Liu, Zhiyuan, Longjun Zhang, Wei-Jun Cai, Liang Wang, Ming Xue
Transcription
Liu, Zhiyuan, Longjun Zhang, Wei-Jun Cai, Liang Wang, Ming Xue
Limnol. Oceanogr., 59(2), 2014, 413–426 2014, by the Association for the Sciences of Limnology and Oceanography, Inc. doi:10.4319/lo.2014.59.2.0413 E Removal of dissolved inorganic carbon in the Yellow River Estuary Zhiyuan Liu,1 Longjun Zhang,1,* Wei-Jun Cai,2 Liang Wang,1 Ming Xue,1 and Xiangshang Zhang 1,3 1 Key Laboratory of Marine Environment and Ecology, Ministry of Education, College of Environmental Science and Engineering, Ocean University of China, Qingdao, P.R. China 2 School of Marine Science and Policy, University of Delaware, Newark, Delaware 3 Qingdao Maritime Safety Administration, Qingdao, P.R. China Abstract The Yellow River of China runs mainly through an arid and semiarid midlatitude region that has experienced substantial anthropogenic and climatic change. This area includes the carbonate-rich Loess Plateau and carries water of exceptionally high carbonate content. To investigate the processes by which dissolved inorganic carbon (DIC) is biogeochemically modified as the river approaches the sea, a multipronged field investigation was conducted in the Yellow River estuary, 2005–2009. The project included four research cruises (spring and fall), a year of monthly sampling at a lower-river hydrological station (Lijin), and in situ bottle incubations. Our study revealed that 4–11% of the Yellow River DIC was removed from the water column in the estuarine mixing zone and thus was not transported to the sea. DIC removal was greater in the spring and occurred at a higher salinity range than in the fall. As a unique feature of the Yellow River estuary, calcium carbonate (CaCO3) precipitation was nearly as important as net biological production in the DIC removal. Longer freshwater–seawater mixing distances (and times) and higher DIC concentrations in the freshwater end member also promoted net biological production and CaCO3 precipitation, thus encouraging DIC removal. Estuaries are regions of active land–ocean interaction. Globally, a total of ,0.34 Gt (1015 g) of dissolved inorganic carbon (DIC) is annually exported to the ocean by rivers (Ludwig et al. 1998; Mackenzie et al. 2004; Lerman et al. 2007). DIC usually exhibits nonconservative behavior during estuarine mixing (Cai and Wang 1998; Abril et al. 2003; Cai 2003) due to a number of processes: strong internal biogeochemical activity (Wollast 2003); material exchange with surrounding environments (Hans et al. 2011), in particular coastal wetlands (Cai 2011); and complex sediment dynamics during mixing. As a result, riverine carbon fluxes can be over- or underestimated if these physical and biogeochemical processes are not considered. Previous research has shown that many factors regulate the nonconservative behavior of DIC in river-dominated estuaries. In most cases, biological production (resulting in removal of DIC from the water column) or respiration (resulting in addition of DIC) has been identified as the dominant factor. For instance, in several estuaries along the eastern coast of the United States, DIC has been observed to increase nonconservatively with salinity due to bacterial respiration during mixing (Cai and Wang 1998; Cai et al. 1999; Raymond et al. 2000). Occasionally, however, CaCO3 dissolution (addition of DIC) has been reported to dominate (Abril et al. 2003; Ortega et al. 2005, 2008). Within the Mississippi River plume, Cai (2003) observed that DIC losses occurred where salinity was , 30 and noted the cause might be a combination of biological production and CaCO3 precipitation. Guo et al. (2012) and Huang et al. (2012) subsequently concluded that biological production was the main cause for the plume DIC removal and the effect of CaCO3 precipitation was minor. Identifying the biogeochemical mechanisms responsible for DIC production and removal can be further complicated by the existence of time-varying or multiple riverine end members (Officer 1979). These nonbiogeochemcial influences can produce an apparently nonconservative DIC–salinity relationship, and detailed mixing schedules are often required to elucidate the contributing factors. This type of complexity has been seen in the Mississippi River and Atchafalaya River plumes (Guo et al. 2012; Huang et al. 2012), the Pearl River estuary of China (Cai et al. 2004), and the Scheldt River plume (Hellings et al. 2001). Ammonium oxidation, denitrification, and air–water CO2 exchange may also contribute to nonconservative behavior of DIC in estuaries (Cai et al. 2004; Dai et al. 2008; Guo et al. 2008). Therefore, understanding the biogeochemical processes controlling DIC distributions and quantifying net DIC fluxes and variations in estuaries constitutes a rather challenging step in the study of carbon transport from rivers to oceans (Vanderborght et al. 2002). The Yellow River of China constitutes an especially interesting and relevant case study with respect to estuarine DIC dynamics. The river is located in a midlatitude zone with an arid and semiarid climate, and its middle reach flows through the carbonate-rich Loess Plateau. Its drainage basin is characterized by high evaporation (Yang et al. 2004; Chen et al. 2006), a high land-utilization rate, and severe water and soil losses (Wang et al. 2006, 2007). These characteristics and, most important, intense weathering of the carbonate-containing loess, result in Yellow River HCO{ 3 concentrations that are among the highest of all the world’s large rivers (Cai et al. 2008). In addition, the Yellow River has been dramatically affected by human activities, including over 2000 years of agriculture and, recently, several decades of intensive irrigation and damming and a decade of direct human * Corresponding author: [email protected] 413 414 Liu et al. Fig. 1. Yellow River estuary: (A) study area location and (B) survey station locations. regulation of water and sediment discharge (Wang et al. 2007). The water and sediment regulation (WSR) plan, which is the only one regularly executed in the world, was designed to flush out sediment accumulation in reservoirs and riverbed in the lower reach of the river once a year. The WSR activities constitute an extreme human disruption of river-matter transport (Zhang et al. 2013). In particular, estuarine mixing conditions differ greatly between the preand post-WSR periods. Understanding how WSR events change estuarine DIC dynamics represents an important step to understand and predict how human intervention of natural systems affects carbon cycling and fluxes. DIC dynamics in the Yellow River estuary have not been well characterized. On the basis of data from a single survey, Cauwet and Mackenzie (1993) speculated that DIC removal occurred in the low-salinity (S , 5) area of the estuary, probably caused by CaCO3 precipitation. However, no further explanation or justification was provided, nor has subsequent fieldwork been reported to confirm this preliminary observation and attribution. The effect of biological production on DIC removal has been documented in the Mississippi River plume (Cai 2003; Guo et al. 2012; Huang et al. 2012), but the Yellow River is clearly different from the Mississippi in several important aspects. For instance, for the period 1950–2000, Mississippi River discharge increased by 30%, whereas Yellow River discharge decreased by . 50% (Milliman et al. 2008). DIC removal within the Mississippi River plume occurs where salinity is , 30 (Cai 2003), whereas in the Yellow River estuary, removal occurs in the lower-salinity zone (Cauwet and Mackenzie 1993). Thorough study of DIC removal mechanisms in the distinctive Yellow River estuary will serve to broaden our understanding of inorganic carbon distributions, production and removal processes, and fluxes at the highly variable land–ocean interface. To identify the major processes controlling DIC removal in the Yellow River estuary, we collected lower-river and estuarine concentration data for water-column DIC, total alkalinity (TAlk), and other relevant parameters. DIC consumption due to net biological production and CaCO3 precipitation was also measured by bottle incubation. These data were analyzed and interpreted according to first principles of carbonate chemistry and estuarine mixing. Methods Site description—The Yellow River discharges into the Bohai Sea, which is located north of China’s Shandong Peninsula (Fig. 1). The lower reach of the river has only a single channel with no distributary branches. Due to rapid sediment deposition, the river mouth (delta) area has been advancing seaward at a recent rate of 1.1–2.6 km yr21 (Hu et al. 1998; Qiao et al. 2010). The estuary is a typical riverdominated estuary with weak tides; the tidally affected zone extends approximately 10–20 km upriver (Huang and Lu 1995). The Yellow River ranks low among the world’s large rivers in terms of water discharge (with a 1919–1995 average of 59 3 109 m3 yr21), but it ranks first or second in terms of suspended sediment load (with a 1950s–1970s average of 1.6 3 1015 g yr21; Chen et al. 2005, 2006). In recent decades, annual water and sediment fluxes of the Yellow River have declined dramatically due to climate change, reservoir construction, and irrigation-related withdrawals (Wang et al. 2006, 2007; Dai et al. 2009). Since 2002, discharges have been deliberately manipulated to prevent sediment accumulation in the downstream channel and reservoirs. In that year, the Yellow River Conservancy Commission began conducting WSR trials in the river’s middle reach during the early wet season (late June and early July). In 2005, the program became operational. The annual WSR scheme consists of two periods: the water release period, during which a large amount of water is released from the Xiaolangdi Reservoir (,700 km upstream from Lijin) for the purpose of flushing sediment from the downstream waterway; and the sediment release period, during which water is discharged from two upstream reservoirs (Sanmenxia and Wanjiazhai) for the purpose of flushing sediment from the Xiaolangdi Reservoir. (For more details, see http://www.yellowriver.gov.cn/ and Zhang et al. 2013.) DIC removal in the Yellow River Estuary These activities constitute an extreme human disruption of river-matter transport (Zhang et al. 2013). As a result, estuarine mixing conditions differ greatly between the preWSR (spring) and post-WSR (fall) periods. The summertime reservoir releases deliver to the estuary large quantities of sediment, most of which is deposited to form a cluster of barrier bars (Wang et al. 2005). This major depositional event temporarily confines estuarine mixing to short distances and times. In subsequent months, the barrier bars are gradually flushed away by less turbid river flow or eroded by tidal action. By spring of the next year, the bars have nearly disappeared and estuarine mixing conditions have returned to normal. Research cruises—Four cruises were carried out in the Yellow River estuary over the first 5 yr of the WSR operational program: September 2005, April 2006, May 2009, and September 2009. All investigations started from the Xintan floating bridge (Fig. 1). Because the river channel morphology changed yearly or even seasonally, the sampling locations were not the same for every cruise. The 2009 site locations were different from those in 2005 and 2006, but we do not expect this differnece to cause cruise-to-cruise differences in chemical and biological processes. Station locations during each cruise were chosen according to salinity variations. The river’s most seaward hydrological station (Lijin), where salinity was , 0, was always sampled as the freshwater end member. Average discharges during our cruises were as follows: 454 m3 s21 in September 2005, 372 m3 s21 in April 2006, 270 m3 s21 in May 2009, and 380 m3 s21 in September 2009. On all four cruises, discrete surface-water samples were collected and analyzed for salinity (S) and concentrations of total suspended solids (TSS), chlorophyll a (Chl a), DIC, TAlk, particulate inorganic carbon (PIC), and nutrients (nitrate, ammonia, and phosphate). During the 2009 cruises, surface-water salinity, temperature (T), saturation of dissolved oxygen (DO%), and partial pressure of carbon dioxide ðPCO2 Þ were also measured continuously. Two additional cruises were conducted in April and September 2004, but those data are not extensively discussed here because Chl a and freshwater endmember data are not available from those cruises. Lijin monthly monitoring—Between July 2010 and August 2011, monthly investigations of DIC were conducted at the Lijin hydrological station. Discharge data were obtained from China Hydrology (http://www.hydroinfo. gov.cn/). River discharges were lower in winter and spring than in summer and fall. Sampling and analytical methods—During the cruises of September 2005 and April 2006, surface S was measured at each discrete sampling station with a portable salinometer (3SYA2-2) that had a precision of 0.005. In May and September 2009, underway surface T and S were measured continuously using an SBE 45 MicroTSG (Sea-Bird Electronics). DO% was measured with a YSI 5000 oxygen analyzer (YSI) that had been calibrated using the Winkler titration method. CO2 partial pressure was measured using a nondispersive infrared spectrometer (LI-7000, LI-COR) 415 coupled to a shower-head equilibrator. All four parameters were automatically recorded each second and then averaged over 1 min, yielding nominal precisions of 0.002uC for temperature, 0.005 for salinity, 0.1% for DO%, and , 0.1%, for PCO2 (Zhang et al. 2010, 2012). Discrete water samples were collected into 5 liter Niskin bottles for later analysis of TSS, Chl a, DIC, TAlk, PIC, and nutrients. All samples were filtered in situ and were analyzed within 1 week. TSS samples were collected onto preweighed cellulose acetate membrane filters (pore size 0.45 mm) and preserved at 220uC. The membranes were later dried at 45–50uC and then weighed on an electronic balance (AL104, Mettler Toledo) with a precision of 0.0001 g. Chl a samples were collected onto glass fiber membrane filters (0.7 mm pore size; Whatman GF/F) at a pressure of , 0.05 MPa. Saturated magnesium carbonate solution was added to the membranes, which were preserved at 220uC. Membrane samples were later extracted with 90% acetone, and Chl a was determined using a fluorescence spectrophotometer (F-4500, Hitachi High-Technologies). The standard curve was constructed using Sigma C-5753 chlorophyll (Sigma-Aldrich). DIC and TAlk samples were filtered through cellulose acetate membranes (0.45 mm), poisoned with saturated mercury bichloride (final concentration , 0.02% by volume; Dickson and Goyet [1994]), and preserved at 4uC. DIC was measured using a total organic carbon analyzer (TOCVCPN, Shimadzu). TAlk was determined by Gran titration, using an alkalinity titrator (AS-ALK2, Apollo SciTech). Measurements of DIC and TAlk were both calibrated against Certified Reference Material (provided by A. G. Dickson from Scripps Institution of Oceanography) at a precision and accuracy level of 0.1%. PIC samples were collected onto glass fiber membrane filters (0.7 mm, Whatman GF/F) and preserved at 220uC. PIC was determined using a total organic carbon analyzer coupled to a combustion device for solid samples (TOC-VCPN and SSM-5000A, Shimadzu). The method of phosphoric acid extraction and combustion (200uC) has a measurement precision of 1%. This method necessitated the use of the GF/F filter (Aucour et al. 1999), which has a pore size that is greater than the size that operationally defines ‘‘particulates’’ (0.45 mm). As a result, this method likely underestimates PIC concentrations but to an extent that we believe is insignificant. Samples collected for nutrient analysis (NO3-N, NH4-N, and PO4-P) were filtered through pretreated cellulose acetate membranes (0.45 mm; immersed in 0.1% HCl solution over 24 h, then flushed with Milli-Q water). The filtrate was poisoned with chloroform and preserved at 220uC. The nutrient samples were later quickly thawed, then analyzed using an AutoAnalyzer 3 (Bran + Luebbe). Samples collected for determination of calcium ion (Ca2+) concentrations were filtered through cellulose acetate membrane filters (0.45 mm) and then analyzed by ethylene glycol bis (2-aminoethyl) tetraacetic acid titration. The precision of this method is 0.1% (Tsunogai et al. 1968). In situ incubations—During the two 2009 cruises, in situ incubations were carried out using the dark–light bottle 416 Liu et al. Fig. 2. Distribution of continuously measured surface temperature, salinity, DO%, and PCO2 in the Yellow River estuary: (A) May 2009 and (B) September 2009. The open circles indicate where discrete water samples were collected. method. Spring incubation salinities were 2.4, 5.1, and 10.0; fall salinities were 0.5, 2.5, 5.4, 10.3, and 19.6. For each salinity level, water was drawn into six light bottles and five dark bottles. The bottles were then placed into 0.5 m deep chests for incubation; temperature was maintained with continuously flowing estuarine surface water. For the light bottles, DO%, DIC, and TAlk were measured every 2 h from 06:00 h to 18:00 h (light intensity was . 10,000 lux). For the dark bottles, DO% was measured every 8 h. The total incubation time was 40 h. During the day, in situ light intensity was measured every 0.5 h with an illumination photometer (TES-1339 Light Meter Pro, TES Electrical Electronic). Estimation of DIC and TAlk removal—To calculate in situ removal of DIC and TAlk from the water column, we first used observed end-member concentrations to obtain theoretical dilution lines for constituents within the estuary (Officer 1979; Regnier et al. 1998). Values on the dilution lines (DICT and TAlkT) are the concentrations that would have theoretically been observed in the case of conservative mixing. Removal of DIC and TAlk from the water column (DDIC and DTAlk) were then obtained as the difference between the theoretical (T) and measured (M) constituent concentrations. The PIC analyses were not used to calculate DIC removal rates. Results Underway temperature, salinity, DO%, and PCO2 —In spring 2009, a downstream one-way survey of surface conditions was conducted from the Xintan floating bridge to offshore waters; in the fall, data were collected on the return leg as well (Fig. 1). Distributions of surface T, S, DO%, and PCO2 are shown in Fig. 2. Because all four parameters were nearly constant within the first 7 km downstream of the floating bridge, data for this stretch are not shown. Downstream of the 7 km point, salinity increased seaward within the estuarine mixing zone, with dramatic fluctuations observed in the neighborhood of the 12 km mark (at 12–13 km in the spring and 11–12 km in the fall). This pattern is typical of a river-dominated estuary with weak tides and a clear interface between freshwater and salt water. Within the mixing zone, seawater along the channel bottom moves upstream against the river current in a wedge shape (Huang and Lu 1995) and exhibits reciprocating flow. Variations of salinity, DO%, and PCO2 were similar in the spring and fall (Fig. 2). In the freshwater end member, DO% was relatively low, whereas PCO2 was high (though still generally , 75 Pa; 1 Pa 5 10.1 matm). At the sharp mixing interface (, 12 km), salinity, DO%, and PCO2 all fluctuated markedly over short distances. In general, DO% increased downstream (i.e., in the direction of increasing salinity), whereas PCO2 decreased. On approach to the highsalinity end member (S . 25), DO% remained oversaturated and PCO2 decreased to , 40 Pa. Some differences between the two cruises were also observed. In spring 2009, DO% increased downstream from 93% to 115%, becoming oversaturated where S , 18. In the fall, DO% ranged from 85% to 105%, and saturation was not observed until S . 25; at S , 18, DO% was only 95%. Spring temperatures decreased markedly downstream (from 21.5uC to 17.4uC), showing a negative correlation with salinity (i.e., the seawater was colder than the river water). In the fall, however, surface temperatures were nearly constant over the entire transect, except for a 1uC increase at , 12.0–12.5 km, where the water was very shallow. In the spring, mixing-zone fluctuations were more drastic and the mixing zone extended over a much greater distance than in the fall. DIC and TAlk—DIC and TAlk exhibited strong spatial and seasonal variation (Fig. 3), as well as some features unique to the Yellow River: (1) The Yellow River has extremely high DIC and TAlk values (,3178 mmol L21 and DIC removal in the Yellow River Estuary 417 Fig. 3. DIC, TAlk, and salinity in the Yellow River estuary. The solid lines are theoretical dilution lines for DIC and TAlk under conditions of conservative behavior; the dotted lines represent extrapolation of the high-salinity line segments back to S 5 0. (Lines are not shown for 2004 because freshwater end-member data are not available from these cruises.) (A, C, E) show spring data and (B, D, F) show fall data. Note the difference in vertical scale between the spring and fall graphs. 3242 mmol L21, respectively), likely the highest among large rivers worldwide (Cai et al. 2008). (2) DIC and TAlk concentrations were consistently higher in spring than in fall, opposite the pattern of river freshwater discharge (which is low in spring and maximal in fall). Spring DIC ranges were 2567–3718 and 2411–3384 mmol L21 (in 2006 and 2009, respectively); fall ranges were 2482–2682 and 2155–2927 mmol L21 (in 2005 and 2009). (3) DIC and TAlk decreased seaward with increasing salinity but with a nonlinear relationship that clearly indicates removal of DIC and TAlk in the low-salinity zone. This phenomenon was observed during all our surveys. The solid lines in Fig. 3 show the theoretical dilution lines for DIC and TAlk, calculated assuming conservative mixing of the end-member waters (Officer 1979; Regnier et al. 1998). Constituent removals and additions (DDIC and DTAlk) at any survey station can be obtained as the difference between the theoretical (T) values on the line and the measured (M) concentrations—i.e., on Fig. 3, the vertical distances between the dilution lines and the DICM and TAlkM data points. DDIC and DTAlk values thus calculated for the Yellow River estuary indicate that removals were consistently greater than additions, with net changes being larger in the spring than in the fall. The salinity ranges where DIC removal occurred were higher (i.e., more marine in character) in the spring than in the fall: S ,18 and 13 during the spring cruises (2006 and 2009, respectively) but , 10 during the fall cruises. DIC removal and TAlk removal were not coincident in space: the salinity ranges where DIC removal occurred were slightly higher (more marine) than where TAlk removal occurred (Fig. 3). At the seaward end of the mixing zones, DIC and TAlk varied linearly with salinity, indicating no removal or addition of these constituents in the later (more downstream) stages of mixing (Fig. 3). Extrapolation of the DIC conservative-mixing line segments from the high-salinity areas back to S 5 0 (shown as dotted lines in Fig. 3) yields the so-called effective concentrations of DIC exported from freshwater end members (DICe) when nonconservative removals and additions are considered. The difference between DICe and actual DIC measured in the freshwater end member is equal to the DIC that was removed during mixing and therefore not transported to the sea (Boyle et al. 1974; Cai and Wang 1998). On the basis of this approach, we estimate that approximately 5.3% (6 0.09%) and 11% (6 0.35%) of the Yellow River DIC was removed from the water column (not transported to the sea) for the fall 2005 and spring 2006 cruises, respectively. Approximately 5.7% (6 0.04%) and 3.8% (6 0.03%) of river DIC was removed for the spring and fall 2009 cruises. Thus, in the Yellow River estuary, DIC in situ percentage losses are greater in spring than in fall. TSS and PIC—Concentrations of TSS were consistently higher in the freshwater end members than in seawater (Fig. 4A), but clear differences in the TSS–S relationships 418 Liu et al. Fig. 4. Correlations between (A) TSS and salinity and (B) PIC% and salinity in the Yellow River estuary. were observed between the earlier and later cruises. During the first two cruises (fall 2005 and spring 2006), TSS concentrations in the freshwater end members were very high (1407 mg L21 and 1970 mg L21, respectively); concentrations during the final two cruises were significantly lower (900 mg L21 and 708 mg L21 in spring and fall 2009). On all four cruises, TSS declined sharply where S , 0.5, then decreased almost conservatively downstream as salinity increased. PIC% (i.e., PIC : TSS 3 100%) also decreased with increasing salinity (Fig. 4B) but with a different downstream pattern than TSS. Also unlike TSS, PIC% exhibited strong seasonal differences. During the spring cruises, PIC% remained high until appreciable levels of salinity were reached (S . 23 for spring 2006 and S . 15 for spring 2009). This pattern of sustained high PIC% suggests that PIC production was occurring in the low- and mid-salinity regions of the estuary. During the fall cruises, in contrast, PIC% deceased rapidly where S , 5. (As noted in the Methods section, the PIC% numbers may be slight underestimates. The patterns and trends are, however, robust because the same method was used on all cruises. We do not use absolute PIC% values to quantify DDIC or other derived quantitites.) We have no direct evidence that spring PIC production in the low-to-moderate salinity area was caused by CaCO3 precipitation, but the overlapping salinity ranges of PIC production and DIC and TAlk Fig. 5. removal strongly suggest in situ CaCO3 precipitation as a common cause. Nutrients—Nitrogen, phosphorus, and other nutrients are important in influencing phytoplankton growth. During the 2009 cruises, nitrate (NO3-N) concentrations were as high as 265–284 mmol L21 in the freshwater end members, but then decreased with increasing salinity (Fig. 5A). Concentrations of ammonia (NH4-N) and phosphate (PO4-P) were very low (Fig. 5B, C). The fall distributions indicate that some in situ generation of these constituents may have been occurring in conjunction with mixing (Fig. 5B, C). Fall NH4-N and PO4-P concentrations were higher than spring concentrations, consistent with the observation of lower DO% in the fall (Fig. 2). The highest NH4-N concentration was 5.52 mmol L21 (much lower than NO3-N); NH4-N could not have been the main nitrogen source for phytoplankton growth. PO4-P concentrations ranged from 0.03 to 0.40 mmol L21, sufficient for phytoplankton photosynthesis. Chl a—During the fall 2005 and spring 2006 cruises, the observed Chl a concentration ranges were 0.35–5.96 and 0.88–10.5 mg L21, respectively (Fig. 6A). Concentrations in the mixing zones were higher than in either end member, with the highest values observed at S 5 5 (fall 2005) and S 5 8 (spring 2006). For the 2009 cruises, Chl a Nutrient concentrations as a function of salinity in the Yellow River estuary: (A) nitrate, (B) ammonia, and (C) phosphate. DIC removal in the Yellow River Estuary 419 Fig. 6. Chlorophyll a concentrations as a function of salinity in the Yellow River estuary: (A) cruises of September 2005 and April 2006, and (B) cruises of May and September 2009. concentrations were much higher in freshwater than in seawater. Spring values generally declined with increasing salinity; fall values remained relatively high until the point where S .12 (Fig. 6B). DO%, DIC, and TAlk changes during in situ incubations— From variations in measured DO%, DIC, and TAlk in the 2009 light and dark bottle incubations, rates can be calculated for these processes: oxygen production in the light bottles (FDO(light); net primary production), oxygen uptake in the dark bottles (FDO(dark); respiration), and DIC and TAlk consumption in the light bottles (FDICM and FTAlkM). (Day lengths for the calculated rates are 12 h for light-bottle quantities and 24 h for dark-bottle quantities.) Several interesting features were observed (Table 1): (1) net primary production rates, FDO(light), were consistently much higher than respiration rates, FDO(dark); (2) rates of DIC consumption, FDICM, were quite high in low-salinity waters and decreased with increasing salinity; (3) TAlk removal, FTAlkM, was observed only in low-salinity (S , 5) waters; and (4) for a given salinity, the DIC consumption rate attributable to net biological production was greater in the spring than in the fall. Seasonal DIC variations—DIC concentrations were measured at the Lijin hydrological station from July 2010 to August 2011 (Fig. 7). These concentration data agree very well with DIC concentrations measured in freshwater end members during our four cruises (open circles on Fig. 7). In general, DIC was negatively correlated with discharge except during the midsummer WSR periods, when large quantities of water were being released from upstream reservoirs. DIC concentrations were low during the wet, high-discharge season of late summer and early fall, then increased steadily. Highest DIC was observed during the dry, low-discharge period of winter and early spring. Concentrations thereafter decreased into the summer. Negative correlations of DIC with discharge have been observed in many rivers (Cai et al. 2008). WSR releases occurred on 19 June–02 July (water release period) and 03 July–10 July (sediment release period) in 2010 and 19 June–03 July (water release) and 04 July–12 July (sediment release) in 2011. In association with these releases, DIC concentrations at Lijin were observed high on 06 July 2010 and again the following year on 25 June and 08 July (Fig. 7). During the 2011 WSR period, for instance, a large quantity of water (, 2600 m3 s21) was released from the Xiaolangdi Reservoir on 19 June. This water reached the Lijin station on 25 June, as seen in the roughly fivefold increase in discharge. Sediment that had been previously deposited in the Yellow River watercourse was flushed out and resuspended, leading to a sharp increase in TSS on 25 Table 1. Results of the 2009 in situ incubation experiments. FDO(light) and FDO(dark) give measured net primary production and respiration rates. FDICM and FTAlkM give measured consumption rates of DIC and TAlk in the light bottles. Dashes represent no experimental data. (Day lengths: 12 h for light bottles and 24 h for dark bottles.) Salinity Spring Fall 2.4 5.1 10.0 0.5 2.5 5.4 10.3 19.6 FDO(light) (mmol O2 L21 d21) 188 151 119 123 91 68 48 13 (610.0) (63.1) (612.5) (60.8) (60.4) (60.5) (61.0) (60.9) FDO(dark) (mmol O2 L21 d21) 223 223 211 217 215 214 210 24 (60.2) (60.2) (60.2) (60.4) (60.2) (60.4) (60.2) (60.2) FDICM (mmol C L21 d21) 2202 2153 2108 2212 2118 272 249 212 (61.7) (60.7) (60.9) (62.5) (66.6) (66.2) (62.9) (60.9) FTAlkM (mmol C L21 d21) — — — 2207 (63.7) 284 (60.4) 223 (61.4) 27 (61.0) 22 (60.7) 420 Liu et al. Fig. 7. Discharge rates and DIC concentrations measured at the Lijin station from July 2010 to August 2011. The open circles (with dates) show DIC concentrations measured in the freshwater end members of the four earlier cruises. June (up to 6685 mg L21, about 10 times as much as preWSR). DIC increased 25%, to achieve a concentration of 3458 mmol L21. Discharge was still high on 08 July, but sediment and DIC concentrations had by then begun to decline. After the WSR ended on 12 July, discharge rates and DIC concentrations returned to normal. This pattern of high DIC in association with high discharge is contrary to the pattern of natural seasonal variability. On 01 August 2010, very low DIC was observed at Lijin, probably due to a dilution effect associated with heavy rainfall. During 24 July–03 August, seven heavy downpours were recorded at the middle or lower reaches of the Yellow River. At the Huayuankou gauging station (in the middle Yellow River basin) and at the Lijin station, associated discharges were 68% and 108% greater than during the flood period of 2009 (http://www.yellowriver.gov.cn/). The annual average concentration of DIC at Lijin was 3158 mmol L21. This value is higher than the 2591 mmol L21 estimate given in the synthesis of Cai et al. (2008), which was based on earlier observations. estuaries (e.g., those of the Changjiang, Pearl, Mississippi, and Hooghly rivers; Mukhopadhyay et al. 2002; Zhai et al. 2005, 2007). WSR events, which can change DIC and TAlk transport patterns within a short period, all fell outside the times of our four cruises. Thus, we conclude by the process of elimination that internal removal processes such as net biological production and CaCO3 precipitation must be responsible for the observed nonconservative behavior of DIC in the Yellow River estuary. Due to the chemical changes that occur when river water encounters seawater—i.e., changing pH and ionic strength and especially rapidly increasing Ca2+—calcium carbonate precipitation is likely to occur within the estuary. Gan et al. (1983), on the basis of calculations of calcite saturation index (SICalcite), noted that calcite was oversaturated in the Yellow River mainstream. We calculated SICalcite from our 2005–2009 estuarine cruise data: Discussion where [ ] represents concentration and KCalcite is the solubility product of calcite (a polymorph of the mineral calcium carbonate). For water in an equilibrium state with respect to calcite, log10(SICalcite) 5 0. A value , 0 indicates undersaturation, and a value . 0 indicates oversaturation (Mucci 1983; Neal 2002). On all four cruises, CaCO3 was oversaturated in lowsalinity areas (Fig. 8). Highest values of SICalcite were observed during the spring cruise of 2006, when calcite oversaturation occurred over a broad range of salinity, from 0 to 18. DIC removal was also greatest during this cruise (Fig. 3). During the two fall cruises (2005 and 2009), SICalcite and DIC removal were much lower, with the zones of calcite oversaturation (Fig. 8) and DIC removal (Fig. 3) restricted to much lower (less marine) salinity ranges. These Causes of estuarine DIC removal—Many processes can contribute to nonconservative DIC behavior in estuaries (Abril et al. 2003; Cai et al. 2004)—e.g., net biological production and respiration, CaCO3 dissolution and precipitation, ammonium oxidation, denitrification, tributary or local anthropogenic inputs (e.g., sewage), and CO2 air–sea exchange. For the Yellow River estuary, some of these processes can be eliminated from consideration. This estuary has no tributaries and no direct anthropogenic inputs. Its waters are characterized by high DO% and relatively low CO2 degassing rates. During the 2009 cruises, DO% ranged between 85% and 115%, whereas PCO2 ranged between 35 Pa and 75 Pa, far less than observed in some other large log10 ðSICalcite Þ~ {log10 ðKCalcite Þ log10 Ca2z | CO2{ 3 ð1Þ DIC removal in the Yellow River Estuary { z 106CO2 z16NO{ 3 zH2 PO4 z122H2 Oz17H ~ðCH2 OÞ106 ðNH3 Þ16 ðH3 PO4 Þz138O2 Fig. 8. Relationship between the log of the calcite saturation index (SICalcite) and salinity. The dotted line, where log10(SICalcite) 5 0, represents saturation with respect to calcite. Values . 0 indicate oversaturation; values , 0 indicate undersaturation. results suggest that CaCO3 precipitation is probably responsible for DIC removal in the low-salinity areas of the estuary. However, the salinity zones of linked processes were not always coincident. The salinity range within which DIC removal occurred was a little broader than the salinity range of calcite oversaturation. The salinity zones for TAlk removal and calcite oversaturation were almost the same. Other factors besides chemical precipitation of CaCO3 (e.g., biological production) clearly contributed to the removal of DIC from estuary waters. The ratio DDIC : DTAlk can be used to determine which processes—CaCO3 precipitation or net biological production or both—may be influencing the removal of DIC from the water column (Cai et al. 2004; Ortega et al. 2005, 2008). CaCO3 precipitation alone removes DIC and TAlk in a ratio (DDIC : DTAlk) of 1 : 2. When CaCO3 precipitation and net biological production co-occur, DDIC : DTAlk . 1 : 2 because net biological production results in a sharp decrease in DIC but very little change in TAlk (Redfield et al. 1963): 421 ð2Þ The greater the DDIC : DTAlk ratio, the greater the contribution from net biological production. In the Yellow River estuary (Fig. 9), DDIC : DTAlk values were 0.946 (fall 2005), 1.29 (spring 2006), 1.70 (spring 2009), and 1.88 (fall 2009). The fact that DDIC : DTAlk was consistently greater than the benchmark value of 0.5 suggests that not only CaCO3 precipitation but also net biological production contributed to DIC removal during all four cruises. The relative effect of net biological production vs. CaCO3 precipitation was strongest during the 2009 cruises. Biological production consumes not only DIC but also NO3-N, with a DIC : NO3-N ratio of 106 : 16 (Eq. 2). In the Yellow River estuary, evidence of NO3-N removal was seen in the low-salinity DIC-removal area (Fig. 5A), but the NO3-N effect was subtle because river-water concentrations were so high. If DIC and TAlk removal in the low-salinity area (i.e., where S , 18) is attributed to net biological production and CaCO3 precipitation, the following equations can be derived: DDIC~DDICOC zDDICCaCO3 ð3Þ DTAlk~DTAlkOC zDTAlkCaCO3 ð4Þ CaCO3 precipitation removes DIC and TAlk in a 1 : 2 ratio, DDICCaCO3 =DTAlkCaCO3 ~1=2 ð5Þ whereas according to Eq. 2, the DIC and TAlk change ratio due to biological production is: DDICOC =DTAlkOC ~{106=17 ð6Þ Thus, DIC and TAlk variations caused by biological production (DDICOC and DTAlkOC) and CaCO3 precipitation Fig. 9. Relationship between DDIC and DTAlk: (A) fall 2005 and spring 2006; (B) spring and fall 2009. 422 Liu et al. Fig. 10. Contributions of net biological production and CaCO3 precipitation to changes in DIC concentration: (A) April 2006, (B) September 2005, (C) May 2009, and (D) September 2009. Negative DDIC indicates DIC removal. (DDICCaCO3 and DTAlkCaCO3) can be calculated by solving Eqs. 3–6. It is important to note that DDICOC and DDICCaCO3 are accumulative properties that include removal signals inherited from processes occurring upstream of the station location (i.e., between the riverine end member and the station in question). Calculated DDICCaCO3 and DTAlkCaCO3 values (Fig. 10) demonstrate that DIC removal was strongest during the spring cruises, with DDIC decreasing in the following order: April 2006, May 2009, September 2009, and September 2005. The salinity ranges over which DIC removal occurred also decreased in the same order. These rankings are consistent with our qualitative assessment above (Fig. 3). In the first two cruises (Fig. 10A,B), the contribution of net biological production to DIC removal was nearly equal to or only slightly greater than that of CaCO3 precipitation. During the latter two cruises, however, the contribution of net biological production to DIC removal was almost twice that of CaCO3 precipitation (Fig. 10C,D). Insights from in situ incubations—Our in situ bottle incubations provide direct evidence for a mechanistic interpretation of the DIC removal signals (Table 1). Measured consumption rates of DIC (FDICM) and TAlk (FTAlkM) decreased with increasing salinity, indicating more intense DIC and TAlk removal in the low-salinity area of the estuary. The fact that FDO(light) was always larger than FDO(dark) suggests that biological production (which consumes DIC and produces DO) was the primary biological process. TAlk removal, like DIC removal, was greatest in the low-salinity (S , 5) incubations, indicating that CaCO3 precipitation also plays an important role in DIC removal. The rate of DIC consumption due to net biological production (FDICOC) can be calculated from measurements of FDO(light) and the 106 : 138 ratio of Eq. 2: FDICOC ~{ð106=138ÞFDOðlightÞ ð7Þ The rate of DIC consumption due to CaCO3 precipitation (FDICCaCO3 ) can be similarly calculated from measurements of FTAlkM, the 1 : 2 ratio of DDIC : DTAlk caused by CaCO3 precipitation (Eq. 5), and the 17 : 138 ratio of DTAlk : DDO caused by photosynthesis (Eq. 2): FDICCaCO3 ~FTAlkCaCO3 =2 ~ðFTAlkM {FDICOC Þ=2 ð8Þ ~FTAlkM =2{ð17=276ÞFDOðlightÞ For our Yellow River estuary incubations, these calculations indicate that the total DIC removal rate (FDICM) and both of its component rates (FDICOC and FDICCaCO3 ) were highest in the low-salinity bottles (where S , 5). All three rates decreased with increasing salinity (Fig. 11). This result further indicates that both net biological production and CaCO3 precipitation were responsible for the removal of DIC from the waters of the Yellow River estuary. Figure 11 also shows that as salinity increased, the relative importance of CaCO3 precipitation decreased and the relative importance of biological removal increased. In the freshest waters, the two component contributions were nearly equal, but by the location (condition) where S , 20, biological removal was clearly dominant. Our incubation conditions were more similar to surfacewater conditions as water was pumped from the surface into the 0.5 m deep incubation chamber. Nevertheless, solar DIC removal in the Yellow River Estuary Fig. 11. DIC consumption during the in situ incubations of fall 2009. FDICCaCO3 (white) is the rate of DIC consumption attributed to CaCO3 precipitation, and FDICOC (black) is the rate attributed to net biological production. FDICM (gray) is the DIC consumption rate (total) measured in the light bottles. insolation at the chamber was greater than at lower water column deeper than 0.5 m. Thus, depth-integrated O2 production rates based on incubation-derived light-bottle rates may be higher than actual water-column O2 production rates. If so, then our incubation-derived assessment of the relative importance of CaCO3 precipitation in DIC removal would be a conservative estimate. Overall, the FDICCaCO3 : FDICOC ratio (or DDICCaCO3 : DDICOC ratio) is somewhat lower from the incubation data (Fig. 11) than from the field data (Fig. 10). Primary productivity (as reflected in FDO(light)) was much lower in fall 2009 than in the preceding spring (Table 1), consistent with the extent of water-column DIC removal measured during the two cruises (Fig. 3). Light intensity differed little between the cruises (13,000–92,950 lux vs. 10,100– 93,280 lux), and water temperatures were actually higher in the fall than in the spring (Fig. 2). The reasons for the lower primary productivity in the fall merit further investigation. Seasonal variations of DIC—Because net biological production and CaCO3 precipitation are important mechanisms of DIC removal in the Yellow River estuary, DIC distributions were strongly correlated with Chl a concentrations and, as noted above, SICalcite. Observed Chl a concentrations give only the standing crop and may have a complicated relationship with in situ biological production rates. Nevertheless, Chl a distributions (Fig. 6) were consistent with distributions of DIC removal due to biological production (Fig. 10). For instance, Chl a concentrations were highest in spring 2006, especially at the mixing interface. DIC removal due to net biological production (DDICOC) was also highest during this cruise. Chl a concentrations were lowest in fall 2005, as was the contribution of net biological production to DIC removal. Looking at spatial (salinity) distributions in 2009, DDICOC can be seen to reach a maximum where Chl a concentrations were highest (i.e., where S , 5). 423 As mentioned above, the salinity range where DIC removal occurred was generally a little broader than the range where calcite oversaturation was observed (Figs. 3, 8). On three of the four cruises, there were locations in the estuary where waters were undersaturated with respect to calcite and TAlk behavior was nearly conservative (where S . 6 in September 2005, S . 18 in April 2006, and S . 5 in September 2009). This combination indicates that chemical precipitation of calcite was unlikely in these areas. However, DIC removal due to biological production (as indicated by high Chl a) continued in the undersaturated, mid-salinity waters. On the fourth cruise (May 2009), where CaCO3 was undersaturated, Chl a was relatively low. Our study suggests that biological and chemical removals of DIC are also regulated by DIC concentrations in the freshwater end members and by freshwater–seawater mixing distances (and times). Higher DIC concentrations in freshwater were accompanied by more severe DIC removal. Data from our four core cruises plus two earlier ones (Fig. 3) and the monthly Lijin observations (Fig. 7) show that freshwater DIC concentrations were higher in spring than in fall. The proportion of DIC removal was also greater in the spring, as were the corresponding salinity ranges. This pattern is probably due to the fact that higher freshwater DIC concentrations resulted in higher SICalcite values and more spatially extensive zones (wider salinity ranges) of calcite oversaturation (Fig. 8). Mixing distances and times are influenced mainly by estuary geomorphology (shape) and magnitude of discharge. Longer mixing distances and times provide conditions favorable for net biological production and CaCO3 precipitation, thus leading to greater DIC removal and a wider range of salinity encompassed by the DIC removal zone. For a given season, greater discharges alone can lead to longer mixing distances. For comparing across different seasons, though, the effect of estuary shape cannot be ignored. For example, during our fall cruises, barrier bars in the estuary limited mixing to a very short distance and time. In spring, the bars were absent, and mixing distances and times were much longer (e.g., May 2009, Fig. 2). In summary, the Yellow River estuary data indicate that higher Chl a concentrations and higher SICalcite values generally correlate to greater DIC removal through the processes of net biological production and CaCO3 precipitation. To the extent that freshwater end-member DIC concentrations and freshwater–seawater mixing distances (and times) influence production and precipitation, these factors may also influence DIC removal from estuarine waters. Causes of DIC increase during WSR events—An important question is why DIC increased during the WSR flood periods but decreased during natural flood periods (Fig. 7). This phenomenon is most likely caused at least in part by the different water origins for the different types of floods. During the WSR period, Lijin floodwaters came mainly from upstream reservoirs, which held waters of relatively high DIC content. Natural floodwaters, however, originated primarily as rainwater of very low DIC. As observed in other rivers (Cai et al. 2008), dilution effects associated with heavy precipitation can be significant. 424 Liu et al. High-DIC source water alone is not sufficient to explain the DIC increase observed at Lijin during the WSR period. DIC in the originating Xiaolangdi Reservoir is generally lower than DIC at Lijin. In July 2007, for example, Xiaolangdi DIC was 3370 mmol L21, whereas Lijin DIC was 3540 mmol L21. In July 2009, Xiaolangdi DIC was 3040 mmol L21 and Lijin DIC was 3180 mmol L21 (our unpubl. data). Other factors must be invoked to explain the increase in DIC between the reservoir and the river mouth. During the initial water-release phase of the annual WSR program, a large amount of clear water is discharged from the Xiaolangdi Reservoir, and downstream TSS concentrations increase sharply due to resuspension of riverbed sediments. Suspended sediments in the Yellow River originate mainly from the Loess Plateau (Chen et al. 2005), which has a carbonate content as high as 15% (Cai et al. 2008). Near-bottom and sediment-interstitial waters in the watercourse must therefore have very high DIC concentrations due to prolonged contact with carbonate-rich particles. Vigorous flushing associated with the WSR release also likely accelerates weathering and encourages the transformation of PIC to DIC. Thus, we attribute the high DIC observed at Lijin in association with the Xiaolangdi clearwater release to not only high-DIC source water but also to flushing out of high-DIC bottom and riverbed water and to enhanced PIC dissolution. This scenario is similar to situations in which severe soil loss and erosion lead to intensified chemical weathering (Van Oost et al. 2007). During the subsequent sediment-release phase, Xiaolangdi Reservoir sediments and near-bottom and sedimentinterstitial waters are flushed from the reservoir and transported downriver. These sediments and waters are high in DIC content, and intense chemical weathering during the long-distance transport serves to keep riverwater DIC concentrations relatively high. At Lijin, sediment discharge started to decline on 08 July 2011, even though water discharge was still high. DIC concentration declined at that same time, further linking high DIC concentrations to sediment-related processes. Following the shutoff of the WSR release on approximately 12 July, Lijin discharge rates and concentrations of TSS and DIC declined to normal levels. We considered the possibility that organic matter degradation may contribute to the DIC increase observed at Lijin during WSR events but concluded that this is unlikely. The ratio of particulate organic carbon (POC) to Chl a during recent releases was high—as high as , 4000 in the Yellow River mainstream and , 960 in the reservoirs— which indicates low biological production and POC degradation (Zhang et al. 2013). Considerations of composition also argue against an organic contribution to high DIC. The POC content of suspended sediments during WSR periods was nearly the same as that of loess (0.6% ; Zhang et al. 2013), and POC is typically composed mainly of refractory natural humus of low degradability (Chen et al. 2003, 2004). Indirect contributions from the pool of dissolved organic carbon (DOC) also seem unlikely. DOC concentrations in the Yellow River reservoirs (228– 337 mmol L21 ; Zhang et al. 2013) were much lower than the observed DIC increase at Lijin (,600 mmol L21; Fig. 7). Possible WSR effects on estuarine DIC removal—Reservoir construction is often called for along rivers in arid and semiarid regions due to scarce water resources and the demands of agricultural activity. In the Yellow River, turbidity is extremely high and a great deal of transported particulate material is retained within its reservoirs (Zhang et al. 2013). In recent decades, sediment discharge from the river has declined dramatically, from average 1.6 3 1015 g yr21 in the 1950s–1970s (Chen et al. 2005, 2006) to as low as average 0.49 3 1015 g yr21 during the 1980s– 1990s (http://www.yellowriver.gov.cn/). During WSR release events, Lijin water discharges and DIC concentrations increase. These factors undoubtedly result in greater DIC removal within the estuary and removal over a wider range of salinity. DIC removal due to chemical precipitation of CaCO3 in particular increases, but removal due to biological production may be limited during these periods by the extremely high TSS concentrations. The sharply increased discharge shifts the mixing zone and the DIC removal zone seaward. On the other hand, most of the sediment flushed from the reservoirs and the watercourse is deposited in the river mouth or on reclaimed tidal flats, thereby reducing the opportunity for the newly precipitated CaCO3 to be transported to the sea. In conclusion, we speculate that Yellow River WSR activities likely enhance DIC removal in the estuary. More research is needed to understand the effects of this acute human intervention within the context of simultaneous long- and medium-term human interventions (e.g., land use changes, irrigation, damming) on land–river–ocean carbon cycling. Implications—DIC removal in other large estuaries has been attributed primarily to biological production or mixing. In the Mississippi River plume, for example, DIC removal has been controlled mainly by biological production (Cai 2003; Guo et al. 2012; Huang et al. 2012). In the Scheldt Estuary, removal processes have been dominated primarily by the mixing of waters contributed from tributaries (Hellings et al. 2001). To our knowledge, the Yellow River estuary is the only example where CaCO3 precipitation has been demonstrated to play a very important role in estuarine DIC removal. Milliman et al. (2008) concluded that 34 representative world rivers experienced 30% declines in discharge during the last half of the 20th century due to damming, irrigation, and interbasin water transfers. Of those, 18 (including the Indus and Yellow rivers) experienced discharge declines of . 50%. Most of the 34 rivers drain arid and semiarid regions in Africa, Asia, and Australia, and all of them are characterized by high DIC concentrations (Cai et al. 2008). If the patterns and processes observed in the Yellow River are generally applicable, these decreased discharges (due to irrigation and more evaporation than precipitation) may lead to yet higher DIC concentrations and thus enhanced CaCO3 precipitation in the estuarine zone. This alteration may in turn lead to a general trend of increasing importance of CaCO3 precipitation in DIC removal at the river–ocean interface in arid and semiarid regions. Understanding DIC and DIC removal in the Yellow River Estuary TAlk removal mechanisms in the Yellow River estuary may aid our understanding of DIC distributions at the land–ocean interface in general and of DIC fluxes to the ocean in an important category of rivers. Acknowledgments We thank Dongmei Liu, Qizhen Liu, Peng Yin, Min Wang, Baosen Wang, Chunchao Xiao, and Yan Cui for sampling and measurement work. We also thank Tonya Clayton for her help in language editing. Extensive and thoughtful comments and guidance from the reviewers and editors are highly appreciated. This work was supported by the National Science Foundation of China (grant 41173107) and the Foundation for Innovative Research Groups of the National Science Foundation of China: Marine Organic Biogeochemistry (grant 41221004). References ABRIL, G., H. ETCHEBER, B. DELILLE, M. FRANKIGNOULLE, AND A. V. BORGES. 2003. Carbonate dissolution in the turbid and eutrophic Loire estuary. Mar. Ecol. Prog. Ser. 259: 129–138, doi:10.3354/meps259129 AUCOUR, A. M., S. M. F. SHEPPARD, O. GUYOMAR, AND J. WATTELET. 1999. Use of 13C to trace origin and cycling of inorganic carbon in the Rhône river system. Chem. Geol. 159: 87–105, doi:10.1016/S0009-2541(99)00035-2 CAI, W. J. 2003. Riverine inorganic carbon flux and rate of biological uptake in the Mississippi River plume. Geophys. Res. Lett. 30: 1–4, doi:10.1029/2002GL016312 ———. 2011. Estuarine and coastal ocean carbon paradox: CO2 sinks or sites of terrestrial carbon incineration? Annu. Rev. Mar. Sci. 3: 123–145, doi:10.1146/annurev-marine-120709-142723 ———, L. R. POMEROY, M. A. MORAN, AND Y. C. WANG. 1999. Oxygen and carbon dioxide mass balance for the estuarine– intertidal marsh complex of five rivers in the southeastern U.S. Limnol. Oceanogr. 44: 639–649. ———, AND Y. C. WANG. 1998. The chemistry, fluxes, and sources of carbon dioxide in the estuarine waters of Satilla and Altamaha Rivers, Georgia. Limnol. Oceanogr. 43: 657–668, doi:10.4319/lo.1998.43.4.0657 ———, AND OTHERS. 2004. The biogeochemistry of inorganic carbon and nutrients in the Pearl River estuary and the adjacent northern South China Sea. Cont. Shelf Res. 24: 1301–1319, doi:10.1016/j.csr.2004.04.005 ———, AND ———. 2008. A comparative overview of weathering intensity and HCO 3 flux in the world’s major rivers with emphasis on the Changjiang, Huanghe, Zhujiang (Pearl) and Mississippi Rivers. Cont. Shelf Res. 28: 1538–1549, doi:10.1016/j.csr.2007.10.014 CAUWET, G., AND F. T. MACKENZIE. 1993. Carbon inputs and distribution in estuaries of turbid rivers: The Yang Tze and Yellow Rivers (China). Mar. Chem. 43: 235–246, doi:10.1016/ 0304-4203(93)90229-H CHEN, J. S., D. W. HE, AND Y. ZHANG. 2003. Is COD a suitable parameter to evaluate the water pollution in the Yellow River? Environ. Chem. 22: 611–614. ———, F. Y. WANG, AND D. W. HE. 2006. Geochemistry of water quality of the Yellow River basin. [In Chinese.] Earth Sci. Front. 13: 58–73, doi:10.3321/j.issn:1005-2321.2006. 01.009 ———, ———, M. MEYBECK, D. W. HE, X. H. XIA, AND L. T. ZHANG. 2005. Spatial and temporal analysis of water chemistry records (1958–2000) in the Huanghe (Yellow River) basin. Global Biogeochem. Cycles 19: GB3016, doi:10.1029/ 2004GB002325 425 ———, Y. ZHANG, T. YU, AND D. W. HE. 2004. A study on dissolution and bio-degradation of organic matter in sediments from the Yellow River. Acta Sci. Circumst. 24: 1–5. DAI, M. H., L. F. WANG, X. H. GUO, W. D. ZHAI, Q. LI, B. Y. HE, AND S. J. KAO. 2008. Nitrification and inorganic nitrogen distribution in a large perturbed river/estuarine system: The Pearl River Estuary, China. Biogeosciences 5: 1227–1244, doi:10.5194/bg-5-1227-2008 DAI, S. B., S. L. YANG, AND M. LI. 2009. The sharp decrease in suspended sediment supply from China’s rivers to the sea: Anthropogenic and natural causes. Hydrol. Sci. 54: 135–146, doi:10.1623/hysj.54.1.135 DICKSON, A. G., AND C. GOYET [EDS.]. 1994. U.S. Department of Energy (DOE). Handbook of methods for the analysis of the various parameters of the carbon dioxide system in sea water, v. 2. ORNL/CDIAC-74. Available from http://cdiac.esd.ornl. gov/oceans/DOE_94.pdf GAN, W. B., H. M. CHEN, AND Y. F. HAN. 1983. Carbon transport by the Yangtze (at Nanjing) and Huanghe (at Jinan) Rivers, People’s Republic of China, p. 459–470. In E. T. Degens [ed.], Transport of carbon and minerals in major world rivers (Part 2). SCOPE/UNEP. GUO, X. H., W. J. CAI, W. D. ZHAI, M. H. DAI, Y. C. WANG, AND B. S. CHEN. 2008. Seasonal variations in the inorganic carbon system in the Pearl River (Zhujiang) estuary. Cont. Shelf Res. 28: 1424–1434, doi:10.1016/j.csr.2007.07.011 ———, AND OTHERS. 2012. Carbon dynamics and community production in the Mississippi River plume. Limnol. Oceanogr. 57: 1–17, doi:10.4319/lo.2012.57.1.0001 HANS, H. D., G. L. GOULVEN, M. K. CHERYL, P. S. CAROLINE, M. MICHEL, AND M. HANS. 2011. Worldwide typology of nearshore coastal systems: Defining the estuarine filter of river inputs to the oceans. Estuaries Coasts 34: 441–458, doi:10.1007/s12237-011-9381-y HELLINGS, L., F. DEHAIRS, S. V. DAMME, AND W. BAEYENS. 2001. Dissolved inorganic carbon in a highly polluted estuary (the Scheldt). Limnol. Oceanogr. 46: 1406–1414, doi:10.4319/ lo.2001.46.6.1406 HU, C. H., Z. W. JI, AND T. WANG. 1998. Dynamic characteristics of sea currents and sediment dispersion in the Yellow River Estuary. Int. J. Sed. Res. 13: 20–30. HUANG, S., AND Q. M. LU. 1995. Estuary hydraulics. Water Resources and Electric Power Press. [In Chinese.] HUANG, W. J., W. J. CAI, R. T. POWELL, S. E. LOHRENZ, Y. WANG, L. Q. JIANG, AND C. S. HOPKINSON. 2012. The stoichiometry of inorganic carbon and nutrient removal in the Mississippi River plume and adjacent continental shelf. Biogeosciences 9: 2781–2792, doi:10.5194/bg-9-2781-2012 LERMAN, A., L. WU, AND F. T. MACKENZIE. 2007. CO2 and H2SO4 consumption in weathering and material transport to the ocean, and their role in the global carbon balance. Mar. Chem. 106: 326–350, doi:10.1016/j.marchem.2006.04. 004 LUDWIG, W., P. AMIOTTE-SUCHET, G. MUNHOVEN, AND J. L. PROBST. 1998. Atmospheric CO2 consumption by continental erosion: Present-day controls and implications for the last glacial maximum. Glob. Planet. Change 16–17: 107–120, doi:10.1016/S0921-8181(98)00016-2 MACKENZIE, F. T., A. LERMAN, AND A. J. ANDERSSON. 2004. Past and present of sediment and carbon biogeochemical cycling models. Biogeosciences 1: 11–32, doi:10.5194/bg-1-11-2004 MILLIMAN, J. D., K. L. FARNSWORTH, P. D. JONES, K. H. XU, AND L. C. SMITH. 2008. Climatic and anthropogenic factors affecting river discharge to the global ocean, 1951–2000. Glob. Planet. Change 62: 187–194, doi:10.1016/j.gloplacha.2008.03.001 426 Liu et al. MUCCI, A. 1983. The solubility of calcite and aragonite in seawater at various salinities, temperatures, and one atmosphere total pressure. Am. J. Sci. 283: 780–799, doi:10.2475/ajs.283.7.780 MUKHOPADHYAY, S. K., H. BISWAS, T. K. DE, S. SEN, AND T. K. JANA. 2002. Seasonal effects on the air–water carbon dioxide exchange in the Hooghly estuary, NE coast of Bay of Bengal, India. J. Environ. Monit. 4: 549–552, doi:10.1039/b201614a NEAL, C. 2002. Calcite saturation in eastern UK rivers. Sci. Tot. Environ. 282: 311–321, doi:10.1016/S0048-9697(01)00921-4 OFFICER, C. B. 1979. Discussion of the behaviour of the nonconservative dissolved constituents in estuaries. Estuar. Coast. Mar. Sci. 9: 91–94, doi:10.1016/0302-3524(79)90009-4 ORTEGA, T., R. PONCE, J. FORJA, AND A. GÓMEZ-PARRA. 2005. Fluxes of dissolved inorganic carbon in three estuarine systems of the Cantabrian Sea (north of Spain). J. Mar. Syst. 53: 125–142, doi:10.1016/j.jmarsys.2004.06.006 ———, ———, ———, AND ———. 2008. Benthic fluxes of dissolved inorganic carbon in the Tinto–Odiel system (SW of Spain). Cont. Shelf Res. 28: 458–469, doi:10.1016/j.csr.2007.10.004 QIAO, S. Q., X. F. SHI, A. M. ZHU, Y. G. LIU, N. S. BI, X. S. FANG, AND G. YANG. 2010. Distribution and transport of suspended sediments off the Yellow River (Huanghe) mouth and the nearby Bohai Sea. Est. Coast. Shelf Sci. 86: 337–344, doi:10.1016/j.ecss.2009.07.019 RAYMOND, P. A., J. E. BAUER, AND J. J. COLE. 2000. Atmospheric CO2 evasion, dissolved inorganic carbon production, and net heterotrophy in the York River estuary. Limnol. Oceanogr. 45: 1707–1717, doi:10.4319/lo.2000.45.8.1707 REDFIELD, A. C., B. H. KETCHUM, AND F. A. RECHARDS. 1963. The influence of organisms on the composition of seawater, p. 26–77. In M. N. Hill [ed.], The sea (2). Interscience. REGNIER, P., A. MOUCHET, F. WOLLAST, AND F. RONDAY. 1998. A discussion of methods for estimating residual fluxes in strong tidal estuaries. Cont. Shelf Res. 18: 1543–1571, doi:10.1016/ S0278-4343(98)00071-5 TSUNOGAI, S., M. NISHIMURA, AND S. NAKAYA. 1968. Complexometric titration of calcium in the presence of larger amounts of magnesium. Talanta 15: 385–390, doi:10.1016/00399140(68)80247-4 VAN OOST, K., AND OTHERS. 2007. The impact of agricultural soil erosion on the global carbon cycle. Science 318: 626–629, doi:10.1126/science.1145724 VANDERBORGHT, J. P., R. WOLLAST, M. LOIJENS, AND P. REGNIER. 2002. Application of a transport-reactive model to the estimation of biogas fluxes in the Scheldt estuary. Biogeochem. 59: 207–237, doi:10.1023/A:1015573131561 WANG, H. J., Z. S. YANG, N. S. BI, AND H. D. LI. 2005. Rapid shifts of the river plume pathway off the Huanghe (Yellow) River mouth in response to water–sediment regulation scheme in 2005. Chin. Sci. Bull. 50: 2878–2884, doi:10.1360/982005-1196 ———, ———, Y. SAITO, J. P. LIU, AND X. X. SUN. 2006. Interannual and seasonal variation of the Huanghe (Yellow River) water discharge over the past 50 years: Connections to impacts from ENSO events and dams. Glob. Planet. Change 50: 212–225, doi:10.1016/j.gloplacha.2006.01.005 ———, ———, ———, ———, ———, AND Y. WANG. 2007. Stepwise decreases of the Huanghe (Yellow River) sediment load (1950–2005): Impacts of climate change and human activities. Glob. Planet. Change 57: 331–354, doi:10.1016/ j.gloplacha.2007.01.003 WOLLAST, R. 2003. Biogeochemical processes in estuaries, p. 61–67. In G. Wefer, F. Lamy, and F. Mantoura [eds.], Marine science frontiers for europe. Springer. YANG, D., AND OTHERS. 2004. Analysis of water resources variability in the Yellow River of China during the last half century using historical data. Water Resour. Res. 40: W06502. ZHAI, W. D., M. H. DAI, W. J. CAI, Y. C. WANG, AND Z. H. WANG. 2005. High partial pressure of CO2 and its maintaining mechanism in a subtropical estuary: The Pearl River estuary, China. Mar. Chem. 93: 21–32, doi:10.1016/j.marchem.2004. 07.003 ———, ———, AND X. H. GUO. 2007. Carbonate system and CO2 degassing fluxes in the inner estuary of Changjiang (Yangtze) River, China. Mar. Chem. 107: 342–356, doi:10.1016/ j.marchem.2007.02.011 ZHANG, L. J., L. XUE, M. Q. SONG, AND C. B. JIANG. 2010. Distribution of the surface partial pressure of CO2 in the southern Yellow Sea and its controls. Cont. Shelf Res. 30: 293–304, doi:10.1016/j.csr.2009.11.009 ———, M. XUE, AND Q. Z. LIU. 2012. Distribution and seasonal variation in the partial pressure of CO2 during autumn and winter in Jiaozhou Bay, a region of high urbanization. Mar. Pollut. Bull. 64: 56–65, doi:10.1016/j.marpolbul.2011.10.023 ———, L. WANG, W. J. CAI, D. M. LIU, AND Z. G. YU. 2013. Impact of human activities on organic carbon transport in the Yellow River. Biogeosciences 10: 2513–2524, doi:10.5194/bg10-2513-2013 Associate editor: Markus H. Huettel Received: 15 May 2013 Accepted: 29 October 2013 Amended: 18 November 2013