Harmful Algae 64 (2017) 42–50

Targeted deep sequencing reveals high diversity and variabledominance of bloom-forming cyanobacteria in eutrophic lakes

Yongguang Jiang a,1, Peng Xiao a,1, Yang Liu b, Jiangxin Wang a, Renhui Li c,*a Shenzhen Key Laboratory of Marine Bioresource and Eco-environmental Science, College of Life Sciences and Oceanography, Shenzhen University, Shenzhen

518060, PR Chinab College of Life Sciences, Henan Normal University, Xinxiang 453007, PR Chinac Key Laboratory of Algal Biology, Institute of Hydrobiology, Chinese Academy of Sciences, Wuhan 430072, PR China

A R T I C L E I N F O

Article history:

Received 29 December 2016

Received in revised form 27 February 2017

Accepted 6 March 2017

Keywords:

Cyanobacterial bloom

Diversity

cpcBA-IGS

Pyrosequencing

Eutrophication

A B S T R A C T

Cyanobacterial blooms in eutrophic lakes are severe environmental problems worldwide. To

characterize the spatiotemporal heterogeneity of cyanobacterial blooms, a high-throughput method

is necessary for the specific detection of cyanobacteria. In this study, the cyanobacterial composition of

three eutrophic waters in China (Taihu Lake, Dongqian Lake, and Dongzhen Reservoir) was determined

by pyrosequencing the cpcBA intergenic spacer (cpcBA-IGS) of cyanobacteria. A total of 2585 OTUs were

obtained from the normalized cpcBA-IGS sequence dataset at a distance of 0.05. The 238 most abundant

OTUs contained 92% of the total sequences and were classified into six cyanobacterial groups. The water

samples of Taihu Lake were dominated by Microcystis, mixed Nostocales species, Synechococcus, and

unclassified cyanobacteria. Besides, all the samples from Taihu Lake were clustered together in the

dendrogram based on shared abundant OTUs. The cyanobacterial diversity in Dongqian Lake was

dramatically decreased after sediment dredging and Synechococcus became exclusively dominant in this

lake. The genus Synechococcus was also dominant in the surface water of Dongzhen Reservoir, while

phylogenetically diverse cyanobacteria coexisted at a depth of 10 m in this reservoir. In summary,

targeted deep sequencing based on cpcBA-IGS revealed a large diversity of bloom-forming cyanobacteria

in eutrophic lakes and spatiotemporal changes in the composition of cyanobacterial communities. The

genus Microcystis was the most abundant bloom-forming cyanobacteria in eutrophic lakes, while

Synechococcus could be exclusively dominant under appropriate environmental conditions.

� 2017 Published by Elsevier B.V.

Contents lists available at ScienceDirect

Harmful Algae

jo u rn al h om epag e: ww w.els evier .c o m/lo cat e/ha l

1. Introduction

Cyanobacteria are photosynthetic microorganisms adaptable tovarious terrestrial and aquatic environments (de Marsac andHoumard, 1993; Whitton, 2012). Water blooms caused by themassive proliferation of cyanobacteria in freshwater ecosystemshave drawn considerable attention around the world (Harke et al.,2016; Paerl and Otten, 2013). Eutrophication of water bodies is themain factor promoting the development of cyanobacterial bloomin the past decades (Bonilla et al., 2012; Gobler et al., 2016; O’Neilet al., 2012). Recently, climate warming and rising atmosphericcarbon dioxide have significantly increased the occurrence,

* Corresponding author at: Institute of Hydrobiology, Chinese Academy of

Sciences, Wuhan 430072, Hubei Province, PR China.

E-mail address: reli@ihb.ac.cn (R. Li).1 These authors contributed equally to this work.

http://dx.doi.org/10.1016/j.hal.2017.03.006

1568-9883/� 2017 Published by Elsevier B.V.

intensity and duration of cyanobacterial blooms (Deng et al.,2014; Paerl and Huisman, 2009; Sinha et al., 2012; Visser et al.,2016; Wiedner et al., 2007). Certain species of bloom-formingcyanobacteria are capable of producing toxic metabolites, such ashepatotoxic microcystins and cylindrospermopsins and neurotoxicsaxitoxins and anatoxins (Dittmann et al., 2013; Neilan et al., 2013;Pearson et al., 2010). Therefore, harmful cyanobacterial blooms areof serious concern for public health due to the toxic effects ofcyanotoxins (Havens, 2008; Paerl and Otten, 2013). Thus, dailymonitoring of cyanobacterial composition in freshwater bodies isimportant for predicting and controlling harmful cyanobacterialblooms to ensure drinking water safety.

Aquatic cyanobacterial community is usually composed ofhighly diverse cyanobacterial species with some of them dominat-ing in the community (Kim et al., 2006; Liu et al., 2016b; Miller et al.,2013). The most prevalent bloom-forming species are Synechococ-

cus, Microcystis, Dolichospermum, Aphanizomenon, Cylindrospermop-

sis, and Planktothrix. Although morphological characteristics are

Y. Jiang et al. / Harmful Algae 64 (2017) 42–50 43

widely used in traditional identification of cyanobacteria in watersamples, extensive training is required and identification procedureis also labor-consuming considering the high diversity of cyano-bacterial morphologies (Whitton, 2012). Besides, microscopicalidentification is much difficult for small-sized picocyanobacteria(e.g., Synechococcus), which play an important role in freshwaterecosystems (Callieri, 2008). To solve this problem, molecular toolshave been developed to efficiently investigate cyanobacterialdiversity. Several housekeeping loci on the cyanobacterial genomeare selected as molecular markers, such as 16S rRNA gene (Nubelet al., 1997), RNA polymerase gene (rpoC1) (Palenik and Swift, 1996),intergenic spacer between phycocyanin subunit genes cpcB and cpcA

(cpcBA-IGS) (Neilan et al., 1995), and internal transcribed spacerbetween the 16S rRNA and 23S rRNA genes (ITS) (Huang et al.,2014). In particular, the 16S rRNA gene and cpcBA-IGS were provenuseful for phylogenetic analysis of cyanobacteria at differenttaxonomic levels (Dadheech et al., 2010; Dojani et al., 2014; Dybleet al., 2002; Ishida et al., 2001; Robertson et al., 2001; Teneva et al.,2005; Wang et al., 2013). PCR amplification of partial 16S rRNA genefollowed by denaturing gradient gel electrophoresis (DGGE) canprovide a general overview of the cyanobacterial diversity inenvironmental samples (Boutte et al., 2006; Ye et al., 2011). Theclone libraries of cpcBA-IGS or 16S rRNA gene can also beconstructed and sequenced to roughly reveal cyanobacterialcomposition (Dojani et al., 2014; Kim et al., 2006). However, onlylimited information regarding cyanobacterial community could beobtained by these methods.

In recent years, high-throughput sequencing has been widelyused for detailed and accurate understanding of microbialcommunity structure in ecosystems by producing a large sequencedataset of molecular markers (Cole et al., 2005). For example,Kleinteich et al. (2014) and Zhang et al. (2016) investigated thecyanobacterial composition in the meltwater ponds and Gurban-tunggut Desert, respectively, by pyrosequencing the 16S rRNAgene. Unfortunately, their sequence datasets were contaminated

Fig. 1. Map of three lakes showing s

by abundant bacterial sequences although the primers used toamplify the 16S rRNA gene were considered to be cyanobacteria-specific (Kleinteich et al., 2014; Zhang et al., 2016). In anotherstudy, plastid 23S rRNA gene was applied for pyrosequencinganalysis of phytoplankton in eutrophic lakes, but cyanobacterialpopulations can only be delineated at order-level unsuitable foraccurate identification of bloom-forming species (Steven et al.,2012). Phycocyanin is an exclusive compound of cyanobacteria,and thus genes related to phycocyanin, for instance, cpcBA-IGScould be a more specific and reliable molecular marker in deepsequencing compared with 16S rRNA gene. The purpose of thisstudy is to provide a comprehensive understanding of the cyano-bacterial community in freshwater ecosystems suffering annualcyanobacterial blooms. Therefore, pyrosequencing method based oncpcBA-IGS was developed to reveal cyanobacterial diversity andcomposition in three eutrophic lakes at molecular level.

2. Materials and methods

2.1. Collection of cyanobacterial samples

Taihu Lake is a large shallow freshwater lake in eastern Chinaand located in a temperate region (308560–31833 N, 1198530–1208360 E; water area = 2338 km2; mean depth = 1.9 m). DongqianLake is also a shallow freshwater lake in eastern China (298440–298470 N, 1218370–1218410 E; water area = 20 km2; meandepth = 2.2 m). A sediment dredging project was conducted inDongqian Lake from 2009. Dongzhen Reservoir is a deep reservoirin southeastern China and located in a subtropical region (258280–258300 N, 1188540–1188590 E; water area = 17.8 km2; maximumdepth = 36 m). All these three lakes have been suffering fromeutrophic problems and annual cyanobacterial blooms (Guo, 2007;Jing et al., 2013; Lv et al., 2013). Surface water (0–0.5 m) wassampled at 9 locations, 5 in Taihu Lake, 2 in Dongqian Lake and 2 inDongzhen Reservoir (Fig. 1). Water samples at 10 m depth in

ampling locations in this study.

Table 1Detailed information of samples and number of robust reads.

Sample number Sampling site Sampling time Robust reads

THN11101 Taihu Lake N1 Jan. 2011 9567

THN11105 Taihu Lake N1 May 2011 8491

THN11108 Taihu Lake N1 Aug. 2011 6447

THN11111 Taihu Lake N1 Nov. 2011 7369

THN21101 Taihu Lake N2 Jan. 2011 7037

THN21105 Taihu Lake N2 May 2011 12,140

THN21108 Taihu Lake N2 Aug. 2011 15,110

THN21111 Taihu Lake N2 Nov. 2011 8430

THW1101 Taihu Lake W Jan. 2011 9611

THW1105 Taihu Lake W May 2011 7425

THW1108 Taihu Lake W Aug. 2011 10,444

THW1111 Taihu Lake W Nov. 2011 11,623

THS11101 Taihu Lake S1 Jan. 2011 11,759

THS11105 Taihu Lake S1 May 2011 10,666

THS11108 Taihu Lake S1 Aug. 2011 6314

THS11111 Taihu Lake S1 Nov. 2011 11,494

THS21101 Taihu Lake S2 Jan. 2011 7386

THS21105 Taihu Lake S2 May 2011 5941

THS21108 Taihu Lake S2 Aug. 2011 8836

THS21111 Taihu Lake S2 Nov. 2011 8662

DQN0907 Dongqian Lake S Jul. 2009 8066

DQN1107 Dongqian Lake S Jul. 2011 11,856

DQS0907 Dongqian Lake N Jul. 2009 8296

DQS1107 Dongqian Lake N Jul. 2011 8316

DZWS1207 Dongzhen Reservoir W Jul. 2012 9129

DZWD1207 Dongzhen Reservoir W, 10 m depth Jul. 2012 4488

DZES1207 Dongzhen Reservoir E Jul. 2012 11,661

DZED1207 Dongzhen Reservoir E, 10 m depth Jul. 2012 6709

Y. Jiang et al. / Harmful Algae 64 (2017) 42–5044

Dongzhen Reservoir were also collected. Sampling was performedquarterly in Taihu Lake in 2011 and in Dongqian Lake from 2009 to2011, and once in Dongzhen Reservoir in July 2012. A volume of300 mL water was filtered through a 0.45 mm filter from whichtotal DNA was extracted subsequently. The samples used forpyrosequencing were numbered according to sampling locationsand time as depicted in Table 1. Before further processing, all thefilters were preserved at �80 8C. Analysis of water qualityparameters, such as total nitrogen (TN) and total phosphorus(TP), were conducted as previously described (Wu et al., 2006). Thewater transparency was represented by Secchi depth (SD)measured by use of a black and white Secchi disk (Cole, 1994).

2.2. DNA preparation, 454 pyrosequencing and sequence quality

control

Genomic DNA were extracted from cyanobacterial strains andfilters with a modified CTAB method involving phenol-chloroformextraction (Porebski et al., 1997). The filters were cut into piecesfirst before subjected to the extraction process. Purified DNAsamples were dissolved in TE buffer (pH 8.0) and stored at �20 8C.A pair of primers, PCbF (50-GGCTGCTTGTTTACGCGACA-30) target-ing cpcB locus and PCaR454 (50-GCTTCGGTRAKKGGRGTTTTCAT-30) targeting the 50 end of cpcA locus was used for amplification ofthe cyanobacterial cpcBA-IGS sequence. PCbF was a forwardprimer designed previously (Neilan et al., 1995) and PCaR454 wasa reverse degenerate primer designed in this study based oncyanobacterial cpcBA-IGS sequences deposited in GenBank data-base (Figure S1, in the supplementary data). The primer sequenceswere checked for homology to bacterial and algal genomicsequences using BLAST on the website of the National Centerfor Biotechnology Information. The specificity of these primers wasfurther examined by PCR using the genomic DNA templates fromvarious cyanobacterial strains belonging to Chroococcales, Nosto-cales, and Oscillatoriales (Table S1, in the supplementary data).Sterile water and the genomic DNA of a cryptomonad strainCryptomonas obovata FACHB-1301 and an aquatic bacterium

Novosphingobium sp. THN1 (Jiang et al., 2011) were used asnegative controls in PCR amplification.

For 454 pyrosequencing, reverse barcoded primers containingadapter sequence were designed based on PCaR454 and synthe-sized (Table S2, in the supplementary data). PCR reactions wereprepared in triplicate with 50 mL volumes by adding 25 mL of 2�Gflex PCR Buffer (Mg2+, dNTP plus), 10 pmol of each primer, 100 ngof genomic DNA and 1.25 U of Tks Gflex DNA polymerase (Takara,Japan). Thermal cycling was performed at 94 8C for 3 min, followedby 30 cycles of 98 8C, 10 s; 54 8C, 30 s; 68 8C, 1 min; and finally onecycle of 68 8C, 10 min. The PCR products were analyzed on a 1%agarose gel stained with ethidium bromide. The target bands wereexcised and purified using a gel extraction kit (MiniBEST AgaroseGel DNA Extraction Kit Ver.4.0, Takara). Triplicate products fromeach environmental DNA sample were pooled to minimize randomPCR bias. High-throughput pyrosequencing was performed usingthe GS-FLX Titanium platform according to the manufacturer’sprotocols (Roche 454 Life Sciences, Branford, USA). Low-qualityreads were removed as described previously (Huber et al., 2007).Barcode and primer sequences were trimmed from the 50 end, andthe reads with sequence errors in their barcode and/or primerregions were discarded. The sequences with a length less than200 bp were all excluded and the chimeras were removed usingUchime algorithm (Edgar et al., 2011). The remaining robustsequences were used for further analyses. Tag sequences have beendeposited in the European Bioinformatics Institute Sequence ReadArchive under accession number PRJEB15368.

2.3. Sequence analysis

The robust sequences were de-replicated to create a simplifieddata set containing unique sequences and the amount of eachunique sequence was recorded. To improve the efficiency andaccuracy of operational taxonomic unit (OTU) classification for thelarge data set, a strategy of template and repeat alignment wasused. In detail, (i) representative OTU sequences were firstidentified by Uparse (Edgar, 2013) at a distance of 0.05;

Table 2Cyanobacterial diversity indices based on OTUs of cpcBA-IGS. eH0 = natural

exponential of Shannon index (H0), D�1 = inverse of Simpson index (D), E = Simpson’s

evenness index.

Sample OTU richness Coverage Chao 1 index eH0 D�1 E

THN11101 250 0.97 457 31 14 0.06

THN11105 220 0.98 424 19 6 0.03

THN11108 271 0.97 530 26 9 0.03

THN11111 239 0.97 535 30 12 0.05

THN21101 244 0.97 388 37 16 0.06

THN21105 243 0.97 581 19 5 0.02

THN21108 143 0.98 270 5 2 0.01

THN21111 201 0.98 351 20 8 0.04

THW1101 282 0.97 584 47 24 0.09

THW1105 142 0.98 269 10 5 0.03

THW1108 236 0.98 417 38 17 0.07

THW1111 220 0.98 354 18 8 0.04

THS11101 204 0.98 397 20 10 0.05

THS11105 147 0.99 226 12 6 0.04

THS11108 217 0.97 527 28 14 0.07

THS11111 262 0.97 638 45 16 0.06

THS21101 268 0.97 539 27 10 0.04

THS21105 243 0.97 493 23 9 0.04

THS21108 220 0.98 343 28 10 0.05

THS21111 217 0.98 366 20 8 0.04

DQN0907 346 0.96 638 64 21 0.06

DQN1107 234 0.98 407 28 13 0.05

DQS0907 462 0.95 845 110 48 0.10

DQS1107 180 0.98 278 26 11 0.06

DZWS1207 245 0.98 355 47 23 0.10

DZWD1207 336 0.97 558 84 40 0.12

DZES1207 163 0.98 277 24 10 0.06

DZED1207 346 0.96 731 71 31 0.09

Y. Jiang et al. / Harmful Algae 64 (2017) 42–50 45

(ii) alignment of these representative sequences was generated byMuscle 3.8.31 (Edgar, 2004) and used as a template to create analignment of all unique sequences by Mothur v.1.33.0 (Schlosset al., 2009); (iii) the alignment was trimmed to the length of theshortest cpcBA-IGS sequence and used to identify representativeOTU sequences for the second time by Mothur at the same distanceof 0.05; (iv) step ii was repeated using the representativesequences obtained in step iii and a new alignment of all uniquesequences was generated for further analyses. The OTUs of cpcBA-IGS were finally defined by the average neighbor clusteringalgorithm at a distance of 0.05. The sequencing depth of thesamples were normalized to the lowest one among them (4488 inDZWD1207) by subsampling implemented in Mothur. Afterward,the alpha diversity indices were estimated as follows: OTUrichness, Coverage, Chao 1 index, Shannon index, Simpson index,and Simpson’s evenness index. Rarefaction values were generatedin Mothur after sample normalization and were imported intoMicrosoft Excel for generation of rarefaction curves.

Euclidean distance between samples were calculated based onthe abundance of shared abundant OTUs. A heatmap wasgenerated by R 3.0.2 (http://www.r-project.org, accessed 2015)using the pheatmap package. The color displaying OTU size wasrepresented by a value calculated using the formula lg(x + 0.1)where x is the OTU size.

2.4. Phylogenetic analysis

For phylogenetic analysis of cyanobacterial species, a data setconsisting of cpcBA-IGS sequences from 58 cyanobacterial strainsof 13 genera was constructed. Multiple sequence alignment wasperformed using the ClustalW v1.4 program and manuallycorrected. Neighbor-joining tree was created by the MEGA5software (Tamura et al., 2011) using the Kimura 2-parametermodel with 1000 bootstrap replicates. For phylogenetic assign-ment of each OTU, a neighbor-joining tree was created using a dataset consisting of representative cpcBA-IGS sequences from238 abundant OTUs (OTU size �45) and reference cyanobacterialspecies aforementioned. Each OTU was classified into the closestcyanobacterial genus in the phylogenetic tree.

3. Results

3.1. Phylogenetic analysis of cyanobacteria based on cpcBA-IGS

A phylogenetic tree of cyanobacteria was constructed using 58cpcBA-IGS sequences derived from representative cyanobacterialspecies (Figure S1, in the supplementary data). These species wereclassified into nine groups, including the bloom-forming Synecho-

coccus, Microcystis, Cylindrospermopsis/Raphidiopsis, mixed Nosto-cales species, Planktothrix, mixed Osillatoriales species, Spirulina/

Arthrospira, Pseudanabaena, and Synechocystis.

3.2. Pyrosequencing of cpcBA-IGS

A primer pair of PCbF and PCaR454 was used to amplify thecyanobacterial cpcBA-IGS fragment of approximately 372 bp(determined from M. aeruginosa PCC 7806, GenBank accessionno. AM778951). No effective matches exist between primersequences and genomic sequences of bacteria and algae. Theprimers were confirmed to be efficient using DNA templates fromcyanobacterial strains of Chroococcales, Nostocales, and Oscilla-toriales. No targeted amplicons were observed in the negativecontrols. Twenty-eight filter samples were collected from ninesites in three lakes from 2009 to 2012 (Table 1). High-qualitycpcBA-IGS fragments were successfully obtained using reversebarcoded primers and genomic DNA extracted from the filters.

After pyrosequencing and sequence processing, a total of 253,273robust cpcBA-IGS sequences, including 4488–15,110 sequencesfrom each sample, were obtained. Ninety-five percent of all thesequences have a length within 321 bp to 401 bp (Figure S2, in thesupplementary data), indicating a high reliability of the pyrose-quencing results.

3.3. Cyanobacterial diversity based on cpcBA-IGS

The diversity indices based on OTUs of cpcBA-IGS are listed inTable 2. Overall, 2585 OTUs, including 142–462 OTUs per sample,were clustered at a distance of 0.05 after normalization. Therarefaction curves were in proximity to plateau (Figure S3, in thesupplementary data), and a good coverage (>0.95) of OTU richnesswas also observed for each sample (Table 2), suggesting that thenumber of sequences reached a hypothetical saturation value. Theaverage OTU number was 305 in Dongqian Lake, 272 in DongzhenReservoir, and 223 in Taihu Lake in decreasing order. The Chao1 index values were 226–845 per sample and the average valueswere 542 for Dongqian Lake, 480 for Dongzhen Reservoir, and434 for Taihu Lake.

The Shannon and Simpson indices were transferred into theirtrue diversity parameters, eH0 and D�1, respectively, representingthe effective number of equally common OTUs with the same valueof diversity index as the original sample (Jost, 2006). The highestcyanobacterial diversity was detected in the summer sample ofDongqian Lake, DQS0907 (OTU = 462, eH0 = 110, D�1 = 48). More-over, the lowest diversity was observed in the two samples ofTaihu Lake, namely, THW1105 (OTU = 142, eH0 = 10, D�1 = 5) andTHN21108 (OTU = 143, eH0 = 5, D�1 = 2). In Taihu Lake, the diversityindices between sites or months were not significantly different (P

value >0.05, Paired sample t-test). Regarding Dongqian Lake, thediversity indices in 2011 (average OTU = 207, eH0 = 27, D�1 = 12)were lower than those in 2009 (average OTU = 404, eH0 = 87,D�1 = 35). In Dongzhen Reservoir, the diversity indices in deep

Fig. 2. Clustered heatmap of samples based on shared abundant OTUs. The letter x in the formula represents OTU size.

Y. Jiang et al. / Harmful Algae 64 (2017) 42–5046

(10 m) water (average OTU = 341, eH0 = 78, D�1 = 35) were higherthan those in surface water (average OTU = 204, eH0 = 36, D�1 = 17).The values of Simpson’s evenness index were 0.01–0.12 persample.

3.4. Comparison of OTU composition

The total abundance of each OTU from all the samples wascalculated after normalization and OTUs with a size less than 1% of4488 were excluded in the following analysis. Thus, only238 abundant OTUs containing 92% of the total sequencesremained. These OTUs were then defined according to phyloge-netic analysis of representative OTU sequences with cpcBA-IGSsequences of cyanobacterial species as reference (Figure S4, in thesupplementary data). Abundant OTUs were classified into sixgroups of cyanobacteria, including Synechococcus (98 OTUs),Microcystis (94 OTUs), mixed Nostocales species (21 OTUs),Cylindrospermopsis/Raphidiopsis (6 OTUs), Planktothrix (1 OTUs),and unclassified cyanobacteria (18 OTUs). The conservation of OTU

sequences was also verified by homologous search using the BLASToption.

As displayed in the clustered heatmap (Fig. 2), OTU compositionvaried significantly among the samples. The samples from theNorthern region of Taihu Lake (N1 and N2 sites) were more closelyrelated and divergent from the samples from the Southern regionof this lake (S1 and S2 sites). AMOVA analysis also revealed that theOTU composition in N1, N2, and S1 sites were significantlydifferent from those in S2 site (P value <0.05, Table S3 in thesupplementary data). In addition, OTU composition was notsignificantly different among the four months (P value >0.05) inTaihu Lake. DQN0907, DQS0907, and DZWD1207 samples wereclustered together with high similarities among them (96–98%,Table S4, in the supplementary data). DZED1207 sample was alsorelated to these three samples with middle-level similarities (48–63%) and highly diverse OTUs from various cyanobacterial generawere detected in these four samples. DQN1107, DQS1107,DZWS1207, and DZES1207 samples were also clustered togetherwith high similarities among them (82–99%), and the majority of

Fig. 3. Spatiotemporal alterations of cyanobacterial composition in eutrophic lakes. Only the 238 abundant OTUs (OTU size �45) were summarized. (A) Taihu Lake; (B)

Dongqian Lake; (C) Dongzhen Reservoir.

Y. Jiang et al. / Harmful Algae 64 (2017) 42–50 47

OTUs in these samples belonged to Synechococcus. By contrast, lowsimilarities were observed for samples between 2009 and 2011 inDongqian Lake (5.9–30%) and between surface and deep waterfrom Dongzhen Reservoir (0.71–18%).

3.5. Spatiotemporal changes in cyanobacterial composition

The cyanobacterial composition of each sample was calculatedbased on the sequence number of abundant OTUs (Fig. 3). In TaihuLake, Microcystis was the most abundant cyanobacterial genuswith high percentages (59–100%) in 13 of 20 samples (Fig. 3A). Thegenus Synechococcus maintained a low proportion in the samplesexcept THW1108 (41%) and THS21108 (93%). Similarly, a highpercentage was observed for the mixed Nostocales species group inthree samples, namely, THN11108 (54%), THN21111 (53%), and

THS11111 (54%). Some OTUs cannot be classified into any knowncyanobacterial genus by phylogenetic or BLAST methods and theywere denominated as unclassified cyanobacteria with a percentageof 0.82–88% in several samples of Taihu Lake. In addition,Planktothrix constituted only a minor percent of total cyanobac-teria (0.25–0.75%) in six samples of Taihu Lake.

As displayed in Fig. 3B, similar cyanobacterial composition wasobserved for the two 2009 samples (DQN0907 and DQS0907) fromDongqian Lake. The cyanobacterial composition of the two deep-water samples (DZWD1207 and DZED1207) from DongzhenReservoir was also similar (Fig. 3C). All these four samples werecomposed of Synechococcus (5.3–29%), Microcystis (20–47%), mixedNostocales species (5.3–21%), Cylindrospermopsis/Raphidiopsis

(4.3–24%) and unclassified cyanobacteria (3.0–40%). A highpercentage of Synechococcus (>99%) was observed in the two

Fig. 4. Nutrient concentration (A) and Secchi depth (B) in Dongqian Lake. Error bars represent standard deviations.

Y. Jiang et al. / Harmful Algae 64 (2017) 42–5048

2011 samples from Dongqian Lake (DQN1107 and DQS1107) andthe two surface samples from Dongzhen Reservoir (DZWS1207 andDZES1207).

3.6. Water quality parameters in Dongqian Lake

TN fluctuated in a narrow concentration range of 0.70–1.1 mg L�1 from April 2009 to January 2011, while it reachedthe highest concentration in April 2011 (Fig. 4A). Afterward, TNconcentration decreased to 0.19–0.39 mg L�1 in October 2011. TPconcentration decreased from 0.11 mg L�1 to 0.04 mg L�1 from2009 to 2011 (Fig. 4A). The TN: TP ratios fluctuated during thesampling period, but the ratios in spring and summer seasonsincreased in 2011 (Fig. 4A). Fluctuations of the SD values were alsoobserved but the SD values in summer season increased in 2011(Fig. 4B). Moreover, SD values in July obviously increased from2009 to 2011. The TN, TP, and SD values in four sequencing sampleswere as follows: DQN0907, TN = 0.87 � 0.08 mg L�1, TP = 0.13 �0.02 mg L�1, TN: TP = 6.7, SD = 50 � 5.0 cm; DQS0907, TN = 1.1 �0.19 mg L�1, TP = 0.14 � 0.02 mg L�1, TN: TP = 7.6, SD = 43 � 12 cm;DQN1107, TN = 0.58 � 0.09 mg L�1, TP = 0.03 � 0.01 mg L�1, TN:TP = 17, SD = 89 � 16 cm; DQS1107, TN = 0.83 � 0.11 mg L�1,TP = 0.02 � 0.01 mg L�1, TN: TP = 45, SD = 138 � 18 cm. Approxi-mately, 23% and 33% of TN were reduced at N and S sites in July2011 compared with July 2009, while 77% and 86% of TP were reducedfrom July 2009 to July 2011. The TN: TP ratios increased in July2011 as against July 2009 by 2.5- and 5.9-fold at N and S sites,respectively. The SD values increased by 1.8- and 3.2-fold at N and Ssites, respectively, in July 2011 compared with July 2009.

4. Discussion

In this study, a targeted deep sequencing method wasdeveloped to reveal the diversity and dynamics of bloom-formingcyanobacteria based on cpcBA-IGS. With this method, thespatiotemporal heterogeneity in the composition of cyanobacterialcommunities was investigated in three eutrophic lakes in China.

According to phylogenetic analysis based on cpcBA-IGS,Cylindrospermopsis and Raphidiopsis are clustered together. Actu-ally, these two genera are closely related species with highlysimilar genome sequence (Stucken et al., 2010) and are constantlyintermixed on the phylogenetic trees (Li et al., 2016). The genusRaphidiopsis could be distinguished from Cylindrospermopsis with alonger fragment of ITS (Li et al., 2016), which may be useful for the

further exploration of the Cylindrospermopsis/Raphidiopsis group.Similarly, the intermixed cluster of mixed Nostocales species wasactually divided into five different clades, including Dolichosper-

mum, Aphanizomenon, Chrysosporum, Sphaerospermopsis, andCuspidothrix in previous phylogenetic analyses, based on cpcBA-IGS or 16S rRNA gene (Cires and Ballot, 2016; Wang et al., 2013).The difference may be caused by the relatively shorter sequenceused in the present study after multiple alignment and trimming ofcpcBA-IGS sequences from a wider range of cyanobacterial species.Therefore, cpcBA-IGS is a flexible molecular marker to investigatecyanobacterial diversity at different taxonomic levels. Osillator-iales species were clustered into four divergent clades, Planktothrix,mixed Osillatoriales species, Spirulina/Arthrospira, and Pseudana-

baena, indicating polyphyletic origin of the members of Oscilla-toriales as suggested by Ishida et al. (2001).

The large number of OTUs in the samples revealed a highdiversity of cyanobacterial populations in eutrophic lakes. For allthe samples, Chao 1 index values were larger than OTU richness.Considering that Chao 1 index is sensitive to the existence of rareOTUs, the larger values of this index thus coincided with the lowabundance of a major proportion of OTUs. The low Simpson’sevenness index values implied the heterogeneity of cyanobacterialcommunity in each sample. As displayed in Figs. 2 and 3, distinctdistribution of abundant OTUs and cyanobacterial populationsamong samples also revealed the spatiotemporal heterogeneity ofcyanobacterial communities in eutrophic lakes.

While relatively stable diversity indices were observed in TaihuLake, the cyanobacterial diversity in Dongqian Lake was obviouslyreduced from 2009 to 2011. After sediment dredging in DongqianLake from 2009, the nitrogen and phosphorus balance in thesediment–water microcosms have been disturbed (Jing et al., 2013,2015). In addition, cyanobacterial community composition wascorrelated with nitrate and phosphorus conditions in lake water(Miller et al., 2013). Therefore, changes in nutrient conditions maybe the cause of dramatic reduction of cyanobacterial diversity inDongqian Lake.

In Dongzhen Reservoir, the cyanobacterial diversity wasapparently high in deep water compared with surface water. Ina previous investigation, the water layer at 10 m was close tometalimnion in Dongzhen Reservoir, and the microbial diversitywas higher in metalimnion and hypolimnion water compared withepilimnion water in this reservoir (Yu et al., 2014). Consequently,these results emphasized the important effect of thermalstratification on cyanobacterial diversity (Miller et al., 2013).

Y. Jiang et al. / Harmful Algae 64 (2017) 42–50 49

The phylogenetic assignment of abundant OTUs demonstratedthat bloom-forming cyanobacteria, such as Synechococcus andMicrocystis, are the main cyanobacterial populations in eutrophiclakes. The high abundance of Microcystis in Taihu Lake wasconsistent with the perennial Microcystis blooms observed in thislake (Chen et al., 2003). In a previous study, a large biomass ofDolichospermum was only detected during winter and spring inTaihu Lake (Wang et al., 2013). Therefore, the dominance of mixedNostocales species in the summer sample (THN11108) of this lakemay be caused by other bloom-forming Nostocales species, such asAphanizomenon.

A much similar pattern of cyanobacterial composition wasobserved for the two 2009 samples from Dongqian Lake and thetwo deep-water samples from Dongzhen Reservoir (Fig. 3B and C),consistent with the topological structure of the clustered dendro-gram in Fig. 2. These four samples were composed of severalbloom-forming cyanobacterial populations. On the contrary,Synechococcus was completely dominant in the two 2011 samplesfrom Dongqian Lake and the two surface samples from DongzhenReservoir. Considering the variations of OTU composition in thesetwo lakes, these results together indicated that significant changeshave occurred in the cyanobacterial community structure from2009 to 2011 in Dongqian Lake, and dramatic reorganization ofcyanobacterial community existed along the depth of watercolumns in Dongzhen Reservoir.

The dominance of Synechococcus has been revealed in bothocean waters and freshwater bodies (Li, 1998; Takasu et al., 2015).Regarding the success of this species, the dramatic increase ofSynechococcus has been observed under high molar TN: TP ratios inlake water (Stockner and Shortreed, 1988). The sediment dredgingprocess could reduce nitrogen removal from lake systems overyears (Jing et al., 2013) and suppress the release of soluble reactivephosphorus from the sediment (Liu et al., 2016a). Combining theseaspects, the dominance of Synechococcus in Dongqian Lake in2011 may be caused by the increased TN: TP ratio after sedimentdredging as observed in this study. In addition, Synechococcus

includes divergent populations that were adaptable to differentlight conditions (Kulkarni and Golden, 1994; Wilhelm et al., 2006).The water transparency in Dongqian Lake in 2011 was much highercompared with that in 2009. Therefore, the dominant Synecho-

coccus populations may be more adaptable to high light conditionsas described by Kulkarni and Golden (1994). Similarly, a previousstudy observed high TN: TP ratios (>25) in the epilimnion ofDongzhen Reservoir in July 2012 (Yu et al., 2014). Thus, thedominance of Synechococcus in the surface water of this reservoircould also be ascribed to the adaptation of this genus to high TN: TPratios and high light intensities.

The transition of the dominant cyanobacterial populations inthese eutrophic lakes revealed the existence of diverse cyano-bacterial species adaptable to different environmental conditions,such as nutrients, light, and temperature. Therefore, the presentstrategy of mitigating cyanobacterial blooms through eliminationof eutrophication should be implemented by reducing bothphosphorus and nitrogen, as previously suggested (Gobler et al.,2016), particularly for dredging project that may weaken nitrogenremoval from freshwater systems (Jing et al., 2013). In addition, thedistribution of bloom-forming cyanobacteria in the deep water ofDongzhen Reservoir emphasized that different strategies for waterquality management should be adopted in shallow and deep lakes.

In conclusion, the cpcBA-IGS locus was an efficient molecularmarker for high-throughput characterization of cyanobacterialcomposition in freshwater bodies. Regarding the coverage andspecificity of molecular identification, the deep sequencingmethod based on cpcBA-IGS was more efficient than previousmolecular methods, such as DGGE, clone libraries and 16S rRNAdeep sequencing. However, a common limitation exists for all

sequencing methods that only relative concentration of eachcyanobacterial species could be obtained from the results. Acombination of deep sequencing and quantitative real-time PCRexperiments would be suitable to comprehensively elucidate thecyanobacterial community composition.

A high heterogeneity existed in the cyanobacterial communityof eutrophic lakes and they were dominated successively bydifferent bloom-forming species. The genus Microcystis was themost abundant bloom-forming cyanobacteria in eutrophic lakes,while Synechococcus, Nostocales species, and some unclassifiedcyanobacteria could also be dominant in the surface water underappropriate environmental conditions. It should be noted that thefactors driving the succession of bloom-forming cyanobacteriacannot be determined based on these results. This problem couldbe solved by considering the physiological characteristics ofdominant cyanobacterial species and the changes of environmen-tal conditions in eutrophic lakes in future studies.

Author contributions

Renhui Li designed the investigation. Yongguang Jiang and PengXiao performed the experiments and prepared the manuscript.Yang Liu collected the samples from Taihu Lake. Jiangxin Wangrevised the manuscript.

Acknowledgments

This research was supported by the National Natural ScienceFoundation of China (31500071 and 31470310) and ChinaPostdoctoral Science Foundation (2016T90801).[CG]

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.hal.2017.03.006.

References

Bonilla, S., Aubriot, L., Soares, M.C.S., Gonzalez-Piana, M., Fabre, A., Huszar, V.L.,Lurling, M., Antoniades, D., Padisak, J., Kruk, C., 2012. What drives the distribu-tion of the bloom-forming cyanobacteria Planktothrix agardhii and Cylindros-permopsis raciborskii? FEMS Microbiol. Ecol. 79 (3), 594–607.

Boutte, C., Grubisic, S., Balthasart, P., Wilmotte, A., 2006. Testing of primers for thestudy of cyanobacterial molecular diversity by DGGE. J. Microbiol. Methods 65(3), 542–550.

Callieri, C., 2008. Picophytoplankton in freshwater ecosystems: the importance ofsmall-sized phototrophs. Freshw. Rev. 1 (1), 1–28.

Chen, Y., Qin, B., Teubner, K., Dokulil, M.T., 2003. Long-term dynamics of phyto-plankton assemblages: Microcystis-domination in Lake Taihu, a large shallowlake in China. J. Plankton Res. 25 (4), 445–453.

Cires, S., Ballot, A., 2016. A review of the phylogeny, ecology and toxin production ofbloom-forming Aphanizomenon spp. and related species within the Nostocales(cyanobacteria). Harmful Algae 54, 21–43.

Cole, G.A., 1994. Textbook of Limnology, 4th ed. Waveland Press, Prospect Heights,IL.

Cole, J.R., Chai, B., Farris, R.J., Wang, Q., Kulam, S., McGarrell, D.M., Garrity, G.M.,Tiedje, J.M., 2005. The Ribosomal Database Project (RDP-II): sequences and toolsfor high-throughput rRNA analysis. Nucleic Acids Res. 33 (Suppl. 1), D294–D296.

Dadheech, P.K., Ballot, A., Casper, P., Kotut, K., Novelo, E., Lemma, B., Proschold, T.,Krienitz, L., 2010. Phylogenetic relationship and divergence among planktonicstrains of Arthrospira (Oscillatoriales, Cyanobacteria) of African, Asian andAmerican origin deduced by 16S-23S ITS and phycocyanin operon sequences.Phycologia 49 (4), 361–372.

de Marsac, N.T., Houmard, J., 1993. Adaptation of cyanobacteria to environmentalstimuli: new steps towards molecular mechanisms. FEMS Microbiol. Rev. 10 (1–2), 119–189.

Deng, J., Qin, B., Paerl, H.W., Zhang, Y., Ma, J., Chen, Y., 2014. Earlier and warmersprings increase cyanobacterial (Microcystis spp.) blooms in subtropical LakeTaihu, China. Freshw. Biol. 59 (5), 1076–1085.

Dittmann, E., Fewer, D.P., Neilan, B.A., 2013. Cyanobacterial toxins: biosyntheticroutes and evolutionary roots. FEMS Microbiol. Rev. 37 (1), 23–43.

Y. Jiang et al. / Harmful Algae 64 (2017) 42–5050

Dojani, S., Kauff, F., Weber, B., Budel, B., 2014. Genotypic and phenotypic diversity ofcyanobacteria in biological soil crusts of the Succulent Karoo and Nama Karoo ofSouthern Africa. Microb. Ecol. 67 (2), 286–301.

Dyble, J., Paerl, H.W., Neilan, B.A., 2002. Genetic characterization of Cylindrosper-mopsis raciborskii (Cyanobacteria) isolates from diverse geographic originsbased on nifH and cpcBA-IGS nucleotide sequence analysis. Appl. Environ.Microbiol. 68 (5), 2567–2571.

Edgar, R.C., 2004. MUSCLE: multiple sequence alignment with high accuracy andhigh throughput. Nucleic Acids Res. 32 (5), 1792–1797.

Edgar, R.C., 2013. UPARSE: highly accurate OTU sequences from microbial ampliconreads. Nat. Methods 10 (10), 996–998.

Edgar, R.C., Haas, B.J., Clemente, J.C., Quince, C., Knight, R., 2011. UCHIME improvessensitivity and speed of chimera detection. Bioinformatics 27 (16), 2194–2200.

Gobler, C.J., Burkholder, J.M., Davis, T.W., Harke, M.J., Johengen, T., Stow, C.A., Van deWaal, D.B., 2016. The dual role of nitrogen supply in controlling the growth andtoxicity of cyanobacterial blooms. Harmful Algae 54, 87–97.

Guo, L., 2007. Doing battle with the green monster of Taihu Lake. Science 317(5842), 1166.

Harke, M.J., Steffen, M.M., Gobler, C.J., Otten, T.G., Wilhelm, S.W., Wood, S.A., Paerl,H.W., 2016. A review of the global ecology, genomics, and biogeography of thetoxic cyanobacterium, Microcystis spp. Harmful Algae 54, 4–20.

Havens, K.E., 2008. Cyanobacteria blooms: effects on aquatic ecosystems. In:Hudnell, H.K. (Ed.), Cyanobacterial Harmful Algal Blooms: State of the Scienceand Research Needs. Springer, pp. 733–747.

Huang, S., Liu, Y., Hu, A., Liu, X., Chen, F., Yao, T., Jiao, N., 2014. Genetic diversity ofpicocyanobacteria in Tibetan lakes: assessing the endemic and universal dis-tributions. Appl. Environ. Microbiol. 80 (24), 7640–7650.

Huber, J.A., Welch, D.B.M., Morrison, H.G., Huse, S.M., Neal, P.R., Butterfield, D.A.,Sogin, M.L., 2007. Microbial population structures in the deep marine biosphere.Science 318 (5847), 97–100.

Ishida, T., Watanabe, M.M., Sugiyama, J., Yokota, A., 2001. Evidence for polyphyleticorigin of the members of the orders of Oscillatoriales and Pleurocapsales asdetermined by 16S rDNA analysis. FEMS Microbiol. Lett. 201 (1), 79–82.

Jiang, Y., Shao, J., Wu, X., Xu, Y., Li, R., 2011. Active and silent members in the mlrgene cluster of a microcystin-degrading bacterium isolated from Lake Taihu,China. FEMS Microbiol. Lett. 322 (2), 108–114.

Jing, L., Liu, X., Bai, S., Wu, C., Ao, H., Liu, J., 2015. Effects of sediment dredging oninternal phosphorus: a comparative field study focused on iron and phosphorusforms in sediments. Ecol. Eng. 82, 267–271.

Jing, L., Wu, C., Liu, J., Wang, H., Ao, H., 2013. The effects of dredging on nitrogenbalance in sediment–water microcosms and implications to dredging projects.Ecol. Eng. 52, 167–174.

Jost, L., 2006. Entropy and diversity. Oikos 113 (2), 363–375.Kim, S.-G., Rhee, S.-K., Ahn, C.-Y., Ko, S.-R., Choi, G.-G., Bae, J.-W., Park, Y.-H., Oh, H.-

M., 2006. Determination of cyanobacterial diversity during algal blooms inDaechung Reservoir, Korea, on the basis of cpcBA intergenic spacer regionanalysis. Appl. Environ. Microbiol. 72 (5), 3252–3258.

Kleinteich, J., Hildebrand, F., Wood, S.A., Cirs, S., Agha, R., Quesada, A., Pearce, D.,Convey, P., Kpper, F., Dietrich, D., 2014. Diversity of toxin and non-toxincontaining cyanobacterial mats of meltwater ponds on the Antarctic Peninsula:a pyrosequencing approach. Antarct. Sci. 26 (5), 521–532.

Kulkarni, R.D., Golden, S.S., 1994. Adaptation to high light intensity in Synechococcussp. strain PCC 7942: regulation of three psbA genes and two forms of the D1protein. J. Bacteriol. 176 (4), 959–965.

Li, W.K.W., 1998. Annual average abundance of heterotrophic bacteria and Syne-chococcus in surface ocean waters. Limnol. Oceanogr. 43 (7), 1746–1753.

Li, X., Li, S., Kong, R., Li, R., 2016. Molecular separation of two long taxonomicallydebated cyanobacterial genera Cylindrospermopsis and Raphidiopsis (Nosto-cales) based on the ITS-L phylogeny. Harmful Algae 57, 88–97.

Liu, C., Zhong, J., Wang, J., Zhang, L., Fan, C., 2016a. Fifteen-year study of environ-mental dredging effect on variation of nitrogen and phosphorus exchangeacross the sediment–water interface of an urban lake. Environ. Pollut. 219,639–648.

Liu, Y., Xu, Y., Wang, Z., Xiao, P., Yu, G., Wang, G., Li, R., 2016b. Dominance andsuccession of Microcystis genotypes and morphotypes in Lake Taihu, a large andshallow freshwater lake in China. Environ. Pollut. 219, 399–408.

Lv, H., Yang, J., Liu, L.M., 2013. Temporal pattern prevails over spatial variability inphytoplankton communities from a subtropical water supply reservoir. Ocea-nol. Hydrobiol. Stud. 42 (4), 420–430.

Miller, T.R., Beversdorf, L., Chaston, S.D., McMahon, K.D., 2013. Spatiotemporalmolecular analysis of cyanobacteria blooms reveals Microcystis–Aphanizomenoninteractions. PLOS ONE 8 (9) .

Nubel, U., GarciaPichel, F., Muyzer, G., 1997. PCR primers to amplify 16S rRNA genesfrom cyanobacteria. Appl. Environ. Microbiol. 63 (8), 3327–3332.

Neilan, B.A., Jacobs, D., Goodman, A.E., 1995. Genetic diversity and phylogeny oftoxic cyanobacteria determined by DNA polymorphisms within the phycocya-nin locus. Appl. Environ. Microbiol. 61 (11), 3875–3883.

Neilan, B.A., Pearson, L.A., Muenchhoff, J., Moffitt, M.C., Dittmann, E., 2013. Envi-ronmental conditions that influence toxin biosynthesis in cyanobacteria. En-viron. Microbiol. 15 (5), 1239–1253.

O’Neil, J., Davis, T.W., Burford, M.A., Gobler, C., 2012. The rise of harmful cyano-bacteria blooms: the potential roles of eutrophication and climate change.Harmful Algae 14, 313–334.

Paerl, H.W., Huisman, J., 2009. Climate change: a catalyst for global expansion ofharmful cyanobacterial blooms. Environ. Microbiol. Rep. 1 (1), 27–37.

Paerl, H.W., Otten, T.G., 2013. Harmful cyanobacterial blooms: causes, conse-quences, and controls. Microb. Ecol. 65 (4), 995–1010.

Palenik, B., Swift, H., 1996. Cyanobacterial evolution and prochlorophyte diversityas seen in DNA-dependent RNA polymerase gene sequences. J. Phycol. 32 (4),638–646.

Pearson, L., Mihali, T., Moffitt, M., Kellmann, R., Neilan, B., 2010. On the chemistry,toxicology and genetics of the cyanobacterial toxins, microcystin, nodularin,saxitoxin and cylindrospermopsin. Mar. Drugs 8 (5), 1650–1680.

Porebski, S., Bailey, L.G., Baum, B.R., 1997. Modification of a CTAB DNA extractionprotocol for plants containing high polysaccharide and polyphenol compo-nents. Plant Mol. Biol. Rep. 15 (1), 8–15.

Robertson, B.R., Tezuka, N., Watanabe, M.M., 2001. Phylogenetic analyses of Syne-chococcus strains (cyanobacteria) using sequences of 16S rDNA and part of thephycocyanin operon reveal multiple evolutionary lines and reflect phycobilincontent. Int. J. Syst. Evol. Microbiol. 51 (3), 861–871.

Schloss, P.D., Westcott, S.L., Ryabin, T., Hall, J.R., Hartmann, M., Hollister, E.B.,Lesniewski, R.A., Oakley, B.B., Parks, D.H., Robinson, C.J., 2009. Introducingmothur: open-source, platform-independent, community-supported softwarefor describing and comparing microbial communities. Appl. Environ. Microbiol.75 (23), 7537–7541.

Sinha, R., Pearson, L.A., Davis, T.W., Burford, M.A., Orr, P.T., Neilan, B.A., 2012.Increased incidence of Cylindrospermopsis raciborskii in temperate zones – isclimate change responsible? Water Res. 46 (5), 1408–1419.

Steven, B., McCann, S., Ward, N.L., 2012. Pyrosequencing of plastid 23S rRNA genesreveals diverse and dynamic cyanobacterial and algal populations in twoeutrophic lakes. FEMS Microbiol. Ecol. 82 (3), 607–615.

Stockner, J.G., Shortreed, K.S., 1988. Response of Anabaena and Synechococcus tomanipulation of nitrogen: phosphorus ratios in a lake fertilization experiment.Limnol. Oceanogr. 33 (6), 1348–1361.

Stucken, K., John, U., Cembella, A., Murillo, A.A., Soto-Liebe, K., Fuentes-Valdes, J.J.,Friedel, M., Plominsky, A.M., Vasquez, M., Glockner, G., 2010. The smallestknown genomes of multicellular and toxic cyanobacteria: comparison, minimalgene sets for linked traits and the evolutionary implications. PLoS ONE 5 (2),e9235.

Takasu, H., Ushio, M., Leclair, J., Nakano, S., 2015. High contribution of Synechococcusto phytoplankton biomass in the aphotic hypolimnion in a deep freshwater lake(Lake Biwa, Japan). Aquat. Microb. Ecol. 75 (1), 69–79.

Tamura, K., Peterson, D., Peterson, N., Stecher, G., Nei, M., Kumar, S., 2011. MEGA5:molecular evolutionary genetics analysis using maximum likelihood, evolu-tionary distance, and maximum parsimony methods. Mol. Biol. Evol. 28 (10),2731–2739.

Teneva, I., Dzhambazov, B., Mladenov, R., Schirmer, K., 2005. Molecular and phylo-genetic characterization of Phormidium species (cyanoprokaryota) using thecpcB-IGS-cpcA locus. J. Phycol. 41 (1), 188–194.

Visser, P.M., Verspagen, J.M., Sandrini, G., Stal, L.J., Matthijs, H.C., Davis, T.W., Paerl,H.W., Huisman, J., 2016. How rising CO2 and global warming may stimulateharmful cyanobacterial blooms. Harmful Algae 54, 145–159.

Wang, Z.J., Liu, Y., Xu, Y., Xiao, P., Li, R.H., 2013. The divergence of cpcBA-IGSsequences between Dolichospermum and Aphanizomenon (Cyanobacteria) andthe molecular detection of Dolichospermum flos-aquae in Taihu Lake, China.Phycologia 52 (5), 447–454.

Whitton, B.A., 2012. Ecology of Cyanobacteria II: Their Diversity in Space and Time,1st ed. Springer, Netherlands.

Wiedner, C., Rucker, J., Bruggemann, R., Nixdorf, B., 2007. Climate change affectstiming and size of populations of an invasive cyanobacterium in temperateregions. Oecologia 152 (3), 473–484.

Wilhelm, S.W., Bullerjahn, G.S., Eldridge, M.L., Rinta-Kanto, J.M., Poorvin, L., Bour-bonniere, R.A., 2006. Seasonal hypoxia and the genetic diversity of prokaryotepopulations in the central basin hypolimnion of Lake Erie: evidence for abun-dant cyanobacteria and photosynthesis. J. Great Lakes Res. 32 (4), 657–671.

Wu, S., Xie, P., Liang, G., Wang, S., Liang, X., 2006. Relationships between micro-cystins and environmental parameters in 30 subtropical shallow lakes along theYangtze River, China. Freshw. Biol. 51 (12), 2309–2319.

Ye, W., Tan, J., Liu, X., Lin, S., Pan, J., Li, D., Yang, H., 2011. Temporal variability ofcyanobacterial populations in the water and sediment samples of Lake Taihu asdetermined by DGGE and real-time PCR. Harmful Algae 10 (5), 472–479.

Yu, Z., Yang, J., Amalfitano, S., Yu, X.Q., Liu, L.M., 2014. Effects of water stratificationand mixing on microbial community structure in a subtropical deep reservoir.Sci. Rep. 4 .

Zhang, B., Li, R., Xiao, P., Su, Y., Zhang, Y., 2016. Cyanobacterial composition andspatial distribution based on pyrosequencing data in the Gurbantunggut Desert,Northwestern China. J. Basic Microbiol. 56, 308–320.