Oligopeptide chemotypes of the toxic freshwater cyanobacterium Planktothrix can form subpopulations with dissimilar ecological traits

Oligopeptide chemotypes of the toxic freshwater cyanobacterium Planktothrix can form

subpopulations with dissimilar ecological traits

Thomas Rohrlack1

Norwegian Institute for Water Research, Gaustadalleen 21, NO-0349 Oslo, Norway; Norwegian Institute for PublicHealth, P.O. Box 4404 Nydalen, NO-0403 Oslo, Norway

Bente EdvardsenNorwegian Institute for Water Research, Gaustadalleen 21, NO-0349 Oslo, Norway; University of Oslo,Department of Biology, P.O. Box 1066 Blindern, NO-0316 Oslo, Norway

Randi Skulberg and Camilla B. HalstvedtNorwegian Institute for Water Research, Gaustadalleen 21, NO-0349 Oslo, Norway

Hans C. UtkilenNorwegian Institute for Public Health, P.O. Box 4404 Nydalen, NO-0403 Oslo, Norway

Robert Ptacnik and Olav M. SkulbergNorwegian Institute for Water Research, Gaustadalleen 21, NO-0349 Oslo, Norway

Abstract

Nonribosomal oligopeptides were used as qualitative and quantitative markers to test whether populations ofthe toxic freshwater cyanobacterium Planktothrix comprise subpopulations with dissimilar ecological traits. Afield program was conducted in Lake Steinsfjorden (Norway), where Planktothrix has dominated thephytoplankton community for decades, allowing the present study to disregard other potential producers ofnonribosomal oligopeptides. Four chemotypes with distinct cellular oligopeptide patterns were found in the lake.The chemotypes occurred largely unaltered throughout a period of up to 33 yr and differed with respect toseasonal dynamics, depth distribution, and participation in loss processes. Changes in the relative abundance ofchemotypes occurred almost constantly and could not be explained with fluctuations in light, temperature, orconcentration of macronutrients but might have been due to differences among chemotypes in depth regulation orinteraction with grazers or pathogens. Chemotypes correlated weakly with taxonomic groups and genotypesdefined on the basis of phycocyanin operon deoxyribonucleic acid (DNA) sequences. Our findings suggest thatfirst, oligopeptide chemotypes can have dissimilar ecological traits and therefore interact differently with theirenvironment; second, populations of toxic freshwater cyanobacteria can comprise multiple ecologically distinctsubpopulations; and, third, the relative abundance of these may vary, causing a high variability in whole-population properties. The latter was demonstrated for the microcystin-related toxicity of Planktothrix. Theconsequences of the present findings for the taxonomy of Planktothrix are discussed.

In phytoplankton ecology the species has typically beenthe lowest taxonomic level studied, and in many investiga-tions on the interactions of phytoplankton organisms withtheir environment it has been assumed that the species isthe main ecological unit. Studies on the cyanobacterialpicoplankter Prochlorococcus and Synechococcus havechallenged this view. These organisms form populations

that comprise coexisting subpopulations with dissimilarecological traits (Becker et al. 2002; Casamayor et al. 2002;Rocap et al. 2002). The relative subpopulation compositionof a given population is variable and can change inresponse to environmental fluctuations. This allows forrapid adjustments to a wide range of ambient conditionsand has been suggested as one reason for the remarkablesuccess of Prochlorococcus and Synechococcus species on aglobal scale (Postius and Ernst 1999; Johnson et al. 2006).

Given the significance of the above findings, the searchfor ecologically distinct subpopulations in cyanobacteriaoutside Prochlorococcus and Synechococcus is a priority forcurrent research on phytoplankton. Of particular interestare toxic freshwater cyanobacteria, including Anabaena,Microcystis, and Planktothrix species, which cause ecolog-ical and health problems around the world (Codd et al.2005). Populations of these cyanobacteria can be polymor-phic with respect to, for example, cellular content of

1 Corresponding author ([email protected]).

AcknowledgmentsWe gratefully acknowledge the logistical and financial support

of Hole and Ringerike municipalities. We also appreciate theexcellent assistance of Jozsef Kotai, Hege Hansen, and SisselBrubak, and the helpful comments of three anonymous reviewers.The work was funded in part by the Norwegian Institute forWater Research, the Norwegian Research Council (grants 148070/730 and 157338/140), and the European Community (grantQLK4-CT-2002-02634).

Limnol. Oceanogr., 53(4), 2008, 1279–1293

E 2008, by the American Society of Limnology and Oceanography, Inc.

1279

biologically active compounds (Fastner et al. 2001; Welkeret al. 2004), gas vesicle properties (Beard et al. 2000), andsusceptibility to grazers (Rohrlack et al. 2005). Attempts toelucidate the consequences of the polymorphism under fieldconditions are often hampered by the inability of classicbiomass markers to discriminate among groups of individ-uals within a population. Two research groups havesuggested to approach this problem with a real-timepolymerase chain reaction (PCR) assay that subdivides apopulation into producers and nonproducers of certaintoxins (Kurmayer et al. 2003; Vaitomaa et al. 2003). Othermethods have higher resolution, but produce only qualita-tive data (Janse et al. 2005).

The objective of the present study was to develop anduse a novel quantitative method that allows studies at thesubpopulation level with a reasonable resolution. For thatpurpose, nonribosomal oligopeptides were used as pheno-typic markers. Many freshwater cyanobacteria share theability to synthesize oligopeptides not as usual on theribosome, but by involving large modular multienzymecomplexes and the thio-template mechanism (Kaebernickand Neilan 2001). These so-called nonribosomal oligopep-tides, hereafter referred to as oligopeptides, are widespreadin at least four of the five orders of cyanobacteria (Welkerand von Dohren 2006). Of the ca. 600 known structuralvariants, most can be assigned to major classes, which areproduced by separate multienzyme complexes encoded bylarge gene clusters. Cyanobacterial cells may contain anycombination of these gene clusters and the correspondingmultienzyme complexes (Welker and von Dohren 2006).

Oligopeptides are for several reasons suitable for studiesat the subpopulation level. Individuals of the samepopulation can differ in their intracellular qualitativeoligopeptide composition (Fastner et al. 2001; Welker etal. 2004). These differences are genetically determined bythe occurrence and/or absence of oligopeptide synthetasegene clusters, mutations within these clusters (Christiansenet al. 2006), as well as gene cluster organization (Welkerand von Dohren 2006) and have already been used tosubdivide populations into oligopeptide chemotypes (e.g.,Fastner et al. 2001; Welker et al. 2004). Oligopeptides mayalso be suitable to quantify cyanobacterial biomass becausetheir synthesis can neither be initiated nor turned off byextracellular stimuli and the cellular amount varies only bya factor of 1–5 when comparing optimal growth conditionswith severe nutrient or light limitation (e.g., Chorus andBartram 1999; Repka et al. 2004; Rohrlack and Utkilen2007). Similar properties apply to established biomassmarkers such as chlorophyll a (Chl a) (Woitke et al. 1997;Schagerl and Muller 2006). Once produced, oligopeptidesstay largely in the cyanobacterial cells and are only releasedin significant amounts after cell death (Rohrlack andHyenstrand 2007).

In the present study, we identified oligopeptide chemo-types of the toxic freshwater cyanobacterium Planktothrixin Lake Steinsfjorden (Norway) and used selected oligo-peptides as chemotype-specific biomass markers. Thechemotypes were studied with respect to, first, theirpresence and variability over a period of up to 33 yr;second, the relationship to taxonomic groups and those

defined on the basis of phycocyanin operon deoxyribonu-cleic acid (DNA) sequences; third, seasonal dynamics anddepth distribution in relation to abiotic environmentalfactors; and fourth, the role in determining the propertiesof the entire population such as its toxic capabilities.Finally, we used the results to assess whether oligopeptidechemotypes interact differently with their environment, i.e.,whether they form subpopulations with dissimilar ecolog-ical traits.

Materials and methods

Study area—Lake Steinsfjorden is situated in Southeast-ern Norway (60u089N, 10u209E). It occupies an area of13.9 km2, has a maximum depth of 20 m, has an averagewater retention time of 4.6 yr, and has a catchment area of63.7 km2. The lake belongs to the boreal dimictic type andis stratified in summer and winter. The mean totalphosphorus, total nitrogen, and Chl a concentrations haveremained rather stable throughout the last decades andcurrently average at 10 mg L21, 262 mg L21, and4.5 mg L21, respectively (Halstvedt et al. 2007).

According to current taxonomy (Suda et al. 2002), twoPlanktothrix species, Planktothrix rubescens and Plankto-thrix agardhii, occur in the lake. These have dominated thephytoplankton community for several decades and regu-larly form blooms in the thermocline during summer orunder the ice during the winter months (Halstvedt et al.2007). High Planktothrix biovolume concentrationshave also occurred during periods with total circulationof the water body. Other cyanobacteria contribute less than5% to the phytoplankton biomass in the upper water layerand less than 1% in deeper parts of the lake (Skulberg andSkulberg 1985; Halstvedt et al. 2007). This allowed us tostudy Planktothrix in Lake Steinsfjorden using cyanobac-terial oligopeptides without interference from other poten-tial producers of these compounds.

Clonal isolates of Planktothrix—Clonal Planktothrixisolates were used to identify and characterize majoroligopeptide chemotypes to be found in Lake Steinsfjorden.Clonal isolates were established by culturing single Plankto-thrix filaments in Z8 medium (Kotai 1972). Several isolationexperiments were undertaken between 1965 and 1998. Amore intensive isolation program was carried out in 2004 toobtain fresh isolates representing the period during which weconducted field work in Lake Steinsfjorden (Fig. 1). Thetypical success rate was higher than 50%, i.e., on an isolationdate more than 50% of the isolated Planktothrix filamentsgrew and a culture was obtained. Species determination wasbased on Suda et al. (2002). The isolates were maintained asnonaxenic clonal cultures at the Norwegian Institute ofWater Research (NIVA) culture collection of algae untilused in the present study. Conditions for cultures in thecollection were glass flasks containing 50 mL Z8 medium,17uC, light at a photon flux density of about 10 mmol m22

s21, and a light : dark cycle of 12 : 12 h.

Field program in 2003 and 2004—Water samples werecollected at the deepest part of the lake on 13 occasions

1280 Rohrlack et al.

Fig. 1. Properties of clonal Planktothrix isolates of this study. For each isolate theidentification number (ID) in the NIVA culture collection of algae is given. The species names P.rubescens and P. agardhii are abbreviated with P. rub. and P. aga., respectively. The statisticalbasis for the oligopeptide chemotype and the cpcBA genotype assignment is shown in Figs. 3 and4, respectively.

Subpopulations in Planktothrix 1281

from May 2003 to August 2004. Samples were taken fromsurface to 19-m depth at 1-m intervals. Depending on thedensity of Planktothrix, 10 to 100 mL of the water sampleswas filtered (25 mm Whatman GF/C glass-fiber filters),and the filters were stored at 220uC until analysis ofoligopeptides by liquid chromatography tandem massspectrometry (LC-MS/MS). A second set of subsampleswas filtered through cellulose nitrate membrane filters(40 mm, 0.45-mm pore size, 1 mm 3 1 mm grid, SartoriusAG) that, after drying overnight, were used for themicroscopic determination of the biovolume concentrationof P. rubescens and P. agardhii. The species weredistinguished by the occurrence of the accessory pigmentphycoerythrin in P. rubescens and the resulting red color ofits filaments. All filaments on a given filter were countedusing a stereomicroscope (SMZ-10, Nikon). Filaments werecounted as sections of 750-mm length. The cell diameter wasdetermined once by microscope measurements of 100filaments per Planktothrix species. Finally, the filamentcounts and diameter were used to calculate the biovolumeconcentration of P. rubescens and P. agardhii as a measureof their abundance.

Temperature profiles were determined using a model 58probe (YSI). Data on nutrient and light availabilitythroughout the water column were taken from Halstvedtet al. (2007).

Oligopeptide analysis—Oligopeptides were extractedfrom filters with field samples or cultured Planktothrixafter lyophilization using 50% MeOH as describedpreviously (Rohrlack et al. 2003).

For the identification and quantification of oligopep-tides, LC-MS/MS was used. The instrumental setupincluded a Waters Acquity ultraperformance liquid chro-matography (UPLC) system equipped with a WatersAtlantis C18 column (2.1 3 150 mm, 5-mm particle size)and directly coupled to a Waters Quattro Premier XEtandem quadrupole MS/MS detector. The UPLC systemwas set to deliver a linear gradient from 20% to 60%acetonitrile in water, both containing 0.1% formic acid,within 8 min at a flow rate of 0.25 mL min21. The columnand auto sampler temperatures were 20uC and 4uC,respectively. At all times, the MS/MS detector was run inpositive electrospray mode (ESI+). Other general settingsincluded a source temperature of 120uC, a desolvationtemperature of 350uC, a drying gas flow rate of 800 L h21,a gas flow at the cone of 50 L h21, and standard voltagesand energies suggested by the manufacturer for the ESI+mode. Only the cone voltage and the settings for thecollision cell were adapted to the various types of analysis.Nitrogen, continuously delivered by a nitrogen generator(NG 11, Parker Balston), served as drying, nebulizing, andcone gas.

To screen extracts for cyanobacterial oligopeptides, thedetector was run in total scan mode for the mass range of500 to 1,200 Da during the entire UPLC gradient. At thisstage, the cone voltage was 60 V and the time for one scan2 s. Afterward, all mass signals that represented com-pounds with a molecular mass within the range of 500–1,200 Da were analyzed in fragmentation experiments. To

this end, the detector was ran in daughter ion scanningmode, and the cone voltage and collision cell settings wereoptimized to obtain as many fragments of the respectivecompound as possible. In all cases, argon served ascollision gas. For the identification of a given compound,its fragmentation spectrum was screened for immoniumions of amino acids and other analytical fragments to beexpected when working with peptides. Larger fragmentswere identified by comparison with fragment patterns ofalready elucidated compounds or available standardmaterial. Further information was gained from using thefragmentation simulation software HighChemMass Fron-tier (version 3). Finally, identified fragments, the molecularmass of the respective compound, its isotope spectrum, andfragmentation patterns of already elucidated oligopeptideswere used to gradually develop a putative model of thecompound’s plain structure, which then was compared withstructures of oligopeptides already described in theliterature (Welker and von Dohren 2006). This way ofstructural elucidation of cyanobacterial oligopeptides onthe basis of MS fragmentation experiments has beensuccessfully used in several earlier studies (e.g., Fastner etal. 2001; Welker et al. 2004; Tooming-Klunderud et al.2007). These papers also give examples for peptidefingerprints of clonal isolates along with the respectiveraw data.

Selected oligopeptides in extracts of Planktothrix isolatesand field samples were quantified by running the MS/MSdetector in selective ion monitoring mode with a dwell timeof 0.05 s. The cone voltage was optimized for eachcompound using direct infusion of concentrated Plankto-thrix extracts. Since we were interested in relative values formost oligopeptides, we quantified them against a standardof microcystin LR, kindly provided by G. Codd (Universityof Dundee, Scotland, U.K.). This also compensated for thelack of standardized material. Desmethyl-microcystin RR,which was the only oligopeptide found in Lake Steinsfjor-den to form double charged ions in the MS/MS detector,was quantified against purified microcystin RR purchasedfrom Sigma-Aldrich. Extensive preanalysis tests proved thereproducibility of the quantification method for the entirespectrum of compounds and their concentration rangesfound in the cyanobacterial extracts of this study.

Chemotyping of Planktothrix isolates—All SteinsfjordenPlanktothrix isolates were grown as batch cultures underconstant conditions in a walk-in environmental chamber(culture volume of 50 mL, complete Z8 medium, 17uC, anight : day cycle of 12 : 12 h with light at a mean photonflux density of 5 mmol m22 s21 measured outside theculture vessels). At the late exponential growth phase,samples were collected for the LC-MS/MS-based oligopep-tide screening, done in accordance with the methodsdescribed above. Only oligopeptides of the classes aerugi-nosins, anabaenopeptins, cyanopeptolins, microcystins,and microginins were considered, since these are the mostabundant and best studied oligopeptides occurring inPlanktothrix (Welker et al. 2004). Other peptides such asmicroviridins were disregarded because of their unknowngenetic background and mode of production. The oligo-


peptide patterns of all Planktothrix isolates were alignedand analyzed using the PHYLIP program package version3.6 and according to the neighbor joining method on Jukes-Cantor pairwise distances. Chemotypes were identifiedaccording to the clades and bootstrap values of theresulting tree. In some cases it was necessary to combinePlanktothrix isolates with similar but not identical oligo-peptide patterns in one chemotype. The reason for that wasthe impossibility to reliably quantify such outliers in fieldsamples using specific oligopeptides, either because suchcompounds did not occur at all or only in very smallcellular amounts.

Estimation of chemotype biovolume concentration infield samples—The biovolume of chemotypes in a givensample from Lake Steinsfjorden was estimated by using thecell-bound fraction of chemotype-specific oligopeptides asa quantitative biomass marker. The conversion factorbetween the amount of an oligopeptide in a sample and thebiovolume of the corresponding chemotype was obtainedin batch culture experiments. Only Planktothrix isolatesacquired after 1990 were considered, since these wereregarded as most representative for the current situation inLake Steinsfjorden. The growth conditions were identicalfor all cultures (culture volume of 50 mL, complete Z8medium, 20uC, a light : dark cycle of 12 : 12 h with light at amean photon flux density of 12.5 mmol m22 s21 measuredoutside the culture vessels, cultures placed on a shaker). Atthe late exponential growth phase, samples for themicroscopic determination of the biovolume of Plankto-thrix were collected, which was carried out according to themethod described above for field samples, and for the LC-MS/MS-based oligopeptide quantification. Finally, foreach chemotype the amount of its specific oligopeptidesthat corresponded to a certain biovolume was determined.The experiments were also used to determine the meancellular amount of major cyanobacterial toxins (desmethyl-microcystins LR and RR) by LC-MS/MS.

To quantify a chemotype in a field sample, the amountof its unique oligopeptides was determined by LC-MS/MS.From this amount the corresponding biovolume of therespective chemotype was calculated by using the aboveconversion factors. Finally, the biovolume concentrationper liter of water was calculated. Since there did not existan independent way of quantifying chemotypes in fieldsample, the above method was validated by comparing thesum of individual chemotype biovolume concentrationswith the total Planktothrix biovolume concentration thatwas determined microscopically for all samples takenduring the field program in 2003 and 2004. This was donein two ways, by testing for linear correlation between bothvalues and by calculating the absolute difference for allsampling dates.

Statistical analysis of chemotype dynamics—The depthmaxima of individual chemotypes were determined byanalyzing the cumulative frequency distribution (CFD) as afunction of depth. A sigmoid or exponential relationshipbetween CFD and depth suggested a maximum in thevertical distribution. To test for such nonlinearity, linear

and nonlinear regression models were compared. If anonlinear regression was significantly better as determinedby the Akaka’s information criterion (Sakamoto et al.1986), a depth optimum was assumed. The actual depthmaximum was then calculated by weighted averaging ofdepth, using the biovolume concentration of the respectivechemotype at each depth as weights. Depth distributionsamong chemotypes were compared by pairwise Wilcoxoncorrelation tests. Responses of chemotypes to environmen-tal factors were analyzed by one-sided Spearman rankcorrelations with chemotype-specific growth rates asresponse variables. For each sampling interval, specificnet growth rates were calculated from the cumulativechemotype abundances over the entire water column. Forthe environmental variables, first the average between eachpair of sampling dates was calculated. For each samplinginterval and chemotype, we then calculated a weightedmean over the water column using chemotype concentra-tion as weights.

DNA extraction, PCR amplification, sequencing, andsequence analysis—The present study used the intergenicspacer and flanking coding regions cpcB and cpcA of thephycocyanin operon (cpcBA) to group Planktothrix iso-lates. The cpcBA DNA-region was introduced by Neilanand coworkers (1995) as a phylogenetic marker and hasbeen amply and successfully used ever since. The length ofcpcBA varies among cyanobacterial species and strains andranges typically from 500 to 740 base pairs (Neilan et al.1995). For DNA extraction, 15 mL of each Planktothrixisolate was taken directly from stock cultures of the NIVAculture collection of algae. After centrifugation (7,000 gfor 10 min, Sorvall RT7), DNA was extracted by theE.Z.N.A. SP plant miniprep kit (Omega-Bio-Tek) accord-ing to the manufacturer’s protocol. The cpcBA region wasamplified using primers described by Neilan et al. (1995).The PCR reactions were run in total volumes of 50 mL,consisting of sterile Milli-Q water, 5 mL of 310 PCR buffercontaining 1.5 mmol L21 of MgCl2, 100 nmol L21 of eachprimer, 200 mmol L21 didesoxyribonukleosidtriphosphate,1 unit of Taq polymerase (MasterTaq), 5 mL MasterTaqenhancer, and 5 mL template DNA. Amplification wasperformed in an Eppendorf Mastercycler Gradient thermalcycler as follows: initial denaturation at 94uC for 3 minfollowed by 35 cycles of denaturation at 94uC for 1 min,annealing at 50uC for 1 min, extension at 72uC for 2 min.PCR products were sequenced directly. Phylogenetic treeswere constructed by using the neighbor joining (NJ) andminimum evolution (ME) methods on Jukes-Cantorpairwise distances and maximum parsimony (MP) inMEGA version 3.1.

Results

Planktothrix species composition and dynamics in LakeSteinsfjorden 2003 and 2004—P. rubescens and P. agardhiiwere present throughout the field program of this study(Fig. 2). Both exhibited similar patterns of seasonaldynamics and depth distribution with growth duringcirculation and accumulation in the metalimnion during


summer stratification. In May and June 2003, largeamounts of P. rubescens were found directly above theground and so well below the euphotic zone, suggesting asignificant loss of biomass due to sedimentation. Therelative abundance of P. agardhii increased from 7% inMay 2003 to above 60% in 2004.

Planktothrix chemotypes in Lake Steinsfjorden—Between1965 and 2004, a total of 46 clonal Planktothrix cultureswere isolated from Lake Steinsfjorden. They produced 33individual oligopeptides within the mass range of 500–1,200 Da, and these compounds could be identified asmembers of the oligopeptide classes anabaenopeptins(variants A, B, C, F, and oscillamide Y), aeruginosins,cyanopeptolins (oscillapeptin G, diverse new cyanopepto-lins), microcystins, and microginins (oscillaginins A, B;Fig. 1). Four distinct chemotypes, henceforth abbreviatedCht1–4, were identified (Fig. 3). All chemotypes werereisolated throughout several years or decades and thusappear to be persistent in Lake Steinsfjorden (Fig. 1). Thelongest record was found for Cht2, which was isolatedunaltered in 1965, 1977, 1982, and 1998. Some isolates hadslightly different oligopeptide patterns as compared withtypical members of the respective chemotypes (NIVA-CYA55, 56/1, 137, 536, 538, and 540; see Fig. 1). Ascreening of field samples from Lake Steinsfjorden taken inJune 2003 and April 2004 detected only oligopeptides thatalso occurred in clonal Planktothrix isolates, suggestingthat Cht1–4 were indeed the main chemotypes.

Taxonomic affiliation and results of genotyping with thecpcBA marker—Isolates were identified either as P.rubescens or P. agardhii (Fig. 1). Genetic analyses basedon cpcBA sequences revealed four genotypes, Gt1–4(Fig. 4). All P. rubescens isolates (except NIVA-CYA392,which had an aberrant cpcBA sequence termed Gt4)clustered together with one P. agardhii isolate (NIVA-CYA532) and were well separated from the majority of theP. agardhii isolates (bootstrap values 100/99/99 using NJ/MP/ME). Some P. agardhii isolates (NIVA-CYA56/1, 56/3,137, termed Gt2) had an uncertain placement in the tree asindicated by low bootstrap support values (53/63/48, NJ/MP/ME). While the taxonomic affiliation and results of thecpcBA genotyping were in reasonable agreement, theyshowed rather weak correlations to the oligopeptidechemotypes described above (Figs. 3 and 4). This culmi-nated in the occurrence of the chemotype Cht4 thatincluded P. rubescens along with P. agardhii isolates, bothwith distinct cpcBA sequences, but identical or very similaroligopeptide patterns.

Determination of the quantitative chemotype compositionin lake samples: method development and validation—Eachof the four chemotypes, which were found in LakeSteinsfjorden, featured a set of unique oligopeptides(Fig. 1). The most dominant among these oligopeptideswere selected to quantify Cht1–4 in lake samples: Cht1-oscillaginin B and oscillapeptin G; Cht2-cyanopeptolin(1098.7); Cht4-aeruginosin (583.5)-1 and cyanopeptolin

Fig. 2. Seasonal dynamics and depth distribution of (A) P. rubescens and (B) P. agardhii.The colors represent square root transformed biovolume concentrations. The red dots mark thedepth of the maximum abundance at each sampling date, if such maximum existed. Temperaturecontour lines are shown in black. The lake was ice covered from December 2003 to the middle ofApril 2004.


(1093.7). Cht3 lacked oligopeptides specific to all of itsisolates and with a high enough concentration in the cells(cyanopeptolin (1142.7) occurred only in traces). However,Cht3 shared methyl-anabaenopeptin C with Cht2, which,as already mentioned, was quantified on the basis ofanother oligopeptide. Therefore, to estimate the biovolumeof Cht3 in a given sample, we first calculated the biovolumeof Cht2 and, on the basis of this number, the fraction ofmethyl-anabaenopeptin C in the sample originating from

Cht2. Then we used the remaining fraction of thisoligopeptide to estimate the Cht3 biovolume. Table 1shows the conversion factors between the amount ofchemotype-specific oligopeptides and the biovolume ofCht1–4.

A highly significant linear correlation was foundbetween the cumulative biovolume concentration ofCht1–4 and the total Planktothrix biovolume concentrationdetermined microscopically for all samples taken during the

Fig. 3. Distance tree (unrooted) constructed according to the neighbor joining method based on cellular oligopeptide patterns in 46Planktothrix isolates from Lake Steinsfjorden. The numbers represent bootstrap support values for the respective nodes (1,000 bootstrapresamplings in total). The abbreviations Cht1–4 refer to the oligopeptide chemotypes described in Fig. 1.



field program in 2003 and 2004 (Fig. 5). The absolutedifference between both values varied over the study period(Fig. 6). Good to very good agreement was found in spring2003 and from autumn 2003 to spring 2004, but theoligopeptide method overestimated the total Planktothrixbiovolume during summer 2003 and summer 2004. Themean absolute difference between estimated and measuredtotal Planktothrix biovolume for the entire study periodwas 0.47 mm3 L21.

Observations regarding the quantitative chemotype com-position in Lake Steinsfjorden in 2003 and 2004—Cht1–4were present during the entire study period, but theydiffered in their dynamics and behaved differently in 2003compared with 2004 (Fig. 7). Two major shifts in therelative chemotype composition were observed, bothcharacterized by a steady decline of one chemotype overa period of several months (Fig. 8). Between June andNovember 2003, the relative abundance of Cht1 wasreduced from ca. 80% to below 5% of the Planktothrixpopulation. In this period, Cht3 and Cht4 increased in theirabsolute and relative abundances (Figs. 7 and 8). Betweenwinter 2003 and early spring 2004, another shift occurredfrom a dominance of Cht4 to that of Cht3. An interestingobservation was made with regard to this event. In April2004, large flakes of Planktothrix were released from themelting ice and transported by a water current directlyunder the ice toward the littoral zone. This biomassaccumulated near the shore of Lake Steinsfjorden and,according to microscopic and oligopeptide analyses,contained both P. rubescens and P. agardhii and consistedof 96% Cht4. The biomass decomposed during thefollowing weeks. We believe therefore that a selectiveaccumulation of Cht4 in the ice during winter followed by a

transportation to the littoral zone in early spring is a likelyexplanation for the decline of Cht4 in favor of Cht3.

To determine whether various environmental factorscorrelated with the abundance and growth of chemotypes,a series of statistical analyses was performed. A Wilcoxontest with Cht1–4 data from the field program as inputshowed that the depth distribution of Cht1 and Cht4 wassignificantly different from that of Cht2 and Cht3 (p ,0.05) when considering the entire study period. A one-sidedSpearman rank correlation test revealed no significantrelationship between light, temperature, or dissolvedmacronutrients and the specific net growth rate ofindividual chemotypes (Table 2, temperature data shownin Fig. 7, nutrients and light data in Halstvedt et al. 2007).The pooled data on Cht1–4 growth, on the other hand,were positively correlated to dissolved inorganic nitrogenand light.

Chemotype composition and toxin concentration inlake water—Cht1–4 produced different variants andcellular amounts of highly toxic oligopeptides belongingto the microcystin class (Fig. 1, Table 1). The samecompounds were found in field samples from LakeSteinsfjorden. With the entire quantitative dataset onchemotype biovolume concentrations in Lake Steinsfjordenand the mean cellular amount of major microcystins(desmethyl-microcystins LR and RR) in Cht1–4, theconcentration of cell-bound microcystins in Lake Steinsf-jorden was estimated for all samples taken during the fieldprogram in 2003 and 2004. This was done by multiplyingthe biovolume concentration of a chemotype at a givenpoint in space and time with its average cellular content ofmajor microcystins, which is shown in Table 1. Bycalculating the sum over all four chemotypes, an estimate

Table 1. Cellular amount of chemotype-specific oligopeptides and major microcystins in Cht1–4. The numbers read 10% percentile,mean value, 90% percentile. Mean values for the chemotype-specific oligopeptides were used as conversion factors to estimate Cht1–4biovolume concentrations in field samples. Only Planktothrix isolates obtained after the year 1990 were taken into account. n values aregiven in column headings and refer to the number of isolates analyzed for each chemotype.

OligopeptideCht1 (n54)(ng mm23)

Cht2 (n52)(ng mm23)

Cht3 (n51)(ng mm23)

Cht4 (n525)(ng mm23)

Oscillaginin B 404, 471, 552Oscillapeptin G 377, 448, 509Cyanopeptolin (1098.7) 614, 902, 1,189Methyl-anabaenopeptin C 263, 298, 333 85Aeruginosin (583.5)-1 261, 392, 545Cyanopeptolin (1093.7) 259, 369, 504Desmethyl-microcystin RR 622, 806, 981 475, 496, 516 1,462 1,261, 1,973, 3,577Desmethyl-microcystin LR 25, 28, 31 288, 319, 349 44 4, 9, 18

r

Fig. 4. Phylogenetic tree constructed according to the neighbor joining method based on cpc BA sequences of the phycocyaninoperon from 46 Planktothrix isolates from Lake Steinsfjorden. The Oscillatoria sp. and Nodularia sequences served as outgroup. Thenumbers represent bootstrap support values (NJ/MP/ME) for the respective nodes (1,000 bootstrap resamplings in total). Theabbreviations Gt1–4 refer to different genotypes. European Molecular Biology Laboratory (EMBL) accession numbers for the 46Planktothrix DNA sequences are AM490087-AM490131 and AM490971 (NIVA-CYA537).


for the microcystin concentration was obtained and couldbe compared with the direct mass spectrometric measure-ment. Linear correlation analysis revealed highly significantrelationships for desmethyl-microcystin LR and RR(Fig. 9A,B), showing that 94% of the variation inconcentration of these two toxins in Lake Steinsfjordencan be explained by considering individual chemotypes andtheir different properties. The numbers drop to 57%(desmethyl-microcystin LR) and 83% (desmethyl-micro-cystin RR) when estimating the toxin concentration on thebasis of the total Planktothrix abundance. It is worthwhilementioning that the oligopeptide markers used for theestimation of Cht1–4 abundances are not identical to andoccur independently of microcystins (Welker and vonDohren 2006).

Discussion

The application of oligopeptides as biomass markers isnew and involves uncertainties, deriving from possibleeffects of fluctuations in growth resource availability onoligopeptide production and the use of laboratory culturesin calibrating a field method. Nevertheless, the goodagreement between the cumulative chemotype abundanceand the total Planktothrix abundance determined micro-scopically validates the oligopeptide method for LakeSteinsfjorden. Further support comes from the fact thatsignificant declines of chemotypes correlated well withlosses of Planktothrix biomass actually observed in the fieldin spring 2003 (sedimentation) and winter and early spring2004 (embedment into ice, transport to shore). Moreover, a

parallel study in Lake Steinsfjorden has dismissed asignificant effect of growth recourses on oligopeptideproduction, since light and macronutrients together ac-counted only for up to 10% of variations in cellular contentof oligopeptides (Halstvedt et al. in press).

Fastner et al. and Welker et al. have found thatfreshwater cyanobacterial populations typically comprisea number of coexisting oligopeptide chemotypes (Fastner etal. 2001; Welker et al. 2004). According to our study,chemotypes can differ with regard to seasonal dynamics,depth distribution, and participation in loss processes.These differences may be a consequence of distinct cellularoligopeptide patterns or related features of chemotypesrather than a result of their taxonomic or cpcBA genotypeaffiliation. Cht1 and Cht2, for example, belong to P.rubescens and have identical cpcBA sequences but behaveddifferently under field conditions. Cht2 and Cht3 differedtaxonomically but showed similarities in oligopeptidepatterns as well as in their seasonal dynamics and depthdistribution. Furthermore, during winter and early spring2004, P. rubescens and P. agardhii belonging to Cht4 werelost in large amounts, while P. agardhii of the Cht3chemotype escaped that fate. The resulting shift from adominance of Cht4 to that of Cht3 is consistent with thehigh relative abundance of P. agardhii in 2004. Thedissimilar behavior of Cht1–4 in nature indicates differ-ences in their interaction with the environment, suggestingthat oligopeptide chemotypes possess dissimilar ecologicaltraits. This demonstrates that populations of toxic fresh-water cyanobacteria can comprise ecologically distinctsubpopulations.

Suda and coworkers (2002) could not discriminate P.rubescens from P. agardhii on the basis of 16s rDNAsequences, fatty acid composition, and cell dimensions.

Fig. 5. Linear regression between the Planktothrix biovol-ume concentration estimated with the oligopeptide method (sumover Cht1–4) and determined microscopically. The data werebased on determinations throughout the water column on 13occasions between May 2003 and August 2004. All data weretransformed using the natural logarithm. Zero values wereomitted. Statistical facts: y 5 1.08x 2 0.42, r2 5 0.88, p ,0.0001, n 5 242.

Fig. 6. Mean biovolume concentration of Planktothrix in theentire water column estimated with the oligopeptide method (sumover Cht1–4) and determined microscopically. The data werebased on determinations on 13 occasions between May 2003 andAugust 2004.


Both groups were described as separate species due to aDNA–DNA hybridization rate lower than 70% and theproduction of phycoerythrin by P. rubescens. Some parallelstudies supported this separation (Davis and Walsby 2002),while others did not (Humbert and Le Berre 2001). In LakeSteinsfjorden, P. rubescens and P. agardhii overlappedsignificantly in ecological properties and chemotype affil-

iation. The cpcBA marker discriminated most P. rubescensand P. agardhii isolates, but outliers (NIVA-CYA392, 532)and isolates with uncertain phylogeny (NIVA-CYA137,56/1, 56/3) emphasized once again the unclear taxonomicstatus of both groups. In the light of recent findings (Kanget al. 2007), even the DNA–DNA hybridization datareported by Suda and coworkers (2002) appear inconclu-

Fig. 7. Depth distribution and seasonal dynamics of Cht1–4 in Lake Steinsfjorden in 2003and 2004. The colors represent square root transformed biovolume concentrations. The red dotsmark the depth of the maximum abundance at each sampling date, if such maximum existed.Temperature contour lines are shown in black. The lake was ice covered from December 2003 tothe middle of April 2004.


sive. Taken together, available data indicate a considerableoverlap between P. rubescens and P. agardhii. Therefore,we consider them henceforth as conspecific and Cht1–4 assubpopulations of the same population.

Because chemotypes differ from each other, properties ofa given population may depend on its chemotype compo-sition, and any changes in this parameter may have aneffect on whole-population characteristics. This leavesroom for considerable variation and may explain the highintrapopulation and interpopulation diversity of freshwatercyanobacteria. A good example is the toxicity of theseorganisms. Ever since the first cyanobacterial toxins havebeen discovered, the vast variability in toxicity among andwithin populations of the same species has given rise tomany speculations (Carmichael and Gorham 1981; Ekman-Ekebom et al. 1992; Vezie et al. 1998). In LakeSteinsfjorden, more than 90% of the spatial and temporalvariation in concentration of cell-bound toxins could beexplained with fluctuations in the absolute and relativeabundance of individual Planktothrix chemotypes. Chemo-type composition was thus the main determinant ofcyanobacterial toxicity in that particular lake.

The repeated isolation of Cht1–4 from Lake Steinsfjor-den throughout our study period encompassing 33 yr

suggests a coexistence of these four chemotypes fordecades. The low number of chemotypes and their lowvariability over longer periods of time are somewhatsurprising, given the considerable frequency of geneticrecombination in cyanobacteria in general (Rudi et al.1998; Hayes et al. 2002) and in cyanobacterial oligopeptidesynthetase gene clusters in particular (Christiansen et al.2003; Mikalsen et al. 2003; Kurmayer and Gumpenberger2006). A high rate of recombination causes a continuousreorganization of genetic information and therefore shouldhave favored the evolution of new chemotypes in LakeSteinsfjorden. The low number of chemotypes is also incontrast to much higher counts reported from Microcystisand Planktothrix populations in eutrophic lakes in CentralEurope (Fastner et al. 2001; Welker et al. 2004). Overall,the above findings suggest that chemotype evolution isdriven by mechanisms that allow both stability anddiversity. We are currently conducting a study in LakeSteinsfjorden to identify the nature of these mechanisms.

The present study identified the relative chemotypecomposition of freshwater cyanobacterial populations asa dynamic parameter that can vary in space and time.These variations could not be explained with momentaryfluctuations in resource availability. In fact, while our studyproduced evidence for the control of Planktothrix growthin general by nutrient availability and light in LakeSteinsfjorden (Table 2), it failed to show any direct effectof these factors on the relative abundance of chemotypes.Growth resources may therefore determine the size of apopulation as a whole but not necessarily its chemotypecomposition or parameters that depend on chemotypecomposition. Previous studies on Prochlorococcus ecotypeshave produced contrasting data (West et al. 2001; Johnsonet al. 2006), suggesting that cyanobacterial subpopulationscan be controlled by different factors.

It is possible that shifts in the relative chemotypecomposition in Lake Steinsfjorden were related to differ-ences in depth regulation among Cht1–4. Planktothrixpossesses gas vesicles as a basis of a sophisticated buoyancyregulation system, enabling the organism to stratify indepths with favorable environmental conditions (Walsby2005). In early summer 2003, Cht1 had its maximalabundance directly above the sediment when it started todecline. This suggests a significant sedimentation of Cht1 asa reason for its decline in favor of Cht3 and Cht4.Differences in depth regulation may also explain whyCht4 accumulated in the ice in winter 2003 and 2004, whileother chemotypes did not. In this respect it is interesting to

Fig. 8. Relative chemotype composition of the Planktothrixpopulation in Lake Steinsfjorden (mean values for the entirewater column).

Table 2. Results from one-sided Spearman rank correlations between growth rates and light, temperature, dissolved inorganicphosphorus (DIP) or nitrogen (DIN). For each test, the p value and the rank coefficient are given. Number of observations per test isgiven in parentheses in the first column. Results with a significance level higher than 0.05 are printed in bold.

Chemotype (n) Light Temperature DIP DIN

Cht1 (12) 0.21, 0.34 0.49, 0.01 0.44, 0.05 0.11, 0.42Cht2 (12) 0.08, 0.44 0.6, 20.08 0.47, 0.03 0.12, 0.37Cht3 (12) 0.24, 0.22 0.62, 20.1 0.39, 0.09 0.15, 0.33Cht4 (12) 0.1, 0.4 0.54, 20.03 0.48, 0.01 0.16, 0.31Chtspooled (48) ,0.01, 0.39 0.47, 0.01 0.38, 0.05 0.01, 0.36


know that gas vesicle genotypes defined by Beard andcoworkers (2000) and oligopeptide chemotypes of thepresent study are in agreement. Whether this is coincidenceor reflects an actual functional coupling between depthregulation and oligopeptide production should be elucidat-ed in future investigations. Also, the association betweenoligopeptide chemotypes and gas vesicle genotypes impliesthat the changes in relative abundances of chemotypesobserved by us may also be caused by selection for differentgas vesicle characteristics.

Another possible explanation for variations in therelative chemotype composition is a differential effect ofzooplankton grazers and/or pathogens on Cht1–4. Mostoligopeptides are bioactive (Welker and von Dohren 2006).Some are toxic to zooplankton, while others may inhibitvirus replication (Rohrlack et al. 2001; Zainuddin et al.2002; Blom et al. 2003). Distinct cellular oligopeptidepatterns of Cht1–4 might thus have resulted in differentinteractions with herbivores and viruses. In fact, a previousstudy has shown Cht1 and Cht4 to contain potentially toxicinhibitors of Daphnia proteases, while no such compoundshave been found in Cht2 and Cht3 (Rohrlack et al. 2005).

In summary, our findings and those by authors whoworked with Prochlorococcus and Synechococcus species(Becker et al. 2002; Casamayor et al. 2002; Johnson et al.2006) suggest that populations comprising ecologicallydistinct subpopulations are common in cyanobacteria. Ahigher level of understanding of cyanobacteria may thusdemand studies that recognize the subpopulation level as amajor basis for biological processes.

References

BEARD, S. J., P. A. DAVIS, D. IGLESIAS-RODRIGUEZ, O. M. SKULBERG,AND A. E. WALSBY. 2000. Gas vesicle genes in Planktothrix spp.from Nordic lakes: Strains with weak gas vesicles possess alonger variant of gvpC. Microbiol. U.K. 146: 2009–2018.

BECKER, S., M. FAHRBACH, P. BOGER, AND A. ERNST. 2002.Quantitative tracing, by Taq nuclease assays, of a Synecho-coccus ecotype in a highly diversified natural population.Appl. Environ. Microbiol. 68: 4486–4494.

BLOM, J. F., B. BISTER, D. BISCHOFF, G. NICHOLSON, G. JUNG, R.D. SUSSMUTH, AND F. JUTTNER. 2003. Oscillapeptin J, a newgrazer toxin of the freshwater cyanobacterium Planktothrixrubescens. J. Nat. Prod. 66: 431–434.

CARMICHAEL, W. W., AND P. R. GORHAM. 1981. The mosaic natureof toxic blooms of cyanobacteria p. 161–172. In W. W.Carmichael [ed.], The water environment—algal toxins andhealth. Plenum.

CASAMAYOR, E. O., C. PEDROS-ALIO, G. MUYZER, AND R. AMANN.2002. Microheterogeneity in 16S ribosomal DNA-definedbacterial populations from a stratified planktonic environ-ment is related to temporal changes and to ecologicaladaptations. Appl. Environ. Microbiol. 68: 1706–1714.

CHORUS, I., AND J. BARTRAM. 1999. Toxic cyanobacteria in water:A guide to their public health consequences, monitoring andmanagement. Spon.

CHRISTIANSEN, G., J. FASTNER, M. ERHARD, T. BORNER, AND E.DITTMANN. 2003. Microcystin biosynthesis in Planktothrix:Genes, evolution, and manipulation. J. Bacteriol. 185: 564–572.

———, R. KURMAYER, Q. LIU, AND T. BORNER. 2006. Transpo-sons inactivate biosynthesis of the nonribosomal peptidemicrocystin in naturally occurring Planktothrix spp. Appl.Environ. Microbiol. 72: 117–123.

CODD, G., S.M.F.O. AZEVEDO, S. N. BAGCHI, M. D. BURCH, W. W.CARMICHAEL, W. R. HARDING, K. KAYA, AND H. C. UTKILEN.2005. Cyanonet—A global network for cyanobacterial bloomand toxin risk management. Initial situation assessment andrecommendations. Technical Documents in HydrologyNo. 76. UNESCO.

DAVIS, P. A., AND A. E. WALSBY. 2002. Comparison of measuredgrowth rates with those calculated from rates of photosyn-thesis in Planktothrix spp. isolated from Blelham Tarn,English Lake District. New Phytol. 156: 225–239.

Fig. 9. Linear regression between the concentration of majormicrocystins in Lake Steinsfjorden calculated on the basis ofCht1–4 abundances and the mean cellular microcystin amounts ofCht1–4 and measured directly by LC-MS/MS. The data werebased on determinations throughout the water column on 13occasions between May 2003 and August 2004. All data weretransformed using the natural logarithm. Zero values wereomitted. (A) desmethyl-microcystin LR (y 5 1.00x + 0.32, r2 50.94, p , 0.0001, n 5 236), (B) desmethyl-microcystin RR (y 51.07x 2 0.17, r2 5 0.94, p , 0.0001, n 5 242).


EKMAN-EKEBOM, M., M. KAUPPI, K. SIVONEN, M. NIEMI, AND L.LEPISTO. 1992. Toxic cyanobacteria in some Finnish lakes.Environ. Toxicol. Water Qual. 7: 201–213.

FASTNER, J., M. ERHARD, AND H. vON DOHREN. 2001. Determina-tion of oligopeptide diversity within a natural population ofMicrocystis spp. (Cyanobacteria) by typing single colonies bymatrix-assisted laser desorption ionization-time of flight massspectrometry. Appl. Environ. Microbiol. 67: 5069–5076.

HALSTVEDT, C. B., T. ROHRLACK, T. ANDERSEN, O. SKULBERG, AND

B. EDVARDSEN. 2007. Seasonal dynamics and depth distribu-tion of Planktothrix spp. in Lake Steinsfjorden (Norway)related to environmental factors. J. Plankton Res. 29:471–482.

———, ———, R. PTACNIK, AND B. EDVARDSEN. In press. On theeffect of abiotic environmental factors on production ofbioactive oligopeptides in field populations of Planktothrixspp. (Cyanobacteria). J. Plankton Res.

HAYES, P. K., G. L. BARKER, J. BATLEY, S. J. BEARD, B. A.HANDLEY, P. VACHARAPIYASOPHON, AND A. E. WALSBY. 2002.Genetic diversity within populations of cyanobacteria as-sessed by analysis of single filaments. Antonie Van Leeuwen-hoek 81: 197–202.

HUMBERT, J. F., AND B. LE BERRE. 2001. Genetic diversity in twospecies of freshwater cyanobacteria, Planktothrix (Oscilla-toria) rubescens and P. agardhii. Arch. Hydrobiol. 150:197–206.

JANSE, I., W. E. A. KARDINAAL, M. KAMST-VAN AGTERVELD, M.MEIMA, P. M. VISSER, AND G. ZWART. 2005. Contrastingmicrocystin production and cyanobacterial population dy-namics in two Planktothrix-dominated freshwater lakes.Environ. Microbiol. 7: 1514–1524.

JOHNSON, Z. I., E. R. ZINSER, A. COE, N. P. MCNULTY, E. M. S.WOODWARD, AND S. W. CHISHOLM. 2006. Niche partitioningamong Prochlorococcus ecotypes along ocean-scale environ-mental gradients. Science 311: 1737–1740.

KAEBERNICK, M., AND B. A. NEILAN. 2001. Ecological andmolecular investigations of cyanotoxin production. FEMSMicrobiol. Ecol. 35: 1–9.

KANG, C. H., AND oTHERS. 2007. Relationship between genomesimilarity and DNA-DNA hybridization among closelyrelated bacteria. J. Microbiol. Biotechnol. 17: 945–951.

KOTAI, J. 1972. Instructions for the preparation of modifiednutrient solution Z8 for algae. Publication B-11/69. Norskinstitutt for vannforskning. [in Norwegian.]

KURMAYER, R., G. CHRISTIANSEN, AND I. CHORUS. 2003. Theabundance of microcystin-producing genotypes correlatespositively with colony size in Microcystis sp. and determinesits microcystin net production in Lake Wannsee. Appl.Environ. Microbiol. 69: 787–795.

———, AND M. GUMPENBERGER. 2006. Diversity of microcystingenotypes among populations of the filamentous cyanobac-teria Planktothrix rubescens and Planktothrix agardhii. Mol.Ecol. 15: 3849–3861.

MIKALSEN, B., G. BOISON, O. M. SKULBERG, J. FASTNER, W.DAVIES, T. M. GABRIELSEN, K. RUDI, AND K. S. JAKOBSEN.2003. Natural variation in the microcystin synthetase operonmcyABC and impact on microcystin production in Micro-cystis strains. J. Bacteriol. 185: 2774–2785.

NEILAN, B. A., D. JACOBS, AND A. E. GOODMAN. 1995. Geneticdiversity and phylogeny of toxic cyanobacteria determined byDNA polymorphisms within the phycocyanin locus. Appl.Environ. Microbiol. 61: 3875–3883.

POSTIUS, C., AND A. ERNST. 1999. Mechanisms of dominance:Coexistence of picocyanobacterial genotypes in a freshwaterecosystem. Arch. Microbiol. 172: 69–75.

REPKA, S., M. KOIVULA, V. HARJUNPA, L. ROUHIAINEN, AND K.SIVONEN. 2004. Effects of phosphate and light on growth ofand bioactive peptide production by the cyanobacteriumAnabaena strain 90 and its anabaenopeptilide mutant. Appl.Environ. Microbiol. 70: 4551–4560.

ROCAP, G., D. L. DISTEL, J. B. WATERBURY, AND S. W. CHISHOLM.2002. Resolution of Prochlorococcus and Synechococcusecotypes by using 16S-23S ribosomal DNA internal transcribedspacer sequences. Appl. Environ. Microbiol. 68: 1180–1191.

ROHRLACK, T., K. CHRISTOFFERSEN, AND U. FRIBERG-JENSEN. 2005.Frequency of inhibitors of daphnid trypsin in the widelydistributed cyanobacterial genus Planktothrix. Environ. Mi-crobiol. 7: 1667–1669.

———, E. DITTMANN, T. BORNER, AND K. CHRISTOFFERSEN. 2001.Effects of cell-bound microcystins on survival and feeding ofDaphnia spp. Appl. Environ. Microbiol. 67: 3523–3529.

———, AND P. HYENSTRAND. 2007. Fate of intracellular micro-cystins in the cyanobacterium Microcystis aeruginosa. Phyco-logia 46: 277–284.

———, AND H. C. UTKILEN. 2007. Effects of nutrient and lightavailability on production of bioactive anabaenopeptins andmicroviridin by the cyanobacterium Planktothrix agardhii.Hydrobiologia 583: 231–240.

———, AND oTHERS. 2003. Isolation, characterization, andquantitative analysis of Microviridin J, a new Microcystismetabolite toxic to Daphnia. J. Chem. Ecol. 29: 1757–1770.

RUDI, K., O. M. SKULBERG, AND K. S. JAKOBSEN. 1998. Evolutionof cyanobacteria by exchange of genetic material amongphylogentically related strains. J. Bacteriol. 180: 3453–3461.

SAKAMOTO, Y., M. ISHIGURO, AND G. KITAGAWA. 1986. Akaikeinformation criterion statistics. D. Reidel.

SCHAGERL, M., AND B. MULLER. 2006. Acclimation of chlorophylla and carotenoid levels to different irradiances in fourfreshwater cyanobacteria. J. Plant Physiol. 163: 709–716.

SKULBERG, O. M., AND R. SKULBERG. 1985. Planktic species ofOscillatoria (Cyanophyceae) from Norway—characterizationand classification. Arch. Hydrobiol. Suppl. 71: 157–174.

SUDA, S., M. M. WATANABE, S. OTSUKA, A. MAHAKAHANT, W.YONGMANITCHAI, N. NOPARTNARAPORN, Y. D. LIU, AND J. G.DAY. 2002. Taxonomic revision of water-bloom-formingspecies of oscillatorioid cyanobacteria. Int. J. Syst. Evol.Microbiol. 52: 1577–1595.

TOOMING-KLUNDERUD, A., T. ROHRLACK, K. SHALCHIAN-TABRIZI,T. KRISTENSEN, AND K. S. JAKOBSEN. 2007. Structural analysisof a non-ribosomal halogenated cyclic peptide and its putativeoperon from Microcystis: Implications for evolution ofcyanopeptolins. Microbiol. SGM 153: 1382–1393.

VAITOMAA, J., A. RANTALA, K. HALINEN, L. ROUHIAINEN, P.TALLBERG, L. MOKELKE, AND K. SIVONEN. 2003. Quantitativereal-time PCR for determination of microcystin synthetase Ecopy numbers for Microcystis and Anabaena in lakes. Appl.Environ. Microbiol. 69: 7289–7297.

VEZIE, C., L. BRIENT, K. SIVONEN, G. BERTRU, J. LEFEUVRE, AND

M. SALKINOJA-SALONEN. 1998. Variation of microcystincontent of cyanobacterial blooms and isolated strains in lakeGrand-Lieu (France). Microb. Ecol. 35: 126–135.

WALSBY, A. E. 2005. Stratification by cyanobacteria in lakes: Adynamic buoyancy model indicates size limitations met byPlanktothrix rubescens filaments. New Phytol. 168: 365–376.

WELKER, M., G. CHRISTIANSEN, AND H. vON DOHREN. 2004.Diversity of coexisting Planktothrix (Cyanobacteria) chemo-types deduced by mass spectral analysis of microcystins andother oligopeptides. Arch. Microbiol. 182: 288–298.

———, AND H. vON DOHREN. 2006. Cyanobacterial peptides—nature’s own combinatorial biosynthesis. FEMS Microbiol.Rev. 30: 530–563.


WEST, N. J., W. A. SCHONHUBER, N. J. FULLER, R. I. AMANN, R.RIPPKA, A. F. POST, AND D. J. SCANLAN. 2001. Closely relatedProchlorococcus genotypes show remarkably different depthdistributions in two oceanic regions as revealed by in situhybridization using 16S rRNA-targeted oligonucleotides.Microbiol. SGM 147: 1731–1744.

WOITKE, P., K. HESSE, AND J. G. KOHL. 1997. Changes in thelipophilic photosynthetic pigment contents of differentMicrocystis aeruginosa strains in response to growth irradi-ance. Photosynthetica 33: 443–453.

ZAINUDDIN, E. N., S. MUNDT, U. WEGNER, AND R. MENTEL. 2002.Cyanobacteria a potential source of antiviral substancesagainst influenza virus. Med. Microbiol. Immunol. 191:181–182.

Received: 18 June 2007Accepted: 31 January 2008

Amended: 19 February 2008
