Due to magmatic activity beneath the Cheb Basin, large-scale degassing of mantle-derived CO₂ (>99%) occurs. The diffuse cold gas emanations at the surface (diffuse degassing structures, DDS) can be distinguished as dry and wet mofettes (Kämpf et al., 2013). Mofettes provide insights into life under elevated CO₂ concentrations, low pH, and anoxic conditions comparable to the Earth’s ancient atmosphere (Emiliani, 1992; Raven, 1995; Young et al., 2012). Moreover, mofettes are used as model ecosystems for studying the response of soil microorganisms to a potential CO₂ leakage of underground carbon capture and storage systems (Krüger et al., 2009, 2011; Frerichs et al., 2013; Morales and Holben, 2013).

CO₂ degassing leads to hypoxia and acidification of the soil (Beaubien et al., 2008; Blume and Felix-Henningsen, 2009; Rennert and Pfanz, 2016). Additionally, an increase in metal mobilization was observed, which may affect the availability of soil nutrients (Mehlhorn et al., 2014, 2016). As shown by several studies, these direct influences of elevated CO₂ concentrations on the environment are affecting the mofette biota and its biological matter cycling.

First studies on the effects of CO₂ on the soil biota were focused on the plant vegetation. These studies point to a decelerated growth, an increased plant C/N ratio but also physiological adaptations that provide advantages in hypoxic or even anoxic environments (Pfanz et al., 2004; Vodnik et al., 2007; Rennert and Pfanz, 2016). A common feature of mofette soils are increased carbon and nitrogen contents (Ross et al., 2000; Rennert et al., 2011). By δ¹³C analyses of plant and microbial lipids, Oppermann et al. (2010) demonstrated that within a CO₂ vent in the Latera Caldera (Central Italy) a substantial amount of geothermal CO₂ is incorporated into the microbial, plant, and soil carbon pools. Recently, Beulig et al. (2016) performed radiocarbon analyses and showed that up to 67% of mofette soil carbon content originated from the assimilation of geogenic CO₂ via plant primary production and microbial CO₂ fixation. The authors pointed out that the almost undegraded organic material found in the mofette soil is facilitated by the permanent exclusion of meso- to macroscopic eukaryotes rather than an impaired biochemical potential for soil organic matter decomposition. Complementary, Nowak et al. (2015) estimated through combined δ¹⁴C and δ¹³C isotope mass balances that around 8–27% of the bulk soil organic matter (SOM) derived from microorganisms. DNA stable isotope probing allowed the identification of chemolithoautotrophic microorganisms such as methanogenic archaea and acetogens as well as sulfate reducing bacteria (SRB) to be involved in the assimilation of CO₂ (Oppermann et al., 2010; Beulig et al., 2015). However, a quantification of cbbL genes, encoding for the large subunit of RuBisCO, a carboxylase which is of crucial importance for carbon assimilation in chemolithoautotrophic microbes, revealed that only a part of the autotrophic CO₂-fixing microorganisms could adapt to the very high CO₂ concentrations found in a mofette in Slovenia (Videmšek et al., 2009).

In the last decade, several studies have focused on the impact of elevated CO₂ concentrations on the microbial community structure in mofettes. A shift to anaerobic/microaerophilic and acidophilic community compositions has been reported in CO₂ mofette soils located near the Laacher See (Germany) (Krüger et al., 2009; Krüger et al., 2011), the Cheb Basin (Czech Republic) (Beulig et al., 2015), and in Stavešinci (Slovenia) (Videmšek et al., 2009). Furthermore, at the Laacher See site, the abundance of several functional and group-specific gene markers revealed a decrease in Geobacteraceae and an increase in SRBs in the vent center and biomarker analysis revealed a predominance of Thaumarchaeota as possible indicator organisms for elevated CO₂ concentrations in soils (Frerichs et al., 2013). Also for a mofette in Latera, Italy, it was shown that strictly anaerobic SRBs are abundant in mofettes, whereby the ATP biomass and total bacterial cell counts decreased (Beaubien et al., 2008). The highest sulfate reducing activity was observed in the center of the vent. Also Methanogenic archaea showed higher activities in the center of the vent compared to a transition zone site. Contrary to the results from Frerichs et al. (2013) in the Laacher See mofette, Geobacteraceae were in Latera mainly found at the CO₂ vent and only minor quantities were found at the reference site (Oppermann et al., 2010).

So far, only a few population datasets from dry mofettes are based on high throughput sequencing. The 454 pyrosequencing analyses of 16S rRNA genes from a natural mofette in central-southern Spain revealed that community richness, evenness, and diversity decreased with increasing CO₂ flux (Sáenz de Miera et al., 2014). An increase in abundance was thereby observed for OTUs related to the Chloroflexi phylum. Interestingly, Beulig et al. (2015) also showed that besides Methanoregulaceae, unclassified Chloroflexi might be involved in acetogenesis by DNA-SIP. The second 16S rRNA pyrosequencing dataset was established by Beulig et al. (2015) for a mofette located in the Plesná floodplain in the Cheb Basin. The community was dominated by methanogens (e.g., Methanosarcinales and Methanomicrobiales) and subdivision 1 Acidobacteria, which likely thrived under stable hypoxia and acidic pH. Recently, Krauze et al. (2017) investigated wet mofettes in the Cheb Basin by Illumina 16S rRNA amplicon sequencing and found a unique microbial community highly adapted to the anoxic conditions in the mineral and thermal waters and highlighted the connection between the groundwater or mineral waters and the deep biosphere. Deeper insights into the influence of CO₂ on the microbial community within the top 40 cm of a mofette in the Cheb Basin were gained by a metatranscriptomic and metagenomic approach (Beulig et al., 2016). One outcome of the study was that transcripts related to methanogenesis (mcr) and sulfate reduction (dsr, cys, apr) were remarkably increased in frequency.

However, until now, microbiological studies have focused only on near-surface soil layers (<70 cm) of dry mofettes, and deeper sediments have not yet been analyzed in detail. The present study aims to characterize the influence of elevated CO₂ concentrations inside a CO₂ conduit on the geochemistry, microbial abundance, and community composition. Furthermore, this study intends to determine significant community-shaping environmental factors. To study the influence of mantle-derived CO₂, two 3-m drillings were performed, one of which was located in the center of the DDS and the other at the undisturbed border of the mofette field. To unravel the community structures a high resolution sampling (every 5 to 10 cm) and Illumina 16S rRNA gene amplicon sequencing was conducted.

Dry mofettes – such as the HMF – allow for an investigation of geo–bio interactions that result from the permanent degassing of mantle-derived CO₂. In the uppermost soil layers – which have been the focus of other studies (Krüger et al., 2009; Blagodatskaya et al., 2010; Sáenz de Miera et al., 2014; Beulig et al., 2015) – the conditions are not necessarily permanent strictly anoxic, which also allows aerobic or microaerophilic heterotrophic microorganisms to grow. The input of oxygen can derive from meteoric water as well as horizontally via groundwater flow. The sampling depth and lithological profile are therefore of major importance in unraveling the influence of the CO₂ on the microbial communities. For the first time, the present study provides insights into the community structure in 3-m-deep sediments of a mofette and a nearby reference site. The combination of geochemical analyses and Illumina MiSeq high-throughput sequencing of 16S rRNA genes reveals the complexity of geo–bio interactions in the CO₂-influenced habitat.

We observed a strong influence of the emanating CO₂-rich fluid on crucial soil parameters, such as pH, water content, and ion composition (Figure 2). The relatively low pH may be an indication for the influence of CO₂ also at the reference site, and reflects the characteristics of the lithological profile. However, the CO₂ flux measurements at the surface of the reference site did not reveal any mofette activity; this is supported by the typical TOC profile and low sulfate concentration and conductivity. Both sampling sites were located in a wet land area (floodplain) with a high groundwater level. We assume that depending on the season (and flooding) an influence by CO₂-rich groundwater at both sites is possible. It is known that the groundwater level at the HMF may shift around 20 cm per day (Nickschick et al., 2017). Within the first 80 cm of the depth profile the pH of the reference site (pH > 4) was higher than at the mofette site (pH < 4). At the reference site the groundwater level is situated within a gravel-rich sand layer at a depth of 80 to 100 cm. The groundwater, which may migrate horizontally, presumably causes a decrease of the pH due to dissolved CO₂. Thereby, the fine- to medium-grained clayey sand in the reference site might function as a natural vertical barrier for the groundwater and CO₂ flow, especially in depths deeper than the core sequence. In the mofette, the groundwater level was shallower (25–70 cm) than at the reference site. On the other hand the pH in the mofette was highest in the gravel-rich layer between depths of 130–140 cm. The high conductivity and sulfate concentration (Figure 2) may indicate that mineral water from greater depth is admixed with groundwater in the mofette. Mineral water found in the same mofette in a depth of ca. 82 m had a pH of 6.4 and a sulfate concentration 1470 mg L^-1 (Bussert et al., 2017).

The different microbial community structures found in each depth and each core at the reference site and in the mofette cannot be explained by the variances in the lithology of the core material, however, the lithological setting is from a major importance for both habitats in greater depth (78 m) where a cap-rock like carbonate-rich layer largely seals the CO₂-rich aquifer and allows for a channelized CO₂ degassing (Bussert et al., 2017). Instead, the pH, TOC, and sulfate concentration formed the optimal subset of parameters to explain the abundance and distribution of the taxa at both sites. Multivariate statistical analyses and qPCR results revealed that at the reference site, the microbial abundance was positively correlated with pH (R = 0.709, p = 0.0002), while a significant negative correlation was observed at the mofette site (R = -0.655, p = 0.0009; Figures 2C,G and Supplementary Tables S5, S6). Therefore, several taxa of the microbial community at the mofette site seem to be adapted to the acidic conditions. However, the diversity and the evenness at the mofette site are positively correlated with the pH (R = 0.582, p = 0.0044, and R = 0.550, p = 0.008; Supplementary Tables S5, S6), indicating that it is mainly specialists that can withstand the extreme environmental conditions. It is remarkable that solely the pH value explains 38.7% of the distribution pattern of taxa at the mofette site (Figure 7A). Since the pH value at the reference site is hardly affected by the emanating CO₂, it only explains 16.3% of the OTU distribution (Figure 7B). However, since the pH is generally co-correlated with CO₂ concentration, it is necessary to consider the fact that the abundance of autotrophic microorganisms in the mofette site is also positively correlated with the amount of CO₂-rich emanating fluids. The results of the geochemical analyses and the 16S sequencing lead to the assumption that in addition to pH, sulfate also plays an important role in microbial matter cycling under anaerobic conditions in the mofette site. The sulfate concentration – which is five to fifteen times higher in the mofette site – explains 12.7% of the distribution pattern of microorganisms at the mofette (Figure 7A). In comparison with the mofette, the sulfate concentration at the reference site explains only 8.4% of the community structure (Figure 7B).

In comparison with the reference site, the TOC values were up to 100 times higher in the peat layers of the mofette (Figures 2A,D). While pH and ion concentrations are directly influenced by the emanating CO₂, the high TOC content at the mofette site can be explained by the consequences of acidification and anoxia on mesoscopic or macroscopic eukaryotes involved in the degradation of complex organic matter; additionally it was shown that acetogenesis is a prominent process in mofettes (Beaubien et al., 2008; Beulig et al., 2015, 2016; Fernández-Montiel et al., 2016). The TOC values explain 11.4% of the distribution pattern at the reference site (Figure 7B) and only 4.5% at the mofette (Figure 7A). Moreover, the top 10 abundant taxa at the reference site were heterotrophic microorganisms (Supplementary Tables S8, S9). In contrast, the high abundance of chemolithoautotrophic taxa in the mofette highlights the importance of an autotrophic- rather than a heterotrophic lifestyle in habitats with strongly elevated CO₂ concentrations (Oppermann et al., 2010). Autotrophic microorganisms can fix significant amounts of carbon from geogenic CO₂ (Nowak et al., 2015), whereas organic substrates remain undegraded and accumulate in the sediment. An interesting side effect can be seen in a long-term perspective because such an enormous accumulation of organic substances in sediments may be the precondition for the development of a paleo organic layer at a late stage.

The bacterial 16S rRNA gene copy numbers in the mofette varied from 10⁵ to 10⁹ copies g^-1 of sediment. The highest gene copy numbers were found in the topsoil of the mofette (1.2 × 10⁹ gene copies g^-1 at a depth of 12 cm; Figure 2G). While the abundance of microorganisms at the reference site followed a classical trend and decreased with depth, the abundance of microorganisms in the mofette did not decrease with depth beyond 50 cm. The present study suggests for the first time that at a depth greater than 100 cm, the microbial abundance in the mofette even partially exceeded that of the reference site. This trend was also observed for SRB and methanogenic archaea (Figures 2H,I). As discussed later, presumably the uprising CO₂-rich fluid feeds the ecosystem from underneath with substrates such as CO₂ and sulfate. However, compared with the reference site, the gene copy numbers at the mofette were about one order of magnitude lower for the uppermost 100 cm. Higher cell numbers in the surface soil of the reference site can be explained by the predominance of aerobic processes and a rather moderate pH. In general, the gene copy numbers per gram of subsurface sediment at the HMF were similar to soils with highly elevated CO₂ concentrations (10⁹ to 10¹⁰ copies g^-1 sediment) (Beaubien et al., 2008; Krüger et al., 2009; Oppermann et al., 2010; Frerichs et al., 2013; Beulig et al., 2015; Fernández-Montiel et al., 2016). The top 40 cm of the mofette site at the HMF was an exception, for here, the gene copies were one order of magnitude lower (10⁸ to 10⁹ copies g^-1 sediment) compared with the Plesná floodplain site, which is located 1.8 km to the NW (Beulig et al., 2015). Reasons for the discrepancy could be the higher CO₂ flux at the HMF or different soil characteristics (Kämpf et al., 2013). Additionally, seasonal effects have to be considered. In April (Beulig et al., 2015), the surface water level is often higher, and substrates might be more easily accessible in comparison with September, when the HMF drilling took place.

Both diversity and evenness were lower at the mofette than at the reference site (Figures 2J,K). A lowered diversity was also observed at the La Sima mofette in Spain (Sáenz de Miera et al., 2014). The low evenness indicates a dominance of specialists, such as anaerobic acidophilic and acido-tolerant taxa in the mofette community. In the mofette, the alpha diversity and the evenness were higher at depths between 85 and 125 cm. This shift can be explained by the surface water table and the admixture of surrounding aerobic communities.

The results of the beta diversity of the community highlight the assumption that only the deeper sediments are almost unaffected by oxygen. The PCoA plot of the weighted UniFrac distance metric revealed that the mofette harbors two distinct communities (Figure 4). One cluster includes communities from shallow depths between 0 and 95 cm (Cluster A), and the other cluster includes communities in the deep anoxic sediments from 100 to 275 cm (Cluster B). The microbial communities from the reference site cluster apart from all mofette samples, which highlights the fact that the mofette community is strongly influenced by the degassing phenomenon. Site-specific effects – such as a low pH and changes in ion composition – do not sufficiently explain the differences between Cluster A and Cluster B, but the oxygen availability could be identified as the major stress factor. In the mofette, the oxygen-dependent depth gradient was clearly reflected in the antagonistic shift of the relative abundance of aerobic and anaerobic taxa in Cluster A and Cluster B. The relative abundance of obligate and facultative anaerobic microorganisms affiliated with Bacteroidetes, Bathyarchaeota, Caldiserica, and Parcubacteria gradually increased with depth. At the same time, the relative abundance of versatile Proteobacteria strongly decreased with increasing depth (Figure 3). Strictly anaerobic and facultative aerobic microorganisms affiliated with Acidithiobacillus, Clostridiaceae 1, Bacillus, and Ignavibacteriales that were observed in Cluster A were much more abundant at the mofette than at the reference site. In Cluster B, strictly anaerobic microorganisms within the taxa Thermoanaerobaculum, Bacteroidetes vadin HA17, Desulfobacca, Caldisericum, Anaerolineaceae, and Sulfurovum shaped the community and explained most of the differences (14.5% dissimilarity) between the communities of the deep sediments of the mofette and reference site (Figures 5A,B). Neither in Cluster A nor Cluster B were any of the strictly aerobic taxa more abundant in the mofette than at the reference site, except for Acidobacteriaceae (Subg. 1). Additionally, the family Acidobacteriaceae (Subg. 1) was the most abundant taxa in both study sites (mofette: 12.4%; reference site: 10.5%). Members of this group are heterotrophic, aerobic, or microaerophilic, and some species are facultative anaerobes (Pankratov et al., 2012). Therefore, the occurrence at both sites under both aerobic and anaerobic conditions is plausible. The HMF multivariate statistics indicate a strong negative correlation of the taxon with pH, and the same correlation has been observed by Männistöet al. (2007) in tundra soils. Moreover, the proportion of unassigned taxa (4.0%) was much higher in the mofette compared with the reference site, indicating that mofettes are unique environments that harbor a large fraction of novel, as-of-yet undescribed organisms.

The sulfate-rich mofette with a pH < 4.0 offers ideal growth conditions for obligate acidophilic bacteria such as Acidithiobacillus (Kelly and Wood, 2000). Accordingly, these bacteria’s abundance is positively correlated with the sulfate concentration (R = 0.699, p = 0.0003; Supplementary Tables S5,S6). In both clusters, taxa potentially involved in sulfur cycling had a more substantial proportion (14.4%) and a higher diversity (e.g., of Desulfobacca, Desulfosporosinus, Sulfurovum, and Sulfuriferula) in the mofette in comparison with the reference site (3.0%; Figure 6A). Desulfobacca was also recently found in a Plesná floodplain site (Beulig et al., 2016). A higher abundance of SRB was also observed in CO₂ vents (Beaubien et al., 2008) and CO₂-affected soils (Frerichs et al., 2013). Striking differences between the mofette and reference site were also observed for taxa potentially involved in iron-cycling. The results lead to the assumption that in addition to sulfate reduction, iron-cycling is also an important feature under aerobic conditions at the reference site and under anaerobic conditions in the mofette. The aerobic or microaerophilic taxa – such as Gallionella and Sideroxydans (Hallbeck and Pedersen, 1990; Emerson and Moyer, 1997) – were more abundant at the reference site, whereby acidophilic obligate and facultative anaerobic genera – such as Ferrithrix and Ferrovum (Johnson et al., 2009; Johnson et al., 2014) – dominated in the mofette (Figure 7B).

The extreme environmental conditions in the mofette also favor the growth of methanogenic archaea. Methanosphaerula (Cadillo-Quiroz et al., 2009) occurred solely in the mofette, whereby other genera – such as Methanoregula (Bräuer et al., 2011), Methanosaeta (Patel and Sprott, 1990), and Methanosarcina (Zinder et al., 1985) – were abundant at all depths in the mofette core but occurred only in low abundances close to the surface in the reference site, presumably because oxygen is spatially depleted by aerobic processes. Other archaea potentially involved in acetogenesis and methane cycling – such as Bathyarchaeota and ammonia-oxidizing Thaumarchaeota affiliated with South African Gold Mine Gp 1 (SAGMCG-1) (Takai et al., 2001) – were found at the mofette and the reference site in almost the same relative abundances (1.7 and 0.2%). These organisms have been found in marine sediments, deep aquifer waters, a CO₂ vent, as well as in water from wet mofettes and in thermal water in the Eger region (Kubo et al., 2012; Frerichs et al., 2013; Evans et al., 2015; Krauze et al., 2017).

Additionally, several other microorganisms found in the deep mofette sediments (e.g., Thermoanaerobaculum, Caldisericum, Sulfurovum, and Mobilitalea) have been isolated from hot springs, hydrothermal sediments, or thermal water from a 2.8-km-deep well (Oude Elferink et al., 1999; Inagaki et al., 2004; Losey et al., 2013; Podosokorskaya et al., 2014). These microorganisms most likely derive from the deep biosphere, indicating that the mofette is connected to the deep subsurface via ascending fluids. The environmental conditions in the mofette at 1–3 m in depth are similar to deep sediments or deeply originating thermal waters with respect to the availability of oxygen and ion composition and may therefore also provide adequate growth conditions for such taxa.

The microbial communities from both Cluster A and Cluster B at the HMF share many abundant taxa with the communities described by Sáenz de Miera et al. (2014) in the shallow sediments at the La Sima CO₂ gas vent (top 10–20 cm; Supplementary Table S12) and a CO₂-influenced floodplain site at Plesná (Beulig et al., 2015). Although the relative abundances of these common taxa display clear differences, they represent an outline of the characteristic core community of CO₂-influenced surface habitats. The occurrence of unshared site-specific taxa can be explained by the differing environmental conditions. For example, higher oxygen concentrations (<7%) (Peinado et al., 2009) and historical thermal anomalies might specifically trigger the growth of aerobic or facultative anaerobic thermophiles – such as members of Ktedonobacteria (e.g., Thermogemmatispora) – at the La Sima mofette. The high resemblance of the Cluster A communities to the microbial communities at the CO₂-influenced Plesná floodplain site (Beulig et al., 2015) lead to the assumption that the environmental conditions at both mofettes located in the Cheb Basin are rather similar compared with those the La Sima site. The low pH at the HMF and the floodplain particularly favors the growth of acidophilic taxa (e.g., Acidobacteria, Subg. 1), which have a rather low abundance at the La Sima site.

The high relative abundance of sulfate reducers, methanogens as well as further autotrophs indicate that the community is supported by hydrothermal originating substrates, delivered by the mantle-derived CO₂. The geochemical data demonstrate that electron acceptors such as sulfate and CO₂ are sufficient available. This is in good accordance with the findings from Beulig et al. (2016) who showed that in the mofette transcripts related to methanogenesis (mcr) and sulfate reduction (dsr, cys, apr) were remarkably increased in the frequency. In the Eger Rift, hydrogen, which is a key electron donor in the deep biosphere (Stevens and McKinley, 1995; Anderson et al., 2001; Chapelle et al., 2002; Nealson, 2005; Spear et al., 2005; Hinrichs et al., 2006), becomes available during radiolytic decay in the underlying fissured granite or stress-released during earthquake swarms (Bräuer et al., 2005, 2007).

Our study of a dry CO₂ degassing mofette in Hartoušov, NW Bohemia, as central part of a CO₂ conduit deepens the knowledge of geo–bio interactions in extreme environments with elevated CO₂ concentrations. The mofette ecosystem is characterized by anoxic conditions, a low pH, high TOC content and due to the admixing of mineral waters a relatively high sulfate concentration and conductivity. Our study shows that the exceptional environmental conditions provoke a decrease in diversity and favor the occurrence of anaerobic, acidophilic taxa, whereby sulfate reduction and methanogenesis become distinct processes. However, the deeper mofette sediments alone provide strictly anaerobic conditions, and accordingly, two distinct community clusters were found at different depth intervals. Electron acceptors such as CO₂ and sulfate are provided by the permanently ascending fluid which is admixed with deep thermal waters, thereby forming a kind of anoxic deep biosphere habitat close to the surface level. Hydrogen, originating from deep fissured granites may thereby function as an electron donor. This study is limited to taxonomical assignments, further studies focusing on deep mofette sediments should implement metagenomic or transcriptomic approaches to unravel the functional repertoire of the communities.

All authors have taken part in the interpretation of the results and writing the manuscript. QL wrote the manuscript, performed sampling, pore water analyses, DNA extractions and purification, gene quantification, and bioinformatical based statistical analyses. HK and TN performed CO₂ soil gas flux measurements. RB led the drilling campaign and performed sedimentological analyses. FH and PK were involved in statistical analyses. BP performed TOC and isotopic analyses. DW contributed to the interpretation of the results and valuable discussion. MA designed and supervised the study and led the writing of the present manuscript. All authors have taken part in the manuscript revisions and agreed with its scientific content.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.