1 INTRODUCTION

Nitrogen (N) is essential in the biosynthesis of DNA, protein, phospholipid, etc., with global storage of 2.2×10¹⁰ Tg (Kuypers et al., 2018). Dissolved inorganic nitrogen (DIN) accounts for a small portion of total N inventory in the biosphere (Kuypers et al., 2018), but it is characterized by active cycling and transport. Biological and industrial nitrogen fixation introduces ca. 425 Tg N-DIN per year into the biosphere; while only 300 Tg N-DIN per year can be transformed to dinitrogen gas (Kuypers et al., 2018). Excessive DIN in terrestrial ecosystems, mainly nitrate (NO₃^ˉ), is transported by surface rivers into coastal oceans through estuaries (47.8 Tg N/a; Galloway et al., 2004). The enrichment of land-borne NO₃^ˉ in coastal waters frequently triggers harmful algae blooms and subsequent hypoxia, causing significant economic losses (Justić et al., 2003). Therefore, evaluating the magnitude of riverine NO₃^ˉ fluxes and exploring the reaction pathways related with NO₃^ˉ concentration variations in estuaries have received great attention from coastal managers and researchers (Wong et al., 2014; Loken et al., 2016; Yan et al., 2017; Domangue and Mortazavi, 2018).

The Changjiang River, length of ca. 6 300 km, delivers 9.24×10¹¹ m³/a river water to adjacent coastal oceans, ranking the fifth largest rivers on a global scale (Milliman and Farnsworth, 2011). In the 1980s, the Changjiang delivered ca. 6.0×10¹⁰ mol NO₃^ˉ/a to the East China Sea and the Yellow Sea (Edmond et al., 1985). Afterwards, the NO₃^ˉ concentration in the Changjiang River water increased more than three folds, indicating a marked enhancement in riverborne NO₃^ˉ input. Land-use changes due to rapid urbanization, overuse of fertilizers and sewage discharge are deemed to be key contributors of riverine NO₃^ˉ (Dai et al., 2011). The Changjiang diluted water (CDW), consisting of the Changjiang River runoff and coastal saline water, could extend to more than 200-km distance from the river mouth and cover 104 km2 surface area (31‰ isohaline; Chang et al., 2014). A series of research projects have been launched to explore the spatial/temporal distribution of riverine NO₃^ˉ and reactions related to the concentration variation (Zhang, 1996; Yao et al., 2014; Yu et al., 2015; Liu et al., 2016; Yan et al., 2017). In the Changjiang estuary, nitrification, biological uptake, denitrification and anaerobic ammonium oxidation (Anammox) in the water column and/or benthic sediments have been determined (Song et al., 2013; Yan et al., 2017). These reaction pathways associated with NO₃^ˉ addition (nitrification) and removal (Anammox, denitrification, assimilation, etc.) fundamentally influence NO₃^ˉ concentration and the concentration-based impacts. Stable isotope compositions such as δ¹⁵N-NO₃^ˉ and δ¹⁸O-NO₃^ˉ are valuable for identifying reaction pathways of NO_3.^ˉ In estuaries, mineralization of pelagic organic matter and subsequent nitrification decreases both δ¹⁵N-NO₃^ˉ and δ¹⁸O-NO_3.^ˉ In contrast, biological consumption leads to an increase in δ¹⁵NNO₃^ˉ and δ¹⁸O-NO_3.^ˉ Theoretically, the ratio for the enhancement in δ¹⁵N-NO₃^ˉ and δ¹⁸O-NO₃^ˉ is 1꞉1 (Granger et al., 2010). Denitrification in the water column could produce a similar enrichment in δ¹⁵NNO₃^ˉ and δ¹⁸O-NO₃^ˉ; while the enhancement in the isotope compositions caused by denitrification in the benthic sediment may not be observed if NO₃^ˉ is fully converted (Yan et al., 2017). Accordingly, the combination of NO₃^ˉ concentration and its stable isotope composition could improve our understanding of NO₃^ˉ production and removal in the CDW.

Environmental factors in the Changjiang estuary, such as salinity, total suspended matter (TSM), and dissolved oxygen (DO), are highly dynamic because of the co-variation of tidal amplitude and the Changjiang River discharge rate, as well as stratification (Zhang, 1996). In addition, there are several outlets for the Changjiang River runoff (Fig. 1). Discharge rates, tidal amplitudes, and related water residence time among these outlets are significantly different (Wu et al., 2010), which also deeply influences the related environmental parameters in the regions outside of these outlets. Biogeochemical reactions, including nitrification, denitrification, and assimilation are sensitive to the changes in these factors (Kuypers et al., 2018). Up-tonow, the published documents only revealed the variation of NO₃^ˉ concentration and its isotope compositions in surface water (Liu et al., 2009) or on a single transect (Yan et al., 2017) in the Changjiang estuary. However, studies on NO₃^ˉ distribution and reactions, as well as dynamic linkage between environmental factors and NO₃^ˉ production/removal based on a comparison between multiple transects with different water depths are limited.

In the present study, a cruise (R/V Zheyuke Ⅱ) in theChangjiang estuary was conducted during summer 2017 when the Changjiang River discharge rate was high ((6-7)×10⁴ m³/s gauged at Datong Station; http://www.cjh.com.cn). The objectives of this study were (1) to identify NO₃^ˉ sources in the Changjiang River runoff prior to river-sea mixing; (2) to determine reactions with regard to NO₃^ˉ production and removal along the CDW pathway; and (3) to explore the relationship between the observed reaction pathways and environmental parameters, such as salinity, DO, and TSM in the estuary.

2 MATERIAL AND METHOD 2.1 Sample collection

The cruise in the Changjiang estuary was conducted on two transects (A and B) from 1^st July to 8^th July 2017, extending from river channels at the south branch (low salinity water) to the outer edge of the CDW as well as the Taiwan Warm Current (TWC) site (Fig. 1). Besides, the river water at the Xuliujing (XLJ) site was obtained on 4^th July 2017 (Fig. 1). The tidal range gauged at the Zhongjun station (Fig. 1) varied from 1.8 m on 4^th July to 2.4 m on 8^th July 2017. Transect A is the south outlet of Changjiang River runoff and adjacent to the deep-water channel, a channel for marine transport vessels. It has been surveyed in previous research projects related to NO₃^ˉ transport and transformation (Wang et al., 2017; Yan et al., 2017). Transect B extends from another outlet with relatively high discharge rate. It is separated with transect A by the Changxing Island and Hengsha Island. During the survey, phytoplankton bloom (dominant species, Skeletonema costatum) was observed in the CDW outer plume on both transects (outlined in Fig. 1). The water residence in these regions ranged from 3 to 15 days, estimated by drifters released in the river mouth at the beginning of the cruise.

Water samples during the cruise were collected from a CTD rosette sampler, equipped with 12-L Niskin bottles, a conductivity-temperature-depth (SBE 911, Sea-Bird Co.) unit and a fluorescence detector. In each site, 2-4 layers of the CDW, ranging from 3- to 53-m depth, were sampled. After collection, the measurement of DO concentration was immediately conducted using a detector (JENCO^®, Model 9173). Approximately 1-L water was filtered through a polycarbonate membrane (pore size: 0.4 μm, Whatman^®) for the quantification of dissolved nitrogen content, δ¹⁵N-NO₃^ˉ and δ¹⁸O-NO_3.^ˉ The filtrate was stored at -20 ℃ until laboratory processing. Another fraction of water (0.3-3 L) was filtered by pre-combusted GF/F filter (average pore size: 0.7 μm, Whatman^®). The solids retained on the filters were determined as the TSM concentration.

2.2 Laboratory analysis

Concentrations of DIN species (NH₄⁺, NO₂^ˉ, and NO₃^ˉ) were determined on a flow injection analyzer (Skalar Analytical B.V., The Netherlands) using standard colorimetric procedures (Hansen and Koroleff, 1999). The method determination limit for these species was approximately 0.1 μmol/L. The analytical accuracy was approximately ±3%. The concentration of total dissolved nitrogen (TDN) was determined by the potassium peroxydisulfate (K₂S₂O₇) digestion method with the determination limit of 0.2 μmol/L and method precession of ±3% (Ebina et al., 1983). The level of dissolved organic nitrogen (DON) was obtained as the difference between TDN and DIN concentrations. The δ¹⁵N-NO₃^ˉ and δ¹⁸O-NO₃^ˉ were determined by bacterial reduction method (NO₃^ˉ to N₂O, using Pseudomonas aureofaciens, Sigman et al., 2001) following the procedure described by Weigand et al. (2016). NO₂^ˉ in samples was removed by the prepared sulfanilamide solution prior to bacterial reduction (Weigand et al., 2016). The generated N₂O was purified and concentrated in a Finnigan Precon System (Thermo Fisher, USA). Subsequently, N₂O and trace amount of CO₂ was separated by a chromatographic loop (GC Isolink; Thermo Fisher, USA). Afterwards, the δ¹⁵N and δ¹⁸O were analyzed in an isotope ratio mass spectrometer (Delta V; Thermo Fisher, USA). Measurements of δ¹⁵N were referenced to atmospheric N₂ and δ¹⁸O to Vienna Standard Mean Ocean Water (VSMOW) with standards of IAEA-NO-3, USGS-34, and USGS-35. The detection limit was approximately 1 μmol/L NO₃^ˉ with the analytical precession of 0.2‰ for δ¹⁵N-NO₃^ˉ and 0.4‰ for δ¹⁸O-NO_3.^ˉ The reproducibility of δ¹⁵NNO₃^ˉ and δ¹⁸O-NO₃^ˉ, based on long-run quality assurance tests, was ca. 0.22‰ and 0.51‰, respectively. The calibration of δ¹⁵N-NO₃^ˉ and δ¹⁸ONO₃^ˉ was done according to Casciotti and McIlvin (2007).

2.3 Data analysis

Apart from the Changjiang River runoff, the Changjiang estuary receives seawater from the Taiwan Warm Current that intrudes into the estuary along 50-m isobath and the Kuroshio Current from East China Sea surface (Zhu et al., 2004). Kuroshio waters are oligotrophic (Zhang et al., 2007) while the TWC waters contain regenerated nutrients. Consequently, a three-endmember mixing model was developed to estimate water sources in the CDW (identified by salinity and water temperature, Supplementary Fig.S1). Based on the identified water sources, calculation of the conservative distribution of NO₃^ˉ concentrations, isotopic compositions, as well as NH₄⁺ concentrations were conducted (more details in Yan et al., 2017). According to the difference in concentration or isotope compositions between the value derived from the model estimation and the observation, offsets for solute concentrations or isotope compositions were obtained. The positive offset for solute concentration indicates the production behavior, while a negative value is a mirror of removal. For isotope compositions, the positive reflects heavy isotope enrichment and vice versa. In the present study, the water parameters derived from the XLJ station (Fig. 1), i.e. the upstream of both transects, were used to represent the Changjiang River water. For TWC water, the NO₃^ˉ concentration and related isotope compositions, as well as NH₄⁺ concentration were from the bottom of the southeast sampling site in the same cruise (Fig. 1). For Kuroshio waters in the East China Sea, the NO₃^ˉ concentration was obtained from Wang et al. (2016) and NH₄⁺ concentration from Zhang et al. (2007). These values are outlined in Supplementary Table S1. Statistical analyses, such as regression and student's t-test, were carried out for the measurement of variables using Minitab 17.0 software (Minitab Inc., Pennsylvania State University). The significant level for all analyses was assumed to be α=0.05. The spatial distribution of parameters along each transect was plotted in Surfer 14.0 (Golden Software Inc., USA) and the dot plots were done in Sigmaplot 12.5 (Systat Software Inc., USA).

3 RESULT 3.1 Water chemistry parameters

Salinities of the cruise samples ranged from 0.11 to 34.5 on transect A (Fig. 2a). On transect B, a similar range (0.13-34.4) was observed (Fig. 2b). The salinity increased from the surface to bottom water. Additionally, the salinity difference between the surface and bottom increased from the river mouth to the outer plume, reaching ca. 8 due to strong stratification. The temperature of the Changjiang River surface water was approximately 26 ℃ and decreased to 18.9 ℃ at bottom water in the CDW outer plume (Fig. 2c). TSM concentration in the estuary water ranged from 9.0 to 75.6 mg/L. On transect A, TSM rapidly increased in the salinities of 5 to 10 regardless of seawater dilution. Afterwards, the TSM concentration sharply decreased, dropping to approximately 10 mg/L at the outermost site. Fluorescence intensities, an indicator of phytoplankton biomass, ranged from 0.34 to 12.8 μg/L in the estuary water. On both transects, fluorescence intensities were low when water salinities < 20. In the outer plume, fluorescence intensities rapidly increased and the watercolor darkened due to enrichment of phytoplankton, while the bottom water was fluorescence deficient (Fig. 2g-h). DO concentrations were ca. 6 mg/L at the inner sites (all depth). In the surface water, the DO concentration gradually increased and peaked at the sites with high fluorescence intensity. When water depth increased, the DO concentration in the saline CDW dropped (minimum 3.9 mg/L) (Fig. 2i-j).

3.2 N species and isotopes

NO₃^ˉ was the major DIN species, and the concentrations of NO₃^ˉ at the starting site (in the river channel) on transect A was approximately 119 μmol/L (Fig. 3a). Together with salinity increases, NO₃^ˉ decreased to ca. 5 μmol/L in the seawater surface (Fig. 4a). On transect B, NO₃^ˉ concentrations were ca. 108 μmol/L in river channels and dropped to 0.1 μmol/L in the outer plume (Fig. 3b). At the starting site, NO₃^ˉ concentrations in the surface and bottom waters were similar. The vertical difference in the NO₃^ˉ concentration increased in the outer plume with enhanced stratification. In contrast to NO₃^ˉ, NH₄⁺ concentration in Changjiang River surface water was much lower (0.1-0.6 μmol/L). Concentrations of NH₄⁺ increased to a maximum of 6.3 μmol/L at the higher salinity water (Fig. 3c-d).

For the distribution of NO₂^ˉ, its concentration on both transects was frequently below 1 μmol/L, indicating its instability (Fig. 3e-f). The NO₂^ˉ concentration showed a peak near the river mouth on transect A (Fig. 3e), while no significant enrichment was found on transect B (Fig. 3f). The DON level varied between 5.5 and 42.6 μmol/L. In the Changjiang River runoff (salinity < 2), DON concentration was < 20 μmol/L (Fig. 3g-h). In the surface water on transect A (3-5-m depth), an increase in DON concentration, followed with decreases, was found. A similar peak in DON concentration was observed at the bottom of saline water (Fig. 3g). On transect B, three DON concentration peaks were identified, mainly occurred in the bottom water (10-40-m depth) (Fig. 3h).

Along transect A, δ¹⁵N-NO₃^ˉ ranged from 6.5‰ to 10.0‰ (Fig. 5a). In the surface water, the elevation of δ¹⁵N-NO₃^ˉ along the salinity gradient from the river mouth to the offshore areas was observed (Fig. 4a). In the outer plume of the CDW, δ¹⁵N-NO₃^ˉ decreased in the deeper water. The range of δ¹⁸O-NO₃^ˉ fell into -0.4‰ to 6.1‰ on transect A and displayed a similar trend along the salinity gradient as δ¹⁵N-NO₃^ˉ (Fig. 5c). On transect B, δ¹⁵N-NO₃^ˉ in the Changjiang River runoff was 6.6‰. In the high salinity CDW, the surface water manifested a significant enrichment in heavy isotopes, reaching 17.4‰ for δ¹⁵N-NO₃^ˉ (Fig. 5b). Nearly identical enrichment was found for δ¹⁸O-NO₃^ˉ, peaking at 14.7‰ (Fig. 5d). Unlike the surface water, both isotope compositions decreased in the bottom layers from the estuary toward the ocean. On both transects, positive correlations between δ¹⁵N/ δ¹⁸O-NO₃^ˉ and fluorescence intensity were observed (Fig. 4c-d). For the correlation between isotope compositions and ln(NO₃^ˉ) (natural logarithm of NO₃^ˉ concentration), the surface and bottom water clearly separated into two clusters (Fig. 4e-f), suggesting a stronger isotope fractionation in the surface water and regeneration in the bottom water. In addition, the slope of linear regression based on the surface water samples was large on transect B (Fig. 4f).

3.3 Offsets for DIN and isotopes

The offset for NO₃^ˉ concentration, i.e. the difference in NO₃^ˉ concentration between the observed value and the values calculated from the three-endmember mixing model (Fig. 6a), was outlined along the salinity gradient (Fig. 6b, Supplementary Tables S2 & S3). On transect A, an increase in NO₃^ˉ concentration offset was found at inner stations (high turbidity water). In the outer plume, the offset for NO₃^ˉ concentration in the surface water was negative, suggesting an active consumption. The bottom water (high salinity) showed a patchy distribution and both increase and decrease in the offset values were obtained. Along transect B, offsets of NO₃^ˉ concentration in low salinity water (ca. 0-20) were frequently smaller than values obtained from transect B. For NH₄⁺ (Fig. 6c), positive offsets were found in the outer plume, while negative values were observed in the low salinity water, especially on transect A. For isotope compositions, the water with the salinity below 20 showed a slightly positive offset for both δ¹⁵N/δ¹⁸O-NO₃^ˉ (Fig. 6d-e). In high salinity water, a positive offset in the surface was found while it decreased dramatically in the bottom water. The linear correlation between δ¹⁵N-NO₃^ˉ and δ¹⁸O-NO₃^ˉ was significant (slope=1.22, R²=0.90, P < 0.05)(Fig. 6f).

4 DISCUSSION 4.1 NO₃^ˉ sources in the Changjiang River runoff

In the past forty years, NO₃^ˉ concentrations in the Changjiang River runoff have markedly increased. At Datong station (600-km distance upstream of the first site on transect A), the NO₃^ˉ concentration increased from ca. 30 to 120 μmol/L (cf. Fig. 7a; data from Dai et al., 2011), accounting for the major component in the riverine DIN inventory (Fig. 3a-b). In the current study, the NO₃^ˉ concentration at the start of both transects (A: 119 μmol/L; B: 108 μmol/L) are comparable with published records from Yan et al. (2017). Enhanced NO₃^ˉ inventory in the Changjiang River has been attributed to weak biological assimilation due to high turbidity (Zhang et al., 2007) and great contributions from multiple sources, including chemical fertilizer leakage from terrestrial aquifers, wastewater discharge, atmospheric deposition and degradation of terrestrial organic nitrogen (Li et al., 2010; Dai et al., 2011).

Given the isotope compositions in the Changjiang River runoff (outlined in Fig. 7b), both degradation of land-borne organic nitrogen (δ¹⁵N-NO₃^ˉ: 4.5‰ to 7.5‰; δ¹⁸O-NO₃^ˉ < 10‰) and leakage of chemical fertilizer (δ¹⁵N-NO₃^ˉ: 0 to 7‰; δ¹⁸O-NO₃^ˉ < 10‰) are likely to be important contributors (Chang et al., 2002; Li et al., 2010). In the Rajang River (Malaysia), the NO₃^ˉ concentration was 7.6 μmol/L and the major source of riverine NO₃^ˉ was the biological degradation of terrestrial organic matter due to the small population in its watershed (Jiang et al., 2019). In the St. Lawrence River (USA), another river with less anthropogenic disturbance, NO₃^ˉ concentration was 30.4 μmol/L, lower than that of Changjiang River (Thibodeau et al., 2013). Consequently, the proportion of NO₃^ˉ from organic matter decomposition may only account for a small portion in the riverine NO₃^ˉ inventory. Instead, fertilizer leakage, especially NH₄⁺/ urea fertilizer, from terrestrial aquifers to the Changjiang River can be significant due to the intensive usage in agriculture to feed the large population. NO₃^ˉ enrichments were also found in the Seine River (France), Zhujiang (Pearl) River (China) and River Thames (UK), where substantial chemical fertilizer was applied to support the large population in the watershed (Sebilo et al., 2006; Bowes et al., 2012; Cai et al., 2015). The temporal correlation between N fertilizer application amount in China and NO₃^ˉ concentration in the Changjiang River water from the 1980s supported this conclusion (Fig. 7a).

Interestingly, in the upstream of Changjiang River (Zhangjiagang), both δ¹⁵N-NO₃^ˉ and δ¹⁸O-NO₃^ˉ were 8.3‰ and 2.6‰, respectively (Li et al., 2010), which were comparable to sewage-borne NO₃^ˉ enriched rivers, such as the Beijiang (China; Chen et al., 2009) and the Potomac River (USA; Pennino et al., 2016). In the river mouth, isotope compositions dropped. Such a significant decrease is likely due to active nitrification in the river water (Hsiao et al., 2014), which increases NO₃^ˉ concentration while decreases δ¹⁵N/δ¹⁸O-NO₃^ˉ (Sanders et al., 2018). The branches of the Changjiang River at the downstream, such as the Huangpu River, cover several megacities (e.g. Shanghai, Suzhou, Hangzhou). Different with the Changjiang River water, the mean concentration of NH₄⁺ was similar to that of NO₃^ˉ in these branches (e.g. 1.4 mg/L for NH₄⁺; 1.1 mg/L for NO₃^ˉ in the Huangpu River, Yang et al., 2007), suggesting the sewage loading. Coupled with nitrification, the sewage related NO₃^ˉ input from adjacent cities also requires attention from managers and stakeholders. In Fig. 7c, a global comparison among several rivers subject to intensive human activities was made. Due to the relatively large discharge rate (9.24×1011 m³/a; Milliman and Farnsworth, 2011), NO₃^ˉdelivered from the Changjiang River is on the top of the list (Fig. 7c), even the yield of NO₃^ˉ in the entire watershed was much smaller than the Zhujiang River and the River Thames. Currently, the Chinese government puts substantial efforts into restoring the land cover in the watershed and reducing the application of fertilizer (Huang et al., 2008); therefore, a gradual reduction in the river-borne NO₃^ˉ flux can be expected.

4.2 NO₃^ˉ production in the CDW

In the estuary, apart from potential errors introduced from sampling and irregular water mixing (e.g. small eddies), NO₃^ˉ addition was found in salinity < 20 (existing from the Changjiang River channels to the high turbidity water) and the outer plume (salinity >26), because of the positive offsets referenced to the conservative mixing status (Fig. 6b). The increment in NO₃^ˉ concentration at low/moderate salinity is in line with the findings reported by Liu et al. (2009) and Yao et al. (2014). The input from Changjiang River branches, especially the Huangpu River, to the Changjiang estuary might be the first reason for the increase. However, the NO₃^ˉ concentration in the Huangpu River was lower than the level in the Changjiang River runoff (Yang et al., 2007). More importantly, compared with the Changjiang River discharge rate in summer ((6-7)×10⁴ m³/s), the Huangpu River loading was two orders of magnitude smaller (Zhu et al., 2018). Adding these together, the NO₃^ˉ concentration increase may not be the product of inputs from Changjiang River branches. In addition, the weak adsorption potential of NO₃^ˉ on particles in estuary water (Eyre, 1994) and limited variation in the adsorption-desorption balance from pH 5 to 9 (Zhang, 2007) indicate a minor influence from the adsorption-desorption process. Alternatively, the increase in NO₃^ˉ concentration may result from intensive nitrification, as highlighted in Fig. 8. On the one hand, the estuary directly receives riverine particles (Zhang et al., 2007) and microbes attached on particle surfaces. Given the high nitrification capability in the Changjiang River water, as aforementioned, it is not surprising to expect active nitrification on particle surface (Hsiao et al., 2014). On the other hand, seawater intrusion and desorption from the suspended particles, especially when salinity increases (Zhang, 2007), also add NH₄⁺ for nitrification. Furthermore, TSM concentration at the freshwater endmember reached 55 mg/L. The proportion of N in TSM, measured by Gao et al. (2012) at XLJ, ranged from 0.12% to 0.24%, indicating a potential N source for NO_3.^ˉ Moreover, an increase in DON content in the high turbidity water was observed, likely due to degradation from particles or desorption, which benefits NO₃^ˉ production. For electron acceptors, DO was nearly saturated due to intensive mixing, which supported the transformation of organic N to NH₄⁺ and eventually to NO_3.^ˉ Notably, the variation in δ¹⁵N in low salinity water was minor. In the Changjiang estuary, the reported mean δ¹⁵N in suspended particle matter measured at XLJ was 5.1‰ (Zhang et al., 2007) and 6.2 ‰ (Gao et al., 2012), which were in line with the δ¹⁵N-NO₃^ˉ value at the Changjiang River runoff (Fig. 4a-b). This similarity supported the reaction chain that includes organic nitrogen mineralization and nitrification, especially on particles (Fig. 8).

Intriguingly, between two transects, the concentration of NO₂^ˉ, the intermediate product in the nitrification reaction chain, differed in the low/ moderate salinity water. On transect A, the NO₂^ˉ concentration peak was observed near the river mouth (Fig. 3e), while an enrichment in NO₂^ˉ was not observed on transect B (Fig. 3f). Moreover, the peak value in positive NO₃^ˉ concentration offset obtained on transect A (ca. 32 μmol/L) was higher than the value on transect B (17 μmol/L). Apart from the bias introduced from the endmember selection, reasons for these differences may be linked to the TSM concentration. In particular, on transect A, TSM concentration markedly increased from the river water to brackish water (ca. salinity < 20) (Fig. 2e), and a large area of turbidity maximum zone (TMZ) was observed by a turbidity probe equipped on a towed system during the cruise, mainly due to relatively low discharge rate and strong tidal swash. The location of TMZ on transect A fits the historical observation (Yang et al., 2015). The marked increase in TSM concentration likely stimulates nitrification by providing substrates for remineralization (a positive link between TSM and NO₃^ˉ offsets in Supplementary Fig.S2). In addition, terrestrial particles in the TMZ were continuously suspended, prolonging the reaction time with oxidants, such as DO, in the water column. In contrast, a strong river plume rapidly injects and spreads in the East China Sea under the Coriolis force on transect B (Wu et al., 2010). Therefore, the typical TMZ along the transect was not captured during the cruise. Consequently, moderate NO₃^ˉ generation was observed because the particle-dependent nitrification may have been influenced by low concentrations of TSM.

Elevations in NO₃^ˉconcentration were also observed at the bottom of high salinity water (e.g. A9 and B6 in Supplementary Tables S2 & S3). Due to wide coverage of permeable sediments, it is possible to assume that submarine groundwater discharge may have contributed to this increase. Given the high concentration of NO₃^ˉ in groundwater, a small fraction could deeply change the NO₃^ˉ concentration in the receiving water (Moore, 2010). Generally, contaminated groundwater was enriched in δ¹⁸O-NO₃^ˉ due to continuous denitrification (15‰ to 90‰), as summarized by Xue et al. (2009). Those values are significantly higher than the records obtained in the present study, indicating that the high concentration of benthic NO₃^ˉ was not the result of groundwater injection. Alternatively, the increase in NO₃^ˉ concentration was triggered by bottom water nitrification (Fig. 8), which was supplied by the algae debris from the upper layer (Yan et al., 2017; Li et al., 2018). Compared to the nitrification rate in low salinity water, accumulation of NH₄⁺ in the outer plume was found (Fig. 7d). Nitrification is conducted by ammonia-oxidation archaea (AOA) and ammoniaoxidation bacteria (AOB) and the community structure in ocean water differs from that of terrestrial water (Martens-Habbena et al., 2009). Usually, the marine-derived AOA (the dominant contributor to nitrification) show a higher affinity to NH₄⁺ (Kuypers, 2017), indicating active nitrification at the high salinity CDW (Hsiao et al., 2014). Given the positive linkage between nitrification rate and NH₄⁺ concentration in the CDW (Wang et al., 2018), minor accumulation of NH₄⁺ should be observed, which is against the current observation. Moreover, the δ¹⁵NNO₃^ˉ fractionation was higher in the bottom of saline water than that in the low/moderate salinity CDW. In particular, the nitrification potential was introduced, which was defined as the slope between δ¹⁵N-NO₃^ˉ and f_nit/(1-f_nit)×ln(f_nit), where f_nit is the concentration ratio between NH₄⁺ and DIN (Wang et al., 2017). This potential in the water column is positively linked to nitrification produced NH₄⁺ in the benthic waters under the same substrate condition. It also helps the identification of reactant sources for nitrification process. Clearly, the bottom water at outer plume showed a significantly positive correlation between δ¹⁵N-NO₃^ˉ and f_nit/(1-f_nit)×ln(f_nit) (Fig. 7e); whereas the trend in low salinity water was weak. The reason for such distribution can be attributed to the supply of oxidants. Near the river mouth, the shallow depth and extensive mixing lead to an active exchange of DO between the atmosphere and the estuary water. In the outer plume, stratification and enhanced water depth hampered DO exchanges, as evidenced by high values of apparent oxygen utilization (AOU) in the bottom water (Fig. 7f). In the CDW, a linear correlation between δ¹⁵N-NO₃^ˉ and AOU was observed (Fig. 7f), indicating the significance of DO in benthic nitrification. Furthermore, the suspended particles in the TMZ might contain high levels of trace metal, such as iron and magnesium, due to the adsorption and flocculation (Zhu et al., 2018). These metals could also act as the electron acceptor during nitrification (Hsiao et al., 2014); while the concentration is usually low in pelagic debris and saline water (Zhu et al., 2018). In summary, the DO supply and reactive trace metals constrained the nitrification rate in the bottom of saline water though reaction potentials can be high.

4.3 NO₃^ˉ removal in the CDW

The NO₃^ˉ removal appeared in two regions in the outer plume (Fig. 8). In the surface water, turbidity decreased at seaward of the TMZ, which resulted from stratification, reduced turbulence and vertical mixing, benefiting the light availability for phytoplankton. It agrees with the observed phytoplankton bloom in the CDW front (Gao and Song, 2005). Hence, the NO₃^ˉ consumption coupled with utilization by primary producers was observed, as depicted by the significantly positive correlation (R²>0.54) between fluorescence intensity and δ¹⁵NNO₃^ˉ (Fig. 4c-d). In the present study, marked increases in NO₃^ˉ isotope compositions occurred at the 20 isohaline (Fig. 4a-b), which was identical to the value from Zhang et al. (2015) on δ³⁰Si enrichment in the estuary. Consequently, 20 isohaline can be treated as a starting boundary for the rapid uptake of nutrients by primary producers. As previously mentioned (Section 2.1), Skeletonema costatum was dominant in the outer plume on both transects. This assimilation increased isotope compositions, leading to strong correlations between NO₃^ˉ concentration and δ¹⁵N/ δ¹⁸O-NO₃^ˉ in surface water (Fig. 4e-f).

Though with identical phytoplankton species, NO₃^ˉ depletion (below 1 μmol/L) was observed on transect B (Fig. 4b). In comparison, approximately 5 μmol/L NO₃^ˉ still existed at the surface of transect A. Moreover, the correlation slopes of δ¹⁵N/δ¹⁸O-NO₃^ˉ to ln(NO₃^ˉ) in the surface water were higher on transect B (Fig. 4f), indicating strong isotopic fractionation during the consumption of NO_3.^ˉ The surface water is usually limited in denitrification carriers due to high DO content (Wong et al., 2014). Moreover, the fluorescence intensity and water residence time on both transects were similar. Adding these together, the difference in the stable isotope composition between transects A and B in this study likely results from strong stratification outside of the TMZ. The potential energy anomaly, an indicator of stratification, displays a peak on transect B (Supplementary Fig.S3), indicating a slow water exchange between the surface and bottom. Such strong stability is beneficial for the growth of phytoplankton in the surface water, which may have resulted in the continuous utilization of NO₃^ˉ and elevation of δ¹⁵N/δ¹⁸O-NO₃^ˉ in the CDW plume. Interestingly, the enhanced NO₃^ˉ utilization by Skeletonema costatum did not trigger a significant increase in biomass, as outlined by the similar CTD fluorescence intensity between transects A and B (Fig. 2g-h). One reason is that strong currents in the Changjiang River plume had flushed the phytoplankton offshore. More importantly, this phenomenon may be linked to density effect, which suggests that algae concentration could not continuously increase; whereas algae communities, especially for the nutrient-sensitive species, such as Skeletonema costatum (Li et al., 2018), could continuously accumulate nutrient from ambient environment into cells (luxury consumption). Collos et al. (1992) revealed that intracellular accumulation of NO₃^ˉ in Skeletonema costatum reached 17 mmol/L, i.e. several orders of magnitude higher than the ambient NO₃^ˉ concentration. The excess N may be stored in the vacuole of the algae as amino acids or protein (Ketchum, 1939). When N concentration in the ambient environment decreases, the stored N could sustain the cell function and algae biomass, leading to a continuous bloom in the CDW. In the benthic water, decrease in NO₃^ˉ concentration was also observed (Fig. 3a-b). The concentration decrease mainly results from denitrification since Skeletonema costatum cannot actively move to the bottom water due to lack of flagellates (Li et al., 2018). Denitrification could occur in both the water column and sediments (Yan et al., 2017). Usually, 2 mg/L DO concentration was assumed to be the highest level that denitrifiers could bear in seawater (Codispoti and Christensen, 1985). The DO concentration in the present study was higher than this threshold. Therefore, sediments are the major rector for denitrification. In the Changjiang estuary, high denitrification potential in benthic and intertidal sediments, especially in cohesive sediments (muddy sediment), has been reported (Song et al., 2013). Coupled with exchanges between sediment porewater and overlying water, a significant fraction of NO₃^ˉ was injected into sediments, especially the permeable sediments. Compared with the biological assimilation, denitrification in the sediment may not significantly alter stable isotope values (< 0.5‰ here) because of the completed transformation (Yan et al., 2017). Moreover, sediment denitrification might be coupled with nitrification (Fig. 8), effectively removing the produced NO₃^ˉ and hence preventing the transport of NO₃^ˉ from the deep water to surface water.

5 CONCLUSION

In the present study, the source of NO₃^ˉ in the Changjiang River runoff before the river-sea mixing was mainly attributed to inputs from chemical fertilizer (based on δ¹⁵N/δ¹⁸O-NO₃^ˉ values), indicating the importance of management on fertilizer application and reforestation in the watershed. Sewage-related NO₃^ˉ may also be important to the riverine inventory while nitrification alters the isotope compositions. In the coastal zone that receives the Changjiang River runoff, when salinities < 20, positive offsets for NO₃^ˉ concentration were obtained, likely due to the strong nitrification supplied by remineralization of N on particles and DON. The NO₃^ˉ production peaked at the TMZ because of resuspension of particles and high levels of DO. In the high salinity CDW (salinity >20), the surface water became a sink for NO₃^ˉ due to diatom assimilation. Consequently, 20 isohaline can be regarded as a boundary for the rapid uptake of nutrients by primary producers, benefiting the predication of algae bloom in coastal management. Compared with transect A, the strong stratification at the outer plume of transect B markedly stimulated phytoplankton utilization. The boost in the biological assimilation did not increase diatom biomass, therefore the assimilation may be linked to phytoplankton luxury consumption. In the bottom water, denitrification and nitrification coexisted. Nitrification was supplied by the degradation of algae debris, which increased NO₃^ˉ concentrations but decreased stable isotope compositions. Denitrification likely occurred at the water-sediment interface that continuously removed NO₃^ˉ, but without significant alternation on NO₃^ˉ due to completed transformation. The production and removal for NO₃^ˉ in the CDW should be taken into consideration by coastal managers and modeling workers. Furthermore, microorganism analyses and incubation experiments to quantitatively determine the dynamic between environmental drivers and NO₃^ˉaddition/removal rates are necessary.

6 DATA AVAILABILITY STATEMENT

The data used in the present study are available from the corresponding author upon request.

7 ACKNOWLEDGMENT

Technical support by Prof. GAO Yonghui, Prof. LI Bo, Ms. ZHENG Wei, Dr. ZHU Xunchi, Ms. QI Lijun, Mr. LIU Zhengbo, Ms. CAO Wanwan, Ms. JIANG Shuo, Mr. ZHU Kun, Mr. DAI Jinlong, and the R/V Zheyuke Ⅱ crew during the cruise are acknowledged. The authors also appreciate the great assistance from Prof. YU Zhiming at the Institute of Oceanology, Chinese Academy of Sciences for the assistance in stable isotope analyses. Thanks are also due to the editor and anonymous reviewers, whose comments helped to improve the original manuscript.

Supplementary material (Supplementary Tables S1–S4, Figs.S1–S3) is available in the online version of this article at https://doi.org/10.1007/s00343-020-0149-8.