Microbial life in deep granitic rock


  • Karsten Pedersen

    Corresponding author
    1. Institute of Cell and Molecular Biology, Department of General and Marine Microbiology, Göteborg University, Medicinaregatan 9C, S 413 90 Göteborg, Sweden
    Search for more papers by this author

Tel.: +46 (31) 773 2578; Fax: +46 (31) 773 2599; E-mail: pedersen@gmm.gu.se


Granitic rock has aquifers that run through faults and single or multiple fracture systems. They can orientate any way, vertically or horizontally and usually, only parts of hard rock fractures are water conducting. The remaining parts are filled by coatings of precipitated minerals, and clay and gouge material. Sampling hard rock is difficult and the risk of contamination due to intrusion of drilling fluids and cuttings in aquifers is obvious. A recent investigation of the potential for contamination of boreholes in granite during drilling operations, using molecular and growth methods, showed that predominating microorganisms in the drilling equipment were absent in groundwater from the drilled boreholes. The total number of bacteria found in subterranean granitic environments ranges from 103 up to 107 cells per ml groundwater, but the number of cultivable microorganisms is usually much lower. We have used culturing techniques with numeric taxonomy for the identification of cultivable microorganisms and the 16S rRNA gene technique to determine bacterial diversity in granitic groundwater. Members of the genera Bacillus, Desulfovibrio, Desulfomicrobium, Eubacterium, Methanomicrobium, Pseudomonas, Serratia and Shewanella have been found. Several biogeochemical processes in granitic rock have been demonstrated where microorganisms seem to be of major importance. One process is the mobilization of solid phase ferric iron oxy-hydroxides to liquid phase ferrous iron by iron reducing bacteria with organic carbon as electron donor. Another biogeochemical process found to be important is the reduction of sulfate to sulfide by sulfate reducing bacteria. They frequently appear in granitic aquifers at depths, and seem to prefer a moderate salinity, approximately 1%. When groundwater rich in ferrous iron, manganese(II) and reduced sulfur compounds reaches an oxygenated atmosphere such as an open tunnel, gradients suitable for chemolithotrophic bacteria develop. A third process is the conversion of carbon dioxide to organic material with hydrogen as the source of energy, possibly formed through radiolysis, mineral reactions or by volcanic activity. Recent results show that autotrophic methanogens, acetogenic bacteria and acetoclastic methanogens all are present and active in deep granitic rock. These observations announce the existence of a hydrogen driven deep biosphere in crystalline bedrock that is independent of photosynthesis. If this hypothesis is true, life may have been present and active deep down in the earth for a very long time, and it cannot be excluded that the place for the origin of life was a deep subterranean igneous rock environment (probably hot with a high pressure) rather than a surface environment.


In recent years, published papers on various aspects regarding the microbiology of subterranean environments have increased in numbers. There has been a significant expansion in the understanding of the bacterial ecology of shallow groundwater systems down to some 50–100 m, accurately reviewed by Ghiorse and Wilson [1]and Matthess et al. [2], and our knowledge is currently increasing about environments deeper down into the crust of the earth [3–6]. Deep subsurface environments vary considerably in composition, from soft sandstone and hardened sedimentary rocks to very hard igneous rock types. The main purpose of this review is to present hypotheses, theories and results on microbial life in one of the hardest and most common rock types of the earth crust, granite.

The Swedish research program on subterranean microbiology [5, 7, 8]has been performed on two sites, the Stripa research mine in the middle of Sweden and the Äspö area, next to the Baltic sea in the southeastern part of Sweden. The Stripa mine is situated 250 km west of Stockholm and was an iron mine until 1976. A total of 16.5×106 tons of iron ore has been mined out since 1448. The ore consisted of a quartz-banded hematite and occurred in a lepatite formation. Adjacent to the lepatite is a large body of 1.7 billion year old medium-grained granite, in which the Stripa project experiments have been performed. The mine was used as a deep underground research facility between 1976 and 1994 [9–11]. The Äspö investigation area, situated on the south-east coast of Sweden, is a part of the Precambrian bedrock in southeastern Sweden where the Småland granites predominate the older, Sveocokarelian complexes. This is where the Äspö Hard Rock Laboratory (HRL) is situated, at 460 m below the surface of the island Äspö (Fig. 1). The Äspö HRL has been constructed as a part of the Swedish nuclear waste disposal program and the work has been divided into three phases: the pre-investigation (1986–1990), the construction (1990–1995) and the operating (1995–) phases. Microbial investigations have been performed during all three phases [12–15]and the research is currently continuing.

Figure 1.

A: The sampling situation at the Äspö Hard Rock Laboratory in June 1996. The sample sites are depicted with their respective borehole names. These names show the type of drilling (HBH, HA=percussion drilled; KAS, SA=core drilled), the tunnel length where they were drilled and whether they were drilled on the left (A) or the right (B) side of the tunnel when going down. Major fracture zones are marked with dashed lines and with their given names, generally indicating their geographic orientation. Possible flow directions of groundwater are indicated with arrows and the estimated inflow rates of groundwater via the fractures to the tunnel are shown in brackets as l s−1. B: A fracture zone (RZ) with boreholes that were drilled in order to follow shallow groundwater intrusion through this major fracture zone into the tunnel. A side vault was constructed (not shown) from which the boreholes KR0012, 13 and 15 were drilled perpendicular through the zone. Note that these three boreholes all sampled at 68 m below sea level (not shown).

2Geology, hydrology and geochemistry of Swedish crystalline bedrock


Sweden is part of the 1.6–3.1 billion years old Fennoscandian Shield and a number of places in Sweden have been investigated as study sites for the deep disposal of spent nuclear fuel. The crystalline rock considered has generally been of granitic composition with quartz, feldspars and mica as the bulk rock minerals. In addition to that, there are accessory minerals which influence the hydrochemical conditions such as calcite (pH and HCO3), pyrite (redox), apatite (HPO2−4), fluorite (F) and clay minerals (ion exchange). Many of these occur as fracture filling minerals and some of them have been formed as a result of weathering reactions. Minor amounts of iron(III) oxy-hydroxy minerals are found in the fractures, especially in the shallow (<100 m) part of the rock.


The distribution of flow has an influence on groundwater composition. The hydraulic conductivity varies considerably between different locations in the rock, and structures like fracture zones may act as conductors and have a dominating influence. Vertical conductive zones are important for groundwater recharge at depth. Horizontal zones may act as hydraulic shields and separate groundwater with different composition. Especially deep groundwater with a relatively high salinity will have a higher density which helps to stabilize the layering. An example of that has been studied in Finnsjön [16]. The openings in rock fractures are potential channels for groundwater. Model studies have been made on flow and transport in fractures with variable apertures [17]. The results suggested that considerable channelling is to be expected in such fractures and that there is a tendency for some pathways to carry much more water than others. In a limited mass of rock, one or a few channels will dominate flow and transport of nutrients and microorganisms. Hydraulic conductivities have been measured in boreholes at different depths and this information together with the groundwater surface topography, which in Sweden is approximately the ground surface topography, is used to calculate the groundwater flow field. Groundwater flow at some 500 m depth is calculated to be in the range of 0.01–1 l m−2 year−1[7]. Hydraulic conductivity and flow increase near the surface. At or below sea level the hydraulic gradient is evened out and therefore the flow rate is very small there.

The hydraulic gradients increase considerably in the vicinity of a tunnel causing a different flow pattern compared to before tunnelling, when the groundwater flow is small due to the small gradients caused by natural water levels and the hydraulic head distribution at deeper levels. Fig. 2 shows the flow pattern and hydraulic head distribution of a hypothetical homogeneous case for the Äspö HRL tunnel at 200 m below sea level (Fig. 1). No hydraulic resistance was assumed around the tunnel. Even though the rock is definitely not a homogeneous porous medium, the water flow into the tunnel can still be illustrated by this flow model. The flow lines in Fig. 2 are regular because of the assumed homogeneity. In reality the flow path will be irregular, but on a scale greater than some 50 m the flow lines will most probably have approximately the same pattern as in the simplified model. Note that a significant part of the inflow comes from aquifers that are situated deeper than the tunnel position considered.

Figure 2.

Calculated hydraulic head distribution and flow lines around the Äspö HRL tunnel. Modelled hydraulic head distribution (m) is shown as isobars around the tunnel when it passes 200 m below sea level. No hydraulic resistance around the tunnel is assumed. The flow lines for particle traces (backtracked from the tunnel) are evenly distributed around the tunnel.


Groundwater under land in Sweden has in general a meteoric origin. The infiltrating water is almost ‘pure water’ from rain or melting snow with dissolved air as an important component. The processes in the biologically active soil zone are therefore very important for the composition of recharge water. Oxygen will be consumed and carbon dioxide added. The carbonic acid will react with minerals such as calcite and feldspars and form carbonate ions and release calcium and alkali ions to the water. Ion exchange with clay minerals may affect the proportions between cations. Organic materials such as humic and fulvic acids and other substances will be added to the water from the soil. The biological processes will also have a similar influence if seawater infiltrates through organic rich sea sediments.

At great depths or under the sea bottom, saline water is found where chloride is the dominating anion. The most common cation in saline groundwater is either sodium or calcium. The saline water may have a marine origin but other end members are also possible, depending on location and other conditions. Very deep, at depths of 1000–1500 m or more, the salinity can be very high and reach well above ocean seawater and even approach brine composition, i.e., 10% or more. It is also common that saline groundwater is found at shallower depth in coastal regions than further inland. This may of course be relict seawater that infiltrated several thousands of years ago, when land near the coast in Sweden was covered by the sea due to the glacial depression (land pressed down by the ice cover). The infiltration of seawater continued until land was reclaimed by the land uplift, which is still continuing in Sweden. However, an alternative explanation can be found in the lack of driving hydraulic force under the ‘flat’ surface of the sea. With no, or very low, hydraulic gradient in the groundwater beneath the sea bottom, fossil saline conditions can be preserved for very long time periods and it must not always be the result of a relatively recent infiltration of seawater. In other words, saline water may have originated even far before the last glaciation some 10 000 years ago.

Some typical groundwater compositions to be expected at different depths and locations encountered in the course of research and exploration within the Swedish radioactive waste management program are given in Table 1. It is obvious that major constituents such as the cations sodium and calcium and the anions bicarbonate and chloride can vary considerably in concentration depending on where and at which depth the samples have been taken. Chloride behaves conservatively but many other ions obviously interact with the minerals. This is particularly evident in groundwater with a marine origin. An example of that is the ion exchange of calcium for sodium and vice versa. A further observation is that ions like potassium and magnesium, which are common in seawater, are evidently suppressed in groundwater – presumably by reactions with the minerals. Even sulfate is partly consumed, probably by sulfate reducing bacteria. Carbonate is less common at depth. Possible explanations are that slow reactions with the rock minerals cause precipitation of carbonate as calcite and autotrophic microbial organic carbon and methane formation.

Table 1.  Chemical parameters of granitic Swedish ground water from boreholes in Finnsjön [42], Klipperås [43]and Äspö[44]
Area FinnsjönKlipperåsÄspö
Borehole KFI09KFI09KKl01KKl09KAS03KAS03
Na+mg l−1415150047156003050
K+mg l−1671127
Ca2+mg l−1115170014291624400
Mg2+mg l−11684232050
Sr2+mg l−1375
Fe2+mg l−10.560.340.
Mn2+mg l−10.190.360.100.20
HCO3mg l−128532801206111
Fmg l−1394322
Clmg l−16805200456123012 300
Brmg l−12270.40.05585
Img l−
SO2−4mg l−11753401.54.332720
HSmg l−
NO2 (N)mg l−10.0010.001
NO3 (N)mg l−
NH+4 (N)mg l−10.040.01
HPO2−4 (P)mg l−10.0010.0040.0010.0030.0020.002
SiO2 (Si)mg l−
TOCmg l−1181.
Uμg l−

The pH of granitic groundwater in Sweden is buffered by the carbonate system. Calcite is abundant as mineral and feldspars can also react with acids. Therefore ‘acid rain’ or any similar disturbance of pH does not propagate very far down underground. Deep groundwater does not contain any oxygen. Measurements of redox potential with Eh electrodes give values between −100 and −400 mV. There is a dependence of Eh on pH and Fe2+ concentration but the low concentrations of redox active species in groundwater make the measurement of Eh a delicate operation. In situ measurement has been found to offer the best quality [18]. The low content of for example Fe2+ gives the water only a low redox buffer capacity. However, a considerable capacity is contained in the rock and its content of iron(II) minerals and pyrites [19]. Groundwater contains dissolved gases such as nitrogen, carbon dioxide, methane, hydrogen, helium (Table 2), neon, argon, krypton and radon. Oxygen is only found at relatively shallow depths.

Table 2.  The content of nitrogen, hydrogen, helium and carbon containing gases, and the total volumes of gas extracted from groundwater from the Stripa borehole V2, the Laxemar borehole KLX01 and the Äspö boreholes KR0012, 13 and 15 [8, 9, 13]
BoreholeSampling depth (m)N2 (μl l−1)H2+He (μl l−1)CO (μl l−1)CO2 (μl l−1)CH4 (μl l−1)C2H6 (μl l−1)C2H2–4 (μl l−1)aVolume of extracted gas (%)
  1. aThe content of C2H2+C2H4.

V2799–80725 000<10<1    32 245  0.3<0.12.4
V2812–82131 000<10<1    11 170  0.6<0.13.4
V2970–124024 500<10<1    10 290  2.9<0.12.7
KLX01830–84146 5004600  0.5   460  26<0.1<0.15.7
KLX01910–92137 0003500  0.1   500  27<0.1<0.14.4
KLX01999–107818 0002450  0.7 1 600  31<0.1<0.13.5
KR0012 6822 000  40  0.1 6 0501030<0.1  0.12.9
KR0013 6825 000 110  0.2 9 6401970<0.1  0.13.7
KR0015 6822 000  64  0.115 0374070<0.1  0.14.0

3Getting access to the deep granitic environment

3.1Drilling and tunnelling

All sampling of subterranean environments requires substantial efforts in drilling or tunnelling. The possibility of microbial contamination of the sampled specimens by access operations is indisputable and must be considered when interpreting the obtained results. Conditions like the geological formation, the history of a borehole or a tunnel, available equipment and the type of sample considered are variables that will influence the prospect of getting non-contaminated samples. Depending on these prevailing conditions, realizable precautions against contamination vary from virtually none to specific devices aimed at sterile sampling [4]. Coring crystalline bedrock requires vigorous drilling action with high drilling fluid pressures. The risk of microbial contamination of the aquifers with the drill water used to transport the drill cuttings out of the borehole during drilling is obvious. Some different measures can be applied to reduce such contamination. Clean drill water as free from microbes as possible is an essential prerequisite. Pumping of a borehole to measure its maximum hydraulic water capacity is often done and will concurrently clean the aquifers and the borehole from drill water, mud and cuttings. In addition, a control of the mixing of drill water in the groundwater can be made by introducing different tracers in the drill water that can subsequently be analyzed for in the groundwater samples [4]. The necessity of clean drilling equipment free from contaminations is evident, but not always achievable.

3.2Culturing and molecular control for contamination

The use of culturing methods alone for the control of potential contamination of boreholes is not enough due to the well documented ‘great plate count anomaly’ of environmental samples as reviewed by Amann et al. [20]. Non-culturing techniques such as extraction and sequencing of the 16S rRNA gene can be applied to overcome that problem [5, 10, 15]. Recently, culturing and molecular techniques were used to investigate possible and lasting contamination of boreholes drilled in crystalline bedrock at the ‘SELECT’ site in the Äspö HRL, at depths of 300–440 m (Fig. 1). Samples were collected from the drill water, the drilling equipment and from the drilled boreholes and analyzed. Total numbers of bacteria, viable aerobic and anaerobic plate counts and most probable numbers (MPN) of sulfate reducing bacteria (SRB) were performed parallel with the analysis of 16S rRNA gene diversity of the samples (Table 3).

Table 3.  Drilling and sampling schedule and counts of microorganisms in groundwater from newly drilled boreholes and the drilling equipment
Sampled siteSection sampled (m)Sampling dateDrilled borehole16S rDNA extraction and sequencingTotal number of bacteria ml−1×105 (S.D.)Aerobic viable count of bacteria ml−1 (S.D.)Anaerobic viable count of bacteria ml−1 (S.D.)MPN SRB ml−1
  1. Borehole HD0025A (Fig. 1) was used as a source of drill water during drilling of the other boreholes.

  2. aS.D., standard deviation.

  3. bNo data.

  4. cImpossible to count due to background fluorescent precipitates.

  5. dNot relevant.

HD0025A0–17.094/11/30KA3005Ayes0.806 (0.20)80 (61)75 (34)100
HD0025A0–17.094/12/14KA3105Ano0.851 (0.12)b
HD0025A0–17.095/01/17KA2858Ayesic70 (28)60 (14)100
Tubingn.r.d95/01/17KA2858Ayes22.3 (0.60)>100 000>100 000100
Drill water containeran.r.94/12/14KA3105Ayes1.030 (0.81)44 500 (22 700)1 930 (808)10
Drill water containeran.r.95/01/17KA2858Ayes4.34 (0.33)7 4003 70010
Drilling machinean.r.95/01/17KA2858Ayes3.94 (0.25)6 6003 100100
KA2858A0–59.795/02/02n.r.yes0.12 (0.007)10
KA3005A0–58.194/12/14n.r.no0.210 (0.01)9 (8)10 (10)0
KA3105A0–7095/01/17n.r.yes0.30 (0.01)13 (10)<2>1 000
KA3105A17.0–19.596/02/02n.r.no0.92 (–)10 000

The measures taken to avoid contamination of the boreholes included steam cleaning of all temporary and permanent equipment with a hot water high pressure cleaner. Using deep groundwater from borehole HD0025 at site level as drill water (Fig. 1) excluded the possibility of introducing contaminating microorganisms with surface originated drill water. This possibility to use surface water as drill water was considered during the planning process but was abandoned due to the obvious risk of introducing oxidized water into the rock aquifers. Part of the SELECT site will be used for studies on the reducing capacity of fracture minerals and oxygenated water would have spoiled that series of experiments. An obvious source of microbial contamination was the tubing taking drill water from HD0025A to the drill water container as can be judged from significantly elevated total and viable counts in samples from the tubing compared to the other sampled sites (Table 3). This tubing was a flexible, reinforced rubber tube of a type that could withstand the significant pressure of groundwater in HD0025A, approximately 40 atmospheres. The distances between this borehole and the drilled boreholes KA2558A and KA3105A (Fig. 1) were 642 m and 95 m respectively, which required at least the same tube lengths. Rubber tubing material contains organic component such as softeners and stabilizers which slowly leak out to the water in the tubing. Therefore, such tubing is susceptible to biofilm formation of microorganisms that can grow with these compounds [21, 22]. Cleaning efforts will not be lasting (unless the tubing is totally sterilized, which was impossible) as new biofilms will develop as soon as water enters the tubing again.

The source of microbial contamination from the tubing offered an excellent possibility to evaluate whether large numbers of certain species in drilling equipment will cause lasting contamination of a borehole with the same species. The culturing technique could not detect such contamination of the drilled boreholes (Table 3). A similar result was obtained when the diversity of 16S rRNA genes in the samples was compared (unpublished results, not shown). Two strains of the genus Shewanella were indicated to be dominating in the drill water and in HD0025, but they could not be detected in water from the drilled boreholes. In conclusion, it could not be proved that microorganisms in the drilling equipment contaminated the drilled boreholes. This is an expected result, because the growth conditions are different in the rock compared to the drilling equipment. Microorganisms that are adapted to grow on the nutritious tube walls would probably not survive in the nutrient poor rock environment and vice versa.

4Life in hard rock tunnels

Excavation for tunnels, mining etc. introduces several changes into the subterranean environment that will induce activities in the tunnel by microorganisms other than those present in the fractured rock. Oxygen is normally introduced into tunnels by ventilation which makes growth of aerobic bacteria possible. As the groundwater at depth is usually anoxic with a low redox potential (Tables 1 and 2), marked redox and oxygen gradients will develop when such groundwater reaches the oxygenated tunnel atmosphere. Typical redox pairs participating in these gradients are manganese(II) oxidizing to manganese(IV), ferrous iron to ferric iron, sulfide to sulfate (Table 1) and probably also methane to carbon dioxide (Table 2). Such gradients are the habitats for many different lithotrophic and also heterotrophic bacteria. Among them are the iron, manganese, sulfur and methane oxidizing bacteria that generate chemical energy for anabolic reactions through the oxidation of reduced inorganic compounds and methane with oxygen. The energy gained by the lithotrophs is used to reduce carbon from CO2 to organic carbon and this is the first step in an environmental succession that eventually ends as a reduced environment again.

Commonly, seeps of groundwater from fractures intersected by the Äspö tunnel or flows of groundwater from boreholes turn light brown to dark brown from precipitates that sometimes can be very voluminous. They usually appear within some weeks after excavation/drilling and have in some cases reached a thickness of 10 cm or more. The most frequently occurring inhabitant in these precipitates is the lithotrophic iron oxidizing bacterium Gallionella ferruginea[23–26]. It forms moss like covers on rocks and sediments in ponds in the tunnel and is very abundant close to the outflow of groundwater from rock wall fractures [7]. At many such outflows, white, threadlike structures are observed. Microscopic observation has revealed them to be sulfur oxidizing bacteria of different types, both extracellular and intracellular deposition of sulfur has been observed. Especially tunnel sections below the sea bed with ongoing sulfate reduction harbor this type of bacteria [7]. Sequencing the 16S rRNA gene from one of these sites has indicated the genus Thiotrix to be present (not published).

5Diversity and distribution of bacteria in granitic groundwater

5.1Culturing methods

Total numbers of microorganisms in subsurface granitic environments range from some 103 up to 107 cells ml−1[5]. We have used culturing techniques with numeric taxonomy for the phenotypic characterization and the 16S rRNA technique for genotype characterization to determine bacterial diversity in granitic groundwater. Facultatively anaerobic, heterotrophic bacteria were identified from boreholes KAS02 and 03 during the pre-investigation phase at Äspö as belonging to the genera Pseudomonas and Shewanella[12]. Later, identification of heterotrophic facultative bacteria from the Äspö tunnel demonstrated that members of the Serratia, Bacillus, Desulfovibrio, Desulfomicrobium, Eubacterium and Methanomicrobium genera are also present [15, 27].

5.2Sequencing 16S rRNA genes

The finding of many new and unknown bacterial 16S rRNA sequences in natural environments is a commonly reported result [10, 28–30]. This was also the case when 16S rRNA genes from the Äspö HRL tunnel boreholes were compared with DNA databases [15]. There is not any accepted value of the percent identity at which two 16S rRNA genes can be concluded to belong to the same genus or species. It can be quite different for different genera [31]and is also due to whether total or partial 16S rRNA genes are compared. It has been suggested, based on a comparison of rRNA sequences and on DNA-DNA reassociation, that a relation at the species level does not exist at less than 97.5% identity in the 16S rRNA sequence. At higher identity values, species identity must be confirmed by DNA-DNA hybridization [32]. Fig. 3 shows a phylogenetic tree for 48 Äspö clone group sequences. Six distinct groups of phylogenetically related bacteria were found [33], the alpha, beta, gamma, delta and epsilon groups of the Proteobacteria, and Gram-positive bacteria. The remaining sequences were only very distantly related to known, named and sequenced bacteria reported to the databases. Accepting the level of 97.5% conservatively, as identifying a sequence approximately at the genus level, some conclusions can be drawn about the sequences from the Äspö granitic groundwater. The Bacillus (A5g, 98.6%), Desulfovibrio (A6-7hq, 97.7%) and Acinetobacter (A24optmn, 98.6%) like sequences had identities higher than 97.5% with 16S rRNA sequences in the database, and may be regarded as identified at the genus level. One of the clone groups could be identified as a member of the domain Eukarya, a yeast, Saccharomyces (A61upm, 97.6%) [15]. The only isolate whose sequence was also found in the clone libraries (clone A1ghq) was Aspo-4. The 16S rRNA of Aspo-4 showed 91.7% identity with the Gram-positive bacterium Eubacterium limosum, which is too low for identification. However, preliminary phenotypic characterization indicates this isolate to be a homoacetogenic species. It was isolated from SA813B (Fig. 1) and its 16S rRNA sequence was found in groundwater and surfaces from this borehole, from KR0013 groundwater and from several of the SELECT boreholes (300–440 m).

Figure 3.

Evolutionary distance tree based on the 16S rRNA gene sequences of clones from different boreholes in the Äspö HRL tunnel. Major phylogenetic groups of bacteria have been designated with their generally accepted names. As references, some 16S rRNA gene sequences of known bacteria from the EMBL database have been added to the tree and are indicated with their Latin names. The branch lengths are proportional to calculated evolutionary distances.

When PCR amplification is used for the determination of species diversity, the result may be biased due to methodological problems, such as uneven extraction of DNA and biased PCR due to differences in genome size [34]. One of the most important causes of bias is that organisms belonging to the domain Archaea have only one or a few gene copies of the 16S rRNA gene while bacteria can have several copies, 5–7 or more, and this will bias PCR amplification towards bacteria [20, 29]. Therefore, using PCR primers that are specific for archaean 16S rRNA gene sequences in parallel with universally conserved ones will enhance the detection of microorganisms belonging to the domain Archaea. The results presented in Fig. 3 were obtained using the universal primers only and should therefore be expected to reveal mainly bacterial diversity and distribution.

5.3In situ hybridization with group specific nucleic acid probes

The inability of our universal primers to detect members of the domain Archaea led us to use nucleic acid probes [20]for the possible in situ detection of methanogens in Äspö groundwaters, inferred by the presence of methane (Table 2) with a biogenic signature [19]. A difficult problem using fluorescent microscopy on fracture surfaces from granite is the intensive background autofluorescence from various accessory fracture coating minerals and precipitates at short wavelengths typical for DAPI, FITC, acridine orange and rhodamine fluorochromes. We have solved this problem by using a fluorochrome for the infrared part of the spectrum, Cy-5, with an excitation maximum at 647 nm and emission above 660 nm. The background fluorescence from various fracture surfaces is very low with this stain (Fig. 4A). Groundwater from the Bockholmen site (Fig. 1) carried biogenic methane [19]. One of the boreholes, KR0013 (68 m), was connected to a 50 l canister filled with 0.5–1 cm large crushed granite that would act as a substrate for attachment of microorganisms in the groundwater slowly passing at flow rates below 0.1×10−3 m s−1. In situ hybridization with a nucleic acid probe for the domain Archaea on the granitic surfaces after 3.5 years of exposure revealed a positive signal (Fig. 4B). Most likely, Fig. 4B shows attached methanogens growing in a chain, as other members of the domain Archaea generally prefer more extreme pH, salinity or temperature conditions than prevailing in the Äspö granite.

Figure 4.

A: Granite rock coupons were exposed overnight to a growing culture of Shewanella putrefaciens and washed with a buffer. Subsequent in situ hybridization with a Cy-5 labelled probe for the domain Bacteria (EUB-338) revealed attached bacteria on the surface. The Cy dyes are based on the cyanine fluor and all seven different fluors offer intense colors with narrow emission spectra (Amersham Life Science). A Molecular Dynamics 2010 confocal laser scanning microscope equipped with a Kr/Ar laser was used for observation with the software Image Space running on a Silicon Graphics UNIX based computer. The hybridization signal obtained was maximal with virtually no background at all, as can be seen from the intensity diagram reflecting a section over an attached bacterium. B: Small stones of granite were exposed to flowing groundwater for 3.5 years and in situ hybridized with a Cy-5 labelled probe for the domain Archaea (ARC-915). A chain of growing archaeal microorganisms is displayed from a top (0° relative to the light path in the microscope) and a side view (90° relative to the light path in the microscope) using the image processing program. The scanned depth was 21.6 μm. The depth resolution is about three times less than the side resolution which gives the microorganisms a three times too thick appearance in the side view. Considering this artefact, it can be concluded that the observed signal is emitted from a threadlike structure with the size of typical prokaryotic cells growing in chains, presumably a methanogen (see text for details).

6Biogeochemical processes

6.1Iron reducing bacteria

Iron reducing bacteria were discovered to be of major biogeochemical importance in granitic rock during a block scale redox experiment at the Äspö HRL. The unexpected redox stability of the studied system could only be explained by the mobilization of solid phase ferric iron oxy-hydroxides to liquid phase ferrous iron by iron reducing bacteria with organic carbon as electron donor [14, 19, 35]. We have isolated several different bacteria from this habitat able to reduce ferric iron to ferrous iron, including Shewanella putrefaciens[15]. The 16S rRNA gene sequences show that several of the dominating species sampled from the Bockholmen fracture zone (Fig. 1) have a 95% or more identity with known IRB like Pseudomonas medosina[7]. Our results imply that much of the ferrous iron in anoxic groundwater (Table 1) may be a product of microbial iron reduction and not only due to pure inorganic redox reactions.

6.2Sulfate reducing bacteria

Sulfate reducing bacteria frequently appear in the Äspö HRL environments at depths greater than approximately 100 m; isolates as well as 16S rRNA genes related to sulfate reducing bacteria have been found [15]. Sulfide production is of particular interest for the disposal of spent nuclear fuel in copper canisters because sulfide is the only substance present in deep groundwater that will cause corrosion of copper. Oxygen, another copper corrodant, is not present in deep groundwater and sulfate will not react with copper unless microbes reduce it to sulfide. Therefore, evidence and indications of sulfate reduction based on geological, hydrogeological, groundwater, isotope and microbial data in and around the Äspö HRL tunnel were evaluated by a multidisciplinary research group [36]and the most important conclusions are given below.

Geological data were evaluated to find the amount of sulfide which could be calculated to result from the sulfate reduction. The conclusion is that the amount of pyrite normally occurring in the fracture coatings could explain the amount produced. However, there are other processes in the geological time span which have also produced pyrite. Therefore, the existence of pyrite is not a conclusive evidence of sulfate reduction.

The hydrogeological conditions were evaluated in order to describe possible transport phenomena related to the sulfate reduction. The questions to be answered were: Can sulfate reduction take place in the sea bottom sediments and the resulting sulfide be transported with groundwater to the tunnel? Could the groundwater flow conditions in the tunnel either increase or decrease the effect of biological sulfate reduction? The answer to the first question is yes, the process can occur in the sea bed sediments and the effect on hydrochemistry can be observed in the water inflow in the tunnel. Hydrogeological calculations imply a transport time of approximately 100–400 days for the water passing through the sediments to reach the tunnel in a proportion of 25%. The answer to the second question is that the relatively simple groundwater flow conditions around the tunnel would not affect the biological process directly. However, if the sulfate reduction had been an ancient process, then the effects would soon be washed out, which has not been the case. In addition, the existence of high bicarbonate and low sulfate concentrations in the probing holes on the very first sampling occasion after the tunnel was excavated strongly imply that the process is ongoing.

The groundwater chemistry was evaluated by multivariate mixing and mass balance calculations. The calculations demonstrated that an understanding of the fluxes of compounds, rather than measurements of concentrations only, is necessary for modelling sulfate consumption and bicarbonate production by SRB. These calculations defined the specific conditions where the process could be ongoing. The results show that a salinity range of 4000–6000 mg l−1 of chloride is the optimal one. Sulfate reduction seems to occur in anaerobic brackish groundwater with access to dissolved sulfate and organic carbon or hydrogen. These conditions are mainly found in the sea bed sediments, in the tunnel section under the Baltic Sea and in some deep groundwaters, such as those in the SELECT tunnel section.

Isotope data were expected to give a definite answer to where the sulfate reduction takes place, since the bacterial processes always result in an enrichment of the lighter isotopes. Concerning both the δ-13C and the δ-34S isotopes the results generally point towards the existence of bacterial sulfate reduction. However, there are several processes in the geological evolution which could have given the same isotopic signatures as well. Therefore, the isotopic data provide indications of biological sulfate reduction but no evidence.

Microbiological data were collected in boreholes where the hydrochemistry indicated an ongoing or previously ongoing sulfate reduction. The results show that sulfate reducing bacteria are present, sometimes in large quantities (Table 3), and that they can be correlated to a groundwater composition with high bicarbonate and low sulfate concentrations.

7Hydrogen-dependent microorganisms

7.1Hydrogen and methane in deep groundwater

Hydrogen is expected to act as an inert gas in most geochemical reactions and it is therefore usually overlooked and not analyzed for. Some data on hydrogen in hard rock were recently published [37, 38]. Values of 2.2–1574 μM hydrogen in groundwater from Canadian shield and Fennoscandian shield rocks was found. Most granitic rocks shows low but significant radioactivity which can generate hydrogen by radiolysis of water. Anaerobic mineral reactions (e.g. anaerobic corrosion of iron) will also create hydrogen [6]. Finally, deep mantle gases contain hydrogen. Methane occurs frequently in subterranean environments all over the globe and the stable isotope profile commonly indicates a biogenic origin of the methane. Values of 1.3–18 576 μM methane in groundwater from Canadian shield and Fennoscandian shield rocks [37, 38]and 1–181 μM methane in Swedish groundwater have been published previously (Table 3) [8, 9, 13]. Recent data indicate up to 720 μM methane down to 440 m depth at Äspö HRL [45]. More support for an ongoing methane generating process in deep Swedish granite is provided by the upflow of gas, mainly methane, from fracture zones below sea bottom sediments [39–41]. Pockmarks in Baltic sea sediments have been observed, indicating gas eruption from fracture systems in the underlying granite.

7.2Acetogenic bacteria in deep granitic groundwater

Acetogenic bacteria have the capability of reacting hydrogen with carbon dioxide to acetate, thereby producing ATP and reducing power for metabolism. One acetogen has been isolated from the Äspö tunnel groundwater (Aspo-4). It is a Gram-positive strictly anaerobic Eubacterium-like species (see Pedersen et al. [15]for details). We have recently found 16S rRNA sequences identical to Aspo-4 also in groundwater at the SELECT site in the Äspö HRL tunnel. Pilot experiments with hydrogen addition to such groundwater resulted in rapid acetate production. The acetate produced can be used by acetoclastic methanogens, iron and sulfate reducing bacteria and other heterotrophic microorganisms, thereby constituting a transformation route of inorganic carbon to organic carbon with hydrogen as the reductant (Fig. 5).

Figure 5.

The deep hydrogen driven biosphere hypothesis, illustrated by its carbon cycle. At relevant temperature and water availability conditions, subterranean microorganisms are theoretically capable of performing a life cycle that is independent of sun driven ecosystems. Hydrogen and carbon dioxide from the deep crust of the earth or from sedimentary deposits of organic carbon can be used as energy and carbon sources. Phosphorus is available in minerals like apatite and nitrogen for proteins, nucleic acids etc. can be obtained via nitrogen fixation; this gas predominates in most groundwaters (Table 2).

7.3Methanogens in deep granitic groundwater

Our results on the presence, diversity and activity of methanogens in 19 different boreholes at Äspö HRL (10–440 m depth) are being prepared for publication. Briefly, the following was found [45]. Pure cultures of autotrophic, rod like methanogens have been isolated and 16S rRNA sequencing indicates them to be related to the genus Methanobacterium[27]. Viable cell counts (MPN) varied from 10 to 4.3×105 methane producing archaea ml−1. Direct counts of autofluorescent cells (fluorescence of the archaea specific coenzyme F420) varied from 1.4×102 to 7.4×105 cells ml−1. Comparisons of total numbers of acridine orange stained cells indicated that methanogens can constitute up to 60–80% of the cell population inhabiting the investigated granitic rock system. The preliminary indication is that the distribution, numbers and physiological diversity of the found methanogens are governed by the carbon dioxide concentration, salinity and organic carbon content. Oligotrophic, methylotrophic and acetoclastic methanogens dominate in shallow rock (0–190 m) with relatively high organic carbon content (7.1–18 mg l−1), while autotrophic methanogens prevail in deep boreholes (190–440 m) with a lower (0.9–4.0 mg l−1) content of organic carbon.

8Conclusions – The deep hydrogen driven biosphere

We have been working on the deep subterranean biosphere (down to 1240 m) for almost 10 years now and the list of references shows our progress. Throughout our work, numerous results have indicated the presence of autotrophic microorganisms utilizing hydrogen as a source of energy in the deep environment [12]. We indicated the possibility of a hydrogen driven biosphere in deep granite in 1992 [8, 46], but solid evidence was lacking. Recent results obtained by us in the Äspö HRL tunnel now show that autotrophic methanogens, acetogenic bacteria and acetoclastic methanogens are all present and active in the investigated groundwaters.

Fig. 5 schematically depicts possible routes of carbon and energy in a subterranean hydrogen driven biosphere. Our present research task is collecting evidence for this model, and we concentrate on archaean organisms and homoacetogenic bacteria. The presence and activity of iron and sulfate reducing bacteria are well documented [7, 14, 36]as described above, and have been included in the subterranean biosphere model earlier [5].

Until recently, it has been a general concept that all life on earth depends on the sun via photosynthesis, including most of the geothermal life forms found in deep sea trenches as they use oxygen for the oxidation of reduced inorganic compounds (almost all oxygen on earth is produced via photosynthesis). Here, it is suggested that a deep subterranean granitic biosphere exists, driven by the energy available in hydrogen formed through radiolysis, mineral reactions or by volcanic activity. Knowledge on this biosphere is just beginning to emerge and it will expand the spatial borders of life from a thin layer on the surface of the planet Earth and in the seas to a several kilometers thick biosphere reaching deep below the ground surface and the sea floor. If this hypothesis is true, life may have been present and active deep down in Earth for a very long time, and it cannot be excluded that the place for the origin of life was a deep subterranean igneous rock environment (probably hot with a high pressure) rather than a surface environment. Some of the species closest to the root of the phylogenetic 16S rRNA tree, as known today, are obligately hydrogen utilizing thermophiles, Aquifex pyrophilus and Methanopyrus belonging to the domains Bacteria and Archaea respectively, supporting the idea of a deep hot origin of life. A rather spectacular conclusion is that life on other planets should probably not be searched for only on the surface but rather deep down in the subsurface.


The author wishes to thank Johanna Arlinger, Susanne Ekendahl, Lotta Hallbeck, Louise Holmquist, Nadi Jahromi, Fred Karlsson, Svetlana Kotelnikova and Marcus Laaksoharju, for their contributions and also numerous colleagues at the Äspö Hard Rock Laboratory and at the Swedish Nuclear Fuel and Waste Management Co., Stockholm, Sweden, who made our research possible. This work is supported by the Swedish Natural Science Research Council, The Swedish Nuclear Fuel and Waste Management Co., and Knut and Alice Wallenbergs Foundation for Scientific Research.