The Siberian shelf seas cover large shallow areas that receive substantial amounts of river discharge. The river runoff contributes nutrients that promote marine primary production, but also dissolved and particulate organic matter. The coastal regions are built up of organic matter in permafrost that thaws and result in coastal erosion and addition of organic matter to the sea. Hence there are multiple sources of organic matter that through microbial decomposition result in high partial pressures of CO2 in the shelf seas. By evaluating data collected from the Laptev and East Siberian Seas in the summer of 2008 we compute an excess of DIC equal to 10 · 1012 g C that is expected to be outgassed to the atmosphere and suggest that this excess mainly is caused by terrestrial organic matter decomposition.
 The Siberian Shelf Seas is a highly dynamic region where significant carbon transformation occur that impact the transport to the deep central Arctic Ocean as well as the atmospheric exchange. However the substantial input of both dissolved and particulate organic matter to the shelf seas [Macdonald et al., 2008], especially during the ice free summer season, hampers light penetration and results in moderate marine primary production outside the river mouths of the great Siberian rivers. Hence, terrestrial organic matter dominates over marine produced organic matter as a source of CO2 to the atmosphere in this area.
 Here we show that significant microbial decay of terrestrial organic matter occurs in the Siberian Shelf Seas, which results in a substantial flux of CO2 to the atmosphere. Our findings suggest that the release of CO2 from decaying terrestrial organic matter within the Siberian Shelf Seas could be an important and so far underestimated source of CO2 to the atmosphere. This source of CO2 will likely increase, as predicted results of future warming include increased permafrost thawing, river discharge and coastal erosion.
 In the summer of 2008 (August 15 to September 26) the International Siberian Shelf Study was conducted with the objective to investigate the flux and transformation of carbon from land over the shelf seas and into the deep central basins of the Arctic Ocean. An extensive sampling program was undertaken on board the Russian vessel Yacob Smirnitskyi in the waters of the Laptev Sea (LS), East Siberian Sea (ESS) and Chukchi Sea (CS) (Figure 1). At 96 stations, depth profiles of nutrients, oxygen, pH, total alkalinity (TA) and dissolved inorganic carbon (DIC) were collected and determined on board using state of the art analytical techniques. The data are archived at the PANGEA information system under the EU project European Project on Ocean Acidification (EPOCA).
 DIC was determined by coulometric detection of CO2 extracted from a known mass of acidified seawater [Johnson et al., 1987], with the precision as determined by duplicate samples typically being ±1 μmol kg−1. The accuracy was set by analysis of Certified Reference Material (CRM), supplied by A. Dickson, Scripps Institution of Oceanography (USA) and should hence be about twice that of precision. Total alkalinity was determined by potentiometric titration [Haraldsson et al., 1997], precision ∼1 μmol kg−1, the accuracy set with CRM. pH was determined by a spectrophotometric method [Clayton and Byrne, 1993; Lee and Millero, 1995] with a precision ∼0.003 pH units. The partial pressure of CO2 (pCO2) was computed from pH and total alkalinity using the software CO2SYS [Lewis and Wallace, 1998]. The carbonate dissociation constants (K1 and K2) used were those of Roy et al.  as they show the best internal consistency in the low temperature waters of the Arctic Ocean when using any two of pH, DIC or TA as input parameters. Oxygen was determined using an automatic Winkler titration system, giving a precision of ∼1 μmol kg−1. The [O2]saturation was calculated using the equation of Weiss . The nutrients were determined on board using an automatic spectrophotometric system (SmartChem from Westco). The samples were filtered before measured and evaluated by a 6 to 8-points calibration curve, precision being about 1%.
 The excess DIC in μmol kg−1 was computed as the difference between the DIC of a water in equilibrium with an atmospheric pCO2 of 385 μatm, all other properties being as observed, and the measured DIC. This excess was then integrated throughout the water column to get the unit gC m−2.
3. Results and Discussion
 Most of the surface waters of the LS were oversaturated relative to atmospheric CO2 levels (>385 μatm, green in Figure 1). This was also the situation for the waters of the western ESS, while the ones in the eastern part were under-saturated (Figure 1). The latter waters had relatively high salinities (>25 psu) that are typical for Arctic shelf surface waters strongly impacted by inflow from the Pacific Ocean through Bering Sea.
 The oxygen concentration was oversaturated in the surface water of the eastern ESS where pCO2 was under-saturated, illustrated by the Apparent Oxygen Utilization (AOU = [O2]saturated − [O2]measured) concentration being negative (Figure 2). The phosphate concentration was high, around 1 μmol kg−1, while the nitrate concentration was low, below 0.5 μmol kg−1. The nutrient concentrations were about the same in the surface water of western ESS, while the AOU was close to zero (oxygen being saturated) and pCO2 were oversaturated. In the LS the phosphate concentration was below 0.1 μmol kg−1 in the surface water but the nitrate concentrations were about the same as in the ESS. In all sections there was a signature of mineralization of organic matter in the bottom waters, with high values of nutrients, AOU and pCO2, even if this was strongest in the ESS.
 The surface water pattern of (1) low nutrient concentration, oxygen at saturation and high pCO2 in the LS, (2) high nutrient concentration, oxygen at saturation and high pCO2 in the western ESS, and (3) high nutrient concentration, oxygen over-saturation and low pCO2 in the eastern ESS, strongly indicate that more than marine primary productivity and the corresponding decay of the produced organic matter was active, at least in the first two regions. However, this does not mean that marine primary production does not occur in all regions, only that the typical chemical signatures during the productive season has been modified by yet another process.
 The organic matter that is mineralized at the sediment surface results in high values of nutrients, AOU and pCO2 that can be of both terrestrial and marine origin. However, since CO2 is oversaturated in the surface waters of the LS and in the western ESS, either the decay of organic matter exceeds that of marine primary production, or this signal is a result of vertical mixing of oversaturated sub-surface water. Either way, a corresponding nutrient supply to the surface would be expected. As this is not observed, the only plausible explanation is decay of terrestrial organic matter with a nutrient content comparatively low relative to marine organic matter.
 In the eastern ESS, it is clear that marine primary productivity is the main process behind the surface water signature. The front between over- and under-saturation of pCO2, marked as a solid line in Figure 1, agrees with the position of the δ13Corg isoline of −24.5 ‰ (Figure 3). This line represents the boundary between the “typical terrestrial” δ13Corg values (lighter than −24.5 ‰) in the western ESS and the “typical marine” values in the eastern ESS [Naidu et al., 2000; Semiletov et al., 2005]. As the western ESS and the LS is the area of high surface sediment content of terrigenous particulate material [Semiletov et al., 2005] and low surface water salinity, we argue that the observed pCO2 signature is dominated by decay of terrestrial organic matter added both by runoff and by coastal erosion. Decay of terrestrial organic matter has earlier been suggested to cause high pCO2 levels [Semiletov et al., 2007], but then primarily in the bottom waters of the LS with values up to more than 2000 μatm.
 In order to quantify the carbon released from decaying organic matter, the excess of DIC was computed (Figure 4). The result shows extensive excess in the western ESS and in the LS but a deficit in the south-eastern ESS and southern CS (above zero in Figure 4). These signals also to a great extent reflect the observations of the surface layer. However, in the northern CS and north-eastern ESS the extreme excess is dominated by deep and bottom waters, which illustrates the export of dissolved carbon to the deep central basin. No sampling of the slopes of the western ESS or the LS was done and hence it cannot be ruled out that export occurs also in these regions.
 If integrated, the excess of DIC that is observed in waters shallower than 50 m, i.e., waters that are largely homogenized in the autumn due to wind mixing and brine release from sea ice formation gives an excesses for half of the ESS of ∼5 · 1012 g C and about the same for the whole LS. Hence a potential CO2 out-gassing of 10 · 1012 g C to the atmosphere could occur within a year if the water mixes and open water persist during the summer season. This number is of the same order as the reported input of total organic carbon to the LS and ESS ∼10 · 1012 g C yr−1 from runoff and ∼4 · 1012 g C yr−1 from coastal erosion [Rachold et al., 2004], to which marine produced organic carbon in the order 45 · 1012 g C yr−1 is added [Sakshaug, 2004]. However, the latter number is highly uncertain as it for the ESS is calculated by extrapolation of data from the neighboring seas.
 We have shown that in the summer of 2008 there was an excess of DIC equal to 10 · 1012 g C in the LS and ESS of the Arctic Ocean as a result of decaying organic matter. The fate of this excess will ultimately be an out-gassing to the atmosphere as it is found in the low salinity surface water. Some of the excess will likely be trapped under the sea ice as the water moves into the central Arctic Ocean, but released when the water is exposed to the atmosphere again. Furthermore we suggest that the transport and decay of particulate eroded terrestrial organic carbon plays a significant role in this CO2 excess. The rationale being that the nutrient and oxygen signatures indicate a limited marine organic matter decomposition and that the eroded organic matter to a large degree is biodegradable [Guo et al., 2004; van Dongen et al., 2008], whereas riverine DOC is more stable and mainly composed of soil-derived humic substances [Dittmar and Kattner, 2003]. However, recent findings suggest terrestrial DOC to be variable in lability [Raymond et al., 2007].
 It is plausible that the previous estimates of terrestrial particulate organic matter are on the low side in view of the retreating summer sea ice coverage (http://nsidc.org/arcticseaicenews/) and the increased coastal and river bank erosion. The increased river discharge [Savelieva et al., 2000; Peterson et al., 2002] furthermore might bring more humic substances that will decrease transparency and primary production. Some of this effect could be compensated for by increasing nutrient supply, mainly phosphate and silicate, by the runoff [e.g., Frey and McClelland, 2009] that could boost primary production. A continuing warming adds more terrestrial organic matter to the Arctic Shelf Seas, which increases pCO2, at the same time as decreased transparency lowers primary production, which reduce consumption of CO2. Both these effects result in a positive feedback by out-gassing CO2 over the Siberian Shelf, which comprises one half of the entire Arctic Ocean shelf area.
 This work was carried out by logistic support from the Knut and Alice Wallenberg Foundation and from the Swedish Polar Research Secretariat. The science was supported by the Swedish Research Council and the European Union projects, CarboOcean (contract no 511176-2), DAMOCLES (contract 018509-2) and EPOCA (contract 211384). Publication 27 from Tellus, The Centre of Earth Systems Science at University of Gothenburg.