For a review and references

Introduction Conclusions References Tables Figures Back Close Full Screen / Esc Abstract Introduction Conclusions References Tables Figures Back Close Full Screen / Esc Abstract The coastal ocean constitutes the crucial link between land, the open ocean and the atmosphere. Furthermore, its shallow water column permits close interactions between the sedimentary and atmospheric compartments, which otherwise are decoupled at short time scales (<1000 yr) in the open oceans. Despite the prominent role of the 5 coastal oceans in absorbing atmospheric CO 2 and transferring it into the deep oceans via the continental shelf pump, the underlying mechanisms remain only partly understood. Evaluating observations from the North Sea, a NW European shelf sea, we provide evidence that anaerobic degradation of organic matter, fuelled from land and ocean, generates alkalinity (A T) and increases the CO 2 buffer capacity of seawater. At both the basin wide and annual scales anaerobic A T generation in the North Sea's tidal mud flat area irreversibly facilitates 7–10%, or taking into consideration benthic denitrification in the North Sea, 20–25% of the North Sea's overall CO 2 uptake. At the global scale, anaerobic A T generation could be accountable for as much as 60% of the uptake of CO 2 in shelf and marginal seas, making this process, the anaerobic 15 pump, a key player in the biological carbon pump. Under future high CO 2 conditions oceanic CO 2 storage via the anaerobic pump may even gain further relevance because of stimulated ocean productivity.


Introduction
Shelf and marginal seas constitute biogeochemically active environments linking fluxes of energy and matter between land, the open ocean and the atmosphere.These areas host high biological activity, which is fuelled by nutrient inputs from all three interfacing compartments (e.g.Wollast, 1998;Thomas et al., 2003;Pätsch and Kühn, 2008;Liu et al., 2009).Furthermore, shallow seas, in which the water column is mixed at the seasonal or annual time scale, establish a close link between surface sediments and the atmosphere.This permits relatively direct interactions between both the sedimentary and atmospheric compartments, which are otherwise strictly separated at long time scales ( ∼ =1000 yr) in the open oceans.It has been proposed that shelf and marginal seas, such as the East China Sea and North Sea, act as continental shelf pumps, transferring atmospheric CO 2 into deeper layers of the open ocean via physical or biological processes, or a combination of the two (Tsunogai et al., 1999;Thomas et al., 2004).However, as recently summarized (Borges, 2005;Borges et al., 2005Borges et al., , 2006;;Cai et al., 2006), a conclusive understanding has yet to be achieved on the role of shelf and marginal seas as sinks or sources for atmospheric CO 2 and of the underlying mechanisms.
Recent investigations describe the North Sea, in the NW European Shelf, as a strong continental shelf pump, facilitated through intense interaction between the deeper northern North Sea and the adjacent North Atlantic (Thomas et al., 2004(Thomas et al., , 2005a;;Bozec et al., 2006).On the other hand, the North Sea's shallow southern part is strongly affected by terrestrial influences such as riverine inputs (Borges and Published by Copernicus Publications on behalf of the European Geosciences Union.Frankignoulle , 2002;Schiettecatte et al., 2006Schiettecatte et al., , 2007)).Earlier studies indicate that the Wadden Sea, a tidal mud flat area, bordering the southeastern region of the North Sea, might influence carbon cycling in the Southern North Sea (Hoppema, 1990;Brasse et al., 1999;Reimer et al., 1999).
In the present study, we unravel the seasonal variability of total alkalinity (A T ) and pH in the North Sea, assess A T generation in tidal mud flat areas, and evaluate the effects of this A T generation on CO 2 uptake in the North Sea at basin wide, annual scales.

Methods
This study is based on an extensive field data set collected in 2001/2002 with spatial coverage over the whole North Sea region.The data set includes the full carbonate system and related parameters.The cruises were carried out on R/V Pelagia during all four seasons consecutively in August/September and November 2001 and February and May 2002.The entire North Sea was sampled by an adapted 1 • by 1 • grid of 97 identical stations resulting in highresolution data sets appropriate for assessing seasonal variability (e.g.Bozec et al., 2006).Approximately 750 samples per cruise were analyzed for A T by potentiometric determination and for dissolved inorganic carbon (DIC) by coulometric determination (Johnson et al., 1993).Uncertainties were estimated in the range of 2-3 µmol kg −1 ( ∼ =0.1%) for A T and 1.5 µmol kg −1 ( ∼ =0.08%) for DIC.Surface water partial pressure of CO 2 (pCO 2 ) was determined continuously (Körtzinger et al., 1996) with approximately 20 000 data points per cruise, and the uncertainty was estimated to be 1ppm (≈0.3%).Atmospheric pCO 2 was measured hourly.
The pH was calculated from DIC and pCO 2 (Dickson and Millero, 1987).More detailed descriptions of the methods used have been reported elsewhere (e.g.Bozec et al., 2006).
The variation of A T ( A T ) in the southeastern bight of the North Sea (Fig. 1) was computed with the following equation: where δA T (mix) represents the exchange between the southeastern bight and the adjacent central North Sea.The δA T (mix) was computed by considering that the water in the southeastern bight is mixed at time scales of six weeks with water of the adjacent open North Sea (Lenhart et al., 1995).
The southeastern bight encompasses areas south of 57.5 • N and east of 5 • E (Figs. 1, 2).The concentrations of the adjacent Central North Sea were averaged from data obtained in an area between 2 • E to 5 • E and 55 • N to 57.5 • N. The A T was linearly interpolated between the measurements obtained every 3 months.The δA T (Riv) represents the riverine inputs and was computed for the observational time period according to Pätsch and Lenhart (2004).The alkalinity inputs from the Wadden Sea (δA T (WaddenSea)) are computed as closing term of Eq. (1).Furthermore, we consider the effect of nitrate (NO − 3 ) uptake and release during new production and aerobic respiration of organic matter on A T (Goldman and Brewer, 1980).In the view of Dickson's definition (Dickson, 1981), A T is altered by the uptake/release of H + , paralleling the uptake/release of NO − 3 because of the required conservation of electrical charges.Sources of NO −  Panel (e) shows the annual mean distribution of A T , relative to which the anomalies have been computed.The box borders the southeastern bight (see methods for details).The color scale is identical in all panels (a-d).The surface area and volume of the southeastern bight are 150 392 km 2 and 4512 km 3 , respectively, computed from the bottom depth distribution (Thomas et al., 2005b).
The surface area and volume of the remainder of the North Sea surface waters are 424 908 km 2 and 12 747 km 3 , assuming a 30 m deep surface layer and a total surface area of the North Sea of 575 300 km 2 (Thomas et al., 2005a).The area of the Wadden Sea has been estimated at 5000 km 2 .

Results and discussion
The A T shows relatively homogenous distributions during each of the seasons in most parts of the North Sea (Fig. 2).
In general, the seasonal variability of A T is relatively low  compared to the seasonal variability of DIC as reported by Thomas et al. (2005b) or Bozec et al. (2006).This signal is perturbed in the southeastern bight, in particular during spring and summer, when high values of A T were observed.Furthermore, reduced A T concentrations were observed during spring, and to a lesser extent, in summer in the northeastern area, which can be attributed to enhanced river and Baltic Sea runoff due to the low A T concentrations of Scandinavian rivers discharging into the Baltic Sea (Thomas et al., 2003).The seasonal cycle of A T in the southeastern bight of the North Sea reveals increasing concentrations from spring to late summer, with maximum values observed in August (Figs.2, 3).In autumn and winter, the A T concentrations decline to minimum values observed in February.In order to evaluate the processes governing the seasonal variability of A T in the southeastern bight, we consider the terms of Eq. ( 1) as the relevant drivers (Fig. 3, Table 1).River inputs make the highest contributions during late winter and spring, Table 1.Seasonality of total alkalinity in the Southern Bight of the North Sea.A T release from the Wadden Sea and from rivers entering the southeastern bight to the North Sea according to Eq. (1) (see also Fig. 2).For comparison, we give the A T changes due to nitrate uptake during (new) nitrate assimilation for selected nitrate sources.Negative values correspond to export to the adjacent North Sea and to NO − 3 release during aerobic respiration of organic matter, respectively.We assumed that 50-80% of the riverine nitrate is available for new production, while 20-50% undergoes denitrification (Beusekom and de Jonge, 2002) Sea.Input from the Wadden Sea makes a positive contribution to A T from late spring until late autumn, while the Wadden Sea acts as a sink for A T during winter and early spring.Assimilation of new nitrate increases A T from late winter until late spring, while aerobic respiration releases nitrate and consequently lowers A T during the remainder of the year (δA T (column NO − 3 )).The contribution from nitrate assimilation or respiration is weaker than, and approximately six months (i.e., 180 degrees) out of phase with the Wadden Sea A T release.The final two terms in Eq. ( 1), uptake of riverine and atmospheric nitrate, play a minor role in controlling A T in the southeastern bight of the North Sea.
Several processes can account for the generation of A T in the Wadden Sea, part of which is linked to the anaerobic degradation of organic matter.In the relatively wellventilated Wadden Sea, only anaerobic processes can irreversibly generate A T , which yield products, either resisting or escaping re-oxidation by oxygen.The primary candidates for these processes are a) denitrification, which generates quasi-inert elemental nitrogen (N 2 ), which in turn can then escape to the atmosphere; and b) sulfate reduction, which generates dihydrogensulfide (H 2 S), which can either escape to the atmosphere, when sediments are exposed at low tide (Kristensen et al., 2000), or which can be buried as pyrite.
Sulfate reduction releases 1.98 mol A T per mol SO 2− 4 reduced (Chen and Wang, 1999) and has been estimated in a range of 6-13 Gmol S yr −1 in the Wadden Sea area (Kristensen et al., 2000;de Beer et al., 2005), which corresponds to a production of 12-26 Gmol A T yr −1 (1 Gmol=10 9 mol).Denitrification, which releases 0.99 mol A T per mol N denitrified (Chen and Wang, 1999), has been estimated in the Wadden Sea to be approximately 100 Gmol N yr −1 (Jensen et al., 1996), yielding a release of 99 Gmol A T yr −1 .Denitrification in the Wadden Sea is partly fuelled by riverine nitrate inputs (37 Gmol N yr −1 from major rivers, Pätsch and Lenhart, 2004), of which approximately 20-50% are denitrified within the Wadden Sea itself (Beusekom and de Jonge, 2002;Seitzinger et al., 2006), thus contributing up to 19 Gmol A T yr −1 to the total of 99 Gmol A T yr −1 .Further nitrate can be supplied from the open North Sea, smaller rivers or ground water (Slater and Capone, 1987).A further process, which releases A T , is the dissolution of calcium carbonate (CaCO 3 ).This dissolution is due to the acidification of pore waters related either to the release of CO 2 by respiration, or to the release of H + from the oxidation of H 2 S within the oxic sediment layers (Jahnke and Jahnke, 2000).The estimated production of calcareous shells, that could be dissolved in the Wadden Sea, is quite low -on the order of 1 mol CaCO 3 m −2 yr −1 , or 5 Gmol CaCO 3 yr −1 (Beukema, 1982).Moreover, the autochthonous production of CaCO 3 would equal more or less the dissolution of CaCO 3 , hence would not yield any net A T gain of the Wadden Sea at the complete annual cycle.Otherwise to the best of our knowledge, the supply of allochthonuous CaCO 3 to the Wadden Sea has not been reported and therefore is not further considered.
The estimates of the potential A T sources from denitrification and sulfate reduction in the Wadden Sea slightly exceed the identified A T fluxes (Table 1).However, there is some uncertainty inherent to these estimates as well as to our assessment, for example related to the exchange time of the southern bight or to the temporal coverage of our study (e.g., Schiettecatte et al., 2007).As pointed out by Hofmann ( 2008), Chen and Wang (1999) neglect the contribution of ammonia release to the A T generation.Accordingly, the release of ammonia by denitrification and sulfate reduction would add approximately another 10% to the A T release.On the other hand, part of this ammonia would be subject to oxidation (nitrification) or biological assimilation, which in turn consumes A T .Moreover, our results indicate that the southeastern bight acts as a sink for A T during autumn and winter, which can be attributed to the re-oxidation of the products of sulfate reduction (Luff and Moll, 2004;de Beer et al., 2005).In autumn and winter, weather conditions enforce the intrusion of oxygen into the shallow surface sediments, permitting the re-oxidation of H 2 S to become the dominant process, while denitrification and sulfate reduction activity is lower due to the decrease in temperature and supply of organic matter.Earlier studies for example show that the oxidation of H 2 S may range from 80-95% of the H 2 S generated by sulfate reduction (e.g.Jorgensen, 1982), however it has to be noted here that this particular study focuses on sediments, which are continuously covered by water, thus preventing H 2 S from escaping to the atmosphere.We therefore argue that the current assessment might serve as a conservative, lower bound of the A T release from the Wadden Sea to the open North Sea.It can be concluded that in the North Sea, and particularly in its shallower areas, A T exerts a much stronger seasonal control on pH (Fig. 4) and surface pCO 2 than in open ocean areas, where the seasonality of DIC dominates.During the transition from winter to spring in the southeastern bight of the North Sea, the DIC uptake (Bozec et al., 2006), and the related A T increase (due to NO − 3 uptake) synergistically raise the pH and lower the pCO 2 .In spring, when large amounts of freshly produced organic matter are available, anaerobic degradation stimulates the release of large amounts of A T .This supports the efficient DIC drawdown in maintaining low pCO 2 concentrations, as seen in late spring (Thomas et al., 2004).In the summer period, the effect of DIC release by decaying phytoplankton blooms is buffered by the increasing release of A T from the tidal flat area.This buffering prevents a strong pH decline and acts to maintain low to neutral air-sea fluxes of CO 2 , despite higher summer temperatures and degradation of organic matter (Bozec et al., 2006).Outside the southeastern bight, i.e. in areas remote from the tidal mud flat areas, the control of DIC on pH and pCO 2 becomes more significant.This is evident, for example, in the enhanced summer values of pH in the Northern North Sea (Fig. 4) (see also Thomas et al., 2004), or from the declining summer values of pH in the Southwestern North Sea, respectively: in the  Northern North Sea the biologically driven CO 2 export out of the surface layer maintains higher pH conditions, while the southwestern region is hardly affected by any A T release and thus reveals declining pH conditions because of organic matter degradation in a permanently mixed compartment.
The difference between anaerobic and aerobic degradation of organic matter plays a crucial role in governing CO 2 air-sea fluxes.This is of particular importance on shelves and in shelf sea regions, where the surface sediments, acting as sinks for organic matter and sources of CO 2 , are in direct or close contact to the atmosphere.This is in contrast to the open ocean situation, where there is no interaction www.biogeosciences.net/6/267/2009/Biogeosciences, 6, 267-274, 2009 between the atmosphere and the sediment at short time scales (<1000 yr).Aerobic degradation has only a minor effect on A T through nitrate release: considering Redfield stoichiometry, aerobic degradation decreases A T by 16/106 mol per mol organic carbon remineralized.Aerobic degradation thus releases DIC that is virtually unbuffered, i.e., without a significant change of A T , thereby largely cancelling out biological DIC drawdown during the preceding production of organic matter at the annual scale.In case of anaerobic degradation of organic matter, for example by denitrification or sulfate reduction, the DIC release is well buffered, i.e., DIC is released together with A T .Since denitrification, and to some extent, sulfate reduction, are irreversible, if their products are buried or escape to the atmosphere, anaerobic degradation of organic matter constitutes a net A T gain at the annual scale.This net gain of A T then facilitates net CO 2 uptake from the atmosphere.
The release of 73 Gmol A T from the Wadden Sea, which has an area of approximately 1% of the North Sea surface area, facilitates approximately 7-10% (0.2 mol C m −2 yr −1 ) of the annual CO 2 uptake of the North Sea (Thomas et al., 2004).Considering the recent estimate for benthic denitrification in the entire North Sea (excluding the Wadden Sea) of 120 Gmol N yr −1 (Pätsch and Kühn, 2008), this corresponds to an A T release of 119 Gmol yr −1 , or in turn may account for circa 15% of the annual CO 2 uptake.In total, anaerobic organic matter degradation has the potential to irreversibly facilitate as much as 20-25% of the CO 2 uptake of the North Sea.The global marine denitrification in shelf areas has recently been estimated at 250 Tg N yr −1 (18 Tmol N yr −1 ) (Seitzinger et al., 2006), which corresponds to a potential CO 2 uptake of 15 Tmol CO 2 , or 0.6 mol CO 2 m −2 yr −1 , assuming a Revelle factor of 11.Our estimate thus covers the lower bound of the independent estimate by Chen (2002).This corresponds to circa 60% of the total CO 2 uptake in shelf and marginal seas (0.3 Pg C yr −1 ) (Borges, 2005;Borges et al., 2005;Cai et al., 2006;Liu et al., 2009), thus exceeding the effects of both atmospheric reactive nitrogen and sulfur depositions on the carbon cycle in the coastal ocean by one to two orders of magnitude (Doney et al., 2007).Benthic A T generation may therefore constitute an anaerobic pump that is a major driver for the uptake of atmospheric CO 2 on shelves and in shelf sea regions.
The evolution of riverine A T inputs to the ocean under future anthropogenic or climate change scenarios remains uncertain.Damming of rivers tends to hinder A T transfer to the coastal ocean (e.g.Humborg et al., 2002), while on the other hand higher erosion in the catchment area related to land use changes might enhance A T delivery to these regions (Raymond and Cole, 2003).With regard to benthic A T release, recent findings reporting a stimulation of organic matter production by rising CO 2 concentrations, become relevant (Riebesell et al., 2007).This enhanced organic matter supply has the potential to stimulate benthic, anaerobic respiration with subsequent A T release and (net−) CO 2 uptake.
Furthermore, global warming has the potential to intensify stratification.This would favour the expansion of oxygen depleted zones and in turn stimulate anaerobic processes.Finally, primary production in near-shore shallow coastal environments is expended to continue to increase due to eutrophication (e.g.Mackenzie et al., 2004) with strong impacts on the carbonate chemistry and air-sea CO 2 fluxes (Mackenzie et al., 2004;Gypens et al., 2008).The increase of primary production due to eutrophication is also expected to enhance anaerobic A T generation.The enhancement of benthic A T generation would thus result in negative feedbacks to rising atmospheric CO 2 conditions, by enhancing the capacity of coastal oceans to uptake anthropogenic CO 2 establishing a further pathway for the continental shelf pump.
et al., Enhanced ocean carbon storage from anaerobic alkalinity Fig. 1

Fig. 1 .
Fig. 1.Total Alkalinity budget of the southeastern bight of the North Sea.The upper panel shows the investigation area.The lower panel reveals the major fluxes, contributing to the alkalinity (A T ) budget.Direct A T transports are shown as normal font, while A T changes due to uptake/release of NO − 3 are shown in italics.

Fig. 2 .
Fig. 2. Seasonality of total alkalinity.Seasonal anomalies of A T , (a) winter (February), (b) spring (May), (c) summer (August/September) and (d) autumn (November) from observations.Panel (e) shows the annual mean distribution of A T , relative to which the anomalies have been computed.The box borders the southeastern bight (see methods for details).The color scale is identical in all panels (a-d).The surface area and volume of the southeastern bight are 150 392 km 2 and 4512 km 3 , respectively, computed from the bottom depth distribution(Thomas et al., 2005b).The surface area and volume of the remainder of the North Sea surface waters are 424 908 km 2 and 12 747 km 3 , assuming a 30 m deep surface layer and a total surface area of the North Sea of 575 300 km 2(Thomas et al., 2005a).The area of the Wadden Sea has been estimated at 5000 km 2 .
., Enhanced ocean carbon storage from anaerobic alkalinity generation in coastal sediments Fig.3

Fig. 3 .
Fig. 3. Processes controlling the seasonal variability of total alkalinity in the southeastern bight of the North Sea according to Eq. (1).The seasonal cycle of A T (black line and symbols), as well as the seasonal cycle of the Wadden Sea A T input (δA T (Wadden Sea), green line and symbols), riverine A T inputs (δA T (Riv), blue line and symbols) as well as the A T release to the Central North Sea (δA T (mix) , red lines and symbols) are shown.The grey lines and symbols denote A T changes due to biological uptake of the water column NO − 3 , (δA T (column NO − 3 ), riverine NO − 3 , (δA T (riv.NO − 3 ), and atmospheric NO − 3 inputs (δA T (atm.NO − 3 ).
with declining relevance from late spring until the following winter.With the exception of the rather homogenous winter situation, the open North Sea reveals lower A T concentrations than the southeastern bight throughout the entire year as indicated by the negative values of the δA T (mix) term, hence, the southeastern bight releases A T to the open North www.biogeosciences.net/6/267/2009/Biogeosciences, 6, 267-274, 2009 ., Enhanced ocean carbon storage from anaerobic alkalinity..., Fig.4

Fig. 4 .
Fig. 4. Seasonality of pH.Seasonal anomalies of pH, (a) winter (February), (b) spring (May), (c) summer (August/September) and (d) autumn (November) calculated from observations.The box borders the southeastern bight (see methods for details).The color scale is identical in all panels (a-d).Panel (e) shows the annual mean distribution of pH, relative to which the anomalies were computed.
. All units are in [Gmol A T ].