Occurrence and mass loads of N-nitrosamines discharged from different anthropogenic activities in Desheng River, South China

N-nitrosamines are widespread in various bodies of water, which is of great concern due to their carcinogenic risks and harmful mutagenic effects. Livestock rearing, domestic, agricultural, and industrial wastewaters are the main sources of N-nitrosamines in environmental water. However, information on the amount of N-nitrosamines these different wastewaters contribute to environmental water is scarce. Here, we investigated eight N-nitrosamines and assessed their mass loadings in the Desheng River to quantify the contributions discharged from different anthropogenic activities. N-nitrosodimethylamine (NDMA) (< 1.6–18 ng/L), N-nitrosomethylethylamine (NMEA) (< 2.2 ng/L), N-nitrosodiethylamine (NDEA) (< 1.7–2.4 ng/L), N-nitrosopyrrolidine (NPYR) (< 1.8–18 ng/L), N-nitrosomorpholine (NMOR) (< 2.0–3.5 ng/L), N-nitrosopiperidine (NPIP) (< 2.2–2.5 ng/L), and N-nitrosodi-n-butylamine (NDBA) (< 3.3–16 ng/L) were detected. NDMA and NDBA were the dominant compounds contributing 89% and 92% to the total N-nitrosamine concentrations. The mean cumulative concentrations of N-nitrosamines in the livestock rearing area (26 ± 11 ng/L) and industrial area (24 ± 4.8 ng/L) were higher than those in the residential area (16 ± 6.3 ng/L) and farmland area (15 ± 5.1 ng/L). The mean concentration of N-nitrosamines in the tributaries (22 ng/L) was slightly higher than that in the mainstem (17 ng/L), probably due to the dilution effect of the mainstem. However, the mass loading assessment based on the river’s flow and water concentrations suggested the negligible mass emission of N-nitrosamines into the mainstem from tributaries, which could be due to the small water flow of tributaries. The average mass loads of N-nitrosamines discharged into the mainstem were ranked as the livestock rearing area (742.7 g/d), industrial area (558.6 g/d), farmland area (93.9 g/d), and residential areas (83.2 g/d). In the livestock rearing, residential, and industrial area, NDMA (60.9%, 53.6%, and 46.7%) and NDBA (34.6%, 33.3%, and 44.9%) contributed the most mass loads; NDMA (23.4%), NDEA (15.8%), NPYR (10.1%), NPIP (12.8%), and NDBA (37.8%) contributed almost all the mass loads in the farmland area. Photodegradation amounts of NDMA (0.65 ~ 5.25 µg/(m3·day)), NDBA (0.37 ~ 0.91 µg/(m3·day)), and NDEA (0 ~ 0.66 µg/(m3·day)) were also calculated according to the mass loading. Quantifying the contribution of different anthropogenic activities to the river will provide important information for regional river water quality protection. Risk quotient (RQ) values showed the negligible ecological risks for fish, daphnid, and green algae.

Previous studies have detected N-nitrosamines in urine, feces, and grey water (Ohshima et al. 1987;Zeng and Mitch 2015), with the most frequently detected species being NDMA (2.8-4.9 ng/L), NDEA (3.8-4.5 ng/L), and NMOR (avg. 4.0 ng/L) (Zeng and Mitch 2015). Endogenously formed N-nitrosamines excreted by the body are the primary contributors to N-nitrosamines loads in domestic wastewater that affect river water (Gushgari and Halden 2018;Hrudey et al. 2013). In addition, extremely high concentrations of N-nitrosamines were detected in different industrial effluents (Chen et al. 2019), including textile printing and dyeing wastewater (NDMA: 3290 ± 137 ng/L; NDEA: 79 ± 4.6 ng/L), electroplating wastewater (NDMA: 1780 ± 192 ng/L; NMOR: 51 ± 3.0 ng/L), metal surface treatment wastewater (NDMA: 21 ng/L), fine chemical wastewater (NDMA: 28 ng/L), and pulping wastewater (NDMA: 97 ng/L) (Chen et al. 2019). These N-nitrosamines were generally formed in the industrial wastewater due to high concentrations of organic matters and the use of disinfector and then discharged into receiving rivers. Agricultural surface runoff may also emit N-nitrosamines into the environment (Leavey-Roback et al. 2016). In addition, a high concentration of up to 1110 ng/L NDMA was detected in swine wastewater (Chen et al. 2019). However, few studies have quantified the contributions of N-nitrosamines from these typical anthropogenic sources to environmental rivers on a river scale. Qiu et al. (2021) quantitatively investigated the source and fate of N-nitrosamines and their precursors in an urban water system in East China. However, they did not calculate the mass loadings of N-nitrosamines and the contributions of different anthropogenic activities. In a natural river, N-nitrosamines accumulate because of continuous emissions from industrial, agricultural, livestock rearing, and human daily life activities. They would also be attenuated through photodegradation and biodegradation Sørensen et al. 2015;Wang et al. 2015). The composition and concentrations of N-nitrosamines in surface water may vary depending on the land use types and human activities in surrounding areas. To the best of our knowledge, few studies have focused on this field. Some articles have evaluated the ecological risk of N-nitrosamines in rivers Li et al. 2020Li et al. , 2021. However, these articles found that the risk quotient (RQ) values of N-nitrosamines were lower than 0.1, indicating insignificant ecological risks to water. Could the same conclusion be drawn in our study area? Evaluating the ecological risk of N-nitrosamines in river water is essential for answering such questions.
The Desheng River is located in the most developed region in China, the Pearl River Delta. This study investigated the occurrence, distribution, mass loads, and ecological risk of eight N-nitrosamines in the river and quantitatively evaluated the discharges in different anthropogenic areas. The eight N-nitrosamines include NDMA, NMEA, NDEA, NDPA, NPYR, NMOR, NPIP, and NDBA. The results will provide essential information for regional river water quality protection.

Study area
The Pearl River Delta in Southern China is one of the biggest economic clusters in China. The study Desheng River and its tributaries are located in Shunde District, Foshan City. It is one of the leading tertiary tributaries of the Pearl River. The Pearl River Delta has a monsoon climate with an average annual precipitation of over 1,500 mm (X. . However, the temporal distribution of precipitation in the region is highly uneven, with distinctly rainy summers and dry winters. According to the National Bureau of Statistics of China, as of 2020, the population of Shunde District was approximately 3.23 million. The Shunde District is the manufacturing base of the Pearl River Delta, and modern services here are very intensive. Furthermore, it is known for its manufacturing industry, particularly in the production of household appliances, furniture, building materials, thriving food and beverage industry, and so on. The annual output of livestock husbandry in Shunde District was 11,901 tons in 2019 (Shunde gov 2021). The Desheng River is the main wastewater-receiving river in the study area.

Water sampling
The study areas and sampling sites are shown in Fig. 1. The distribution of human activities in this basin is apparent. Twenty-six river water samples were collected in the Desheng River in June 2020, including six in the livestock rearing area, eight in the residential area, five in the farmland area, and seven in the industrial area (Fig. 1). There were no rainfall events during the week before sampling to avoid the dilution effects of rain. At each site, 500 mL of water was collected and stored in an amber glass bottle, and 50 mg of sodium thiosulfate (Na 2 S 2 O 3 ) was immediately added as stabilizer. The samples were collected in amber glass bottles and then were delivered under refrigerated conditions (4 °C in cooling boxes) within 48 h to the lab. Once arrived at the lab, the samples were stored in the dark at 4 °C and analyzed within 7 days (Chen et al. 2019).

N-nitrosamines pretreatment and analysis
Water samples were prepared with solid-phase extraction. Sample pretreatment and analysis procedures have been described in detail in our previous studies (Chen et al. 2019(Chen et al. , 2017. Briefly, after filtration (0.22 µm, fiberglass filter paper), sodium bicarbonate (NaHCO 3 ) (2 g; pH of approximately 8) and NDMA-d6 (recovery surrogate; 25 ng) were added to each water sample (500 mL), and the samples were extracted through a coconut charcoal cartridge (2 g/6 mL; 80-120 mesh) at a flow rate of < 20 mL/min. Cartridges were pretreated with dichloromethane (3 mL), methanol (6 mL), and Millipore water (15 mL) before use. The analytes were then eluted with dichloromethane (12 mL) and concentrated to a final volume of 500 μL under a stream of ultrahigh purity nitrogen. After concentration, each extract was spiked with internal standard NDPA-d 14 (25 ng) and stored at -20 °C. All solid reagents (Na 2 S 2 O 3 and NaHCO 3 , AR grade), fiberglass filter paper, coconut charcoal cartridge, N-nitrosamine standard products, and two isotope-labeled N-nitrosamines (NDMA-d 6 and NDPA-d 14 ) used in the water sampling, pretreatment, and analysis processes were purchased from ANPLE Laboratory Technologies (Shanghai, China), while liquid reagents (dichloromethane and methanol, chromatographic grade) were obtained from Thermo Fisher Scientific (Shanghai, China). Millipore water was produced by a Milli-Q system from Sartorius (Göttingen, Germany).
The samples were analyzed on an Agilent 7890B gas chromatograph (GC) coupled with an Agilent 7000C triple quadrupole mass spectrometer (MS/MS) with tandem capillary columns of Agilent DB-35 MS (Agilent; 30 m, 0.32 mm i.d., and 0.25 μm film) and Agilent HP-5 MS (Agilent; 15 m, 0.25 mm i.d., and 0.25 μm film). These two columns were serially coupled by a zero-dead-volume fitting (the purge union). Electron ionization (EI) mode (70 eV) was carried out for mass spectrometric ionization. The ion source and transfer line temperatures were set as 230 °C and 280 °C, respectively. The first MRM transition shown for each molecule was used for quantification, while the second transition shown was monitored for confirmation of molecular identification (Supporting Information, Table S1).

Quality assurance and quality control
Individual N-nitrosamines were quantified by corresponding linear calibration curves (r 2 > 0.999) with a series of N-nitrosamines standard solutions (2,5,10,20,40,100,200, and 500 µg/L). The absolute recoveries of NDMA-d6 ranged from 68 to 83% (average ± standard deviation: 78 ± 4%). A laboratory blank was incorporated in every batch of 10 samples and treated with the same analytical procedures. No N-nitrosamines were detected in the were calculated by multiplying the standard deviation of seven replicates by the Student's T value of 3.14 (oneside T distribution for six degrees of freedom at the 99% confidence level). American Public Association, American Water Works Association, and Water Environment Federation issued this MDL calculation method in the Standard Method 1030 C for the Examination of Water and Wastewater (Eaton et al. 2005). The MDLs of target N-nitrosamine species are 0.50 ng/L for NDMA, 0.60 ng/L for NMEA, 0.90 ng/L for NDEA, 0.9 ng/L for NDPA, 0.70 ng/L for NMOR, 1.1 ng/L for NPYR, 0.90 ng/L for NPIP, and 1.0 ng/L for NDBA.

Basic parameters
Total nitrogen (TN) was detected by Alkaline Potassium Persulfate Digestion Ultraviolet Spectrophotometry (HACH DR3900 Ultraviolet Spectrophotometer, China). The detection limit of this method is 0.05 mg/L, and the determination range is 0.20 ~ 7.00 mg/L. Ammonium (NH 3 -N) was determined by Nessler reagent spectrophotometry with a detection limit of 0.02 mg/L. In addition, nitrate (NO 3 -N) and nitrite (NO 2 -N) were detected by UV spectrophotometry, and the detection limits were 0.08 mg/L and 0.001 mg/L, respectively. All water quality parameters are presented in Table S3.

Mass loading calculation
The mass load of individual N-nitrosamine at each sampling site were estimated based on the following equations Huang et al. 2020;Qiu et al. 2021): where M i represents the mass load of individual N-nitrosamine, and C i represents the concentration of individual N-nitrosamine in site D i . T represents the time period (1 day). A i and v i represent the cross-sectional area and the water velocity in site D i , respectively. Q i represents the water amount flowing through site D i . M site is the mass loads of all the detected N-nitrosamines in a river sample.
The cross-sectional area was calculated according to the river cross-sectional area model in Figure S1. The flow rate at each point was less than 0.1 m/s, the minimum threshold of the current meter used in the actual sampling process. The float method measured the river flow rate as 0.01 m/s at each point, and the flow estimation of each point is shown in Table S2. In the float method, an object of low density is allowed to float for a known distance in the stream, and the time taken by the float is measured and water speed calculated (Dobriyal et al. 2017).

Ecological risk assessment
The ecological risk posed by N-nitrosamines in river water was accessed by using the RQ of three taxa (fish, daphnia, and green algae) at acute levels (Aydin et al. 2019;Godoy et al. 2018). The RQs of different N-nitrosamines in different taxa can be calculated by the following equations: where C i is the concentration of individual N-nitrosamine in the Desheng River (ng/L), and PNEC is the predicted maximum no-effect concentration (ng/L). LC 50 and EC 50 are the lethal concentration and effect concentration for 50% of the population of exposed organisms, respectively, obtained from the ECOSAR Program presented by the USEPA Office of Chemical Safety and Pollution Prevention (Table 1). SF is the safety factor for the acute toxicity of N-nitrosamine, and the SF value is commonly set as 100 in freshwater (Brink 1999;Hela et al. 2005). RQ site is the ecological risk from a river water sample. The RQ values were classified into four indicating different risk levels: negligible risk (RQ < 0.01), low risk (0.01 ≤ RQ < 0.1), medium risk (0.1 ≤ RQ < 1), and high risk (RQ > 1) (Figuière et al. 2022). (4)

Concentrations of N-nitrosamines
N-nitrosamine concentrations in the Desheng River are shown in Table 2 and Fig. 2. At least one N-nitrosamine was detected in all sampling sites, indicating its prevalence in the river water. Among the eight N-nitrosamines, NDMA, NDEA, NPYR, NMOR, and NDBA were detected with mean concentrations of 9.7 ± 4.5, 1.2 ± 1.0, 1.5 ± 4.2, 0.7 ± 1.1 and 7.7 ± 3.8 ng/L, respectively ( Table 2, Fig. 2). NDPA was not detected in any water sample, which might be because NDPA is not an essential industrial or commercial chemical product, and it is also not detectable in swine wastewater (Chen et al. 2019) and domestic sewage (Zeng and Mitch 2015). NDMA and NDBA were the dominant compounds contributing 89% and 92% to the total N-nitrosamine concentrations. Our previous study showed that NDMA (1110 ± 110 ng/L) and NDBA (226 ± 36 ng/L) are the predominant compounds in swine wastewater, possibly resulting from the usage of antibiotics and frequent disinfection during livestock breeding (Chen et al. 2019). Therefore, their dominance in river water might mainly be associated with the intensive and large-scale livestock-rearing activities upstream of this area. The dominance of NDMA and NDBA was also reported in other rivers in China and other countries (Kulshrestha et al. 2010;Van Huy et al. 2011;Wang et al. 2011). For instance, NDMA and NDBA were detected at 37.1 ± 49.5 and 6.2 ± 4.6 ng/L in the Jialu River water samples in northern China, accounting for 44.7% and 26.4% of the total N-nitrosamines, respectively (Ma et al. 2012). Similarly, NDBA was the most frequently detected compound in the Yangtze River in China, with a detection rate of 88% . Likewise, Wang et al. (2016) found that NDMA and NDBA were prevalent in the Songhuajiang River in China. However, we observed lower total concentrations of N-nitrosamines in this study compared to a previous investigation conducted in the same area in December 2013 (up to 276 ng/L) (Chen et al. 2017), which could be attributed to water quality improvement in this river due to the policy "Ten Measures for Water" launched by the Chinese government in April 2015 to improve water quality and conservation. As we conducted the investigation in June (the wet season), the influence of seasonality could also be the reason. A previous study found the lower N-nitrosamines in source water in the wet season than in the dry season (Maqbool et al. 2020).

The influence of anthropogenic activities
The influence of different anthropogenic activities along this river was investigated by analyzing and comparing N-nitrosamines collected from different land use types (as shown in Table 2 and Fig. 2). The mean concentrations of N-nitrosamines in the livestock rearing area (26 ± 11 ng/L) and industrial area (24 ± 4.8 ng/L) were higher than those in the residential area (16 ± 6.3 ng/L) and farmland area (15 ± 5.1 ng/L) ( Table 2). This result indicated higher pollution caused by N-nitrosamines was found in livestock rearing and industrial areas than in domestic and agricultural areas. A previous study also found higher N-nitrosamine concentrations in river samples collected near a industrial area (> 10,000 ng/L) than in those collected from the non-industrial areas (1000 ~ 10,000 ng/L) (Lee and Oh 2016). In our study, the Kruskal-Wallis test showed that the concentrations of seven N-nitrosamines did not significantly vary between different regions (p > 0.05). This result could be due to the influence of various factors on the distribution and fate of N-nitrosamines.
Livestock rearing area High concentrations of NPYR (up to 18 ng/L) were observed in the livestock rearing area ( Fig. 2(a)). This was consistent with our previous study (Chen et al. 2019), in which high levels of NPYR (245 ± 22 ng/L) were detected in swine wastewater. In addition, Zeng et al. (2016) also observed high concentrations of NDMA (37 ng/L) and NPYR (20 ng/L) in source water affected by animal husbandry facilities. This result might be attributed to the use of antibiotics containing pyrrolidine structures and disinfection in farming processes (Bhat et al. 2023;Chen et al. 2019). Pyrrolidine, which is the precursor of NPYR, is an important structural motif in drug design and development (Bhat et al. 2023). Our results showed that NPYR could be an indicator of livestock-rearing wastewater discharge into the river water.
), residential area ((c), (d)), farmland area ((e), (f)), and industrial area ((g), (h)). The black lines and black points in the subgraphs in Fig. 2 (a, c,  e, and g) showed the river flow directions and the sampling sites, respectively Further systematic investigation on livestock-rearing wastewater is needed to confirm this hypothesis.

Farmland area
The average concentrations of NDMA, NDBA, and NDEA in the river water sampling sites near the farmland area were 5.6 ± 6.6, 5.3 ± 3.2, and 1.2 ± 1.1 ng/L, respectively ( Fig. 2(e)), which were comparable to those observed in residential areas. The reaction of nitrite with certain herbicides can form N-nitrosamines in agricultural soils (Pitts et al. 1978). Therefore, the low concentrations of N-nitrosamines could be the result of a reaction between pesticides/herbicides and nitrite generated from nitrogen fertilizer. The low concentrations of N-nitrosamines in the farmland and residential areas suggested the small contribution of N-nitrosamines from domestic and farmland wastewater to river water, which is consistent with a previous study (Zeng et al. 2016).

Industrial area
In recent years, the influence of industrial wastewater on N-nitrosamines in rivers has been an important research focus (Chen et al. 2019;Qiu et al. 2021). The average concentrations of NDMA, NDEA, NMOR, and NDBA in the river water of the industrial area were 11 ± 4.0, 1.3 ± 1.2, 1.3 ± 1.4, and 9.4 ± 3.9 ng/L, respectively ( Fig. 2(g)). Previous studies have also shown that industrial wastewaters discharged into river water carry high concentrations of N-nitrosamines and their precursors (Chen et al. 2019). Textile printing and dyeing wastewater contained high concentrations of NDMA (up to 10,000 ng/L) and its precursors during washing, dyeing, and cleaning (Kosaka et al. 2009). The decolorization and disinfection processes using sodium hypochlorite might be the main reason for NDMA formation in textile printing and dyeing wastewater (Chen et al. 2019). The electroplating wastewater displayed high concentrations of NDMA (1780 ± 192 ng/L), NPYR (up to 279 ng/L), and NMOR (51 ± 3.0 ng/L) due to the use of dithiocarbamate solutions in printed-circuit board manufacturing processes and the application of sodium hypochlorite as an oxidizing agent (Chen et al. 2019). The occurrence of N-nitrosamines in industrial wastewater is complicated. They might originate from the chemical material used in industrial production or might be produced during industrial processes from their precursors. For example, nitrite and secondary amines can form N-nitrosamines during manufacturing in food processing plants (Hinuma et al. 1990). Drugs and personal care products in pharmaceutical factory wastewater are precursors to form NDMA and NMOR (Kemper et al. 2010;Shen and Andrews 2011). However, the concentrations of N-nitrosamines in our study were lower than those detected in industrial wastewater, which could be due to the dilution and photodegradation in river water . Industrial effluent is generally not allowed to be discharged directly into environmental water bodies. However, heavy rain events may flush industrial effluent containing N-nitrosamines and their precursors into rivers (Zhao et al. 2022), and the illegal discharge of industrial effluent outside regulation cannot be excluded. Spatially, the highest mean concentration was observed in D5 (43 ng/L) upstream of this river, followed by D6 (36 ng/L), D21 (31 ng/L), and D23 (28 ng/L) (Fig. 2). These sites were located in livestock rearing and industrial areas. Therefore, the surrounding discharged wastewater might be the primary source of N-nitrosamines. The lowest mean concentration was observed in D8 at 1.8 ng/L, located at a residential area. The mean concentration of N-nitrosamines in the tributaries (22 ng/L) was slightly higher than that in the mainstem (17 ng/L), probably due to the dilution effect of the mainstem. Additionally, the compositions of N-nitrosamines in tributaries and the mainstem were highly similar (Fig. 3). These results were consistent with their close hydraulic connection in this region.
The heatmap interprets the distribution patterns of N-nitrosamines (Fig. S2). In the vertical direction, N-nitrosamines were divided into two main clusters. Cluster 1, including NDMA and NDBA, was the dominant pollutant in all the samples. Cluster 2A, which includes NMEA, NPIP, NMOR, and NDEA, was little loaded in all the samples. Cluster 2B includes NPYR, which was highly loaded at sites D5 and D6. In the horizontal direction, the sampling sites were divided into two clusters. Cluster I contains sites D5 and D6 with NPYR (> 10 ng/L) detected. Cluster II contains the rest of the other sampling sites. The lack of clusters in the specific anthropogenic area suggested that various factors other than land use type, contributed to the occurrence of N-nitrosamines in rivers.

The influence of water quality parameters
Water quality parameters (including NH 3 -N, NO 2 -N, NO 3 -N, TN, dissolved oxygen (DO), pH, and redox potential (Eh)) were analyzed to reveal their influence on N-nitrosamine concentrations based on Spearman correlation analysis, as shown in Fig. 4. There were no significant relationships between the seven detected N-nitrosamines in each area, which suggested a variety of N-nitrosamine sources in our study area. We observed significant negative correlation between NDEA and DO (r = -0.473, p < 0.05). This might be attributed to the values of DO reflecting the river water pollution extent (Nong et al. 2020), and NDEA mainly occurred in polluted urban rivers with low DO values (Bei et al. 2016). On the other hand, the significant negative correlation between NDEA and DO could be due to degradation with the participation of DO. The large amounts of dissolved organic matter in water could increase the activity of microorganisms and cause DO consumption (Findlay et al. 2003). In some cases, dissolved organic matter interferes with the indirect photolysis of organic contaminants by screening reactive wavelengths of light and by scavenging reactive oxygen species (Xue et al. 2019). Generally, the lower the DO, the higher the dissolved organic matter, the stronger the light-screening effect, the weaker the photodegradation rate of NDEA, and the higher the NDEA concentration in water. A significant relationship was observed between NPYR and pH (r = 0.512, p < 0.05), while the correlation for NMEA and pH was negative. Based on the previous reference, we concluded that there is no obvious N-nitrosamine formation in environment water due to the low levels of precursors, including secondary amines and nitrite, and low formation rates of N-nitrosamines (Krasner et al. 2013). We suspected the relationships between N-nitrosamines and pH could be attributed to the different degradation rates in different water pH (Zhou et al. 2012). Furthermore, a significantly positive correlation was observed between NMEA and nitrite. As nitrite could significantly increase the formation of specific N-nitrosamine, further research is needed based on the formation potential in river water.

Input loads and possible degradation loads of N-nitrosamines based on mass loading calculation
The mass loadings of N-nitrosamines at each sampling site were calculated based on the river's flow and the concentration of each compound, with results shown in Fig. 5. Tributaries contributed only a tiny fraction (< 10 -2 ) of N-nitrosamines ( Fig. 5(b), (d), and (h)) to the mainstem due to their low water flow, implying the primary contributions of wastewaters discharging or surface runoff along the mainstem. As we can see from the results (Fig. 5), even without tributary input, the N-nitrosamine loads would increase in the mainstem, suggesting the emissions of wastewater into the mainstem. The average mass loads of N-nitrosamines discharged into the mainstem were as follows: the livestock rearing area (742.7 g/d), industrial area (558.6 g/d), farmland area (93.9 g/d), and residential area (83.2 g/d). In the livestock rearing, residential, and industrial areas, NDMA (60.9%, 53.6%, and 46.7%) and NDBA (34.6%, 33.3%, and 44.9%) contributed the most mass loads; NDMA (23.4%), NDEA (15.8%), NPYR (10.1%), NPIP (12.8%), and NDBA (37.8%) contributed all the mass loads in the farmland area.
In the livestock rearing area, the highest inputs of NDMA (1642.8 g/d) and NDBA (866.0 g/d) were detected at site D1. The mass loads of N-nitrosamines decreased continuously in the mainstem in this section, with a mass loss of 1605.7 g/d for NDMA and 815.7 g/d for NDBA in site D6. At site D11 in the residential area, the inputs of NDMA and NDBA were observed with mass loadings of 72.5 and 40.4 g/d, respectively, which are higher than the mass loads at site D9 (15.0 g/day for NDMA and 37.8 g/day for NDBA). The results indicated the new inputs of NDMA and NDBA in the residential area due to the discharge of domestic wastewater. At site D19 in the farmland area, new inputs of NPYR (19.0 g/d), NPIP (24.0 g/d), and NDBA (71.1 g/d) were observed compared to no detection of these N-nitrosamines at site D16. This result suggested that the primary inputs of N-nitrosamines in the farmland area were NPYR, NPIP, and NDBA. At site D23 in the industrial area, higher inputs of NDMA (313.7 g/d), NDEA (44.8 g/d), NMOR (48.9 g/d), and NDBA (160.1 g/d) were also detected compared to lower mass loads of these N-nitrosamines at site D19. This result suggested that the primary inputs of N-nitrosamines in the industrial area were NDMA, NDEA, NMOR, and NDBA.
There are no tributaries in the three sections from D3 to D4, D6 to D9, and D11 to D12. The distances between them were short (< 5000 m). We suspected that photodegradation could be the main reason for the reduction of N-nitrosamines, as the biodegradation rates are much lower than the photodegradation rates. For example, the photochemical attenuation half-life for NDMA is 16 min and 12-15 min for the other N-nitrosamines in surface water (Plumlee and Reinhard 2007). However, the biodegradation half-life for NDMA in surface water is as long as 31 ± 19 days (Plumlee and Reinhard 2007). The photodegradation amounts of NDMA in the three sections were 1.03, 0.65, and 5.25 µg/(m 3 ·day). The values were 0.58, 0.37, and 0.91 µg/(m 3 ·day) for NDBA and 0.05, -0.06, and 0.66 µg/ (m 3 ·day) for NDEA. Since there was little difference in river width between the two points in each section, the average value of the cross-sectional area between the two points was used to calculate the photodegradation amount. The negative value of NDEA photodegradation amounts in section D6 to D9 could be attributed to the unaccounted surface runoff inputs. Therefore, we assumed that the photodegradation amount of NDEA in this section was 0 µg/(m 3 ·day). The different N-nitrosamine attenuations between different sections can be attributed to the light-screening effect of organic matter (Mostafa et al. 2016). The first two sections were located in the livestock rearing area with widespread and higher contents of natural organic matter, which could slow the NDMA photolysis rate (Plumlee and Reinhard 2007;Sørensen et al. 2015). From the comparison of different N-nitrosamine compounds, NDMA and NDBA in the three sections display similar photodegradation amounts, while the photodegradation amounts of NDEA were lower than those of NDMA and NDBA. A previous study found that NDMA and NDBA had the same degradation rate constant (2.6 × 10 -2 L/W-min); however, NDEA was lower (1.8 × 10 -2 L/W-min) (Afzal et al. 2016). The difference in degradation rates could be due to their chemical structures (Afzal et al. 2016). For example, Xu et al. (2009) concluded that the higher degradation rate constant of NPIP in drinking water treated by UV irradiation was due to the lower electron density and N-N band energy than those of NPYR.

Ecological risk caused by N-nitrosamines
NDBA had higher RQ values for green algae than daphnid and fish, and these values were higher than other N-nitrosamines. Fig. 6 The RQ values of the detected N-nitrosamines for fish, daphnid, and green algae in all sampling sites However, all the individual N-nitrosamines were found to have negligible ecological risks for fish, daphnid, and green algae due to their RQ values being much lower than 0.01 (Fig. 6). The ΣRQ site in the industrial area was higher but still far lower than 0.01 (Table S4). Although the ecological risks caused by N-nitrosamines are negligible, there is the potential for synergistic risks with other compounds. Besides, the drinking water concentrations at the 10 -5 cancer risk level for NDMA, NDEA, NPYR, and NDBA are 7, 2, 200, and 60 ng/L, respectively (Andrzejewski et al. 2005). In this study, 80.8% of water samples exceeded the corresponding thresholds. Specifically, 65.4% of samples for NDMA and 23.1% for NDEA exceeded the corresponding threshold. Therefore, the regulation of N-nitrosamines in the environment is still needed.

Conclusions
Our study investigated the occurrence of N-nitrosamines in the Desheng River and characterized N-nitrosamines emitted from different anthropogenic activities. In particular, of the eight investigated N-nitrosamines, NDMA, NMEA, NPYR, NMOR, NPIP, and NDBA were detected. NDMA and NDBA were the most common N-nitrosamines with the highest the concentrations and detection rates in Desheng River. Higher N-nitrosamine concentrations were found in river water in the livestock rearing area (26 ± 11 ng/L), followed by industrial (24 ± 4.8 ng/L), residential (16 ± 6.3 ng/L), and farmland areas (15 ± 5.1 ng/L). The mean concentration of N-nitrosamines in the tributaries (22 ng/L) was slightly higher than that in the mainstem (17 ng/L). However, mass loading assessment suggested the tiny contributions of tributaries to the N-nitrosamines in the river which could be due to the lower water flow in the tributaries. The average mass loads of N-nitrosamines discharged into the mainstem ranked as the livestock rearing area (742.7 g/d), industrial area (558.6 g/d), farmland area (93.9 g/d), and residential area (83.2 g/d). The N-nitrosamine types with high mass loads differed among the anthropogenic activities. In the livestock rearing, residential, and industrial areas, NDMA (60.9%, 53.6%, and 46.7%) and NDBA (34.6%, 33.3%, and 44.9%) contributed the most mass loads; NDMA (23.4%), NDEA (15.8%), NPYR (10.1%), NPIP (12.8%), and NDBA (37.8%) contributed all the mass loads in the farmland area. Photodegradation amounts of NDMA, NDEA, and NDBA were calculated according to the mass loading calculation. In this study, we hypothesized that variation in the degradation of N-nitrosamines is related to the light-screening effect of dissolved organic matter. In future research, the analysis of dissolved organic matter (e.g., dissolved organic carbon and the total of organic carbon) and other precursors such as secondary amines is important to understand the formation potential of N-nitrosamines in river water. Further studies need to pay attention to the co-existence and co-degradation of dissolved organic compounds and N-nitrosamines in different anthropogenic activities. Although the RQ values showed negligible ecological risks for fish, daphnid, and green algae, further research is still needed to examine the chronic individual teratogenic dose of N-nitrosamines to aquatic organisms, which may pose a threat to the survival of aquatic organisms. This study will contribute to understanding the environmental impact of N-nitrosamines produced by different anthropogenic activities.