SHI Limei (施麗梅) HUANG Yaxin (黃亞新) LU Yaping (盧亞萍)CHEN Feizhou (陳非洲) ZHANG Min (張民) YU Yang (于洋) KONG Fanxiang (孔繁翔)

1State Key Laboratory of Lake Science and Environment,Nanjing Institute of Geography and Limnology,Chinese Academy of Sciences,Nanjing 210008,China

2Biological Experiment Teaching Center,College of Life Sciences,Nanjing Agricultural University,Nanjing 210095,China

AbstractCyanobacterial blooms occur in eutrophic lakes worldwide, and greatly impair these ecosystems.To explore inf l uences of cyanobacterial blooms on dynamics of both particulate organic matter (POM) and dissolved organic matter (DOM), which are at the base of the food chain, an investigation was conducted from December 2014 to November 2015 that included various stages of the seasonal cyanobacterial blooms(dominated byMicrocystis) in a large-shallow eutrophic Chinese lake (Taihu Lake). Data from eight sites of the lake are compiled into a representative seasonal cycle to assess general patterns of POM and DOM dynamics. Compared to December, 5-fold and 3.5-fold increases were observed in July for particulate organic carbon (POC, 3.05-15.37 mg/L) and dissolved organic carbon (DOC, 5.48-19.25 mg/L), respectively, with chlorophylla(Chla) concentrations varying from 8.2 to 97.7 μg/L. Approximately 40% to 76% of total organic carbon was partitioned into DOC. All C, N, and P in POM and DOC were significantly correlated with Chla. POC:Chlaratios were low, whereas proportions of the estimated phytoplankton-derived organic matter in total POM were high during bloom seasons. These results suggested that contributions of cyanobacterial blooms to POM and DOC varied seasonally. Seasonal average C:P ratios in POM and DOM varied from 79 to 187 and 299 to 2 175, respectively. Both peaked in July and then sharply decreased.Redundancy analysis revealed that Chlaexplained most of the variations of C:N:P ratios in POM, whereas temperature was the most explanatory factor for DOM. These fi ndings suggest that dense cyanobacterial blooms caused both C-rich POM and DOM, thereby providing clues for understanding their inf l uence on ecosystems.

Keyword:C:N; N:P; stoichiometry; phytoplankton blooms; eutrophic lake

1　INTRODUCTION

In aquatic environments and especially eutrophic waters, phytoplankton are major components of the seston and play a major role in the biogeochemical cycling of carbon (C), nitrogen (N), and phosphorus(P). In particular, phytoplankton contribute to autochthonous particulate organic matter (POM)through cell proliferation, or cell detritus (Hessen et al., 2003) and to dissolved organic matter (DOM) by direct extracellular release, cellular lysis, or zooplankton sloppy grazing (Nagata, 2000; Zhang et al., 2009). Given that POM is a major substrate for grazers and DOM is an important substrate for microbial communities (Mei et al., 2005), their concentrations greatly inf l uence energy transfer in the food web, and thus control ecosystem structure.However, partitioning of organic matter appears to vary by ecosystem. Respectively, about 89% and 78%of accumulated organic matter were partitioned into POM duringPhaeocystisblooms in the Ross Sea or during the exponential growth of coastal diatom blooms in Oregon deck incubations (Carlson et al.,2000; Wetz and Wheeler, 2003). More than 56% of total organic carbon loading was partitioned into DOC with cyanobacterial blooms in Lake Soyang,Korea (Kim et al., 2000).

As the three most important elements of organic matter, C, N, and P can provide basic stoichiometric information on food resources. Lower C:N:P ratios indicate good food quality, while higher C:N:P ratios suggest poor food quality (Sterner et al., 1993; Elser et al., 2001). Thus, the nutrient ratio of organic matter determines the features, roles and fates of organic matter in ecosystems. In fact, the relative proportions of detritus and various phytoplankton species comprising the seston greatly affect the C:N:P ratio(Hecky and Kilham, 1988; Geider and La Roche,2002; Hessen et al., 2003; Klausmeier et al., 2004).Even for a certain type of phytoplankton, the elemental stoichiometry within a cell is regulated by the biochemical allocation of resources to different growth strategies (Elser et al., 2003; Klausmeier et al., 2004). DOC concentrations with strong impacts on the C:N:P ratio of DOM are greatly inf l uenced by phytoplankton blooms or allochthonous input (Zhang et al., 2009). Therefore, patterns of that ratio can vary geographically towing to variations in temperature,nutrients and phytoplankton composition among various ecosystems (Sterner and Elser, 2002; Martiny et al., 2013). Moreover, the C:N:P ratio patterns may vary seasonally, even within Specific ecosystems.

The occurrence of cyanobacterial blooms dramatically inf l uences carbon and nutrient cycling and ecosystem structure in eutrophic lakes worldwide.However, information about the entire seasonal cycle of both POM and DOM in cyanobacterial bloomforming lakes is still lacking. In the present study, we monitored C, N, and P contents in both POM and DOM over a full year in Taihu Lake. This is a typical eutrophic lake, with high concentrations of nitrogen and phosphorus caused by human activities that has long experienced from cyanobacterial blooms (Chen et al., 2003; Qin et al., 2015). There are heavy cyanobacterial blooms in warm seasons, which generally decline in winter. Our objective here was to(1) evaluate the dynamics of concentrations and stoichiometry of C, N and P in POM and DOM in the eutrophic lake, and (2) access key water parameters contributing to the dynamics of POM and DOM.

Fig.1　The map of Taihu Lake and sampling locations as indicated

2　METHOD AND MATERIAL

2.1　Study area

Taihu Lake, which is located between 30°56′-31°33′N and 119°53′-120°36′E in eastern China, is the third largest lake in the country. With a surface area of 2 338 km2, a mean depth of 1.9 m, and a water residence time of approximately 284 days (Qin et al.,2007), the lake is used as the drinking water source of several cities, such as Shanghai, Suzhou, Wuxi, and Huzhou; serves as an important water resource for irrigation in farming and industry; and is an important recreational and tourist attraction (Qin et al., 2007). Along with the rapid development of the economy, the lake water has been seriously polluted.Presently, the lake as a whole is eutrophic and suffers from cyanobacterial blooms, which are overwhelming dominated byMicrocystisspp. (Chen et al., 2003; Ma et al., 2016).

2.2　Sample collection and handling

Monthly samplings were carried out between December 2014 and November 2015 at eight stations in the lake (Fig.1). Water samples were collected from the surface to a depth of 0.5 m using a 5-L acidcleaned plastic bucket. Because Taihu Lake is shallow and well mixed (McCarthy et al., 2007), sampling of the top 0.5 m water column was considered representative of the entire column. Phytoplankton(mainly composed of cyanobacterial colonies) was collected by towing a phytoplankton net (64 μm mesh) through the cyanobacterial surface bloom from a slowly moving motorboat around the sampling station. The water physical parameters including the temperature, pH and dissolved oxygen (DO) were measured at each sample site using a multiparameter meter (model 6600V2; Yellow Springs Instruments,Yellow Springs, OH, USA). An adequate water sample with a volume of 100-300 mL was fi ltered through a precombusted (450°C for 4 h) Whatman GF/F glass fi ber fi lter (nominal pore size 0.7 μm), and the fi lters and fi ltrate were used for the determination of POM and DOM, respectively. In addition, 50-100 mL water samples were fi ltered through Whatman GF/C glass fi ber fi lters (nominal pore size 1.2 μm),and the fi lters were used to determine the chlorophylla(Chla) concentration. All the fi lters and fi ltrates collected in acid washed glass bottles were stored frozen until analysis.

2.3　Chemical analyses

Chlawas extracted from the fi lters with 90%acetone and determined using a spectrof l uorophotometer (RF-5301PC, Shimadzu, Japan). For the particulate organic carbon (POC) and particulate nitrogen (PN) analysis, the fi lters were freeze-dried,weighed, and then fumed with concentrated HCl for 4-5 h in a desiccator to remove inorganic carbon. The carbon and nitrogen contents on the fi lters were measured using a CN elemental analyzer EA3000(EuroVector, Italy). Particulate phosphorus (PP) was determined by inductively coupled plasma- atomic emission spectrometry (ICP-AES, Prodigy, Teledyne Leeman Labs, Hudson NH, America) after complete digestion with HCl-HNO3-HF-HClO4. The C:N:P molar ratios were obtained as POC:PN:PP from the bulk particulate fraction.

DOC was determined using a TOC analyzer(Teledyne Tekmar, Torch, USA). The concentrations of nutrients in the GF/F fi ltrate, including ammoniumnitrate (NO3ˉ), nitrite (NO2ˉ), and phosphatewere analyzed using a continuous flow analyzer (Skalar San++, Netherlands). The concentrations of total dissolved nitrogen (TDN) and total dissolved phosphorus (TDP) were determined using alkaline potassium persulfate digestion (Ebina et al., 1983), followed by spectrophotometric analysis similarly to the determination ofandDON and DOP were determined as the differences between TDN and inorganic Nbetween TDP and inorganic Prespectively.The C:N:P molar ratios were obtained as DOC:DON:DOP from the bulk dissolved fraction.

2.4　Phytoplankton derived carbon, nitrogen and phosphorus composition

To estimate the contribution of phytoplankton(mainly composed of cyanobacterial colonies) to total POM, C, N and P compositions in the collected phytoplankton were determined as described above.The dry weight of phytoplankton was calculated based on a linear equation describing the relationship between the dry weight of the phytoplankton and the Chlafor Taihu Lake:

C, N and P derived from phytoplankton were estimated by the dry weight of phytoplankton,multiplied by the relative contents of phytoplankton C, N, and P. The proportion of phytoplankton derived organic carbon (phyC), nitrogen (phyN) and phosphorus (phyP) in total POM was calculated by the contents of C, N, and P of phytoplankton divided by the concentrations of POC, PN, and PP respectively.

2.5　Data analysis

Relationships between POM, DOM and water quality variables such as Chla, temperature, pH, DO,inorganic nitrogen and phosphate were analyzed using Pearson’s correlation and regression analyses.The correlations were considered statistically significant at 95% confi dence intervals (P＜0.05).Redundancy analyses were performed using the vegan package in R software to identify the major water quality variables affecting C:N:P ratios in POM and DOM.

3　RESULT

3.1　Seasonal changes in concentrations of Chl a and phytoplankton derived C, N and P

The Chlaconcentration grew in fluctuations from winter to late autumn, with the median values at the total eight stations reaching two peaks at 97.7 μg/L in July and 45.3 μg/L in October, and subsequently dropping to 10.5 μg/L in November, which was similar to the level measured at the beginning of the sampling in December (8.2 μg/L) (Fig.2a).

The estimated proportion of phytoplankton (mainly composed of cyanobacterial colonies) derived carbon(phyC), nitrogen (phyN) and phosphorus (phyP) in total POM varied from 15% to 33%, 20% to 45%, 3%to 24% respectively. PhyC and phyN peaked in May,while phyP peaked in July (Fig.3).

Fig.2　Temporal variations in chlorophyll a (Chl a) (a), particulate organic carbon (POC) (b), particulate nitrogen (PN) (c),particulate phosphorus (PP) (d), dissolved organic carbon (DOC) (e), dissolved organic nitrogen (DON) (f), dissolvedorganic phosphorus (DOP) (g), and POC: Chl a ratio (h) from December 2014 to November 2015 in Taihu Lake Red line, black square and box indicated median, mean, and 25%-75% values at eight stations, respectively. Diamond indicated outliers and whisker indicated the maximum and minimum values.

Fig.3　Estimated proportion of phytoplankton (mainly composed of cyanobacterial colonies) derived organic carbon (phyC), nitrogen (phyN), and phosphorus(phyP) in total POMBlack square and error bar represent average values and standard deviations at eight stations in Taihu Lake, respectively.

3.2　Dynamics of C, N, P in POM

Although the concentrations of POC, PN, and PP varied widely among different sampling sites, changes in POC were strongly coincident with PN and PP. All POC, PN and PP concentrations in December were extremely low (i.e., median values of 3.05 mg/L,0.35 mg/L, and 0.11 mg/L, respectively) (Fig.2b-d).However, coinciding with the increases and peak values of Chla, all POC, PN and PP values also increased, with two peaks in July and October. A respective of 5-fold, 8.4-fold, and 1.9-fold increase in POC, PN and PP, which corresponded to ?POC, ?PN and ?PP values of approximately 12.31 mg/L,2.59 mg/L, and 0.10 mg/L, respectively, was observed by comparison of the maximal median value in July to that in December.

Fig.4　Temporal variations in molar C:N:P elemental ratios of particulate and dissolved pools of organic matter collected during the study periods in Taihu LakePOC:PP (a), POC:PN (b), PN:PP (c), DOC:DOP (d), DOC:DON (e), DON:DOP (f). Red line, black square and box indicated median, mean, and 25%-75% values at eight stations, respectively. Diamond indicated outliers and whisker indicated the maximum and minimum values.

The molar POC:PN ratios fluctuated between 5.4 and 12.5, with a mean value of 8.5 (±1.6); this was slightly higher than the Redfi eld C:N ratio of 6.6.However, the molar POC:PP ratios spanned a range of nearly 20 times, with an overall mean of 122.1(±59.8). The mean molar PN:PP ratio was 15.5 (±9.8).Nonetheless, neither the POC:PP nor the PN:PP ratios were statistically different from the corresponding Redfi eld ratios. The seasonal average molar POC:PP ratios and PN:PP ratios were in the range of 79 to 187 and 8 to 28, respectively, and they exhibited similar seasonal trends (Fig.4a, c). That is, from December to May, the median molar POC:PP and PN:PP ratios were consistently lower than 100 (except January)and 13, respectively, and then increased along with bloom development (as indicated by the increased Chlaconcentrations). The median values of POC:PP and PN:PP showed two peaks in July and October,which coincided with the Chlaconcentration peaks;by contrast, the median POC:PN ratio decreased gradually from 10 in December to 7 in July, and then began to increase (Fig.4b).

3.3　Dynamics of C, N, P in DOM

In contrast to the wide spatial variation in POM,the concentrations of DOM among the different sampling sites were quite similar to each other.However, the temporal changes of the DOC, DON,and DOP concentrations were quite varied (Fig.2e-g).The DOC concentrations increased significantly from winter to summer and reached maximal value in July,which was broadly synchronous with the increase and maximum of Chla. The ?DOC was approximately 13.8 mg/L, which corresponded to a 3.5-fold increase of DOC concentrations. However, during the study period, DON tended to fluctuate approximately 0.6 mg/L, whereas the DOP concentrations fluctuated approximately 0.02 mg/L during most months, with three peaks in January, April, and September.

On average, the molar DOC:DON:DOP ratio of Taihu Lake was 832:52:1, which was substantially higher than the Redfi eld ratio (106:16:1). Both molar DOC:DOP and molar DOC:DON ratios increased greatly from spring to summer and peaked in July,reaching maximal average values of 2 175 and 30,respectively (Fig.4d-f). This coincided with the maximal Chlaconcentrations. However, the temporal changes in DON:DOP were quite different from the temporal changes in DOC:DON and DOC:DOP.There were significant increases in the DON:DOP ratio in February, August and November, and the latter two months occurred directly after the increase of Chla.

Fig.5　Correlations among C, N, P in POM and DOM, and scatterplots of Chl a and POC, PN, PP, DOC, DON, and DOPNote logarithmic scale of POC, PN, PP and Chla. Samples in winter, spring, summer and autumn were indicated by solid square, open circle, solid up triangle, and open down triangle, respectively.

3.4　Relationships between POM, DOM, and water quality variables

The POC:Chlaratio (mg/mg) varied seasonally and spanned from 69 to 692 (Fig.2h). The POC:Chlaratio peaked in December (with a median value of 434), decreased thereafter until April, and then increased gradually from May to September. The median POC:Chlaratio displayed a pattern of spring＜summer＜ autumn＜ winter, with median values in the ranges of 115-146, 131-165, 166-212, and 246-434,respectively.

Regression analyses revealed that significant relationships (P＜0.001 in all regressions) were observed for POC versus PN, POC versus PP, and PN versus PP withR2values of 0.98, 0.74, and 0.73,respectively (Fig.5a-c). Despite this, significant correlation relationships (P＜0.001 in all regressions)were also observed for POC versus Chla(R2=0.73),PN versus Chla(R2=0.76), and PP versus Chla(R2=0.55) (Fig.5g-i). No significant relationships were found for DOC versus DON and DOC versus DOP (P＞0.05), while only a weak correlation was observed for DON versus DOP (R2=0.007,P=0.004)(Fig.5d-f). There was a significant correlation between DOC and Chla(R2=0.30,P＜0.001), but DON and DOP were not correlated with Chla(P＞0.05) (Fig.5j-l).

Results of the redundancy analysis illustrated that the measured water quality variables explained 34%of the variations of C:N:P ratios in POM. Chlawas the most significant explanatory variable (adjustedR2=0.21,P=0.002) (Fig.6a). The water quality variables explained 30.4% of the variations of C:N:P ratios in DOM. Temperature was the most significant explanatory variable (adjustedR2=0.14,P=0.002),followed by NO3ˉ (adjustedR2=0.04,P=0.036) and pH(adjustedR2=0.03,P=0.028) (Fig.6b).

Fig.6　Redundancy analysis biplot showing composition of C:N:P ratios of POM (a) and DOM (b) in relation to water quality variables

Fig.7　Partitioning of DOC in total organic carbon in Taihu LakeRed line, black square and box indicated median, mean, and 25%-75% values at eight stations, respectively. Diamond indicated outliers and whisker indicated the maximum and minimum values.

3.5　Partitioning of organic carbon stocks

In Taihu Lake, both the POC and DOC increased from low values in winter to their maximal values in July. These increases occurred with the increasing Chlaconcentration. The ?POC was almost equal to?DOC, indicating that the cyanobacterial bloom contributed relatively equally to POC and DOC.Aside from the two lowest DOC:(DOC+POC) ratios of 5% and 10% in the samples (which had extremely high POC values, as described in 3.2), most of the DOC:(DOC+POC) ratios ranged widely from 14% to 85% (Fig.7), indicating large spatial-temporal variance. However, the seasonal median values of the DOC:(DOC+POC) ratio were in the range of 40% to 76%, among which only the value in October was lower than 50%. Overall, the partitioning of total organic C into DOC was a little higher than into POC,with overall median values of 62%.

4　DISCUSSION

During the study period, the Chlaconcentration increased from winter to late autumn, with median values of the eight sampling sites varying from 8.2 to 97.7 μg/L. According to the threshold of visible surface cyanobacterial blooms in Taihu Lake, defi ned as Chlaconcentration 20 μg/L (Qin et al., 2015), the Chlaconcentrations we determined verified that cyanobacterial blooms occurred from April through October, and were most serious in July.

4.1　POM concentrations and contribution of phytoplankton to total POM

The total mean POC concentration in Taihu Lake was similar to those of the shallow lake Kyoga and Prairie Lake 885 (Hecky et al., 1993), which also have high algal biomass. The changes in POC, PN and PP were significantly correlated to each other,indicating the tight coupling of C, N, and P in POM.Moreover, all of their concentrations were significantly correlated with Chla, which suggested that the POM pool was mainly inf l uenced by phytoplankton cell constituents during the study period. In addition, the sharp increase of POC concentrations in summer and autumn (July and October) coincided with the peak values of Chla, indicating the cyanobacterial blooms contributed greatly to the increase in POC.

The POC to Chla(POC:Chla) ratio, which represents the relative proportion of particulate detritus, is also usually used as a marker of the quality of particulate organic matter (Parsons et al., 1977).That is, a POC:Chlavalue lower than 200 indicates a greater contribution of “fresh” living algae, while a value higher than 200 is usually considered to indicate detrital or degraded organic matter (Parsons et al.,1977; Cifuentes et al., 1988). The seasonal changing trend of the POC:Chlaratios in Taihu Lake was similar to a parabola, which was similar to that observed in Lake Kinneret (Berman and Pollingher,1974), Jiaozhou Bay in China (Lü et al., 2009), and also observed during a phytoplankton bloom induced by in situ iron enrichment in the western subarctic Pacific (Yoshimura et al., 2009). In summary, newly formed organic matter of predominantly planktonic origin (seston with a POC:Chlaratio lower than 200)was found during March and September, whereas most of the detrital or degraded organic matter (seston with a POC:Chlaratio higher than 200) was observed during late autumn and winter.

It was diffcult to directly separate phytoplankton from seston including detritus, and zooplankton, etc.Here we estimated contribution of phytoplankton(mainly composed of cyanobacterial colonies) to total POM based on the C, N, and P contents of phytoplankton collected from Meiliang Bay (N2 in Fig.1) and the evaluated dry weight of phytoplankton according to Zhang (Zhang et al., 2011b). The proportion of phytoplankton derived organic matter in total POM were higher during bloom season than late autumn and winter, coinciding with that estimated based on values of the POC:Chlaratio.

4.2　C:N:P ratio in POM

The average C:N:P ratio of 122:16:1 in POM in Taihu Lake was similar to that of 137:18:1 in warm,nutrient-rich upwelling zones (Martiny et al., 2013).Because the C:N ratio in POM, which is often indicative of the predominant source of organic matter in a system, and that of phytoplankton origin ranged from 7.7 to 10.1 (Meyers, 1994), the average POC:PN ratios (8.5) in Taihu Lake also indicated the major contribution of phytoplankton to POM. This was consistent with the redundancy analysis result that Chlawas the most significant explanatory factor for the variance of C:N:P ratio in POM.

The average value of POC:PN:PP increased from 97:11:1 during winter and spring to 157:21:1 during June to October, when there were large cyanobacterial blooms. This did not correspond with seasonal changes of POC:Chla, a result similar to that observed by Hessen (Hessen et al., 2005), which was because POM was mainly composed of phytoplankton. There are three possible explanations for the lower values of POC:PP and PN:PP in winter and spring. One may be due to luxury consumption and storage of phosphate by phytoplankton in winter, as autotrophs accumulate P while photosynthetic activity is low (Hessen et al.,2005). The second possible reason was that phytoplankton proliferated rapidly along with the rising temperature during spring (Kong and Gao,2005). Fast-growing phytoplankton synthesize large amount of P-rich ribosomal RNA, allocating resources toward the production of growth machinery by reducing their N:P ratio (Elser et al., 2000; Arrigo,2005; Hillebrand et al., 2013). Lastly, the phytoplankton community was dominated by diatoms in late winter and early spring (data not shown).Diatoms generally have lower C:P, and N:P ratios,whereas cyanobacteria have higher C:P and N:P ratios(Arrigo et al., 1999; Bertilsson et al., 2003). Thus,changes in the phytoplankton community may be the third reason for the variation of the C:N:P ratios in the seston. Given that POM is a substrate for grazers (Mei et al., 2005), the nutrient ratio of POM can provide basic stoichiometric information about the food resources. Thus these POM of different quality may inf l uence community structure of zooplankton through a bottom-up manner (Elser et al., 1998; Cross et al., 2007).

4.3　DOM concentrations and contributions of cyanobacterial blooms

Taihu Lake had a strong gradient of DOC (2.1-21.1 mg/L) that varied by over 10-fold over the course of the observation period; this DOC gradient was comparable to the typical DOC concentration of 3-34 mg/L in eutrophic lakes (Thurman, 1985). The DOC concentration, which increased after winter and reached a maximum in July, when there was a dense cyanobacterial bloom, was similar to the high DOC concentrations during cyanobacterial blooms observed in Lake Soyang in Korea (Kim et al., 2000),and in Lake Kasumigaura and Lake Suwa in Japan(Hama and Handa, 1983; Fukushima et al., 1996).These results indicated that the cyanobacterial blooms introduced substantial amounts of DOC into water column. However, in contrast to the synchronous increase of POM and DOC along with Chla, both DON and DOP fluctuated independently. In the case of a freshwater lake, much of DON is often from terrestrial leaching and runoff, consisting mainly of humic substances and atmospheric deposition(Berman and Bronk, 2003). DOM from surrounding rivers may also contribute to the DOM in Taihu Lake(Zhang et al., 2011a). Furthermore, the fact that both DON and DOP can be assimilated and utilized by cyanobacteria under certain conditions (Glibert et al.,2004; Davis et al., 2010; Shi et al., 2011) may have also led to their fluctuations.

4.4　C:N:P ratio in DOM

On average, the C:N:P ratio of 832:52:1 for bulk DOM in Taihu Lake was lower than that of the ultraoligotrophic freshwater Lake Puma Yumco located in the pre-Himalayas of Tibet in China(2100:140:1) (Mitamura et al., 2003) and the mesotrophic Lake Biwa (1978:147:1) (Kim et al.,2006). However, it was higher than the average ratio of C:N:P in the DOM (374:22:1) of the N-limited,cyanobacteria-dominated East/Japan Sea (Kim and Kim, 2013). This suggests that the C:N:P ratios in DOM varied among different ecosystems.

Both C:P and C:N increased from spring until July,reaching maximum average values of 2 175 and 30,respectively. These maxima coincided with the maximum Chlaconcentration, suggesting that the newly produced DOM was quite C-rich and defi cient in both nitrogen and phosphorus (Norrman et al.,1995; Williams, 1995; S?ndergaard et al., 2000).Among the environmental factors determined,temperature, NO3ˉ, and pH contributed to the variance of C:N:P in DOM. Given that temperature greatly affects carbon excretion from cyanobacteria (Zlotnik and Dubinsky, 1989), the rising temperature leading to increased DOC may be one reason for the high C:P and C:N ratios in DOM. NO3ˉ may also inf l uence the metabolism of cyanobacteria, thereby affecting DOC excretion (Huang et al., 2007). Moreover, DOM and NO3ˉ are closely coupled (Zhang et al., 2014), whereas pH inf l uences the mineralization of DOM (Roth et al.,2013). Thus, both NO3ˉ and pH had a role in DOM composition. Furthermore, in situ preferential remineralization of the N- and/or P-rich compounds(Hopkinson et al., 1997, 2002) may be another reason for the high C:N:P ratios in the bulk DOM.

However, as the cyanobacterial bloom proceeded,there was a decrease of both DOC:DON and DOC:DOP after their maximum values in July.Similar to this study, substantial evidence from previous studies (and the references therein) suggests that production of the DOC fraction may be exceeding decomposition during summer bloom months, while verse visa in the autumn (Williams, 1995). Because cyanobacteria-derived DOC is biodegradable and is preferentially utilized by bacteria (Ye et al., 2015), the decreased C:N and C:P ratios of DOM from summer to autumn in Taihu Lake indicated this DOM appears to be rapidly assimilated. Thus, this autochthonous organic matter appears to be an important carbon and energy source for the food web.

4.5　Partitioning of POC and DOC in Taihu Lake

In Taihu Lake, the partitioning of total organic carbon into DOC varied from 40% to 76% and tended to fluctuate by approximately 62%. Thus, the partitioning of total organic carbon into DOC was slightly higher than that of POC. This is unlike the Ross Sea with occurrence ofPhaeocystisblooms or deck incubations of coastal diatom blooms in Oregon,in which 89% and 78% of the accumulated organic matter were partitioned into POM, respectively(Carlson et al., 2000; Wetz and Wheeler, 2003).However, it was similar to those in Lake Soyang in Korea, a deep reservoir with cyanobacterial blooms,in which monthly DOC loading comprises more than 56% of the TOC loading (Kim et al., 2000). In addition, along with the increase of Chlaconcentration from a low level in winter to a maximum level in July,?POC was comparable to ?DOC, indicating that the cyanobacterial bloom contributed nearly equally to POC and DOC. These results further revealed the inf l uence of phytoplankton community structure on the differential partitioning of organic carbon within various water systems (Carlson et al., 2000).

5　CONCLUSION

Our fi ndings that POC, PN, PP, and DOC were significantly and positively correlated with Chlaindicated the substantial contribution of cyanobacterial blooms to POM and DOC in Taihu Lake. Moreover,the partitioning of total organic carbon into DOC was slightly higher than POC. Compared to winter and spring, elevated C:P and N:P ratios in POM and C:P and C:N ratios in DOM were observed during the dense cyanobacterial blooms in July. These ratios were representative of the algal production of C-rich POM and DOM. However, their subsequent decreases indicated that these C-rich POM and DOM can be easily assimilated. Furthermore, redundancy analysis revealed that Chlaexplained most of the variations of C:N:P ratios in POM, whereas temperature, NO3ˉ, and pH were significant explanatory factors for the variations of C:N:P ratios in DOM. The present study provides basic information on the food web, and may be helpful for understanding ecosystem structure in these eutrophic lakes. Future studies are needed of the mechanisms as to how the cyanobacterial bloom derived organic matter is cycled in the food web.

6　DATA AVAILABILITY STATEMENT

The data that support the fi ndings of this study are available from the corresponding author upon reasonable request.

7　ACKNOWLEDGEMENT

We thank the anonymous reviewers for their valuable comments in the previous version of this manuscript. We thank Taihu Laboratory for Lake Ecosystem Research (TLLER) for providing the monitoring data of phytoplankton. We appreciate CHEN Chao for his help for sample collection in the fi eld.