A significant net sink for CO2 in Tokyo Bay

Most estuaries and inland waters are significant source for atmospheric CO2 because of input of terrestrial inorganic carbon and mineralization of terrestrially supplied organic carbon. In contrast to most coastal waters, some estuaries with small freshwater discharge are weak source or sometimes sink for CO2. Extensive surveys of pCO2 in Tokyo Bay showed that the overall bay acts as a strong net sink for atmospheric CO2. Although small area was a consistent source for CO2, active photosynthesis driven by nutrient loading from the land overwhelmed the CO2 budget in the bay. Here we show a comprehensive scheme with a border where air-sea CO2 flux was ±0 between nearshore waters emitting CO2 and offshore waters absorbing CO2. The border in Tokyo Bay was extremely shifted toward the land-side. The shift is characteristic of highly urbanized coastal waters with an extensive sewage treatment system in the catchment area. Because highly urbanized coastal areas worldwide are expected to quadruple by 2050, coastal waters such as Tokyo Bay are expected to increase as well. Through extrapolation of Tokyo Bay data, CO2 emission from global estuaries would be expected to decrease roughly from the current 0.074 PgC year−1 to 0.014 PgC year−1 in 2050.

indicated that a total of 16,345 points were under-saturated with respect to the atmospheric equilibrium; oversaturation was found only in 4,731 data points.
The air-sea CO 2 fluxes were calculated from all of the pCO 2 data (see Methods). The results obtained from different times were binned into a 500 m × 500 m horizontal resolution grid and then averaged together (Fig. 2a). The northwestern head of the bay was the only area with consistent positive CO 2 flux to the atmosphere, whereas the central bay and bay mouth were a sink for atmospheric CO 2 . The location where CO 2 flux was ± 0 roughly coincided with the location where the average annual surface salinity was 25 (Fig. 2b). The areas where salinity was above and below 25 were estimated according to methodology described in Ninomiya et al. 8 . The average CO 2 flux (positive values indicate efflux into the air) for areas where salinity was less than 25 (81 km 2 ) was 15.2 mmolC m −2 day −1 . In contrast, the average CO 2 flux for areas where salinity was greater than 25 (1239 km 2 ) was -10.6 mmolC m −2 day −1 . The area weighted annual CO 2 flux in Tokyo Bay had a rate of -8.8 mmolC m −2 day −1 ; the bay as a whole was a strong net sink for atmospheric CO 2 . The annual CO 2 flux in Tokyo Bay was calculated to be -5.2 × 10 10 gC year −1 .
The seasonal variations of pCO 2 and related parameters at three representative stations in the bay are presented in Fig. 3. The parameters include the observed pCO 2 , pCO 2 normalized to average temperature (19 °C), temperature, salinity and chlorophyll a (Chl a) concentration. In the northwestern head of the bay (St. TPE), where salinity was lowest among the three stations, pCO 2 values generally exceeded the atmospheric equilibrium. There was no distinct pattern observed for the seasonal variation of pCO 2 (93-1920 μ atm), and the variation of pCO 2 was not correlated with Chl a (R 2 = 0.03, P > 0.1, n = 39). At the central bay (St.F6) and the bay mouth (St.06), low values of pCO 2 were observed during spring and summer (70-336 μ atm), whereas pCO 2 values were close to atmospheric equilibrium values during autumn and winter (264-449 μ atm). In terms of the seasonal   variation of pCO 2 in the central bay and bay mouth, a negative correlation was found between pCO 2 and Chl a (R 2 = 0.64, P < 0.001, n = 77). The correlation suggests that active photosynthesis reduced pCO 2 during spring and summer, hence suggesting that pCO 2 in surface waters is mainly controlled by biological activity. The stratified water column prevents the CO 2 supply from the deeper waters with high CO 2 levels. In autumn and winter, the low photosynthetic rate along with the well-mixed water column resulted in pCO 2 values close to the atmospheric equilibrium. Temperature did not significantly contribute to the seasonal variation of pCO 2 . The pCO 2 values normalized to annual average temperature (19 °C), which accounted for the temperature dependence of equilibrium constants and the solubility coefficient 9 , did not affect the seasonal variation of pCO 2 .
In June 2010, pCO 2 decreased to a low of 10 μ atm, and Chl a concentration was > 300 μ g L −1 , which we found to be the lowest reported value in a marine environment. Based on the low observed CO 2 concentrations, the exhaustion of gaseous CO 2 in seawater resulted in the limitation of the CO 2 supply to algal cells, which may limit the cells growth. In such CO 2 limiting conditions, phytoplankton would need to take up bicarbonate using the proton pump mechanism and the carbonic anhydrase 10 .
The CO 2 absorption of 5.2 × 10 10 gC year −1 in Tokyo Bay was in accordance with the overall carbon budget of the bay. A mass balance model estimated the total organic carbon (TOC) influx from the rivers to Tokyo Bay to be 8.1 × 10 10 gC year −1 and the TOC efflux from the bay to the open ocean to be 9.4 × 10 10 gC year −1 11 . On the basis of actual observations, the amount of organic carbon burial was estimated to be 4.2 × 10 10 gC year −1 12 . A different box model estimated the dissolved inorganic carbon (DIC) influx to Tokyo Bay to be 11.2 × 10 10 gC year −1 and the DIC efflux from the bay to be 13.4 × 10 10 gC year −1 13 . By combining these budget estimates, an additional carbon input of 7.7 × 10 10 gC year −1 will be required, a value roughly equal to our estimate of net CO 2 uptake. In addition, the above analysis also suggests that the eventual sink of the fixed carbon from CO 2 absorption in Tokyo Bay would be both from its burial in bay sediments and export to outer oceanic areas.
Because our observations were conducted on monthly basis, our estimates of CO 2 flux may have failed to report on events on a daily and/or weekly time-scale. We acknowledge that our results regarding the net CO 2 uptake in the bay might be compromised if these short-term temporal events contributed to significant CO 2 emission or outgassing from surface waters. We postulated two types of outgassing events in Tokyo Bay and found that such outgassing events should be insignificant to our carbon budget. First, we postulated a sudden vertical mixing event in early autumn when stratification was weakened. The observed pCO 2 values of the bottom waters during the stratification season from June to September were 400-940 μ atm 7 , with a typical water-column average of approximately 600 μ atm. Even if the outgassing from seawater at this level of CO 2 occurred in the entire area of the bay for 30 days, the amount of CO 2 efflux under typical wind speed would be 0.6 × 10 10 gC, which would account for 12% of the air-sea CO 2 exchange in the bay (5.2 × 10 10 gC). Second, we postulated coastal upwelling events, which are generally observed in the northeastern part of the bay in late summer. These upwelling events cause conspicuous milky turquoise waters 14  This study clearly demonstrated that Tokyo Bay as a whole is a net sink for atmospheric CO 2 ; however, this finding may contradict those of many studies on coastal waters. Most inland waters and estuaries have been reported to be significant sources of CO 2 to the atmosphere due to respiration of terrestrial organic carbon 2,3,16 and terrestrial input of freshwater CO 2 17 . In contrast, continental shelf areas, which are laid offshore of coastal areas, have generally been reported to be sinks for atmospheric CO 2 due to nutrient input through coastal waters and from pelagic deep waters 2,3,18,19 . Oceanic basins that are further offshore are either weak sinks or weak sources of atmospheric CO 2 , depending on the biogeochemical settings of the basin 20 . Although it is confined to a very small inner part of the bay, Tokyo Bay certainly has an area of CO 2 emission, and the CO 2 emission mechanism in this area is common to that of other coastal areas. The net CO 2 absorption in the main body of the bay is driven either by a mechanism similar to that of continental shelf regions or by biological CO 2 fixation with a terrestrial supply of nutrients. On the basis of these considerations, we propose a generalized scheme for the CO 2 budget in a continuing water system composed of nearshore water emitting CO 2 , an outer water absorbing CO 2 , and pelagic water with a neutral CO 2 budget (Fig. 4).
In this scheme, there would be a border where the air-sea CO 2 flux is ± 0 between the nearshore waters emitting CO 2 and an outer waters absorbing CO 2 . On the land-side of this border, CO 2 emission due to biological degradation of terrestrial organic matter would exceed CO 2 uptake due to photosynthesis. The CO 2 emission would decrease toward the offshore side, and photosynthetic CO 2 uptake would exceed emissions on the offshore side of this border. The location of this border may shift either offshore or inshore. In fact, several studies have observed that some waters in estuaries with small freshwater discharge are weak sources or sometimes weak sinks of CO 2 [21][22][23] .
In addition, some continental shelves with large freshwater discharge have been reported to be sources of CO 2 to the atmosphere 19 . We interpret that these rather atypical observations are associated with the shift of the aforementioned border. The shift is likely caused by the different terrestrial organic carbon load accompanied by freshwater discharge 24 . Other factors that may affect the border shift and intensity of CO 2 flux include hydrographic and geomorphological characteristics such as a stratified estuarine system 22 , a microtidal estuarine system 25 , an open/enclosed nature of the coast 23,26 , and a submerged aquatic vegetation in shallow coastal waters 27 .
In the case of Tokyo Bay, the border is extremely shifted inshore. The shift reflects the relatively low organic carbon supply from land and the active organic matter production driven by the massive nutrient supply from land. The seasonal stratification and semi-enclosed nature of the embayment should further facilitate the net uptake of CO 2 in Tokyo Bay. The low organic matter supply and large nutrient supply from land may seem Scientific RepoRts | 7:44355 | DOI: 10.1038/srep44355 contradictory. We believe that this imbalance between organic matter and nutrient supply is largely derived from the secondary sewage treatment in the catchment area of Tokyo Bay 4 . Currently, secondary-treated effluent flowing into Tokyo Bay accounts for 50% of the total freshwater discharge 28 . Sewage treatment plants along Tokyo Bay remove organic carbon from freshwater at a rate of 7.7 × 10 10 gC year −1 28 . This value is comparable to the uptake of CO 2 observed in this study (5.2 × 10 10 gC year −1 ). As the quantity of dissolved organic carbon (DOC) flowing into the bay has decreased by 60% in the last 40 years, the quality of DOC has become more recalcitrant due to improved sewage treatment 29 . The decrease in turbidity due to the decrease in organic carbon inflow has also enhanced phytoplankton activity from the improvement of light availability 30 . The effect of sewage treatment has also been reported in other estuary systems; Amann et al. 31 have recently reported that pCO 2 decreased from 7000 to 2500 μ atm at the oxygen minimum zone in 1986 and 2007 due to the installation of sewage treatment plants in the Elbe estuary.
Urbanized coastal waters that have a relatively large coverage of secondary sewage treatment tend to act as a strong net sink for atmospheric CO 2 . The progress made in the urbanization of coastal areas with improved sewage treatment is common in many parts of the world 32 . This development will have an impact on the future budget of marine CO 2 . Moreover, approximately 40% of the world's population is currently settled in coastal zones, and developed urban areas with completed sewer systems are expected to expand rapidly 32 . According to a UNEP report 32 , highly developed coastal areas occupied only 15% of the world coastal zones in 2002, but this figure is expected to rise to 60% in 2050. Therefore, the CO 2 budget characteristic of Tokyo Bay is expected to be observed in more marine coastal areas in the future. Although it may be difficult to predict future trends, we assume that highly urbanized coastal waters currently occupy 15% of the world coastal zones and are expected to occupy 60% by 2050. When the CO 2 fluxes in highly urbanized coastal waters and ordinal coastal waters are the same values as that of Tokyo Bay (-3.2 molC m −2 year −1 ) and previously published data 3 (7.7 molC m −2 year −1 ), respectively, the CO 2 emissions from global coastal waters will be estimated to be approximately 0.074 PgC year −1 at present and 0.014 PgC year −1 in 2050. The emission of CO 2 to the atmosphere from estuaries is calculated to be 21% and 85% lower than previous estimates. As a result, the CO 2 budget of the global coastal waters would be expected to be a sink rather than a source. , which is estimated from the "CO 2 derived from respiration and terrestrial sources (dotted brown line)" plus the "CO 2 consumption by photosynthesis (dotted green line)" The lower panel shows net CO 2 flux in Tokyo Bay (dotted red line). The dotted gray line denotes the atmospheric CO 2 equilibrium (air-sea CO 2 flux is ± 0). The orange arrow indicates the border shift where air-sea CO 2 flux is ± 0 between the nearshore waters emitting CO 2 and outer waters absorbing CO 2 . In the case of Tokyo Bay, the border is extremely shifted inshore because of the low organic carbon supply and expansive coverage of secondary sewage treatment plants (STPs) in the catchment area.

Methods
The ship routes of all cruises were presented in Fig. 3 (a). Each point in this figure represents a pCO 2 value measured at a one-minute interval. Observations of pCO 2 , salinity, and temperature were conducted during 40 cruises of the R/V Seiyo-maru and 9 cruises of the R/V Hiyodori. Measurements of pCO 2 , salinity, and temperature were taken with a sampling frequency of one minute. Surface seawater was pumped up from the ship's bottom at ca. 2-m depth. Our pCO 2 measuring system consisted of a NDIR analyzer (LI-820, Li-Cor) and a membrane equilibrator. The membrane equilibrator was composed of multi-layered composite hollow-fiber membrane modules 33 (MHF module, Mitsubisi Rayon Co., Ltd.). The equilibrator was made of 6 MHF modules to create more surface area and a more rapid response. The response time and the standard error of this system were approximately 100 seconds, and smaller than 0.4 μ atm, respectively. Atmospheric pCO 2 (pCO 2 air ) was measured every three hours. In situ surface water salinity and temperature were measured using a thermosalinograph (Tsurumi Seiki Co., Ltd. at Seiyo-maru and YSI 6920 at Hiyodori). The net flux of CO 2 across the air-sea interface was calculated as the product of the solubility of CO 2 in seawater 34 , the gas transfer piston velocity of CO 2 , and the difference between pCO 2 sea and pCO 2 air . The gas transfer piston velocity was calculated using the equation given by Wanninkhof 35 . Wind speed data were obtained from the Japan Coast Guard (http://www6.kaiho.mlit.go.jp/ tokyowan/) from stations (Tsurugisaki, Kannonzaki, Honmoku, and Tokyo 13 gouchi) close to the observation areas. The samples for chlorophyll a (Chl a) measurement were collected at each stations using bucket and filtered through precombusted (450 °C, 3 h) GF/F filters. After filtration, chlorophyllous pigments were extracted using N, N-dimethylformamide, and the concentrations of Chl a were determined by the fluorometric method 36 (fluorometer used TD-700, Turner Desings).