Microbial Biomass and Community Composition Involved in Cycling of Organic Phosphorus in Sediments of Lake Dianchi, Southwest China

ABSTRACT Organic phosphorus (Po) was a major fraction of phosphorus (P) in sediments of lakes, and microbes were involved in most of its relevant biogeochemical cycling. Forms and quantification of Po were investigated by sequential fractionation in 18 sediments of Lake Dianchi, Southwest China. Microbial biomass and community structure in these sediments were determined by phospholipid fatty acids (PLFAs). Distribution of Po fractions were in the rank order that humic Po > nucleic acid and polyphosphate > residual P > Ca-Al-Po > Fe-Po > sugar Po > acid soluble Po > H2O-Po. The recoveries of Po and Pi in these detailed sequential fractions including residual P shows that the total contents of Po in sediments of lakes were overestimated by the Standards, Measurements and Testing (SMT) protocol (ignition method). Microbial biomass including Gram-positive bacteria (14.4–20.0%), Gram-negative bacteria (32.7–38.4%), microeukaryotes (14.9–24.4%), aerobic bacteria (43.6–55.8%), anaerobic bacteria (0–2.9%) and type І methanotrophs (17.6–24.4%) were assigned. Microbial mass and their composition were strongly correlated with H2O-Po, Fe-Po, nucleic acid and polyphosphate, and humic Po, though residual P was likely inert for microbes in sediments. The formation and degradation of Po was closely related with microbial activities in sediments. These findings have implications for understanding the role of microbes on cycling of Po and organic matter in sediments of lakes.


Introduction
Phosphorus (P) is widely recognized as a key element resulting in eutrophication of lakes (Schindler et al. 2008), and microbes drive most of its relevant biogeochemical cycling (Søndergaard et al. 2003;McMahon and Read, 2013). The sediment plays an important role in the P dynamics and microbial activities of lakes (Qian et al. 2011). Internal P loading in the sediments could be a significant source for bioavailable P of algae and other organisms in overlying water, especially after reducing external loading in a eutrophic lake (Søndergaard et al. 2003;Zhu et al. 2013a). Organic P (P o ) can constitute 12-80% of this internal P load in sediments of lakes (Ding et al. 2010;Torres et al. 2014). Especially, biogeochemical cycling of P o in sediments played an important role in the eutrophic status of lakes, such as Lake Dianchi and Lake Taihu in China (Zhu et al. 2013a;Zhu et al. 2015a).
Forms and labilities of P o (generally including condensed phosphate) in sediments of lakes have been investigated by sequential extraction (Golterman, 1996;Zhang et al. 2008), 31 P-NMR (Zhang et al. 2013;Torres et al. 2014) and enzymatic hydrolysis (Zhu et al. 2013a;Giles et al. 2015). These P o compounds and condensed P include phosphate monoester (e.g., AMP, inositol phosphates, a-glycerophosphate, bglycerophosphate and glucose phosphate); diester phosphate (e.g., phospholipids, RNA and DNA); phosphonates; pyrophosphate; and polyphosphate (Ding et al. 2010;Jørgensen et al. 2011;Zhu et al. 2015b). The biogeochemical cycling of these P o and condensed P including generation, degradation and transformation is closely related with activities of microbes in sediments of lakes (Søndergaard et al. 2003;Hupfer et al. 2004;McMahon and Read, 2013). However, there were still significant gaps in knowledge of relationships between internal cycling of P o and microbial activities in sediments of lakes.
Phospholipid fatty acid (PLFA), as a biomarker for microbial biomass and community structure of microorganisms, has been applied to investigate the microbial biomass and community composition in sediments of lakes (White et al. 1979;Findlay et al. 1989;Steger et al. 2011;Zhao et al. 2011). Extraction of PLFA allows quantification of viable microbial biomass in sediments without cultivation, which could avoid biases of the true community composition (White et al. 1979;Zhao et al. 2011). Thus, the relationships between microbial biomass and other quantified parameters such as dissolved organic carbon, total nitrogen (TN) and total P (TP) in the water and sediments of lakes could be analyzed further (Steger et al. 2011;Zhao et al. 2011). Although there are many methods for quantification of P o forms in the sediments, relationships between these P o forms with microbial biomass characterized by PLFA have not, to the knowledge of the authors, been discussed previously. Sequential extraction can be used to fractionate P forms and evaluate their bioavailabilities in sediments (Golterman, 1996;Golterman et al. 1998;Zhang et al. 2008). A sequential extraction procedure had been developed by Golterman and coworkers (1996;1998), which has been tested and calibrated for specific chemical P forms such as sugar bound phosphates and Ca-associated P (McDowell et al. 2005). Thus, P o forms were characterized by the sequential extraction procedure developed by Golterman and coworkers (1996;1998) for sediments of Dianchi Lake, a eutrophic lake, Southwest China. Microbial biomass and community composition in those sediments were analyzed by PLFA. Relationships between P o forms and microbial biomass in sediments were analyzed further.

Study area and sampling
Lake Dianchi (24 40 0 -25 02 0 N, 102 36 0 -102 47 0 E) is the sixth largest freshwater lake in China (Figure 1). It has an area of approximately 306 km 2 and is located in the southwest of Kunming, Yunnan Province, China. The average depth is approximately 5 m, and the maximum depth is 9.7 m. Lake Dianchi is very important as a source of drinking water supply, for industrial production, climate regulation and ecological protection. However, in recent decades Lake Dianchi has become eutrophic with frequent and massive blooms of algae (Wang et al. 2009). Phosphorus is the nutrient that currently limits productivity in Lake Dianchi (Gao et al. 2005). Lake Dianchi is a relatively closed system with a residence period of 3-8 years, which might exacerbate retention and accumulation of nutrients in the lake. Now that external inputs of nutrients have been controlled to a certain extent (Lu et al. 2012), large loads of internal nutrients are a major factor that continue to influence the trophic status of Lake Dianchi (Gao et al. 2005;Zhu et al. 2013a). In May 2010, surface sediments were collected from 18 locations (identified as S1-S24; Figure 1) by use of a Peterson grab sampler. Sediments were transported to the laboratory in air-sealed plastic bags and cold storage with ice. Sediment samples were lyophilized and ground to powder and stored at ¡20 C before analysis.

Analysis of sediment properties
Total concentrations of Al, Ca, Fe and Mn were measured using inductively coupled plasma optical emission spectrometry after microwave acid (HNO 3 -HCl-HF) wet digestion of sediment samples. General characteristics of P forms including contents of TP, inorganic P (P i ) and P o in sediments were determined by the method that had been harmonized and validated by use of the SMT program of the European Commission (Ruban et al. 1999;Ruban et al. 2001). Concentrations of P were determined by molybdenum blue method (Murphy and Riley, 1962). Sediments were pretreated by an excess of 1 mol L ¡1 HCl to remove carbonates, and then analyzed for total organic carbon (TOC) and TN using an elemental analyzer (Vario EL S, Elementar, Germany).

Sequential fractionation of sediment phosphorus
Phosphorus in sediments was extracted using the sequential fraction scheme developed by Golterman et al. (1996;1998) ( Figure 2). Briefly, 0.5 g of each lyophilized sediment was sequentially extracted via shaking with 15 mL of deionized water (2 h), 0.05 M Ca-EDTA (C1% Na-dithionite, pH 7.5), 0.1 M Na-EDTA (pH 4.5), 0.5 M H 2 SO 4 , cold 0.5 M trichloroacetic acid (TCA; 0 C, 4 h), hot 0.5 M TCA (95 C, 30 min), and finally 2 M NaOH (90 C, 1 h) before the remaining P was released by digestion with K 2 S 2 O 8 . These fractions represent, in sequential order, water soluble or sediment interstitial water P (H 2 O-P), Fe associated P (Fe-P), Ca and Al associated P (Ca-Al-P), acid soluble inorganic and organic P (ASIP or ASOP), sugar bound P (Sugar P) after digestion by K 2 S 2 O 8 (cold TCA), nucleic P and polyphosphate (NP and PP) after digestion by K 2 S 2 O 8 (hot TCA), humic bound P and phytate (NaOH P i and humic P o ), and residual P. Following extraction, each suspension of sediment was centrifuged (8000 £ g) for 10 min, decanted, and an aliquot was taken for quantification of P. To determine the P i from the chelating reagents extracts, a 25-50 time dilution was used to prevent interference of chelating reagents with the Mo-P colorimetric reaction. In addition, the P o fraction was defined as the difference between detectable molybdate reactive P before and after digestion with K 2 S 2 O 8 at 121 C for 30 min.

Fatty acid analysis
For the PLFA extraction, 1.5 g freeze-dried sediment samples were analyzed by a modified one-phase Bligh-Dyer method (Frostega rd et al. 1991). Total microbial biomass was determined as total phospholipids phosphate (PLP) by digestion of K 2 S 2 O 8 followed by quantification using phosphomolybdatemalachite green analyzed at 610 nm (Findlay et al. 1989). Structure of microbial community was determined by quantifying fatty acid methyl esters (FAMEs) from phospholipids. Total lipids were fractionated into neutral, glyco-, and polar lipids using solid phase extraction with silicic acid columns (Supelclean TM LC-Si SPE Tubes, Supelco, Inc.). The polar fraction containing the phospholipids was subjected to a mild alkaline methanolysis for conversion to FAMEs. Organic fractions were dried by a stream of nitrogen and stored at ¡20 C prior to gas chromatography (GC) analysis.
FAMEs were dissolved in hexane for GC analysis. The fatty acid 19:0 (nonadecanoic acid methyl ester) was added to the samples as an internal standard. The FAMEs were identified by GC (Agilent 6890N) with an autosampler, an Agilent 7683B injector and a flame ionization detector. The following temperature program was used in the column oven: 170 C at the starting point, followed by an increase of 5 C/min to 260 C, and then increase of temperature at 40 C/min to 310 C, where temperature was maintained for 90 s. Vaporization chamber and detector temperature were 250 C and 300 C respectively. Volume injected was 1 ml, and the split ratio was 20:1. The FAMEs were identified by use of Sherlock MIS software (v. 4.5), and qualified relative to the internal standard.
Standard nomenclature for molecular formula as "X: YvZ(c/ t)," was used for FAMEs (Table S1): "X" is the total number of carbon atoms and "Y" is the number of unsaturated double bonds. The position of the first double bond is indicated by "v" and the number of carbon atoms from the aliphatic end is indicated by "Z." Furthermore, the suffixes "c" and "t" specify the cis and trans conformations of the double bond respectively. Methyl branching at the iso and anteiso positions was designated by the prefixes "i" and "a" respectively. The prefix "cy" denotes cyclopropane fatty acids. Additionally, the suffixes "G" and "Alcohol" specify the uncertain double bond and fatty alcohol. Certain PLFAs were assigned to microbial groups based on previous studies (Vestal and White, 1989;White et al. 1996;Liu et al. 2009;Steger et al. 2011). Thus, all iso-and anteisobranched fatty acids with 14-19 carbons plus 15:0 were considered as representing Gram-positive bacteria (G C ); all monounsaturates containing 14-19 carbons (except 16:1v9 and 18:1v9) plus cyclopropane fatty acids were considered to be markers for Gram-negative bacteria (G ¡ ). The sum of 14:0; 16:1v9, 17:0, 18:0, 18:2, 18:3, 18:1v9, 20:4, 20:5 and 24:0 were considered as representing microeukaryotes; all monounsaturates containing 14-19 carbons were regarded as aerobic bacteria; cyclopropane fatty acids (cy 17:0 and cy 19:0) were considered as representing anaerobic bacteria. Furthermore, monounsaturated 16 carbon fatty acids indicate type y methanotrophs, a type of methane-oxidizing bacteria. Moreover, Gram-positive bacteria would concert some monounsaturated carbon fatty acids to cyclopropane fatty acids under the stress of hunger. Therefore, contents of cyclopropane fatty acids could be stress biomarker for nutrients ).

Statistical analysis
Data were checked for deviations from normality of variance before analysis. To check whether there were significant correlations between sequential P o forms and microbial community structure in the sediments, Pearson correlation coefficients (r values, two-tailed) at P < 0.01 and P < 0.05 were determined using SPSS 11.5 for windows. Factor analysis including principal component analysis (PCA) was further carried out by SPSS 11.5.

Sediment characteristics
Sediment characteristics of Lake Dianchi including TP, P i , P o , TOC, TN and TOC/TN (Table 1) were also discussed in a the previous study (Zhu et al. 2013a). Concentrations of TP in sediments of Lake Dianchi ranged from 1574 to 2623 mg kg ¡1 . Inorganic P was the main component of TP, concentrations of which ranged from 860 to 1847 mg kg ¡1 and accounted for 51.7-70.6% of TP. However, the HCl extractable P i was the main composition of P i in these sediments, which was P primarily bound to calcium that was released from sediments with difficulty (Zhu et al. 2013a;Zhu et al. 2013b). Concentrations of P o ranged from 393 to 630 mg kg ¡1 , which accounted for 20.6% to 29.8% in sediments of Lake Dianchi. Contents of TOC and TN ranged from 3.31% to 8.29% and 0.42% to 1.00% respectively. Molar ratios of TOC/TN ranged from 8.2 to 9.7, which  Golterman et al. (1996;1998). indicated that the organic matter derived mainly from the autochthonous sources such as algae, bacteria and aquatic macrophytes (Meyers and Ishiwatari, 1993). This was also supported by the molecular compositions of organic matter, d 13 C and d 15 N in sediments of Lake Dianchi from the previous studies (Wang et al. 2009;Xiong et al. 2010). Contents of Al ranged from 63.5 to 80.2 g kg ¡1 . Contents of Ca ranged from 30.5 to 84.5 g kg ¡1 . Contents of Fe were varied from 58.3 to 77.0 g kg ¡1 . Contents of Mn ranged from 0.8 to 1.2 g kg ¡1 . Contents of Al, Ca and Fe, especially Ca, was greater than that of other lakes in China, such as Lake Taihu and Lake Poyang (Ding et al. 2010). This is likely due to the fact that sediments from Lake Dianchi are calcareous sediments.

Distribution and recoveries of P o and P i in sequentially extracted fractions
Contents of H 2 O-P o ranged from 1.4 to 5.1 mg kg ¡1 , which accounted for 0.1-0.3% of TP in the sediments ( Figure 3). H 2 O-P o was loosely adsorbed to sediment particles or in interstitial water of sediments, which was transferred easily across  the sediment-water interface (Zhu et al. 2013a;Zhu et al. 2015b). Though there was only a small proportion of H 2 O-P o in TP, large proportions of H 2 O-P o extracted from the sediments of Lake Dianchi could be hydrolyzed by phosphatase (e.g., alkaline phosphatase, phosphodiesterase and phytase), thus bioavailable for algal blooming (Zhu et al. 2013a). Contents of Fe-P o ranged from 9.2 to 45.3 mg kg ¡1 and accounted for 0.6-2.3% of TP in sediments. Fe associated P was an important internal source of P in sediments that could be released to support blooms of cyanobacteria in Lake Dianchi Zhu et al. 2015a). Fe-P o in sediments could be released into the overlying water under anoxic conditions, thus some dissolved P o (e.g., phytate) could be hydrolyzed by phosphatase (Golterman et al. 1998;Zhu et al. 2013a). Some P o (e.g., glucose phosphate) associated with Fe oxides such as goethite could be hydrolyzed by phosphatase on the surface of mineral directly without desorption (Olsson et al. 2011), and thus could be as a proportion of Fe-P i (Figure 4). Contents of Ca-Al-P o ranged from 13.8 to 73.2 mg kg ¡1 , and accounted for 0.7-3.5% of TP in sediments. Aluminum hydroxide has been widely used to precipitate P from overlying water of lakes and to immobilize P in sediments (Jensen et al. 2015). Ca-Al-P o in sediments was thought to be released with difficulty and thus hardly bioavailable to support algal blooming (Zhu et al. 2015a). Contents of ASOP ranged from 9.0 to 40.3 mg kg ¡1 , and accounted for 0.3-2.5% of TP in sediments. ASOP was thought to be a component of moderately labile P o (Zhang et al. 2008). Contents of sugar P o ranged from 22.2 to 31.8 mg kg ¡1 with relative contributions to TP in sediments of 1.1-1.7%. Sugar P o , such as glucose phosphate, which would be hydrolyzed by phosphatase and release bioavailable phosphate (Zhu et al. 2013a), has also been detected by 31 P-NMR (Giles et al. 2015). Humic P o , NP and PP defined in this procedure of sequential fractionation were the main constituents of P o in sediments from Lake Dianchi ( Figure 3). Contents of NP and PP ranged from 51.0 to 107.3 mg kg ¡1 , and accounted for 2.9-4.9% of TP in sediments. Nucleic acid P o (e.g., DNA, RNA) and PP (e.g., pyrophosphate) could be hydrolyzed by phosphatase, thus was bioavailable for algal blooming (Zhu et al. 2013a;Zhu et al. 2015b). Contents of humic P o ranged from 51.5 to 143.0 mg kg ¡1 , and accounted for 3.2-7.5% of TP in sediments. Humic P o , characterized by 31 P-NMR and hydrolysis by phytase, included orthophosphate, phosphonates, monoester P (e.g., glucose phosphate, inositol phosphates, a-and b-glycerophosphate) and diester P (Golterman et al. 1998;He et al. 2015). When evaluated by ultraviolet irradiation and enzymatic hydrolysis, only 10-63% humic bond P was considered to be labile (He et al. 2015). Bioavailability of humic P o would be reduced by interaction between P o and humic substances in sediments of lakes (Zhu et al. 2015b). Additionally, contents of residual P ranged from 31.5 to 72.4 mg kg ¡1 and accounted for 1.4-4.0% of TP in sediments. Residual P after sequential fractionation would be a component of P o or P i in sediments from Lake Dianchi, which could be refractory P o or P i . Concentrations and distributions of P i fractions extracted simultaneously with the P o fractionation are shown (Figure 4). Ca-Al-P i was the dominant fraction in sediments from Lake Dianchi, ranging from 732.6 to 1297.7 mg kg ¡1 , with contributions of 41.4-50.5% of TP in sediments. This is likely due to abundant phosphate rock in the area of Lake Dianchi. For example, concentrations of Ca-Al-P i were greater in sediments from south part of Lake Dianchi (S1, S2, S3 and S4), which was the result of phosphate rock mining in this area in the past. H 2 O-P i and Fe-P i were the important internal source of bioavailable P in sediments of Lake Dianchi Zhu et al. 2015a). Contents of H 2 O-P i ranged from 0 to 1.4 mg kg ¡1 and accounted for only 0-0.1% of TP in sediments; contents of Fe-P i ranged from 188.1 to 524.1 mg kg ¡1 and accounted for 9.7-20.0% of TP in sediments. Contents of ASIP ranged from 212.1 to 283.7 mg kg ¡1 and accounted for 9.6-17.2% of TP in sediments. ASIP is likely the proportion of Ca-P not extracted by Na 2 EDTA in the previous step. Contents of NaOH-P i ranged from 198.6 to 323.0 mg kg ¡1 and accounted for 9.0-19.0% of TP in sediments. NaOH-P i is likely the proportion of Fe-P i or Al-Ca-P i not extracted in previous steps. Residual P was also likely a component of P i fractions, with relatively low proportions in P i fractions (Figure 4).
Recoveries of TP, P o and P i from sediments of Lake Dianchi by this sequential extraction procedure are given in Table 2. Recoveries of TP were close to 100%, which ranged from 98.5% to 110.9% in these sediments. However, recoveries of P o varied from 60.0% to 80.5% in these sediments. Recoveries of P i were 116.3-172.3% in these sediment samples. The reason for this might have been due to labile P o being hydrolyzed during extraction and measurement. For example, solutions including H 2 SO 4 and NaOH could result in hydrolysis of P o (Turner et al. 2005). Second, total contents of P o in these sediments were overestimated by use of the SMT protocol. Total contents of P i in these sediments were underestimated by use of the SMT protocol. Overestimation of P o in sediments by use of the SMT protocol was also supported by results of sequential extraction and 31 P-NMR analysis (Zhu et al. 2016). It is likely that some P i is immobilized strongly by sediments-due to this the solution of HCl in the SMT protocol could not recover all of the P i from sediments in the first step. Thus, the remaining P i in sediments would be determined as a proportion of P o after ignition and extraction by HCl solution again (Zhu et al. 2016). Also, the ignition method tends to overestimate contents of P o in soils by increasing solubility of P i after ignition (Condron et al. 1990;Turner et al. 2005). Thus, caution is needed in quantification of P o in sediments by use of the ignition method.

Microbial biomass and community structure characterized by PLFA
Microbial biomass in sediments from Lake Dianchi, as measured by the PLP, ranged from 420.0 nmol PLP g ¡1 to 1247.9 nmol PLP g ¡1 dm (dry mass) ( Figure 5). Total amounts of PLP in sediments of Lake Dianchi were compared with contents of PLP in the sediments of eight lakes from Sweden (Steger et al. 2011) and an eutrophic Lake Acton (Smoot and Findlay, 2001). This result indicated that the microbial biomass in sediments from eutrophic Lake Dianchi was similar to those in sediments of other lakes described previously. Microbial community structure in Lake Dianchi was further characterized by use of the compositions of FAMEs from PLFA (Table 3 and  Table S1). In sediments from Lake Dianchi, the proportion of G C ranged from 14.4% to 20.0%, while the proportion of G ¡ ranged from 32.7% to 38.4%. The proportion of microeukaryotes ranged from 14.9% to 24.4%. The proportion of aerobic bacteria accounted for 43.6-55.8% of the microbial biomass in sediments of Lake Dianchi. However, based on relative proportions of FAMEs, the proportion of anaerobic bacteria accounted for only 0-2.9% of the microbial biomass in sediments of Lake Dianchi. Additionally, the proportion of type y methanotrophs accounted for 17.6-24.4% of microbial biomass in sediments. Proportions of G C , G ¡ and microeukaryotes in sediments of Lake Dianchi were similar with those in sediments Table 2. Recoveries (%) of P including TP, P i and P o by sequential extraction in sediments of Lake Dianchi.
Recoveries S1 S2 S3 S4 S6 S9 S10 S13 S17 S18 S19 S20 TP in the sediments were determined by the SMT procedure, and recovery of TP including sequential extraction TP and residual P. b Contents of total P o were determined by the SMT procedure, and recovery of P o including sequential extraction P o and residual P. c Contents of total P i were determined by the SMT procedure, and recovery of P i including sequential extraction P i , except for residual P. from boreal lakes of Sweden (Steger et al. 2011). Ratios of G C to G ¡ (G C /G ¡ ) ranged from 0.4 to 0.6 (Table 3), which suggested that G ¡ was the predominant composition of microbial biomass in sediments from Lake Dianchi. These results further showed that Gram-negative bacteria were the most important contributors to community composition in sediments of lakes (Steger et al. 2011).
Proportions of aerobic bacteria were relatively great (Table 3), though relatively large amounts of organic matter have been accumulated in sediments due to frequent blooms of algae. Values of DO in the overlying water of Lake Dianchi now range from 5 to 10 mg L ¡1 ; hypoxia (DO 2 mg L ¡1 ) occurred rarely in the overlying water (Liu et al. 2014). Based on the composition of FAMEs, proportions of anaerobic bacteria were small in sediments of Lake Dianchi. Branched fatty acids containing 14-16 carbons also likely represented some anaerobic bacteria (Zaady et al. 2010). However, only cyclopropane fatty acids (cy 17:0 and cy 19:0) were selected as representing anaerobic bacteria based on previous studies (Vestal and White, 1989). Thus, the proportion of anaerobic bacteria might underestimate in this study. Type y methanotrophs are physiologically and phylogenetically distinct and fall into the group of Gammaproteobacteria. Type y methanotrophs are an important methane-oxidizing bacteria in sediments of Lake Dianchi (Table 3), which also support by the abundance of Gammaproteobacteria (Bai et al. 2012) and quantification of type y methanotrophs (Yang et al. 2016) in sediments of Lake Dianchi. Type y methanotrophs play an important role in cycling of carbon, nitrogen and oxygen in lakes, and are also widespread in other lakes such as Lake Washington (Auman et al. 2000;Costello et al. 2002). Lake Dianchi, is a hypereutrophic lake, where debris derived from cyanobacteria and aquatic plants accumulated in sediments (Dong et al. 2006;Qu et al. 2013). Under suitable conditions in Lake Dianchi, methane would be produced from accumulated organic matter (Dong et al. 2006;Yang et al. 2016). The flux of methane was detected at the water-air interface (Chen et al. 2007). Type y methanotrophs would play a key role in oxidizing and mitigating the methane emission in sediments of Lake Dianchi (Yang et al. 2016). Although the methane was the only source of carbon for methanotrophs, they could degrade other organic matter in sediments, which would also play an important role on cycling of organic matter in the sediments of Lake Dianchi. Additionally, the "stress" of nutrients for microbes was low, which indicated that the organic matter accumulated in the sediments of Lake Dianchi received sufficient nutrients.

Relationships between P fractions and microbial community structure
Bulk compositions of organic matter including TOC, TN and TP; some P o fractions including H 2 O-P o , Fe-P o , NP and PP; and humic P o were significantly correlated with microbial biomass and other microbial composition except for anaerobic bacteria (Table 4). Greater concentrations of TOC in sediments would result in greater abundances of microbial mass in sediments of Lake Dianchi. This indicated that microbial activities would be important for decomposition of organic matter in sediments. Decomposition of organic matter with microbial activities would result in degradation or release of nutrients such as P o and N complexed with organic matter (Wu et al. 2001;Wu et al. 2010). Microbes have been shown previously to be the key factor for cycling of N in sediments of lakes (Keeney, 1973). Contents of TN were significantly correlated with microbial mass in sediments from Lake Dianchi. Contents of TP were also significantly correlated with microbial mass in sediments, which indicated that microbes also play an important role on biogeochemical cycling of P in Lake Dianchi. Bacterial communities and their roles in mineralization of P in sediments of Lake Dianchi has also been studied previously (Xia et al. 2004).
Results of this study suggested that numbers of P-accumulating bacteria were greater in areas where P was accumulating, and that bacteria can absorb soluble P and transport it into the form of polyphosphate after self-synthesis in the sediments of Lake Dianchi (Xia et al. 2004). After death of P-accumulating bacteria, the P would be accumulated in sediments as various minerals. Additionally, P-decomposing bacteria were much greater (approached 490 times) in the P o culture medium rather than that in the P i culture medium (Xia et al. 2004). This result indicated that accumulating P o in sediment during eutrophication of Lake Dianchi resulted in increased growth of bacteria in sediments. H 2 O-P o was significantly correlated with microbial mass, G C , G ¡ , microeukaryotes, aerobic bacteria and type y methanotrophs (Table 4), which was likely that the decomposition of organic matter by bacteria results in increasing labile P o in sediments of lakes. Additionally, bacteria play an important role on bioavailability of H 2 O-P o that dissolved into overlying water, which was readily available for algal blooming (Zhao et al. 2012). Bacteria such as Gordonia sp. and Burkholderia sp. can degrade constituents of labile H 2 O-P o such as glucose phosphate and mononucleotide phosphate and release bioavailable orthophosphate to support blooming of algae in eutrophic lakes (Zhao et al. 2012). Fe-P o was also closely related with microbes such as microeukaryotes and aerobic bacteria (Table 4). These microorganisms would consume O 2 , NO 3 -and so on during organic matter decomposition, thus providing necessary conditions for reduction of Fe(S) and subsequent release of Fe-P o (G€ achter et al. 1988). There were no significant relationships between microbial biomass and Ca-Al-P o , ASOP and sugar P o ( Table 4). The distribution of microbial biomass was significantly correlated with NP and PP in the sediments of Lake Dianchi. Contents of NP and PP characterized by the same sequential fractions in this study were significantly correlated with contents of diester such as DNA and RNA characterized by 31 P-NMR in the soils (P < 0.01) (McDowell et al. 2005). DNA or RNA also could represent the microbial mass. Thus, NP and PP could be directly related to microbial mass; it is likely that NP and PP were derived from microbial mass in the sediments of Lake Dianchi. The NP and PP attributed by microbial mass would play an important role on cycling of P in lakes (Hupfer et al. 2004;Shinohara et al. 2012). Humic P o was also closely related with microbial biomass (Table 4). Humic P o includes monoester P, diester P and pyrophosphate complexed with humic substances (He et al. 2006). Some labile P o in humic substance could be hydrolyzed by activities of microbial biomass, such as enzymatic hydrolysis (Lovley et al. 1996). Some P o derived from debris of microbial biomass would be complexed with humic substances in sediments (He et al. 2015;Zhu et al. 2015b). Further investigation was needed for understanding the role of microbial biomass on the formation of humic P o in sediments. Additionally, there were no relationships between residual P and microbial biomass, which further indicated that residual P would be refractory and could not be recycled by the activities of microbes in sediments of Lake Dianchi.
Relationships between the P i fractions and microbial biomass were also analyzed (Table 5). There were no relationships between H 2 O-P i and microbial biomass. However, Fe-P i was significantly correlated with microbial mass, G C , G ¡ , which indicated that cycling of Fe-P i was closely related with microbes in sediments of Lake Dianchi. Ca-Al P i was also significantly correlated with microbial biomass. Fe-P i and Ca-Al P i were the main component of P i in the sediments of Lake Dianchi (Figure 4). It has been reported that microbes were the key factor in formation of Fe-P i and Ca-Al P i in sediments of Lake Dianchi (Xia et al. 2004). Phosphorus-concentrating bacteria assimilated soluble P into P o or condensed P i , which would form precipitates with Fe, Ca or Al hydrated P i after death of microbes and diagenesis in sediments of Lake Dianchi (Xia et al. 2004). The negative correlation between ASIP and G ¡ , microeukaryotes and aerobic bacteria indicated that carbonic acid, nitric acid or organic acids would be important for dissolution of ASIP in sediments (Xia et al. 2004). Additionally, there was no significant correlation between NaOH P i and residual P with microbial biomass (Table 5).
In order to present an in-depth analysis of correlations between P o fractions and microbial biomass in sediments, a PCA was carried out ( Table 6). Results of the PCA showed that the proportion of total variance explained by PC1-PC6 was 91.549%, which explained the majority of relationships were closely related with decomposition or accumulation of organic matter and activities of microbes in sediments of Lake Dianchi. PC2, which accounted for 13.607% of the total variance, was negatively loaded by Ca-Al-P o and sugar P o , and was positively loaded by ASOP and residual P. These P o fractions were likely more influenced by other factors rather than microbes in sediments of Lake Dianchi. PC3, which accounted for 8.327% of the total variance, was primary correlated with sugar P o and NP and PP. This indicated that the origin of sugar P o and NP and PP was likely similar, such as from microbe or plankton. Origin and decomposition of sugar P o were also related with microbe in the sediments. PC4, accounted for 6.550% of the total variance, was negatively loaded by Fe-P o and positively loaded by residual P. This further indicated that biogeochemical cycling of Fe-P o and residual P was different; an anaerobic condition would result in release of Fe-P o . However, residual P would be preserved in sediments. PC5, which accounted for 6.162% of the total variance, was correlated with anaerobic bacteria. This is likely that the anaerobic bacteria could not be characterized well by the PLFA in the sediments in this study. Thus, the role of anaerobic bacteria in sediments of Lake Dianchi needs further investigation. PC6 accounted for 4.338% of the total variance and was negatively loaded by sugar P o and positively loaded by humic P o . This likely indicated that sugar P o as operationally defined here was labile and could be hydrolyzed or assimilated by microbes in lakes, whereas large humic P o was likely formed under the activities of microbe in sediments of Lake Dianchi.
The microbial mass and their composition were closely related with TOC, TN and TP in sediments of Lake Dianchi. The P o fractions including H 2 O-P o , Fe-P o , nucleic acid and polyphosphate, and humic P o were closely related with the activities of microbial mass in the sediments of Lake Dianchi. Residual P in sediments was not being recycled by the activities of microbes. Accumulation of organic matter and nutrients in sediments of eutrophic lakes would result in more activities of microbes, thus biogeochemical cycling of P o in those lakes. However, microbes involved in cycling of P o in sediments of lakes, such as molecular mechanisms controlling metabolism of P o should be further investigated (McMahon and Read, 2013).

Funding
This research was jointly supported by the National Natural Science Foundation of China (No. 41403094, 41130743, 41261140337). Prof. Giesy was supported by the program of 2012 "High Level Foreign Experts" (#GDT20143200016) funded by the State Administration of Foreign Experts Affairs, the P.R. China to Nanjing University and the Einstein Professor Program of the Chinese Academy of Sciences. He was also supported by the Canada Research Chair program and a Distinguished Visiting Professorship in the School of Biological Sciences of the University of Hong Kong.