Massive marine methane emissions from near-shore shallow coastal areas

Methane is the second most important greenhouse gas contributing to climate warming. The open ocean is a minor source of methane to the atmosphere. We report intense methane emissions from the near-shore southern region of the North Sea characterized by the presence of extensive areas with gassy sediments. The average flux intensities (~130 μmol m−2 d−1) are one order of magnitude higher than values characteristic of continental shelves (~30 μmol m−2 d−1) and three orders of magnitude higher than values characteristic of the open ocean (~0.4 μmol m−2 d−1). The high methane concentrations (up to 1,128 nmol L−1) that sustain these fluxes are related to the shallow and well-mixed water column that allows an efficient transfer of methane from the seafloor to surface waters. This differs from deeper and stratified seep areas where there is a large decrease of methane between bottom and surface by microbial oxidation or physical transport. Shallow well-mixed continental shelves represent about 33% of the total continental shelf area, so that marine coastal methane emissions are probably under-estimated. Near-shore and shallow seep areas are hot spots of methane emission, and our data also suggest that emissions could increase in response to warming of surface waters.

Methane (CH 4 ) is the second most important greenhouse gas (GHG) after CO 2 , accounting for 32% of the anthropogenic global radiative forcing by well-mixed GHGs in 2011 relative to 1750 1 . Yet, there remains an important uncertainty on estimates of the sources and sinks of CH 4 2 , and how their variations can affect the atmospheric CH 4 growth rate and burden 3 . The atmospheric CH 4 increase (34 TgCH 4 yr −1 for 1980-1989 and 6 TgCH 4 yr −1 for 2000-2009 1 ) is calculated from the measured increase of the CH 4 concentration in the atmosphere, but results from the net balance between the sum of sources and of sinks which are one to two orders magnitude larger. The open ocean is a very modest source of CH 4 to the atmosphere (0.4-1.8 TgCH 4 yr −1 4 ) compared to other natural (220-350 TgCH 4 yr −1 ) and anthropogenic (330-335 TgCH 4 yr −1 ) CH 4 emissions 2 . Coastal regions are more intense sources of CH 4 to the atmosphere than open oceanic waters 5 . Continental shelves emit about 13 TgCH 4 yr −1 5 and estuaries emit between 1 and 7 TgCH 4 yr −1 5-8 . The high CH 4 concentrations in surface waters of continental shelves are due to direct CH 4 inputs from estuaries and from sediments where methanogenesis is sustained by high organic matter sedimentation 5,6,9 . Natural gas seeps from continental shelves contribute additionally between 16 and 48 TgCH 4 yr −1 10,11 . Biogenic or thermogenic CH 4 can accumulate in large quantities in sub-surface seabed (gassy sediments) in deep and shallow areas, and can be released as bubbles (gas flares) or by pore water diffusion. However, the estimates of CH 4 "emission" from marine seeps 10,11 correspond to CH 4 release from sediments to bottom waters and not to the actual transfer from surface waters to the atmosphere, which is probably much lower 12 . Bubbles dissolve in water leading to high dissolved CH 4 concentrations in bottom waters (from tens of nmol L −1 up to several μ mol L −1 ), but removal by microbial CH 4 oxidation and lateral dispersion by physical transport leads usually to much lower CH 4 concentrations in surface waters (5-20 nmol L −1 ) even in the shallow areas of continental slopes and shelves [13][14][15][16][17][18][19] .
In this study, we report a data-set of CH 4 concentrations in surface waters of the Belgian coastal zone (BCZ) in spring, summer and fall 2010 and 2011 (Fig. S1). This is a coastal area with multiple possible sources of CH 4 such as from rivers and gassy sediments. The BCZ is also a site of important organic matter sedimentation and accumulation unlike the rest of the North Sea 20 .

Results and Discussion
The CH 4 concentrations in surface waters of the BCZ in spring, summer and fall 2010 and 2011 ( Fig. 1) were high, with about 43% of the observed values above 50 nmol L −1 , and a maximum concentration of 1,128 nmol L −1 in July 2011. The near-shore area (within 15 km of the coastline) was characterized by CH 4 concentrations in surface waters between 3 and 13 times higher than the more off-shore area (> 15 km away from the coastline). The overall average CH 4 concentration in the BCZ near-shore area (139 nmol L −1 ) was ~6 times higher than in the off-shore area (24 nmol L −1 ), and in both areas distinctly above atmospheric equilibrium (~2 nmol L −1 ). These values are one to two orders of magnitude higher than the CH 4 concentrations in surface waters of most of the North Sea with values typically < 5 nmol L −1 5,9 that are mainly influenced by inputs of water from the North Atlantic, where CH 4 is close to atmospheric equilibrium 4 . Values in the BCZ were also high compared to estuarine plumes of the North Sea where maximal CH 4 concentrations in surface waters range between 60 and 90 nmol L −1 , such as for the Elbe 9 and the Rhine 6,21 . Our own CH 4 data in the Thames river plume were below 25 nmol L −1 (Fig. S2), distinctly lower than values in the BCZ. Values in BCZ were consistent with the high values (up to 372 nmol L −1 ) reported 22 further north along the Dutch coast in March 1989 in a near-shore area with similar settings (well mixed waters overlying peat-rich sediments). The highest CH 4 concentration in the BCZ (1,128 nmol L −1 ) was higher than any other previous report in (natural) surface waters of the North Sea, and nearly equals the value reported above an abandoned borehole in the Northern North Sea of 1,453 nmol L −1 9 . The highest CH 4 concentration in the BCZ is comparable to the maximal value in surface waters (~1,800 nmol L −1 ) in the Santa Barbara Channel (Coal Oil Point), one of the most intense marine seep area in the world 13 .
High CH 4 concentrations in near-shore coastal areas have been frequently attributed to estuarine inputs of CH 4 6,9,21,22 . This could explain the higher CH 4 concentrations in the lower salinity region of the Thames river plume (Fig. S2). The inputs from Scheldt estuary have been shown to influence a variety of biogeochemical variables in the BCZ, such as CO 2 23 . However, during most cruises, maximal CH 4 concentrations measured in the BCZ were not located at the mouth of the Scheldt estuary ( Fig. 1), and were higher than in the freshwater region of the Scheldt estuary (Fig. 2). Also, the CH 4 concentrations in the near-shore BCZ were above the theoretical dilution line between the lower Scheldt (salinity > 25) and the outer BCZ (Fig. 2), except for April and September 2010. This indicates that a local additional source of CH 4 contributes to the observed high values in the near-shore BCZ.
Extensive areas of the North Sea have sediments with seismic/acoustic characteristics indicative of shallow gas accumulation, that is assumed to be mainly CH 4 24 . In the BCZ, a four to twelve km wide band parallel to the coastline contains sediments with shallow gas, associated to a peat-rich layer from the late Pleistocene 25 . The high near-shore CH 4 concentrations in surface waters were observed within this band of gassy sediments (Figs 1 and S1) that was most probably the source of CH 4 . The nearshore BCZ has similar sediment characteristics than Norton Sound (Alaska), an area of intense shallow submarine gas seepage 26 . However, occurrence of actual gas flaring has not been investigated in the BCZ in a systematic way, but there are some indications of local seepage of bubbles 25 . In the Scheldt estuary, an increase of CH 4 was observed in the lower estuary (salinity > 25) compared to the mid estuary (salinity ~15) (Fig. 2) which has been attributed to the presence of extensive tidal flats 7 , where gassy sediments also occur 27 . Hence, CH 4 seepage from shallow gassy sediments could be the main reason for elevated CH 4 concentrations in surface waters of both the nearshore BCZ and lower Scheldt. Concentrations of CH 4 between 15 to 300 nmol L −1 have been reported in bottom waters at Tommeliten, a prominent CH 4 macro-seep area in the Central North Sea 17 , yet, in surface waters, CH 4 concentrations were below 5 nmol L −1 . This was attributed to removal by microbial CH 4 oxidation and lateral dispersion by physical transport, favored by thermal stratification 17 . Similarly, in another gas seepage area in the North Sea, south of the Dogger Bank, surface waters were characterized by lower concentrations (4-518 nmol L −1 ) than bottom waters (40-1,628 nmol L −1 ) 14 .
Due to the shallowness (< 30 m) and strong tidal currents, thermal or haline stratification never occurs in the BCZ (Fig. S3). Due to the strong tidal currents, dissolved O 2 values remain close to atmospheric equilibrium (Fig. S4), with no gradients between surface and bottom. The O 2 and CH 4 concentrations were uncorrelated. While CH 4 in bottom waters was statistically higher than in surface waters (Wilcoxon matched-pairs signed rank test p = 0.0002, n = 48), the difference was very small (on average ~14%) (Fig. S3). Hence, due to the shallowness and well-mixed water column there is little loss of CH 4 between bottom and surface waters unlike deeper and stratified areas such as Tommeliten and south of the Dogger Bank. Indeed, summertime average CH 4 concentration in surface waters showed a regular decreasing pattern across the North Sea as a function of depth, from the vertically mixed BCZ towards the stratified and deeper regions south of the Dogger Bank and Tommeliten (Fig. 3).
The dissolved CH 4 concentration in the BCZ showed distinct seasonal variations with higher values in summer than spring and fall. Inter-annual variations were also observed with higher values in summer 2010 than 2011, but conversely lower values in spring and fall 2010 than 2011 in the near-shore area ( Fig. 1; Table 1). In the near-shore BCZ, the lower CH 4 concentrations were associated with lower water temperatures (April 2010) and the highest CH 4 concentrations were associated with the higher water temperatures (June 2010) (Fig. 4). The relationship between CH 4 concentration and temperature was non-linear with distinctly different slopes of the linear regressions for data above and below 19 °C. We interpret the positive relationship between dissolved CH 4 and water temperature as resulting from enhanced CH 4 release from the seafloor in response to warming. Due to the well-mixed nature of the water column in the BCZ, the amplitude of the seasonal variation of temperature in bottom waters was very large (~15 °C) 23 compared to bottom waters in seasonally thermally stratified regions (~1 °C). In Cape Lookout Bight, enhanced bubble accumulation in sediments as well as CH 4 diffusion and ebullition were observed in summer 28 . Increase in temperature stimulates microbial CH 4 production 29 and decreases CH 4 solubility 30 , both processes contributing to releasing CH 4 from sediments to the water column. Hence, increasing temperature could enhance a passive release of CH 4 from gassy sediments due to the decrease of gas solubility, but this does not exclude an increase of CH 4 production by methanogens also in response to higher temperature, and organic matter availability. Indeed, the maximal CH 4 concentrations were observed in summer, when the sediment was enriched in organic matter produced by spring phytoplankton bloom.
The air-sea CH 4 emissions ranged seasonally between 1 and 160 μ mol m −2 d −1 in the off-shore BCZ and between 2 and 426 μ mol m −2 d −1 in the near-shore BCZ (Table 1). Wind speed was lower during summer than during the other two seasons, yet, seasonal variations of the air-sea CH 4 emissions were mainly driven by variations in CH 4 concentrations rather than wind speed (Fig. S5). Annual air-sea CH 4  To envisage the impact of our findings on the marine CH 4 emission budget, it is necessary to evaluate the representativeness of our study site for coastal areas in general. This is not an easy task since there are no global spatial datasets of gassy sediments and of submerged peat deposits. Regions corresponding to drowned coastlines (drowned forests and peatland) have been identified among the coastal environments most likely to have gas-rich sediments, in addition to estuaries, bays, rias and deltas 31 . Due to the global sea-level rise of the past 20,000 yr, it is probable that most near-shore coastal areas are drowned former land and that most of the Quaternary peat layers are now inundated and situated on the continental shelf, buried under marine sediments 32 . Yet, extensive or global spatial data-sets of submerged peat deposits are unavailable because it is difficult to identify them from seismic data alone and verification is required with coring 32 . In continental shelves where the presence of gassy sediments and seepage sites have been systematically investigated, such as around the United Kingdom, very extensive areas of gassy sediments associated with Quaternary peat deposits have been mapped 33 . In addition, permanently well-mixed water columns could represent a large fraction of continental selves. By analogy with the European continental shelf, if we assume that regions shallower than 35 m are permanently well-mixed by tidal action 34 , they would represent 33% of the total surface area of continental shelves (< 200 m, that is 26,400 km 2 ) 35 .  Table 1. Wind speed and air-sea CH 4 fluxes in the near-shore (<15 km from coastline) and offshore (>15 km from coastline) Belgian coastal zone (BCZ) in spring, summer and fall 2010 and 2011 (mean ± standard deviation). Annual fluxes were calculated assuming a zero flux in winter (based on the very low CH 4 concentrations measured at low temperature, Fig. 4). The distinctly different CH 4 concentrations in well-mixed and seasonally stratified continental shelves (Fig. 3) should then be accounted when budgeting CH 4 emissions. These emission estimates for the near-shore BCZ are most likely underestimated since they only account for diffusive CH 4 fluxes, although there are some indications of local seepage of bubbles 25 . While in deeper continental shelf areas CH 4 bubbles dissolve as they rise, and dissolved CH 4 is removed by microbial oxidation and by horizontal physical transport 17 , in very shallow areas such as the BCZ (< 30 m) bubbles from seepages could avoid dissolution 36 and be directly emitted to the atmosphere. While the emissions from seeps should be considered as natural sources in the global CH 4 budget, our data (Fig. 4) suggest that further warming of surface waters could increase CH 4 emissions and provide a positive feedback on warming climate. This feedback will be expected to be acute in shallow gassy areas such as the BCZ since they are natural hotspots of CH 4 emission, and the well-mixed water column will allow an efficient propagation of additional heat to the sediment that will be buffered by seasonal thermal stratification in deeper seep areas. The increase of temperature will stimulate the biogenic CH 4 production, as well as, decrease Henry's constant promoting bubbling from sediments.  (Fig. S1). Sampling was carried out with a 10L Niskin bottle coupled to a conductivity-temperature-depth (CTD) probe (Sea-bird SBE19 on the Belgica and YSI 6600 on the Luctor), in surface waters (1 m depth) and on some occasions ~3 m above the seafloor. When CTD data were unavailable on the Belgica, we used salinity and temperature measurements from an underway instrument (Sea-bird SBE21) connected to a seawater supply (pumped at 2.5 m). Water samples were collected in borosilicate serum bottles (50 ml) with a tubing, left to overflow, poisoned with a saturated solution of HgCl 2 (100 μ l), sealed with a butyl stopper, crimped with an aluminum cap, and stored at ambient temperature in the dark until analysis. Dissolved oxygen was measured by titration with the Winkler method 37 .

Methods
The concentration of CH 4 was determined with the headspace equilibration technique (20 ml N 2 headspace in 50 ml serum bottles and overnight equilibration in a thermostated bath after initial manual vigorous shaking) and a gas chromatograph 38 equipped with a flame ionization detector (SRI 8610C) calibrated with CH 4 :CO 2 :N 2 O:N 2 mixtures (Air Liquide Belgium) of 1, 10 and 30 ppm CH 4 . Each of the three standards was analyzed in triplicate at the start and the end of the daily batch of samples (typically 30) and the calibration curve was computed by linear regression forced through zero (r 2 ≥ 0.999). The slope of the calibration regression line was interpolated linearly from initial and final values for the whole batch of samples, although no statistical difference was ever observed between the start and end calibrations. About 10 ml of the headspace (or standard) was injected through a 6-way valve from which a 2 ml subsample (loop) was injected into a 2 ml column of magnesium perchlorate (water vapor trap), and then into a packed column (Hayesep D, 5.0 m length, mesh 80/100) kept at 50 °C, using N 2 as carrier gas. The 10 ml volume of headspace was sampled with a plastic syringe with a steel needle through the septum, and the retrieved gas volume was replaced by a hyper-saline solution (about 60 g NaCl L −1 ) injected with another syringe in the bottom of the serum bottle, in order to keep the sampled gas sample at atmospheric pressure. Chromatographic peak areas were integrated and logged using the Peaksimple software (version 4.44 for Windows TM XP). The in-situ CH 4 concentration was computed 39 from the volume of water and headspace (determined from the weight of bottles empty, and before and after making the headspace), the measured partial pressure of CH 4 and Henry's constant 40 . Precision estimated from multiple injections of gas standards was better than ± 3.0% for the 1 ppm standard and better than ± 0.5% for the other two standards. The precision estimated from duplicated samples was ± 3.9%.
The air-sea CH 4 flux (F) was computed according to: where k is the gas transfer velocity and Δ CH 4 is the air-sea CH 4 concentration gradient computed from the measured dissolved CH 4 concentration in seawater and the concentration at equilibrium with an atmospheric CH 4 partial pressure value of 1.8 ppm, computed with Henry's constant 40 .
The k values were computed from the parameterization as a function of wind speed based on dual deliberate tracer ( 3 He/SF 6 ) experiments in the Southern Bight of the North Sea 41 , and the Schmidt number of CH 4 in seawater computed from temperature 42 . Wind speed data were obtained from the National Centers for Environmental Prediction reanalysis daily averages surface flux (http://www.cdc.noaa.gov/) at 2 grid points covering the sampled region (3.7500°E 52.3799°N; 0.0000°E 50.4752°N). F was computed using daily wind speed values (average of the 2 grid points) for a time interval of 30 days centered on the date of the middle of the cruises.