Variations in seawater-dissolved calcium ion (Ca2+) are small but substantial, which provides information essential to establish the oceanic calcium carbonate (CaCO3) dissolution flux. In this study, high-precision data of Ca2+ were collected in the South China Sea (SCS), the largest marginal sea of the western North Pacific, and its precursor, the West Philippine Sea (WPS), on the basis of two cruises conducted in 2007 and in 2008. An excess Ca2+ of 13 ± 5 μmol kg−1 was observed in the SCS subsurface water at 200–800 m relative to the WPS, and we suggest that such an excess is attributed to in situ CaCO3 dissolution at a rate of ∼0.5 mmol m−2 d−1 in the SCS shallow subsurface water. Through subsurface water outflow, this shallow-depth CaCO3 dissolution may provide a Ca2+ export flux of (0.8 ± 0.3) × 1012 mol yr−1 from the SCS to the WPS, establishing it as an important Ca2+ source from the SCS to the interior Pacific Ocean. This study indicates, for the first time, that along with the benthic processes, CaCO3 dissolution in waters at shallow depth in marginal seas could also contribute to Ca2+ and total alkalinity accumulations in the upper layer of the open ocean, which would ultimately enhance the buffer capacity of the world ocean in the context of rising anthropogenic CO2.
 The formation and dissolution of calcium carbonate (CaCO3) is an important component of the oceanic carbon cycle. Recent estimates of both CaCO3 production and export at a global ocean scale were roughly at the same order, at 0.4–1.8 Gt C yr−1, suggesting relatively high carbon export efficiency in the form of inorganic carbon [Berelson et al., 2007]. Together with organic carbon metabolism (photosynthesis and respiration), the CaCO3 cycle modulates the oceanic dissolved inorganic carbon system and eventually the CO2 buffer capacity of the world ocean. Relevant issues have also emerged concerning the fate of CaCO3 and calcifiers in the context of ocean acidification that has occurred since the last century [Intergovernmental Panel on Climate Change, 2007]. Several studies have indicated that biogenic calcification will decline and CaCO3 dissolution will increase under rising atmospheric CO2 and lowered seawater pH [Kleypas et al., 1999; Riebesell et al., 2000; Zondervan et al., 2001; Andersson et al., 2005; Delille et al., 2005; Orr et al., 2005]. As a consequence, the oceanic CaCO3 export will decrease, which might further weaken its ballast effect for vertical transfer of organic carbon to the deep ocean [Armstrong et al., 2002; Barker et al., 2003].
 In a simplified way, the process of CaCO3 formation and dissolution, or inorganic carbon metabolism, can be illustrated as
When 1 mole of CaCO3 is produced or dissolved, the corresponding variations of dissolved calcium ions (Ca2+) and total alkalinity (TAlk) are 1 and 2 moles, respectively. Therefore, both Ca2+ and TAlk are used to study CaCO3 production and/or dissolution in the ocean. Ca2+, one of the eleven major ions in seawater, has been recognized for a long time to be nonconservative in the ocean [Millero, 2006; Pilson, 1998]. The observed low ratios of calcium to chlorinity or salinity (Ca2+/Cl or Ca2+/Sal ratio) in the surface water remain an outstanding question. Prior to early 1970s, studies indicated that the surface low ratios were due to calcareous shell construction via biotic Ca2+ uptake, and elevated ratios in deep waters were caused by increasing solubility and carbonate dissolution owing to higher pressure, lower temperature and increasing CO2 loading [Dittmar, 1884; Culkin and Cox, 1966; Riley and Tongudai, 1967; Tsunogai et al., 1968; Sagi, 1969; Tsunogai et al., 1971, 1973]. Latter on, the proton flux associated with nutrients release during organic matter decomposition was appreciated to contribute to the titration of seawater TAlk [Brewer and Goldman, 1976], which implies that changes in the observed TAlk within a water mass will underestimate the amount of CaCO3 dissolved; thus, the “potential alkalinity” (PTA) was introduced to assess the Ca2+-TAlk relationship via proton correction, which can be expressed as ΔPTA = ΔTAlk + ΔNO3 + ΔPO4 (Δ denotes the difference between two water masses) [Brewer et al., 1975].
 Unlike TAlk, which is affected by both organic and inorganic carbon production/consumption, the variations of Ca2+ in the ocean interior are almost solely controlled by CaCO3 formation and dissolution on day-to-decade time scales. The advantage of using Ca2+ to look into the marine CaCO3 cycle is thus obvious. However, reports of Ca2+ distribution are less popular since the change in seawater Ca2+ is relatively small, typically representing only ∼1% of its ambient concentration (∼10,280 μmol kg−1 at a salinity of 35 [Pilson, 1998]). Alternatively, abundant studies tend to provide high-quality TAlk data, which forms a strong knowledge base of present estimates of oceanic CaCO3 production, export, and dissolution (see the review by Berelson et al. ).
 Lysocline is defined as the depth in the water column where a critical undersaturation state with respect to aragonite or calcite results in a distinct increase in the CaCO3 dissolution rate [Morse, 1974]. The majority of CaCO3 dissolution is thus believed to occur only below the depth of the lysocline, typically below 1000 m. However, Milliman et al.  proposed the concept of supralysocline dissolution. This concept of shallow-depth CaCO3 dissolution above the lysocline contradicts the conventional understanding and provides a new vision on the marine CaCO3 cycle. At A Long-Term Oligotrophic Habitat Assessment (ALOHA) Station in the North Pacific, Milliman et al.  revealed that a significant fraction as high as 60–80 μmol kg−1 of Ca2+ was in excess above the aragonite lysocline, defined as the difference between measured Ca2+ (Ca2+_meas) and Ca2+ calculated from a conservative Ca2+/Sal ratio. Considering the preformed value which is the Ca2+ content in the preformed water from the Southern Ocean to the North Pacific, however, Chen  pointed out that this excess was reduced to only 10 μmol kg−1, which was not much larger than the analytical uncertainty level. Thus, when using Ca2+ to evaluate in situ CaCO3 dissolution for a given water mass, the influence from its preformed water must be removed and the choice of reasonable preformed values becomes tricky.
 On the other hand, Berelson et al.  summarized the CaCO3 dissolution flux estimates via examining TAlk patterns of major oceans. The total dissolution flux in upper 200–1500 m of the world ocean was estimated to be ∼1.0 Gt CaCO3 yr−1, which is ∼2.5 times higher than that below 2000 m. As a consequence, Berelson et al.  concluded that dissolution of settling carbonate particles should occur to a greater extent in the upper water column where waters are mostly oversaturated with CaCO3. However, Friis et al. [2006, 2007] suggested possible overestimation of shallow-depth CaCO3 dissolution, also using the TAlk evaluation. The TAlk approach usually employs a TA* term, which denotes the excess TAlk and is obtained by subtracting the preformed TAlk value from the observed one and calibrating changes induced by the protons flux [see Feely et al., 2002, equation 5]. And the assumption for using TA* to estimate the CaCO3 dissolution flux is that the only source of TA* in seawater is CaCO3 dissolution. This approach should be applied with further caution, because additional processes such as denitrification that are apparently unrelated to CaCO3 dissolution can also contribute to TA* [Chen, 2002; Wolf-Gladrow et al., 2007]. Studies have also shown that TAlk measurements are subject to interferences from dissolved organic bases, which might add additional complications in the TAlk application [Hernández-Ayon et al., 2007; Kim and Lee, 2009]. Ca2+ does not suffer from such potential issues. The excess Ca2+ (Ca2+_ex) in the ocean interior, if it exists, can only have originated from CaCO3 dissolution, especially in waters at shallow depth being hardly influenced by hydrothermal inputs [de Villiers and Nelson, 1999]. There is conjecture that dissolution or ion change of silicate material in the water might increase Ca2+ in the intermediate water [Tsunogai et al., 1973], however, until now there is no report of such processes.
 While we also recognize the potential of shallow-depth CaCO3 dissolution from many trap studies [e.g., Martin et al., 1993; Honjo et al., 1995; Wong et al., 1999], the magnitude and the causes accounting for shallow-depth CaCO3 dissolution warrant further examination, in particular in the context of better predicting the response of oceanic buffer capacity to increasing penetrated anthropogenic CO2. We argue that high-quality Ca2+ data from a proper study site with distinguishable preformed waters are required to better examine this shallow-depth CaCO3 dynamics.
 The South China Sea (SCS) holds an only source of deep water from the adjacent West Philippine Sea (WPS) and the water residence time in the SCS is also well studied [Chen et al., 2001; Qu et al., 2006]. This unique circulation pattern provides us a unique opportunity to explore the potentially different Ca2+ behavior between the two seas with the WPS as the preformed water mass, and subsequently to estimate the CaCO3 dissolution rate. Note that previous studies have suggested that marginal seas can significantly contribute Ca2+ and TAlk exports to the pelagic ocean [Milliman, 1993; Milliman and Droxler, 1996]. Our study in the SCS, the largest marginal sea in the western North Pacific, aims to further explore such processes associated with marginal sea–open ocean interactions.
2. Material and Methods
2.1. The Study Area
 With a total area of 3.5 × 106 km2, the SCS is the world's largest tropical-subtropical marginal sea. The East Asian Monsoon prevails in the SCS, resulting in a seasonal reversal in surface circulation with a cyclonic gyre in the winter and an anticyclonic gyre during the summer (Figure 1). Due to input from some of the major world rivers (e.g., the Mekong River and the Pearl River; see Figure 1) and influence by intensive upwelling, internal waves, and strong mixing [Chao et al., 1996; Liu et al., 2006], the SCS receives a large supply of nutrients to the euphotic zone, leading to a relatively higher biological productivity compared to the adjacent western North Pacific [Liu et al., 2002; Chen and Chen, 2006]. The SCS and the western North Pacific exchange their water via the 2200 m deep Luzon Strait, through which, the Kuroshio Branch Water intrudes from the WPS into the SCS basin area (Figure 1). A “middepth front” between 350 and 1350 m, where the abrupt transition occurs between SCS and WPS proper waters, exists in the Luzon Strait near 122°E. East of this middepth front the water mass belongs to the WPS whereas the SCS water mass mostly lies to the west [Chen and Huang, 1996]. Previous studies [Chao et al., 1996; Li and Qu, 2006; Qu et al., 2006; Tian et al., 2006] have indicated a “sandwich-like” structure of water exchanges across the Luzon Strait, with a net inflow from the Pacific in the surface and deep layers and a net outflow from the SCS in the intermediate layer. In terms of mass balance, the rapid replenishment of the SCS deep water from the WPS is maintained by fast ventilation with the shallower intermediate water, as well as a persistent net outflow at an intermediate depth [Chao et al., 1996; Chen et al., 2006; Li and Qu, 2006]. This unique flow pattern therefore provides an opportunity to study both the material exchange across different depths between the two seas and the independent biogeochemical processes in the SCS relative to the WPS [Dai et al., 2009].
2.2. Sampling and Analyses
 Seawater samples were collected in September 2007 at the South East Asian Time-series Study (SEATS) station established in the SCS basin on board R/V Dongfanghong II and in September 2008 at stations BY01, K404, K503, and K506 on board R/V Shiyan III. K503 and K506 were located right on the middepth front, and other stations were located to its west at different distances (Figure 1). Water column samples were collected with Niskin bottles attached on a Rosette sampler. Samples for Ca2+ analysis were stored in 60 mL acid-cleaned polyethylene bottles with Parafilm wrapped around the cap. Samples for TAlk and dissolved inorganic carbon (DIC) analysis were stored in 100 mL polyethylene bottles and 40 mL borosilicate glass Environmental Protection Agency (EPA) vials, respectively. Both TAlk and DIC samples were poisoned with HgCl2-saturated solution upon sample collection, with 100 μL for TAlk and 50 μL for DIC.
 In the laboratory, Ca2+ was determined using the classic ethylene glycol tetraacetic acid (EGTA) titration modified from Lebel and Poisson . About 4 g of seawater and 4 g of HgCl2 solution (∼1 mmol L−1) were accurately weighed out; then about 4 g of a concentrated EGTA solution (∼10 mmol L−1, also by weighing) was added to complex Hg2+ completely and nearly 95% of Ca2+. After adding 4 mL of borate buffer, the remaining Ca2+ was titrated by a diluted solution of EGTA (∼2 mmol L−1) up to the end point potential, which was given by the Methrom 809 TITRANDO potentiometer with an amalgamated silver combined electrode (Methrom Ag Titrode). Instead of using the autogenerated value given by the Methrom TiAMO 1.1 software, the volume of the diluted solution of EGTA necessary for the remaining ∼5% of Ca2+ was obtained by manually fitting the first derivative of the titration curve, which proved to be optimal for better measurement precision (10,006 ± 95 μmol kg−1 versus 10,190 ± 6 μmol kg−1, 1σ, n = 24; the latter method yielded a precision of ∼0.06%). EGTA solutions were standardized daily against International Association for the Physical Sciences of the Oceans (IAPSO) standard seawater (Batch P147, Salinity = 34.993). Standard seawater of the same batch was calibrated against a standard Ca2+ solution in a synthetic seawater matrix. The Ca2+/Sal ratio of Batch P147 was determined to be 292.3, which is higher than that of Batch P61 (291.3 [Shiller and Gieskes, 1980]) and Batch P79 (290.5 [Olson and Chen, 1982]), but lower than Batch P67 (293.0 [Kanamori and Ikegami, 1980]). The reasons causing the variations of the Ca2+/Sal ratio among different batches of IAPSO standard seawater are unclear but could be related to the different analytical methods of Ca2+ analysis used by the different investigators. Alternatively, there might be small variations in Ca2+ in the IAPSO standard seawater because they are prepared on the basis of Atlantic Ocean surface seawater. Note that our replicated analysis did not show measurable difference between filtered and unfiltered samples within the limit of the precision. Data reported here were from unfiltered samples.
 DIC was determined by acidification of 0.5 mL of a water sample and the subsequent quantification of CO2 with a nondispersive IR detector (Li-Cor 7000). This method has a precision of 0.1% [Cai et al., 2004]. TAlk was determined by Gran titration on a 25 mL sample with a Kloehn digital syringe pump, with a precision of 2 μmol kg−1. Both DIC and TAlk were calibrated against the certified reference material provided by Dr. A. Dickson of the Scripps Institution of Oceanography. Depth profiles of temperature and salinity were determined shipboard with the SBE9/11 Conductivity-Temperature-Depth (CTD) recorder (Sea-Bird Co.).
2.3. Ratio of CaCO3 Dissolution and Organic Carbon Decomposition
 During the settling of surface produced particles, both CaCO3 dissolution and organic carbon decomposition contribute to the TAlk and DIC addition in deeper waters, which are fundamental to the ocean's capacity to absorb CO2. It is, therefore, important to determine the ratio of the DIC addition between CaCO3 dissolution and organic carbon decomposition (IC/OC ratio).
Equations (2)–(8) have been used previously to calculate the IC/OC ratio in the Pacific Ocean and the WPS and SCS [Chen et al., 1982, 2006]. In this study, TAlk, DIC and potential temperature (PT) data from station SEATS were applied to calculate the IC/OC ratio in deep waters of the SCS.
For deep and bottom waters:
For salinity minimum waters and waters above:
The superscript “meas” and the degree symbol denote measured and preformed values, respectively, and their difference is the addition of DIC, TAlk, or Ca2+ and is represented as Δ. N denotes the value normalized to a salinity of 35. Sal and PT are salinity and potential temperature.
Equation (2) is deduced on the basis of the stoichiometric relationship during the inorganic and organic carbon metabolism. DIC and Ca2+ increase by 1 mol, and TAlk increases by 2 moles per mole CaCO3 dissolution (equation (1)). On the other hand, during 1 unit organic carbon decomposition the change in TAlk with increase in DIC is minor (−17 versus 106; (CH2O)106(NH3)16H3PO4 + 138O2 ↔ 106CO2 + 122H2O + 16HNO3 + H3PO4). Assuming that x μmoles of CaCO3 dissolve and y μmoles of organic carbon decompose in 1 kg of seawater, the additions of TAlk and DIC can be written as ΔTAlk = 2x − 17y and ΔDIC = x + 106y. After solving for x and y with ΔTAlk and ΔDIC, equation (2) can be yielded from IC/OC = x/106y [Chen et al., 1982]. Note that the value of 45 (μmol kg−1) in equation (2) is the estimated average amount of anthropogenic CO2 input in the SCS and the WPS by Chen et al.  based on field data collected around 1997 with an uncertainty of ±30%. After 10 years till our sampling period, the anthropogenic CO2 content should increase but only to a small extent when considering that the surface NDIC was predicted to ∼60 μmol kg−1 higher in 2050 than in 1997 [Chen et al., 2006]. Therefore, we still employed the value of 45 in this study, and this anthropogenic CO2 contribution should be subtracted from DIC° to better quantify the portion originating from the CaCO3 dissolution and organic carbon decomposition [Chen et al., 2006].
 The preformed value is fixed when a given water mass was last in contact with the atmosphere and can be calculated from PT because the two properties are well correlated in a given ocean basin [Chen and Pytkowicz, 1979]. Equations (5)–(8) were obtained by fitting surface DIC or TAlk to PT in the North Pacific by Chen et al.  on the basis of revised GEOSECS data (T. Takahashi et al., Carbonate chemistry of the Atlantic, Pacific, and Indian Oceans: The results of the GEOSECS expeditions, 1972–1978, unpublished document, 1980). The numbers in parentheses give 1 standard deviation of the least squares fit.
2.4. The Addition of Ca2+ From CaCO3 Dissolution
 On the basis of previous studies, the addition of Ca2+ (ΔCa2+) from CaCO3 dissolution in deep waters can be estimated on the basis of the combination of the addition of TAlk (ΔTAlk) and the addition of DIC (ΔDIC [Chen et al., 1982]):
or directly obtained from Ca2+_meas subtracting the preformed Ca2+ value (NCa2+° [Chen, 2002]):
Equation (9) is also deduced on the basis of the stoichiometric relationship during the inorganic and organic carbon metabolism. The estimated ΔCa2+ equals x which has been solved during the deduction of equation (2) [Chen et al., 1982]. In this study, we added the value of 45 (μmol kg−1) for calibrating the anthropogenic CO2 input.
 In this study, both estimated ΔCa2+ and measured ΔCa2+ were obtained from the deep water at station SEATS.
 As shown in the Temperature-Salinity (T-S) diagram (Figure 2), all surface waters collected during the present study (PT > 25°C, top 100 m) were mixtures of SCS and WPS proper waters with varied combinations of both. The lower surface salinity in the SCS interior is attributed to the basin-wide net precipitation and large river input (i.e., the Pearl River). At the subsurface, the T-S distribution pattern at station K503 was mostly reflective of WPS proper waters, whereas at stations SEATS and BY01 it showed characteristics of SCS proper waters. The observed patterns for T-S distribution at stations K506 and K404 represented a mixture of SCS and WPS waters with a higher contribution from the SCS. Both the salinity maximum and minimum could still be found at station SEATS, less pronounced though as compared to station K503. These salinity extremes in the SCS and the WPS are influenced by the North Pacific Tropical Water (NPTW) and the North Pacific Intermediate Water (NPIW) sourced from the subpolar region, respectively [Dai et al., 2009, and references therein], whereas the intensive upwelling and vertical mixing in the SCS interior must have reduced those extreme signals [Chen et al., 2006]. The overall SCS water T-S property, therefore, showed a less curved inverse S shape as compared to that of the WPS (Figure 2).
3.2. Vertical Ca2+ Distribution
 The Ca2+ distribution largely followed the salinity (Figure 3). At station SEATS in the inner basin of the SCS, Ca2+ increased considerably from surface to up to 150 m coinciding with the zone of shallow salinity maximum layer at a density level (σθ) of ∼25.5. Below 150 m, Ca2+ concentrations showed a sharp decrease until the intermediate salinity minimum layer at a depth of ∼450 m with σθ ∼ 26.5–27.0. Below 500 m, Ca2+ concentration increased again until it reached an almost constant value in the bottom water (>1500 m, σθ > 27.5). The Ca2+/Sal ratio in the top 100 m water at station SEATS was constant. It increased rapidly from 200 m to up to 800 m and showed a slight increase below 800 m (Figure 3). The average Ca2+/Sal ratios in the subsurface (200–800 m) and deep waters (below 1000 m) were ∼0.3% and ∼0.7% higher than that of the top 100 m water, corresponding to an additional Ca2+ content of ∼31 and ∼72 μmol kg−1, respectively.
 The values of surface salinity and Ca2+ from stations K503 and K506 were significantly higher than those of stations BY01 and K404 (Figure 3), suggesting the dilution by fresh water input at the latter two stations in the SCS interior. Around the 200 m depth (the shallow salinity maximum layer of the WPS at σθ ∼ 24.5), K503 and K506 also displayed much higher salinity and Ca2+ values, indicating the relatively strong influence of the NPTW on these two sites located at the middepth front. However, the observed salinity and Ca2+ at the intermediate salinity minimum layer at K503 (∼500 m, σθ ∼ 26.5) were lower, whereas in the case of K506 these values were much closer to those of BY01 and K404, indicating the possible influence of the intermediate SCS outflow waters on the station K506. Below the depth of ∼1000 m (σθ > 27.3), there was no noticeable difference in the content of salinity and Ca2+ among the four study stations primarily due to the rapid renewal of the SCS deep water.
 In the surface layer (top 100 m), Ca2+/Sal ratios were higher at stations BY01 and K506 than K503 and K404. At the depth between 200 and 600 m, the Ca2+/Sal ratio of K503 was remarkably lower, whereas below 600 m, no apparent difference was observed in Ca2+/Sal ratios. The surface Ca2+/Sal ratios of K404 were more similar to K503 than BY01 located in the west of the front (Figure 3). The reason for this discrepancy is unclear, but possibly due to the fact that K404 is located in the slope area (with a water depth of 1250 m) under dynamic environmental settings. Some local processes such as mixing with other end-members or strong biouptake of Ca2+ might lower the surface Ca2+/Sal ratios. This might be also true in the case of K503 and K506, where the water depth of K503 (1990 m) is much shallower than that of K506 (4650 m).
4.1. Excess Ca2+ in the SCS Subsurface Water and Shallow-Depth CaCO3 Dissolution
 The T-S property patterns (Figure 2) suggest that in the subsurface and intermediate layers, the hydrographic characteristic of the SCS waters (i.e., SEATS) were very different from those of the WPS waters (i.e., K503). A closer look at the PT and Ca2+ profiles at stations SEATS and K503 (Figure 4) revealed that the subsurface water (200–800 m) of the SCS was relatively cold and Ca2+ enriched. The relatively higher Ca2+ in the SCS than the WPS might be a consequence of the in situ CaCO3 dissolution and/or upwelled Ca2+-rich deeper waters in the SCS interior.
 To distinguish the two potential processes leading to the higher Ca2+ in the SCS, we first constructed the relationship between Ca2+ and PT for the K503 waters between 200 and 1000 m as PT is conservative during the transport of a given water mass (Figure 5a). We then substituted the measured PT of the SEATS subsurface water into the above relationship to obtain their preformed Ca2+ value (denoted as Ca2+_pre to distinguish itself from the term NCa2+° in equations (10) and (11)) produced via physical processes such as upwelling and vertical mixing in the SCS interior. As a consequence, Ca2+_ex was the difference between Ca2+_meas and Ca2+_pre, which signifies the amount of Ca2+ enriched through CaCO3 dissolution (Figure 5b). As shown, Ca2+ remained higher in the subsurface layer (corresponding to the PT range from ∼5 to 20°C) at SEATS than at K503 (Figure 5a). The average concentration of Ca2+_ex in the subsurface waters at 200–800 m at SEATS was estimated to be 13 ± 5 μmol kg−1 (Figure 5b). It is reasonable to argue that it is impossible for processes such as hydrothermal input to affect the Ca2+ behavior within this depth range of the SCS. The significant amount of Ca2+_ex found in the SCS thus should be reflective of Ca2+ additions from the dissolution of sinking carbonate particles produced in the surface layer of the SCS. The total additional Ca2+ content in the SCS subsurface water relative to the local top 100 m water was ∼31 μmol kg−1. In this context, almost 40% (approximate 13/31) of the increased Ca2+ was due to in situ CaCO3 dissolution with the remainder 60% (approximate (31 − 13)/31) coming from the upwelled Ca2+-rich deeper waters.
 With the residence time of ∼50 years for the SCS intermediate water [Chen et al., 2001], we obtained a CaCO3 dissolution rate of ∼0.5 mmol m−2 d−1 in the waters at 200–800 m, which is similar to the range of dissolution rates in the Pacific shallow depth [see Feely et al., 2002, Table 1] and is close to the average CaCO3 lost between two traps deployed at 124 and 620 m, respectively, in the SCS basin (station KK3: 18.3°N, 115.7°E [Cai, 2007]). Direct reports on the CaCO3 export flux of the SCS basin are scarce. However, the global CaCO3 export flux estimated from models is 0.3–1.1 mmol m−2 d−1 [Berelson et al., 2007]. Thus, it is reasonable to contend that more than half of biogenic CaCO3 particles escaped from the euphotic zone would dissolve in the SCS upper 1000 m water columns, which coincides with the anthropogenic CO2 penetration depth in the SCS [Chen et al., 2006]. CaCO3 dissolution above this depth would convert this CO2 to HCO3− (equation (1)), thereby further stimulating the onsite oceanic uptake of CO2.
 Due to the fact that waters between 200 and 800 m in the SCS are mostly oversaturated with respect to both aragonite and calcite [Chen et al., 2006], it is impossible that the thermodynamically predicted CaCO3 dissolution at depths below the lysocline leads to the Ca2+_ex. Together with the significant DIC increase in the similar water depth range (DICbio between 100 and 600 m [Chou et al., 2007; Sheu et al., 2009]), the Ca2+_ex observed in the SCS subsurface water may support the second “possible mechanism” envisioned by Milliman et al. , i.e., supralysocline dissolution due to the microbial oxidation of organic matter that produces a microenvironment conducive to carbonate dissolution. Wollast and Chou [1998, 2001] also demonstrated that shallow-depth CaCO3 dissolution could happen in oversaturated waters in the Gulf of Biscay, and proposed that the “microenvironment” might be the interior of fecal pellets or the surface of biogenic aragonite or calcite. Through examining Mg/Ca ratio in the shell of planktonic foraminifers at four water depths, Huang et al.  proposed partial dissolution at depths well above the calcite lysocline in the SCS. In this study, by straightly analyzing seawater Ca2+, we demonstrated that shallow-depth CaCO3 dissolution does occur in the SCS.
 Moreover, other investigators have suggested CaCO3 dissolution in the euphotic zone. Moore et al. [2002, 2004] proposed models showing that 30%–50% of the CaCO3 produced in surface waters would dissolve before escaping the euphotic zone. By estimating the weight of coccoliths from two sediment traps deployed at the same tropical Atlantic site, but separated by 2000 m water depth, Beaufort et al.  found no significant coccolith dissolution during settling between 250 and 2500 m, but proposed that most of the dissolution occurs in the euphotic zone and possibly in the guts of grazers. In this study, although normalized Ca2+ concentrations were higher in the top 100 m waters at SEATS than at K503 (Figure 4), whether CaCO3 was dissolved in the euphotic zone of the SCS is not certain because the signature of normalized Ca2+ (against a salinity of 35) may well be compounded by the riverine Ca2+ input.
4.2. Ca2+ Export Out of the SCS Through Shallow-Depth CaCO3 Dissolution
 With an annual outflow of 1.9 ± 0.4 Sv for the SCS subsurface waters [Chen et al., 2001; Chou et al., 2007; Sheu et al., 2009] and the average Ca2+_ex concentration of 13 ± 5 μmol kg−1, we estimated that the SCS subsurface outflow would export (0.8 ± 0.3) × 1012 mol yr−1 of Ca2+ out of the SCS. This flux is ∼10% of the global annual input of Ca2+ from the marginal sea to the open oceans [Milliman, 1993; Milliman and Droxler, 1996] and represents the Ca2+ and TAlk additions from the water column dissolution in the continental margins. Moreover, along with the northwardly flowing Kuroshio Current (Figure 1), the Ca2+/TAlk-laden SCS subsurface water would travel to high-latitude region in the western North Pacific, even upwell onto the East China Sea (ECS) [Chen, 1996], which is finally conducive to the CO2 uptake of waters around the Kuroshio regime.
 Thus far, it is believed that the ocean margin contribution of TAlk to the open ocean is mainly related to the processes occurring in the ocean margin sediments [Chen, 2002; Berelson et al., 2007]. Both benthic carbonate dissolution and anaerobic processes such as sulfate reduction involving precipitation of iron sulfides [Hammond et al., 1999] and denitrification [Fennel et al., 2008; Thomas et al., 2009] generate TAlk, which might be transported horizontally to contribute to TA* in the upper water column of the open ocean. Our results indicated, for the first time, that along with benthic processes, CaCO3 dissolution in waters at shallow depth in marginal seas might also contribute to the TA* in the upper layer of the open ocean.
4.3. IC/OC Ratio and ΔCa2+ in the SCS Deep Water
 Due to the rapid renewal of the SCS deep water, water properties in the SCS deep layer are almost homogenous to those of the WPS [Chou et al., 2007]; thus, deep waters of the SCS and the WPS have the same precursor as the Pacific Ocean, i.e., waters from the Southern Ocean. At station SEATS (Figure 6), the IC/OC ratio increased rapidly from 500 m to 2000 m, indicating the relatively high decomposition rates of organic matter in the upper water. Below 2000 m, the IC/OC ratio agreed well with the value of the SCS estimated by Chen et al. , showing a constant value at 0.31. The IC/OC ratios of the WPS [Chen et al., 2006] are lower than those of the SCS in the waters below 1000 m (Figure 6), suggesting that the deep SCS water is more corrosive and more CaCO3 has been dissolved. Taking 0.29 as the average IC/OC ratio for the water below 1000 m, we estimated that during the water transport from the Southern Ocean, CaCO3 dissolution contributed approximately 22% of the DIC accumulation in the deep SCS. This percentage is close to that for the deep North Pacific (∼26%, IC/OC = 0.35 [Chen, 1990]), and both of them are higher than in the deep South Pacific (∼12%, IC/OC = 0.135 [Chen et al., 1982]).
 As shown in Figure 7, the value of the estimated ΔCa2+ increased with depth below 500 m at station SEATS, reached a plateau of ∼45 μmol kg−1 at 2000 m and then remained relatively constant below 2000 m. Although the pattern looked similar for the measured ΔCa2+, their values were much higher. Note that the observed additional Ca2+ content below 1000 m at SEATS was ∼72 μmol kg−1, relative to the local top 100 m water. It is unlikely that the ΔCa2+ level would be above this value, suggesting that our estimated ΔCa2+ were reasonable. We contend that the exceptional higher ΔCa2+ measured was mainly due to the systematic deviation of our measured Ca2+ from NCa2+° (equation (10)), which is summarized from Ca2+ determined using other methods [Tsunogai et al., 1973; Chen, 1983]. In fact, the difference of the Ca2+ concentration of the IAPSO standard seawater measured by us and by Olson and Chen , respectively, is 63 μmol kg−1 (10,230 μmol kg−1 for Batch P147 and 10,167 μmol kg−1 for Batch P79). If our measured Ca2+ data were shifted down by 63 μmol kg−1, the difference between the estimated and measured values would be reduced to ∼10–20 μmol kg−1 (Figure 7). Therefore, the average ΔCa2+ value of 51 ± 8 μmol kg−1 below 1000 m at the SEATS station reflected the real CaCO3 flux during the journey of the deep waters from the Southern Ocean to the SCS, which is comparable to the amount of ΔNCa2+ (36 ± 6 μmol kg−1) in the deep North Pacific Ocean relative to the Weddell Sea [Chen, 1990].
4.4. Ca2+ Versus TAlk in the Estimates of CaCO3 Dissolution
 In waters below 100 m where CaCO3 dissolves in the SCS (Figure 8), the slope of the linear regression line of Ca2+ to TAlk was 0.54 ± 0.03, slightly higher than the theoretical value of 0.5. Considering the effect of proton flux, the slope of the linear regression line fitted to Ca2+ and PTA was 0.46 ± 0.03, slightly lower than 0.5. Both of these slopes agreed well with the theoretical value within the limit of uncertainties, suggesting that Ca2+ and TAlk variations from subsurface to deep waters in the SCS are overall controlled by CaCO3 dissolution. We note that dissolution of magnesium (Mg) calcites (in the form of Ca(1-x)MgxCO3) also releases Ca2+ and TAlk [Andersson et al., 2007, equation 5], which could represent another potential mechanism inducing supralysocline dissolution [Millero, 2007]. However, the Ca2+/TAlk change ratio during Mg calcites dissolution should always be <0.5 with variations dependent upon different Mg content therein. Our data with Ca2+ to TAlk ratio of >0.5 thus suggested that the influence of Mg calcites dissolution on Ca2+ should be negligible in the SCS.
 Elsewhere in the North Pacific [Kanamori and Ikegami, 1982; Ikegami and Kanamori, 1983], a comprehensive data set also displayed a good positive linear relationship between Ca2+ and TAlk throughout the whole water column (Ca2+ = (0.69 ± 0.02)TAlk + (8609 ± 39), r = 0.98, n = 86; Ca2+ = (0.54 ± 0.01)PTA + (8963 ± 24), r = 0.98, n = 86. All of the data were normalized to a salinity of 35), which could also be simply interpreted as the CaCO3 formation and dissolution.
 Besides qualitative analysis of Ca2+-TAlk relationship within water column, quantitative evidence can be obtained by evaluating Ca2+_ex and excess PTA (PTAex) independently through point-by-point calculations. For example, on the basis of the preformed value obtained by de Villiers  and data available from other investigations, she demonstrated that in deep waters close to mid-ocean ridges, Ca2+_ex values were greater than the expected values for corresponding changes in PTA during CaCO3 dissolution (i.e., Ca2+_ex > 0.5PTAex), and proposed that the middepth excess Ca2+ was supplied by hydrothermal input.
 In this study, by applying the same calculation method as Ca2+_ex, we estimated PTAex in the subsurface water at SEATS relative to K503. There was no correlation at all between Ca2+_ex and PTAex, and Ca2+_ex was overall higher than half of PTAex (Figure 9b). The average value of PTAex between 200 and 800 m was nearly the same as Ca2+_ex (10 ± 5 μmol kg−1 versus 13 ± 5 μmol kg−1), rather than twofold higher as expected from the good positive linear relationship between Ca2+ and TAlk (Figure 9a). As Ca2+_ex in the SCS subsurface water could only originate from CaCO3 dissolution, the discrepancy of Ca2+_ex > 0.5PTAex in this study might imply the TAlk removal through other processes which have no influence on Ca2+ variations, and the removal content should be too minor to deteriorate the apparent Ca2+-TAlk relationship (Figure 9a). On the other hand, underestimation of CaCO3 dissolution might occur if solely using the TAlk data and our study emphasizes the necessity of distinguishing processes influencing Ca2+ and TAlk variations. In any case, due to its relatively simpler biogeochemical behavior, comprehensive Ca2+ data can give the actual CaCO3 dissolution flux in the ocean potentially advantageous over TAlk, especially since high-precision Ca2+ measurements have become routine with advanced instruments such as new generation of automatic potentiometer now commercially available.
 In the SCS, physical mixing dominated Ca2+ distribution, whereas CaCO3 dissolution contributed to the observable gradient. A Ca2+ excess of 13 ± 5 μmol kg−1 was revealed in the SCS subsurface water relative to the WPS, representing an in situ shallow-depth dissolution rate of ∼0.5 mmol m−2 d−1. This shallow-depth CaCO3 dissolution might account for as much as half of the total water column dissolution and facilitate the uptake of anthropogenic CO2 in the upper ocean of the SCS. Moreover, through net subsurface water outflow, this shallow-depth CaCO3 dissolution in the SCS provides significant Ca2+ and TAlk export to the open Pacific, which would further improve the capacity to absorb CO2 in waters along the Kuroshio travel path. In the deep SCS below 1000 m, the amount of Ca2+ added from CaCO3 flux relative to its source water (the Southern Ocean) was estimated to be 51 ± 8 μmol kg−1, and this inorganic CaCO3 dissolution contributes ∼22% of the DIC increase in the deep water, relative to the decomposition of organic carbon. We envisaged that the significant shallow-depth CaCO3 dissolution in the SCS oversaturated water might be mediated by the microbial oxidation of organic carbon. Further work is needed to understand the coupling and interactions between the oceanic inorganic and organic carbon metabolism.
 This work was supported by the National Basic Research Program of China (973 Program) through grants 2009CB421201 and 2009CB421206 and by the National Science Foundation of China (NSFC) through grants 40821063 and 90711005. We would like to thank Nan Zheng, Xianghui Guo, Lifang Wang, Yan Yang, and Feifei Meng for help with the ancillary data collection and the crew of R/V Dongfanghong II and Shiyan III for their assistance in sample collection. The 2008 cruise was supported by the South China Sea Institute of Oceanology, the Chinese Academy of Sciences. We thank Dongxiao Wang, Huabin Mao, and Jianyu Hu for providing the CTD data. We are also grateful to the inspiring comments from Shuh-Ji Kao, Wei-Jun Cai, and Chen-Tung Arthur Chen. John Hodgkiss and Harish Gupta are thanked for their help with English. Comments from two anonymous reviewers have significantly improved the quality of the paper.