O–NO3 isotope approach

To trace NO−3 sources and assess NO − 3 dynamics in salinized rivers and estuaries, three rivers (Haihe River: HH River, Chaobaixin River: CB River and Jiyun River: JY River) and two estuaries (HH Estuary and CJ Estuary) along the Bohai Bay (China) have been selected to determine dissolved inorganic nitrogen (DIN: NH 4+, NO − 2 and NO − 3 . Upstream of the HH River, NO−3 was removed 30.9 ± 22.1 % by denitrification, resulting from effects of the floodgate: limiting water exchange with downstream and prolonging water residence time to remove NO −3 . Downstream of the HH River NO−3 was removed 2.5 ± 13.3 % by NO − 3 turnover processes. Conversely, NO−3 was increased 36.6 ± 25.2 % by external N source addition in the CB River and 34.6 ± 35.1 % by instream nitrification in the JY River. The HH and CY Estuaries behaved mostly conservatively excluding the sewage input in the CJ Estuary. Hydrodynamics in estuaries has been changed by the ongoing reclamation projects, aggravating the loss of the attenuation function of NO −3 in the estuary.


Introduction
Increasing population, extensive agricultural activities and rapid development of urbanization in coastal areas have dramatically increased N loading to rivers and coastal waters (Seitzinger and Kroeze, 1998;Jennerjahn et al., 2004;Umezawa et al., 2008).Estuaries play a prominent role for delivery of terrestrially derived N to coastal water through physical, chemical and biological processes (Mulholland, 1992;Bernhardt et al., 2003;Sebilo et al., 2006;Hartzell and Jordan, 2012).
Many estuarine studies have focused on tracing N sources and assessing N dynamics in large estuarine systems, such as the Elbe Estuary (Dähnke et al., 2008) and the Atlantic coast (Middelburg and Herman, 2007) in Europe, the San Francisco Bay estuary (Wankel et al., 2006), the Mississippi River estuary (Rabalais et al.,1996), and the mid-Atlantic coast (Dafner et al., 2007) in the United States, and the Yangtze River estuary (Chai et al., 2009) and Pearl River estuary (Dai et al., 2008) in China.Compared to these large estuarine systems with high discharge of freshwater, the levels of freshwater discharge are relatively low in the small estuaries, which are characterized by salinization from seawater intrusion for rather long distances upstream (Graas and Savenije, 2008).How do these salinized estuaries respond to increased N loading?How do physical and biological processes control dissolved inorganic nitrogen (DIN: NH + 4 , NO − 2 and NO − 3 ) concentration variations?
To answer these questions, an intensive study was conducted in three rivers and the corresponding estuaries characterized by different levels of salinization in a coastal municipality (Tianjin) along the Bohai Bay (China).Two investigated rivers with mean salinities around 0.5 and 0.7 flow through a rural area and converge before entering into the estuary.The third one with mean salinity around 2.2 flows through Tianjin municipality and is separated into three parts by two floodgates crossing the river, for providing water supply for the residents living along the river bank.Since the The three rivers investigated are located in a coastal municipality, Tianjin, China (Fig. 1).The study region is influenced by a warm temperate semi-humid monsoon climate with an average annual temperature of 11.4-12.9• C. The annual precipitation is 520-660 mm, with 75 % of the total precipitation occurring in June, July and August (Yue et al., 2010).The population of Tianjin municipality is ca.16 million and the density is 1100 inhabitants km −2 .The survey took place in the dry season on three rivers along a salinity gradient: the Haihe River (HH River) on 7 November 2012, the Chaobaixin River (CB River) on 9 November 2012 and the Jiyun River (JY River) on 10 November 2012 (Fig. 1).Water samples were also taken along the estuary of the HH River (HH Estuary) and the mixing estuary of the CB River and the JY River (CJ Estuary) on 16 November 2012 to study reactive N transformation processes from the river to the estuary (Fig. 1).The HH River is characterized by 72 km in length, ca. 100 m in width, 3-5 m in depth, and a watershed area of 2066 km 2 (Liu et al., 2001).Since the separation by the floodgate F1, the upstream part of the HH River has served as a river-type reservoir for the purpose of supplying water to the residents living along the river bank.The other floodgate F2 is located at the end of the HH River serves as flood discharging, tidal blocking and ship traffic.Although there were eight sewage outlets along the HH River, they were all forbidden to discharge.The average runoff of the HH River was 12.36 × 10 8 m 3 a −1 ; the average tidal amplitude was 2.43 m; and the average flow velocity was 0.3-0.4m s −1 in 2000-2004 (Wen and Xing, 2004).The CB River flows through a rural area and is characterized by 81 km in length, ca.700 m in width, 5-7 m in depth and a watershed area of 1387 km 2 (Gburek and Sharpley, 1998).Animal manure could be a potential dominant NO − 3 source in the CB River as this watershed is an important livestock breeding base for the municipality (Shao et al., 2010).The JY River flows through the agricultural area and is considered to be a significant water source for agricultural and domestic use.The JY River is characterized by 144 km in length, ca.300 m in width, less than 7 m in depth and a watershed area of 2146 km 2 (Chen et al., 2000).The average runoff of the converged river mouth of the CB River and the JY River was 16.03 × 10 8 m 3 a −1 , the average tidal amplitude was 2.45 m, and the average flow velocity was 0.5-0.7 m s −1 from 1990-1997 (Liang and Xing, 1999).Unfortunately, we have no hydrological data for these rivers during the study period.

Sampling and analysis
Water samples were taken on a bridge using a bucket serially from upstream downwards for the rivers and on a ship for estuarine water.The bucket was put into the river/estuary water until it reached ∼ 0.5 m below the surface to sample water.Water samples were stored frozen in 1 L HDPE (high-density polyethylene) bottles for determination of physicochemical properties and δ 15 N-NO − 3 and δ 18 O-NO − 3 .Salinity, temperature (T ), pH and dissolved oxygen (DO) were measured by a portable water quality probe (Thermo Orion, USA).Laboratory analyses included NO − 3 , NO − 2 and NH + 4 .All samples were filtered through 0.45 µm membrane filters and stored at 4 • C until analysis.Nitrate (NO − 3 ), NO − 2 and NH + 4 concentrations were analyzed on a continuous flow analyzer (Auto Analyzer 3, Seal, Germany).
The δ 15 N-and δ 18 O-NO − 3 values were determined by the "bacterial denitrification method" (Sigman et al., 2001;Casciotti et al., 2002;Xue et al., 2010) in the UC Davis Stable Isotope Facility of California University, which allows for the simultaneous determination of δ 15 N and δ 18 O of N 2 O produced from the conversion of NO − 3 by denitrifying bacteria, which naturally lack N 2 O-reductase activity.Isotope ratios of δ 15 N and δ 18 O are measured using a Thermo Finnigan GasBench + PreCon trace gas concentration system interfaced to a Thermo Scientific Delta V Plus isotope-ratio mass spectrometer (Bremen, Germany).The N 2 O sample is purged from vials through a double-needle sampler into a helium carrier stream (25 mL min −1 ) and CO 2 is removed using scrubber (Ascarite ).By cryogenic trapping and focusing, the N 2 O is compressed onto an Agilent GS-Q capillary column (30 m × 0.32 mm, 40 • C, 1.0 mL min −1 ) and subsequently analyzed by isotope ratio mass spectrometry (IRMS).
Stable isotope data were expressed in delta (δ) units in per mil ( ‰) relative to the respective international standards: where R sample and R standard are the 15 N /

Mixing model
The concentration of a mixture can be calculated via a basic mixing model (Liss, 1976): where C represents concentration; the subscripts R and M represent riverine and marine end-members, respectively; f represents the fraction of freshwater in each sample calculated from salinity (Dähnke et al., 2008): where MAX is taken as the maximum measured salinity of marine end-member for coastal water and MEA is taken as the measured salinity of the mixture.Isotopic values of mixed estuarine samples (δ MIX ) were calculated using concentration-weighted isotopic values for riverine and marine end-members, respectively (Fry, 2002;Dähnke, 2008): where C represents concentration; δ represents isotopic value; the subscripts R and M represent riverine and marine end-members, respectively; and f represents the fraction of freshwater in each sample.The salinity-based isotopic mixing does not follow linear conservative mixing but shows curvilinear mixing that reflects concentration-based weighting of end-member isotopic contributions.When a conservative mixing appears between the riverine and marine end-members, DIN distribution is expected to fall on the linear mixing line.When an enriched external source or biological transformation (e.g., mineralization, nitrification) contributes to the river, DIN distribution is expected to fall above the mixing line.In turn, when a depleted external source or the internal removal process (e.g., denitrification, assimilation) appears in the river, DIN distribution is expected to fall below the mixing line (Wankel et al., 2006).The curvilinear mixing curves of determined δ 15 N and δ 18 O of NO − 3 provide better information for transformation processes: an isotopic enriched NO − 3 source or internal removal processes (e.g., denitrification, assimilation) will result in a distribution of δ 15 N and δ 18 O falling above the mixing lines, while an isotopic depleted nitrate source or internal nitrification will result in a distribution of δ 15 N and/or δ 18 O falling below the mixing line.

Nitrate removal efficiency
Variation percentages of the measured NO − 3 concentrations compared to that of the calculated mixing lines were computed to assess the NO − 3 removal efficiency for the rivers and estuaries as follows: where C measured represents the measured NO − 3 concentration, and C theoretical represents the theoretical NO − 3 concentration calculated based on the mixing line.A variation percentage > 0 represents a source, a variation percentage < 0 represents a sink, and a variation percentage equal to 0 represents a mixing.

Physicochemical properties
Table 1 summarizes the data of physicochemical properties collected in this study from the rivers and estuaries.Obviously, the salinities of the HH River (ranging from 0.7 to 4.9 with a mean value of 2.2) and its estuary (ranging from 18.6 to 24.1 with a mean value of 21.2) are higher than the rivers of CB (ranging from 0.5 to 0.6 with a mean value of 0.5) and JY (ranging from 0.6 to 0.8 with a mean value of 0.7) and the corresponding estuary (ranging from 2.0 to 20.0 with a mean value of 7.7).The municipality has suffered multiple seawater intrusions and regressions, which resulted in the  salinization of the rivers and soil (Wang, 2004); moreover, the greater salinization level of the HH River is also related to seawater intrusion over the floodgate until upstream of the HH River for a relatively long distance.The rivers and the estuaries showed similar pH values between 7.5 and 8.6.The temperature of HH River varied around 12.3 • C, which was slightly higher than the CB River (mean is 10.9 • C) and the JY River (11.5 • C).The mean temperature of the HH Estuary (9.7) was also higher than that of the CJ Estuary (6.7).DO concentrations were relatively enriched in this study (higher than 7.2 mg L −1 ), excluding the DO depleted area upstream of the HH River (lower than 5.0 mg L −1 ).

DIN species
Wide concentration variations were noticeable for DIN (NH + 4 , NO − 2 and NO − 3 ) species in Table 1.In the HH River, the NH + 4 concentrations varied from 124.1 to 332.6 µmol L −1 , the NO − 3 concentrations varied from 62.5 to 219.0 µmol L −1 and the NO − 2 concentrations varied from 7.2 to 20.8 µmol L −1 .The DIN concentrations of the HH Estuary varied smoothly (5.6-6.7 µmol L −1 for NO − 2 , 7.1-25.7 µmol L −1 for NO − 3 and 65.7-88.1 µmol L −1 for NH + 4 ) and were quite low compared to the HH River.Nitrate concentrations in the CB river were relatively elevated (120.0-171.5 µmol L −1 ) with a continuous accumulation along the entire salinity gradient, while NO − 2 concentrations decreased from 12.0 to 6.0 µmol L −1 .Ammonium concentrations in the CB River varied from 143.9 to 380.0 µmol L −1 .The JY River also showed NO − 3 accumulation (increased from 40.0 to 83.3 µmol L −1 ) along the entire salinity gradient, while a decreasing trend was observed for both NO − 2 (decreased from 7.0 to 2.1 µmol L −1 ) and NH + 4 (decreased from 72.8 to 11.1 µmol L −1 ) concentrations.The CJ Estuary displayed a seaward decreasing trend with relatively elevated concentrations in NH + 4 (328.4-43.2µmol L −1 ), NO − 2 (7.8-3.4 µmol L −1 ) and NO − 3 (153.4-6.1 µmol L −1 ).Compared to the other rivers and estuaries, DIN results of this study are similar to those in the Pearl River estuary (Dai et al., 2008) in South China Sea, but higher than those in the Elbe Estuary (Dähnke et al., 2008) in Europe and the San Francisco Bay estuary (Wankel et al., 2006) in the United States.The specific reasons causing such variations could potentially be linked to internal/external N source contributions and different N dynamics in the rivers and the estuaries.

3
The isotopic composition of NO − 3 varied spatially among the rivers and the estuaries (Table 1).The δ 15 N-NO − 3 values in the HH River varied from −0.2 to 8.4 ‰ and the δ 18 O-NO − 3 values varied from -0.5 to 1.5 ‰.The isotopic composition of NO − 3 in the HH Estuary remained stable around 8.1 ‰ for δ 15 N-NO − 3 and 5.6 ‰ for the δ 18 O-NO − 3 .In the CB River, the δ 15 N-NO − 3 values were enriched with a mean of 13.6 ‰, and the δ 18 O-NO − 3 values ranged between 3.9 and 5.6 ‰.A decrease in δ 15 N-NO − 3 (from 6.5 to 4.4 ‰) and an increase in δ 18 O-NO − 3 (from 0.8 to 5.3 ‰) values along the salinity were observed in the JY River.The CY Estuary demonstrated a wide range of δ 15 N-NO − 3 (from 7.1 to 15.0 ‰) but a narrow range of δ 18 O-NO − 3 (from 5.9 to 6.9 ‰). 4 in fertilizer and rain (NFR), soil N (Soil) and manure and sewage (M&S).To provide a more wide and clear range of δ 18 O-NO − 3 values, the upper limit of NP reaches 50 ‰.

Potential dominant NO − 3 sources
To derive qualitative information on the predominant NO − 3 sources in the rivers and the corresponding estuaries, a classical dual isotope approach (δ 15 N-NO − 3 vs.δ 18 O-NO − 3 ) has been applied (Fig. 2).It is clear that the isotope signatures of all the sampling locations are shown in a scattered distribution, indicating a different NO − 3 source influence in the rivers and the estuaries.Upstream of the HH River at a salinity of 1.0, a floodgate F1 separates the river into two parts, and at the end of the river at a salinity of 4.9, the other floodgate F2 controls the connection of the river to the HH Estuary.Hence, the δ 15 N-and δ 18 O-NO − 3 values of the HH River behaved quite differently, which moved from the overlapping area of the "NH + 4 fertilizer" and "soil N" source boxes for the majority of the upstream sampling locations, to the overlapping area of the "soil N" and "manure and sewage" source boxes at the end of the river.In this study, the majority of the sampling locations were potentially influenced by the source of "soil N" or "sewage" not the "mineral fertilizer", as the HH River flows through the municipality without agricultural activities.In addition, it can not be excluded from the influence from saltwater intrusion from the estuary, which showed similar isotopic values to those at the end of the HH River.The distribution of the HH Estuary does not show a landward trend due to the floodgate F2 at the end of the HH River, but falls into the range of marine NO − 3 reported by Kendall et al. (2007).
Animal manure could be a potential dominant NO − 3 source in the CB River as this watershed plays the role of an important livestock breeding base for the municipality (Shao et al., 2010).Furthermore, the δ 15 N-NO − 3 values were enriched and varied around 14 ‰, indicating anthropogenic NO − 3 derived from manure (Kendall et al., 2007;Xue et al., 2009).The isotope signatures of the JY River were mainly concentrated in the "soil N" source box.The δ 15 N-NO − 3 and δ 18 O-NO − 3 values of the CJ Estuary suggested an influence of the CB River.In addition, quite high DIN concentrations (Table 1) appeared in this estuary, due to sewage discharge of mooring ships in the vicinity of the sampling area.Thus, the influence of sewage and the CB River was considered as the dominant NO − 3 source.

Nitrate dynamics in the HH River and its estuary
A mixing line (HH1-E) was set up between the most upstream sampling location in the HH River and the most downstream sampling location in the HH Estuary (Fig. 3).
After the separation of the floodgate F1, the upstream of the HH River serves as a river-type reservoir.Thus, a new mixing line (HH2-E mixing line) was re-calculated between the sampling location after the floodgate and estuarine water (Fig. 3).
The salinity gradient sampled in the HH River and its estuary showed a seaward decreasing trend in DIN (NH + 4 , NO − 2 and NO − 3 ) concentrations and an increasing trend in δ 15 N-NO − 3 and δ 18 O-NO − 3 values throughout the entire salinity gradient (Fig. 3).However, the DIN and isotopic trends did not behave conservatively, as most of the measured data deviated from the calculated mixing lines.
It is clear that in the upstream part of the HH River before the floodgate F1, NO − 2 and NH + 4 were above (a source) while NO − 3 was below (a sink) the HH1-E mixing line.Normally, the reductive removal of NO − 3 due to denitrification and assimilation is accompanied by N and O isotope fractionations.The kinetic isotope effects are responsible for preferentially utilizing the lighter isotopes 14 N and 16 O, causing an enrichment of the heavy isotopes in the remaining NO − 3 (Mariotti et al., 1981;Mayer et al., 2002;Fukada et al., 2003).Some studies reported that a linear relationship indicating an enrichment of 15 N relative to 18 O by a factor between 0.8 and 2.0 gives evidence for denitrification (Aravenaand and Robertson, 1998;Fukada et al., 2003;Xue et al., 2009) and 1.0 for assimilation (Granger et al., 2004).In our study, the ratio of N and O isotopic enrichment is 0.8, apparently implying that the removal process of NO − 3 in this river was dominated by denitrification rather than assimilation.From another aspect, elevated NH + 4 compared to NO − 3 will inhibit NO − 3 assimilation by phytoplankton (Dugdale and Hopkins, 1978;Dugdale and MacIsaac, 1971;Dugdale et al., 2006); thus, assimilation process is unlikely significant.The linear relation between the isotopic values and the logarithm of residual NO − 3 indicated that denitrification with constant enrichment factors (ε = −1.8‰ for δ 15 N and ε = −1.4‰ for δ 18 O) was responsible for the increases in δ 18 O and δ 15 N as well.The relatively small enrichment factors were potentially linked to sedimentary denitrification, as diffusion limits the effects of fractionations in the sediments on the δ 15 Nand δ 18 O-NO − 3 in the overlying water column (Sebilo et al., 2003;Lehmann et al., 2004;Sigman et al., 2005).The NH + 4 species was accumulated as a source, potentially originating from organic matter decomposition not sewage discharge, as the δ 15 N-NO − 3 values (−0.2-1.1 ‰) were out of the sewage range.Denitrification could also be the potential process for NO − 2 accumulation in the upstream part of the HH River.
However, nitrification can not be excluded, especially at relatively low DO levels which may favor ammonium oxidizers (NH + 4 →NO − 2 ) rather than nitrite oxidizers (NO − 2 →NO − 3 ), promoting NO − 2 accumulation (Helder and De Vries, 1998).For the HH2-E mixing line after the floodgate F1 (Fig. 3), salinity gradient sampled in the downstream of the HH River illustrated NO − 3 turned from a source (above the HH2-E mixing line) to a sink (below the HH2-E mixing line), while NO − 2 and NH + 4 turned from a sink (below the HH2-E mixing line) to a source (above the HH2-E mixing line) at the end of the river.Nitrate accumulation may be linked to an in-stream nitrification process, in which NO − 2 and NH + 4 were consumed to produce NO − 3 .In nitrification, the conversion of NH + 4 to NO − 2 and NO − 3 is accompanied by marked N isotope fractionation effects, resulting in 15 N-depleted NO − 3 (Delwiche and Steyn, 1970;Mariotti et al., 1981;Macko and Ostrom, 1994).For δ 18 O-NO − 3 values, NO − 3 produced by nitrification in aquatic environments usually takes similar δ 18 O values to the ambient water (Casciotti et al., 2002;Sigman et al.2005).There is evidence that O can exchange between H 2 O and intermediate compounds of nitrification (Andersson et al., 1982;DiSpirito and Hooper, 1986;Kool et al., 2007).Since the δ 18 O of estuarine water is expected to be higher than that of river water (Miyajima et al., 2009), δ 18 O-NO − 3 should increase along the salinity gradient when in situ nitrification is occurring.Thus, a decrease in δ 15 N-(4.6-3.9 ‰) and an increase in δ 18 O-NO − 3 (0.6-1.2 ‰) occurred downstream of the HH River and confirmed the in-stream nitrification process as a NO − 3 source.The NH + 4 concentrations increased at the end of the HH River (a maximum turbidity zone), probably from the release of particle-bound NH + 4 (Seitzinger et al., 1991;Schlarbaum et al., 2010).Results (Kranck, 1984;Eisma, 1986;Schlarbaum et al., 2010) have been reported that this NH + 4 could originate from the mineralization of 15 N-enriched DON adsorbed onto the particles and was released with the estuarine turbidity maximum.The 15 N-enriched NH + 4 was further converted to 15 N-enriched NO − 3 .Thus, the δ 15 N-NO − 3 increased sharply from 3.9 to 8.4 ‰ while the δ 18 O-NO − 3 only increased slightly from 1.2 to 1.5 ‰, resulting from taking similar δ 18 O values to the ambient water.Another possible cause for a sharp increase in NH + 4 concentration could be sewage discharge.Sewage is enriched in 15 N relative to other N sources, as ammonia volatilization causes a large enrichment of 15 N in the residual NH + 4 .This NH + 4 is subsequently converted into 15 N-enriched NO − 3 .When salinity achieves 5, nitrifying bacterial was potentially inhibited and reduced the conversion rate from NO − 2 to NO − 3 (Pollice et al., 2002).Hence, the NO − 2 was accumulated and NO − 3 was declined in this zone.The DIN concentrations and δ 15 N-and δ 18 O-NO − 3 in the coastal water behaved conservatively with regards to mixing.Since the separation of the floodgate F2 at the end of the HH River, the salinity has demonstrated a sudden increase from 4.9 (before the floodgate) to 18.6 (after the floodgate) in 1 km, potentially indicating that the HH river discharge was limited due to the floodgate F2.The δ 15 N-NO − 3 value of the last sampling location in the HH River was close to that of the estuarine water; hence δ 15 N-NO − 3 values remained stable at ∼ 8.0 ‰.The δ 18 O-NO − 3 values increase seaward because of the high percentage of coastal water.

Nitrate dynamics in the CB River and JY River and their estuary
Compared to the HH River, the salinity of the CB and JY rivers varied in a relatively small range, from 0.5 to 0.6 for the CB River and from 0.6 to 0.8 for the JY River.Mixing lines were calculated between the CB and JY rivers and the estuarine water, respectively (Fig. 4).Both CB and JY rivers demonstrated a NO − 3 source along the salinity gradient, indicating a NO − 3 input from either in-stream nitrification or external loading.
Nitrate concentrations in the CB River were elevated with a continuous accumulation along the river.The CB River flows through a rural area with intensive livestock www.biogeosciences.net/11/5957/2014/Biogeosciences, 11, 5957-5967, 2014

Table 1 .
Physicochemical properties and isotopic composition of NO − 3 for the three investigated rivers and the corresponding estuaries.