Agricultural Activities and Hydrological Processes Drive Nitrogen Pollution and Transport in Polder Waters: Evidence from Hydrochemical and Isotopic Analysis

Abstract

Excessive nitrogen export from lowland polders is a key contributor to cultural eutrophication in downstream aquatic ecosystems. This study investigated the spatiotemporal characteristics, migration pathways, and sources of nitrogen pollution in a typical polder system. Eight surface water sampling campaigns were conducted at 13 sites in Quyuan Polder, Dongting Lake, from 2022 to 2023, combining ArcGIS spatial analysis, multivariate statistics, and dual-isotope (δ¹⁵N-NO⁻), δ¹⁸O-NO₃⁻) techniques. Nitrate and ammonium nitrogen dominated the nitrogen pool, accounting for ~76% of total nitrogen. Concentrations were higher in the dry season (2.48 mg/L) than in the wet season (1.89 mg/L) and differed significantly among hydrological periods (p < 0.05). Within the polder, total nitrogen and ammonium nitrogen were elevated, whereas nitrate nitrogen was higher at the outlet, reflecting distinct nitrogen profiles along the hydrological gradient. Nitrogen transport patterns were largely consistent with flow direction, driven by both upstream inputs and in situ generation. Isotopic signatures indicated that nitrate originated mainly from ammonium fertilizer and soil nitrogen, with contributions from manure and sewage. These findings enhance understanding of nitrogen dynamics in lowland catchments and provide a scientific basis for targeted pollution control in polder waters.

1. Introduction

Agricultural non-point source nitrogen (N) pollution has emerged as a major environmental concern in lowland catchments, where intensive cultivation, excessive fertilizer use, and poor drainage conditions contribute to deteriorating water quality and ecological degradation [1]. Among these landscapes, polder systems—hydraulically managed low-lying farmlands enclosed by dikes—play a prominent role due to their spatial extent, agricultural productivity, and altered hydrological regimes [2,3]. In China, over 200 polders in the Dongting Lake Basin account for nearly 70% of the lake’s area and support intensive rice cultivation and aquaculture [4]. However, the semi-enclosed nature of polder hydrology, characterized by weak water exchange and long residence times, facilitates the accumulation and in situ transformation of reactive nitrogen, increasing the risk of downstream eutrophication [5]. Despite their ecological vulnerability and importance to food security, current studies have rarely examined the internal transport processes, transformation pathways, and seasonal dynamics of nitrogen pollution within these semi-closed systems. This knowledge gap limits our ability to design effective, region-specific mitigation strategies for nitrogen management in polder-dominated lowland basins [6].

The stable nitrogen (N) and oxygen (O) isotope compositions (δ¹⁵N-NO₃⁻, δ¹⁸O-NO₃⁻,) of NO₃⁻ provide unique isotopic fingerprints that can be utilized to discriminate among nitrogen sources. Accurate measurement of these isotopic ratios in aquatic systems enables the identification of multiple pollution sources, including chemical fertilizers, livestock and poultry wastewater, and domestic sewage [7]. However, in regions where multiple nitrate sources coexist, their isotopic signatures may overlap, making it difficult to differentiate among sources., which hinders the distinction between sources. As a result, dual-isotope analysis has not always yielded definitive results in source identification [8,9,10]. Recent studies have shown that combining stable isotopes with hydrogeochemical parameters is regarded as an effective screening tool to reduce the uncertainties in the sources of NO₃⁻ in water [11]. In practice, nitrogen pollution sources are highly complex, influenced by a range of factors including agricultural practices and climatic conditions. The dynamic spatiotemporal variations in these factors further result in significant diversity in the spatial and temporal distribution of water pollutants. This diversity reflects the intricate interactions between anthropogenic sources and environmental drivers underlying nitrogen pollution. Consequently, the integration of multiple approaches has become an essential trend in nitrogen pollution research.

Due to differences in fertilizer types [12], crop types [13] and application amounts [14], changes in land use and farming patterns [15], irrigation methods [16], and other factors related to rainfall [17] also affect the migration and transformation of nitrogen pollution. For example, in the Yiluo River Basin, the concentrations of TN, NO₃⁻-N, NH₄⁺-N, and NO₂⁻-N are higher in the wet season than in the dry season, and the proportions of various nitrogen sources differ among the upper, middle, and lower reaches during the dry and wet seasons [18]. In the Ru River Basin, the total nitrogen load shows significant seasonal variations, reaching its maximum in summer and minimum in spring [19]. Therefore, sampling strategies should account for hydrological periods, incorporating regional climatic patterns and the timing of agricultural activities. When integrated with multivariate statistical analysis and ArcGIS-based spatial analysis, this approach facilitates the identification of critical periods, key source areas, dominant pollution types, and transmission pathways of nitrogen in polder surface waters. To address the limitations of isotopic overlap in nitrate source identification, this study integrates hydrochemical parameters, multivariate statistical analysis, and spatial mapping. These methods complement isotope tracing by distinguishing pollutant types, revealing temporal-spatial pollution patterns, and improving source apportionment reliability.

The polders around Dongting Lake are important agricultural production areas in the middle and lower reaches of the Yangtze River, occupying 69.1% of the water area of Dongting Lake. There are 226 polders of various sizes within the basin. The protected area of the polders is 16,200 km², and the cultivated area is 6080 km². As a typical polder in the South Dongting Lake Basin, Quyuan Polder’s land is predominantly used for cultivation, with cultivated land accounting for 58%. It supports diverse agricultural activities, including double-cropping rice cultivation, aquaculture (crayfish and fish), watermelon farming, and tea production. These agricultural activities largely coincide with the high-temperature and heavy rainfall season under the subtropical monsoon climate, during which the annual TN concentrations in polder surface water are 0.82–3.43 mg/L [20]. The water then flows into the lake area through ditches and tributaries. According to the monitoring data up to 2022, the annual average total nitrogen (TN) value of Dongting Lake ranges from 1.21 to 2.38 mg/L. TN levels show significant spatial and temporal variation across different regions of the lake and have emerged as the primary pollutant of concern in Dongting Lake [4].

This study was conducted in the Pingjiang River section of Quyuan Polder in the Dongting Lake Basin. By combining hydrogeochemical indicators, dual isotopes, and statistical methods, we investigated the spatiotemporal migration and sources of nitrogen pollution in the surface water within the polder. Our aims were: (1) To characterize the seasonal variation of nitrogen pollution in polder surface water through targeted sampling during the dry and wet periods; (2) to identify the transport pathways of inorganic nitrogen pollutants by using multivariate statistical analysis of hydrochemical indicators; and (3) to use stable nitrogen and oxygen isotopes to analyze the sources of NO₃⁻ during the dry and wet seasons. The research results provide a basic understanding of nitrogen pollution in agricultural areas of lowland catchments and necessary scientific evidence for improving water quality.

2. Materials and Methods

2.1. Study Area

The Pingjiang River, a natural watercourse within Quyuan Polder, is located between 28°51′ to 28°59′ N and from 112°55′ to 112°59′ E. Its catchment covers 74.80 km², accounting for approximately one-third of the total area of Quyuan Polder. This region lies within the lowest elevation zone of the alluvial-lacustrine plains in northern Hunan Province. The terrain is primarily composed of man-made plains interspersed with low hills and uplands. The highest point is Leishi Mountain, with an altitude of 90.72 m. The lowest point is the bottom of Qiaomai Lake, with an altitude of 22 m. It is a typical lowland catchment area of plain river networks. There are three types of soil-forming parent materials in the region: sandstone metamorphic rock, Quaternary red clay, and river-lake sediments. The region experiences a humid subtropical monsoon climate, with an average annual temperature of 16.9 °C and average annual precipitation of 1406.7 mm. Precipitation is concentrated from April to August, accounting for 62.3% of the annual total. Historically, the highest average rainfall occurs in May, while June experiences the greatest frequency of rainstorms. Since the 1950s, under the combined influence of natural conditions and anthropogenic activities, certain sections of the Pingjiang River have been transformed into ponds or irrigation and drainage ditches. Currently, it mainly receives sewage from three north–south ditches (see Figure 1). Land use types include cultivated land, orchards, forests, grasslands, bare land, and residential areas. The low-lying and gentle terrain of the basin makes cultivated land the dominant land use type, accounting for 58% of the total land area. The region hosts demonstration bases for high-quality rice covering approximately 6.67 km², as well as for watermelon cultivation and integrated rice-crayfish farming. Except for the integrated rice-crayfish farming mode, all other rice-planting modes are double-cropping rice. Aquaculture is well-developed within the polder, with most ponds used for crayfish farming, lotus root cultivation, and freshwater fish production. Field observations revealed the presence of small-scale poultry farming, such as Muscovy duck breeding, facilitated by abundant water resources. There are no factories or processing enterprises within the study area, so the influence of industrial pollution sources can be excluded. In 2017, black and odorous water bodies appeared within the polder. The test results showed that the degree of blackness and odor of the black and odorous water bodies in the Gutangcha area was severe, and the levels of COD, ammonia nitrogen, and total phosphorus seriously exceeded the Chinese Surface Water Quality Standard (GB3838-2002).

2.2. Field Sampling and Laboratory Analysis

By taking into account different land use patterns, potential pollution sources, and the requirements for spatial distribution uniformity, a total of 13 sample points were selected within the Pingjiang River Basin to represent the surface water in the study area. Among them, 8 sample points (H1–H8) were set on the main channel of the Pingjiang River, 3 sample points (H9–H11) were set at the estuary where the Pingjiang River flows into the lake, and 2 sample points (H12 and H13) were set on the tributary irrigation and drainage ditches. In general, H1–H8, H12, and H13 represent the surface water within the polder, while H9–H11 represent the water bodies outside the polder. Influenced by the monsoon climate, the hydrological characteristics of rivers and lakes in the Dongting Lake area exhibit significant annual differences. The period from April to September and from October to March of the following year represent the wet season and the dry season of Dongting Lake, respectively [21]. To capture interannual variations in river–lake hydrological characteristics and differences in fertilization timing, a total of eight sampling campaigns were conducted during both the wet and dry seasons in 2022 and 2023.

Before sampling, polyethylene bottles were rinsed three times with the target water sample. Samples were stored in a portable refrigerator in the field and transported to the laboratory as soon as possible, then refrigerated at 4 °C. For parameters other than total nitrogen and total phosphorus, samples were filtered through a 0.45 μm membrane. Water quality parameters, including pH, temperature, dissolved oxygen, conductivity, total nitrogen, total phosphorus, ammonia nitrogen, major ions, and stable isotope ratios (δD, δ¹⁸O, δ¹⁵N-NO₃⁻, δ¹⁸O-NO₃⁻), were measured using standard laboratory procedures. All analyses were conducted in triplicate to ensure accuracy and reliability. Any missing or outlier data were carefully examined and, if necessary, extrapolated using standard procedures (e.g., exponential smoothing). Detailed QA/QC information, including instrument calibration, detection limits, method precision, and reference standards, is provided in Supplementary Material Table S1.

2.3. Data Analysis

Descriptive statistics were used to summarize the elemental concentration data. One-way analysis of variance (ANOVA) at a 95% confidence level (p < 0.05) was used to evaluate the significant differences in element concentrations between wet and dry seasons. The Spearman coefficient was calculated to quantitatively study the correlations between element concentrations, with a significance level set at p < 0.05 (two-tailed). Hierarchical cluster analysis (HCA) was used to analyze the hydrochemical dataset. HCA identifies statistical similarities among physicochemical parameters to classify samples into distinct groups. For the identified pollutants, correlations among samples from different sampling sites were examined to evaluate similarity levels and intrinsic relationships. This facilitated the identification of primary water pollutants and their transport pathways [22]. In this study, Q-type clustering was applied using SPSS (version 26.0, System Cluster) with the between-groups average linkage method and the squared Euclidean distance metric. The between-groups average linkage was chosen because it defines cluster distances as the mean of all pairwise distances, thereby avoiding chaining and overly compact clusters while emphasizing overall similarity. This approach is particularly suitable for environmental and hydrochemical datasets, where correlated variables reflect hydrological gradients, and it provided spatial groupings consistent with the hydrological connectivity and functional differences in the polder system.

To further identify hotspots and coldspots of nitrogen pollution, spatial interpolation was performed using the Inverse Distance Weighting (IDW) method in ArcGIS (version 10.8). Nitrate pollution sources in aquatic environments are complex. Traditional hydrochemical methods utilize the emission data of various pollution sources, the mass concentration of NO₃⁻, and the concentration characteristics of other ions to analyze the sources of NO₃⁻ pollution in water bodies. Advancements in technology have enabled the widespread application of stable nitrogen and oxygen isotope techniques in environmental pollution research [23,24]. These techniques have proven highly effective in tracing the sources, transport, and transformation of nitrate pollution in aquatic systems [25]. Based on the distinct nitrogen and oxygen isotope signatures associated with different nitrate pollution sources, and in combination with other environmental isotope and chemical analyses, this study quantified the contribution rates of various NO₃⁻ sources in surface water, groundwater, and precipitation. Additionally, nitrification and denitrification processes were assessed. Five potential nitrate sources were effectively distinguished: atmospheric deposition (AD), manure, and sewage (M and S), nitrogen fertilizer (NHF), aquaculture, and soil organic nitrogen (SN). Local measurements of δ¹⁵N-NO₃⁻ and δ¹⁸O-NO₃⁻ were carried out for the five potential sources of NO₃⁻ (Supplementary Materials Table S2) [26,27].

3. Results

3.1. Hydrogeochemical Characterization

The basic physicochemical characteristics of surface water in the study area are shown in Supplementary Materials Table S3. The average pH during both the dry and wet seasons is 7.6, indicating weakly alkaline conditions. Dissolved oxygen (DO) concentrations range from 1.53 to 14.6 mg/L, with an average of 7.04 mg/L. These values comply with the Chinese Surface Water Quality Standard (GB3838-2002), indicating that the surface water is relatively well oxygenated. The average detected value of TN is 2.48 mg/L during the dry season and 1.89 mg/L during the wet season. According to the Chinese Surface Water Quality Standard (GB3838-2002), the surface water in the study area is classified as Class IV–V water. Compared to the overall water quality of Dongting Lake, which has declined from Class III to Class IV in recent years, previous studies report TN concentrations ranging from 1.15 to 1.70 mg/L in the lake area [4]. Therefore, the TN pollution of the surface water within the polder is more serious. The concentration of total phosphorus (TP) ranges from 0.02 to 0.2 mg/L (with an average value of 0.07 mg/L), which is also higher than the detected range of 0.036–0.128 mg/L in Dongting Lake [4].

The proportion of dissolved inorganic nitrogen in surface water samples (DIN/TN) ranges from 79% to 96%, with an average of 86% (Supplementary Materials Figure S1). These results indicate that dissolved inorganic nitrogen is the predominant form of nitrogen nutrient in the surface water of the study area. In the water samples of the area within the polder, the proportion of nitrate nitrogen in total nitrogen (NO₃⁻-N/TN) reaches 57% during the wet season, as shown in Supplementary Materials Figure S1, and is lower at 42% during the dry season. In the water samples outside the polder, the measured range of NO₃⁻-N/TN reaches 84% during the wet season and 60% during the dry season. The NO₃⁻-N/TN in the water samples of the entire basin reaches 50%. Nitrate nitrogen is the dominant form of nitrogen pollution in surface water within the study area, followed by ammonium nitrogen. The nitrogen concentration in surface water varies significantly. By comparing the annual average concentrations of three inorganic nitrogen forms in the surface water, NO₃⁻-N is the highest (1.13 mg/L), followed by NH₄⁺-N (0.57 mg/L) and NO₂⁻-N (0.08 mg/L). NO₃⁻-N concentrations peaked in May 2023, with an average of 2.07 mg/L and a maximum of 2.48 mg/L, and in March 2023, with an average of 1.62 mg/L and a maximum of 2.49 mg/L.

3.2. Water Sample Grouping

Pollutants enter the water body and are gradually transported downstream along the flow direction, undergoing dilution and dispersion. Concurrently, physical and chemical processes—including adsorption, sedimentation, degradation, and transformation—affect pollutant dynamics [22]. The hierarchical cluster analysis (HCA) dendrogram (Figure 2a) showed that cutting at four clusters corresponded to the first clear increase in rescaled distance, indicating a natural separation of the sampling sites. The four-cluster partition was also consistent with the hydrological structure of the polder system, separating the upstream reaches, tributary—middle sections, confluence zones, and the outlet segment. This classification highlighted both the statistical distinctness of the groups and their functional differences, which are crucial for interpreting the spatial heterogeneity of hydrochemical characteristics. Moreover, it provided a robust basis for subsequent analysis of nitrogen transport pathways (Section 4.2), as the identified groups reflected the actual connectivity and flow directions within the polder system.

Specifically, the first group consists of sampling points H1–H3 in the upper reaches of the Pingjiang River within the study area, as well as the ditch confluence point H7. The second group includes inlet and outlet points H12 and H13 of the ancient lake wetland, the confluence point H4 of the second tributary, and the terminal point H8 of the Pingjiang River within the study area, covering the middle tributary and lower main channel sections. The third group only contains two ditch confluence points H5 and H6 on the main channel. The fourth group is made up of three sampling points H9, H10, and H11 located at the estuary where the Pingjiang River flows into the lake.

The concentrations of Na⁺, K⁺, Ca²⁺, Mg²⁺, HCO₃⁻, Cl⁻, SO₄²⁻, TP, TN, NO₃⁻-N, NO₂⁻-N and NH₄⁺-N in each group of water samples are different (Figure 2b). The average concentrations of TP, TN and NH₄⁺-N in the first group of water samples are higher than those in the other three groups, which are 0.076 mg/L, 2.406 mg/L and 0.794 mg/L, respectively. The sampling points of the first group are mainly located at the junction of farmland irrigation ditches and river chan2nels, as well as in the high-quality rice demonstration bases and watermelon planting bases. Therefore, the high concentration may be related to the application of ammonium nitrogen fertilizer and phosphate fertilizer. The second group has a relatively high load in terms of the indicators of NO₂⁻-N (0.106 mg/L), Cl⁻ (33.653 mg/L), SO₄²⁻ (34.953 mg/L), and Na⁺ (21.738 mg/L). Since H12 and H13 are irrigation and drainage ditches, they converge with the river channel at H4 through tributaries. According to the investigation, the ditches connected to this tributary collect wastewater from lobster aquaculture and farmland. Consequently, nitrite concentrations tend to accumulate and increase in these waters; the third group comprises sampling points located along the main river channel. Water from the first and second groups converges at these points. Following mixing, no significantly elevated concentrations of the measured indicators were observed. The first to the third groups represent the surface water within the polder, while the fourth group is composed of the surface water outside the polder, with relatively high loads of NO₃⁻-N (1.438 mg/L) and HCO₃⁻ (80.118 mg/L).

3.3. Characteristics of Nitrate Isotopes

The measured δ¹⁵N-NO₃⁻ and δ¹⁸O-NO₃⁻ values in surface water are summarized in Supplementary Materials Table S4. During the dry season, δ¹⁵N-NO₃⁻ ranged from 1.11‰ to 8.45‰ (mean = 4.82‰, SD = 2.54‰), and δ¹⁸O-NO₃⁻ ranged from 1.41‰ to 11.42‰ (mean = 5.70‰, SD = 2.97‰). In the wet season, δ¹⁸O-NO₃⁻ ranged from −2.71‰ to 7.09‰ (mean = 3.64‰, SD = 3.15‰), while δ¹⁸O-NO₃⁻ ranged from 2.38‰ to 9.10‰ (mean = 5.53‰, SD = 1.99‰). The overall values varied within relatively broad ranges in both seasons (Figure 3a,b).

To support qualitative nitrate source identification, the measured isotope values were compared with typical δ¹⁵N-NO₃⁻ and δ¹⁸O-NO₃⁻ ranges reported for major nitrogen sources [26,28]. The δ¹⁵N-NO₃⁻ values of 1–8‰ and δ¹⁸O-NO₃⁻ values of 2–11‰ observed in most samples generally fall within the overlapping ranges associated with ammonium-based fertilizers, soil organic nitrogen, and manure/sewage. A few negative δ¹⁵N-NO₃⁻ values (<0‰) appeared in the wet season, which may reflect minor influence from atmospheric deposition or recent fertilizer input. Overall, the isotope composition suggests that nitrate in surface water is likely derived from multiple sources with mixed contributions.

4. Discussion

4.1. Spatiotemporal Changes of Nitrogen in the Study Area

4.1.1. Temporal Variation Characteristics of Nitrogen Pollution in Surface Water

The concentrations of nitrogen species in surface water exhibited noticeable temporal patterns across the sampling periods. In general, TN and NO₃⁻-N showed higher concentrations during the dry seasons compared to the wet seasons, while NH₄⁺-N and NO₂⁻-N displayed less consistent seasonal trends. Peaks in NH₄⁺-N occurred sporadically, particularly in early wet-season months, indicating episodic inputs or transformation suppression. These variations suggest that hydrological conditions, agricultural schedules, and in situ biogeochemical processes jointly influence nitrogen dynamics over time. Field investigations have found that in the study area, the feeding of fish feed and application of fish fertilizers in fish ponds began in March, and the sowing and seedling raising of cash crops and early rice also started at this time. The fertilization of early rice ended by the end of May. Intensive agricultural activities, combined with increasing precipitation from March to May under the influence of the monsoon climate, and irrigation-related water diversion, facilitated the transport of nitrogen-rich substances from farmland into the basin’s surface water [29]. The peak in NO₃⁻-N concentrations suggests that fertilization, irrigation, and subsequent precipitation events may have led to substantial nitrate leaching [30]. NH₄⁺-N concentrations peaked in May 2022, with an average of 0.94 mg/L and a maximum of 2.43 mg/L, while NO₃⁻-N levels remained relatively low during the same period (Figure 4c). This may be because the main land use patterns within the polder are paddy fields and ponds. Aquaculture wastewater and livestock and poultry excreta contain compounds with high contents of organic nitrogen such as urea and proteins. In early summer (May), elevated temperatures promote microbial activity, facilitating the degradation of organic nitrogen compounds into NH₄⁺. However, a substantial portion of the resulting ammonium remains in the water column [31]. This pattern may reflect the timing of fertilizer application, but also suggests limited nitrification during this period. According to Supplementary Table S3, the dissolved oxygen (DO) concentration was 5.79 ± 2.25 mg/L in the wet season of 2022, with values at some sites approaching the lower threshold (<4 mg/L) for effective nitrification [32]. Under sub-optimal DO conditions, the microbial oxidation of NH₄⁺ to NO₃⁻ may be inhibited, leading to NH₄⁺ accumulation and reduced NO₃⁻ production. Therefore, both external inputs and in situ biogeochemical processes may have contributed to the observed nitrogen species distribution.

In comparison of two-year sampling data, the concentrations of TN and NO₃⁻-N in the second year are generally higher than those in the first year (Figure 4d,f). Zhao et al. reported that 39–43% of applied fertilizer nitrogen is taken up by crops in paddy soils, while approximately 50% is rapidly lost during the growing season, posing considerable risks to environmental safety [33]. The unabsorbed nitrate is prone to leaching or surface runoff, which increases TN concentrations in surrounding water bodies. The increase in annual average concentration indicates that surface water may not be able to recover quickly from nitrogen pollution [34]. Additionally, 2022 experienced a severe drought across the Dongting Lake Basin, whereas precipitation increased significantly in 2023. Greater rainfall in 2023 likely enhanced surface runoff and mobilized nitrogen from agricultural fields into adjacent water bodies. These findings suggest that both residual fertilizer nitrogen and interannual hydrological variability jointly contributed to the observed differences in nitrogen concentrations across years.

Significant differences in total nitrogen (TN) and nitrate nitrogen (NO₃⁻-N) concentrations were observed between the dry season (February, March, and November) and the wet season (May, June, and August). During the wet season, intensive rainfall promotes NO₃⁻ leaching from soil into groundwater and enhances surface runoff. This process can dilute pollutant concentrations in rivers due to the inflow of rainwater [35]. In contrast, during the dry season, the decomposition of aquatic plant residues may contribute to nutrient release into surface waters [36]. Moreover, elevated temperatures during the wet season enhance microbial activity, facilitating the decomposition and microbial assimilation of nitrogen compounds. Moreover, the growth and metabolic activities of phytoplankton will also accelerate the removal of NO₃⁻-N, causing the nitrogen concentration in the wet season to be lower than that in the dry season. As shown in Supplementary Table S3, average dissolved oxygen (DO) concentrations during the dry season are higher than those in the wet season. Additionally, slower flow rates during the dry season prolong water residence time in river channels. These conditions are favorable for nitrification, thereby increasing NO₃⁻-N concentrations in surface water.

A similar seasonal trend has been reported in the Taihu Lake Basin, where polders are widely distributed: total nitrogen (TN) concentrations in surface water tend to be higher in spring and winter, and lower in summer [37]. In the lower reaches of the Mun River Basin in northeastern Thailand, the concentration of NO₃⁻ in the dry season is significantly higher than that in the rainy season [38]. This pattern may not only be attributed to dilution by precipitation during the wet season but also to the coincidence of the rainy period with high summer temperatures, which mark the peak of crop growth. During this period, elevated nitrate uptake by crops reduces the amount of nitrate that migrates into surface water [39]. Enhanced biogeochemical activity during the wet season may also promote denitrification, further reducing NO₃⁻ concentrations. In contrast, nutrient release from the decomposition of aquatic vegetation during the dry season may contribute to elevated nitrate levels [40]. In addition to source attribution, the isotopic composition of nitrate provides indicative evidence of transformation processes within the aquatic system. The majority of δ¹⁵N-NO₃⁻ values were below 10‰, suggesting limited denitrification across most sites. However, relatively enriched δ¹⁵N-NO₃⁻ values (e.g., >7‰) observed in some dry-season samples may imply partial denitrification under suboxic conditions, which is consistent with the lower dissolved oxygen (DO) concentrations recorded at several locations (Supplementary Table S3). Conversely, samples exhibiting lower δ¹⁵N-NO₃⁻ and δ¹⁸O-NO₃⁻ values likely reflect nitrification from ammonium-based inputs.

4.1.2. Spatial Variation Characteristics of Nitrogen Pollution in Surface Water

The spatial distribution of nitrogen concentration in the surface water at 13 sampling points are shown in Figure 5a–d). During the dry season, when TN concentrations peak, the highest values were observed at H12 and H13, followed by H2 and sites H7 through H11. NO₂⁻-N concentrations during the dry season were highest at H5, with elevated values also detected at sites H8 to H11. In the wet season, NH₄⁺-N concentrations peaked at H1 and H3, followed by H2 and H7.

Overall, nitrogen concentrations in surface water exhibited distinct spatial heterogeneity across the study area. During the dry season, TN and NO₃⁻-N concentrations displayed similar spatial distribution patterns. The critical source areas are located at sampling points H12, H13, and H2. Field investigations revealed that H12 and H13 correspond to the inlet and outlet of a historical lake wetland, adjacent to which lies an active lobster aquaculture zone. The dry season aligns with key aquaculture activities such as aquatic vegetation planting and the stocking of lobster fry. Wastewater discharge from aquaculture operations is likely the primary contributor to elevated nitrogen levels at these sites. Sample points H2 and H1 are situated in the upstream section of the Pingjiang River, where land use is dominated by agriculture and the river is intersected by multiple artificial irrigation channels. The elevated nitrogen concentrations in this area may be attributed to runoff and return flows from agricultural irrigation [41].

The concentration of NH₄⁺-N shows a similar spatial distribution trend, and the concentration is also higher in the upper half of the river section. This pattern is likely associated with fertilizer application in paddy fields. During the rice season, ammonia volatilization from fertilizer sources accounts for 50.0~93.4% [42]. As for the concentration distribution of NO₂⁻-N, sampling point H5 serves as the critical source area. Effluents from lobster aquaculture are discharged into downstream waters via ditches, introducing elevated levels of nitrogen-rich metabolites and increasing the nitrogen burden in the aquatic environment. Improper feeding practices may lead to the accumulation of nitrogenous compounds beyond the assimilation capacity of aquatic microbial communities. The lack of oxygen will affect the nitrification process, preventing nitrite from being smoothly converted into nitrate [43]. This interpretation is further supported by variations in dissolved oxygen (DO) observed during the study period. As shown in Supplementary Table S3, DO concentrations fluctuated seasonally, with several sites showing reduced DO levels (<5 mg/L) during the dry season. Such suboxic conditions are known to limit nitrification and promote partial denitrification, potentially contributing to the observed accumulation of NH₄⁺-N and the relative reduction in NO₃⁻-N. Therefore, DO dynamics provide additional evidence supporting the occurrence of redox-sensitive nitrogen transformation processes in the system. Consequently, NO₂⁻-N pollution is more pronounced in the downstream segments of the river.

Dongting Lake is hydraulically connected to polder water bodies through a network of ditches and ponds, resulting in reciprocal influences on their aquatic ecological environments [44]. By analyzing nitrogen concentrations in water samples collected during both wet and dry seasons (Supplementary Materials Figure S2), it was found that the average concentrations of TN and NH₄⁺-N in the Pingjiang River section within the polder (H1–H8, H12, H13) were higher than those at the estuarine sites outside the polder (H9–H11), with a statistically significant difference observed in NH₄⁺-N concentrations (p < 0.01). Non-point source pollutants from agricultural activities are transported into internal rivers via surface runoff and drainage channels. For instance, excessive fertilizer application followed by heavy rainfall can mobilize large amounts of nitrogen into the water bodies, increasing concentrations of non-point source pollutants such as TN and NH₄⁺-N in the polder’s river system. These rivers ultimately discharge into the external lake waters. The high assimilative capacity of the Xiangjiang River facilitates pollutant dilution, thereby reducing their concentrations downstream. However, NO₃⁻-N concentrations were lower within the polder compared to the external lake sites. This is likely due to the dominant land use types within the polder, namely paddy fields and aquaculture ponds. The ammonium nitrogen fertilizer applied during rice cultivation, as well as the manure excreted by aquatic animals, livestock, and poultry, lead to a relatively high concentration of NH₄⁺-N pollutants in the surface water within the polder. But NO₃⁻-N needs to be generated from NH₄⁺-N through the process of nitrification under suitable conditions. As NH₄⁺-N-enriched water is discharged from the polder into the lake, the enhanced availability of ammonium promotes nitrification in the oxygen-rich external waters, resulting in elevated NO₃⁻-N concentrations downstream [45].

4.2. Characterization of Nitrogen Pollutant Transmission Paths

Pollutant concentrations at adjacent sampling sites often exhibit varying degrees of internal correlation. For instance, once pollutants enter the water body, they are gradually transported downstream with the flow, undergoing dilution and diffusion processes. Simultaneously, physicochemical processes such as adsorption, sedimentation, degradation, and transformation may occur. In addition, the potential role of biological uptake should not be overlooked [46]. The combined effects of these processes typically result in a gradual reduction in pollutant concentrations over time and space. Alternatively, when pollutant inputs are continuous or intense, enrichment may occur along the downstream flow direction, facilitating pollutant accumulation in lower reaches [47]. This mechanism partly explains the commonly observed pattern of higher pollutant loads in downstream sections compared to upstream areas.

Based on hierarchical cluster analysis and channel connectivity, three primary nitrogen transport pathways were identified within the polder system: Path I (H1 → H2 → H3 → H4), Path II (H12 → H13 → H4 → H5 → H6 → H7 → H8), and Path III (H8 → H9 → H10 → H11), corresponding to upper tributaries, central mixed-flow routes, and downstream discharge outlets, respectively (Figure 6). These pathways reflect the dominant hydrological and land use controlled routes of nitrogen migration across the system.

Path I represents a relatively short upstream tributary pathway. NH₄⁺-N was the dominant nitrogen species at upstream sites (H1–H3), with elevated concentrations likely attributed to localized agricultural inputs and aquaculture discharges (Figure 7). Limited hydraulic exchange and low flow velocity in this region may lead to ammonium accumulation. As water progresses downstream to H4, TN and NH₄⁺-N concentrations decrease gradually, suggesting dilution and potential in-stream transformation (e.g., nitrification). This trend is consistent with previous findings in enclosed lowland systems, where long residence times promote NH₄⁺ buildup and transformation under aerobic conditions [3,5]. Path II is the most complex and hydrologically active pathway, integrating water from aquaculture ponds (H12–H13), main channels (H5–H6), and adjacent agricultural areas. Nitrogen concentrations along this path showed significant variability. At H13 and H7, TN and NH₄⁺-N concentrations peaked, reflecting input from shrimp and lobster farming, as well as fertilizer-rich surface runoff. The appearance of elevated NO₃⁻-N concentrations toward H8 suggests active nitrification during downstream transport, likely facilitated by increased dissolved oxygen levels (see Supplementary Table S3) and turbulence in the wider main channel. This aligns with prior studies showing enhanced nitrification in open, well-oxygenated lowland waterways [5]. The mixing of multiple pollution sources along Path II also highlights its vulnerability to diffuse agricultural pressure.

Path III is a downstream export route draining into Dongting Lake. NO₃⁻-N concentrations along this path were consistently high, particularly at H10–H11, which are directly influenced by discharge from upstream pathways and surrounding croplands. This suggests that nitrate becomes the dominant nitrogen form exported from the polder system, likely due to upstream nitrification processes. The transition from NH₄⁺-N dominance in upstream/inner regions to NO₃⁻-N dominance at the outlet supports the existence of an internal nitrogen conversion continuum. This is further reinforced by spatial isotope evidence (see Section 4.3), indicating ammonium-based fertilizers and soil nitrogen as major sources. This spatial transition from ammonium to nitrate dominance in Path III reflects a typical nitrogen transformation pattern driven by increased oxygen availability and flow connectivity. Similar downstream accumulation of nitrate along extended flow paths has been observed in lowland drainage systems, where prolonged travel times and artificial channel networks facilitate nitrification and nitrate export [48].

In summary, nitrogen migration in the polder system exhibits spatial heterogeneity, shaped by the interaction of land use intensity, water flow connectivity, and biogeochemical conditions. The three identified pathways show functional differentiation: Path I as an input-accumulation section, Path II as a mixed and transitional zone, and Path III as a nitrate-export route. These findings provide a basis for spatially targeted management strategies, such as inlet control, aquaculture discharge regulation, and downstream buffer construction.

4.3. Identifying Nitrate Sources Using Stable Nitrogen and Oxygen Isotope Compositions

Spearman correlation analysis (Figure 8) revealed a significant positive correlation between TN and both NO₃⁻-N and NO₂⁻-N concentrations in surface water within the polder (p < 0.01). Outside the polder, the temporal variation in TN concentration closely mirrors that of NO₃⁻-N, suggesting that changes in nitrate levels largely drive TN dynamics in these regions. Given its stability, strong correlation with TN, and substantial contribution to total nitrogen, NO₃⁻-N was selected as the representative nitrogen species for source analysis. Through the study of δ¹⁵N-NO₃⁻ and δ¹⁸O-NO₃⁻, the source of nitrogen in water bodies can be effectively analyzed. A significant negative correlation was observed between NH₄⁺-N and NO₃⁻-N concentrations within the polder (p < 0.05), possibly reflecting redox-driven transformation processes between these nitrogen species.

As shown in Figure 3, δ¹⁵N-NO₃⁻ and δ¹⁸O-NO₃⁻ values during the dry and wet seasons fall into distinct source zones, indicating notable seasonal differences in nitrate sources among sampling points. Compared to the dry season, lower δ¹⁵N-NO₃⁻ and higher δ¹⁸O-NO₃⁻ values observed during the wet season may result from the dilution and mixing effects of atmospheric deposition [40].

The four groups shown in Figure 3 were derived from hierarchical cluster analysis (HCA) based on nitrogen species concentrations and related water quality variables. Although some intra-group variation is present, the grouping reflects underlying similarities in pollution characteristics. This classification highlights potential functional units within the polder system where nitrogen dynamics share common patterns, and provides a meaningful basis for interpreting differences in nitrogen transport and accumulation. During the wet season, the isotopic composition of nitrate in Group 1 is primarily distributed within the source domain of ammonium-based fertilizers and atmospheric precipitation. In the dry season, δ¹⁵N-NO₃⁻ values increase and gradually shift toward the isotopic signature of soil-derived nitrogen. Most samples fall within the overlapping source region of ammonium fertilizer, precipitation, and soil nitrogen. During the dry season, ammonium fertilizers are either taken up by crops or immobilized as organic nitrogen in soils. Simultaneously, NH₄⁺-N volatilization leads to isotopic enrichment, increasing δ¹⁵N values in the remaining ammonium and its nitrification products [49]. The source of nitrate in the second group is complex. In the wet season, nitrate originates from a mixture of ammonium fertilizer, soil nitrogen, and manure and sewage inputs. The sources remain the same during the dry season, which may be related to lobster aquaculture and paddy field cultivation near the sampling points of the second group. In Group 3, nitrate in the wet season also originates from a mixed source zone comprising ammonium fertilizer, soil nitrogen, and manure and sewage. However, during the dry season, the δ¹⁵N ratio decreases to within the range of ammonium fertilizer sources. As shown in Figure 1, the residential areas in rural townships are primarily distributed along the riverbanks. Consequently, during the wet season, sewage and manure from residential areas are more likely to be mobilized by surface runoff and discharged into the receiving water bodies. This phenomenon diminishes during the dry season due to reduced precipitation and surface runoff. The fourth group represents the water body outside the polder. In the wet season, isotopic values in this group cluster within the overlapping zone of soil nitrogen and manure/sewage sources, indicating mixed inputs. In the dry season, nitrate sources include ammonium fertilizers, manure, and sewage, as well as soil nitrogen.

Overall, the isotopic compositions of most surface water samples fall within the characteristic signature ranges of ammonium-based fertilizers and soil nitrogen, indicating that nitrate pollution primarily originates from ammonium fertilizers and soil nitrogen, with additional contributions from manure and sewage. Identifying nitrate sources provides a scientific basis for targeted nitrogen pollution control and ecological restoration, and is essential for protecting water quality and ensuring sustainable water resource management.

5. Conclusions

This study examined the spatiotemporal distribution, transport pathways, and sources of nitrogen pollution in surface waters of a typical polder system in the Dongting Lake basin, using hydrochemical indicators, dual-isotope analysis, and statistical methods. Results showed that NO₃⁻-N and NH₄⁺-N were the dominant nitrogen forms, with concentrations higher during the dry season. Spatially, NH₄⁺-N was enriched in inner tributaries, while NO₃⁻-N dominated near outlet sections. Three nitrogen transport routes were identified—upstream ammonium accumulation (Path I), mixed forms in confluence zones (Path II), and nitrate-dominated export pathways (Path III)—reflecting in-stream transformation and pollutant convergence. Isotopic evidence pointed to ammonium fertilizers and soil nitrogen as the main nitrate sources, with localized inputs from manure and sewage. These patterns reflected differences in land use and nitrogen transformation. The findings underscore the need for targeted management strategies, such as optimized fertilizer use, aquaculture effluent regulation, and buffer zones near outlets, tailored to the identified transport and source patterns.

Despite offering new insights, this study is limited by the lack of continuous temporal and hydrological data. Future research should incorporate high-frequency monitoring and coupled isotope–hydrodynamic modeling to support refined nitrogen management in complex lowland agricultural systems.