Lipid biomarker signatures as tracers for harmful cyanobacterial blooms in the Baltic Sea

The recent proliferation of harmful cyanobacterial blooms (cyanoHABs) in the Baltic and other marginal seas poses a severe threat for the health of infested ecosystems as e.g. the massive export and decay of cyanobacterial biomass facilitates the spread of bottom water hypoxia. There is evidence that cyanoHABs occurred repeatedly in the Baltic Sea but knowledge of their spatiotemporal distribution and the cyanobacteria that contributed to them is limited. In this study, we examined representatives of the major bloom-forming heterocystous cyanobacteria (i.e. Aphanizomenon, Dolichospermum (formerly Anabaena) and Nodularia) to establish lipid fingerprints that allow tracking these environmentally important diazotrophs in the modern and past Baltic Sea. The distribution of normal and mid-chain branched alkanes, fatty acid methyl esters, bacteriohopanepolyols and heterocyst glycolipids permitted a clear chemotaxonomic separation of the different heterocystous cyanobacteria but also indicated a close phylogenetic relationship between representatives of the genera Aphanizomenon and Dolichospermum. Compared to the discontinuous nature of phytoplankton surveys studies, the distinct lipid profiles reported here will allow obtaining detailed spatiotemporal information on the frequency and intensity of Baltic Sea cyanoHABs as well as their community composition using the time-integrated biomarker signatures recorded in surface and subsurface sediments. As heterocystous cyanobacteria of the genera Aphanizomenon, Dolichospermum and Nodularia are generally known to form massive blooms in many brackish as well as lacustrine systems worldwide, the chemotaxonomic markers introduced in this study may allow investigating cyanoHABs in a great variety of contemporary environments from polar to tropical latitudes.


Abstract
The recent proliferation of harmful cyanobacterial blooms (cyanoHABs) in the Baltic and other marginal seas poses a severe threat for the health of infested ecosystems as e.g. the massive export and decay of cyanobacterial biomass facilitates the spread of bottom water hypoxia. There is evidence that cyanoHABs occurred repeatedly in the Baltic Sea but knowledge of their spatiotemporal distribution and the cyanobacteria that contributed to them is limited. In this study, we examined representatives of the major bloom-forming heterocystous cyanobacteria (i.e. Aphanizomenon, Dolichospermum (formerly Anabaena) and Nodularia) to establish lipid fingerprints that allow tracking these environmentally important diazotrophs in the modern and past Baltic Sea. The distribution of normal and mid-chain branched alkanes, fatty acid methyl esters, bacteriohopanepolyols and heterocyst glycolipids permitted a clear chemotaxonomic separation of the different heterocystous cyanobacteria but also indicated a close phylogenetic relationship between representatives of the genera Aphanizomenon and Dolichospermum. Compared to the discontinuous nature of phytoplankton surveys studies, the distinct lipid profiles reported here will allow obtaining detailed spatiotemporal information on the frequency and intensity of Baltic Sea cyanoHABs as well as their community composition using the time-integrated biomarker signatures recorded in surface and subsurface sediments. As heterocystous cyanobacteria of the genera Aphanizomenon, Dolichospermum and Nodularia are generally known to form massive blooms in many brackish as well as lacustrine systems worldwide, the chemotaxonomic markers introduced in this study may allow investigating cyanoHABs in a great variety of contemporary environments from polar to tropical latitudes. PLOS

Introduction
Massive accumulation of unicellular and/or filamentous cyanobacteria-known as harmful cyanobacterial blooms (cyanoHABs)-have significantly increased in abundance, intensity and duration in both brackish and freshwater environments over the last decades [1][2][3]. In the modern Baltic Sea, cyanoHABs are particularly frequent and during summer can cover large areas of surface water with profound impacts on the health of the aquatic ecosystem [4,5]. This is because the decay of organic matter and its associated O 2 consumption has resulted in a fourfold increase in the area affected by bottom water hypoxia and the spread of ecological 'dead zones' in the recent past [6,7]. As a consequence, the modern Baltic Sea experiences a drastic loss of viable habitat of both benthic and pelagic faunal communities, as well as significant changes in nutrient cycling and ecosystem structure [8,9]. The negative feedback on the Baltic Sea ecology is amplified by the fact that many bloom-forming cyanobacteria express the ability to synthesize strong hepato-and neurotoxins (e.g. anatoxin, microcystin, nodularin, or saxitoxin), which pose a significant health risk for both animals and humans and have led to the repeated poisoning of wildlife along the coasts of Sweden and Finland [10]. The major bloom-forming diazotrophs in the contemporary Baltic Sea belong to three genera of filamentous heterocystous cyanobacteria including Aphanizomenon, Dolichospermum (previously classified as Anabaena) and Nodularia [11,12]. Representatives of these genera are considered important sources of combined N, as they add an estimated 180-420 Gg N yr -1 to the surface waters of the Baltic Sea, which is similar in magnitude to the entire riverine load and twice the atmospheric input [12,13]. Long term monitoring programs indicate that the frequency and abundance of cyanoHABs have increased considerably since the early 1900s due to the massive anthropogenic loading of nutrients (in particular phosphorous) to the Baltic Sea [4,14]. Given an undiminished high nutrient input, in combination with an anticipated increase in surface water temperature by 3-4˚C until the end of the 21 st century [15], it has been predicted that cyanobacterial bloom-formation will be significantly enhanced in the future, along with intensification and greater frequency of hypoxia [9]. Current estimates on the future development of cyanobacterial blooms and their consequences for the Baltic Sea show, however, large uncertainties due to our incomplete understanding of the spatiotemporal distribution of cyanoHABs in the Baltic Sea over time and the anthropogenic as well as environmental controls that promote bloom formation.
Sediment-hosted lipid biomarkers constitute a powerful tool in ecosystem studies as they i) provide a time-integrated signal of the occurrence and composition of cyanoHABs and ii) allow characterizing the past occurrence and composition of such blooms in a high spatiotemporal resolution over geological timescales, provided that they can be linked unequivocally to their biological sources [16,17]. Biomarkers considered to be characteristic for cyanobacteria and potentially suited for tracing cyanoHABs in modern and past ecosystems include midchain branched alkanes [18, 19], C-2 methylated bacteriohopanepolyols (BHPs) as well as their diagenetic products [20], and BHP structures (such as C-30,32,33,34,35-pentol BHP) only observed in cyanobacteria to date [21]. In addition, the polyunsaturated fatty acids (PUFAs) 18:2ω6, 18:3ω6 and 18:4ω3 have been reported in high abundances in certain genera of both unicellular as well as filamentous heterocystous cyanobacteria [22,23]. These components are frequently used to investigate the composition of cyanobacterial communities in microbial mats [24] and extreme environments [25]. Heterocyst glycolipids (HGs) represent a novel suite of biomarkers that have been reported exclusively from the heterocyst cell envelope of heterocystous cyanobacteria and as such constitute the first and only biomarkers that allow tracing for the process of cyanobacterial N 2 fixation in natural systems [26,27]. Interestingly, the distribution of HGs has been shown to vary at least on a family level in cultured cyanobacteria [26,27], so HGs may hold great promise for providing detailed information on the composition of cyanoHABs in aquatic ecosystems (such as the Baltic Sea) over time.
Although the morphology and molecular diversity of cyanobacteria in general and of heterocystous species in particular have been extensively studied in the past [28][29][30][31][32], there is surprisingly little comprehensive information on their biomarker inventory. Detailed lipid profiles, focusing mainly on cellular fatty acids and mid-chain branched alkanes, have been obtained from a few key genera, including among others Anabaena, Nostoc, Synechocystis or Oscillatoria [18,22,33,34], but the great majority of cyanobacteria has yet only been inadequately studied. Furthermore, these investigations are largely biased towards freshwater cyanobacteria, whereas little is known about the lipid inventory of cyanobacteria growing in marine and brackish environments such as the Baltic Sea. In this study, we qualitatively assessed the lipid profiles of eight bloom-forming heterocystous cyanobacteria in order to (i) determine whether the distribution of specific lipids or individual lipid classes allows the chemotaxonomic separation of the studied genera and (ii) identify biomarkers suitable for tracing N 2 fixing heterocystous cyanobacteria in the Baltic Sea and possibly other aquatic environments.

Cyanobacterial cultures
Eight strains of heterocystous cyanobacteria belonging to the genera Dolichospermum (315, BIR53, BIR169), Aphanizomenon (TR183) and Nodularia (AV1, BY1, F81, HEM) had been isolated from the Baltic Sea and were obtained from the Culture Collection of Cyanobacteria (HAMBI) maintained at the Department of Food and Environmental Sciences at the University of Helsinki. Culture conditions are provided in S1 Table. Briefly, they were grown in 0.5 l Erlenmeyer flasks containing 0.2 l Z8X culture medium that was free of combined N and supplemented with NaCl at 8.75 g l -1 [35]. All strains were grown under continuous white fluorescent light at an irradiance of 8-9 μE m -2 s -1 and at a temperature of either 18˚C (BIR53, BIR169, BY1, F81) or 23˚C (315, TR183, AV1, HEM), respectively. All cultures were harvested towards the end of the logarithmic growth phase by centrifugation and stored frozen afterwards. Upon arrival at Christian-Albrechts-University, cells were lyophilized and kept at -20˚C prior to solvent extraction. Dolichospermum BIR53 and BIR169 as well as Nodularia BY1 and F81 were non-axenic but on the basis of microscopy showed only low levels of bacterial contamination.

Solvent extraction
Freeze dried biomass (ca. 80 mg) was extracted using a modified Bligh and Dyer technique [36]. Briefly, the cell material was extracted ultrasonically 3x for 10 min in a mixture of MeOH, dichloromethane (DCM) and phosphate buffer (2:1:0.8; v/v/v). After each sonication step, the mixture was centrifuged at 1,500 × g for 6 min and the supernatant transferred to a centrifuge tube. The combined supernatants were phase separated by adding DCM and phosphate buffer to a final solvent ratio of 1:1:0.9 (v/v/v). The organic bottom layer was transferred to a round bottom flask and reduced under vacuum using a rotary evaporator. Each Bligh and Dyer extract (BDE) was transferred to a pre-weighed glass vial using DCM:MeOH (9:1; v/v) and dried under a gentle stream of N 2 . respective eluents. The apolar fraction, containing aliphatic hydrocarbons, was dried under a gentle stream of N 2 and dissolved in n-hexane to a concentration of 1 mg ml -1 prior to analysis. Gas chromatography-mass spectrometry (GC-MS) was performed using an Agilent 7820A gas chromatograph equipped with a Phenomenex ZB-5 fused silica column (30 m × 0.25 mm i.d.; 0.25 μm film thickness). Helium (He) was used as carrier gas at a constant 1 ml min -1 . The injector was at 70˚C and the oven programmed to 130˚C at 20˚C min -1 and then at 4˚C min -1 to 320˚C (held 10 min). The GC instrument was interfaced with an Agilent 5975 MSD mass spectrometer operated at 70 eV and scanning m/z 50 to 650 with a cycle time of 1.7 s (resolution 1000). Straight chain and branched alkanes were assigned from retention time indices as described in Kissin et al. [37] and comparison with published mass spectra [38,39].

Fatty acid methyl ester analysis
Freeze-dried and homogenized cell material (3 mg) was extracted and transesterfied as detailed in Mudimu et al. [40]. The extract was dissolved in DCM to a concentration of 0.5 mg ml -l and subjected to the analysis of FAMEs using an Agilent 7820A gas chromatograph equipped with a Phenomenex ZB-Wax column (30 m × 0.25 mm i.d.; 0.50 μm film thickness) and connected to a flame ionization detector (FID). Hydrogen was the carrier gas at 1 ml min -1 . The oven temperature was programmed from 50˚C (1 min) to 200˚C at 25˚C min -1 , then to 230˚C (held 18 min) at 3˚C min -1 . FAMEs were assigned by comparison of retention times with those of the "Supelco 37 Component FAME mix" (Sigma-Aldrich, Germany).
FAMEs not included in the standard mixture were assigned from their mass spectral characteristics using an Agilent 7820A gas chromatograph coupled to an Agilent 5975 MSD mass spectrometer. Analytical conditions were as above. Assignment of FAMEs was accomplished by comparison of mass spectra with those in the "Lipid Library" of the American Oil Chemists' Society (http://lipidlibrary.aocs.org).

Heterocyst glycolipid analysis
An aliquot of the BDE (ca. 1 mg) was dissolved in DCM:MeOH (9:1, v/v) to a concentration of 0.5 mg ml -1 and filtered through a 0.45 μm regenerated cellulose syringe filter. Heterocyst glycolipids were analyzed as described by Bauersachs et al. [41] with some modifications. Briefly, normal phase high performance liquid chromatography (HPLC) was carried out using a Waters Alliance 2690 HPLC system fitted with a Phenomenex Luna NH 2 column (150 × 2 mm; 3 μm particle size) and a guard column of the same material, which both were maintained at 30˚C. Separation of HGs was accomplished with the following gradient profile: 95% A/5% B to 85% A/15% B in 10 min (held 7 min) at 0.5 ml min -1 , followed by back flushing with 30% A/ 70% B at 0.2 mg ml -1 for 25 min and re-equilibrating the column with 95% A/5% B for 15 min. Detection of HGs was performed using a Quattro LC triple quadrupole mass spectrometer (Micromass, UK) equipped with an electrospray ionization (ESI) interface and operated in positive ion mode. Source conditions were: capillary 3.5 kV, cone 20 V, desolvation temperature 230˚C, source 120˚C; cone gas 80 l h -1 and desolvation gas 230 l h -1 . BDEs were initially screened in data-dependent mode with two scan events, where a scan of m/z 300-1000 was followed by a product ion scan of the base peak in the spectrum of the first scan event. Assignment of HGs was achieved by comparison of mass spectra published in the literature [42,43]. As some of the eight identified target compounds occurred only in low abundance, all cultures were re-analyzed using multiple reaction monitoring (MRM) to increase sensitivity. MRM experiments were performed as outlined by Bauersachs et al. [44] with transitions monitored as follows: m/z 547 ! 415 (pentose HG 26 diol), m/z 561 ! 415 (deoxyhexose HG 26 diol), m/z 575 ! 413 (HG 26 keto-ol), m/z 577 ! 415 (HG 26 diol), m/z 603 ! 441 (HG 28 keto-ol), m/z 605 ! 443 (HG 28 diol), m/z 619 ! 457 (HG 28 keto-diol) and m/z 621 ! 459 (HG 28 triol). HGs of higher chain length were not detected in the investigated Baltic Sea cyanobacteria.
As the structure of the HG backbones cannot be unequivocally determined by HPLC/MS alone, cell material of each heterocystous cyanobacterium was acid hydrolyzed and analyzed following the procedure detailed in Bauersachs et al. [42]. Briefly, freeze-dried biomass (~30 mg) of each cyanobacterium was suspended in a mixture of MeOH:10.2M hydrochloric acid (95:5; v/v) and refluxed for 3 h at 80˚C to cleave the aglycone moieties from their sugar head groups. HPLC-grade water was then added and the reaction products were extracted thrice with DCM:MeOH (2:1; v/v). The organic bottom layer was transferred to a round bottom flask and the remaining aqueous phase extracted twice with DCM. Subsequently, the hydrolyzed total lipid extracts were reduced under vacuum and dried under a gentle stream of N 2 . The liberated polyhydroxyalcohols were converted to trimethylsilyl ethers by heating at 60˚C for 2 h with 25 μl of each bis(trimethylsilyl)trifluoroacetamide (BSTFA) and pyridine prior to GC-MS analysis.
The derivatized polyhydroxyalcohols were analysed using an Agilent 7820A gas chromatograph fitted with a Phenomenex ZB-5 fused silica column (30 m × 0.25 mm i.d.; 0.25 μm film thickness). Helium (He) was used as carrier gas at a constant 1 ml min -1 . The injector was at 70˚C and the oven programmed to 130˚C at 20˚C min -1 and then at 5˚C min -1 to 320˚C (held 20 min). The GC instrument was interfaced with an Agilent 5975 MSD mass spectrometer operated at 70 eV and scanning m/z 50 to 850 with a cycle time of 1.7 s (resolution 1000). Identification of HG degradation products was performed by comparison with published mass spectra [42,45].

Bacteriohopanepolyol analysis
Analysis of BHPs was carried out at Newcastle University. Freeze-dried biomass (ca. 80 mg) was extracted using a modified Bligh and Dyer technique as described previously [46]. In brief, the biomass was weighed into a Teflon centrifuge tube and extracted using 19 ml of a monophasic mixture of MeOH, chloroform (CHCl 3 ) and H 2 O (10:5:4; v/v/v). The mixture was sonicated (40˚C, 10 min) and centrifuged at 17,200 × g for 15 min. The supernatant was then transferred to a new centrifuge tube and the residue was re-extracted twice using the same procedure. The tube containing the supernatant was phase-separated by adding 5 ml of each CHCl 3 and H 2 O and subsequent centrifugation at 17,200 × g for 5 min. The organic bottom layer was rotary evaporated to near dryness and transferred to a pre-weighed glass vial using CHCl 3 :MeOH (2:1; v/v) and dried under a gentle stream of N 2 . An aliquot of the BDE (ca. 6 mg) was acetylated by heating with 0.5 ml acetic anhydride (Ac 2 O):pyridine (1:1; v/v) at 50˚C for 1 h and letting it react at room temperature overnight. The derivatization agents were removed under N 2 and the acetylated TLE dissolved in 0.5 ml MeOH:2-propanol (3:2; v/v) prior to analysis.
Analysis of BHPs was achieved using reversed phase HPLC-atmospheric pressure chemical ionization mass spectrometry (HPLC-APCI-MS n ) with a ThermoFinnigan surveyor HPLC system equipped with a Phenomenex Gemini C 18 column (150 × 3.0 mm, 5 μm particle size) and a security guard column cartridge of the same material, coupled to a Finnigan LCQ ion trap mass spectrometer. BHPs were eluted at a constant 30˚C and using the following gradient: 90% A/10% B (0 min) to 59% A/1% B/ 40% C (at 25 min), then isocratic to 45 min, returning to the starting conditions in 5 min and stabilizing for 10 min with A = MeOH, B = H 2 O, C = 2-propanol. Solvent flow was 0.5 ml min -1 . The mass spectrometer was equipped with an APCI source operating in positive ionization mode with the following settings: capillary 150˚C, APCI vaporizer 490˚C, corona discharge 8 μA, sheath gas flow 40 and auxiliary gas flow of 10. MS n analysis was carried out in data dependent mode with 2 scan events. Scan event 1 was a full MS scan from m/z 300-1300. Scan event 2 was data dependent and utilized the most intense ions from scan event 1 to perform MS 2 scans. Assignment of BHPs was achieved by comparison of mass spectra with those published in the literature [46,47].

Statistical analysis
Correspondence analysis (CA) was performed to investigate similarities/differences in whole lipid profiles and distribution pattern of individual lipid classes using XLSTAT (v. 18.06) software package for Microsoft Excel. The axes of the CA plots essentially measure the amount of variation from the average, with the most common lipids located closest to the origin and those being highly specific for a certain species or genera visualizing at the plot extremes [48]. The axes of the graphs were chosen to show the largest fraction of variability in the multidimensional data matrix. Thus, the most informative axes are in the direction of maximum variance of the scatter of points. The inertia value of an axis is the proportion of variance explained by it.

Straight chain and branched hydrocarbons
The hydrocarbons included straight chain as well as mono-and dimethyl alkanes with 16 to 19 carbons and distinct distribution patterns among the genera (Fig 1; S2 Table). In species belonging to Dolichospermum (formerly Anabaena), n-heptadecane (n-C 17:0 ) was the only hydrocarbon detected, with the exception of strain BIR53, in which traces (<1%) of n-hexadecane (n-C 16:0 ) and n-heptadecene (n-C 17:1 ) were also present. These results are well in line with the previously reported hydrocarbon distribution in freshwater representatives of this genus [34]. n-Heptadecane has been reported to be among the most prominent hydrocarbons in various cultured unicellular and filamentous heterocystous cyanobacteria [49][50][51]. It also occurred in high abundances in hypersaline microbial mats [52] as well as hot springs [53,54] and in such extreme environments is generally considered to be of cyanobacterial origin. However, n-C 17:0 is present in many aquatic algae and photosynthetic bacteria [55] and in nonextreme environments it may not serve as an indicator for a cyanobacterial contribution to the sedimentary organic matter. Biological markers that are more specific for a cyanobacterial origin are mid-chain branched mono-, di-, and trimethylalkanes [18, 49,56]. These components were not detected in any of the investigated Dolichospermum strains, although 7-methylheptadecane (7Me-C 17:0 ) and 8-methylheptadecane (8Me-C 17:0 ) have been reported in a representative of this genus before [34,57]. None of the Dolichospermum strains isolated from the Baltic Sea seems, however, to be a potent source of mid-chain branched alkanes.
Strain TR183 was the only heterocystous cyanobacterium of the genus Aphanizomenon investigated here and to the best of our knowledge is the first representative of this genus analyzed for its hydrocarbon composition. It contained n-C 17:0 as the most dominant hydrocarbon (89.7%) and traces of n-C 16:0 (0.3%; S2 Table). Small quantities of monomethyl alkanes (MMAs; 10.0%) were also detected with 8Me-C 17:0 the most dominant (7.1%) followed by 7Me-C 17:0 (2.9%). Interestingly, 8Me-C 17:0 was not found in any of the other heterocystous cyanobacteria and it may thus be of chemotaxonomic value in paleoenvironmental studies addressing the role of N 2 -fixing heterocystous cyanobacteria in the Baltic Sea. It should be noted, however, that 8Me-C 17:0 and 7Me-C 17:0 constitute the most frequently reported MMAs from extant cyanobacteria, including both unicellular and filamentous species [34,58], and at this point it cannot be excluded that diazotrophs other than Aphanizomenon TR183 may provide a source of 8Me-C 17:0 in the Baltic Sea.
The apolar fraction of Nodularia spp. contained only a low relative proportion of n-C 17:0 (0.6 to 3.9%) but abundant MMAs and dimethyl alkanes (DMAs) that constituted >95% of all hydrocarbons. The major MMAs were 7Me-C 17:0 and 6-methylheptadecane (6Me-C 17:0 ), which together accounted for 65.3 to 87.8% of all hydrocarbons, respectively (Fig 1; S2 Table). We also observed traces (<1%) of 7-methylhexadecane (7Me-C 16:0 ) in all strains and 7-methylpentadecane (7Me-C 15:0 ) exclusively in Nodularia F81. A unique feature of Nodularia spp. was the presence of DMAs that occurred in comparatively high abundance (7.6 to 32.9%; S2 Table). The 7,11Me-C 17:0 was usually dominant, ranging from 5.9 to 27.5%, while 6,12Me-C 17:0 occurred only in small quantity (<5%). DMAs have been reported from some pure cultures of filamentous cyanobacteria including Calothrix scopulorum [56] and Phormidium luridum [19] but they usually occur in low abundance and, compared with MMAs, seem to be less widespread in cyanobacteria. However, they are frequently recognized in modern coastal and hot spring microbial mats, in which cyanobacteria represent a major constituent of the microbial community [39, 53,54] and in the Baltic Sea they may represent valuable biomarkers for tracing heterocystous cyanobacteria of the genus Nodularia.
In addition to the suite of straight-chain and branched alkanes and alkenes, we also detected the pentacyclic triterpenoid hop-22(29)-ene (diploptene) in the hydrocarbon fraction of all Dolichospermum and Aphanizomenon strains. This component has been reported from a vast number of cultured unicellular and filamentous cyanobacteria [18,59] and it is also known to occur ubiquitously in modern hot spring microbial mats, where it is commonly attributed to a cyanobacterial origin [58,60]. However, it should be pointed out that it is widely distributed within the bacterial domain [59] and has been reported, among others, from sulfate reducing bacteria [61] as well as methanotrophic bacteria [62]. It is interesting to note that diploptene and bacteriohopanepolyols (see below) were not detected in any of the four Nodularia strains because both are considered essential in regulating membrane rigidity and permeability in the prokaryotic cell [63]. Our results are, however, in agreement with the observation that <10% of all prokaryotes may in fact synthesize hopanoid structures [64], which raises the questions about whether components other than hopanoids regulate the fluid transport across the cell membrane of Nodularia spp. or whether these components are indeed essential for the biology of the prokaryotic cell.

Saturated, monounsaturated and polyunsaturated fatty acids
Fatty acids (FA) constitute a major component of the bacterial and eukaryotic cell membrane. They are ubiquitously present in sedimentary environments, where their characteristic distribution patterns are studied to obtain information on the structure and composition of modern and fossil microbial communities [65]. Pure cultures of cyanobacteria are generally characterized by relatively simple cellular FA distributions dominated by 16:0, 16:1ω9, 18:2ω6 or 18:3ω3 [33, 66,67]. The dominance of these components in e.g. microbial mats [24, 68] is usually taken as evidence for the presence of cyanobacteria; in particular in extreme settings where the harsh environmental conditions limit the activity of eukaryotes [25].
Here, 16:0 (25.1±2.7%), 18:2ω6 (9.5±4.6%) and 18:3ω3 (31.9±7.8%) were the dominant FAMEs in all of the Dolichospermum strains (Fig 2; S3 Table). Other FAMEs occurring in lower abundance included 14:0 (5.6±2.1%), 16:1ω9 (4.8±0.3%), 16:3ω3 (7.7±3.0%) and 18:1ω9 (5.2±2.9%). Taken together, these seven components accounted for the majority of the identified FAMEs (!80%). A common characteristic of all Dolichospermum strains was the comparatively high content of polyunsaturated fatty acids (PUFAs), which accounted for 53.5±5.1% of the total FAMEs. Saturated fatty acids (SAFAs) and monounsaturated fatty acids (MUFAs) each occurred in significantly lower abundance, constituting 31.3±2.3% and 15.2±5.7%, respectively (S3 Table). In terms of distribution and relative abundance, the FAME content of our Dolichospermum strains is thus in good agreement with those previously reported for freshwater representatives of this genus [23, 33, 69] and isolates obtained from the Baltic Sea [67,70]. All Dolichospermum strains exclusively contained higher relative proportions of 14:0, which occurred in trace amount only in the Aphanizomenon and Nodularia strains (<1%) with the exception of Nodularia HEM, where it constituted 2.7% (S3 Table). In a previous study on the FAME content of Baltic Sea cyanobacteria, 14:0 was found to be abundant only in hepatotoxic Dolichospermum (previously defined as Anabaena) strains [67]. Although this finding is not supported by our study, as both non-toxic (BIR53 and BIR169) and hepatotoxic (315) Dolichospermum strains contained a similar proportion of 14:0, its generally low abundance in the other cyanobacterial genera investigated here suggests that it is of chemotaxonomic relevance and may be suited to tracing the presence of Dolichospermum spp. in Baltic Sea surface waters and sediments.
The FAME profile of Aphanizomenon TR183 was dominated by 16:0 (31.9%), 18:2ω6 (12.9%) and 18:3ω3 (31.4%) which, taken together, constituted > 76.2% of all FAMEs. In addition, we detected 16:1ω7 (4.1%), 16:3ω3 (5.2%) and 18:1ω9 (6.4%) in moderate abundances and low or trace amounts of 14:1ω9 (0.2%), 16:1ω9 (2.9%), 16:2ω6 (3.6%), 18:0 (0.6%) and 18:1ω7 (0.7%). PUFAs constituted 53.0%, while SAFAs and MUFAs accounted for 32.8% and 14.2%, respectively (S3 Table). The FAME profile thus resembles those reported for Aphanizomenon TR183 [67] and other representative of this genus isolated from the Baltic Sea [70]. Comparison of the distribution pattern and relative abundance of individual FAMEs in Aphanizomenon TR183 and the three Dolichospermum strains showed an overall strong similarity, except for the virtual absence of 14:0 from Aphanizomenon TR183, suggesting a close phylogenetic relationship between both genera. This result is generally in line with data obtained from the analysis of 16S rRNA sequences, showing that species of Aphanizomenon and Dolichospermum occur interspersed in phylogenetic trees [32, 69,71] and that Aphanizomenon TR183 clusters together with neurotoxic Dolichospermum strains isolated from Finnish freshwater lakes [30]. As only one strain of the genus Aphanizomenon was investigated here, it is of course difficult to generalize our results. However, molecular analysis suggests that the genetic diversity in Aphanizomenon is low, with only one phenotype dominating in the Baltic Sea [72,73]. Hence, FAME profiles for representatives of the genus Aphanizomenon are expected to show little intraspecies variation. Such a perception is in line with the analysis of FAME profiles in five planktonic Aphanizomenon strains (including TR183) that in general show a very similar distribution pattern [67].
Based on variation in FA content, cyanobacteria are currently divided in four distinct groups [33,75]. The first group is devoid of PUFAs and contains only SAFAs and MUFAs. The second and third groups are characterized either by the presence of 18:3ω3 or 18:3ω6, respectively, while the fourth also has 18:4ω3. According to this classification, all Dolichospermum and Aphanizomenon strains investigated here can be assigned to group 2. On the contrary, heterocystous cyanobacteria belonging to the genus Nodularia fall within group 4 of the Kenyon-Murata classification scheme. Hence, compositional differences in the FAME distribution pattern between bloom-forming Baltic Sea cyanobacteria allow a clear separation of representatives of the genus Nodularia from the genera Dolichospermum and Aphanizomenon and again suggest a close phylogenetic relationship between the latter two genera.

Heterocyst glycolipids
Heterocyst glycolipids occur exclusively in the heterocyst cell envelope of N 2 -fixing heterocystous cyanobacteria [76], where they are considered to act as a gas diffusion barrier that limits the entry of atmospheric gases into the heterocyst [77]. The structure of HGs consists of a sugar head group (e.g. hexose, deoxyhexose or pentose) glycosidically bound to an even num- Dolichospermum BIR169 is interesting given that both components have to date been reported in major abundances only from representatives of the genus Calothrix (Rivulariaceae) [27] and the freshwater cyanobacterium Cylindrospermopsis raciborskii (Nostocaceae) [43]. In the Baltic Sea, HG 28 triol and HG 28 keto-diol have been found in surface sediments of the Landsort Deep [78] but their biological origin was elusive. Given that species belonging to Calothrix have only been described from benthic microbial mats of the Gulf of Finland [10], planktonic heterocystous cyanobacteria of the genus Dolichospermum may represent a potent source of HG 28 triol and HG 28 keto-diol in the Baltic Proper. In addition, we also detected traces of a HG 26 diol with a pentose instead of the more commonly reported hexose head group (pentose HG 26 diol) in all Dolichospermum strains (0.2±0.1%). This component has only been reported from Aphanizomenon ovalisporum and C. raciborskii so far [43]. Its subordinate presence in the Baltic Sea Dolichospermum strains as well as in representatives of the other genera investigated here (see below) suggests, however, that it may be more widespread among heterocystous cyanobacteria.
Aphanizomenon TR183 showed a HG distribution that similar to the other investigated component classes revealed a strong resemblance to those observed in the Dolichospermum strains (Fig 4). Two structural isomers of the HG 26 diol (together comprising 68.2%) and HG 26 keto-ol (13.2%) were the most dominant components, constituting 81.6% of all HGs (S4 Table). In addition to HG 26 diol (60.1%) eluting at 9.3 min, we observed an early eluting structural isomer (HG 26 diol II; 8.1%) at 6.8 min that has been reported from an axenic culture of Anabaena CCY9613 [44] and water column filtrates taken from a small freshwater lake in northern Germany [79]. Acid hydrolysis of cell material of Aphanizomenon TR183 released high amounts of a polyhydroxyalcohol with MS characteristics corresponding to those reported for 1,3,25-hexacosanetriol [42,45], the degradation product of HG 26 diol. Structural isomers of this component were not detected, suggesting that both HG 26 diols in Aphanizomenon TR183 differed in the type of head group attached to the aglycone moiety. Indeed, the composition and configuration of the sugar functionality in heterocystous cyanobacteria is known to be highly variable, with α-and β-glucosides usually dominating over α-mannosides and/or α-galactosides [26]. Other HGs detected in Aphanizomenon TR183 included small quantities of HG 28 diol (2.9%) and traces of HG 28 keto-ol (0.4%) together with pentose HG 26 diol (0.3%). The suite of HGs was completed by the presence of comparatively high relative abundances of HG 28 triol (12.7%) and keto-diol (2.3%), constituting 15.0% to all HGs. Although both components occurred in lower proportions than in Dolichospermum BIR169, it seems plausible that Aphanizomenon TR183 constitutes a second important biological source of HG 28 triol and HG 28 keto-diol in the Baltic Sea. In fact, the HG distribution pattern as well as the relative abundance of individual HGs in Aphanizomenon TR183 closely resemble those reported from surface sediments of the Landsort Deep [78]. This strong similarity suggests that Aphanizomenon TR183 may be a major bloom former in the Baltic Sea, in full agreement with phylogenetic analysis and studies of the species composition of cyanoHABs in the modern Baltic Sea [10,80].
All Nodularia strains expressed a comparatively simple HG distribution pattern with only four different types of HGs present. HG 26 diols (89.8±12.8%) and HG 26 keto-ol (9.0±7.7%) were again the most abundant HGs, which together accounted for 97.7 to 99.4% of all HGs (Fig 4; S4 Table). Such a predominance of HG 26 diols and HG 26 keto-ol has been observed before in Nodularia chucula and other representatives of this genus [27]. A unique characteristic of Nodularia spp. was the presence of two structural isomers of the HG 26 diol, of which the latter eluting isomer (HG 26 diol III; eluting around 10.1 min) was not detected in any of the other Baltic Sea cyanobacteria. In Nodularia spp., however, it was generally the most abundant of all HGs, ranging from 48.8 to 62.9% (55.6±5.8%). GC-MS analysis of the total lipid extract liberated after acid hydrolysis of cell material of Nodularia BY1 revealed the presence of large quantities of the same polyhydroxyalcohol observed in Aphanizomenon TR183. Structural isomers of 1,3,25-hexacosanetriol were, however, again not detected, indicating that both HG 26 diol isomers differed in the composition and/or configuration of their sugar moiety. Although the exact structure of HG 26 diol III remains to be determined, it may hold promise for tracing heterocystous cyanobacteria of the genus Nodularia in surface waters and sediments of the Baltic Sea, as it was absent from the other cyanobacteria investigated. In addition, pentose HG 26 diol was detected in all Nodularia strains in trace amount (0.3±0.1%). All representatives of the genus Nodularia also contained small quantities of 1-(O-deoxyhexose)-3,25-hexacosanediol (deoxyhexose HG 26 diol) that constituted up to 1.8% of all HGs in Nodularia BY1 but which was not detected in the other Baltic Sea cyanobacteria. At present, there exists little information on the distribution of this component in heterocystous cyanobacteria. It was for the first and only time reported from A. aphanizomenoides and C. raciborskii [43]. Given that deoxyhexose HG 26 diol thus occurs in at least three different genera, it may be more widespread among heterocystous cyanobacteria but further cultures studies are necessary to fully elucidate the chemotaxonomic potential of this component.

Bacteriohopanepolyols
Bacteriohopanepolyols are highly functionalized pentacyclic triterpenoids present in a wide range of prokaryotes [59]. Due to the large diversity of BHP structures in pure cultures of bacteria and, given their high taxonomic value, these components offer potential for fingerprinting bacterial communities and processes in modern and fossil ecosystems [81]. In cyanobacteria, the most frequently reported BHPs include bacteriohopanetetrol (BHT), bacteriohopanepentol (BHpentol) and composite BHPs such as BHT cyclitol ether [21]. BHPs with an additional methyl group at C-2 are considered most diagnostic for cyanobacteria [20], although only selected unicellular and filamentous cyanobacteria indeed synthesize 2Me-BHPs [21] and that C-2 methylation also occurs in other prokaryotes including methylotrophs [82], N 2 -fixing bacteria [83] and two strains of the purple non-sulfur bacterium Rhodopseudomonas palustris [84].
BHP distributions were largely similar in the Dolichospermum and Aphanizomenon strains (except for Dolichospermum BIR53), with BHT cyclitol ether (CE) being most dominant and accounting for 74.4 to 91.6% (84.6±9.0%) of all BHPs (Fig 5; S5 Table). This component has not been reported from heterocystous cyanobacteria of the genera Dolichospermum and Aphanizomenon before, but it is common in marine and freshwater cyanobacteria, including unicellular species such as Anacystis montana and Crocosphaera sp. as well as filamentous forms like Trichodesmium sp. and Chlorogloeopsis sp. [21]. BHT CE is also among the most dominant composite BHPs found in freshwater lakes [81] and is ubiquitously distributed in temperate to tropical soils [85,86]. Although cyanobacteria may constitute a biological source of BHT CE in these environments, it should be pointed out that this component is also synthesized by a large variety of bacterial species, including the facultative methanotroph Methylobacterium organophilum [87], the purple non-sulfur bacterium Rhodoblastus acidophilus [88], the acetic acid bacterium Frateuria aurantia [89] and some δ-proteobacteria belonging to the Geobacteraceae [90]. In the Baltic Sea, BHT CE (together with BHT and BHT glucosamine) has been reported from surface waters of the Gotland Deep collected during the formation of a cyanobacterial bloom dominated by Aphanizomenon sp. and-to a lesser extent-Anabaena sp. [91]. Although none of the heterocystous cyanobacteria investigated here seem to be capable of synthesizing BHT under the chosen culture conditions, the dominance of BHT CE and BHT glucosamine (see below) in Dolichospermum and Aphanizomenon is in agreement with the overall BHP distribution pattern observed during this bloom event.
Guanidine-substituted BHT CE comprised the second most abundant BHP in Dolichospermum 315 and BIR169 as well as in Aphanizomenon TR183, constituting 8.6±2.5% of all BHPs (Fig 5). Compared with BHT CE, it is less widespread in natural environments and has been reported in minor concentration only from Western Canadian soils [92] and a New Zealand geothermal vent system [93]. Guanidine-substituted BHT CE was initially described from some methylotrophic bacteria [87,94], the α-proteobacterium Zymomonas mobilis [95] and more recently the δ-proteobacterium Geobacter sulfurreducens [90]. Given the lack of direct evidence for biological sources other than facultative and obligate anaerobes, in combination with an ubiquitous distribution of Geobacter in aquatic and terrestrial anoxic sediments, it has been put forward that guanidine-substituted BHT CE may be more specific for tracing Geobacter spp. than other BHPs, in particular when accompanied by a high amount of BHT CE [90]. Our results demonstrate, however, that BHP signatures similar to those found in Geobacter may also derive from cyanobacterial sources, which is in agreement with the presence of guanidine-substituted BHT CE in an enrichment culture of the unicellular cyanobacterium Gloeocapsa sp. [21]. As cyanobacterial biomass may be readily exported from the photic zone and incorporated into sediment sequences, care should be taken when interpreting BHP profiles in stratified aquatic systems with seasonally or permanently anoxic bottom waters and sediments.
BHT glucosamine was present in minor abundance (6.8±7.9% of all BHPs) in all Dolichospermum and Aphanizomenon strains with the exception of Dolichospermum BIR53 (Fig 5). This component has been described from mostly the same groups of bacteria as those producing BHT CE, including methanotrophs [87], the thermoacidophilic bacterium Alicyclobacillus acidocaldarius [46], the α-proteobacterium Z. mobilis [95] and two representatives belonging to Geobacter [90] but importantly BHT glucosamine has not been report from cultured cyanobacteria so far. It is less common in natural environments but has been reported in minor abundance from surface sediments of tropical to subtropical freshwater lakes [81] and surface water samples collected from the Baltic Sea [91]. BHT glucosamine is particularly abundant in Dolichospermum BIR53, in which it accounted for ca. 97% of all BHPs, making this cyanobacterium an environmentally significant source of BHT glucosamine in the Baltic Sea. Dolichospermum BIR53 also contained a low abundance (ca. 3%) of guanidine-substituted BHTglucosamine. This component was reported from subsurface sediments of Lake Druzhby, Antarctica [21] but its biological source was unknown until present. Although the isolate of Dolichospermum BIR53 was not axenic and guanidine-substituted BHT-glucosamine may thus potentially also derive from bacterial contamination, its presence in a pure culture of Nostoc CCY9926 (Bauersachs, unpublished data) suggests that heterocystous cyanobacteria indeed constitute a biological source of guanidine-substituted BHT glucosamine in the Baltic Sea but also in freshwater environments.
The overall BHP distribution patterns in our Baltic Sea cyanobacteria differed markedly from those reported in the literature. BHPs commonly found to dominate the BHP distribution in pure cultures of cyanobacteria such as BHT, BHPpentol and BHP with additional C-2 methylation [21] were not detected in the heterocystous cyanobacteria investigated here. This may suggest that the restricted conditions of the Baltic, with its limited access to the open ocean and freshwater environments, may have favored the development of an endemic population of heterocystous cyanobacteria with characteristic BHP distribution patterns. Alternatively, the dramatic environmental conditions in the Baltic, with its strong salinity, temperature and nutrient gradients, may have resulted in a unique adaptation of the cyanobacterial cell membrane and the synthesis of BHPs that differ from those commonly found in freshwater and marine cyanobacteria.

Chemotaxonomy of bloom-forming heterocystous cyanobacteria
Applying CA on whole lipid profiles (Fig 6a) and individual lipid classes (Fig 6b-6d) allowed (i) a clear separation of cyanobacteria belonging to the genus Nodularia from the other two genera of Baltic Sea cyanobacteria examined here and (ii) the suggestion of a close phylogenetic relationship between cyanobacteria of the genera Dolichospermum and Aphanizomenon. The distinct position of the four Nodularia strains in all CA plots is primarily associated with the large contributions of MMAs (6Me-and 7Me-C 17:0 ), DMAs (6,12-and 7,11Me-C 17:0 ), the PUFAs 18:3ω6 and 18:4ω3 as well as the heterocyst glycolipids HG 26 diol III and deoxyhexose HG 26 diol. These components-with the exception of 7Me-C 17:0 occurring in minor abundance in Aphanizomenon TR183 -were not detected in the other heterocystous cyanobacteria investigated here and are thus considered most diagnostic for tracing planktonic cyanobacteria of the genus Nodularia in the Baltic Sea. Moreover, the close clustering of Nodularia AV1, BY1, F81 and HEM in the CA plots indicates little intraspecies variation in the lipid biomarker inventory between strains and argues for a close relationship of these heterocystous cyanobacteria. Our lipid biomarker results are thus in good agreement with molecular analyses, including whole genome [96] and 16S rRNA sequences [97], which have indicated a close overall relationship of Baltic Sea Nodularia strains and in fact support the finding that only one planktonic genotype of the genus Nodularia occurs in the Baltic Sea [80,98].
The significant contribution of HG 28 diol and HG 28 keto-ol as well as HG 28 triol and HG 28 keto-diol, together with the presence of n-C 17:0 , BHT CE and guanidine-substituted BHT CE in the whole lipid CA plot, is a unique characteristic of the genera Dolichospermum and Aphanizomenon (Fig 6a). Whole lipid profiles as well as hydrocarbon, FAME and BHP distribution patterns of non-toxic Dolichospermum BIR169 and hepatotoxic Dolichospermum 315 show high overall resemblances but significantly differ in their HG content due to the exceptional high loading of HG 28 triol and HG 28 keto-diol to axis 2 in the CA plot (Fig 6d). The lipid inventory of BIR53 is, for the most part, similar to those in the other two Dolichospermum strains. Its unusual BHP signature, with the dominance of BHT glucosamine and guanidinesubstituted BHT-glucosamine, however, sets it apart from other representatives of the genus Dolichospermum in the CA plots of both whole lipid profiles (Fig 6a) and BHP distribution patterns (Fig 6c). Although our data thus indicate that the distribution pattern of individual lipid classes may vary to some extent between the Dolichospermum strains, the generally close clustering of the three strains in the CA plots suggests a close overall relationship of representatives of this genus. Indeed, molecular analysis indicates that the planktonic Dolichospermum strains examined in this study share !97.8% 16S rRNA gene similarities and from a genetic perspective thus belong to the same species [32]. Similar to our results, however, the molecular data indicate a closer phylogenetic relationship between BIR169 and 315 and a more distant relationship of both strains from BIR53.
Aphanizomenon TR183 shows lipid signatures closely resembling those of the three Dolichospermum strains and it is generally associated with both the non-toxic Dolichospermum BIR169 and the hepatotoxic Dolichospermum 315 in the different CA plots (Fig 6a-6d). This similarity in lipid profiles, in particular of whole cell FAMEs (Fig 6b), BHPs (Fig 6c) and HGs (Fig 6d), suggests a close phylogenetic relationship between species of both genera, which is in contrast to the current separation of both genera based on morphological traits [99]. Previous phylogenetic studies have, however, shown that the genera Dolichospermum and Aphanizomenon are not monophyletic and occur intermixed in 16S rRNA trees [30,71], in agreement with the lipid-based data obtained here and FAME distribution pattern reported previously [67]. Yet, there are subtle differences in the lipid biomarker inventory of Aphanizomenon TR183 that set it apart from all the investigated Dolichospermum strains. First, it is characterized by the presence of a small quantity of MMAs (such as 7Me-C 17:0 and notably 8Me-C 17:0 ), which were both absent from Dolichospermum spp. (Fig 1). Second, HG 26 diol II was not observed in any of the Baltic Dolichospermum strains and thus is a unique characteristic of Aphanizomenon TR183 although it has been reported from benthic Anabaena CCY9613 isolated from a coastal microbial mat of the southern North Sea before [44]. Therefore, the high loading of 8Me-C 17:0 and HG 26 diol II onto axis 2 clearly distinguishes Aphanizomenon TR183 from the Baltic Sea Dolichospermum strains. The systematic differences in the lipid profiles of Aphanizomenon, Dolichospermum and Nodularia thus allow a separation of the major bloom-forming genera of heterocystous cyanobacteria and they may ultimately constitute a means of studying the community composition of cyano-HABs in the Baltic Sea and potentially other aquatic environments.

(Paleo)environmental considerations
Blooms of N 2 fixing heterocystous cyanobacteria have increased significantly in frequency and abundance in the Baltic Sea [4] as well as in many other brackish and freshwater environments worldwide [2]. Information on the spatiotemporal distribution and composition of cyano-HABs as well as their annual and interannual variation in modern and past ecosystems is often lacking but essential to improve our understanding on how bloom formation will proceed with changing environmental conditions (such as global warming, freshwater acidification or eutrophication) in the future. A factor that aggravates studying cyanoHABs in modern-day aquatic environments is their transitory nature and the resultant high frequency water column sampling required to adequately describe seasonal or annual variations in the intensity of cya-noHABs and their community composition. In this respect, the application of the here established lipid biomarker signatures to sediment traps, surface and subsurface sediments of the Baltic Sea (and possibly also other brackish and freshwater environments worldwide) will provide detailed information on present and past activity and diversity of bloom-forming heterocystous cyanobacteria. The time integrated nature of such records, averaging out the high seasonal and annual fluctuations of cyanoHABs, will allow obtaining a more comprehensive picture of cyanoHABs over time than can be achieved by the analysis of water column samples alone. We thus consider the here presented biomarker profiles with their high chemotaxonomic potential a valuable addition to the existing tools suited to investigate cyanoHABs in modern and moreover past ecosystems.
It has to be emphasized though that isolates stored in culture collections may adapt morphological and/or cellular traits in response to laboratory growth conditions and thereby show features departing from those found in nature, in particular when they are maintained in culture over multiple generations or over long time periods [71]. Growth conditions may also directly affect the lipid content and relative abundance of individual lipids and lipid classes as has been previously reported from a variety of algae and cyanobacteria [100,101]. In order to minimize biases between the lipid signature of cultured isolates and natural populations of bloom-forming cyanobacteria, growth parameters (such as temperature, salinity, light intensity, etc.) were carefully chosen to match those typically encountered during the development of cyanoHABs in the Baltic Sea. In addition to culture-dependent biases, diagenetic reactions may alter the originally synthesized organic matter and not all of the established lipid signatures for Aphanizomenon, Dolichospermum and Nodularia may unequivocally be applied to the geological archive in order to qualitatively assess past abundances and species composition of cyanoHABs. Particularly susceptible to alteration are polyunsaturated and monounsaturated fatty acids that are readily utilized by microorganisms and in sediments show a ten times faster degradation compared to their saturated counterparts [102,103]. Hence, fatty acid profiles such as those obtained in this study are well suited to examine bloom-forming heterocystous cyanobacteria in modern water column samples and surface sediment layers but major alterations of the originally-synthesized biological signal may be experienced over geological time scales. Information on the diagenetic stability of BHPs and HGs is currently only sparse. Both component classes have been reported from sediments of Holocene to at least Eocene age [78,104,105] and thus they show potential to preserve well in the sedimentary archive, in particular under strongly reducing conditions. Although additional studies on degradation rates and selective preservation of these components are needed, BHPs and HGs thus appear robust markers for the study of cyanoHABs in the Holocene Baltic Sea and likely other brackish and freshwater environments of comparable age. Normal and mid-chain branched alkanes are amongst the most refractory of organic matter and do not experience major alterations during diagenesis [106]. Given their exceptional high preservation potential with numerous reports from Middle Proterozoic to Phanerozoic sediments [107,108] and the high specifity of midchain branched alkanes demonstrated in this study, these components are considered best suited to examine for the presence and community composition of cyanoHABs in the modern and past Baltic Sea and possibly also other aquatic ecosystems.

Conclusions
Eight planktonic heterocystous cyanobacteria known to regularly from extensive cyanoHABs in the Baltic Sea were investigated for their lipid biomarker inventory. In brief, the hepatotoxic and non-toxic Dolichospermum strains were characterized by the presence of n-heptadecane and comparatively high abundances of myristic acid and BHT CE. The non-toxic Aphanizomenon strain showed lipid distributions very similar to those of Dolichospermum spp. but also contained HG 26 diol II, MMAs (including 7Me-C 17:0 and 8Me-C 17:0 ) in minor proportion and lacked a higher relative abundance of myristic acid. In contrast, all Nodularia strains were recognized as important sources of MMAs (6Me-C 17:0 and 7Me-C 17:0 ) and DMAs (6,12Me-C 17:0 and 7,11Me-C 17:0 ). They also contained HG 26 diol III, deoxyhexose HG 26 diol and the PUFAs 18:3ω6 and 18:4ω3, which were absent from the other heterocystous cyanobacteria investigated. These differences in lipid distributions support the current taxonomic separation of bloom-forming heterocystous cyanobacteria into three distinct genera but they also indicate a close phylogenetic relationship between species of the genera Dolichospermum and Aphanizomenon. Compared with the classical biomarker approach, in which a single component or a component class is employed to trace for a certain group of organism, the analysis of multiple lipid classes in water column samples, surface and subsurface sediments will allow a more detailed reconstruction of the community composition of cyanoHABs in the Baltic Sea and their variation over time. In addition, our study highlights the importance of dedicated culture studies to unequivocally link biomarker profiles found in the geosphere to their biological sources and thus improve our understanding on the spatiotemporal distribution of microbial communities (such as bloom-forming cyanobacteria) and their response to environmental changes in modern and fossil ecosystems.
Supporting information S1