Biomarkers in the stratified water column of the Landsort 1 Deep ( Baltic Sea ) 2 3

1 The water column of the Landsort Deep, central Baltic Sea, is stratified into an oxic, suboxic 2 and anoxic zone. This stratification controls the distributions of individual microbial 3 communities and biogeochemical processes. In summer 2011, particulate organic matter was 4 filtered from these zones using an in situ pump. Lipid biomarkers were extracted from the 5 filters to establish water column profiles of individual hydrocarbons, alcohols, phospholipid 6 fatty acids, and bacteriohopanepolyols (BHPs). As a reference, a cyanobacterial bloom 7 sampled in summer 2012 in the central Baltic Sea Gotland Deep was analyzed for BHPs. The 8 biomarker data from the surface layer of the oxic zone showed major inputs from 9 cyanobacteria, dinoflagellates and ciliates, while the underlying cold winter water layer was 10 characterized by a low diversity and abundance of organisms, with copepods as a major 11 group. The suboxic zone supported bacterivorous ciliates, type I aerobic methanotrophic 12 bacteria, sulfate reducing bacteria, and, most likely, methanogenic archaea. In the anoxic 13 zone, sulfate reducers and archaea were the dominating microorganisms as indicated by the 14 presence of distinctive branched fatty acids, archaeol and PMI derivatives, respectively. Our 15 study of in situ biomarkers in the Landsort Deep thus provided an integrated insight into the 16 distribution of relevant compounds and describes useful tracers to reconstruct stratified water 17 columns in the geological record. 18


Introduction
The Baltic Sea is a brackish marine marginal sea with a maximum depth of 459 m in the Landsort Deep (western central Baltic Sea; Matthäus and Schinke, 1999;Reissmann et al., 2009;Fig. 1).A positive freshwater budget and saltwater inflows from the North Sea through Skagerrak and Kattegat lead to a permanent halocline that stratifies the water column of the central Baltic Sea at about 60 m water depth (Reissmann et al., 2009).Major saltwater inflows, as detected in 1993 and 2003, sporadically disturb the stratification in the eastern central Baltic Sea and oxygenate the suboxic zone and deep water.These inflows, however, rarely reach the western central Baltic Sea.Even the strong inflow from 1993 had only minor effects on Landsort Deep, where stagnating conditions prevailed throughout (Bergström and Matthäus, 1996).Therefore, the Landsort Deep offers stable environments for microbial life within the oxic, suboxic, and anoxic zones and provides an excellent study site for the investigation of biomarker inventories that specify stratified water columns.
The Black Sea, although much larger in size, is comparable with the Landsort Deep with respect to the existence of a permanently anoxic deep-water body.Two comprehensive in situ biomarker reports gave a wide-ranging overview of various biomarkers and their producers in the Black Sea water column and identified a close coupling of microorganisms to biogeochemically defined water layers (Wakeham et al., 2007(Wakeham et al., , 2012)).Several other in situ biomarker water-column studies exist, but they were usually focused on certain aspects, for example anaerobic and aerobic methanotrophy Published by Copernicus Publications on behalf of the European Geosciences Union.(Schouten et al., 2001;Schubert et al., 2006;Blumenberg et al., 2007;Sáenz et al., 2011;Xie et al., 2014, and others).
For the Baltic Sea water column, biomarker knowledge is limited as most studies so far were focused on pollutionrelated compounds (e.g., Beliaeff and Burgeot, 2001;Lehtonen et al., 2006;Hanson et al., 2009).Recently, we reported the water-column distributions and 13 C-content of individual bacteriohopanepolyols (BHPs) and phospholipid fatty acids (PLFA) from the Gotland Deep, located about 150 km SE of the Landsort Deep in the eastern central Baltic Sea.These studies were aimed at microbial methane turnover and confirmed the importance of the Baltic Sea suboxic zone for bacterial methane oxidation (Schmale et al., 2012;Berndmeyer et al., 2013;Jakobs et al., 2014).The theoretical possibility of sulfate-dependent methane oxidation in the anoxic zone was also stated (Jakobs et al., 2014) but still remains to be proven for the central Baltic Sea water column.
Because the eastern central Baltic Sea is regularly disturbed by lateral intrusions in intermediate water depths (Jakobs et al., 2013), we chose the more stable Landsort Deep in the western central Baltic Sea as a sampling site for this biomarker study.Furthermore, published genetic studies reporting on prokaryotes and the related metabolisms in the water column of the Landsort Deep (Labrenz et al., 2007;Thureborn et al., 2013) provide a background to which the organic geochemical results can be advantageously related.The depth profiles of biomarkers from this setting not only reveal how actual biogeochemical processes are reflected by lipid abundances, distributions, and stable carbon isotope signatures, they also provide reference data for the reconstruction of past water columns using biomarkers from the sedimentary record.
A cyanobacterial bloom was sampled in summer 2012 on cruise M87/4 of R/V Meteor at the Gotland Deep (57 • 19.2 N, 20 • 03.0 E; Fig. 1), east of Gotland.Water samples of 10 L were taken at 1 m water depth and filtered with a 20 µm net.The samples were centrifuged and the residue freeze-dried.Samples were kept frozen at −20 • C until analysis.

Bulk CNS analysis
Three pieces (∅ 1.2 cm) from different zones of the filters were combusted together with V 2 O 5 in a EuroVector Eu-roEA Elemental Analyzer.Particulate matter in the Baltic Sea was reported to be free of carbonate (Schneider et al., 2002), and, thus, the filters were not acidified prior to analysis.C, N, and S contents were calculated by comparison with peak areas from standards.Standard deviations were ±2 % for C and ±5 % for N and S.
An aliquot of each filter extract and the bloom extract was acetylated using Ac 2 O and pyridine (1 : 1, v : v) for 1 h at 50 • C and then overnight at room temperature.The mixture was dried under vacuum and analyzed for BHPs using liquidchromatography-mass-spectrometry (LC-MS).
Another aliquot of each filter extract was separated into a hydrocarbon (F1), an alcohol and ketone (F2), and a polar fraction (F3) using column chromatography.The column (∅ ca. 1 cm) was filled with 7.5 g silica gel 60; samples were dried on ca.500 mg silica gel 60 and placed on the column.

Gas-chromatography-mass-spectrometry (GC-MS) and GC-combustion isotope ratio mass spectrometry (GC-C-IRMS)
GC-MS was performed using a Varian CP-3800 chromatograph equipped with a Phenomenex Zebron ZB-5MS fused silica column (30 m × 0.32 mm; film thickness 0.25 µm) coupled to a Varian 1200L mass spectrometer.Helium was used as a carrier gas.The temperature program started at 80 • C (3 min) and ramped up to 310 • C (held 25 min) with 4 • C min −1 .Compounds were assigned by comparing mass spectra and retention times to published data.Concentrations were determined by comparison with peak areas of squalane (F2 and F3) and n-eicosane-D42 (F1) as internal standards.
Compound-specific stable carbon isotope ratios of biomarkers in F2 and F3.3 were measured (twice) using a Thermo Trace gas chromatograph coupled to a Thermo Delta Plus isotope ratio mass spectrometer.The GC was operated under the same conditions and with the same column as for GC-MS.The combustion reactor contained CuO, Ni, and Pt and was operated at 940 • C. Isotopic compositions are reported in standard delta notation relative to the Vienna PeeDee Belemnite (V-PDB) and were calculated by comparison with an isotopically known CO 2 reference gas.GC-C-IRMS precision and linearity was checked daily using an ex-ternal n-alkane isotopic standard (provided by A. Schimmelmann, Indiana University).

Liquid-chromatography-mass-spectrometry (LC-MS)
LC-MS was performed using a Varian Prostar Dynamax high-performance liquid chromatography (  ibration).Amino BHPs had a 7 times higher response factor than nonamino BHPs, and concentrations in the samples were corrected accordingly.Comparisons with elution times of previously identified compounds further aided in BHP assignment.The quantification error is estimated to be ±20 %.

Principle component analysis (PCA)
PCA was based on the relative abundance of individual components in different water depths and was performed using R (version 3.0.2, 25 September 2013) with the "princomp" module (The R Foundation, 2014).

Physicochemical parameters of the water column
In summer 2011, the Landsort Deep showed a strong vertical stratification (Fig. 2).The oxic zone consisted of the uppermost 80 m and was divided by a strong thermocline into a warm surface layer (∼ 0-10 m) and a cold winter water layer (∼ 10-60 m).The halocline was located between 60 and 80 m.O 2 concentrations rapidly decreased from > 8 mL L −1 at ∼ 50 m to < 0.2 mL L −1 at ∼ 80 m, defining the upper boundary of the suboxic zone (Tyson and Pearson, 1991)  increased.The upper suboxic zone also showed a sharp peak in turbidity that is possibly caused by the precipitation of Fe and Mn oxides (Dellwig et al., 2010) or zero-valent sulfur (Kamyshny Jr. et al., 2013) and can be used as an indicator of the O 2 -H 2 S transition (Kamyshny Jr. et al., 2013).The anoxic zone extends from 90 m to the bottom and is characterized by the complete absence of O 2 and high concentrations of H 2 S and CH 4 .
CH 4 was highest in the deep anoxic zone and decreased strongly towards the suboxic zone but was still present in minor concentrations in the oxic zone.A small CH 4 peak was detected at the suboxic-anoxic interface (Fig. 2).Particulate organic carbon (POC) was highest at 10 m (380 µg L −1 ), decreased to a minimum in the cold winter water layer (48 µg L −1 ), and showed almost constant values of ∼ 70 µg L −1 in the suboxic and anoxic zones.
Generally, we follow the zonation of the Landsort Deep water column as given in Jakobs et al. (2014).We regarded the onset of H 2 S as the top of the anoxic zone, however, as this is better supported by our biomarker data (see below).

Lipid analysis
The PCA analysis separated six groups of biomarkers according to their distribution in the water column (Fig. 3,.Out of these groups, 18 compounds were selected as representative biomarkers, specifying inputs from individual prokaryotes and eukaryotes (with phototrophic, chemotrophic, and/or heterotrophic metabolisms).These biomarkers and their distributions are discussed in detail in Sect. 4.
The concentrations of these compounds are shown in Fig. 4, and compound-specific δ 13 C values are given in Table 1.Apart from the biomarker families revealed by PCA, two compound classesn-alkanes and n-alkenes in the sea surface layer -and individual BHPs obtained from the water

Group 1: surface maximum
The first group is defined by a strong maximum in the surface layer and only minor concentrations in greater depths.

Group 3: cold winter water layer maximum
The third group showed compounds that peaked in the cold winter water layer at 65 m water depth (Fig. 3).17   only occurred from 65 to 80 m, with a maximum at 65 m (287 and 228 ng L −1 , respectively).Of Group 3, the 16 : 0 − 18 : 1 wax ester was included in the discussion.δ 13 C values of the wax esters were ∼ −28 ‰ (Table 1).

n-Alkanes and n-alkenes in the sea surface layer
The concentrations of n-alkanes and n-alkenes in the surface sample (10 m water depth) are given in Fig. 5.The longest n-alkane chain was n-C 36 , and odd carbon numbers dominated over even.Highest concentrations were found for n-C 27 (21 ng L −1 ), n-C 29 (30 ng L −1 ), and n-C 31 (26 ng L −1 ).The longest n-alkene chain was n-C 26 : 1 , and highest nalkene concentrations were measured for n-C 23 : 1 (3 ng L −1 ) and n-C 25 : 1 (3 ng L −1 ).
For comparison, the major phytoplankton species from a cyanobacterial bloom in the Gotland Deep (2012) were determined by microscopy (HELCOM manual, 2012) and the particulate organic matter (POM) was analyzed for BHPs.This reference biomass contained mainly Aphanizonemon and, to a smaller extent, Anabaena and Nodularia, which were accompanied by dinoflagellates.Three BHPs were observed in the bloom POM (Fig. 6b).Among these compounds, the most abundant was BHT (∼ 86 %), followed by BHT cyclitol ether (∼ 10 %) and BHT glucosamine (∼ 4 %).

Discussion
In the following, we discuss several aspects of the biomarker profiles with respect to their significance as tracers for the relevant biota and biogeochemical processes in stratified water columns.

Water-column redox zones as reflected by cholestanol / cholesterol ratios
Different redox states of the Landsort Deep water column and the associated microbial processes are reflected by the profiles of cholesterol and its diagenetic product, cholestanol (Fig. 4; Groups 1 and 6, respectively).Cholesterol is synthesized by various eukaryotic phyto-and zooplankton and higher plants (Parrish et al., 2000) and is abundant in water columns and sediments.In sediments as well as in stratified water columns, stanols are produced from sterols by anaerobic bacterial hydrogenation (Gaskell and Eglinton, 1975;Wakeham, 1989) and by the abiotic reduction of double bonds by reduced inorganic species such as H 2 S (Hebting et al., 2006;Wakeham et al., 2007).Therefore, cholestanol / cholesterol ratios typically increase under more reducing conditions.In the Black Sea, low ratios of ∼ 0.1 were associated with oxygenated surface waters, and the suboxic zone showed ratios between 0.1 and 1, whereas the anoxic zone revealed values > 1 (Wakeham et al., 2007).
In the Landsort Deep, the cholestanol / cholesterol ratios showed a slight increase with depth from the surface towards the suboxic zone but always remained < 0.1 (Fig. 4).Below, the values increased to ∼ 0.3 in the suboxic zone and further to a maximum of 0.45 in the anoxic zone.Whereas the ratios in the Landsort Deep are considerably lower than in the Black Sea, the depth trend still clearly mirrors the changes from oxic to suboxic, and further to anoxic conditions.It is also interesting to note that total cholesterol and cholestanol concentrations in the Landsort Deep were ten-and fourfold higher, respectively, than in the Black Sea (Wakeham et al., 2007).

Phototrophic primary production
As expected, in situ biomarkers for phototrophic organisms were most abundant in the surface layer and are pooled in PCA Group 1. 20 : 4ω6 PLFA is a biomarker traditionally assigned to eukaryotic phytoplankton (Nanton and Castell, 1999;Lang et al., 2011) and organisms grazing thereon, such as protozoa (Findlay and Dobbs, 1993;Pinkart et al., 2002;Risse-Buhl et al., 2011).20 : 5ω3 PLFA is known to be a major compound in diatoms (Arao and Marada, 1994;Dunstan et al., 1994), and high concentrations of these PLFAs, as observed in the surface layer of the oxic zone, are in good agreement with such an autochthonous plankton-based source.7-Methylheptadecane is a characteristic marker for cyanobacteria (Shiea et al., 1990;Köster et al., 1999) most likely source are members of the subclass Nostocophyceae that were often reported to produce isomeric midchain branched alkanes, including 7-methylheptadecane (Shiea et al., 1990;Hajdu et al., 2007;Liu et al., 2013).Nostocophyceae are key members of the photoautotrophic community in the Baltic Sea.Particularly the filamentous genera Nodularia and Aphanizonemon (see Sect. 3.2.8)and the picocyanobacterium Synechococcus play a major role in blooms during summer time (Stal et al., 2003;Labrenz et al., 2007).
The importance of cyanobacteria in the surface layer of the Landsort Deep is further reflected by the presence of C 21 : 1 , C 23 : 1 , and C 25 : 1 n-alkenes (Fig. 5).These compounds have been reported from Anacystis (Gelpi et al., 1970) and Oscillatoria (Matsumoto et al., 1990).Oscillatoria vaucher is also known to occur in the Baltic Sea but is of only minor abundance (Kononen et al., 1996;Vahtera et al., 2007).Unlike the n-alkenes that only occurred in the surface layer, long-chain n-alkanes were present in the whole water column, with high abundances in the oxic zone.Long-chain n-alkanes with a strong predominance of the odd-numbered n-C 25 to n-C 36 homologues (Eglinton and Hamilton, 1967;Bi et al., 2005) and β-sitosterol (Volkman, 1986) are typical components of higher-plant lipids, thus indicating continental runoff and/or aeolian input of terrigenous organic matter into the Landsort Deep.n-C 27 , n-C 29 , and n-C 31 showed surface maxima (not shown), indicating similar sources as for β-sitosterol and a contribution of land plant leaf waxes.Other than β-sitosterol, most n-alkanes peaked between 65 and 70 m (n-C 25 for example; Fig. 4).Apart from the surface peaks, this is also true for n-C 27 , n-C 29 , and n-C 31 .A possible explanation is the accumulation of terrigenous higherplant particles accumulating at the pycnocline, where density differences were highest (MacIntyre et al., 1995)

Phototrophic vs. heterotrophic dinoflagellates, and ciliates
The distribution of dinoflagellates and, most likely, ciliates in the water column is reflected by two specific biomarkers: dinosterol and tetrahymanol (see Sect. 3.2.2,Fig. 4).Dinosterol is mainly produced by dinoflagellates (Boon et al., 1979), although it was also reported in minor abundance from a diatom (Navicula sp., Volkman et al., 1993).The dinosterol concentrations in the Landsort Deep showed a bimodal distribution.The strong peak in the surface layer of the oxic zone probably represents contributions from phototrophic dinoflagellates.Plausible candidates are Peridiniella catenata and Scrippsiella hangoei, both of which are involved in the spring phytoplankton blooms in the central Baltic Sea (Wasmund et al., 1998;Höglander et al., 2004).The latter species was previously reported to produce dinosterol (Leblond et al., 2007).However, P. catenata as well as S. hangoei are virtually absent below 50 m water depth (Höglander et al., 2004) and can thus not account for the second peak of dinosterol at the suboxic-anoxic transition zone.An accumulation of surface-derived dinosterol at the bottom of the suboxic zone is unlikely, as the pycnocline, and thus the strongest density discontinuity, is located at 60-70 m water depth, i.e., about 20 m further above.Dinosterol is absent in the pycnocline and only occurs from the bottom of the suboxic zone downwards and below.Instead, a likely source of dinosterol at this water depth is represented by heterotrophic dinoflagellates that are abundant in the suboxic zones of the central Baltic Sea (Anderson et al., 2012).Due to their enhanced productivity, these environments provide good conditions to sustain communities of eukaryotic grazers (Detmer et al., 1993).A possible candidate, Gymnodinium beii, was described from the suboxic zones of the central Baltic Sea (Stock et al., 2009).Indeed, several Gymnodinium species are known to be heterotrophs (Strom and Morello, 1998) and some have been reported to produce dinosterol (Mansour et al., 1999).Like cholesterol and β-sitosterol, dinosterol was also found in the anoxic zone at 400 m water depth.The production of these compounds at this depth is unlikely, as the synthesis of sterols requires oxygen (Summons et al., 2006).Hence, the observed sterol occurrences probably reflect transport through the water column.
A similar concentration distribution as for dinosterol was observed for tetrahymanol.Tetrahymanol is known to be produced by ferns, fungi, and bacteria, such as the purple nonsulfur bacterium Rhodopseudomonas palustris (Zander et al., 1969;Kemp et al., 1984;Kleemann et al., 1990;Sinninghe Damsté et al., 1995;Eickhoff et al., 2013).Moreover, ciliates ubiquitously produce tetrahymanol as a substitute for cholesterol when grazing on prokaryotes instead of eukaryotes, such as algae (Conner et al., 1968;Boschker and Middelburg, 2002).This is also a feasible scenario for the Baltic Sea, where the ciliate genera Metopus, Strombidium, Metacystis, Mesodinium, and Coleps are abundant in the suboxic zone and at the suboxic-anoxic interface (Detmer et al., 1993;Anderson et al., 2012).Unidentified ciliates also occurred in the anoxic waters of the Landsort Deep (Anderson et al., 2012).Members of the genus Rhodopseudomonas, a possible alternative source of tetrahymanol, have so far not been identified in the suboxic zone (Labrenz et al., 2007;Thureborn et al., 2013).We therefore regard bacterivorous ciliates living under suboxic to anoxic conditions as the most likely source of tetrahymanol in the suboxic zone and below.Likewise, ciliates feeding on chemoautotrophic bacteria were assumed as producers of tetrahymanol in the suboxic zone of the Black Sea (Wakeham et al., 2007).The situation is somewhat different in the surface waters, where tetrahymanol shows its maximum concentrations at 10 m water depth.Although Rhodopseudomonas and other purple nonsulfur bacteria usually occur under oxygen-deficient conditions, they have been genetically identified in the surface water of the Landsort Deep (Farnelid et al., 2009) and thus have to be considered as potential producers of tetrahymanol.Furthermore, cholesterol is abundant in the surface waters and could be incorporated by ciliates instead of tetrahymanol.However, some ciliates seem to prefer prokaryotes as a prey.Sinking agglomerates of cyanobacteria and other bacteria are known to be covered by feeding ciliates (Gast and Gocke, 1988).Hence, in addition to R. palustris, ciliates grazing selectively on cyanobacteria would plausibly explain the abundance of tetrahymanol in the shallow waters of the Landsort Deep.
δ 13 C values of tetrahymanol revealed an opposite trend to dinosterol.While dinosterol became isotopically more negative with depth (−29.9 to −32.0 ‰), tetrahymanol became more positive (−28.7 to −25.9 ‰) and showed its highest δ 13 C values in the anoxic zone.Although ciliates and dinoflagellates are both grazers at the suboxic-anoxic interface, they seem to occupy different ecological niches and feed on different bacterial sources.

Heterotrophs in the cold winter water layer
The only biomarkers with enhanced concentrations in the deep cold winter water layer are wax esters (e.g., 16 : 0−18 : 1 wax ester, Fig. 4) and, to a minor extent, cholesterol and 20 : 5ω3 PLFA.As the pycnocline, and thus a strong density discontinuity, is also located at this depth, an accumulation of settling organic debris containing these compounds has to be considered (MacIntyre et al., 1995).Living organisms, however, may be also be plausible sources.Copepods are known producers of wax esters and cholesterol (Lee et al., 1971;Sargent et al., 1977;Kattner and Krause, 1989;Nanton and Castell, 1999;Falk-Petersen et al., 2002), which are often abundant in density layers, where they feed on accumulated aggregates (MacIntyre et al., 1995).These organisms synthesize wax esters with total chain lengths of between 28 and 44 carbon atoms (Lee et al., 1971;Kattner and Krause, 1989;Falk-Petersen et al., 2002); several of wax esters were present in the Landsort Deep (data not shown in Fig. 4), with roughly the same distribution as the most prominent 16 : 0 − 18 : 1.Although copepods migrate through the water column, particularly those rich in wax esters prefer deep water or nearsurface cold water (Sargent et al., 1977), which is in full agreement with the high amounts of these compounds in the cold winter water layer.Copepods are abundant and diverse in the Baltic Sea, with major species being Pseudocalanus elongatus, Temora longicornis, and Acartia spp.(Möllmann et al., 2000;Möllmann and Köster, 2002).Like the wax esters, the 20 : 5ω3 PLFA shows higher concentrations in the cold winter water layer, but it is also abundant in the surface and at the suboxic-anoxic interface (Fig. 4).Copepods are also known to feed on diatoms and incorporate their specific fatty acids, such as 20 : 5ω3 PLFA, largely unchanged into their own tissue (Kattner and Krause, 1989).Dinoflagellates are also known producers of 20 : 5ω3 PLFA (Parrish et al., 1994;Volkman et al., 1998) and may be an alternative source in the surface layer and at the suboxic-anoxic interface; this is supported by a good correlation with dinosterol at these depths.
Unlike the abovementioned compounds, all other selected biomarkers show particularly low concentrations in the cold winter water layer.This is also true for widespread compounds such as the 16 : 1ω7c PLFA, which is produced by eukaryotes (Pugh, 1971;Shamsudin, 1992) as well as prokaryotes (Parkes and Taylor, 1983;Vestal and White, 1989).While a mixed origin of 16 : 1ω7c PLFA has to be assumed for the oxic zone, a bacterial source is more probable in the suboxic zone and in the anoxic zone.Regardless of the biological source, a very low amount of this ubiquitous fatty acid (Fig. 4) indicates that the cold winter water layer of the Landsort Deep does not support abundant planktonic life.Based on microscopy, similar observations have been made for the cold winter water layers of the Gotland, Bornholm, and Gdansk basins (Gast and Gocke (1988) and citations therein).

BHPs as indicators for aerobic and anaerobic metabolisms
Bacteria are the only known source of BHPs (Kannenberg and Poralla, 1999).Although the biosynthesis of BHPs and their precursor, diploptene (both part of Group 2), does not require oxygen, the production of hopanoids was long assumed to be restricted to aerobic bacteria, as reports from facultatively or strictly anaerobic bacteria were initially lacking.More recently, however, planctomycetes (Sinninghe Damsté et al., 2004), metal-reducing Geobacter (Fischer et al., 2005), and sulfate-reducing Desulfovibrio (Blumenberg et al., 2006(Blumenberg et al., , 2009(Blumenberg et al., , 2012) ) were identified as anaerobic producers of BHPs.In the Landsort Deep, cyanobacteria are abundant in the surface water layer and may be considered as a major source of BHPs (cf.Talbot et al., 2008;Welander et al., 2010).Evidence for such cyanobacterial BHP contributions may come from our analysis of a Gotland Deep bloom from summer 2012 (see Sect. 3.2.7).BHPs identified in this bloom were BHT, BHT cyclitol ether, and BHT glucosamine (Fig. 6b), which is in line with the BHP composition of the Landsort Deep surface layer (Fig. 6a).These three cyanobacterial BHPs were present throughout the Landsort Deep water column, although they were present in minor amounts in the suboxic zone and below.In addition, the surface layer contained aminotriol that was also present in the whole water column.Aminotriol is an abundant BHP produced by various bacteria (e.g., Talbot and Farrimond (2007) and references therein), indicating that organisms other than cyanobacteria may contribute BHP to the surface layer.
A further notable feature is the occurrence of BHT II at 70 m and below.The source of BHT II is not fully resolved yet.It was recently related to planctomycetes, especially those performing anaerobic ammonium oxidation (anammox) in sediments (Rush et al., 2014).Anammox bacteria can also be traced by 10-me16 : 0 PLFA and ladderane PLFAs (not studied here; Sinninghe Damsté et al., 2005;Schubert et al., 2006).10-me16 : 0 PLFA does indeed show a peak in the lower suboxic zone, where BHT II is abundant.However, 10-me16 : 0 PLFA may also be contributed by sulfate-reducing bacteria (see Sect. 4.6), and no evidence for anammox has been observed in the water column of the Landsort Deep from molecular biological studies so far (Hietanen et al., 2012;Thureborn et al., 2013).Regardless of the biological source, BHT II was described in stratified water columns of the Arabian Sea, Peru Margin and Cariaco Basin (Sáenz et al., 2011), and the Gotland Deep (Berndmeyer et al., 2013) and has therefore been proposed as a proxy for stratified water columns.This hypothesis has been adopted to reconstruct the development of water-column stratification in the Baltic Sea during the Holocene (Blumenberg et al., 2013).
Like BHT II, aminotetrol and aminopentol are absent from the surface layer (Fig. 6a).Whereas both BHPs are biomarkers for methanotrophic bacteria, the latter typically occurs in type I methanotrophs (Talbot et al., 2001).The presence of type I methanotrophic bacteria is further supported by the co-occurrence of the specific 16 : 1ω8c PLFA (Nichols et al., 1985;Bowman et al., 1991Bowman et al., , 1993) ) and its considerably depleted δ 13 C value (−45.4 ‰).
Whereas a major in situ production of BHPs in the suboxic zone is evident from our data, the sources of BHPs in the anoxic zone are more difficult to establish.BHPs in the anoxic zone may partly derive from sinking POM as well as being newly produced by anaerobic bacteria.Sinking POM as a source may apply to BHT cyclitol ether and BHT glucosamine, which seem to derive from cyanobacteria thriving in the oxic zone, as discussed above.Aminotriol, aminotetrol, and aminopentol, however, are known products of sulfate-reducing bacteria (Blumenberg et al., 2006(Blumenberg et al., , 2009(Blumenberg et al., , 2012) ) and may have their origin within the anoxic zone.This interpretation is supported by the close correlation of the total BHPs with the ai-15 : 0 PLFA, which is considered as indicative of sulfate reducers (see Sect. 4.6; both compounds were part of the same PCA Group 2).Thus, the anoxic zone of the Landsort Deep is likely an active source for BHPs rather than solely being a pool for transiting compounds.
In addition to the bacterial FA, two archaeal in situ biomarkers, archaeol and PMI, were identified.Archaeol is the most common ether lipid in archaea but is especially abundant in euryarchaeotes, including methanogens (Tornabene and Langworthy, 1979;Koga et al., 1993).Likewise, PMI and its unsaturated derivatives are diagnostic for methanogenic euryarchaeotes (Tornabene et al., 1979;De Rosa and Gambacorta, 1988;Schouten et al., 1997).In the Landsort Deep, both compounds are virtually absent in the oxic zone and increase in abundance with depth through the suboxic zone (Fig. 3).The same trend has been described for PMI in the Black Sea (Wakeham et al., 2007), and the presence of Euryarchaeota in Landsort Deep anoxic waters has recently been proven by Thureborn et al. (2013).
Given the available sample resolution, it is impossible to further elucidate the exact distribution of archaea in the anoxic zone of the Landsort Deep.Likewise, δ 13 C values could not be obtained for archaeol and PMI due to low compound concentrations, which makes statements on inputs of these lipids from archaea involved in the sulfate-dependent anaerobic oxidation of methane (AOM) impossible (cf.Hinrichs et al., 1999;Thiel et al., 2001).Whereas it has been shown that AOM is theoretically possible in the anoxic zone of the Landsort Deep and anaerobic methane consumption has recently been demonstrated to occur (Jakobs et al., 2013), clear evidence for abundant AOM is as yet lacking and requires further investigations focused on the anoxic water bodies of the Baltic Sea.

Conclusions
The Landsort Deep in the western central Baltic Sea is characterized by a stratified water column.Marine microbial organisms have adapted to the vertical chemical limitations of their ecosystems, and their distributions in the water column can be reconstructed using diverse in situ biomarkers.According to their behavior in the water column, PCA analysis revealed six groups of biomarkers for distinct groups of (micro)organisms and the related biogeochemical processes.Within the oxic zone, a clear preference for the surface layer became obvious for distinctive biomarkers.Among these compounds, 7-methylheptadecane, different alkenes, BHT cyclitol ether, and BHT glucosamine were indicative of the presence of bacterial primary producers, namely cyanobacteria.Dinosterol concentrations and −δ 13 C values revealed a phototrophic dinoflagellate population in the surface waters and a second, heterotrophic community thriving at the suboxic-anoxic interface.Similarly, abundant tetrahymanol at the surface indicated ciliates feeding on cyanobacterial agglomerates, but a second maximum at the suboxic-anoxic interface suggested a further ciliate population that grazed on chemoautotrophic bacteria.The cold winter water layer at the bottom of the oxic zone showed only low concentrations of biomarkers and seemed to be avoided by most organisms, except copepods.In contrast, biomarkers obtained from the suboxic zone reflected a high abundance and diversity of eukaryotes and prokaryotes.Whereas 16 : 1ω8 PLFA and aminopentol revealed the presence of type I aerobic methaneoxidizing bacteria, ai-15 : 0 PLFA, 10-me-16 : 0, and total BHPs indicated the distribution of sulfate-reducing bacteria in the Landsort Deep water column.The close coupling of ai-15 : 0 PLFA with total BHPs suggests that these bacteria represent a major in situ source for hopanoids in the anoxic zone.The anoxic zone was also inhabited by most likely Euryarchaeota, as shown by the presence of archaeol and PMI and its derivatives.Our study in the water column of the Landsort Deep gives insights into the recent distributions and actual sources of organic matter as reflected by lipid biomarkers.The results may also aid the interpretation of organic matter preserved in the sedimentary record and thus help to better constrain changes in the geological history of the Baltic Sea.

Figure 1 .
Figure 1.Map showing the sampling locations in the central Baltic Sea.

Figure 2 .
Figure 2. Physicochemical characteristics of the Landsort Deep water column in summer 2011.The suboxic zone is shaded light grey.Temperature and methane data were partially taken from Jakobs et al. (2014).

Figure 3 .
Figure 3. PCA of the relative abundances of compounds in different water depths.Group 1: surface maximum; a subgroup of compounds exclusively occurring at the surface are listed in the box.Group 2: surface and lower suboxic zone maxima.Group 3: cold winter water layer maximum.Group 4: oxic-zone high concentrations.Group 5: suboxic zone maximum.Group 6: absent in oxic zone, bottom layer maximum.Compound names are given in Table 2.

Figure 4 .
Figure 4. Vertical distribution of biomarkers in the Landsort Deep water column.The suboxic zone is shaded grey.
n-C 21 to n-C 36 as well as 26 : 0 PLFA (43).26 : 0 PLFA only occurred at 80 m, whereas all other compounds were abundant from the surface to the upper suboxic zone at 80 m (data not shown).The homologues n-C 27 (74), n-C 29 (76), and n-C 31 (78) show maxima at the surface (21-30 ng L −1 ).For the other compounds, maxima were either located at 65 or 70 m, with highest concentrations for n-C 25 − n-C 36 (10-23 ng L −1 ).Below 80 m, concentrations dropped to constantly low values.As an example, the depth profile of n-C 25 (71) is shown in Fig.4.δ 13 C values for these compounds could not be obtained.

Figure 5 .
Figure 5. Concentrations of n-alkanes and n-alkenes in the Landsort Deep surface layer (oxic zone, 10 m water depth).

Figure 6 .
Figure 6.Relative abundances of individual BHPs (as percent of the total) of (a) the Landsort Deep water column and (b) the Gotland Deep cyanobacterial bloom.Note that [%] axes start at 85 %.An asterisk indicates data taken from Jakobs et al. (2014).

Table 1 .
δ 13 C values of the compounds chosen from the PCA groups.No δ 13 C values were available for Group 4. N.d.stands for "not detectable".

Table 2 .
Compounds sorted by number as shown in Fig.3.Compounds chosen for further discussion are marked bold.