Source Identification and Control of Eutrophication in Large Shallow Freshwater Lakes: A Case Study of Lake Taihu

Abstract

Lake Taihu, a large, shallow freshwater lake in China, has experienced severe eutrophication for decades under intense human activities occurring around cities. Through long-term water quality management since 1995, the eutrophication of Lake Taihu has been controlled. This review examines the eutrophication characteristics, source identification methods, and control measures in Lake Taihu. Phosphorus is a primary driver of eutrophication, correlating strongly with chlorophyll a. The lake exhibits significant temporal and spatial variability in nutrient dynamics, influenced by human activities and the climate. Historical data show fluctuating nutrient levels and persistent algal blooms despite government efforts. A critical assessment of various source apportionment methods, including statistical analysis, physical modeling, and empirical models, is presented to elucidate the relative contributions of different nutrient sources. These methods identify agricultural non-point and urban point sources as major external contributors, with sediment nutrient release as a significant internal source. Implemented controls, including wastewater treatment plants and non-point-source management, have had limited success. Increased sewage and sediment nutrients necessitate integrated watershed management. Future research should prioritize advanced source tracking, sediment dynamics, climate impacts, and integrated ecological models. Sustainable eutrophication management in Lake Taihu requires integrated science, policy, and public engagement to ensure ecosystem health.

1. Introduction

Nutrient enrichment in water bodies is a natural phenomenon that can occur in any aquatic system. However, this process is now being exacerbated on a global scale due to the escalating influx of human-derived nutrients in recent decades [1,2]. Lakes play an important role in maintaining a stable freshwater supply. The expansion of agriculture, industry, and urbanization has contributed to the significant discharge of nutrients, particularly nitrogen and phosphorus, into water bodies, resulting in detrimental impacts on the ecological structures, functions, and esthetic value of lakes [3]. This alarming trend poses a serious threat to ecosystem health and the sustainability of human society [4,5], demanding immediate attention and intervention.

Under eutrophication, phytoplankton biomass exhibits pronounced growth, with cyanobacteria often dominating [6,7]. This leads to a significant increase in the concentration of chlorophyll a (Chl a) on the water surface, ultimately resulting in reduced light penetration [8]. Consequently, the growth and community structures of aquatic plants are adversely affected [9]. The excessive proliferation of cyanobacteria can deplete dissolved oxygen levels during the night, leading to a decline in fish populations [10]. Additionally, the reproduction of cyanobacteria is associated with the production of algae toxins that pose various risks to human health, ranging from mild skin irritation to gastrointestinal distress, liver ailments, neurological impairments, and even fatalities [11,12]. However, many lakes around the world exhibit significant eutrophication issues, such as Erie (America), Winnipeg (Canada), Peipsi (Estonia), Balaton (Hungary), Taihu (China), Kasumigaura (Japan), and Tana (Ethiopia) [1,13,14,15]. Globally, eutrophication has affected over 40% of lakes in recent years, posing a huge threat to aquatic systems [7]. In particular, the occurrence of cyanobacterial blooms has garnered significant worldwide attention. In North America and Europe, the prevalence of cyanobacteria in nearly 60% of lakes has exhibited a substantial increase since the onset of the Industrial Revolution [16].

In China, there are 2693 natural lakes, with areas of over 1.0 km² [17]. However, environmental challenges caused by eutrophication have become a substantial hindrance to China’s sustainable development [18]. Lake Taihu, as the third-largest freshwater lake in China, performs an array of pivotal functions, e.g., fishery support, transportation facilitation, tourism attraction, and flood control measures, while serving as a crucial drinking water source for the neighboring communities [19]. However, cyanobacterial blooms in Lake Taihu have been observed with an increasing frequency since the late 1980s [20,21]. The deteriorating water quality, exacerbated by the surge in village enterprises within the Taihu Basin during the early 1990s, has elicited increased concern regarding eutrophication control. In 2007, a severe outbreak of cyanobacteria in Lake Taihu rendered approximately two million inhabitants incapable of accessing clean water. This spurred public awareness and scrutiny regarding the imperative for comprehensive eutrophication management strategies pertinent to Lake Taihu [22].

Since the occurrence of cyanobacterial blooms in 1990, the Chinese government has prioritized the management of Lake Taihu, and stringent discharge limits for wastewater in the Taihu Basin have been in effect since 1999 [22]. After the outbreak of cyanobacterial blooms in 2007, the government launched a series of comprehensive water treatment plans, resulting in an unprecedented, intensive, and large-scale effort to control the eutrophication and cyanobacterial blooms in Lake Taihu. The concentrations of nutrients in Lake Taihu exhibited a downward trajectory, although recent years have witnessed the resurgence of cyanobacterial blooms [23]. In 2017, the area affected by cyanobacterial blooms reached its maximum [24]. A number of strategies have been employed to mitigate the eutrophication issue in Lake Taihu, including wetland restoration [25], mechanical algae removal [26], and freshwater infusion from the Yangtze River [23]. However, the effectiveness of these strategies has been limited by a lack of understanding of the sources regarding lake eutrophication [5]. Ascertaining the sources of these nutrients should be regarded as the first step in mitigating eutrophication within the lake’s ecosystem [27]. The possibility of future large-scale blooming phenomena in Lake Taihu remains a concern [28].

As is the case for many large lakes [29], the issue of eutrophication in Lake Taihu has attracted the attention of researchers and become an interdisciplinary and comprehensive challenge [17]. This review summarizes the distinct characteristics and key issues associated with eutrophication in Lake Taihu, evaluates the methodology for the identification of nutrient pollution sources in the Taihu Basin, and assesses the efficacy of the source control and interception measures implemented over the years. Based on this, future research topics and remediation measures targeting eutrophication in Lake Taihu and other lakes around the world are discussed.

2. Characteristics of Eutrophication in Lake Taihu

Over the past few years, a multitude of studies have revealed that the occurrence, strength, and duration of cyanobacterial blooms in various aquatic environments worldwide are expected to rise due to increasing eutrophication and climate change [30]. Lake Taihu, as a typical large shallow lake, exhibits severe eutrophication. The dynamics of the nutrient concentrations and cyanobacterial blooms in the lake are regulated by complex interactions between several factors, including the economic development of the basin, pollution control measures, climate change, and hydrological conditions [3,31]. Hence, identifying the key pollutants that are related to cyanobacterial blooms, and analyzing the temporal and spatial characteristics of the eutrophication in Lake Taihu, are necessary to prevent the outbreak of cyanobacterial blooms and other potential hazards.

2.1. Characteristics of Nutrients

The underlying cause of eutrophication lies in the accumulation of nutrients. It is widely accepted among researchers that a nutrient concentration in water bodies that surpasses a specified threshold is necessary to trigger cyanobacterial blooms [15,32,33]. The “Guidelines for Lakes and Reservoirs Technical Guidelines—Nutrient Standards”, issued by the United States Environmental Protection Agency, stipulate that, when the concentrations of total nitrogen (TN) and total phosphorus (TP) exceed 150 µg/L and 1 µg/L, respectively, the occurrence of cyanobacterial blooms becomes likely [34]. It is now widely recognized that a TN level of >0.2 mg/L and a TP level of >0.02 mg/L indicate eutrophication in water bodies [35]. Compared to single pollution with nitrogen or phosphorus, the combined enrichment of nitrogen and phosphorus exhibits a more pronounced influence on the occurrence of cyanobacterial blooms [36].

Although both nitrogen and phosphorus play crucial roles in the occurrence of cyanobacterial blooms in lakes [37], it has been reported that the concentrations of TP in the water columns of eutrophic lakes often correspond to the levels of cyanobacterial blooms [38]. The significant correlation between the eutrophication status and TP concentration has been observed in many lakes worldwide [39,40,41,42,43,44]. A study showed that 90% of global lakes exhibit phosphorus retention [45]. In the case of Lake Taihu, TP has been identified as the primary factor influencing the spatial distribution of cyanobacterial blooms and thus requires prioritized attention in pollution control efforts [46]. As such, phosphorus has emerged as a paramount element in the preservation and ecological management of lakes [47].

Chl a serves as a key indicator in assessing water eutrophication and provides valuable insights for the identification of algal species and their abundance [48]. A strong positive correlation has been observed between the Chl a and TP concentrations in Lake Taihu, yielding a correlation coefficient (r) of 0.76 (p < 0.05). In contrast, some studies have found no evident relationship between TN and Chl a in many lakes, including Lake Taihu [3,15,49]. A possible explanation is that nitrogen is often not a primary limiting factor for cyanobacteria growth in some regions [46].

2.2. Temporal Heterogeneity

2.2.1. Historical Variation

Prior to the 1980s, the concentration of TN in Lake Taihu had reached approximately 60 μg/L [50]. In the late 1970s and early 1980s, the rapid development of township enterprises around Lake Taihu led to the production of a large amount of industrial wastewater. Because of the lack of centralized management in this sector, the generated wastewater was discharged into the water body without proper treatment. By 1981, 16.9% of Lake Taihu’s area was classified as eutrophic [51], triggering the occurrence of frequent cyanobacterial blooms in Meiliang Bay in the northern region of the lake and subsequently expanding northwestward [52]. The eutrophication of Lake Taihu progressed at an alarming rate after 1990 [50]. From 1991 to 1996, the annual average concentrations of TN, TP, and Chl a in Lake Taihu showed an initial upward trend. In 1995, the local government imposed mandatory control on wastewater discharge within the catchment area, and the eutrophication issue of Lake Taihu appeared to exhibit signs of recovery from 1996 onwards [53].

Unfortunately, the TN and TP concentrations in Lake Taihu have rebounded since 2001 [51]. In 2007, the annual average concentration of TN in the western waters reached its highest value, coinciding with the water supply crisis at Gong Bay [54]. Subsequently, the government intensified its pollution control efforts, leading to a decline in the TP concentrations in the rivers entering Lake Taihu [55]. From 2009 to 2017, the concentration of Chl a in Lake Taihu exhibited an upward trend. Assessment through satellite images indicated that there was no decreasing trend in the average level and maximum area of cyanobacterial blooms between 2007 and 2017. In fact, the largest area of bloom coverage was recorded in the lake in 2017, which was due to the synergistic effect of climate warming and nutrient enrichment [15]. Black patch events are unique ecological disasters caused by eutrophication in Lake Taihu. An analysis using field data from 2009 to 2017 revealed that, in 2017, the highest number of black patch events occurred (17), as well as the longest total occurrence time (47 d) and single duration (3.76 d). Correspondingly, the TP and TN loads in Lake Taihu reached their peaks during this year [56].

2.2.2. Seasonal Variation

The seasonal variations in the nutrient concentrations and cyanobacterial blooms in Lake Taihu are strongly influenced by both the meteorological conditions and anthropogenic activities.

Drawing upon an analysis of monthly water quality data spanning 1985 to 2015, it was observed that the concentration of TN exhibited distinct seasonal patterns, with the highest levels occurring in March and April and the lowest in August and November [51]. The substantial reduction in nitrogen during the summer can be attributed to its absorption and utilization by cyanobacteria and a marked increase in denitrification capacity during the period of bloom [26,57]. In the northern region of Lake Taihu, serving as a primary inflow area, the nitrogen pollution pattern was found to be significantly influenced by the discharge of urban wastewater, spring fertilization practices, and land disturbance [58].

Researchers hold different opinions on the seasonal variations in the TP concentration. The study conducted by Dai et al. indicates that there was no significant seasonal variation in the TP concentration from 1985 to 2015 [51]. Zhang et al. analyzed the seasonal changes in phosphorus concentrations in the whole lake from 2005 to 2018 [59], and the result showed that TP reached its peak in summer, while the average values of TP in other seasons did not differ significantly. This does not mean that these research conclusions are contradictory, as the monitored TP includes phosphorus in organisms such as algae, and the algal reproduction status of water bodies in different years is greatly influenced by the water regime and meteorological conditions [60].

Regarding the nutrient composition in algae- and macrophyte-dominated regions of Lake Taihu, significant variations were observed among the different seasons. Lv et al. reported that nitrate nitrogen and organic phosphorus in algal-dominated regions were the predominant nutrients in winter, while, for the other seasons, particulate nitrogen and phosphorus dominated [32]. On the other hand, in macrophyte-dominated regions, particulate nitrogen and phosphorus prevailed during winter, whereas ammonia nitrogen and organic phosphorus were prominent during other seasons [32].

2.3. Spatial Heterogeneity

The pattern of Chl a distribution in Lake Taihu is intricately linked to the influx of river water and the presence of surrounding urban and industrial areas. The area division of Lake Taihu is shown in Figure 1. Most river inflows into Lake Taihu are from the western and northern sectors [49]. The water quality in these areas is largely impacted by industrial activities [61], resulting in a higher level of eutrophication compared to other regions [62]. The bay area, characterized by high nutrient loads and low water flow, becomes highly susceptible to cyanobacterial blooms. When examining different zones of Lake Taihu, it is evident that the highest frequency of cyanobacterial blooms occurs in the northwestern lake area and Meiliang Bay, followed by Zhushan Bay, Gong Bay, the central lake area, and the southwestern lake area, with the lowest frequency of cyanobacterial bloom occurrences in the eastern lake. Notably, from 2013 to 2017, there was a trend of an increase in the cyanobacterial bloom frequency spreading toward the central lake area [34].

The spatial distribution of Chl a is also related to the lake flow dynamics caused by the wind. The prevailing southeasterly wind above Lake Taihu in summer and the resulting wind-driven current cause the surface layer of the lake to flow from southeast to northwest. Accordingly, Chl a and cyanobacteria accumulate in the northwest of Lake Taihu, and the spatial difference across the entire lake increase [63,64]. It has been observed that black patch events caused by cyanobacterial blooms mainly occur in the western coastal zone and northern part of Lake Taihu, such as Meiliang Bay and Gong Bay, rather than in the eastern lake [19,56].

3. Source Identification of Eutrophication in Lake Taihu

The sources and transport of nutrients in Lake Taihu can be summarized as in Figure 2. Non-point-source nutrients such as rural sewage, livestock manure, chemical fertilizers, etc., are transported to surface water via septic tanks and farmland. Other point-source nutrients (such as industrial wastewater and sanitary sewage) are partly transported to rivers after wastewater treatment. Then, the nutrients are introduced into Lake Taihu via inflowing rivers [65]. Moreover, the atmosphere and fisheries can also lead to nutrient input in Lake Taihu.

Due to the long-term eutrophication of Lake Taihu, the sediments within the lake mainly act as recipients of nutrients derived from external sources (e.g., sewage, livestock manure, fertilizers, and the atmosphere), as well as local sources (e.g., plankton and benthic plants). Over time, these sediments undergo a transition from nutrient “sinks” to significant nutrient “sources”, releasing substantial amounts of N and P under both undisturbed and disturbed conditions, and these releases bring new pollution challenges. As illustrated in Figure 2, the sources are categorized into two groups: external sources and endogenous sources.

The proliferation of cyanobacteria in Lake Taihu is influenced by a multitude of factors, with the excessive input of nutrients being a root cause. To effectively curb the growth of cyanobacteria in Lake Taihu, it is imperative to reduce the external load of nutrients [65,66,67]. Many methods have been developed to identify the nutrient sources and driving factors of eutrophication in Lake Taihu. The most noteworthy research studies conducted in recent years are provided in

Table 1. Studies on source identification of eutrophication in Lake Taihu.

Pollutants	Analytical Methods	Demand Data	Source Types or Driving Factors	Reference
TN, TP	Machine learning	Water quality, main rivers entering the lake, sewage treatment plants, industrial output, meteorological conditions, etc.	Temperature, SS and TN in sewage, fertilizer yield, etc.	[68]
TN, TP	Spatial relationship model	Sewage treatment plant emissions, rural sewage, agricultural activities, livestock and poultry breeding activities, land use, etc.	Farmland, rural sewage, livestock, etc.	[69]
TN, TP	Direct calculation	River flux, pollutant concentration, etc.	Surrounding rivers, atmosphere, sediment, etc.	[70]
TP	PSCI model	Meteorological, hydrological, and water quality	Sediment, inflow river	[71]
TP	SPARROW model	River phosphorus flux, land cover, etc.	Land cover, river length, runoff depth, etc.	[72]
Sediment Ns	Monte Carlo simulations	TOC content and isotopic ratios (δ13C and δ15N) of potential source samples	Aquatic macrophytes, algae bloom, inflow/outflow rivers, hydrodynamic changes	[73]
TDN, TDP	MARINA-Lake model	Activity level data within the basin	Synthetic fertilizers, livestock manure, human waste, etc.	[5]
TN, TP, NO3–N, NO2–N, PO43−, etc.	Multivariate statistical analysis	CODMn, TN, TP, Cl, and DO	Agriculture, domestic sewage, industry, etc.	[62]
SOM	Fatty acid composition analysis	Composition of fatty acid biomarkers	Land plants, aquatic plants, algae, etc.	[74,75]

.

Researchers have employed diverse models to identify the causes of lake eutrophication. The models can be categorized into several distinct modes of analysis, including multivariate statistical analysis, physical model analysis, empirical model analysis, and multidisciplinary comprehensive analysis.

3.1. Identification of External Pollution Using Multiple Analyses

3.1.1. Multivariate Statistical Analysis

With the water quality observation data, the results of multivariate statistical analysis have shown considerable potential in understanding the spatiotemporal patterns and source distribution of water eutrophication [76,77,78]. Principal component analysis, cluster analysis, and discriminant analysis have been extensively employed to evaluate the temporal and spatial variations in surface water, thereby enabling the inference of pollution sources of lakes [79,80,81].

Based on the water quality data of the outlets of rivers from 2006 to 2010, Chen et al. analyzed the potential pollution sources of two regions surrounding Lake Taihu through principal component analysis [82]. The results indicated that, in highly developed areas, most industrial and domestic wastewater was directly discharged into rivers. Therefore, the area was mainly affected by runoff related to industrial and domestic pollution, while the Tiaoxi River Basin, originating from mountainous and moderately developed rural areas, was affected by both point sources and non-point agricultural sources. In the northwestern area of Lake Taihu, where the eutrophication was the most severe, absolute principal component score–multiple linear regression and positive matrix factorization models were used to calculate the contribution rate of each pollution source. The results showed that agricultural non-point source pollution (26.6%) was the major contributor, followed by domestic sewage discharge (23.5%) [62].

Multivariate statistical analysis employs receptor models and dimensionality reduction techniques to identify the primary sources of pollution based on extensive water quality data. Various parameters, such as nutrients, organic matter, heavy metals, and other pollutants, are analyzed by this method, providing a comprehensive understanding of the pollution sources. However, the results obtained are limited to determining the contribution of overall pollution and are insufficient in precisely identifying the cause of water eutrophication.

3.1.2. Physical Model Analysis

Non-point-source pollutants, characterized by uncertain emissions from multiple outlets, play an important role in the eutrophication of water bodies. Due to the spatial and temporal variability in nutrient loads, non-point-source pollution has arbitrary and irregular processes and complex mechanisms [83]. Model simulation is a common approach to monitoring and evaluating non-point-source pollution. Physical and empirical models are two typical types of measures for such simulation. Physical models have been extensively employed to simplify the intricate natural processes involved in the generation and transformation of non-point-source pollutants [84,85,86]. The estimation of the loads of non-point-source pollutants is achieved through watershed models with hydraulic response simulations and receiving water models that simulate water quality and hydrodynamic transmissions [87].

In a study conducted in the southern part of the Taihu Basin, the Soil and Water Assessment Tool (SWAT) model was employed to analyze hydrological and water quality parameters. Agricultural fertilizers—specifically nitrogen and phosphorus carried by runoff—were identified as the primary non-point-source pollutants within the lake area. Thus, it is suggested to address the excessive application of nitrogen and phosphorus fertilizers and their synergistic impacts with manure, as these measures can significantly control nutrient pollution within the lake ecosystem [72]. In a study concerning Tianmu Lake (situated in the upper reaches of the Taihu Basin), using the Spatially Referenced Regressions On Watershed Attributes (SPARROW) model, variables such as the land cover, river length, runoff depth, and pond density were found to effectively explain 94% of the temporal and spatial variations in the TP load [88]. On average, ponds intercepted 24% of the phosphorus output from the watershed landscape. These findings underscore the profound importance of landscape characteristics in mitigating TP losses in environmentally sensitive hilly watersheds [88].

3.1.3. Empirical Model Analysis

Export coefficient models (ECMs) have been widely employed in studies of non-point-source pollution within large basins over the past two decades [89,90,91,92,93]. The initial ECM estimates the overall loss from diverse sources, such as land use, livestock, and rural activities, based on the nutrient load discharged from the watershed [94]. In recent years, significant advancements have been made in ECMs regarding parameterization and application scope [74]. Notably, to incorporate the influence of the distance between the source and sink on nutrient discharge (including hydrological, transport, and terrain impacts), the transmission [95], rainfall impact, and watershed loss coefficient have been introduced to the ECM [29].

Previous research on estimating ECM coefficients has relied on field experiments or investigations to determine nutrient load allocation across various pollution sources. While these studies increased the accuracy of the results, their limitations included small-scale watersheds and a reduced temporal scope. A study conducted in Xueyan Town, a first-grade protection zone of the Taihu Basin, during the rice-growing season of 2000 revealed that nitrogen emissions from farmland, village residents, town residents, and livestock contributed 72.7%, 18.9%, 7.2%, and 1.2%, respectively. Phosphorus emissions from these sources accounted for 56.2%, 18.9%, 22.2%, and 2.7%, respectively [96,97,98]. In 2004, it was observed that nitrate nitrogen and particulate nitrogen were the primary forms of nitrogen loss from dry land to water bodies in the Taihu Basin. Therefore, controlling these forms of nitrogen should be a priority. In this investigation, particulate phosphorus emerged as the major form of phosphorus loss, representing 76% of the total phosphorus [99]. Additionally, the land use composition significantly impacts nutrient outputs, as demonstrated in the Xitiaoxi watershed of the Taihu Basin, where an increase in the ratio of farmland to forest land was correlated with intensified nutrient outputs [100].

When the export coefficients obtained from field investigations exceed a specific threshold, they can be applied to large-scale calculations, enabling expanded research coverage, enhanced pollution source identification, and accelerated data analysis [101]. Via an ECM, the total nutrient loads of TN and TP originating from industry, livestock farming, agriculture, household activities, and atmospheric deposition in the Taihu Basin in 2008 were estimated to be 33,043.2 t/a and 5254.4 t/a, respectively. Notably, household consumption exhibited the highest impact, accounting for 46% of the TN load and 47% of the TP load. Atmospheric deposition and agriculture contributed 18% and 15% of the TN load, respectively, while livestock farming was the second-largest contributor to the TP load, accounting for 32% [102]. Leveraging an ECM, a study estimated pollutant emissions from non-point sources in the rural areas of the south lake region and utilized the optimized maximum quantity of a pollutant program to determine the environmental capacity of water bodies in the small rural watershed. The findings supported the formulation of related measures for pollutant reduction [103].

3.1.4. Multidisciplinary Analysis

The limitations of simple ECMs in terms of accuracy have prompted researchers to incorporate geographic information systems (GISs) into pollution source accounting models to increase the precision in land use classification. The Agricultural Non-Point-Source Pollution Potential Index (APPI) system was developed to address this need. By considering factors such as the sediment production index, runoff index, human and animal load index, and chemical use index, the APPI system effectively quantifies non-point-source pollution. Research focused on Xueyan Town within the Taihu Basin revealed that rural residents contributed 33% of TP and 40% of TN, while residents in the town center contributed 25% of TP and 10% of TN [97]. The integration of a GIS with an ECM, along with the consideration of watershed-specific rainfall patterns, resulted in an improved model with excellent spatiotemporal suitability and generalization [104]. Based on GIS technology, a study integrated hydrological models into the ECM by using meteorological data, the watershed topography, land use, and river network data as inputs, thereby constructing the Monthly Export Coefficient Model (MECM). One significant achievement of this research is its monthly time scale, surpassing the prevalent annual time scale used in most ECM studies [105]. Zhou et al. used GIS spatial analysis capabilities to construct a semi-distributed export coefficient model, determining that the TN load from non-point sources in the entire Taihu Basin in 2011 was 398,100 t/year, with the corresponding TP load reaching 55,900 t/year [69]. Additionally, a spatial relationship model based on a GIS was developed in 2014 to describe the relationship among point sources, river segments, and catchments in the Taige Canal watershed of the Taihu Basin [106]. This model enables in-time source tracking triggered by predefined water quality thresholds, offering a fast response.

The MARINA-Lake model was recently developed to accurately assess nutrient inputs from river systems into the marine environment and to quantify nitrogen and phosphorus outputs from sub-basins to lakes [71]. Using this model, the Taihu Basin was divided into five sub-basins, and the pollution loads in each sub-basin were calculated by an ECM. The nutrient load discharged from each sub-basin into the lake was determined using the MARINA-Lake model, with the consideration of non-point-source losses, providing estimates that aligned closely with real values [5]. The Phosphorus Source Contribution Index (PSCI) model characterizes phosphorus sources, sinks, and transport in both horizontal and vertical directions within the lake, derived from a comprehensive three-dimensional hydrodynamic and water quality model. This model has been used to track the phosphorus sources of two drinking water intakes in Lake Taihu, and the results indicate that the internal phosphorus load in sediments is an important phosphorus source, with contribution rates of 47.1% and 30.4%, respectively [68]. Furthermore, artificial intelligence (AI) technologies have great potential in effectively identifying sources of eutrophication and pollution within Lake Taihu. Using 13 years of data, Hu et al. constructed six integrated machine learning models, and the results indicated that the TN concentration was primarily influenced by the endogenous load and incoming water quality [107], while TP was mainly affected by the endogenous load. The model shows potential in tracking and predicting pollution sources, aiding in the early warning and effective control of lake eutrophication [107].

3.2. Identification of Endogenous Pollution

It has been reported in many cases that, despite the extensive efforts being made to reduce external nutrient inputs, the decrease in the phosphorus concentration in the overlying water remains negligible [108,109]. This lack of reduction could be attributed to the release of phosphorus from sediment sources [23,110,111]. In this process, labile phosphorus in interstitial water is supplemented by sediment solids, subsequently becoming concentrated, and is released into the overlying water. Such release can be notably intensified during cyanobacterial blooms [112]. Furthermore, the occurrence of cyanobacterial blooms contributes to the accumulation of active organic matter and a decrease in the dissolved oxygen concentration. Consequently, nutrient release is enhanced in anoxic environments, leading to exacerbated algal growth and perpetuating a detrimental cycle [113]. As a shallow lake, Lake Taihu is heavily influenced by wind-induced disturbances, which also enhance the recycling efficiency of phosphorus [23]. As a result, the algal concentration in the lake sharply increases, further impeding the restoration process [70].

3.2.1. Nutrient Suspension Flux

Studies on endogenous phosphorus release loads have focused on simulating sediment suspension. By establishing a quantitative relationship between the wind speed and sediment suspension rate, it was calculated that the annual flux loads of TN and TP released from the sediments in Lake Taihu were 4577 t and 1101 t, respectively, accounting for 15.4% and 39.3% of the total sources [114]. Under static conditions, the diffusion-driven release of nutrients from sediment into the overlying water is primarily governed by concentration gradients, and these nutrients can be easily used by alga. The annual phosphorus release of Lake Taihu sediments measured without wind disturbances was estimated to be 899 t, and the annual nitrogen release was 10,000 t [115]. In contrast, the annual phosphorus release measured using diffusive gradients in thin films (DGT) analysis was approximately 700 t [108]. These results indicate that endogenous release approximates one third of the external input. In fact, the intensity of dynamic nutrient release caused by wind and wave conditions is much higher than that under static conditions [23]. Although the annual flux of endogenous release might be offset by the sedimentation load [116], the rapid increase in nutrient concentrations in a water body during a short period of time can still trigger cyanobacterial blooms [114].

3.2.2. Relation to Sedimentary Organic Matter

The excessive endogenous phosphorus load in Lake Taihu primarily stems from the long-term accumulation of external phosphorus load in sediments [68]. Due to the presence of sedimentary organic matter (SOM), sediment acts as a reservoir for pollutants in the overlying water, and its association with Lake Taihu eutrophication has been well established [73]. The degradation of SOM, accompanied by nutrient release, is a key driver of eutrophication. Furthermore, the sources and composition of SOM significantly affect both the habitats of aquatic plants and the food quality of benthos organisms [95]. Researchers have employed the stable isotope analysis of nitrogen, phosphorus, and other elements in SOM [75,117,118], as well as the determination of the fatty acid composition [95,119], to identify the contributions of land, macrophytes, algae, and other sources to SOM in the lake.

The formation of authigenic phosphorus mainly occurs through organic matter mineralization and apatite precipitation, as indicated by the oxygen isotope ratios of phosphate [75]. In both estuaries and open waters, the proportion of land organic matter in sediments (46.8%~55.0%) exceeds those of algal sources (13.8%~23.4%) and macrophytes (20.0%~30.0%) [117]. This finding aligns with the evaluation results regarding organic matter sources in sediments in different zones of Lake Taihu, suggesting a substantial contribution from terrigenous matter [118].

By utilizing source-specific fatty acid biomarkers, the contributions of different potential sources of organic matter in the western region of Lake Taihu were evaluated, with terrestrial plants being identified as the primary source [95]. The eutrophication status of Lake Taihu significantly impacts the sources and composition of organic matter. Increased eutrophication leads to a higher contribution of aquatic organic matter sources compared to terrestrial sources [119], further supporting the positive feedback relationship between organic matter accumulation and lake eutrophication. To mitigate the impact of endogenous pollution on the eutrophication of Lake Taihu, it is crucial to not only control nutrient release from the sediment to the overlying water but also to reduce the concentrations and total amounts of pollutants in the sediment.

4. Source Control of Eutrophication in Lake Taihu

During the 1990s, the local government invested billions of Chinese Yuan each year to combat eutrophication in Lake Taihu [120]. A long-term fixed-point water quality monitoring system was implemented in Lake Taihu, leading to the significant accumulation of extensive and intricate datasets [109,121]. Moreover, a large-scale investment in ecological environment protection and restoration projects commenced in 2000 [122]. Since 2007, comprehensive treatment measures have been implemented in the Taihu Basin, including the establishment of new wastewater treatment facilities, the closure of chemical-intensive and highly polluting factories, the establishment of cyanobacteria salvaging sites and water–algae separation stations surrounding the lake, and sediment remediation efforts, as well as wetland protection and restoration initiatives [56]. The control of water pollution in Lake Taihu has exerted a notable influence on the spatiotemporal distribution of nutrient concentrations [24]. By 2017, significant reductions in nutrient indices were observed in the rivers surrounding the lake and the lake itself, with 12 of the 15 major rivers meeting or surpassing the Class III water quality standard [123]. Remarkable achievements in nutrient concentration (NH₄⁺-N, TN, TP) control were witnessed in most of the studied areas. Notably, source-targeted interventions demonstrated greater efficacy regarding ecosystem restoration in comparison to those focusing on pollution reduction [122].

4.1. Control of Non-Point-Source Pollution

Non-point-source pollution poses a significant threat to water quality in various global regions, bringing challenges in effective management and mitigation efforts (Adu and Kumarasamy). The Environmental Protection Law of China, enacted in 1989, established guidelines regarding strengthening rural environmental protection, adopting agricultural inputs (such as pesticides and fertilizers), and advancing the concept of agricultural non-point-source pollution control. However, the effective collection and treatment of wastewater, such as domestic wastewater, in the rural area, as well as township industrial wastewater, agricultural wastewater, and livestock farming wastewater, are still lacking. Wetlands or ponds have been widely employed globally to intercept non-point pollution [23], and the captured nutrients can be viably utilized if appropriately recycled [124]. However, the current pollution load of Lake Taihu surpasses the environmental capacity of the surrounding wetlands, leading to a delay in non-point-source pollution control. Moreover, climate change in the future is predicted to increase the frequency and intensity of extreme rainstorms, intensifying the challenge of mitigating non-point-source pollution and cyanobacterial blooms [125].

The source of non-point pollution in the Taihu Basin has undergone changes due to the rapid development of agriculture. The fruit, vegetable, and tea planting system has surpassed the rice planting system and become the major contributor to nitrogen and phosphorus losses in farmland [126]. Therefore, optimizing the planting structure of fruit, vegetables, tea, and rice, as well as the fertilization strategy, is important in reducing and controlling agricultural non-point-source pollution in the Taihu Basin [127,128].

Freshwater aquaculture in the Taihu Basin accounts for over 85% of inland fishery in China [129]. However, there is a lack of systematic control over the aquaculture wastewater around Lake Taihu. The supervision and management of livestock and poultry farms since 2001 has not effectively reduced eutrophication in many water bodies in China [130]. To address these issues, a case study was conducted on a town in the Taihu Basin with three non-point pollution sources: crop farming, livestock, and aquaculture. The results showed that reusing all livestock wastewater and manure together with partial aquaculture wastewater is the most effective method to control non-point-source pollution in the Taihu Basin, considering cost-effectiveness among the studied four scenarios [131].

4.2. Control of Point-Source Pollution

The expansion of wastewater treatment infrastructure constitutes the primary strategy to reduce the amounts of nutrients transported from point sources to Lake Taihu. During 2007 to 2020, the wastewater treatment facilities in the Taihu Basin expanded from 129 to 312 plants [132]. Meanwhile, discharge limits for pollutants in sewage are becoming increasingly stringent [133]. In 2018, the nearby provinces issued more stringent local standards than the National Discharge Standards (GB 18918-2002) [134] for sewage treatment plants and required corresponding technological upgrading to promote water resource protection [59].

Although point-source pollution has been effectively controlled, the water quality of Lake Taihu is still far from the expected level [40,135]. After the implementation of the Discharge Standard of Pollutants for Municipal Wastewater Treatment Plants (GB 18918-2002), the sewage treatment plants in China had already reduced their nutrient loads by around half; however, the load of pollutants discharged to the water bodies has increased compared with the 1980s because of the increase in the total sewage volume [136]. From 2008 to 2018, the average annual TN export to Lake Taihu was around 40,000~50,000 t, and that of TP was around 2000 t [23]. The lack of sewage treatment facilities and the low efficiency of existing facilities still exist [137]. The inappropriate industrial structure and distribution and the dominance of secondary industries, especially considering the excessive pollutant emissions from the effluents of sewage treatment plants, also lead to the severe pollution of river tributaries and poor water quality in the Lake Taihu area [120,138].

5. Conclusions and Perspectives

This review has synthesized the current understanding of eutrophication dynamics and source identification in Lake Taihu, a critical freshwater resource facing substantial environmental challenges. The escalating influx of human-derived nutrients has exacerbated eutrophication globally, with Lake Taihu serving as a prominent case study. Our analysis reveals that phosphorus emerges as the key driver of eutrophication in Lake Taihu, exhibiting a strong correlation with Chl a concentrations. The lake’s nutrient dynamics are characterized by significant temporal and spatial heterogeneity, influenced by both anthropogenic activities and meteorological conditions. Historical trends indicate fluctuating nutrient concentrations and cyanobacterial bloom occurrences, despite government-led control efforts. The resurgence of blooms highlights the complex interplay among factors governing eutrophication in this large, shallow lake.

The effective management of Lake Taihu’s eutrophication necessitates a comprehensive understanding of nutrient sources. This review has evaluated various methodologies, including multivariate statistical analysis, physical models, empirical models, and multidisciplinary approaches integrating GISs and machine learning. These methods have collectively identified agricultural non-point-source pollution and urban point-source discharges as major external nutrient contributors. Furthermore, the role of sediment nutrient release as an endogenous source, influenced by sedimentary organic matter and wind-induced resuspension, is increasingly recognized.

The evolution of source identification techniques underscores the shift from traditional methods to more sophisticated, spatially explicit assessments. ECMs provide valuable insights into nutrient loads from various sources, while physical models such as SWAT and SPARROW enhance our understanding of hydrological and water quality processes. Multidisciplinary approaches, incorporating GISs and machine learning, offer improved accuracy and predictive capabilities. Notably, the application of AI has shown promise in identifying complex relationships between nutrient concentrations and environmental factors.

Despite substantial investments in wastewater treatment plants and non-point-source pollution control, Lake Taihu continues to experience eutrophication challenges. The increasing volume of sewage discharge and the legacy of nutrient accumulation in sediments necessitate a paradigm shift towards integrated watershed management. Future research should prioritize the following: (a) refined source tracking—further refine source tracking methodologies, particularly for non-point-source pollution, using advanced techniques like stable isotope analysis and high-resolution spatial modeling; (b) sediment dynamics—enhance the understanding of sediment nutrient release mechanisms and develop effective sediment remediation strategies; (c) climate change impacts—investigate the impacts of climate change on nutrient loading and cyanobacterial bloom dynamics in Lake Taihu; (d) integrated modeling—develop integrated models that couple hydrological, water quality, and ecological processes to simulate the complex interactions within the lake ecosystem; (e) AI applications—expand the application of deep learning architectures for predictive modeling and eutrophication monitoring.

Ultimately, achieving the sustainable management of Lake Taihu’s eutrophication requires a holistic approach that integrates scientific research, policy implementation, and public engagement. This review provides a foundation for future efforts aimed at restoring and protecting this vital aquatic ecosystem.