Air–water fluxes and sources of carbon dioxide in the Delaware Estuary: spatial and seasonal variability

Distributions of surface water partial pressure of carbon dioxide (pCO2) were measured on nine cruises in the Delaware Estuary (USA). The Delaware River was highly supersaturated in pCO2 with respect to the atmosphere during all seasons, while the Delaware Bay was undersaturated in pCO2 during spring and late summer and moderately supersaturated during mid-summer, fall, and winter. While the smaller upper tidal river was a strong CO2 source (27.1± 6.4 mol-C m yr), the much larger bay was a weak source (1.2± 1.4 mol-C m yr), the latter of which had a much greater area than the former. In turn, the Delaware Estuary acted as a relatively weak CO2 source (2.4± 4.8 mol-C m yr), which is in great contrast to many other estuarine systems. Seasonally, pCO2 changes were greatest at low salinities (0≤ S < 5), with pCO2 values in the summer nearly 3-fold greater than those observed in the spring and fall. Undersaturated pCO2 was observed over the widest salinity range (7.5≤ S < 30) during spring. Near to supersaturated pCO2 was generally observed in midto high-salinity waters (20≤ S < 30) except during spring and late summer. Strong seasonal trends in internal estuarine production and consumption of CO2 were observed throughout both the upper tidal river and lower bay. Positive correlations between river-borne and air–water CO2 fluxes in the upper estuary emphasize the significance of river-borne CO2 degassing to overall CO2 fluxes. While river-borne CO2 degassing heavily influenced CO2 dynamics in the upper tidal river, these forces were largely compensated for by internal biological processes within the extensive bay system of the lower estuary.


Introduction
While, globally, the surface area of estuaries is only about 4 % that of continental shelves, recent studies have concluded that the carbon dioxide (CO 2 ) degassing flux from estuarine waters is as large as the CO 2 uptake by the continental shelf (Borges, 2005;Borges et al., 2005;Cai et al., 2006;Chen and Borges, 2009;Cai, 2011).Global estuarine waters are estimated to emit 0.10-0.45Pg C yr −1 , while continental shelves take up 0.20-0.40Pg C yr −1 (Borges, 2005;Borges et al., 2005;Cai, 2011;Chen et al., 2013;Regnier et al., 2013;Laruelle et al., 2015).Such large estuarine CO 2 degassing suggests that much of the terrestrial organic carbon, including that from coastal wetlands, is respired to CO 2 during transport through the estuarine zone, though the relative importance of river-supplied CO 2 and organic carbon versus those from the coastal wetlands is debatable (Borges and Abril, 2011;Cai, 2011).In turn, estuarine waters are a major source of CO 2 to the atmosphere, with partial pressures of CO 2 (pCO 2 ) ranging from 350 to 10 000 µatm and air-water CO 2 fluxes ranging from −5 to 80 mol C m −2 yr −1 (Raymond et al., 1997;Cai and Wang, 1998;Frankignoulle et al., 1998;Borges, 2005;Borges et al., 2006;Borges and Abril, 2011;Cai, 2011).
There is rising concern that global estuarine CO 2 degassing flux may be overestimated (Cai, 2011).Although substantial progress has been achieved over the past decade (Borges and Abril, 2011;Chen et al., 2013;and references therein), our knowledge of CO 2 degassing fluxes and their controlling processes in estuaries remains insufficient.Globally, the majority of past estuarine CO 2 studies have been conducted on small estuarine systems, which typically have  (Chen and Borges, 2009;Cai, 2011;Borges and Abril, 2011).Specifically, in the US east coast, high pCO 2 has been found in estuaries along the southeastern (Cai and Wang, 1998;Jiang et al., 2008a) and northeastern (Salisbury et al., 2008;Hunt et al., 2010) coastal regions.While high pCO 2 has also been found in small estuaries along the US mid-Atlantic coast (Raymond et al., 1997;Raymond et al., 2000), only a few estuarine CO 2 studies have been conducted in this region, such as Crosswell et al. (2012) in the Neuse River, NC; Raymond et al. (1997) in Hudson River, NY; and Raymond et al. (2000) in the York River, VA.Thus, there is limited research on CO 2 dynamics in large estuaries or bay systems with long freshwater residence times in the US mid-Atlantic coast (most notably the Chesapeake and Delaware estuaries).Presumably, these large estuaries have lower pCO 2 than small estuaries or bay systems with rapid freshwater transit times (Borges and Abril, 2011;Cai, 2011).Except for a few recent studies and the pioneering work of Sharp and Culberson, over the past 30 years there have been few inorganic carbon studies in the Delaware Estuary (Culberson, 1988;Sharp et al., 2009).Air-water CO 2 fluxes, total DIC fluxes, and ongoing evaluations of water acidification have not been consistently (via annual and seasonal surveys) studied.Overall, there is a lack of data and a pressing need to synthesize and expand global research to larger estuaries.Furthermore, of past estuarine CO 2 studies, many lack spatial and seasonal coverage of surface water pCO 2 and air-water CO 2 fluxes, making flux estimates highly uncertain.
The Delaware Estuary is divided into an upper 100 km long tidal Delaware River and lower Delaware Bay (Fig. 1; Sharp, 2010).With a relatively simple hydrology, the Delaware Estuary is fairly easy to characterize, and because of this it has served as a model estuary for biogeochemical studies (Cifuentes et al., 1988;Sharp et al., 2009).The tidal freshwater portion of the Delaware River flows from the head of the tide near Trenton, NJ, through the greater Philadelphia area, the sixth largest municipal region of the US, before passing into the saline Delaware Bay (Fig. 1; Sharp et al., 2009;Sharp, 2010).In turn, the upper Delaware River is heavily influenced by major industrial activity and continuously responding to a rapidly changing environment.For example, in the mid-20th century, the urban river of the Delaware Estuary suffered from severe hypoxia, with average summer dissolved oxygen (DO) concentrations near zero (Sharp, 2010).Fortunately, the implementation of the Clean Water Act (CWA) in the early 1970s helped promote efforts to improve water quality conditions in the Delaware River.With major upgrades to large sewage treatment plants, DO concentrations since the early 1990s have consistently been above the CWA standard of 3.5 mg L −1 (∼ 219 µmol L −1 ; Sharp, 2010).Nonetheless, high pCO 2 is still expected to be associated with strong respiratory O 2 consumption in the upper estuary.In contrast, the Delaware Bay is a large shallow embayment surrounded by salt marshes with minimal industrial or municipal inputs (Cifuentes et al., 1988).Thus, the Delaware Estuary is governed by the dynamic interaction between a river-dominated upper estuary and an oceandominated lower bay.This feature, typical for other large estuaries and, depending on river flow and geomorphology, smaller estuarine systems as well, provides us the opportunity to examine how contrasting geographical settings, physical mixing processes, and ecosystem metabolism in an extensive bay system can affect CO 2 gas exchange.
In this paper, we report the first seasonal distribution of pCO 2 and air-water CO 2 flux in the Delaware Estuary, which was surveyed nine times via various day-to week-long surveys from 2013 through 2014.We further assess the temperature and biological effects on pCO 2 distributions as well as the overall contribution of internal versus riverine sources on CO 2 inputs to the estuarine system.Finally, we present a summarized pCO 2 distribution over the study area and provide a conceptual model to illustrate the control mechanisms on surface water CO 2 dynamics in the Delaware Estuary.

Field measurements
The Delaware Estuary was surveyed on nine cruises: 8-10 June 2013, 8-15 August 2013, 17 October 2013, 17-22 November 2013, 23-24 March 2014, 3 July 2014, 27 August-1 September 2014, 30 October-2 November 2014, and 5 December 2014.Distributions of pCO 2 , dissolved inorganic carbon (DIC), total alkalinity (TA), and pH were measured from the mouth of the bay to the near-zero salinity of the estuary in five of the nine cruises.During the August and October 2013 cruises, only surface water pCO 2 was measured.
To monitor levels of pCO 2 , surface water was directly pumped from 1 to 2 m below the sea level through an underway pCO 2 analyzer (AS-P2, Apollo Scitech) installed in the shipboard laboratory (Huang et al., 2015).Surface water flowed into a 1 L volume shower head equilibrator at a minimum rate of 1.7 L min −1 to facilitate rapid gas exchange.A specifically designed water-drain system is attached to the equilibrator to insure that the pressure inside and outside remains balanced (Jiang et al., 2008b).The equilibrated gas was pumped through a water trap (Peltier cooler), which removed most of the water vapor, and then into a drying tube packed with magnesium perchlorate (Mg(ClO 4 ) 2 ) or Nafion tubing.Surface water CO 2 (mole fraction of dry air (xCO 2 )) was measured approximately every 1.5 min using an underway flow-through system equipped with a non-dispersive infrared (NDIR) gas analyzer (Li-Cor, model LI-7000, Lincoln, NE, USA).The LI-7000 was calibrated, every 3-6 h, against three or four CO 2 gas standards (151.5, 395.4, 982.6, and 1969 ppm CO 2 in air) referenced against standards traceable to those of the National Institute of Standards and Technology (NIST).Atmospheric xCO 2 was measured every 3-6 h using the same CO 2 system.In order to avoid contamination from the ship's stack gases or other possible sources of air pollution, the inlet of the atmospheric CO 2 pipe was installed on the highest platform in the front of the ship.An onboard Sea-Bird thermosalinograph (SBE-45) measured surface water temperature and salinity.To calculate surface water and atmospheric pCO 2 values, all xCO 2 measurements were corrected to 100 % saturation of water vapor pressure and the in situ surface water temperature (Dickson et al., 2007).
DIC and TA water samples were collected throughout the salinity gradient.Multiple samples were taken at near-zero salinity and at the mouth of the bay to obtain river and ocean end-member values.Samples for DIC and TA measurements were filtered through a cellulose acetate filter (0.45 µm) into 250 mL borosilicate bottles and then fixed with 100 µL of saturated mercury bichloride solution (Cai and Wang, 1998;Jiang et al., 2008a).When collecting water, all bottles were overflowed for at least twice its volume to minimize contact with the atmosphere.Afterwards, sample bottles were kept at 4 to 10 • C for future analysis.DIC was determined by acidifying 0.5-1.0mL samples with phosphoric acid.The extracted CO 2 gas was subsequently quantified via an infrared gas analyzer (AS-C3 Apollo Scitech).TA was measured by Gran titration (Gran, 1952) using the open-cell method with a semi-automatic titration system (AS-ALK2, Apollo Scitech; Cai et al., 2010a;Huang et al., 2012).Both DIC and TA measurements were calibrated against certified reference material (CRM, provided by A. G. Dickson from Scripps Institution of Oceanography) at a precision level of about ± 2 µmol kg −1 (Huang et al., 2012).

Air-water CO 2 flux estimation
In this study, air-water CO 2 fluxes (F , mmol m −2 d −1 ) at pixel i of a 0.01 longitude × 0.01 latitude grid were calculated as follows: where k i (cm h −1 ) is the gas transfer velocity of CO 2 ; K oi is the solubility coefficient of CO 2 (mol L −1 atm −1 ), which can be calculated from in situ temperature and salinity (Weiss, 1974); and pCO 2(water)i and pCO 2(air)i (µatm) are the partial pressure of CO 2 in the water and the air, respectively.The mean atmospheric [xCO 2 ] during each cruise and the sea surface temperature, salinity, and pressure were used to calculate the pCO 2(air)i .A positive F value indicates CO 2 transfer from water to the atmosphere.Generally, two main issues arise when trying to accurately determine air-water CO 2 fluxes in coastal waters: (1) how to accurately represent surface turbulence and (2) obtaining spatial and temporal heterogeneity of pCO 2 distributions.One of the greatest uncertainties when calculating air-water CO 2 fluxes is estimating gas transfer velocities (Wanninkhof et al., 2009).While gas transfer velocities primarily depend on wind regime in the open ocean, in coastal and shallower estuaries it is probably more complicated as other factors such as tidal currents, bottom stress, wave slope, turbidity, surface films, and fetch limitation can also influence gas exchange rates (Raymond and Cole, 2001;Borges et al., 2004;Zappa et al., 2007;Jiang et al., 2008a;Abril et al., 2009).Unfortunately, because there have not been many studies on gas transfer velocities in estuaries, we relied on wind speed dependence to estimate gas exchange rates.Moreover, limited research has been conducted at wind speeds less than 4 m s −1 .In turn, quadratic relationships that estimate k often extrapolate to zero at low wind speeds (Wanninkhof et al., 2009).Increasing evidence suggests that k does not approach zero at low wind speeds but rather asymptotes to a finite value due to various external factors such as buoyancy effects, chemical enhancements, and physical mixing processes (McGillis et al., 2001;McGillis et al., 2004;Wanninkhof et al., 2009).To avoid gas transfer velocities of zero in river and inland waters where wind speeds are typically low, we adopted the gas transfer relationship as proposed by A. Joesoef et al.: Air-water fluxes and sources of carbon dioxide in the Delaware Estuary Wanninkhof et al. (2009): where k 660 is the gas transfer velocity at the Schmidt number of 660, which can be calculated from in situ sea surface temperature (Wanninkhof, 1992), and U 10 is the wind speed at 10 m above the water surface.Another challenge to accurately determining air-water CO 2 fluxes is obtaining reliable spatial and temporal pCO 2 distributions.Unfortunately, while seasonal distributions of pCO 2 were measured from the mouth of the bay to the near-zero salinity of the estuary (north to south), our lack of cross-bay transects (east to west) limits our knowledge of CO 2 dynamics in shallow water regions of the estuary.Thus, there is a pressing need to conduct more research near these shallow water boundaries.
In addition, because the relationship between k and mean wind speeds is nonlinear, temporal distributions of wind speeds influence gas transfer velocities (Wanninkhof, 1992;Wanninkhof et al., 2002).To accurately determine the effect of variability of winds over a month, Wanninkhof (1992) introduced the nonlinearity coefficient of the wind speeds (C 2 ), which is calculated as follows (Wanninkhof et al., 2002;Jiang et al., 2008b): where C 2 is the nonlinearity coefficient for quadratic terms of gas transfer relationships, U j is the high-frequency wind speed collected at the buoys, U mean is the monthly mean wind speed, and n is the total number of available wind speeds during that month.We used high-frequency wind speed data (measured every 6 min) obtained from four National Oceanic and Atmospheric Administration (NOAA) buoys (LWSD1, CMAN4, SJSN4, and DELD1) to calculate the nonlinearity coefficients at each buoy and extrapolate them to the entire estuary.Using the calculated nonlinearity coefficients, gas transfer relationships were corrected to obtain the most accurate relationship between gas transfer velocities and wind speeds during each month.
In order to calculate area-averaged CO 2 flux throughout the Delaware Estuary, the system was divided into five geographic zones as defined by Sharp et al. (2009).However, due to rapid change in pCO 2 values across the mid-bay, this region was split into an upper and mid-bay zone to allow for a more robust comparison of pCO 2 and CO 2 fluxes throughout the system (Fig. 1).Surface water pCO 2 , temperature, salinity, wind speed, and pressure were interpolated onto a 0.01 × 0.01 grid.Following the same method as presented in Jiang et al. (2008b), flux F i at each pixel was calculated: where S i is the total area surrounding pixel i, Lon and Lat are the longitude and latitude intervals of the grid, respectively, Lat i is the latitude at pixel i, and R is the radius of the earth.The area-averaged CO 2 flux was calculated as follows (Jiang et al., 2008b): Because there is no precise method to account for the uncertainties of air-water CO 2 fluxes, we followed the same approach as described in Jiang et al. (2008b).Atmospheric measurements for each cruise and gas transfer velocities of Wanninkhof et al. (2009) and Wanninkhof (2014) were used to estimate standard deviations of the atmospheric CO 2 and CO 2 flux, respectively.

Temperature-normalized pCO 2 estimation
Temperature changes are important as they influence surface water pCO 2 by governing the thermodynamic equilibrium of the inorganic carbon system (Takahashi et al., 1993).If only controlled by temperature change and no other physical (mixing) or biogeochemical changes, pCO 2 in surface seawater would double for every 16 Takahashi et al., 1993).The temperature constant above determined by Takahashi et al. (1993) works well for open-ocean waters with salinities between 34 and 36 as physical mixing with freshwater is generally minor.After temperature normalization, one may attribute the remaining pCO 2 change to non-thermal processes (mostly biological activity but possibly also mixing processes).However, in coastal oceans mixing is often serious and influences the interpretations of observed temperature dependences.For example, Jiang et al. (2008a) found that values of (∂ ln pCO 2 /∂T ) in river-and marine-dominated estuaries were less (about 0.027-0.042• C −1 ) than that determined by Takahashi et al. (1993).We suggest that a thermodynamic prediction for estuarine water should be used for such comparisons (Bai et al., 2015).We first derived temperature constants for a general estuarine system using the Excel macro CO2SYS (Pierrot, 2006) and inorganic carbon dissociation constants from Millero et al. (2006) for estuarine waters (S < 30) and from Mehrbach et al. (1973) refit by Dickson and Millero (1987) for high-salinity waters (S >30).
Based on data collected over the past two years, river and ocean end-members of TA (900 and 2300 µmol kg −1 , respectively) and of DIC (960 and 2000 µmol kg −1 , respectively) were used.Calculated pCO 2 varied among different temperatures, from 5 to 30 • C, with the largest difference in low salinities (0 to 5; Fig. 2).In turn, when binning salinities to intervals of 5 units, the greatest variability in temperature constants was observed in salinities 0-5 and 5-10 (Table 1).Averaged values of (∂ ln pCO 2 /∂T ) for salinity intervals between 0 and 35 ranged from 0.0332 to 0.0420 • C −1 (Table 1).Similar to the results found in Jiang et al. (2008a), temperature-derived constants were lower than the isochemical seawater constant 0.0423 • C −1 determined by Takahashi  (1993).Thus, knowing the extensively complex nature of estuarine systems, it is important to note that derived variances in temperature-normalized pCO 2 provide only a relatively simple analysis of seasonal pCO 2 fluctuations due to temperature and biological processes as derived variances in temperature-normalized pCO2 neglect the impact that various physical processes, turbulent forces, and tidal mixing scenarios have on pCO 2 dynamics.Using a similar approach to that in Takahashi et al. (2002), we also attempted to separate the temperature effect from other non-thermal effects on seasonal pCO 2 change.We first normalized the pCO 2 at in situ temperature to the 10-year (2004-2014) annual mean temperature of 13.3 • C via the following (Takahashi et al., 2002): where T is temperature ( • C), C s is the averaged (∂ ln pCO 2 /∂T ) value for the salinity interval, and subscripts "mean" and "obs" indicate the annual mean and observed values.Through this approach, we attributed any differences between calculated and observed pCO 2 values to be the result of biological activity and/or physical mixing processes (non-thermal).Because salinity gradients down the estuary vary greatly depending on the season, river discharge, tidal cycle, precipitation, and other circulation processes, salinitybinned climatologies can provide crucial insight into and a different perspective of the various physical and biological controls behind observed pCO 2 distributions that geographic boundaries may not.In turn, pCO 2 values from each survey were constructed into salinity-binned climatologies (intervals of 5 salinity units from 0 to 30) to better isolate and interpret the thermal versus non-thermal effects on seasonal pCO 2 fluctuations.Observed pCO 2 values during months with no surveys were estimated by linearly regressing data from adjacent months with sample measurements.In contrast, to best analyze the effect of temperature changes on observed pCO 2 values, annual mean pCO 2 values across each salinity interval were used in conjunction with the mean and observed temperatures via the following equation (Takahashi Using this method, we attributed any differences between calculated mean versus observed pCO 2 values as a result of seasonal temperature changes.To remove the temperature effect from observed in situ pCO 2 , the observed pCO 2 values were normalized to a constant temperature of 13.3 • C, which was the 10-year annual mean water temperature measured in the Delaware Estuary from 2004 to 2014.

Estuarine and river CO 2 contributions
Due to various CO 2 sources such as the degradation of organic matter, discharge of sewage effluents, soil-induced respiration, freshwater runoff, and addition of humic substances, river water flowing into estuarine systems is typically supersaturated in CO 2 with respect to the atmosphere (Raymond et al., 2000;Abril and Borges, 2004;Borges et al., 2006).To investigate the influence of river-borne CO 2 input to overall air-water CO 2 fluxes, we used similar methods to those performed in Jiang et al. (2008a).In situ DIC and TA measurements were coupled using the Excel macro CO2SYS (Pierrot, 2006) and inorganic carbon dissociation constants from Millero et al. (2006) for estuarine waters to calculate dissolved CO 2 concentrations.We first estimated the contribution of the ocean end-member to the estuarine DIC alone as follows (Jiang et al., 2008a): where DIC mixing w/o is the DIC concentration after the ocean end-member is diluted by fresh water with zero DIC and S i and S ocean are in situ and ocean end-member salinities, respectively (Fig. 3a).When DIC inputs from both the river and the ocean end-members were considered, estuarine DIC was estimated using a two-end-member mixing model as fol-lows (Jiang et al., 2008a): where DIC mixing w/R is the DIC concentration after mixing of river and ocean end-members and DIC river is the river endmember (Fig. 3a).With much of the DIC pool dominated by carbonate and bicarbonate ions, Sharp et al. (2009) observed small seasonal influences on DIC concentrations due to temperature affects and biological activity.They suggest that the majority of variability in DIC in the upper tidal river of the Delaware Estuary is due to the combined interaction of varying precipitation rates and prior meteorological conditions.This is expected as river DIC and TA are largely a dilution of weathering production by rain (Cai et al., 2008).However, at higher salinities, any drawdown of DIC relative to salinity is small since less than 1 % of the DIC pool exists as pCO 2 (Sharp et al., 2009).Thus, while total DIC concentrations illustrate some fluctuations in biological activity (which occurred mostly at the highly productive mid-bay), it is an integrated measurement of freshwater and seawater mixing (Sharp et al., 2009).TA mixing w/o and TA mixing w/R were also estimated using similar equations by replacing DIC with TA (Fig. 3b).Because CO 2 concentrations do not change linearly during mixing, they were estimated using corresponding DIC and TA mixing values (Fig. 3c; Jiang et al., 2008a).Moreover, since CO 2 concentrations fluctuate with temperature change, the 10-year (2004-2014) annual mean temperature of 13.3 • C was used in this work.Thus, the CO 2 contribution due to river input ( [CO 2 ] riv ) was estimated as follows: Calculated river CO 2 inputs ([CO 2 ] riv ) and combined river discharges from the Schuylkill and Delaware rivers for each month were used to compute river-borne CO 2 fluxes in the upper tidal river.
To further investigate the influence of CO 2 inputs from the river (external) versus production from within the estuary (internal), we used a similar but modified method as performed in Jiang et al. (2008a).The CO 2 contribution from within the estuarine zone ([CO 2 ] est ) was estimated as follows: where [CO 2 ] i is the in situ CO 2 concentration, τ i is the flushing time, and F i is the air-water CO 2 flux.Specifically, [CO 2 ] i was calculated using in situ DIC and TA concentrations and τ i was estimated using river discharge rates and volume of each region (Table 3; Sheldon and Merryl, 2002).Surveys that did not contain sufficient river end-member DIC and TA measurements were excluded.Alternatively, Eq. ( 11) suggests that integrated CO 2 degassing (τ i × F i ) is supported by the deficit or excess CO 2 concentration ([CO 2 ] mixing w/R -[CO 2 ] i ) plus the internal estuarine CO 2 production or consumption ([CO 2 ] est ) exhibited across each region.3 Results

Hydrographic conditions
Measured surface water temperatures and river discharge during each cruise were compared with the 10-year (2004-2014) and 30-year (1980-2014)   vey (Fig. 5a-i).Salinity < 1.0 was reached on six of the nine cruises (Fig. 5a, b, c, e, g, and h).The July 2014, August 2013, and October 2013 cruises only transected as far north as the Chesapeake and Delaware Canal (about 39.55 • N; Fig. 1).Salinity < 1.0 (a minimum of 0.98) was only observed during the July 2014 excursion, which had the highest river discharge of the three partial surveys (Fig. 5c).Generally, high-salinity waters (25-32.5)were observed in the lower bay and salinities around 20 to 25 in the mid-bay.The upper bay had a much broader scale ranging from salinities of 10 to 20 and during the high-flow months of March 2014 and June 2013 salinities < 10 were observed (Fig. 5a and b).Salinities did not reach less than 0.25 in the turbidity maximum zone.Salinity distributions in the urban river were limited due to the lack of surveys conducted in this region.

Surface water pCO 2
Generally, surface water pCO 2 in the Delaware Estuary increased from the ocean to the river end-member, with pCO 2 values ranging from about 150 to over 4000 µatm (Fig. 6a-i).Moreover, pCO 2 exhibited strong seasonal variations across both river and bay portions.The most pronounced shifts in surface water pCO 2 were observed within the lower urban river and turbidity maximum river zones of the Delaware River, with pCO 2 being lowest in the cool months (March, October, and November) and highest in the warm months (June, July, and August; Table 2).During all months, the turbidity maximum zone was supersaturated in CO 2 with respect to the atmosphere (atmospheric pCO 2 : 375-398 µatm) except during March 2014 (Fig. 6a).Throughout the summer and early fall (June, July, and August), pCO 2 ranged from about 650 to over 4000 µatm across the turbidity and lower urban river zones (Fig. 6b-e).In late fall (October and November), pCO 2 dropped to as low as 500 µatm in the turbidity maximum zone and reached 1400 µatm within the lower urban river zone (Fig. 6f-h).However, the decrease in pCO 2 values was not always observed as temperatures cooled.During the winter (December), surface water pCO 2 values increased across the turbidity maximum zone, ranging from about 650 to 1000 µatm (Fig. 6i).As discussed later, this shift in pCO 2 during winter is likely a result of opposing timing of seasonal temperature cycles and respiration versus that of river discharge rates.Surface water pCO 2 exhibited strong seasonal variations in the Delaware Bay as well (Fig. 6a-i).In March 2014, most likely due to a strong biological bloom and low temperature (Fig. 4a), the entire bay system (upper, middle, and lower) was undersaturated in CO 2 with respect to the atmosphere (Table 2).In particular, pCO 2 reached as low as 160 µatm in the mid-bay (Fig. 6a).During the warmer summer months (June, July, and August), pCO 2 in the bay remained around 400 to 500 µatm, with occasional undersaturation occurring in the mid-bay region (Fig. 6b-e).In August 2014, low pCO 2 ranging from about 200 to 350 µatm was observed throughout much of the mid-and lower bay regions (Fig. 6e).In contrast, during the late fall, pCO 2 values were fairly homogenous throughout the mid-and lower bay (400-450 µatm in October 2013 and 2014 and 375-415 µatm in November 2013) and slightly higher pCO 2 occurring in the upper bay (Fig. 6f-h).In December 2014, pCO 2 increased throughout all regions of the bay, with pCO 2 values ranging from 500 to 650 µatm (Fig. 6i).While it remains unclear what supports these elevated pCO2 values, stratification of subsurface waters in late fall followed by strong winter mixing during winter (December 2014) and a 2-fold increase in river discharge could explain the elevated pCO 2 values observed throughout the mid-and the lower bay systems (Fig. 4b).

Air-water CO 2 fluxes
The urban river and turbidity maximum zone served as strong sources of CO 2 to the atmosphere and were positive during all months (Table 2).Across the upper to lower bay portions of the estuary, uptake of CO 2 from the atmosphere was greatest during spring (March), ranging from F CO 2 = −12.1 to −20.0 mmol m −2 d −1 (Table 2).The CO 2 uptake flux was highest in March 2014 in the mid-bay (−20.0 mmol m −2 d −1 ), while the highest CO 2 degassing flux occurred in June 2014 in the urban river (144.8 mmol m −2 d −1 ; Table 2).Air-water CO 2 fluxes in the upper to lower bay regions decreased in early winter (December) to a minimum in early spring (March), followed by an increase to an annual maximum in early summer (June).In the turbidity maximum zone and urban river, area-averaged CO 2 fluxes followed the same seasonal decrease in spring and increase in summer but reached an annual minimum in late fall instead of early spring.In winter (December), the mid-and lower bays, which were typically sinks or weak sources of CO 2 , exhibited relatively strong CO 2 fluxes to the atmosphere.

CO 2 distribution across the salinity gradient
To further investigate pCO 2 variations along the Delaware Estuary, we examined distributions of pCO 2 across the salinity gradient.Due to limited area and salinity coverage, surveys conducted in August and October 2013 were excluded for this assessment.In all months, pCO 2 versus salinity followed a concave upward trend towards the river end-member (Fig. 7).The seasonal variation between pCO 2 values was largest at low salinities around 0 to 5, with pCO 2 values in the summer (June, July, and August) nearly 2-fold greater than those observed in the spring (March) and fall (October and November) seasons (Fig. 7).In all seasons, pCO 2 was supersaturated with respect to the atmosphere from salinities 0 to 5. In spring, undersaturated pCO 2 was observed over the widest salinity range from 7.5 to 30.In summer, undersaturated pCO 2 was generally not observed except at moderate salinities around 17 to 28 in August.In fall, pCO 2 values were near atmospheric concentrations around mid-salinity waters and were only undersaturated at salinities greater than 25.In winter (December), pCO 2 values were always supersaturated with respect to the atmosphere across the entire salinity range.Seasonally, the Delaware Estuary served as a strong CO 2 sink (−5.0 ± 6.0 mol-C m −2 yr −1 ) in the spring, a strong source (4.9 ± 8.1 mol-C m −2 yr −1 ) in the summer, a weak source (1.0 ± 2.4 mol-C m −2 yr −1 ) in the fall, and a strong source (5.7 ± 1.9 mol-C m −2 yr −1 ) in the winter.While low-salinity waters were strong CO 2 sources, proportionally these upper regions (0 ≤ S < 10) were small in comparison to the total estuarine study area.In turn, their areaaveraged contribution (27.1 ± 6.4 mol-C m −2 yr −1 ) to overall regional flux (2.4 ± 4.8 mol-C m −2 yr −1 ) is minor.Thus, the Delaware Estuary as a whole acts as a relatively weak CO 2 source (2.4 ± 4.8 mol-C m −2 yr −1 ), which is in great

Seasonal variations in temperature-normalized pCO 2
Seasonal distributions of pCO 2obs at 13.3 • C, which indicate impacts of non-thermal processes (biological and mixing), varied noticeably throughout the year and across salinity intervals (Fig. 8).Typically, pCO 2obs at 13.3 • C was greatest during the early and mid-winter season (December and January) except in the 0-5 salinity interval (mostly turbidity maximum zone and urban river) when pCO 2obs at 13.3 • C reached its maximum in June.Coupled with decreasing flow, in the 0-5 salinity interval, pCO 2obs at 13.3 • C decreased from June to an annual minimum in October.In the midsalinity waters (5 ≤ S ≤ 20), pCO 2obs at 13.3 • C decreased from mid-winter to an annual minimum in March, followed by an increase to a secondary maximum in June.In contrast, in the high-salinity waters (20 ≤ S ≤ 30) of the lower bay, where biological removal of CO 2 was generally strong, annual minimums were observed in August.The annual distribution of pCO 2mean at T obs , which indicates the impact of the seasonal thermal cycle, followed typical bell-shaped curves across all salinity intervals, with the lowest values occurring in winter and an annual maximum occurring in July.

Discussion
The seasonal and spatial distributions of estuarine pCO 2 are governed by the dynamic interaction between water temperature; horizontal and vertical mixing processes; biological processes; and CO 2 contributions from the river, ocean, and estuarine zone (Jiang et al., 2008a;Borges and Abril, 2011;Hunt et al., 2014).In the estuarine zone, the addition or removal of CO 2 includes net ecosystem metabolism, DIC exchange with intertidal marshes, sediments, groundwater inputs, air-water gas exchanges, and other estuarine contributing processes (Jiang et al., 2008a

Temperature vs. biological effects on pCO 2
Similar to other estuaries (Borges and Abril, 2011), seasonal temperature changes provided a first control on the observed seasonal changes in pCO 2obs (low in the winter and high in the summer, Figs.4a and 7).This is further reflected in the fact that temperature-normalized pCO 2 was always higher than in situ pCO 2 in the winter but lower than in situ pCO 2 in the summer (Fig. 8).Presumably, then, seasonal patterns of the temperature-normalized pCO 2 reflect how non-thermal processes (mixing and biological) influence in situ pCO 2 .For example, in the urban river and turbidity maximum zones (S < 5), high pCO 2obs at 13.3 • C in the spring and winter may reflect both river inputs and strong respiratory CO 2 production.Low pCO 2obs at 13.3 • C during the warmer months likely reflects the removal of CO 2 due to various non-thermal processes.During the warmer months from May to October, Yoshiyama and Sharp (2006) found elevated nitrite (NO 2 ) concentrations in the urban river when nitrification and primary production were highest.In addition, high NO 2 concentrations were observed in the mid-bay in summer, when primary production was maximal (Pennock and Sharp, 1994).
Comparably, pCO 2mean at T obs (changes due to the seasonal thermal cycle) trends were opposite to the trend of pCO 2obs at 13.3 • C with lower than pCO 2obs values in the winter and higher than pCO 2obs values in the summer.These opposing signals suggest that increases in surface water pCO 2 due to winter-to-summer warming are partially compensated for by the reduction of surface water pCO 2 due to mixing processes and/or biological removal of CO 2 (Takahashi et al., 2002).Sharp et al. (2009) found that, during the March-April period, ammonium (NH 4 ), phosphate (PO 4 ), and silicate (Si) concentrations were heavily depleted in the mid-and lower bay regions due to extensive spring blooms.Similarly, but in the opposite direction, the reduction in surface water pCO 2 due to fall-to-winter cooling is partially compensated for by the elevation of surface water pCO 2 caused by various nonthermal processes (Fig. 8).
We further examine the relative importance of the temperature and biological effects in each salinity interval by calculating the ratio of pCO 2temp to pCO 2bio (T /B).Using similar methods as performed in Takahashi et al. (2002), we calculate the thermal effects on surface water pCO 2 in each salinity interval as follows: where (pCO 2mean at T obs ) max and (pCO 2mean at T obs ) min are the maximum and minimum pCO 2mean at T obs values, respectively.In other words, the thermal effects on the mean annual pCO 2 value are represented by the seasonal amplitude of (pCO 2mean at T obs ) values computed using Eq. ( 7).
Likewise, the non-thermal effects (biological and mixing processes) on surface water pCO 2 were calculated as follows (Takahashi et al., 2002): where (pCO 2obs at 13.3 • C) max and (pCO 2obs at 13.3 • C) min are the maximum and minimum pCO 2obs at 13.3 • C values, respectively.Thus, the non-thermal effects on surface water pCO 2 (pCO 2obs at 13.3 • C) are represented by the seasonal amplitude of pCO 2 values corrected to the 10-year (2004-2014) annual mean temperature using Eq. ( 6).The relative importance of these effects in each salinity interval can be expressed as the difference between pCO 2thermal and pCO 2non-thermal (T − B) or the ratio of pCO 2thermal to pCO 2non-thermal (T /B).In estuarine regions where thermal effects on surface water pCO 2 exceed non-thermal effects, the (T /B) ratio is greater than 1 or (T − B) is positive, whereas in areas where non-thermal effects dominate, the (T /B) ratio is less than 1 or (T − B) is negative.Based on our results, temperature was a dominant factor in controlling surface water pCO 2 in low-salinity waters (0 ≤ S ≤ 10; mainly the urban river and turbidity maximum zone), with T /B ratios ranging from 1.30 to 1.68 (Table 4).As salinity increased, both pCO 2thermal and pCO 2non-thermal decreased (Table 4).The decrease in pCO 2thermal may be attributed to the reduction in river water temperatures at the ocean end-member (Hunt et al., 2014).In comparison to the upper tidal river, low T /B ratios ranging from 0.69 to 0.80 were observed in mid-salinity waters (15 ≤ S ≤ 25; mainly the mid-and lower bay), suggesting that pCO 2 distributions in the Delaware Bay are largely governed by biological and/or mixing processes.

Influence of river-borne CO 2 on estuarine degassing
The potential emission of river-borne CO 2 was estimated based on the concept of excess CO 2 , the difference between the in situ DIC at zero salinity and a theoretical DIC value at atmospheric equilibrium ( DIC; Abril et al., 2000;Borges et al., 2006).The theoretical DIC was computed using in situ TA values and an atmospheric pCO 2 of 395 µatm.Riverborne CO 2 fluxes were calculated as the product of DIC and the combined river discharges from the Schuylkill and Delaware rivers for each month divided by the estuarine surface area.Generally, as freshwater residence time increases (river discharge decreases), river-borne CO 2 fluxes decrease (Borges et al., 2006).As more river-borne CO 2 is released into the atmosphere in the upper estuary due to increased residence time, leaving less river-borne CO 2 for degassing in the lower estuary, the overall contributions of CO 2 emissions are largely shaped by the net community production in the mixed layer (ML NCP) in the mid-to high-salinity estuarine zones (Abril et al., 2000;Borges et al., 2006).In comparison, as freshwater residence time decreases (river discharge increases), DIC enrichment from ML NCP is reduced and river-borne CO 2 fluxes increase.In certain cases, such as the Rhine Estuary or other systems with extremely rapid flushing times, residence time is so short that not all of the riverborne CO 2 is ventilated to the atmosphere in the estuarine zone (Borges and Frankignoulle, 2002;Borges et al., 2006).
In turn, the potential emission of river-borne CO 2 is higher than the actual observed air-water CO 2 fluxes from the estuary (Borges et al., 2006).
Positive correlations between river-borne and air-water CO 2 fluxes illustrate the importance of river inputs to CO 2 degassing fluxes (Fig. 9).In the Delaware Estuary, the largest river-borne CO 2 flux was observed during the highest flow month of June 2013, with river CO 2 flux accounting for 119 and 60 % of the overall CO 2 degassing flux in the urban river and turbidity maximum zone, respectively (Fig. 9).Moreover, during the high-flow month of March 2014, riverborne CO 2 fluxes exceeded 200 and 150 % of the overall CO 2 degassing fluxes in the urban river and turbidity maximum zone, respectively (Fig. 9).Presumably, the higher riverborne CO2 fluxes than overall CO 2 fluxes in March are due to the combined influence of increased river discharge coupled with large CO 2 consumption in the estuary (Figs.4b and 5a).This is consistent with the observed low pCO 2 and high O 2 values (Fig. 6a; Cai, unpublished data).In contrast, in July and August 2014, air-water CO 2 fluxes exceeded riverborne CO 2 fluxes, indicating strong estuarine CO 2 production.Such internal estuarine CO 2 production is most likely due to respiration in the water column, but may also include other inputs such as benthic respiration and net respiration from surrounding intertidal marshes.In turn, while correlations between river-borne and air-water CO 2 fluxes were exhibited, differences between the two fluxes suggest that the input of CO 2 from other estuarine sources is important.

Internal estuarine production versus river CO 2 input
Our results illustrate that both the river and the estuarine zone contribute to CO 2 inputs in the Delaware Estuary (Fig. 10).Combined river CO 2 input and internal estuarine production was highest in the urban river seasonal trends, with CO 2 contributions being lowest in the spring (March), highest in the summer (June and August), and medium in the fall (October and November).Strong seasonal trends in internal estuarine production were also observed in the bay regions.During spring and late summer (March and August 2014), internal estuarine CO 2 signals were negative in the mid-and lower bay zones and reached as much as 8-fold greater than total river CO 2 inputs, ranging from −22.9 to −100.4 µmol L −1 (Fig. 10).Thus, the majority of river CO 2 input was heavily compensated for by the biological removal of CO 2 in the bay waters.In addition, during the spring season (March), high CO 2 consumption was also observed in the upper bay, with internal estuarine CO 2 signals (−30.7 µmol L − 1) exceeding total river CO 2 contribution (25.7 µmol L −1 ; Fig. 10).Depending on river discharge rates, the freshwater residence time in the Delaware Estuary ranges from about 40 to 90 days (Ketchum, 1952).Due to smaller physical sizes, freshwater residence time in the upper tidal river is much shorter (Table 3).Thus, the percentage of river-borne CO 2 in the upper Delaware Estuary is large (Fig. 10), and that percentage decreases in the midand lower bays, which have longer residence times and high biological CO 2 removal (Sharp, 1983).

Assumptions and limitations
While this study serves as the first air-water CO 2 flux product in the Delaware Estuary, there are several limitations.First, the lack of cross-bay transects (east to west), except in December 2014, limits our knowledge of surface water pCO 2 distributions in shallow water regions of the bay system.Due to various biological and physical processes (i.e., influence from nearby tidal marshes, tributaries, or estuarine circulation forces), surface water pCO 2 may vary from within the main channel to the perimeters of the estuary.Jiang et al. (2008a) found that surface water pCO 2 and airwater CO 2 flux in the marine-dominated Sapelo and Doboy sounds paralleled seasonal temperature changes and net CO 2 inputs from within the estuarine zone.Due to intense productivity of vegetation in the surrounding salt marshes, extensive accumulation of organic carbon occurs during spring and early summer (Dai and Wiegert, 1996;Jiang et al., 2008a).
During late summer and early fall, increased surface water temperatures coupled with tidal flushing of intertidal marsh waters and the decomposition of dead plants contribute to high CO 2 degassing in these estuaries (Dai and Wiegert, 1996;Cai and Wang, 1998;Cai et al., 1999;Neubauer and Anderson, 2003;Wang and Cai, 2004).However, due to the much broader geographic size of the Delaware Bay compared to the marine-dominated Sapelo and Doboy sounds, in-water biological processes are most likely important.In turn, the impact from the growth and decay of marsh plants on surface water pCO 2 and CO 2 flux dynamics may not be as influential in the Delaware Bay, except near the shorelines, where tides regularly flush marsh boundaries.Culberson et al. (1987) and Lebo et al. (1990) performed several cross-bay transects sampled at various depths, over diel cycles, within tributaries, and periodically offshore.Results showed that cross-bay gradients were inconsistent and relatively small, except in shallow waters near the shoreline, where total suspended sediment and chlorophyll concentrations were frequently elevated (Culberson et al., 1987;Lebo et al., 1990;Sharp et al., 2009).Thus, the impact from marsh input of DIC to the Delaware Bay on overall pCO 2 distributions and associated CO 2 degassing fluxes are most likely small.During December 2014, pCO 2 measurements were collected not only in the main channel but also near the Delaware and New Jersey perimeters of the bay (Fig. 6i).While slight variability was observed across the bay, pCO 2 values from the lower to upper bay regions remained within about 150 µatm (Fig. 6i and Table 2).
In addition to the lack of cross-bay transects, there is a pressing need to conduct more winter and early spring surveys to fully cover seasonal ranges in key properties such as temperature and river discharge rates.Moreover, cruises or moored sensor studies at or around large discharge events are needed.A recent study by Voynova and Sharp (2012) found that in the past century there have been a recorded 54 extreme discharges (defined by the average daily discharge as recorded in Trenton, NJ, from 1 October 1912 to 30 September 2011 plus 10 standard deviations), with 46 % of these occurring in the past decade (Voynova and Sharp, 2012).With increasing evidence suggesting that extreme weather events will occur more frequently with climate change, it is important to maintain routine seasonal surveys to learn how such subsequent conditions (i.e., increased summer stratification, riverine CO 2 fluxes, removal of oxygen in bottom waters) impact various coastal environments (Allan and Soden, 2008;Voynova and Sharp, 2012).Furthermore, more research is needed in the urban and upper river sections of the estuary to better understand CO 2 dynamics throughout the whole estuarine gradient.The lack of inorganic carbon data in these upper regions limits syntheses of regional CO 2 fluxes and generalizations to underlying mechanisms.Routine sampling along small tributaries and river systems could provide crucial insight into the biogeochemistry in the upper tidal river.
There are also several limitations to the temperaturenormalized and end-member mixing models that need to be addressed.First, knowing the extensively complex nature of estuarine systems, it is important to note that derived vari-ances in temperature-normalized pCO 2 provide only a relatively simple analysis of seasonal pCO 2 fluctuations due to thermal and non-thermal processes as it neglects the impact that various physical processes, turbulent forces, and tidal mixing scenarios have on pCO 2 dynamics.However, as mentioned before, since salinity fluctuates greatly depending on factors such as season, river discharge, and tidal cycle, salinity-binned climatologies can provide crucial insight into various physical and biological controlling mechanisms behind pCO 2 distributions that geographic boundaries may not.Unfortunately, due to the lack of winter surveys and unusually high pCO 2 values in December, interpolated temperature-normalized pCO 2 during cooler months may be biased and slightly overestimated.Moreover, the temperature-derived constants (∂ ln pCO 2 /∂T ) derived in this study were based on river and ocean end-member TA and DIC concentrations collected in the Delaware Estuary over the past 2 years.Thus, it is important to note that derived temperature constants here are applicable for general estuarine systems and may not be suitable for coastal environments with different hydrological and/or geochemical characteristics.
In situ DIC and TA measurements were coupled using the Excel macro CO2SYS (Pierrot, 2006) and inorganic carbon dissociation constants from Millero et al. (2006) for estuarine waters to calculate dissolved CO 2 concentrations.While river and ocean end-members were obtained at nearzero salinity and at the mouth of the bay, respectively, no fixed end-member sampling locations were established.This marginal difference in end-member location could slightly increase or decrease estimated CO 2 concentrations.In the chemical model of the CO2SYS, NH 3 , NH + 4 , and organic matter contribution to TA were not included (Cai et al., 1998;Cai et al., 2010b), which were likely high in low-salinity waters.Thus, lower calculated than observed CO 2 was expected as the observed TA included other acid-base components (Fig. 3c).However, due to the very high pCO 2 , such uncertainty is deemed unimportant in our consideration.Another factor that may contribute to the lower calculated than observed CO 2 could be the use of mercuric chloride as a preservative in low-salinity samples (S < 10; Trabalka and Reichle, 2013).Excess alkalinity generated via the dilution of mercuric chloride could contribute to conservative CO 2 flux estimates (Trabalka and Reichle, 2013), although, due to the relatively high TA in the Delaware River, we believe this effect is small.

Summary and concluding remarks
While the urban river and turbidity maximum zone are strong CO 2 sources to the atmosphere, these upper regions are small in comparison to the bay regions of the Delaware Estuary.Thus, overall the Delaware Estuary acts as a relatively weak CO 2 source (2.4 ± 4.8 mol-C m −2 yr −1 ) in comparison to many other estuarine systems that serve as strong CO 2 sources to the atmosphere (26 ± 21 mol-C m −2 yr −1 ; Borges and Abril, 2011).Of the 62 estuaries compiled in Borges and Abril (2011), only the Aby Lagoon, a permanently stratified system, served as a sink for atmospheric CO 2 .Seasonal temperature cycles influence the rise and fall of surface water pCO 2 throughout the Delaware Estuary, but these effects are partially compensated for by opposing cycles of biological removal and addition of CO 2 .Moreover, positive correlations between river-borne degassing to overall CO 2 fluxes in the upper subsections of the estuary (the urban river and turbidity maximum zone) illustrate the importance of river-borne CO2 to overall CO2 degassing fluxes.Such features are typical for rapidly flushing river-dominated estuaries.While river-borne CO 2 degassing fluxes heavily impact CO 2 dynamics throughout the upper Delaware Estuary, these forces are largely compensated for by internal biological processes within the extensive bay system of the lower estuary.
Along the eastern Georgia (USA) coast, Jiang et al. (2008a) identified the Altamaha Sound as a riverdominated estuary with CO 2 fluxes driven by river discharge.Comparably, the Kennebec Estuary, located on the central Maine (USA) coast, exhibited high river CO 2 inputs and short freshwater residence times (∼ 4 days), suggesting that CO 2 sources in the estuary were mainly controlled by the degassing of river-borne DIC (Hunt et al., 2014).The upper Delaware Estuary showed similar results, with high river CO 2 contributions and rapid freshwater transit times during all months (Fig. 10 and Table 3).In contrast, in systems with long freshwater residence times (i.e., the Delaware Bay and Scheldt Estuary), much, if not all, of the river-borne CO 2 is released into the atmosphere (Abril et al., 2000;Borges et al., 2006).In turn, overall CO 2 emission from the estuary is largely controlled by net community production in the mixed layer (ML NCP; Borges et al., 2006).In the case of the Euro-pean Scheldt Estuary, long freshwater residence time (30-90 days) leads to extensive DIC enrichment in the water column and high CO 2 emissions to the atmosphere (Abril et al., 2000;Borges et al., 2006).Similarly, and in contrast to the rapidly flushing Altamaha Sound, Jiang et al. (2008a) identified the marsh-surrounded Sapelo Sound as a marine-dominated estuary with CO 2 fluxes driven by seasonal temperature and metabolic cycles.
With its extensive geographic size, the Delaware Estuary features both a river-dominated upper estuary and an oceandominated lower bay.In this case, air-water CO 2 fluxes in the heterotrophic upper estuary are significantly influenced by intense river-borne CO 2 degassing akin to the riverdominated Altamaha Sound and Kennebec Estuary.The autotrophic lower estuary is governed by water-column biological processes and seasonal temperature cycles akin to the marine-dominated Sapelo Sound and Scheldt Estuary (though the Delaware Estuary and other large estuarine systems are orders of magnitude more productive than smaller marine-dominated estuaries).
The continuation of research cruises on estuarine and coastal margins can provide crucial insight into the physical and biological changes in the past, present, and future ocean systems.With such extensive surveys, collection of carbonate parameters, and comparison of carbonate parameters over time, we can significantly broaden our understanding of the processes that govern these coastal zones.In turn, such knowledge can be used to help predict and hopefully regulate the rise of current and future threats to our coastal ocean systems.

A.
Joesoef et al.: Air-water fluxes and sources of carbon dioxide in the Delaware Estuary high pCO 2 .

Figure 1 .
Figure 1.Map of the Delaware Estuary divided into six zones from the head of the tide in Trenton, NJ to the mouth of the bay as defined in Sharp et al. (2009).The gray diamonds indicate the position of four NOAA buoys (LWSD1, CMAN4, SJSN4, and DELD1).The white star shows the location of the USGS gauging station (#01463500).

Figure 2 .
Figure 2. Simulated surface water pCO 2 against salinity grouped by temperature bins.Surface water pCO 2 values were calculated using river and ocean end-member TA and DIC values of 900 and 960 µmol kg −1 and 2300 and 2000 µmol kg −1 , respectively.

Figure 3 .
Figure 3. Concentrations of (a) DIC, (b) TA, and (c) dissolved CO 2 in the Delaware Estuary during March 2014.Open circles represent in situ concentrations.Solid lines represent values after the ocean end-member is diluted by freshwater with a concentration of zero units.Dotted lines represent concentration after mixing of river and ocean end-members.CO2SYS was used to calculate pCO 2 from measured DIC and TA.

Figure 4 .
Figure 4. (a) Surface water temperatures and (b) Delaware River discharge rates recorded in the Delaware Estuary during each sampling month.Error bars represent standard deviations of the 10-year (2004-2014) and 30-year (1980-2014) monthly averages for surface water temperatures and Delaware River discharge rates, respectively.

Figure 5 .
Figure 5. Spatial distributions of surface water salinity in the Delaware Estuary measured during each sampling month.The map was designed with the ODV software by R. Schlitzer (Ocean Data View software, 2015, http://odv.awi.de/).

Figure 6 .
Figure 6.Spatial distributions of surface water pCO 2 in the Delaware Estuary measured during each sampling month.Black and red arrows show surface water pCO 2 values at the Chesapeake and Delaware Canal and the northern end-member of each survey, respectively.The map was designed with the ODV software by R. Schlitzer (Ocean Data View software, 2015, http://odv.awi.de/).

Figure 7 .
Figure 7. Measured surface water pCO 2 against the salinity gradient during each sampling month in the Delaware Estuary.

Figure 8 .
Figure 8. Salinity-binned intervals of temperature-normalized observed pCO 2 values at 13.3 • C, annual mean, area-averaged pCO 2 values at in situ temperature, and observed pCO 2 values in the Delaware Estuary over the year.Red arrows indicate increases in pCO 2 and blue arrows indicate decreases in pCO 2 .The symbol T represents changes in pCO 2 due to thermal processes and the symbol B represents fluctuations in pCO 2 due to non-thermal processes.Error bars represent 1 standard deviation of the mean value for each month.

Figure 9 .
Figure 9. Air-water CO 2 fluxes against river-borne CO 2 fluxes in the urban river and turbidity maximum zone of the Delaware Estuary.Note the different axes used for the urban river and turbidity maximum zone.

Figure 10 .
Figure 10.Dissolved CO 2 concentrations (normalized to 13.3 • C, area-averaged) due to river inputs and internal estuarine sources in each region of the Delaware Estuary.Note the different axes used across all regions of the estuary.Error bars represent 1 standard deviation of the mean value for each month.

Table 1 .
Averaged temperature coefficients (∂ ln pCO 2 /∂T ) for each salinity bin.Simulated surface water pCO 2 values at varying salinities were computed using river and ocean end-member TA and DIC values of 900 and 960 µmol kg −1 and 2300 and 2000, respectively.

Table 2 .
Area average, standard deviation, and range of pCO 2 and CO 2 flux (F CO 2 ) in five of the six zones in the Delaware Estuary during each cruise.
* Months when surveys did not extend into urban river.Area average was estimated by linearly regressing data from adjacent months with sample measurements.Standard deviation and range not available.

6101, 2015 A. Joesoef et al.: Air-water fluxes and sources of carbon dioxide in the Delaware Estuary and
river discharge rates have on surface water pCO 2 distributions' river and estuarine CO 2 inputs, and river-borne CO 2 fluxes throughout the Delaware Estuary.

Table 3 .
Flushing time in five of the six zones in the Delaware Estuary during each cruise.

Table 4 .
Calculated pCO 2thermal , pCO 2non-thermal , T − B, and T /B values for each salinity interval in the Delaware Estuary.