Assessment of the in situ biomethanation potential of a deep aquifer used for natural gas storage

Abstract The dihydrogen (H2) sector is undergoing development and will require massive storage solutions. To minimize costs, the conversion of underground geological storage sites, such as deep aquifers, used for natural gas storage into future underground hydrogen storage sites is the favored scenario. However, these sites contain microorganisms capable of consuming H2, mainly sulfate reducers and methanogens. Methanogenesis is, therefore expected but its intensity must be evaluated. Here, in a deep aquifer used for underground geological storage, 17 sites were sampled, with low sulfate concentrations ranging from 21.9 to 197.8 µM and a slow renewal of formation water. H2-selected communities mainly were composed of the families Methanobacteriaceae and Methanothermobacteriaceae and the genera Desulfovibrio, Thermodesulfovibrio, and Desulforamulus. Experiments were done under different conditions, and sulfate reduction, as well as methanogenesis, were demonstrated in the presence of a H2 or H2/CO2 (80/20) gas phase, with or without calcite/site rock. These metabolisms led to an increase in pH up to 10.2 under certain conditions (without CO2). The results suggest competition for CO2 between lithoautotrophs and carbonate mineral precipitation, which could limit microbial H2 consumption.


Introduction
Our societies are facing the challenges of climate change, the need to massiv el y de v elop r ene wable ener gies, the ener gy sov er eignty, and the cost of energy.The ongoing de v elopment of the dihydrogen (H 2 ) sector and the imminent arrival of green H 2 from rene wable ener gies in the gas grid (Le Duigou et al. 2017 ) have led many industrialists and academic researchers around the world to examine the consequences for surface infr astructur e (DBI GUT 2017 ) and under gr ound geological stor a ge (UGS) sites used to balance the grid and secure supplies.Ultimately, the aim is to transform UGS into under gr ound H 2 stor a ge (UHS; Dopffel et al. 2021, Heinemann et al. 2021, Kr e vor et al. 2023 ).
The question of the future of UGS in general, and UHS in particular, is central to (i) developing future massive energy storage to accommodate seasonal variations; (ii) securing countries' energy r eserv es; and (iii) avoiding a possible fr a gmentation of a global gas network that would hinder the de v elopment of the H 2 energy sector or even renewable energies (Rabiee et al. 2021 ).In the petr oc hemical and chemical sectors, H 2 storage in salt caverns has been in use for se v er al decades, and specialists a gr ee that the technology is reliable (Aftab et al. 2022, Réveillère et al. 2022, Bradshaw et al. 2023 ).Ho w e v er , the number , volume, and geogr a phical distribution of these salt caverns are far from sufficient to store H 2 on a massive scale, given current production projections, which explains the strong interest in porous reservoirs (Barison et al. 2023 ).Most of the work focusing on H 2 stor a ge in porous reservoirs, such as depleted reservoirs or deep aquifers , in volv es sim ulations using various models that generally do not take microbial activity into account; se v er al examples of suc h sim ulations can be found in the r e vie w by Al-Shafi et al. ( 2023 ).Ho w e v er, models that take micr oor ganisms into account hav e shown that micr oor ganisms ar e likel y to hav e a str ong impact on the evolution of these futur e stor a ge sites (Iv anov a et al. 2007, P anfilov 2010, Ebigbo et al. 2013, Amid et al. 2016, Hemme and van Berk 2018, T ha ysen et al. 2021, Tremosa et al. 2023 ).
Man y deep envir onments ar e home to micr obial comm unities that use H 2 as an energy source .T hese communities are referred to by the acr on ym SLiME, standing for SubLithoautotrophic Microbial Ecosystem (Stevens and McKinley 1995, Fry et al. 1997, Chapelle et al. 2002, Takai et al. 2003, Lin et al. 2005, Crespo-Medina et al. 2014 ).Se v er al works have demonstrated, directly or indir ectl y, that UHS in por ous r eservoirs, particularl y in deep aquifers, could lead under certain conditions to in situ biomethanation by hydr ogenotr ophic methanogenic arc haea that natur all y e volv e in these ecosystems (Amigá ň et al. 1990, Buzek et al. 1994, P anfilov 2010, Liebsc her et al. 2016, Gr egory et al. 2019, Str obel et al. 2020, Haddad et al. 2022a, Molíková et al. 2022 ).These initial biomethanation results obtained for UGS in deep aquifers are at-tr activ e because they allow us to envisage a disruptive innovation.UGS in deep aquifers would combine the ca ptur e and injection of CO 2 with the production of nonfossil methane, thereby reducing the consumption of fossil hydrocarbons and curbing the quantities of greenhouse gases released while enabling more virtuous carbon-neutr al ener gy pr oduction, i.e. mor e homogeneous on a territorial scale (Zav ark o et al. 2021, Chai et al. 2023 ).Indeed, deep aquifers are found all over the globe in sedimentary basins, and those in the first 2 km of depth could r eac h a cum ulativ e volume of 22.6 million km 3 (Gleeson et al. 2016 ).The use of these storage sites would be conditional on the presence of a decarbonated H 2 pr oduction ar ea (r ene w able and nuclear), an accessible sour ce of CO 2 , ideally captured from industry or even the atmosphere (Gutknecht et al. 2018, Hou et al. 2022 ), and a geological stor a ge reservoir equipped with injection and production wells, as well as a gas network enabling biomethane distribution (Bellini et al. 2022 ).By overcoming a number of scientific hurdles, these deep, secure sites would represent a biomethanation potential at a scale se v er al times larger than that of conventional catalytic or biological methanation reactors due to the very large reservoir volumes (Molík ov a et al. 2022, Vít ȇzová et al. 2023 ).
The concept of biomethanation in por ous r eservoirs has its origins, on the one hand, in the discovery that part of the methane present in natural gas reservoirs has a biogenic origin (Davis and Updegraff 1954 ) and, on the other hand, in the realization that in situ biomethanation in a geological reservoir could be performed at time-scales compatible with industrial exploitation, as demonstrated by a study of town gas stor a ge in Lobodice (Czech Republic), with an estimated conversion of 17% H 2 (associated with CO 2 /CO present at the site) to CH 4 in just 7 months (Šmigá ň et al. 1990).In 2021, a study on the Olla Oil Field, which was operated using CO 2 injection (CO 2 -EOR) until 1986, sho w ed a conversion of 13%-19% of CO 2 into CH 4 via methanogenesis (Tyne et al. 2021 ).At pr esent, thr ee pr ojects ar e attempting to pr ov e the feasibility of biomethanation in depleted hydr ocarbon r eservoirs and ar e being tested under real conditions: the Hychico-BRGM pilot project in Argentina, the Underground Sun Conversion-Flexible Storage Project, and the Bio-UGS Project "Biological conversion of carbon dioxide and hydrogen to methane in porous underground gas stora ge facilities: Anal ysis of under gr ound biomethanation potential" (see r e vie ws by Str obel et al. 2020, Dopffel et al. 2021, Bellini et al. 2022 ).
This study targets a deep aquifer used for natural gas UGS and featuring formation water with low sulfate concentrations (Ranchou-Peyruse et al. 2019 ).In the context of its potential use as a UHS site, CO 2 will also be present due to its natural presence in the aquifer, as a coconstituent of the natural gas already present in the storage site or even as a result of voluntary injection (Delshad et al. 2022, Wang et al. 2022 ).In UHS sites in porous reservoirs, hydr ogenotr ophic methanogenic archaea compete mainly with sulfate reducers for H 2 and CO 2 , although a smaller proportion of these substrates may also be consumed by homoacetogens (Haddad et al. 2022b, Mura et al. 2024 ).Her e, cultur al and molecular biological a ppr oac hes wer e used to assess the effect of the two dominant hydr ogenotr ophic functional gr oups in this aquifer, methanogens and sulfate reducers, on gas-phase H 2 .Formation waters fr om v arious monitoring wells in the aquifer were collected to (i) assess whether the quantity of methanogens present is a good proxy for the potential for methanogenesis; (ii) e v aluate competition and possible inhibition between the two functional groups; and (iii) assess the impact of the r oc k, specificall y calcite, on the metabolic activities of interest.

Sampling campaigns
The deep aquifer used as a UGS site is located in the sedimentary basin of southw est F rance .T he various wells sampled were the subject of a pr e vious study, and the well names have been retained (Ranchou-Peyruse et al. 2019 ).All formation water sampling was carried out at the head of monitoring wells after they had been purged 10 times with the volume of each well.
The first part of the study was carried out using formation water from 17 wells (Fig. 1 ) sampled between September and October 2020.For each well, the water was collected in four 1 l flasks filled to the brim (three flasks for molecular biology analyses and one for culture methods).Upon return to the laboratory, the w ater w as either stored at 4 • C for microbial enrichment (1 l) or filtered in triplicate through 0.1-μm pore size filters (PES, Sartorius Stedim) at a rate of 1 l of water per filter.The filters were then stored at −20 • C to preserve the DNA until use.
In the second part of the study, five 1 l bottles of wellhead w ater w er e sampled a gain at se v en sites in May and October 2021 based on the results obtained in the first campaign (Ab_L_1, Ab_L_3, Ab_L_7, Ab_L_8, Ab_l_10, Ab_Y_2, and Ab_P_1).For each site, two 1 l flasks were stored at 4 • C until used for microbial enric hment.On site, thr ee 1 l flasks wer e filter ed immediatel y thr ough thr ee differ ent 0.1 μm filters (PES, Sartorius Stedim) and pr eserv ed dir ectl y in liquid nitr ogen until r eturned to the labor atory, wher e they were then stored at −80 • C until use .T he formation water composition and physico-chemical characteristics ar e pr esented in Table S1 ( Supporting Information ) (SOBEGI, Lacq, France).

Culture methods
Using water sampled during the first campaign, microbial enrichments were performed in 500 ml flasks in an anaerobic chamber (nitr ogen atmospher e; J acomex).A v olume of 250 ml of formation water containing the indigenous micr oor ganisms was incubated with a gas phase consisting of H 2 /CO 2 (80/20; 1.5 bar) at 37 • C. For eac h site, ther e was onl y one assay and one abiotic control was carried out by filter sterilizing the formation water (0.1 μm, Sartorius Stedim) inside the anaerobic chamber.
The water sampled at se v en sites during the second campaign was subjected to microbial enrichment in a similar way to that described abo ve .T his time , se v er al conditions wer e tested in duplicate for each site: incubation in the presence of a gas phase consisting of either H 2 /CO 2 or H 2 .Eac h time, tr eatments with the presence or absence of calcite (mineral particles < 150 μm) were tested.An abiotic control was prepared for each of the conditions tested.For site Ab_P_1, 1 g of crushed cores from the site was used in place of calcite in the presence of H 2 .For site Ab_Y_2, an additional condition was pr epar ed by adding 0.05 g of barite ( < 150 μm) to the calcite.From the duplicates, one of the tests was dedicated to the analysis of taxonomic diversity, while the second was used to measure the modification of the physico-chemical composition of the water, as well as the effect of incubation on the r oc k when it was present.
Calcite, r eservoir r oc k, and barite wer e gr ound manuall y using a mortar and pestle before being sieved to < 150 μm.The resulting po wders w er e autoclav ed befor e use.Roc ks in the abiotic and the biotic r eplicates r eserv ed for ionic anal ysis wer e weighed befor e and after the experiment using a balance (A&D Company, Limited FZ-5001) with the accuracy of ± 0.001 g.

Gas measurement
The evolution of the gas phase (CH 4 , CO 2 , H 2 , and H 2 S) was monitored for all enrichment samples throughout the experiment (GC-μTCD, Micr o GC Fusion; Cheml ys, Fr ance).Measur ements wer e taken in duplicate at each time-point, with an uncertainty of ±5%.Detection limits were 5 ppm for CO 2 and H 2 and 20 ppm for CH 4 .The headspace volume was determined for each vial (total vial v olume −v olume of liquid added).After each gas sample was taken for analysis, the pressure in the flask was measured with a manometer (Digitron 2022P, Farnell).P er cent obtained with the GC-μTCD was converted into concentrations (mmol) for each gas.The quantity of gas moles was determined according to the ideal gas law (PV = nRT) with R = 8.3145 J mol −1 K −1 and T = 310.15K.

pH and Eh measurements
At the initial and final time points, 1 ml was taken from all biotic and abiotic enrichments to measure pH and oxidation-reduction potential (Inlab Ultr a-micr o ISM and Redox micr oelectr odes, Mettler Toledo, Se v en Compact, Columbus, Ohio, USA).

Chemical measurements
For each test condition, water from a biotic replicate and the abiotic contr ol wer e used in their entirety to quantify ions c har acteristic of the dissolution or precipitation of minerals of interest (barite, calcite).Ba 2 + and Ca 2 + ions were monitored by inductiv el y coupled plasma-atomic emission spectroscopy (ICP-AES) with a limit of quantification of 0.05 mg l −1 and 0.1 mg l −1 , respectiv el y; HCO 3 − and CO 3 2 − wer e determined by titrimetry with a limit of quantification of 40 mg l −1 and 60 mg l −1 , respectiv el y (UT2A, P au, Fr ance).For all enric hment samples, sulfate was measured using ion chromatography (Dionex Integrion HPIC, Thermo Fisher Scientific) with the accuracy of ±5%.To quantify sulfide, 20 μl of cell cultures from each sample were mixed with 480 μl of zinc acetate at 2% concentration for sulfide quantification (Cline 1969 ).A standard curve was constructed using a sulfide solution, and along with the samples, 200 μl of DMPD ( N , Ndimethylpar a phen ylenediamine sulfate) at 0.2% (w/v) were added After 20 min in the dark at room temperature, the absorption at 670 nm was measured with a V-1200 spectrophotometer (VWR).

Molecular approaches
During the first campaign, molecular analyses were carried out onl y on DNA extr acted fr om formation water filtered in the laboratory (0.1 μm, Sartorius , Stedim).T he 16S rRN A (v4-v5), dsrB and mcrA genes were quantified to estimate the presence of all pr okaryotes, sulfate r educers and methanogenic arc haea, r espectiv el y.During the second campaign, the water was filtered directly on site (0.1 μm, Sartorius, Stedim), and the filters were preserved in liquid nitrogen until they r eac hed the labor atory, wher e they wer e stor ed at −80 • C until use.

Nucleic acid extraction and RT-PCR
All filters were ground manually using a mortar and pestle with liquid nitrogen.For samples from the first sampling campaign, DN As w er e extr acted using the DNeasy Po w er Soil kit (Qiagen) according to the supplier's recommendations.For the second part of the study, all nucleic acids wer e extr acted using the Fast RNA Pr osoil Dir ect kit (MP BIO Medicals), and DN As and RN As w ere separated using the All Prep RN A/DN A kit (Qiagen), accor ding to the supplier's recommendations.Nucleic acids were quantified using Quant-it TM dsDNA HS (High sensibility) and Quant-it TM RiboGreen kits (Invitrogen).RN As w ere converted to cDNAs using the M-MLV r e v erse tr anscription kit (Invitr ogen) according to the supplier's recommendations.

PCR, qPCR, and sequencing
F rom the DN As and cDN As obtained, the 16S rRNA , dsrB and mcrA genes were targeted using the primer pairs 515F-928R, dsr2060F-dsr4R, and mlasF-mcrAR (Wagner et al. 1998, Geets et al. 2006, Steinberg and Regan 2008, 2009, Wang and Qian 2009 ).To reduce inhibition, bovine serum albumin (BSA, NEV-B9200S) was used in the PCRs at a concentration of 1 mg ml −1 .For amplification of the 16S rRNA and dsrB genes, the Taq PCR kit (Roche) was used, while for the mcrA gene, the Fidelio ® Hot Start PCR kit (Ozyme) was used.The pr ocedur es hav e pr e viousl y been described in mor e detail in Haddad et al. ( 2022b ).Genes, their transcripts and associated standards were quantified by quantitative PCR (qPCR; Biorad CFX Connect) with 41 Tak y on NO R O X SYBR 2X MasterMix blue dTTP (Eurogentec), as pr e viousl y described (Haddad et al. 2022b ).

Targeting UGS sites that are already functional for upgrading to UHS sites
There is an urgent need to assess the potential of the various aquifers used as UGS sites to determine their possible future uses (H 2 stor a ge, CO 2 stor a ge, in situ biomethanation, geothermal energy, and so on).For this study, we selected an aquifer used for natural gas storage (Fig. 1 ) that has already been the subject of se v er al studies and could be converted into a UHS site (Ranchou-Peyruse et al. 2019, Haddad et al. 2022a ).The aquifer is configured into two anticlines (Fig. 1 A, in turquoise blue in the center of the ima ge; Fig. 1 B), whic h enables two stor a ge zones to be accommodated in the submolassic sand layer composed mainly of quartz, some calcite, and occasionally dolomite and K-feldspar (André et al. 2002 ), with a por osity v arying between 25% and 35% and a low concentration of sulfate (from 21.9 to 197.8 μM; Table S1 , Supporting Information ); consequently, the site is a priori favorable for H 2 stor a ge (Bo et al. 2021 ).Of the 17 sampling sites, 13 are located in the Eocene-Lutetian stratum dated from −40 to −46 My and coded Ab_L_1 to Ab_L_14 (except Ab_L_6) at depths ranging from −10 to −874 m above mean sea le v el (AMSL).Thr ee wells pr ovide access to water from the lower Eocene-Ypr esian le v el ( −46 to −53 My) at depths ranging from −475 to −949 m AMSL.Finally, formation water was sampled in the lower Paleocene/Danian layer ( −59 to −65 My) at −595 m AMSL; stored gas does not r eac h this layer, and a new UHS site could be considered here .T he a verage age of the w ater cir culating in this zone has been estimated by the GAIA project using 14 C and some 36 Cl dating to be between 20 000 and 50 000 years old ( http://infoterr e.br gm.fr/r a pports/RP-69126-FR.pdf),with por e-le v el cir culation estimated at ∼5 m y ear −1 (Labat 1998 ).

Quantification of sulfate-reducing and methanogenic microorganisms at all sites to assess biomethanation potential
An initial sampling campaign was carried out in October 2020 to screen the 17 sites for their estimated 16S rRNA gene copy concentr ations (total pr okaryotes), dsrB (sulfate r educers) and mcrA (methanogens), as presented in Fig. 2 .For prokaryotes, the avera ge concentr ation for all sites studied was 4.8 × 10 5 ± 3.3 × 10 4 copies of the 16S rRNA gene ml −1 , with the lo w est concentrations found at sites Ab_L_10 and Ab_Y_3 with 1.1 × 10 3 ± 1.5102 and 8.3 × 10 2 ± 1.7 × 10 2 copies of the 16S rRNA gene ml −1 , r espectiv el y.Conv ersel y, the sites with the highest concentrations were Ab_L_4, Ab_Y_2, and Ab_P_1, with 2.1 × 10 6 ± 1.9 × 10 5 , 2.4 × 10 6 ± 1.5 × 10 5 , and 2.5 × 10 6 ± 1.5 × 10 5 copies of the 16S rRNA gene ml −1 , r espectiv el y.All sites sho w ed the presence of sulfate-reducing micr oor ganisms, whic h often dominate microbial communities.On the other hand, based on mcrA gene detection and quantification, the presence of methanogens was observed at only 12 sites with variable and low concentrations ranging from 1.6 × 10 0 to 4.3 × 10 2 ± 8.3 × 10 1 mcrA gene copies ml −1 .The corresponding formation w aters w er e also incubated in the labor atory with a mixtur e of H 2 /CO 2 (80/20; 1 bar), and the asterisk in Fig. 2 identifies the samples showing methanogenesis activity: Ab_L_1 (2% of CH 4 in 49 days of incubation), Ab_L_3 (6% in 40 days), Ab_L_7 (6.2% in 32 days), Ab_L_8 (0.7% in 17 days), Ab_L_9 (0.2% in 254 days), Ab_L_10 (2.8% in 193 days), Ab_Y_1 (2.7% in 48 days), Ab_Y_2 (1.9% in 138 days), and Ab_P_1 (3.9% in 97 days).As expected, the absence of detection of the mcrA gene in water was corr obor ated by a systematic absence of methanogenesis, and quantification of this gene is ther efor e a good proxy for this metabolic capacity.Of the 12 formation water samples with the mcrA gene, nine sho w ed methane pr oduction.Ov er a 1-year monitoring period, samples Ab_L_2, Ab_L_4 and Ab_L_13 sho w ed no methane production.Low concentrations of mcrA gene copies alone cannot explain these latest r esults, since methane pr oduction in other assays was sometimes ac hie v ed at lo w er concentr ations.We ther efor e hypothesize that the methanogens encountered at these three sites, such as members of the families Methanomicrobiaceae and Methanosarcinaceae, could be nonhydr ogenotr ophic and use acetate , formate , alcohols and methylated compounds identified at some sites in this aquifer (Ranchou-Peyruse et al. 2019 ).This could also imply a low acetogenic activity (i.e.production of acetate and/or formate), preventing sustained activity of acetotrophic methanogens.
It is important to note that, with the exception of sites Ab_L_2 and Ab_L_4, all the other sites that did not exhibit methanogenic archaea and/or methane production are remote from current gas stor a ge locations (Fig. 1 A).The redox potential may partly explain some of these results, since several of these sites had redox potential unfavorable to hydrogenotrophic methanogenesis, such as Ab_L_5 ( −73.9 mV), Ab_L_11 ( −71.2 mV), Ab_L_12 ( + 130.9 mV), and Ab_L_13 ( −50.1 mV), instead of the optimal −200 to −400 mV (Reebur gh 1983, Hir ano et al. 2013 ).Ab_L_12 sho w ed gr eat v ariation in redox potential over the years, ranging from −119.0 to 161 mV since June 2020.
Based on the results of this first sampling campaign, a panel of sites was selected for further study: Ab_L_1, Ab_L_3, Ab_L_7, Ab_L_8, Ab_Y_2, and Ab_P_1.The choice took into account the geological layer of the formation water (L: Eocene-Lutetian, Y: Eocene-Ypr esian, and P: P aleocene-Danian), methane pr oduction from the H 2 /CO 2 gas mixture, and the quantity of mcrA genes.On this last point, site Ab_L_10 was also selected as it had few copies of mcrA genes but ne v ertheless sho w ed methanogenesis potential in cultivation trials .T he rest of the study focused on demonstrating the effect of physico-chemical and microbiological parameters on methanogenesis.

Physico-chemistry of water samples from seven selected sites
Two new sampling campaigns were carried out in May and September 2021 to resample the se v en selected formation waters ( Table S1 , Supporting Information ).These waters had low salinity c har acterized by an electrical conductivity of ∼300 μS cm −1 and negativ e r edox potential between −40.3 and −351 mV.Av er a ge sulfate concentrations ranged from 21.9 to 197.8 μM.Nitrate and nitrite concentrations were all below the detection limits of 1.6 μM and 0.3 μM, r espectiv el y.For dissolv ed ir on, w e w ere unable to distinguish between Fe 2 + and Fe 3 + , but the redox potential at the sites favors its more reduced form.André et al. ( 2007 ) suggested an equilibrium between Ca-HCO 3 facies and the dissolution of carbonate miner als suc h as calcite (CaCO 3 ).These carbonates may r epr esent a source of carbon accessible to autotrophic microorganisms under the pH and redox potential conditions pr e v ailing in the aquifer.Ther e ar e complex balances between carbonate miner als, CO 3 2 − and HCO 3 − , CO 2 dissolv ed in water and CO 2 in the gas phase (gas stor a ge).By consuming dissolv ed CO 2 , lithoautotr ophic micr oor ganisms significantl y alter these balances .T he metabolic gr oups likel y to dominate micr obial comm unities ar e sulfate r educers , methanogens , homoacetogens and fermenters .Given the nature of our study (i.e. in situ biomethanation), we decided to focus on sulfate reducers and methanogens .T hese two functional groups comprise H 2 -consuming lithoautotrophic organisms.We consider homoacetogens and fermenters as complementary, but nonetheless secondary, in the functioning of these communities in the context of massive H 2 injection.Indeed, it is expected that homoacetogens will consume part of the H 2 and CO 2 to form acetate and/or formate (Stoll et al. 2018, Haddad et al. 2022b, Mura et al. 2024 ), which will be consumed by the rest of the microbial community and, in particular, by methanogenic archaea to form methane (Pan et al. 2016 ) and heter otr ophic sulfate r educers (Weijma et al. 2002, Dai et al. 2022 ).The experiment carried out by Haddad et al. ( 2022b ), which aimed to simulate H 2 injection into an aquifer similar to the one in this study in terms of sulfate concentr ation (ar ound 150 μM), sho w ed that the main consumers of H 2 wer e methanogens (ar ound 80% of the H 2 ), while homoacetogens accounted for onl y 4%.The r emaining H 2 lost (around 16%) could be considered as consumed by sulfate reducers.In these oligotrophic en vironments , where organic carbon concentr ations ar e lo w (1.1 mg l −1 ) or belo w the detection limit, the impact of fermenters is expected to be str ongl y constr ained, participating in particular in the recycling of microbial necromass into H 2 , CO 2 , and other organic acids that can be used by sulfate reducers and methanogens.Finally, the low detected concentrations of ammonium (between 3.3 and 26.6 μM) and dissolved phosphates ( < 2.1 μM) suggest a low capacity of these ecosystems to sustain a m uc h higher biomass concentr ation than befor e H 2 injection.These molecules hav e alr eady been cited se v er al times as limiting nutrients in the deep biosphere (Madigan et al. 1997, Head et al. 2003 ).Similarly, the concentrations of certain metals such as nickel, essential for the proper functioning of enzymes such as hydrogenases, could be limiting factors for hydr ogenotr ophic In our case, we belie v e that with a slow r ec har ge of the aquifer (5 m year −1 ; Labat 1998 ) and in the context of a r oc k ov erwhelmingl y composed of quartz and few minerals that could serve as a source of phosphorus or nitrogen, the cell concentration will remain constant to within one log.Unsurprisingly, these two parameters (i.e.ammonium and phosphate) do not appear to be the only ones driving the communities, since their concentr ations wer e not dir ectl y corr elated with those of pr okaryotes in the collected formation water ( Table S1 , Supporting Information ; Fig. 2 ).

Enrichment in the presence of H 2 /CO 2 gas phase (80/20; 1 bar)
Following this first part of the study, incubations in the presence of calcite (CaCO 3 ) were carried out.Calcite plays a role in methanogenesis as an indirect carbon source via its dissolution, but its presence in the geological structures used for storage can lead to changes in porosity and permeability (Haddad et al. 2022b, Saeed et al. 2023 ).In these UGS sites, CO 2 is pr esent natur all y or artificially (coinjected with natural gas to ∼2%; Burgers et al. 2011 ).Added to this is a complex balance between gaseous and dissolved CO 2 , on the one hand, and between carbonates (CO 3 2 − ) and bicarbonates (HCO 3 − ) in water and carbonate minerals on the other.
Cultivation trials were carried out at near-atmospheric pressure (1 bar at the start of the experiment) in flasks to screen a wide r ange of conditions, whic h would not be possible with pressurized experiments .T he pr essur es encounter ed on these sampling sites (between 40 and 80 bars) are relatively low, compared to abyssal pr essur es , e .g. and are thought to have little effect on the microor-ganisms that e volv e ther e; no piezophile has e v er been discovered in deep continental systems.A priori , the microorganisms revealed in this study at atmospheric pressure would be the same at pr essur es sim ulating those in situ (i.e.high pr essur e).On the other hand, it is certain that manipulations at high pr essur e hav e an effect on the solubility of gases in water and ther efor e their accessibility to microbial populations, particularly in the case of lithoautotr ophs.Her e, the aim was not to assess yields but rather to e v aluate a potential for hydr ogenotr ophy.For eac h site, the formation water and its indigenous micr obial comm unity (without nutrient supplementation) were brought into contact with a gas phase of H 2 /CO 2 (80/20) or H 2 alone, with or without calcite.In the case of Ab_Y_2, an additional condition was added with the presence of barite (BaSO 4 ) as a potential sulfate source for sulfate reducers (Haddad et al. 2022b ).For Ab_P_1, calcite was replaced with r oc k fr om the r eservoir to mimic in situ conditions as closely as possible.
The most critical and quantifiable physico-chemical data measured at the start of the experiment and after 26 to 193 days of incubation ar e r epr esented in the principal component analysis (PCA) shown in Fig. 3 , explaining 67.6% of the sample distribution.As expected, the "Bicarbonate, " "Calcium, " and "Calcite" vectors are associated and correlate very well with Axis 1.They a ggr egate the controls and the H 2 /CO 2 tests (both with calcite) at the end of the experiment.The "sulfate" vector correlates well with axis 2 and is opposite to the "CH 4 " vector.The almost right angle formed by the "sulfate" vector and the group of the three "bicarbonate"-"calcium"-"calcite" vectors indicates that these two sets of variables are independent of each other.Finally, the "Eh " vector is logically opposed to the "CH 4 " vector.A summary of the information fr om the se v en anal yzed v ariables shows that, r egardless of the conditions tested, sulfate is the factor with the greatest influence at the start of incubation (as indicated by the empty geometric Figur e 3. PC A of the main physico-chemical parameters (bicarbonate, calcium, calcite, CH 4 , pH, Eh , and sulfate) before and after incubation of the micr obial comm unities fr om the six formation waters studied (Ab_L_1, Ab_L_3, Ab_L_7, Ab_L_8, Ab_L_10, and Ab_Y_2).Incubations were carried out in the presence of a gas phase consisting of either H 2 /CO 2 (80/20; 1 bar) or H 2 only and with or without calcite .T he results from site Ab_P_1 are not shown in this figure ( Fig. S4 , Supporting Information ).shapes in Fig. 3 ).While the sulfate concentration remained constant in all the abiotic controls, sulfate reducers consumed sulfate in all the biotic trials, although this consumption was not total over the incubation period ( Table S2 , Supporting Information ).All biotic tests sho w ed methane production and the presence of sulfate at the end of the experiment, suggesting that methanogens were able to thrive and be active at the same time as sulfate reducers ( Table S3 , Supporting Information ).Black iron sulfide precipitates wer e observ ed in all biotic assa ys .In the conditions without r oc k and without barite, the sulfide concentr ations did not exceed 5 μM at the end of incubation.With the r oc k fr om the site (Ab_P_1), these concentr ations incr eased fr om 2.3 ± 1.3 μmol of sulfide to 11.3 ± 0.9 μmol of sulfide.For the Ab_Y_2 formation water, the sulfide production increased from 2.9 ± 1.6 μmol of sulfide to 7.2 ± 0.0 μmol with the barite supplementation (Ab_Y_2) without there being any more sulfate-reducing agents quantified ( Table S3 , Supporting Information ).Low sulfate concentrations allow methanogens to compete for H 2 with sulfate reducers.In their work, Lupton and Zeikus (1984) set a concentration limit of ∼5 mM, well above the concentrations found at the various sites in the present aquifer.H 2 tests, with and without calcite, ar e gr ouped together in the upper left quadrant (Fig. 3 ).These trials ar e str ongl y marked by their highest pH v alues, since the acidity generated by CO 2 solubilization from the gas phase is absent, and calcite dissolution was observ ed.Clearl y, in the condition with CO 2 in the gas phase, CO 2 was the carbon source for methanogens and other c hemolithoautotr ophs.In the condition without CO 2 but with calcite, calcite dissolution enabled methanogenesis (CaCO 3 ↔ CO 3 2 − / HCO 3 − ↔ CO 2(aq) ).Biotic tests with H 2 alone in the gas phase natur all y sho w ed higher initial pH v alues (av er a ge pH 8.1 ± 0.1) than those under H 2 /CO 2 conditions (av er a ge pH 6.3 ± 0.2).At the end of incubation in CO 2free conditions with calcite, the highest pH was 10.2, av er a ging 9.3 ± 0.7 across all sites .T he ionic Ca 2 + concentration de-creased under the action of micr oor ganisms (0.93 ± 0.08 mM in the abiotic controls versus 0.32 ± 0.19 mM under biotic conditions).Initiall y, calcite-carbonate-dissolv ed CO 2 -gas equilibrium was ac hie v ed.The methane pr oduction indicates that the dissolved CO 2 was consumed by methanogenic archaea.This consumption led to increased calcite dissolution, the release of calcium ions and an increase in pH with the appearance of hydroxyl ions.Methanogenesis and sulfate reduction are associated with alkalinization (Berta et al. 2018, Dopffel et al. 2023 ), and this can lead to conditions deviating from the optimal growth conditions of methanogenic arc haea, whic h ar e gener all y at a ppr oximatel y pH 6.5 to 8.5, but their resistance can reach pH 10 for some (Gerardi 2003, Liu and Whitman 2008, T ha yssen et al. 2021 ).In the assays without CO 2 , we assume that when the pH of the enric hments became v ery alkaline , conditions became unfa vorable for micr oor ganisms, but not necessaril y because of a toxic effect on micr oor ganisms, as has alr eady been observ ed (Bassani et al. 2015 ).Calcium ions could then complex with the organic matter of the necromass (Kloster et al. 2013, Zhang et al. 2019 ), contributing to lo w er concentrations of this ion in biotic tests than in abiotic tests ( Table S2, Supporting Information ).This hypothesis will r equir e dedicated experiments to confirm or dispr ov e it under experimental conditions simulating environmental parameters, in particular those related to pressure , salinity, temperature , rock type, and micr oor ganisms.In the context of in situ biomethanation, this point is crucial, as it assumes that during methanogenesis, alkalinization initiates a new thermodynamic equilibrium that induces competition for CO 2 between lithoautotrophs (i.e.methanogens , sulfate reducers , and homoacetogens) and carbonate precipitation.After the depletion in sulfate, this would represent a potential brake on methanogenesis and imply a possible decrease in porosity/permeability as a function of Ca 2 + , Mg 2 + , or Fe 2 + concentr ation, whic h would induce calcite (CaCO 3 ), ma gnesite (MgCO 3 ), or siderite (FeCO 3 ) pr ecipitation, r espectiv el y.For the CO 2 -and calcite-free conditions, we can only hypothesize CO 2 pr oduction by fermentativ e and heter otr ophic functional gr oups (i.e.sulfate reducers) growing on the necromass of part of the micr obial comm unity.

Methane production as a function of test conditions
All the test conditions for the se v en formation waters, apart from the abiotic controls, sho w ed methane production (Fig. 4 ).It should be noted that in the formation waters closest to the stored natural gas bubble (Ab_L_1 and Ab_L_7), methane may still have been dissolved when the experimental tests began, which explains some of the results .T he highest methane production was observed under H 2 /CO 2 conditions (80/20) and without calcite (CaCO 3 ; Fig. 4 , Part 1).Methane production was also observed for Ab_L_10 formation water, which had a barely detectable amount of mcrA genes.
Logically, in incubations with only H 2 in the gas phase (Fig. 4 Part 2), methanogenesis was gener all y less efficient than in the presence of H 2 /CO 2 (80/20).In all these assa ys , an increase in pH was measur ed, fr om ar ound 8.0 at the start of the incubations to 10.2 (in particular, Ab_L_7 with calcite).The methanogenesis in the tests without calcite (and without CO 2 ) implies that a significant proportion of the carbon used to produce methane did not come from calcite.We hypothesize that the source carbon could be bicarbonate ions in the waters, with concentrations ranging from 2.5 to 3.2 mM ( Table S2, Supporting Information ), and by the fermentation and heter otr ophy of the microbial necromass, producing H 2 , CO 2 , and organic acids that feed methanogenic archaea.Ab_Y_2 formation water in the presence of barite (BaSO 4 ), a potential source of sulfate, did not show an increase in the concentration of the sulfate reducers ( Table S3, Supporting Information ), but rather in their activity (i.e. more sulfide produced).We deduce that for such an aquifer with relatively low sulfate concentrations between 0.02 and 0.2 mM ( Table S2, Supporting Information ), methanogenesis can take place at the same time as sulfate reduction, and the latter is not limiting for the de v elopment of methanogens.Based on the hydr ogenotr ophic methanogenesis reaction (4H 2 + CO 2 → CH 4 + 2H 2 O) and the quantities of methane detected at the end of incubation, the theoretical H 2 consumption by this metabolism has been estimated at between 0.1% and 13.4% of the H 2 consumed under H 2 /CO 2 conditions, and between 0.3% and 3.8% under conditions with only H 2 in the gas phase.
As for the other hydr ogenotr ophic metabolisms, their theoretical H 2 consumption was estimated at between 20% and 65% when the gas phase was composed of the H 2 /CO 2 mixture, and between 15% and 72% when only H 2 was present.The taxonomic diversity results (Fig. 5 ) suggest that hydrogenotrophic sulfate reducers are the k e y players in this consumption.In deep aquifers with slo w w ater turnover, lo w sulfate concentr ations ar e expected to be r a pidl y consumed, allowing methanogens and acetogens to become the dominant metabolisms in a second phase.In the context of natural gas storage, annual monitoring at site Ab_L_1, the interface between stored natural gas and formation water, between 1992 and 2017 showed that increased microbial activity had reduced the sulfate concentration from 18 mg l −1 to less than 7 mg l −1 (190 to 73 μM; Ranchou-Peyruse et al. 2019 ).This increase in sulfate-reducing activity was explained by the solubilization of organic molecules present in the natural gas and available to heter otr ophic micr oor ganisms pr esent in the water of the oligotrophic aquifer.This same research article also suggested that the effect of a massi ve arri val of H 2 in such an ecosystem could impact microbial diversity, and by indirect effect on the physico-chemistry of water by maintaining low sulfate concentrations in particular, ov er se v er al decades; and this e v en when the H 2 storage was finished.

Final microbial taxonomic di v ersity of cultiv a tion trials with H 2 /CO 2
Pr okaryotic taxonomic div ersity was studied at the end of incubation in biotic assays with a gas phase consisting of H 2 or H 2 /CO 2 .These biomethanation conditions str ongl y selected for micr obial comm unities, as pr e viousl y r eported (Bellini et al. 2022 ).In order to test a large number of conditions, it was decided not to use cultur e r e plicates for taxonomic di v ersity anal yses, whic h can make it difficult, if not impossible, to inter pr et the e volution of complex micr obial comm unities.Bearing this limitation in mind, we can only note the astonishing maintenance of a few pr okaryotic gener a pr esent on all the sites tested, and that it is not possible to draw general conclusions on community behavior and changes without appropriate re plication.The relati ve abundances of the 50 dominant ASVs obtained fr om high-thr oughput sequencing of the 16S rRNA genes of the different communities ar e r epr esented in the form of a heatma p (Fig. 5 A).Although eac h condition tested was only in a single replicate for taxonomic div ersity anal yses, hydr ogenotr ophic methanogenesis was systematically carried out by members of the Methanobacteriaceae famil y, whic h includes the genera Methanobacterium and Methanobrevibacter , and the Methanothermobacteriaceae family (i.e.Methanothermobacter spp.), as confirmed by analysis of the mcrA genes in the same samples ( Fig. S2, Supporting Information ).The corresponding 16S rRNA gene transcripts sho w ed activity until the end of incubation (Fig. 5 B ).These archaeal families are regularly highlighted in microbial communities in deep aquifers (K otelniko va et al. 1998, Ma et al. 2019, Kadnikov at et al. 2020, Ranchou-Peyruse et al. 2019, 2021, Molik ov á et al. 2022 ) and wer e assumed to be responsible for in situ biomethanation in the case of town gas stor a ge at the Lobodice site (Czechia; Buzek et al. 1994, Molik ov á et al. 2022 ).The growth conditions interfered with the r epr esentativeness of ASVs but ultimately had little influence on the results at the genus le v el.In the majority of trials, methanogenesis was carried out by members of the genus Methanobacterium (ASV16-16S , ASV4-16S , ASV13-16S , ASV25-16S , ASV46-16S , ASV48-16S, and ASV58-16S).In samples Ab_L_7, Ab_L_10, Ab_Y_2, and Ab_P_1, members of the Methanothermobacter genus were also repr esented.Their pr esence is unexpected, because of formation water temper atur es at the bottom of the wells range from around 37 • C to 40 • C.These temper atur es ar e deduced fr om temper atur e gr adient measur ements taken during logging oper ations (Gal et al. 2021 ).In 2019, archaea belonging to the Methanopyraceae family, a group of exclusively hyperthermophilic microorganisms, were identified at sites Ab_L_1, Ab_L_3, Ab_L_7, and Ab_L_10 (Ranchou-Peyruse et al. 2019 ).Faults allowing fluid circulation between the different superimposed aquifers could explain these results in the context of a sedimentary basin str ongl y impacted by the proximity of the Pyrenean mountain range and could explain the frequent detection of a priori strictly thermophilic organisms in the shallo w er mesothermal aquifers.Ho w e v er, this hypothesis does not explain why thermophilic micr oor ganisms could be active and thrive at temperatures so far from these optima, even in this study with an incubation temper atur e of 37 • C (Fig. 5 ).We hypothesize that these arc haea ar e eurythermal or simply mesophile.
The same was true of the order Thermotogales , which includes thermophiles .En vironmental sequences of this order had been detected in mesothermal en vironments , such as a UGS in the Paris  For the results based on the 16S rRNA gene and its transcripts, while sulfate reducing conditions were constant in all cultivation trials, eac h enric hment cultur e seemed to be exclusiv el y dominated by a phylogenetic group of sulfate reducers such as the genera Desulfovibrio (Ab_L_3, Ab_L_8, Ab_Y_2, and Ab_P_1), Thermodesulfovibrio (Ab_L_10) or Desulforamulus (Ab_L_1, Ab_L3, Ab_L_7, and Ab_L_10), suggesting competition between these different taxa (Fig. 5 ).The diversity of this group based on the dsrB gene ( Fig. S3 , Supporting Information ) is more nuanced but could be explained by the persistence of spores in the assays and the greater specificity of the primers targeting the dsrB gene than the more generalist primers targeting the 16S rRNA gene.This presumed higher specificity would also explain the detection of genera not identified by 16S rRNA -based approaches ( Desulfosporosinus , Desulfosarcina , Desulfobulbus , and LA-dsrAB), and therefore justifies the systematic use of the dsrB gene for the study of this functional group.Sporulating sulfate reducers are regularly found in deep continental environments and often described as lithoautotrophic (Aüllo et al. 2013 ).These bacteria are represented in all trials by one or two phylogenetic groups close to the genera Desulfosporosinus , Desulfotomaculum , and Desulforamulus or e v en the LA-dsrAB gr oup (Müller et al. 2015 ).While these sulfate r educers hav e alr eady been identified in this aquifer, some have also been identified in other UGS sites in aquifers, such as members of the Desulforamulus genus and micr oor ganisms close to the strain formerly named Desulfotomaculum profundi Bs107 (Aüllo et al. 2016, Berlendis et al. 2016 ).In addition to these micr oor ganisms, others persist in these simplified communities and hav e alr eady been identified in a pr e vious study carried out on this aquifer ( Burkhoderiaceae , Pseudomonadaceae , and Rhizobiaceae ; Ranchou-Peyruse et al. 2023 ).Their survival can be explained by a fermentative metabolism, as in the case of members of the genus Pseudoclostridium (ASV44-16S) and the phylogenetic group DTU014 (ASV68-16S; Dyksma et al. 2020 ).Under the conditions studied, no ASVs could be matched to any of the homoacetogenic bacteria pr e viousl y described.

Microbiological assessment of dedicated H 2 storage in the lower Eocene
The formation water for well Ab_P_1 comes from a reservoir located at a lo w er le v el than the aquifer curr entl y used as a UGS site, which itself evolved in a geological layer dating from the Eocene (Fig. 1 B ).A r oc k sample from the same horizon as Ab_P_1, at the boundary of the Eocene and Dano-Paleocene, was obtained and used in the tests in place of calcite .T his r oc k is composed of 63% quartz, 13% calcite, 16% clay, and 7% pyrite (DRX/FluoX analysis, TEREGA data).During the first sampling campaign, this formation water had one of the highest concentrations of methanogenic archaea, with 2.62 × 10 2 ± 5.7 × 10 1 mcrA gene copies ml −1 .The methane production from Ab_P_1 formation water in the presence of H 2 /CO 2 gas (80/20; 1 bar) was among the highest and did not increase in the presence of rock (Fig. 4 ; Fig. S4, Supporting Information ).After 2 months of r oc k-fr ee incubation with a gas phase composed of H 2 /CO 2 , the test carried out with formation water from site Ab_P_1 showed a production of 3.3 × 10 −1 mmol of CH 4 in 2 months with a total consumption of 3.8 mmol of H 2 .With r oc k incubation, H 2 consumption almost doubled (6.4 mmol) and CH 4 pr oduction decr eased (1.4 × 10 −1 mmol), r e v ealing incr eased activity of metabolisms other than methanogenesis .Con v ersel y, when the gas phase was composed solely of H 2 (1 bar), the yield was among the lowest.For the other sites, the highest pH values were obtained in the absence of CO 2 in the gas phase and were associated with the lo w est Ca 2 + and HCO 3 − concentrations ( Fig. S4 and Table S2 , Supporting Information ).The results presented in Fig. S5 ( Supporting Information ) clearl y illustr ate the str ong similarity between the taxonomic diversity obtained from the 16S rRNA genes and that obtained from their transcripts.For batch cultures with very limited available nutrients, this result is interesting, as it suggests a restructuring of the microbial community with strong recycling of the necromass constituted by micr oor ganisms that are not adapted to the experimental conditions and leave no remnant DN A. F rom an initial state mainly dominated by sporulating sulfate-reducing Firmicutes affiliated with the Desulfurispora genus, the communities wer e subsequentl y all dominated by hydr ogenotr ophic methanogenic archaea belonging to the Methanobacteriaceae or Methanothermobacteriaceae families .T he results suggest that the members of Methanothermobacteriaceae are not all thermophilic since the environmental factor selecting them was not temperatur e, but r ather the acid pH induced by the addition of CO 2 associated with one or more nutrients released into the rock.
We note that while the addition or nonaddition of calcite or r oc k did not have any effect on the structuring of microbial communities based on H 2 and CO 2 consumption and production and dominated by methanogens and sulfate reducers.Regarding calcite, r oc k, or e v en barite supplementation, the diversity may differ between communities at the ASV level, but this variation is very low, or even nonexistent, at the microbial genus level.Howe v er, these miner als r epr esent a carbon source (calcite dissolution), sulfate source (barite dissolution), and buffer for microorganisms, they had little impact on the structure of the sulfater educing functional gr oup and none on that of methanogens.These results suggest that the ecological valence of these micr oor ganisms is stronger than expected.For example, members of the genus Methanobacterium show activity at pH values ranging from ∼6 (conditions with H 2 /CO 2 ) to around pH 10.While alkalinization is often associated with methanogenesis and sulfate r eduction, a shar p incr ease in pH has been shown to be r esponsible for the cessation of methanogenesis .Here , methane production yields were lower when the gas phase was composed solely of H 2 (without CO 2 ), e v en in the presence of calcite as an indirect carbon source.On deep aquifers with miner alogicall y mor e complex reservoir rocks, a recent study experimentally simulating H 2 injections into a high-pr essur e thr ee-phase r eactor (gas-r oc kwater) with indigenous micr oor ganisms suggested similar alkalinization during physicochemical modeling (Mura et al. 2024 ).Her e, the r oc k of the aquifer studied is essentially composed of quartz (81%), while calcite was estimated at around 12% (Haddad et al. 2023 ).It is reasonable to assume that the buffering effect of the aquifer r oc k is greater than that of our test media, but over the lifetime of such a storage facility (i.e.several decades), it seems likel y that miner als suc h as calcite will be almost completely dissolv ed, giv en their low concentrations.On the other hand, bearing in mind that e v en at the highest pH and based on the study of the 16S rRNA , dsrB , and mcrA genes transcripts, methanogenic archaea continued to be active, we hypothesize that the low methane yields may be more related to a limitation of CO 2 solubilization rather than to a deleterious effect of pH on the physiological activity of the hydr ogenotr ophs pr esent.

Conclusion
As the first study of its kind on this aquifer, whic h serv es as a UGS for natural gas, these experiments are intended to assess the hydr ogenotr ophic potential of indigenous communities in general, and of methanogens in particular.Interestingly, it was shown that the hydr ogenotr ophy ca pacity linked to sulfate reduction was pr esent ov er the entir e aquifer used as UGS and hydr ogenotr ophic methanogenesis was only present near the current natural gas stor a ge.It is obvious than these batch experiments at atmospheric pr essur e under estimate H 2 consumption because of its low disso-lution and low quantity available for the microbial gro wth.Ho we v er, this study has made it possible to identify certain sites and conditions to be tested from now on under conditions closer to r eality (high pr essur e , monitoring o ver time , and so on) in order to determine H 2 consumption (or e v en CO 2 ) and methane and sulfide production yields, and to assess the economic relevance of a future UHS in this deep aquifer.Finally, the strong alkalization initiated by lithoautotrophic microbial metabolisms is a k e y parameter to take into consideration.In the context of a UHS sure, this phenomenon could consider abl y curb microbial consumption of H 2 by mineralizing CO 2 dissolved in carbonates and thus making this CO 2 inaccessible to autotr ophic micr oor ganisms.In the context of in situ biomethanation, alkalinization could be counterbalanced by CO 2 coinjection, enabling active in situ biomethanation to be maintained.

Figure 1 .
Figure 1.Representations of the deep aquifer targeted for study.(A) Structural map of the aquifer, (B) 2D seismic section of the geological layers constituting the aquifer used as a UGS site along the broken line shown on map A. The interpreted data were acquired from a 3D seismic campaign as well as from older 2D seismic lines acquired in the storage influence zone during oil exploration.For each well, depths are indicated in meters above mean sea le v el (AMSL).

Figure 2 .
Figure 2. Comparison of prokaryote (bacteria and archaea) quantifications in the 17 formation waters sampled from the three levels of the aquifer.Concentrations of prokaryotes, sulfate reducers, and methanogens were estimated by qPCR in copy numbers per milliliter of water of the 16S rRNA , dsrB , and mcrA genes, r espectiv el y. * : formation water that showed methane production after incubation in the presence of H 2 /CO 2 (80/20; 1 bar).

Figure 4 .
Figure 4. Monitoring of the gas phase evolution in microbial tests on the seven formation waters with and without calcite.Part 1: tests in the presence of a H 2 /CO 2 gas phase (80/20; 1 bar); Part 2: tests in the presence of a H 2 gas phase (1 bar).Test Ab_Y_2 also featured an additional condition with added barite (BaSO 4 ).Test Ab_P_1 was carried out with aquifer r oc k r ather than calcite.C: abiotic contr ols; A/A': trials used for molecular biology anal ysis; B: trials used for physicochemical analysis.X%/Y%: written on histograms; X% indicates the theoretical percentage of H 2 consumed by methanogens as a function of the number of mmol of CH 4 produced based on the hydrogenotrophic methanogenesis reaction (4H 2 + CO 2 → CH 4 + 2H 2 O); Y% indicates the theoretical percentage of H 2 consumed by other hydrogenotrophic microorganisms (total H 2 disappeared-H 2 consumed by methanogens).

Figure 5 .
Figure 5. Taxonomic diversity of prokaryotes based on the 16S rRNA gene in different enrichment cultures at the end of incubation in the presence of H 2 .(A) Heatmap representing taxonomic diversity results based on the 16S rRNA gene.The 50 dominant phylotypes, r epr esenting between 86% and 99% of sequences in each culture trial, are indicated.(B) Heatmap showing the taxonomic diversity results based on 16S rRNA gene transcripts .T he 50 dominant phylotypes, r epr esenting between 83% and 99% of sequences in eac h cultiv ation trial, ar e shown.H 2 : incubation with H 2 in the gas phase; H 2 CO 2 : incubation with H 2 /CO 2 (80/20; 1 bar) in the gas phase; C: incubation with calcite; Wt: incubation without calcite; R: incubation with r oc k; WtR: incubation without r oc k; and Ba: incubation with barite.