Gridded and normalized to an area of 100 km 2 , the daily sea−air methane flux from each area ranged from 0.5 kg to 8.8 kg per 100 km 2 , with highest values in the nearshore ( SI Appendix, Table S1 ). In the context of a well-constrained global atmospheric methane source (e.g., ruminants), the flux from the shallow-water continental shelf seep field [6.1 kg CH 4 (100 km −2 )⋅d −1 , SI Appendix, Table S1 ] is equivalent to that from ∼320 sheep, each emitting 18.9 g CH 4 d −1 ( 26 ). To match the methane output of the 3 × 10 7 sheep in New Zealand alone would require more than 90,000 multiseep clusters of the type investigated here. Although tens of thousands of discrete seeps likely remain undiscovered on global margins ( 8 ), there is no evidence that such a large number of multiseep clusters exists. Even if there were, the annual cumulative atmospheric methane flux would be ∼0.15 Tg CH 4 y −1 , a negligible (0.03%) quantity relative to the 580 Tg of methane emitted to the atmosphere annually ( 18 ).

For the deep-water and shelf-edge systems ( Fig. 1A and SI Appendix, Fig. S3 ), surface water methane concentrations ranged from 3.2 nM to 4.3 nM, corresponding to saturation anomalies of −1.5 to 36%. Sea−air fluxes ranged from 0.0 μmol⋅m −2 ⋅d −1 to 2.8 μmol⋅m −2 ⋅d −1 ( Table 1 and SI Appendix, Fig. S3 ). The low saturation anomalies and fluxes for the 240-m water depth region are comparable to those in the open ocean ( 23 ) and are similar to those previously reported for this site ( 24 ), confirming that this setting is not a significant source of methane to the atmosphere.

Our data show that shallow arctic methane seeps like those we investigated on the WSM emit negligible methane to the atmosphere. However, in comparison with the seeps we investigated, methane fluxes from the shallow East Siberian Arctic Shelf (ESAS), which may be underlain by thawing subsea permafrost, are ∼100 times greater. Shakhova et al. ( 2 ) report average fluxes of 229 μmol·m −2 ·d −1 for “background” areas and 738 μmol·m −2 ·d −1 from “hotspots” during the ice-free summertime. These values are comparable to an average ice-free flux for the middle and outer East Siberian Arctic shelf of 238 μmol·m −2 ·d −1 , as measured by Thornton et al. ( 17 ). Given that (i) gas flares (evidence of seafloor gas ebullition) are prominent features on the WSM (ref. 15 and Fig. 1C ) and ESAS ( 2 ), (ii) maximum bottom-water methane concentrations at the WSM seeps ( Fig. 1B ; ∼300 nM) are comparable to summertime bottom-water concentrations in ESAS hotspots ( 2 ), and (iii) WSM gas could also have a component derived from thawing subsea permafrost ( 4 , 19 ), it is difficult to reconcile why the diffusive fluxes we report from the WSM differ so greatly from those of the ESAS. Methane may be more rapidly oxidized from the WSM water column ( 13 , 24 ); however, a more likely explanation is that lateral transport of methane from the relatively small and narrow WSM shelf dilutes and disperses methane into the deeper ocean. Similar dispersion and dilution on the shallow, expansive ESAS is not possible, which could permit a greater fraction of methane released from the seafloor to transfer to the atmosphere.

Surface water time series data for the shallow shelf survey. (A) Dissolved methane concentration and pCO 2 ; (B) δ 13 C−CH 4 and δ 13 C−CO 2 ; (C) SST temperature [dark green, hull-mounted sensor; light green, EXO2 sensor (YSI Incorported)] and fDOM; and (D) pH and DO. Seep crossings (highlighted with gray bars) are characterized by colder water containing elevated concentrations of 13 C-depleted methane and lower concentrations of 13 C-enriched CO 2 . Isotopic excursions are demarcated by dashed lines that connect δ 13 C values from the margins of the seep crossings. Within the seep crossings, fDOM, pH, and DO are elevated. The combined evidence suggests upwelling of cold, methane-charged (and presumably nutrient-rich) bottom water originating from the seep-stimulated phytoplankton activity in the surface water that enhanced the consumption of CO 2 . Similar trends occur within the nearshore coastal zone.

Within the shallow-water gas seep field, pCO 2 in the surface water was substantially less than in the surrounding area ( Figs. 2 C and D and 5A ) and correlates negatively with methane concentration (r 2 = 0.61; SI Appendix, Fig. S4 ). These undersaturated pCO 2 values support a CO 2 influx rate of −33,300 ± 7,900 μmol⋅m −2 ⋅d −1 ( Table 1 ), which is about twice that of the surrounding background area (−16,000 ± 6,000 μmol⋅m −2 ⋅d −1 ) and more than 1,900 times greater than the efflux of methane (17.3 ± 4.8 μmol⋅m −2 ⋅d −1 ). Taking into account the 25 times greater global warming potential of methane relative to CO 2 for a 100-y timescale on a per unit mass basis ( 18 ), the strongly negative CO 2 flux at the seep offsets the positive effect of methane expelled by a factor of 231 despite methane’s greater global warming potential. Even on a 25-y timescale, for which methane has stronger GWP of 84 ( 18 ), the cooling effect of CO 2 uptake is 69 times greater than methane’s warming effect. Our comparisons consider only the dissolved phase gas fluxes. However, hydroacoustic imaging ( Fig. 1C ) and bubble modeling ( SI Appendix, Fig. S8A ) suggest minimal direct bubble transport to the atmosphere. Furthermore, a recent study from the ESAS suggesting that turbulence-driven diffusive methane flux (not ebullition) is the primary transport mechanism for sea−air methane flux ( 17 ) supports our assessment that bubble transport of methane to the atmosphere is not important at this setting.

Stimulation of CO 2 Uptake over Shallow-Water Methane Seeps.

At least two processes could be responsible for the reduced concentrations of CO 2 observed over the shallow-water methane seeps: (i) Methane bubbles ascending from the seafloor dissolve methane, strip CO 2 from the water column, and transport this CO 2 to the sea−air interface and release it to the atmosphere (12), or (ii) a physical and/or biological mechanism stimulates photosynthesis, and thus CO 2 drawdown, above the seep area. To test the first hypothesis, we applied a numerical bubble-stripping model (12). Reproducing the low CO 2 concentrations requires (i) bubble diameters of 14 mm, which is much larger than the most frequent diameter of ∼6 mm (range 2 mm to 16 mm) observed in the area (27), and (ii) a volumetric gas flux of 34 L⋅m−2⋅min−1 from the seabed at 90 m (∼13.6 mol/min, at 4 °C), compared with reported values of 3 mL⋅min−1 to 41 mL⋅min−1 per seep at 385 m (5.4 mmol/min to 74.5 mmol/min, at 4 °C) (19). Bubble stripping is therefore not a plausible mechanism for removing CO 2 .

The alternate hypothesis for lower surface-water pCO 2 is that upwelling of cold, nutrient-rich water stimulated CO 2 assimilation by phytoplankton, a phenomenon also observed in areas of strong upwelling associated with eastern boundary currents of major ocean basins (28). Surface water within the high-methane, low-CO 2 seep area was 0.65 °C colder than the surrounding surface water (Figs. 4A and 5C and SI Appendix, Table S1), and the estimated δ13C of the seabed-sourced methane measured at the sea surface (−54.6‰; SI Appendix, Methane Isotopic Mass Balance for Determination of Seabed) was similar to that reported at the seafloor (29) and emanating from seeps downslope (19). We are therefore confident the cold and methane-rich surface water originated from near the seafloor close to the seep area. Furthermore, CO 2 uptake rates we measured (2,200 μmol⋅m−2⋅d−1 to 42,000 μmol⋅m−2⋅d−1; Table 1) are comparable to primary production rates reported from nearby Kongsfjorden (30) (600 μmol⋅m−2⋅d−1 to 184,000 μmol⋅m−2⋅d−1), confirming the plausibility that phytoplankton-related processes altered the surface water CO 2 budget. A possible subsurface manifestation of high surface productivity is that benthic chlorophyll and phaeopigment concentrations at this seep were the highest among nine stations investigated in the western Svalbard−Barents Sea region (31).

Upwelling on the WSM shelf is driven by Ekman transport during northerly or onshore wind events that can occur during any season (32). On a smaller scale, the topographically steered Spitsbergen Polar Current encountering the high-relief glacial moraine may upwell locally along steeply tilted isopycnals (Fig. 4D). Bubble-driven buoyancy and entrainment of bottom waters may also transport bottom water to the photic zone from depths as great as 1,000 m (33), a mechanism invoked to explain elevated surface-water chlorophyll above a Gulf of Mexico hydrocarbon seep (34). The relatively low seafloor methane flux at the WSM seep sites between 240 m and 385 m water depth (19) renders it unlikely that bubble-associated buoyancy caused the upwelling, supporting the assumption that physical oceanographic processes alone are responsible for upwelling, independent of the presence of gas seepage.

Regardless of the upwelling mechanism, multiple lines of evidence support the interpretation that primary production and consequent CO 2 drawdown are enhanced where methane-charged bottom water emerges: (i) Chlorophyll-fluorescence, a proxy for photosynthesis, is elevated (Fig. 4E); (ii) DO, a product of photosynthesis, is ∼1 mg/L higher in surface waters with high methane and low CO 2 concentrations (Fig. 5D); (iii) pH, which increases when CO 2 is removed from solution by photosynthesis, is elevated by as much as 0.6 units compared with background (Fig. 5D); and (iv) δ13C−CO 2 , a metric that becomes more positive when algae preferentially remove 12CO 2 during photosynthesis, is 13C-enriched (more positive) by as much as 2‰ within the upwelling area of methane-charged bottom water (Fig. 5B).

Similar, yet more pronounced, patterns of high methane, low CO 2 , and changes in water chemistry indicative of upwelling-induced photosynthesis were observed in the nearshore coastal zone (Fig. 5 and SI Appendix, Fig. S4). However, the coastal zone lacks pervasive discrete bubble-releasing methane seeps (Fig. 2A). Most methane in that region (up to 150 nM in bottom waters; Fig. 1B) likely originates from in situ production in organic-rich, anoxic sediment. Elevated methane in marine surface waters can also be a product of dimethylsulfoniopropionate demethylation (35), but the high bottom-water methane content and δ13C signature of the nearshore methane are most consistent with a sediment source.

Despite the spatiotemporal coincidence between high concentrations of methane and enhanced CO 2 uptake at seeps on the WSM continental shelf (Fig. 5A) and in some other settings such as the Santa Barbara Basin seep field (36), we suggest that high methane concentrations are an indicator of, but not a necessary condition for, enhanced CO 2 drawdown. Instead, the surface-water methane observed on the WSM is a chemical tracer for cold, nutrient-rich upwelled water that supports enhanced photosynthesis within the euphotic zone. A relationship of higher methane efflux and CO 2 influx that correlated with colder surface waters was also observed near the >2,000-m deep-water gas hydrate site (SI Appendix, Fig. S5). This observation suggests enhanced CO 2 drawdown is likely to occur whenever deep nutrient-rich (and perhaps methane-charged) waters are upwelled to the surface, and conditions for photosynthesis are suitable.