Meiliang bay is in the northern area of Taihu lake, a cyanobacteria-dominant area. In June and December 2021, ten sites in Meiliang bay were selected to collect water samples with the water depths of .5, 1.5, and 2.0 m, as shown in Figure 1. At site-1 of Meiliang bay, three parallel samples of surface sediments and sediment cores were obtained by columnar sediment sampling. A small parts of surface sediments in each sample were stored in a sterile centrifuge tube on-site with dry ice for bacteria community analysis, and other sediments were brought back to the laboratory within 12 h.

The pH, dissolved oxygen (DO) and water temperature were measured by a water quality meter (Alalis PD320) on site. In the laboratory, Chlorophyll-a (Chla) content in the water sample was determined by using the spectrophotometry method (Editorial board of Water and Wastewater Monitoring and Analysis Methods, 2002). The original water sample and filtered water sample through fiber filter membrane (pore size of .45 μm) were digested with potassium persulfate solution at 120°C, .1 MPa, and then were used to determine total P (TP) and total dissolved P (TDP) through molybdenum blue method, respectively (Editorial board of Water and Wastewater Monitoring and Analysis Methods, 2002). The orthophosphate (PO₄ ^3-) in the filtered water sample was also determined by using the molybdenum blue method. For each of the water samples, the concentration of TP minus TDP was total particulate P (TPP), and the concentration of TDP minus PO₄ ^3- was dissolved organic P (DOP).

Surface sediment samples were air-dried, ground and sieved by screen mesh with a pore size of .15 mm. The SMT protocol (Ruban et al., 1999) was used to measure sediment P forms, which were divided into soil total P (S-TP), inorganic P (IP), organic P (OP), iron-bound P (FeP) and calcium-bound P (CaP).

DGT is an in situ technique for kinetic passive sampling measurement of analytes in sediments or wet soils over the deployment period (Davison and Zhang, 1994; Krom et al., 2002). In this study, the concentrations of labile PO₄ ^3-, Fe²⁺ and S^2- in the sediment cores were simultaneously determined using the DGT technique at depths of -110–10 mm with a vertical resolution of 2.0 mm. The ZrO-Chelex DGT probe was applied for the simultaneous measurements of labile PO₄ ^3- and Fe²⁺, and the AgI DGT probe was applied for the measurement of labile S^2-.

HR-Peeper is an in situ sampling technique for analytes in the pore water based on the theory of diffusion equilibrium of ion concentration (Ding et al., 2010). HR-Peeper technique was used to determine concentrations of dissolve PO₄ ^3- and Fe²⁺ in the pore water at depths of −135–15 mm with a vertical resolution of 4.0 mm. These DGT and HR-Peeper probes were provided by Easysensor Co. In China (www.easysensor.net), and the details for their preparations and assemblies can be seen from previous studies (Xu et al., 2012; Xu et al., 2013).

All of the DGT and HR-Peeper probes were deoxygenated overnight with nitrogen flushing. When the sediments cores at site-1 were delivered to the laboratory, first the HR-Peeper probe was inserted into each of the sediment cores, after 24 h the ZrO-Chelex and AgI DGT probes were also inserted into the same sediment core, and at another 24 h all these probes were carefully retrieved from the sediment core and wiped clean thoroughly using lens wiping paper (Xu et al., 2012).

After being disassembled, the ZrO-Chelex binging gel was sliced at a 2 mm interval using a multi-blade ceramic cutter. Each of the sliced gel pieces was extracted using 1 mol/L HNO₃ for elution of Fe²⁺, and then was extracted using 1 mol/L NaOH for elution of PO₄ ^3-. The concentrations of the extracted PO₄ ^3- and Fe²⁺ were detected using the molybdenum blue and phenanthroline colorimetric methods, based on which labile PO₄ ^3- and Fe²⁺ concentrations in the sediment cores can be calculated (Xu et al., 2013; Wang et al., 2017). The S^2- reacted in the AgI binding gel was measured by the computer imaging densitometry technique, which was used to calculate labile S^2- concentrations in the sediment cores (Ding et al., 2012).

After being disassembled, about 200 μL pore water sample was immediately sampled from each chamber of the HR-Peeper probe, and then the concentrations of PO₄ ^3- and Fe²⁺ in the pore water were determined by using the molybdenum blue and phenanthroline colorimetric methods (Ding et al., 2010; Xu et al., 2013), respectively.

Total DNA was extracted from .5 g freeze-dried surface sediment sample by E.Z.N.A. ® soil DNA spin kit (Omega Bio-tek, Norcross, GA, United.States). The primers set 338F (5’-ACT​CCT​ACG​GGA​GGC​AGC​AG-3’) and 806R (5’-GGACTACHVGGGTWTCTAAT-3’) were used in the PCR amplification for targetting the V3∼V4 region of bacteria 16S ribosomal RNA gene. PCR product was purified and quantified, sequencing was performed on an Illumina MiSeq PE300 platform (Majorbio Bio-Pharm Technology Co., Ltd., Shanghai, China) according to the standard protocols (https://cloud.majorbio.com/).

The diversity of the bacteria community was investigated using the Mothur software, which was displayed as the indices of Shannon, Simpson, ACE, Chao, and Coverage. The heat map of the bacterial community structure was made through the Origin software. The Pearson correlation analysis was performed using the SPSS software.

Based on the concentrations of PO₄ ^3- or Fe²⁺ in the pore water determined by the HR-Peeper technique, the apparent diffusion flux (F _d ) of PO₄ ^3- or Fe²⁺ across the interface between pore water and overlying water (the P-O interface) was estimated using the following equations (Fan et al., 2018). φ was the sediment porosity (Harper et al., 1998), D₀ was the diffusion coefficient of PO₄ ^3- or Fe²⁺ in the water (10^-6 cm²/s) (Li and Gregory, 1974), θ was the sediment tortuosity calculated by φ, C was the solute concentration of PO₄ ^3- or Fe²⁺ measured by the HR-Peeper probe (mg/L), x is the sediment depth (m), ∂ C ∂ X was the concentration gradient of PO₄ ^3- or Fe²⁺ in the pore water core at depths of −20–5 mm (from pore water to overlying water). F d = − φ ⋅ D 0 θ 2 ∂ C ∂ X (1) θ 2 = 1 − I n φ 2 (2)

Water quality parameters including pH, DO, temperature and Chla in Meiliang bay in June and December 2021 are shown in Supplementary Table S1 and Supplementary Figure S1. From June to December, water temperature in the water of Meiliang bay decreased from 26°C to 10°C, and DO contents increased from 6.63 mg/L to 11.25 mg/L, because the low temperature can promote oxygen solubility in the water (Zaker, 2007). The pH values in the water of Meiliang bay varied little between the two periods.

During the sampling period of June 2021, dense blue-green algae were obviously observed on the surface water in Meiliang bay (Supplementary Figure S2). In June 2021, the average content of Chla reached 103.17 μg/L in the surface water, which decreased rapidly as the water depth deepened. In December 2021, Chla contents were stabilized at 22–26 μg/L in the vertical direction of the water.

Concentrations of P forms containing PO₄ ^3-, DOP, TPP and TP in the water of Meiliang bay during the two periods are shown in Figure 2. TP and TPP concentrations in the water in June 2021 (.13–.25 mg/L) were much higher than those in December 2021 (0–.13 mg/L), while concentrations of dissolved P contained PO₄ ^3- and DOP were lower than those in December 2021. TPP was the predominant P form in the water in June 2021 which occupied 74%–84% of TP. Dissolved PO₄ ^3- and DOP were the predominant P forms in the water in December 2021. Similar to the vertical distribution of Chla contents in June 2021, concentrations of TP and TPP in the water in June 2021 also decreased as the water depth deepened.

Concentrations of P forms containing IP and OP, FeP, and CaP in the sediments of Meiliang bay during the two periods were shown in Figure 3. The temporal variations of IP, OP and FeP forms in the sediments were different from those of TP and TPP forms in the water. Both IP, OP and FeP concentrations in the sediments in June 2021 (.23, .05 and .06 mg/g) were lower than those in December 2021 (.31, .11 and .09 mg/g), while CaP concentrations in June 2021 (.29 mg/g) was higher than that in December 2021 (.18 mg/g).

Concentrations of labile PO₄ ^3-, Fe²⁺ and S^2- in the sediment cores at depths of 10∼-104 mm were determined by DGT probes with the vertical resolution of 2.0 mm (Figure 4). Concentrations of labile PO₄ ^3-, Fe²⁺ and S^2- in the sediment cores in June 2021 were .01–.35, .01–1.27 and 0–.06 mg/L, both of which were relatively higher than those in December 2021 (.01–.31, .01–.25 and 0–.06 mg/L).

In June 2021, labile PO₄ ^3-, Fe²⁺ and S^2- in the sediment cores had significant correlations (Supplementary Table S2) and similar vertical distributions, presenting first increases at depths of 0∼-40 mm and then decreases at depths of −40 to −60 mm (Figure 4). In the deeper depth of −60 to −104 mm, labile PO₄ ^3- and Fe²⁺ concentrations in June increased first and then decreased again, but labile S^2- in June increased unsteadily as the sediment depth deepened.

In December 2021, there was only a significant correlation between labile PO₄ ^3- and S^2- in the sediment cores (Supplementary Table S2). With the deepening of sediment depth from 0 mm to 104 mm, labile PO₄ ^3- and S^2- concentrations in December exhibited unsteady increases, while labile Fe²⁺ in December increased limitedly.

In the pore water and overlying water at depths of 15 to −135 mm, PO₄ ^3- and Fe²⁺ concentrations were determined by HR-Peeper probes with a vertical resolution of 5.0 mm (Supplementary Figure S3). Similar to labile Fe²⁺ in the sediment cores, Fe²⁺ concentrations in the pore water in June (.20–2.03 mg/L) were also higher than that in December (.33–1.16 mg/L). Compared with PO₄ ^3- in the pore water in December 2021, PO₄ ^3- concentrations at depths of −20 to −60 mm were higher but those at depths of −80 to −135 mm were lower in June. In the two periods, both PO₄ ^3- and Fe²⁺ concentrations in the pore water were higher than those in the overlying water, suggesting PO₄ ^3- and Fe²⁺ ions can spontaneously diffuse from pore water to overlying water. The diffusion fluxes of PO₄ ^3- and Fe²⁺ from pore water to overlying water in June (.13 and .33 mg/m²·d) were relatively higher than those in December (.08 and .17 mg/m²·d) (Supplementary Figure S4).

The sequence information calculated the coverage percentage, richness estimator (Chao and ACE), and diversity index (Simpson and Shannon) of the samples in Meiliang bay over the two periods, which were listed in Supplementary Table S3. Based on the ACE and Chao indices, higher bacteria community richness in the surface sediments was found in June 2021. Furthermore, the Simpson and Shannon indices revealed that the surface sediments in June 2021 had higher bacteria diversity than that in December 2021. There were detected 47 phyla in June and December 2021 with the relative abundance above .01% in the surface sediments of Meiliang bay. Supplementary Figure S5 showed the top thirty phyla. At the phylum level, the dominant bacterial community comprised Proteobacteria (24.63% in June and 22.63% in December), Chloroflexi (13.28% and 12.12%), Acidobacteriota (10.53% and 16.80%), Actinobacteriota (8.90% and 9.96%), Desulfobacterota (8.00% and 8.61%), Bacteroidota (6.23% and 3.04%), Nitrospirota (5.93% and 6.08%) and Nitrospinota (2.47% and 2.86%), etc.

Proteobacteria are commonly found in freshwater sediments, and sedimentary degradation and metabolism rely mainly on the Proteobacteria phylum (Chaudhry et al., 2012). Furthermore, most of organic P-solubilizing bacteria are predominant in Proteobacteria and Actinobcateriota, which primarily participate in OP mineralization in sediments (Tu et al., 2022). The relative abundance of Proteobacteria and Actinobcateriota in June 2021 (33.53%) was a bit higher than that in December 2021 (32.59%). In addition, the relative abundance of Desulfobacterota was also ahead of the top thirty ranks of phyla in Meiliang bay, suggesting the sediments had a strong ability for sulfate reduction. The relative abundance of Desulfobacterota in December 2021 (8.61%) was higher than that in June 2021 (8.00%).

At the genus level, some typical iron-reducing bacteria (FRB) in the surface sediments of Meiliang bay in the two periods, including Anaeromyxobacter, Geothermobacter, Deferrisoma and Crenothrix (Fan et al., 2018), were also detected in this study. The relative abundances of these typical FRBs are shown in Figure 5. The relative abundance of FRBs in the surface sediments in June 2021 (1.4%) was higher than that in December 2021 (.9%), suggesting the stronger ability for iron oxides reduction in June 2021.

In Taihu Lake, Chla content of 20–40 μg/L had been defined as a threshold of algae blooms, exceeding which was seen as the occurrence of algae bloom (Xu et al., 2015). In this study, the maximum content of Chla in the surface water of Meiliang bay in June 2021 reached 103.17 μg/L, and it decreased rapidly at the water depths of 1.5 m and 2 m. These reflected the serious state of algae bloom in Meiliang bay in June 2021. Furthermore, algae cells were conditioned to concentrate on the surface water body to compete for sunlight and oxygen (Huisman et al., 2005). Interestingly, during the sampling period in June 2021, we observed that the concentrated algae cells weren’t uniformly flat on the surface water, but presented a lot of strips, as shown in Supplementary Figure S2. In December 2021, Chla contents in the water of Meiliang bay were 22–26 μg/L, so no algae bloom occurred in Meiliang bay during this period. Thus, June 2021 was in the algae bloom period and December 2021 was in the algae collapse period.

The vertical distributions of TP and TPP concentrations in the water of Meiliang bay in June 2021 were similar to Chla, both of which showed extremely higher concentration levels in the surface water. Whereas, in December 2021, there was little TPP in the surface water with the limited algae biomass. A statistical study, covering data from 1992 to 2012, also indicated that annual mean TP concentrations were significantly correlated to Chla contents in Taihu Lake (Xu et al., 2017). In addition, it had indicated that P form composition in the water body was greatly controlled by the distribution of suspended sands for their strong abilities for P adsorption (Yao Q. Z. et al., 2016; King et al., 2022). In the algae bloom period (June 2021), the elevated TPP and TP concentrations in the water of Meiliang bay seemed to be not contributed by sand-adsorbing P but intracellular P in the algae cells.

Cyanobacteria is a kind of ancient bacteria with a strong ability for P absorption (Jeppesen et al., 2005). In the processes of algae growth and reproduction, amounts of PO₄ ^3- are assimilated into algae cells, deriving the transformation of dissolve P to particulate P in the water. When PO₄ ^3- loads in the water were not sufficient to supply algae growth, algae cells can even utilize the products of DOP enzymolysis as their nutritional source (Ni et al., 2022). Accordingly, dissolved PO₄ ^3- and DOP concentrations in the water of Meiliang bay in June 2021 were much lower than those in December 2021. Therefore, algae appearing and subsiding in the water strongly regulated form composition of dissolved P and particulate P in the water over the different periods. Furthermore, it was implied that algae vertical motion and algae horizontal banding aggregation during the algae bloom period can make spatial heterogeneity of TPP and TP distribution in the water.

In contrast to the periodic variations of TPP and TP in the water, sediment P forms such as IP, OP and FeP concentrations in December 2021 (the algae collapse period) were higher than those in June 2021 (the algae bloom period). (Ding et al., 2018) observed a similar phenomenon that sediment TP concentrations during the algae collapse period were relatively higher than those during the algae bloom period in Taihu Lake. Thus, during the algae collapse period, the accumulated P forms in the sediments were dominant in OP and FeP, as well as IP since most FeP was included in the IP (Tu et al., 2022).

From the algae bloom period to the algae collapse period, sediment OP accumulation occurred first due to the deposition of dead algae cells accompanied by the migration of algae intracellular P from water to sediments (Kristensen, 1994; Zhang et al., 2021). These biogenic sediments were abundant in organic matter, resulting in the initial OP accumulation in the sediments. It can be verified by the periodic variation of OP proportion in the sediment P of Meiliang bay, which rose from 19% in June 2021 to 27% in December 2021 (Figure 6).

After that sediments have undergone the processes of OP mineralization and FeP formation during the algae collapse period. The proportion of FeP in the sediment P of Meiliang bay in December 2021 (33%) was about 2 times to that in June 2021 (17%) (Figure 6). The increase of FeP proportion in the sediments in December 2021 was unlikely to directly source from the deposition of algae intracellular P, because algae intracellular P was dominant in the form of OP (Kristensen, 1994; Zhang et al., 2021). The increase of FeP proportion in the sediment in December 2021 might be formed through the dissolved PO₄ ^3- adsorption by iron oxides/hydroxides. There were large abundances of Proteobacteria and Actinobcateriota in the sediments of Meiliang bay (Supplementary Figure S5), which contained many kinds of organic P-solubilizing bacteria. Proteobacteria and Actinobcateriota in the sediments would accelerate the conversion of OP into soluble inorganic P (Tu et al., 2022). Then, the released soluble inorganic can be adsorbed onto iron oxides/hydroxides in the sediments (Wang et al., 2019), thus FeP was formed and accumulated in the sediments during the algae collapse period.

Besides sediment P accumulation in December 2021, sediment P release also occurred in Meiliang bay for the positive diffusion fluxes of PO₄ ^3- from pore water to overlying water. However, the diffusion flux of PO₄ ^3- in December 2021 was lower than that in June 2021 (Supplementary Figure S4), suggesting sediment P release was relatively weaker during the algae collapse period than that during the algae bloom period. Both OP mineralization and iron redox cycling are recognized as the main factors controlling P regeneration in the sediments (Ni et al., 2022; Xiao et al., 2022). In the surface sediments of Meiliang bay over the two periods, Proteobacteria and Actinobcateriota occupied above 30% of bacterial community composition, which played important roles in sediment degradation and metabolism, as well as OP mineralization. FeP in the sediments was easily reduced under anaerobic-reductive conditions or by FRBs activities, resulting in the co-release of Fe²⁺ and PO₄ ^3- (Fan et al., 2018). Due to the occurrence of FeP reductive dissolution in the sediments, significant correlations between PO₄ ^3- and Fe²⁺ were often observed in the sediments and pore water of lakes, rivers and oceans (Yao Y. et al., 2016; Sun et al., 2016; Sun et al., 2017; Pan et al., 2019).

During the algae collapse period (December 2021), labile PO₄ ^3- concentrations were not correlated with labile Fe²⁺ but correlated with labile S^2- in the sediment cores of Meiliang bay (Supplementary Table S2), and the vertical distribution of PO₄ ^3- was also different from Fe²⁺ in the pore water (Supplementary Figure S3). Thus, FeP reductive dissolution wasn’t the main cause of sediment P regeneration in Meiliang bay during the algae collapse period. OP mineralization driven by organic P-solubilizing bacteria activities (Proteobacteria and Actinobcateriota) should predominate sediment P regeneration during the algae collapse period. In addition, concentrations of labile PO₄ ^3- and labile S^2- in the sediment cores of Meiliang bay in December 2021 were significantly correlated (Supplementary Table S2), both of which unsteadily increased as the sediment depth deepened (Figure 4). The previous study indicated that rich organic matter (OM) can fuel sulfate reduction under anaerobic sediment conditions (Ma et al., 2017). Usually, active sulfate reduction fueled by OM prevails in the deep sediment zones where oxygen has been depleted (Burdige, 2006). Desulfobacterota works for sulfate reduction in sediments (Fan et al., 2018). The relative abundance of Desulfobacterota in the sediments of Meiliang bay in December 2021 (8.61%) was higher than that in June 2021 (8.00%). Thus, it was inferred that sulfate reduction accompanied by OM degradation might also accelerate the conversion of OP into soluble inorganic P in the sediments during the algae collapse period. Overall, during the algae collapse period, sediment P regeneration in Meiliang bay was mainly attributed to the OP mineralization process driven by organic P-solubilizing bacteria activities and accelerated by the sulfate reduction process.

During the algae bloom period (June 2021), the larger diffusion flux of PO₄ ^3- from pore water to overlying water, as well as the larger Fe²⁺ diffusion flux, can be attributed to FeP reductive dissolution in the sediments. It can be confirmed by the significant correlation between labile PO₄ ^3- and labile Fe²⁺ in the sediments and their similar vertical distributions in June 2021 (Supplementary Table S2; Figure 4). The active area of FeP reductive dissolution in the sediments of Meiliang bay in June 2021 was concentrated at the sediment depths ranging from −15 mm to −90 mm. Generally, either the reducing environment conditions can drive chemical iron reduction (CIR), or microbe-mediated activities can drive microbial iron reduction (MIR) (Ma et al., 2017). The previous studies indicated that the high electrical conductivity above 300 mV in the upper 35 mm surface sediment layer in Taihu Lake may exclude the CIR process (Christophoridis and Fytianos, 2006; Ding et al., 2018). We observed that the relative abundance of typical FRBs in the sediments of Meiliang bay in June 2021 was higher than that in December 2021 (Figure 5), reflecting the stronger FeP reductive dissolution through MIR during the algae bloom period. Moreover, the promotion of FeP reductive dissolution by FRBs activities may lead to the relatively lower concentrations of FeP and OP in the sediment in June 2021 (Figure 3). In addition, the release and diffusion of PO₄ ^3- through sediment FeP reductive dissolution may also trigger the appearance of algae bloom due to energy saving (Ding et al., 2018). Therefore, although PO₄ ^3- in the surface water was lower in June 2021 (Figure 2), it should be believed that large PO₄ ^3- diffused from pore water to overlying water can be quickly absorbed by algae again in terms of high PO₄ ^3- in the pore water and high TPP in the surface water during the algae bloom period (Figure 2; Supplementary Figure S4).

Cutting off loads of exogenous P input and internal P release to eliminate algae bloom events in the eutrophic lakes have been expected (Glibert and Burford, 2017), but there have been greatly underestimated the self-living cycles of P in aquatic environments driven by the processes of algae bloom and algae collapse. This study obtained direct evidence to illustrate the periodic variations of P migration and transformation in Meiliang bay of Taihu Lake, the eutrophic and cyanobacteria-dominant area, which was shown in Figure 7.

During the algae bloom period, the main route of P migration was from sediment P to water P: abundant PO₄ ^3- released from sediments by FeP reductive dissolution were migrated upward and then transported into algae cells by algae uptake of PO₄ ^3-, being TPP form in the surface water. Whereas, P migration from water to sediment was dominant in the algae collapse period, during which sediment P was mainly accumulated in the forms of OP and FeP. In the algae collapse period, the early formed TPP due to algae bloom settled in the sediments resulting in OP accumulation at first, and then sediment FeP was formed and accumulated through the processes of OP mineralization and P adsorption of iron oxides/hydroxides. Furthermore, the PO₄ ^3- existed in the pore water and PO₄ ^3- released from pore water to overlying water in Meiliang bay were based on the different sediment P regeneration routes in the two periods. During the algae bloom period, FeP reductive dissolution driven by FRBs activities was the main factor responsible for the dissolution and release of PO₄ ^3- in the sediments. Sediment P regeneration during the algae collapse period was mainly attributed to the OP mineralization process driven by organic P-solubilizing bacteria activities and accelerated by the sulfate reduction process.

A lack of understanding of the periodic variations of P migration and transformation patterns in aquatic environments with frequent algae bloom events can lead to inefficient treatments of lake eutrophication and algae bloom. For instance, the flocculation of algae cells using modified local soils, sands or bio modified activated carbon can rapidly make cyanobacteria or Microcystis cells disappear during the algae bloom period (Shi et al., 2016; Wang et al., 2016), but they just accomplish sediment P accumulation several months ahead of the algae collapse period, and they cannot prevent the development of the next algae bloom because of sediment P regeneration. Accordingly, it is best to implement capping and stabilization of treatments for sediment P (Douglas et al., 2016; Gu B.-W. et al., 2019) after the algae collapse period and before the next algae bloom period. Therefore, for eutrophic lakes, treatment approaches and management practices require special consideration about how to break the self-living cycles of P in aquatic environments driven by algae bloom and algae collapse processes.