The Characteristics and Source Contribution Analysis of Nutrients in Water Bodies of Small Watersheds in the Pearl River Delta

Abstract

With the rapid development of urbanization in rural areas of China, various environmental issues have become increasingly prominent, particularly the water pollution problems in small rural watersheds, which have garnered considerable attention. Comprehensive management of small watersheds requires an initial analysis of the sources and characteristics of water pollution. This study focuses on small rural watersheds in the Pearl River Delta. Based on the characteristics of the watersheds, 35 water quality monitoring stations were set up to collect water quality data. Cluster analysis was used to study the spatial distribution characteristics of water quality indicators at each monitoring point. Further, factor analysis methods (PCA/FA) and Absolute Principal Component Scores-Multiple Linear Regression (APCS-MLR) models were employed to identify water quality influencing factors and quantify pollution source contributions. Finally, the comprehensive index method for eutrophication assessment was used to evaluate and analyze the potential eutrophication pollution risk in the watersheds. The results indicate significant pollution in the water quality of rural small watersheds in the study area, with varying degrees of pollution over time and space. During the wet season, water quality is mainly influenced by agricultural nutrients, followed by biochemical factors. In the normal and dry seasons, water quality is primarily affected by oxygen-consuming organic pollutants, followed by eutrophication factors. The comprehensive eutrophication evaluation shows that the overall water quality in the watershed is better during the wet season, with a lower risk of eutrophication; during the normal season, the overall water quality is poorer, with the highest eutrophication risk in the midstream; during the dry season, the upstream and midstream water quality is better, while the downstream water quality is poorer. In contrast, the pond water exhibits a higher risk of eutrophication during the wet season compared to the normal and dry seasons. This is mainly due to the peak of fish farming during the wet season, which results in a heavier load on the water body. This study provides effective data support for the water environment management of rapidly developing rural small watersheds.

1. Introduction

China’s rural areas are currently experiencing a period of rapid urbanization and economic development, bringing opportunities to the countryside. However, this development process has also led to increasingly prominent environmental issues in small rural watershed [1]. Small watersheds, typically less than 50 square kilometers in size, are predominantly located in county-level towns and villages [2,3]. These watersheds encompass multiple critical elements of modern rural watershed systems, including mountains, water bodies, forests, fields, roads, and villages. Therefore, the health of water bodies in small watersheds plays a crucial role in constructing a good rural ecological environment. Balancing the protection of small watershed ecosystems with economic development is not only an intrinsic requirement for harmonious coexistence between humans and nature but also an important guarantee for sustainable rural development. Studying the overall condition of water quality and the temporal and spatial variations of pollutants in small rural watersheds as well as analyzing their causes are essential prerequisites for the management and protection of these watersheds.

The Pearl River Delta (PRD) region, as a pioneer area of China’s reform and opening-up, has gradually developed into one of the most economically vibrant regions in the country and has become a global manufacturing and import-export base [3,4]. In recent years, with the expansion of urbanization and the increasing population in the PRD [3], the discharge of industrial and domestic sewage, as well as the use of agricultural fertilizers, have brought nutrients into water bodies through various pathways, leading to the deterioration of river and lake water quality. Numerous studies [5,6,7] have shown that the main causes of river pollution are excessive human activities and discharges, such as agriculture, livestock farming, urbanization, and industrialization. These activities lead to significant nutrient inputs into rivers, resulting in eutrophication problems. Therefore, analyzing the nutrient status of water bodies in the watershed is an important aspect of understanding the water quality in the PRD and a necessary step for managing the watershed environment.

Many scholars have analyzed the water environment of rural watersheds in the PRD. For example, Chen et al. [8] used a comprehensive water quality index to analyze and evaluate the current state of water pollution in rural rivers in a town in the PRD, finding severe pollution in these rivers. Long et al. [9] conducted a systematic analysis of the trends and causes of water environment quality changes in the PRD from 2001 to 2015 based on the Guangdong Provincial Environmental Quality Report, discovering continuous improvement in river water quality during this period, but high discharge intensity and significant water environment pressure, with rural domestic pollution becoming the primary source of water pollution. Huang et al. [10] analyzed the impact of rural domestic pollution, agricultural runoff, and livestock farming pollution on the rural water environment in four regions of Guangdong Province (the PRD, eastern Guangdong, western Guangdong, and northern Guangdong) using the Nemerow comprehensive index method at the county level, finding a higher number of heavily polluted and severely polluted counties in the PRD. These studies have provided important insights into the characteristics of water environment pollution in rural watersheds in the PRD in terms of water quality analysis, water resource carrying capacity and utilization, pollution transmission, and ecological environment protection needs and pressures. However, rural small watersheds in the PRD have their unique characteristics, such as complex pollution sources, overlapping impacts of industrial and agricultural activities, and diverse water body types. For example, ponds and small reservoirs (collectively referred to as ponds) are also widely distributed in rural areas. Ponds, a type of small inland wetland, not only serve functions in agricultural production, such as water storage for irrigation and aquaculture, but also have ecological functions in intercepting and assimilating nitrogen and phosphorus pollutants in the watershed [11]. However, against the background of global climate change and rapid urbanization, the rapid development of industrial and agricultural activities has led to a continuous reduction in the number of ponds [12], degradation of their ecological functions such as flood storage, drought resistance, and pollution interception, and serious deterioration of water quality [13], weakening their ability to purify the surrounding environment. Therefore, monitoring the water quality and analyzing the nutrient status of ponds are also essential parts of studying the eutrophication status of water bodies in the PRD. However, there is still a certain gap in the research on changes in water quality and its influencing factors in small rural watersheds in the PRD in recent years, especially in the spatial and temporal analysis of various water bodies. Eutrophication is an ecological process in water bodies where excessive input of nutrients such as nitrogen and phosphorus leads to abnormal algae growth. Dissolved oxygen and organic matter play key roles in this process, while water temperature influences these metrics through physical, chemical, and biological processes, ultimately regulating the eutrophication process. Specifically, eutrophication is a complex ecological process driven by the overaccumulation of bioessential elements in water bodies, resulting in accelerated primary production, algal blooms, and subsequent degradation of aquatic ecosystems. The core of this phenomenon is bioessential elements, particularly nitrogen (N) and phosphorus (P), which are limiting nutrients for phytoplankton growth. The interactions between these nutrients, dissolved oxygen (O₂), and organic matter control the nutrient status of aquatic systems, affecting water quality and ecosystem health. Bioessential elements are necessary for the biosynthesis of organic compounds in aquatic organisms. Nitrogen, present in the form of nitrate (NO₃⁻-N), nitrite (NO₂⁻), and ammonium (NH₄⁺), is a fundamental component of proteins and nucleic acids, whereas phosphorus, primarily as orthophosphate (PO₄³⁻-P), is crucial for energy transfer (ATP) and cell structure (phospholipids). Organic matter, sourced from both autochthonous (algae, macrophytes) and allochthonous (terrestrial runoff, wastewater) origins, serves both as a carbon substrate for microbial metabolism and as a sink for dissolved oxygen through decomposition. The microbial mineralization of organic matter consumes O₂, leading to hypoxic or anoxic conditions, which exacerbate eutrophication by releasing sediment-bound phosphorus, further promoting a positive feedback loop of productivity.

Therefore, a comprehensive analysis of the water quality status and spatial and temporal distribution characteristics of various water bodies in typical rural small watersheds in the PRD and a scientific analysis of their pollution sources and characteristics are of great significance for watershed pollution prevention and control as well as sustainable development. This study focuses on evaluating a set of chemical indicators for eutrophication dynamics: dissolved oxygen (DO) is crucial for aerobic aquatic organisms, and DO levels fluctuate daily due to photosynthesis (O₂ production) and respiration (O₂ consumption). Hypoxia (DO < 2 mg/L) is typically caused by the excessive decomposition of organic matter. Nutrients (N and P) are measured as total nitrogen (TN) and total phosphorus (TP), and their concentrations are associated with algal biomass. Inorganic nitrogen (NH₄⁺, NO₃⁻-N) has particularly high bioavailability. pH is influenced by the CO₂ flux from respiration and photosynthesis, affecting the chemical form of nutrients.

To obtain the temporal and spatial variations of the physical and chemical characteristics of water bodies, continuous and periodic monitoring is required. However, the resulting database is large and complex, necessitating reliable statistical analysis tools. Multivariate statistical techniques, such as Principal Component Analysis (PCA) and Factor Analysis (FA), are widely used for evaluating temporal and spatial variations and interpreting large and complex water quality datasets. PCA/FA is a dimensionality reduction technique that provides information through simpler data representations by extracting the most important factors. PCA/FA is commonly used to determine data structure and provide qualitative information on potential pollution sources [14]. However, PCA/FA alone cannot quantify the contribution of pollution sources to each variable [15]. Receptor-based models, such as Absolute Principal Component Scores-Multiple Linear Regression (APCS-MLR), can be used for this purpose. In atmospheric environment research, the APCS-MLR method is primarily used for pollution source identification and allocation. In recent years, this technique has been increasingly applied to the allocation of water environment pollution sources. For example, Du et al. [16] used the APCS-MLR method to determine the main pollution sources affecting the Chanba River, and Du et al. [17] applied APCS-MLR to determine the contribution of different pollution sources to water quality based on years of water quality monitoring data from Caohai Lake and Waihai Lake in Dianchi Lake. Therefore, this study combines the two methods using PCA/FA to extract water quality influencing factors and APCS-MLR to determine the contribution rates of various influencing factors to water quality indicators in rivers and ponds.

Based on the above considerations, the main objective of this study was to gain an in-depth understanding of the temporal variations, spatial distribution, pollution sources, and pollution characteristics of various water bodies in small watersheds in the economically developed PRD region to infer the causes and processes of pollution. Specifically, observation points were set up in the study area of small rural watersheds in the PRD, water samples were collected and analyzed, and methods such as the comprehensive eutrophication assessment index, PCA/FA, and APCS-MLR were used to analyze a large data matrix of nine water quality parameters (approximately 600 observations) from 35 different locations in the small watersheds in 2022. The aims were to (1) determine the temporal and spatial distribution characteristics of water quality indicators in the watershed; (2) identify and interpret possible pollution sources and estimate their contributions to the concentrations of water quality indicators; and (3) evaluate the potential distribution of eutrophication pollution in the watershed. The goal was to reveal the characteristics and causes of pollution in small rural watersheds and provide data support for the water environment management of rural small watersheds in the PRD.

2. Study Area and Methods

2.1. Overview of the Study Area

The study watershed is located in the Pearl River Delta (PRD), covering an area of 70.70 km². The two main streams in the watershed are the Zhucun Canal and the Nangang River, which converge at the outlet and flow into the Xifu River, a primary tributary of the Dongjiang River. The study area has a subtropical monsoon climate, with an average annual temperature of 21.9 °C and annual rainfall ranging from 1623.6 to 1899.8 mm. The frost-free period is about 346 days. The topography of the study area is characterized by low hills and basins, with agriculture being the main land use. The primary economic crops include citrus, lychee, and longan. Due to limited groundwater recharge, the watershed mainly relies on precipitation for water supply. According to the Guangdong Yearbook, rainfall is concentrated between April and September, accounting for about 80% of the annual total, while the period from October to March of the following year is the dry season. Additionally, December to February is the cool winter season. Based on these characteristics, the study divides the watershed into three periods: wet season (April to September), normal season (March, October, November), and dry season (December to February). Based on the characteristics and geographical location of the watershed, 35 representative sampling points were selected using GPS positioning in the study watershed (specific locations are shown in Figure 1. Among these, there are 7 pond sampling points (P) and 28 river sampling points (Z), the overview of the 7 ponds is shown in

Table 1. Overview of the ponds in the study area.

Number	Perimeter/m	Area/m2	Location	Management Situation
P01	213.56 ± 5.64	3091.03 ± 23.36	Downstream (113.6967, 23.2557)	Raising ducks and fish in enclosed ponds, with residential areas directly discharging domestic sewage into them.
P02	262.68 ± 3.21	3264.06 ± 2.42	Midstream (g113.6966, 23.2793)	Raising chickens, ducks, and fish, with residential areas and road. Standing water drainage included.
P03	176.93 ± 0.24	1435.70 ± 10.66	Midstream (113.7040, 23.2868)	Fish farming, with agricultural planting areas surrounding the site, and lychee trees planted around the perimeter.
P04	292.76 ± 0.90	5878.50 ± 33.84	(g113.7009, 23.2991) is a coordinate point in the midstream sector.	The farm raises chickens, ducks, cattle, sheep, and fish, surrounded by farmland.
P05	96.14 ± 0.71	558.76 ± 3.59	Midstream (g113.7084, 23.2942)	Raising ducks and fish, residential areas, with domestic sewage directly discharged.
P06	348.31 ± 1.30	5103.03 ± 52.13	Midstream (g113.7281, 23.2932)	Fish farming, surrounded by ponds, farmland, and banana trees, with water hyacinths covering the water surface.
P07	240.10 ± 1.10	2478.83 ± 15.52	Midstream (g113.7314, 23.2860)	Fish farming, surrounded by farmland and residential areas, with upstream drainage from farmland and a paper mill.

. There are no large livestock farms in the study area; almost all households have small-scale scattered livestock. Aquaculture is concentrated in the middle and lower reaches of the watershed. Residents in the watershed obtain their water supply through a pipeline system. There are no large wastewater treatment plants for domestic sewage; instead, there are two small wastewater treatment facilities using anaerobic + constructed wetland processes. According to on-site investigations, each facility has a treatment capacity of 120 m³/day, serving a total population of approximately 600 people, which is insufficient to cover the entire watershed. Therefore, in addition to agricultural wastewater, domestic wastewater generated by residents may also be directly discharged into the river or enter the river through ponds for dilution and natural purification.

2.2. Research Methods

2.2.1. Sampling Methods

This study conducted location-based sampling from January 2022 to December 2022. The specific process involved the following: setting one sampling point at each cross-section, collecting surface water samples (30 cm deep) using a long pole or directly, placing the water samples in polyethylene plastic bottles, preserving the samples by acidification or storing them in a refrigerator at 0–4 °C, and then immediately analyzing them in the laboratory after settling. Sampling was conducted in mid-month, with increased sampling frequency to twice per month from April to August due to abundant rainfall, resulting in a total of 17 sampling events. Phosphorus indicators were monitored starting from late April. For missing or uncollected water quality parameters, interpolation methods were used to generate data for the respective months to facilitate subsequent analysis. For water quality parameters below detection limits, the detection limit was used to supplement the parameter data. The collection, storage, and transportation of water samples strictly adhered to the fourth edition of the national environmental protection standards (GB 3838-2002 [18], https://wzq1.mee.gov.cn/ywgz/fgbz/bz/bzwb/shjbh/shjzlbz/200206/t20020601_66497.shtml, accessed on 1 January 2025) for water sample testing methods.

2.2.2. Indicator Measurement Methods

This study selected nine conventional water quality monitoring indicators: water temperature (WT), pH, electrical conductivity (EC), dissolved oxygen (DO), ammonia nitrogen (NH3-N), nitrate nitrogen (NO₃⁻-N), total nitrogen (TN), orthophosphate (PO₄³⁻-P), and total phosphorus (TP) for detection and analysis. During water sample collection, a portable multi-parameter water quality analyzer was used to measure on-site indicators such as water temperature (WT), pH, and dissolved oxygen (DO). Other indicators were measured using Smartchem 200. All water quality indicators detection equipment is shown in

Table 2. Water quality indicators detection equipment.

Water Quality Monitoring Indicators	Unit	Instrument and Equipments
WT	°C	YSI (YSI Inc., Yellow Springs, OH, USA)
pH	-
EC	μS/cm
DO	mg/L
NH3-N	mg/L	Smartchem 200 [19] (Westco Scientific Instruments, Brookfield, CT, USA)
NO3--N	mg/L
TN	mg/L
PO43−-P	mg/L
TP	mg/L

.

2.3. Data Treatments

2.3.1. Cluster Analysis (CA)

Cluster Analysis (CA) is an unsupervised pattern detection method that clusters large-scale data of each entity into several groups to identify the characteristics of each group. In this study, we performed clustering based on the order of distances between objects. The analysis of the standardized dataset was conducted using Ward’s method and squared Euclidean distance, which provided spatial similarity information for the rivers in the Zhucun Small Watershed. Ward’s method recommends merging pairs with the minimum variance in a connected manner based on the variance of the entities that make up each cluster. Euclidean distance generally represents the similarity between two samples; “distance” refers to the “difference” between the analytical values of the two samples. CA uses one-way Analysis of Variance (ANOVA) (p < 0.05) to analyze the significant differences in spatiotemporal water quality parameters and post hoc analysis to analyze the significant differences between clusters.

2.3.2. Principal Component Analysis/Factor Analysis (PCA/FA)

Principal Component Analysis/Factor Analysis (PCA/FA) was used in this study to extract water quality influencing factors. PCA/FA generates “principal components” (PCs) by linearly combining potentially correlated variables in a complex dataset and then polarizes the variable loadings on the principal components through maximum variance rotation, forming new principal components called “variance factors” (VFs). This method better explains the data. Variable loadings are defined as low, medium, and strong when values fall within the ranges of 0.3–0.5, 0.5–0.75, and 0.75–1, respectively (Huang et al., 2010) [20]. Multiple variables with strong loadings on the same variance factor typically indicate that they are influenced by the same source. For PCA/FA, the dataset requires sufficient degrees of freedom (sample size-number of variables > 50) and should be assessed using the Kaiser–Meyer–Olkin (KMO) test and Bartlett’s test of sphericity (p): KMO > 0.5 and p < 0.05 indicate that the dataset is suitable for PCA/FA [21].

2.3.3. Absolute Principal Component Scores-Multiple Linear Regression Model (APCS-MLR)

Since the dataset is normalized, the sample factor scores generated by PCA/FA cannot be directly used to calculate the contribution of factors. In 1985, Thurston and Spengler [22] proposed that by subtracting the factor scores of zero-concentration samples, a multiple linear regression equation can be constructed to quantify the actual contribution of each source to the variables. The observed parameter concentration can be expressed as a linear combination of different sources as follows:

$$C_i={\displaystyle \sum _{j=1}{a}_{ji}APC{S}_{ji}+b_i}$$

(1)

where $C_i$ is the observed concentration of parameter i, $b_i$ is the constant term of the equation (representing the contribution of undefined sources), $a_{ji}$ is the regression coefficient of source j for element i, and $a_{ji}APC{S}_{ji}$ represents the contribution of source j to the concentration of element i.

Compared to the PMF and Unmix models, the APCS-MLR model performs more stably because it is less sensitive to outliers in the data and more easily identifies source spectra obtained through orthogonal transformation. Additionally, to address potential negative contributions from sources due to possible negative coefficients in multiple linear regression calculations, Haji et al. [23] proposed reallocating pollution sources using the absolute value of negative contributions. Therefore, the source contribution calculation in this study is as follows:

$$P{C}_i=\left\{\begin{array}{c}\left|a_{ji}\overline{APC{S}_{ji}}\right|/\left|b_i\right|+{\displaystyle \sum _{j=1}^p\left|a_{ji}\overline{APC{S}_{ji}}\right|}, \mathit{identified\; sources}\\ \left|b_i\right|/\left|b_i\right|+{\displaystyle \sum _{j=1}^p\left|a_{ji}\overline{APC{S}_{ji}}\right|}, \mathit{undefined\; sources}\end{array}\right.$$

(2)

where PC_i is the contribution of the i source, b_i is the contribution of undefined sources, and p is the number of identified sources.

2.3.4. Eutrophication Assessment Method

This study used a logarithmic universal index formula (Equation (3)) [24] to calculate the comprehensive eutrophication assessment index ($EI$) to evaluate the potential eutrophication level of the water bodies.

$$EI={\displaystyle \sum _{j=1}^n{W}_j\times E{I}_j=10.77}\times {\displaystyle \sum _{j=1}^n{W}_j}\times {\left(In{X}_j\right)}^{1.1826}$$

(3)

where j is the indicator index, n is the total number of indicators, W_j is the normalized weight value of indicator j. This study selected TN, TP, orthophosphate (PO₄³⁻-P), ammonia nitrogen (NH3-N), and dissolved oxygen (DO) for evaluation, each with a weight of 0.2, $E{I}_j$ is the universal index for eutrophication assessment of indicator j and $X_j$ is the standardized value of the monitoring result of indicator j, with the conversion method referenced from the literature [24].

According to this literature, the eutrophication level of freshwater bodies is divided into five levels: $EI$ ≤ 20.00 is Oligotrophic, 20.00 < $EI$ ≤ 39.42 is Mesotrophic, 39.42 < $EI$ ≤ 61.29 is Eutrophic, 61.29 < $EI$ ≤ 76.28 is Hypertrophic, 76.28 < $EI$ ≤ 99.77 is Extremely hypertrophic.

3. Results and Analysis

3.1. Temporal and Spatial Variations in Water Quality Characteristics of the Watershed

In this study, based on the meteorological condition and hydrological characteristics of the watershed, a hydrological year is divided into three periods: wet season (April to September), normal season (March, October, November), and dry season (December to February). Spatially (as shown by the vertical red line in Figure 2), the 28 river monitoring sections can be categorized into three groups in the research area:

Group one (G1): Located in the upper reaches of the Zhucun Canal and the upper reaches of the Nangang River, including sections Z05 to Z13, Z18, and Z22 to Z28.

Group two (G2): Located in the mid- to lower reaches of the Zhucun Canal and the mid-reaches of the Nangang River, including sections Z01, Z04, Z14 to Z16, and Z19 to Z21.

Group three (G3): Located in the lower reaches of the Zhucun Canal and the lower reaches of the Nangang River, including sections Z02, Z03, and Z17.

Spatial differences in watershed water bodies can reflect trends in pollutant variations. Therefore, this study analyzes the differences in water quality monitoring indicators on both temporal and spatial scales. As shown in Figure 3a, on the temporal scale, water temperature and nitrate concentration are highest during the wet season and lowest during the dry season; ammonia nitrogen and total nitrogen concentrations are highest during the normal season and lowest during the dry season; total phosphorus concentration is highest during the normal season and lowest during the wet season; electrical conductivity (EC), pH, and dissolved oxygen concentrations are highest during the dry season and lowest during the wet season; phosphate concentration is highest during the dry season and lowest during the normal season. At the same time, in terms of spatial scale (as shown in Figure 3b, G1~G3 are the standardized analysis results for the river sections, and G4 is the analysis result for the seven pond cross-sections), the temperature and nitrate concentration in Group G1 show greater variability compared to the other groups. Group G2 has higher concentrations of nitrate nitrogen, phosphate, and conductivity. Group G3 has higher concentrations of total nitrogen, ammonia nitrogen, and conductivity compared to the other groups, with significant differences. The water temperature, pH value, and dissolved oxygen concentration in the G4 pond group are higher than those in the other groups, while the nutrient concentration is lower than in the other groups.

The main water quality indicator data from the monitoring stations in the watershed of this study are shown in

Table 3. Main water quality index of river basin water monitoring station.

Period	Value	pH	EC/ (μS/cm)	Water Quality Index Concentration/(mg/L)
				DO	TN	NH3-N	NO3−-N	TP	PO43−-P
Whole year	MIN	6.69	45.10	0.86	ND	ND	ND	ND	ND
	MAX	13.24	840.70	15.94	12.41	12.80	2.97	2.94	0.95
	AVG	8.79	175.68	6.90	1.58	0.74	0.61	0.19	0.07
	CV	0.10	0.22	0.28	0.69	0.78	0.76	0.66	0.84
Wet season	MIN	6.69	45.10	0.86	0.09	ND	ND	ND	ND
	MAX	12.78	710.83	15.79	10.94	6.39	2.31	0.82	0.61
	AVG	8.50	162.08	6.24	1.51	0.70	0.66	0.17	0.06
	CV	0.07	0.20	0.23	0.65	0.67	0.60	0.54	0.75
Normal season	MIN	6.82	51.90	1.39	0.42	0.01	ND	ND	ND
	MAX	13.24	764.20	15.94	12.41	12.80	2.97	2.94	0.51
	AVG	8.83	187.76	6.85	2.25	0.92	0.65	0.25	0.06
	CV	0.13	0.19	0.24	0.39	0.67	0.72	0.65	0.65
Dry season	MIN	7.93	54.70	1.84	ND	ND	ND	-	-
	MAX	13.15	840.70	15.54	10.42	4.69	2.60	-	-
	AVG	9.32	190.78	8.28	1.05	0.64	0.48	0.21	0.12
	CV	0.09	0.16	0.16	0.71	0.57	0.81	-	-

Note: ND indicates values below the detection limit.

. The coefficient of variation (CV) is used as an indicator to assess variability, reducing the impact of differences in units and means across different datasets. From the table, it can be observed that the CV values for all water quality parameters range from 0.07 to 0.84, indicating substantial variability in the water quality parameters within the study area. Among these, the variability of total nitrogen, ammonia nitrogen, nitrate nitrogen, total phosphorus, and phosphate is generally greater than 0.5 across different periods, suggesting that these five water quality parameters exhibit uneven temporal and spatial distribution throughout the watershed and are influenced to varying degrees by external pollution sources. A detailed analysis of each water quality indicator is as described further.

pH is an important indicator reflecting the acidity or alkalinity of water. The study area shows considerable fluctuation in pH values (ranging from 6.69 to 13.24), with an average of 8.79, indicating a weakly alkaline condition. Pond waters have the highest average pH value of 10.14, while the average pH value across river and canal sections, excluding ponds, is 8.45.

EC is a comprehensive reflection of the salinity concentration in water and is one of the important indicators of water quality [25]. Overall, the variability of EC within each group across different periods is small, while the variability between groups is large, especially in ponds. The highest EC values during the wet and normal seasons are found in this area. The maximum value during the dry season (840.7 μS/cm) comes from the watershed’s outlet area, followed by ponds (820.4 μS/cm).

DO refers to the molecular oxygen dissolved in water and is a crucial basis for understanding the self-purification ability of water. Lower concentrations indicate less oxygen content in the water, suggesting poorer water quality. The table shows that the DO concentration at the small watershed observation points in the study area varies significantly across the watershed. Among the samples, 26.90% have DO concentrations below Class III water quality standards (5 mg/L), with 6.67% and 2.14% of samples falling below Class IV (3 mg/L) and Class V (2 mg/L) standards, respectively. This indicates poor water quality and significant pollution in the target watershed.

To better represent the temporal and spatial differences in nutrient water quality, the study uses ArcGIS 10.2 mapping to display the spatial distribution of nutrient water quality indicators at each sampling section during the wet, normal, and dry seasons (Figure 4). Overall, in terms of the temporal and spatial distribution characteristics of nutrients, the average values of total nitrogen (TN) during the wet, normal, and dry seasons are 1.51, 2.45, and 1.05 mg/L, respectively; the average values of ammonia nitrogen (NH3-N) are 0.70, 0.92, and 0.64 mg/L, respectively; and the average values of nitrate nitrogen (NO₃⁻-N) are 0.66, 0.65, and 0.48 mg/L, respectively, all showing significant differences.

Comparing the upstream regions, due to the accumulation of pollutants, nutrients generally show higher concentration values downstream. Specifically, for nitrogen nutrients, the total nitrogen (TN) values are higher in residential areas, with higher TN and ammonia nitrogen concentrations in the downstream outlet areas. Except for the reservoir outflow, nitrate nitrogen values are generally high throughout the watershed. For phosphorus nutrients, phosphate (PO₄³⁻-P) and total phosphorus (TP) values are highest in residential areas but decrease towards the downstream outlet, possibly due to some dilution and assimilation effects from tributary convergence, although they generally remain at high levels.

This analysis highlights the significant temporal and spatial variability in nutrient concentrations across different regions and seasons within the watershed, which has shown the influence of both natural processes and human activities on water quality. To identify the main factors contributing to water pollution, the study employs a combined analysis using PCA/FA and APCS-MLR. Based on the previously mentioned temporal (wet, normal, dry seasons) and spatial (G1, G2, G3, G4) divisions of the watershed, the influencing factors of water quality and pollution sources can be extracted and identified separately for both temporal and spatial scales. This provides effective strategies for managing and mitigating water pollution in the watershed.

3.2. Temporal and Spatial Factor Analysis and Pollution Source Indentification of the Watershed

3.2.1. Data Standardization and Correlation Testing

This paper selected nine water quality indicators, used mean imputation to handle missing original data in order to improve data quality, conducted standard transformation of the original data through range standardization, and performed KMO and Bartlett’s sphericity test to examine the correlation between variables, with the results shown in

Table 4. KMO and Bartlett test results.

The KMO Measure of Sampling Adequacy		0.797
	Test value χ2	1522.691
Bartlett’s sphericity test	Degrees of freedom (df)	136
	Significance Level Sig	0

. The KMO measure was 0.797, close to 0.8, indicating that the data are suitable for factor analysis. Meanwhile, the Bartlett’s sphericity test statistic had a p-value close to 0, suggesting a strong correlation between the nine variables.

3.2.2. Temporal Factor Analysis and Pollution Source Identification

Based on the PCA/FA method, it is first necessary to extract the number of factors influencing water quality during each period. The factor scores are calculated using the factor score coefficient matrix from the factor analysis, and the total scores of the factors are computed based on the variance contribution rates of each factor. According to the principle of eigenvalue > 1, this study extracted two, two, and three factors for the wet, normal, and dry seasons, respectively, as shown in

Table 5. Rotation factor load matrix for three periods.

Indicators	Wet Season		Normal Season		Dry Season
	Factor 1	Factor 2	Factor 1	Factor 2	Factor 1	Factor 2	Factor 3
WT	−0.02	0.90	−0.28	0.81	0.10	−0.73	0.24
pH	−0.13	0.87	−0.14	0.86	−0.26	−0.02	0.82
EC	0.89	0.14	0.81	−0.01	0.28	0.68	−0.02
DO	−0.39	0.78	−0.36	0.83	−0.50	−0.19	0.73
NH3-N	0.71	−0.33	0.89	−0.19	0.78	0.55	−0.02
NO3−-N	0.11	−0.80	−0.04	−0.73	−0.08	0.48	−0.63
TN	0.93	−0.19	0.96	−0.19	0.71	0.57	−0.14
PO43−-P	0.63	−0.53	0.92	−0.18	0.89	−0.11	−0.29
TP	0.92	−0.20	0.45	−0.16	0.94	0.04	−0.21
Eigenvalue	4.77	2.11	4.50	1.89	4.36	1.45	1.12
Variance Contribution Rate/%	39.73	36.72	40.37	30.62	35.57	21.22	20.14
Cumulative Contribution Rate/%	39.73	76.45	40.37	70.99	35.57	56.79	76.93

Note: The bold number were indicators with factor loadings > 0.7 were selected for interpretation (with some selected > 0.6).

. There were two main factors influencing water quality in the wet season, two in the normal season, and three in the dry season. The factor scores reflect the distribution of different pollution sources at each monitoring section. In the wet, normal, and dry periods, except for Factor 2 in the wet season, Factor 2 in the normal season, and Factor 3 in the dry season, higher factor scores indicated poorer water quality. The cumulative variance contribution rates were 76.45%, 70.99%, and 76.93%, respectively. The results of the rotated factor loadings are shown in Table 5, and water quality monitoring indicators with factor loadings > 0.7 were selected for interpretation (with some selected > 0.6), which were bold number.

During the wet season, Factor 1 has a variance contribution rate of 39.73% and is positively correlated with conductivity, ammonia nitrogen, total nitrogen, and total phosphorus, which is inferred to be agricultural pollution. The watershed is a major agricultural production base for Guangzhou, with year-round suitable conditions for agricultural activities due to its low latitude. Rice production mainly involves double-cropping rice and rice–vegetable rotation. Nitrogen and phosphorus are the main contributors to agricultural water pollution, commonly found in farmland runoff and livestock waste. Agricultural non-point source pollution, including soil erosion, aquaculture, and livestock manure, carries a large amount of nutrients into the rivers. Factor 2 has a variance contribution rate of 36.72% and is positively correlated with temperature and pH, and it is negatively correlated with nitrate nitrogen (NO₃⁻-N), suggesting biochemical pollution of the water body. In the wet season, the high temperatures in South China limit the dissolved oxygen concentration in the water, reducing the water body’s self-purification capacity. The study area is located in the rural watershed of the PRD, characterized by a well-developed river network but relatively shallow river depths. On the one hand, the high-water period coincides with hot weather, and high temperatures reduce the dissolved oxygen saturation point in the water, so water temperature directly affects the dissolved oxygen saturation of the water. On the other hand, although increased water temperatures enhance denitrification efficiency, they also cause the phosphorus release flux from river bottom sediments to double, and the dominant species of algal communities shift from diatoms to green algae and even blue-green algae.

During the normal season, Factor 1 has a variance contribution rate of 40.37% and is positively correlated with conductivity, ammonia nitrogen, total nitrogen, and phosphate concentration, while negatively correlated with dissolved oxygen. This is inferred to be oxygen-consuming organic matter pollution. Factor 2 has a variance contribution rate of 30.62% and is positively correlated with temperature, pH, and DO, and negatively correlated with NO₃⁻-N, suggesting the water body’s self-purification capacity. In this period, the rivers in South China have higher runoff and faster flow rates, enhancing the water body’s self-purification capacity. This period differs from the wet season. On the one hand, faster flowing rivers can accelerate the mixing of pollutants with clean water through turbulence, thereby reducing local pollution concentrations through dilution. On the other hand, the increased surface area of fast-flowing rivers in contact with air enhances the rate of oxygen dissolution, raising the dissolved oxygen concentration, which can promote the decomposition of organic matter by aerobic microorganisms.

During the dry season, Factor 1 has a variance contribution rate of 35.57% and is positively correlated with ammonia nitrogen, total nitrogen, phosphate, and total phosphorus concentrations, which is inferred to be pollution from livestock and poultry farming and agriculture. Factor 2 has a variance contribution rate of 21.22% and is positively correlated with EC and negatively correlated with temperature, suggesting pollution may come from agricultural and domestic sources. Factor 3 has a variance contribution rate of 20.14% and is positively correlated with pH and DO and negatively correlated with NO₃⁻-N, which infers the water body’s self-purification capacity. During the dry season, the rivers have lower runoff and slower flow rates, resulting in a weaker self-purification capacity compared to the normal season.

3.2.3. Spatial Factor Analysis and Pollution Source Identification

Based on the principle of eigenvalue > 1, three factors were extracted for each of Groups G1, G2, G3, and G4, with cumulative variance contribution rates of 71.75%, 79.07%, 86.16%, and 86.40%, respectively. The rotated factor loadings are shown in

Table 6. Rotated factor loading matrix for Groups G1, G2, G3, and G4.

Water Indicator	G1			G2			G3			G4
	Factor 1	Factor 2	Factor 3	Factor 1	Factor 2	Factor 3	Factor 1	Factor 2	Factor 3	Factor 1	Factor 2	Factor 3
WT	0.81	−0.26	0.15	−0.05	−0.80	−0.18	−0.06	−0.96	−0.25	−0.03	−0.11	0.97
pH	−0.78	−0.34	−0.13	−0.09	0.92	0.06	−0.09	0.97	0.08	−0.40	0.78	0.25
EC	0.02	0.78	0.33	0.62	0.51	0.22	0.10	0.62	−0.71	0.91	−0.26	−0.10
DO	−0.83	−0.28	−0.15	−0.42	0.73	−0.16	−0.59	0.04	−0.03	−0.35	0.85	−0.23
NH3—N	0.27	0.29	0.75	0.91	−0.17	0.03	0.97	0.00	0.13	0.76	−0.04	−0.38
NO3−-N	0.37	0.78	−0.32	0.29	−0.17	−0.70	−0.82	−0.46	−0.16	0.21	0.85	−0.13
TN	0.52	0.63	0.31	0.95	−0.15	−0.11	0.87	−0.10	0.06	0.97	−0.14	0.05
PO43−-P	−0.06	0.68	0.49	0.01	0.07	0.91	0.24	0.36	0.88	0.93	−0.09	0.01
TP	0.13	0.03	0.64	0.39	−0.07	0.88	0.13	0.14	0.97	0.92	−0.04	0.05
Eigenvalue	3.87	1.52	1.07	2.87	2.56	1.69	3.54	2.31	1.90	4.75	1.93	1.09
Variance Contribution Rate/%	27.27	26.93	17.54	28.15	26.40	24.52	31.27	29.09	25.81	48.71	24.03	13.67
Cumulative Contribution Rate/%	27.27	54.21	71.75	28.15	54.55	79.07	31.27	60.35	86.16	48.71	72.73	86.40

Note: The bold number were indicators with factor loadings > 0.7 were selected for interpretation (with some selected > 0.6).

. From Table 6, it can be seen that there are three main factors influencing the water quality of Groups G1, G2, G3, and G4. Factor loadings > 0.75 indicate a strong correlation, 0.75–0.5 indicate a moderate correlation, and 0.5–0.3 indicate a weak correlation.

In the upstream G1, Factor 1 is strongly positively correlated with temperature and strongly negatively correlated with pH value and dissolved oxygen concentration, with a variance contribution rate of 27.27%. This is inferred to be oxygen-consuming organic matter pollution. Factor 2 is strongly positively correlated with electrical conductivity (EC) and nitrate nitrogen (NO₃⁻-N), moderately positively correlated with total nitrogen (TN) and phosphate (PO₄³⁻-P), with a variance contribution rate of 26.93%. This is inferred to be related to agricultural pollution. Factor 3 is moderately positively correlation with ammonia nitrogen (NH3-N) and total phosphorus (TP), with a variance contribution rate of 17.54%. This is inferred to be related to livestock farming pollution.

In the midstream G2, Factor 1 is strongly positively correlated with ammonia nitrogen and total nitrogen concentrations and moderately positively correlated with conductivity, with a variance contribution rate of 28.15%. This is inferred to be an indicator of agricultural pollution. Factor 2 is strongly positively correlation with pH, moderately positively correlation with DO, and strongly negatively correlated with temperature, with a variance contribution rate of 26.40%. This is inferred to be related to the water body’s self-purification capacity. Factor 3 is strongly positively correlated with PO₄³⁻-P and TP, strongly negatively correlated with NO₃⁻-N, with a variance contribution rate of 24.52%. This is inferred to be related to livestock farming pollution.

In the downstream G3, Factor 1 is strongly positively correlated with ammonia nitrogen and total nitrogen concentrations, strongly negatively correlated with nitrate nitrogen concentration, and moderately negatively correlated with dissolved oxygen concentration, with a variance contribution rate of 31.27%. This is inferred to be an indicator of oxygen-consuming organic matter pollution. Factor 2 is strongly positively correlated with pH, moderately positively correlated with NO₃⁻-N, and strongly negatively correlated with temperature, with variance contribution rate of 29.09%. This is inferred to be related to biochemical pollution in the water. Factor 3 is strongly positively correlated with PO₄³⁻-P and TP and moderately negatively correlated with EC, with a variance contribution rate of 24.52%. This is inferred to be related to eutrophication pollution.

In pond Group G4, Factor 1 is strongly positively correlated with EC, TN, PO₄³⁻-P, and TP, moderately positively correlated with NH3-N, and weakly negatively correlated with pH and DO, with a variance contribution rate of 48.71%. This is inferred to be related to oxygen-consuming organic pollution. Factor 2 is strongly positively correlated with pH, DO, and NO₃⁻-N, with a variance contribution rate of 24.03%. This is inferred to be related to the water body’s self-purification capacity. Factor 3 is strongly positively correlated with temperature and weakly negatively correlated with NH3-N, with a variance contribution rate of 13.67%. This is inferred to be related to biochemical pollution in the water.

3.2.4. The Contribution Analysis of Pollution Source

This study utilized the Absolute Principal Component Scores-Multiple Linear Regression (APCS-MLR) method to calculate the source contribution rates of various factors to water quality indicators during the wet, normal, and dry seasons. Figure 5 illustrates the source contribution rates for each water quality indicator. Based on the figure, it can be seen that during the wet season, water quality is mainly affected by agricultural pollution, followed by biochemical pollution of the water body. The contribution rates of agricultural pollution to the main impact indicators of conductivity, ammonia nitrogen, total nitrogen, and total phosphorus are 68.76%, 31.66%, 49.89%, and 46.84%, respectively. The contribution rates of biochemical pollution to the main impact indicators of water temperature, pH, DO, and nitrate are 23.45%, 44.70%, 82.06%, and 41.19%, respectively.

During the normal season, water quality in the basin is mainly affected by oxygen-consuming organic matter, followed by the self-purification capacity of the water body. The contribution rates of oxygen-consuming organic matter pollution to the main impact indicators of pH, ammonia nitrogen, total nitrogen, and phosphate are 70.44%, 55.12%, 48.84%, and 52.54%, respectively. The contribution rates of the water body’s self-purification capacity to the main impact indicators of water temperature, pH, DO, and nitrate are 20.97%, 40.41%, 69.96%, and 41.07%, respectively.

During the dry season, water quality in the basin is mainly affected by agricultural pollution, followed by rural domestic pollution and the self-purification capacity of the water body. The contribution rates of agricultural pollution to the main impact indicators of ammonia nitrogen, total nitrogen, phosphate, and total phosphorus are 8.80%, 1.16%, 3.73%, and 0.46%, respectively. The contribution rate of rural domestic pollution to the main impact indicator of conductivity is 46.32%, while the self-purification capacity of the water body contributes 21.63%, 27.41%, and 19.35% to the main impact indicators of pH, DO, and nitrate, respectively.

Similarly, the absolute principal component multiple linear regression method was used to calculate the source contribution rates of various factors within Groups G1, G2, G3, and G4 to water quality monitoring indicators. Figure 6 shows the source contribution rates of various water quality monitoring indicators. It can be seen from the figure that the upstream G1 is mainly affected by oxygen-consuming organic matter, followed by planting and livestock and poultry breeding pollution. The contribution rates of oxygen-consuming organic matter pollution to the main impact indicators of water temperature, pH, and DO are 11.69%, 7.66%, and 8.55%, respectively. The contribution rates of planting pollution to the main impact indicators of conductivity, nitrate nitrogen, total nitrogen, and phosphate are 5.23%, 9.61%, 17.58%, and 13.13%, respectively. The contribution rates of livestock and poultry breeding pollution to the main impact indicators of ammonia nitrogen and total phosphorus are 12.13% and 14.90%, respectively.

The midstream G2 water quality is mainly affected by agricultural pollution, followed by the self-purification capacity of the water body and livestock and poultry breeding pollution. The contribution rates of agricultural pollution to the main impact indicators of conductivity, ammonia nitrogen, and total nitrogen are 61.46%, 39.20%, and 2.63%, respectively. The contribution rates of the self-purification capacity of the water body to the main impact indicators of water temperature, pH, and DO are 3.28%, 3.68%, and 5.54%, respectively. The contribution rates of livestock and poultry breeding pollution to the main impact indicators of nitrate nitrogen, phosphate, and total phosphorus are 0.32%, 1.97%, and 61.65%, respectively.

The downstream G3 water quality is mainly affected by oxygen-consuming organic matter pollution, followed by biochemical pollution of the water body and eutrophication factors. The contribution rates of oxygen-consuming organic matter pollution to the main impact indicators of DO, ammonia nitrogen, nitrate nitrogen, and total nitrogen are 28.24%, 8.50%, 1.15%, and 2.21%, respectively. The contribution rates of biochemical pollution of the water body to the main impact indicators of water temperature, pH, and conductivity are 2.75%, 0.08%, and 10.47%, respectively. The contribution rates of eutrophication factors to the main impact indicators of conductivity, phosphate, and total phosphorus are 43.56%, 0.25%, and 22.35%, respectively.

The water quality of the pond water body in G4 is mainly affected by oxygen-consuming organic matter pollution, followed by the self-purification capacity of the water body and biochemical pollution of the water body. The contribution rates of oxygen-consuming organic matter pollution to the main impact indicators of conductivity, ammonia nitrogen, total nitrogen, phosphate, and total phosphorus are 60.13%, 21.50%, 0.49%, 5.95%, and 57.23%, respectively. The contribution rates of the self-purification capacity of the water body to the main impact indicators of pH, DO, and nitrate nitrogen are 8.46%, 55.41%, and 25.16%, respectively. The contribution rates of biochemical pollution of the water body to the main impact indicators of water temperature and ammonia nitrogen are 3.73% and 3.62%, respectively.

From the pollution source contribution analysis (Section 3.2), it is evident that the main sources of pollution in the watershed include agricultural pollution, oxygen-consuming organic pollution, and biochemical pollution. These factors primarily affect indicators such as ammonia nitrogen (NH3-N), total nitrogen (TN), total phosphorus (TP), and dissolved oxygen (DO). Furthermore, as observed in the analysis of temporal and spatial variations in water quality (Section 3.1), the midstream and downstream areas of the watershed are significantly impacted by nitrogen and phosphorus nutrient pollution, which could potentially lead to eutrophication risks. Therefore, it is necessary to further introduce the comprehensive Eutrophication Index (EI) to evaluate the specific eutrophication status and its temporal and spatial distribution characteristics within the watershed.

3.3. Evaluation of Eutrophication Pollution of the Watershed

3.3.1. Temporal and Spatial Distribution Characteristics of Eutrophication

In this study, the concentrations of TN, TP, orthophosphate, ammonia nitrogen, and DO at all sections of the basin were selected as evaluation indicators for eutrophication pollution in the basin. The potential eutrophication level of the water body was assessed using a logarithmic universal index (EI), and further statistical analysis was conducted according to the three periods (wet, normal, and dry seasons) and spatial clustering results. The results are shown in

Table 7. Level of eutrophication index during all periods and in G1–G4 groups.

Periods/Groups	Sections/Period	The Proportion of Various Water Quality Eutrophication Grades to the Whole (%)
		Oligotrophic	Mesotrophic	Eutrophic	Hypertrophic	Extremely Hypertrophic
Wet season	All sections	0	34.3	65.7	0	0
Normal season	All sections	0	48.6	37.1	14.3	0
Dry season	All sections	5.7	45.7	45.7	2.9	0
G1	Wet season	0	41.2	58.8	0	0
	Normal season	0	64.7	35.3	0	0
	Dry season	0	64.7	35.5	0	0
G2	Wet season	0	0	100	0	0
	Normal season	0	0	50	50	0
	Dry season	0	12.5	87.5	0	0
G3	Wet season	0	0	100	0	0
	Normal season	0	0	66.7	33.3	0
	Dry season	0	0	66.7	33.3	0
G4	Wet season	0	71.4	28.6	0	0
	Normal season	0	85.7	14.3	0	0
	Dry season	28.6	57.1	14.3	0	0

. Overall, the proportions of eutrophic and hypereutrophic during the wet, normal, and dry seasons were 65.7%, 51.4%, and 48.6%, respectively, with hypereutrophic levels appearing during the normal and dry seasons.

For the G1, G2, G3, and G4 groups, the proportions of eutrophic and hypereutrophic levels were 43.1%, 79.2%, 100%, and 19.0%, respectively. Specifically, no hypereutrophic appeared in G1 sections, while hypereutrophic appeared in the G2 sections during the normal season. The G3 mainly exhibited eutrophic condition, with hypereutrophic appearing during the normal and dry seasons. The G4 pond group mainly exhibited mesotrophic, and attention should be paid to the risk of pond eutrophication during the whole year.

Figure 7 illustrates the temporal and spatial distribution characteristics of the eutrophication assessment results for pond water bodies. The figure shows that the eutrophication levels of pond water bodies vary across different periods, with the greatest variability occurring during the dry season. P1 maintains an eutrophic level across all periods. P5 is consistently at a mesotrophic level throughout all periods. P6 exhibits oligotrophic levels during the dry season, mesotrophic levels during the normal season, and eutrophic levels during the wet season. Other ponds were generally at mesotrophic levels during the wet and normal seasons and oligotrophic levels during the dry season. These observations indicate significant seasonal variations in eutrophication status within the ponds, highlighting the need for targeted management strategies to address potential eutrophication risks, especially during the wet season.

3.3.2. Spatial and Longitudinal Variation Characteristics of Watershed Eutrophication

To more comprehensively analyze the overall eutrophication situation of the basin’s water bodies, this study set up water quality monitoring points to collect and analyze water quality conditions at various locations along the main and tributary streams of the basin. Figure 8 shows the spatiotemporal variation characteristics of eutrophication from upstream to downstream sections of the tributary Diaozhong Discharge Channel, the main stream Zhucun Canal, and the Nangang River.

The eutrophication changes in the sections of the Diaozhong Discharge Channel from upstream to downstream (Z09 to Z05) are shown in Figure 8a. As seen from the figure, the eutrophication status of the Diaozhong Discharge Channel changes from mesotrophic to eutrophic from upstream to downstream.

The eutrophication changes in the sections of the Zhucun Canal from upstream to downstream (Z26, Z25, Z23, Z22, Z18 to Z15) are shown in Figure 8b. From the upper to the midstream, the canal is generally at a mesotrophic level. At the midstream residential section Z12, the eutrophication level rises to eutrophic, then returns to mesotrophic, and finally becomes eutrophic at all downstream sections (Z15).

The eutrophication changes in the sections of the Nangang River from upstream to downstream (Z18, Z19, Z14, Z17, Z16, Z15 to Z02) are shown in Figure 8c. The eutrophication level changes from mesotrophic to hypereutrophic from upstream to downstream, primarily passing through densely populated town residential areas and areas with concentrated factories such as paper mills and hardware processing plants.

4. Discussion

Based on the spatiotemporal characteristics of pond water quality and the eutrophication evaluation results, it was found that ponds in the basin can be mainly divided into two types. One type is aquaculture ponds managed by villagers, primarily used for fish farming. It was found that these ponds undergo water extraction and replacement during the dry season. This is because the managers add a large amount of feed into the water for fish farming throughout the year, which easily leads to nutrient overload and subsequent water hyacinth overgrowth. The other type is landscape ponds, which are typically closed ponds with only artificial water extraction and rainwater along with the resulting surface runoff entering them. This leads to the accumulation of nutrients brought by rainwater runoff in the pond. According to Hou et al. [26], oxygen-consuming organic pollutants are the primary pollutants in aquaculture, and ponds are among the top in terms of the number of sudden pollution incidents. This aligns with the findings of this study that the main pollutants in the ponds of the Zhucun basin are oxygen-consuming organic pollutants. Ponds and reservoirs are not only important nodes for intercepting and absorbing non-point source pollution in the basin, but they also play a crucial role in water storage and flood control during extreme precipitation events caused by climate change. Therefore, it is recommended to implement standardized classification management for rural basin ponds. For aquaculture ponds, centralized treatment of fish pond drainage is necessary [8]. After collection, the water should be aerated and retained in a collection pond before being periodically discharged to reduce the impact of eutrophication on rivers. For landscape ponds, they should be included in non-point source pollution prevention and control measures, with strengthened supervision to reduce the input of nitrogen and phosphorus nutrients into the ponds. Additionally, planting aquatic plants that effectively absorb nitrogen and phosphorus in the ponds can help achieve water storage and ensure ecological functions.

The spatiotemporal variation characteristics of water quality in the rural small basin of the study area show that ammonia nitrogen generally exhibits higher concentrations during the dry season and lower concentrations during the wet season. The total nitrogen content is relatively low during the dry season, with the lowest ammonia nitrogen and higher nitrate nitrogen, indicating that the water body is mainly in an oxidized state during the dry season with strong self-purification capacity, which is the opposite during the normal season. Phosphate and total phosphorus (TP) concentrations are higher during the dry season, likely due to the large number of algae in the summer and autumn seasons. After the temperature drops in winter, some algae die and decompose, releasing phosphorus, causing the accumulation of phosphorus in the water body. When the temperature rises, it can easily lead to algal blooms [27]. Additionally, the higher concentrations of nitrogen and phosphorus nutrients during the dry season and lower concentrations during the wet season may also be due to the presence of point source pollution in the basin. Furthermore, according to the spatiotemporal differences in river sections and eutrophication evaluation results, the eutrophication status of river sections during the normal and dry seasons is more severe than during the wet season. The possible reason is that the wet season is the flood season, with large runoff and fast flow rates, which dilute pollutants and improve river water quality. Therefore, the conductivity, ammonia nitrogen, total nitrogen, and total phosphorus concentrations during the wet season are lower than during the normal season. The Zhucun Canal and Nangang River are the two main rivers in the basin, both transitioning from a mesotrophic state in the upper reaches to a eutrophic state at the downstream outlet without purification and dilution. Meanwhile, the impact of pollutant discharge caused by rapid urbanization on the water quality of rural small basins should be taken seriously. In the management of eutrophication pollution in rural small basin rivers, it is recommended to implement segmented management. For areas concentrated with farmland and residential life, sewage treatment stations should be constructed to treat the sewage from both sources during the normal and dry seasons. Additionally, strict measures should be taken to prevent rural domestic waste and agricultural waste from entering the channels and rivers, ensuring the environmental safety of the river. For industrial areas with relatively concentrated distribution in the study basin, it is recommended to strengthen the construction of sewage collection networks and enhance the supervision and control of sewage discharge from factories affecting the rivers.

5. Conclusions

Based on the eutrophication assessment and APCS-MLR analysis, it is evident that the overall watershed in the study area is at a high level of eutrophication, with the most severe conditions observed downstream. The analysis identified the main sources of water pollution as agricultural pollution, oxygen-consuming organic pollution, and biochemical pollution, with agricultural pollution and oxygen-consuming organic pollution being the most significant contributors. The specific conclusions are summarized as follows:

(1)

Seasonal Water Quality Variations: In the Zhucun watershed (South China), water quality is best during the wet season (April to September) due to dilution from high runoff, with ammonia nitrogen as the main pollutant. The worst water quality occurs in the normal season (March, October–November), dominated by ammonia nitrogen and phosphate from organic pollution. The dry season (December to February) shows poor quality due to phosphate and total phosphorus, linked to eutrophication.

(2)

Spatial Water Quality Differences: Upstream areas have the best water quality, affected mainly by farming and livestock pollution. Midstream regions face agricultural and livestock pollution, while downstream areas suffer the worst quality due to organic pollution, eutrophication, and biochemical contamination. Ponds are primarily impacted by organic pollution and human activity.

(3)

Pollution Sources and Management Needs: Pollution in the Zhucun watershed is concentrated in densely populated midstream and downstream zones, highlighting sewage discharge issues. Given rapid rural urbanization, targeted pollution control measures are needed.