Figure 4j). This suggests that seawater is being advected or pumped downward because its value is around 0‰. For 7 cmbsf, the δ13C-DIC values were between −30‰ and −40‰ and showed a slight increase in values during the first OLW CH4 spike (Figure 4k), but were not nearly as prominent as the SWI.

[29] Ethane and propane concentrations were also measured during the time series (Figure 5). Ethane concentrations at the SWI started at ∼0.0002 mM and slowly increased to 0.04 mM in December, decreased to 0.02 mM over the next month, increased back to 0.04 mM right before the first water column spike event, was lower through this event and recovered to the end of the time series (Figure 5a). For propane, concentrations were an order of magnitude lower than ethane, and showed slight variability over time (Figure 5a). The ratio of CH4 to ethane + propane was around 2000, decreased to 1500 by December, increased sharply to 2000 in late December 2009 and then decreased the rest of the time series (Figure 5b). For 7 cmbsf, ethane concentrations increased gradually to 0.03 mM, increased quickly to 0.04 mM right before the second water column CH4 spike, remained at this value, and then decreased back to 0.03 mM (Figure 5c). For propane, concentrations gradually increased over the time series from <0.001 to 0.0015 mM (Figure 5c) and the gas ratios decreased from 2000 to 1500 (Figure 5d).

Figure 5.

Ethane and propane concentrations and concentration ratios (a and c) for SWI and (b and d) for 7 cmbsf. Shaded regions show the timing of two methane concentration spikes.

4. Discussion

[30] CH4 concentrations from OLW, at the SWI, and 7 cmbsf were measured over a 9 month period from a CH4-bearing seep off Vancouver Island, British Columbia. Porewaters were CH4 saturated at the SWI and were nearly saturated at 7 cmbsf. Several measured chemical parameters (CH4, sulfate, DIC) changed over time suggesting a potential coupling between pore-fluid movement, microbial processes, and physical factors.

4.1. Do Earthquakes Influence CH4 Flux From Sediments?

[31] We explored our original hypothesis that earthquakes have a significant influence on CH4 flux from sediments to the water column in the Northern Cascadia Margin. For the purpose of this paper, we have limited our analyses to timing and distribution of local earthquakes within the vicinity of the study area, as well as timing of teleseismic events (and their tsunamis) in the circum-Pacific of magnitude 7.0 and above.

[32] Local earthquakes were queried from “Earthquakes Canada” online search engine (http://www.earthquakescanada.nrcan.gc.ca/) within a 300 km radius of Bubbly Gulch (Figure 6). This distance was chosen arbitrarily. During the time series, there were 842 earthquakes (Figure 6a). The largest occurred on 12 January 2010 and 7 May 2010 and were magnitude 4.4 and 4.5, respectively. Using the entire earthquake data set, there was no correlation between CH4 being released from the seafloor and earthquake magnitude (Spearman's rho = −0.017, P value = 0.081) or distance to the earthquake (Spearman's rho = −0.007, P value = 0.92, Figure 6b).

Figure 6.

(a) Distribution of earthquakes within 300 km of Bubble Gulch (purple dot). The blue line shows the backbone of the NEPTUNE Canada cabled observatory. (b) Correlation plots between methane released into water column and earthquake magnitude and distance.

[33] During the timing of the two spikes of CH4 to the OLW, 7 and 23 February 2010, there were two earthquakes but they were magnitude 1.8 (40 km away from Bubbly Gulch) and 1.6 (78 km away from Bubbly Gulch), respectively, and likely did not contribute to significant shaking at the mPFA site. As seen from the earthquake distribution map (Figure 6a), most earthquake activities with magnitude above 3.0 were more than 80 km away from the mPFA site and along the main transform fault zones (Nootka, Sovanco). Earthquake activities closer to the mPFA site were on the shelf (and further east across Vancouver Island) but were mostly of magnitude 1.0–2.0 and occurred at greater depths (>20 km) within the crust. The accretionary prism itself is seismically quiet, which also has been verified during a long-term passive seismic monitoring experiment in summer of 2010 [Scherwath et al., 2011]. Therefore, there is no strong correlation between methane flux and earthquake activity at Bubbly Gulch.

4.2. Alternative Explanations of Physical Processes Controlling Variable CH4 Flux

4.2.1. Earthquakes Were Not Strong Enough

[34] Our data do not provide support for the hypothesis that the observed CH4 releases from the seabed at Bubbly Gulch are related to earthquake activity. However, the earthquakes that occurred during the time series were fairly minor events. Previous reports which attributed elevated bottom water CH4 concentrations to earthquakes were associated with magnitude ≥4 events [e.g., Mau et al., 2007; Lapham et al., 2008]. Such reports suggest that an alternative explanation for the lack of correlation of earthquakes in our study is that the magnitudes of earthquakes that occurred during the time series were too small. This would suggest a threshold level of earthquake activity that controls CH4 release.

4.2.2. Surface Waves Control CH4 Flux

[35] An extension of our original hypothesis is that CH4 release is linked to teleseismic events and their surface waves. Teleseismic events of magnitude 7.0 and larger could create large enough surface waves that affect local gas-charged sediments (Gary Rogers, Natural Resources Canada, personal communication). We therefore queried the worldwide earthquake catalogue (online USGS data base: http://earthquake.usgs.gov/earthquakes/eqinthenews/) and selected all events of magnitude 7.0 and above. A total of 21 events were found with two large earthquakes of magnitude 8.1 off Samoa (29 September 2009) and 8.8 off Chile (27 February 2010). Both these earthquakes created a measurable tsunami off Vancouver Island and were clearly recorded in all bottom-pressure recorder data connected to the NEPTUNE Canada cable [e.g. Thomson et al., 2011]. The 8.8 event off Chile corresponded to the second OLW CH4 spike (Figure 4). At 7 cmbsf, right after the second OLW CH4 spike, CH4 concentrations decreased which could be interpreted as a single “burp”-like event with a shallow CH4 release into the water column. Right after this degassing event, seawater was pulled into the sediment as seen in the increase in sulfate concentrations and an increase in the δ13C-CH4 record at the SWI and at 7 cmbsf. However, equally large (or larger) changes in all measured chemical constituents also occurred across the period of observation without any coincident teleseismic events (especially the spike in CH4 in the OLW early in February 2010). Extending the test to all 21 selected teleseismic events showed no further evidence for any correlation between those events and the pore-fluid record. Especially in the period between the beginning of the record in August and early December of 2009, where no sudden activity was detected in the pore-fluids, nine earthquakes of magnitude 7.0 and above occurred in the circum-Pacific region and yet no large changes could be identified in the pore-fluid record. Apparently surface waves do not obviously exert a strong control on CH4 flux. It is also possible that the local geological structure of the sampling site and the relative orientation of the surface wave arrivals and associated patterns of seafloor deformation directly at the sampling site are possible influencing parameters, but the data set does not include sufficient amounts of data and coexisting observations from pressure-recorders of seismometers to finally address this potential relationship.

4.2.3. Oceanographic Events

[36] We tested whether or not local oceanography controlled CH4 release from the seabed, mainly related to bottom water temperature variations. Because the solubility of methane decreases with warmer temperatures, the correlation here would be increased bottom water temperature increases the CH4 flux from the seafloor. The temperature record from the mPFA showed a ∼2 week periodicity, but overall the temperatures decreased from 3.00 to 2.75°C over 5 months from August to December, then increased at a faster rate to ∼3.05°C until February (Figure 7a). In early February, there was a sharp drop in temperature from 3.05 to 2.70°C within a week, which recovered to about 2.80°C within a week and stayed constant until mid-March. From this point on, the typical 2 week periodicity returned but two smaller sudden drops in temperature occurred with about half the magnitude as the main event in February. There were no surface seawater temperature anomalies during this time period (supporting information Figure S1).

Figure 7.

(a) Bottom water temperature from Bubbly Gulch (thick black line) and Bullseye Vent (gray line). (b) Bottom water pressure recorded from Bullseye vent using BPR (black) and CTD (red line). Shaded regions show the timing of two methane concentration spikes.

[37] To verify the regional extent of this bottom water temperature record, a temperature record was also obtained from ∼750 m away near the Bullseye Vent site using NEPTUNE Canada's bottom pressure recorder (BPR), which was intended as a tsunami monitoring device [NEPTUNE, 2011]. The BPR at Bullseye Vent sits at 1258 m water depth, 6 m shallower than Bubbly Gulch. The temperature here was a few tenths of a degree warmer than Bubbly Gulch, probably due to the depth difference, and also exhibited the same pattern, including the temperature drop in February (Figure 7a). This sudden drop in seafloor temperatures appears to correspond to the first large methane spike in the SWI sampler (Figures 4 and 7). It is unclear why CH4 flux would increase with a temperature decrease.

[38] One possible explanation for the increased CH4 fluxes could be the dissociation of gas hydrate within the sediments. In the Gulf of Mexico, a temperature increase was correlated to an increase in the bubble volume from a gas seep due to hydrate dissociation [MacDonald et al., 2005]. However, our time series shows a decrease in temperature, which would have stabilized the shallow hydrate even more, and does not support the explanation of a CH4 spike from dissociating hydrate.

4.2.4. Seasonal Storm Pattern on West Coast Off Vancouver Island

[39] Pressure decreases can also stimulate gas hydrate dissociation, and thus CH4 flux. While the impact of seafloor pressure on gas hydrate has generally been viewed as a long-term effect [Paull et al., 1991], short term variations may occur. So, the bottom pressure data was analyzed for any anomalous events, such as occurrence of seasonal storms (supporting information Figure S2). Typically, strong storms occur during the winter months off Vancouver Island from September/October to March/April. These storms last several days and together with large pressure changes, they do increase wave action and thus a pressure signal in the water column. Storms can be easily identified by a much larger energy in the frequency spectrum obtained from the BPR data, especially in the period range above 50 s (Martin Heesemann, NEPTUNE-Canada, and Earl Davis, Natural Resources Canada, personal communication). The largest storm event happened in early November and lasted for about 4 days from 6 November to 10 November 2009 (see supporting information Figure S2). There were no other anomalous bottom water pressure changes during the time period as recorded by two different instruments at Bullseye Vent as part of NEPTUNE Canada's monitoring (Figure 7b). Thus pressure changes from such storm events do not appear to explain the two peaks in CH4 concentration measured in the OLW during February 2010.

4.2.5. Suggestions on Possible Controlling Factors Not Tested

[40] From our study, we can conclude that there is no correlation between methane flux out of the seabed at Bubbly Gulch and the parameters we tested (earthquakes, surface waves, bottom water temperature, and storm activity). The bottom water temperature drop did correlate to the first CH4 spike, yet the mechanism is unclear. A factor that was not tested was bottom currents. Recent work has suggested a relationship between bottom currents and CH4 release from the seabed at a nearby gas hydrate site [Thomsen et al., 2012]. Unfortunately the bottom water currents were not monitored simultaneously with the mPFA measurements. However, an Acoustic Doppler Current Profiler (ADCP) installed on the NEPTUNE Canada cable did monitor currents during the month of May 2010 with the broad-band seismometer at Bullseye Vent (∼750 m west of the mPFA site). This ADCP allowed us to verify that the bottom currents are mostly tidal dominated (data not shown). The pore fluid data from the mPFA are not sampled at high enough frequency to resolve tidal cycles, as each data point represents an average condition over a 4 day period. In order to resolve these remaining possibilities, higher frequency sampling with the mPFA is required.

4.3. Sedimentary Microbial Processes

[41] The time series data set presented here is unique, and gives us an opportunity to monitor how variable CH4 fluxes might influence sedimentary microbial processes. The variations in sulfate concentration captured at Bubbly Gulch suggested seawater was periodically drawn into the sediment pore fluids. This observation was also supported by the δ13C-DIC time series data (Figure 4j). OLW had a δ13C-DIC value around 0‰, which was mixed with the pore-water DIC of −40‰ to yield an average value around −20‰. A correlation between the first methane spike and the heavier δ13C-DIC value was observed, but did not hold for the second methane spike. Admittedly, the second CH4 spike was not as strong as the first. This pattern of heavier DIC was similar at 7 cmbsf, yet the signal was not as strong (Figure 4k). This active pumping of seawater down into pore waters has several implications discussed below.

4.3.1. Implications of OLW Pumping Into Seawater

[42] Since sulfate is the main electron acceptor for organic matter mineralization and anaerobic CH4 oxidation, this renewal in supply suggests microbial communities are no longer dependent solely upon diffusion to supply more sulfate. Bioirrigation processes have also been shown to resupply sulfate to these communities [e.g., Fossing et al., 2000] but there were no conspicuous bioirrigators (clams, tubeworms, etc.) visible on the seafloor at Bubbly Gulch during the deployment. At a southern Cascadia gas hydrate site, Tryon et al. [1999] found that periodic changes in fluid flow directions, apparently controlled by episodic gas release, draw seawater (and thus sulfate) into the sediments. According to their fluid flux model, when the sample collection began, Bubbly Gulch was at a stage where gas vents were active through a buoyancy-driven gas discharge event. Over time, the system relaxed back to a stage where the gas built up in the deeper reservoir until another event released it.

[43] The oscillations in the advective flow direction will impact the microbial community as sulfate is pumped down and CH4 released. We saw an indication of this in the data at the SWI (Figure 4). Thus, the essential components required for microbial anaerobic oxidation of CH4 (AOM) are resupplied within these sediments at a rate that greatly exceeds replenishment via diffusion. Such pumping was also suggested at mud volcano sites in the Gulf of Mexico [Joye et al., 2005]. One implication for more sulfate being pumped into the sediments could be that more CH4 will be oxidized through AOM.

[44] Seawater pumping will also infuse oxygen (O2) into the sediment. O2 is the highest energy yielding process for organic matter remineralization [Froelich et al., 1979]. With more O2, rates of organic matter decomposition should also increase, along with aerobic CH4 oxidation, as CH4 concentrations are so high at the SWI. Such further implications need to be addressed at finer temporal resolution. The implications mentioned for the seawater pumping need to be addressed with more focus and directed research projects.

4.3.2. Methane Source

[45] Using stable carbon isotopes, we show that the CH4 measured at Bubbly Gulch was predominately from microbial methanogenesis, as opposed to thermogenesis. CH4 with carbon isotope values less than −50‰ indicated a predominately biogenic origin [Bernard et al., 1976]. This was consistent with what was measured at the nearby Bullseye Vent, where CH4 was also shown to be biogenic using a dual stable isotope approach between carbon and hydrogen [Riedel et al., 2006; Pohlman et al., 2009a, 2009b]. There was an apparent inverse relationship between methane concentration and the stable isotope value (supporting information, Figure S3), although it was not strong (Spearman's rho = −0.411, P value = 0.000247 for SWI). Bubbly Gulch δ13C-CH4 values were within this biogenic range (all values were between −55 and −70‰) and ∼99% of the dissolved gases consisted of CH4. The temporal variability of these higher-hydrocarbon gases also behaved similarly to the CH4 concentrations.

5. Conclusions

[46] From a gas seep off Northern Cascadia Margin, a long-term sampling device collected OLW and pore-fluid samples from 25 cm above seafloor, at the SWI, and 7 cmbsf over a 9 month period (August 2009–May 2010). The source of this gas was found to be predominantly biogenic, with measurable amounts of ethane and propane. Results showed temporal variability in in situ CH4 concentrations as well as other chemical parameters, suggesting a complex interaction between microbial responses and a still unidentifiable outside physical factor (or factors). Possible controls on this temporal variability from seismic activity (local and teleseismic events), regional oceanography, storm-weather patterns, and microbial activity were considered. The CH4 flux from the seabed was variable over time but a single source or physical driver for this flux was not identified. Future studies should focus on higher resolution sampling to address tidal variations and bottom water currents. While the temporal resolution was not high enough to resolve the tidal or current controls, it was sufficient to begin to assess how sedimentary microbial processes might be affected by the variable methane flux. This work also highlights the importance of online observatories or lander deployments to obtain higher temporal resolution [e.g., Friede et al., 2003; Kasaya et al., 2009; Barnes et al., 2011].


[47] This work was supported by many groups giving in-kind support. Specifically, we would like to thank NEPTUNE Canada, the pilots of ROVs ROPOS, and Doc Ricketts, Captain and crews of the R/V Thomson, R/V Western Flyer, and R/V Tully for graciously deploying and retrieving the mPFA. The David and Lucile Packard Foundation provided support. We also thank the generous support of Dr. Peter Brewer from MBARI for time on the Pacific Northwest Cruise. Claire Langford helped with laboratory analysis of all water samples. Laura Lapham was supported by the DOE Gas Hydrate Post-doctoral fellowship, the Center for Geomicrobiology at Aarhus University, and by a grant (in part) from BP/the Gulf of Mexico Research Initiative to support consortium research entitled “Ecosystem Impacts of Oil and Gas Inputs to the Gulf (ECOGIG)” administered by the University of Mississippi during the preparation of this manuscript (GRIIDC ID R1.x132.134:0041). We also thank Martin Heesemann, NEPTUNE-Canada, and Earl Davis, Natural Resources Canada, for thoughtful discussions. This is UMCES contribution #4756.