Seasonal dynamics of the microbial methane filter in the water column of a eutrophic coastal basin

Abstract In coastal waters, methane-oxidizing bacteria (MOB) can form a methane biofilter and mitigate methane emissions. The metabolism of these MOBs is versatile, and the resilience to changing oxygen concentrations is potentially high. It is still unclear how seasonal changes in oxygen availability and water column chemistry affect the functioning of the methane biofilter and MOB community composition. Here, we determined water column methane and oxygen depth profiles, the methanotrophic community structure, methane oxidation potential, and water–air methane fluxes of a eutrophic marine basin during summer stratification and in the mixed water in spring and autumn. In spring, the MOB diversity and relative abundance were low. Yet, MOB formed a methane biofilter with up to 9% relative abundance and vertical niche partitioning during summer stratification. The vertical distribution and potential methane oxidation of MOB did not follow the upward shift of the oxycline during summer, and water–air fluxes remained below 0.6 mmol m−2 d−1. Together, this suggests active methane removal by MOB in the anoxic water. Surprisingly, with a weaker stratification, and therefore potentially increased oxygen supply, methane oxidation rates decreased, and water–air methane fluxes increased. Thus, despite the potential resilience of the MOB community, seasonal water column dynamics significantly influence methane removal.


Introduction
Coastal ecosystems are the major contributor to global ocean methane emissions, despite only covering about 15% of ocean surface area (6-12 Tg CH 4 yr −1 ) (Borges et al. 2016, Weber et al. 2019 ).Due to the presence of methane-oxidizing microorganisms in the water column and in the sediment, these methane emissions are onl y a fr action of the methane pr oduced in the sediment.The efficiency of this so-called methane biofilter is one of the greatest uncertainties in methane-emission predictions (Dean et al. 2018 ).In coastal basins, a consortium of anaerobic methane-oxidizing arc haea with sulfate-r educing bacteria builds the methane biofilter in the sediment (Wallenius et al. 2021 ), and aerobic methaneoxidizing bacteria (MOB) dominate the methane biofilter in the water column (Reeburgh 2007, Steinle et al. 2017, Steinsdóttir et al. 2022a, Venetz et al. 2023 ).Deoxygenation and eutrophication can lead to increased methane production in the sediment that can exceed the methane oxidation ca pacity, whic h can r esult in high benthic methane fluxes into the water column (Egger et al. 2016, Zhang et al. 2020, Lenstra et al. 2023, Żygadłowska et al. 2023a ).As coastal ecosystems ar e especiall y affected by eutrophication and hypoxia (Diaz andRosenber g 2008 , Br eitbur g et al. 2018 ), benthic methane fluxes ar e likel y to incr ease in the futur e (Dean et al. 2018 ).A better understanding of the fate of methane in the water column is ther efor e crucial for better predictions of methane emissions to the atmosphere and k e y for adequate policymaking.
Coastal basins are highly influenced by seasonal water column dynamics, which affects the water column chemistry and ultimately the microbial community structure and activity (Gilbert et al. 2012, Wang et al. 2020 ).For instance, increasing surface water temper atur es in summer can lead to a density stratification, which lo w ers the oxygen supply to the bottom and results in an oxycline and anoxic bottom water.Multiple potential metabolic pathways within the ambient methanotrophic commu-nity can ensure the functionality of the methane biofilter along this v ertical r ange of oxygen (and methane) concentr ations (Hernandez et al. 2015, Venetz et al. 2023 ).Such niche partitioning over a range of oxygen concentrations likely aids the overall resilience of the methane biofilter to w ar d r egular c hanges in oxygen availability in the water column (Venetz et al. 2023 ).For instance, thanks to the r a pid succession of the MOB community, the abundance of methanotrophs and methane oxidation capacity during autumn turnov er r emained high and could mitigate methane emissions in a freshwater lake (Mayr et al. 2020 ).
Another study observed increased methane oxidation accompanied by an increase of MOB abundance from spring to summer (Gründger et al. 2021 ).T hus , the succession of the MOB community from a mixed water column in spring to summer stratification with an oxygen gradient as well as the succession upon water column mixing is crucial for the functioning of the methane biofilter.To reduce the uncertainties in methane-emission predictions, it is thus important to understand how the methanotr ophic comm unity structur e and potential r emov al activity at differ ent depths r espond to the seasonal changes in methane, o xygen, and n utrient availability and how this ultimately affects methane release to the atmosphere (Louca et al. 2016, Grossart et al. 2020 ).
Here, we studied the seasonal dynamics of the microbial methane biofilter in the water column of a marine coastal basin in the southwest Netherlands.In nine sampling campaigns between March and October 2021, we investigated the vertical distribution of methane and oxygen, the potential methane oxidation rates, the microbial community structure, and the methane fluxes to the atmosphere.

F ield w ork loca tion and sampling methodologies
Lake Gr e v elingen is a highl y eutr ophic former estuary with a total surface area of 115 km 2 and an average water depth of 5.1 m.The water column of the main channel is stratified during the summer months with hypoxic or anoxic bottom waters (Wetsteijn 2011, Ha gens et al. 2015 ).A mor e detailed description of the system and the study site can be found else wher e (Egger et al. 2016, Sulu-Gambari et al. 2017 ).At the deepest point of Lake Gr e v elingen, the Sc har endijke basin (51.742 • N; 3.849 • E, 45 mbs), the high sedimentation rates and anoxic bottom water lead to high benthic methane fluxes to the water column during summer stratification (0.6-2.7 mmol m −2 d −1 ) (Egger et al. 2016(Egger et al. , Żygadłowska et al. 2023a ).To monitor seasonal water column methane dynamics in the Sc har endijke basin, we conducted nine r esearc h cruises with the RV Navicula betw een Mar ch and October 2021.During these campaigns, we measured in situ water-air methane fluxes, constructed depth profiles of water column chemistry, and took samples for microbial analysis, including incubation experiments to measure potential aerobic methane oxidation rates.
The extent of water column stratification during each sampling campaign was determined with a CTD unit (SBE 911 Plus, Sea-Bird Electr onics, Belle vue, WA, USA).In addition, the oxygen distribution was sim ultaneousl y r ecorded by a seabird sensor (Seabird SBE43).Because of the limit of detection commonly observed in oxygen sensor data of oxygen-depleted w aters, w e here consider hypoxia at concentrations < 63 μmol l −1 and anoxia once concentr ations ar e < 3 μmol l −1 and do not further decr ease with depth.
Water samples were taken at 30 depths with a 10-l Niskin bottle.Subsequentl y, unfilter ed w ater w as collected in 1-l sterile plastic bottles for DNA analysis, and 0.5-l sterile Schott bottles for incubation experiments, which were stored in the dark at 4 • C until further pr ocessing.Furthermor e, 120-ml bor osilicate serum bottles were filled for the determination of the methane concentration.To avoid air contamination, the bottles were filled from the bottom via gas-tight tubing while letting the w ater overflo w three times, after which they were crimp-capped with an aluminum cap and a butyl stopper.To stop microbial activity, 0.25 ml of HgCl 2 (sat.) was added.Samples were stored upside down at room temper atur e until further processing.

Methane concentr a tion measurements
To determine methane concentrations in the water column, 5 ml of N 2 gas was added to all borosilicate bottles, while simultaneousl y r emoving the same volume of the water.After equilibrating for at least 2 h, methane concentrations were measured with a Thermo Finnigan Trace ™ gas chromatograph equipped with a flame ionization detector (detection limit: 0.02 μmol l −1 ).

DN A extr action, 16S rRN A gene sequencing, and da ta anal ysis
Water samples were filtered in the lab, within 24 h after sampling on Supor ® PES 0.22-μm filters with a vacuum pump setup.After immediate freezing at −80 • C, the samples were stored at −20 • C until extraction.The DN A w as extracted with the FastDNA™ SPIN Kit for Soil DNA Isolation Kit (MP Biomedicals) according to the protocol.
The V3-V4 region of the 16S rRNA gene was sequenced by Macrogen with an Illumina MiSeq platform (Macrogen, Amsterdam, The Netherlands) to analyze the micr obial comm unity composition.The primer pairs Bac341F (CCTA CGGGNGGCWGCA G) (Herlemann et al. 2011 ) and Bac806R (GGA CTA CHV GGGTWTC-T AA T) (Ca por aso et al. 2012 ) were used for bacteria, and the primer pairs Arch349F (GYGC ASC AGKCGMGAAW) (Ken et al. 2000 ) and Arch806R (GGA CTA CVSGGGTATCTAAT) (Ken et al. 2000 ) were used for archaeal 16S sequencing.Sequencing data were processed with RStudio.Primers wer e r emov ed with cutadapt (Martin et al. 2012 ) with the options -g, -G, and discar d-untrimmed.Lo wquality reads ( < Q20 forw ar d and < Q30 reverse) were removed by truncating reads to a length of nt 270 forw ar d and nt 240 reverse with the D AD A2 pipeline (Callahan et al. 2017 ).Finally, amplicon sequence variants (ASVs) were inferred, forw ar d and rev erse r eads wer e mer ged, and c himaer as r emov ed.For taxonomic assignment, the 254 Silva non-redundant train set v138 ( https: // zenodo.org/record/ 3731176#.XoV8D4gzZaQ) was used.For further clustering and calculation of r elativ e abundances, the Phyloseq pac ka ge w as used, and data w ere visualized with ggplot2, and gr a phs wer e adjusted with Adobe Illustr ator.Raw r eads of the 16S amplicon sequencing data can be accessed on the National Center for Biotechnology Information (NCBI) website under the accession number PRJNA1053269.

Potential methane oxidation rates
To investigate the seasonal dynamics of aerobic methane removal potential by the methanotrophic methane filter in the water column, w e incubated w ater samples from 14 depths during each sampling campaign within 24 h after sample r etrie v al.For eac h depth, 100 ml of unfilter ed, air-equilibr ated sample, was put into an autoclaved 120-ml borosilicate bottle (in triplicate) and closed with bromobutyl stoppers, and crimped with an aluminum cap.
To each incubation, 1 ml of 13 C-CH 4 (99%) was added, which resulted in a partial pr essur e of 5% in the headspace.All bottles were incubated under constant shaking (150 r/m) in the dark, at room temper atur e for the duration of the incubation.
For each time point, 3-ml liquid sample was taken from the incubation bottle and replaced with 1 ml of air.As the carbonate balance at the pH range of our sample is susceptible to small changes in pH, we acidified subsamples to a pH of < 2. The liquid sample was tr ansferr ed into a gas-tight air-equilibrated 3-ml vial (Labco, exetainer, UK) containing 50 μl of 0.1 M HCl.The produced 13 C-CO 2 was measur ed dir ectl y fr om the exetainer headspace with a gas c hr omatogr a phy-mass spectr ometer (GC −MS) (Agilent 5975C inert MSD).Liquid 13 CO 2 concentr ations wer e calculated with the Henry coefficient ( Supplementary data ).The linear increase in 13 C-CO 2 after the lag phase was used to determine methane oxidation rates.

In situ water-air flux measurements
In situ water-air fluxes of methane were measured during each sampling campaign.Fluxes were determined using a transparent, cylindrical floating chamber (ø: 390 mm, height 270 mm, Techno-Centrum, Radboud University, Nijmegen, The Netherlands) connected in a closed loop to a LICOR trace gas analyzer (LI-7810, LI-COR En vironmental-UK Ltd., Cambridge , UK).T he chamber frame was stabilized to withstand wave turbulence using a bespoke raft ( Supplementary Fig. S1 ), as the surface water of the Sc har endijke basin can be quite turbulent.To ensure a closed loop between the chamber and the LICOR gas analyzer, the input and output connectors of the LICOR were connected to the top of the floating chamber using gas-tight polyurethane tubes [ø: 4 mm (inside), F esto, 5 m].T he chamber was gently placed on the water surface, and the accumulation of methane was measured with the trace gas analyzer for at least 3 min in triplicate .T he chamber was aerated until atmospheric methane concentrations were r eac hed, befor e starting each new measurement.
Methane fluxes were then calculated with the following equation: where CH 4 / t is the linear increase of the concentration of methane (mmol m −3 ) in the chamber over time ( t ), V is the volume of the chamber (m 3 ), and A is the area of the chamber (m 2 ).
The measured partial pressures (ppb) in the chamber were converted to methane (mmol m −3 ) using the ideal gas law and the ambient air temper atur e during each deployment.

Sta tistical anal ysis
To assess the seasonal dynamics of the bacterial community, Shannon diversity and Chao1 richness were calculated to illustrate the bacterial alpha diversity of untransformed ASV counts of eac h sample.Befor e anal ysis, ASVs that were assigned to archaea, mitoc hondria, and c hlor oplasts wer e r emov ed fr om the dataset.The dissimilarity of the bacterial community between each sample was calculated as the Bray-Curtis distance of r ar efied ASV counts, and distance matrices wer e illustr ated by nonmetric multidimensional scaling (NMDS).Based on the stress factor assessment of the ordination with different dimensions, the ordination was performed on two dimensions (stress = 0.092, nonmetric fit R 2 = 0.992, and linear fit R 2 = 0.965).Vectors for the environmental variables O 2 , CH 4 , H 2 S, NO 2 -, NO 3 -, Fe tot , and depth were determined with the env_fit() function of the vegan pac ka ge.

Results and discussion
In this study, we monitored the seasonal dynamics of the microbial methane filter in the water column of a marine basin during nine sampling campaigns in 2021.In the following sections, we will show and discuss our results for the seasonal succession and distribution of MOB, methane oxidation potential along the depth gradient, the potential for anoxic methane r emov al, and the water-to-air fluxes of methane in marine Lake Gr e v elingen in 2021.We show that there is (i) a vertical distribution and seasonal succession of the v ersatile methanotr ophic comm unity, (ii) that this is related to water column dynamics and the availability of oxygen and nutrients, and (iii) that this is ultimately related to methane removal and in situ methane fluxes to the atmosphere.

Seasonal succession and vertical distribution of the methanotrophic community
The methanotr ophic comm unity w as dominated b y methaneoxidizing Gamma pr oteobacteria ( γ -MOB) belonging to Methylomonadaceae .Although alpha pr oteobacteria ( α-MOB) wer e pr esent, their r elativ e abundance ne v er exceeded 0.07% of the bacterial 16S rRNA reads .T his could be due to the eutrophic state of the basin, as α-MOB are more adapted to oligotrophic conditions and could be continuously outcompeted b y Meth ylomonadaceae (Ho et al. 2013, Kaupper et al. 2020 ).The main methanotrophic genera were Milano-WF1B-03 ( M1B ), Ma-rine_Meth ylotrophic_Group2 ( MMG2 ), and Meth yloprofundus ( MP ).These ( γ -MOB) together built a methanotrophic community, i.e. potentiall y v ersatile in terms of oxygen metabolism, denitrification capacity, and potential for sulfur transformation (Venetz et al. 2023 ).Despite the low div ersity of the methanotr ophic community, our seasonal data demonstrated a high adaptation potential enabling both v ertical nic he partitioning and seasonal succession.While the r elativ e abundance of MOB was < 0.3% of the bacterial 16S rRNA reads in the mixed, oxygenated water column in March (Fig. 1 A), it increased up to 9% at 39 m depth by July when the water column was stratified (Fig. 2 A).This drastic increase can be attributed to the summer str atification, whic h resulted in the formation of chemical niches along the methaneoxygen counter gr adient (Amar al and Knowles 1995 , Mayr et al. 2020 ) and was accompanied by a shift in the MOB community structure with depth.In the oxygenated water, MOB r elativ e abundance is low and appears to consist mostly of M1B (Fig. 2 A, Supplementary Fig. S2 ).In the anoxic water, MOB r elativ e abundance is up to 9% and consists of mostly MMG2 and MP (Fig. 2 A).A similar niche partitioning was observed at the end of summer stratification in September 2020 (Venetz et al. 2023 ).T here , the metabolic potential of these genera indicated metabolic adaptations to oxygen limitation with high-affinity oxidases and thr ough potential nitr ate, ir on, and sulfur r eduction, whic h could explain the shift in community composition along the oxycline.Similarl y, these metabolic ada ptations ar e also important for seasonal succession.Our seasonal data sho w ed that despite having low diversity, the methanotrophic community could adapt to both water column mixing and summer stratification.
Ho w e v er, eutr ophication and stratification in coastal ecosystems will likely intensify in the future (Dominovi ć et al. 2023 ), which may induce more pronounced shifts in the methanotrophic community during summer stratification and further decrease di versity.The di versity of the entire water column bacterial community indeed shows a drastic decrease in alpha diversity from March to October 2021 (Fig. 2 C, Supplementary Fig. S3 ).Prolonged anoxia and warming could ar guabl y pr omote e v en slow-gr owing anaer obic methanotr ophs, suc h as anaer obic methanotr ophic archaea or NC10 bacteria (Su et al. 2023 ).Yet, the pool of other methanotr ophic micr oor ganisms than Methylomonadaceae in marine Lake Gr e v elingen is v ery low.T hus , the low diversity of methanotrophic microorganisms can impair the resilience to w ar d changes in oxygen availability that exceed the amplitude of the current seasonal dynamics.

Methane remo v al b y methanotrophic bacteria
Potential aerobic methane oxidation rate measurements (up to 0.60 μmol l −1 h −1 ) indicate that benthic methane is (partially) oxidized in the water column before reaching the atmosphere (Fig. 1 C).These measurements confirmed that there is potential for methane r emov al thr oughout the year.Mor eov er, methane oxidation rates follo w ed a seasonal pattern and varied with depth along the water column.Methane oxidation in the mixed water column in March was low ( < 0.002 μmol l −1 h −1 ) compared to May (0.008-3.6 μmol l −1 h −1 ) upon the onset of summer water column stratification and the development of an oxycline.During the stratification period, methane oxidation rates follo w ed a vertical pattern.In J une , a clear methane-oxygen counter gradi-ent had formed with an oxycline between 35 and 36 m and high methane concentrations in the bottom water (up to 17 μmol l −1 ).
The high methane oxidation rates below 9 m (0.8-2.9 μmol l −1 h −1 ), indicated active aerobic methane removal.Ho w ever, canonical aerobic methane oxidation alone cannot explain the observed biogeoc hemical and micr obiological pattern, whic h is especiall y pronounced in July, August, and September.Between June and July, an additional thermocline formed in between 12 and 16 m (Fig. 1 B).This induced the formation of the oxycline higher up and resulted in a vertical decoupling of the methane and oxygen gradients and the formation of an oxygen-and sulfide-free (suboxic) zone ( Żygadłowska et al. 2023b ).Inter estingl y, the potential aerobic methane oxidation rates and the r elativ e abundance of MOB were highest far below the oxycline (Figs 1 C and 2 A).If this de-coupling had induced a diminishing of the methane biofilter, w e w ould hav e expected a decr eased r elativ e abundance of MOB, lo w er methane oxidation rates, and a higher methane flux to the atmosphere.Ho w ever, w e did not observe any of the mentioned changes .T his points to a potentiall y activ e methane biofilter e v en under anoxic conditions.Notabl y, potential methane oxidation rates and relative MOB abundance were consistently highest below the oxycline, and both dr asticall y decr eased by the onset of water column mixing and reoxygenation in autumn.

MOB responsible for anaerobic methane oxidation
The high r elativ e abundance of putativ el y aer obic MOB in the anoxic water column and their potential activity is an emerging paradox in a variety of aquatic ecosystems.Anaerobic methane r emov al is commonly attributed to slo w-gro wing methanotrophic archaea in the sediment, where total microbial biomass is high and substr ate r esidence time can be long.Arc haea can also be involved in anaerobic methane removal in the water column.For example, in the Black Sea, methanotrophic archaea are responsible for anaerobic methane removal in the suboxic zone (Schubert et al. 2006 ).The r elativ e abundance of methanotr ophic arc haea in the water column of marine Lake Gr e v elingen was highest in July with 2% (of archaeal 16S rRNA reads) at 23 m and 1.8% at 40 m and mainly belonged to the family of Methanoperedenaceae .
Ho w e v er, their depth distribution did not follow any coherent pattern, and based on meta genomic anal ysis, in September 2020, the total abundance of archaea in the water column was 50-100 × lo w er than the abundance of bacteria (based on phyloflash analysis of the metagenome, data not shown).Moreover, the slowgro wing ar chaea might struggle to establish a population in shallow ecosystems where the water column is fully mixed in spring and autumn and changes in water column c hemistry ar e fast (Su et al. 2023 ).Ther efor e, we estimate the ov er all contribution of methanotrophic archaea to the anaerobic removal of methane in the water column to be low.It is more plausible that the domi-nant methanotrophic bacteria use alternative electron acceptors during oxygen limitation (Steinsdóttir et al. 2022a ).The methanotr ophic comm unity in the suboxic zone in July consisted of MP and MMG2 .In the euxinic zone below that, the methanotrophic community was dominated by MP alone .T he shift in the MOB community to w ar d MP and MMG2 in the anoxic zone may be explained by their versatile metabolism and adaptation potential.Methanotr ophic meta genome-assembled genomes (MAGs) pr e viousl y r etrie v ed fr om Lake Gr e v elingen implied the genomic signature of low-oxygen adaptation; an MP MAG contained the highaffinity bd oxidase, which indicates the ability to scavenge oxygen at very low concentrations, and in a MAG attributed to MMG2, the potential use of nitrate or metal oxides as alternative electron acceptors was described (Venetz et al. 2023 ).Recent studies further indicate the high potential for anaerobic methane r emov al by MOB in the water column of marine ecosystems (Thamdrup et al. 2019, Steinsdóttir et al. 2022b ), and laboratory incubations indicate the possibility of methane oxidation with iron oxides by methanotrophic bacteria (Zheng et al. 2020, Li et al. 2021 ).Suboxic zones harbor electron acceptors other than oxygen, such as metal oxides or nitrate (Murray et al. 1999 ).Although cryptic cycling could ar guabl y obscur e the suppl y of oxygen and alternativ e electr on acceptors, nitr ate is depleted in the suboxic zone (except for August/September), but iron oxides are potentially available ( Żygadłowska et al. 2023b ).Furthermor e, r ecent studies suggest that external electron transfer to or with dissolved organic matter (DOM) might increase the methane oxidation capacity in wetland and br ac kish sediments and e v en in the water column of humic bog lakes (Valenzuela et al. 2019, Olmsted et al. 2023, Pelsma et al. 2023 ).Considering the high sedimentation in the Sc har endijke basin (Egger et al. 2016 ) and eutrophic state of the lake, external electr on tr ansfer linked to DOM and metal oxides might be an additional mechanism supporting anaerobic methane oxidation, especially in early summer (J uly).T herefore , methane oxidation with alternative electron acceptors would potentially enable the removal of methane in anoxic water.

Seasonal dynamics of water-air methane fluxes
Calculations based on benthic methane flux measurements and the distribution of methane in the water column show that ebullition is a major contributor to the methane flux all year long ( Żygadłowska et al. 2023b ).These bubbles can bypass the microbial methane filter dir ectl y, or dissolv e while tr av eling upw ar d through the water and can considerably lo w er the methane filtering efficiency throughout the year ( Żygadłowska et al. 2023b ).
In contrast, the in situ diffusive water-air fluxes of methane follo w ed a clear seasonal pattern (Fig. 1 A).While the fluxes increased fr om Marc h to April, they wer e low in J une , when the water column was stratified and methane concentrations had increased in the bottom waters.Together with the high r elativ e abundance of MOB in the water and high potential methane oxidation rates (Figs 1 C and 2 A), this suggests that the activ e micr obial methane filter during summer stratification mitigated most of the diffusive methane emissions to the atmosphere.Ho w ever, in September, the diffusive fluxes of methane were higher despite the relative abundance of methanotrophs.While the methane water-air fluxes w ere lo w er during summer stratification (0.12-0.22 mmol m −2 d −1 ), the fluxes started to increase to w ar d the end of the stratification period (Fig. 1 A).Although the r elativ e methanotr ophic abundance in the anoxic water column was high at the end of August (up to 10%), methane still bypassed the MOB filter, and methane fluxes to the atmosphere increased at the end of August (0.96 mmol m −2 d −1 ) and in September (1.15 mmol m −2 d −1 ).While in the preceding months, the methane concentrations right below the oxycline ne v er exceeded 0.5 μmol l −1 , methane concentrations in the oxycline at 50% oxygen saturation were 1.6 μmol l −1 in September ( Supplementary data "water_column_chemistry"). Inter estingl y, the methanotr ophic comm unity did not decr ease significantl y, whic h indicates that the seasonal dynamics in water column chemistry affected methane oxidation rates more than abundance or comm unity structur e (Kaupper et al. 2020 ).We suggest a combination of reasons that contributed to increased methane fluxes to the atmosphere from the end of August onw ar d, despite the high r elativ e abundance of methanotrophs: (i) the weakening of stratification increased the turbulent flux, which resulted in less time for microbial oxidation, (ii) nitrite accumulation inhibited methanotrophic activity and linked to that (iii) a shift in microbial community structure induced a change in the interactions between the methanotrophic bacteria and other members of the micr obial comm unity.The weakening of the water column methane biofilter coincided with the weakening of summer stratification.An intrusion of oxygen through a lateral influx of oxygenated water (Hagens et al. 2015, van Haren 2019 ) and a weakening of the stratification due to warming could increase the do wnw ar d flux of oxygen and enhance the upw ar d methane flux ( Żygadłowska et al. 2023b ).The high r elativ e abundance of M1B in the anoxic water might be the result of oxygen intrusion prior to sampling at the end of August.The MAG associated with M1B , r etrie v ed fr om this basin in 2020, did not r e v eal an y genomic indication for the oxidation of methane during oxygen limitation: neither high-affinity oxidases were found, nor key genes for the utilization of alternative electron acceptors (Venetz et al. 2023 ).
Ther efor e, we suggest that cryptic oxygen intrusions between July and August could have provoked a shift away from MOB adapted to oxygen limitation.This may have led to a temporary weakening of the methane oxidation filter once the introduced oxygen was depleted.In the anoxic water, MP and MMG2 were similarly abundant in August, while MP dominated in J uly. T he MAGs associated with MMG2 and MP , r etrie v ed fr om this basin in 2020, sho w ed different adaptations to oxygen limitation.While the MAG associated with MP harbored a high-affinity bd oxidase, the MAG associated with MMG2 sho w ed almost full potential for denitrification but onl y harbor ed the low-affinity oxidase (Venetz et al. 2023 ).This shift further indicates that intrusions of oxygen-rich water could have affected the community in the anoxic water as well.In September, the methane oxidation r ates wer e m uc h lo w er compared to the previous month.Notably, at the end of August, nitrite accumulated in the suboxic zone ( Żygadłowska et al. 2023b ), which could be caused by oxygen inhibition of nitrite oxidation (Sun et al. 2021 ).The NMDS plot based on the Bray-Curtis distance shows that while oxygen and methane were most predictive of the bacterial community composition ( R 2 = 0.86), nitrite concentrations correlated with a shift in microbial community structure (Fig. 2 B; Supplementary data "NMDS_output").It is known that nitrite can inhibit methane oxidation, especially in communities not adapted to nitrite, and in some cases, this inhibition appears to be even irreversible (King andSylvia 1994 , Dunfield andKnowles 1995 ).The r elativ e abundance of nitrifiers was very low in the micr obial comm unity, and nitrite concentr ations ne v er exceeded 2 μmol l −1 in the proceeding months below the oxycline ( Supplementary data "water_column_c hemistry").Ther efor e, the methanotr ophic comm unity was likel y poorl y ada pted to nitrite, and nitrite accumulation might hav e irr e v ersibl y inhibited methanotrophic activity and resulted in low methane oxidation rates in September (Fig. 1 B).This could have been accompanied by a cascading effect within the plankton community due to the above-mentioned oxygenation event.For example, it has been shown that methane oxidation can be positiv el y or negativ el y influenced by the total microbial community (Gilbert et al. 2012 ), through nutrient competition and cross-feeding, and either the production or removal of toxic compounds by heterotrophic bacteria (Ho et al. 2014, Krause et al. 2017, Veraart et al. 2018 ).Therefor e, factors accompan ying the oxygenation e v ent could hav e inhibited methane oxidation activity of MP and MMG2 in the anoxic water.All findings together suggest that a weaker stratification and increased oxygen supply together with potential nitrite accumulation and associated shifts in the microbial community might hav e r esulted in the malfunctioning of the micr obial methane filter and resulted in even higher diffusive methane fluxes to the atmosphere.

Conclusion
We conclude that the efficiency of the microbial methane filter in the water column of a shallow marine basin is str ongl y influenced by seasonal water column dynamics .T he microbial community consists of potentially metabolic versatile Methylomonadaceae and counteracts high benthic methane fluxes during summer stratification.This is strikingly reflected in high water column methane oxidation potential and lo w er in situ diffusive water-air methane fluxes.Inter estingl y, a high r elativ e abundance of MOB and a high potential for aerobic methane oxidation were even found in the anoxic water.We suggest that methanotrophic bacteria dominate anoxic methane r emov al in the water column due to the availability of iron oxides and/or external electron transfer in the suboxic zone.Ho w e v er, methane incr easingl y bypasses the biofilter as stratification weakens and the contribution of the ebullitive flux from the sediment directly to the atmosphere is high.Hence, notwithstanding the large capacity of the microbial methane filter to oxidize methane under varying redox conditions, water column mixing can decrease the methane filtering efficiency: either dir ectl y by higher v ertical turbulent flux of dissolved methane or by altering other potential drivers for methane o xidation, such as n utrient limitation, c hanges in micr obial comm unity structur e , or inhibition.T his contradicts the assumption that mixing e v ents incr ease methane oxidation by r e-oxygenating the system.Our study demonstrates that the oxygenation state of the water column is not the ultimate factor that determines the functioning of the microbial methane filter.To accurately predict methane emissions from seasonally dynamic eutrophic coastal basins, a more complex network of drivers should be considered.

Figure 1 .
Figure 1.(A) Diffusive methane water-air fluxes measured with an in situ floating chamber.Boxes indicate the first and third quartiles, lines indicate the median, and whiskers indicate outer data points if < 1.5 interquartile range from quartiles.(B) Depth profiles of oxygen (blue lines) and temper atur e (r ed lines), and (C) methane concentr ations (blac k circles) together with potential aer obic methane oxidation r ates (MOx) (or ange circles) determined by incubation experiments between March and October 2021 in the Grevelingen Scharendijke basin.Error bars of MOx rates show the standard deviation between the biological replicates .T he hypoxic zone ( < 63 μmol l −1 ; Br eitbur g et al. 2018 ) is shaded light and the anoxic zone ( < 3 μmol l −1 ) is shaded dark.

Figure 2 .
Figure 2. (A) Relative abundance of methanotrophic bacteria retrieved through 16S rRNA sequencing.Light shaded areas indicate the hypoxic zone ( < 63 μmol l −1 h −1 ), and dark shaded areas indicate the anoxic zone ( < 3 μmol l −1 ).(B) Beta Diversity of all samples was calculated as Bray-Curtis distance and ordinated via two-dimensional NMDS.Short arrows indicate weak correlations and long arrows indicate strong correlations .T he R 2 values can be found in the Supplementary data ("NMDS_output").(C) Alpha diversity measures Chao1 (richness) and Shannon (richness and evenness) of the total water column bacterial community of each month.