Comparative studies of pelagic microbial methane oxidation within the redox zones of the Gotland Deep and Landsort Deep ( central Baltic Sea )

Pelagic methane oxidation was investigated in dependence on differing hydrographic conditions within the redox zone of the Gotland Deep (GD) and Landsort Deep (LD), central Baltic Sea. The redox zone of both deeps, which indicates the transition between oxic and anoxic conditions, was characterized by a pronounced methane concentration gradient between the deep water (GD: 1233 nM, 223 m; LD: 2935 nM, 422 m) and the surface water (GD and LD < 10 nM). This gradient together with a 13C CH4 enrichment (δ13C CH4 deep water: GD−84 ‰, LD −71 ‰; redox zone: GD−60 ‰, LD −20 ‰; surface water: GD−47 ‰, LD −50 ‰; δ13C CH4 vs. Vienna Pee Dee Belemnite standard), clearly indicating microbial methane consumption within the redox zone. Expression analysis of the methane monooxygenase identified one active type I methanotrophic bacterium in both redox zones. In contrast, the turnover of methane within the redox zones showed strong differences between the two basins (GD: max. 0.12 nM d −1, LD: max. 0.61 nM d−1), with a nearly four-times-lower turnover time of methane in the LD (GD: 455 d, LD: 127 d). Vertical mixing rates for both deeps were calculated on the base of the methane concentration profile and the consumption of methane in the redox zone (GD: 2.5 × 10−6 m2 s−1, LD: 1.6× 10−5 m2 s−1). Our study identified vertical transport of methane from the deep-water body towards the redox zone as well as differing hydrographic conditions (lateral intrusions and vertical mixing) within the redox zone of these deeps as major factors that determine the pelagic methane oxidation.


Introduction
Methane is an atmospheric trace gas that influences the climate and the atmospheric chemistry (Wuebbles and Hayhoe, 2002).It is generated in terrestrial, limnic and marine ecosystems by biotic (Segers, 1998;Reeburgh, 2007) and abiotic mechanisms (Berndt et al., 1996;Keir et al., 2008).Although methane is trapped in aquatic sediments in vast amounts, marine environments only represent a negligible source for atmospheric methane (Reeburgh, 2007).In the marine system, microbial methane oxidation provides a control mechanism on the release of methane into the atmosphere (Bange et al., 1994;Valentine et al., 2001;Reeburgh, 2007;Schmale et al., 2012).Environmental studies regarding microbially mediated methane oxidation have been carried out for various aquatic systems.From marine water column studies it is known that methane is oxidized in the aerobic environment by methanotrophic bacteria (MOB) of type I, II and X, whereas in the anoxic waters this process is mediated by methanotrophic archaea (ANME) I and II (Durisch-Kaiser et al., 2005;Schubert et al., 2006b;Blumenberg et al., 2007).Different electron acceptors are used for the consumption of methane (Modin et al., 2007;Knittel and Boetius, 2009).In the sediments, oxidation processes have been proven by oxygen, sulfate, nitrate, nitrite, iron and manganese, and in the water column only by sulfate and oxygen (Reeburgh, 2007;Beal et al., 2009;Ettwig et al., 2010).According to their carbon assimilation, which can either be performed by the ribulose monophosphate (RuMP) or serine pathway, the aerobic MOB are separated into the main groups type I and type II, respectively.The thermophilic MOB type X represent a subgroup of type I and are capable of fixing methane as well Published by Copernicus Publications on behalf of the European Geosciences Union.

G. Jakobs et al.: Comparative studies of pelagic microbial methane oxidation
as carbon dioxide.Aerobic methanotrophs are able to oxidize methane via oxygen due to the key enzyme methane monooxygenase (MMO), which catalyzes the formation of methanol during the first step of methane oxidation.Only the particulate form of this enzyme (pMMO) is present in all three groups.The functional pmoA gene encoding the alpha subunit of pMMO is used as a marker to identify methanotrophs in different environments (Bourne et al., 2001;Mc-Donald et al., 2008).The anaerobic oxidation of methane (AOM) via sulfate takes place by a consortium of ANME (I, II and III) and sulfate-reducing bacteria with a very low energy yield of G 0 = −16.6 kJ mol −1 CH 4 compared to the energy yield of G 0 = −773 kJ mol −1 CH 4 obtained by the aerobic oxidation of methane (Boetius et al., 2000;Valentine and Reeburgh, 2000).Methanotrophic archaea harbor the methyl coenzyme M reductase (MCR), which is also present in methanogenic archaea to catalyze the formation of methane (Hallam et al., 2003).
The bottom topography of the Baltic Sea consists of a series of subbasins that are separated by narrow sills, which hamper the continuous water exchange with the North Sea.The hydrographic situation of the Baltic Sea is characterized by a freshwater discharge from the rivers and episodic inflows of saline water from the North Sea (estuarine circulation) (Lass and Matthäus, 2008).Ventilation of the deep water by saline oxygenated water from the North Sea (socalled major Baltic inflows) has occurred irregularly over the last decades, leading to long periods of deep-water stagnation in the central Baltic Sea (Matthäus et al., 2008;Reissmann et al., 2009); the last major Baltic inflow took place in year 2003.The more frequent but less intense inflows contain water masses with a density too low to displace the old stagnant water in the deep basins and propagate laterally into intermediate water layers of the central Baltic Sea (Matthäus et al., 2008).
In the central and southern Baltic Sea, the less saline surface-and more saline deep water cause the formation of a permanent density boundary, the halocline, which has a crucial effect on the vertical exchange of matter and the renewal of deep water (Reissmann et al., 2009).Especially the downward diffusion of oxygen is affected by this density boundary, leading to a vertical biogeochemical zonation with oxygen-limiting conditions in the intermediate-and deep-water body and the microbial turnover of organic matter by, for example, sulfate reduction or denitrification (Lass and Matthäus, 2008).The oxygenated surface water and the anoxic deep water are separated by the redox zone.This zone is a smooth transition between oxic and anoxic conditions with an overlap of oxygen-and hydrogen sulfide-containing waters (Nausch et al., 2008;Labrenz et al., 2010;Dellwig et al., 2012).The pronounced chemical gradients in redox zones of stratified water columns are known to provide favorable conditions for different microbial processes such as ammonia oxidation, denitrification or methane oxidation (Bret-tar and Rheinheimer, 1992;Schubert et al., 2006a;Labrenz et al., 2010).
In the Gotland basins, comprehensive water column investigations revealed a widespread release of methane from the sediments with strong methane enrichments in the stagnant anoxic water bodies (eastern Gotland Basin (Gotland Deep) and western Gotland Basin (Landsort Deep); max.504 nM and 1086 nM, respectively; Schmale et al., 2010).Compared to the atmospheric equilibrium, only slightly elevated methane concentrations (3-5 nM CH 4 ) prevail in the surface waters (Bange et al., 1994;Schmale et al., 2010;Gülzow et al., 2012).High-resolution gas chemistry studies in the water column of the Gotland basins showed a pronounced methane gradient and an enrichment of 13 C CH 4 within the pelagic redox zone that indicates microbial activity related to the aerobic oxidation of methane in that water depth (Schmale et al., 2012).This finding is supported by earlier methane oxidation rate measurements, indicating that methane is preferentially oxidized under low oxygen concentrations (16 to 63 µM O 2 ; Dzyuban et al., 1999).Schmale et al. (2012) also showed that aerobic methane oxidation within the pelagic redox zone of the Gotland Deep is performed by members of type I methanotrophic bacteria.Furthermore, biomarker analysis could identify specific bacteriohopanepolyols (BHP) and phospholipid fatty acids (PLFA) that are characteristic for the presence of active type I methanotrophic bacteria (Berndmeyer et al., 2013).
Although microbial oxidation of methane in the water column represents an important sink of methane, the impact of dynamic environmental conditions like methane source strength and hydrodynamic forces on the efficiency of that mechanism is still insufficiently described (Reeburgh, 2007;Schmale et al., 2012).With its distinct oxic and anoxic stratified zones that are episodically perturbed by oceanographic events, the Baltic Sea provides an ideal field to investigate these effects on the methane cycle.In this study we focus on the influence of different hydrographic conditions (lateral intrusions and vertical mixing) on the methane turnover within the pelagic redox zone.For this scope, comparative interdisciplinary studies were carried out in the Gotland Deep and Landsort Deep.The combined data on methane chemistry, methane oxidation rates and molecular biological analysis will gain first insights into the temporal stability and regional transferability of microbial processes related to pelagic microbial methane consumption in the central Baltic Sea.

Study area
Investigations were carried out in the water column of the Gotland Deep (57 • 19.2 N, 20 • 03.0 E; water depth 223 m) and Landsort Deep (58 • 35.0 N, 18 • 14.0 E; water depth 422 m) located in the central Baltic Sea (Fig. 1).Both sampling sites are characterized by different basin structures and hydrographic conditions (see cross section through the central Baltic Sea, Fig. 1c).The Landsort Deep represents the deepest areal in the western Gotland Basin (max.depth 460 m) with a relatively small spatial dimension.In contrast, the eastern Gotland Basin represents the largest basin of the Baltic Sea, with a maximum water depth of about 250 m at the Gotland Deep.
During an inflow of saline waters from the North Sea towards the central Baltic Sea, this water mass propagates first into the eastern Gotland Basin.Depending on the strength of the inflow, the deep-water mass can continue its path via a northern passage towards the western Gotland Basin.The travel of inflowing saline water from the North Sea along different basins and sills promotes the mixing between saline bottom water and less saline overlying water masses, resulting in a decreasing salt content of the intruding water along its way into the central Baltic Sea.Accordingly, the Landsort Deep is characterized by less frequent and weaker lateral intrusions, resulting in a more stable oxic-anoxic transition zone (redox zone) and a lower deep-water salinity compared to the Gotland Deep (Matthäus et al., 2008).

Sampling strategy
Samples were taken during cruise 06EZ/12/13 on R/V Elisabeth Mann Borgese in July/August 2012.Sampling for molecular biological analysis and radiotracer experiments were performed after first inspection of the physical parameters of the water column (e.g., salinity, turbidity) using a CTD probe (CTD system Sea-Bird sbe911+ and a turbidity sensor ECO FLNTU, WET Labs) and methane concentration analysis onboard to identify the most relevant depths intervals for methane oxidation.To obtain enough sample volume for all analysis, water samples were taken within five casts on two consecutive days at each sampling site.Methane, oxygen and hydrogen sulfide analyses were performed on the first day of sampling.Water samples for determination of methane oxidation rates and molecular biological analysis were obtained on the following day.Between these two samplings the physical parameters of the water column showed no significant differences.

Methane, oxygen and hydrogen sulfide
For the determination of methane concentrations, water samples were transferred from the rosette water sampler into 250 mL glass flasks and poisoned with 500 µL of saturated mercury chloride solution.Methane concentrations were determined directly after the cruise using a modified purge and trap procedure (Michaelis et al., 1990;Thomas, 2011).This modified method includes the following process steps: the samples were stripped with helium to remove volatile constituents (purge step), dried using a Nafion ® trap (Perma Pure LLC., USA), cryofocussed at −120 • C (ethanol / nitrogen) on HayeSep D ® (Valco Instruments Company Inc., Switzerland) desorbed by heating at 85 • C and analyzed using a gas chromatograph (Shimadzu GC-2014) equipped with a flame ionization detector.The concentration of oxygen was determined by titration method (precision ± 0.9 µM O 2 ) and the hydrogen sulfide concentrations were measured calorimetrically using the methylene blue method (precision ± 1.0 µM H 2 S) (Grasshoff et al., 1999).Oxygen concentrations were only determined for samples virtually devoid of hydrogen sulfide.

Stable carbon isotope ratio of methane
Gas samples for stable carbon isotope analysis of methane (δ 13 C CH 4 ) were gained using a vacuum degassing system (described in Keir et al. (2009) in detail).Following this method, a water sample of 600 mL was transferred into a 1100 mL pre-evacuated glass bottle, leading to a partial degassing of the water sample.The extracted gas was decompressed to atmospheric pressure and subsequently transferred to a gastight burette.An aliquot of the extracted gas was conserved in a 10 mL pre-evacuated glass crimp vial containing a saturated sodium chloride solution poisoned with mercury chloride.The isotope ratio was determined using an isotoperatio mass spectrometer (MAT 253, Thermo Electron, Germany) according to the method described by Schmale et

Sampling and 14 CH 4 tracer preparation
Immediately before sampling, the sample bottles were flushed with argon.According to the sampling procedure by Reeburgh et al. (1991), water samples were directly transferred from the rosette sampler into transfusion bottles and sealed air-free with aluminum screw caps and natural rubber septa.Three times the volume of the sample bottle (600 mL) was filled into the bottle in overflow to avoid any entry of oxygen into the water samples.In contrast to methods described by Reeburgh et al. (1991) and Durisch-Kaiser et al. ( 2005), a larger sample volume (600 mL) was chosen in order to account for the low specific 14 CH 4 tracer activity ( 14 CH 4 dissolved in sterile anoxic water, injection volume 100 µL, activity 3 kBq) used in this procedure and to obtain sufficient labeled oxidation products even in case of a low methane turnover.The 14 CH 4 activity of the injected tracer was determined by stripping of the tracer liquid with synthetic air and a following combustion step on a Cu(II) oxide catalyst at 850 • C (Treude et al., 2003).The resulting 14 CO 2 was trapped with a mixture of phenylethylamine and ethylene glycol monomethyl ether (1 : 7 v/v), and the activity was measured by liquid scintillation counting.The final concentration of the injected tracer (each sample labeled with 5 nM 14 CH 4 ) was lower than the in situ methane concentrations of the water samples to prevent any artificial stimulation of microbial methane oxidation by increased methane concentrations.

Sample processing
After tracer injection, the bottles were incubated in the dark for 3 days near in situ temperature (∼ 4 • C).The microbial activity was terminated by injection of sodium hydroxide (500 µL, 50 % (w/w)), which led to the precipitation of dissolved inorganic carbon compounds (CO 2(aq) , HCO − 3 , CO 2− 3 ).For trapping of the microbial-formed 14 CO 2 a method described by Treude et al. (2005) was used.Briefly, the samples were transferred into gastight bottles equipped with scintillation vials positioned in the gas headspace below the bottle cap.These vials contained a mixture of sodium hydroxide (2.5 % w/w) and phenylethylamine in equal amounts to capture the 14 CO 2 released from the sample after acidification with hydrochloric acid (5 mL, 25 %) under stirring for 24 hours.The activity of the scintillation vials was measured on a Tri-Carb 2910 TR (Perkin Elmer Inc., USA) liquid scintillation counter using a scintillation cocktail (Irga-safe Plus, Perkin Elmer).For selected water depths, incubation experiments were performed in triplicates to determine the standard deviation (s in percentage of the average, n = 3) of the measured oxidation rates for the upper oxic-, deep anoxic-and oxic-anoxic transition zone in intermediate water depth (80-145 m).For the Gotland Deep, the standard deviation for the oxic zone is 12.4 %, the oxicanoxic transition zone 8.8 % and the anoxic zone 134.4 % derived from triplicates sampled in 70, 85 and 175 m water depth, respectively.For the Landsort Deep, the standard deviation for the oxic zone is 11.7 %, oxic-anoxic transition zone 11.2 % and the anoxic zone 173.2 % determined by triplicates sampled in 70, 80 and 175 m water depth, respectively.Blank values were obtained by terminating with sodium hydroxide immediately after tracer addition and sample treatment as described above.Assuming a first-order kinetic during methane degradation, the methane oxidation rates (r ox ) were calculated according to Eq. ( 1), where 14 CO 2 is the radioactivity (dpm = disintegrations per minute) of the microbial-formed carbon dioxide, 14 CH 4 is the radioactivity (dpm) of the injected tracer, CH 4 is the in situ methane concentration of the water sample (nmol L −1 ), t is the incubation time (d) and 0.9 the experimentally determined recovery factor (recovery factor of 1 indicates a quantitative release and capture of 14 CO 2 produced by the turnover of CH 4 ).This factor was determined by a known amount of H 14 CO − 3 dissolved in water, followed by acidification and the capture of 14 CO 2 as described above.The turnover rate constant (k) in Eq. ( 2) expresses the fraction of 14 CH 4 that is oxidized per unit time, whereby the turnover time of methane is represented by its reciprocal (1/k).
Integrated methane oxidation rates (ir ox ) were calculated according to the trapezoid rule displayed in Eq. ( 3).Therefore we subdivided the redox zone into depth intervals.Each depth interval was defined by two consecutive sampling depths, whereas dz (m) is the vertical distance between the two samples.f (z) and f (z+ dz) are the corresponding oxidation rates (r ox ) [µmol d −1 m −3 ] for each sampling depth.The total oxidation rate within the redox zone (ir ox ) was obtained by summing up the individual oxidation rates of each depth interval.

pmoA gene and transcript analysis
Of each sample, 1000 mL was filtered on a Durapore filter (0.2 µm pore size, GVWP, Merck Millipore, USA), frozen in liquid nitrogen and stored at −80 • C. DNA and RNA were extracted from frozen filters as described by Weinbauer et al. (2002).pmoA transcripts were analyzed with reverse transcriptase PCR (30-33 cycles) using the primer system A189f/mb661r (Holmes et al., 1995;Costello and Lidstrom, 1999) followed by DGGE fingerprinting as described in Schmale et al. (2012).Visible bands were cut out of the gel, reamplified and sequenced by LGC Genomics GmbH (Berlin, Germany).Sequences were checked for quality using SeqMan software (DNASTAR Inc., USA).Phylogenetic affiliations of the partial sequences were initially estimated with the program BLAST (Altschul et al., 1990).In addition 50 ng DNA of each water sample was processed via PCR (30-35 cycles) and DGGE under the same conditions to determine the total methanotrophic assemblages within the entire water column.

Gotland Deep
During our field study the temperature and salinity profiles (Fig. 2a) demonstrated strong vertical gradients, revealing thermohaline stratification with pronounced density differences of individual water bodies.The thermocline extended from 15 to 30 m water depth and the halocline was located in a depth interval from 60 to 80 m.The vertical turbidity profile showed two maxima -one in the surface water and the other in around 127 m water depth.According to the content of oxygen (O 2 , Fig. 2b), the water column can be separated into different depths intervals (Tyson and Pearson, 1991;Rabalais et al., 2010), an oxic zone from 0 to 81 m (349-9 µM O 2 ), a redox zone from 81 to 143 m (< 9 µM O 2 ) and an anoxic zone below 143 m with no detectable O 2 concentrations.Hydrogen sulfide concentrations (H 2 S, Fig. 2b) build up slightly between 93 and 127 m (3-17 µM H 2 S) but show a stronger increase below that depth interval (max.172 µM H 2 S).The turbidity anomalies covered a depth interval of approximately 62 m (81-143 m), reflecting the redox zone and the subsequent transition to anoxic conditions (chemocline) (Kamyshny et al., 2013).For the Gotland Deep it is known that O 2 and H 2 S can co-occur at the redox zone (Labrenz et al., 2010).However, since O 2 concentrations were only measured until first detection of H 2 S, the co-occurrence of O 2 and H 2 S is not documented in our data set.Assuming that the suboxic water layer is mirrored by the depth interval of turbidity anomalies (Kamyshny et al., 2013), the redox zone is positioned between 81 m and 143 m water depth (Fig. 2b).
The methane profile (Fig. 2b) revealed concentrations lower than 10 nM CH 4 in the surface water.Below 80 m water depth, the methane concentration increased with increasing water depth with a methane maximum of 1233 nM CH 4 in the bottom water at 223 m water depth.The stable carbon isotope ratios in the deep-water body are characterized by relatively low δ 13 C CH 4 values (−84 ‰ at 223 m water depth) and a continuous enrichment of 13 C CH 4 towards the redox zone with strongly increasing δ 13 C CH 4 values between 90 and 80 m water depth (δ 13 C CH 4 from −67 to −45 ‰).Above the redox zone the isotopic ratio shifted back to more negative values (δ 13 C CH 4 from −45 to −68 ‰) and showed values at the sea surface that are close to atmospheric δ 13 C ratios of methane (−47.6 ‰, http://www.esrl.noaa.gov/gmd/dv/iadv/) measured in the Baltic Sea.
Methane oxidation rates were measured in the oxic, suboxic and anoxic part of the water column.The major consumption of methane occurred in a depth interval between 80 and 130 m (Fig. 2c).The highest rate was measured within the redox zone (0.12 nM d −1 at 90 m water depth) with a methane turnover time of 455 d (k = 0.0022 d −1 , Fig. 4).
Expression of the pmoA gene was restricted to only one type I methanotroph in a depth interval from 85 to 125 m.Sequence analysis revealed a similarity of 100 % to an uncultured bacterium named Uncultured GotDeep_pmoA1 (accession number KC188735), which had already been detected earlier in the central Baltic Sea (Schmale et al., 2012).pmoA genes could not be detected in this study.

Landsort Deep
Water column data from the Landsort Deep display thermohaline stratification similar to that observed in the Gotland Deep (Fig. 3a).The thermocline extended from 10 to 30 m water depth, and the halocline was located in a depth interval from 50 to 80 m.Apart from the turbidity signal in the surface water, a pronounced signal was present in a depth interval between approximately 91 and 130 m (Fig. 3a).The oxic zone extended from 0 to 84 m (351-9 µM O 2 ; see Sect.4.1 above).Again following the assumption, that turbidity anomalies can be used as an indicator for the depth interval of O 2 -H 2 S transition zone (Kamyshny et al., 2013), the redox zone is positioned between 84 and 130 m water depth (Fig. 3b).H 2 S concentrations increased only slightly within the lower redox zone and remained relatively constant towards the sediment.The deep water revealed maximal H 2 S concentrations of about 14 µM.
The methane concentration profile showed a strong enrichment close to the bottom (2935 µM CH 4 at 422 m), a pronounced methane gradient within the redox zone (from 80 to 124 m) and methane concentrations lower than 10 nM CH 4 in the surface water.At the redox zone, a sharp stable carbon isotopic shift was found between 99 and 85 m (δ 13 C CH 4 , from −71 to −20 ‰ in the upper part of the redox zone).
Methane oxidation was detected throughout the water column, where the major consumption of methane took place between 80 and 115 m water depth (Fig. 3c).The highest rate was measured within the redox zone (0.61 nM d −1 at 90 m water depth) with a methane turnover time of 127 days (k = 0.0079 d −1 , Fig. 4).
As determined for the Gotland Deep, the expression of the pmoA gene was restricted to the Uncultured Got-Deep_pmoA1 in a depth interval from 70 to 115 m (Fig. 3c).pmoA genes were detected between 80 and 115 m water depth.

Discussion
Our physical, chemical and microbiological results (summarized in Table 1) show considerable differences between both deeps.In the following chapter, we will discuss the different factors controlling the fate of methane in the Landsort Deep (LD) and Gotland Deep (GD) based on data of the methane dynamics and the hydrodynamic situation.

Concentration and stable isotope pattern
The measured δ 13 C CH 4 values (GD: −84 ‰, 223 m; LD: −71 ‰, 347 m) in the anoxic waters of both deeps are in the range of those which are representative for a biogenic origin of methane (Whiticar, 1999).This stable isotope signature is modified by microbial methane turnover in the overlain water column as this consumption impacts the concentration distribution of methane and its stable carbon isotope pattern (Whiticar, 1999;Reeburgh, 2007;Schmale et al., 2010).Our data set points to prominent microbial methane consumption within the redox zones in both deeps (Figs. 2 and 3).The significance of this specific depth interval is emphasized by a constant methane concentration decrease and a pronounced change of the δ 13 C CH 4 values.Here, the observed enrichment of 13 C CH 4 within the redox zone in both deeps can be explained by isotopic discrimination during microbial methane oxidation due to the kinetic isotope effect that leads to an enrichment of 13 C CH 4 in the residual methane pool (Whiticar, 1999).In the LD, the steeper methane gradient within the redox zone together with the stronger 13 C CH 4 enrichment in the lower oxic zone indicates more pronounced and efficient methane consumption compared to the GD assuming similar eddy diffusion.

Methane oxidation rates
The outstanding position of the redox zone in the LD and GD as an important depth interval for microbial methane oxidation is also supported by measured elevated oxidation rates Biogeosciences, 10, 7863-7875, 2013  in this specific depth.However, remarkable differences are obvious between both deeps.Compared to the redox zone of the LD, which is characterized by the highest oxidation rates (max.rate 0.61 nM d −1 ), the rates in the GD (max.rate 0.12 nM d −1 ) were much lower.Oxidation rates measured in the suboxic zone of anoxic basins like the central Black Sea: Reeburgh et al., 1991;Durisch-Kaiser et al., 2005)  high as the integrated rate in the GD (1.77 µmol d −1 m −2 , 81-143 m).At this zone the maximum turnover rate constant (k) calculated for the LD (0.0079 d −1 ) is nearly four times as high as the constant in the GD (0.0022 d −1 , Fig. 4).Assuming that the turnover rate constant is reflecting the population size of methanotrophic microorganisms (Kessler et al., 2011), the growth of these organisms is possibly stimulated by the stronger stability of the intermediate water body that is less perturbed by lateral intrusions in the LD compared to the GD (see Sect. 5.2.2).Furthermore, elevated turnover rate constants were also detected in the oxic zone of the LD and not only within the redox zone as it was observed in the GD.This is in accordance with the transcript analysis, which could confirm the expression of the pmoA gene above the redox zone of the LD, indicating active methanotrophs also in the oxic zone.
In addition to these perturbations, the concentration of methane also influences the abundance of the microorganisms involved in the turnover of CH 4 within the redox zones of both deeps.Previous studies showed that increased substrate availability results in a dynamic adaption of the population size of the aerobic methanotrophic community (Valentine et al., 2001;Kessler et al., 2011).In conjunction with our study, this would imply that in comparison to the GD the higher methane concentrations within the redox zone of the LD (GD: 200 nM CH 4 , 139 m, LD: 799 nM CH 4 , 124 m) promote microbial methane oxidation and the growth of the methanotrophic community.The different population sizes within the redox zones are also supported by our DNA analysis.In contrast to the GD, where no pmoA genes could be amplified in PCR reaction and thus probably were below the detection limit, DNA analysis for the identification of pmoA genes on the samples obtained in the LD yielded in PCR products.Thus, the pmoA gene copy number at GD was below the detection limit of our approach.Within this study we were not able to determine this limit directly.But for soil methanotrophs it ranged between 10 1 and 10 2 copies of the pmoA gene per reaction (Kolb et al., 2003), which could be a realistic number also for our study.
To determine the kinetic fractionation factor (α) of microbial methane oxidation within the redox zones we created a plot of δ 13 C CH 4 versus 1 / CH 4 (Fig. 5) according to Mau et al. (2012).To avoid any side effects in these calculations, which could be caused by lateral intrusions (i.e., input of CH 4 from other regions with a different δ 13 C CH 4 signature), we used in a first step the data set obtained in the LD, which is compared to the GD characterized by a less disturbed intermediate water layer (Dellwig et al., 2012).Motivated by the apparent restriction of methane oxidation within the redox zone, as observable from the methane concentration gradient and 13 C CH 4 enrichment in the LD redox zone (Fig. 3b), we applied a closed-system Rayleigh fractionation approach (Coleman et al., 1981).The methane oxidation trend between the deep water (1192 nM CH 4 , δ 13 C −71 ‰) and the water depth, which is characterized by the strongest 13 C CH 4 enrichment within the redox zone (δ 13 C CH 4 : −20 ‰ at 85 m water depth), fitted best with a fractionation factor (α) of 1.012 (Fig. 5a).In a second step, we used this derived α to calculate the oxidation trend in the GD, which is characterized by a more disturbed intermediate water layer (Dellwig et al., 2012).The assumption of a similar α is justified because of comparable temperatures and chemical conditions, and in particular by the identification of the same single methanotrophic bacterium in both deeps (see Sect. 5.1.3).The calculated theoretical oxidation trend can be used to classify each CH 4 data point into three main groups: below the mixing line (CH 4 affected by mixing processes), between the mixing line and oxidation trend (CH 4 influenced by mixing and partly influenced by oxidation) and above the oxidation trend (CH 4 clearly related to oxidation processes).Our results show that the CH 4 data points in both basins fit reasonably well in the 13 C CH 4 -depleted part of the oxidation trend.However, within the redox zone of the eastern Gotland Basin a deviation is visible in the 13 C CH 4 enriched part (Fig. 5B).Based on oceanographic studies, which indicated a stronger perturbation of the redox zone in the eastern compared to the western Gotland Basin (Dellwig et al., 2012;Kamyshny et al., 2013), we assume that the observed deviation from the oxidation trend results from enhanced mixing within the redox zone of the eastern Gotland Basin.

Aerobic methanotrophs in the redox zone
Sequence analysis revealed that in both deeps, methane oxidation in the redox zone is restricted to type I methanotrophic bacteria, indicating that the different environmental settings (i.e., methane concentrations and disturbances within the redox zone by lateral intrusions) did not influence the microbial diversity of aerobic methanotrophs.The exclusive detection of type I methanotrophs confirms recently published biomarker analysis, which indicates the absence of type II methanotrophic bacteria in the water column of the GD (Schmale et al., 2012;Berndmeyer et al., 2013).The restricted diversity of type I methanotrophs in the central Baltic Sea is in agreement with studies conducted by Schubert et al. (2006c) in the Black Sea, who identified type I methanotrophic bacteria as the most important methane consumer in the redox zone zone.However, other studies conducted in the oxic zone as well as in the redox zone of the Black Sea also proved the existence of type II and X methanotrophs (Durisch-Kaiser et al., 2005;Blumenberg et al., 2007).The methanotrophic bacterium identified in the GD and LD is restricted to the phylotype Uncultured Got-Deep_pmoA1 (Schmale et al., 2012).A phylogenetically affiliated phylotype of the Uncultured GotDeep_pmoA1 was also detected in a meromictic crater lake (Lake Pavin) that is characterized by a permanently stratified water column (Biderre-Petit et al., 2011).Apart from the identified phylotype in the present study, two main phylotypes of type I methanotrophs have been found in the meromictic lake.We assume that the reduced diversity in the Baltic Sea re-dox zone is related to an overlap of sulfide-and oxygencontaining waters as a result of lateral intrusions (Dellwig et al., 2012).The influence of sulfide-containing waters on the microbial diversity has been shown by Labrenz et al. (2010).
Based on this relation it is assumed that the toxicity of sulfide to many organisms may inhibit the activity of other methanotrophs than the detected phylotype (Schmale et al., 2012).However, a higher diversity of active aerobic methanotrophic bacteria in the compared to the GD less disturbed redox zone of the LD could not be detected.
The detection of the mcrA gene for analyzing anaerobic methanotrophs was not performed in our studies due to the main focus on aerobic methane oxidation.From a theoretical point of view, sulfate-dependent methane oxidation (AOM) in the anoxic water layer should be possible as the ambient sulfate concentrations (LD 0.81 g kg −1 and GD 0.93 g kg −1 ; derived from the averaged salinity in the anoxic water layer; calculated after Bruland, 1983) would enable this process (Reeburgh, 2007).This assumption is supported by our methane oxidation rate measurements, which indicate anaerobic turnover of methane in both basins.However, the determined oxidation rates are significantly affected by high standard deviations and a potential underestimation due to the chosen 14 CH 4 tracer amount.The method developed for this study was designed to analyze microbial methane oxidation within the oxic and redox zone, and as such these oxidation rates cannot provide clear evidence for the existence of AOM processes.

Hydrodynamic controls on the fate of methane in
the Gotland and Landsort Deep

Vertical mixing
The vertical transport of matter (e.g., nutrients, gases, particles) in the central Baltic Sea is strongly influenced by vertical mixing processes.The different intensities of mixing directly impact the concentration distribution pattern of chemical species in the water column (Nausch et al., 2008;Holtermann et al., 2012).Especially the transport across the chemocline influences important biogeochemical processes as reduced and energy-rich chemical species like CH 4 , H 2 S, iron (Fe 2+ ) and manganese (Mn 2+ ) are abundant in high concentrations within the deep water and can drive microbial reactions at the redox zone (Labrenz et al., 2005;Dellwig et al., 2012;Schmale et al., 2012).
Between 300 m water depth and the lower edge of the redox zone (130 m) the LD is characterized by a uniform methane profile, whereas the methane concentrations in the GD decrease constantly with decreasing depth.The same relationship is also reflected by the H 2 S profiles and the δ 13 C CH 4 distribution in the anoxic water bodies.The shapes of the methane profiles are related to the different vertical mixing rates (K ρ ; see also Eq. 4) between the two basins, which are increased nearly 6 times in the LD compare to the GD (K ρ : GD 1.1 × 10 −5 m 2 s −1 , LD 6.2 × 10 −5 m 2 s −1 , annual mean at 150 m water depth; Axell, 1998).The uniform shape of the LD methane profile in a depth range between approximately 150 and 300 m as well as the increase towards the bottom can be nicely explained by the vertical structure of K ρ , which shows a maximum (K ρ : 9.5 × 10 −5 m 2 s −1 ) in the depth range between 200 and 300 m and a decrease (K ρ : 1.0 × 10 −5 m 2 s −1 ) to the bottom.
Assuming that the flux of methane from the anoxic deep water to the redox zone and the consumption of methane within the redox zone are in steady state, we can use our data set to calculate vertical mixing rates (K ρ ) for the upper anoxic zone (GD: 143-200 m, LD: 130-250 m).Using the integrated methane oxidation rates (ir ox ) in the redox zones (see Sect. 5.1.2) and the methane concentration gradients within the upper anoxic zone, K ρ can be calculated according to Eq. ( 4), where c is the in situ concentration of methane [µmol m −3 ] and z the water depth [m].
The mixing rates calculated by this method for the LD are one order of magnitude higher than those for the GD (GD: 2.5 × 10 −6 m 2 s −1 , LD: 1.6 × 10 −5 m 2 s −1 ).This higher vertical mixing rate in the LD is in accordance with the observations of Axell (1998), who reported a 6-times-larger K ρ for the LD than for the GD at the 150 m depth level.However, the mixing rates calculated within the present study are 4 times lower compared to the values reported by Axell (1998).
The enhanced vertical transport of methane implies a higher supply of substrate to the methanotrophic community within the redox zone of the LD.

Lateral intrusions
The turbidity in anoxic basins is often used as a marker to determine the depth of the chemocline.Turbidity anomalies near the chemocline reflect an incompletely understood phenomenon.It is often explained by the diffusion of reduced chemical species across the redox zone, which lead to the precipitation of metal oxides (e.g., oxides of Fe and Mn) or the oxidation of H 2 S to elemental sulfur S 0 (Dellwig et al., 2010;Kamyshny et al., 2013).Another theory is that the energy-rich compounds transported from the anoxic zone across the chemocline increase microbial turnover of matter and thus the abundance of bacteria in a discrete depth interval (Prokhorenko et al., 1994;Dellwig et al., 2010;Labrenz et al., 2010).However, the distribution pattern can also be used to discuss the stability of that chemical boundary as lateral intrusions of external water masses into the redox zone will perturb the established chemical stratification (Kamyshny et al., 2013) and may inject O 2 / H 2 S enriched water into this specific depth interval (Dellwig et al., 2012).The turbidity profile of the GD (Fig. 2a) reveals different distinct peaks along the redox zone.Previous studies in the GD indicate that these turbidity anomalies are related to lateral intrusions into the redox zone (Lass et al., 2003;Dellwig et al., 2012).Therefore, lateral intrusions or internal waves produce perturbations in the intermediate water column and prevent the formation a clearly defined redox zone.In contrast, the LD reveals only one pronounced turbidity signal on top of the redox zone with a constant decrease with increasing water depth (Fig. 3a).Even if this decrease cannot currently be explained, the clear trend without turbidity spikes (in contrast to the observation in the GD) point to a more undisturbed situation in the LD redox zone (Dellwig et al., 2012;Kamyshny et al., 2013).Temperature-salinity diagrams (T -S) are useful approaches to identify the strength of intrusions in the central Baltic Sea (Dellwig et al., 2012).Signatures of intrusions can be identified in the GD by slight variations in the T -S profiles across the entire redox zone (see insert Fig. 6a).The LD is characterized by a relatively smooth T -S pattern, indicating the undisturbed situation of this intermediate water layer (see insert Fig. 6b).

Conclusions
In the present study, methanotrophic processes within the stratified water columns of the Gotland Deep and Landsort Deep (central Baltic Sea) were investigated to reveal their dependence on different hydrographic conditions.Within the redox zones of both deeps, microbial consumption of methane was identified by the distribution patterns of the concentration of methane, 13 C CH 4 and elevated methane oxidation rates.In this intermediate depth interval one potentially active type I methanotrophic bacterium was identified at both sampling sites, indicating that the different hydrographic conditions apparently do not impact the diversity of methanotrophic communities.In contrast, the microbial turnover of methane in the redox zones reveals considerable differences with lower methane oxidation rates in the Gotland Deep compare to the Landsort Deep.The intensity of lateral intrusions and the vertical transport rate of methane from the anoxic zone into the redox zone are different between both deeps, and seem to represent the key processes that control the turnover of methane within the redox zone.
Our results confirm that pelagic microbial methane oxidation within redox gradients represent an efficient methane sink that prevents the escape of methane from the deep anoxic zone into the atmosphere.The comparative investigations in the Gotland Deep and Landsort Deep also indicate that this microbially mediated process can react on different environmental conditions by adapting the population size of methanotrophs and/or the rate of methane oxidation.Open questions regarding the involvement of other electron acceptors besides oxygen, the temporal dynamic of methane degradation and the verification of anaerobic methane oxidation in the deep anoxic waters need to be clarified in further studies.

Fig. 1 .
Fig. 1. (A) Sampling sites in the central Baltic Sea (red dots).(B) Bathymetric map of the central Baltic Sea with the Gotland Deep in the eastern Gotland Basin and the Landsort Deep in the western Gotland Basin.(C) Cross section from the eastern (a) to the western Gotland Basin (b).

Fig. 2 .
Fig. 2. Gotland Deep.(A) Vertical profiles of salinity (green), temperature (red) and turbidity (black); (B) oxygen (blue squares), hydrogen sulfide (black squares), methane (red circles) and δ 13 C values of methane (green triangles); (C) methane oxidation rates (light-blue diamonds) and sampling depths of pmoA gene expression analysis (black bars denote the occurrence and white bars the absence of active type I methanotrophs).To obtain the standard deviation (s) for methane oxidation rates, triplicates were taken in three water depths (70, 85, 175 m).The solid error bars indicate the standard deviation from these triplicates, whereas the dashed error bars show the standard deviation calculated from these triplicates for the single water samples.The redox zone is indicated by the gray-shaded area.

Fig. 3 .
Fig. 3. Landsort Deep.(A) Vertical profiles of salinity (green), temperature (red) and turbidity (black); (B) oxygen (blue squares) and hydrogen sulfide (black squares), methane (red circles) and δ 13 C values of methane (green triangles); (C) methane oxidation rates (light-blue diamonds) and sampling depths of pmoA gene expression analysis (black bars denote the occurrence and white bars the absence of active type I methanotrophs).To obtain the standard deviation (s) for methane oxidation rates, triplicates were taken in three water depths (70, 80, 175 m).The solid error bars indicate the standard deviation from these triplicates, whereas the dashed error bars show the standard deviation calculated from these triplicates for the single water samples.The redox zone is indicated by the gray-shaded area.The insert in (C) illustrates the depth interval of the redox zone in higher vertical resolution.

Fig. 4 .
Fig. 4. Turnover rate constants (k) from the Gotland Deep (red) and Landsort Deep (black).Indicated are the redox zones of the Gotland Deep (gray-shaded area) and the Landsort Deep (white rectangle).

Fig. 5 .
Fig. 5. Landsort Deep (A) and Gotland Deep (B).δ 13 C CH 4 versus 1 / CH 4 .Please note different scales for δ 13 C CH 4 .The solid lines denote mixing between deep and surface water, the dashed lines represent the oxidation trends based on a fractionation factor of α = 1.012, the open triangles indicate the anoxic zone, the red circles the redox zone and the open circles the oxic zone.

Fig. 6 .
Fig. 6.Temperature-salinity diagrams.The inserts denote the depth interval where the redox zone is located.