Zhanjiang Bay, surrounded by the developing coastal city of Zhanjiang in Guangdong Province, South China, has a population of approximately 7.3 million (Chen et al., 2022b). The bay spans an area of about 490 km², with depths ranging from 2 m in the upper bay to 32 m in the lower bay (Fig. 1). Zhanjiang Bay is geographically and hydrodynamically complex, mainly influenced by the local discharge from the Suixi River at the top of the upper bay and the intrusion of high-salinity water from the SCS through a narrow channel at the bay mouth (~2 km wide) (Lao et al., 2022b). Zhanjiang Bay is famous for its mariculture industry, which includes cage culturing and oyster farming. Oyster culture is mainly located in the upper bay, while cage culture is primarily in the lower bay (Chen et al., 2022b). However, due to the influence of intensive human activities and weak hydrodynamic conditions, seawater pollution in the bay, such as eutrophication, is gradually intensifying (He et al., 2023). Affected by the East Asian Monsoon, the annual rainfall in this region is approximately 1731 mm, with over 85% of the annual rainfall occurring in the rainy season (April to October) and only 15% of the annual rainfall in the dry season (November to March of the following year) (Chen et al., 2019, 2021).

Two cruises were conducted in Zhanjiang Bay, China in September 2017 and March 2018 (Fig. 1). A total of 26 seawater sampling stations were established from the top of the upper bay to the outer bay during these two periods (Fig. 1). Seawater samples were collected using 10-L Niskin bottles. The temperature, salinity, and depth of the water were determined onsite using an RBR Maestro multiparameter water quality monitor (RBRmaestro3, RBR, Canada). For the collection of Chl-a, PN, and δ¹⁵N-PN samples, approximately 1000 mL of seawater was filtered immediately after sampling through precombustion (450℃, 4 h) GF/F (glass fiber filters, 47-mm diameter) and then frozen at −20℃ until further analysis. For the samples of nutrients and nitrate dual isotopes (δ¹⁵N-${\mathrm{NO}}_3^- $ and δ¹⁸O-${\mathrm{NO}}_3^- $), the seawater was filtered through a cellulose acetate membrane, and the filtrate was transferred into acid-washed polyethylene bottle and stored at −20℃ for laboratory analysis.

Dissolved oxygen (DO) was measured using Winkler titration with a precision of 0.07 mg/L. Chl-a samples were extracted using 90% acetone, and the levels were determined using the fluorometric method. Nutrient samples (${\mathrm{NO}}_3^- $, ${\mathrm{NO}}_2^- $, ${\mathrm{NH}}_4^+ $, ${\mathrm{PO}}_4^{3-} $, and ${\mathrm{SiO}}_3^{2-} $) were analyzed using a San++ continuous flow analyzer (Skalar, Netherlands).

The PN and δ¹⁵N-PN samples were tightly packed into a tin cup and acidified with concentrated HCl vapor for at least two days to remove carbonate. Subsequently, the HCl was removed in a caustic soda dryer for 48 h (Lao et al., 2023a). The samples were then analyzed using an elemental isotope ratio mass spectrometer (EA Isolink-253 Plus, Thermo Fisher Scientific, USA). Atmospheric N₂ references were utilized for δ¹⁵N-PN measurements. The average standard deviation of PN was ±0.1%, and the precision for δ¹⁵N-PN was ±0.2‰.

The analysis of δ¹⁵N-${\mathrm{NO}}_3^- $ and δ¹⁸O-${\mathrm{NO}}_3^- $ was conducted using a chemical method (cadmium-azide) (McIlvin and Altabet, 2005). Before determination, the nitrite in the seawater was removed using sulfamic acid (Granger and Sigman, 2009). The nitrate in the sample was then reduced to nitrite by Cd and further reduced to nitrous oxide using sodium azide buffered with acetic acid at pH 4–5. The TraceGas was then used to purify and separate the nitrous oxide, and the compositions of δ¹⁵N-${\mathrm{NO}}_3^- $ and δ¹⁸O-${\mathrm{NO}}_3^- $ were determined by a GasBench II-MAT 253 (MAT 253 Plus, Thermo Scientific, United States). Isotope standard materials, including USGS34 (δ¹⁵N = −1.8‰, δ¹⁸O = −27.9‰), IAEA-NO3 (δ¹⁵N = 4.7‰, δ¹⁸O = 25.6‰), and USGS32 (δ¹⁵N = 180‰, δ¹⁸O = 25.7‰), were used in this study. USGS32 and USGS34 were used to create nitrogen and oxygen isotope standard curves, and IAEA-NO3 was used for quality control monitoring. The reproducibility of duplicates for δ¹⁵N-${\mathrm{NO}}_3^- $ and δ¹⁸O-${\mathrm{NO}}_3^- $ was <0.3‰ and <0.6‰, respectively, with mean differences of ±0.1‰ and ±0.3‰, indicating high precision in the measurements.

In this study, a conservative mixing model was used to calculate the mixing between diluted riverine water and seawater in the bay to reveal the behavior of various N species along the salinity gradient (Fry, 2002; Lao et al., 2019; Ye et al., 2016). The formula is as follows:

where N_d and N_m denote the concentrations of various N species in the diluted water endmember and marine endmember, respectively, and q denotes the contribution of diluted water to each sample, calculated from salinity as follows:

where S_mix, S_d, and S_m denote the salinity of the sample, diluted water end-member, and marine end-member, respectively.

Additionally, the isotopic composition (δ_mix) of various N species in a sample is the concentration-weighted mean of values derived from diluted water and marine end-members during physical mixing. The formula is as follows:

where δ_d and δ_m denote the isotopic values of various N species in the diluted water and marine end-members, respectively.

In this study, the diluted water end-member was selected from the lowest-salinity water at the top of the upper bay (Station Z25), and the marine end-member was selected from the highest-salinity water in the outer bay (Stations Z13 and Z14). Endmember values are listed in Table 1.

The seasonal distributions of the physicochemical parameters are illustrated in Fig. 2. Seawater temperature was higher in the rainy season (29.9–32.9℃) and lower in the dry season (22.2–29.4℃). Significant variations in salinity were observed in the bay (t-test, p<0.01), with lower values in the rainy season (15.04–27.89) and higher values in the dry season (20.59–30.46). There was a remarkable salinity gradient from the upper bay to the outer bay, with low salinity occurring at the top of the upper bay, while high salinity was observed in the outer bay during both seasons. The DO values in the rainy season (3.79–7.33 mg/L) were much lower than those in the dry season (6.99–10.50 mg/L). Low DO values were recorded at the top of the upper layer during the rainy season. However, the Chl-a concentration in the rainy season (2.09–22.33 μg/L) was remarkably higher than that in the dry season (1.52–11.37 μg/L).

The seasonal distributions of the dissolved nutrients (${\mathrm{NO}}_3^- $, ${\mathrm{NO}}_2^- $, ${{\rm {NH}}_4^+} $, ${{\rm {PO}}_4^{3-}} $, and ${{\rm {SiO}}_3^{2-}} $) and PN are presented in Fig. 3. They varied significantly between the seasons. Generally, high concentrations of dissolved nutrients were observed in the upper bay, and the concentrations gradually decreased towards the outer bay. The concentrations of dissolved nutrients [(15.34 ± 9.60) μmol/L for ${\mathrm{NO}}_3^- $, (2.87 ± 1.40) μmol/L for ${\mathrm{NO}}_2^- $, (5.87 ± 2.28) μmol/L for ${{\rm {NH}}_4^+} $, (4.07 ± 1.42) μmol/L for ${{\rm {PO}}_4^{3-}} $, and (20.07 ± 12.25) μmol/L for ${{\rm {SiO}}_3^{2-}} $] and PN [(0.075 ± 0.022) mg/L] in the rainy season were higher than those in the dry season [(6.83 ± 14.93) μmol/L for ${\mathrm{NO}}_3^- $, (0.74 ± 1.21) μmol/L for ${\mathrm{NO}}_2^- $, (0.03 ± 0.03) μmol/L for ${{\rm {NH}}_4^+} $, (1.28 ± 0.78) μmol/L for ${{\rm {PO}}_4^{3-}} $, (4.56 ± 5.97) μmol/L for ${{\rm {SiO}}_3^{2-}} $, and (0.057 ± 0.021) mg/L for PN], which was in agreement with our previous observation in the bay (He et al., 2023), reflecting more terrestrial nutrient discharge in the rainy season. In addition, PN varied similarly to Chl-a during both seasons.

The δ¹⁵N-${{\rm {NO}}_3^-} $ values ranged from 0.5‰ to 7.3‰ in the rainy season and from 1.0‰ to 7.9‰ in the dry season, showing similar ranges between seasons and generally decreasing from the upper bay to the outer bay (Fig. 4). However, the values of δ¹⁵N-${{\rm {NO}}_3^-} $ in the upper bay during the dry season were higher compared to the rainy season (Fig. 4). The δ¹⁸O-${{\rm {NO}}_3^-} $ values ranged from –3.5‰ to 9.1‰ in the rainy season and from –1.4‰ to 18.9‰ in the dry season, with higher values in the dry season (an average of 8.7‰) than in the rainy season (an average of 1.9‰). Unlike the δ¹⁵N-${{\rm {NO}}_3^-} $, the lower δ¹⁸O-${{\rm {NO}}_3^-} $ values occurred in the upper bay, with an increasing trend towards the outer bay (Fig. 4).

The δ¹⁵N-PN values in the rainy season (ranging from 3.1‰ to 11.1‰, an average of 5.6‰) were lower than those in the dry season (ranging from 6.5‰ to 11.6‰, an average of 9.2‰). Except for the higher δ¹⁵N-PN values in the outer bay during the rainy season, the values generally decreased from the upper bay to the outer bay (Fig. 4). The distinct seasonal and spatial distributions of isotopic values suggest seasonal N sources and biogeochemical processes in the bay.

In Zhanjiang Bay, nutrient concentrations in the upper bay were remarkably higher than those in the lower bay during both seasons (Fig. 3), consistent with observations by He et al. (2023) and Li et al. (2020). Similarly, low salinity was recorded in the upper bay (Fig. 2), indicating that elevated nutrient levels were predominantly influenced by terrestrial input. Heavy rainfall in this region typically occurs from April to October (Lao et al., 2022b). Such rainfall can erode and wash away land-based pollutants into the rivers surrounding Zhanjiang Bay, ultimately entering the coastal waters (He et al., 2023). The nutrient concentrations increased significantly during the rainy season, indicating the significant impact of heavy land-based sources discharged from anthropogenic activities (He et al., 2023; Zhang et al., 2021). Over the past decades, increasing terrestrial nutrient inputs have contributed to enhanced eutrophication in the bay (He et al., 2023). However, in the lower bay, the lower nutrient levels could be attributed to the intrusion of high-salinity water from the outer bay. Water mass transport dramatically affects nutrient distribution, thus impacting local marine ecosystems (Lao et al., 2022a, 2023b, 2023c). Human activities have significantly increased the intrusion of high-salinity water from the outer bay over the past two decades (Lao et al., 2022b), resulting in the dilution of nutrients in the lower bay. Although the intrusion of high-salinity water was more pronounced in summer owing to the stronger west-Guangdong coastal current during this period (Lao et al., 2022b), the nutrient concentration in the lower bay during the rainy season was still significantly higher than that during the dry season (t-test, p<0.01) (Fig. 3). This suggests that during the rainy season, terrestrial inputs had a more pronounced effect on nutrient distribution than water mass transportation from the outer bay.

Notably, the N/P ratio in the two seasons (ranging from 4.0 to 11.2, an average of 6.2 in the rainy season, and from 0.7 to 22.6, an average of 5.1 in the dry season) was significantly lower than the Redfield ratio (16.0). High N and P concentrations suggest that N and P did not act as limiting nutrients in the bay and that the environmental conditions were favorable for phytoplankton blooms (Yang et al., 2018). Over recent decades, eutrophication and harmful algal blooms have increased substantially in the bay (He et al., 2023; Zhang et al., 2022). Before the 1980s, harmful algal blooms rarely occurred in Zhanjiang Bay, but have occurred periodically and frequently since the 2000s (Zhang et al., 2022). This is mainly due to the increasing nutrient input in the bay, particularly P input (He et al., 2023; Zhang et al., 2022). The dissolved inorganic nitrogen concentration increased threefold from 1990 to 2019, while the P concentration increased 21-fold owing to the continuous input of high-concentration phosphorus from industrial factories around Zhanjiang Bay (He et al., 2023; Zhang et al., 2022). The faster rate of increase in the P concentration in the bay has been responsible for the decrease in the N/P ratio over the past decades (He et al., 2023). Therefore, the ecosystem in Zhanjiang Bay has shifted from P-limited oligotrophic conditions before the 2020s to N-limited eutrophic conditions (Zhang et al., 2022). This implies that the N input plays an important role in the health of the Zhanjiang Bay ecosystem. However, in addition to the impact of the watershed, sewage outlets around Zhanjiang Bay can also input large amounts of anthropogenic nitrogen into the bay (Zhang et al., 2021), resulting in a complex and diversified source of N in the bay.

In the rainy season, the negative offset of ${{\rm {NO}}_3^-} $, along with slightly positive offsets in δ¹⁵N-${{\rm {NO}}_3^-} $, and offset-δ¹⁸O-${{\rm {NO}}_3^-} $ in the upper bay (areas with lower salinity areas) (Figs 5a and c) indicated the consumption of ${{\rm {NO}}_3^-} $. The processes of denitrification and phytoplankton assimilation should be ruled out for the ${{\rm {NO}}_3^-} $ consumption in the upper bay during the rainy season due to the high DO level (>5 mg/L) and decoupling of δ¹⁵N-${{\rm {NO}}_3^-} $ and δ¹⁸O-${{\rm {NO}}_3^-} $ (Fig. 6c). In addition, a high concentration of ${{\rm {NH}}_4^+} $ (>5 μmol/L) was observed in the upper bay in the rainy season, and phytoplankton preferentially assimilated ${{\rm {NH}}_4^+} $ under sufficient ${{\rm {NH}}_4^+} $ conditions (Chen et al., 2022c; Glibert et al., 2016). Therefore, the assimilation of ${{\rm {NH}}_4^+} $ (rather than ${{\rm {NO}}_3^-} $), by phytoplankton should be responsible for the negative offset-δ¹⁵N-PN value, while the positive offset-¹⁵N-${{\rm {NO}}_3^-} $ values could be influenced by the intense physical sediment-water interaction.

Active ${{\rm {NO}}_3^-} $ consumption induced by denitrification in the sediments results in an efflux of ${{\rm {NO}}_3^-} $ from the upper overlying seawater to the sediments, thereby consuming ${{\rm {NO}}_3^-} $ and enriching δ¹⁵N-${{\rm {NO}}_3^-} $ in the water column (Chen et al., 2022c; Ye et al., 2016; Zhang et al., 2013). Indeed, the upper bay is narrow and shallow, and is substantially influenced by tidal pumping. In particular, tidal pumping is stronger in September (Wang et al., 2021), which can promote the bidirectional exchange of materials, including nutrients, between the sediment pore water and the overlying water. This process can increase the δ¹⁵N-${{\rm {NO}}_3^-} $ value but lead to the consumption of ${{\rm {NO}}_3^-} $ in seawater due to denitrification-induced enrichment of nitrate isotopes in sediments (Chen et al., 2022c).

Both offset-${{\rm {NO}}_3^-} $ and offset-δ¹⁵N-${{\rm {NO}}_3^-} $ values in the lower bay fluctuated on the theoretical mixing line during the rainy season (Fig. 5), indicating that the ${{\rm {NO}}_3^-} $ is in a conservative mixing state. However, similar to the upper bay, significant positive correlations were found between Chl-a and PN in the lower bay during the rainy season (Fig. 6), suggesting that phytoplankton biomass was the dominant component of PN. In addition, negative offset-δ¹⁵N-PN values occurred in the lower bay during the rainy season (Fig. 5b), which could be influenced by the assimilation of ${{\rm {NH}}_4^+} $ by phytoplankton due to the sufficient ${{\rm {NH}}_4^+} $ (an average of 5.04 μmol/L). Sufficient ${{\rm {NH}}_4^+} $ may be responsible for the higher Chl-a concentrations in both the upper and lower bays during the rainy season (Fig. 2g). Unlike the upper bay, the lower bay is broader and deeper owing to the influence of terrain and is thus less influenced by tidal pumping (Wang et al., 2021). Moreover, the greater depth in the lower bay prevents exchange between the surface and bottom water because of strong stratification in the rainy season. This is supported by the higher temperature (an average of 31.39℃) and lower salinity (an average of 25.05) in the surface water, whereas the lower temperature (an average of 30.70℃) and higher salinity (an average of 26.74) in the bottom water.

In the dry season, a negative offset of ${{\rm {NO}}_3^-} $ was observed in both the upper and lower bays (Fig. 5a), indicating that removal of ${{\rm {NO}}_3^-} $ occurred within the bay. Different from the rainy season, extremely low ${{\rm {NH}}_4^+} $ concentration (<0.2 μmol/L) (Fig. 3f) was observed in the bay, suggesting that ${{\rm {NH}}_4^+} $ is almost completely consumed in the dry season. The overall positive offset δ¹⁵N-${{\rm {NO}}_3^-} $ and δ¹⁸O-${{\rm {NO}}_3^-} $ values (average of 1.5‰ and 3.5‰, respectively) and the significantly positive relation between them suggested that ${{\rm {NO}}_3^-} $ assimilation by phytoplankton occurred in the bay during this season. However, the slopes were >1 in both regions (Figs 6c and d), indicating that there may have been other processes or sources during this period. The significant positive correlations between Chl-a and PN in the two regions indicated that phytoplankton biomass was the major component of PN. However, unlike the rainy season, most offset δ¹⁵N-PN values were above the mixing line (positive values) in the two regions during the dry season (Fig. 5b), suggesting the loss of PN (such as decomposition/mineralization) simultaneously occurred in the bay. This was consistent with the decrease in Chl-a concentration during the dry season (Fig. 2h). Interestingly, the offset δ¹⁵N-PN values (−2.2‰–2.5‰) were within the range of offset δ¹⁵N-${{\rm {NO}}_3^-} $ values (−3.8‰–3.7‰) in the dry season, probably suggesting a strong coupling between assimilation and decomposition/mineralization (Ye et al., 2016).

Indeed, in mid-March 2018, a red tide of Phaeocystis globosa in Zhanjiang Bay, with a maximum cell density of 1.13 ×10⁹ cells/L (2018 South China Sea Region Marine Disaster Bulletin; http://scs.mnr.gov.cn/scsb/gbytj/201911/b90b49d8b13e4f609ef31615fdde69c6.shtml), occurred. The dry season, particularly late winter and early spring, is characterized by frequent algal blooms and red tides in Zhanjiang Bay (Zhang et al., 2022). The outbreak of red tides consumes large amounts of nutrients, especially ammonium. However, due to the sharp decrease in runoff during the dry season, the input of nutrients from terrestrial sources has significantly diminished (Lao et al., 2022b). The outbreak of red tides consumes a large amount of nutrients in the bay, resulting in decreased nutrients in the bay owing to the lack of external nutrient input during this period (He et al., 2023; Zhang et al., 2022). Owing to the high density of algae and insufficient nutrient supply for their continued growth, a large number of algae gradually began to die. The lower Chl-a concentration during the sampling period (at the end of March, during the late red tide period) is consistent with this phenomenon. Therefore, these algal deaths and gradual mineralization could contribute to elevating δ¹⁵N-PN values in Zhanjiang Bay during the dry season. Thus, the coupling between assimilation and decomposition/mineralization could be the dominant process in the bay during the dry season.

Due to the reduced input of terrestrial nutrients and the dominance of biological processes (assimilation and mineralization) in isotopic fractionation, the isotopic values of nitrate cannot be used for source apportionment during the dry season. In contrast, despite some fluctuations, nitrate concentrations mainly aligned with the theoretical mixing line, indicating that nitrate was in a conservative mixing state during the rainy season, and that nitrate isotopes should be similar to the source. Therefore, we quantified the sources of nitrate in Zhanjiang Bay during the rainy season. Because Zhanjiang Bay is surrounded by cities and rivers, heavy rainfall can erode and wash away land-based pollutants from the basin into rivers, which then flow from the top of the bay to the lower bay during the rainy season. Thus, four potential nitrate sources, namely manure and sewage, fertilizer, soil N, and atmospheric deposition, were considered in this study. The values of δ¹⁵N-${{\rm {NO}}_3^-} $ and δ¹⁸O-${{\rm {NO}}_3^-} $ for the four potential nitrate sources are presented in Table 2. To quantify the proportional contributions of these nitrate sources, a Bayesian mixing model was used, and the results are shown in Fig. 7b. Soil N (36%) was the dominant nitrate source in Zhanjiang Bay during the rainy season, followed by manure and sewage (33%), and fertilizer (30%), while atmospheric deposition (only 1%) contributed less to the nitrate pool in the bay. These results were consistent with the findings from a classical nitrate dual isotopic approach depicted in Fig. 7. The δ¹⁵N-${{\rm {NO}}_3^-} $ and δ¹⁸O-${{\rm {NO}}_3^-} $ values in Zhanjiang Bay mainly fell in the sources of soil N and manure and sewage. Zhanjiang Bay experiences high rainfall and frequency during the rainy season (Chen et al., 2021). Rainwater erosion can transport a substantial amount of soil N from the basin into the bay, similar to findings in Qinzhou Bay, which shares similar climatic characteristics (Chen et al., 2022c). Moreover, many sewage outlets around Zhanjiang Bay directly discharge large amounts of urban and industrial wastewater into the bay (Chen et al., 2022b; He et al., 2023; Zhang et al., 2021). As shown in Fig. 7, the contribution of manure, sewage, and fertilizer to nitrate loads is very close to the contribution of soil N, suggesting that heavy rainfall in the rainy season can also input massive amounts of sewage and fertilizer N into Zhanjiang Bay. The elevated nutrient concentrations in Zhanjiang Bay during the rainy season may be attributed to abundant contaminants discharged from soil, sewage, and fertilizers. Due to intensive human activities, the intrusion of high-salinity water from the outer bay into the inner Zhanjiang Bay has significantly increased over the past two decades (Lao et al., 2022b). Particularly in the summer (rainy season), the stronger west-Guangdong coastal current in outer Zhanjiang Bay can increase high-salinity water intrusion (Lao et al., 2022b). In addition, Zhanjiang Bay is frequently affected by typhoons, during which heavy rainfall can input large amounts of nutrients into Zhanjiang Bay (Lao et al., 2023c, 2023d; Zhou et al., 2021). However, typhoon storm surges can also intrude large amounts of high-salinity seawater into Zhanjiang Bay (Lao et al., 2023d). The mixing of high-salinity seawater intrusion and diluted freshwater in the bay forms a strong ocean front, thereby trapping a significant amount of pollutants within the bay (Lao et al., 2023d). This is also an important factor that exacerbates the eutrophication of seawater in Zhanjiang Bay (He et al., 2023), which must be considered by environmental managers.