Multi-Level Driving Mechanisms: Cascading Relationships Among Physical Factors, Nutrient Cycling, and Biological Responses in the Yangtze River–Lake Ecosystems

Abstract

River–lake systems in the middle and lower reaches of the Yangtze River Basin function as critical ecological interfaces for maintaining regional water security and biodiversity. However, the complex interplay between environmental factors and biological communities in these systems remains poorly understood, limiting evidence-based management strategies essential for achieving sustainable development goals. This study investigated the spatiotemporal heterogeneity of environmental variables and their relationships with biological communities across 36 sampling sites (4 rivers, 5 lakes) from January to November 2022. Significant spatial differences were observed between river and lake systems, with lakes exhibiting higher concentrations of biochemical oxygen demand (BOD₅), chlorophyll-a, permanganate index (CODMn), loss on ignition (LOI), sediment total nitrogen (STN), total organic carbon (TOC), and turbidity, while rivers showed elevated total nitrogen levels. These patterns reflected fundamental differences in hydrodynamic mechanisms, particularly water retention time and sedimentation processes. Environmental parameters displayed distinct seasonal variations, with BOD₅ increasing markedly in autumn, and chlorophyll-a showing system-specific peaks in lakes (May) and rivers (September). Multivariate analyses revealed that water temperature, organic matter, and nutrients (particularly phosphorus) were key drivers shaping both phytoplankton and benthic communities. The structural equation model identified a strong cascade pathway from turbidity through phosphorus to phytoplankton richness, and uncovered a “phosphorus paradox” wherein total phosphorus exhibited contrasting effects on different biological components—positive for phytoplankton richness but negative for benthic richness. The positive relationship between phytoplankton and benthic macroinvertebrate richness provided evidence for benthic–pelagic coupling in this river–lake system. These findings advance our understanding of the complex mechanisms linking physical factors, nutrient dynamics, and biological communities across river–lake continuums, providing a quantitative framework for ecosystem-based management that supports sustainable development in the Yangtze River Basin and similar freshwater systems globally.

1. Introduction

River–lake systems are integral components of large river basins, functioning as critical ecological interfaces and wetland reserves and play a vital role in maintaining regional water security and ecological balance [1,2,3]. A prime example of the significance of river–lake systems can be found in the middle and lower reaches of the Yangtze River Basin, where large river-connected lakes such as Dongting Lake and Poyang Lake are prominent features [1]. Furthermore, the influence of river–lake systems extends far beyond their immediate boundaries, significantly impacting biodiversity and ecosystem health on a broader scale [2]. These systems offer a range of essential ecosystem services, including but not limited to water storage, flood regulation, and the provision of habitats for diverse aquatic species [3]. The interconnectivity between rivers and lakes fosters dynamic environments that support unique ecological processes and intricate nutrient cycling patterns [1,4,5].

Benthic–pelagic coupling emerges as a fundamental process regulating ecosystem function within these intricate river–lake systems [6,7]. This coupling facilitates the exchange of energy, nutrients, and organisms between benthic and pelagic zones, profoundly influencing nutrient cycling, primary production, and food web dynamics [7,8]. Notably, the intensity and nature of this coupling exhibit spatial and temporal variations, reflecting the heterogeneity of river–lake environments and their responses to environmental fluctuations [9,10,11]. Furthermore, Sediment organic thresholds function as critical triggers for ecosystem decoupling in river–lake systems, significantly affecting biogeochemical processes and ecological stability [12,13]. The accumulation of organic matter in sediments can significantly alter biogeochemical processes, potentially leading to shifts in ecosystem functioning [12]. These thresholds may act as tipping points, beyond which the benthic–pelagic coupling becomes disrupted, affecting nutrient cycling, primary production, and food web dynamics [13]. Within this context, the strength of these interactions directly and indirectly affects the overall stability and resilience of aquatic ecosystems [14], potentially triggering cascading effects throughout the food web [11,15].

The complex interplay between multidimensional environmental factors and biological communities in river–lake systems presents a significant challenge in understanding ecosystem dynamics [16]. Recent studies have highlighted the intricate relationships and heterogeneity that shape these systems, revealing a nuanced web of interactions that extend beyond simple cause-effect paradigms [17,18]. The concept of spatiotemporal heterogeneity has emerged as a fundamental characteristic of river–lake ecosystems, profoundly influencing biodiversity patterns and ecological processes [16,18]. This heterogeneity manifests as a mosaic of physical, chemical, and biological gradients that fluctuate across both space and time, creating a dynamic landscape of ecological niches and environmental conditions [19,20]. The interplay between these gradients and the organisms that inhabit these ecosystems gives rise to intricate patterns of species distribution and community composition [20]. The multifaceted nature of environmental variables and their synergistic effects on biotic components necessitate a comprehensive approach to unraveling the underlying mechanisms driving ecosystem functioning [21].

The concept of cascade effects has emerged as a powerful framework for elucidating the intricate cause-and-effect chains that permeate aquatic ecosystems [22]. This theoretical construct posits that changes in one component of the ecosystem can trigger a series of interconnected responses, ultimately leading to far-reaching consequences for the entire system [22,23]. The historical development of causal chain concepts in aquatic ecology has paved the way for more sophisticated analytical approaches. Among these, piecewise Structural Equation Modeling (SEM) has emerged as a particularly promising tool for quantifying cascade mechanisms in complex river–lake ecosystems [24,25]. This technique allows to decompose intricate ecological relationships into more manageable components, facilitating the identification of key pathways and feedback loops that drive ecosystem dynamics [24,25,26].

The middle and lower Yangtze River Basin (MLYRB) represents a complex and dynamic ecological interface, offering an ideal model system for investigating river–lake connectivity and ecological processes. This region, encompassing diverse geomorphological features, harbors a rich tapestry of biodiversity and ecosystem services [27,28]. The intricate network of lakes, wetlands, and river channels within the MLYRB forms a highly interconnected hydro-ecological system, the dynamics of which profoundly influence regional ecological security and economic development [29,30]. Recent years have witnessed significant spatiotemporal changes in the ecological landscape of the MLYRB, driven by both natural factors and intensifying human activities [30]. These changes not only reflect the region’s inherent environmental variability but also underscore the far-reaching impacts of anthropogenic interventions on ecosystem functioning, such as the construction of dam [31,32].

Understanding river–lake ecosystem dynamics has direct implications for achieving multiple Sustainable Development Goals (SDGs). The Yangtze River Basin, as China’s most critical freshwater ecosystem supporting over 600 million people, exemplifies the intricate linkages between ecological integrity and sustainable development [33]. Recent assessments reveal that water quality improvement in China’s rivers has progressed unevenly, with lakes lagging behind rivers despite intensive governance efforts, indicating that sustainable management requires integrated approaches addressing both external catchment loads and internal lake processes [34]. The deterioration of aquatic ecosystems in the Yangtze Basin has been declared a significant constraint to meeting sustainable development objectives, with the river system being characterized as “sick” and struggling to fulfill its ecological functions [33]. Achieving SDG 6 (Clean Water and Sanitation), SDG 14 (Life Below Water), and SDG 15 (Life on Land) in this region necessitates a comprehensive understanding of the physical-chemical-biological cascade mechanisms that govern ecosystem functioning. Water sustainability evaluations across China have identified the Yangtze River Delta and surrounding areas as exhibiting relatively poor performance in water conservation and aquatic ecology from 2010 to 2023, underscoring the urgent need for evidence-based management strategies [35].

This study aligns with China’s national strategic shift toward “prioritizing ecological conservation and green development” in the Yangtze River Basin, a policy transformation that has fundamentally reshaped river basin governance since 2016 [36]. The implementation of the Yangtze River Protection Law and associated regulations has established a comprehensive policy framework targeting sustainable development through ecosystem-based management. However, translating ecological research findings into actionable management strategies remains a critical gap. Recent analyses demonstrate that accounting for interactions between different SDGs is essential for effective water pollution control, as interventions targeting one goal may inadvertently impede progress toward others [37]. Therefore, this research adopts a transdisciplinary perspective that not only quantifies ecological processes but also explicitly considers their implications for integrated river–lake governance. By revealing the cascade mechanisms linking physical drivers to biological responses, our findings provide a quantitative foundation for designing management interventions that can simultaneously advance multiple sustainability objectives while avoiding unintended trade-offs.

This study employed a comprehensive, multi-faceted approach to unravel the complex dynamics of the middle and lower Yangtze River Basin (MLYRB) ecosystem. Our methodology integrated multi-temporal and multi-site environmental and biological monitoring with advanced statistical techniques, creating a robust framework for ecological analysis. By combining multivariate statistical analyses with piecewise Structural Equation Modeling (SEM), we have developed a novel framework that captures the intricate relationships between physical, chemical, and biological factors across hierarchical levels. The research design encompassed three primary objectives: (1) to quantify the spatiotemporal heterogeneity patterns of environmental variables within the MLYRB river–lake system, (2) to elucidate the mechanisms linking environmental factors to phytoplankton and benthic communities, and (3) to construct and validate a cascade effects model that illustrates the “physical drivers–nutrient cycling–biological response” pathway. This multi-pronged approach allows for a nuanced understanding of ecosystem dynamics, moving beyond traditional single-factor analyses to capture the complex interplay of environmental and biological processes. This research advances the understanding of river–lake ecosystem dynamics, providing valuable insights for the development of integrated management strategies in the Middle and Lower Yangtze River Basin and similar aquatic systems globally.

2. Methods

2.1. Study Area

The study was conducted in the middle and lower reaches of the Yangtze River Basin in Hubei Province, China, encompassing representative river–lake systems that span longitudinally from 111.78° E to 116.08° E and latitudinally from 29.56° N to 30.46° N (Figure 1). This region is characterized by a subtropical monsoon climate with distinct seasonal variations, featuring an annual mean temperature of approximately 15–17 °C and average annual precipitation of 1100–1300 mm [38]. The investigated water bodies form a complex interconnected hydrological network influenced by both natural processes and anthropogenic modifications, particularly the operation of the Three Gorges Dam located upstream.

We selected sampling sites following a “highlight priorities, combine points with areas” approach to ensure comprehensive coverage of both upstream-downstream gradients and left-right bank variations. The study area encompassed two distinct hydrological systems: (1) river systems, including the Southern Jingzhou Four Rivers network (Songzi River, Hudu River, Ouchi River, and Tianxian River); and (2) lake systems, comprising Sihu Region lakes (Lake Hong and Lake Chang) on the upper Yangtze’s left bank, Lake Liangzi (including Liangzi, Niushan, and Tuzhen lakes) on the middle Yangtze’s right bank, and Lake Longgan and Lake Futou on the lower Yangtze’s left bank.

Beyond geographical coverage, site selection explicitly considered the morphological and ecological characteristics of each water body to ensure spatial representativeness. For lake systems, sampling sites were strategically positioned to capture both littoral (nearshore) and pelagic (offshore/central) zones, which represent ecologically distinct habitats with different hydrodynamic regimes, light availability, and benthic–pelagic coupling intensity. For river systems, sampling sites were distributed longitudinally along each of the four rivers (Songzi, Hudu, Ouchi, Tianxian) from upstream to downstream sections, with consideration of channel morphology variations (e.g., meandering vs. straight reaches, shallow vs. deep sections). This longitudinal gradient design captures spatial variations in flow velocity, sediment transport capacity, and cumulative anthropogenic inputs along river corridors. The combined river–lake sampling framework thus encompasses multiple spatial scales: (1) within-lake heterogeneity (littoral-pelagic gradients), (2) among-lake variability (different morphometric and trophic characteristics), (3) longitudinal riverine gradients (upstream–downstream), and (4) river–lake contrasts (lotic vs. lentic systems).

To complement the field sampling design with systematic information on anthropogenic disturbance patterns, we conducted a characterization of each sampling site using remote sensing and geospatial data. Land use/land cover (LULC) within a 500 m buffer zone around each sampling site was characterized using Landsat 8 Operational Land Imager (OLI) imagery acquired during the 2022 sampling period (30 m spatial resolution), obtained from the United States Geological Survey (USGS) Earth Explorer database (https://earthexplorer.usgs.gov/; accessed on 24 March 2025). Image classification was performed using a supervised maximum likelihood classifier in ArcGIS 10.8, with training samples selected based on Google Earth high-resolution imagery (2022) and field photographs. Land cover was classified into five categories following the Current Land Use Classification (GB/T 21010-2017 [39]): (1) cropland, (2) forest/grassland, (3) water bodies, (4) built-up land (urban/industrial areas, roads), and (5) bare land.

For each site, we calculated the percentage coverage of each land use type within the 500 m buffer and characterized primary anthropogenic disturbance sources based on: (1) proximity to urban centers or industrial facilities (<2 km = high urban/industrial influence), (2) percentage of agricultural land (>50% = high agricultural influence), (3) visible point sources (wastewater discharge pipes, aquaculture facilities) identified from high-resolution imagery and field notes, and (4) information from local environmental management agencies when available. Finally, it was confirmed that all sampling points cover various gradients from low anthropogenic disturbance to high anthropogenic disturbance, exhibiting a certain degree of environmental heterogeneity and representativeness of the sample.

A total of 36 sampling sites were established across the study area based on hydrological connectivity and geographical representation (Figure 1). The river systems comprised 13 sampling sites distributed across four major rivers. The Songzi River was monitored at five locations extending from its upstream section (30.35° N, 111.78° E) to its downstream reaches (29.82° N, 112.08° E), capturing the longitudinal gradient of this waterway. The Hudu River was sampled at three locations that spanned from its upper reaches (30.22° N, 112.12° E) to its downstream section (29.67° N, 112.17° E), encompassing its primary flow path. Three sampling points were positioned along the Ouchi River course (29.79° N–29.61° N, 112.32° E–112.39° E) to characterize its environmental conditions. The Tianxian River was monitored at two locations within the Hubei provincial section (29.67° N–29.63° N, 112.59° E–112.57° E), representing the terminal portion of the Southern Jingzhou river network.

The lake systems were more extensively sampled with 23 sites distributed across five major lakes. Lake Chang was monitored at five locations (30.39° N–30.46° N, 112.34° E–112.49° E) strategically positioned to capture spatial heterogeneity across its basin. Similarly, Lake Hong was sampled at five sites (29.81° N–29.92° N, 113.28° E–113.43° E) covering different functional zones including central pelagic areas and near-shore regions. The Lake Liangzi system, which encompasses the main lake body and connected water bodies, was monitored at five locations (30.19° N–30.32° N, 114.45° E–114.59° E) distributed to represent its morphometric and ecological diversity. Lake Longgan, situated in the easternmost portion of our study area, was sampled at three sites within the Hubei provincial section (29.91° N–29.95° N, 116.03° E–116.08° E). Futou Lake was monitored at five locations (29.99° N–30.06° N, 114.19° E–114.26° E) encompassing its various ecological zones from littoral to pelagic areas.

Field surveys were conducted on a bimonthly basis from January to November 2022, with sampling in January, March, May, July, September, November. Each survey was synchronized to ensure that sites within the same water body were sampled within a 48 h window to minimize temporal variability between comparative samples. To account for potential diel variations, all sampling activities were performed during daylight hours (9:00–16:00), and the sampling sequence for sites within each water body was maintained consistent throughout the study period. Weather conditions were recorded for each sampling event, and sampling was avoided during heavy rainfall events to minimize the influence of short-term weather disturbances. All field activities followed standardized operating procedures to ensure data quality and comparability across sampling events. Site coordinates were recorded using a handheld GPS device with an accuracy of ±3 m.

2.2. Environmental Variables and Biological Community

Water and sediment samples were collected bimonthly from January to December 2022 at 36 sampling sites. At each site, in situ physical-chemical parameters were measured using a multi-parameter probe (YSI) at 0.5 m below the water surface, including water temperature (WT, °C), pH, dissolved oxygen (DO, mg/L), electrical conductivity (EC, μS/cm), turbidity (Turb, NTU), and oxidation-reduction potential (ORP, mV). Water transparency was measured using a standard 30 cm Secchi disk (SD, cm), and water depth (WD, m) was determined with a digital depth sounder. All in situ measurements were conducted in triplicate, and the mean values were used for analysis.

Water samples for chemical and biological analyses were collected at 0.5 m below the surface using a 2.5 L Plexiglas water sampler. For each site, 2 L of water was collected in acid-cleaned polyethylene bottles for chemical analysis, and 1 L was preserved with Lugol’s solution (final concentration 1%) in amber glass bottles for phytoplankton analysis. Sediment samples were collected using a Peterson grab sampler (1/16 m²). The upper 5 cm of sediment was collected using a plastic core tube, homogenized, and stored in sealed polyethylene bags at 4 °C for subsequent physicochemical analysis. The selection of the upper 0–5 cm sediment layer for sampling in Yangtze River basin studies is scientifically justified based on biogeochemical and ecological considerations. Previous investigations in the Yangtze River system have consistently adopted this sampling depth for assessing organic carbon dynamics, with forest stream sediment studies in the upper Yangtze River successfully employing 0–5 cm samples to quantify organic carbon concentrations and stocks across seasonal variations [40]. Studies in similar Chinese shallow lakes have consistently demonstrated that benthic macroinvertebrates predominantly inhabit the surface sediment layer. In Lake Taihu, the third-largest freshwater lake in China with comparable depth and substrate characteristics to our study lakes, 39.25% and 24.87% of total benthic abundance occurred within the 0–5 cm and 5–10 cm sediment layers, respectively, with density markedly declining with increasing depth [41]. In other study, additional Peterson grab sediment samples were collected at each sampling station in Lake Taihu to determine the total nitrogen and total phosphorus content in the surficial sediments, with the sampling depth specified as approximately the upper 5 cm [42]. Moreover, in coastal sediment research, the study of the La Paz lagoon in Northwest Mexico demonstrated that 0–5 cm surface samples effectively capture pharmaceutical residues associated with wastewater inputs, supporting this interval as an appropriate depth for assessing contaminant retention [43]. These findings collectively support the 0–5 cm sampling protocol as an ecologically meaningful and methodologically sound approach for characterizing active sediment biogeochemistry in river and lake systems.

Water samples were transported to the laboratory in refrigerated containers (4 °C) and analyzed within 24 h of collection. Total nitrogen (TN) and total phosphorus (TP) were determined using the alkaline potassium persulfate digestion-UV spectrophotometric method and ammonium molybdate spectrophotometric method, respectively, following standard procedures [44]. Ammonia nitrogen (NH₃-N) was measured using the Nessler’s reagent colorimetric method. Chemical oxygen demand (COD) was determined by the dichromate method, biochemical oxygen demand (BOD₅) by the 5-day incubation method, and permanganate index (CODMn) by the acidic potassium permanganate method. Chlorophyll-a (Chl-a) was extracted with 90% acetone after filtration through Whatman GF/C filters and quantified spectrophotometrically.

Sediment samples were air-dried at room temperature, ground, and passed through a 100-mesh sieve for homogenization. Sediment moisture content (SMC) was determined by the weight loss after drying at 105 °C for 24 h. Loss on ignition (LOI) was measured by combustion at 550 °C for 4 h. Sediment total nitrogen (STN) and sediment total phosphorus (STP) were determined after H₂SO₄-HClO₄ digestion using the same methods as for water samples. Total organic carbon (TOC) was measured using the K₂Cr₂O₇ oxidation method. All analytical procedures followed standard methods [44], and quality control was performed using certified reference materials and analytical blanks.

Phytoplankton samples (1 L) were preserved with Lugol’s solution in the field and allowed to settle for 48 h in the laboratory. After careful siphoning of the supernatant, the concentrated samples (30 mL) were homogenized and examined using an Olympus BX51 microscope (Olympus Corporation, Tokyo, Japan) at 400× magnification. Phytoplankton were identified to the lowest possible taxonomic level (primarily genus). A minimum of 300 individuals from each sample were counted in random fields. Species abundance was expressed as cells/L, and community metrics including species richness and evenness were calculated.

Benthic macroinvertebrates were collected using a Peterson grab sampler (1/16 m²) with three replicates at each site. Samples were washed through a 425 μm mesh sieve in the field, and the retained material was preserved in 75% ethanol. In the laboratory, benthic organisms were sorted, identified to the lowest possible taxonomic level (genus or species) according to taxonomic keys, and counted under a stereomicroscope. Taxonomic richness was determined as the total number of taxa present in each sample. All specimens were deposited in the collection of the Institute of Hydrobiology, Chinese Academy of Sciences (Wuhan, China).

All water quality, sediment, and biological analyses were performed in triplicate, and mean values were used for statistical analysis. Standard deviations were typically less than 10% of the mean values for all measured parameters, indicating acceptable analytical precision.

2.3. Statistical Analyses

Spatial variations in environmental variables between river and lake systems were tested using non-parametric Kruskal–Wallis tests due to non-normal distribution of most variables. Temporal patterns were analyzed through time series plots to visualize monthly variations across systems. All statistical analyses were considered significant at p < 0.05.

Principal Component Analysis (PCA) was performed to identify patterns of environmental heterogeneity among river and lake systems. Data were standardized to zero mean and unit variance prior to analysis to account for different measurement scales. Redundancy Analysis (RDA) was used to examine relationships between environmental variables and community composition data (phytoplankton and benthic macroinvertebrates). Prior to RDA, environmental variables also were log-transformed as needed to improve normality and standardized to zero mean and unit variance. Moreover, Detrended Correspondence Analysis (DCA) was conducted to determine the appropriate ordination method; as all gradient lengths were <3, the linear RDA model was selected rather than unimodal Canonical Correspondence Analysis [45]. Environmental variables were selected using a forward selection procedure with Monte Carlo permutation tests (999 permutations) to retain only significant explanatory variables (p < 0.05).

A piecewise Structural Equation Model (SEM) was constructed to quantify the cascading relationships among physical drivers, nutrient cycling, and biological responses. Prior to model construction, multicollinearity among environmental variables was assessed using Variance Inflation Factors (VIFs), and variables with VIF > 10 were removed to minimize redundancy. Environmental variables were then organized into three hierarchical layers: (1) physical drivers (WT, Turb, EC, pH, DO), (2) nutrient cycling (TP, TN, COD, TOC, SMC, STP), and (3) biological responses (phytoplankton richness, phytoplankton evenness, and benthic macroinvertebrate richness).

The piecewise SEM approach was selected to accommodate the complex, hierarchical dataset structure while allowing for the inclusion of random effects. The initial model was constructed based on theoretical expectations, then refined by removing non-significant pathways (p > 0.05). Model fit was evaluated using Fisher’s C statistic, with p > 0.05 indicating adequate fit. Standardized path coefficients were calculated to facilitate comparison of effect sizes among variables. Both marginal R² (variance explained by fixed effects) and conditional R² (variance explained by both fixed and random effects) were calculated to assess model explanatory power. The piecewise SEM was implemented using the “piecewiseSEM” package [24].

All statistical analyses were performed in R 4.1.0 [46].

3. Results

3.1. Spatiotemporal Heterogeneity Patterns

The boxplots (Figure 2) revealed significant spatial differences in environmental variables between lake and river systems. Variables such as BOD5 (5-day Biochemical Oxygen Demand), Chl-a (Chlorophyll-a), CODMn (Permanganate Index), LOI (The Loss on Ignition), STN (Sediment Total Nitrogen), TOC (Total Organic Carbon), and Turb (Turbidity) exhibited markedly higher values in lakes compared to rivers (p < 0.05). For example, BOD5, which measures organic matter decomposition, was substantially elevated in lakes, indicating the accumulation of organic material likely due to reduced hydrodynamic flushing. Similarly, STN (sediment total nitrogen) was significantly higher in lakes, reflecting nutrient retention and organic matter deposition in these systems. Chl-a, an indicator of phytoplankton biomass, was significantly elevated in lakes compared to rivers (p < 0.05), highlighting the nutrient-rich conditions that promote primary productivity in these systems. Additionally, CODMn and Turbidity were higher in lakes, suggesting increased suspended organic matter and particles, which is consistent with the stagnant hydrological conditions in lakes. In contrast, TN (Total Nitrogen) showed significantly higher values in rivers, likely due to higher nutrient transport from upstream sources. Other variables, such as pH, ORP (Oxidation-Reduction Potential), and NH₃-N (Ammonia Nitrogen), exhibited no significant differences between the two systems, indicating relatively stable chemical conditions across both environments.

Temporal analysis revealed distinct seasonal patterns across multiple parameters (Figure 3). BOD₅ values remained relatively stable in both systems from January to July, followed by a marked increase in September and November, with lakes consistently showing higher values than rivers. This pattern suggests seasonal accumulation of organic matter, particularly pronounced in lake environments during autumn months. Chl-a exhibited notable fluctuations, with peak concentrations in May for lakes and September for rivers. The consistently higher Chl-a levels in lakes compared to rivers throughout most of the year indicate enhanced primary productivity in lentic systems. LOI parameter showed similar trends in both systems, with relatively stable values from January to July, followed by a sharp increase in September and November, reaching approximately 50% in both systems. This synchronous pattern indicates comparable organic matter accumulation processes in the sediments of both environments during autumn. Nutrient dynamics revealed system-specific patterns, with NH₃-N showing higher concentrations in rivers during winter months (January–March), followed by convergence between the two systems for the remainder of the year. TN was consistently higher in river systems throughout the year, with peak values in May, while TP displayed similar temporal patterns in both systems with maximum concentrations in lake system of September. These temporal dynamics highlight the distinct biogeochemical processes operating in river and lake systems, with lakes generally exhibiting greater seasonal variability in most parameters, particularly those related to primary productivity and organic matter cycling.

Principal component analysis (PCA) of environmental variables revealed distinct patterns between lake and river systems (Figure 4). The first two principal components explained 36.5% of the total variance (PC1: 19.9%, PC2: 16.6%). Lake samples showed greater dispersion throughout the ordination space compared to river samples, which formed a more compact cluster, indicating higher environmental heterogeneity in lake systems. PC1 was characterized by strong positive loadings of productivity-related parameters (Chl-a, CODMn, TP) and physical variables (Turb, COD, WT), while Secchi depth (SD) showed a strong negative loading. PC2 was primarily defined by variables associated with organic matter and sediment characteristics (BOD5, LOI, SMC, pH) in the positive direction, with dissolved oxygen (DO) and water depth (WD) loading negatively.

Lake environments were more strongly associated with higher productivity indicators (Chl-a, TP), organic matter content (BOD5, LOI), and pH, while river environments correlated with greater water clarity (SD) and dissolved oxygen (DO). Nutrient variables including EC and NH3-N showed relatively weak influence in differentiating between the two systems, as evidenced by their shorter vector lengths positioned near the origin of the ordination.

3.2. Biotic-Environment Linkages

Redundancy analysis (RDA) revealed significant relationships between environmental variables and biological communities across seasonal patterns (different month). For phytoplankton (Figure 5), the first two RDA axes explained 70.33% of the total variance, with RDA1 accounting for 50.16% and RDA2 explaining 20.17%. For benthic organisms (Figure 6), the first two axes explained 78.75% of the variance (RDA1: 65.34%, RDA2: 13.41%). Both RDA analyses demonstrated that water temperature, organic matter (indicated by BOD5, COD, CODMn), and nutrients (particularly phosphorus and nitrogen compounds) were the primary environmental factors influencing the community structures in this lake system, with clear seasonal differentiation observed in both phytoplankton and benthic communities.

The phytoplankton community exhibited clear seasonal clustering patterns. November samples were predominantly distributed in the upper left quadrant and associated strongly with BOD5 and LOI. May samples were scattered in the lower portion of the plot, correlating with WT, CODMn, TP, and Chl.a. January samples were positioned mainly in the right portion, showing associations with STP, while March and July samples displayed more scattered distributions. Diatom species including Melosira varians, Cyclotella sp., and Synedra sp. were positioned near the center of the ordination plot, indicating their widespread distribution across different environmental conditions, whereas Raphidiopsis sp. showed stronger associations with higher water temperatures and phosphorus concentrations. Furthermore, Melosira granulata exhibited preferential distribution in environments characterized by reduced BOD5, TOI, and pH values, suggesting a potential ecological affinity for less contaminated aquatic habitats.

For benthic organisms, the RDA ordination also revealed distinct seasonal patterns. STN and SD vectors pointed toward the upper right quadrant, correlating with winter samples, particularly those from January. BOD5 and pH vectors were associated with November samples. TN, COD, and WT vectors pointed toward the left side of the plot, demonstrating their influence on summer samples from May and September. The benthic taxa showed varying environmental preferences. Tanypus sp. was positioned toward areas with higher BOD5, pH, and STN values, indicating their preference for more polluted conditions. Propsilocerus sp. and Hydrobaenus sp. showed stronger associations with elevated STN and STP concentrations, also suggesting their tolerance to certain types of pollution, albeit with different specific preferences than Tanypus sp. The majority of other benthic species were clustered near the center of the ordination plot, suggesting no strong environmental preferences among the measured variables.

3.3. Cascade Mechanism Quantification

To investigate the cascading relationships among physical drivers, nutrient cycling, and biological responses in the River and Lake system, we developed a piecewise structural equation model (SEM). The model integrated multiple environmental variables into three hierarchical layers: physical drivers (including WT, Turb, EC, pH, DO), nutrient cycling (TP, TN, COD, TOC, SMC, STP), and biological responses (phytoplankton richness, phytoplankton evenness, and benthic macroinvertebrate richness).

The final SEM model exhibited excellent fit with the observed data (Fisher’s C = 43.132, p = 0.261, df = 38), indicating that the proposed causal structure adequately represented the relationships among variables. The model passed all tests of directed separation except for one relationship between phytoplankton richness and evenness (p = 0.0108). The SEM explained substantial variation in the environmental and biological variables, with conditional R² values (including both fixed and random effects) ranging from 0.90 to 1.00. The marginal R² values (fixed effects only) were highest for TP (0.52) and phytoplankton richness (0.22), while lower for benthic richness (0.12), TN (0.08), and phytoplankton evenness (0.03).

The SEM revealed significant cascading effects among the three hierarchical layers (Figure 7). Within the physical-nutrient relationship, turbidity showed a strong positive effect on TP (standardized path coefficient = 0.717, p < 0.001), while TOC had significant negative effects on both TP (std. coef. = −0.156, p < 0.01) and TN (std. coef. = −0.156, p < 0.05). Additionally, TP positively influenced TN concentrations (std. coef. = 0.212, p < 0.01) (Table S1).

In the nutrient-biological response layer, DO and TP were the strongest predictors of phytoplankton richness, with significant positive effects (std. coef. = 0.369, p < 0.001 and std. coef. = 0.339, p < 0.001, respectively). Turb negatively affected phytoplankton evenness (std. coef. = −0.222, p < 0.05), while TP showed no significant effect on phytoplankton evenness (p = 0.447).

For benthic macroinvertebrate richness, multiple significant pathways were identified. TOC and TP exhibited negative effects (std. coef. = −0.234, p < 0.001 and std. coef. = −0.194, p < 0.01, respectively), while SMC showed a positive influence (std. coef. = 0.159, p < 0.05). Notably, phytoplankton richness positively affected benthic macroinvertebrate richness (std. coef. = 0.232, p < 0.01), suggesting a bottom-up trophic linkage between these communities.

4. Discussion

4.1. Heterogeneity in the River–Lake Systems

The study revealed significant spatial differentiation in environmental gradients between river and lake systems in the middle and lower reaches of the Yangtze River. Lakes exhibited higher levels of BOD5, Chl-a, CODMn, LOI, STN, TOC, and turbidity compared to rivers, while TN concentrations were elevated in riverine environments. Similar spatial variations in microplastic pollution was found between river and lake sediments, indicating that environmental gradients influenced both conventional water quality parameters and pollutant distribution patterns [47]. Differences in hydrodynamic mechanisms, particularly water retention time and sedimentation processes, largely accounted for the observed spatial differentiation [48,49]. Complex river–lake interactions in the middle Yangtze River Basin involved stronger flow dynamics in rivers and relatively slower water exchange rates in lake systems [29,50]. The extended hydraulic retention time in lakes facilitates increased sedimentation and accumulation of organic matter and nutrients, resulting in elevated levels of BOD5, Chl-a, and other indicators. In contrast, the continuous flow in rivers promoted nutrient transport (possibly from industrial wastewaters), potentially explaining the higher TN concentrations observed in riverine environments.

This spatial heterogeneity significantly impacted ecosystem functioning. For example, fish composition and diversity in shallow lakes of the Yangtze River Basin correlate closely with environmental variables [51]. The diverse habitats created by environmental heterogeneity support higher biodiversity. However, human-induced environmental changes may disrupt this naturally formed heterogeneity, posing a threat to ecosystem stability [52].

Time series analysis revealed distinct seasonal dynamics in environmental factors. BOD5 showed a marked increase in autumn, while Chl-a peaks occured at different times in lakes (May) and rivers (September). These seasonal variations reflected intricate interactions between ecological processes and environmental factors. For example, Significant spatial and temporal patterns affecting water quality parameters and algal community composition and diversity also have been observed in the Dongting Lake, a typical interconnected river–lake system [17]. The asynchronous seasonal patterns observed between rivers and lakes have important ecological implications. Complex interactions between Poyang Lake and its surrounding rivers, including the Yangtze River, created a unique hydrological regime supporting diverse aquatic communities [53]. This temporal heterogeneity provides various ecological niches and resources throughout the year, potentially enhancing biodiversity and ecosystem resilience [54].

While our bimonthly sampling design (six events per year) successfully captured broad spatiotemporal patterns across multiple river–lake systems, we acknowledge that higher temporal resolution could have revealed additional ecological dynamics. Comparative studies have demonstrated that bimonthly sampling regimes (8 sampling events per year) may capture >75% of seasonal variance, whereas traditional seasonal sampling (4 events per year) typically explains <65% of total variance [55]. Moreover, increased sampling frequency has been shown to strengthen correlations between biotic data and seasonal variables such as temperature, salinity, and nutrients [55]. For rapidly growing planktonic taxa, population abundances can fluctuate by a factor of 10 or more over periods of 1–2 weeks, imposing important constraints on sampling design and indicating that monitoring at longer intervals may omit critical ecological information [56,57]. Additionally, episodic carbon flux events in benthic–pelagic systems, such as those associated with large diatom colonies, are temporally and spatially scarce and therefore difficult to capture through fixed-interval sampling [58]. Consequently, our bimonthly sampling frequency may not have fully captured rapid interactions between benthic and pelagic communities or their short-term responses to environmental fluctuations, potentially underestimating seasonal variations in benthic–pelagic coupling strength and missing transient but ecologically significant coupling events.

4.2. Environmental Factors Shaping Aquatic Biotic Communities

Redundancy Analysis (RDA) revealed that water temperature, organic matter, and nutrients (particularly phosphorus) were key factors shaping phytoplankton communities, exhibiting distinct seasonal differentiation. Water temperature emerged as a critical driver of phytoplankton community structure, consistent with findings from other studies in the Yangtze River Basin [59]. However, the mechanism by which temperature influences phytoplankton communities is multifaceted. Temperature not only directly affected phytoplankton metabolism but also indirectly influenced nutrient cycling and stratification patterns in river–lake systems [60].

The role of nutrients, especially phosphorus, in shaping phytoplankton communities was significant but varied across different studies. While our findings emphasized phosphorus, other studies found that both nitrogen and phosphorus were not primary drivers of phytoplankton functional group variations in the Yangtze River Basin [61,62]. This discrepancy might be attributed to differences in nutrient limitation patterns across the Yangtze River Basin. The relative importance of nitrogen versus phosphorus in limiting phytoplankton growth can shift seasonally and spatially, emphasizing the need for site-specific management strategies [63]. In this study, temperature exerted a significant regulatory effect, whereby elevated thermal conditions appeared to augment the relative importance of phosphorus limitation in phytoplankton growth kinetics [64], suggesting complex interactive mechanisms between thermal regimes and nutrient bioavailability in aquatic ecosystems.

The preference of phytoplankton communities observed in our study reflect complex interactions between environmental factors and biological processes. Our results showed that cyanobacteria, particularly Raphidiopsis, exhibited a strong affinity for warmer waters with elevated phosphorus levels. Our thermal-dependency observations align with recent studies showing cyanobacterial proliferation in response to elevated water temperatures within urbanized subtropical lake systems [65]. However, this study demonstrated a more pronounced relationship between cyanobacteria and phosphorus levels, suggesting that in our river–lake system, phosphorus may play a more critical role in cyanobacterial dominance than in isolated lake systems.

The RDA revealed distinctive associations between benthic macroinvertebrates and environmental gradients across our study sites, with certain taxa demonstrating clear environmental preferences. Particularly noteworthy was the distribution of Tanypus sp., which showed strong affinity for environments characterized by elevated BOD₅, pH, and STN. This chironomid taxon appears well-adapted to organically enriched conditions, aligning with previous observations in Agro-industrial District of Brazil [66]. Their prevalence in high-BOD₅ environments likely reflects specialized physiological adaptations—including enhanced hemoglobin expression and respiratory efficiency—that permit survival in periodically hypoxic conditions frequently associated with organic enrichment [67,68]. Interestingly, our analysis revealed differential responses among chironomid genera. While Tanypus exhibited tolerance to organic pollution indicators, Propsilocerus sp. and Hydrobaenus sp. showed distinctive associations with nutrient parameters, particularly elevated sediment nitrogen and phosphorus. This pattern suggests niche differentiation among seemingly related taxa, with each genus exploiting specific biogeochemical microhabitats within the broader benthic environment [69]. The mechanistic basis for such differentiation may involve feeding specializations.

The clustering of most benthic taxa near the ordination center merits particular attention. Rather than indicating analytical weakness, this pattern reveals an important ecological characteristic of the Yangtze floodplain benthic communities—the prevalence of environmental generalists. Such distributed positioning suggests adaptations to the inherent hydrological variability of these systems, where successful taxa must tolerate substantial fluctuations in physicochemical conditions. This contrasts markedly with findings from more stable lentic systems, where pronounced environmental filtering typically produces clearer taxon-specific associations [70]. The generalist strategy predominant in our study system likely represents an evolutionary response to the historical disturbance regime characterized by monsoon-driven water level fluctuations that create naturally dynamic conditions in these lake–river networks [71].

4.3. Benthic–Pelagic Cascade Mechanisms and Coupling

The piecewise structural equation modeling approach unveiled complex cascading relationships among physical drivers, nutrient cycling, and biological communities in the middle-lower Yangtze lake–river systems. Unlike traditional correlation analyses that merely identify associations, our SEM quantified direct and indirect pathways through which environmental factors shape biological communities (Fisher’s C = 43.132, p = 0.261). The revealed cascade strength, particularly the robust pathway from turbidity through phosphorus to phytoplankton richness, demonstrates how physical perturbations can propagate through multiple ecological compartments with varying degrees of signal attenuation. The pronounced turbidity to TP pathway (std. coef. = 0.717, p < 0.001) constitutes the strongest link in our model, substantiating the fundamental importance of sediment resuspension in shallow subtropical systems. Phosphorus buffer mechanisms are intensified by increasing turbulence [72], which is further amplified by the distinct hydrological regime of the Yangtze river–lake system, where pronounced water level fluctuations enhance physical–chemical coupling processes.

Several counter-intuitive relationships revealed by our SEM warrant particular attention, as they challenge conventional understanding of nutrient–biology relationships in shallow lake systems. The contrasting effect of TP was positive for phytoplankton richness (std. coef. = 0.339, p < 0.001) but negative for benthic richness (std. coef. = −0.194, p < 0.01). This divergence creates what we term a “phosphorus paradox” in shallow lake management: interventions targeting phosphorus reduction to improve benthic conditions may simultaneously reduce phytoplankton diversity [73]. The positive TP and TN relationship (std. coef. = 0.212, p < 0.01) further contradicts the nitrogen–phosphorus decoupling frequently reported in eutrophic lakes [74]. A 37-year whole-lake experiment demonstrated that despite severe nitrogen limitation created by decreasing nitrogen inputs while maintaining phosphorus additions, eutrophication persisted [74]. Our contrasting finding suggests that in the middle-lower Yangtze system, common sources or synchronous release mechanisms likely drive both nutrients simultaneously, such as sediment resuspension events (turbidity to TP pathway), creating an coupling that transcends their biogeochemically distinct cycles.

The relationship between phytoplankton richness and benthic richness (std. coef. = 0.232, p < 0.01) revealed by our SEM presents an intriguing divergence from the widely established paradigm in aquatic ecology. While we observed an overall positive association between these communities, recent evidence from partial least squares regression studies indicates that this relationship is highly dependent on phytoplankton taxonomic composition rather than simply richness [75]. In Nanwan Reservoir, excessive growth of certain phytoplankton groups (particularly Cyanophyta) and associated organisms (crustaceans and protozoans) was detrimental to macroinvertebrate development, while the presence of Bacillariophyta (diatoms), Chrysophyta, and Euglenophyta positively influenced benthic assemblages [75]. In the middle-lower Yangtze river–lake systems, the positive coefficient between phytoplankton and benthic richness may reflect the predominance of beneficial phytoplankton groups (particularly diatoms) that serve as high-quality food resources for benthic communities [76,77]. The system-specific conditions of the middle-lower Yangtze lakes, particularly their high turbidity and shallow nature, likely enhance the importance of settling phytoplankton detritus as a food resource for benthic organisms.

4.4. Implications for Sustainable Management

The cascade mechanisms identified in this study, particularly the turbidity–phosphorus–biology pathway and the “phosphorus paradox”, have profound implications for achieving Sustainable Development Goals (SDGs) in the Yangtze River Basin. Our structural equation model reveals that total phosphorus exhibited contrasting effects on phytoplankton richness (positive, standardized coefficient = 0.339) versus benthic richness (negative, standardized coefficient = −0.194), suggesting that simplistic nutrient reduction strategies may inadvertently compromise biodiversity objectives. This complexity aligns with recent findings that accounting for interactions between different SDGs is essential for effective water pollution control in China, as interventions targeting one goal may inadvertently impede progress toward others [37]. Despite intensive management efforts over the past decade, lake ecosystems in the Yangtze Basin continue to exhibit persistent eutrophication, with water quality improvement lagging significantly behind rivers [34]. Moreover, the strong turbidity-to-phosphorus pathway (standardized coefficient = 0.717) identified here indicates that physical processes governing sediment resuspension must be addressed alongside nutrient loading reduction. This finding is particularly relevant given that benthic–pelagic coupling regulates essential ecosystem functions including nutrient cycling, primary production, and food web dynamics in shallow aquatic systems [8]. Therefore, sustainable management requires coordinated strategies addressing water quality (SDG 6), aquatic biodiversity (SDG 14), and terrestrial ecosystems (SDG 15) simultaneously [34].

Some priority management domains emerge from our cascade framework, each requiring specific interventions informed by the quantified relationships in our structural equation model. In terms of the Three Gorges Dam (TGD), adaptive operations must account for downstream ecological consequences propagating through altered hydrological and sediment regimes. Our finding that water temperature and turbidity significantly regulate nutrient dynamics and community composition provides a quantitative basis for ecological flow protocols. The TGD has fundamentally altered downstream sediment transport, with reservoir sedimentation reducing sediment loads to downstream reaches and triggering substantial riverbed erosion [78]. Dam operation has indirectly driven temporal changes in riparian soil nutrients through water level fluctuations, with elevation, soil layer, and flood cycle explaining 56.64% of changes in soil properties [79]. Recent assessments demonstrate that TGD construction has significantly altered hydrological connectivity between the Yangtze River and its connected floodplain lakes, particularly affecting “off-stream” floodplains during receding periods, with the connectivity function decreasing substantially due to dam operations [80]. Large-scale ecological programs implemented alongside TGD operation, including terrestrial ecosystem rehabilitation and water pollution prevention, have demonstrated measurable improvements, with 2118.47 km² of sloping cropland returned to forest or grassland and soil erosion curbed in 2196 km² areas [81]. Our cascade mechanism framework suggests three operational strategies: (1) seasonal flow pulse management maintaining turbidity within thresholds preventing excessive sediment resuspension and phosphorus mobilization; (2) thermal regime maintenance through strategic water release from different reservoir depths to support native species while limiting thermophilic cyanobacteria proliferation; and (3) coordinated regulation with downstream connected lakes to prevent extreme water level fluctuations disrupting benthic–pelagic coupling.

In terms of the agricultural non-point source (NPS) pollution, the control can be informed by the mechanistic understanding revealed by our cascade framework. Our finding that total nitrogen (TN) concentrations were significantly elevated in rivers compared to lakes suggests that rivers may function as rapid conduits for nitrogen transport (e.g., agricultural sources), aligning with evidence that diffuse sources account for 72% of total dissolved nitrogen inputs to Chinese rivers [82]. The strong turbidity-to-phosphorus pathway (standardized coefficient = 0.717) identified in our structural equation model provides critical mechanistic insight into how agricultural sediment inputs drive phosphorus dynamics in river–lake systems. This relationship indicates that phosphorus mobilization is tightly coupled to sediment resuspension processes, which may be exacerbated by agricultural erosion. Laboratory simulations demonstrate that sustained sediment disturbance enhances dissolved phosphorus concentrations in overlying water, with legacy phosphorus accumulated in lake sediments representing a large nutrient pool that can sustain eutrophication even after external loading reduction [83,84]. The “phosphorus paradox” revealed by our model, wherein total phosphorus positively affects phytoplankton richness but negatively affects benthic richness, challenges simplistic nutrient reduction strategies and demands nuanced management approaches that consider differential responses across biological compartments. Thus, the control need the necessitating nuanced management interventions such as strategically positioned retention structures that can reduce total nitrogen and phosphorus transport for water bodies (e.g., riparian wetlands) [85].

Achieving sustainable river–lake management requires an integrated governance framework operating at watershed scales. China’s recent policy evolution toward “prioritizing ecological conservation and green development” in the Yangtze Basin provides an institutional foundation, with the relevant laws of the Yangtze River establishing a national coordination mechanism marking a significant shift toward ecosystem-based management [36]. Based on our cascade mechanism framework, we propose five policy recommendations explicitly linked to SDG achievement: (1) Develop ecosystem-based TGD operation protocols maintaining hydrological variability within ranges supporting native biodiversity while minimizing sediment resuspension, particularly avoiding prolonged high-turbidity discharge during ecologically sensitive periods given the strong turbidity-to-TP pathway (standardized coefficient = 0.717; SDG 15.1, 15.5); (2) Implement integrated nutrient management strategies and establish watershed-level targets based on lake-specific thresholds rather than uniform standards simultaneously addressing point sources, agricultural NPS, and internal loading (SDG 6.3, 6.6); (3) Prioritize restoration of small wetlands and buffer zones as cost-effective nature-based solutions, targeting high-nutrient-input agricultural regions where small wetlands have experienced greatest losses (SDG 14.1, 15.8); (4) Strengthen cross-jurisdictional collaboration and science-policy integration mechanisms, establishing regular dialogues and accessible decision support tools that translate research findings into management actions (SDG 17.16, 17.17); (5) Integrate climate change adaptation through diversified approaches including macrophyte restoration, wetland conservation, and hydrological regime maintenance, developing early warning systems using the cascade framework to anticipate propagating impacts (SDG 13.1, 13.3). The middle and lower Yangtze River Basin stands at a critical juncture where decades of degradation meet renewed commitment to restoration. Our quantified cascade mechanisms provide a scientific foundation for designing interventions that harness positive feedbacks, interrupt negative feedback loops, and steer river–lake systems toward sustainable states simultaneously supporting biodiversity conservation, water security, and human well-being.

5. Conclusions

This study revealed significant spatial heterogeneity in environmental parameters across river–lake systems in the middle and lower reaches of the Yangtze River, with lakes showing higher concentrations of BOD₅, chlorophyll-a, CODMn, LOI, STN, TOC, and turbidity, while rivers exhibited elevated total nitrogen levels. Environmental variables displayed distinct seasonal patterns reflecting differences in hydrodynamic processes and sedimentation rates. Through multivariate analyses and structural equation modeling, we identified water temperature, organic matter, and nutrients (particularly phosphorus) as key drivers shaping biological communities. Our findings uncovered a “phosphorus paradox” wherein total phosphorus exhibited contrasting effects on different biological components, positive for phytoplankton richness but negative for benthic richness, highlighting the complex nature of nutrient–biology interactions in these shallow ecosystems. The positive relationship between phytoplankton and benthic macroinvertebrate richness provided evidence for benthic–pelagic coupling, with phytoplankton community composition influencing benthic diversity through sedimentation processes. The findings are particularly relevant for sustainable governance of the Yangtze River Basin, where integrated approaches addressing dam operations, agricultural pollution, and internal nutrient cycling are necessary to achieve water quality and biodiversity objectives aligned with national river protection policies and international sustainable development goals.