The LRE is located in the Liaodong Bay, and is the northernmost estuary in China (Figure 1). The region is characterized by a semi-humid temperate monsoon climate, with an average annual precipitation of 611.6 mm, approximately 70~80% of which occurs between July and September. The LRE features a distinctive funnel-shaped bay formed by the combined influence of tidal action and river runoff. The river runoff into the LRE is greatly regulated by the Panshan gate, situated approximately 45 km upstream. The gate is typically closed from January to July to store water for agricultural and industrial purposes, resulting in low runoff into the estuary. In contrast, it opens during the remaining time and allows normal to high runoff discharge (Supplementary Text S1) (Ye et al., 2013; Hu et al., 2023).

Two cruise surveys were conducted in June and November 2022, both during neap-tide periods, but with marked differences in inflowing runoffs due to gate operation. June experienced low runoff and represented the summer season, whereas November exhibited normal runoff conditions and characterized a near-winter environment before the onset of ice formation. Runoff regulation and seasonal variation jointly led to distinct environmental differences between the two observation periods, including water temperature (T), dissolved oxygen (DO), and salt. Thirteen sampling sites were deployed along the river-to-marine salt gradient (Figure 1), representing three hydro-sedimentary environments as informed by previous studies (Zhou et al., 2022; Douglas et al., 2025) and our laboratory physicochemical analyses (detailed sampling conditions are provided in Supplementary Text S1). Specifically, the estuarine sandy mouth area (L1, L2, L3) was chosen to capture the impact of intense terrestrial sediment input and relatively low salt levels. The muddy area (L4, L5, L6, L7, L12, L13) was identified by its fine-grained sediments and enrichment of organic carbon. Finally, the outer estuary area (L8, L9, L10, L11) represented a marine-dominated environment with higher primary productivity and salt levels. These areas can adequately capture the spatial heterogeneity of sediment properties, the salt gradient, and the varying influence of terrestrial P input. Bottom water was collected approximately 0.5 m above the sediment using a Niskin bottle (Hydro-Bios Apparatebau GmbH, Germany). Upon arrival at the laboratory, the samples were filtered through a 0.45μm Teflon membrane. The filtrates were then stored at 4 °C for subsequent analysis of SRP, total dissolved P (TDP), dissolved inorganic carbon (DIC), and dissolved organic carbon (DOC) concentrations. T, pH, oxidation-reduction potential (ORP), DO, and salt of the bottom water were measured in situ using a multi-parameter water quality analyzer (Multi 3630 IDS, WTW, Weilheim, Germany). Surface sediment (approximately the top 0~20 cm) was collected at each sampling site using a grab sampler (Hydro-Bios Apparatebau GmbH, Germany) and immediately transported to the laboratory on ice. The sediment samples were freeze-dried, passed through a 2-mm sieve in the laboratory to remove gravel, plant materials, and other debris, and stored in a desiccator for subsequent analyses of metal concentrations, grain‐size distributions, and P fractionations, and for adsorption experiments.

The concentration of SRP in the filtered water samples was measured by the molybdenum blue method (Murphy and Riley, 1962). TDP was converted to SRP through high-temperature digestion with potassium persulfate and then measured using a UV spectrophotometer (Shimadzu UV-3900 UV-VIS Spectrophotometer, Tokyo, Japan). The concentrations of DIC and DOC in the filtered water were measured using a total organic carbon (TOC) analyzer (TOC-L, Shimadzu, Japan). The pretreated sediment samples were digested by HNO₃-HF-HClO₄ at 180 °C, and the Na (sodium), Mg (magnesium), Ca (calcium), Al (aluminum), and Fe (iron) concentrations of the digested samples were determined using an inductively coupled plasma optical emission spectrometer (ICP-OES, Agilent 5110, Santa Clara, USA) (Liu et al., 2017). The organic matter (OM) was measured using the loss on ignition (LOI) method after calcination in a muffle furnace at 550 °C for 5 h (Dias et al., 2010). The particle size of the sediment samples was determined using a laser diffraction particle size analyzer (S3500, micro trac MRB, Montgomery, USA), and the percentages of clay (<2 µm), silt (2-50 µm), and sand (>50 µm) particles were calculated. The concentrations of different P forms in sediment were determined using the five-step method (Rydin, 2000) (Supplementary Figure S1), which involved the sequential extraction of NH₄Cl-P, BD-P, NaOH-rP, NaOH-nrP, and HCl-P. TP in the sediment was determined using the Standards, Measurements, and Testing (SMT) protocol(European Commission), and Res-P was calculated as the difference between TP and the sum of other extracted P forms (Ruban et al., 2001; Pardo et al., 2004; Shan et al., 2016; Zhao et al., 2019; Zhuo et al., 2023). All experiments were performed in triplicate to ensure analytical precision, with an average coefficient of variance (C_V) consistently below 10%. The accuracy of P forms in water was verified using nationally certified phosphate reference materials (GSB04-2835-2011), while the accuracy of TP and metal concentrations in sediment was verified using marine sediment reference materials (GBW-07333). Recoveries of analyzed elements were within the range of 85–107%, and relative standard deviations were below 10% (Supplementary Table S1). Procedural blanks were analyzed in parallel and were consistently below the detection limits.

Thermodynamic adsorption experiments of the sediments were carried out to elucidate the P adsorption capacity in LRE sediments. A series of KH₂PO₄ solutions with varying P concentrations of 0, 0.02, 0.04, 0.05, 0.1, 0.5, 1, 3, 5, 10, and 30 mg/L were prepared using deionized water. The pH of all solutions was adjusted to 8, and the salt was modified to match field conditions (Supplementary Text S1). Salt was adjusted using artificial seawater, prepared by dissolving commercial sea salt in deionized water at a concentration of 1 g/L to obtain a salt of 1‰. 0.5 g of pre-treated surface sediment samples (L1~L12) were mixed with 30 mL of respective KH₂PO₄ solutions and incubated in a thermostatic shaker at 25 °C for 24 hours. Following incubation, the suspensions were centrifuged and filtered through a 0.45 µm membrane. The SRP concentration of supernatants was determined using a Skalar Continuous-Flow Analyzer (SKALAR San++, Skalar Co., Netherlands).

The amount of P adsorbed by the sediment at time t (Q_t-abs) was calculated with Equation 1 (Liao et al., 2020):

where Q_t-abs is the amount of P adsorbed by sediment at time t in the adsorption kinetics experiments, or the equilibrium adsorption amount of P at 24h in the adsorption isotherm experiment, mg/kg; C₀ is the initial P concentration in the supernatant, mg/L; C_t-abs is the P concentration in the supernatant at time t in the adsorption kinetics experiments, or the equilibrium concentration of P in the supernatant at 24h in the adsorption isotherm experiment, mg/L; V is the volume of supernatant added, mL; and W is the weight of the added sediment, g.

Adsorption isotherms were plotted with C_t-abs (x-variable) and Q_t-abs (y-variable) data and fitted using the modified Langmuir model and the Freundlich crossover-type model.

The modified Langmuir equation is expressed as Equations 2, 3:

Where Q_max is the theoretical maximum P adsorption capacity of sediment at equilibrium, mg/kg; K_f is the equilibrium constant of P adsorption between the sediment and the water body, L/mg; NAP is the original P content adsorbed on the particles of the sediment, mg/kg; EPC₀ is the equilibrium concentration of P in the water body at Q = 0, mg/L.

The Freundlich crossover-type equation is expressed as Equations 4–7 (Pan et al., 2013):

Where K_d is the adsorption coefficient associated with the adsorption capacity of particles; β are empirical constants; C is the SRP concentration in the bottom water, mg/L; λ, EPC_sat, and δ are all coefficients of determination.

The equilibrium concentration of P was determined as the X-axis intercept of the adsorption isotherm, and the coefficients of determination were calculated by the Levenberg-Marquardt nonlinear least squares algorithm. The sediment was identified as a P source when EPC₀ > SRP or δ< 0, and as a P sink when EPC₀< SRP or δ > 0 (Wang et al., 2009; Pan et al., 2013). The Q_max obtained from the Langmuir model was used in Equation 8 to calculate the degree of P saturation (DPS), which serves as a quantitative indicator of the P adsorption capacity of sediment (Pöthig et al., 2010):

Where TP is the total P content in sediment, mg/kg.

In addition, adsorption kinetics experiments were conducted to investigate the P adsorption rate in sediments. 0.5 g of pretreated surface sediment samples (L1~L12) were mixed with 30 ml of a 10 mg/L KH2PO4 solution and incubated in a thermostatic shaker. Supernatants were collected at sequential intervals (0 min, 5 min, 10 min, 30 min, 1h, 2h, 8h, 16h, and 24h) to record the concentration of SRP. The P adsorption kinetics curves were then plotted with t (x-variable) and Qt-abs (y-variable) data and fitted using the Lagergren, Elovich, and ExpAssoc kinetic equations, as detailed in Supplementary Text S2. Meanwhile, the supernatant samples that reached adsorption equilibrium after 24h satisfy the criteria for single-point equilibrium adsorption data. The equilibrium P concentration in the supernatant after 24 h (C24h-abs) was therefore substituted into Equation 9 to calculate the P sorption index (PSI) (Bache and Williams, 1971; Bolster et al., 2020):

Where X is the amount of P adsorbed by 100 g of sediment, converted from Q_24h-abs, mg P/100 g sediment.

The P eutrophication risk index (ERI) is calculated using Equation 10 to quantitatively assess the risk posed by P in sediment (Xu et al., 2020):

Data analysis and visualization were performed using Origin Pro 2021, SPSS 26.0, and ArcGIS 10.2. The relationships among sediment properties, sediment P forms, and P adsorption parameters were analyzed using the Spearman correlation coefficient, with significances labelled at p< 0.05 and p< 0.01 levels. The adsorption experimental data were model-fitted using MATLAB. Seasonal variations (summer versus winter) in bottom water and sediment properties were tested via non-parametric Kruskal–Wallis.

To identify the key drivers of sediment P forms and P adsorption, a partial least squares path modelling (PLS-PM) was employed. Based on previous research and the results of correlation analysis in this study, observed variables associated with P adsorption capacity were selected and grouped into four latent variables for the model: bottom water properties (salt and SRP), sediment properties (Fe, Al, OM, and the proportion of fine particles), P forms (NaOH-rP, NaOH-nrP, and HCl-P), and P adsorption capacity (Q_max, the equilibrium adsorption capacity a₂, and adsorption rate a₃) (Zhang and Huang, 2011; Li et al., 2016). The following hypothetical paths were established: (1) bottom water properties, sediment properties, and P forms directly affect the P adsorption capacity; (2) bottom water and sediment properties directly influence P forms; and (3) bottom water properties directly affect sediment properties. The Goodness of fit (GOF) index (0.647 in summer and 0.675 in winter, section 3.5) confirmed the adequate explanatory power for both measurement and structural model (Wetzels et al., 2009). Path coefficients were calculated after 1000 bootstraps, with significance levels indicated by p< 0.05, p< 0.01, and p< 0.001. The construction and validation of the PLS-PM model were conducted using the “plspm” package in R software.

The physicochemical properties of the bottom water in the LRE showed distinct spatial and temporal variations under different runoff input conditions (Table 1). During the summer (June) and winter (November) observations, the average DO concentrations across all sites were 8.06 ± 0.68 mg/L and 11.92 ± 0.45 mg/L, respectively. The average ORP was 117.21 ± 34.33 mV in summer and 201.99 ± 32.31 mV in winter, indicating that oxidizing capacity was relatively lower in summer than in winter. Under low runoff input conditions in summer, the salt of the bottom water maintained consistently high values (19.8 to 25.2‰) with minimal spatial variation; while the winter higher runoff resulted in a marked salt gradient from the sandy mouth area (0.7‰) to the outer estuary (22.6‰), demonstrating pronounced terrestrial runoff influence on estuary salt distribution. The DIC concentrations in summer and winter observations were 33.14 ± 2.34mg/L and 50.70 ± 11.87mg/L, respectively.

The sediment OM concentrations averaged 3.46 ± 0.01%, with the highest values observed in the muddy areas. The average concentrations of Fe, Al, Ca, and Mg across all sediment samples during the two surveys were 25022 ± 3949mg/kg, 24637 ± 3899 mg/kg, 15600 ± 5633 mg/kg, and 39008 ± 7135mg/kg, respectively. The outer estuarine sediments exhibited the highest Ca concentration (Supplementary Table S2), indicating a greater propensity for Ca precipitation and accumulation in this zone (Zhang et al., 2022b). Sediment composition analysis (Supplementary Figure S2) revealed that the sediments in LRE were predominantly composed of silt, with fine-grained particles (silt and clay) comprising 75.45% of the total sediment.

Figure 2 presents the concentrations of SRP and TDP in the bottom water. The average concentrations in summer were 0.053 ± 0.01 mg/L for SRP and 0.111 ± 0.02 mg/L for TDP, while those in winter were 0.128 ± 0.05 mg/L and 0.152 ± 0.05 mg/L, respectively. Accordingly, dissolved organic P (DOP), calculated as the difference between TDP and SRP, averaged 0.058 ± 0.02 mg/L in summer and 0.025 ± 0.01 mg/L in winter. Notably, approximately 70% of our sampling sites in summer and all sites in winter exceeded the Class IV standard for seawater quality in China (SRP: 0.045 mg/L) (GB 3097-1997).

Both SRP and TDP showed significantly higher levels in winter than in summer (p<0.01, Figure 2C), likely due to greater P inflowing fluxes and intensified sediment resuspension in winter. This is consistent with previous studies indicating that riverine input maintains relatively high SRP concentrations, and sediment resuspension promotes P release into the bottom water (Walve et al., 2018; Zhang et al., 2022b; Bi et al., 2011; Yang et al., 2024). The DOP concentrations exhibited reverse trends with higher values in summer than in winter. This may be explained by the enhanced P utilization by algae in summer (Bronk and Ward, 2005; Lin et al., 2024), while in winter, DOP mineralization by alkaline phosphatase under a P-limited environment likely leads to its depletion (Deborde et al., 2007; Liu et al., 2009; Li et al., 2017). Spatially, significant variations in SRP and TDP concentrations were observed among sampling sites in winter observations (p<0.05), with concentrations decreasing from the sandy mouth to the outer estuary zone.

Furthermore, the Redfield ratio (C:P = 106:1) is widely used in marine systems as a benchmark for assessing nutrient limitation status. A DOC: DOP ratio exceeding 250 is generally indicative of DOM with a predominantly terrestrial origin (Moore et al., 2013; Karl and Björkman, 2015; Letscher and Moore, 2015). In the LRE, DOC: DOP ratios ranged from 45~314 in summer and from 113~1325 in winter. More than 85% of winter sampling sites exceeded the threshold value of 250, suggesting a predominant role of terrestrial DOM input and a higher likelihood of P-limited conditions during this period.

The TP concentrations of all sediment samples are displayed in Figure 3. The average TP concentrations were 495.26 ± 69.33 mg/kg in summer and 399.62 ± 39.29 mg/kg in winter, respectively, exhibiting a significant difference (p<0.01) between the two observations (Figure 3C). The lower TP concentrations observed in winter may be attributed to enhanced sediment resuspension, which facilitates the release of labile P fractions from the sediment into the bottom water (Søndergaard et al., 2003). No significant spatial variation in TP was detected during summer (p > 0.05), while statistically significant spatial differences were observed during winter observation (p<0.05). In both seasons, the highest TP concentrations occurred in the muddy area, likely due to the high proportion of fine particles, with silt and clay composing on average 79.54%, which provided greater capacity for P adsorption and retention (Wang et al., 2021).

Figure 4 further shows the concentrations of various P forms in sediments. In summer, the average proportions of P were HCl-P (38.19%), Res-P (27.36%), NaOH-rP (16.54%), NaOH-nrP (15.71%), BD-P (1.82%), and NH₄Cl-P (0.38%). In winter, the proportions were HCl-P (50.40%), NaOH-nrP (21.75%), NaOH-rP (18.90%), Res-P (5.33%), BD-P (3.25%), and NH₄Cl-P (0.37%). HCl-P was the dominant form of P in both seasons, which aligned with the findings in Jiaozhou Bay, the Yangtze River estuary, and the Yellow River estuary (Cao et al., 2017; Bai et al., 2020b; Zhang et al., 2022b). It is generally believed that debris and biological material transported by runoff erosion and weathering are the primary sources of HCl-P; therefore, higher terrestrial inputs in winter likely contributed to the elevated HCl-P proportions (Ni et al., 2020).

BD-P and NaOH-nrP are regarded as active P forms and potential sources for algal growth. The average concentrations were 8.53 ± 5.89mg/kg for BD-P, 79.15 ± 24.51mg/kg for NaOH-nrP in summer, and 12.94 ± 3.21mg/kg and 87.63 ± 17.83mg/kg in winter. Previous studies have reported that under anoxic conditions in summer, iron-reducing bacteria converted BD-P to Fe²⁺ and soluble P, leading to a decline in BD-P level (Kraal et al., 2015). Meanwhile, intensified biological activity in summer would enhance the mineralization of NaOH-nrP, whereas decomposition of plant and animal remains in winter resulted in elevated NaOH-nrP concentrations (Knudsen-Leerbeck et al., 2017; Pan et al., 2017). Spatially, significant spatial variation of NaOH-rP was observed in summer (p< 0.01), while other P fractions showed no significant spatial differences in either season (p> 0.05). The highest NaOH-rP concentrations were observed in the outer estuary zone during summer (average 102.08 mg/kg), likely resulting from elevated salt that promoted the formation of metal oxides (Al–O–Si complexes) and thereby enhanced P adsorption capacity (Zhao et al., 2019).

Bioavailable P (BAP) serves as an indicator of P release potential from sediment, as it can be transformed into SRP, which is readily available to aquatic organisms and potentially contributes to eutrophication. BAP is generally recognized as the sum of NH₄Cl-P, BD-P, NaOH-rP, and 60% of NaOH-nrP (Rydin, 2000), which averaged 140.48 ± 37.22 mg/kg in summer and 142.80 ± 23.63 mg/kg in winter. BAP exhibited significant spatial variation in summer (p< 0.01) (Supplementary Figure S3), with the highest concentration (159.90 mg/kg) observed in the outer estuary, suggesting a strong potential for P release in this region. In contrast, no significant spatial differences of BAP were observed during winter observation (p > 0.05).

The fitting parameters of P adsorption thermodynamics from the Langmuir model and the Freundlich crossover-type model are presented in Supplementary Table S3 and Supplementary Table S4, respectively. The adsorption isotherms for P in sediment are shown in Supplementary Figure S4. The thermodynamic process of P adsorption was well described by both the modified Langmuir model and the Freundlich crossover-type model, as evidenced by high correlation coefficients (R² > 0.9). The maximum adsorption capacity (Q_max) and native adsorbed P (NAP) are key indicators of sediment P adsorption capacity and nutrient status (Cao et al., 2017). Across all sampling sites, Q_max averaged 327.13 ± 118.21 mg/kg and 472.15 ± 182.66 mg/kg in summer and winter, while NAP were 3.42 ± 1.73mg/kg and 3.79 ± 1.96 mg/kg, respectively. The higher NAP and Q_max observed in winter indicated enhanced P adsorption capacity and P accumulation, likely driven by increased terrestrial runoff during this period that transported greater loads of P, sediment particles, and Fe-Al reactants into the estuary. Additionally, intense sediment resuspension during winter can promote the disaggregation of aggregates and expose more mineral surfaces, thereby increasing the availability of binding sites and enhancing the Q_max (Wang et al., 2013). The EPC₀, representing the equilibrium exchange capacity between sediment and bottom water, averaged 0.22 ± 0.10 in summer and 0.24 ± 0.11 mg/L in winter. Approximately 90% of sediment samples (21 out of 24) exhibited δ< 0 and EPC₀ > SRP (Supplementary Tables S3, S4), indicating that the sediments in the LRE were predominantly in a state of P release. This finding is consistent with those reported for Jiaozhou Bay and Laizhou Bay (Zhang et al., 2022a, b). Spatially, EPC₀ decreased from the muddy area to the outer estuary (Figure 5B) in winter, implying a stronger external P buffering capacity and lower release potential in the outer estuary (Zhang and Huang, 2011).

The amount of P adsorbed by the sediment at time t (Q_t-abs in Equation 1) was determined based on the sediment adsorption kinetics experiments (section 2.3), as shown in Figure 6. The Q_t-abs trend revealed a rapid adsorption stage within the first 2 hours, achieving approximately 75%~95% of the total adsorption capacity. This was followed by a slower adsorption stage, and the system gradually approached equilibrium over the period from 8 to 24 hours. The fitting parameters derived from the Lagergren first-order and second-order, Elovich, and ExpAssoc models are shown in Supplementary Tables S5 and S6. Among these, the Elovich and second-order kinetics models provided the best fit to the experimental data (R² > 0.9). Both models have been previously reported as more suitable for describing chemical adsorption where adsorption dominates over desorption (Xia et al., 2020). The equilibrium adsorption capacity (a₂ in the second-order kinetics model, Supplementary Text S2) and adsorption rate (a₃ in the Elovich model, Supplementary Text S2) averaged 105.01 ± 17.36mg/kg and 93.52 ± 16.83mg/(kg·h) in summer, and 112.64 ± 24.19mg/kg and 100.55 ± 20.58mg/(kg·h) in winter, respectively. These results indicated a higher adsorption capacity in winter, favoring P fixation in sediment.

The degree of P saturation (DPS in Equation 8) is a key parameter for evaluating the P adsorption state in sediments, with higher values indicating adsorption closer to saturation. The DPS averaged 61.81% ± 0.08 in summer and 48.18% ± 0.11 in winter, indicating greater P adsorption saturation in summer sediment. The P sorption index (PSI in Equation 9), reflecting sediment P adsorption capacity, was 4.74 ± 0.84 in summer and 5.11 ± 1.22 in winter. Following previous studies, the ratio of DPS to PSI was used to calculate the P eutrophication risk index (ERI), where ERI< 10% indicates low P potential release risk, 10–20% moderate risk, and 20–25% high risk (Jiang et al., 2023). Summer ERI values ranged from 9.30% to 23.60%, with 83.33% of samples classified as moderate risk. In contrast, winter ERI values ranged from 5.90% to 19.95%, with 66.67% of samples classified as low risk. Overall, sediments exhibited a higher potential for P release in summer. Elevated salt during summer may increase the concentration of electrolyte anions (such as Cl^-), which compete for active adsorption sites and weaken electrostatic outer-sphere interactions between phosphate and sediment (Wang et al., 2022). To illustrate the underlying processes, a schematic diagram was developed to summarize the mechanisms governing P adsorption and desorption in the LRE during summer and winter (Figure 7).

To identify the key factors influencing P adsorption behaviors in sediment, correlation analyses were conducted among sediment physicochemical properties, P forms, and adsorption parameters (Figure 8). In summer, the concentrations of Al, Fe, and OM showed significant negative correlations with BD-P (p< 0.01), while showing significant positive correlations with NaOH-nrP (p< 0.01). BD-P represents a redox-sensitive and more reactive P form associated with amorphous Fe (III) oxides (Jensen and Thamdrup, 1993). Under anaerobic conditions in summer, the reduction and dissolution of Fe (III) oxides reduce the availability of binding sites for P, thereby increasing the negative surface charge of solid phases and reactivating BD-P (Jensen et al., 1992; Rozan et al., 2002; Kraal et al., 2015). Moreover, high OM concentrations in sediment compete for binding sites on Fe/Al oxides, promoting the formation of OM-oxide aggregates. This weakened the adsorption strength between P and Fe/Al oxides, and ultimately led to a reduction in BD-P concentrations (Kaiserli et al., 2002; Zhang et al., 2022b). Additionally, OM acts as an important carrier of NaOH-nrP, which is released upon OM mineralization (Zhang et al., 2008). In winter, Fe exhibits a significant positive correlation with NaOH-rP (p< 0.01). Unlike BD-P, NaOH-rP predominantly represents P bound to Al oxides and crystalline Fe oxides, and reflects a more stable P form (Hupfer et al., 1995; Lukkari et al., 2007). The low-temperature oxidizing environment in winter favors the stability and accumulation of the crystallization of Fe and Al oxides, providing abundant adsorption sites for phosphates and thereby facilitating the formation of NaOH-rP in sediment (Jensen et al., 1992; Roden and Zachara, 1996; Hyacinthe and Van Cappellen, 2004).

In summer, the concentrations of HCl-P and NaOH-nrP showed significant positive correlations with silt proportion (p< 0.01), but significant negative correlations with sand proportion (p< 0.01). Silt has a larger specific surface area and a higher proportion of clay minerals and organic matter, which provide more adsorption sites and stronger P fixation capacity (Ruttenberg, 1992; Kaiser and Guggenberger, 2003). Moreover, complexes formed between organic matter and clay minerals can inhibit P release, thereby enhancing the stability of HCl-P and NaOH-nrP. In contrast, sand has a smaller specific surface area and fewer adsorption sites, leading to a diminished P fixation capacity (Froelich, 1988). EPC₀, a₂, and a₃ (Supplementary Text S2) are key adsorption parameters representing the P release potential, adsorption capacity, and adsorption rate of P, respectively. In summer, EPC₀ showed a significantly negative correlation with the concentrations of HCl-P and NaOH-nrP in sediments (p< 0.05), whereas a₂ and a₃ exhibited significant positive correlations with NaOH-nP and NaOH-nrP (p< 0.01) (Figure 8A). These P forms were considered to have an important influence on the P adsorption capacity and adsorption rate in the LRE sediments.

The PLS-PM results further reveal the directional causal pathways among various latent variables during summer and winter, including bottom water properties (salt and SRP), sediment properties (Fe, Al, OM, the proportion of clay and silt), sediment P forms (NaOH-rP, NaOH-nrP, and HCl-P), and P adsorption capacity (Q_max, a₂, and a₃) (Figure 9). During summer observation (Figure 9A), sediment properties showed a significant positive effect on P forms in sediment with a path coefficient of 0.894 (p< 0.001). None of the variables showed a significant effect on P adsorption capacity during summer, while sediment properties and sediment P forms exhibited relatively larger positive direct and total effects. This finding, aligned with correlation analysis, suggested that sediment properties and sediment P forms served as key regulatory factors of P adsorption.

In winter, bottom water properties exerted an overall significant negative influence on P adsorption capacity (path coefficient = -0.651, p< 0.01) (Figure 9B). Within this construct, salt showed a strong negative loading (-0.917) (Supplementary Figure S5), resulting in a positive indirect influence on P adsorption capacity. Specifically, reduced salt levels lead to a decrease in divalent cations(Ca²⁺ and Mg²⁺), which weakens the cation bridging effect and thereby reduces P adsorption capacity (Liu et al., 2002; Wang et al., 2024). Conversely, SRP exerted a negative indirect effect on P adsorption, as elevated SRP concentrations in winter lead to the saturation of adsorption sites (Flower et al., 2016), further inhibiting the P accumulation.

P is widely recognized as a key limiting nutrient in estuarine systems, prompting extensive research dedicated to elucidating its biogeochemical dynamics. Comparative analyses across diverse estuarine systems offer valuable insights into the specific water environmental conditions, sediment properties, and terrestrial inputs of study areas. The DIC content in the bottom water of LRE (41.92 ± 12.26mg/L) was higher than that reported for the Yellow River estuary (25.86 ~ 40.61mg/L) and the Yangtze River estuary (24.79mg/L) (Yan et al., 2011; Liu et al., 2014), suggesting more pronounced terrestrial inputs and carbon accumulation in the LER. Moreover, the average SRP concentration in the LRE (0.090 ± 0.051mg/L) was substantially elevated compared to the Jiaozhou Bay (0.002mg/L) and the Yellow River estuary (0.003mg/L), while remained comparable to levels observed in the Pearl River estuary (0.029mg/L), Green Bay (0.045mg/L), and the Jiujiang River estuary (0.075mg/L) (Xu et al., 2016; Yang et al., 2021; Ke et al., 2022; Zhang et al., 2022b, c). In terms of sediment geochemistry, Fe, Al, and Ca concentrations in the LRE were lower compared with those reported for other estuaries worldwide (Supplementary Table S7), while Mg concentration was relatively higher in the LRE. This pattern may be attributed to the variations in sediment loads and transport dynamics. Elements such as Fe and Al are typically enriched in fine-grained sediment. The annual sediment load of the LRE (6.94 × 10⁶ t yr^-¹) is substantially lower than that of other estuarine systems, such as the Yellow River estuary (3.9 × 10⁸ t yr^-¹), potentially limiting the accumulation of Fe and Al (Yu et al., 2019; Zhang et al., 2024). In addition, the proportion of fine-grained material in the LRE (75.45%) exceeds that of the Yellow River estuary (73.21%) and the Yangtze River estuary (58.74%) (Yan et al., 2011; Wang et al., 2021). While the higher proportion of fine particles provides a larger specific surface area conducive to phosphate adsorption in sediment, the TP levels in sediment are collectively affected by sediment composition, the degree of chemical weathering, and the magnitude of terrestrial P inputs (Hupfer et al., 1995; Hartmann et al., 2014). Consequently, TP levels in the LRE sediment remained generally lower than those reported for estuaries in southern China and the global sediment average (Supplementary Table S8).

The results of this study provide practical implications for P management in estuaries under runoff regulation. Integrated statistical analyses, including PLS-PM model and correlation analysis, indicate that sediment P adsorption processes are jointly governed by aquatic environmental conditions (salt and SRP) and sediment properties (Fe, Al, OM, and the proportion of clay and silt).

With respect to hydrological regulation, the PLS-PM results reveal that elevated SRP levels combined with reduced salt in winter likely constrain P accumulation in the sediment. This suggests that freshwater discharge during winter may enhance the risk of internal P loading. Accordingly, optimizing the timing and magnitude of runoff release (such as avoiding large-volume runoff inputs within short periods and maintaining moderate salt levels) may help alleviate the risk of internal P loading. In contrast, sediment in the summer exhibited a moderate P release risk, representing a critical period for controlling P pollution. Appropriately increasing runoff through gate regulation can shorten water residence time, thereby diluting P concentrations and reducing eutrophication risk. From the perspective of sediment management, Fe, Al, OM, and fine-grained particles were identified as the primary controls on NaOH-rP, NaOH-nrP, and HCl-P. This finding highlights the critical role of sediment transport processes in regulating the estuarine P reservoir. Accordingly, measures such as soil conservation practices can effectively reduce the transport of fine-grained and organic-rich sediment to the estuary. Such measures would limit the accumulation of reactive P fractions (such as NaOH-rP and NaOH-nrP) in estuarine sediment, thereby lowering the risk of internal P release. Overall, these results provide a mechanistic basis for mitigating internal P loading and managing P pollution in gate-controlled estuarine systems.