Distribution of culturable microorganisms in Fennoscandian Shield groundwater

Abstract

Microbial populations in 16 groundwater samples from six Fennoscandian Shield sites in Finland and Sweden were investigated. The average total cell number was 3.7×10⁵ cells ml⁻¹, and there was no change in the mean of the total cell numbers to a depth of 1390 m. Culture media were designed based on the chemical composition of each groundwater sample and used successfully to culture anaerobic microorganisms from all samples between 65 and 1350 m depth. Between 0.0084 and 14.8% of total cells were cultured from groundwater samples. Sulfate-reducing bacteria, iron-reducing bacteria and heterotrophic acetogenic bacteria were cultured from groundwater sampled at 65–686 m depth in geographically distant sites. Different microbial populations were cultured from deeper, older and more saline groundwater from 863 to 1350 m depth. Principal component analysis of groundwater chemistry data showed that sulfate- and iron-reducing bacteria were not detected in the most saline groundwater. Iron-reducing bacteria and acetogens were cultured from deep groundwater that contained 0.35–3.5 mM sulfate, while methanogens and acetogens were cultured from deep sulfate-depleted groundwater. In one borehole from which autotrophic methanogens were cultured, dissolved inorganic carbon was enriched in ¹³C compared to other Fennoscandian Shield groundwater samples, suggesting that autotrophs were active. It can be concluded that a diverse microbial community is present from the surface to over 1300 m depth in the Fennoscandian Shield.

Acetogenesis, Groundwater, Iron reduction, Methanogenesis, Principal component, Subsurface, Sulfate reduction

1 Introduction

Microorganisms are widespread in the Earth's subsurface [1]. It has been estimated that the microbial biomass of the Earth's subsurface equals that in the surface biosphere [2]. The knowledge of subsurface ecosystems and the impact they have on the biogeochemical cycles of our planet is growing. Subsurface sediments and sedimentary rocks contain microorganisms from the time of their formation. Conversely, igneous rock is formed at very high temperatures and microorganisms can only populate the rock once it has cooled and water-bearing fractures have been formed. Microorganisms have been observed in fractures in both terrestrial and subseafloor igneous rock [3–6.

Igneous rock of the Fennoscandian Shield extends over Scandinavia and western Russia [7]. Subsurface microorganisms in the Fennoscandian Shield have been studied for many years at the Äspö Hard Rock Laboratory (HRL) in southeastern Sweden, and the Stripa mine in central Sweden [8]. Subsurface microbial communities have been detected in Äspö HRL and Stripa groundwater by culturing, by autoradiographic methods, and by DNA extraction and 16S rRNA sequencing [9,10–15. These subsurface microorganisms include methanogenic archaea, homoacetogenic bacteria, sulfate-reducing bacteria (SRB), and iron-reducing bacteria (IRB). Culturable microbial populations of groundwater in boreholes drilled at four sites in Finland included SRB, IRB and acetogens [16]. SRB and IRB were cultured from boreholes that contained sulfur- and iron-containing fracture minerals, respectively [16].

The purpose of this research was to study groundwater microbial populations from several sites and from a wide range of depths in the Fennoscandian Shield. These groundwater samples were analyzed using the same methods, which allowed microbial populations from geographically distant sites and varying depths to be compared. Anaerobic microorganisms were cultured and enumerated from 16 groundwater samples from six Fennoscandian Shield sites. Principal component analysis was used to determine patterns in a complex groundwater chemistry data set, and microbiological results were superimposed on the principal component plot.

2 Materials and methods

2.1 Sample collection

The main rock types in the Fennoscandian Shield are granites, granodiorites and gneisses, which range in age from 1.5 to 3.5 Ga [7]. Groundwater distribution in igneous rock is heterogeneous, with flow occurring in fractures. Boreholes drilled from the surface or from subsurface tunnels allow access to groundwater. Five to 30 meter long water-conducting sections of boreholes were isolated using inflatable packers and groundwater was collected from within these sections [17]. Samples were collected at four sites in Finland and two sites in Sweden (Fig. 1).

1

Investigation site locations in Sweden and Finland.

At Olkiluoto, Hästholmen, Kivetty, and Romuvaara in Finland, groundwater was sampled from boreholes that extend from the surface to depths of up to 1000 m (Table 1). Finnish samples were collected using the PAVE (a Finnish acronym for pressure vessel) groundwater sampling system (Posiva OY, Helsinki, Finland). This system can be sterilized for microbiological sampling and groundwater samples are collected at in situ pressure [16].

1

Fennoscandian Shield sites, borehole numbers and depths from which samples were collected

Site	Borehole number	Sample depth (m)
Olkiluoto	3	441
	4	863
	9	473
Hästholmen	4	686
	5	385
	6	65
Kivetty	13	497
Romuvaara	11	543
Äspö HRL	3566:2	450
	3573:1	450
	3573:2	450
	3600:1	450
	3600:2	450
Laxemar	2	1160
	2	1350
	2	1390

In Sweden, samples were collected at Äspö HRL and Laxemar (Fig. 1). Äspö HRL is a tunnel excavated to a depth of 450 m below the surface, and the boreholes investigated in this study were drilled from the bottom of the tunnel. The subsurface boreholes are artesian, so groundwater was collected directly. Boreholes were opened to flush several borehole volumes before samples were collected into sterile N₂-flushed bottles. Laxemar borehole KLX02 extends from the surface to a depth of 1700 m in Laxemar, which is located on the mainland, 2 km west of Äspö Island. Laxemar samples were collected with the PAVE system.

Methods for sterilization of the PAVE system for microbiological sampling have been described elsewhere [16]. Briefly, PAVE samplers were flushed with chlorine dioxide, and then rinsed with deionized, filtered water. Control samples were analyzed to ensure the sterility of the PAVE system. The control samples included deionized, filtered water used in sterilization, as well as, on two separate occasions, PAVE samplers sterilized and containing deionized, filtered water. Total numbers of cells were counted in the control samples as described below. Four of the deionized, filtered water samples tested had fewer cells than the detection limit of 1.4×10³ cells ml⁻¹. The final deionized, filtered water sample contained 1.7 (±1.6)×10³ cells ml⁻¹, where the variability is standard deviation. The two PAVE control samples contained 3.8 (±2.2)×10⁴ cells ml⁻¹ and 6.3 (±1.4)×10⁴ cells ml⁻¹. These total cell numbers are smaller than in all but one groundwater sample tested in this study (Hästholmen borehole 5, see below). Water from the two PAVE samplers was inoculated into freshwater culture medium described previously [16]. No microbial growth was observed after incubation for 3 months at 17°C.

2.2 Principal component analysis

The evolution of groundwater is generally strongly related to present and past flow conditions. As this is a continuous process, the type of infiltrating water, as well as that already existing in the rock, changes groundwater compositions. The solute and isotopic content of groundwater samples can be interpreted either as the result of geochemical reactions between the groundwater and the minerals it contacts, or as the mixing of groundwater types of different origins (and hence different chemical signatures), or as a combination of both processes. One of the software tools addressing this topic is the Multivariate Mixing and Mass balance (M3) model [18], an interpretative technique used to perform a cluster analysis (using multivariate principal component analysis) in order to simplify and summarize groundwater data. M3 identifies waters of different origins and infers the mixing ratio of these mixing reference waters (end-members) to reproduce the chemistry of each sample. In addition it identifies any deviations between the chemical measurements of each sample and the theoretical chemistry from the mixing calculation, and interprets these deviations as resulting from groundwater reactions.

The groundwater chemical data from the boreholes investigated for culturable microorganisms were compared with other chemical data from Fennoscandian Shield sites to obtain information on the types of water found in the studied boreholes and on the origin of the groundwater. The M3 model consists of three steps: standard multivariate analysis (principal component analysis, PCA), mixing and mass balance calculations. Multivariate techniques are useful in gathering information from many chemical variables, and the information is used to model the mixing and reactive processes affecting the obtained groundwater composition. Here, the first two steps were applied, namely PCA and some mixing calculations.

2.3 Total cell numbers

Total numbers of cells were determined in four aliquots of each groundwater sample. Methods for staining cells and for counting by epifluorescence microscopy have been described previously [12,16. Total numbers of cells were calculated as the average of the four aliquots for each sample, and variability was calculated as sample standard deviation.

2.4 Culture media

Two different methods were used to prepare culture media modified such that chemical compositions close to those of each groundwater sample were obtained. All media were prepared anaerobically.

2.4.1 Synthetic media

Methods for design of synthetic media based on groundwater chemistry data were described previously [16]. All synthetic media contained (g l⁻¹ double-distilled water): resazurin, 0.0002; KH₂PO₄, 0.01; Na₂SO₄, 0.002; FeCl₂.4H₂O, 0.001; cysteine HCl.H₂O, 0.25; Na₂S.9H₂O, 0.25; trace element solution [16], 10 ml l⁻¹; and vitamin solution [19], 5 ml l⁻¹. Salt and buffer concentrations were varied for each sample according to groundwater chemistry data. Synthetic medium for Hästholmen borehole 5 contained (g l⁻¹): NaCl, 2.7; CaCl₂·2H₂O, 2.9; MgCl₂.6H₂O, 1.3; NH₄Cl, 0.4; KCl, 0.02; NaHCO₃, 1.72. Synthetic medium for Kivetty borehole 13 contained (g l⁻¹): CaCl₂·2H₂O, 0.07; MgCl₂.6H₂O, 0.036; NH₄Cl, 0.01; NaHCO₃, 0.86; Tris–HCl, 1.21. Synthetic medium for Olkiluoto borehole 3 contained (g l⁻¹): NaCl, 2.8; CaCl₂·2H₂O, 1.2; MgCl₂.6H₂O, 0.21; NH₄Cl, 0.4; KCl, 0.02; NaHCO₃, 1.29; Tris–HCl, 1.82. Synthetic medium for Olkiluoto borehole 9 contained (g l⁻¹): NaCl, 6.5; CaCl₂·2H₂O, 6.8; MgCl₂.6H₂O, 0.5; NH₄Cl, 0.4; KCl, 0.02; NaHCO₃, 1.72. Synthetic medium for the five Äspö HRL samples contained (g l⁻¹): NaCl, 2.8; CaCl₂·2H₂O, 1.7; MgCl₂.6H₂O, 0.6; NH₄Cl, 0.4; KCl, 0.02; NaHCO₃, 1.68. The pH of the media was adjusted to borehole pH after autoclaving.

2.4.2 Groundwater-based media

Groundwater-based media were prepared with filter-sterilized groundwater. Groundwater was pumped from each borehole until pH, redox potential, electrical conductivity, oxygen concentration and temperature readings stabilized. A sterile 5-l polycarbonate bottle was filled with groundwater and shipped on ice to the laboratory at Göteborg University, Göteborg, Sweden, within 24 h after collection. Groundwater was put into an anaerobic chamber (Coy Laboratory Products, Grass Lake, MI, USA) with an atmosphere of approximately 4% H₂, 5% CO₂ and 91% N₂. To sterilize the groundwater, it was filtered with 0.22-μm nitrocellulose filters (Millipore Corporation, Bedford, MA, USA). The filtered groundwater was kept overnight in the anaerobic chamber to equilibrate with the anaerobic atmosphere and then refiltered before use.

Medium components were added from sterile, anaerobic stock solutions. Groundwater-based media contained (g l⁻¹ filter-sterilized groundwater): resazurin, 0.0002; NH₄Cl, 0.4; KH₂PO₄, 0.01; Na₂SO₄, 0.002; cysteine HCl.H₂O, 0.25; Na₂S.9H₂O, 0.25; trace element solution [16], 10 ml l⁻¹; and vitamin solution [19], 5 ml l⁻¹. Buffers were added from sterile, anaerobic stock solutions depending on borehole pH. Medium for Laxemar borehole 2 1390 m depth contained 6.5 g l⁻¹ piperazine-1,4-bis(2-ethanesulfonic acid) and 0.084 g l⁻¹ NaHCO₃, to give a pH of 6.6. Medium for groundwater with a pH of 7.0–8.0 contained 1.72 g l⁻¹ NaHCO₃. Medium for groundwater with a pH of 8.0–9.0 contained 0.86 g l⁻¹ NaHCO₃ and 1.21 g l⁻¹ Tris–HCl. Medium for Romuvaara borehole 11 contained 1.46 g l⁻¹ NaHCO₃ and 0.32 g l⁻¹ Na₂CO₃, to give a pH of 9.2. The pH of the media was adjusted to the desired pH, where necessary, with sterile, anaerobic HCl and NaOH solutions.

2.5 Most probable number

Substrates were added to separate bottles of media for each of the six physiological groups of microorganisms investigated: heterotrophic IRB, SRB, acetogens and methanogens, and autotrophic acetogens and methanogens. Media for IRB contained 125 mM amorphous ferric iron and 11 mM lactate. Media for SRB contained 14 mM Na₂SO₄ and 6 mM lactate. Media for heterotrophic acetogens contained 74 mM formate, 10 mM trimethylamine (TMA), 0.2% yeast extract and 50 mM bromoethanesulfonic acid (BESA). Media for heterotrophic methanogens contained 10 mM acetate, 10 mM TMA, 50 mM methanol and 74 mM formate. In media for autotrophic acetogens and methanogens, the sole carbon source was the NaHCO₃ in the buffer (10–30 mM; see Sections 2.4.1 and Sections 2.4.2), and oxygen-free H₂ was provided in the headspace at 200 kPa overpressure. In addition, the media for autotrophic acetogens contained 50 mM BESA. Media containing selective substrates were dispensed in 9-ml aliquots in sterile test tubes and stoppered with sterile blue rubber stoppers (Bellco Glass, Vineland, NJ, USA). Since this was done inside the anaerobic chamber, the tubes contained approximately 4% H₂, 5% CO₂ and 91% N₂ in the gas phase.

The most probable number (MPN) of each physiological group of microorganisms in each sample was determined using the media prepared with selective substrates. Groundwater samples were diluted decimally in media four to six times. Five replicates of each dilution series were prepared. After inoculation, a 1-ml aliquot of filter-sterilized groundwater was added to each synthetic medium dilution to provide any growth factors present in the groundwater but not in the medium. Groundwater was filtered with 0.2-μm DynaGard filters (Microgon, Laguna Hills, CA, USA). Negative controls for synthetic media have been described previously [16]. For the groundwater-based media, negative controls were tubes with medium only as well as medium inoculated with 1 ml groundwater and immediately killed with 2% formaldehyde. To MPN tubes and controls for autotrophic methanogens and autotrophic acetogens, 200 kPa oxygen-free H₂ was added.

MPN inoculation was completed within 4–6 h after removal of water from the PAVE samplers, and within 2 h of sample collection for Äspö HRL samples. The MPN tubes were incubated on their sides in the dark at 17°C (30°C for the Laxemar 1390 m sample) for at least 8 weeks. Tubes were observed for several months after inoculation to verify that no additional growth occurred after 8 weeks. MPN tubes were analyzed for products of metabolism as described previously [16], and tubes were marked positive if metabolic products were present at levels higher than in negative controls. MPN was calculated as described elsewhere [20]. The detection limit for MPN was 0.2 cells ml⁻¹.

3 Results

3.1 PCA

The results of the PCA analysis are based on the geochemical data set from many Fennoscandian Shield sites. The constituents used for the modelling are major components (Na, K, Ca, Mg, HCO₃, Cl and SO₄), tritium (³H) and stable isotopes (δ²H and δ¹⁸O). It is known that these components contain most of the variability of the information in groundwater data. The variability for the first and second principal components shown in Fig. 2 is 67%, which means that the first and second principal components summarize 67% of the groundwater data information. The weight for the different elements is shown in the equations for the first and second principal components respectively. From these weights the importance of the individual elements in the total analysis can be tracked. The samples from these sites and those investigated earlier plot throughout the PCA plot indicating that these waters represent a large variation in composition, affected by a mixing contribution from several end-member reference waters such as glacial, brine and precipitation. The selected reference waters identified from the PCA (Fig. 2) for the current modelling are:

2

Principal component plot generated with groundwater chemistry data for 25 samples analyzed for microbiology (closed circles) and for other Scandinavian groundwater samples (open circles). Samples 1–7 are from Olkiluoto, 8–9 from Kivetty, 10–11 from Romuvaara, 12–17 from Hästholmen, 18–22 from Äspö HRL and 23–25 from Laxemar. The plot is based on seven major components, tritium and stable isotopes (δ²H and δ¹⁸O). The first and second principal components account for 67% of the variability in the data set analyzed. The sites listed were sampled in this investigation (16 samples, Table 1) and in an earlier investigation (nine samples [16]).

Brine reference water, which represents the brine type of water found in Laxemar borehole 2 at 1631–1681 m depth, with a measured Cl content of 47 200 mg l⁻¹. Brines are waters with significantly higher salinity than ocean water (Cl content of 19 800 mg l⁻¹).

Baltic Sea reference water, which represents modern Baltic Sea water, with ocean water as the saline component.

Altered marine reference water, which represents seawater altered by bacterial sulfate reduction. This water type is obtained in the Äspö HRL tunnel below marine sediments (borehole SA0813B: 5.6–19.5 m depth).

Precipitation reference water, which represents dilute shallow groundwater of the type found in a shallow borehole in the Äspö area called HLX06 (871103): 45–100 m depth.

Glacial reference water, which is a precipitation water composition where the stable isotope values (δ¹⁸O=−21 SMOW and δ²H=−158 SMOW) are based on measured values of δ¹⁸O in the calcite surface deposits, interpreted as subglacial precipitates, collected from different geological formations on the west coast of Sweden. The water represents a possible melt water composition from the last glaciation >13 000 years BP.

The chemical composition of all groundwater samples within the lines between the end-members can be described by mixing of different proportions of the end-members. It is important to note that the PCA plot does not account for microbial or geochemical reactions that may have affected groundwater composition. The deepest groundwater is mainly composed of the brine end-member. The origin of the brine is unknown, but it has been estimated that it was introduced into Fennoscandian Shield rock at least 1.5 million years ago [21].

3.2 Total cell numbers

Total numbers of planktonic cells in groundwater samples ranged from 3.0×10⁴ to 2.8×10⁶ cells ml⁻¹ (Fig. 3). The average total cell number was 3.7×10⁵ cells ml⁻¹. There was no change in the mean total cell numbers with depth; rather cell numbers remained stable between 65 and 1390 m depth. The main cell morphologies were small rods and cocci, and the majority of cells were 0.5–1 μm in length.

3

Depth distribution of total numbers of planktonic microorganisms in Fennoscandian Shield groundwater. Each point is the average of two to four separate results. Error bars are 95% confidence intervals.

3.3 Comparison of groundwater-based and synthetic media

Four of the 16 samples were directly compared by MPN analysis with both synthetic and groundwater-based media. In all cases, if IRB, SRB or heterotrophic acetogens were cultured with synthetic medium, they were also cultured with groundwater-based medium. No autotrophic acetogens or autotrophic or heterotrophic methanogens were cultured from the four samples directly compared with synthetic and groundwater-based media. The ratio of the number of cells cultured with each medium type was calculated by dividing the number cultured in groundwater-based medium by the number cultured in synthetic medium (Fig. 4). The ratios were distributed over unity for SRB and heterotrophic acetogens, indicating that the two types of media worked equally well. For IRB, all ratios were less than 1, meaning that more cells were cultured in synthetic media. However, three of the four ratios were between −0.5 and 1 (Fig. 4), which means that results were similar for synthetic and groundwater-based media. Overall, the results give no indication that one method of medium preparation was significantly better than the other.

4

Ratio of number of cells cultured with groundwater-based and synthetic media for three physiological groups: IRB, SRB and heterotrophic acetogens (HA). For values greater than 10⁰, more cells grew in groundwater-based media, and for values less than 10⁰, more cells grew in synthetic media.

3.4 MPN

Microorganisms were cultured from every sample except that from the Laxemar borehole at a depth of 1390 m. The percentage of total cells cultured was calculated as the sum of the MPN results for the six physiological groups divided by the total cell numbers for each sample (Table 2). Between 0 and 1% of total cells were cultured from the majority of samples with both synthetic and groundwater-based media. The median percentage of cells cultured was 0.18% for groundwater-based media and 0.20% for synthetic media (Table 2). Overall, the percentage of total cells cultured ranged from 14.8% in Kivetty borehole 13 to 0.0084% in Olkiluoto borehole 4.

2

Distribution of percentage of total cells cultured by medium type in Hästholmen, Kivetty, Laxemar, Olkiluoto, Romuvaara and Äspö HRL samples

Medium	Number of samples	Percentage of total cells cultured	Median % cultured
0–1%	>1–5%		>5–10%	>10%
Groundwater-based	11		8	2	1	0	0.18
Synthetic	9		7	0	1	1	0.20

The presence or absence of culturable microorganisms belonging to each physiological MPN group for all 16 samples is shown in Fig. 5. IRB, SRB, and heterotrophic acetogens were cultured from over 73% of samples tested, and from all six sites. IRB were cultured from all but four of the samples tested (one sample was not tested for IRB), SRB from all but four samples and heterotrophic acetogens from all but one sample (Fig. 5). Autotrophic acetogens were cultured from four samples, from all sites except Olkiluoto and Kivetty. Heterotrophic methanogens were cultured from one Olkiluoto and two Äspö HRL samples, while autotrophic methanogens were detected in one sample from Olkiluoto. Table 3 shows MPN results for autotrophic acetogens and methanogens and heterotrophic methanogens. MPN of these groups ranged from 10⁻¹ to 10² cells ml⁻¹ (Table 3).

5

Occurrence of culturable microorganisms belonging to six different physiological groups in 16 Fennoscandian Shield groundwater samples.

3

MPN results for autotrophic and heterotrophic methanogens and autotrophic acetogens cultured from Fennoscandian Shield groundwater samples

Physiological group	OL-KR4	RO-KR11	HH-KR4	KA3566G02:2	KA3573A:2	KA3600F:2	KLX02
	863 m	543 m	686 m	450 m	450 m	450 m	1160 m
Autotrophic methanogens (cells ml−1)	7.0	<0.2	<0.2	<0.2	<0.2	<0.2	<0.2
Heterotrophic methanogens (cells ml−1)	2.3	<0.2	<0.2	24	0.92	<0.2	<0.2
Autotrophic acetogens (cells ml−1)	<0.2	3.3	0.2	<0.2	<0.2	0.78	110

Samples not included in this table (Olkiluoto boreholes 3, 9; Hästholmen boreholes 5, 6; Kivetty borehole 13; Äspö boreholes 3573:1, 3600:1; and Laxemar borehole 2, 1350 and 1390 m depths) contained <0.2 cells ml⁻¹ of all of these physiological groups.

MPN results for IRB, SRB and heterotrophic acetogens were superimposed on the PCA plot from Fig. 2 (Fig. 6). Also included in Fig. 6 are MPN results for nine groundwater samples from the four Finnish sites, which were published previously [16]. IRB were cultured from samples from 65–1160 m depth. However, IRB were not cultured from the samples that consisted of 69% or more of the brine end-member: Olkiluoto borehole 4, and Laxemar samples from 1350 and 1390 m depth (Fig. 6A). SRB were cultured from samples from 65–721 m depth (Fig. 6B). No SRB were detected in the six deepest and most saline samples: all three Laxemar samples, Olkiluoto borehole 4, and Hästholmen boreholes 1 and 2 [16]. These samples contained at least 31% brine and came from depths of 860 m or greater. Sulfate concentrations in the samples in which SRB were not found were 0.35–3.5 mM in Hästholmen and Laxemar samples, and 0.01 mM in Olkiluoto borehole 4. Heterotrophic acetogens were cultured from all but four samples (Fig. 6C): Laxemar 1390 m depth, Olkiluoto boreholes 8 and 9, and Hästholmen borehole 1 [16]. Heterotrophic acetogens were cultured in samples with up to 78% brine, and to a depth of 1350 m. IRB, SRB and heterotrophic acetogens were cultured at levels up to 10⁴ cells ml⁻¹ (Fig. 6), in contrast to a maximum of 10² cells ml⁻¹ in media for autotrophic acetogens and autotrophic and heterotrophic methanogens (Table 3).

6

Numbers of microorganisms enumerated by MPN, superimposed on the PCA plot (cf. Fig. 2). A: IRB. B: SRB. C: Heterotrophic acetogens. For sample identity, see Fig. 2.

4 Discussion

Groundwater samples originating from depths of 65–1350 m from six sites in the Fennoscandian Shield all contained viable microorganisms. Total cell numbers ranged from 10⁴ to 10⁶ cells ml⁻¹ groundwater, and no change in the mean of the total cell numbers was observed in relation to depth to 1390 m (Fig. 3). The situation in the igneous subsurface is different from that in marine sediments, where a significant decrease in total cell numbers was observed to a depth of 1000 m below the seafloor [22]. Stable cell numbers with increasing depth suggest that there must be sufficient energy and carbon sources to support or maintain the microbial population.

In this study, two approaches were used to produce culture media with the same composition as groundwater. No significant difference was observed between the two approaches (Fig. 4). A median of 0.2% of total cells was cultured with both types of media (Table 2). The small proportion of total cells cultured must be kept in mind when drawing conclusions from the results. Culturing should be supplemented with culture-independent methods to give more information about community structure. However, positive culturing results do give valuable information about physiological groups that are present.

The complex chemistry and mixing of Fennoscandian Shield groundwater can be understood using PCA (Fig. 2). Superimposing MPN results on the PCA plot (Fig. 6) showed that SRB and IRB were cultured from all but the most saline samples, i.e. those with the largest proportion of the over 1.5 million year old brine end-member [23]. Less saline groundwater contained the same physiological groups in spite of large differences in chemical composition.

Redox environments in sediments and groundwater can be classified as oxic, post-oxic, sulfidic and methanic [24]. Oxic environments contain >10⁻⁶ M oxygen, which acts as the main electron acceptor in microbial metabolism. Post-oxic, sulfidic and methanic environments contain <10⁻⁶ M oxygen, so alternative electron acceptors must be used. In post-oxic environments nitrate, Mn(IV) or Fe(III) may be the terminal electron acceptors. Sulfidic environments contain >10⁻⁶ M H₂S, and sulfate reduction is the dominant electron-accepting process. Finally, in methanic environments, sulfide concentrations are <10⁻⁶ M, and CO₂ is the terminal electron acceptor. Olkiluoto groundwater is post-oxic, sulfidic and methanic with increasing depth. In Hästholmen, post-oxic and sulfidic groundwater has been observed to 1000 m depth, but sulfate is present in the deepest samples investigated and methanic conditions have not been observed [25]. The deepest groundwater samples from Laxemar, Äspö HRL, Kivetty and Romuvaara are sulfidic as they also contain sulfate. Using these classifications, the 16 groundwater samples investigated in this study along with nine samples previously analyzed from four Finnish sites [16] can be divided into three different groups, each with different microbial populations: (1) post-oxic and sulfidic groundwater (down to 721 m depth) containing SRB, IRB and acetogens; (2) sulfidic groundwater (from more than 900 m depth) containing IRB and/or acetogens; (3) methanic groundwater containing methanogens and acetogens.

The first group of samples consists of 19 post-oxic and sulfidic groundwater samples from depths of 65–721 m in Olkiluoto, Kivetty, Romuvaara, Hästholmen, and Äspö HRL. These samples all contained culturable IRB, SRB and heterotrophic acetogens, with the exception of three negative results out of 57 (Fig. 6). The microbial community cultured from these depths probably live by oxidizing organic carbon that originates on the surface. The second group includes four sulfidic groundwater samples from over 900 m depth in Hästholmen and Laxemar. IRB and/or acetogens were cultured from these samples but no SRB or methanogens were detected (Fig. 6,Table 3). It is unclear why SRB were not cultured from these samples. Sulfate is present at concentrations of 0.35–3.5 mM. Electron donors do not appear to be limiting since viable heterotrophic acetogens and IRB as well as 10⁵ total cells ml⁻¹ were found. SRB were cultured in samples containing up to 1.6% NaCl, which is considerably lower than the 11% NaCl at which SRB isolated from other subsurface environments were active [26,27. Since IRB and acetogens were cultured from the samples in which SRB were not found, and SRB were cultured from all shallower groundwater samples as discussed above, it seems that culture conditions were appropriate. Acetogens are often able to compete with other microorganisms due to their metabolic versatility [28]. Use of iron as an electron acceptor gives more energy than sulfate [29], so the lack of SRB at a shallower depth than IRB may indicate an energetic limitation with the available organic carbon. The final type of groundwater is methanic groundwater containing methanogens and acetogens. Only one sample of this type was analyzed, from Olkiluoto borehole 4 at a depth of 863 m, which contained autotrophic and heterotrophic methanogens as well as heterotrophic acetogens, but no culturable IRB, SRB or autotrophic acetogens (Table 3,Fig. 6). The carbon stable isotope ratio in dissolved inorganic carbon (δ¹³C-DIC) in Olkiluoto borehole 4 groundwater was +16.8‰ Pee Dee Belemnite (PDB), while the ratios for the other Finnish samples ranged from −25‰ to −4‰ PDB. The positive δ¹³C value indicates that the lighter ¹²C has preferentially been removed by biological activity, for example by the autotrophic methanogens that were cultured from this sample. Culturing results and stable isotope data indicate that some of the microorganisms in methanic groundwater use inorganic carbon sources for autotrophic growth. In Columbia River basalt aquifers, DIC was also increasingly enriched in ¹³C with depth, which was attributed to the activity of autotrophic methanogens, but as in the Fennoscandian Shield, this was only observed in groundwater with low sulfate concentrations [3]. It has been suggested that some subsurface ecosystems may get their energy from hydrogen gas produced by inorganic processes in the Earth's crust [3,5,3,30. The hydrogen- and carbon dioxide-utilizing autotrophic methanogens observed in Olkiluoto borehole 4 could be a part of such a subsurface ecosystem. Further studies of sulfate-depleted groundwater at depths of over 750 m in the Fennoscandian Shield should be carried out to determine whether the microbial community observed in Olkiluoto borehole 4 is widespread.

Several authors have suggested that there are two separate subsurface microbial ecosystems in the igneous subsurface, one dependent on organic carbon being transported down from the surface and the other dependent on hydrogen or other inorganic energy sources produced in the geosphere [3,5,30,31. The results of this study indicate that a deep hydrogen-driven biosphere may be present in the Fennoscandian Shield, but it is not clear that it occurs in all sites. More samples of methanic groundwater should be studied to better understand processes at great depth that may occur independent of the sun's energy.

Acknowledgements

Posiva OY, the Swedish Nuclear Fuel and Waste Management Company and the Swedish Natural Science Research Council (08466-331) supported this work. Berit Ertman Ericsson and Emma Larsdotter Nilsson provided laboratory assistance. Posiva OY and äspö HRL personnel assisted with sample collection. Margit Snellman and Paula Ruotsalainen coordinated transfer of samples and data for the Finnish samples. Marcus Laaksoharju and Cecilia Andersson did the PCA analysis.