MICROBIOLOGY ECOLOGY. Introduction RESEARCH ARTICLE. Lotta Hallbeck & Karsten Pedersen. Abstract

Transcription

1 RESEARCH ARTICLE Culture-dependent comparison of microbial diversity in deep granitic groundwater from two sites considered for a Swedish final repository of spent nuclear fuel Lotta Hallbeck & Karsten Pedersen Microbial Analytics Sweden AB, Mölnlycke, Sweden MICROBIOLOGY ECOLOGY Correspondence: Karsten Pedersen, Microbial Analytics Sweden AB, Mölnlycke, Sweden. Tel.: ; fax: ; kap@micans.se Received 26 September 2011; revised 13 December 2011; accepted 13 December Final version published online 18 January DOI: /j x Editor: Christian Griebler Keywords Forsmark; Laxemar; most probable number; ATP; subsurface; redox potential. Introduction Abstract Fennoscandian Shield granite formations are proposed as host rocks for deep (~500 m) spent nuclear fuel (SNF) repositories in Sweden and Finland. The SNF waste will be encapsulated in copper canisters that must resist corrosion for at least years (SKB, 2010). Detailed site-selection procedures for future repositories have been carried out. The Swedish site was selected in 2011 by the Swedish Nuclear Fuel and Waste Management Co. (SKB) and is located on the Baltic Sea coast in Forsmark, approximately 170 km north of Stockholm (SKB, 2011). The Finnish site was selected by Posiva Oy in 2002 and is located on the Baltic Sea coast in Olkiluoto, 300 km north of Helsinki (Posiva Oy, 2009). A microbiological analysis programme including quantification and culturing methods was developed during the Finnish site investigations (Haveman et al., 1999; Haveman & Pedersen, Site selection for a spent nuclear fuel (SNF) repository required analysis of microbial abundance and diversity at two Swedish sites, Forsmark and Laxemar-Simpevarp. Information about sulphate-reducing bacteria (SRB) was required, as sulphide could corrode copper SNF canisters. Total number of cells (TNC) and ATP were analysed, and plate counts and most probable number (MPN) analyses were conducted using eight media based on different electron donors and acceptors for specific microorganism physiological groups. Groundwater chemical composition and E h were analysed; sampling depths were m below sea level. TNC was to cells ml 1, correlating with ATP concentrations. Culturability in TNC percentage was , averaging Culturable numbers varied greatly between sample positions and uncorrelated with depth. SRB were found in 29 samples and were below detection in three; the MPN of SRB correlated negatively with E h, as did the MPN of acetogens. Data indicated that microbial sulphate reduction was ongoing in many sampled aquifers; published stable isotope data and modelling results supported this observation. The sites did not differ significantly, but the large data range suggested that analysis of more samples would enable detailed evaluation of microbial processes and their relationship with geochemical information. 2002). This microbiological programme was adopted and further developed during the Swedish site investigation programme for two candidate sites, Laxemar-Simpevarp and Forsmark. The site investigations started in 2002, and the major field activities were completed in 2007 (Ström et al., 2008). Oxygen and sulphide are corrosive to copper, so the concentrations of these substances in groundwater must be low for the long-term safe functioning of the repositories (Tullborg et al., 2010). Microbial processes generally reduce oxygen with organic carbon, methane or hydrogen, and sulphide can be produced from sulphate by sulphate-reducing bacteria (SRB) under anoxic conditions. Information about the abundance of microorganisms that can influence oxygen and sulphide concentrations by direct or indirect activities was consequently requested. An example of indirect activity is the reduction by ironreducing bacteria (IRB) of ferric iron to ferrous iron, ª 2011 Federation of European Microbiological Societies FEMS Microbiol Ecol 81 (2012) 66 77

2 Culture-dependent evaluation of microbial diversity 67 which precipitates with sulphide and reduces oxygen. Intruding oxygen could also be reduced by facultative anaerobes, such as most nitrate-reducing bacteria (NRB), that survive in anoxic groundwater in the absence of oxygen. Quantification of functional genes using qpcr was not as straightforward at the start of the microbiology analysis programme in 2003 as it is today. In addition, the largest available volume of water per sample was 190 ml, containing at most 10 5 cells ml 1, so the amount of extractable DNA was very small. The potentially low yields of many extraction methods (Lloyd et al., 2010) might have limited the amount of extracted DNA to amounts too small for qpcr quantification. Instead, the samples were treated using classic quantification and anoxic culturing methods that can detect as few as 0.2 culturable cells ml 1 (Greenberg et al., 1992). The unculturability of microorganisms from environmental communities is commonly used to justify DNA analysis; this problem can be avoided by using various solid and liquid media that enrich for specific microorganism groups using various electron donors and acceptors and varying the growth conditions (Vartoukian et al., 2010). The aim of this paper was on the use of a cultivationbased approach for the analysis of abundance and diversity on microorganisms in deep groundwater. All microbiology data obtained from the completed Swedish site investigations were used for this purpose. The total microbial population was quantified as total number of cells (TNC) and adenosine-tri-phosphate (ATP) concentrations were analysed. Aerobic heterotrophic microorganisms were quantified using plate counts. Anoxic most probable number (MPN) cultures with liquid media included growth media for NRB, IRB, manganesereducing bacteria (MRB), SRB, acetogenic bacteria and methanogens. The microbial data were compared with geochemical and physical data, and correlations between data were identified. Data from the two sites were also compared and evaluated separately. Materials and methods Site descriptions Two sites have been considered for a final Swedish repository for SNF, the Forsmark and Laxemar-Simpevarp locations, both situated on the Swedish Baltic Sea coast. The Forsmark site is located in the northern part of the province of Uppland, Sweden, in the municipality of Östhammar, approximately 170 km north of Stockholm and in the immediate vicinity of the Forsmark nuclear power plant and the subsea-floor (50 m) repository for low- and intermediate-level radioactive waste. The site is flat and low lying, covering an area of approximately 10 km 2. Because of glacial isostasy, almost the whole area was covered by seawater until some 2500 years ago, after which the process of land uplift led to the gradual formation of islands in what is now a coastal archipelago. Today s landscape in Forsmark contains many small shallow lakes and bays with a thick layer of boulder clay. The Laxemar-Simpevarp site consists of two subsites (Laxemar and Simpevarp) and is located in the province of Småland, in the municipality of Oskarshamn, approximately 350 km south of Stockholm. The Simpevarp subsite area is located near the Baltic Sea coast, immediately adjacent to the Oskarshamn nuclear power plant and the Swedish central interim storage facility for SNF. The Laxemar subsite, which is flat and covers an area of approximately 10 km 2, is located inland from the Simpevarp peninsula on which the power plant is located. The overburden in this area is thin and consists mostly of moraine; large parts of the area lack overburden and have exposed basement rock and boulders. Details of the hydrogeochemical characterizations of these two sites can be found in reports published at and in a special issue of Applied Geochemistry (Gascoyne & Laaksoharju, 2008). Groundwater sampling Samples for microbiological analysis were taken from packed-off sections of six core-drilled boreholes in Laxemar-Simpevarp and of 10 core-drilled boreholes in Forsmark to depths of and m below sea level, respectively (Table 1). The flushing water used for drilling was groundwater taken from shallow boreholes and stored in nitrogen-filled plastic tanks. The flushing water was sterilized with UV light, and the fluorescent dye uranin was added to serve as a tracer before the flushing water entered the drilling system. Culturable heterotrophic aerobic bacteria (CHAB) were intermittently analysed to determine the sterilization efficiency (e.g. Hallbeck et al., 2004; Pedersen, 2006a, b). Before installing the packers, the borehole walls were investigated using a borehole image processing system, and sections with open fractures were selected for sampling. The differential flow of groundwater in each sampled section was subsequently monitored to confirm that the fracture was water conducting. Groundwater samples for microbiology were taken using a 190-mL sample volume pressure vessel sampler (PVB) modified by SKB from the PAVE sampling system developed in Finland by Posiva Oy (Haveman et al., 1999). Sampling using this equipment overcame the problems associated with decreased pressure and reduced the obvious risks of oxygen and microbial contamination when groundwater was pumped from packedoff borehole sections through the 1000 m of tubing used during the site investigations. Groundwater was pumped FEMS Microbiol Ecol 81 (2012) ª 2011 Federation of European Microbiological Societies

3 68 L. Hallbeck & K. Pedersen for 2 3 weeks, bypassing the PVB vessels, until the flushing water content was below 0.1%. The PVB sample vessels were then opened, filled, closed and hoisted to the surface for analysis. The sampling method, quality controls for reproducibility and decontamination procedures for the PVB sampling equipment were described in detail elsewhere (Hallbeck & Pedersen, 2008). The PVB vessels containing pressurized groundwater samples were sent immediately to the microbiology laboratory for analysis that commenced the same day the sampler reached the ground surface. The microbiology sampling and analysis programme continued for 5 years from January 2003 to October Chemical analyses of groundwater Physical and chemical parameters were determined according to the analytical programme designed for site investigations (SKB, 2001). Redox potential (E h ) was measured continuously using flow-through cells installed in line with three PVB vessels in the packed-off sections and in the mobile chemistry laboratory at ground level. The methodology and equipment for measuring E h in deep boreholes were described previously (Grenthe et al., 1992). Briefly, E h was continuously logged simultaneously by Au, Pt and glassy C electrodes, both in the borehole and at the surface, against Ag/AgCl double junction gel-filled reference elec- Table 1. Borehole information including sampling date, sampled borehole section and sampling depth. Several boreholes were inclined, resulting in a shallower depth than indicated by the sampled section. Groundwater chemistry data were analysed on samples pumped to the surface, and E h was measured using electrodes installed in the sampled sections Borehole designation Depth (mbsl) Sampled section (m) Sampling date (year-month-day) Simpevarp KSH01A KSH01A KSH01A Laxemar KLX MD 266 KLX KLX KLX17A KLX KLX MD KLX13A KLX15A MD KLX17A* KLX KLX < MD Forsmark KFM01A MD 195 KFM01A KFM10A < KFM06A KFM10A < KFM01D KFM11A KFM03A KFM01D KFM02A KFM08D < KFM08A MD KFM03A KFM06A MD KFM08D < 0.4 MD KFM07A MD KFM03A KFM03A MD SO 2 4 (mm) Cl (mm) Fe 2+ (lm) Mn 2+ (lm) CH 4 (lm) E h (mv) mbsl, m below sea level; MD, missing data because of technical problems. *The groundwater chemistry was not representative of the sampled depth. ª 2011 Federation of European Microbiological Societies FEMS Microbiol Ecol 81 (2012) 66 77

4 Culture-dependent evaluation of microbial diversity 69 trodes. The data measurement and treatment are described in detail elsewhere (Auqué et al., 2008). Major and minor dissolved solids and trace elements were analysed in samples pumped to the mobile chemistry laboratory at the ground (SKB, 2001). Cl was analysed using titration (Swedish Institute of Standards SIS 28120), Mn 2+ using inductively coupled plasma atomic emission spectroscopy, Fe 2+ using spectrophotometry with ferrozine, sulphate using ion chromatography and methane using gas chromatography with flame ionization detection (Table 1). Total number of cells The TNC was determined using the acridine orange direct count method as devised by Hobbie et al. (1977) and modified by Pedersen & Ekendahl (1990). ATP analysis The ATP Biomass Kit HS (no ; BioThema, Handen, Stockholm, Sweden) was used to determine total ATP in living cells. The ATP biomass method (Lundin, 2000) used here has been described, tested in detail and evaluated for use with Fennoscandian Shield groundwater, including Laxemar and Forsmark groundwater (Eydal & Pedersen, 2007). ATP analysis was introduced into the analytical programme in Culturable heterotrophic aerobic bacteria Petri dishes containing agar with nutrients were prepared as described elsewhere (Pedersen & Ekendahl, 1990) for determining the CHAB numbers in the flushing water used for drilling and in groundwater samples. Briefly, 10-times dilution series of water samples were made in sterile analytical grade water (AGW) containing 1.0 g L 1 of NaCl and 0.1 g L 1 K 2 HPO 4 ; 0.1-mL portions of each dilution were spread with a sterile glass rod on the plates in triplicate. The plates were incubated for 7 9 days at 20 C in the dark, after which the number of colonyforming units was counted; plates with colonies were counted. CHAB analysis was introduced into the analytical programme at the end of Preparation of media, inoculations and analysis for MPN of culturable anaerobic microorganisms The procedures described by Widdel & Bak (1992) for preparing anoxic media for determining the MPN of microorganisms were modified and applied as described elsewhere (Hallbeck & Pedersen, 2008). Five tubes were used for each 10-times dilution, resulting in an approximate 95% confidence interval lower limit of 1/3 of the obtained value and an upper limit of three times the value (Greenberg et al., 1992). Media were prepared for NRB, IRB, MRB, SRB, autotrophic acetogens (AA), heterotrophic acetogens (HA), autotrophic methanogens (AM) and heterotrophic methanogens (HM). The cultivation time was about 8 weeks to ensure that slow growing microorganisms were included in the results. Analyses of MRB and NRB were introduced into the analytical programme in 2004 and 2005, respectively. The specific media compositions were based on previously measured chemical data from Laxemar-Simpevarp and Forsmark. The sodium chloride concentration was adjusted to obtain a medium salinity corresponding to the salinity of the sampled borehole water. This allowed the formulation of artificial media that most closely mimicked in situ groundwater chemistry for optimal microbial culturing, as was previously successful (Haveman et al., 1999). Statistical analyses Data graphics design and statistical analyses were performed in STATISTICA 10 (Statsoft Inc., Tulsa, OK). Results Chemistry and E h The salinity as reflected by the concentration of Cl generally increased with depth at both sites, reaching approximately 300 mm at a depth of 1 km (Table 1). The sulphate concentration also increased with depth in Laxemar-Simpevarp, but displayed a decreasing tendency with depth in Forsmark. However, these general trends with depth were patchy, reflecting the heterogeneity in chemistry commonly observed in groundwater from fractures in hard rock. The other three species shown in Table 1, Fe2+, Mn 2+ and methane, and all varied significantly from sample to sample without any relationship with depth. Analysis of nitrate was usually in the range of lm in Laxemar (Laaksoharju et al., 2009b) and in the range of lm in Forsmark (Laaksoharju et al., 2009a). E h values were negative and ranged from 303 to 143 mv with no clear relationship with depth. Total numbers of microorganisms and ATP The TNC varied by approximately two orders of magnitude in the analysed groundwater samples, ranging from the lowest number, cells ml 1 in KLX m, to the highest number, cells ml 1 in KFM10A-328 m (Table 2). There was no correlation with depth (not shown) at any of the sites, and the variation FEMS Microbiol Ecol 81 (2012) ª 2011 Federation of European Microbiological Societies

5 70 L. Hallbeck & K. Pedersen in TNC from depth to depth was often large. The amount of ATP also varied by approximately two orders of magnitude, ranging from the lowest amount, amol ATP ml 1, found in KFM08A-546 m to the highest amount, amol ATP ml 1, found in KLX08A- 150 m (Table 2). The ATP values generally correlated with the TNC values [ 10 Log(ATP) = Log (TNC); r = ; P = ], as was previously demonstrated for deep groundwater, but the average amount of ATP per cell was larger in the present data set (1.54 amol ATP cell 1, n = 25) than that previously found for ATP-TNC analyses of Fennoscandian Shield groundwater (0.43 amol ATP cell 1, n = 166) (Eydal & Pedersen, 2007). There was no correlation between the TNC or ATP analysis results and sampling date, excluding biases owing to variations in executing analysts over the 5-year reparation, sampling and analysis period. Numbers of culturable microorganisms The results of the CHAB analyses ranged from 100 cells ml 1 in KFM08D-664 m to cells ml 1 in KFM01D-445 m without any relationship with depth or sampling date (Table 2). NRB varied by approximately three orders of magnitude in the analysed groundwater samples, ranging from the lowest number, 50 cells ml 1, found in KLX m, to the highest number, cells ml 1, found in KFM01D-445 m. The CHAB values correlated with the NRB values [ 10 Log(NRB) = Log(CHAB); r = 0.775; P = ]. IRB and MRB numbers ranged from below the detection limit (< 0.2 cells ml 1 ) to at most cells ml 1 and cells ml 1, respectively, in KFM01A-112 m. SRB were detected in all but three of the 32 samples analysed. The largest MPN of SRB, cells ml 1, was obtained from the KFM01D-445 m sample. AA and HA displayed the largest value range of all MPN analyses, from below the detection limit (< 0.2 cells ml 1 ) to above the upper detection limit of cells ml 1 with six 10-times dilution steps in the MPN analysis. Most of the AM and HM determinations were below or just above the detection limit (< 0.2 cells ml 1 ). Two samples, KLX m and KFM03A-931 m, had AM and HM numbers significantly above the detection limit. The MPN results were not correlated with sampling date, excluding biases owing to variations in executing analysts over the 5-year sampling period. The percentage of the TNC that could be cultured using MPN analysis was under the conservative assumption that IRB and MRB, AA and HA, and AM and HM cultures counted similar populations (Table 2). The addition of NRB analysis in 2005 increased these percentages significantly from an average of 5.12 for all samples (n = 30) to 8.78% for samples including NRB analyses (n = 16). The CHAB determinations cultured % of the TNC values. Distribution of analysis results per investigated area The distribution of the logarithm of all analysis results per investigated site is shown in Fig. 1a. The range of values was larger for all analyses from Forsmark (n = 18), except for AM and HM, than from Laxemar-Simpevarp (n = 14), as indicated by the length of the standard deviation bars. The mean of the logarithm of the values did not differ significantly between sites for any of the analyses, as judged from the overlapping standard deviations. The mean and range of the percentage of TNC culturable using MPN analysis for the Forsmark samples were approximately twice the Laxemar-Simpevarp values (Fig. 1b). A similar result was obtained for the percentage of TNC culturable for CHAB. Correlation analysis A correlation matrix of all analysed chemistry and microbiology data was estimated and evaluated (not shown). The significant correlations between TNC and ATP, and between CHAB and NRB, have been presented above. In addition, there were good correlations between IRB and MRB (Fig. 2a), AA and HA (Fig. 2b), SRB and AA (Fig. 2c), and AM and HM (not shown). The MPN of AA (Fig. 2d) and SRB (Fig. 2e) both correlated with the E h values best fitting with the AA vs. E h data. A good correlation was also found between the Mn 2+ data and E h. Discussion Contamination control The potential for lasting contamination of groundwater with flushing water during drilling operations was previously investigated in the underground Äspö Hard Rock Laboratory (HRL) when drilling three boreholes at a depth of approximately 400 m (Pedersen et al., 1997). The results indicated, using 16S rdna gene sequencing and culturing methods, that although large numbers of contaminating bacteria were introduced into the boreholes during drilling, they did not become established in the borehole groundwater at detectable levels. The site investigations presented here incorporated UV treatment of the flushing water and meticulous drilling hygiene involving steam cleaning of the drilling system. This generally reduced the number of contaminating cells in the flushing water more than 100-fold, from above 10 5 CHAB ml 1 in the Äspö HRL investigation to approximately 10 3 ª 2011 Federation of European Microbiological Societies FEMS Microbiol Ecol 81 (2012) 66 77

6 Culture-dependent evaluation of microbial diversity 71 Table 2. Microbiology data on microscopic counts, ATP determinations and cultures of physiological groups of microorganisms having different electron acceptors and donors. The percentages of the total number of cells (TNC) cultured with MPN [(NRB + (IRB + MRB)/2 + SRB + (AA + HA)/2 + (AM + HM)/2) 9 100/TNC] and with CHAB are also shown. The approximate 95% confidence interval limits for the MPN values were lower = 1/3 of the value and upper = three times the value Borehole designation Depth (mbsl) Sampling date (year-month-day) TNC cells SD ml 1 (n = 3) ATP amol SD ml 1 (n = 3) CHAB CFU ml 1 SD (n = 3) Simpevarp KSH01A ND ND KSH01A ND ND KSH01A ND ND Laxemar KLX KLX ND KLX KLX17A KLX MD ND KLX KLX13A KLX15A KLX17A* KLX ND KLX ND Forsmark KFM01A ND ND KFM01A ND ND KFM10A KFM06A ND KFM10A KFM01D KFM11A KFM03A ND KFM01D KFM02A ND ND KFM08D KFM08A KFM03A MD ND KFM06A ND KFM08D KFM07A ND KFM03A ND KFM03A ND ND Borehole designation Cells ml 1 % of TNC NRB IRB MRB SRB AA HA AM HM MPN CHAB Simpevarp KSH01A ND 2.1 < < 0.2 < 0.2 < 0.2 < KSH01A ND 2.1 < MD 900 < KSH01A ND 3.3 < < Laxemar KLX KLX03 ND < KLX < 0.2 < KLX17A < KLX03 ND MD KLX08 ND KLX13A < KLX15A < 0.2 < KLX17A* < FEMS Microbiol Ecol 81 (2012) ª 2011 Federation of European Microbiological Societies

7 72 L. Hallbeck & K. Pedersen Table 2. Continued Borehole designation Cells ml 1 % of TNC NRB IRB MRB SRB AA HA AM HM MPN CHAB KLX KLX03 ND < < < Forsmark KFM01A ND ND ND MD KFM01A ND 4.0 < < 0.2 < 0.2 < KFM10A < KFM06A ND KFM10A KFM01D < 0.2 < KFM11A < KFM03A KFM01D < 0.2 < KFM02A ND 11.0 < < 0.2 < 0.2 < KFM08D < 0.2 < KFM08A MD < KFM03A ND MD KFM06A ND < 0.2 < KFM08D < 0.2 < KFM07A 220 < 0.2 < 0.2 < < < KFM03A ND MD < KFM03A ND < 0.2 < < mbsl, m below sea level; SD, standard deviation; n, number of observations; CFU, colony-forming units; ND, no data due to analysis not performed; MD, missing data because of technical problems. *The groundwater chemistry was not representative of the sampled depth. CHAB ml 1 or fewer in the site investigations. Before the site investigation samples were collected, groundwater was pumped until the flushing water content was below 0.1%, minimizing the risk of contamination with flushing water, analogous to what was observed after extended borehole draining in the Äspö HRL contamination study. The successful cleaning and decontamination of the PVB sampling system has been described elsewhere (Hallbeck & Pedersen, 2008). Evaluation of the sampling methodology The reproducibility of the sampling and analysis methods was tested on groundwater samples from the KFM m section with two PVB connected in series (Hallbeck & Pedersen, 2008). There was no significant difference between any of the analyses, as the observed differences between the sample pairs were within the 95% confidence intervals in all analyses. A similar test was performed on two occasions using groundwater from two borehole sections at a depth of 450 m in the Äspö HRL having very different population sizes; again, very small differences were observed between the pairs of analysis results (Hallbeck & Pedersen, 2008). The reproducibility of the PVB sampling system and the culture and quantification methods between parallel samples and over time was excellent. Taken together, the documented reproducibility and the contamination control procedures used during drilling and sampling, which were demonstrated to be successful, should ensure that the results presented here reflect the microbial populations of the sampled aquifers. The largest sample volume obtained using the PVB sampler was 190 ml of groundwater, which is a limited amount. The alternative for obtaining a larger sample would have been to pump water to the surface via the 1000-m-long tubing of the sampling system. However, this would have introduced contaminating microorganisms into the samples from the tubing, pump and other installations in the flow line that were impossible to decontaminate properly. In addition, the pumping rate was generally low, which would have meant transport times of several hours before the groundwater reached a sampling point at the surface, irrespective of sampling depth, because the tube length was fixed. The composition and diversity of pumped samples would likely have been severely altered compared with the aquifer conditions, leaving the PVB sampler as the only alternative for obtaining high-quality samples. Numbers and culturability of the studied groundwater populations The average TNC was below 10 5 cells ml 1 at both sites, meaning on average approximately cells per ª 2011 Federation of European Microbiological Societies FEMS Microbiol Ecol 81 (2012) 66 77

8 Culture-dependent evaluation of microbial diversity 73 (a) (b) Fig. 1. (a) Distribution of the 10 logarithm of the results of each analysis performed on groundwater samples from Laxemar-Simpevarp ( ) and Forsmark ( ) and the mean of the log numbers for each analysis for Laxemar-Simpevarp ( ) and Forsmark ( ). TNC, total number of cells; ATP, adenosine triphosphate; CHAB, culturable aerobic heterotrophic bacteria; NRB, nitrate-reducing bacteria; IRB, iron-reducing bacteria; MRB, manganese-reducing bacteria; SRB, sulphate-reducing bacteria; AA, autotrophic acetogens; HA, heterotrophic acetogens; AM, autotrophic methanogens; and HM, heterotrophic methanogens. Data below the detection limits for the MPN analyses (< 0.2 cells ml 1 ) were set to 10 log(0.1) = 1. Whiskers indicate standard deviations of the means. (b) Distribution of the percentage of the TNC that could be cultured using MPN and aerobic (CHAB) plate counts for each analysis performed on groundwater samples from Laxemar-Simpevarp ( ) and Forsmark ( ) and the mean of the log numbers for each analysis for Laxemar- Simpevarp ( ) and Forsmark ( ). Culturable MPN of the TNC was calculated as [(NRB + (IRB + MRB)/2 + SRB + (AA + HA)/2 + (AM + HM)/2) 9 100/TNC]. Whiskers indicate standard deviations of the means. sample. In choosing between DNA- and culture-based methods for diversity analysis, the culturing approach was selected for reasons explained in the Introduction. It is often claimed that more than 99% of bacteria in environmental samples cannot be cultured, as a justification for using culture-independent, nucleic acid-based methods (Sharma et al., 2005). However, this problem can arguably be overcome using various culture procedures developed specifically for the sampled environment (Vartoukian et al., 2010). This approach was applied in developing the culturing methods used in the work presented here during the Finnish site investigations for an SNF repository, preceding the Swedish investigations (Haveman et al., 1999; Haveman & Pedersen, 2002). Since then, the array of methods has been supplemented by ATP, CHAB, NRB and MRB analyses in which NRB analysis increased the overall average proportion of culturable microorganisms in Laxemar-Simpevarp and Forsmark groundwater samples from 5.12% to 8.78%. The MPN culturability range after 8-week incubation time was large, % (Table 2). High culturability generally corresponded to relatively high numbers for all MPN analyses and CHAB, suggesting that the viability of the sampled physiological groups (grouped based on utilized electron donors and acceptors) correlated crosswise. The reasons for the low culturability of some samples are not obvious, as culturability did not correlate with any of the chemistry parameters (Table 1). This leaves two speculative possibilities: the viability of the low-culturability samples was reduced during sampling and sample transport to the laboratory, or the microorganisms were not culturable with the methods used. Low culturability might indicate that the microbial activity in the sampled aquifer was very low and that the cells were accordingly in some kind of dormancy from which the culturing methods could not recover them. Likewise, high culturability might suggest that the sampled populations were active in the aquifers. The large data ranges observed here (Fig. 1a) have been observed previously in the Äspö HRL (Hallbeck & Pedersen, 2008). Obviously, the number of samples examined here should be increased significantly before good distribution statistics can be obtained, and the correlations with environmental parameters presented in Table 1 can be justified. A general correlation with depth was observed in Olkiluoto, Finland (Pedersen et al., 2008). That site, however, displayed a very clear layering of different groundwaters, which was absent in the Laxemar-Simpevarp and Forsmark sites, and in addition, shallow groundwater samples were included in the Olkiluoto investigation. The TNC and ATP values correlated well, as has been found previously for Fennoscandian Shield groundwater (Eydal & Pedersen, 2007). This suggests that the microbial populations were metabolizing at some level and had access to energy. The average numbers obtained using the CHAB and NRB media for samples from the two sites were the highest of all culture media results (Fig. 1a). Most NRB are facultative anaerobes, so CHAB and NRB were expected to correlate well. A future SNF repository will be sensitive to oxygen intrusion because the copper storage canisters can be corroded by oxygen. The presence of facultative anaerobes suggests that there is a biological barrier against oxygen, assuming that organic carbon is available as a reductant for oxygen. Such biological oxygen-reducing activity was previously FEMS Microbiol Ecol 81 (2012) ª 2011 Federation of European Microbiological Societies

9 74 L. Hallbeck & K. Pedersen (a) (b) (c) (e) (d) (f) Fig. 2. The correlation between the logarithm of the numbers of (a) iron-reducing bacteria (IRB) and manganese-reducing bacteria (MRB), (b) autotrophic acetogens (AA) and heterotrophic acetogens (HA), (c) sulphate-reducing bacteria (SRB) and AA, (d) E h and AA, (e) E h and SRB, and (f) E h and Mn 2+ for groundwater samples from Laxemar-Simpevarp ( ) and Forsmark ( ). Linear regression lines are shown for each correlation plot together with the equation, correlation coefficient (r) and probability (P) for the correlation. demonstrated in shallow groundwater from the Finnish repository site (Pedersen et al., 2008). The second most abundant group of culturable microorganisms was the acetogens (Fig. 1a). This is a metabolically very diverse group of facultative heterotrophic bacteria that can generally utilize both autotrophic and heterotrophic metabolic ª 2011 Federation of European Microbiological Societies FEMS Microbiol Ecol 81 (2012) 66 77

10 Culture-dependent evaluation of microbial diversity 75 pathways, depending on the environmental circumstances (Drake et al., 2002). The excellent correlation between AA and HA (Fig. 2b) was therefore expected, and their presence in deep Fennoscandian Shield groundwater has been well documented (Kotelnikova & Pedersen, 1998; Pedersen et al., 2008). The acetate produced by acetogens either from degradable organic carbon or from hydrogen and carbon dioxide is an excellent carbon source for IRB, MRB and many SRB. The good correlations between AA and SRB (Fig. 2c) and between HA and SRB (not shown) could indicate that acetogenic acetate production in the sampled aquifers was important for SRB growth and activity. Influence of microbial processes on groundwater geochemistry and E h The sulphate reducers were slightly less abundant than the acetogens (Fig. 1a). The first are the most important physiological group of microorganisms with respect to SNF repositories because of their production of sulphide, which is very corrosive to copper. Their general presence in the investigated groundwater, with only three observations below the detection limit (Table 2), suggests that the safety analyses for the Forsmark repository will require more data on their numbers (i.e. a larger number of samples) as well as reliable data on in situ sulphatereduction rates. The average numbers of IRB and MRB were in the same range as SRB numbers and, as expected, correlation between IRB and MRB was good (Fig. 2a). There were no significant correlations between IRB and Fe 2+ or MRB and Mn 2+ (Tables 1 and 2), though such correlations should not be expected because Fe 2+ and Mn 2+ concentrations are influenced by their participation in inorganic precipitation and dissolution processes that depend on ph and the presence of other inorganic ions such as sulphide and carbonate. However, it is plausible that some of the observed ferrous Fe 2+ and Mn 2+ may have resulted from IRB and MRB respiration, respectively. Methanogens were very rare, many observations being below the detection limit. This was in line with the observed concentrations of methane, which were usually only a few lm. Methanogens were obviously not very active in the investigated aquifers. The reasons for this are unclear: first, it is not a given that methanogenesis should occur in deep groundwater, and second, SRB and AA may compete successfully with AM for the small amount of hydrogen present (most hydrogen analyses indicated levels below the detection limit of 1 lm), and IRB, MRB and SRB may have competed successfully with HM for acetate produced by acetogens. When each microbiology variable was tested separately for correlations between the geochemistry data (Table 1) and culturability, two relationships appeared. The MPN of AA and SRB displayed significant correlations with the E h measured in the borehole sections (Fig. 2d,e). Sulphate concentrations displayed opposite trends with depth at Laxemar-Simpevarp and at Forsmark (Table 1): in Laxemar-Simpevarp, the sulphate concentration increased with depth, while in Forsmark, it decreased. Because of the risk of canister corrosion from sulphide, the issue of whether microbial sulphate reduction is ongoing and, if so, at what rates, is crucial from a repository safety perspective. The dependent variable in the AA and SRB correlations with E h remains to be determined, but most results suggest that E h depends on the activity of AA and SRB. Sulphate reduction with acetate as the carbon source and electron donor will produce sulphide that exerts a strong influence on E h measurements. Values between 300 and 200 mv are commonly recorded in sulphate-reducing environments (Stumm & Morgan, 1996), and most E h values found in this investigation fell within this range. Consequently, it is reasonable to assume that the number of SRB observed here reflected ongoing sulphate reduction in the sampled aquifers, which would have influenced the measured E h. Sulphide was found in low concentrations ( lm, data not shown, see Tullborg et al., 2010 for data) in groundwater samples analysed in the chemistry laboratory at the ground surface after pumping via a 3-m-long borehole installation of stainless steel and 1000 m of tubing. It is inescapable that sulphide likely reacted with the metal in the borehole installation and precipitated as iron sulphide in the tubing, making the sulphide values very uncertain and too low relative to the levels present in the aquifers. The correct analysis of sulphide in the sampled aquifers is still a technical puzzle that awaits a solution. Meanwhile, results obtained using other methods can be utilized to interpret the relationship found between E h and the SRB and AA values. Thermodynamic calculations of E h for different redox couples in the groundwater of Laxemar and Forsmark have been performed (Auqué et al., 2008). These calculations indicated that the sulphate sulphide redox pair dominated the groundwater in the studied sites. It was also concluded from calculations of ionic activity products that there was equilibrium with respect to amorphous iron monosulphide; this suggests the formation of solid metal sulphides on fracture surfaces, which would remove sulphide from the groundwater. The positive correlation between E h and Mn 2+ (Fig. 2f) could indicate increasing sulphide production with decreasing E h, with the formation of MnS that will remove sulphide and Mn2+ from solution, analogous to what was found by the ionic activity calculations that suggested the formation of FeS. Thus, a microbial sulphate-reduction process may have been ongoing without an observed increase in FEMS Microbiol Ecol 81 (2012) ª 2011 Federation of European Microbiological Societies

11 76 L. Hallbeck & K. Pedersen the sulphide concentration. Evidence of sulphate reduction could also be derived from stable isotope data (Smellie et al., 2008). The d 34 S signature of the sulphur atom in sulphate in Forsmark groundwater was 30 40& relative to the Canyon Diablo troilite (CDT) standard. This was heavier than the d 34 S signature of the marine source term signature of sulphate, which was 21& CDT, indicating that microbial sulphate reduction had occurred in Forsmark groundwater in the past. In Laxemar, sulphate in deep groundwater had a light d 34 S signature, as low as 9& CDT, which was far below the marine signature of 21& CDT (Laaksoharju et al., 2009b). However, the very high sulphate concentration at depth in Laxemar may have masked sulphate reduction owing to an isotope dilution effect. The sulphate in shallower Laxemar groundwater, at depths down to 350 m, had a heavier signature than 25& CDT, indicating that sulphate reduction had occurred in the past, at least to this depth. Taken together, with most MPN of SRB above detection, E h values in the sulphide regime of 300 up to 200 mv, 34 S data indicating past microbial sulphate reduction, and modelling work supporting these observations, it seems reasonable to conclude that microbial sulphate-reduction processes have existed and still persist in the deep groundwater of the investigated sites. Comparison of the two studied sites In the selection process for the Swedish SNF repository, microbial activity as judged from TNC and ATP and culturable diversity did not differ between the sites. Although the average number of SRB in the Laxemar-Simpevarp site was somewhat higher than in the Forsmark site selected for the Swedish SNF repository, the range of SRB was narrower (Fig. 1a). A small advantage for Forsmark can be found in the lower sulphate concentrations at repository depth than at Laxemar-Simpevarp (Table 1), rendering less sulphide in Forsmark if all sulphate is conservatively assumed to be reduced in a safety analysis. Otherwise, the sites did not differ significantly with respect to the overall numbers retrieved with the methods used in this study, judging from the averages and the standard deviations shown in Fig. 1a. In the detailed site investigations, which will commence when SKB is licensed by authorities to construct the repository, more groundwater investigations will be performed. In addition, the presence and activity of microorganisms on fracture surfaces was long ago identified as meriting investigation (Pedersen & Ekendahl, 1992a, b). Methods for sampling and analysing fracture surface microbiology in drill cores are currently being developed and tested on samples from the Äspö HRL and Forsmark, and the first results from Äspö HRL were recently published (Jägevall et al., 2011). Acknowledgements The authors are grateful to Annika Kalmus at the Department of Cell and Molecular Biology, Gothenburg University, and to Johanna Arlinger, Anna Hallbeck, Jessica Johansson and Lisa Rabe at Microbial Analytics Sweden for their excellent laboratory work. This work was funded by the Swedish Nuclear Fuel and Waste Management Co. and the Swedish Research Council. References Auqué LF, Gimeno MJ, Gómez JB & Nilsson A-C (2008) Potentiometrically measured Eh in groundwater from the Scandinavian Shield. Appl Geochem 23: Drake HL, Küsel K & Matthies C (2002) Ecological consequences of the phylogenetic and physiological diversities of acetogens. Antonie Van Leeuwenhoek 81: Eydal HSC & Pedersen K (2007) Use of an ATP assay to determine viable microbial biomass in Fennoscandian Shield groundwater from depths of m. J Microbiol Meth 70: Gascoyne M & Laaksoharju M (2008) High-level radioactive waste disposal in Sweden: hydrogeochemical characterisation and modelling of two potential sites. Appl Geochem 23: Greenberg AE, Clesceri LS & Eaton AD (1992) Estimation of bacterial density. Standard Methods for the Examination of Water and Wastewater, 18th edn, pp American Public Health Association, Washington, DC. Grenthe I, Stumm W, Laaksoharju M, Nilsson A-C & Wikberg P (1992) Redox potentials and redox reactions in deep ground water systems. Chem Geol 98: Hallbeck L & Pedersen K (2008) Characterization of microbial processes in deep aquifers of the Fennoscandian Shield. Appl Geochem 23: Hallbeck L, Pedersen K & Kalmus A (2004) Forsmark Site Investigation: Control of Microorganism Content in Flushing Water Used for Drilling of KFM05A. Report no. P (available from Swedish Nuclear Fuel and Waste Management Co., Stockholm. Haveman SA & Pedersen K (2002) Distribution of culturable anaerobic microorganisms in Fennoscandian Shield groundwater. FEMS Microbiol Ecol 39: Haveman SH, Pedersen K & Routsalainen P (1999) Distribution and metabolic diversity of microorganisms in deep igneous rock aquifers of Finland. Geomicrobiol J 16: Hobbie JE, Daley RJ & Jasper S (1977) Use of nucleopore filters for counting bacteria by fluorescence microscopy. Appl Environ Microbiol 33: Jägevall S, Rabe L & Pedersen K (2011) Abundance and diversity of biofilms in natural and artificial aquifers of the Äspö Hard Rock Laboratory, Sweden. Microbial Ecol 61: ª 2011 Federation of European Microbiological Societies FEMS Microbiol Ecol 81 (2012) 66 77

12 Culture-dependent evaluation of microbial diversity 77 Kotelnikova S & Pedersen K (1998) Distribution and activity of methanogens and homoacetogens in deep granitic aquifers at Äspö Hard Rock Laboratory, Sweden. FEMS Microbiol Ecol 26: Laaksoharju M, Smellie J, Tullborg E-L, Gimeno M, Hallbeck L, Molinero J & Waber N (2009a) Bedrock Hydrogeochemistry Forsmark: Site Descriptive Modelling SDM-Site Forsmark. Report no. SKB R-08-47, Swedish Nuclear Fuel and Waste Management Co., Stockholm. Laaksoharju M, Smellie J, Tullborg E-L et al. (2009b) Bedrock Hydrogeochemistry Laxemar: Site Descriptive Modelling SDM- Site Laxemar. Report no. SKB R-08-93, Swedish Nuclear Fuel and Waste Management Co., Stockholm. Lloyd KG, MacGregor BJ & Teske A (2010) Quantitative PCR methods for RNA and DNA in marine sediments: maximizing yield while overcoming inhibition. FEMS Microbiol Ecol 72: Lundin A (2000) Use of firefly luciferase in ATP-related assays of biomass, enzymes, and metabolites. Methods Enzymol 305: Pedersen K (2006a) Oskarshamn Site Investigation: Double Control of Microorganism Content in Flushing Water Used for Drilling of KLX17A. Report no. P (available from Swedish Nuclear Fuel and Waste management Co., Stockholm. Pedersen K (2006b) Oskarshamn Site Investigation: Triple Control of Microorganism Content in Flushing Water Used for Drilling of KLX13A. Report no. P (available from Swedish Nuclear Fuel and Waste management Co., Stockholm. Pedersen K & Ekendahl S (1990) Distribution and activity of bacteria in deep granitic groundwaters of southeastern Sweden. Microbial Ecol 20: Pedersen K & Ekendahl S (1992a) Incorporation of CO 2 and introduced organic compounds by bacterial populations in groundwater from the deep crystalline bedrock of the Stripa mine. J Gen Microbiol 138: Pedersen K & Ekendahl S (1992b) Assimilation of CO 2 and introduced organic compounds by bacterial communities in ground water from Southeastern Sweden deep crystalline bedrock. Microbial Ecol 23: Pedersen K, Hallbeck L, Arlinger J, Erlandson A-C & Jahromi N (1997) Investigation of the potential for microbial contamination of deep granitic aquifers during drilling using 16S rrna gene sequencing and culturing methods. J Microbiol Meth 30: Pedersen K, Arlinger J, Hallbeck A, Hallbeck L, Eriksson S & Johansson J (2008) Numbers, biomass and cultivable diversity of microbial populations relate to depth and borehole-specific conditions in groundwater from depths of 4 to 450 m in Olkiluoto, Finland. ISME J 2: Posiva Oy (2009) Olkiluoto Site Description 2008, vols. I II. Posiva report no (available from Posiva Oy, Eurajoki, Finland. Sharma R, Ranjan R, Kapardar RK & Grover A (2005) Unculturable bacterial diversity: an untapped resource. Curr Sci 89: SKB (2001) Site Investigations: Investigation Methods and General Execution Programme. Report no. SKB-TR (available from Swedish Nuclear Fuel and Waste management Co., Stockholm. SKB (2010) Design and Production of the KBS-3 Repository. SKB technical report no. TR (available from www. skb.se), Swedish Nuclear Fuel and Waste Management Co., Stockholm, Sweden. SKB (2011) Long-term Safety for the Final Repository for Spent Nuclear Fuel at Forsmark: Main Report of the SR-Site Project, vols. I II. SKB technical report no. TR (available from Swedish Nuclear Fuel and Waste management Co., Stockholm, Sweden. Smellie J, Tullborg E-L, Nilsson A-C, Sandström B, Waber N, Gimeno M & Gascoyne M (2008) Explorative Analysis of Major Components and Isotopes SDM-Site Forsmark. SKB report no. R (available from Swedish Nuclear Fuel and Waste management Co., Stockholm. Ström A, Andersson J, Skagius K & Winberg A (2008) Site descriptive modelling during characterization for a geological repository for nuclear waste in Sweden. Appl Geochem 23: Stumm W & Morgan J (1996) Aquatic Chemistry, 3rd edn. John Wiley, New York. Tullborg E-L, Smellie J, Nilsson AC, Gimeno MJ, Auqué LF, Bruchert V & Molinero J (2010) SR-Site: Sulphide Content in the Groundwater at Forsmark. SKB Technical report no. TR (available from Swedish Nuclear Fuel and Waste management Co., Stockholm, Sweden. Vartoukian SR, Palmer RM & Wade WG (2010) Strategies for culture of unculturable bacteria. FEMS Microbiol Lett 309: 1 7. Widdel F & Bak F (1992) Gram-negative, mesophilic sulphatereducing bacteria. The Prokaryotes, Vol. 4. (Balows A, Truper HG, Dworkin M, Harder W & Schleifer KZ, eds), pp Springer-Verlag, New York. FEMS Microbiol Ecol 81 (2012) ª 2011 Federation of European Microbiological Societies