Seasonal Sources and Cycling of Nitrogen Revealed by Stable Isotopes in the Northeastern Beibu Gulf, China

: Isotope measurements were performed on dissolved nitrate (NO 3− ) and ammonium (NH 4+ ) in the coastal waters of the northeastern Beibu Gulf, China, to investigate the seasonal nitrate sources and their biogeochemical processes, which are due to the rapid development of local industrialisation and urbanisation. The high N/P ratio observed in the coastal bay during both fall and spring suggests that P is a limiting nutrient, which in turn indicates that increasing P causes conditions favourable for algal blooms. Higher nutrient concentrations and δ 15 N-NO 3− and δ 15 N-NH 4+ values were found in the nearshore area in the fall, suggesting that nutrients originated mainly from land-based pollution. A Bayesian isotope mixing model was used to calculate the contribution of potential NO3- sources and the results showed that in the nearshore area, NO 3− originated mainly from manure and sewage (58%). In the spring, however, in addition to the impact of urban sewage effluents, the exchange of sediment and water was another important factor causing higher nutrient concentrations and positive NO 3− isotopes in the nearshore area. There were lower concentrations of nutrients and an increase in δ 15 N-NO 3− and δ 15 N-NH 4+ values in the offshore area in the fall, and the NO 3− loss in the surface water was mainly caused by the process of assimilation. However, the exchange of sediment and water was the dominant factor causing higher nutrient concentrations (except for NO 3− ) and positive dual nitrate isotopes but lower NO 3− concentration in the offshore area during the spring. Overall, isotope analysis of NO 3− and NH 4+ helps to illustrate the major sources of the former and their biological transformation in the northeastern Beibu Gulf.


Introduction
Coastal systems globally often receive various nutrient sources, which are primarily driven by anthropogenic activities, resulting in serious nutrient load in coastal waters [1][2][3][4][5][6]. Moreover, coastal areas play a significant role in the biogeochemical processing of nutrients [4]. Nitrogen is an important element in the regulation of marine primary productivity and climate change [7]. However, the global discharge of dissolved inorganic nitrogen (DIN) is predicted to increase to 47 × 10 6 tons per year in 2050, which is over double the estimated discharge rate in 1990 [4,8]). Increasing N loads in coastal systems have caused several environmental problems including eutrophication, hypoxia, and harmful algal blooms [3,[9][10][11][12][13][14]; therefore, they have become a focus of increasing public concern. Thus, tracing N sources and N recycling processes are important to designing effective management practices to protect coastal ecosystems.
The traditional methods for the determination of the nitrate contamination source are to identify by investigating the land use type of the contaminant area, combined with the analysis of hydrochemical characteristics. However, because of the diversity of the potential sources, which mixed the point and non-point sources, and the complex chemical, physical and biological transformation processes in the nitrate cycle, the results obtained by this traditional method are relatively rough [15]. Among the many methods available, the stable nitrogen isotope δ 15 N-NO3 − from various N pools and the oxygen isotope of nitrate (δ 18 O-NO3 − ) are powerful "tracers" and/or "integrators" in N cycle processes and have been used successfully to reveal N sources and their biogeochemical processes in aquatic systems [9,[16][17][18][19]. Various N sources can generally be distinguished by the range of δ 15 N-NO3 − and δ 18 O-NO3 − values they contain [20]. For example, nitrate from manure and sewage is characterised by relatively high δ 15 N-NO3 − values compared to that from fertiliser and atmospheric deposition; this is because of the volatilisation of 15 N-depleted ammonia [3,21]. Furthermore, the δ 18 O-NO3 − values from atmospheric nitrates are generally high (>50‰) relative to those from other sources (<25‰) [9,22]. A Bayesian isotope mixing model has previously been applied successfully to quantify the relative proportion of various N sources in a sample by using different isotopic signals from the former [23][24][25][26]. In addition, isotopic signals in nitrates also reflect biogeochemical N processes in aquatic systems, including nitrate assimilation, nitrification, denitrification, and N2 fixation [18,27,28]. However, biological processes of N often cause isotopic fractionation due to the preferential transformation or assimilation of lighter isotopes ( 14 N and 16 O) [9,20]. For example, nitrification can cause the formation of 15 N-depleted nitrate while denitrification can cause the enrichment of δ 15 N and δ 18 O by a ratio of 1:1 [18,28,29]. Similarly, assimilation can cause an increase in δ 15 N and δ 18 O in a residual nitrate pool [27], but the remineralisation of sinking organic N increases the percentage of light N in water [30]. The processes that occur in marine environments make it more complicated to determine N sources; therefore, a better understanding of N sources and the processes they undergo in marine ecosystems could be achieved by combining the variations in and distribution of the δ 15 N (and δ 18 O) of various N.
The Beibu Gulf (17.0°-22.0° N, 105.5°-110.0° E), located in the north-western part of the South China Sea (SCS), is a developing industrial area in southern China. Its high productivity and rich biological diversity have also made the Beibu Gulf a key fishing ground and source of fisheries products for that area [31,32]. However, the increasing population, intensification of agricultural activity, and rapid industrialisation and urbanisation that took place along the coast of Guangxi Province in recent decades have greatly increased N loading to this gulf [33][34][35][36][37][38][39][40]. Eutrophication fuelled by anthropogenic nutrient inputs is now a serious problem along the coast [33,34,38], which have led to many harmful algae bloom events in the coastal gulf and port [41]. Moreover, the index of nutritional status increased significantly over past decades, which was caused by the increased inputs of nutrients from the coastal Guangxi Beibu Gulf [37,38]. Particularly in the northeast of the gulf, the pollution level in that region is higher than that in the northwest due to the rapid development of industrialisation and urbanisation [34,42,43]. However, despite this, information on nitrate sources and biogeochemical transformations in the area is very limited.
To provide a comprehensive overview of nitrate sources and cycles in the northeastern Beibu Gulf, we investigated the concentrations of the nutrients NO3 − , NO2 − , NH4 + , and PO4 3− and the isotopes δ 15 N-NO3 − , δ 18 O-NO3 − , and δ 15 N-NH4 + in the seawater during different seasons (fall and spring). These data enable us to qualitatively characterise the main N sources in the coastal area and understand the factors controlling the distribution of the nutrients studied.

Study Area and Sampling
The Beibu Gulf is a semi-enclosed gulf with an area of approximately 130 × 10 3 km 2 and is located in the tropical and sub-tropical areas of southern China. It is influenced by the East Asian monsoon, which in turn is influenced by cold air from mainland China during the winter-spring season, with wind rising from the tropical ocean during the summer-autumn period. Freshwater discharged from coastal cities carry large quantities of anthropogenic pollutants to coastal waters [32,33,44]. For this study, two cruises were conducted in the north-eastern part of the Beibu Gulf, along the coast of Beihai City, in November (autumn) 2018 and March (spring) 2019 ( Figure 1). Samples were collected from the surface of the seawater (0.5 m) at 20 stations using a rosette sampler fitted with 10 L Niskin bottles (Figure 1).

Chemical Analysis
Temperature, salinity (S), and dissolved oxygen (DO) were measured on-site, with DO being measured using the Winkler titration method. The seawater samples were filtered through glass-fibre filters (Whatman, 0.7 μm, GF/F) to determine their chlorophyll a (Chl a) levels, and the filtered samples were stored at −20 °C before further processing and analysis. Measurements of nutrients concentrations and isotope ratios were conducted after the seawater samples were pre-filtered through pre-combustion (450 °C, 4 h) GF/F membranes (47 mm diameter, Whatman). The filtered samples were stored at −20 °C until analysis.
Chl a levels were determined using a spectrophotometer. NO3 − , NO2 − , and PO4 3− concentrations were determined using a San++ continuous flow analyser (Skalar, Breda, the Netherlands) and NH4 + concentrations were determined using spectrophotometry. The detection limits for NO3 − , NO2 − , and NH4 + , were 0.1 μmol L −1 each while that for PO4 3− was 0.02 μmol L −1 . For isotopic analysis, sulfamic acid was added to the samples to remove NO2 − , and the method used to determine the levels of nitrogen and oxygen isotopes of nitrate (δ 15 N-NO3 − and δ 18 O-NO3 − ) was modified from Mcilvin et al. [45]. Briefly, NO3 − was reduced to NO2 − by the addition of Cd and then reduced to N2O with the addition of NaN3 in an acetic buffer of pH 4-5. For isotopes of NH4 + , the NH4 + was first quantitatively oxidised to NO2 − with the addition of BrO − at pH 12. Following this, excess BrO − was consumed by NaAsO2 and the yield was verified via colorimetric NO2 − determination. Afterwards, the NO2 − was further reduced to N2O by the addition of a buffer solution of 1:1 NaN3 and acetic acid [46]. Finally, the N2O was purified and separated using TraceGas (Isoprime) and the isotope was determined using a GasBench II-MAT 253 (MAT 253 Plus, Thermo Scientific, United states). The IAEA-N3 international standard was utilised to calibrate the isotope values for δ 15 N-NO3 − and δ 18 O-NO3 − , and the international standards IAEA-N1, USGS 25, and USGS 26 were used to calibrate δ 15 N-NH4 + . The deviations from the standard in the δ 15 N-NO3 − , δ 18 O-NO3 − , and δ 15 N-NH4 + analyses were less than 0.2‰, 0.5‰, and 0.2‰, respectively. The reproducibility of duplicate samples for the δ 15 N-NO3 − , δ 18 O-NO3 − , and δ 15 N-NH4 + analyses were, respectively, less than 0.3‰ (average ± 0.1‰), 0.6‰ (average ± 0.3‰), and 0.3‰ for δ 15 N (average ± 0.1‰).

Mixing Model
A conservative salinity-based mixing model was used to calculate the mixing of NO3 − concentrations and dual nitrate isotopic values resulting from simple physical mixing of two endmembers [3,9,47,48]. The equations used are as follows: where f denotes the proportion of two-endmember water masses (with the subscripts 1 and 2 representing the two endmembers); S, N, and δ denote salinity, NO3 − concentration, and dual nitrate isotopic value (i.e., the parameters being mixed); and SMix, NMix, and δMix denote the measured values in a mixed water sample. Using the above equations, the following results are obtained.
These expressions show that NO3 − concentrations change in a linear manner along the mixing gradient under steady-state conditions, whereas salinity-based isotope mixing shows curvilinear behaviour (Equation (7)) that reflects the NO3 − concentration-based weighting of the isotopic proportions of endmembers. In this study, the salinity at station BH01 was the lowest measured although it was still high (S = 27.98) relative to freshwater, suggesting that these waters originated mainly from coastal water inputs. The highest salinity was recorded at station BH20, which is a reflection of the water from further out to sea being mixing with coastal waters as an input. Thus, the endmember parameters from stations BH01 and BH20 were chosen as the two endmembers for this study, as shown in Table 1. Table 1. Endmember parameters, including salinity, nutrient concentrations and dual nitrate isotope values, used in the two-endmember mixing model in Equations (5)(6)(7). Station BH01 represents water at the head of nearshore areas and station BH20 represents water carried into coastal areas from the water further out to sea.

Quantification of the Mixing Model
A Bayesian stable isotope mixing model was used to quantify the proportional contributions of each nitrate source along the coastal area. The model was run in the stable isotope analysis in the R (SIAR) software package. The framework of the model is as follows: where Xij and Sjk are the dual-isotope value of nitrate (δ 15 N-NO3 − and δ 18 O-NO3) of a mixed sample and the isotope value of NO3 − sources, respectively; Sjk is normally distributed with mean μjk and standard deviation ωjk; Pk denotes the contribution of NO3 − sources; cjk is the factor of fractionation for δ 15 N-NO3 − and δ 18 O-NO3 in NO3 − sources, normally distributed with mean λjk and standard deviation τjk; and εjk denotes the residual error of the additional unquantified change between samples, normally distributed with mean 0 and standard deviation σj. A detailed description of this model was reported by Blake et al. [16], Moore and Semmens [24], Xue et al. [25], and Zhang et al. [26]. However, it is difficult to quantify the source of nitrate by using this model if the nitrate biogeochemical process occurs, because the biological processes can cause isotope fractionation. Thus, this model is only used in the absence of nitrate biological processes.

Chemical Parameters
The seasonal distribution characteristic of physicochemical parameters were presented in Figure 2. Salinity and temperature increased seaward, i.e., from the nearshore to the offshore areas, ranging overall from 27.98 to 31.45 and 23.8 to 26.00 °C during the fall, and from 29.31 to 32.32 and 20.8 to 24.7 °C during the spring, respectively; the minimum values were at the nearshore area during both two seasons (Figure 2a-d). A slightly higher Chl a level was found in the spring (average of 1.43 μg L −1 ) compared to that in the fall (average of 1.31μg L −1 ). A relatively low Chl a level was found in the nearshore area whereas a higher Chl a level was found in the offshore area during both two seasons (Figure 2e,f). The DO level during the spring (ranged from 7.22 to 8.18, average of 7.45 mg L −1 ) was higher than that during the fall (ranged from 5.94 to 7.73 mg L −1 , average of 6.71 mg L −1 ). The minimum DO (5.94 mg L −1 ) was recorded at the nearshore area at the top of the gulf during the fall, while the DO level was relatively uniform during the spring. The level of total suspended particulate matter (TSPM) during the spring (average of 24.29 mg L −1 ) was obviously higher than that in the fall (average of 16.05 mg L −1 ), particularly in the offshore area (Figure 2i,j). Moreover, a higher TSPM level was found in the nearshore area during the spring (Figure 2j). Figure 3 shows that, in contrast to the distributions of physiochemical parameters, the concentrations of NO3 − , NO2 − , NH4 + , and PO4 3− decreased seaward-with high concentrations in the nearshore area and low concentrations in the offshore area. Generally, the concentrations of NO3 − (average of 6.59 μmol L −1 ), NO2 − (average of 1.66 μmol L −1 ), NH4 + (average of 4.12 μmol L −1 ), and PO4 3− (average of 0.11 μmol L −1 ) in the spring were higher than that in the fall (average of 2.80 μmol L −1 , 0.45 μmol L −1 , 1.24 μmol L −1 and 0.05 μmol L −1 for NO3 − , NO2 − , NH4 + , and PO4 3− , respectively). Among the types of dissolved inorganic nitrogen (DIN), the concentration of NO3 − was the highest, followed by NH4 + , and the lowest by NO2 − . The ratio of N/P ( ranged from 34 to 206 (with an average of 79) during the fall and from 50 to 374 (average of 152) during the spring throughout the coastal area, and they were significantly higher than the Redfield ratio (i.e., the ratio of these nutrients utilised by marine phytoplankton) of 16. This suggested that P could be largely limited to producing phytoplankton along the coastal areas; this is consistent with the results of a previous study [34,38].

Isotopic Composition
During the fall, the δ 15 N-NO3 − , δ 18 Figures 2 and 3 show that there are clear differences in the hydrographic characteristics of the nearshore and offshore areas; the former had lower temperatures and salinity levels with higher nutrient concentrations, while higher temperatures and salinity and lower nutrient concentrations were recorded in the latter during both fall and spring. We believe that these water conditions in the nearshore area are mainly influenced by the diluted freshwater inputs from a nearby city. The distribution pattern seen in our study is similar to one seen in a previous study of that coastal city; the latter found that there were high nutrient concentrations in estuaries and coastal waters but low concentrations of the same further offshore [34,38]. We classified the sample sites into two zones, based on the distribution of hydrographic characteristics of the coastal area, to enable discussion of nutrient sources and dynamics. Figure 1 shows that zone 1, the nearshore area that included stations BH01 to BH08, B05, B02 and B08, was characterised by the influence of low temperature and salinity level and high nutrient concentrations from local terrestrial water discharge. The same figure shows that zone 2, the offshore area that included stations from BH09 to BH20, was characterised by high salinity and low nutrient concentrations.

Nutrient Sources in the Nearshore Area
In the nearshore area, the concentrations of DIN  [34]. The dominant proportion of NO3-N in DIN varied from 50% to 71% (average of 64%) in the fall and from 53% to 76% (average of 65%) in the spring; followed by NH4-N, which varied from 16% to 44% (average of 27%) in the fall and from 15% to 32 (average of 22%) in the spring, and NO2-N, which varied from 7% to 15% (average of 9%) in the fall and from 3% to 18% (average of 12%) in the spring. Significantly low PO4 3− concentrations and high N/P ratios were recorded in the coastal area during both two seasons. However, the same measurements were significantly higher along the entirety of the coast of the Beibu Gulf (0.40 μmol L −1 PO4 3− , N/P ratio of 11) during the same period [34], suggesting that P acted as a limiting nutrient in this ecosystem and that the increase in P would be favourable for the proliferation of phytoplankton. In addition, there is a counterwind current (a northeastward current) in the bottom layer in the Beibu Gulf during the winter boreal, and there will be an upwelling when it comes to the coastal Beibu Gulf [49]. Moreover, influenced by the southwestward current off the western coast of Guangdong, the high concentration of nutrients from the Pearl River Estuary may be brought into the Beibu Gulf via the westward current in the Qiongzhou Strait [23,50]. These processes may carry high concentrations of PO4 3 , resulting in algal blooms due to P limits. Many red tides and phytoplankton bloom events, which are mainly caused by seasonal eutrophication, have been reported on the coast of Beihai City [51,52]. Moreover, the index of the nutritional status of those waters has increased significantly in recent years, and this was caused by the increased amounts of inorganic N and P present along the coastal area of the Beibu Gulf [34,37,38]. As both the highest nutrient levels and lowest salinity values all appeared in the nearshore area, we speculated that land-based pollution discharge from coastal cities may have been responsible for the former. This shows why it is important to trace nutrient biological processes and their sources in coastal areas.
Positive δ 15 N-NO3 − and δ 15 N-NH4 + values and higher nutrient concentrations were observed in the nearshore area during the fall, while higher nutrient concentrations and higher δ 15 N-NO3 − values were found during the spring (Figures 3 and 4). Among the biological processes, the nitrification of NH4 + could increase NO3 − concentrations and generate high δ 15 N-NH4 + values [9]. During the process of nitrification, the microorganisms preferentially used light N in the presence of excess NH4 + , which would lead to heavier δ 15 N-NH4 + values in the residual NH4 + pool and add δ 15 N-NO3 − depleted N to the NO3 − pool [17,18], which contrasted with our results in the fall. This difference suggests that nitrification is unlikely to have caused the significantly positive δ 15 N-NO3 − values during the fall in the nearshore area and that other processes or sources may therefore exist. Such high δ 15 N-NO3 − can be caused by denitrification and/or assimilation in which the lighter isotope δ 14 N is preferentially consumed [18,29]. Denitrification generally leads to losses of N to the atmosphere as N2 or N2O [53]. Previous studies have reported evidence for significant nitrate loss in the water in the study area, and this loss could be directly influenced by denitrification in the sediments and pore water [9,17,54]. However, high DO values (which ranged from 5.94 to 6.96 mg L −1 with an average of 6.37 mg L −1 ) were observed throughout the nearshore area during the sampling period; such values do not favour the process of denitrification in the water column [9,17]. The process of elimination thus suggests that the assimilation of phytoplankton seems to be a likely explanation for the observations. Furthermore, a relatively high Chl a level (which ranged from 3.66 to 5.36 μg L −1 ) was observed in the nearshore area (Figure 2d), and an increase in phytoplankton would have resulted in a large quantity of nutrients being consumed. If assimilation was the major control on isotopic composition, the assimilative uptake along would cause δ 15 N-NO3 − and δ 18 O-NO3 − to increase at a rate of 1 [3,27]; however, the observed increase in the dual nitrate isotope is defined by a slope <1 and a lack of correlation between these isotopes in the nearshore area ( Figure 5). Such deviations suggest that phytoplankton assimilation is unlikely to be the dominant process increasing the isotopic composition of NO3 − , although the high δ 15 N-NH4 + values could likely have been caused by this process. Given sufficient NH4 + in the water along the coast, phytoplankton usually preferentially uptake this nutrient over nitrate due to the overall lower energy cost of the former process [9]. In this study, a relatively high NH4 + concentration was recorded in the nearshore area and residual NH4 + in the seawater was still not completely consumed by the phytoplankton. This suggested that the phytoplankton continued to assimilate NH4 + but not NO3 − in the nearshore area, causing the elevated δ 15 N-NH4 + levels therein. Moreover, the relationship between NO3 − , NH4 + , and salinity could further confirm that NO3 − was a mixture of coastal discharge water and seawater although this was not true for NH4 + in the nearshore area ( Figure 6). Therefore, biological processes may be responsible for the weaker relationship between NH4 + and salinity in the nearshore area, suggesting that the high δ 15 N-NO3 − value may not have been caused by assimilation during the fall, although a high Chl a level was recorded in the nearshore area.
In the spring, higher nutrients were found in the nearshore area compared with that in the fall (Figure 3), which was similar to the distribution pattern of the coastal Beibu Gulf [34]. This suggested that there may be additional nutrients input in the nearshore area during spring. Land input may be an important factor causing the increase of nutrients in spring [34]. However, the rainfall in the spring (March and April, 251.9 mm) was relatively higher than that in the fall (October and November, 55.1 mm), and heavy rainfall could bring land-based pollutants into the nearshore environment. In addition, wind forcing can induce resuspension of sediments [33]. The surface water temperature in the spring (average of 22.9℃) was lower than that in the fall (average of 25.1℃), indicating that the water mixing was more uniform in the spring, which could re-suspend the bottom matter to the surface water. These processes increased the TSPM level, resulting in the TSPM and nutrient levels in the spring being higher than that in the fall. Similar to the fall, the isotope values (average of 9.94‰ for δ 15 N-NO3 − and 14.48‰ for δ 18 O-NO3 − ) were still high in the nearshore area during the spring. There could be several factors for higher dual nitrate isotope patterns observed, including assimilation, denitrification and intense physical sediment-water interaction, or other sources [55]. However, the influence of water denitrification could be ruled out as the factor of the higher dual nitrate isotope due to the high DO level (average of 7.42 mg L −1 ) in the nearshore area during spring. In addition, DIN assimilation by phytoplankton is unlikely to be the cause of the positive nitrate isotope in the spring, because a significantly low Chl a level (most below 1 μg L −1 ) was observed in the nearshore area. Thus, the physical sediment-water interaction may be responsible for the positive dual nitrate isotope in the nearshore area during spring. Previous studies suggested that the advection flux returning nitrate from the sediment to the overlying water, the δ 15 N-NO3 − and δ 18 O-NO3 − values of pore water are substantially high, which reflects more closely the enzyme level nitrate isotopic fractionation by denitrification [56,57]. Such a dynamic environment was previously reported in the Pearl River Estuary, particularly during the winter monsoon with severe mixing of water columns, and this nitrate exchange between sediments and water would finally lead to elevated dual nitrate isotope values but decreased nitrate concentration in the water column because of denitrificationinduced isotopic enrichment in sediments [55]. However, nitrate concentrations were still high in the nearshore, indicating that other sources may occur in spring. The positive dual nitrate isotope and high nutrient levels in the nearshore area during the spring are possibly associated with municipal sewage effluents from Beihai City. Since nitrate from sewage is characterised by relatively high δ 15 N-NO3 − (7‰ to 25‰) because of the volatilisation of depleted δ 15 N ammonia produced from human and animal waste, and this process leaves the residual nitrate enriched in isotope [3,21].
As NO3 − was less affected by the biological processes in the fall, the isotopic characteristics may provide fingerprint information on the various sources that contributed to the mixture. As shown in Figure 5, the overall ranges of δ 15 N-NO3 − and δ 18 O-NO3 − in the nearshore area (which ranged from 5.8 to 12.9‰ for δ 15 N-NO3 − and from 8.8 to 14.8‰ for δ 18 O-NO3 − ) would seem to suggest-according to a classical dual nitrate isotope approach [21]-that manure and sewage (M&S) and soil N might be the dominant nitrate sources in this reach. Previous studies have proposed sewage effluent as a significant nitrate source in coastal areas [9,16,55,58]. Furthermore, in addition to the sources from M&S and soil N, the high δ 18 O-NO3 − values could be partly influenced by atmospheric deposition and synthetic nitrate fertiliser (SNF) sources. However, the contribution from SNF can be eliminated readily as it only accounts for <2% of all fertiliser applied in China [17]. Moreover, the nitrogen fertiliser (NF) that is heavily applied as part of coastal agriculture may be running off into the water in that area [33]. Therefore, a Bayesian mixing model was applied to quantify the proportional contribution of the nitrate sources in the nearshore area. The δ 15 N-NO3 − and δ 18 O-NO3 − values of the potential sources are presented in Table 2. The results from the application of the model are presented in Figure 7; M&S was the predominant source of nitrate in the nearshore area (58%); it was followed by soil N (17%), NP (15%), and NF (10%). Thus, the high contribution of M&S may be responsible for the high δ 15 N-NO3 − and δ 15 N-NH4 + values in the water of nearshore areas. This is because the isotope fingerprint of M&S is characterised by high values due to the volatilisation of 15 Ndepleted NH3 produced from human and animal waste [3,9]. This result is similar to that from the western coast of Guangdong Province, where nitrate sources originated mainly from local sewage inputs (53%) due to the rapid development of the local economy [16]. However, although NF is heavily applied in the agricultural practices of coastal cities [33], its contribution only accounts for 10% of the total used in the coastal area; this may be due to a lack of river input in the latter. The NF applied in catchment-based agriculture is transported to coastal waters by rivers [9,17].

Nitrate Biogeochemical Process in the Offshore Area
In the fall, the chemical compositions were characterised by high salinity values and significantly low nutrient concentrations in the offshore area (Figures 2 and 3). The δ 15 N-NO3 − and δ 15 N-NH4 + values ranged from 5.6‰ to 8.8‰ and 8.8‰ to 15.3‰ in the fall, respectively, and showed an increasing trend in the offshore area; however, these values were generally lower than those in the nearshore area ( Figure 4). In addition, the distribution of NO3 − concentrations in the offshore area was not conservative, their deviations from the calculated mixing line (Figure 8a). The highest NO3 − concentrations fell below the mixing line, suggesting that NO3 − loss or consumption occurred in the surface water in the offshore area and that the processes of denitrification and/or assimilation may be responsible. However, denitrification can be ruled out due to the significantly high levels of DO in the waters of the offshore area ( Figure 2c). Thus, the increase in δ 15 N-NO3 − and δ 15 N-NH4 + values could likely be caused by assimilation. DIN assimilation by phytoplankton could cause isotopic enrichment of residual NO3 − and NH4 + [9], while the parallel increase in δ 15 N-NO3 − and δ 18 O-NO3 − values further confirm that assimilation occurred in the surface waters of the offshore area ( Figure 8). Therefore, because assimilation could lead to an increase in δ 15 N-NO3 − and δ 18 O-NO3 − in the residual nitrate pool, the 18 ε: 15 ε for this process is ~1 [27]. However, the slope of their linear relationship is ~1.56, and most of the δ 15 N-NO3 − and δ 18 O-NO3 − values are above the assimilation line (1:1 line) (Figure 8). This means that there may be other sources simultaneously contributing to nutrient levels in the offshore area. Atmospheric deposition may be the source of nitrate responsible for values above the assimilation line, considering that the extremely high δ 18 O-NO3 − values were also observed in the surface waters of the offshore area (Figure 4d). Neither phytoplankton assimilation nor mixing with other water masses from surrounding water would be significant enough to cause δ 18 O-NO3 − values greater than 10‰ [9,55]. Thus, such high δ 18 O-NO3 − values are likely to be sourced from atmospheric deposition because the δ 18 O-NO3 − values measured from atmospheric deposition in the South China Sea is >50‰ on average [22,60]. A simple steady-state isotope mass balance model was used to quantify the contribution of NO3 − from atmospheric deposition in the offshore area. This model is based on nitrate anomalies and thus deviates from conservative mixing. Therefore, because of this, we assumed that the processes of assimilation and atmospheric deposition are responsible for the nitrate isotope anomalies. The equations used in our calculations are as follows: 15 Natmosphere × Natmosphere/(Natmosphere + Nmixing) + 15 εassimilation × fNO3, (9) △δ 18 O =δ 18 Oatmosphere × Natmosphere/(Natmosphere + Nmixing) + 18 εassmilation × fNO3 , where △δ 15 N and △δ 18 O denote the dual nitrate isotope anomalies relative to the expected mixing values. Assimilation increases in δ 15 N-NO3 − and δ 18 O-NO3 − values ( 15 ε = 18 ε) [27]. Therefore, according to Equations (9) and (10): where fatmosphere denotes the proportion of nitrate contributed by atmospheric deposition. More detailed information on this model can be found in Ye et al. [9]. In this model, the δ 15 N-NO3 − and δ 18 O-NO3 − values of atmospheric deposition (0.8 for δ 15 N-NO3 − and 52.4 for δ 18 O-NO3 − values) observed in the neighbourhood of the city in Zhanjiang were used [60]. Our estimates suggest that the proportion of nitrate from atmospheric deposition in the offshore area ranged from 1% to 11% with an average of 6%. The importance of atmospheric NO3 − deposition for new production in the SCS has been suggested previously [9,16,22,23]. The nitrate source from atmospheric deposition in the nearshore area of the gulf was lower than that in the high-salinity waters in the outer area of PRE (17%) during winter [9]. This would occur particularly during the dry seasons (autumn and winter) when prevailing northern winds can carry anthropogenic pollutants such as those from industrial activities and fossil fuel combustion from mainland China to the SCS [9,60]. However, during transport, some atmospheric nutrients would gradually be deposited out of the air; this may be responsible for the slightly low contribution of atmospheric NO3 − deposition in the offshore area (6%) compared to in the nearshore area (15%). In the spring, higher concentrations (except for NO3 − ) and positive δ 15 N-NO3 − and δ 18 O-NO3 − were still found in the offshore area (Figures 3 and 4). Denitrification in the water column can also be ruled out as high DO levels (average of 7.46 mg L −1 ) were observed in the offshore area. Phytoplankton assimilation is also unlikely responsible for the In addition, higher NH4 + concentrations were found in the offshore area, and phytoplankton would preferentially uptake NH4 + over NO3 − under sufficient NH4 + in the water column [9]. Since the terrestrial input has relatively little effect on the nutrients in the offshore area, the positive dual nitrate isotopes may be caused by the intense interaction between sediments and seawater. In this season, with wind-induced mixing and tidal pumping, tending to uniform mixing of the offshore seawater, which accelerated the exchange of sediment and water. This also resulted in the increase of TSPM concentration in the offshore area (average of 21.12 mg L −1 in spring, and higher than that of 15.59 mg L −1 in fall). Such a dynamic process would cause bidirectional exchange of solutes, including nitrate, between sediment pore water and the overlying water column, and causing elevated dual nitrate isotope values while decreased nitrate concentration in the water column because isotopic enrichment caused by denitrification in sediments [9]. Thus, the intense exchange of sediments and water in the offshore area was responsible for the higher concentration of other nutrients and the lower concentration of NO3 − .

Conclusions
Our study on stable isotopes of dissolved NO3 − and NH4 + provided detailed information on seasonal nitrate sources and their biogeochemistry in the coastal area of the northeast Beibu Gulf. A high N/P ratio was recorded in the coastal bay during both fall and spring, suggesting that P was a limiting nutrient and indicating that increasing P would be favourable for algal blooms. In the nearshore area, the high concentrations of nutrients and δ 15 N-NO3 − and δ 15 N-NH4 + values in the fall suggested that the former originated mainly from terrestrial pollution. A sufficient amount of NH4 + that could support biological processes was observed in the nearshore area owing to intensive human activities in that region. Therefore, the weak effect of biological processes on nitrate, and the characteristics of its isotopes, may provide fingerprint information in water mixed from various sources in the fall. The contributions of various potential NO3 − sources were calculated using a Bayesian isotope mixing model, and its results showed that manure and sewage (58%) were the predominant sources of nitrate in the nearshore area in the fall, followed by soil N (17%), nitrogen in precipitation (15%), and nitrogen fertiliser (10%). In the spring, however, in addition to the impact of urban sewage effluents, the exchange of sediment and water was also responsible for causing higher nutrient concentrations and dual nitrate isotope values in the nearshore area. The lower nutrient concentrations and increased δ 15 N-NO3 − and δ 15 N-NH4 + values observed in the offshore area in fall, suggested that the process of assimilation was responsible for the NO3 − loss; however, the greater enrichment of δ 18 O-NO3 − than δ 15 N-NO3 − during the sampling period suggests that atmospheric deposition may also contribute to NO3 − concentrations. In the spring, the exchange of sediment and water was the dominant factor causing higher other nutrient concentrations and positive dual nitrate isotope but lower NO3 − concentration in the offshore area.
Author Contributions: Q.L. analyzed the data, and wrote the manuscript; J.G. conceived and designed the study, and collaborated in discussing the manuscript and modified the manuscript; G.L., Y.S., Q.S. and C.C. performed sample collection and contributed to the experiment and participated in the taking of measurements; F.C. helped perform the statistical analysis. All authors have read and agreed to the published version of the manuscript.

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The study did not report any data.