A carbonate corrosion experiment at a marine methane seep: The role of aerobic methanotrophic bacteria

Methane seeps are typified by the formation of authigenic carbonates, many of which exhibit corrosion surfaces and secondary porosity believed to be caused by microbial carbonate dissolution. Aerobic methane oxidation and sulfur oxidation are two processes capable of inducing carbonate corrosion at methane seeps. Although the potential of aerobic methanotrophy to dissolve carbonate was confirmed in laboratory experiments, this process has not been studied in the environment to date. Here, we report on a carbonate corrosion experiment carried out in the REGAB Pockmark, Gabon‐Congo‐Angola passive margin, in which marble cubes were deployed for 2.5 years at two sites (CAB‐B and CAB‐C) with apparent active methane seepage and one site (CAB‐D) without methane seepage. Marble cubes exposed to active seepage (experiment CAB‐C) were found to be affected by a new type of microbioerosion. Based on 16S rRNA gene analysis, the biofilms adhering to the bioeroded marble mostly consisted of aerobic methanotrophic bacteria, predominantly belonging to the uncultured Hyd24‐01 clade. The presence of abundant 13C‐depleted lipid biomarkers including fatty acids (n‐C16:1ω8c, n‐C18:1ω8c, n‐C16:1ω5t), various 4‐mono‐ and 4,4‐dimethyl sterols, and diplopterol agrees with the dominance of aerobic methanotrophs in the CAB‐C biofilms. Among the lipids of aerobic methanotrophs, the uncommon 4α‐methylcholest‐8(14)‐en‐3β,25‐diol is interpreted to be a specific biomarker for the Hyd24‐01 clade. The combination of textural, genetic, and organic geochemical evidence suggests that aerobic methanotrophs are the main drivers of carbonate dissolution observed in the CAB‐C experiment at the REGAB pockmark.


| INTRODUC TI ON
The dynamics of methane seeps have been studied in great detail since this carbon pool may play an important role in global climate change (Kennett et al., 2003) and potentially as energy source (Sloan, 2003).
Methane seeps occur along passive and active continental margins, associated with geological features, as for example, gas hydrates, pockmarks, and mud volcanoes (Suess, 2014). It has been shown that many seeps are related to gas hydrate dynamics (Elvert et al., 1999;Suess et al., 1999). The stability of gas hydrate depends on the conditions at the seafloor and in the subsurface, such as temperature, pressure, and the availability of water and gas, predominantly methane (Bohrmann & Torres, 2006). Whether gas hydrate stability and dissociation is controlled by climate change, and how relevant it is for the overall methane seepage dynamics (Kennett et al., 2003) and methane export to the water column and atmosphere, is still a matter of debate (Ruppel & Kessler, 2016). In any case, methane seeps are active over long time periods and are further characterized by the formation of authigenic methane-derived carbonate (e.g., Bohrmann et al., 1998;Greinert et al., 2013;Peckmann et al., 2001), commonly precipitating in anoxic sediments affected by seepage. Seep carbonate is the result of an increase in alkalinity induced by the anaerobic oxidation of methane (AOM; Peckmann & Thiel, 2004), a process mediated by a consortium of methane-oxidizing archaea and sulfate-reducing bacteria (Boetius et al., 2000). Seep carbonates have a global distribution, including numerous modern and ancient deposits (Campbell, 2006 for a review). Although AOM is favoring carbonate precipitation, many seep carbonates exhibit corrosion surfaces and secondary porosity, likely caused by microbially induced carbonate dissolution (e.g., Campbell et al., 2002;Crémière et al., 2012;Himmler et al., 2011;Matsumoto, 1990;Natalicchio et al., 2015).
Carbonate dissolution in the marine realm has received much attention because it is a key process in global carbon cycling (Naviaux et al., 2019;Subhas et al., 2017). However, the role of microbes in carbonate dissolution has received little attention to date (Krause et al., 2014). The two biological processes that have been put forward to induce carbonate corrosion at seeps are aerobic methane oxidation (Cai et al., 2006;Crémière et al., 2012;Himmler et al., 2011;Matsumoto, 1990;Natalicchio et al., 2015) and hydrogen sulfide oxidation (Crémière et al., 2012;Leprich et al., 2021). Both processes can locally lower pH, resulting in carbonate dissolution (Crémière et al., 2012). In particular, aerobic methane oxidation leads to the production of carbon dioxide, with a subsequent increase of pCO 2 , favoring the dissolution of carbonate (Equation 1; Aloisi et al., 2000).
The influence of aerobic methanotrophy on carbonate dissolution has been confirmed in closed-system experiments with the Type II methanotrophic bacterial strain Methylosinus trichosporium OB3b, demonstrating that aerobic methanotrophs have the potential to enhance carbonate dissolution and induce corrosion (Krause et al., 2014).
Biomarkers of aerobic methanotrophic bacteria are typically 13 Cdepleted to different degrees (Jahnke et al., 1999), providing additional evidence for an assignment to this group of source organisms.
One example of seep carbonates with the coincidence of extensive corrosion features and abundant lipids of aerobic methanotrophic bacteria stems from northern Istria, Croatia (Natalicchio et al., 2015).
Here, we report on a corrosion and bioerosion experiment using limestone cubes deployed for 2.5 years at three locations within the REGAB seep field of the Gabon-Congo-Angola passive margin, including a mussel bed, a site with outcropping methane hydrate and active seepage, and a site with no apparent seepage for comparison.
After sample recovery, the surfaces of cubes were inspected for carbonate corrosion and bioerosion, by identifying traces of microbioerosion on the limestone surfaces and in epoxy casts thereof using scanning electron microscopy. Adhering biofilms were scraped off the cubes and analyzed for their lipid biomarker inventories; additionally, samples showing signs of microbioerosion were analyzed by 16S rRNA gene sequencing targeting the bacterial population. The combination of obtained data allows us to make a case for carbonate corrosion caused by aerobic methanotrophic bacteria, supporting the hypothesis of aerobic methanotrophy as one of the triggers of carbonate corrosion at marine methane seeps.

| G EOLOG IC AL S E T TING
The giant REGAB Pockmark is an active seep field located at approximately 3160 m water depth on the Gabon-Congo-Angola passive margin, about 10 km to the north of the Congo deep-sea channel ( Figure 1; Ondréas et al., 2005). The REGAB pockmark is above (1) CH 4 + 2O 2 → CO 2 + 2H 2 O the carbon compensation depth (CCD), currently approximately at 5000 m in the Atlantic Ocean (Woosley, 2016), hence no physicalchemical dissolution occurs at the study site and carbonates can accumulate over long periods of time (Pop Ristova et al., 2012). The elliptical structure with diameters ranging between 700 and 950 m is composed of numerous depressions that are 0.5-15 m deeper than the surrounding seafloor (Marcon, Ondréas, et al., 2014). Remotely operated vehicle-borne micro-bathymetry and backscatter data of the entire structure, and high-resolution photo-mosaicking, sidescan sonar mapping of gas emissions, and maps of faunal distribution as well as of carbonate crust occurrence were combined to provide detailed information on fluid flow regimes in the subsurface (Marcon, Ondréas, et al., 2014). One prominent feature of the REGAB Pockmark is the presence of abundant authigenic carbonate (Ondréas et al., 2005;Pierre & Fouquet, 2007). Seismic profiles showed that the REGAB Pockmark is linked to a deep channel system that acts as a reservoir for seeping fluids by a 300-m deep chimney rooted in the channels, along which gas escapes (Ondréas et al., 2005). Gas seepage has been observed at several locations, and gas hydrates have been detected in the shallow subsurface at depths of 6 m or outcropping on the seafloor (Olu-Le Roy et al., 2007). The gas hydrates are composed mainly of biogenic methane (up to 99%) and traces of other gases, such as carbon dioxide, ethane, and hydrogen sulfide (Charlou et al., 2004), which sustain an abundant and diverse population of megafauna and microbial communities, including aerobic and anaerobic methanotrophs (Bouloubassi et al., 2009;Marcon, Sahling, et al., 2014). The methane gas was found to exhibit δ 13 C values of −69.3‰ (Charlou et al. (2004).

| ME THODS
Non-seep carbonate samples were deployed in three habitats within the REGAB seep field (Congo area) at ca. 3160 m water depth as part of the "corrosion and bioerosion" (CAB) experiment launched during cruise GUINECO M76/3b with the research vessel Meteor in July-August 2008 (Figure 2; Zabel et al., 2012). Experiment CAB-B was deployed in a mussel bed, experiment CAB-C was placed on an active seepage site with abundant chemosymbiotic benthos (mussels and tubeworms) and underlying gas hydrates; experiment CAB-D was deployed on the seafloor at a site where no evidence of seepage was recognized ( Figure 2). Data on pore water geochemistry (hydrogen sulfide and methane fluxes) of the studied habitats at the REGAB pockmark were published by Olu-Le Roy et al. (2007) and Pop Ristova et al. (2012). The latter study used sediment samples and in situ geochemical measurements collected during the M76/3b cruise, when our CAB samples were deployed. These measurements included a mussel patch close to CAB-B, and bare sediments close to the carbonate deployment station CAB-D. Unfortunately, no sediment cores were taken and no biogeochemical measurements were made right at the carbonate site of CAB-C due to the hard substrate at this location and outcropping gas hydrates, accompanied F I G U R E 1 Location of the study area within the REGAB Pockmark (after Marcon, Sahling, et al., 2014). by vigorous methane gas efflux. Pore-water concentrations of hydrogen sulfide as well as hydrogen sulfide fluxes generally increased with increasing sediment depth at all study sites; with maximum concentrations detected at mussel beds close to CAB-B, where the hydrogen sulfide flux was 23 mmol m −2 d −1 and the methane efflux 334 mmol m −2 d −1 , representing average values in 0-10 cm sediment depth. At the site close to CAB-D, the hydrogen sulfide flux was 7 mmol m −2 d −1 , methane efflux was not detectable (Pop Ristova et al., 2012). The hydrogen sulfide gas fluxes in the bottom water at REGAB were low or undetectable (Olu-Le Roy et al., 2007). Methane concentrations in the bottom waters (10 cm above seafloor), as well as methane fluxes and methane consumption rates, were highest in the vicinity of the mussel habitat (CAB-B), suggesting substantial transport of methane to the bottom water. The reference site (CAB-D) showed low contents of methane gas compared with CAB-B (Pop Ristova et al., 2012).
Each experiment was comprised of four limestone cubes with 10 cm side length, including two cubes of marble and two cubes of fossiliferous limestone, all of which were regularly pierced by vertical drill holes ( Figure 2). Before deployment, the limestone cubes had been fixed on a square-cut plastic board (edge length 25 cm), which was regularly pierced by drill holes to allow for fluid flow through the equidistantly pierced limestone cubes. Each of the limestone cubes was attached to the plastic board by a central metal screw.
The samples were retrieved in January-February 2011 during the West Africa Cold Seeps (WACS) cruise aboard the R/V Pourquoi Pas? (Olu, 2011). For this study, only the marble cubes have been used.
The marble cubes were divided into subsamples for petrographic and organic geochemical analyses.

| Lipid biomarkers analyses
For lipid biomarker analyses, biofilms attached to the carbonate surfaces were scraped off with a razor blade. The surface coatings were gently grounded and hydrolyzed with 6% KOH in methanol to cleave ester-bond lipids, and then extracted with a mixture of dichloromethane/methanol (3:1, v/v) by ultrasonication until the solvents became colorless. An aliquot of the resulting lipid extracts was separated by column chromatography (using aminopropyl-bonded silica gel column) into four fractions: (1) hydrocarbons (n-hexane), (2) ketones (n-hexane/ dichloromethane, 3:1, v/v), (3) alcohols (dichloromethane/acetone, 9:1, v/v), and (4) carboxylic acids (2% formic acid in dichloromethane; cf. Cordova-Gonzalez et al., 2020). Alcohols were analyzed as trimethylsilyl-ether derivatives and fatty acids as fatty acid methyl ester derivatives (cf. Birgel & Peckmann, 2008). The double bond position of monounsaturated fatty acids was determined by analysis of their dimethyl disulfide adducts following Nichols et al. (1986). The GC temperature program was: injection at 60°C (1 min) to 150°C at 10°C min −1 , then from 150°C to 320°C at 4°C min −1 , 37 min isothermal. The identification by GC-MS was based on GC retention times, comparison of mass spectra with published data, and coelution experiments. Internal standards (cholestane, 1-nonadecanol, 2-methyl-octadecanoic acid) with known concentrations were added prior to sample hydrolysis and extraction. Compound-specific carbon isotope analysis was performed using a Thermo Fisher Trace GC Ultra connected via a Thermo Fisher GC Isolink Interface to a Thermo Fisher Delta V Advantage isotope-ratio-monitoring mass F I G U R E 2 Carbonate experiments with four limestone cubes each, deployed for 2.5 years within the REGAB Pockmark in the Lower Congo Basin at ca. 3160 m water depth. (a) Upon deployment of CAB-B in a mussel bed, gas escaped from the sediment. (b) CAB-C placed onto an active hydrate site with mussels and tubeworms, yellow arrow indicating a shrimp, white arrow pointing to a crab. (c) CAB-D located in a reference area with no apparent seepage activity. The conditions for GC − IRM − MS analyses were identical to those used for GC analyses. Each measurement was calibrated using several pulses of carbon dioxide gas with known composition at the beginning and the end of the run. The precision of measurements was checked with a mixture of n-alkanes (C 14 -C 38 ) with known isotopic composition. The analytical standard deviation was smaller than 0.4‰.

| 16S rRNA gene sequencing
16S rRNA gene sequence analysis was performed on biofilms adhered to carbonates surfaces showing evidence of microbial corrosion (CAB-C). DNA was extracted from 0.2 g material using a phenol-chloroform extraction protocol previously described (Urich et al., 2008). 16S rRNA gene libraries of bacteria were constructed with primers 616 V and 1492R (Juretschko et al., 1998;Lane, 1991).
The gel-purified PCR products were cloned into a pGEM-T Easy vector system (Promega, Mannheim, Germany) and positive clones were further processed and Sanger-sequenced at LGC Genomics (Berlin, Germany). Sequences were taxonomically classified with CREST (Lanzén et al., 2012). The sequences reported in this paper have been deposited in the GenBank database under accession numbers OM761078 -OM761163.

| Composition of the microbial community
16S rRNA gene sequencing was carried out to decipher the bacterial community in the biofilms attached to the corroded surface of the CAB-C marble cube from a hydrate-bearing active seep site ( Figure 4), for which SEM analyses revealed evidence of microbioerosion. Biofilms were adhering to all cube surfaces exposed to seawater, including the upper surface. A total of 86 bacterial clones were detected (Table S1). Phylogenetic information shows that Proteobacteria predominate, accounting for 70% of the total clones.  (Table S1).
Members of the class Gammaproteobacteria, known to be sulfideoxidizing bacteria, were detected as well, accounting for 21% of the total clones. Detected sulfide-oxidizing bacteria include members of the order Thiotrichales (e.g., genera Leucothrix and Thiothrix of the family Thiotrichaceae) and Chromatiales (e.g., genus Granulosicoccus of the family Granulosicoccaceae).

| Lipid biomarkers
The lipid biomarker compound inventories of the hydrocarbon, alcohol, and fatty acid fractions of biofilms from the CAB-B, CAB-C, and CAB-D marble cubes are presented in Table 1.
Overall, highest contents of lipid biomarkers were detected for CAB-C (277.96 μg/g wet weight), followed by the reference sample CAB-D (61.84 μg/g), while lowest contents were found for CAB-B (48.48 μg/g). In all samples, fatty acids are the most abundant compounds, representing 87% of all lipid biomarkers in the CAB-B cube, 82% in CAB-C, and 96% in CAB-D. The second largest group F I G U R E 3 Scanning electron imaging of experimental marble substrates CAB-C (a) and epoxy casts thereof (b-f), showing distinctive signs of microbioerosion: (a) intense surface pitting in the coarsely crystalline carbonate matrix (m); (b) epoxy cast with circumradial microboring structure; (c-d) dense clusters of the same type of microboring; (e-f) close-ups of individual tunnels radiating and ramifying from a central point of entry toward slightly widened terminations.
F I G U R E 4 Diversity and relative abundance (%) of bacterial groups in the corroded surface of CAB-C. Each color represents the percentage of the total sample contributed by each taxon group at (a) phylum level, except for Proteobacteria and Bacteroidetes, which are shown by class; (b) order level (dominant phylotypes with relative abundance ≥3%), (c) family level (dominant phylotypes with relative abundance ≥3%).
are alcohols, representing 12% in CAB-B, 17% in CAB-C, and 3% in CAB-D. Hydrocarbons were detected in trace amounts in all samples, making up less than 3% of the total lipid biomarkers.
Hopanoids and steroids dominate the alcohol and hydrocarbon fractions. Among fatty acids (including combined free and phospholipid derived fatty acids), saturated and monounsaturated TA B L E 1 Contents (μg/g) and δ 13 C (‰ vs V-PDB) values of lipid biomarkers from biofilms attached to CAB-B, CAB-C, and CAB-D cubes.
*Excluded from calculations due to coelution of n-C 16:1ω7t , and n-C 16:1ω5c/t with n-C 16:2 on GC-FID. compounds with 16 and 18 carbon atoms are of particular interest ( Figure 5). These compounds are the dominant fatty acids of aerobic methanotrophs (Hanson & Hanson, 1996), and the most abundant fatty acids in the samples (>75% of fatty acid content).
The δ 13 C values of fatty acids vary widely from −71‰ to −30‰.

F I G U R E 6
Relative abundance of lipids in the alcohol (a) and fatty acid (b) fractions of biofilms from CAB-B, CAB-C, and CAB-D marble cubes. Values within each stacked column represent δ 13 C values of lipids. δ 13 C value of n-C 16:1ω6c not provided due to coelution with C 16:1ω7c on GC − IRM − MS.
No δ 13 C values were obtained for compounds of the hydrocarbon fraction due to too low contents.

| DISCUSS ION
Some biogeochemical processes strongly affect the carbonate system in sedimentary environments by the production or consumption of protons and the production of carbonate species (Coleman, 1993), causing either the formation of authigenic carbonate minerals or the dissolution of skeletal or authigenic carbonate in the sediment or at the sediment surface. Apart from such effects driven by the metabolism itself, extracellular polymeric substances (EPS) are known to affect the carbonate system (Dupraz et al., 2009). Microbial sulfate reduction is an example to highlight that the effect of a biogeochemical process is not straightforward. It depends on factors including (1) environmental conditions (e.g., presence and speciation of iron), (2) the types of electron donors metabolized, and (3) EPS composition if sulfate reduction results in carbonate formation or dissolution (Baumann et al., 2016;Gallagher et al., 2014;Meister, 2013).
The effects of other biogeochemical processes on the carbonate system are more straightforward. This includes the calcium pump (i.e., ATPase-mediated transcellular Ca 2+ transport) of cyanobacteria (Garcia-Pichel, 2006;Garcia-Pichel et al., 2010), which is one of the mechanisms, apart from local acidification caused by respiration, that might also apply for other agents of microbial bioerosion in marine environments. With respect to marine methane seeps, aerobic methanotrophy and sulfide oxidation have been held responsible for carbonate dissolution (Himmler et al., 2011;Matsumoto, 1990). Yet, experiments to assess the effect of biogeochemical processes on the carbonate system in marine methane-rich environments are scarce.

| Were aerobic methanotrophic bacteria corroding carbonate at the REGAB seep field?
The SEM analyses revealed evidence of microbioerosion associated with the presence of biofilms adhering to the carbonate substrate of cube CAB-C, which was exposed to active methane seepage and outcropping hydrates for 2.5 years. As expected, reference site CAB-D revealed no evidence of active seepage and respectively no signs of bioerosion. For CAB-B, where methane and hydrogen sulfide flux was measured when the samples were deployed (see Olu-Le Roy et al., 2007;Pop Ristova et al., 2012), no bioerosion was recognized.
Possibly, the lack of bioerosion for the latter site was caused by temporal variability or even episodical cessation of seepage, indicated by a lower density of the local mussel population with smaller size of mussels compared with other mussel-dominated sites at the REGAB Pockmark (cf. Olu-Le Roy et al., 2007). In fact, CAB-B might represent an older mussel assemblage related to lower methane flux (cf. Olu-Le Roy et al., 2007), which could be one of the reasons why only the marble cube at the high flux, mussel-inhabited CAB-C site was affected by microbioerosion. Such pattern provides circumstantial evidence for a methane-dependent bioerosion process in case of the CAB-C cube, given the dominance of aerobic methanotrophs in the adhering biofilm suggested by 16S rRNA genes and lipid biomarker data. This line of reasoning receives further support from the observation of a nearly monoichnospecific bioerosion trace assemblage with the dominant microbioerosion trace being new to science and having a very high abundance in the samples studied. These microbioerosion traces have been formed in a purely inorganic substrate (marble), being of limited attraction for organotrophic microendoliths such as marine fungi, which would have been expected in a much higher abundance and diversity in a biogenic substrate (e.g., skeletal carbonate) after more than 2 years of exposure in this aphotic environment (cf. Golubic et al., 2005;Wisshak et al., 2011).
Together, the ichnological observations suggest that the dominant agent of microbioerosion observed in the marble cubes exposed to active seepage is a bacterium with a methane-or hydrogen sulfidebased metabolism.
Even at the most active seeps, hydrogen sulfide tends to be completely oxidized at the seafloor, and its concentration in the bottom water is typically much lower than the concentration of methane (Boetius & Wenzhöfer, 2013;Niemann et al., 2005;Sahling et al., 2002), Aerobic methanotrophy commonly occurs at methane seeps (Birgel et al., 2011;Birgel & Peckmann, 2008;Cordova-Gonzalez et al., 2020;Elvert & Niemann, 2008;Himmler et al., 2015;Kellermann et al., 2012). The composition of a community of aerobic methanotrophs can be assessed by phylogenetic analyses and lipid biomarkers (Hanson & Hanson, 1996). To date, no pure cultures of bacteria of the Hyd24-01 clade exist; however, this clade has been reported from other methane-rich sites, including surface sediments of Haakon Mosby Mud Volcano (Lösekann et al., 2007). The uncultured clade IheB2-23 was first detected in the water column of the northern South China Sea (Mau et al., 2020).
Apart from aerobic methanotrophs, the CAB-C biofilms contained abundant Thiotrichales. These and other sulfide-oxidizing bacteria are potential candidates other than aerobic methanotrophs that may have caused the observed microbioerosion of marble. The biomarker inventory of the CAB-C biofilms helps to further evaluate the phylogenetic affiliation of the microbioeroder. Although, lipid biomarkers are affected by taphonomic processes (e.g., Xie et al., 2013), the circumstance that active biofilms have been sampled renders it unlikely that the obtained lipid inventory was compromised by degradation processes. Biomarkers are therefore suitable to determine the microorganisms that dominated the CAB-C biofilms and to identify the probable bioeroder.
The lipid biomarker inventory of the biofilm adhering to the CAB-C marble chiefly includes fatty acids, accompanied by 4-methyl and 4,4-dimethyl sterols and hopanoids, most of them commonly reported from aerobic methanotrophic bacteria. The fatty acid profile of the CAB-C biofilm is dominated by n-C 16:1ω8 (δ 13 C: −69‰).
This lipid biomarker is known to be produced only by Type I methanotrophic bacteria (Bowman, 2006;Willers et al., 2015), including members of the genus Methylomonas (Hanson & Hanson, 1996), detected by 16S rRNA gene sequencing. Another fatty acid related to Type I methanotrophs of the genus Methylomonas is n-C 16:1ω5t , which represents an excellent environmental marker for aerobic methanotrophy (Bowman, 2006;Willers et al., 2015).
Abundant n-C 16:1ω7 (δ 13 C: −60‰), a fatty acid found in high concentration in Methylomonas strains (Bowman, 2006), is a ubiq-  (Knief, 2015). Since similar fatty acid profiles have been described for Type I and II methanotrophs (e.g., Knief, 2015), additional evidence is needed for such comparison. The assignment of n-C 18:1ω7c fatty acid to a source organism is further complicated by the circumstance that this compound is also produced by sulfide-oxidizing bacteria (McCaffrey et al., 1989), and possibly by other bacteria identified in the biofilm attached to the carbonate, since this fatty acid is produced by many organisms. Its low δ 13 C value rather agrees with aerobic methanotrophs as dominant source microorganisms (cf. Kellermann et al., 2012;Pond et al., 1998), although mixed input from both groups cannot be excluded.
Hopanoids commonly occur in aerobic methanotrophic bacteria, especially bacteriohopanepolyols (Talbot et al., 2001). In the CAB-C biofilm, the only abundant hopanoid was diplopterol, which is similarly 13 C-depleted as the associated 4-methyl and 4,4-dimethyl sterols and therefore is also assigned to methanotrophic bacteria.
Biomarkers of aerobic methanotrophs are typically 13 C-depleted due to fractionation relative to the already 13 C-depleted methane source. Thus, compound-specific isotope composition represents a valuable tool for the interpretation of biomarkers of methanotrophs in recent and ancient environments (Cordova-Gonzalez et al., 2020;Jahnke et al., 1999). Considering a δ 13 C methane value of ca. −69‰, measured in shallow gas hydrates recovered from the REGAB Pockmark (Charlou et al., 2004), the fractionation (Δδ 13 C) between methane and biomarkers of aerobic methanotrophs averages −12‰  (Jahnke et al., 1999), in accordance with the Δδ 13 C terpenoids-methane found for the CAB-C biofilms herein. Yet, the Δδ 13 C of fatty acids relative to methane is +6‰ on average (from 0‰ to +13‰). Such 13 C enrichment is best explained by cellular physiology and intermediate metabolites causing fatty acids to be 13 C-enriched relative to hopanoids in methanotrophs (cf. Jahnke et al., 1999) or by additional input of ω7 fatty acids from sulfide-oxidizing bacteria.
Together with the predominance of aerobic methanotrophs revealed by 16S rRNA gene analyses and the supposedly greater potential for methane oxidation rather than sulfide oxidation several centimeters above the seafloor on the surface of the exposed marble cubes (cf. Boetius & Wenzhöfer, 2013;Niemann et al., 2005;Sahling et al., 2002), the observed bioerosion below the CAB-C biofilm was probably caused by methanotrophic bacteria, whereas sulfide-oxidizing bacteria are apparently subordinate members of the CAB-C biofilm community.

| Carbonate microbioerosion and aerobic methanotrophy
Our carbonate corrosion experiment in the REGAB seep field confirms that carbonate rocks are susceptible to corrosion when exposed to methane seepage in an oxic environment. The corrosion surfaces, in this case microbioerosion traces, were found to be associated with biofilms dominated by aerobic methanotrophic bacteria. After 2.5 years of exposure to a seepage environment, the CAB-C marble blocks were found to exhibit densely pitted surfaces resulting from microbioerosion. The CAB-C trace fossil (Figure 3) has not previously been reported elsewhere. Pitting was apparently a chemically driven process, supposedly relying on a metabolism that produced acidity. It cannot be excluded that the trace maker used organic acids or chelating compounds in addition to metabolites or, possibly, took advantage of the chemical environment created in the biofilm (low pH caused by aerobic methanotrophy and subordinate sulfide oxidation), but was not a methanotroph or thiotroph itself.
Based on our data, the phylogenetic affiliation cannot be assessed with certainty and a better understanding of this new type of microbioerosion requires further studies. However, the new data of the CAB experiment, specifically the bioerosion observed in CAB-C, support the hypothesis that aerobic methanotrophy at seeps can be closely associated with substantial carbonate corrosion.
Abundant corrosion surfaces had already been identified in ancient seep carbonates also found to contain biomarkers of aerobic methanotrophs Natalicchio et al., 2015). In some instances, lacking biomarker evidence, corrosion of seep carbonates was attributed to aerobic methanotrophy (Aloisi et al., 2000), sulfide oxidation (Campbell et al., 2002), or a combination thereof (Himmler et al., 2011). Examples of carbonate corrosion in modern seep environments stem from the Mediterranean Sea, such as the Amsterdam and Athina Mud Volcanoes (Himmler et al., 2011) and the Nadir Brine Lake (Aloisi et al., 2000). Laboratory experiments conducted by Krause et al. (2014)  a site unaffected by seepage. Shells exposed to seepage were substantially affected by carbonate corrosion. Carbon mass balances led the authors to conclude that aerobic oxidation of methane was the dominant mechanism lowering the porewater pH and consequently causing carbonate dissolution. Our experiment in the REGAB seep field supports this interpretation, providing new microtextural, phylogenetic, and biomarker evidence for the role of aerobic methanotrophy in carbonate corrosion at methane seeps.
Given the current concern about climate change and resultant ocean acidification, it becomes pertinent to pay more attention to the mechanisms involved in carbonate dissolution including microbially driven corrosion. More research, and in particular more laboratory studies (e.g., Leprich et al., 2021), should be conducted to understand the mechanisms and the dynamics of carbonate corrosion and microbioerosion. With seawater on a projected trend of declining pH, the demand on quantitative estimates of the influence of aerobic methanotrophy and other biogeochemical processes on carbonate corrosion -a process sequestering carbon dioxide as seawater bicarbonate -in marine environments is increasing.

| CON CLUS IONS
A carbonate corrosion experiment conducted over the course of 2.5 years in the REGAB seep field confirms that carbonate rocks exposed to methane seepage are susceptible to corrosion when exposed to oxic conditions. After retrieval, purely inorganic marble cubes were found to be covered by biofilms dominated by aerobic methanotrophic bacteria. Underneath biofilms, marble surfaces of the CAB-C experiment, exposed to active methane seepage, were affected by corrosion, more precisely microbioerosion. The discovered microbioerosion trace with its circumradial tunnels, ca. 2-3 μm in diameter and radiating from a central point of entry while repeatedly bi-or trifurcating in acute angles, is new to science. The genetic data obtained from biofilms reflect the dominance of Type I aerobic methanotrophs of the order Methylococcales; among these, most methanotrophs are relatives of the uncultured clade Hyd24-01.
Similar to the genetic data, the biofilm's lipid biomarker inventory including abundant 13 C-depleted fatty acids (n-C 16:1ω8c , n-C 18:1ω8c , n-C 16:1ω5t ), 4-methyl and 4,4-dimethyl sterols, and diplopterol suggests a close tie between methanotrophs and their acidity-generating metabolism and microbioerosion; however, our data do not allow for the assignment of the trace maker to a certain phylogenetic affiliation. This study is only the second report of the uncommon sterol 4α-methylcholest-8(14)-en-3β,25-diol. As for the first report of this compound, it is associated with abundant methanotrophs of the Hyd24-01 clade. Such coincidence renders it likely that 4αmethylcholest-8(14)-en-3β,25-diol is a biomarker for the Hyd24-01 clade. The microtextural, phylogenetic, and biomarker evidence compiled during the examination of the REGAB carbonate corrosion experiment supports the concept of aerobic methanotrophy as a potent trigger of carbonate corrosion at marine methane seeps.

ACK N OWLED G M ENTS
We thank the German Academic Exchange Service -DAAD, which

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.