Methane- and hydrogen-dependent prokaryotic deep biosphere at the Suwa Basin, Japan: impacts of hydrogeological processes on subsurface prokaryotic ecology at the boundary between the North American and the Eurasian Plates

Abstract

The subsurface biosphere hosts diverse prokaryotes whose metabolic activities and roles in biogeochemical cycles remain unexplored. Specifically, interactions between subsurface prokaryotes and environmental characteristics are poorly constrained. The Suwa Basin, Japan, is a suitable setting for investigating the impacts of hydrogeological features on subsurface microbiology, as it lies at the boundary between the North American and Eurasian Plates and is associated with geological events. This basin consists of sedimentary layers overlying bedrock. Previous studies have reported active methane seepage from the deep subsurface, presumably supporting the metabolic activities of subsurface prokaryotes. Moreover, faults and hot springs in the basin contribute to the complex subsurface geochemistry. Consequently, diverse methane-dependent ecosystems are expected to arise in response to heterogeneous subsurface conditions. We conducted geochemical and microbiological community analyses on groundwater samples from the sedimentary layer (10–100 m below the surface, mbs) and hot spring samples from bedrock region (max. 1000 mbs). The isotopic profiles (δ¹³C and δD) of methane indicated a primary microbial origin. However, subsurface community analysis revealed that methanogens were not significant at 10–100 mbs in the sedimentary layer, whereas aerobic methane-oxidizing bacteria were predominant. These results suggested the presence of an ecological niche for methanogens in deeper, reducing environments. The microbial communities in the hot spring samples were dominated by hyperthermophilic hydrogenotrophs. H₂ was contained in gas phase collected at hot spring sites (55.5 ppm), and its isotopic composition (− 736‰ VSMOW) suggested that H₂ generation was driven by tectonic activity. Subsurface hydrogeological processes were investigated based on the ¹⁴C concentration of dissolved inorganic carbon, revealing the significant intrusion of young groundwater from surrounding mountainous areas into the basin. The Li/Na molar ratios of the groundwater and hot spring water samples suggested the recharge of deeply derived hydrothermal fluids into the sedimentary layer. Considering the potential role of fault surfaces as conduits for vertical material transport, the unique geological setting of the Suwa Basin potentially supports the growth of prokaryotes in the sedimentary layer through an increased nutrient supply (e.g., H₂) from deep fluids. These findings provided new insights into subsurface methane-related microbial ecology and its hydrogeological controls.

1 Introduction

Prokaryotes (domains bacteria and archaea) are widely present in diverse environments on Earth. Through metabolic activities, prokaryotes drive chemical reactions that modify the properties (e.g., species and valence state) of elements (Falkowski et al. 2008; Rosenberg et al. 2013; Flemming and Wuertz 2019). Therefore, prokaryotes are a critical component of biogeochemical cycles in various habitats on Earth (Shu and Huang 2022; Templeton and Caro 2023). A significant proportion of prokaryotes resides deep within Earth’s terrestrial and oceanic crust, so-called deep biosphere (Whitman et al. 1998; Bar-On et al. 2018; Ruff et al. 2024). Despite its substantial biomass, the deep subsurface remains one of the last frontiers of microbial ecology (e.g., Jørgensen and Boetius 2007; Edwards et al. 2012; Colwell and D’Hondt 2013; Teske 2013; Inagaki et al. 2015; Hoshino et al. 2020; Morono 2023). Thus, it is crucial to elucidate the metabolic activities driven by subsurface prokaryotes and the environmental factors that affect them. In particular, subsurface lineages involved in methane dynamics are critically important, as their metabolic activities regulate the atmospheric methane concentration, thereby exerting a significant impact on the global climate.

Methane has a significant greenhouse effect with more than 27 times greater global warming potential than does carbon dioxide for the 100-year time horizon (Forster et al. 2021). While methane can be generated through abiotic processes, an archaeal group known as methanogens produces methane in subsurface reducing environments as the last step of buried organic matter degradation (Dean et al. 2018). Other subsurface lineages related to methane cycling, anaerobic methanotrophic archaea (ANMEs), and aerobic methane-oxidizing bacteria (MOBs), have been reported from subsurface environments enriched with methane (e.g., Knittel and Boetius 2009; Ahmadi and Lackner 2024; Reis et al. 2024). These microbes play a key role in mitigating methane emissions into the atmosphere by metabolizing methane as a carbon substrate (e.g., Thauer et al. 2008; Valentine 2011).

Over the past few decades, the establishment and continual improvement of next-generation sequencing techniques have significantly accelerated the accumulation of knowledge regarding the metabolic capacities of microbes in subsurface environments (e.g., Anantharaman et al. 2016; Momper et al. 2017). Since the cultivation of most subsurface microbes is technically challenging under laboratory conditions, culture-independent sequencing is principally advantageous and informative to the processes driven by prokaryotes in subsurface environments (e.g., Teske and Sørensen 2008; Imachi et al. 2020; Kapinusova et al. 2023). While sequencing data reveal equipped gene sets, their potential functions, and metabolic networks of microbes, demonstrating the adaptation of subsurface prokaryotes to environmental conditions and disturbances (e.g., geological events) based solely on sequencing data remains challenging. Therefore, the integration of other disciplines to provide direct evidence of microbial processes in the subsurface is important for assessing subsurface metabolism and its potential role in biogeochemical cycles (e.g., D’Hondt et al. 2019; Beulig et al. 2022; Mara et al. 2023). In this context, geochemistry serves as a key discipline; for example, light elements and isotope systematics offer crucial insights into what actually occurs in subsurface and deep aquifers (e.g., Ohkouchi and Takano 2014; Close 2019; Hoefs 2021 and references within; Ruiz-González et al. 2021; Lalk et al. 2023). Furthermore, geochemical factors are vital for understanding subsurface material cycles, which strongly influence microbial diversity and distribution (Shu and Huang 2022).

In this study, we investigated the subsurface microbial communities and geochemical/hydrogeological properties of subsurface environments at depths of 10–1000 m below the surface (hereafter “mbs”) of the Suwa Basin, central Japan. The basin is situated along the Itoigawa–Shizuoka tectonic line (ISTL), which is recognized as the boundary between the North American Plate and Eurasian Plate. The northern part of the ISTL hosts a strongly alkaline, H₂-rich, serpentinite-hosted hydrothermal field, the Hakuba Happo hot springs. Previous studies have described carbon and hydrogen dynamics, abiotic hydrocarbon synthesis during serpentinization reactions, and H₂-dependent lithotrophic microbial communities in this region (Suda et al. 2017, 2022; Nobu et al. 2024). In the southern part of the ISTL, a methane-rich subsurface biosphere inhabits the oceanic sedimentary layers of the accretionary prism at the convergent margin. Geochemical and microbiological analyses revealed regional variations in the origins of methane, including methanogenesis and thermogenic processes (Matsushita et al. 2016). In the central part of the ISTL, the Suwa Basin exhibits active biogenic methane emission from the deep subsurface (Urai et al. 2022) along with normal faults in the pull-apart basin, suggesting the potential for diverse methane-dependent prokaryotic communities adapted to geochemical conditions. However, microbial insights into the subsurface environment in the Suwa Basin remain an undescribed issue.

We considered the importance of the correlation between the two-component mixing of meteoric and geothermal fluids in deep aquifers and ongoing prokaryotic microbial activities (e.g., Colman et al. 2019; Bregnard et al. 2023; Beaver and Neufeld 2024). To investigate the relationships between the prokaryotic distribution and diversity in deep aquifers and the unique geological setting of the Suwa Basin, we analyzed microbial communities in groundwater and hot springs alongside geochemical characterizations of the subsurface. Furthermore, we evaluated the groundwater mixing process to assess the contributions of deeply derived hydrothermal fluids to subsurface communities. The present findings illuminated the vertical stratification of subsurface microbial diversity, with aerobic methanotrophs predominating in the sedimentary layer and anaerobic hydrogenotrophs dominating in the underlying bedrock layer. Our results suggested that H₂ production driven by tectonic activity in the bedrock layer played a crucial role in supporting subsurface microbial activity related to the methane cycle across the 10–1000 mbs region of subsurface environments.

2 Methods/experimental

2.1 Geological setting of the Suwa Basin

The Suwa Basin is in the central part of Honshu Island, Japan (Fig. 1). The geological characteristics of the basin are represented by its location at the intersection of the median tectonic line (MTL) and the ISTL, which is the boundary of the North American and Eurasian Plates. The basin is associated with left-lateral strike–slip faults along the ISTL and normal faults subsiding the basin (Fujimori 1991; Fujimori and Ota 1992; Yamazaki 1994). These faults enable the transport of gas, water, and dissolved chemical species from the deep subsurface to the surface. Such transport can be observed in the geysers and Kamisuwa hot springs located around Lake Suwa, a eutrophic lake located in the center of the basin. This feature contributes to a complex subsurface hydrogeological context, potentially leading to a variety of geochemical conditions throughout the Suwa Basin.

Fig. 1

© JAMSTEC. B Geological map of central Japan with the Itoigawa–Shizuoka tectonic line (ISTL), which represents the boundary between the Northeastern (NE) and Southwestern (SW) parts of Honshu Island, Japan. The Suwa Basin is at the region of the red square where the median tectonic line (MTL) and ISTL are crossing. The figure was modified after Panayotopoulos et al. (2010). Copyright 2010 Springer. C Geological map of Suwa Basin. The faults are indicated by red and black dotted lines. The locations of sampling sites are also shown. The figure was modified from “Seamless Geological Map of Japan at 1:200,000 (https://gbank.gsj.jp/seamless/).”—Ito: Itoigawa City; Hak: Hakuba Village; Mat: Matsumoto City; Suw: Suwa City; Kof: Kofu City; Shi: Shizuoka City; NFM: Northern Fossa Magna Rift basin; CUZ: Central Uplift Zone; Quat: Quaternary

A Geological settings of Japan with plate boundaries of the North American Plate, Eurasia Plate (EUP), Philippine Sea Plate (PSP), and Pacific Plate.

This basin is geologically composed of a Quaternary sedimentary layer deposited upon bedrock consisting of Quaternary andesite and Neogene granodiorite (Suwa city, private communication; Geological Survey of Japan 2023). The Suwa Basin is a pull-apart basin developed as a result of plate tectonics (Fujimori 1991). Surrounded by mountains rising to thousands of meters in height, the basin has a large catchment area, which has led to the deposition of an organic-rich layer over 380 m in thickness (Motojima et al. 1952; Oshima et al. 1997). This layer provides a high-resolution record of past depositional events and related environmental changes, such as climate change (Hatano et al. 2023). The layer consists of alternating beds of sand and mud, which may host deep aquifers and the prokaryotic biosphere.

In aquifers, past or ongoing methanogenesis by archaeal lineages has been proposed. Iwata and coworkers investigated the composition of gas bubbles steadily emitted from the bottom of Lake Suwa (Iwata et al. 2020) and measured the carbon isotope ratio of methane in seep gas. These results indicated the presence of a methane-rich gas phase, most likely originating from the biological methanogenesis of deeply buried carbon compounds (Urai et al. 2021b). A similar methane-rich gas phase is observed in non-lake areas of the basin (Miyabara 2012), suggesting the widespread occurrence of subsurface methanogenesis and a methane-dependent biosphere within the sedimentary layer.

2.2 Sampling procedure for water, gas, and microbes

Field sampling was conducted in June 2023 and July 2024 in the Suwa Basin. Groundwater samples from sedimentary layers were collected at three sites using observation wells and bailer samplers (Wells #1–#3) at targeted depths of 10 mbs, 50 mbs, and 100 mbs (Fig. 2a–c). The sampling sites were selected to be an intersection with the directions of faults. The Kamisuwa hot spring samples were obtained from eight sources (HSs #1–#8), with a maximum well depth of 1000 mbs (Fig. 2d). Detailed information, including locations, well depths, and sampling dates, is provided in Table S1.

Fig. 2

Photographs of sampling sites utilized in this study. A–C The observational wells used for the acquisition of groundwater samples from the sedimentary layer up to 100 mbs. The recordings of blowouts and degassing observed at wells are presented in Supplementary Movies. D The HSs #6 and #7 sources of Kamisuwa hot spring. E, F The sampling site of mountain spring water

We measured pH, redox potential (E_h), and electrical conductivity (EC) immediately after sampling using a pH/E_h meter (PH72; Yokogawa, Tokyo, Japan) and an EC meter (SC72; Yokogawa). To avoid alteration of these parameters during sampling and measurement due to degassing and oxidation induced by exposure to air, we employed a multiparameter water quality meter (ProDSS; YSI, OH, USA) to measure in situ physicochemical properties and their vertical profiles during the 2024 fieldwork.

Water samples for dissolved species analyses were filtered through hydrophilic PTFE membrane filters (pore size 0.2 µm; Advantec, Tokyo, Japan). To minimize oxidation during storage, aliquots for ion chromatography were stored without headspace, while those for inductively coupled plasma‒mass spectrometry (ICP–MS) were amended with ultrapure HNO₃ (Tamapure-AA-100, Tama Chemicals, Kanagawa, Japan). These samples were stored at 4 °C. Storage vials were treated with hydrochloric acid and nitric acid to remove inorganic salts prior to use. Prokaryotic cells were obtained from water samples by filtration through 0.3-µm pore-size glass fiber filters (GF-75; Advantec), and the filters were preserved at − 80 °C. For total cell counting, unfiltered water samples were collected and amended with formalin.

When gas emission was significant in the observational wells, gas samples were collected directly from the outlet of the well into vacuumed glass vials tightly sealed with butyl rubber stoppers. The gas emission rate was simultaneously measured using a gas flowmeter (ProFLOW 6000; Restek, PA, USA). When gas emission was not significant, groundwater samples were collected into vacuum-sealed vials through 0.2-µm pore-size filters to analyze dissolved gas and dissolved inorganic carbon (DIC). The dissolved gas species were analyzed using headspace gas of the vials. For the hot spring sources, gas samples were obtained via the water displacement method and stored in glass vials. These samples were pressurized with high-purity helium to prevent air contamination, amended with saturated mercuric (II) chloride (HgCl₂) solution to inhibit biological activities, and stored at 4 °C.

2.3 Analytical procedures

2.3.1 Cation and anion quantification

The concentrations of ionic species were quantified via ion chromatography (Metrohm 850 Professional IC system; Metrohm AG, Herisau, Switzerland) at Japan Agency for Marine–Earth Science and Technology (JAMSTEC) in Yokosuka, Japan. Cations were analyzed via a Metrosep C 6 - 250/4.0 column (Metrohm). The eluent was 8 mM ultrapure HNO₃ (TAMAPURE AA-100; Tama Chemicals) at a flow rate of 0.9 mL min⁻¹. Anion concentrations were measured via a Metrosep A Supp 19 - 250/4.0 column (Metrohm) and a chemical suppressor module (MSM; Metrohm). A mixture of 8.0 mM Na₂CO₃ and 0.25 mM NaHCO₃ was used as the eluent at a flow rate of 0.7 mL min⁻¹. Throughout the analysis, the column temperature was set at 40 °C, and an electrical conductivity detector was utilized. The quantification was performed using an external standard method with authentic standards.

The total alkalinity was measured using the Pack Test (WAK-MAL; Kyoritsu Chemical-Check Lab. Co., Kanagawa, Japan) and a portable multiparameter water analyzer (DPM-MTSP; Kyoritsu Chemical-Check Lab. Co.). The bicarbonate ion concentration was calculated from the total alkalinity and pH using the PyCO2SYS program with default settings (version 1.8.3.3; Lewis and Wallace 1998; Humphreys et al. 2022).

2.3.2 Trace element analysis

The trace element concentrations were quantified using a single quadrupole inductively coupled plasma mass spectrometer (iCAP-Qc; Thermo Fisher Scientific, MA, USA) at JAMSTEC via a previously described procedure (Yoshimura et al. 2021, 2023). The samples were diluted with ultrapure HNO₃ (TAMAPURE AA-100; Tama Chemicals) in polypropylene vials. The diluent was amended with internal standards (Be, Sc, Y, and Tl) to calibrate the instrumental drift during analysis. Additionally, we measured laboratory standards in every five samples during analysis for data correction. The analysis was performed in kinetic energy discrimination mode with helium gas.

2.3.3 Measurement of CH₄ and CO₂ concentrations

The concentrations of CH₄ and CO₂ in the gas and dissolved gas samples were measured using a gas chromatograph (7890A; Agilent Technologies, CA, USA) equipped with a thermal conductivity detector. The analyses were performed using a micropacked column (ShinCarbon ST 100/120 mesh, 2 m × 1 mm internal diameter; Restek) at a flow rate of 7.0 mL of helium carrier with the oven temperature programmed as follows: start at 100 °C for 2 min, ramp at 15 °C min⁻¹ to 300 °C, and hold for 6 min.

2.3.4 Stable isotope measurements for CH₄, H₂, and CO₂

The carbon and hydrogen stable isotopic compositions were measured for CH₄ and H₂ in gas and dissolved gas samples from the sedimentary layer and for gas samples from the Kamisuwa hot spring. The detailed pretreatment and analytical procedures were described elsewhere (Okumura et al. 2016). First, an aliquot of gas containing CH₄ was introduced into the gas preparation line filled with ultrapure helium using a gas-tight syringe. This helium was further purified using a Molecular Sieve 5A column held in a liquid N₂ bath prior to use. Within the preparation line, CH₄ was separated from associated CO₂ and H₂O using a stainless coil trap held at − 110 °C (ethanol/liquid-N₂ bath) and a chemical trap of Mg(ClO₄)₂ and Ascarite II (sodium hydroxide-coated silica; Thomas Scientific, NJ, USA). The purified CH₄ was then condensed using a trap filled with HayeSep-D porous polymer (60/80 mesh; Hayes Separations Inc., TX, USA) held at − 130 °C (n-hexane/liquid N₂ bath). CH₄ was subsequently released and re-concentrated on a capillary trap filled with PoraPLOT Q (20 cm long, 0.32 mm i.d.) held at − 196 °C (liquid N₂ bath) for cryofocusing. After release at room temperature, CH₄ was introduced into a gas chromatograph (6890; Agilent Technologies) equipped with an HP-PLOT Molesieve capillary column (30 m long, 0.32 mm i.d.) operated at 40 °C. The effluent CH₄ was converted to H₂ or CO₂ using pyrolysis or combustion units (Thermo Fisher Scientific) held at 1430 °C and 1000 °C, respectively. For isotope analyses, CH₄-derived H₂ and CO₂ were introduced into an isotope ratio mass spectrometer (IRMS; MAT253; Thermo Fisher Scientific) via an open-split interface (GC Combustion III; Thermo Fisher Scientific). The analytical precision was better than 0.3‰ and 5‰ for the carbon and hydrogen stable isotope ratios, respectively.

The hydrogen isotope analysis of H₂ was conducted via a previously established procedure (Kawagucci et al. 2010). Similar to the preparation method for CH₄, an aliquot of gas sample containing H₂ was introduced into the preparation line filled with ultrapure helium as a carrier gas. Purification was achieved using columns packed with Molecular Sieve 5A, which were held at − 100 °C (ethanol–liquid N₂ bath) for the removal of H₂O and CO₂ and at − 196 °C (liquid N₂ bath) for the condensation of H₂. Finally, H₂ was released from the column at room temperature and introduced into a gas chromatograph (6890; Agilent Technologies) connected to an IRMS (MAT253; Thermo Fisher Scientific) for quantification and isotope ratio analysis. The analytical precision was better than 5‰ for the stable hydrogen isotope ratio of H₂.

A slightly modified elemental analyzer (FlashEA1112 series) coupled with Conflo IV interface and IRMS (Delta plus Advantage; Thermo Fisher Scientific) was used for δ¹³C analysis of CO₂ in gas samples. A manual syringe injection port was additionally installed between the downstream part of the reduction reactor and the upstream part of the dehydration tube of the elemental analyzer, and the sample and standard gas were injected from the injection port. The moisture of the samples was removed via a dehydration tube, and the air components (N₂, O₂, Ar, CH₄, etc.) were separated from the CO₂ gas by a gas chromatograph column of the elemental analyzer. The Oztech CO₂ gas standard was used as the working standard, and the analytical precision of repeated analyses of the working standard was better than ± 0.3‰ (n = 3). The DIC in the filtered groundwater samples was converted to CO₂ by a reaction with phosphoric acid (H₃PO₄), and δ¹³C analysis was conducted using the same procedure as that used for the gas samples. All measurements were conducted at JAMSTEC. The carbon and hydrogen stable isotopic compositions are reported relative to international standards expressed as follows:

$$\delta ^{{13}} \mathrm{C}\left(\permil \right) = \left[ {{{\left( {^{{13}} \mathrm{C}/^{{12}} \mathrm{C}} \right)_\mathrm{{sample}} } \mathord{\left/ {\vphantom {{\left( {^{{13}} C/^{{12}} C} \right)_{{sample}} } {\left( {^{{13}} C/^{{12}} C} \right)_{{standard}} }}} \right. \kern-\nulldelimiterspace} {\left( {^{{13}} \mathrm{C}/^{{12}} \mathrm{C}} \right)_\mathrm{{standard}} }} - 1} \right] \times 1000$$

(1)

$$\delta {\text{D}} \left(\permil \right)= \left[ {{{{{\left( {{\text{D}}/{\text{H}}} \right)}_{{\text{sample}}}}} \mathord{\left/ {\vphantom {{{{\left( {{\text{D}}/{\text{H}}} \right)}_{{\text{sample}}}}} {{{\left( {{\text{D}}/{\text{H}}} \right)}_{{\text{standard}}}}}}} \right. \kern-0pt} {{{\left( {{\text{D}}/{\text{H}}} \right)}_{{\text{standard}}}}}} - 1} \right] \times 1000$$

(2)

where δ¹³C (‰ VPDB: Vienna Peedee Belemnite) and δD (‰ VSMOW: Vienna Standard Mean Ocean Water) in delta (δ) notations.

2.3.5 Radiocarbon measurements

The radiocarbon content was measured for CH₄, CO₂, and DIC following the method described in a previous study (Urai et al. 2021b, 2022). The details about sample preparation are presented in the supplementary information. In brief, CH₄ was converted to CO₂ as described previously (Kawagucci et al. 2020). DIC was extracted as CO₂ by the reaction with 85% phosphoric acid and bubbling with ultrahigh-purity helium gas. The resulting CO₂, equivalent to 1 mg of carbon, was cryogenically purified, transferred into a 9-mm-diameter Pyrex glass ampoule, and flame sealed. The ampoules containing CO₂ were connected to the vacuum line, and graphitization was conducted at Atmosphere and Ocean Research Institute (AORI), the University of Tokyo (Yokoyama et al. 2022). Finally, the radiocarbon content was measured using a single-stage acceleration mass spectrometry (AMS) at AORI (Yokoyama et al. 2019, 2022). The results are denoted in delta notation (Δ¹⁴C, ‰) and pMC (percent modern carbon) using the following equations (Stuiver and Polach 1977):

$${\Delta^{14}}{\text{C}}\left(\permil \right) = {\delta^{14}}{\text{C}} - 2\left( {{\delta^{13}}{\text{C}} + 25} \right)\left( {1 + {\delta^{14}}{\text{C}}/1000} \right)$$

(3)

$${\text{pMC}} = \left( {{\Delta^{14}}{\text{C}}/1000 + 1} \right) \times 100$$

(4)

2.3.6 16S rRNA gene tag sequencing and diversity analysis

Genomic DNA was extracted from samples collected on glass fiber filters via the DNeasy PowerSoil Pro Kit (Qiagen, CA, USA) following the manufacturer's protocol. The concentration of the extracted DNA was measured using a Quant-iT dsDNA High-Sensitivity Assay Kit (Thermo Fisher Scientific). PCR amplification was performed using TaKaRa LA Taq polymerase (TaKaRa Bio Inc., Shiga, Japan), and the reaction mixture for PCR was prepared according to the manufacturer's instructions. For PCR amplification, a universal primer set 530F/907R (Nunoura et al. 2012) was used to amplify the V4–V5 regions of the 16S rRNA genes. The PCR primers contained overhang adapters at the 5′ ends. The PCR amplification conditions and detailed sequencing procedures were described previously (Hirai et al. 2017; Imachi et al. 2019). Amplicon sequence variants (ASVs) were generated from raw sequencing data using DADA2 (Callahan et al. 2016) and QIIME2-2022.11 (Bolyen et al. 2019), and taxonomic assignments were made via the SILVA SSURef NR99 Database (version 138.2; Quast et al. 2013).

On the basis of the analysis output, the alpha diversity indices (Chao1, ACE, Simpson, and Shannon) were calculated for each sample using R (version 4.4.1; R Core Team 2024) and the VEGAN package (version 2.6.8; Oksanen et al. 2024) in the R Studio platform. The obtained alpha diversity values, together with the rarefaction curves, are provided in the Supplementary Information. Principal component analysis (PCA) and nonmetric multidimensional scaling (NMDS) analyses based on the Bray‒Curtis distance (Bray and Curtis 1957) were performed using the same platform.

3 Results

3.1 Subsurface physicochemical parameters

The vertical profiles of temperature, pH, and E_h, which were measured in situ at observational wells during the 2024 survey, are shown (Fig. 3a). We observed that the strong blowout of groundwater and gas and floating of the sensor during measurement caused the shoulders to be approximately 40 mbs (see supplementary movies) at Well #1. Despite the location of the wells, the pH trend was similar across all the wells, ranging from 6.5 to 7.0. Although the differences in pH along with depth were small, it became more acidic by 40–50 mbs, and the deeper region (> 50 mbs) was more stable. Except for the shallowest region, which was heated by sunlight, the temperature increased gradually from 15 °C at approximately 5 mbs to 25 °C at approximately 95 mbs. The observed geothermal gradient (~ 110 °C km⁻¹) was relatively higher than the Japanese mean geothermal gradient in regions without thermal anomalies, such as volcanic settings (20–50 °C km⁻¹; Uyeda and Horai 1963; Horai 1964; Tanaka et al. 2004). This finding suggested the influence of subsurface heat sources beneath the sedimentary layer. Similar to the pH trends, the E_h values decreased with increasing depth at Wells #1 and #2. Although the E_h values could have been artificially perturbed by the drilling of well holes, the E_h at the 95 mbs region of Well #1 (32.7 mV vs. standard hydrogen electrode, SHE) indicated microaerobic conditions (< 10 µmol L⁻¹; the redox classification followed Berg et al. 2022 and references therein) down to 100 mbs region of the sedimentary layer. In contrast, the vertical redox trend at Well #3 differed significantly, with values ranging from 90 to 95 mV versus SHE at depths greater than 5 mbs. As shown in Fig. 1, Well #3 is located at the transition zone between the sedimentary layer and surrounding mountainous areas, where fresh, oxygen-rich groundwater from the mountains likely influences the redox profiles. The E_h measured at the nearby mountain spring site near Well #3 was 426 mV versus SHE, supporting this assumption. The data obtained after sampling at ground level are presented in Table S2. In general, the properties were consistent with the in situ measurements, except for E_h at Well #1. The EC and dissolved oxygen (DO) profiles measured in situ simultaneously with other parameters are also provided in Fig. 3b. Briefly, the EC values at the deepest points of Wells #1, #2, and #3 were 1.2 mS cm⁻¹, 3.3 mS cm⁻¹, and 0.4 mS cm⁻¹, respectively. The apparently low EC at Well #3 could be explained by its location being similar to that of E_h. Oxygen depletion was confirmed at depths of 10 mbs, 22 mbs, and 6 mbs for Wells #1, #2, and #3, respectively.

Fig. 3

Vertical physicochemical profiles of a temperature, pH, redox potential (E_h), b dissolved oxygen (DO) and electrical conductivity (EC) at three observational wells. The measurement was conducted in situ using a multiparameter water quality sensor in July 2024. The shoulders of vertical trends at 40 mbs depth of Well #1 are disturbed by strong blowout of groundwater and gas, resulting in floating of the sensor. The data were obtained every five seconds equivalent to around 0.5 m depth. The scale intervals of horizontal axes were adjusted individually for each panel

The properties of the Kamisuwa hot spring sources are provided in Table S2. Among the spring sources, the highest water temperature of 88 °C was recorded at HS #6. Although both HSs #1 and #6 were pumped from the same depth of 1000 mbs, the temperature of HS #1 was 55 °C, which was significantly lower than that of HS #6. This difference suggested the mixing of groundwater within the sedimentary layer during the pumping process. The pH was approximately 8–9, and the E_h was − 182 mV versus SHE at HS #2.

3.2 Water chemistry of groundwater and hot springs

The compositions of major cations and anions were measured using ion chromatography and visualized in a Piper plot (Fig. 4; Piper 1944). In the groundwater samples, the dominant anion was HCO₃⁻, accounting for more than 70% of the total anions. In contrast, the dominant anions in the hot spring samples, except for those in HS #5, were Cl⁻ and SO₄²⁻. Regarding cations, hot spring samples were notably depleted in alkaline earth metals (Ca²⁺ and Mg²⁺). The ion composition of the mountain spring sample was similar to that of the groundwater samples, with HCO₃⁻ constituting 94% of the total anions and SO₄²⁻ being nearly absent.

Fig. 4

Piper diagram showing the composition of major cations and anions of groundwater, hot spring, and mountain spring samples. For groundwater and hot spring samples, open symbols correspond to data obtained in 2023, while the filled symbols correspond to data of 2024. The data of mountain spring was obtained in 2024. For hot spring samples, the spring source numbers (#) are indicated together. Sakakibara et al. (2025) also measured the concentrations of ionic species for groundwater and spring water samples collected from the same sampling sites as this study and reported similar compositions

The abundances of trace elements differed significantly between the groundwater and hot spring samples. In the groundwater samples, the concentrations of Fe (9.6–12.3 mg L⁻¹) and Mn (0.28–0.31 mg L⁻¹) were relatively high, whereas these elements were almost absent in the hot spring samples. Conversely, Li, B, Cs, and W were enriched exclusively in the hot spring samples. The details concerning the concentrations of major ions and trace elements are provided in the supplementary information (Tables S3 and S4).

3.3 Chemical characteristics of the gas component

The concentrations and stable isotope profiles for CH₄, CO₂, and H₂ are summarized in Table S5. At steady state, the gas emission rates were > 100 mL min⁻¹, > 750 mL min⁻¹, and > 0.8 mL min⁻¹ from the wellheads (approximately 20 cm²) of Wells #1, #2, and #3, respectively. The gas and dissolved gas phases at Wells #2 and #3 were predominantly composed of CH₄ and CO₂. The concentrations and isotope data for CH₄ and CO₂ were consistent across the observational wells, despite their locations within the basin, and they remained stable over the years (Fig. 5). The CH₄-specific δ¹³C values were consistent with those reported for gas seep sites in Lake Suwa, which were attributed to microbial metabolism (Nakai et al. 1974; Iwata et al. 2020; Urai et al. 2021a). The isotope profiles of the hot spring sources varied significantly. The δ¹³C-CO₂ values of Wells #2 and #3 were similar to the values reported from Lake Suwa (Urai et al. 2022), although slight ¹³C depletion was observed in Well #1.

Fig. 5

The carbon and hydrogen isotope profiles of CH₄ and CO₂ obtained at the observation wells within the sedimentary layer and Kamisuwa hot spring sources. The data were plotted on the genetic diagrams based on A δ¹³C-CH₄ versus δD-CH₄ and B δ¹³C-CH₄ versus δ¹³C-CO₂. The genetic fields were originally proposed by Whiticar et al. (1986) and Whiticar (1999) and revised by Milkov and Etiope (2018).—CR: CO₂ reduction; Ferm: fermentation; Therm: thermogenic

3.4 Radiocarbon profiles

The ¹⁴C contents of DIC contained in deep aquifers within the sedimentary layer are summarized in Table S6. At all locations, DIC was predominantly composed of relict carbon (Fig. 6), indicating isolation from surface material cycles. In addition to DIC, radiocarbon contents were measured for CH₄ and CO₂ in the gas phase of Well #2. These species were similarly depleted in ¹⁴C, which was consistent with the values reported from Lake Suwa (Urai et al. 2022).

Fig. 6

The diagram showing stable carbon isotope (δ¹³C) and radioisotope composition (Δ¹⁴C) for dissolved inorganic carbon (DIC) in groundwater samples obtained from the sedimentary layer. The isotope profiles for CH₄ and CO₂ in the gas phase collected at Well #2 were also indicated. The reported isotope compositions for benthic bubble gas of Lake Suwa (CH₄ seep sites #1 and #2 in Urai et al. 2022) are indicated together

3.5 Microbial communities in the groundwater and hot spring samples

The microbial composition of subsurface aquifers within the sedimentary layer of the Suwa Basin (10–93 mbs) and Kamisuwa hot spring sources, pumped from the bedrock region (~ 1000 mbs), was revealed by 16S rRNA gene amplicon analysis (Fig. 7). The groundwater samples were abundant in aerobic and methane-oxidizing bacteria (MOBs) and in methylotrophic bacteria affiliated with the classes Alphaproteobacteria (Type II methanotrophs, Serine pathway) and Gammaproteobacteria (Type I methanotrophs, RuMP pathway). A notable presence of fermenters affiliated with the phyla Bacteroidota and Chloroflexota (mainly the class Anaerolineae) was observed in the groundwater samples. Among the phylum Bacteroidota, the most abundant orders were Bacteroidales, Chitinophagales, Flavobacteriales, and Sphingomonadales, a characteristic shared among observational wells. While the majority of the observed lineages belonging to the class Anaerolineae were affiliated with the order Anaerolineales (primarily the family Anaerolineaceae), the order SJA-15, which is thought to include the candidate genera Ca. Sarcinithrix and Candidatus (Ca.) Amarolinea (Speirs et al. 2019), constituted more than 10% of the total community at both depths of Well #1. Among the three wells, Well #2 exhibited a clear compositional distinction between the two depths, with the deep region (52 mbs) showing enrichment with lineages belonging to the genus Clostridium of the phylum Bacillota. Archaea was not highly abundant across the groundwater samples. Methanogenic archaea belonging to the genus Methanothrix were almost exclusively detected from 52 mbs at Well #2, comprising approximately 3% of the total community. Regarding ANME, lineages belonging to the family Ca. Methanoperedenaceae (formerly ANME-2d), a terrestrial group of ANME known for anaerobic methane oxidation using versatile electron acceptors, including nitrate (Haroon et al. 2013) and iron and manganese oxides (Ettwig et al. 2016; Cai et al. 2018; Leu et al. 2020), were detected at 10 mbs in Well #2 and at both depths of Well #3. In all the cases, the abundance of ANME was approximately 1–2% of the total microbial community.

Fig. 7:

16S rRNA gene amplicon-based microbial composition of groundwater and hot spring water samples obtained in 2023. The depths for groundwater samples are indicated relative to the water surface

The microbial compositions among the eight hot spring sources varied considerably, even at nearby sites. In contrast to the predominance of heterotrophs in the sedimentary layer groundwater, the hot spring samples were dominated by chemolithoautotrophs. For example, the genus Hydrogenobacter, a hyperthermophilic hydrogen-oxidizing bacterial group, was one of the most abundant taxa in HSs #4 and #6. The remaining Aquificota-affiliated lineages in HS #4 were composed of the genus Sulfurihydrogenibium, another thermophilic genus capable of utilizing molecular hydrogen and reduced sulfur compounds as electron donors (Takai et al. 2003; Nakagawa et al. 2005). In HSs #1, #3, #5, and #7, the class Gammaproteobacteria was abundantly represented, as observed in the groundwater samples. For HSs #5 and #7, the dominant group was MOBs of the order Methylococcales. In HS #5, the abundance of MOB lineages belonging to the genus Methylosinus (within the class Alphaproteobacteria) was significant. In HS #7, a moderately thermophilic aerobic methanotroph, the genus Methylothermus (Tsubota et al. 2005), constituted approximately 20% of the total community. Additionally, a candidate genus, Ca. Methylomirabilis, a bacterial group known to perform nitrite-dependent anaerobic methane oxidation (Ettwig et al. 2010), was detected at HS #7. In contrast, in HSs #1 and #3, the lineages were affiliated not with methanotrophs but with the family Hydrogenophilaceae, a thermophilic group that could utilize molecular hydrogen as an electron donor. In HS #3, the autotrophic iron-oxidizing bacterium Ferriphaselus (Kato et al. 2014), from the family Gallionellaceae, was detected as an abundant member of Gammaproteobacteria.

The archaeal abundance was notably high in several hot spring sources. In particular, at HSs #6 and #8, the relative abundance of Thermoproteota exceeded 35% of the total community at each site. The predominant genera affiliated with Thermoproteota were Ignisphaera and Pyrobaculum at HS #6 and Ca. Caldiarchaeum and Pyrobaculum at HS #8. In contrast, the archaeal communities in HSs #3 and #4 were dominated by the classes Bathyarchaeia and Nitrososphaeria, respectively.

The microbial composition of HS #2 was markedly different from that of the other sources. In this sample, the genus Caldimicrobium of the phylum Desulfobacterota and the class Thermodesulfovibrionia of the phylum Nitrospirota constituted 31% and 50% of the total community, respectively. The genus Caldimicrobium is an extreme thermophile capable of chemolithoautotrophic growth, utilizing molecular hydrogen in the presence of thiosulfate or elemental sulfur (Miroshnichenko et al. 2009). On the basis of the sequencing data, we conducted diversity analyses. The rarefaction curves and alpha diversity indices (Fig. S2 and Table S8) clearly revealed differences in the richness of diversity between the hot spring and groundwater samples. The highest Shannon index (Shannon 1948) was 6.01 at 52 mbs in Well #3, whereas the lowest value, 1.66, was obtained from HS #2. The PCA distinctly separated HS #2 along the first principal component and HS #6 along the second principal component (Fig. 8A). The other hot spring sources clustered closely with the groundwater samples, except for HSs #4 and #8. NMDS based on Bray‒Curtis distances was performed to visualize the diversity patterns among the microbial communities (Fig. 8B). The results highlighted the differences in the dominant lineages between the groundwater and hot spring samples, with MOBs and fermenters prevailing in the former and with chemolithoautotrophs dominating in the latter. HSs #3 and #5 samples were plotted in intermediate positions, reflecting potential mixing between these two endmembers. Depth-dependent shifts in community composition were not significant in the groundwater samples.

Fig. 8

A Principal component analysis (PCA) charts based on 16S rRNA gene reads. B Nonmetric multidimensional scaling (NMDS) charts based on Bray‒Curtis distance. For Kamisuwa hot data points, spring source numbers (#) are presented together. Among hot spring sources, samples with a high abundance of hydrogenotrophs were highlighted by double-circle symbols

4 Discussion

4.1 Subsurface hydrogeology and the source of groundwater

From the boundary region with the mountains to the center of the pull-apart basin, a clear decreasing trend in the ¹⁴C concentration of DIC was observed (Fig. 6), reflecting the horizontal flow direction of groundwater within the sedimentary layer. The water sources for Well #3 were inferred to be a mixture of two components: (1) young groundwater originating from mountainous areas and (2) aged groundwater initially present at the location. On the basis of this assumption, the mixing ratios of mountain groundwater and sedimentary layer groundwater were calculated using the following two equations with respect to the ¹⁴C and DIC concentrations:

$${\text{DI}}{{\text{C}}_{\text{M}}}f + {\text{DI}}{{\text{C}}_{\text{S}}}\left( {{1} - f} \right) = {3}.{2}$$

(5)

$${\text{DI}}{{\text{C}}_{\text{M}}}f \times {\text{pM}}{{\text{C}}_{\text{M}}} + {\text{DI}}{{\text{C}}_{\text{S}}}\left( {{1} - f} \right) \times {\text{pM}}{{\text{C}}_{\text{S}}} = {3}.{2} \times {12}.{8}$$

(6)

where f represents the proportion of young groundwater recharged from the surrounding mountain region during the mixing process, whereas DIC_x and pMC_x (x = M or S; “M”ountain groundwater and “S”edimentary layer groundwater, respectively) are the concentrations of DIC (mM) and radiocarbon (¹⁴C), respectively, and the subscripts M and S correspond to the parameters for mountain groundwater and sedimentary layer groundwater, respectively. In the calculation, the ¹⁴C concentration of Well #1 was utilized as the upper limit of the pMC_S value because the location of Well #1 was minimally affected by mountain groundwater in terms of the ¹⁴C contribution. The lower limit of the pMC_S value was set to 0 pMC. For the pMC_M value, 100 pMC was assumed, on the basis of the reported SF₆ concentrations (Sakakibara et al. 2025) and the estimated ages derived from the data (15–20 years). The DIC concentration (3.2 mM) and ¹⁴C concentration (12.8 pMC), measured at 51 mbs in Well #3 in 2023, were used in the calculations. The DIC concentration of the mountain spring water (124.4 mg L⁻¹; Table S3) was adopted as the value for DIC_M.

The results indicated that the proportion of mountain groundwater ranged from 12 to 15%. As there were potential contributions of other sources of dead carbon, such as methane, the calculated proportion should be regarded as a minimum estimate. Given the redox potential of mountain spring water (+ 426 mV vs. SHE), the substantial influx of oxidative groundwater from mountainous areas into the sedimentary layer of the Suwa Basin could play a significant role in forming microaerobic conditions even at 100 mbs in the subsurface (Fig. 3). This influx could contribute to the proliferation of aerobic methanotrophs by providing essential electron acceptors.

The aquatic ecosystem at Lake Suwa has been reported to be influenced by biogenic methane originating from deep subsurface regions (Urai et al. 2022). Similarly, not only methane but also deep fluids from bedrock regions could alter the physicochemical and geochemical conditions of the sedimentary layer and the prokaryotic communities harbored therein. While the above discussion on the local mixing process between mountain groundwater and sedimentary groundwater illuminates the significance of mountain groundwater in determining physicochemical conditions at the boundary region, a distinct basin-scale mixing model is required to evaluate the contributions of deeply derived fluids to ecosystems within the basin. In the model, we assumed that the groundwater of the sedimentary layer was a mixture of two endmembers: (1) young groundwater from mountainous areas and (2) hydrothermal fluid, which has undergone high-temperature water‒rock interactions. The hydrothermal fluid discharged from the bedrock region was assumed to originate from meteoric water, as evidenced by water isotopes (δ¹⁸O, δD), which were plotted along the Suwa local meteoric water line (Fig. S4; Nakano, unpublished data; Craig 1961; Sakakibara et al. 2025). The δD-depleted water isotope profile (− 80 to − 86‰ VSMOW) suggested that additional contributions from volcanic gases and magma were negligible. Thus, the assumption regarding the water source for subsurface aquifers is highly plausible. In the analysis, the chemical profiles of the hot spring samples were considered to represent hydrothermal fluid. Among the eight sources, HS #6 was selected as the endmember for the hydrothermal fluid, as it was sourced from the deepest region (1000 mbs), exhibits the highest water temperature (89.1 °C), and contained relatively high concentrations of ions (e.g., Li⁺, Cl⁻, Br⁻, and SO₄²⁻; Table S3) and elements (B, Mo, and Cs; Table S4) unique to hot spring samples. Additionally, the microbial composition and results of multivariate analysis, which revealed a distinctive microbial community composition predominated by hyperthermophilic hydrogenotrophs (e.g., the genera Hydrogenobacter and Pyrobaculum) in HS #6 (Figs. 7, 8), further supported its selection. For the mixing index, the lithium‒sodium (Li/Na) molar ratio was used, as Li is known to be mobilized from rocks into the fluid under elevated temperatures and to remain in the fluid during the cooling process (You et al. 1996).

4.2 Chemistry of groundwater for cation and anion properties

The Li/Na ratios and concentrations of ions (Cl⁻ and Br⁻) and trace elements (Mo and Cs) of groundwater and hot spring samples are summarized in Fig. 9. Although the data points of hot spring samples were scattered, the concentrations of both ions and trace elements exhibited significant correlations with the Li/Na ratio, which serves as a mixing index. This result suggests that the mixing process between mountain groundwater and hydrothermal fluid is a major factor constraining the subsurface water chemistry of the Suwa Basin. Among the eight sources, the HSs #1 and #5 were the hot spring samples most strongly influenced by groundwater derived from the sedimentary layer, as supported by the Li/Na index. The proportions of groundwater at HSs #1 and #5 were approximately 20% and 30%, respectively (Table S9). This is consistent with the ionic species composition and prokaryotic community structure. The chemical and microbiological characteristics of HS #5 were distinctly different from those of the other hot spring sources. Both the microbial community associated with a high abundance of MOBs (Figs. 7, 8) and the major ion species composition suggested the substantial mixing of hydrothermal fluids and groundwater within the subsurface. This scenario is particularly plausible, considering the strainers installed at the depth corresponding to the unconsolidated sedimentary layer in the HS #5 borehole (Table S1). The ionic composition of HS #1 was similar to that of other sources. However, both the absolute concentrations of ions (e.g., Na⁺, Cl⁻, and SO₄²⁻) and trace elements (e.g., B, Mo, and Cs) that are typically enriched in the hot spring samples were lower at HS #1 than at other HSs (Fig. 9). Although the borehole depth of HS #1 reaches the greatest depth, the water temperature (55.3 °C) was lower than that of most other sources. These results suggested the influence of groundwater intrusion into the hot spring water at HS #1.

Fig. 9

The scatter plots with regression lines showing the relationship between Li/Na molar ratio and concentrations of ions or trace elements. Shaded areas represent 95% confidence intervals. The correlation coefficients (r) for each pair are shown together. For hot spring samples, the source numbers (HS #) are also indicated

In Well #3, the Li concentration was below the quantification limit at both depths, indicating that the groundwater at this site was not significantly influenced by deeply derived fluids. In contrast, the remaining wells showed an approximately 10–20% contribution of hydrothermal fluids (Fig. 9). The apparent difference in mixing ratios between Wells #1 and #2 could result from heterogeneity in subsurface hydrogeology and the distribution of faults. In addition to Li, the concentrations of ion species typically enriched in hot springs (e.g., Na⁺ and Cl⁻) were higher at Well #2 than at Wells #1 and #3. Conversely, SO₄²⁻ was almost depleted at Well #2, whereas it was present at approximately 200 mg L⁻¹ at HS #6. This depletion was explained by the microbial consumption of SO₄²⁻ as a sulfur source and/or an electron acceptor. This consumption suggested that deep fluid intrusion could activate microbial communities in the shallow subsurface by introducing nutrients. Furthermore, the relatively high concentration of bicarbonate at Well #2, which could result from the oxidation of organic matter, including methane, supported this assumption. To better understand the adaptation of subsurface microbial communities to the intrusion of deeply derived nutrients and altered geochemical settings, further analyses, including biomass quantification, isotope analyses of biomolecules and metabolic substances, and biological omics analyses (e.g., genome-resolved metagenomics and transcriptomics), would be needed. Additionally, it should be noted that the current fluid mixing model is based on a simplified assumption with two endmembers, whereas mixing processes in natural system are likely more complex due to additional fluid sources, diverse transport pathways, and biological and chemical reactions (e.g., metabolic activities, water–rock interactions). Toward the more rigorous and quantitative interpretations of the relationship between subsurface biogeochemical cycles and prokaryotic ecology, more comprehensive hydrogeological and geochemical models will be required in future studies.

4.3 Diversity of ecosystems by redox constrain with methane and hydrogen

The microbial composition analysis clarified the community distribution down to 1000 mbs in the subsurface of the Suwa Basin, highlighting the predominance of aerobic bacterial lineages involved in methane oxidation or fermentation at depths reaching 100 mbs in the sedimentary layer. The communities were associated with methylotrophs, such as the genera Methylotenera and Methylobacillus of Gammaproteobacteria, which do not possess methane monooxygenase. The growth of these non-methanotrophic methylotrophs could be dependent on methanol or other one-carbon compounds secreted by MOBs (Xu et al. 2020).

Aerobic methanotrophs are generally assumed to utilize O₂ as an electron acceptor, which was nearly depleted in the subsurface aquifers of the Suwa Basin. Nevertheless, several studies have documented the persistence of MOBs in terrestrial and oceanic environments with very low or undetectable oxygen levels (Li et al. 2024 and references therein). These unconventional ecological adaptations of MOBs could be facilitated by the following two strategies: (i) the production or acquisition of O₂ from coexisting partners (Dershwitz et al. 2021; Kraft et al. 2022) or (ii) the utilization of alternative electron acceptors, such as iron oxides (Zheng et al. 2020) and nitrate (Kits et al. 2015). Additionally, under oxygen-limited conditions, some MOBs have been reported to adopt fermentation-based pathways, producing organic acids and molecular hydrogen as products of methane oxidation (e.g., Kalyuzhnaya et al. 2013). These findings suggested that MOBs could contribute to reducing net greenhouse gas emissions under a wider range of environmental conditions than previously assumed. Future studies leveraging genomic information are expected to elucidate the metabolic adaptations of MOBs in O₂-depleted groundwater aquifers of the Suwa Basin. Furthermore, such insights could provide a better understanding of their role in mitigating greenhouse gas emissions. This knowledge could contribute to the understanding of the influences of MOBs on methane oxidation and carbon cycling, especially in O₂-limited environments, and of the broad implications of MOBs for mitigating global climate change.

4.4 Factors constraining redox boundaries and deep methanogenesis

The upward and lateral migration processes of deep methane and other gas fluids are considered important factors in defining the redox boundaries of the subsurface environment (Gay et al. 2011; McMahon et al. 2011; Grall et al. 2018). Although the isotopic compositions of CH₄ obtained from two hot spring sites were varied, the ¹³C-enriched δ¹³C-CH₄ value of HS #6 suggested its thermogenic origin and the profile of HS #2 was potentially affected by the mixing of fluid originated from the sedimentary layer (Fig. 5). In contrast, the CH₄-specific δ¹³C and δD values obtained from the well sites indicated a primary biological origin (hydrogenotrophic and methylotrophic methanogenesis) for the seeping CH₄ (Fig. 5A; Whiticar et al. 1986; Milkov and Etiope 2018). The δ¹³C-CH₄ and δ¹³C-CO₂ relationships also supported the microbial origin of the CH₄ seeping from the sedimentary layer (Fig. 5B; Whiticar 1999; Milkov and Etiope 2018), which was previously suggested at the region of Lake Suwa (Urai et al. 2022). Additionally, the composition ratios of hydrocarbons, namely CH₄, C₂H₆, and C₃H₈, supported this assumption of the origins of CH₄ (Fig. S1). Despite these geochemical evidence of biological methanogenesis in the subsurface of the Suwa Basin, the cell population of methanogens was not significant in the groundwater communities present down to 100 mbs. This result suggested that methanogens could inhabit a relatively reducing niche (e.g., deeper sedimentary layers) due to their high sensitivity to oxygen stress. Alternatively, the absence of methanogens could be explained by past methanogenesis, which resulted in subsurface methane reservoirs, accounting for the occurrence of biogenic methane emissions despite the absence of methanogenic populations.

To better understand subsurface methanogenesis, the utilization of the coenzyme factor F430 (hereafter F430) was considered a promising approach (Kaneko et al. 2014). F430 is a prosthetic group of methyl-coenzyme M reductase, utilized by all known methanogens and ANMEs. Since F430 readily undergoes epimerization after the death and lysis of host cells (Kaneko et al. 2021), it serves as a sensitive indicator to distinguish active and relic methanogenesis. Our group previously detected planktonic F430 in the water column of Lake Suwa, suggesting the presence of methanogens (Urai et al. 2021a). Applying F430 analysis to groundwater aquifers could provide valuable insights into the mechanisms underlying the strong methane emissions in the Suwa Basin.

The prokaryotic community in the hot spring, which originated from the bedrock region to a depth of 1000 mbs, was characterized by hyperthermophilic and thermophilic bacteria and archaea, including the genera Hydrogenobacter (Kawasumi et al. 1984), Sulfurihydrogenibium (Takai et al. 2003; Nakagawa et al. 2005), Ignisphaera (Niederberger et al. 2006; Podosokorskaya et al. 2024), and Pyrobaculum (Huber et al. 1987; Völkl et al. 1993). While Ignisphaera is known as a hyperthermophilic chemoorganotroph, all the other genera include hydrogenotrophic lineages. Although in situ metabolic activities could not be determined solely from 16S rRNA gene sequencing data, their chemolithoautotrophic growth, which relied on molecular hydrogen as an electron donor, was likely, considering that the H₂-containing gas phases were associated with the hot spring water (Table S5).

4.5 Origin of H₂: insight from D/H signatures

H₂ can be produced by prokaryotic fermentation in dark subsurface environments. However, the dominant prokaryotes in hot spring samples associated with H₂-containing gas phase were predominated by hydrogenotrophs and hydrogen producers were nearly undetected. This can be explained by the depletion of organic matter in the fluid and limitation on the fermentative metabolism (Schwartz and Friedrich 2006). Similar to the case of past methanogenesis described in the above sections, the possibility that past biological H₂-production and its remaining reservoir cannot be completely denied. Alternatively, H₂ could have originated from abiotic processes, without prokaryotes responsible for H₂-production. Among the abiotic H₂ generation processes, the observed hydrogen isotope profiles fell within the range of serpentinization and mechanoradical reactions. The serpentinization reaction involves the metamorphism of ultramafic minerals (olivine and pyroxene) with water molecules, producing serpentinites, molecular hydrogen, and methane. The reported hydrogen isotopic compositions at terrestrial and oceanic serpentinization sites are relatively close to those observed at the Kamisuwa hot spring (HS #6; δD = − 736‰) in this study and at Hakuba Happo (δD = − 710 to − 700‰; Suda et al. 2014) and the Lost City hydrothermal field (δD = − 689 to − 605‰; Proskurowski et al. 2006). However, the bedrock of the Suwa Basin is composed of andesite and granodiorite, which are not enriched with the minerals required for serpentinization. Alternatively, mechanoradical reactions were considered more suitable for the geological setting of the Suwa Basin. This process could generate H₂ through reactions between fresh silicate minerals and water molecules at the fault surface during crustal movement (Kita et al. 1982). The reported δD-H₂ values for H₂ originating from this process ranges from − 770 to − 470‰ and − 708 to − 618‰ at the Yamasaki fault, Japan (Kita et al. 1980), and San Andreas fault, USA (Wiersberg and Erzinger 2008), respectively. Considering the geological activities related to the ISTL in the Suwa Basin, which is located on the plate boundary, mechanoradical generation was assumed to be the most plausible scenario.

Based on the measured δD-H₂ value and previously reported δD-H₂O value (− 85.65‰ VSMOW for HS #6; Nakano, unpublished data), the in situ fluid temperature was estimated using the following equation developed by Horibe and Craig (1995);

$$\alpha \left( {{{\text{H}}_2}{{\text{O}}_{{\text{aq}}}}-{{\text{H}}_{2}}} \right) \, = \, 1.0473 + 201036/{T^2} + 2.060 \times {10^9}/{T^4} + 0.180 \times {10^{15}}/{T^6}$$

(7)

where α(H₂O_aq – H₂) represents hydrogen isotope fractionation factor between H₂ and H₂O_aq and T represents absolute temperature (K), respectively. The calculated equilibrium temperature for HS #6 was around 42 °C, far below even the measured temperature at the outlet (approximately 90 °C; Table S2). Therefore, the result suggests a disequilibrium between the two compounds. The hot spring samples were likely a mixture of components that had undergone distinctive processes (e.g., water–rock interactions). Indeed, mantle-derived helium has been previously reported at Kamisuwa hot spring (Umeda et al. 2013), indicating an origin and transport pathway distinct from meteoric water-derived hot spring water. If H₂O and H₂ have originally existed in different fluids, the reaction time between them may have been insufficient to achieve isotopic equilibrium. This limitation provides one possible explanation for the disequilibrium.

The H₂ that originated from bedrock could be utilized by chemolithoautotrophs at subsurface depths of ~ 1000 mbs, and also it could support communities inhabiting the overlying sedimentary layer. As shown by the Li/Na index discussed earlier, the intrusion of deep fluids into the sedimentary layer was evident. Additionally, the methane isotope profiles suggested the contribution of hydrogenotrophic methanogenesis, which reduced CO₂ with H₂. Although it is not yet determined whether methanogenesis is still occurring today, the supply of H₂ could enhance methanogenesis, thereby indirectly supporting the growth of methane-dependent MOBs in shallower regions. Therefore, it should be emphasized that the methane-related subsurface communities in the sedimentary layer were ultimately influenced by the active geological processes of the Suwa Basin, potentially producing H₂; H₂ could serve as an important energy source for subsurface ecosystems and increase vertical material flow through fault surfaces (Fig. 10).

Fig. 10

Illustration of gas and water mixing within the subsurface of Suwa Basin inferred in this study. The arrows with numbers indicate (1) intrusion of young groundwater from mountain regions, (2) penetration of groundwater from the sedimentary layer into the bedrock layer, and (3) recharge of hydrothermal fluids into the sedimentary layer following high-temperature water–rock interactions. The distribution of subsurface microbial communities is also indicated. The location of normal faults and offset of the Quaternary tephra layers (AT and Pm-I) are based on the reference (Yamazaki 1994)

5 Conclusions

We characterized the geochemical and microbiological features of the methane-rich basin located on the boundary between the Eurasian and North American Plates. Based on insights derived from subsurface environments at depths of 10–1000 mbs, we documented the hydrogeological processes, deep fluid properties, and their potential impacts on subsurface microbial ecology. The findings of this study, summarized below, could aid in the comprehension of interactions between subsurface geochemical and hydrogeological properties and the prokaryotic ecosystems harbored therein in geologically active settings.

(1) Young groundwater intrudes from the surrounding mountain regions into the sedimentary layer, creating horizontal groundwater flow approaching the center of the basin. In the vertical direction, groundwater partially penetrates the bedrock region where heat sources exist. After undergoing high-temperature water‒rock reactions, hydrothermal fluids are discharged into the sedimentary layer, accounting for 10–20% of the total fluid content. Consequently, there are steady inputs of both water masses with both high and low E_h values to the sedimentary layer. This water mixing creates subsurface niches for the growth of subsurface microbes through the provision of metabolic substrates and the formation of oxic‒anoxic interfaces. This process is likely to be enhanced by faults, which act as potential material pathways.

(2) The predominance of hydrogenotrophs in the hot spring samples, along with the H₂-containing gas phase, indicates that H₂ generation processes, which most likely originated from silicate mineral friction, occur in the bedrock regions. The H₂ produced can be utilized by prokaryotic communities in the overlying sedimentary layers through efficient vertical material transport along faults. Therefore, subsurface microbial ecology is closely linked with the unique geological setting of the Suwa Basin, which is located on the plate boundary.

(3) The sedimentary layer at depths of 10–100 mbs is dominated by MOBs and fermenters, with an insignificant abundance of archaea, including methanogens and ANMEs. In contrast, in the bedrock region at depths reaching 1000 mbs, thermophilic and hyperthermophilic bacteria and archaea responsible for H₂ utilization have been determined to be the dominant members of the microbial communities.