Extraordinary slow degradation of dissolved organic carbon (DOC) in a cold marginal sea

Dissolved organic carbon (DOC) is the largest organic carbon reservoir in the ocean, and the amount of carbon in this reservoir rivals that in atmospheric CO2. In general, DOC introduced into the deep ocean undergoes a significant degradation over a centennial time scale (i.e., ~50 μM to ~34 μM in the North Atlantic and Mediterranean Sea). However, we here show that high concentrations of DOC (58 ± 4 μM) are maintained almost constantly over 100 years in the entire deep East/Japan Sea (EJS). The degradation rate in this sea is estimated to be 0.04 μmol C kg−1 yr−1, which is 2–3 times lower than that in the North Atlantic and Mediterranean Sea. Since the source of DOC in the deep EJS is found to be of marine origin on the basis of δ13C-DOC signatures, this slow degradation rate seems to be due to low temperature (<1 oC) in the entire deep water column. This observational result suggests that the storage capacity of DOC in the world ocean is very sensitive to global warming and slowdown of global deep-water overturning.

significantly higher than those in the major oceans (34-43 μ M) 6,17,18 (Fig. 2) and slightly higher than or similar to those in the Arctic Ocean (54 ± 3 μ M, up to ~4000 m) 19 and the Nordic seas (50 μ M, up to ~3500 m) 20 , where deep water formation occurs. In the Greenland Sea, one of the Nordic seas, the concentrations of DOC decrease from the surface (60 μ M) to the deep layer (53 μ M, < 1500 m depth) and reach a constant DOC concentration (50 μ M) in the deeper layer (> 1500 m depth) 20 . Therefore, the concentration of DOC in the deep bottom water of the EJS seems to be the highest reported so far in the world oceans.
In the Arctic and Nordic Seas, the elevated concentrations of DOC are related to downwelling of waters that are enriched in terrigenous DOC 19,20 . However, there is no large river that directly drains into the EJS. In addition, export of DOC in the deep EJS to the Pacific Ocean can be ignored since the EJS is connected to the Pacific Ocean only through the sills which are shallower than 150 m. The δ 13 C-DOC in the EJS ranged from − 20.4 to − 21.7% (avg.: − 21.3 ± 0.4%), similar to that in the North Pacific (avg.: − 21.2 ± 0.2%) and in the North Atlantic (avg.: − 21.0 ± 0.3%) oceans 21 (Fig. 2). The enriched δ 13 C values in these oceans are different from terrigenous C 3 plant material (− 28 to − 30%) indicating that the main fraction of the DOM is of marine origin (δ 13 C values of plankton organic matter = − 18 to − 20%). However, we cannot exclude the possibility that the Tsushima Current, which brings Yangtze River water into the EJS, may contribute some terrigenous DOC to the EJS. In addition, the DOC production from sinking particulate organic carbon (POC) could result in high DOC in the EJS. However, despite most of the sinking POC (>80%) being generally degraded between 200 and 1600 m 22 , the DOC profiles were uniform throughout the entire depths, down to 3500 m, indicating that DOC production in the deep ocean is not significant over the time scale of deep water mixing. The total export production (including sinking particles) through 200 m in the EJS has been estimated to be 99 Tg yr −1 (8.25 mol C m −2 yr −1 ) by using 3 H-3 He isotopes 22 , which in general is not unusually higher than in other oceans.
The concentrations of DOC in the deep layer of the EJS in May 2007 decreased slightly from the north (59 ± 3 μ M) to the south (55 ± 2 μ M) along the deep-sea current flows, which was associated with biological degradation reflected in apparent oxygen utilization (AOU) (Fig. 3). Considering the fact that these DOC profiles include the entire East Sea through several cruises, including the southwestern EJS stations 23 , over the last 10 years, the DOC concentrations in the deep EJS were remarkably stable, and neither systematic nor significant differences were observed among the stations.
In order to determine the characteristics of DOC degradation in the EJS, it was necessary to compare DOC degradation rates for the same DOM quality and the same age following deep water formation. In this connection, the degradation rate of DOC in the deep EJS was compared with those of the Atlantic Ocean and the Mediterranean Sea, which have similar deep water formation modes (thermohaline circulation). Because the turnover times of the EJS and Mediterranean Sea are about 100 years, we used the DOC difference in the entire regions of these marginal seas. In the case of the Atlantic Ocean, we used the difference in DOC concentration between the Greenland Sea water and the water mass around 50 °N , which was found to be approximately 100 years old based on 14 C dating 24 .
The slopes of DOC concentrations over the time scale of 100 years were obtained from the highest (59, 50, and 48 μ M) to the lowest (55, 42, and 34 μ M) concentrations in the EJS, Atlantic Ocean 25 , and Mediterranean Sea 26 , respectively. A major fraction of the DOC mineralization was assumed to have occurred during downwelling. Then, the degradation rate of DOC was estimated to be 0.04, 0.08, and 0.14 μ mol C kg −1 yr −1 in the EJS, Atlantic Ocean, and, Mediterranean Sea respectively. The degradation rate in the EJS is much lower than those of the Mediterranean Sea and the Atlantic Ocean.
In order to determine the contribution of DOC oxidation to oxygen consumption in deep EJS, DOC is plotted against AOU. The plotting was performed after converting AOU to carbon equivalents [AOU-C eq (μ M C) = AOU (μ M O 2 ) × 0.72] 16 . A Δ C/Δ O ratio of 0.72 was derived from the Redfield stoichiometry (C:O:N:P = 106:42:16:1) 27 . The slope (− 0.13) of the linear regression of DOC concentration versus AOU-C eq in the deep EJS indicates that oxidation of DOC is responsible for 13% of the oxygen utilization in the deep EJS. This value in the deep EJS is similar to that (14 ± 3%) in the North Atlantic Deep Water 25 , but much higher than that (approximately 32%) in the eastern Mediterranean Sea 28 .
The reason behind the low degradation rates remains speculative at this point, but it is most likely associated with the difference in water temperature. The water temperature in the EJS is <1 °C (Fig. 2), which is lower than that in the Atlantic Ocean (3 °C , range from 2.2-3.5 °C ) and in the Mediterranean  32 . Similarly, the rate of anthropogenic CO 2 accumulation in the deep layer of the EJS decreased considerably from 1992 to 2007 owing to the considerable weakening of the overturning circulation 33 . It is unknown whether this recent small change has resulted in DOC reduction in the deep EJS during the last few decades. However, our results from EJS in comparison with other seas suggest that global carbon cycles could be significantly affected by the warming and slow-down of water overturning in marginal seas.

Methods
The samples for DOC and δ 13 C-DOC were filtered onboard through a pre-combusted syringe glass-fiber filter (Whatman, 0.7 μ m, 25 mm) into pre-combusted glass ampoules (550 °C for 5 h). The samples were then acidified to pH 2 with 6 M HCl and stored at 4 °C for preservation until further analysis. All sampling procedures were performed in a clean bench (class 100). The DOC concentrations were measured using a TOC-V CPH analyzer (Shimadzu, Japan), as described elsewhere 34 . Briefly, the acidified seawater sample was bubbled with high-purity air gas (purity: 99.999) to completely purge inorganic carbon species in the injection system. Carrier gas was passed at a controlled flow rate of 150 mL min −1 . Three to five replicates (100 μ L) for each sample were automatically injected into a combustion glass tube filled with a catalyst (Pt-coated Al) at 720 °C in a furnace. The organic matter present in the sample was oxidized on the catalyst to CO 2 . The evolved CO 2 was measured by non-dispersive infrared detection (NDIR) by integrating peak area.
In order to determine DOC concentrations with high accuracy, it is necessary to reduce the system blank as low as possible. The blank run was continued until organic-free distilled water (DIW) was stable within the detection limit (< 5 μ M) of our DOC method. Six-point calibration curves for acetanilide were used daily for DOC standardization. The reliability of the measurements was verified on a daily basis by comparing the measured values with a DOC-certified seawater sample (DSR: 44-46 μ M for DOC, University of Miami) and procedural blanks. Most of our DSR measurement results (n = 34, 92%) were in good agreement with the DSR values (within 2%), although three measured values differed by 3-5% from the DSR values. These DSR measurements were performed for every 10 samples. The average standard deviation value for all DSR measurements was ± 1.2 μ M, which is similar to the value (± 1 μ M) suggested by Carlson et al. (ref. 25). An inter-laboratory comparison between the two laboratories involved (Kim and Dittmar) confirmed the accuracy of measurements to be within 5%.
The stable isotope compositions of dissolved organic carbon (δ 13 C-DOC) were measured by using TOC-IR-MS (Isoprime IR-MS coupled with vario TOC cube), which is commercially available (Isoprime, Elementar). This instrument uses the common high temperature catalytic combustion method, which is the same principle as our DOC method. Briefly, 10 mL of filtered samples, following acidification to pH 2 with 6 M HCL, is placed into an auto-sampler. The instrument is programmed to purge the sample for 20-30 min with O 2 gas to completely remove dissolved inorganic carbon species. Then, 1.5 mL of the sample is injected into Pt-impregnated silica spheres in a quartz tube, where DOC is fully converted to CO 2 at 750 °C . The CO 2 gas flows through a halogen trap and a water trap. After detecting DOC concentrations using an NDIR detector, CO 2 gas enters the interface by O 2 carrier gas. In the interface, a series of valves are set so that the flow from the TOC and the flow toward the IR-MS system can be separated. The CO 2 gas, carried by O 2 carrier gas, is trapped by CO 2 column in the interface. After switching the valves, the trapped gas is released to IR-MS by helium carrier gas. After passing through a reduction furnace to remove any interfering species, including residual oxygen, CO 2 gas is transferred to the IR-MS. The δ 13 C content of the CO 2 is measured using Isoprime IR-MS.
The ion source parameters were fine-tuned every day, and the variations of the reference gas were checked to maintain good stability (standard deviation: < 0.01%). The blank was determined by running low carbon water (University of Miami), containing a DOC level lower than 2 μ M before the analysis of each sample batch. Certified IAEA-CH6 sucrose (International Atomic Energy Agency, − 10.45 ± 0.03%) dissolved in the low carbon water was used for standardization. For every 10 samples, one standard sample was measured to check the stability of measurements. The method used for blank determination and corrections followed the previous method 35,36 . Our measurement result of δ 13 C-DOC for DSR was -21.5 ± 0.1%, which agree well with previous results reported by Lang  At all the stations, dissolved oxygen (DO), potential temperature, and salinity were measured using a rosette system with a mounted CTD (SBE 911+ ). All DO data were from a CTD-DO sensor that was calibrated by the data from grab samples (measured on board using the Winkler titration method). AOU, determined from the same bottle as DOC concentration, was obtained by subtracting the measured DO concentration from the saturated DO concentration which is dependent on temperature and salinity. The temperature sensor used was a SBE-3/F thermometer (resolution 0.0003 °C ). The probes were calibrated at Sea-Bird Electronics, USA, prior to each cruise. Reported error margins were standard errors. Wherever appropriate, error propagation was applied.