Institute of Geographical Science and Natural Resources Research, Chinese Academy of Sciences, Beijing, China
Corresponding author: W. Yan, Institute of Geographical Science and Natural Resources Research, Chinese Academy of Sciences, 11A Datun Rd., Chaoyang District, Beijing 100101, China. (firstname.lastname@example.org)
 This study investigated the variations of dissolved N2O and emissions over diurnal and seasonal temporal scales in 2009, as well as the time series of riverine N2O export to estuary and emission to atmosphere in response to increasing anthropogenic nitrogen loads in the Changjiang River. For the diurnal study, N2O concentrations ranged from 0.26 to 0.34 and from 0.44 to 0.52 μg N–N2O L−1 in August and October 2009, respectively. The dissolved N2O was supersaturated with a mean value of 197%. Studies on N2O emissions, also taken in August and October, ranged from 2.67 to 11.6 and from 6.72 to 15.2 μg N–N2O m−2 h−1, respectively. For the seasonal study (June through December 2009), N2O concentrations ranged from 0.34 to 0.72 μg N–N2O L−1 and were supersaturated in all the samples (average 212%). N2O emissions ranged from 1.87 to 40.8 μg N–N2O m−2 h−1. Our study found no significant differences in diurnal patterns of N2O saturation but detected significant difference in seasonal patterns of N2O saturation: higher during summer while lower during autumn and winter. We found a significant relationship between dissolved N2O and river nitrate, which can predict the variation of N2O concentrations in the River. The net production of N2 ranged from 0.01 to 0.47 mg N–N2 L−1. These excess N2 values were significantly correlated to the N2O production and are suggestive of denitrification in the river. Applying the Global News model to the river system using measures taken during the 1970 to 2002 period, we estimated N2O emissions to atmosphere increased from 330 to 3650 ton N–N2O yr−1. During that same 1970–2002 period, N2O exports to estuary increased from 91 to 470 ton N–N2O yr−1. Taken together, the findings reported here suggest that both the river N2O concentrations and emissions would increase in response to rising anthropogenic nitrogen loads. Our study showed that the mean emission factor based on the ratio of the total N2O flux to NO3− flux is four times greater than the value of 0.0025 obtained with the methodology recommended by the Intergovernmental Panel on Climate Change. Thus, our findings reflect the open river channel rapid exchange of gases with the atmosphere.
 Human activities have greatly impacted nitrogen (N) cycles on both regional and global scales via their increasing anthropogenic N inputs [McIsaac et al., 2001; Yan et al., 2003; Han et al., 2009; Seitzinger et al., 2005]. At the regional level, anthropogenic N elevations could lead to increased river N loading under the human pressures. Studies on large-scale nitrogen budgets showed that an average of just 25% of the N added to the biosphere is exported to the ocean or inland basins from rivers [Howarth et al., 1996; Boyer et al., 2006]. As N inputs increase, potentially more N2O is produced [Bouwman, 1995; Seitzinger, 1988]. Multiple studies have demonstrated that N2O emissions from various rivers respond significantly to increasing anthropogenic N loading [Cole and Caraco, 2001; Garnier et al., 2006]. A recent study showed that, globally, more than 0.68 Tg yr−1 of anthropogenic N inputs were converted to N2O in river networks, equivalent to 10% of the global anthropogenic N2O emission rate. Importantly, this rate of anthropogenic N inputs conversion is three times greater than estimates by the Intergovernmental Panel on Climate Change [Beaulieu et al., 2011]. Taken together these findings highlight the significant contribution that rivers make to atmospheric N2O. In addition, estuaries have been identified as active sites for N2O productions and emissions [Usui et al., 2001; Kroeze et al., 2010; Allen et al., 2011; Seitzinger and Kroeze, 1998; Mortazavi et al., 2000]. However, the process by which riverine N2O exports to estuaries is not well understood.
 In China, river N pollution is ubiquitous, from small streams to large rivers [Yan et al., 2010; Tao et al., 2010; Cao et al., 2005]. The Changjiang River, with a drainage area of 1.8 × 106 km2 and average annual runoff of 10 × 1011 m3, is the largest out-flowing river (6400 km length) in China and the third largest in the world (Figure 1). As the most developed region in China, the Changjiang River basin (24°27′–35°54′N, 90°13′–122°19′E) is the most greatly affected by human activities, and thus arguably serves as the best region for assessing the impact of those activities on N cycling in eastern Asia. The N pollution in the Changjiang River has received a great deal of attention [Li et al., 2010; Yan et al., 2003; Liu et al., 2003; Huang et al., 2006; Duan et al., 2008; Yan et al., 2010]. However, the response of the riverine N2O budgets to the increasing nitrogen loads needs further study [Yan et al., 2004]. Geographically, the studies on N2O concentrations and emissions from the Changjiang River are principally focus on the intertidal zone and estuary areas [Wang et al., 2007; Zhang et al., 2010]. Few studies center on the main stem of the river [Yan et al., 2004; Zhao et al., 2009]. Importantly, research until now has not examined the temporal patterns of N2O concentrations and emissions from the river. Regarding the time series of both N2O emissions and exports to the estuary in response to the increasing N load, there are no direct, consecutive, long-term observations to support such a systematic analysis. However, it is feasible to estimate the N2O emissions that are linked to N leaching from the terrestrial environments [Intergovernmental Panel on Climate Change (IPCC), 2006]. Furthermore, from these estimates, the relationship between N2O emissions and anthropogenic N inputs can be established.
 N2O could be produced through several processes, including denitrification, aerobic ammonium oxidation, dissimilatory nitrate reduction to ammonium, etc., in the aquatic ecosystems. Nitrification has been reported as a source of N2O in rivers and estuaries [Barnes and Owens, 1999]. However, the process that is most studied, and, that has been found to be most important is denitrification. High denitrification rates have been measured in eutrophic aquatic ecosystems. Furthermore, denitrification is highly efficient at reducing NO3− loading in temperate rivers [David and Gentry, 2000]. However, there has been a paucity of research on denitrification in the Changjiang River system. Our first report of denitrification measurement in the Changjiang River [Yan et al., 2004] was limited to temporal variations over diurnal and seasonal scales.
 This study was undertook at the lower reach of the Changjiang River, and the findings reported here included (1) the diurnal and seasonal variations of dissolved N2O and emissions; (2) the time series of both N2O emissions to the atmosphere and exports to estuary from the Changjiang River over 1970–2002; and (3) the diurnal and seasonal changes of net N2 productions in relation to N2O productions in the Changjiang River.
2. Materials and Methods
2.1. Study Area
 In this study, the three sampling locations selected from the lower reach of the river were the Datong hydrological station (DHS), Anqing station (AQS), and Maanshan station (MAS). The DHS (117°11′E and 30°46′N) presents the upstream limit of the estuary that is free from both tidal effects and urban effluents of the nearby city. The section of the Changjiang River above the DHS drains about 94% of the total watershed and delivers more than 95% of the water [Yan et al., 2003]. The AQS (117°02′E and 30°28′N) is about 70 km upstream of the DHS, while the MAS (118°25′E and 31°44′N) is about 150 km downstream of the DHS. Table 1 details the research contents and sampling time at the three sampling stations.
[N2O], dissolved N2O concentration (μg N2O–N L−1); [N2], dissolved N2 concentration (mg N–N2 L−1); RT, in situ river temperature (°C); WS, wind speed at 10-m height (m s−1).
AQS (117°02′E, 30°28′N)
Seasonal variation (monthly sampling)
From Jun. to Dec.
NO3−, DO, [N2O], RT,WS
DHS (117°11′E, 30°46′N)
Seasonal variation (monthly sampling)
From Jun. to Dec.
NO3−, DO, [N2O], RT,WS, [N2]
Diurnal variation (60 h, 6-h intervals)
20–22 Aug., 26–28 Oct.
NO3−, DO, [N2O], RT,WS, [N2]
MAS (118°25′E, 31°44′N)
Seasonal variation (monthly sampling)
From Jun. to Dec.
NO3−, DO, [N2O], RT,WS
2.2. Sample Collection
 Surface water samples, 20 cm below the surface, were collected in the reaches from a boat using a bucket. Three replicate water samples were transferred from the bucket to sample bottles avoiding bubbling for dissolved gas analyses (Ar, N2, N2O). Samples for dissolved Ar and N2measurements were collected using 100-ml glass digests, while samples for dissolved N2O analysis were collected in 60-ml serum bottles. Water samples were preserved under water 1–2°C below the in situ temperature by adding a small volume of saturated HgCl2solution (the final concentration was approximately 3%, v/v) to stop biological activity. At the same times and locations, another surface water sample was collected in 100-mL glass bottles for determining the NO3−concentration. Several auxiliary measures were taken in parallel with the water samples including river temperature and dissolved oxygen (DO). Both were measured with a calibrated portable handheld meter (YSI 550A, Unitek, Ltd, China). In addition, 10-m height wind speeds for AQS, DHS and MAS were collected from meteorological stations in nearby Anqing, Tongling and Maanshan cities, respectively. The distance between the sampling sites and the meteorological stations is about 15 km. Finally, at DHS, river discharge, depth, and water flow rate were measured.
2.3. Chemical Analysis
 The headspace-equilibrium method [Walter et al., 2006; Clough et al., 2007; Huttunen et al., 2002] was used for measurement of initial sample dissolved N2O concentrations in river water. Five milliliters of highly purified N2(purity > 99.999%) was injected into the sampling bottle using an airtight syringe and a 5-ml water sample was displaced. Bottle headspace N2O concentrations were directly analyzed using a gas chromatograph (GC-2014, Shimadzu, Japan) equipped with an electron capture detector (ECD) after the bottles were vigorously shaken for 4 h. The gas chromatograph was calibrated using a standard air sample. Initial N2O concentrations (Cw) in water samples were calculated using methodology described by Johnson et al. . Dissolved Ar and N2 in river water were determined using a Membrane Inlet Mass Spectrometry (MIMS) system (HPR40, UK). For details on the analysis and calculation of dissolved Ar and N2 concentrations see the method described by Kana et al.  and Laursen and Seitzinger . Here, we briefly describe the process of MIMS N2 measurement. We measured N2/Ar using MIMS, then we calculated Ar concentrations at in situ temperatures according to the equation by Weiss and Price , finally, we got dissolved N2 concentrations by multiplying N2/Ar with Ar concentrations at in situ temperatures. The net production of N2 (ΔN2) were then calculated as: ΔN2 = [N2]measured − [N2]equilibrium. The equilibrium concentrations of N2O and N2 (Ce) between river water and atmosphere were calculated using the relationship with temperature and salinity determined by Weiss and Price  and Weiss , respectively. Dissolved N2O saturation, expressed in %, was calculated as the ratio of the concentration of dissolved N2O and the expected equilibrium water concentration. NO3−concentrations were determined using a Flow Injection Analyzer (FIA-3100, China) after water samples were filtrated through 0.45μm membranes.
2.4.1. The Diurnal and Seasonal Study
 Water surface emissions of N2O observed in the diurnal and seasonal study were calculated by the classic approach based on the gas transfer velocity and the concentration difference [Raymond and Cole, 2001; Garnier et al., 2006]:
where ΔN2O (μg N–N2O L−1) is the concentration difference between the measured and the equilibrium concentration with the atmosphere in river water. kN2O (cm h−1) is the gas transfer velocity. Since no direct measurements of gas transfer velocity were made in the Changjiang River, we used the method of Borges et al.  to estimate K600 by an equation accounting for both wind speed and water flow rate:
where w is the water flow rate (m s−1), h is the depth of river water column (m), u10 (m s−1) is the wind speed at a 10-m height. kN2O was corrected at in situ temperatures based on the relationship kN2O/k600 = (ScN2O/600)−n, where ScN2O is the Schmidt number for N2O and was calculated according to the equation by Wanninkhof  at in situ temperature, and n is the Schmidt number coefficient (2/3 for smooth surfaces). Table 2 details the physical characteristics and transfer velocity at three sampling sites.
Table 2. Physical Characteristics of the Changjiang River and the Gas Transfer Velocity
Mean Flow Rate (m s−1)
Mean River Depth (m)
Mean River Temperature (°C)
Mean Wind Speed (u10) (m s−1)
Mean K600 (cm h−1)
Mean KN2O (cm h−1)
Diurnal Study (60 h)
Seasonal Study (Jun–Dec)
2.4.2. Time Series of N2O Emissions to Atmosphere
 The time series (1970–2002) of N2O emissions to atmosphere linked to nitrogen leaching from the Changjiang River were calculated by the equation developed by the IPCC :
where [NO3−] (ton N yr−1) is considered as the fraction exported from soil to river channel through leaching of the total annual surface N surplus yields from synthetic fertilizer and manure application in the Changjiang river basin. The data for [NO3−] can be obtained from Yan et al. . The IPCC methodology for estimating aquatic N2O emissions is based on the assumption that groundwater derived N2O is the dominant source of N2O in streams while in situ production makes up the dominant N2O source in rivers [Beaulieu et al., 2008]. EF5-r in this study is the coefficient of 0.0025 determined by theIPCC .
2.4.3. Time Series of Riverine N2O Export to Estuary
 Due to the lack of systematic measurement of the dissolved N2O, the riverine N2O exported to the estuary from the Changjiang River over a long time period is still unknown. Here, we first estimated the dissolved N2O through a model established in relation to NO3− concentrations based on the direct measurements from our study:
where N2Oexport, is dissolved N2O flux to estuary, (kg N–N2O yr−1); f (NO3−) is a model describing the relationship between observed NO3− concentration (mg N L−1) and N2O concentration (μg N–N2O L−1), which is discussed in section 3.4; and is water discharge (1010 m3 yr−1). The estimates for the time series (1970–2002) of concentrations of NO3− in this paper can be found in Yan et al. .
 We supposed that the N2O production processes, like denitrification and nitrification, were both correlated with the NO3− concentration. In addition, NO3− ranged from 0.59 to 1.55 mg N L−1, with an average concentration of 1.01 mg N L−1, but NH4+ concentrations were much lower, ranging from 0.09 to 0.25 mg N L−1 in the Changjiang River (DHS) over the 1970–2002 period (Figure 2). Thus, NO3− is the main component of DIN (NH4+ + NO3−), accounting for about 72%–94% of the DIN concentration, and NH4+ accounting for only 6–28% of the DIN. Furthermore, Middelburg and Nieuwenhuize  had reported a rather transient character of NH4+ in the natural aquatic ecosystems, although the relative importance of nitrification as a N2O source in large rivers is still unclear, we can argue that NH4+ is not an important predicting variable because the relationship between their measurements showed no correlation between NH4+ and N2O. Therefore, we thought the model (equation (4)) for predicting dissolved N2O based on the correlation of dissolved N2O to NO3− is acceptable in this article.
2.5. Statistical Analysis
 The SPSS 13.0 software package was employed for all statistical analyses. One-Way ANOVA (LSD test) and Independent-Samples T Test at the 0.05 confidence level were used to test for statistical differences between group mean values. Regression analyses were done to test the relationships between N2O concentrations and environmental factors.
3. Results and Discussion
3.1. Diurnal Variations of Dissolved N2O and Emissions
 Over a 60-h period in August, concentrations of N2O, NO3−, and DO at the DHS ranged from 0.26 to 0.34 μg N–N2O L−1, from 0.79 to 0.82 mg N L−1, and from 7.68 to 7.83 mg L−1, respectively. Over the same time period and in the same location in October, N2O, NO3−, and DO ranged from 0.44 to 0.52 μg N–N2O L−1, from 1.01 to 1.08 mg N L−1, and from 8.79 to 9.24 mg L−1, respectively. Correspondingly, the percent saturation in August for both N2O and DO ranged from 154% to 198% and from 99% to 101%, respectively. In October, the percent saturation for N2O and DO gases ranged from 209% to 235% and from 99% to 101%, respectively (Figure 3). Dissolved N2O was supersaturated with a mean value of 197% (ranged from 154% to 235%), while saturation of DO changed little (with a mean value of 100%) in all the samples (Figure 3). Our study found that there are no significant differences in diurnal patterns of saturation for both N2O and DO gases over a consecutive 60-h period in each observation (LSD test, p> 0.05). However, Independent-Samples T Test showed that mean dissolved N2O in October (0.47 ± 0.02 μg N2O–N L−1, SEM) was significantly higher than that in August (0.29 ± 0.03 μg N–N2O L−1, SEM) (p < 0.001), highlighting the seasonal inhomogeneity in N2O productions [McCrackin and Elser, 2010; Chénier et al., 2006; Stadmark and Leonardson, 2007]. NO3−and DO concentrations also showed no statistically significant differences over the consecutive 60-h periods in each observation (LSD test,p > 0.05). For the diurnal study, the ratios of masses of N–N2O against N–NO3− ranged from 0.3 × 10−4 to 0.5 × 10−4 and were correlated with DO (r2 = 0.58, p < 0.0001) (Table 3). In addition, both the dissolved N2O and ΔN2O (N2Omeasured − N2Oequilibrium) (ranged from 0.09 to 0.30 μg N–N2O L−1) were significantly correlated with NO3− (r2 = 0.94, p < 0.0001; r2 = 0.88, p < 0.0001) and DO (r2 = 0.86, p < 0.0001; r2 = 0.82, p < 0.0001), suggesting that NO3− and DO could be the predominant factors influencing N2O production. River temperatures varied from 19 to 29°C between the 2 seasonal sampling times and showed clear diurnal patterns (Figure 3). River temperature (r2 = 0.74, p < 0.0001) was significantly correlated with dissolved N2O.
Table 3. Regression Analysis Between Dependent and Independent
Best Regression Formula
y = −0.76 + 0.14x
y = −0.27 + 0.71x
y = −0.68 + 0.10x
y = 0.31 + 0.53x
y = −9.82 × 10−5 + 6.02 × 10−5x
y = 0.13 + 0.04x
y = 0.26 + 0.18x
y = −0.26 + 0.06x
y = 0.02 + 0.19x
y = 6.46 × 10−4 − 2.58 × 10−5x
y = −0.12 + 0.06x
log(y) = −0.39 + 0.69 log(x)
y = −0.37 + 0.07x
y = −0.03 + 0.23x
y = 4.9 × 10−4 − 8.99 × 10−6x
y = 0.75 + 7.88x
y = 0.03 + 5.98x
 Estimated N2O emissions in August ranged from 2.67 to11.6 μg N–N2O m−2 h−1 (average 7.36 ± 3.32 μg N–N2O m−2 h−1, SEM) over a 60-h period (Figure 4a). These estimated emissions measured in August were significantly lower than those sampled in October which ranged from 6.72 to15.2 μg N2O–N m−2 h−1 (average 9.37 ± 2.46 μg N2O–N m−2 h−1, SEM) (p = 0.004) (Figure 4b). The dynamics of N2O emissions did not correlate with dissolved N2O, suggested that temporal variations in K may obscure the correlation between the ΔN2O and N2O emissions rates [Zappa et al., 2007].
3.2. Seasonal Variations in Dissolved N2O and Emissions, Riverine N2O Exports to Estuary
 Dissolved N2O in the AQS, DHS, and MAS ranged from 0.43 to 0.56 (average 0.49 ± 0.05, SEM), from 0.34 to 0.48 (average 0.40 ± 0.04, SEM), and from 0.42 to 0.72 μg N–N2O L−1 (average 0.57 ± 0.11 μg N–N2O L−1, SEM), respectively. Correspondingly, N2O saturation in the AQS, DHS, and MAS ranged from 147% to 244%, from 116% to 256%, and from 163% to 339%, respectively (Figure 5a), and were supersaturated in all the samples (range of 116% to 339%, average 212%). Dissolved N2O concentrations showed a significant seasonal variation pattern with concentrations higher during summer and lower during autumn and winter. Our observed N2O concentrations on the Changjiang River, fell within the range reported previously (0.25 to 0.93 μg N–N2O L−1) [Zhao et al., 2009; Zhang et al., 2010; Yan et al., 2004]. NO3− concentrations in the AQS, DHS, and MAS ranged from 0.74 to 1.11, from 0.92 to 1.13, and from 1.39 to 2.08 mg N L−1, respectively (Figure 5b). The DO concentrations in the AQS, DHS and MAS study sites ranged from 6.84 to 9.19, from 7.37 to 11.33, and from 8.27 to 11.5 mg L−1, respectively. Finally, DO saturation at the three sites ranged from 83% to 92%, from 99% to 102%, from 103% to 118%, respectively, (Figure 5c). Like the diurnal study, the dissolved N2O and ΔN2O (range from 0.05 to 0.49 μg N–N2O L−1) was significantly correlated with NO3− and DO concentrations (Table 3). Ratios of masses of N–N2O to N–NO3− ranged from 0.2 × 10−4 to 0.7 × 10−4. In contrast to the diurnal study, no significant correlation between N–N2O/N–NO3− and DO was found.
 Estimated N2O emissions to atmosphere from the AQS, DHS, and MAS ranged from 7.74 to 25.6, from 1.87 to 35.7, and from 8.26 to 40.8 μg N2O–N m−2 h−1, and averaged 15.9 ± 7.62 (SEM), 12.5 ± 13.1(SEM), and 25.6 ± 14.3 (SEM) μg N–N2O m−2 h−1 (Figure 5d), respectively. Other reports on the N2O emissions from a diverse set of rivers span a wide range (Table 4), and the N2O emissions we observed in the Changjiang River fell in the lower end of this range. N2O emissions from three sites exhibited a similar seasonal pattern to the dissolved N2O. Mean N2O emission and concentration in the MAS were significantly higher than those in the AQS and DHS (LSD test, p < 0.05). This could be explained by the high N concentrations in the river channel of MAS by the effluents of urban wastewater from the nearby city of Maanshan. As a consequence, more N2O was produced and emitted than at the AQS and DHS sites.
Table 4. Comparison of N2O Emissions From Various Kinds of Streams and Rivers
Emission rates were measured with the floating chamber technique, were calculated based on dissolved N2O concentrations and a modeled air-water gas exchange rate (k), or k was measured with a gas tracer technique [Beaulieu et al., 2008].
 From June to December of 2009, water discharge through the DHS varied from 30.3 to 113 km3 mon−1 (mean 72.4 km3 mon−1). When multiplied by the N2O concentrations, we estimated that monthly riverine N2O exports to the estuary ranged from 11.5 to 49.8 ton N–N2O. Furthermore, that a total of 206 ton N–N2O was exported to the estuary from the Changjiang River over the 7-month observed period (Figure 6). Given that the NO3− export to estuary varied from 2.79 to 11.7 × 104 ton N mon−1, we estimated that the N2O export represents only about 0.04% of the exported NO3−. Our data suggested a strong relationship between water discharge, N2O and NO3− export. Since interflow is the main link between the land and river, and the base flow is normally a minor contributor, we hypothesized that the strong relationship between water discharge, N2O and NO3− export may be principally regulated by the water discharge, and that this could be an indicator of leaching, groundwater interflow and river transport.
3.3. Denitrification and N2O Production
 In riverine ecosystems, N2O can be generated via the process of denitrification as the intermediate product [Silvennoinen et al., 2008]. N2 is the end product of denitrification, the net production of which is generally considered reflecting the denitrfication level. In our study, the dissolved N2 concentrations were synchronized to N2O observations in the DHS. The net production of N2 (ΔN2, equal to [N2]measured − [N2]equilibrium) were then calculated under the in situ river temperature.
 The diurnal variations of ΔN2 ranged from 0.01 to 0.03 and from 0.23 to 0.43 mg N–N2 L−1 in the August and October measures (Figure 7a), respectively. The seasonal variations in ΔN2 ranged from 0.01 to 0.47 mg N–N2 L−1 (Figure 7b). The excess gas concentration of dissolved N2 strongly suggests robust denitrification in the lower reach of the Changjiang River. However, our measured excess N2 was low relative to that reported in other aquatic ecosystems [Laursen and Seitzinger, 2004; Chang et al., 2010; Weymann et al., 2008; Pribyl et al., 2005], and suggesting a relatively low level of denitrification in the Changjiang River. Kenny et al.  have reported that denitrification is associated with high NO3− availability in aquatic environments. In our study, a significant correlation between ΔN2O and NO3− (r2 = 0.55, p < 0.0001) were identified (Figure 8a). In addition, good linear correlations between ΔN2 and ΔN2O (r2 = 0.79, p < 0.0001), ΔN2 and NO3− (r2 = 0.52, p < 0.0001) were found (Figures 8b and 8c). These correlations suggest that the N2O production could be accounted for by the denitrification process in the Changjiang River.
 The ratios of masses of N–N2O to N–N2 in this study were calculated to determine the magnitude of the N2O production during denitrification. Results showed the ratios ranged from 0.51% to 1.12%. Generally, the ratio of N2O to N2produced via denitrification in water-saturated ecosystems should vary from 0.04% to 5.6% [Beaulieu et al., 2011]. Relative to the result that N2O production was about 0.5–47% of the N2 production reported by Teixeira et al. , our ratios of masses of N–N2O to N–N2 were lower. However, they are consistent with the new data reported by Beaulieu et al. . In considering that an unknown fraction of ΔN2O came from nitrification, our ΔN2O/ΔN2 ratio would represent the maximum fraction of denitrified NO3− converted to N2O. However, the correlation between the ΔN2O and ΔN2 strongly suggests denitrification as a source for N2O. Clearly, additional research is required to partition N2O production between nitrification and denitrification. Finally, the ratios of ΔN2 to NO3− varied between 0.96% and 4.08%, with a mean value of 2.61%, representing the fraction of the N load that has been converted to N gases in the Changjiang River.
 The findings presented here suggest that denitrification plays the predominant role in the Changjiang River. Nevertheless, our finding that NO3− is significantly correlated with DO (Figure 9; r2 = 0.42, p < 0.001) suggests that nitrification may also play a role in the Changjiang River. A role for nitrification seems reasonable given that DO concentrations in the river water were relatively high. A role for nitrification in the Changjiang River appears to contradict our finding that that ΔN2O was significantly correlated with ΔN2. However, the possibility remains that the processes of both nitrification and denitrification occurred in the river water during our experimental periods. Furthermore, that the N2O produced via nitrification may be too small to influence the significant correlation between ΔN2O and ΔN2. Finally, while our measured NO3−concentrations in the Changjiang River are well above Seitzinger's theoretical maximal levels for supporting coupled denitrification-nitrification [Seitzinger et al., 2006], we suggest that the greater significance of nitrification could be enhancing the denitrification process through promoting NO3− concentrations (Figure 10). We will test this alternative explanation for coupled denitrification-nitrification in future detailed studies.
3.4. Time Series of Riverine N2O Exports to Estuary and Emissions to Atmosphere
 To date, there have been no systematic measurements detailing the time series of riverine N2O exports to the estuary and emissions to the atmosphere from the Changjiang River. Here, the time series of N2O exports to the estuary were estimated using the data of predicted N2O concentrations and river water discharge through the DHS. In parallel, we estimated the N2O emissions to atmosphere using calculations based on the IPCC guidelines and by applying the EF5-r (the coefficient of 0.0025 derived by the IPCC) to the [NO3−] data reported by Yan et al. .
 N2O concentrations in the DHS over the period of 1970–2002 were derived using a model (equation (4)), in which NO3−concentrations were taken as the candidate predictor. The model required several assumptions including: (i) river channel morphology did not significantly change over the 32-yr period, and (ii) N2O production could be predicted from NO3− concentrations. The model to predict dissolved N2O can be best (r2 = 52, mse = 0.08, n = 41) described by equation (4) (Figure 11):
where the numbers in parentheses below the equation are the standard errors of their respective parameter estimates. By using DHS NO3− concentrations described by Yan et al. , we estimated that N2O concentrations increased from 0.12 to 0.53 μg N–N2O L−1 during the 1970–2002 time interval. Given that the annual water discharge through DHS ranged from 649 to 1240 km3 during this period, we estimated that N2O exports from the river to the estuary increased from 91 to 566 ton N–N2O yr−1, with an annual growth rate of 6.22% relative to the elevated NO3− loading. In alignment with the seasonal data reported here, the riverine N2O exports during 1970–2002 represent about 0.04% of the NO3− exports.
 The time series of N2O emissions to atmosphere linked to nitrogen leaching were estimated using the emission factor developed by the IPCC  guidelines. In the Changjiang River basin, fertilizer and manure N lost through leaching increased from 132 to 567 kg N km−2 yr−1 during the period of 1970–2002 [Yan et al., 2010], equivalent to 13.3 × 104–126 × 104 ton N yr−1. When N2O emissions were calculated using the IPCC guidelines and the revised EF5-r coefficient, the estimates showed that indirect N2O emissions from the Changjiang River had increased from 330 to 3650 ton N–N2O yr−1. This could be accounted for by an annual growth rate of 7.01%. When expressed with respect to river water area (see Table 5), the N2O emissions had ranged from 3.86 to 36.6 kg N–N2O km−2 yr−1. The N2O exports to the estuary, ranged from 91 to 470 ton N–N2O yr−1, accounting for about 15%–28% of the riverine N2O emissions to atmosphere during the 1970–2002 time period. Yan et al.  reported that significant differences existed for total diffuse N inputs between the periods 1970–1982 and 1983–2003, due to the change in N leaching and runoff from agricultural soils to surface water of the Changjiang River. Here, we identified similar dynamics for estimated N2O emissions and riverine N2O exports. During 1970–1982, N2O emissions and riverine exports showed mean values of 800 and 170 ton N–N2O yr−1, respectively. These values were significantly lower than the N2O emissions and riverine N2O export values for 1983–2002, which averaged 2400 and 385 ton N–N2O yr−1, respectively. This indicates that the potential effect of anthropogenic activities on N2O emissions and riverine exports had increased markedly as the rate of N loading to river water increased.
Table 5. Comparison of N2O Emission Factor Between Those Calculated by the Classic Method (Equation (1)) and the IPCC Method
Total N2O emission per hour calculated by equation (1) = Mean observed N2O emission rate × river surface area. While water surface area of the Changjiang River is reported as 86000 km2, (seeing “The general situation of the Changjiang drainage basin” [http://www.tgenviron.org]).
N2O emission factors calculated in this study = N2O emission (ton N h−1)/NO3− export (ton N h−1). Ratio (%) = total N2O emission calculated by equation (1)/total N2O emission calculated by the IPCC methodology.
Diurnal study (60 h)
Seasonal study (Jun–Dec)
 However, in order to assess the accuracy of N2O emissions by the IPCC methodology, we compared the values of N2O emission factor between this study by using (equation (1)) and the IPCC methodology. We incorporated our parameters for river water area, water discharge and NO3− concentration that we measured in this study (Table 5; equation (1)). We found that the emission factor (ratio of N2O to NO3−) calculated in our study ranged from 0.0050 to 0.0148 with an average of 0.01 (Table 5). This estimate is almost four times greater than the value of 0.0025 obtained by the IPCC methodology. Generally, the emission factor (EF5-r coefficient) developed byIPCC  represents the ratio of dissolved N2O to NO3− concentrations in rivers, which may underestimate the fraction of the N load converted to N2O, because much of the N2O has already been lost to the atmosphere [Beaulieu et al., 2008]. Importantly, the emission factors in our study are based on the ratio of the total N2O flux to NO3− flux through the river reach. Thus, our findings represent the open river channel rapid exchange of gases with the atmosphere. Of course, the uncertainty of calculation for N2O emission factors may exist when we used the data only observed in lower reach of the river to represent the whole river, we consider that a spatial observation along the river system can improve such calculation of N2O emission factor.
 The diurnal variation study showed that, N2O concentrations in the lower Changjiang River ranged from 0.26 to 0.34 μg N–N2O L−1 (average 0.29 ± 0.03 μg N–N2O L−1, SEM) in August, and from 0.44 to 0.52 μg N–N2O L−1 (average 0.47 ± 0.02 μg N2O–N L−1, SEM) in October. Dissolved N2O was supersaturated with a mean value of 197% (ranged from 154% to 235%). N2O emissions ranged from 2.67 to 11.6 μg N–N2O m−2 h−1 (average 7.36 ± 3.32 μg N–N2O m−2 h−1, SEM) in August; and ranged from 6.79 to15.2 μg N–N2O m−2 h−1 (average 9.37 ± 2.46 μg N2O–N m−2 h−1, SEM) in October. From June to December of 2009, N2O concentrations varied from 0.34 to 0.72 μg N–N2O L−1 and were supersaturated in all the samples (range of 116% to 339%, average 212%). N2O emissions ranged from 1.87 to 40.8 μg N–N2O m−2 h−1. Both N2O concentration and emission values were higher in summer and lower in autumn and winter. The seasonal N2O exports to the estuary ranged from 2.79 to 11.7 × 104 ton N mon−1. A significant relationship between dissolved N2O and river nitrate concentrations was established to predict the variation of N2O concentrations in the Changjiang River. The net production of N2 ranged from 0.01 to 0.47 mg N–N2 L−1. These excess N2 values were significantly correlated to the N2O production, and are suggestive of denitrification in the Changjiang River. Relative to N loading of the Changjiang River, we estimated N2O emissions to atmosphere due to N leaching increased from 330 to 3650 ton N–N2O yr−1 during the 1970–2002 time period. This marked increase is equivalent to an annual growth rate of 7.01%. We projected similar marked changes in N2O exports to estuary during the periods of 1970–2002, with values increasing from 91.0 to 470 ton N–N2O yr−1 and an equivalent annual growth rate of 6.22%. Taken together the findings presented here suggest that the potential effect of anthropogenic activities on N2O emissions and exports would increase markedly as the rate of N loading to the Changjiang River increases. Our study showed that the emission factors based on the ratio of the total N2O flux to NO3− flux through the river reach is four times greater than the value of 0.0025 obtained by the IPCC methodology. Thus, our findings reflect the open river channel rapid exchange of gases with the atmosphere.
 This study was supported by the National Natural Science Foundation of China (20777073) and Exploratory Forefront Project for the Strategic Science Plan in IGSNRR, CAS (2012QY001). The authors would like to thank Sun Pu (Hydrological Bureau of Anhui Province) for help with sampling in the field and Dr. Zhang Lu (Nanjing Institute of Geography and Limnology, Chinese Academy of Sciences) for providing analysis of dissolved N2 in the water. We also appreciate the reviewers' constructive comments.