Evidence for methane production by the marine algae Emiliania huxleyi

Abstract. Methane (CH4), an important greenhouse gas that affects radiation balance and consequently the earth's climate, still has uncertainties in its sinks and sources. The world's oceans are considered to be a source of CH4 to the atmosphere, although the biogeochemical processes involved in its formation are not fully understood. Several recent studies provided strong evidence of CH4 production in oxic marine and freshwaters, but its source is still a topic of debate. Studies of CH4 dynamics in surface waters of oceans and large lakes have concluded that pelagic CH4 supersaturation cannot be sustained either by lateral inputs from littoral or benthic inputs alone. However, regional and temporal oversaturation of surface waters occurs frequently. This comprises the observation of a CH4 oversaturating state within the surface mixed layer, sometimes also termed the "oceanic methane paradox". In this study we considered marine algae as a possible direct source of CH4. Therefore, the coccolithophore Emiliania huxleyi was grown under controlled laboratory conditions and supplemented with two 13C-labeled carbon substrates, namely bicarbonate and a position-specific 13C-labeled methionine (R-S-13CH3). The CH4 production was 0.7 µg particular organic carbon (POC) g−1 d−1, or 30 ng g−1 POC h−1. After supplementation of the cultures with the 13C-labeled substrate, the isotope label was observed in headspace CH4. Moreover, the absence of methanogenic archaea within the algal culture and the oxic conditions during CH4 formation suggest that the widespread marine algae Emiliania huxleyi might contribute to the observed spatially and temporally restricted CH4 oversaturation in ocean surface waters.

Abstract.Methane (CH 4 ), an important greenhouse gas that affects radiation balance and consequently the earth's climate, still has uncertainties in its sinks and sources.The world's oceans are considered to be a source of CH 4 to the atmosphere, although the biogeochemical processes involved in its formation are not fully understood.Several recent studies provided strong evidence of CH 4 production in oxic marine and freshwaters, but its source is still a topic of debate.Studies of CH 4 dynamics in surface waters of oceans and large lakes have concluded that pelagic CH 4 supersaturation cannot be sustained either by lateral inputs from littoral or benthic inputs alone.However, regional and temporal oversaturation of surface waters occurs frequently.This comprises the observation of a CH 4 oversaturating state within the surface mixed layer, sometimes also termed the "oceanic methane paradox".In this study we considered marine algae as a possible direct source of CH 4 .Therefore, the coccolithophore Emiliania huxleyi was grown under controlled laboratory conditions and supplemented with two 13 C-labeled carbon substrates, namely bicarbonate and a position-specific 13 C-labeled methionine (R-S-13 CH 3 ).The CH 4 production was 0.7 µg particular organic carbon (POC) g −1 d −1 , or 30 ng g −1 POC h −1 .After supplementation of the cultures with the 13 C-labeled substrate, the isotope label was observed in headspace CH 4 .Moreover, the absence of methanogenic archaea within the algal culture and the oxic conditions during CH 4 formation suggest that the widespread marine algae Emiliania huxleyi might contribute to the observed spatially and temporally restricted CH 4 oversaturation in ocean surface waters.

Introduction
Methane (CH 4 ), the second-most important anthropogenic greenhouse gas after CO 2 , is the most abundant reduced organic compound in the atmosphere and plays a central role in atmospheric chemistry (IPCC, 2013;Kirschke et al., 2013;Lelieveld et al., 1998).The mixing ratio of CH 4 in the atmosphere has been increasing from preindustrial values of around 715 ppbv (parts per billion by volume) to about 1800 ppbv in 2010 (Kirschke et al., 2013).In total, annual CH 4 emissions from natural and anthropogenic sources amount to 500-600 Tg (10 12 g) yr −1 .They derive from various terrestrial and aquatic sources and are balanced primarily by photochemical oxidation in the troposphere (≈ 80 %), diffusion into the stratosphere, and microbial CH 4 oxidation in soils.
Until recently, natural sources of atmospheric CH 4 in the biosphere have been considered to originate solely from Published by Copernicus Publications on behalf of the European Geosciences Union.
strictly anaerobic microbial processes in wetland soils and rice paddies, the intestines of termites and ruminants, human and agricultural waste, and from biomass burning, fossil fuel mining, and geological sources including mud volcanoes, vents and seeps.However, more recent studies have suggested that terrestrial vegetation, fungi, and mammals may also produce CH 4 without an input from methanogens and under aerobic conditions (Bruhn et al., 2012;Ghyczy et al., 2008;Keppler et al., 2006;Lenhart et al., 2012;Wang et al., 2013b;Liu et al., 2015).A fraction of these vegetationderived emissions might be released directly by in situ formation in plants (Bruhn et al., 2012;Keppler et al., 2009;Wang et al., 2013a), and it is now apparent that several pathways exist by which CH 4 is generated under aerobic conditions (Bruhn et al., 2014;Messenger et al., 2009;Wang et al., 2013b).Hence, the biogeochemical CH 4 cycle appears to be even more complex than previously thought.
In particular, the biogeochemical cycle of CH 4 in the oceans is still far from being understood.The world's oceans are considered to be a minor source of CH 4 to the atmosphere with approximately 0.6-1.2Tg CH 4 yr −1 (Rhee et al., 2009).Concentrations of CH 4 in near-surface waters are often 5-75 % supersaturated with respect to the atmosphere, implying a net flux from the ocean to the atmosphere (Conrad, 2009;Reeburgh, 2007;Scranton and Brewer, 1977).Because the surface ocean is also saturated or slightly supersaturated with oxygen, which does not favor methanogenesis, the observed CH 4 supersaturation has been termed the oceanic methane paradox (Kiene, 1991).To explain the source of CH 4 in surface waters, it has been suggested that methanogenesis takes place in anoxic microenvironments of organic aggregates (Grossart et al., 2011;Karl and Tilbrook, 1994;Bogard et al., 2014), the guts of zooplankton or fish (de Angelis and Lee, 1994;Oremland, 1979), and inside bacterial cells (Damm et al., 2015).It has also been shown that contrary to the conventional view, some methanogens are remarkably tolerant to oxygen (Angel et al., 2011;Jarrell, 1985).
A potential substrate for methanogenesis in such anoxic microniches is dimethylsulfoniopropionate (DMSP) (Damm et al., 2008(Damm et al., , 2015;;Zindler et al., 2013), an algal osmolyte that is abundant in marine phytoplankton and serves as a precursor of dimethyl sulfide (DMS) and dimethyl sulfoxide (DMSO) (Stefels et al., 2007;Yoch, 2002) For example, Zindler et al. (2013) measured concentrations of DMS, DMSP, DMSO, and CH 4 , as well as various phytoplankton marker pigments in the surface ocean along a north-south transit from Japan to Australia.Positive correlations between DMSP (dissolved) and CH 4 , and DMSO (particulate and total) and CH 4 , were found along the transit.Based on their data, they concluded that DMSP and DMSO and/or their degradation products serve as substrates for methanogenic archaea in the western Pacific Ocean.Damm et al. (2010) hypothesized that under N limitation and a concomitant availability of phosphorus, marine bacteria use DMSP as a carbon source and thereby release CH 4 as a by-product and its production could yield energy under aerobic conditions.Methanethiol, a further potential degradation product of DMSP, may act as a direct precursor of methane in aerobic environments.By reason of thermodynamic calculations the authors considered it possible for microorganisms to yield energy from the pathway of methanethiol formation operating in its reverse direction, whereby methane is formed.
An alternative non-biological CH 4 formation pathway in seawater might occur via a photochemical pathway due to the formation of methyl radicals; however, photochemical production of CH 4 in oceans is thought to be negligible under oxic conditions (Bange and Uher, 2005).
In addition, Karl et al. (2008) suggested that CH 4 is produced aerobically as a by-product of methylphosphonate (MPn) decomposition when aerobic marine organisms use methylphosphonic acid as a source of phosphorus when inorganic sources of this element are limited.Furthermore, a mechanism has been identified that leads to the formation of CH 4 from MPn via enzyme-catalytic cleavage of the C-P bond (Kamat et al., 2013).The critical issue with this pathway is that MPn is not a known natural product nor has it been detected in natural systems.However, it was recently shown that the marine archaeon Nitrosopumilus maritimus encodes a pathway for MPn biosynthesis and that it produces cell-associated MPn esters (Metcalf et al., 2012).They argued that these cells could provide sufficient amounts of MPn precursor to account for the observed CH 4 production in the oxic ocean via the C-P lyase-dependent scenario suggested by Karl et al. (2008).However, it was not possible to explain the supersaturation state of CH 4 in oxic surface water by the quantification of produced CH 4 from dissolved MPn under natural conditions (del Valle and Karl, 2014).
It remains uncertain whether CH 4 formation from MPn (Karl et al., 2008) or the metabolism of DMSP by methanogens in anoxic microenvironments (Damm et al., 2008(Damm et al., , 2015;;Zindler et al., 2013) is sufficient to provide a permanent increase in the concentration of CH 4 in oxygenated surface waters or whether other pathways are also required to fully explain the CH 4 oversaturation in oxic waters.In this context it is important to note that almost 40 years ago researchers (Scranton and Brewer, 1977;Scranton and Farrington, 1977) already mentioned the possibility of in situ formation of CH 4 by marine algae.These scientists measured CH 4 saturation states in open-ocean surface waters of the west subtropical North Atlantic.They observed 48-67 % higher CH 4 concentrations in surface waters than estimated from atmospheric equilibrium concentration, with a narrow maximum of CH 4 concentration in the uppermost part of the pycnocline.Since the loss of CH 4 from the surface to the atmosphere was calculated to be much larger than diffusion from CH 4 maxima of the pycnocline into the mixed layer, an in situ biological CH 4 formation process within the mixed layer was hypothesized (Scranton and Farrington, 1977; Scranton and Brewer, 1977).However, direct evidence of algae-derived CH 4 formation from laboratory experiments with (axenic) algae cultures is still lacking, and the accumulation of CH 4 in the upper water layer has not yet been directly related to production by algae.
The aim of our study was to quantify in situ CH 4 formation from marine algae such as coccolithophores and to identify precursor compounds of CH 4 via 13 C labeling techniques.Therefore, we used Emiliania huxleyi, a widely distributed, prolific alga.The coccolithophore blooms including E. huxleyi are the major regional source of DMS release to the atmosphere (Holligan et al., 1993).Specific goals in this study were (I) to measure the CH 4 production of a biogeochemically important marine phytoplankton, (II) to screen for methanogenic archaea or bacteria, and (III) to identify methyl sulfides, such as the amino acid methionine, which play a role in metabolic pathways of algae, as possible precursors for CH 4 .
To investigate algae-derived CH 4 formation a closedchamber system was used.Hence, 2 L flasks (Schott, Germany) filled with 1800 mL sterile filtered seawater and with 480 mL headspace volume were used in our investigations.The flasks were sealed with lids (GL 45, PP, 2 port, Duran Group) equipped with two three-way ports (Discofix ® -3, B-Braun), where one port was used for water and the other port (fitted with a sterile filter, 0.2 µm; PTFE, Sartorius) for gas sampling.The cells were grown on a day-night cycle of 16 and 8 h at 20 • C and a light intensity of ≈ 450 µE over a 10day period.The initial dissolved inorganic carbon (DIC) of the culture medium was 2235 µmol L −1 (for details on DIC measurements, see Langer et al., 2009).The different treatments and the number of replicates are provided in Table 1 and Fig. 1.To increase the detectability of CH 4 formation and to exclude a possible contamination with CH 4 from the surrounding air, 13 C-labeled bicarbonate (NaH 13 CO 3 , 99 % purity, Sigma-Aldrich, Germany) was added to the cultures.Bicarbonate (Bic) was used as a C source for biomass production.To gain a 13 C enrichment of 1 % of the total inorganic C (CO 2 , HCO − 3 , and CO 2− 3 ), 22.35 µmol L −1 NaH 13 CO 3 was added, leading to a theoretical δ 13 C value of 882 ‰.
To test methionine (Met) as a precursor of algae-derived CH 4 , Met with only the sulfur-bound methyl group 13 C labeled (R-S-13 CH 3 , 99 % enriched, 1 µmol L −1 ) was added to the cultures.Met has previously been identified as a methyl-group donor for CH 4 biosynthesis in higher plants and fungi (Lenhart et al., 2012(Lenhart et al., , 2015)).Moreover, marine algae use Met to produce DMSP, DMS, and DMSO, substances that can be released into seawater and are known to act as precursors for abiotic CH 4 production.

Sample collection and analysis
Samples were taken daily from day 4 until day 10 (see Table 1).Prior to day 4, algae biomass was too low to allow the measurement of changes in CH 4 mixing ratio.
For gas chromatography (GC) and continuous-flow isotope ratio mass spectrometry (CF-IRMS) analysis samples of headspace (30 mL) were taken from each flask.GC samples were measured within 24 h after sampling, while GC-IRMS samples were stored in 12 mL exetainers until 13 C-CH 4 measurements were carried out.
After gas sampling, samples of medium (25 mL) from each flask were also taken for cell density determination.These samples were supplemented with 0.15 mL Lugol solution (Utermöhl, 1958) and stored in 50 mL Falcon tubes at 4 • C. In order to maintain atmospheric pressure within the flask, the surrounding air was allowed to enter via an orifice fitted with a sterile filter to avoid bacterial contamination.Variable amounts of water and headspace volume as well as the inflow of surrounding air were all taken into consideration when CH 4 production rates were calculated.

Gas chromatography
Gas samples were analyzed for CH 4 mixing ratio within 24 h on a gas chromatograph (Shimadzu GC-14B, Kyoto, Japan) fitted with a flame ionization detector (FID) operating at 230 • C with N 2 as carrier gas (25 mL min −1 ) (Kammann et al., 2009).The GC column (PorapakQ, Fa.Millipore, Schwallbach, mesh 80/100) was 3.2 m long and 1/8 inch in diameter.The length of the precolumn was 0.8 m.The GC gas flow scheme and automated sampling was that of Mosier and Mack (1980) and Loftfield (1997), and peak area integration was undertaken with the software PeakSimple, version 2.66.The standard deviation (SD) of the mean of six atmospheric air standard samples was below 0.2 % for CH 4 .

CF-IRMS for measurement of δ 13 C values of CH 4
Headspace gas from exetainers was transferred to an evacuated sample loop (40 mL).Interfering compounds were separated by GC and CH 4 trapped on Hayesep D. The sample was then transferred to the IRMS system (ThermoFinnigan Delta plus XL, Thermo Finnigan, Bremen, Germany) via an open split.The working reference gas was carbon dioxide of high purity (carbon dioxide 4.5, Messer Griesheim, Frank-furt, Germany) with a known δ 13 C value of −23.64 ‰ relative to Vienna Pee Dee Belemnite (V-PDB).All δ 13 C values of CH 4 were corrected using three CH 4 working standards (isometric instruments, Victoria, Canada) calibrated against IAEA and NIST reference substances.The calibrated δ 13 C-CH 4 values of the three working standards were −23.9 ± 0.2 ‰, −38.3 ± 0.2 ‰, and −54.5 ± 0.2 ‰.Samples were routinely analyzed three times (n = 3) and the average standard deviations of the CF-IRMS measurements were in the range of 0.1 to 0.3 ‰.
All 13 C / 12 C-isotope ratios are expressed in the conventional δ notation in per mil (‰) vs. V-PDB, using the following equation (Eq.1): (1) To determine the δ 13 C signature of the CH 4 source, the Keeling-plot method was applied (Keeling, 1958).

DNA extraction and real-time PCR
Samples for DNA extraction were taken from the stem culture (RCC 1216) during the stationary growth phase (2 × 10 6 cells mL −1 ).After DNA extraction, real-time polymerase chain reaction (qPCR) was used to detect mcrA genes, which are solely found in methanogenic archaea.As positive proof, aliquots of the samples were supplemented with a defined cell density of Methanothermobacter marburgensis (either 10 4 or 10 7 cells mL −1 ) .
The real-time PCR reference standards were produced according to Kampmann et al. (2012).By using the standard solution (5.5 × 10 7 DNA copies µL −1 ), dilution with nucleasefree water was accomplished down to 5.5 × 10 1 copies per µL −1 .All standards and regular samples taken from the flasks were analyzed with four repetitions.
Quality assurance of the real-time PCR product was achieved by melt curve analysis and gel electrophoresis using the fluorescent stain GelRedTM (Biotium).

Cultivation approach
In addition to real-time PCR, a cultivation and enrichment procedure (Kampmann et al., 2012) was conducted to screen for methanogenic archaea in algae cultures.The enrichment medium (Widdel and Bak, 1992) was modified for marine conditions by adding 320 mmol L −1 NaCl, 16 mmol L −1 MgCl 2 , and 1 mmol L −1 NaHCO 3 .At day 10, an aliquot (5 mL) of each cultivation flask was transferred into injection flasks (Ochs, Bovenden-Lenglern, Germany) with the enrichment medium (50 mL) and acetate (10 mM), methanol (5 mM) was added, and in the gas phase H 2 and CO 2 (90 : 10) were provided as substrates.Incubation was carried out over a period of 6 weeks at 20 • C in the dark.

CH 4 mass
The mass of CH 4 (m CH 4 ) per flask was calculated via the ideal gas law from the corrected CH 4 mixing ratio (ppmv), where the changing volume of water and headspace and the inflow of surrounding air were all considered, according to Eq. (3): where p is pressure, T is temperature, R is ideal gas constant, V is volume, and M CH 4 is mol.weight of CH 4 .The solubility of CH 4 in the water phase was calculated according to Wiesenburg and Guinasso (Wiesenburg and Guinasso Jr., 1979) based on the headspace-CH 4 mixing ratio, temperature and salinity of the water phase.

Calculation of CH 4 production
The low CH 4 mixing ratios produced by E. huxleyi during the exponential growth phase precluded the determination of CH 4 production during this period.Therefore, we calculated production from day 7 to day 10, a period representing the transition from exponential to stationary phase.This growth phase features changing growth rates and cellular CH 4 quotas, rendering the dilute-batch method of calculating production inapplicable (Langer et al., 2013).We followed the recommendation of Langer et al. (2013) and calculated incremental (daily) CH 4 production: where P inc is incremental CH 4 production (ng CH 4 cell −1 day −1 ), q inc is incremental cellular CH 4 quota (ng CH 4 cell −1 ), and µ inc is incremental growth rate (day −1 ).Incremental growth rate was calculated according to where t 1 is cell density on the day q inc was determined and t 0 is cell density on the previous day.We present average P inc (SD).In order to compare CH 4 production to literature data it was necessary to normalize to cellular particulate organic carbon (POC) quota as opposed to cell.The POC-normalized CH 4 production is termed "methane emission rate" in the following.Since it was not possible to measure cellular POC quota on a daily basis, we used a literature value determined for the same strain under similar culture conditions, i.e., 10.67 pg POC cell −1 (Langer et al., 2009).We are aware of the fact that the cellular POC quota is likely to change alongside other element quotas when approaching the stationary phase, but this change is well below an order of magnitude (Langer et al., 2013).For our purpose this method is therefore sufficiently accurate to determine POC-normalized CH 4 production.

Statistics
To test for significant differences in cell density, CH 4 mixing ratio, and CH 4 content between the treatments, two-way analysis of variance (ANOVA) (considering repeated measurements) and a post hoc test (Fisher least significant difference (LSD) test; alpha 5 %) were used.

Algae growth
Cell density and growth of the cultures are presented in Fig. 2a, b over the whole incubation period for all treatments.The initial cell density at time 0 (t 0 ) was 3.5 × 10 3 cells mL −1 in all flasks.At day 10 cell density reached its maximum value with 1.37 × 10 6 cells mL −1 (algae), 0.82 × 10 6 cells mL −1 ("algae + 13 C-Bic"), and 1.24×10 6 cells mL −1 ("algae + 13 C-Met").The exponential growth rates (µ) were 0.85 ± 0.2 d −1 for algae + 13 C-Met, 0.98 ± 0.1 d −1 for algae + 13 C-Bic, and 1.06 ± 0.02 d −1 for the control "algae" (n.s., p = 0.286).Significant differences in cell density between the treatments only occurred at days 9 and 10, where the cell density of the control algae was higher than in the treatments where 13 C-Bic or 13 C-Met was added.

Methane mixing ratio
Initial headspace-CH 4 mixing ratios measured at day 4 were in the range of 1899 to 1913 ppbv for all treatments including the controls without algae.From day 4 to day 7 headspace-CH 4 mixing ratios slightly increased in all flasks.Therefore, no significant differences in the CH 4 mixing ratios occurred between the treatments.After day 8 CH 4 mixing ratios in the flasks containing algae were significantly higher compared to the controls without algae (Fig. 2c, d).The highest CH 4 mixing ratios at day 10 corresponded to 2102 ± 62 ppbv (algae + 13 C-Met), 2138 ± 42 ppbv (algae + 13 C-Bic), and 2119 ± 25 ppbv (algae).Hence, from day 4 to day 10 the CH 4 mixing ratios increased by about 192 ppbv (algae + 13 C-Met), 49 ppbv (seawater + 13 C-Met), 235 ppbv (algae + 13 C-Bic), and 67 ppbv (seawater + 13 C-Bic).

Stable carbon isotope values of methane
The δ 13 C signature of headspace CH 4 (δ 13 CH 4 value) is presented in Fig. 2e and f.The addition of 13 C-Bic did not affect CH 4 production of algae, but the δ 13 CH 4 value was clearly different from that of the control algae.The initial value of −47.9 ± 0.2 ‰ increased to 44 ± 13 ‰, whereas in the controls "seawater + 13 C-Bic" and algae no change in the δ 13 CH 4 value was observed.
Based on the initial amount of 13 C-Bic and the total amount of 13 CH 4 at the end of the incubation period, 88.3 ± 17.2 pmol of 22.4 µmol 13 C-Bic were converted to 13 CH 4 .For Met, this was 78.5 ± 18.6 pmol of the initial 1.8 µmol 13 C-Met.
The Keeling plots to determine the 13 C values of the CH 4 source are presented in Fig. 3.For the bicarbonate treatment (algae + 13 C-Bic), the mean δ 13 CH 4 value of the CH 4 source was 811.9 ± 89.9 ‰, which is close to the calculated δ 13 C value of 881.5 ‰ after the addition of NaH 13 CO 3 .
For the treatment algae + 13 C-Met, we applied the Keeling-plot method only for the period from day 5 to day 7, as the increase in the δ 13 C values were not linear after day 7.For this treatment, the δ 13 C values of the CH 4 source range between 967 and 2979 ‰.
The correlation between the growth of the algae cultures and the total amount of CH 4 in the flasks (headspace + water phase) is presented in Fig. 4. For the treatment algae + 13 C-Bic (Fig. 4a), there is an exponential correlation between cell density and CH 4 content (r 2 = 0.994), whereas for the treatment algae + 13 C-Met (Fig. 4b), a linear correlation was observed (r 2 = 0.995).The daily CH 4 content in the flasks for days 8, 9, and 10 is shown in Fig. 5.For all flasks the CH 4 content exceeded the CH 4 content of the respective control, with a continuous increase in the CH 4 content in the flasks containing algae.At day 10, the difference between algae + 13 C-Bic and seawater + 13 C-Bic and between algae + 13 C-Met and "seawater + 13 C-Met" was 65 ± 16 and 54 ± 22 ng, respectively.
The CH 4 production of algae presented in Table 2 shows no major differences between the treatments.Furthermore, for all treatments, the daily CH 4 production rates did not change over time (Fig. 6).

Microbial investigations
Via real-time PCR no mcrA genes could be detected in the flasks containing the CH 4 -producing algae cultures, whereas in the positive control in which the algae culture was supplemented with 10 4 and 10 7 cells mL of the methanogenic archaea Methanothermobacter marburgensis, 9.4 × 10 4 and 4.6 × 10 6 mcrA-gene copies mL −1 have been detected, respectively.
With the cultivation approach, where an aliquot of each flask was taken at day 10 and transferred to the media for the enrichment of methanogenic archaea, no CH 4 production was observed after the 6-week incubation period.In the case of a successful enrichment of methanogenic archaea, the CH 4 -mixing ratio in the headspace would increase over time.

Discussion
Our results of the CH 4 mixing ratio and stable isotope measurements provide unambiguous evidence that E. huxleyi produces CH 4 .In the following we will discuss the relationship between CH 4 production and the growth of the algae, stable isotope measurements, potential precursor compounds, and the exclusion of methanogenic archaea.Finally, we will discuss the implications of our results for the methane paradox in oxic waters.

Growth and CH 4 production
Over the course of the exponential growth phase headspace-CH 4 mixing ratios in treatments containing E. huxleyi were not measurably different from the control treatments.Therefore, it was not possible to determine CH 4 production in the exponential growth phase.However, we conclude that E. huxleyi produces CH 4 throughout all growth phases as will be detailed in the following.In the transitionary growth phase leading up to the stationary phase, we calculated incremental CH 4 production (daily).The transitionary phase features a declining growth rate and often increasing cellular carbon quotas (Langer et al., 2013).Cellular CH 4 quotas also increased (data not shown).On the other hand, CH 4 production remained constant within the measurements of error, displaying a slight downward trend when approaching stationary phase (Fig. 6).Therefore, we conclude that CH 4 production is not a feature of senescent cells only but is probably operational in all growth phases.This is interesting in the context of the ecology and biogeochemistry of E. huxleyi.Contrary to the traditional assumption that E. huxleyi production in the field is dominated by late summer bloom events, it was recently shown that non-bloom production in spring contributes significantly to yearly average production and therefore bloom events are not exceptionally important in biogeochemical terms (Schiebel et al., 2011).Since senescent cells in field samples are mainly a feature of late bloom stages, the exclusive production of CH 4 by such cells would confine any contribution of E. huxleyi to the oceanic CH 4 budget to a relatively short, and biogeochemically less important, period.However, from results found in this study we would propose that E. huxleyi produces CH 4 during all growth phases as part of its normal metabolism.If our findings are confirmed and supported by other research groups, this has considerable implications as it would render this species a prolific aerobic producer of CH 4 , on a par with, for example, terrestrial plants (Bruhn et al., 2012).

Methane emission rates
To calculate CH 4 emission rates of E. huxleyi, we normalized CH 4 production to cellular POC content (see Material and Methods).The CH 4 emissions were 0.7 µg POC g −1 d −1 , or 30 ng g −1 POC h −1 (mean for all treatments, n = 8).
In this study the main aim was (as a proof of principle) to unambiguously provide evidence that E. huxleyi are able to produce methane under aerobic conditions and without the help of microorganisms.
However, we suggest that CH 4 emission rates of E. huxleyi algae are different under changing environmental conditions, e.g., temperature, light intensity, or nutrient supply.The effect of changing environmental parameters should be the focus of future investigations.

Inorganic and organic precursors of CH 4
Based on the addition of bicarbonate ( 13 C-Bic, 1 % enrichment), which is the principal carbon source for the growth of algae, and the measurements of δ 13 CH 4 values it was possible to clearly identify bicarbonate as the principal carbon precursor of CH 4 in E. huxleyi.
In the flasks where algae were supplemented with 13 C-Bic, a significant increase in δ 13 CH 4 values occurred over the incubation period, which shows that algae use bicarbonate as precursor carbon (C) for CH 4 production.As expected, in the controls flasks algae where no 13 C-Bic was added and the control seawater + 13 C-Bic without algae, no change in δ 13 CH 4 values was observed.The initial δ 13 C value of the bicarbonate in the treatment algae + 13 C-Bic (+882 ‰) is within the range of the source δ 13 CH 4 values obtained via the Keeling-plot method (+812 ± 90 ‰).Even though there might be kinetic isotope fractionations involved in each of the several steps during organic matter formation, these data clearly indicate that bicarbonate is the principle inorganic carbon precursor of CH 4 produced in algae.
Bicarbonate is taken up by the algae via autotrophic C fixation (Burns and Beardall, 1987) and might therefore -during several steps of metabolism, i.e., the formation of organic compounds -lead to the formation of CH 4 .It will probably be used as an unspecific C source in many different metabolic pathways, e.g., the synthesis of lignin, pectin, and cellulose (Kanehisa et al., 2014) -components already known as CH 4 precursors from terrestrial plants, where CH 4 can be produced via methyl group cleavage (Keppler et al., 2008;Bruhn et al., 2009;Vigano et al., 2009).However, lignin and pectin are not commonly found in marine algae such as E. hux-leyi.For these organisms, sulfur-bonded methyl groups such as thioethers, sulfoxides, and sulfonium salts (methionine, S-adenosylmethionine (SAM), adenosylmethionine DMSP, DMSO, DMS) are of much more interest.For our experiments, we used 13 C positionally labeled Met where only the sulfur-bond methyl group (-S-CH 3 ) was 99 % enriched in 13 C.Our choice of this compound was partly due to its commercial availability but more importantly because it is known to be involved in a number of metabolic pathways and transmethylation reactions (Stefels, 2000;Bruhn et al., 2012).
In contrast to the ubiquitous C-source bicarbonate -which can also be used to build Met in algae (Stefels, 2000) -Met is incorporated in specific metabolic pathways.Algae use part of the Met for protein synthesis; in E. huxleyi it is also involved in the synthesis of DMSP, a main precursor of DMS and DMSO.
The clear increase in δ 13 CH 4 values of headspace-CH 4 in the treatment algae + 13 C-Met (Fig. 2e, f) shows that the methyl thiol group of Met is a direct CH 4 precursor.The Keeling-plot results (Fig. 3) show higher variability for Met than for Bic.However, Met is almost certainly not the only precursor of CH 4 , as the headspace-CH 4 mixing ratios increased (Fig. 2d), while the 13 C values of headspace-CH 4 showed a saturation curve (Fig. 2f).This indicates either a shift from Met to other CH 4 precursors or to the use of newly synthesized, non-labeled Met.Based on the initial amount and the total amount of 13 CH 4 formed at the end of the incubation, only a small fraction (79 pmol, i.e., 4.0 ‰) of the initial added 13 C-Met (1.8 µmol) was converted to 13 CH 4 .The formation of CH 4 from 13 C-Met explains roughly about 3 % of the total amount of CH 4 formed throughout the incubation period.Possibly, the formation of potential precursors of CH 4 may change under various climatic conditions, leading to varying CH 4 production rates in different pathways.
This observation is in line with the findings of Lenhart and colleagues, who demonstrated that the sulfur-bound methyl group of Met was a precursor of CH 4 in plants (Lenhart et al., 2015) and fungi (Lenhart et al., 2012).The linear increase in headspace-CH 4 mixing ratio (Fig. 2d) together with the nonlinear increase in δ 13 CH 4 signature (Fig. 1f) indicates that the pool of 13 C-Met was either exhausted or was diluted by newly synthesized, non-13 C-enriched Met.
In addition, we also found an indication for a chemical CH 4 formation pathway in the seawater with Met as methylgroup donor as a small increase in 13 CH 4 values in the control treatment seawater + 13 C-Met was observed (Fig. 2f).This CH 4 formation pathway is approximately 10-fold lower when compared to the treatment algae + 13 C-Met and is only observed in the isotopic experiment but not when only the CH 4 mixing ratio is considered (Fig. 2d).However, this observation is in line with some previous findings (Althoff et al., 2010(Althoff et al., , 2014)), who showed that the abiotic formation of CH 4 due to the degradation of methionine or ascorbic acid by light or oxidants such as iron minerals is possible.In the case of methionine, it was shown that the sulfur-bound methyl group of Met was the carbon precursor of CH 4 (Althoff et al., 2014).

Potential implications for the occurrence of CH 4 in oxic marine waters
Several hypotheses with regard to the occurrence of the seasonal and spatial CH 4 oversaturation in oxic surface waters (Bange et al., 1994;Forster et al., 2009;Owens et al., 1991) have been postulated.They include CH 4 formation from methanogenic archaea in anoxic microsites (Karl and Tilbrook, 1994) or CH 4 formation via the C-P-lyase pathway from methylphosphonate (Karl et al., 2008).
In the ocean, both CH 4 production by methanogens and consumption via methanotrophic bacteria occur simultaneously.Therefore, CH 4 production can exceed estimated CH 4 production rates when based solely on CH 4 mixing ratio measurements (Reeburgh, 2007).To provide a noteworthy contribution to oceanic CH 4 production, precursors must either be available in high abundance or be continually synthesized.Algae-derived methylated sulfur compounds such as Met, DMSP, DMS, and DMSO are ubiquitous in the ocean but show a high spatial and temporal variability with high mixing ratios in algal blooms.Therefore, they are potential compounds that might be involved in CH 4 formation in the oceans (Keppler et al., 2009;Althoff et al., 2014).The involvement of methyl moieties from methylated sulfur compounds in CH 4 biosynthesis might therefore play an important role in pelagic CH 4 production.Mixing ratios of DMS and DMSP in seawater during algal blooms were reported in the range of 0.82 to 8.3 nmol L −1 and 1.25 to 368 nmol −1 , respectively (Matrai and Keller, 1993).
The CH 4 emission rates of E. huxleyi may also occur by a second formation pathway, where DMSP is first converted to DMS and subsequently oxidized to DMSO (Bentley and Chasteen, 2004).
However, several studies have afforded evidence for a CH 4 formation pathway via methyl radicals (Althoff et al., 2014;Eberhardt and Colina, 1988;Herscu-Kluska et al., 2008), leading to the hypothesis that algae-derived DMSO can also act as a precursor of CH 4 in oxic seawater (Althoff et al., 2014).A correlation between Met and DMSP synthesis was provided by Gröne and Kirst (1992), who showed that the supplementation of Tetraselmis subcordiformis with 100 µg L −1 Met yielded a 2.6-fold increase in DMSP.For E. huxleyi, DMSO mixing ratios in the stationary growth phase can reach 0.1 pg per cell (Simo et al., 1998).Assuming that a similar DMSO mixing ratio were to be found in our study, this would mean that in every 4 × 10 3 DMSO molecules per day must be transferred to CH 4 to explain the observed increase in CH 4 .Moreover, a positive correlation was observed between chlorophyll a and CH 4 , as well as between DMSP or DMSO and CH 4 (Zindler et al., 2013).
www.biogeosciences.net/13/3163/2016/Biogeosciences, 13, 3163-3174, 2016 Our study provides the first isotope evidence that marine algae such as E. huxleyi produce CH 4 with bicarbonate and the sulfur-bound methyl group of Met as C precursors.
Our results based on real-time PCR and the enrichment of methanogenic archaea make it highly unlikely that there is a contribution of archaea to the observed CH 4 production.It is of interest to note that it is almost 40 years since algae were suggested as a possible direct source of CH 4 in the ocean (Scranton and Brewer, 1977;Scranton and Farrington, 1977).Thus, despite the scientific endeavors of numerous research groups over a considerable period of time the explanation for the frequently monitored CH 4 oversaturation of oxic surface waters in oceans and fresh water lakes is still a topic of debate (Zindler et al., 2013;Tang et al., 2014;Damm et al., 2008).Since our results unambiguously show that the common coccolithophore E. huxleyi is able to produce CH 4 per se under oxic conditions, we thus suggest that algae living in marine environments might contribute to the regional and temporal oversaturation of surface waters.However, our results of the laboratory experiments should be confirmed by field measurements in the ocean.We would encourage further studies in this research area to make use of stable isotope techniques together with field measurements as we consider such an approach well suited to the elucidation of the pathways involved in CH 4 formation in oceanic waters.

Figure 1 .
Figure 1.Experimental setup: the potential precursors of CH 4 , 13 Clabeled bicarbonate ( 13 C-Bic) or a position-specific 13 C-labeled methionine ( 13 C-Met) were added to the flasks containing either a culture of E. huxleyi or seawater only.

Figure 3 .
Figure 3. Keeling plots for the treatment (a) algae + 13 C-Bic and (b) algae + 13 C-Met, where f (0) refers to the 13 C value of the CH 4 source.

Figure 4 .
Figure 4. Correlation between cell density per flask and CH 4 content (sum of headspace and water phase) for the coccolithophore E. huxleyi (a) in seawater only (n = 2; light green) and supplemented with 13 C-labeled bicarbonate (Bic; dark green) or (b) methionine (Met) (n = 3); error bars mark the standard deviation; d is day of incubation.

Table 2 .Figure 5 .
Figure 5. Mean CH 4 content (sum of headspace and water phase) in the flasks of E. huxleyi supplemented with either bicarbonate or methionine (n = 3) or the respective control without algae (n = 2) measured at days 8, 9 and 10; error bars show the standard deviation.

Table 1 .
Overview of sample collection during the incubation of E. huxleyi.